There are a number of persistent anomalies in binding theory. One is the lack of an integrated view of long-distance anaphora. Anaphors generally require an antecedent, but languages show striking differences as to where such antecedents may occur. This volume is a collection of original articles by distinguished contributors on the nature of anaphoric systems in a wide variety of genetically and structurally different languages. It examines the general laws underlying the apparent diversity of data from the perspective of current linguistic theory. There is a surprising degree of convergence in the analyses proposed. A substantive introduction summarizes and discusses the main results, providing an integrative picture of individual and common results. This is the first representative collection of articles on this important topic. It is both conceptually coherent and of fundamental theoretical importance.
Long-distance anaphora
Long-distance anaphora Edited by
Jan Koster and
Eric Reuland Department of Linguistics, University of Groningen
The right of Ihl! Un;)'ersity oj Cambridge to print and sell all malll/er of books
was granted by Henry VIII in 1534.
The Vl/hers;ly has printed alld pi/blished contillllously sillce 1584.
CAMBRIDGE UNIVERSITY PRESS Cambridge New York Port Chester Melbourne Sydney
Published by the Press Syndicate of the University of Cambridge The Pitt Building, Trumpington Street, Cambridge CB2 lRP 40 West 20th Street, New York, NY lOOll, USA 10 Stamford Road, Oakleigh, Melbourne 3166, Australia
© Cambridge University Press 1991 First published 1991 Printed in Great Britain at the University Press, Cambridge
British Library cataloguing ill publication data Long-distance anaphora 1. Anaphora. I. Koster, Jan. II. Reuland, Eric J .
Library of Congress cataloguing in publication data Long-distance anaphora / edited by Jan Koster and Eric Reuland. p. cm. Most of the articles originated as contributions to a workshop organized by the Dept. of General Linguistics of Groningen University, June 18-20, 1987. ISBN 052139111 3 (hardback). ISBN 0 521 400007 (paperback) 1. Anaphora (Linguistics) 2. Government-binding theory (Linguistics). I. Koster, Jan. II. Reuland, Eric J. III. Rijksuniversiteit Groningen. Instituut voor Algemene Taalwetenschap. P299.A5L661991 415-dc20 90-39990 CIP ISBN 0 521391113 hardback ISBN 0 521 400007 paperback
FP
Contents
List of contributors Preface
pageix xi
1.
Long-distance anaphora: an overview Eric Reuland and Jan Koster
2.
Containment and connectedness anaphors Lars Hellan
27
3.
Long-distance reflexives and the typology ofNPs Hoskuldur Thniinsson
49
4.
Contextual determination of the anaphor/pronominal distinction Martin Everaert
77
5.
On the interaction between antecedent-government and binding: the case of long-distance reflexivization Pierre Pica
119
6.
Binding in Polish Ewa Reinders-Machowska
137
7.
Anaphors in binary trees: an analysis of Czech reflexives Jindfich Toman
151
8.
Latin long-distance anaphora Elena Benedicto
171
9.
Prepositions, binding and a-marking Alessandra Giorgi
185
Locality, parameters and some issues in Italian syntax M. Rita Manzini
209
10.
1
vii
viii
Contents
11.
Long-distance binding in Finnish Marlies van Steenbergen
231
The primacy condition of anaphora and pronominal variable binding
245
12.
E. Kiss Katalin 13.
The local nature of the long-distance reflexive in Chinese C.-T. James Huang and C.-c. Jane Tang
263
14.
Anaphors and logophors: an argument structure perspective Tanya Reinhart and Eric Reuland
283
References
323 335
Index
Contributors
Elena Benedicto Universita Autonoma de Barcelona Martin Everaert Institute for Language and Speech, University of Utrecht Alessandra Giorgi IRST, Povo Trento Jan Koster Department of Linguistics, University of Groningen Lars Hellan Department of Linguistics, University of Trondheim C.-T. James Huang Department of Modern Languages and Linguistics, Cornell University
E. Kiss Katalin Linguistic Institute, Hungarian Academy of Sciences M. Rita Manzini Department of Italian, University College London Pierre Pica Department of French, The Pennsylvania State University Ewa Reinders-Machowska Eindhoven Tanya Reinhart Department of Poetics and Comparative Literature, Tel Aviv University Eric Reuland Department of Linguistics, University of Groningen
ix
x
Contributors
Marlies van Steenbergen Dr Neher Laboratory, PTT Telecom C.-c. Jane Tang Department of Modern Languages and Linguistics, Cornell University Jindfich Toman Department of Slavic Languages, University of Michigan Hoskuldur Thniinsson Institute of Linguistics, University of Iceland
Preface
This volume is a collection of original articles on the nature of the anaphoric systems in a variety of languages and from a number of different perspectives. The aim of the editors is to provide a new impetus to the study of long-distance anaphors, a phenomenon with ramifications that are rather puzzling from the perspective of the binding theory in its canonical form (as in Chomsky (1981)). All of the contributions are concerned with extending that theory in a manner that is as restrictive as possible. Some of the solutions solidly remain within the domain of the structural binding theory. In other cases, developing a restrictive theory required recognizing different components in binding, and assigning a specific role to pragmatic factors. Many of the articles converge, however, in important respects, leading to considerable simplification of the overall picture. The patterns found in the languages covered are so consistent that they cannot be accidental (although one would certainly wish to investigate a larger number of languages). Most of the articles originated as contributions to a workshop on long-distance anaphora organized by the Department of Linguistics of Groningen University, 18-20 June 1987. The editors added a first chapter in which the main results of the volume are put together. The original plans were for this chapter to be written by the editors together with Tanya Reinhart, who participated in the workshop as a discussant. At a later stage, it turned out that part of the material collected warranted a separate contribution by Reinhart and Reuland, which is included in chapter 14. Once more, the editors found out that editing a book takes more time than one hopes it will do. A number of minor delays add up, and all of a sudden one finds oneself about a year behind schedule. We are grateful to the authors and to Cambridge University Press for their patience. The workshop from which this volume originated would have been impossible
xi
xii
Preface
without the financial support of the Netherlands Organization for the Advancement of Research (NWO) and of the Faculty of Letters of Groningen University. This support is hereby gratefully acknowledged. We are very much indebted to Hennie Zondervan for the excellent way in which she helped us to organize this workshop. We would also like to thank Liesbeth van der Velden for her help in preparing the references. Eric Reuland Jan Koster
1 Long-distance anaphora: an overview Eric Reuland and Jan Koster
1 Introduction One of the major foci of linguistic research during the last decade has been the development of a theory of binding. Despite a reasonable amount of consensus on major issues, there are a number of persistent anomalies. Especially, an integrated view of so-called long-distance anaphors is lacking. The present book sets out to contribute to the development of such a view. In the individual chapters a number of important issues in the theory of local and long-distance anaphors are analysed. The purpose of this overview is to summarize and interpret the results. In section 2 we provide the necessary background. Section 3 summarizes the individual contributions and puts them into context. Section 4 presents an overview of the facts reported. Section 5 discusses a major result of the book as a whole: the existence of just two main classes of A-anaphors.
2 Binding theory and its parameters The starting point of most current discussions of anaphora is the binding theory (BT) developed by Chomsky in a series of works from 1973 on. (1) gives the formulation in Chomsky (1981). (1) A. B. C.
An anaphor is bound in its governing category. A pronominal is free in its governing category. An R-expression is free.
The definition of 'governing category' is given in (2). (2)
b is a governing category for a if and only if b is the minimal category containing a, a governor of a, and a SUBJECT (accessible to a). 1
1
2
Eric Reuland and Jan Koster
As it has been argued that condition C is not part of the grammar (Chomsky (1982), Reinhart (1983a, 1986)), and none of the contributions in this book discusses it, we will further ignore it. It has been noted for quite some time that for anaphors in many languages condition A does not hold as stated. This is in particular true for long-distance anaphors. There are some well-known discussions in the literature of non-clausebounded reflexives (NCBR) in Icelandic (Thrainsson (1976a,b), Maling (1982) and others), and the Italian anaphor proprio (Giorgi (1984)). This led to a number of proposals to modify the binding theory, for instance Koster (1985, 1987), Manzini & Wexler (1987), and Wexler & Manzini (1987). These theories express two claims: (1) languages may differ in the binding domains of their anaphors; (2) within one language different anaphors may have different binding domains. The common assumption in these studies is that all anaphors share the same type of binding relation. The differences are captured by parametrizing the opacity factor defining the governing category, as in (3). (3)
b is a governing category for a if and only if b is the minimal category containing a, a governor of a, and F (F an opacity factor).
The generally accepted definition of binding is given in (4). (4) a binds b iff a and bare coindexed and a c-commands b. F may assume values such as (accessible) SUBJECT, Tense, Agr, or Compo These opacity factors are taken from a universal set, with particular anaphors differing in the value selected. This choice is represented in the lexical entry. Similarly, languages may differ in the opacity factors they make available. Anaphors with an opacity factor beyond the SUBJECT are classified as long-distance anaphors. So far, however, opacity factors have been represented as arbitrary features; hence, no principled restrictions on the set of opacity factors have been developed. In the absence of such restrictions the theory predicts virtually unlimited possibilities for anaphors to differ. In the survey of the languages discussed in the volume which is presented in section 3, we will see to what extent this prediction is borne out. Condition B raises a question of a different kind, namely, why it is so constant across languages. The original formulation in Chomsky (1981) left some empirical gaps. But Huang's (1982) modification that the opacity factor for pronominals is SUBJECT, rather than accessible SUBJECT, comes close to being descriptively correct. It is presently quite unclear why the opacity factor for pronominals does not vary, and why the complementarity between pronominals and anaphors, which
Long-distance anaphora: an overview
3
is generally quite striking, breaks down in some constructions, especially in languages with long-distance anaphors. Since complementarity facts will bear on the assessment of the anaphoric system in a language, they are included in the survey below.
3 The contributions and their implications Many of the contributions contain features that cannot be easily accommodated in a binding theory modelled on (1-4). However, they are often quite compatible with each other. 3.1 Argument structure and binding One of the important issues in current theory is the relation between argument structure and binding. This issue is addressed in a number of contributions. In chapter 2 Hellan discusses the anaphoric systems of a number of languages, focussing on Norwegian and Icelandic, but also including Italian, Dutch, Japanese and Chinese. He proposes that there are two types of conditions on anaphoric relations: containment conditions and connectedness conditions. Containment conditions are in effect conditions on the binding relation. They can be viewed as generalizations of the c-command requirement in (4). They express that the antecedent a must bear a certain prominence relation to the maximal constituent which contains the anaphor b but does not contain a (max(b)). In the case of c-command this relation is purely structural (sister of). The cases discussed by Hellan involve narrower requirements, for instance, that max(b) must be predicated of a, that max(b) is an argument of the same verb a is an argument of, or the (negative) requirement that max(b) does not contain a Tensed S containing b, etc. Connectedness conditions affect a different part of the binding theory. Hellan argues that for local binding relations binder and bindee must be connected by being co-arguments of a lexical head. This makes the structural notion of a governing category irrelevant for local binding. This modification ties in with other proposals. Everaert establishes that there is a direct connection between a-theory and the selection of anaphors. Kiss shows that anaphor binding must be sensitive to the hierarchy of thematic roles. In a different way, also Giorgi shows that the a-assigning properties of the head governing the anaphor are crucial. A synthesis of these findings is attempted by Reinhart and Reuland, who propose that local binding involves the manipulation of the thematic grid ofthe predicate of which binder and bindee are arguments. Hellan also discusses discourse-dependent anaphora, involving the condition
4
Eric Reuland and Jan Koster
that max(b) must be in the perspective of a. b is in the perspective of a if a is the person from whose point of view the report containing b is made. This instantiates the more general phenomenon of logophoricity: the element binding b must bear one of a number of specific roles in the discourse structure (see Sells (1987) for an analysis of the notion of logophoricity, and Reinhart & Reuland for further discussion in the present book). Logophoric binding does not require that the antecedent is structurally represented in the sentence, so the anaphoric element is allowed to be free (as discussed by Thniinsson). 2 It seems that logophoric binding relations fall outside the domain of the structural binding theory, but their existence is important for determining the number of binding domains that have to be distinguished and the nature of their difference. Standard binding theory just takes it for granted that there is a distinction between anaphors and non-anaphors. But Thrainsson, in chapter 3, provides a detailed analysis of the referential properties of NPs introducing much finer distinctions, which standard binding theory is unable to deal with. He provides a typology of reflexives based on comparison of their behaviour in the group of Scandinavian languages. This leads to a theory in which the anaphoric character of an NP can be accounted for in terms of specific lexical properties. Like Hellan, Thrainsson shows that structural requirements represent necessary, but not always sufficient, conditions on anaphoric binding. By careful consideration of the data he leaves no doubt that in certain cases of LD-reflexives the binding requirements cannot be stated in purely structural terms, and that some need not even be syntactically bound at all. The characteristic of such reflexives is that they are incapable of independent reference for purely lexical reasons. Instances of reflexives that are bound from outside a finite clause, and those that , are not syntactically bound at all, form a natural class. This leads to a typology in which the feature [+/- independent reference] is a separate parameter. In the binding theory of (1-4), the complementarity between pronominals and anaphors is accounted for in a largely stipulative manner. One of Everaert's major insights is that complementarity phenomena are essential for our understanding of how binding works. In his contribution in chapter 4 he discusses facts from a wide range of languages while focussing on a comparison of Dutch and Frisian. He notes that a coherent notion of long-distance anaphora has not been established so far, and starts from the null hypothesis, namely that one binding theory suffices for non-Iogophoric anaphors. The distinction between local and non-local anaphors is represented in the requirement that certain anaphors must be free in their minimum governing category, and in the conditions on complementarity between anaphors and pronominals. Of special importance is the fact that the non-local anaphor zich in Dutch may be locally bound in the same environments in which
Long-distance anaphora: an overview
5
the pronominal 'm may be locally bound in Frisian. His finding that the relevant conditions are stated in terms of 9-government constitutes a major step forward. It provides a crucial link between binding theory and argument structure. These results are quite compatible with much of Hellan's approach, and can be viewed as filling in the connectedness module. Although Everaert's contribution itself still uses the notion of a governing category, it effectively contains the considerations needed to abolish it as an independent theoretical concept. His contribution contains an illuminating discussion of reciprocals. The relation between binding and argument structure is also the topic of Kiss's contribution in chapter 12. Kiss discusses anaphora and variable binding in Hungarian, comparing these with their counterparts in English. Kiss argues that due to the flat structure of Hungarian all arguments in a clause c-command each other. Yet, there are binding asymmetries mirroring those of English. Kiss concludes that in general the binding hierarchy is only in part a reflection of formal properties of the structure, such as c-command and precedence. Relative prominence of roles in the thematic lexical argument structure must be another important factor. She argues that this factor is operative not only in Hungarian, but also more generally. The asymmetries among PP complements in English (with NP versus about NP) cannot be explained in terms of differences in c-command, but must be stated in terms of such a hierarchy. Giorgi's contribution in chapter 9 analyses anaphora in PPs, focussing on the long-distance anaphor proprio and the local anaphors se and se stesso. 3 Proprio can be clause bound, or LD-bound. In the latter case it is subject oriented. Se is always subject oriented. It cannot be governed by a verb, but must be governed by a preposition; se stesso is clause bound, but without further restrictions. Proprio is like se when LD-bound, but like se stesso when clause bound. When subject bound these anaphors are not in complementary distribution with pronominals. The properties of se are reminiscent of Dutch zich. This suggests a connection with the theory of zich developed by Everaert (see Reinhart & Reuland for further discussion) . Discussing anaphora in structures of the form [NP XP], Giorgi makes the following important observation. If the XP contains a subject-oriented anaphor and it is an AP, the NP is a possible antecedent, but if the XP is a PP it is not. On the basis of an extensive discussion of 9-marking properties of APs and PPs, she argues that the effect is due to a difference in the following respect: APs may assign an external 9-role, PPs cannot. Hence, when construed with a PP, the NP cannot count as prominent in the relevant sense. The relation between the NP and the PP cannot be the formal predication relation. Interpreted along these lines her contribution provides further support for the 9-related nature of anaphora. There is also
6
Eric Reuland and Jan Koster
another possible interpretation: APs exhibit agreement, and PPs do not. If subjectoriented anaphors required an Agr, the same pattern would follow. Establishing which is the correct factor will be important for a proper understanding of the Italian system of anaphors. 3.2 Binding and movement Standard binding theory has little to say on the nature of the opacity factors. It essentially stipulates that certain categories are opaque for binding. In recent years, movement theory has become increasingly articulated. Building on earlier proposals (Lebeaux (1983), Chomsky (1986a)), Manzini, Pica, and Huang & Tang set out to derive a number of properties of binding from the theory of movement discussed in Chomsky (1986b). These proposals deal with the mechanics of binding, i.e. the question of how coindexing is brought about. The discussion in Hellan, Thniinsson, Everaert, Kiss, and Giorgi involves higher-level issues. It is an important question to what extent the various higher and lower-level analyses are compatible. Faltz (1977) and Pica (1985, 1987) have observed that there are systematic differences between simplex (mono-morphemic) and complex anaphors. Anaphors of the former type may be non-local, those of the latter type are always local. In his earlier work Pica attempted to give a full explanation of binding in terms of movement. In his present contribution (chapter 5), Pica argues that a full reduction cannot be achieved, but that the differences between local and non-local binding follow from the way in which the theory of movement proposed in Chomsky (1986b) interacts with the binding theory proper. Cross-linguistic variation between reflexives can be reduced to differences in their lexical properties. No reference to parameters is necessary. Central is the idea that reflexives are subject to cyclic movement from Infl to Infl through Compo Mono-morphemic anaphors like sig are full-argument NPs; hence the term 'argument anaphors'. If at LF they move up to a position governed by their prospective antecedent, intermediate traces may delete. So, this movement will not be generally clause bounded. In the case of complex anaphors like himself (adjunct reflexives in his terms), the specifier him moves up. Him is an adjunct, hence intermediate traces may not delete. Therefore, movement will be clause bounded. Similar considerations apply to clitic reflexives, which are not arguments either. a-theory and argument structure are only indirectly involved. It is not clear how Pica's approach is to be reconciled with the facts and insights put forward by Hellan, Everaert, and Kiss. In chapter 10 Manzini investigates the hypothesis that the binding conditions reflect conditions on movement, and that all locality conditions can be ~nified under the notion government. In particular she investigates whether the natural
Long-distance anaphora: an overview
7
minimal binding category for an anaphor is the first maximal projection containing it (see also Koster (1987». She argues that binding of anaphors in verb complement position can be brought under that conception, but for binding of anaphors contained in NPs presence or absence of a subject is relevant. Also, the fact that anaphors may differ as to their binding domain shows that opacity factors must enter into domain specifications to express the parametrization (see (4». Her analysis is based on a detailed discussion of inalienable possession and reciprocals in Italian. She shows that when l'uno l'altro 'each other' is part of an NP, the opacity conditions on l'altro are subject based, those on l'uno barrier based. She then argues that l'altro is an A-anaphor, while l'uno is an A-bar anaphor. Pica and Manzini converge in the conclusion that a full reduction of binding theory to principles of movement has not been achieved. As in the case of Pica, it is still to be determined to what extent Manzini's approach ties in with the results obtained by Hellan, Everaert, and Kiss. This is especially relevant, since these authors have nothing to say about A'-anaphors. However, if it is in general true that A'-anaphors are sensitive to barriers and A-anaphors to other conditions, this raises the question to what extent the theory of the two should be unified. As we noted earlier, most theories have nothing specific to say on the question of what makes an anaphor into an anaphor. To our knowledge, Bouchard (1984) has been the first to claim that NPs must have
8
Eric Reuland and Jan Koster
LF. So, LF-movement into the domain of a higher antecedent does not cause a change of referential index. Hence it is allowed. Just like the referential index, 'Pfeatures once assigned may not be changed. LF-movement into the domain of a higher subject will proceed through the domains of the subjects of intervening clauses. If any of those bears a different set of 'P-features this would cause the features of ziji to change, in violation of the requirement that once assigned they may not change. Hence, intervening subjects must all agree in 'P-features. Both Huang & Tang and Pica assume, with Manzini, that an anaphor must be governed by its antecedent. If at S-structure it is not, it must move up at LF. They assume different mechanisms, though. Huang & Tang subsume non-local anaphora under QR (adjunction to IP). Pica assumes that anaphors are ditics and hence undergo ditic movement (adjunction to I). The most striking difference involves the explanation of the locality of complex anaphors. According to Huang & Tang it is because the anaphor contains an element with 'P-features, according to Pica because it contains an adjunct, which is sensitive to barriers. Like Pica's, Huang & Tang's proposal does not easily accommodate the 0theoretic insights of Hellan, Everaert, and Kiss. Both are intriguing proposals, which require detailed further investigation. 3.3 Properties of anaphoric domains In order to determine the nature of anaphoric domains it is helpful to further investigate a number of languages. The anaphoric system of Polish, discussed by Reinders-Machowska in chapter 6, comes out as rather prototypical. The author discusses the argumental anaphor siebie and the possessive/adjectival anaphor sw6j. Despite appearances, there are two relevant binding domains, as in Finnish. The smallest maximal projection containing both the anaphor and a subject is the local domain; the smallest maximal projection containing the anaphor and Tense is the extended domain. In addition to local binding there is binding across a subject. The separate character of the local domain is obscured by the fact that there is no overt distinction between local and non-local anaphors. Yet, a local domain must be distinguished, as the reciprocal interpretation of siebie and sw6j (available in addition to the standard reflexive interpretation) is only possible within the local domain. Furthermore, anaphors and pronominals are in complementary distribution in the local domain, but not when binding obtains across a subject. As a matter of terminological convenience, we will henceforth refer to binding across subjects as medium-distance binding. In chapter 7, Toman discusses reflexivization in Czech. There is a strong reflexive sebe, a ditic reflexive se, and a possessive reflexive svUj. In Czech these anaphors
Long-distance anaphora: an overview
9
constitute a quite homogeneous group. For all of them the binding domain is the smallest maximal projection containing an accessible subject. So, they qualify as local under the present perspective. They induce the same requirements on possible antecedents. Toman shows that Czech anaphors are all subject oriented. Apparent object antecedents must be construed as subjects of small clauses, along the lines of Kayne's (1981) small-clause approach to argument structure. Unlike in Polish, infinitival clauses are opaque. Small clauses are not, however. Toman reconciles this with the role of the subject in defining the binding domain by proposing that subjecthood is optional when it is not forced by other factors. This is independently necessary for binding within NP. Subjects of NPs are possible antecedents, but do not cause opacity. The reason is that in Czech such subjects are adjectival. Either argumenthood or adjectivality may prevail. The behaviour of the clitic anaphor se seems more complicated. However, Toman convincingly shows that its special behaviour follows from general properties of clitics (e.g. that they must be licensed by Comp or Infl) interacting with a binding requirement which is identical to that of sebe. It should be noted that the Czech reflexives are subject oriented, although they only allow local binding. This is a matter for further investigation. Investigating an extinct language poses a special challenge. In chapter 8 Benedicto presents an attempt to investigate the anaphora system of Latin. Although an unequivocal interpretation of the data cannot always be guaranteed, Benedicto makes clear that se can at least be bound across subjects, i.e. it is a mediumdistance anaphor. Even in non-local contexts se can be bound by direct objects, though. This would seem to go against the general pattern that medium- or longdistance anaphors can only be bound by subjects. It becomes less surprising, however, if one takes into account that se can even be bound by various topicalized material. This suggests that in some of its uses se falls under conditions on logophoricity rather than binding conditions in a strict sense. The properties of the anaphoric system in Finnish are discussed by Van Steenbergen in chapter 11. There is a reciprocal, toiset, two lexical reflexives, itse and han itse, and an empty reflexive possessive, e. Finnish has two binding domains: the smallest maximal projection containing the anaphor and a subject and the smallest maximal projection containing the anaphor and Tense. The former has the properties of a local domain, the latter is a medium-distance domain. As usual, the reciprocal must be bound in the local domain. Similarly, anaphors and pronominals are in complementary distribution only under local binding. Itse, han itse and e can all be bound outside the local domain; the antecedent must be a subject. Itse and e can also be bound within the local domain, but only by the subject. Also han itse can be bound within the local domain, but in that case only by a non-subject.
10
Eric Reuland and Jan Koster
Van Steenbergen proposes a general distinction between two modes of binding: local binding, which involves the direct association of an anaphor with its antecedent, and long-distance binding (medium-distance binding in our terms, since no logophoricity effects can be observed), which involves the formation of a chain of elements linking the local domain to the domain of the antecedent. This chain requires a dependence between the Infl-nodes heading the Ss crossed. The fact that intermediate Infls are crucially involved in this link explains the privileged status ~f the subject as an antecedent (as only the subject is governed by Infl), and also the fact that the local subject cannot bind an anaphor in an adverbial clause (the Infl governing the local subject does not c-command the Infl of the adverbial clause). These results can be directly related to the approach in Reinhart & Reuland; binding in the local domain involves argument binding via the a-grid; medium-distance binding is sensitive to Tense/Agr domains. By way of conclusion, in chapter 14 Reinhart & Reuland sketch an alternative conception of the binding theory. This proposal is based on a number of results obtained inthe various preceding chapters. Its main claims are the following. The complementarity between pronominals and anaphors in the local domain follows from a general condition on a-grids, applied to predicates expressing a reflexive relation. The local character of complex anaphors follows from the conditions on the interpretation of 'self'. The subject orientation of medium-distance anaphors follows from the fact that they must obtain
4 Surveying anaphoric domains 4.1 Background The properties of long-distance anaphors (LD-anaphors/LDAs) as discussed in the current literature can be summarized as in (5), with (SA) as the initial defining characteristic. (5) A. B.
LD-anaphors allow an antecedent outside the governing category as defined in (2). The antecedents of LD-anaphors are subject to a more restrictive promi-
Long-distance anaphora: an overview
11
nence condition than c-command. The most common requirement is that the antecedent must be a subject. C. LDA is restricted to reflexives. Reciprocals are not allowed as LDanaphors (Yang (1984». D. LD-anaphors are morphologically simplex. Morphologically complex anaphors are local (Pica (1985, 1987». E. Outside the local domain there is no complementarity between pronouns and anaphors. In this section we survey the languages discussed in the various chapters. Where necessary for a more complete picture, we also refer to other works. Issues we could not resolve on the basis of the available literature are marked? Together with c-command it implies that possibly a more restrictive notion applies. In general, it should be noted that the discussion in the literature is less systematic than one would have expected. So, the chart in section 4.2 should also serve to stimulate research to fill in the gaps. For each language we give the domains relevant for binding, the anaphors which can be bound in these domains, the prominence requirement to which they are sensitive, and whether they show complementarity with respect to pronominals. In order to facilitate comparison of the different languages, the terminology used in this overview is unified as much as possible, at the risk of some oversimplification. The reader should therefore always consult the original contribution for the precise picture. The domain specifications are given in terms of the definition of governing category in (3). Domain 1 is the governing category with (accessible) Subject as the value of F, domain 2 the governing category with finite Infl as the value of F. The domains are given in the order of increasing size. The specific characteristics of the larger domain will only show up in the area in which it does not overlap with the smaller one. We do not represent variation in the role of the accessibility condition, as none of the contributions discusses this. Note that finite Infl covers both subjunctive and indicative domains. The largest domain is specified as 'beyond domains 1 or 2', without an opacity factor indicated. Binding relations across such domains turn out to involve logophoricity, as discussed in Reinhart & Reuland. There is reason to assume that all languages exhibit some logophoricity effects. They are specifically mentioned, however, only when the discussion refers to it. There is extensive discussion of prominence conditions in the various chapters, especially in the contributions by Hellan, Kiss, and Reinhart & Reuland. In the overview, we use c-command in the standard BT sense, disregarding the problem
12
Eric Reuland and Jan Koster
posed by binding from within PPs, and the fact that it reduces to role command in the domain in which the latter applies (see Reinhart & Reuland). The requirement of role command means that the antecedent must bear a thematic role which ranks higher on some scale of thematic roles than that of the bindee (see Hellan and Kiss for more discussion). Where c-command is not sufficiently discriminating, role command is used explicitly. As indicated in (5), it is commonly held that the antecedent of a long-distance anaphor must be a subject in some sense. The use of the notion 'subject' varies, however. For some authors it means I-subject in the sense of Borer (1986a). For others, any NP is a subject if there is some XP predicated of it, including indirectobject constructions as analysed by Kayne (1981). In the overview the term is used descriptively: subject stands for c-commanding subject of an IP or NP. The overview also uses the notion of predication command (from Hellan): a predication commands b when b is contained in a constituent which is predicated of a .. The requirement that the antecedent be an I-subject often goes together with predication command (but not always, think of adjuncts). The converse does not hold, however. So, these requirements are independent. When there is no subject requirement, but a predication requirement is still involved, the term predication command will be explicitly used. Finally, the requirement of perspective command, discussed by Hellan, expresses that the antecedent must be a perspective holder in the associated discourse environment (see also Reinhart & Reuland).
4.2 The survey
Icelandic Discussed by HeJlan, Thrainsson, and Everaert; see also Maling (1982, 1986), Anderson (1986), Rognvaldsson (1986), and Sigurosson (1986a, b).
Anaphor
Prominence factor of antecedent
Domain 1: first (accessible) Subject sjdlfur sig predication command: 'him self' subject
Complementary with respect to pronominals
yes
hann sjdlfur 'him self'
c-command?: non-subject
yes
sig/sin 'himself/his'
predication command: subject
yes
Remarks
Long-distance anaphora: an overview hvor annar
c-command?
13 yes
'each other'
sig/sin
Domain 2: first finite Inll beyond domain 1 predication command: 'himself/his' subject
yes
Domain 3: beyond domain 2 perspective command 'himself/his'
no
sig/sin
1. The grouping of the data here differs from what has become the standard view of Icelandic. It reflects an insight which has already been expressed in Anderson (1986), namely that infinitival domains should not be grouped together with subjunctive domains. Together with the facts discussed by Hellan, this leads to a picture in which, structurally, subjunctive and indicative domains constitute one group. The literature gives very little information on sjdlfur sig and ham! sjdlfur, except that they are locally bound. None of the literature consulted explicitly gives the environments in which sjdlfur sig occurs and sig is excluded. According to Maling (1982) the conditions on sjdlfur are less clearly part of the grammar than the conditions on selv in Norwegian. Hellan (1988), however, claims that the conditions on the various types of local anaphors in Icelandic are similar to the conditions on the corresponding items in Norwegian. Locally bound sig would then have a distribution similar to Dutch zich. Further research of the environments favouring the sjdlfur anaphors is necessary. Note that, since domain 3 includes the domains 1 and 2, and since interpretation of sig/sin under perspective command does not always require a subject, under the relevant conditions apparent violations of the requirement that the binder be a structural subject may also be observed in the latter two domains; in such cases there is no complementarity. The same may obtain for Norwegian seg.
Norwegian Discussed by Hellan, Thniinsson, and Everaert.
Anaphor
I
Prominence factor of antecedent
Complementary with respect to pronominals
Domain I: first (accessible) Subject predication command: 'him self' subject
yes
ham selv
c-command?: non-subject
yes
predication command: subject
yes
seg selv
'him self'
seg 'himself'
Remarks
14
Eric Reuland and Jan Koster
'his'
predication command: subject non-subject
yes no
hverandre
c-command?
yes
Sin
'each other' Domain 2: first finite Inll beyond domain 1 predication command: subject
seg/sin
no
Dutch Discussed by Everaert and by Hellan; see also Koster (1985, 1987) and Rullmann (1988). Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject c-command 'him self'
yes
2
zich
subject
yes/no
c-command
yes
c-command
yes
Anaphor
Prominence factor of antecedent
zichzelJ
'himself'
'mzelf 'him self'
elkaar 'each other'
Domain 2: first finite Inll beyond domain 1 subject 'himself'
zich
3 no
2. Zichzelf, 'mzelJ and elkaar must all be bound within domain 1, zich may be bound in that domain. They differ in the thematic conditions on the relation they bear to their antecedent. These conditions are discussed in more detail by Everaert and by Reinhart & Reuland. They can be summarized as follows: zichzelf and its antecedent have independent thematic roles; the antecedent of 'mzelJ is not an argument of the latter's governor; zich occurs in all other positions. 4 Zich is not complementary with respect to pronominals when it occurs in PPs, in the other positions it is. Note thatzichzelf, zich and 'mzelJare third person only. First and second person pronouns lack a special form corresponding to zich; in the relevant positions simply the pronoun is used. In positions otherwise requiring zichzelf, zelJ is attached to the regular pronominal form as in English. 3. Observations would suggest that this statement should refer to first Inll, rather than first
Long-distance anaphora: an overview
15
finite Infl. In Reinhart & Reuland it is argued, however, that infinitival clauses are opaque for independent reasons.
Frisian Discussed by Everaert. Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject c-command
yes
4
['m/him]
n.a.
n.a.
inoar
c-command
yes
Anaphor
Prominence factor of antecedent
'msels
Domain 2: beyond domain 1 none, except possible logophoric expressions not discussed 4. The element listed as 'm/him enclosed in brackets is homophonous to the corresponding pronominal in all persons, numbers and genders. 'm/him is just the third person singular member of this paradigm. In inherently reflexive constructions it has an obligatory bound variable reading, hence its inclusion here. In other contexts it behaves as a pronominal but is compatible with such a reading. The distribution of the Frisian forms X versus Xsels mirrors that of Dutch zich versus zichzelf The system seems similar to that of Dutch, but simplified.
English Discussion passim. Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject c-command
yes
5
each other
yes
Anaphor
Prominence factor of antecedent
himself
c-command
Domain 2: beyond domain 1 none, except logophoric expressions 5. In sentences such as Peter saw the snake near him and John too, him behaves very much as its Frisian and Dutch counterparts 'm and zich. However, unlike those there are hardly any contexts in which English him has an obligatory bound variable reading, hence it would not seem appropriate to classify it as an anaphor as well.
16
Eric Reuland and Jan Koster
Polish Discussed by Reinders-Machowska, Everaert, and Toman. Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject c-command siebierefl 'himself'
yes
6
c-command
yes
c-command
yes
7
Domain 2: first independent Tense beyond domain 1 no c-command 'himself'
8
Prominence factor of antecedent
Anaphor
siebierecipr 'each other'
sw6j poss
siebierefi sw6j
c-command
no
poss 6. Given the structures assumed in Reinders-Machowska only the subject will c-command the reflexive, hence a subject requirement would be descriptively equivalent. 7. Abstracting away from a restriction against the antecedent being the specifier of a superordinate NP. 8. If participial clauses are assumed to have Agr, depending on one's theory of finiteness, domain 2 could also be characterized as the domain of the first finite inflection, just as in the previous cases.
Czech Discussed by Toman. Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject predication command selsebe 'himself'
yes
9
svUj
yes
Anaphor
Prominence factor of antecedent
predication command
'his' Domain 2: beyond domain 1 none, except possible logophoric expressions not discussed
Long-distance anaphora: an overview
17
9. Toman formulates the prominence factor as a subject requirement. Objects are possible antecedents if the anaphor is in a constituent that can be construed as predicated of the object. Hence, Hellan's notion predication command seems quite apt here. For independent reasons, the clitic anaphor se is restricted to a subset of the environments of sebe.
Latin Discussed by Benedicto.
Anaphor
Prominence factor of an teceden t
Domain 1: first (accessible) Subject c-command: se subject 'himself'
Complementary with respect to pronominals
yes
10
Domain 2: beyond domain 1
se
predication command
Remarks
no?
11
'himself'
10. The facts discussed do not allow a decision as to whether domain 2 actually constitutes a single domain, or should be split up. Specifically, it is not clear how and when topicalized NPs qualify as antecedents. 11. Equivalently, the antecedent must be a governing subject in Koster's (1987) sense, if this notion is suitably extended so as to take into account certain thematic considerations.
Italian Discussed by Giorgi and by Manzini; see also Napoli (1979) and Hellan.
Anaphor
Prominence factor of antecedent
Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject c-command 'himself'
yes
si
subject
yes
S8 'himself'
subject
no
13
l'uno l' altro
c-command
yes
14
c-command
yes?
se stesso
12
'himself'
'each other'
proprio 'own'
18
Eric Reuland and Jan Koster
Domain 2: beyond domain 1 subject/role command? se 'himself'
proprio
15 no
role command?
'own'
12. Giorgi (1987) argues that 'Minimal Complete Functional Complex' rather than 'first accessible subject' is the required notion. See Reinhart & Reuland for discussion. 13. Si is a ditic linked to a verbal argument; se is a non-ditic, occurring in positions where cliticization is not obligatory (or where it is impossible), for instance in the domain of prepositions. 14. As is shown by Manzini, strictly speaking, only l'altro is an A-anaphor and associated with a locality domain with the subject as its opacity factor. L'uno is an A'-anaphor and sensitive to canonical barrierhood. 15. Since no anaphora appears to be possible beyond indicative clause boundaries, one might consider having indicative Ina as an opacity factor, yielding first indicative Ina beyond domain 1 as the domain definition. However, a number of restrictions discussed in Giorgi (1984) and Napoli (1979) indicate that LDA in Italian, as in Icelandic, involves logophoricity, rather than strict anaphoricity; this is in line with suggestions by Sells (1987) and Hellan. For more discussion, see below. As we see it, the opacity of the indicative domain may follow from general conditions on perspective, and need/should not be stated as an opacity factor. At this point one cannot exclude the possibility that domain 2 should be further divided (as in Icelandic). There is an observation in Napoli (1979) to the effect that a sentence only violating the SSC is preferred over one in which both the SSC and the Tensed S condition are violated. More systematic research is needed for a decision.
Finnish Discussed by Van Steenbergen.
Anaphor
Prominence factor of antecedent
Domain 1: first (accessible) Subject c-command: lziin itse non-subject 'himself'
Complementary with respect to pronominals
yes
itse 'himself'
subject
yes
'zero' 'his'
subject
yes
toiset 'each other'
c-command
yes
Remarks
Long-distance anaphora: an overview
19
Domain 2: first finite Infi beyond domain 1 subject hiill itse 'himself'
no
itse
subject
no
subject
no
16
'himself' 'zero' 'his'
16. Van Steenbergen gives Tense as the relevant factor; finite Infl is equivalent for the present discussion. Note that LD-binding of hiill itse goes against Pica's generalization, unless it turns out to be logophoric.
Hungarian Discussed by Kiss; see also Manicz (1989). Complementary with respect to pronominals
Remarks
Domain 1: first (accessible) Subject role command 'himself'
yes
17
egymas
yes
Anaphor
Prominence factor of antecedent
maga
role command
'each other' Domain 2: beyond domain 1 none, except possible logophoric expressions not discussed 17. The requirement of role command expresses that the antecedent must bear the thematic role which ranks higher on some scale than that of the anaphor. Given the lack of sufficient structure in Hungarian, c-command would not sufficiently discriminate. Note that, in addition to the anaphors mentioned, Hungarian has reflexive morphemes; these are not discussed here (cf. Manicz (1989)).
Chinese Discussed by Huang & Tang; see also Aoun & Li (to appear).
Anaphor
Prominence factor of antecedent
Domain 1: first (accessible) Subject c-command: 'himself' subject
taziJi
Complementary with respect to pronominals
yes
Remarks
20
Eric Reuland and Jan Koster
ziji 'himself'
c-command: subject
yes
Domain 2: beyond domain 1 c-command: subject 'himself'
ziji
18
18. Subject to a local licensing condition discussed in Huang & Tang.
4.3 Discussion By and large we see that the facts found in the survey are in conformity with the correlations expressed in (5). There are systematic differences between the anaphors that must be bound within domain 1, and those allowing an antecedent beyond it. A simple c-command requirement is never sufficient for binding beyond domain 1, although it is generally not sufficient for local binding either. None of the languages allows a reciprocal to be bound outside domain 1. Non-Iogophoric LD-anaphors are not morphologically complex, although some of the local anaphors are simplex too. Outside domain 1 there is generally no complementarity between pronouns and anaphors, with the exception of Icelandic, where the complementarity only disappears in domain 3. Note, however, that complementarity is not even always enforced in domain 1. A theory of anaphora should be able to explain these facts (see Reinhart & Reuland for an attempt). In the present discussion we will concentrate mostly on the status of the binding domains and on the relation between the selection of a domain and the basic characteristics of anaphors. We will not discuss reciprocals here. Instead we refer to the contributions by Everaert and by Manzini, and to Heim, Lasnik & May (1988). We just wish to note that the case of Polish , where the reciprocal and the reflexive are realized by the same lexical item, indicates that the restriction of reciprocals to domain 1 must be closely connected to the interpretation they receive, along the lines set out by Everaert.
5 The binding domains: a result One of the most surprising facts about anaphora is that there is a systematic distinction between local and long-distance amphora at all. As noted earlier, the parametric approach based on the definition of governing category in (3) predicts virtually unlimited possibilities for anaphors to differ. Even if the languages included in this survey are limited in number, the absence of variation is striking. The number of different opacity factors turning up in the
Long-distance anaphora: an overview
21
domain definitions is extremely small. The smallest domain found is the one with the (accessible) subject as its opacity factor. The second domain is characterized as the domain of a finite Inti. The third domain is not characterized by an opacity factor, but is just anything beyond the second domain. Some languages do not employ either the domain of the subject or that of finite Inft, as we saw. In these it is the second domain which lacks an opacity factor. Then it is just characterized as anything beyond the first domain. The result that there is just a small number of structurally delimited binding domains need not come as a surprise, reconsidering previous investigations. On closer scrutiny, the opacity factors listed in Koster (1987), Wexler & Manzini (1987), and Manzini & Wexler (1987) fall into just two natural classes: (accessible) Subject on the one hand, and Agr, Tense, Inft, Comp on the other. This division correlates with a distinction between minimal predication and minimal proposition. Local are the anaphors bound in the minimal predication, long distance those allowed to be bound in a larger domain. This interpretation of the facts may perhaps seem bold, as in the last few years the number of different governing categories established for some languages tended to increase with the growth of detailed factual knowledge. We believe, however, that in all of these cases other factors are involved. For instance, Manzini (this volume) convincingly argues that l'uno (in the reciprocall'uno l'altro) and the anaphoric determiner in constructions of inalienable possession must be bound in the first maximal projection which is a barrier. This is a smaller domain than our domain 1. This fact is irrelevant for our generalization, however, as these two anaphors are A' -anaphors, whereas the generalization applies to A-anaphors. In Koster (1985) it is argued that an analysis of Dutch anaphors requires reference to both a smaller (opacity factor P) and a larger domain (factor Comp). The preposition domain was instantiated by locational and directional PPs, which were seen as governing categories in order to block zichzelf Reference to thematic properties is already implicitly present in this account, and is in fact sufficient by itself. In other environments the choice between zich and zichzelf is also determined by thematic factors (see Everaert), as is the distribution of'mzelf, Hence, there are no reasons to include (subjectless) PPs among the governing categories. The Comp-domain has been motivated by the fact that zich may be bound outside perception verb complements (PVC) (which lack a Comp), but ~ot outside infinitival clauses (which have a Comp). However, as is argued by Reinhart & Reuland, there is an independent reason that zich appears to behave differently in PVCs and infinitival clauses. The second domain is defined by the first finite Inti, as we saw, and can be
22
Eric Reuland and Jan Koster
characterized as a minimal proposition. Polish may turn out to minimally differ, but still the domain defined virtually coincides. That is, if there is variation it appears to reside not so much in a factor picked out by the anaphor, but in the category construed as a minimal proposition. Current views on the nature of the parametrization of long-distance anaphors have been heavily influenced by the fact that certain elements allow binding beyond domain 2. Much of the discussion has concentrated on Icelandic seglsin and Italian proprio. This type of binding appeared to be sensitive to yet another factor, namely Mood, and thus seemed to fit in only with a parametric theory along the lines sketched above. So, it was found that seglsin can be bound across subjunctives, and claimed that it is long-distance anaphors which must be bound in the domain of the minimal indicative Tense. Similar facts were discovered about Italian proprio and discussed in Giorgi (1984). Hellan and Thniinsson convincingly show, however, that the distinction between subjunctive and indicative clauses does not involve an opacity factor in the sense adopted. They show that siglsin also allow binding out of indicative clauses, given suitable pragmatic conditions, and in fact need not be syntactically bound at all. Mood, as they argue, plays a role only indirectly. What is involved in the binding of siglsin is that the antecedent must be a perspective holder; it must 'perspective command' the anaphor. The use of the subjunctive has just a signalling function concerning the perspective of a clause. This is quite in line with the earlier conclusions in Thrainsson (1976a, b) and Maling (1982) that siglsin has logophoric properties and with more recent results in Anderson (1986). The logophoric use of anaphoric pronouns with concomitant long-distance binding effects is not a restricted phenomenon. It is also possible in languages which otherwise only allow local binding. English, for instance, exhibits many prima facie exceptions to the standard binding theory (Ross (1970), Cantrall (1974), ZribiHertz (1987)). The essence of the claim that a certain type of binding relation involves logophoricity is that discourse factors enter into it. That is, syntactic conditions are not sufficient. This does not entail that discourse properties yield sufficient conditions, or that the element involved need not be syntactically bound as well. In fact, as is shown in Reinhart & Reuland, the exceptions to the standard binding theory in English are subject to clearly syntactic constraints. The point is that it would be a mistake to incorporate all constraints on logophoric relations into a structural binding theory. Italian LD-anaphors constitute a case in point. The domain of indicative Tense is opaque for LDA in Italian. This might seem to necessitate including indicative Tense among the opacity factors. There is evidence, however, that Italian LDA have logophoric properties. LD se, for instance, is best when coordinated with
Long-distance anaphora: an overview
23
another NP (Giorgi (1984)). So, it is like the logophoric 'long-distance' himselfin English, and must fall under the same theory (see Reinhart & Reuland for a theory of logophoricity which accounts for this restriction). Napoli (1979) mentions many cases where the judgements on LD se are affected by the point of view expressed. Italian proprio has logophoric properties too (Sells (1987), see also Hellan). If the Italian LDAs are logophors indeed, the relevance of subjunctive versus indicative falls more readily under a theory of perspective, and indicative should not be represented as a structural opacity factor. The fact that Icelandic logophors have more freedom than their Italian counterparts with respect to crossing indicative tense boundaries may be related to the fact that they have more freedom of interpretation in other respects as well. Italian logophors must be syntactically bound, whereas Icelandic logophors need not be. We conclude that Italian does not provide grounds to complicate the simple theory of structural binding domains which has emerged. It is important to note that 'free' anaphors or logophors (like Icelandic sinlsig or English first or second person free anaphors) are completely beyond the descriptive range of a structural binding theory with opacity factors. This follows from the way in which the governing category of an anaphor is calculated. Unless the factor F in (3) has a value, the governing category of an anaphor is just the minimal maximal projection containing it. The largest domain to be specified in this manner is the root cause (although even this domain can only be obtained under the stipulation that root Tense is structurally distinct from just indicative Tense). The fact that certain logophors need not even be bound in the root dause implies that there are no structural limitations to their domain. Hence, no opacity factor should be specified, but under (3) this gives the wrong result as not specifying an opacity factor gives the minimal, rather than a maximal value for the governing category. Therefore, such a binding theory cannot handle logophoricity in its general form at all. 5 We can now formulate one of the main results of this book, which is in line with the three-way split of the anaphoric domains in Icelandic already discovered by Anderson (1986): (A) There are just two different domains for A-binding: 1. a local domain, characterized by the minimal (accessible) subject as its opacity factor; 2. an extended domain, characterized by the minimal finite inflection as its opacity factor. (B) There is no structurally definable domain beyond the extended domain. Binding with the antecedent in the local domain will be referred to as local
24
Eric Reuland and Jan Koster
binding. Binding with the antecedent in the extended domain is medium-distance binding. Referential dependence beyond those domains (,real long distance binding') involves logophoricity. The results indicate that a binding theory based on parametrization of opacity factors is insufficiently restrictive. Many more questions should be answered by an explanatory theory of binding. Among those are the following; 1. 2.
3.
Why are Subject and finite Inft opacity factors? Under what conditions is there complementarity between pronominals and anaphors, and why is this so? What is the source of the subject orientation of medium-distance anaphors?
In the chapter by Reinhart & Reuland possible answers are discussed, including a proposal to account for logophoricity.
Notes 1. Accessibility is defined as in (i). (i) a. *[c ... d ... J where c and d bear the same index. b. a is accessible to b if and only if b is in the c-command domain of a and assignment to b of the index of a would not violate (ia). 2. One could say that in such cases max(b) comprises the whole sentence, which therefore does not contain a. 3. Giorgi does not discuss logophoricity. 4. This classification draws heavily on Everaert's insights. The summary presupposes that zich and Cindy in Cindy schaamt zich, 'Cindy is ashamed of herself', do not have independent thematic roles; see Reinhart & Reuland for discussion. Zich can be bound beyond the domain of its local subject only in perception verb complements, as in (i). See Reinhart & Reuland for a discussion of why this is so. (i) Jan liet Marie voor zich werken Jan let Mary work for himself 'mzelj occurs in expressions such as namens 'mzel!,on behalf of himself', which modify the proposition. The fact that it cannot occur in temporal and locative PPs falls under the generalization in the text if one assumes that these are part of the predicate phrase, which takes the subject as an argument. Koster (1985) presents a number of observations which would appear to contradict the simple generalization presented here. However, it can be shown that in those cases no bound variable binding is involved. Compare, for instance, (ii) with (iii).
(ii) Ik wantrouw John's beschrijving van 'mzelf en die van Piet ook I distrust John's description of himself and Peter's too (iii) Ik wantrouw John's beschrijving van zichzelf en die van Piet ook
Long-distance anaphora: an overview
25
The former case must be construed as referring to Peter's description of John, the latter case only refers to Peter's description of himself. Similar remarks apply to examples like (iv). Peter tevreden over 'mzelf (iv) Mary acht Mary considers Peters satisfied with himself On the face of it, when contained in a specific NP, 'mzel[ allows a binder outside it, as in (v). (v) Jan haat dat schilderij van 'mzelf en Piet ook Jan hates that picture of himself and Piet too Again, this is no variable binding, as the thing hated by Piet is John's picture rather than his own. If the NP is non-specific we get variable binding instead, as in (vi). (vi) Jan haat een schilderij van 'mzelf en Piet ook Jan hates a picture of himself and Piet too Note that 'mzel[ can be bound by a non-subject in cases like (vii). (vii) Jan gaf Piet een schiJderij van 'mzelf terug Jan returned to Peter a picture of himself There is a simple rule of thumb for the distribution of zich and zichzel[in PPs: zich only occurs as the object of a preposition which may be stressed (Rullmann (1988». This may be directly related to the thematic properties of the preposition involved. 5. Unless it is classified as an anaphoric element without a governing category. However, for at least some of them this is impossible as they have full argument status, bearing Case and a thematic role.
2 Containment and connectedness anaphors Lars Hellan
1 Introduction
This chapter articulates the following hypothesis: the conditioning factor for whether an anaphor is 'long-distance' or 'locally' bound resides in a distinction between what will be called containment conditions and connectedness conditions on anaphors. The hypothesis is based mainly on properties of the anaphoric systems of Norwegian and Icelandic (sections 2.1-2.2), and the importance of the distinction is demonstrated also by interpretive facts concerning Norwegian constructions containing anaphors (section 3). Based on preliminary consideration oflong-distance anaphors in certain other languages (section 2.3), our conjecture is that the hypothesis holds universally. The distinction between containment and connectedness conditions can be seen in the perspective of what relation modules exist in grammar: connectedness conditions deal with what we may call the argument module, i.e. relations defined in terms of the relation 'argument of', a module extensively explored in recent work (e.g. Chomsky (1986a) and related research). Containment conditions, in contrast, involve what we will call the command module, a family of relations tied together by a general notion of 'command', to be elucidated as we proceed. The significance of containment conditions for anaphors brings to light the status of the command module as an independent and potentially significant module in the grammar.
2 Long-distance binding and the containment/connectedness distinction 2.1 Notions and the main principle Let us distinguish at the outset between an anaphor and an anaphoric element. By an anaphor, we understand, as is commonly done, an item which has to be
27
28
Lars H etlan
bound, with coreference as the most standard interpretation of the binding relation. Categorically, anaphors are maximal-projection-type items (e.g. NPs). An anaphoric element is an XO-type item whose constituent status in an anaphor is crucial for determining (i) the status of that anaphor as an anaphor (rather than, say, a pronominal), and (n) the binding domain of that anaphor (where by the 'binding domain' we mean the set of configurations where binding is possible). For example, in Norwegian, the binding domain of seg selv is the intersection of the binding domains definable for seg in isolation and selv in isolation, and likewise complementarity effects with regard to these elements. (See Hellan (1988), for an analysis along these lines.) The following distinguishes containment and connectedness conditions. A containment condition on an anaphor A requires that the anaphor occur in a configuration of the type represented in diagram (la), a configuration we call a containment relation; B is the binder of the anaphor, and C a constituent which contains A. No further relations are specified as holding between C and A. A connectedness condition requires that the anaphor occur in a configuration of the type depicted in (lb), what we call a connectedness relation between Band A. Here C has a particular relation, definable in terms of argumenthood, to A. In either case, R is some relation holding between Band C. (1) a.
R
BoE
C
L[' I
\
I
/
I
\
\ \
A \
I
bound by b. R
B"
C
Li bound by
Various relations can instantiate R in (la); examples are predication and be within the perspective of (logophoricity). In (lb), R is solely the relation be an argument of. An illustration of the containment type of relation is given in the Norwegian
Containment and connectedness anaphors
29
examples (2), where R = predication and C is the bracketed constituent; (3) illustrates a connectedness relation, with C = admires. (2) a.
predication
J
I
[ba meg om Ii snakke om seg] asked me to talk about REFL
Jon Jon
I
Lbinding predication
b.
J
Vi gjorde Jon we made Jon
I
[glad i seg selv] fond of REFL
---'I
lbinding _ _ _ r--argumentof~
(3)
~argumentof~
John
admires
himself
1
binding
I
Our main hypothesis in this section has two parts, (4A) and (4B): (4) A. B.
Long-distance anaphors (LDAs) obey containment conditions exclusively. The elements of an LDA obey the same containment conditions, or a subset of them, when they are elements of a more locally bound anaphor as they do in long-distance binding.
To exemplify (4B): whatever containment conditions govern [seg]NP as an LDA should also (all, or a subset of them) restrict the local anaphor [seg selv]NP, since the element seg is part of this anaphor. We note the intuitive plausibility of (4): it is exactly from anaphors whose binding domain is not defined in terms of any independent link between binder and bindee that we will expect long-distance binding. Furthermore, by the claim that it is in virtue of inherent properties of the constituting elements of the anaphors that this freedom becomes available, we locate the source for LDA in independently existing factors in the anaphors. The natural scope of (4) is thereby universal. The empirical scope of the present chapter being rather limited, however, our aim will only be to provide enough initial justification for (4) to make it a viable hypothesis for further investigation.
30
Lars H etlan
The opposite view of (4) will construe apparent non-local relations as being composed of local ones, chained together in a limited number of ways. This view construes all anaphors as what we will call connected: the primary instance of A being connected with C is for A to be an argument of C, as in (3), where 'be an argument of' may be taken in either of the senses 'be syntactically governed by' or 'receive a a-role from'. Hence, the binder B being in turn an argument of C, these are the cases where B and A may be characterized as co-arguments (with the same ambiguity of 'argument of'). Typical examples are Germanic anaphors composed with varieties of -self (Dutch zelf, Norwegian selv, etc.), such as in John admires himself (cf. (3)). On a universal basis, co-argument anaphors may possibly be the type most widely attested, and is also the one covered by principle A of the binding theory of Chomsky (1981), i.e. that an anaphor be bound in its governing category. However, also compositions of 'argument-of' relations will count under the notion 'connectedness', which we thus define as follows:
(5) A.
B.
X is tied to Y if X bears an argument relation to Y, or X bears an argument relation to an item Z which is tied to Y, or heads a constituent tied to Y. An anaphor A is connected if there is an item C such that the binder B is an argument of C, and C is tied to A.
In justifying (4), let us in the first place distinguish between the characterization of an anaphor per se as an LDA, and an actual occurrence of an anaphor as being 'long-distance bound'. The former pertains to binding potential, and is compatible with the circumstance that some occurrences of that anaphor might be what one would call 'short-distance bound'. Still, of course, a substantive number of occurrences would need to be 'long-distance bound' in order for the anaphor as such to count as an LDA. The following examples from Norwegian all instantiate what we call 'long-distance binding':
(6)
a. b. c. d.
Joni hf/lrte oss snakke om segi Jon heard us talk about REFL Joni bad oss snakke om segi Jon asked us talk about REFL Joni likte din artikkel om segi Jon liked your article about REFL Joni bad oss forsf/lke a fa deg til a snakke pent om segi Jon asked us try to get you to talk nicely about REFL
Containment and connectedness anaphors e.
f.
31
Joni snakket om katten sini REFL Jon talked about cat 'Jon talked about his cat' J oni valgte en for sini kone fordelaktig ordning Jon chose a for REFL wife advantageous arrangement
(6a) and (6b) exemplify binding out of an infinitival clausal domain, (6a) being an Ad-construction and (6b) an object control construction. In (6c) seg is bound out of a definite NP, and in (6d), we see that the number of infinitival boundaries across which binding can occur does not seem upward limited. (6e) and (6f) exemplify use of the possessive form. For both the possessive and the non-possessive form, the upward bound on the binding domain is tensed S; thus, (7) is ill-formed: (7)
*J oni var ikke klar over at vi hadde snakket om segi Jon was not aware that we had talked about REFL
In all of the cases in (6), the binder of the reflexive is outside the co-argument domain of the reflexive, and this we take as an initial characteristic of 'longdistance binding'. Given the rich possibilities of such binding demonstrated in (6), we also count seg/sin in general as LDAs. The mere possibility of being bound by a non-eo-argument will not suffice for being classified as an LDA, however: as we will see, an interesting class of anaphors in various languages show a great freedom in long-distance choice of binders, whereas some make use of it only to a very limited extent, and we will reserve the label 'LDA' for the former type. Hence, the anaphors we will count as LDAs in (4) are those which make essential use of the long-distance binding strategy.
2.2 Long-distance anaphors in Scandinavian
2.2.1 Norwegian We now show the validity of (4) for some LDAs, starting with seg/sin in Norwegian. Consider first (4A), the claim that the Norwegian LDAs obey containment conditions exclusively. A condition on seg/sin, used as LDAs, is that they occur as A in (la), with R = 'predicated of' (see Hellan (1980, 1983, 1988) concerning this condition). That is, the reflexive must occur inside a constituent predicated of the binder, a constellation present in all of the examples in (6), schematically represented in (2a) and repeated below:
32
Lars H etlan
(2a)
predication
,JI Jon Jon
I
[ba meg om a snakke om seg] asked me to talk about REFL
-'1
Lbinding _ _ _ _ _ _ _ _ _ _ _ _ _
In this constellation we say that the binder predication commands the anaphor. In (8), this condition is not met, and the result is ungrammatical: (8)
*Vi fortalte Joni om et fors~k pa a hjelpe segi We told Jon about an attempt to help REFL
The predication-command condition clearly looks like a containment relation, in that - judging from (6) - no restrictions seem to apply to the relation between the predicate and the reflexive, apart from the reflexive being contained in the predicate. There is only one negative condition: no tensed S-boundary can be crossed, as illustrated by (7). This, though, is a negative containment condition, its requirement being simply that the reflexive cannot be anywhere inside a tensed S if the binder is outside. Thus, long-distance seg is constrained by one positive and one negative containment condition. To fully establish (4A), we also have to show the absence of any connectedness condition concerning long-distance seglsin. In the first place, it is not clear what such a condition would contribute, since the predication-command condition fully distinguishes the grammatical from the ungrammatical. In effect, replacing this condition with a connectedness condition would seem to both overgenerate and undergenerate: It would overgenerate in a case like (8), since here presumably Jon and the subsequent infinitival expression are co-arguments. It would undergenerate in a case like (9), since the adverbial constituent in which sin is part is presumably not an argument of the verb: (9)
Jon; snakket med meg [utenfor kontoret sitt;] Jon talked with me outside office his
Then consider (4B), the proposition that short-distance bound occurrences of anaphors where seg is an element obey at least a subset of the conditions governing the long-distance anaphor seg. There are two cases of short-distance anaphors with seg as element: the anaphor seg selv, and 'non-argument' occurrences of seg in constructions like skamme seg 'be ashamed'. 1 Both types obey the predicationcommand requirement; (lOa) and (lOb) illustrate well-formed cases with seg selv and (lOc) one with non-argument seg:
Containment and connectedness anaphors (10)
a. b. c.
33
Joni fortalte oss om seg selvi Jon told us about himself Vi gjorde Joni glad i seg SelVi we made Jon fond of REFL J oni skammet segi J on was ashamed
In contrast, in (11), the material following Jon is not predicated of Jon, and ungrammaticality results: (11)
a.
*Vi fortalte Joni om seg selvi we told Jon about REFL
(4B) thus holds as well, the 'subset' in question being identical to the set of positive conditions by virtue of which (4A) is true, i.e. the predication-command condition. Given that seg/sin is counted as a long-distance anaphor, (4) consequently holds true of this anaphor.
2.2.2 Icelandic We next turn to the LDA sig in Icelandic. The element sig obeys a predicationcommand requirement very much like seg in Norwegian (cf. Maling (1986)). That is, both the long-distance bound anaphor [sig]NP and the locally bound anaphor sjalfan sig obey predication command. In this respect Icelandic recapitulates the situation just described for Norwegian, (4B) thus holding. The main difference between the reflexives seg and sig is that sig under certain circumstances can be bound across a tensed S-boundary, the so-called 'non-clausebounded reflexivization' (NCBR) (cf. Anderson (1986), Maling (1984), Rognvaldsson (1986), Sigur5sson (1986a), and Thrainsson (1976a)). The necessary condition for NCBR to arise, in addition to predication command, seems to be that some kind of 'perspective holding' on the part of the (referent of the) binder be understood, in that the reflexive must be contained inside a constituent understood as being under the binder's perspective (wording, point of view, etc.). This is the phenomenon often referred to as 'logophoricity'. Although a general definition of this phenomenon seems hard to obtain (cf., e.g., Sells (1987)), its relevance to Icelandic NCBR seems quite clear. The perspective factor can be construed as a containment condition. For instance, the example in (12) (from Thrainsson (1976a)) can be seen as an instantiation of (la), on analogy to (2a), with 'in perspective of' instead of 'predication' and C = the bracketed constituent. In this constellation we will say that B (Haraldur) perspective commands A (sig). Note that Jon is not a possible binder of sig. This,
34
Lars H ellan
according to the present view (first suggested in Thniinsson (1976a), is because the perspective-command requirement is a necessary condition, and there is no way of construing Jon as a perspective holder with regard to the 'unless'-clause. (12)
in perspective of
f
I
Haraldur segir a5 [cJon komi ekki nema Maria kyssi sig] Harald says that Jon comes not unless Mary kisses REFL
iL_________ binding _ _ _ _ _ _ _ _----'1 On this view, NCBR in Icelandic is thus governed by two containment conditions, predication command and perspective command. Perspective command serves as a necessary condition on binding of sig only when this is an instance of NCBR, i.e. not in short-distance bound occurrences. This situation again satisfies (4B), but this time with a proper subset relation. The two notions of 'command' now introduced (following Hellan (1986b, 1988)) should be seen as forming a family together with c-command and two others to be mentioned below, the common feature being a superiority relation, be it defined in syntactic or semantic terms. Regarding the difference between seg and sig in the possibility of having NCBR, hypothesis (13) suggests itself: (13) Principle for 'strength' of long-distance binding: stronger intervening category boundaries can be crossed in the binding of an anaphor the more containment conditions hold for the anaphor. That is, it is only by making the perspective-command requirement, in addition to the predication-command requirement, available for sig that Icelandic enables this anaphor to enter into NCBR; no such requirement is available in Norwegian. A reason why Icelandic has this possibility and Norwegian not may be that the perspective factor is partly grammaticalized in Icelandic through the existence of subjunctive mood, a device absent in Norwegian. (We thereby imply that the relationship between subjunctive mood and NCBR in Icelandic is far more indirect than a view sometimes held, namely that subjunctive is somehow a direct licenser for NCBR. See further discussion shortly.) The reason why an extra containment condition should 'add strength' in this way may be that such conditions are the only way of ensuring distance from a binder - they, as it were, constitute the only means of 'breaking away' from the local environment of a binder. To establish fully that (4A) holds of sig, we still have to show that sig is not really governed by a connectedness condition. Two circumstances might be taken to
Containment and connectedness anaphors
35
point in the direction that such a condition is in play. One is the fact, illustrated by the impossibility of Jon binding sig in (12), repeated, that configuration (14) is generally ruled out (noted in Maling (1984)): (12)
in perspective of
'" segir a5 [cJI on komi ekki nema Maria kyssi sig] Haraldur Harald says that Jon comes not unless Mary kisses REFL I<------------binding (14)
-----------'1
~S~ /vrnax~
NP
rnax 1 v -
/ADV~ •• '[S[ +tensel •••
REFL ... ] ...
binding--------------->I Rather than attributing this circumstance in (12) to the lack of perspective command between Jon and sig, one might see it as an effect of a connectedness condition on sig, since sig in (12), or REFL in (14), is not inside of an argument of the verb taking the binder as argument. What such a proposal leaves unexplained is how (12) can be good with H araldur as binder. True enough, sig is then inside an argument C of the verb of which Haraldur is external argument, namely the bracketed constituent in (12), but the 'argument-of' chain stops at this argument C: the verb heading this argument (i.e. komi) has no argument-taking relation to what follows, i.e. to the nema-clause containing sig. So, if this alternative approach can hope to dispense with the perspective-command restriction on sig, this can happen only at the cost of introducing another containment condition: that sig be contained inside an argument of the same verb as the binder is an argument of. Let us call this putative relation argument command between the binder and sig. (4), then, clearly holds true also on this analysis. Further examples show that argument command cannot generally replace perspective command. Consider (15): (15)
Joni heldur a5 Mariuj hafi veri5 sagt a5 pu tala5ir urn sigi/*j Jon thinks that Maria has been told that you talked about REFL
36
Lars H ellan
In this example, Manu cannot bind sig, even though the verb segja 'tell' takes the ao-clause as argument, and Manu hence argument commands sig. What is missing in this case is arguably perspective command between Manu and sig: Manu, as a passivized NP, does not carry a role which is naturally construed as corresponding to a perspective holder with regard to the ao-clause. This indicates that perspective command is the necessary condition on NCBR in Icelandic, not argument command. Still further examples seem to indicate that both types of command may in fact playa role in Icelandic NCBR. Maling (1984) notes the contrast between (16a) and (16b) (mentioned also in Sells (1987»: (16)
a.
*Olafuri hefur ekki enn fundiD vinnu sem Seri likar Olafur has not yet found a job that REFL likes b. Joni segir ao Olafurj hafi ekki enn fundiD vinnu sem Seri/j likar Jon says that Olafur has not yet found a job that REFL likes
In (16a), the reflexive is inside a relative clause. Although this clause is part of an NP serving as co-argument with the putative binder, the binding relation is impossible, confirming the position that argument command is neither necessary nor sufficient in order for NCBR to obtain. In (16b),Jon is a possible binder, again as we would expect, since Jon here perspective commands the reflexive. However, in this case, the reflexive can be found by OZajur as well. This contrasts with the illformedness of a binding between Jon and sig in (12). The only difference between these cases is that in (16b), argument command after all does obtain between OZajur and the reflexive, while it does not in (12) between Jon and sig. Thus, it seems that once a certain domain is perspective commanded by some item X, certain operations can be licensed by argument command within this domain, even if the argument commander is not X. This may then indicate that argument command does have a function in Icelandic NCBR, but only a secondary function compared with perspective command. And with regard to the principle (4), since both these command relations are containment relations, (4) is supported. Another circumstance which might seem to suggest that Icelandic NCBR is a connectedness phenomenon is the frequent use of subjunctive mood on verb forms interconnecting binder and bindee. The constellation 'V ... V[ +subj]' might be seen as a manifestation of a government relationship between the two verbs (or between the first verb and the VP headed by the second), and on that view, NCBRs in Icelandic would seem to be connected in exactly the sense (5), once we count any governee as an argument of the governor. 2 However, it appears that subjunctive mood is neither necessary nor sufficient in order for NCBR to occur. That it is not necessary is argued in Sigurosson (1986a); he adduces
Containment and connectedness anaphors
37
examples which look quite like those above, except that no subjunctive occurs. That subjunctive alone is not sufficient is shown by examples like (17), where the subjunctive is induced by nema. (From Thniinsson (1976a); cf. (12).) (17)
*J6ni kemur ekki nema Maria kyssi sigi Jon comes not unless Mary kisses REFL
In (17), in accordance with what we have already said, neither perspective command nor argument command obtains; and in the examples used by Sigurasson, perspective command does obtain, serving as the licenser, according to Sigurasson. Thus, the presence or absence of subjunctive has no effects independently of the factors discussed above, and so is not critical to the use of NCBR. Rather, as indicated earlier, the presence of subjunctive should be seen at least partly as a reflex of the perspective-command relation, and is relevant for NCBR primarily in virtue of being a means by which perspective command has been grammaticalized. This concludes our demonstration that (4) holds both for Norwegian and Icelandic long-distance anaphors. The licensing containment conditions for these anaphors are all phrased in terms of what we have called 'command' relations, which is to say that the binder must have a superiority status with regard to the anaphor, identifiable on either syntactic or semantic grounds. The phenomenon of longdistance anaphora thus seems to highlight in a very clear-cut fashion the existence of these command relations, or what we may call the command module. These command relations interact with connectedness conditions, or the argument . module, in the cases where the LDA-element occurs together with selvlsjrilfan, and the simple insight stated in (4) is that LDA is what results when this extra element is omitted, i.e. when the command relations are the sole domain-defining factors in the anaphors. This having been established for Norwegian and Icelandic, a question is still whether the only way in which an LDA can emerge, cross-linguistically speaking, is through its constituent element(s) being associated with command-type conditions only. Our conjecture, as expressed in (4), is in the affirmative, but limitations of space and time preclude an investigation of this in the present chapter. We confine ourselves to just reviewing data from a few languages in the perspective of (4). 2.3 Other long-distance anaphors
2.3.1 Italian proprio The possessive reflexive proprio in Italian undergoes NCBR, as shown in example (18) (from Giorgi (1984)):
38
Lars H ellan
(18)
Gianni j ritiene che Osvaldoi sia convinto che quella casa Gianni believes that Osvaldo is persuaded that that house appartenga ancora alla propriai/j famiglia belongs still to REFL's family
The main restriction on proprio as an NeBR, according to Giorgi, is that it be in the P-domain of its binder, where for X to be in the P-domain of Y means that X is contained in some item Z such that Y and Z get their O-roles from the same O-roleassigner, and Y's O-role ranges higher than Z's O-role on the hierarchy '(1) Agent, (2) Experiencer, (3) Theme'. Using a term from Hellan (l986a, 1988) we will say that in such a case, Y role commands X. The relevance of this command relation has been demonstrated for more locally bound anaphors in languages such as Hungarian (Kiss (in preparation» and Norwegian (Hellan (1986a, 1988». It is distinct from the command relations established so far, and appears to be the only one in the family relevant for proprio as an LDA. First, in contrast to predication command, it does not entail a c-command relation between binder and bindee, and thus allows (19), a construction whose counterpart with seg or sig is ill-formed in Scandinavian (for seg, the relevant example would have to have an infinitival subject clause): (19)
La salute di quelli che amano la propriai moglie preoccupa molto Osvaldoi a lot Osvaldo the health of those who love REFL's wife worries
Secondly, the role-command relation entails argument command, and so the rolecommand requirement suffices to rule out (20), which exemplifies constellation (14).
(20) *Osvaldoi ritorna III patria prima che il fisco Osvaldo returned to his country before the public treasury sequestrasse il proprioi patrimonio sequestered REFL's estate Role command is presumably entailed by perspective command, since perspective holders are likely to be higher in the O-hierarchy than the material within the perspective. However, the entailment clearly does not go the other way - the higher one among two given roles need not be a perspective holder. In sum, then, role command is an additional member to the set of command relations. Since in all of the well-formed examples considered, the binder could be interpreted as a perspective holder, and in (20) not, a possible alternative analysis might be that proprio be perspective commanded by its binder; this question cannot be decided here, however, so we stick to Giorgi's analysis for the present.
Containment and connectedness anaphors
39
(4A) thus being true of proprio, the question is whether role command is associated with proprio as an essential condition in the way predicted by (4B). The literature does not give any explicit answer so far, so we leave this point open. 3
2.3.2 Japanese zibun The behaviour of the Japanese reflexive zibun as an LDA is discussed in a number of works (e.g. Kuno (1973), Kuno & Kaburaki (1977), Kuroda (1965), Inoue (1976), Farmer (1980), and many others). It can be bound out of clauses, as exemplified in (21) (from Inoue (1976: 163», hence it allows NCBR: (21)
Tarooi ga Hanako ni zibuni Amerika e itta koto Taroo Hanako REFL America to go-PAST that 'Taro did not tell Hanako that he had been to the States'
0
hanasanakatta tell-NEG-PAST
It is commonly assumed (see, e.g., all of the works cited, and Sells (1987» that perspective command plays an essential licensing role for NCBR in Japanese. A perspective-command requirement licenses zibun in (22) (an example similar to (19) in Italian above), the well-formedness of which shows that predication command is not a necessary condition (Inoue (1976: 171»: (22)
Zibuni ga Mary ni karakawareta koto ga Johni 0 zetuboo e oiyatta SELF Mary was-made-fun-of that John desperation to drove 'that self (= he) was made fun of by Mary drove John to desperation'
Perspective command furthermore suffices to rule out (23) (instantiating constellation (14», rendering 'argument command' just as unnecessary as it was for proprio in Italian (Kuroda (1965»: (23)
*Takasii wa [Yosiko ga mizu 0 zibuni no ue ni kobosita toki] nurete simatta Takasi Yosiko water REFL 's on spilled when wet got 'Takasi got wet when Yosiko spilled water on him'
Reportedly (Kuroda (1965); see also Sells (1987», substituting node 'because' for toki 'when' in (23) yields a well-formed result. Assuming that this item turns the adverbial clause into a perspective domain for Takasi, the requirement of perspective command then licenses also this new construction. This time, a role-command requirement would not have such a licensing function, and this will be an indication that it is perspective command and not role command which is the crucial notion for zibun as an LDA. As with Italian, we are now hypothesizing that it is just a single command relation which is relevant for zibun as an LDA, as opposed to the situation in Icelandic. This may be an oversimplified picture, due to the incompleteness of the
40
Lars H ellan
data chosen. Also, if a single relation is to be selected, one might hope for something more tangible than perspective command, which is probably the most elusive of those considered. Let us still, for the present, assume that perspective command is the licensing-containment condition for zibun as an LDA, and role command for
proprio. (4A) being true of zibun, the next question is whether (4B) holds, i.e., whether perspective command acts as a condition also on more locally bound occurrences of zibun. An example like (24) (from Farmer (1981: 178)) indicates that some condition within the family we have discussed holds, but it does not determine whether it is predication command, role command or perspective command: (24)
Tarooi wa kodomoj 0 zibuni/*j no kutu de butta REFL 's shoe with hit Taroo child 'Taroo hit the child with self's shoe'
(25a) lets through the same possibilities, whereas the relative acceptability of the 'j' -indexed version of (25b) is difficult to view as licensed by predication command; both of the other relations obtain, however (from Inoue (1976: 148)): (25)
a. Oyazii wa bokuj ni zibuni/j no kuruma 0 Dad me REFL 's car 'Dad made me wash self's car' b. Bokui wa oyazij ni zibuni/(j) no kuruma 0 1 Dad REFL 's car 'I was-caused-to-wash self's car by Dad'
arawaseta made wash arawaserareta was made wash
At the present point we are not able to press these considerations any further, but there is clearly a possibility that (4B), and thus (4) as a whole, will turn out to be true. (See some further remarks below.) 2.3.3 Chinese ziji As a final example of anaphors with clear long-distance properties, we mention the non-contrastive variety of the Chinese reflexive ziji, basing our discussion on that in Xu (forthcoming). Examples like the following show that ziji can attain NCBR, (b) indicating that only subjects can serve as binders.
(26)
a. Lao Zhangi ting shuo Xiao Iij rna ZlJli/j Lao Zhang hear say Xiao Li swear REFL 'Lao Zhang heard Xiao Li swore at himself/him' b. Johni xiangxin Billj dui Samk shuo zijii/j/*k taoyan Mary John believe Bill to Sam say REFL hate Mary 'John believes that Bill said to Sam that he hated Mary'
Containment and connectedness anaphors
41
According to Xu, the requirement that the binder be a subject holds generally, i.e. also when binder and bindee are inside the same minimal clause. In our terms, these observations will indicate that the containment condition of predication command is the one operative in Chinese NCBR, and, in accordance with (4), that this condition is inherently associated with ziji. Provisionally, then, we interpret the Chinese data too as potentially confirming hypothesis (4).
2.3.4 A possible weaker version of(4), and criteria of being a long-distance anaphor With the discussion of proprio, zibun and ziji in mind, there might be cases of LDAs conforming to (4A), but where the containment condition(s) obeyed by the relevant anaphoric element in local uses is/are different from the one(s) in operation in long-distance uses. For instance, perspective command could be the relevant notion for zibun as an LDA, but role command the one governing locally bound occurrences. If we assume that the command relations constitute a natural class, a weaker, but still interesting version of (4B) would be to say that, if (4A) holds, then some command relation must also govern the locally bound occurrences of the anaphoric element in question. It might in turn be possible to establish a ranking among the possible command relations. Concerning (4A), the question will arise as to exactly what the criterion should be for a binding relation to count as 'long distance'. Consider the Dutch anaphor zich. Apparently, this anaphor allows long-distance binding, judging from the possibility of (27); on the other hand, although putatively long-distance binding of zich must have a subject as its binder, zich does not induce any predication requirement in locally bound occurrences, like in the anaphor zich zelf, as evidenced by the well-formedness of (28): (27)
dat Mariej Piet voor zichj liet werken that Mary Peter for her let work 'that Mary let Peter work for her'
(28)
Ik vroeg Pietj over zichzelfj I asked Peter about REFL
The question is whether zich should be counted as an LDA, and perhaps satisfying (4A), but then refuting (4B) (even in its weaker version). Our proposal is that zich should not count as an LDA in the first place. As witnessed by the ill-formed examples in (29), zich is peculiarly restricted in its binding domain; even in a language with only moderate use of LDA like Norwegian (where Tensed S cannot be crossed), the counterparts of all of these examples are well-formed:
42
Lars H ellan
(29)
a. *Jan; liet Marie zich; ophangen Jan let Mary him hang 'Jan let Mary hang him' b. *Jan; liet Marie zich; respecteren Jan let Mary him respect 'Jan let Mary respect him' c. *Marie; vroeg ons (om) PRO over zich; te praten Marie asked us (COMP) about REFL to talk d. *Jan; vroeg mij Marie voor zich; te laten werken Jan asked me Mary for him to let work 'Jan asked me to let Mary work for him' e. *dat Jan; het schot op zich; betreurde that Jan the shot at REFL regretted 'that Jan regretted the shot at himself' f. *dat Jan; Marie verliefd op zich; achtte that Jan Marie in love with REFL considered 'that Jan considered Marie in love with himself'
Clearly, zich lacks the basic freedom to occur long-distance bound which has been manifest in all the other cases considered so far - it is not 'essentially non-local', in the terms from 1.1, and so does not naturally fall within the group constituting the clear cases of LDA. And if zich is not an LDA, (28) ceases to be a counterexample to (4B). What is zich then? We propose that it is a connectedness anaphor with a binding domain extending beyond co-argumenthood, as our definition (5) renders possible. This is in line with analyses presented in Everaert (l986a) and Koster (1987), a basic point of which is that there must be some kind of uninterrupted connection, describable in terms of government, extending from zich to its binder. Thus, in Koster (1987), on the assumption that 'regret' and 'consider' do not govern their complement, (2ge,f) are accounted for, and on the assumption that complementizers interrupt the relevant connection and that infinitives can only be governed to the left in Dutch, (29c) is accounted for. While (29a, b) require more specific stipulations and (29d) remains unaccounted for, the picture emerging is that zich must be 'connected' to its binder, in the sense of the notion of 'connectedness' stated above. Clearly, now, on the basis of this line of analysis and examples like (27) looked at in isolation, zich might seem like an LDA subject to connectedness conditions, in strict violation of (4A). By the reasonable requirement on the status as LDA just proposed, however, zich cannot count as an LDA, thus saving (4A); (4B) is left
Containment and connectedness anaphors
43
unaffected in turn. 4 (Notice that although the definition of LDA plays a crucial role here, we are not providing a definition which renders (4) automatically true: we are just suggesting a limitation on the empirical range of the notion LDA, about which (4) still makes an empirical claim.) 2.4 Concluding remarks We have seen that all ofthe LDAs seg, sig, proprio, zibun and ziji obey containment conditions when long-distance bound: seg, sig and ziji must all be predication commanded by their binder; sig and zibun must be perspective commanded; proprio must be role commanded. Among these anaphors, seg and sig are those on which hypothesis (4), repeated below, is based. (4) A. B.
Long-distance anaphors (LDAs) obey containment conditions exclusively. The elements of an LDA obey the same containment conditions, or a subset of them, when they are elements of a more locally bound anaphor as they do in long-distance binding.
The results mentioned lend initial plausibility to (4A) as a universal hypothesis, which in turn shows the significance of the command module. (4B) holds for the Scandinavian reflexives, and our cursory survey of the other anaphors indicates that this proposition too may hold universally, but this is much less clear than in the case of (4A). (4) yields one expectation which is fulfilled: the fewer elements which serve as constituents in an LDA, the greater is the possibility that the conditions associated with the anaphor will be purely of one type, here the containment type. From this perspective, the optimal LDA will consist of one element only; and this holds true of all the LDAs we have considered. Non-LDAs, on the other hand, are often complex, like seg seiv, with selv inducing the connectedness conditions. 5
3 Differences between containment and connectedness anaphors in certain interpretive processes We now address the distinction between containment and connectedness conditions from a different angle. Returning to Norwegian, we observe two contrasts between the connectedness anaphor seg selv and the containment anaphors seg and sin, both in the domain of semantic interpretation. We will propose a unifying principle for these contrasts phrased crucially in terms of the containment/connectedness distinction.
44
Lars H etlan
3.1 VP-anaphora It appears that in VP-anaphora constructions involving containment anaphors, like those in (30), it is possible to interpret the VP-anaphor both with a 'sloppy' and a 'strict identity' reading concerning the reference of the reflexive. (30)
a.
b.
Johni snakker ofte om bilen sinh og det gjj2lr de pa John talks often about car his and it do those at verkstedet ogsa the garage too 'John often talks about his car, and so do the people at the garage' Johni hadde hj2lrt meg snakke nedsettende om segi, og det John had heard me talk depreciatorily about him and it hadde de som stod rundt ogsa had those who stood around too 'John had heard me talking depreciatorily about him, and so had those who were standing around'
That is, in (30a), both a reading on which the people at the garage talk about John's car (,strict identity'), and one where they talk about their own car ('sloppy identity'), are possible. In (30b), the people standing around could have heard me talking depreciatorily about John ('strict identity') or about themselves ('sloppy identity'). On the other hand, if seg selv is used in such a construction, like in (31), a 'sloppy identity' reading is highly preferred. (31)
Joni skryter av seg selvi, og det gjj2lr Marit ogsa Jon boasts of himself, and it does Marit too
The way in which the containment-connectedness distinction will be applied in the following analysis of this contrast is tied to a particular formal analysis of binding and VP-anaphora; but the distinction could probably be construed as relevant also with other formal construals. We assume that the difference 'strict''sloppy' identity in such constructions is to be analysed as follows: in a given structure, occurrences of reflexives and pronominals are annotated for their binding relations, and are subject to rules checking whether these relations are admissible. The interpretation of constructions with VP-anaphoraldeletion involves a process of copying the VP from the first conjunct onto the emptyianaphoric VP of the second conjunct. The rules of assigning and checking binding relations can take place either before or after this copying. If they apply before, the reflexive is marked as being bound by an NP in the first conjunct; this means that the 'copied' instance of the reflexive, i.e. the occurrence contained in the copied VP, will be marked as having the NP in the first conjunct as its binder. This yields the 'strict
Containment and connectedness anaphors
45
identity' reading. If the rules for assigning and checking binding relations apply after the copying of the VP, the reflexive inside the copied VP gets assigned an NP in the second conjunct as its binder - this yields the 'sloppy identity' reading. (See Hellan (1988) for details.) Apparently, now, only the latter option is open in a case like (31), since it allows only a 'sloppy identity' reading. The question is why. One imaginable possibility might have been to ascribe seg selv some sort of inherent variable status, while seg/ sin could have a freer status, and pin the possibilities of readings on this difference. Regardless how such a proposal were to be made explicit, however, it would seem to be on the wrong track, since sin can obtain in a 'strict identity' construction even when it is bound by an NP like 'each participant', the prima facie kind of construction where an anaphor would be expected to behave like a 'variable':6 (32)
Hver deltageri var forhindret each participant was prevented naboen hans matte gjpre det neighbour his had to do it
fra a hente premien sini, sa from fetching trophy his, so istedet instead
The approach sketched above provides another possibility of analysis, and this will be our proposal. We will assume that when a binding relation is connected, it is resistant to being 'interrupted' in the way which happens when the copying process removes the reflexive without bringing along the binder. A connected reflexive thus is connected to its binder in a very concrete sense: the bindee cannot be displaced without its binder following it. No such requirement applies to containment anaphors, hence the differences in the interpretation of (30) and (31). This hypothesis has to face the fact that connected anaphors can easily be 'moved away' from their binder in a very overt fashion, like in (33): (33)
a. b.
Seg selvi himself Beundre admire
jeg at vet know I that seg selvi vet himself know
J oni beundrer Jon admires jeg at Joni gjpr I that Jon does
Apparently, the existence of a trace, or the reconstruction process/relation thereby encoded, makes this case count as a non-intrusion in the respect in question. Consequently, the ban against displacement of connected anaphors relative to their binder holds only with regard to 'non-visible' displacement. Once the ban is associated with connected anaphors in general, it follows that also 'non-local' connected anaphors, such as Dutch zich, should be ineligible for strictly read VP-anaphora. According to Everaert in chapter 4, this prediction is
46
Lars H ellan
borne out; but further investigations at a cross-linguistic level are obviously needed to assess fully the validity of our hypothesis. 3.2 Speaker construal An analogous situation to the one now described may be seen to arise in cases where a given part of a structure is subject to construal on the part of the speaker rather than by (the referent of) an NP in the structure? In (34), the part sin egen kone 'his own wife', with the reflexive sin, is most reasonably read as a description supplied by the speaker, not as one representing Jon's or Peter's construal of the situation (that is, Jon observes Peter's interest in a certain woman, without knowing that they are in fact married, and interprets what he sees as a wish on Peter's part to marry the woman); (34)
Jon tror at Peterj viI gifte seg med sinj egen kone Jon thinks that Peter will marry his own wife
Sin is here a containment reflexive. Consider now what happens when a connected reflexive is involved, as in (35): (35)
Jon tror at Peterj snakker om seg selVj Jon thinks that Peter talks about himself
Here it appears very difficult to interpret the description seg selv as supplied by the speaker and not by Jon. (That is, it is difficult to apply (35) to a situation where Jon hears that Peter speaks about somebody, and only the speaker is aware that this somebody is Peter himself.) Various explanations could be imagined for this, but one possibility is that the exclusive speaker construal of seg selv in (35) is blocked by the connected binding relation in the same way as the copying process mentioned above is. This means that 'exclusive speaker construal' must be understood as a process at the interpretive level which somehow 'cuts loose' some expression and assigns it a higher scope than its syntactic position would indicate. Various conceptions of 'raising at LF' suggest themselves as implementations of this idea. For our purposes, it is enough to point to the 'invisibility' of such a process as a property it would share with the copying assumed in VP-anaphora constructions (which could also be taken to be an LF-process), and state the essential characteristic of connected binding relations with regard to both processes: that they cannot displace the anaphor from the domain of its binder. If correct, this analysis lends further support to the distinction between connectedness and containment anaphors, and entails that it be represented at the level (e.g. LF) where scope relations are represented.
Containment and connectedness anaphors
47
4 Concluding remarks The purpose of this chapter has been to provide motivation for drawing a distinction between containment and connectedness conditions on anaphors. While the latter fit into models largely based on locality notions of the type 'govern' or 'argument of', such as the model of Chomsky (1986), the containment conditions represent an entirely different paradigm of principles: their common denominator, apart from 'containment', is the superiority notion command, whose instances comprise predication command, role command, perspective command, c-command, perhaps argument command, and possibly others. We propose that these notions be regarded as constituting a module on their own in the grammar, namely the command module. As has been pointed out, this module is not restricted in its application to long-distance anaphora, but is observable in local anaphora as well; and the main hypothesis of section 1 of this chapter is that long-distance potential of an anaphor presupposes (1) that it obeys command-relational conditions only, and (2) that its constituting element(s) is/are subject to the same commandrelational conditions when it/they constitute(s) locally bound anaphors. Apart from the cross-linguistic investigation of this hypothesis, another topic for further research will be in what other areas of the grammar the command module is operative. The proposals of section 2, likewise, are open for investigation, both cross-linguistically and with respect to the nature of LF.
Acknowledgements This paper has benefited from comments by Lars Johnsen, Anneliese Pitz, Tor Afarli, and the audience at the Workshop on Long-Distance Anaphora at Groningen in 1987.
Notes 1. See Hellan (1988) for a treatment of Norwegian non-argument reflexives. 2. This is the idea of, e.g., Koster (1987). 3. In a full discussion of anaphors in Italian, also the reflexives se and se stesso must be included. See Giorgi (1984 and 1986), the latter addressing facts of NP-internal binding with possible relevance for the proposition (4B). 4. What is not yet accounted for is why the 'stretched' connectedness-type zich apparently only takes a subject as binder. Whether this will follow in any way from principles concerning the connectedness group of relations is a matter we cannot go into here. 5. Given that self in English induces connectedness conditions as well, and is the only reflexive element in English, it follows that English has no long-distance anaphors. English and other reflexives have been noted to display certain long-distance effects
48
Lars H ellan
when they appear as parts of subjects in embedded clauses (see, e.g., Chomsky (1981), Giorgi (1984), and many others). I believe this to be a quite different phenomenon from the cases under discussion, and nothing is said about it here. 6. This example is due to Elisabet Engdahl, personal communication. 7. I take it that this distinction accounts for many of the classical 'de dicto/de re' ambiguities.
3 Long-distance reflexives and the typology of NPs Hoskuldur Thrainsson
1 Introduction Part of the importance of linguistic theories, as well as other scientific theories, is that they tell you which facts are interesting, why they are interesting, and where to look for other interesting facts. The standard binding theory (BT) of Chomsky (1981) has been extremely important in this respect since it has sparked a great deal of cross-linguistic research into the binding properties of pronouns, reflexives, and other NPs, overt and non-overt. As a result, we have learned a lot about the nature of NPs in various languages, what their similarities are and how they differ, both within a given language and cross-linguistically. It is well known that one does not have to look very hard or very far to find, say, reflexives that do not obey the same restrictions as reflexives in English. The socalled long-distance reflexives found in various languages are a case in point (see, e.g., Thniinsson (1976a, c), Napoli (1979), Yang (1983), Hellberg (1984), von Bremen (1984), Giorgi (1984), Maling (1984), Anderson (1986), Rognvaldsson (1986), Everaert (1986a), Barnes (1984, 1986), Sigurosson (1986a), Sportiche (1986), Sells (1987), Manzini & Wexler (1987), Wexler & Manzini (1987), Pica (1987), Koster (1987), Kuno (1987), and references cited there). The question is, however, what the existence of such reflexives implies for BT. There are various possibilities, and different suggestions can be found in the literature cited above. In the present chapter I would like to argue for the following claims among others: (1) a.
b.
Some of the 'long-distance reflexives' described in the literature can be accounted for by assuming parametric variation in the definition of 'binding domain', with minimal changes in the standard binding principles (cf. e.g., Yang (1983), Anderson (1986); see also Harbert (1986). There is a class of 'extra long-distance (logophoric?) reflexives' that need not be syntactically bound at all and thus do not obey principle A of
49
50
Hoskuldur Thrainsson
c.
standard BT nor any extension of such a syntactic binding principle (cf., e.g., Maling (1984), and Sigurosson (1986a». The standard classification of NPs in terms of the two binary features [± anaphoric] and [± pronominal] leaves a few existent categories ofNPs unaccounted for. By assuming the basic lexical feature [± independent reference] (to be explained below) and the (more BT-internal) feature [± R-expression] (cf. Lasnik (1986», it becomes possible to account for the syntactic behaviour of (or explain the existence of) 'truly long-distance reflexives', and such an account also makes some interesting predictions about the possible existence of other NP-types, which seem to be borne out.
The organization of this chapter is as follows: in section 2, I review the main characteristics of standard BT (Chomsky (1981», give a few examples of longdistance reflexives that are not immediately accounted for within it, describe briefly a possible extension of BT along the lines of Anderson (1986) and argue that although it - or other syntactic accounts - will work for a number of cases, there is a residue of cases that does not seem to lend itself to a syntactic account in terms of an extension of binding domains or other similar syntactic approaches (say, movement at LF). In section 3, I argue that we need to account for the fact that (standard BT) anaphors and long-distance reflexives of all types have something in common and this something is the lack of capacity for independent or deictic reference (i.e. they are all referentially dependent in some sense) rather than, say, the requirement that they must all be interpreted as bound variables. In section 4, I present a new classification of NPs, first schematically and then in terms of four binary features. Because of the mutual incompatibility of certain feature combinations, it turns out, however, that this system only predicts the existence of eight categories of NPs and it is argued that this is not too many since they all exist (although not necessarily all within a given language). Finally, section 5 is a conclusion.
2 Standard BT classification of NPs and some extensions 2.1 The standard BT Let us assume something like the following as the standard BT: Table 3.1. Standard BT - feature representation Feature complex
Categories
Some examples
A.
(bound) anaphors NP-trace
himself, each other Johnj was killed ej
[+an, -prJ
Long-distance reflexives and the typology of NPs
B.
[-an, +pr]
51
pronominals
he (she, it)
pro
pro leemos muchos libros (Sp.) '(We) read many books' John, the bastard Whoi did you see ei? John told Peter [PRO to go]
c.
[-an, -prj
lexical NPs wh-trace (variables)
D.
[+an, +pr]
PRO
(2) The standard BT principles: A. A [+anaphoric] NP must be bound in its governing category (or some such well-defined local domain). B. A [+pronominal] NP must be free in its governing category (or some such well-defined local domain). C. A [-anaphoric, -pronomimal] NP must be free. If we concentrate on the overt NPs, this classification could be represented schematically as in (3): _ _ NPs
(3)
must be free .
~
--
must be bound
I
free in gov cat
free in any domain
bound in gov cat
pronominals
lexical NPs
anaphors
2.2 A first revision - the syntactically defined domain extended It is not immediately obvious how, say, reflexives in the Scandinavian languages would fit into this classification since they can occur in infinitival clauses and be bound by the matrix subject whereas their English counterparts have to be bound within the infinitival clause in such structures (see, e.g., Anward (1974), Thrainsson (1979: 289ff.), Vikner (1985), Hellan (1986a), Anderson (1986»: (4)
(5)
a. b. c. a.
Peturi Peteri Peteri Annai Anne b. Annei Anne
baa Jensj urn [PROj aa raka sigi/j] (Icel.) bad Jensj om [PRO j at barbere sigi/j] (Dan.) asked Jensj [PROj to shave himself*i/j] telur [pig hafa svikiO sigi] believes you (Acc) have (Inf) betrayed self hprte [mig snakke med dig om sigi] heard me talk (Inf) to you about self
52
Hoskuldur Thrdinsson
This shows that the binding domain for reflexives in the Scandinavian languages is larger than in English, perhaps something like the minimal tensed clause. Note also that intervening subjects of the infinitives do not seem to have any effect, not even overt ones like the /Jig 'you' in (5a) (assuming an exceptional Case-marking analysis for Accusative-with-Infinitive structures like this) and the mig 'me' in (5b) (if mig is the subject there). Similar facts have been reported for Gothic, for instance, by Harbert in a series of interesting papers (1981, 1982, 1983, 1986). Given what we have found so far, it might not seem so difficult to give a syntactic definition of the extended binding domain in the Scandinavian languages, for instance. There is an added complication, however. What we typically find is that 'semi-long-distance reflexives' of this type are bound by subjects and they cannot be bound by objects. This is illustrated by examples from Danish and Icelandic below (cf. Vikner (1985), Thrainsson (1979: 288ff.)): (6)
a. *Eg; lofaoi 6nnuj [PROj ao kyssa sig j] (Ice!.) b. *Jegj lovede Annej [PROj at kysse sig j] (Dan.) I promised Anne to kiss self
Several attempts have been made to account for facts of this sort in the spirit of standard BT. Since the main objective of this chapter is not to account for the syntactic binding of semi-long-distance anaphors of this type, I will just briefly review here the interesting proposal of Anderson (1986) (for a valuable discussion of other recent approaches and a different proposal see Harbert (1986)). Anderson argues that there is a parametric variation in principle A of BT of roughly the following kind: I (7) Principle A: A [+anaphoric] NP must be 1. bound in its governing category, or 2. bound by a superordinate subject within its anaphoric domain. 2 Al is obviously the standard principle A of BT, and Anderson refers to NPs obeying Al as anaphors, whereas he uses the term reflexive pronouns to refer to NPs obeying A2. The idea is, then, that languages can choose between Al and A2, and English uses Al for reflexives and reciprocals whereas the Scandinavian languages (and Latin etc.) would use A2 for (at least some of) their reflexives 3 but Scandinavian reciprocals would obey Al like their English counterparts, as witnessed by Icelandic examples like those in (8) (for further discussion see Vikner (1985), Anderson (1986), Hellan (1986a), and Hellan & Christensen (1986a)):
Long-distance reflexives and the typology of NPs (8)
53
a. Peiri rokuau hvorn annani4 they shaved each other b. Eg sendi peimi gallabuxur ii hvorn annani I sent them bluejeans for each other c. Peiri lofuau mer [PRO i aa raka hvorn annani] they promised me to shave each other d. *Peiri skipuau mer; [PRO; aa raka hvorn annani] they ordered me to shave each other e. *Peiri telja [aa eg hati hvorn annani] they believe that I hate (sbjnct) each other
Now recall that principle B of standard BT (pronominals must be free in their governing category) is in a sense the inverse of principle A. Given the variation of A expressed in (7), Anderson suggests that we should have a similar variation for principle B, namely:5 (9) Principle B: A [+pronominal] NP must be 1. free in its governing category, or 2. subject free (( = not bound by a superordinate subject) in its anaphoric domain. The assumption would then be that English chooses B1 and the following would seem to suggest that Icelandic chooses B2: (10)
(11)
a. *Haralduri syndi mer fot ii hanni Harold showed me clothes for him b. Eg syndi Haraldii fot ii hanni I showed Harold clothes for him Peturi baa Jens; [PRO; aa raka hann*ii*;] Peter asked Jens to shave him
In this rather ingenious way, Anderson tries to express the well-known (partial) complementary distribution of pronouns and reflexives. The contrast between (11) for Icelandic and (12) for the Mainland Scandinavian languages would then suggest that these choose B1 rather than B2 (cf. Hellan (1983), Vikner (1985), Anward (1974)):
(12)
a. Joni bad oss; [PRO; hjelpe hami] (Norw.) John asked us (to) help him b. Susani bad mig; om [PRO; at ringe til hendei] (Dan.) Susan asked me to call her
54
Hoskuldur Thrtiinsson c. Honj bad migj [PROj klippa henneil (Swed.) she asked me cut her 'She asked me to cut her hair'
The following would also be consistent with Bl (again, see Hellan (1983) and Vikner (1985)): (13)
a. *Vi fortalte we told b. *Vi fortalte we told
Jonj om hamj (Norw.) John about him Annej om hendej (Dan.) Anne about her
Note, however, that if a given language uses both Al and A2, as the Scandinavian languages seem to do, there is nothing to prevent it from choosing both Bl and B2, which could mean, for instance, that within clauses (i.e. inside their governing categories) pronominals had to be completely free whereas within the larger anaphodc domain (say, the minimal tensed clause) they would only have to be subject free. While we have omitted certain details in this discussion of binding in the Scandinavian languages,6 it should be clear from the examples given how Anderson's approach allows for an extension of the standard BT while still keeping its spirit. Note, for instance, that it is still true under this approach that (all types of) reflexives and reciprocals (i.e. [+anaphoric] NPs) must be bound in some syntactically defined domain whereas pronominals ([ + pronominal] NPs) must be free in some (corresponding) domain. But now we will look at some reflexive-like elements that seem more problematic for this standard classification of NPs. 2.3 Truly long-distance reflexives, bound and unbound There are also languages where a reflexive inside a finite clause can be bound by a subject outside the clause. Latin, Icelandic and Faroese are a case in point (cf. Milner (1978), Thniinsson (1976a, c), Anderson (1986: 84-6), Barnes (1986), Kuno (1987: 136ff.)), whereas such constructions are normally not found in the Mainland Scandinavian languages (but see Barnes (1984), Hellberg (1984)): (14)
a. Orat te paterj rut ad sej venias] (Latin) asks you father that to self come (sbjnct) 'The father (your father) asks that you come to him' b. Ariovistusj ad Caesaremj legatos mittit rut ex suis j ambassadors sent that of self's Ariovistus to Caesar legatis aliquem ad sej mitteret] ambassadors somebody to self sent (sbjnct)
Long-distance reflexives and the typology of NPs
55
'A. sent ambassadors to C., in order that he (C.) would send his [refl] (C.'s) ambassadors to him [refl] (A.)' (15) a. J6ni sagoi [ao eg hefOi svikio sigi] (Icel.) b. *Jensi sagde [at jeg havde svigtet sigiJ (Dan.) c. *Johni said [that I had betrayed himselfi] (16) Guoruni skilti [at Martin hevoi skrivao eitt brl£v til sini] (Far.) Gudrun understood that Martin had written a letter to self 'Gudrun understood that Martin had written a letter to her' It should be made clear that this phenomenon is truly long distance, since the antecedent (or binder) of the reflexive pronoun can be arbitrarily far away (cf. Thniinsson (1976a: 226»: (17) J6ni segir [ao Maria telji [ao Haraldur vilji [ao Billi heimsl£ki sigiJ]] John says that Mary believes that Harold wants that Bill visits self Similar facts have been reported for Italian by Napoli (1979), for instance, and truly long-distance reflexives exist in several other languages as well (cf. Yang (1983), Kuno (1987), etc.; see also Giorgi (1984». At first it might seem that we have here a somewhat similar syntactic phenomenon to that observed in section 2.2 - i.e. some sort of syntactically definable extension of the binding domain for reflexives. It would, however, seem that one would not want to have principle B of the BT apply to this extended domain since pronominals do not have to be (subject) free in it, as can be seen from the comparison of (15a) and (18). In other words, we do not have complementary distribution of pronouns and reflexives within this domain: 7 (18)
J6ni sagoi [ao eg hefOi svikiO hanni] John said that I had betrayed him
Note, on the other hand, that here again we find that objects do not seem to be acceptable antecedents, at least not in Icelandic: (19)
*Eg sagoi J6nii [ao ]:nl hefOir svikio sigiJ I told John that you had betrayed self
It has turned out, however, that it is quite difficult to come up with a convincing syntactic account and it is probably worthwhile to review the main properties of this kind of long-distance reflexives here. First, observe that for most speakers of Icelandic there is a correlation between the long-distance reflexive (LDR) and subjunctives. Typically, verbs of saying, thinking etc. take subjunctive complements and it is in such complements that we find the LDR. A (semi-)factive verb
56
Hoskuldur Thrainsson
like vita 'know', on the other hand, takes an indicative complement and then most speakers reject the LDR:8 (20)
*J6ni veit [ao pu hefur svikio sigil John knows that you have betrayed self
There are, however, a few verbs that can take either a subjunctive or an indicative complement. In such cases the indicative seems to imply the speaker's presupposition that the complement is true whereas the subjunctive implies a report from the subject's point of view in some sense and the speaker does not commit himself with respect to its truth (cf. Thrainsson (1976a), Maling (1984) and most extensively Sigurosson (1986a). See also Sells (1987». Hence the speaker can deny the content of a subjunctive complement but not that of an indicative one without creating a contradiction: (21)
a. J6n heyroi [ao Maria hefOi komia] en John heard that Mary had (sbjnct) come but b. J6n heyroi [ao Maria hafOi komii'll *en John heard that Mary had (ind) come but
hUn she hUn she
hafOi had hafOi had
ekki not ekki not
komii'l come komii'l come
In cases like these, the LDR goes together with the subjunctive: 9 (22)
J6ni heyroi [ao eg hefOiI*hafOi svikio sigil John heard that I had (sbjnctlind) betrayed self
It is not the case, however, that LDRs are mechanically conditioned by the subjunctive. We do not find them in adverbial clauses that contain subjunctives signalling unfulfilled conditions, for instance (23a), nor in adverbial clauses that contain sUbjunctives because the conjunction in question requires the subjunctive (23b) (cf. Thrainsson (1976a»: (23)
a. *J6ni John b. *J6ni John
yroi glaour [ef pu hjalpaoir seri] would be glad if you helped (sbjnct) self lykur pessu ekki [nema pu hjalpir Seri] finishes this not unless you help (sbjnct) self
But if sentences containing adverbial clauses are embedded under verbs of saying, for instance, the LDR can occur inside the adverbial clause (cf. Thrainsson (1976a: 230-1»: (24)
a. J6ni sagoi [ao hanni yroi glaour [ef pu hjalpaoir ser]] John said that he would be glad if you helped self
Long-distance reflexives and the typology of NPs
57
b. Joni segir [ao hann Ijtiki pessu ekki [nema pti hjalpir ser]] finishes this not unless you help self John says that he Second, in the light of some recent proposals suggesting some sort of movement analysis (movement in LF) to account for apparent long-distance phenomena involving anaphors (see Lebeaux (1983), Chomsky (1986a), and especially Pica (1987) (who discusses Icelandic)), it is perhaps particularly interesting to note that LDRs seem to occur freely in syntactic islands, as long as these islands are embedded under verbs of saying or believing, for instance. The subjects of such verbs can thus be the antecedents of LDRs that occur inside islands. We have actually already seen an indication of this in (24), since adverbial clauses are normally islands. This is illustrated further in (25-26), where it is shown that relative clauses and complex NP-clauses are extraction islands even when embedded under verbs of saying but in such contexts we can get LDRs inside them (see also Thrainsson (1976a: section 3.2)): (25)
a. Jon segir [ao pti hafir baria konuna sem hafi John says that you have (sbjnct) hit the woman that has (sbjnct) svikio pig]] betrayed you b. *Hverni segir Jon [ao pti hafir bario konuna [sem hafi svikio ti]]? who says John that you have hit the woman that has betrayed 'Who does John say that .. .' c. Joni segir [ao pti hafir bario konuna sigil] [sem hafi svikia John says that you have hit the woman that has betrayed self 'John says that you hit the woman that betrayed him (refl), (26) a. Jon segir [ao Maria trtii ekki peirri fullyroingu John says that Mary believes (sbjnct) not that claim [ao konan hafi svikio pig]] that the woman has (sbjnct) betrayed you 'John says that Mary doesn't believe the claim that the woman betrayed you' b. *I>igi segir Jon [ao Maria trtii ekki peirri fullyroingu [ao you says John that Mary believes not that claim that konan hafi svikio til] the woman has betrayed c. Joni segir [ao Marfa trtii ekki peirri fullyroingu [ao John says that Mary believes not that claim that sigi]] konan hafi svikio the woman has betrayed self
58
Hoskuldur Thrainsson 'John says that Mary doesn't believe the claim that the woman has betrayed him [refl],
Thirdly, as pointed out by Maling (1984), Hellberg (1984) and Barnes (1986), and discussed more extensively by Sigurosson (1986a), it is not the case that all long-distance reflexives in languages like Icelandic and Faroese need to be syntactically bound. Consider the following examples, for instance: Sigvaldii neitaoi jwi, ao petta va:ri vilji pj6oarinnar. Ao rninnsta kosti Sigvaldi denied it that this was will the nation's at least v::eri pao ekki sinni vilji (Icel.) was it not self's will 'Sigvaldi denied that this was the nation's will. At least it was not his [refl] will [he said]' ta iO hanni (28) ... hanni vildi ikki leypa fnt sinarii abyrgd, he would not run from self's responsibility now that he var komin soleiOis fyri vio Sigrid. was come so for with Sigrid Hon hevoi meiri krav upp a segi enn hin (Far.) she had more demand up on self than the other 'He would not run from his responsibility now that he had got into this situation with Sigrid. She had more right to him [refl] than the other [girl]' (29) Maria var alltaf svo andstyggileg. I>egar Olafurj k::emi segoi hUn Mary was always so nasty when Olaf came said she seri/*j areiOanlega ao fara ... (lcel.) himself certainly to leave 'Mary was always so nasty. When Olaf came, she would certainly tell himself [the person whose thoughts are being presented - not Olaf] to leave ... '
(27)
It seems very unlikely that the concept of syntactic binding can be extended in a meaningful way to cover intersentential reflexives of this sort, and they are not limited to Icelandic or Faroese. Similar examples can be found with Japanese zibun (cf. Sells (1987: 455)), for instance. The example in (29) is perhaps particularly interesting, however, since the antecedent is nowhere to be seen in the immediately preceding sentence. Although examples of this sort are not discussed in Anderson's paper (1986), he has later studied them in some detail and argued that they show the necessity of setting up an additional category, namely that of logophoric pronouns (class lectures 1985). This had also been argued by Maling (1984), with reference to West-African logophoric pronouns of the sort discussed
Long-distance reflexives and the typology of NPs
59
by Clements (1975), for instance. Sigurosson (1986a) discusses the (semantic! pragmatic) aspect of these Icelandic long-distance reflexives in some detail. Other discussions of 'logophoricity' in this (or a similar) sense include Kuno (1987: 136ff.) and Sells (1987), both dealing with a number of languages, including Icelandic. (See also note 12 below.) But it is important to note that the semantic conditions for these syntactically unbound cases oflong-distance reflexives in Icelandic (and Faroese) seem to be the same as those for the ones where a reflexive inside a finite (subjunctive) clause is syntactically bound by the subject of a higher clause in the same sentence. This is shown in some detail in Sigurosson (1986a) and it indicates that we do not want a special account of the syntactically unbound long-distance reflexives in these languages. What we need is rather an account that takes care of both the more familiar instances of reflexives inside finite (subjunctive) clauses bound by (subject) antecedents in a higher clause and the intersentential, unbound reflexives just observed. That seems to make any attempt to extend the syntactic binding domain beyond finite-clause boundaries in languages like Icelandic and Faroese, for instance, a dubious enterprise. That does not necessarily mean, however, that there cannot be languages that allow syntactic binding domains to be extended to a larger domain than the minimal finite clause but do not allow intersentential relations between reflexives and their antecedents. Rather than concentrating on what is special about the semantics of the longdistance reflexives in languages like Icelandic and Faroese, as many of the abovementioned studies have, I will in the following sections try to determine what, if anything, all types of anaphors and reflexives have in common and how they differ syntactically from pronominals and lexical NPs. That will lead to a new classification or typology of NPs, but the basic principles of BT will require only minimal modifications if we keep in mind that they are to be interpreted as necessary rather than sufficient syntactic conditions.
3 The common property of 'anaphors' in the extended sense 3.1 Bound-variable reading? Contrasts like the following have frequently been observed: (30) a. Johni shaved himself, and Peterj did too (= 'Peterj shaved himselfj, 'Peter shaved John') b. Johni saw hisi book and Peterj did too (= 'Peter saw his own book', or: 'Peter saw John's book') c. Johni said [that you had betrayed himi] and Peterj did too
*
60
H oskuldur Thrainsson (= 'Peter said that you had betrayed John', or: 'Peterj said that you had betrayed him j')
On the basis of examples of this sort, it has been claimed that anaphors like himself (normally) receive a bound-variable reading (at least in these 'VP-deletion' contexts)lO and hence we only get the 'sloppy identity' interpretation of sentences like (30a), whereas non-anaphors like the possessive his in (30b) and the pronominal him in (30c) may either be interpreted as bound variables with respect to their antecedent or be coreferential with it.ll Hence (30b, c) are ambiguous. Given this, one might expect the long-distance reflexives discussed in the preceding section to pattern with (local) anaphors in this respect. The sentences in (31), however, show that this is not the case for the Icelandic long-distance sig, although locally bound sig seems to behave like himself here: (31)
a. Joni rakaoi sigi og Peturj geroi pao lika John shaved himself and Peter did so too (=1= 'Peter shaved John') b. Joni sagoi [ao pu hemir svikio hannil og Peturj geroi pao lika John said that you had betrayed him and Peter did so too (= 'Peter said that you had betrayed John', or: 'Peterj said that you had betrayed himj') c. Joni sagoi [ao pu hemir svikiO sigi] og Peturi geroi pao lika John said that you had betrayed self and Peter did so too (= 'Peterj said that you had betrayed himj', or: 'Peter said that you had betrayed John')
In sentences like (31c), most speakers seem to find it rather easy to get both the sloppy and strict readings. Interestingly, however, it seems to be more difficult (if not impossible) to get the sloppy (i.e. bound-variable) reading in (32), where the antecedent does not c-command the long-distance reflexive: (32)
Skooun J onsi er [ao pu hafir svikiO sigi] og pao er skooun Peturs lika opinion John's is that you have betrayed self and that is opinion Peter's too 'John's opinion is that you have betrayed him and that is Peter's opinion too' (= 'It is Peter's opinion that you have betrayed John')
It is at least more difficult to get the bound-variable reading here and if it were indeed impossible it would be consistent with Reinhart's claim that 'the antecedent must c-command the pronoun for the sloppy identity interpretation to be obtained' ((1983c: 63) - see also Reinhart (1986: 125ff.)). But the available interpretations of (31c) and (32) show in any case that the long-distance reflexives in Icelandic are not
Long-distance reflexives and the typology of NPs
61
always interpreted as bound variables. 12 One should perhaps point out in this connection, however, that the long-distance reflexives still differ from pronominals in various respects, one being that they do not allow split antecedents: (33)
a.
Jani John b. *Jani John
sagoi told sagoi told
Mariuj Mary Mariuj Mary
[ao that [ao that
pu you pu you
hefOir had hefOir had
svikio betrayed svikio betrayed
paui+j] them sigi+j] self13
With coordinated antecedents, on the other hand, both pronominals and longdistance reflexives are possible, of course: (34)
a
Jani John b. J ani John
og and og and
Mariaj Mary Mariaj Mary
siigou said siigou said
[ao that [ao that
pu you pu you
hefOir had hefOir had
svikio betrayed svikiO betrayed
paui+j] them sigi + j] self
3.2 Incapable of 'independent' or 'deictic' reference? In his discussion of the BT classification of NPs, Chomsky (1981: 188) says that 'intuitively, anaphors are NPs that have no capacity for "inherent reference"'. This property of BT anaphors has frequently been noted and sometimes even used as a defining characteristic of anaphors (cf., e.g., Reinhart (1983c: 70), Giorgi (1984: 309)). If we take 'capacity for independent reference' to mean the capability of 'picking up a definite referent in the world, or [freely] in the previous discourse' (cf. Giorgi (1984: 309)), it should be clear that syntactically bound anaphors of the type Chomsky (1981) was discussing will never be able to do this since they will always be syntactically bound by some (c-commanding, coreferential) antecedent. And it is well known there is a clear contrast in this respect between pronominals and lexical NPs (R-expressions) on the one hand and anaphors like the English reflexive on the other. This can be illustrated by using a diagnostic test ofthe following type (cf. Hankamer & Sag (1976)): (35)
a. [Somebody in the audience gets up and leaves] Speaker: He is weird OR Where is the bastard going? OR Where is John going? OR I guess his patience ran out
62
Hoskuldur Thrainsson b. [Same situation] Speaker: *1 like himself (vs. I like him)
It is perhaps not a priori clear, on the other hand, how the long-distance reflexives would pattern in this respect, especially if we' consider the ones that are not syntactically bound at all. Recall, however, that the long-distance reflexives differ from pronominals in that they cannot just have any old syntactic constituent as an antecedent - if there is one, it seems to have to be the subject (subject of S or of NP) in Icelandic, for instance (cf. the discussion above). But what about cases like (29), repeated here for ease of reference? (29)
Maria var alltaf svo andstyggileg. Pegar Olafurj kfemi segoi hUn Mary was always so nasty when Olaf came said she seri/*j areioanlega ao fara ... himself certainly to leave 'Mary was always so nasty. When Olaf came, she would certainly tell himself [the person whose thoughts are being presented - not Olaf] to leave .. .'
Here there is no syntactic antecedent in sight for the reflexive. One might think, therefore, that cases of this type could be interpreted as instances of 'independent reference' . I want to argue, however, that cases like (29) are not instances of independent reference and there is a clear difference between pronominals and long-distance reflexives of this type with respect to the 'independence' of their reference (cf. also Sigurosson (1986a: 14)). Thus it is impossible to begin a book or a short story or a chapter or any such section of a narrative by something like (29). We need an introduction of some kind - an introduction where it is made clear whose thoughts are being presented. The use of pronominals needs no introduction of that sort. Hence (36) could very well be the beginning of a novel or of a smaller section of some narrative, and the pronominal hann has independent reference in the sense that it needs no previous mention but will become clear from the situation described. Given this introduction, we can then go on as in (29): (36)
Hanni la einn i myrkrinu og hugsaoi. Maria var alltaf ... 'He lay alone in the dark, thinking. Mary was always .. .'
This shows, then, that even the inter sentential reflexives cannot have independent or deictic reference in the sense that pronominals can. It is this property which seems to be common to all 'anaphors' in the extended sense - i.e. the NPs that are
Long-distance reflexives and the typology of NPs
63
not of the lexical NP or the personal-pronoun type - be they locally bound, longdistance bound or syntactically unbound. This must be made explicit in our classification of NPs. 4 Towards a new classification of NPs 4.1 Schematically What we have found so far, then, is that not all 'anaphors' in the extended sense, taking 'anaphors' now to mean the NPs not capable of independent reference, need to be syntactically bound. But it is also important to recall too that not all NPs that are capable of independent reference have to be free in a particular domain. More specifically, non-reflexive possessive pronouns need not be free at all in languages like English and German, for instance: (37)
a. Johni saw hisilj book b. Hansi sah seinilj Buch
The same is true for the 3rd person plural possessive in Danish: (38)
Dei sa deresilj bog they saw their book
These possessive pronouns are, however, like regular pronominals in that they have the capability of independent or deictic reference (cf. (35a) above). 1st and 2nd person personal pronouns in languages that do not have 1st and 2nd person reflexive pronouns behave the same way as the English (and German) possessive pronouns with respect to binding. In other words, they need not be free at all and in that respect they contrast with their English counterparts, for instance. This is illustrated in (39), with examples from Icelandic, Danish, and English: (39)
a. Egi rakaoi migi (IceI.) b. Jegi barberede migi (Dan.) c. *Ii shaved mei
Given this, we could present the observed types schematically, as shown in (40): (40)
_________ NPs-------capable of indep. ref.
~ need not be
must be free in some domain
Domain varies:
free in any domain
not capable of indep. ref.
--------------
must be synt. bound in some domain
Domain varies:
need not be synt. bound in any domain
64
Hoskuldur Thrainsson Examples: 2: his la: he Ib: Icel. hann Ger. sein Icel. eg lc: the bastard
3a: himself 3b: Dan. sig 3c: Ital. se (?)
4: Icel. & Far. sig. & more LDRs
Domains la and 3a:
'Governing category' (or some such welldefined local domain)
Domains Ib and 3b:
Minimal tensed clause (or something like that)
'Domain' lc:
Any domain
Domain 3c:
Minimal indicative clause (or some such definition) - see, e.g., Anderson (1986), Yang (1983), Giorgi (1984, and references cited there)
We have already presented some evidence for the classification given in (40), both with respect to the four basic types of NPs assumed there and the different anaphoric domains. As pointed out in section 2.2, there are several languages where the anaphoric domain for reflexive-like elements is not simply the governing category but rather an extended domain like the minimal tensed clause (domains Ib and 3b in (40)), even though something like governing category may seem the correct domain for reciprocals in these languages (cf., e.g., (8) above). It is also conceivable that rules of anaphora need to refer to a larger domain like the 'minimal indicative clause' (3c), although we saw in 2.3 above that such an extension would not do for Icelandic or Faroese, since the long-distance reflexives in these languages need not be syntactically bound at all. Sentences of the type given in (41) have been presented as arguments for the claim that something like the minimal indicative clause is the relevant domain for (the stressed) se in Italian (cf. Napoli (1979), Yang (1983) - compare also Giorgi (1984)): (41)
a. Gianni; e molto contento di sei/*j Gianni is very happy with self b. La signora; mij dice [PRO j di giacere presso di Sei/*k] the woman orders me to lie near self c. La signora; dice [che io giaccia presso di sei/*j] the woman orders that I lie (sbjnct) near self d. *La signora; dice [che io giaccio presso di sei/j] the woman says that I lie (ind) near self
Long-distance reflexives and the typology of NPs
65
For all I know, however, it may very well be that Italian se is like its Icelandic and Faroese counterparts in that it need not be syntactically bound at all, and that may be true of all 'logophoric' reflexives. It seems to hold for the logophoric ye in Ewe, for instance (cf. Clements (1975». If that is true, then we may not need a syntactically defined anaphoric domain like 3c at all. Now observe that we really should be more explicit about what it means that certain NPs 'need not be free in any domain' or 'need not be syntactically bound in any domain'. What is the relationship of these NPs to other types of NPs? Could they observe certain BT principles but not others? In particular, how could we specify them in terms of features like [±anaphoric] and [±pronominal]? Is the classification suggested here compatible with feature specifications of that sort? In the following subsection I shall show that it is indeed advantageous to continue to use binary feature specifications of that sort, and if we add two more binary features, we will arrive at a quite explicit and interesting typological classification of NPs.
4.2 A (partially) new classification in terms of features First, recall that lexical NPs (or R-expressions) are specified as [-an, -prj in standard BT. Since this seems to be a sort of a default value, it is perhaps somewhat strange that principle C refers specifically to this feature combination. More importantly, however, Lasnik (1986) has argued quite convincingly that we need to be able to express the fact that the so-called pronominal epithets (the bastard, etc.) do have certain things in common with names (John, etc.), as assumed in standard BT, but also behave like pronominals in certain ways. He suggests, therefore, that we need the binary BT-feature [±R-expression], and 'pure' R-expressions, like names, have the feature specification [-an, -pr, + R] but pronominal epithets are [-an, +pr, +R]. Given this, let us now assume that the property of being capable - or not capable - of independent reference in the sense discussed earlier can be expressed by a binary feature too, namely [±ind refJ.lfwe take [+anaphoric] to mean, as before, that the NP in question must be syntactically bound (in some domain), it is obvious that the feature specifications [+ind ref, +an] are incompatible. We can also assume that the feature specifications [-ind ref, + R] are incompatible, since R-expressions (= referential expressions) must be capable of independent reference (by definition, one would think). This will then give us the following possible classes of NPs (where a * indicates an impossible specification):
66
Hoskuldur Thrdinsson
(42)
NPs [-ind ref]
[+ind ref]
~ [-R]
~ [-R]
*[+R]
[+R]
~ [-an]
*[ + an)
/ [-pr] \
[+pr] 1.
2.
A [-an]
*[ + an]
~ [-pr]
[+pr] 3.
4.
~ [-an]
[+an]
A
[+pr] 5.
[-pr] 6.
~ [-pr]
[+pr] 7.
8.
Now let us assume the version of BT principles A and B suggested by Anderson (1986, cf. section 2.2 above; see also the discussion in Yang (1983), Mohanan (1982), and Harbert (1986) for other possible parametrizations of BT) and the version of principle C suggested by Lasnik (1986). That gives the following BT: (43) A revised BT: A A [+anaphoric] NP must be 1. bound in its governing category, or 2. bound by a superordinate subject within its anaphoric domain. B A [+pronominal] NP must be 1. free in its governing category, or 2. subject free (= not bound by a superordinate subject) in its anaphoric domain. C A [ + R-expression] NP must be free. 14 The categories defined by this BT and the feature system shown in (42) will then include the ones given in Table 3.2: Table 3.2. Revised BT - feature representation Feature complex 1. 2.
3. 4. 5. 6. 7. 8.
[+indref, [+indref, [+ind ref, [+ind ref, [-ind ref, [-ind ref, [-ind ref, [-ind ref,
+R, +R, - R, - R, - R, - R, - R, - R,
-an, -an, -an, -an, +an, +an, -an, -an,
+pr] -prJ +pr] -prJ +pr] -prJ +pr] -prJ
Categories
Some examples
pron epith true R-expr pronominals poss prons/l st, 2nd pers prons PRO(?), etc. anaphors pran LDRs(?) (logoph) LDRs
the bastard John he his; sein; eg ham selv himself aapa!J; !aan (?) sig
Long-distance reflexives and the typology of NPs
67
A few of the claims and suggestions made in Table 3.2 need further clarification or discussion. We will proceed in the order given in the table. First, we should say something about the distinction between categories 1 and 2, i.e. pronominal epithets and true R-expressions. For one thing, it has been observed that pronominal epithets can have antecedents like pronouns, as long as they are not syntactically bound by them. Names do not share this property (cf. Lasnik (1986)): (44) a. Johni promised me to come to my talk but the bastardi never came b. ?Johni promised me to come to my talk but Johni never came Thus apart from condition C, there seems to be some sort of a repetition constraint on names that does not hold for pronominal epithets. This difference is not accounted for by the binding theory - it is mainly shown here as an argument for the claim that these types of NPs differ in some respects at least. More importantly, however, there are languages that do not seem to observe condition C (cf. note 14). If these languages do observe condition B of BT, we would expect that pronominal epithets would have to be free in their governing category but names would not have to be, given the feature specifications assumed here. According to Lasnik (1986), this seems to be what we find in Thai: (45)
a. C:l:lni ch:l:lp C:l:lni John likes John b. C:l:lni khit w;a C:l:lni chalaat John thinks that John is smart c. *C:l:lni ch:l:lp ?iiyba~ John likes the nut d. C:l:ln khit wlia ?iiyb;a chalaat John thinks that the nut is smart
The feature specification of pronominals (category 3) needs no further comment, but category 4 is the possessive pronouns and personal pronouns that need not be free in any domain (cf. (37), (38), and (39) above) and these receive here the default specification [ - R, - an, - pr]. That is meant to imply that they are not subject to any of the rules that refer to the plus-values for these features. That seems to be exactly the effect we want. Turning now to category 5 in Table 3.2, the obvious suggestion would seem to be PRO, in accordance with standard BT. This is somewhat questionable, however. First, there has been some controversy as to whether this is the right way to account for the distribution of PRO (cf. Manzini (1983), Bouchard (1985)). In addition, for the PRO-theorem to follow from BT, it is crucial that the anaphoric
68
H oskuldur Thniinsson
domain relevant for [+anaphoric] NPs and [+pronominal] NPs be governing category, as in standard BT, but we have seen ample evidence above that this is not always the case. Consequently, we have parametrized principles A and B of our BT so it is no longer clear that the PRO-theorem would follow. Instead, new possibilities are predicted for the specification [+an, +pr]. It should, for instance, be possible to find an NP that was so specified where [+anaphoric] implied that it had to be bound in its governing category (condition Al in (43) above) but [+pronominal] meant that it had to be subject free in its anaphoric domain (condition B2 above). Something like this seems to be a correct statement of the conditions on Danish ham/hende selv 'him/her self', for instance, as indicated by the following (cf. Vikner (1985»: (46)
a. Susani fortalte Annej om hende selv*ilj Susan told Anne about her self b. *Susani bad Annej om [PROj at ringe til hende selvilj] Susan asked Anne to ring to her self
As the reader can undoubtedly figure out, there should be other possibilities for category 5 too, given the parametrized versions of conditions A and B and the varying domains for pronominals and anaphors. It should also be mentioned here that Mohanan (1982) argues that there are 'pronominal anaphors' in Malayalam, i.e. overt NPs that must be specified as [+an, +pr], and the reason is obviously the same as the one we have found here - i.e. that the domains where anaphors must be bound and pronominals must be free do not always coincide. We will return to Malayalam in our discussion of category 7 below. Nothing further needs to be said about category 6, except that the reader should remember that we are assuming possible variations in the anaphoric domain, so this category will not only include 'regular anaphors' bound in their governing category but also the reflexive pronouns of Anderson (1986), for instance, i.e. NPs obeying condition A2 of (43) above, and possibly other 'long-distance' reflexives, as long as they need to be bound within a syntactically specifiable domain. The feature specification for category 7 in Table 3.2 above would mean that we had an NP that need not be syntactically bound in any domain (that is what we take [-anaphoric] to mean) but had to be free in the domain relevant for pronominals in that language. This seems an entirely plausible possibility. Thus it has been reported in the literature that the 'anaphor' aaparj in Marathi is a long-distance anaphor in that it can occur in a finite clause with an antecedent in the matrix clause but it cannot have its antecedent in the same clause (cf. Wali (1979: 405-6); see also von Bremen (1984: 203, 212»:
Long-distance reflexives and the typology of NPs (47)
a.
69
Minilaai vaatta [ki aapaJ;li cuk keli] Mini thinks that self mistake made
b. *Mininei aaplayaalaai ba<;lavle Mini self to hit (intended sense: 'Mini hit herself') Similarly, Mohanan (1982) observes that there is what he calls a 'pronominal longdistance anaphor' in Malayalam, namely the !aan mentioned in Table 3.2. It must be free in the minimal clause that contains it, just like pronouns (Mohanan (1982: 169)), but it requires an antecedent of some sort (167-8), like an anaphor. This antecedent can, however, 'be removed from the anaphor by any number of clauses' (175). The following sentence is meant to show how the combination of these two conditions works (cf. Yang (1983: 185)): (48) [[Moohani !anne*ilj/k l].uHi el].l].g] ammaj acchanootg paraiiiiu el].l].g] Mohan self pinched that mother father said that raajaawingk !ool].l].i king felt 'The king felt that the mother told the father that Mohan pinched him/her (i.e. pinched the mother or the king)' Now if these LDRs in Marathi and Malayalam just discussed need not be syntactically bound at all, then [-ind ref, - R, -an, +pr] would be the correct specification for them in our framework, as suggested, since they apparently must be free in the domain relevant for pronominals. If they must be syntactically bound, on the other hand, then they are of type 5 in our classification. I do not have any evidence that would determine this for the Marathi case. But while Mohanan (1982) claims that !aan normally obeys a c-command condition, he also points out that the following is good, where the antecedent in fact does not c-command the reflexive (op. cit. 171): (49)
Moohantei wiswaasam [!aani ghiiranaaJ;lg el].l].g] aalJ. g that IS Mohan's belief self brave is 'Mohan's belief is that he is brave'
This is very reminiscent of Icelandic sentences like (50) (cf. Maling (1984: 222); see also (43) above) and the semantic/pragmatic conditions appear to be the same (Mohanan (1982: 171)): (50)
Skooun Siggui er [ao sigi vanti h::efileika] opinion Sigga's is that self lacks talent 'Sigga's belief is that she lacks talent'
70
Hoskuldur Thrainsson
Hence I suspect that the Malayalam long-distance anaphor could very well be like its Icelandic counterpart in that it need not be syntactically bound at all but rather obeys pragmatic (logophoric) restrictions. More evidence is needed, however, to determine this. But if this is the case, then the feature specification of type 7 would be the correct one, as suggested in Table 3.2. Otherwise it would be of type 5. Finally, category 8 includes the long-distance reflexives ofIcelandic and Faroese and these have a minus-value for all features, indicating that they do not obey any of the syntactic binding conditions of (43) above. They do, however, obey certain pragmatic or logophoric conditions, as mentioned in section 2.3 and discussed most recently and extensively by Sigurosson (1986a), Sells (1987), and Kuno (1987). We will return briefly to that issue in the final section.
5 Conclusion I believe that we have seen in this chapter that there are more types of NPs in the languages of the world than the standard BT might lead us to expect. IS Some of the observed cross-linguistic differences can be accounted for by assuming parametrized binding conditions while others involve different specifications (or nonspecifications) of the binary classificatory features assumed here. The only 'new' feature proposed in this chapter is the lexical feature [±independent reference]. We have seen that this is closely related to the BT -feature [±anaphoric] in that no [ + anaphoric] NPs (= NPs that must be syntactically bound) will be [+ independent reference]. Similarly, there is a close relationship between the BT-feature [±R-expression] and our lexical feature [±independent reference], since no NPs having the feature specification [+ R-expression] will have the feature value [-independent reference]. Despite this, we have seen that these features are not identical, since it is not only [+anaphoric] NPs that are [-independent reference] nor is it only NPs marked [+R-expression] that are [+independent reference]. This indicates, I believe, that there is a close relationship between BT and lexical content of NPs but BT is nevertheless autonomous in the sense that not all binding properties of NPs follow from their lexical content. If they did, it would be difficult to imagine how non-overt NPs could have different binding properties. We have, however, said very little about the binding properties of non-overt NPs here. Lasnik (1986) contains some discussion of these within a framework that assumes the binary BT-feature [±R-expression] adopted here. It should be noted, however, that some of his more difficult cases disappear in the present framework because of the interaction of the lexical feature [±independent reference] with the other three features. An interesting suggestion can also be found in Rognvaldsson (1988) to the effect that object gaps in Modern Icelandic could be
Long-distance reflexives and the typology of NPs
71
analysed as being of category 7 in our classification whereas empty subjects (in prodrop languages) would presumably be of category 3, the latter having 'independent reference' in our sense but the former not. Finally, I would like to emphasize that there is a lot more to be said about longdistance (logophoric) reflexives than what has been said here. But I would like to maintain that this is about as much as one would like to say about their syntactic binding properties. That does not mean that they do not obey any syntactic conditions, only that they do not have to observe the syntactic binding conditions on [+anaphoric] NPs. In general, we can look at the BT as a set of necessary syntactic binding conditions that must be observed by the NPs that are subject to them. In the case of regular, clause-bounded reflexives, there may also be certain semantic requirements that need to be fulfilled. Hence not all objects and not even all subjects will be equally good antecedents for reflexives (or anaphors), as has frequently been observed (see, e.g., Maling (1986), Rognvaldsson (1986), Kuno (1987); cf. also Hellan (1986a)). In that sense, there is certainly a semantic/pragmatic side to clause-bounded reflexives in addition to the syntactic binding conditions that these typically observe (see also Sigurosson (1986a) and von Bremen (1984)). But in this chapter I have tried to argue that the syntactic binding conditions on anaphors in the standard BT should not be extended in an attempt to cover the truly long-distance reflexives that need not be syntactically bound at all. That might seem obvious, but then again I am sure that there are ingenious people who can think of fairly abstract syntactic analyses that capture some of the facts discussed here. But it seems to me that one should rather continue to try to get a better grasp of the semantic/pragmatic (logophoric) concepts involved (cf., e.g., Sigurosson (1986a), Sells (1987), and Kuno (1987), while accepting the claim made here that these do not obey the syntactic binding conditions on anaphors, although some of them could obey the condition on pronominals, as we have seen. They may also be sensitive to other syntactic or structural properties not discussed in the present chapter.
Acknowledgements Some of the ideas in this chapter were presented at the Third Workshop on Comparative Germanic Syntax in Abo in June 1986, at the Seventh Biennial Conference of Teachers of Scandinavian Studies in the British Isles at University College London in March 1987, at the Groningen Workshop on Long-Distance Anaphora in June 1987, and at linguistics colloquia in the fall of 1987 at Boston University, University of Massachusetts at Amherst, CUNY Graduate Center in New York, Brandeis University, Cornell University and Harvard University. I would like to thank all the audiences for valuable comments, both public and private ones. Thanks are also due to Guy Carden/Joan Maling, Eirikur Rognvaldsson, Peter Sells
72
Hoskuldur Thrainsson
and Halld6r Armann Sigurosson for written comments on a related paper, and to Pierre Pica, Tanya Reinhart and Noam Chomsky for discussions about points made in this chapter. All these people have saved me from making various mistakes, although I have not always followed their advice. The first draft of this chapter was written while I was a visiting scholar in the Linguistics Department at Harvard University in the fall semester of 1987, and I am grateful for having had the opportunity to work there.
Notes 1. Anderson does not refer to features like [±anaphoric] or [±pronominal] but rather uses the labels anaphor, reflexive pronoun, and pronominal (pronoun), as explained in the text. I have taken the liberty of reformulating his principles in terms of features to make the comparison with (2) above easier. We will return to a discussion of features in the following sections, especially section 4. 2. Anderson gives the following definition of superordinate subject (1986: 76): 'A is a subject superordinate to B if either (a) A is the subject of the S or NP in which B appears; or (b) A is the subject ofthe verb which governs an S or NP in which B appears.' 3. Actually, the facts are somewhat more complicated. First, it seems that we must assume that Icelandic simple reflexives are 'ambiguous', as Anderson realizes (1986: 82), in that they also have the purely anaphoric property of being able to refer to the object (although there are some restrictions on that, cf. the discussion in Maling (1986), and Rognvaldsson (1986)):
(i) a. Eg sendi honumi gallabuxur:i sigi I sent him bluejeans for himself b. Eg baroi hanai mea dukkunni sinnii hit her with doll self's 'I hit her with her doll' Second, there are 'complex' forms in Icelandic and the other Scandinavian languages, involving the element s}dlfur (Ice!.) 'self' (Far. s}dlvur, Dan.lNorw. selv, Swed. s}iilv), and these always seem to be strictly local, i.e. having their antecedent inside their governing category, but their binding properties in other respects seem to depend on what the other element is - i.e. whether it is the pronominal hann 'he' (or the equivalent in the other languages) or the reflexive sig 'self' (or the equivalent in the other languages). We will return briefly to some of these in section 4 below. There are also semantic (or thematic?) constraints that seem to determine when one can use the simple and when the complex reflexive. These will not be discussed here (see Vikner (1985), Hellan (1986a) and Hestvik (1987) for discussion; cf. also Pica (1984), and Everaert (1986a)). Third, one needs to say that the 'anaphoric domain' for reflexives may also vary, since it seems to be the minimal tensed clause for the Mainland Scandinavian languages whereas something like the minimal indicative clause would seem to suggest itself so far for Icelandic and Latin, for instance (perhaps also for Italian se, according to Napoli (1979)). We will return to this question of the extended domain for these languages at the end of this section (but see also Sigurosson (1986a)). Finally, it must be
Long-distance reflexives and the typology of NPs
73
admitted that the requirement that certain (long-distance) reflexives be bound by subjects does not follow from anything in this kind of approach but must be stipulated in the binding conditions. For a rather different way of achieving some long-distance (and subject-) binding effects see Lebeaux (1983, 1985) and Chomsky (1986: 164-84). See also Pica (1987). We will briefly discuss the plausibility of such approaches in section 2.3. 4. I am here assuming the 'dialect' of Icelandic that requires case agreement between the two parts hvor 'each' and annar governed by the verb raka 'shave'. There are also speakers who prefer to have the first part hvor agreeing with the subject, which in this instance would give nominative. These speakers will not allow the first part to occur inside a prepositional phrase, so we get the following contrast: (i)
a. I>eir (N) they b. I>eir (N) they
litu hvor (N) a annan (A) looked each at (the) other litu a hvorn (A) annan (A) looked at each other
As suggested by the English glosses, the first variant is probably more similar to the English each . .. the other, which may not have exactly the same properties as the usual reciprocal. Hence I have chosen the Icelandic version which corresponds more closely to the well-known reciprocal, although there are probably some differences between Icelandic and English reciprocals, which need not be discussed here. 5. Again the formulation is mine rather than Anderson's - cf. note 1 above. 6. Note, for instance, that while (13) suggests that pronominals must be free in their governing category in Norwegian and Danish, and not merely subject free in their anaphoric domain as argued for their Icelandic counterparts, Anward (1974) claims that local (i.e. clause-bounded) 'reflexivization with non-subject antecedents' is 'optional' in Swedish, which would mean in the present framework that pronominals are only subject free in their governing category in Swedish. That would follow neither from Bl nor B2, given sentences like (l2c). All the examples Anward gives (1974: 2lff.) involve possessive constructions like (i), which seem to be good in Norwegian too, for instance (Arild Hestvik p.c.): (i)
Vi gav honomi hansili bOssa (vs. sinil*i bossa) we gave him his gun
If this is true, something more needs to be said, but one must also try to distinguish between pragmatic preferences and syntactic requirements here, as always (see section 5). 7. Anderson (1986) suggests a syntactic account of the LDRs in Icelandic and he assumes that Icelandic subjunctive clauses are 'ambiguous' in the sense that they can either count as anaphoric domains for reflexives or not. When they do, the pronominal inside the subjunctive clause is subjective free (as it is in infinitival clauses like (11», but when they do not we can have pronominals coreferential with the matrix subject. Hence we get the apparent non-complementarity of pronouns and reflexives observed in (15a) vs. (18). 8. There are, however, some speakers for whom this correlation between LDRs and subjunctives is less strict in that they allow LDRs in the indicative complements of some verbs, including vita 'know' (cf. Sigurosson 1986a: 8). I am ignoring this dialect here. 9. But note that there is not a strict correlation between LDRs and non-factivity either,
74
H oskuldur Thrainsson since true factive verbs (or emotive verbs - cf. Kiparsky & Kiparsky (1970» like harma 'regret' typically take (a special type of) subjunctive complements (cf. Thniinsson (1979: 190ff.», and we can find LDRs inside these too: (i)
Joni harmar [ao pu skulir hafa svikiO siga John regrets that you should (sbjnct) have betrayed self
Sigurosson (1986a: 25) claims that the difference between true factive verbs and semifactive verbs like know or predica tes like be obvious is that in the latter the speaker is not only presupposing the truth of the complement but also asserting it or 'taking responsibility' for it. 10. Sells, Zaenen & Zec (1986: 175) argue that English sentences like the following: (i)
John defends himself better than Peter
have both the sloppy identity and strict identity (coreferential) reading (in addition to 'object comparison'). The corresponding Icelandic sentence in (ii) can only have, I believe, the 'sloppy identity' reading: (ii)
Jon varoi sig betur en Haraldur John defended himself better than Harold (Nom) (* 'better than Harold defended John')
(The 'object comparison' reading is ruled out by case marking - to get it we would need the Acc. Harald.) 11. For discussions of bound-anaphora interpretation, sloppy identity and related issues, see e.g. Reinhart (1983c), and Sells (1986, 1987), and references cited there. 12. Sells (1987) presents a very interesting analysis of logophoricity. His main point is that there is 'no unified notion of logophoricity per se and that logophoric phenomena are instead a result of the interaction of [... ] more primitive notions: the source of the report [his SOURCE], the person with respect to whose consciousness (or 'self') the report is made [his SELF], and the person from whose point of view the report is made [his PIVOT - roughly Kuno and Kaburaki's 'empathy' (1977)], (Sells 1987: 445). These primitive 'roles' in discourse may figure in different ways in the 'role-oriented anaphora' found in various languages. Sells maintains that 'PIVOT is the crucial role in zibunbinding: for Icelandic it would seem that SELF is the important one' (1987: 473). He wants to argue, however, that alliogophoric binding (binding to SOURCE, SELF or PIVOT) is variable binding. We have seen that this does not seem to be the case for Icelandic sig. The following sentences and their interpretation (originally from Thrainsson (1976a, c» play an important role in Sells's argumentation for his claim: (i)
a.
Aoeins Only b. Aoeins Only
J oni John Joni John
telur believes telur believes
[ao that lao that
Maria Mary Maria Mary
elski hanna loves him elski siga loves self
Following Thrainsson (1976a), Sells (1987: 467) claims that (ib) can only have the bound-variable interpretation whereas (ia) only has the 'referential' one. While it is true that the bound-variable reading is much more prominent than the referential one for (ib), it is probably possible to get them both. Similarly, it is probably not the case that
Long-distance reflexives and the typology of NPs
75
the reflexive is absolutely required in (ii) as Sells (ibid.) reports (presumably based on a footnote in Thr:iinsson (1976c»: (ii)
Enginni telur [an Maria elski sig)?hannil nobody believes that Mary loves self/him
13. One could argue that this follows from the fact that objects cannot function as antecedents oflong-distance reflexives (cf. (19) above). 14. Actually, Lasnik (1986) argues that condition C needs to be parametrized too since it holds in English but not in Thai, for instance. He suggests, on the other hand, that it may hold universally that pronouns cannot bind R-expressions - this additional requirement possibly being a part of a more general prohibition against the binding of a more referential expression by a less referential one. We need not go into details of this sort here, however. 15. I do not pretend to have discussed all the types of NPs that may exist. Thus we have not discussed expletive pronouns nor considered where they would fit into the proposed typology. 'Indefinite' pronouns like English one have also been left out. Finally, note that Icelandic sig can sometimes be translated by English 'oneself' and then it can have arbitrary PRO as antecedent (/lao er galt [PRO ao raka sig i sturtuJ 'It is good to shave oneself in the shower'). More puzzling are examples like this: (i)
pan er veriO an raka sig there is being shaving oneself 'One is shaving oneself'
I leave these for further research.
4 Contextual determination of the anaphorl pronominal distinction Martin Everaert
1 Introduction
A lot of work on long-distance binding in generative grammar is concerned with the proper formulation of parameter setting in binding theory (BT), i.e. what are the parameters and which values can these parameters have? It seems to me that there are at least four primitive notions in the binding theory as developed in Chomsky's Lectures on Government and Binding which are potential candidates for parametric variation: (1) a. b. c. d.
the classification of pronominal elements the notion 'governing category' the notion 'accessible subject' the notion 'bound/free'
To account for cross-linguistic variation, parameter choices for each of these have been proposed in the literature. In this chapter I will limit myself to discussion of (la), formulated as in A: A.
What kind of anaphor/pronominal types must we distinguish?
In other words, is the classification of pronominal elements in BT subject to parametric variation? Does the language learner have to make a choice between, for instance, b-anaphors vs. d-anaphors (Vikner (1985) following work by Hellan (1983», non-pronominal anaphors vs. pronominal anaphors (Bok-Bennema (1985), Huybregts (1979), Vat (1980», connectedness anaphors vs. containment anaphors (Hellan (1988», or normal anaphors vs. long-distance anaphors (Anderson (1986), Falk (1984), Koster (1987), Rebuschi (1988), Yang (1983»? Question A presupposes taking a particular position with respect to the closely related question B:
77
78
B.
Martin Everaert
Do we have to distinguish several types of reflexivization?
Is there reason to believe that we need to distinguish between non-clause-boundedl logophoric reflexivization and clause-bounded reflexivization, each subject to specific structural, thematic or discourse conditions (cf. Giorgi (1984), Maling (1984), Thrainsson (1976a), Zribi-Hertz (1988))?! In section 2, I will briefly discuss the potential motivation for distinguishing two types of reflexivization. Subsequently, section 3 will give a general outline of the approach I want to pursue in answering the question whether or not we have to distinguish more than one type of anaphor or pronominal. More specifically, I will be concerned with a description of the pronoun system of Dutch and Frisian, a language closely related to Dutch and spoken in the northern part of Holland. Section 4 presents the distributional properties of the Dutch anaphors zich and zichzelJ, potential candidates for an LD-anaphor/SD-anaphor opposition, and section 5 will focus on the Frisian pronoun system where the anaphor/pronominal distinction does not seem to be so clear-cut. On the basis of these facts a reformulation of BT will be proposed in section 6 and it will be shown how this reformulated BT offers a straightforward account of the Dutch and Frisian reflexivization facts of sections 4 and 5. In the last section I will discuss some problems and possible extensions of the analysis proposed.
2 Two types ofreflexivization? In discussing reflexivization facts from Icelandic, Thniinsson (1976a) was the first in generative literature to make a distinction between two types of reflexivization. He argued that Icelandic has an LD-rule which is sensitive to semantic/pragmatic factors which do not seem to play any role in the 'normal' SD-rule. Maling (1984) develops this position and argues that this non-clause-bounded use of the reflexive anaphor is reminiscent of the logophoric pronoun system of West African languages as described in Clements (1975) (cf. Giorgi (1984) for a similar approach to Italian). What are the arguments to distinguish two reflexivization types? I will only briefly mention some of the potential arguments discussed in the literature on Germanic and Romance reflexivization: (2) a. In the case of LD only subjects are potential antecedents; SD also allows object antecedents. b. Adverbial complements are opaque for LD but not for SD. With the former, opacity is relaxed under conditions which seem to be better formulated in semantic/pragmatic than syntactic terms.
Contextual detennination o/the anaphorlpronominal distinction
79
c. The complementary distribution of anaphors and pronominals typical of SD breaks down in the case of LD. d. LD does not allow a derived subject as antecedent, SD does. e. LD but not SD allows discourse antecedents given appropriate setting. These arguments may be challenged on two points. First of all, they are, on an observational level, not always as straightforward as has been suggested. For instance, one might very well question the validity of (2a). It seems that object antecedents are marked, at least in the Germanic languages, even in the case of SD (cf. Everaert (1986a), Grewendorf (1984), Hellan (1988), Maling (1986). Rognvaldsson (1986), Sigurosson (1986a), Vikner (1985)).2 With respect to (2b), Everaert (1986a, b) argues that it is possible to account for the behaviour of reflexives in adverbial complements without resorting to the SD/LD distinction or semantic notions (cf. Van Steenbergen, chapter 11). Similarly, the non-complementary distribution of anaphors and pronominals in LD (2c) can be explained syntactically without crucially making use of an SD/LD distinction (cf. Anderson (1986), Everaert (1986a)). Exceptions to (2d) have been noted as well (Koster (1987: 329), Maling (1984: 239)). More importantly, however, the tests in (2) do not uniformly discriminate between SD and LD. By this I mean that it is not always clear where the demarcation line between SD and LD is to be drawn. Suppose we divide the reflexivization facts as in (3): (3) a. clause-bounded reflexivization b. reflexivization into non-tensed complements (i.e. nominal complements, small clauses, infinitives) c. reflexivization into subjunctive complements In her discussion of Icelandic, Maling (1984) draws a division between (3a, b) on the one hand (i.e. SD), and (3c) on the other (i.e. LD). For Italian, Giorgi (1984) and Napoli (1979) make a distinction between (3a) and (3b, c) for the SD/LD distinction. Furthermore, if test (2a) discriminates between SD and LD, the latter includes (3b, c) but not (3a) (cf. Anderson (1986), Hellan (1988), Napoli (1979)); the same holds for test (2d) (cf. Belletti & Rizzi (1988), Napoli (1979)). Test (2c), however, discriminates (3a, b) from (3c) in Icelandic (cf. Thniinsson (1979)), but (3a) from (3b,c) in the other Scandinavian languages (cf. Everaert (1986a)). At first sight, test (2e) seems to give some motivation for a distinction between SD and LD. For those languages that make an SD/LD distinction (2e) appears to hold for (3c) exclusively. On the other hand, it has been observed that English, generally considered to be a language without an SD/LD distinction, also allows discourse
80
Martin Everaert
antecedents (cf. Kuno (1987: 125), Zribi-Hertz (1988», which casts doubt on the reliability of this test as a means to determine the existence of LD. The conclusions seem to be that the SD/LD distinction has not been clearly defined and that there are no compelling observational reasons to distinguish two types of binding in accounting for the distribution of anaphors and pronominals in the Germanic and Romance languages. If this is the case, there is a priori reason not to make an SD/LD distinction. Postulating two distinct BTs would mean a considerable extension of the primitive notions available in universal grammar for BT. Let me briefly illustrate the point (see also Hoekstra (1987». Giorgi (1984) claims that, apart from the regular notion A-bound for SD, the notion P-bound needs to be introduced for LD. P-bound is defined with the help of the primitive notions prominent argument and P-domain, an extension of the descriptive apparatus. On an explanatory level, it is always best to pursue an approach which reduces the principles necessary, if possible. Following the line of argumentation of Sigurosson (1986a) - although diametrically opposed to his ideas on the semantic nature of reflexivization - and Pica (1987), I will thus assume that no SD/LD distinction should be made. This brings us back to the question whether or not one should distinguish two types of anaphors. It is evident that, once the idea is rejected that there are two separate binding processes SD and LD, there is no principled reason to make an SD-anaphor/LD-anaphor distinction. Still, one might want to argue for a distinction between LD-anaphors and SD-anaphors without directly adopting two separate binding theories. In the next section I will show that such an approach is, apart from considerations of elegance, undesirable because it does not allow one to make the right generalizations.
3 SD-anaphors vs. LD-anaphors It has frequently been observed that only specific anaphors participate in LD (cf. Yang (1983». In the Germanic languages only simple reflexives - elements like Dutch zich, Norwegian seg and Icelandic sig - and reflexive possessives - such as Norwegian sin and Icelandic sinnlsinaletc. - permit long-distance binding (cf. (4»; reciprocals - Dutch elkaar, Norwegian hverandre - and complex reflexives - Dutch zichzelJ, Norwegian seg selv and Icelandic sjdlfan sig - never do (cf. (5»: (4)
a. b.
Olai bad oss snakke om segi Ola asked us to talk about himself Olafuri segir ao b6kin Stni se ennpa till solu Olaf says that his book is still for sale
Contextual determination o/the anaphorlpronominal distinction (5)
a. b.
81
*Dei bad Ola korrigere hverandrei they asked Ola to correct each other *Jani hoorde mij over zichzelfi praten Jan heard me about himself talk
Pica (1984) observes that the French clitic reflexive se and the Italian clitic reflexive si do not participate in LD (cf. (6)). The same holds for those simple reflexives which are used in inherently reflexive constructions (cf. Everaert (1986a), Hellan (1980, 1988), Vikner (1985)): (6)
a. b.
(7)
a. b.
*Oni ne doit pas dire aux gens de sei donner de l'argent One shouldn't say to people to give money to oneself *Carloi mi invita a guardarsii allo specchio Carlo invites me to look at himself *Jani liet mij zich i verslikken Jan made me choke *Peteri overtalte Anne til at sId fra sigi Peter persuaded Anne to defend himself
Suppose one accounted for these facts by assigning to the reflexives in (4) and to the reflexives and reciprocals in (5-7) the features <+ LD-anaphor) and <- LDanaphor) respectively. Although descriptively adequate, it clearly would offer no insight whatsoever. A more principled reason must be found why the facts are as exemplified in (4-7) and not otherwise. Or to put it differently, why is LD limited to a specific class of anaphors (cf. Pica (1987))? Let us begin, on an observational level, by giving a classification of anaphors taking into account their morphological structure and their syntactic behaviour apart from BT. Disregarding unchallenged 'reflexive' morphemes like, for instance, Russian -sja, Swedish -s and Icelandic -sk, this would result in the table shown in (8):3 (8) a. b. c. d. e. f.
special clitic reflexives: French se, Italian si, etc. simple clitic reflexives: Dutch zich, Norwegian seg, etc. non-complex reflexives: French soi, Italian se, German sieh, etc. Dutch zichzel/, Norwegian seg selv, etc. complex reflexives: non-complex reciprocals: Polish siebie, Dutch elkaar, etc. complex reciprocals: 4 Icelandic hvor annar, Italian ['uno . .. l'altro, etc.
On an observational level the following distinction can be made: the anaphors in (8a) and (8d-f) are SD-anaphors only; the anaphors in (8b, c) are LD-anaphors (and occasionally SD-anaphors). The approach I want to pursue is to account for
82
Martin Everaert
the behaviour of anaphors without resorting to the classification in (8) in BT. Instead, the binding theoretic differences between these anaphors should follow from their intrinsic properties, if possible. 5 In the next sections I will develop such an approach, limiting myself to reflexivization facts from Dutch and Frisian. As a consequence I will be concerned only with anaphors belonging to the classes (8b) and (8d). In this section, however, I will briefly say something more about the behaviour of the anaphors in classes (8a), (8c) and (8e, f). If special clitic anaphors of class (8a) like French se or Italian si were to be longdistance bound, they would violate the well-known constraint in (9), from Napoli (1973): (9) No reflexive may cliticize to a verb which has a subject not coreferential with the reflexive at the time of clitic placement. An explanation for this descriptive statement might be the following. It has been argued by Burzio (1986) and Zubizarreta (1985) that selsi is a morpho syntactic element. As such it should be subject to rules of both syntax and morphology. The partly syntactic nature of se makes it subject to binding theory and thus in principle to long-distance binding. However, as a morpheme, se must be lexically related to the subject of the verb it is attached to (cf. Burzio (1986)). It seems reasonable to assume that the outcome of the syntactic constraints on se must, in some sense, be compatible with its behaviour as a morpheme. From this it follows straightforwardly that se must always be bound by the closest subject. A similar treatment might well hold for the inherently reflexive use of the simple clitics in (7), if we assume that these reflexives are semi-morphemes in these cases. Coopmans (1985) defends such a position for Dutch zich. In our view, there is little to say about the 'normal' reflexives of (8c). They should in principle be able to participate in both SD and LD. And, apart from language-specific structural considerations, they seem to do so. In Yang (1983) it is claimed that reciprocals (i.e. anaphors oftype (8e) and (8f)) are SD-anaphors cross-linguistically. If we accept his claim, it remains to be explained why this should be SO. 6 For the reciprocals of class (8f) one might attribute this to their complex status. The reciprocals in this class are formed by a conjunction of two independent elements. English each other, for instance, can be divided into each and other. The same holds for Italian l'uno ... l'altro and Icelandic hvor annar (cf. Everaert (to appear)). Belletti (1983) argues that the anaphoric status of Italian l'uno ... l'altro follows from the conjecture that at LF, l'uno obligatorily has to move to a nearby subject just like subject-oriented quantifiers such as tutti 'all' (see also Heim, Lasnik & May (1988)). Compare the partial LFrepresentations (lOb) and (llb):
Contextual detennination of the anaphor/pronominal distinction (10)
a.
(11)
a.
83
I miei amici hanno parlato tutti dello stesso problema my friends spoke all of the same problem [NP[tutti]i [NP i miei amici]i] hanno parlato [pp ei [pp dello stesso problema]] I miei amici hanno parlato l'uno dell'altro my friends spoke of each other [NP [l'uno]i [NP i miei amici];] hanno parlato [pp ei [pp del' [NP altro]]]
Belletti argues that the trace left behind by movement at LF is an NP-trace and that the anaphoric interpretation of l'uno ... l'altro in (11) has to be attributed to the relation between the moved NP and its trace created at LF. It is a well-known fact that this relation is always clause bound for reasons that probably have nothing to do with BT. Following Belletti's argumentation, complex reciprocals like l'uno ... l'altro have to be clause bound (cf. Van Riemsdijk (1985) for a different approach). However, there seems to be more to it. Even in the case of non-complex reciprocals of class (8e), LD is excluded. This can be illustrated with the following case from Polish. The anaphor siebie in Polish can be interpreted as both a reflexive and a reciprocal: (12)
Kochamy siebie 'We love ourselves/each other'
LD is only allowed in the reflexive reading. Contrast (13a) and (13b) (cf. ReindersMachowska, chapter 6): (13)
a. Cht0PCYi czytali dziewczatj wspomnienia 0 sobieilj the boys read the girls' memories about them/themselves sobie*ilj b. Cht0PCYi czytali dziewczatj wspomnienia 0 the boys read the girls' memories about each other (= the girls)
This shows that the fact that reciprocals only allow SD must be explained otherwise. It also shows that simply marking anaphors as <+ LD-anaphor) or <- LDanaphor) fails to make the correct generalization. I will leave the matter here and return to it in section 6.5. I will now turn to a discussion of the reflexives in classes (8b) and (8d). In fact, I will limit myself to a discussion of reflexivization facts in Dutch (section 4) and Frisian (section 6), two languages which make a distinction between simple reflexives and complex reflexives. In line with the approach sketched above, I will show
84
Martin Everaert
that the difference between these classes of reflexives can be made to follow from their intrinsic properties.
4 SD-anaphors vs. LD-anaphors: the case of Dutch zichlzichzeJf
The Dutch anaphors zich and zichzelJ are possible candidates for the SD-anaphorl LD-anaphor distinction. The examples in (14) show that, in general, zich must have an antecedent but that this antecedent may not be the closest available antecedent, while zichzelj must take the closest available antecedent: 7 (14)
a.
b.
c.
d.
Jani liet mij voor *zichzelfi werken zichi 'Jan made me work for him' Jani zette de vaas voor ??zichzelfi neer zichi 'J an put the vase in front of him/himself' Jani verraste zichzelfi *zichi 'Jan surprised himself' Jani mopperde op zichzelfi *zichi 'Jan grumbled at himself'
This clear distributional opposition is somewhat blurred by the examples in (IS): (15)
a.
b.
Jani droogde zichi af zichzelfi 'Jan dried himself' Jani schaamde zichi *zichzel( 'Jan was ashamed'
On the basis of (14), Falk (1984), Koster (1985, 1987) and Yang (1983), among others, have proposed to account for the zichlzichzelj distinction in terms of an SDanaphor/LD-anaphor distinction. On an observational level, there are at least two arguments against such a distinction. First of all, proponents of such an approach have to disregard the examples in (IS) as exceptional only because zich behaves here as an SD-anaphor and not as an LD-anaphor. It is not explained why zich in inherently reflexive constructions is exceptional nor what exceptional precisely means here. As I have argued in Everaert (1986a), any account of the distribution of zichlzichzelj should treat the examples in (IS) as non-exceptional, at least as far
Contextual determination of the anaphor/pronominal distinction
85
as BT is concerned. Furthermore, such an approach suggests a fundamental distinction between the two anaphors, which is challenged by their morphological similarity. In Everaert (1986a) I have shown that the paradigm of the complex reflexive follows the paradigm of the simple reflexive. This strongly suggests that Dutch has only one reflexive anaphor zich, which allows addition of zelf (emphatic 'himself') under certain (structural) conditions (cf. Van der Leek (1980)). Zichzelf is represented as (16), where zelf is in a base-generated adjunction position: (16)
NP'
A
NP zich
zelf
A slightly different approach is taken by Vat (1980) and Bok-Bennema (1985), in part following unpublished work by Riny Huybregts (Huybregts (1979)). They take zich as a pronominal anaphor, i.e. zich has both pronominal and anaphoric features, while zichzelf is a real anaphor. This contention is not as far-fetched as one might think. If we, again, leave out examples (15a, b) as exceptional, zich seems to behave on a par with the uncontroversial pronouns hem (haar/hen, etc.). Compare (14) with (17):8 (17)
a. Jani liet mij voor hemi werken 'Jan made me work for him' b. ?Jani zette de vaas voor 'mi neer 'Jan put the vase in front of him' c. *Jani verraste hemi 'Jan surprised him' d. *Jani mopperde op hemi 'Jan grumbled at him'
If the distributional properties of the pronominal hem and zich coincide, one might wonder whether zich is not a true pronominal in other respects as well. Let us see what the usual tests - i.e., the possibility of split antecedents and the availability of bound-variable and/or pragmatic-coreference readings - for the anaphor/pronominal distinction indicate. It has frequently been observed that anaphors do not allow split antecedents, while no such restriction holds for pronominals. Compare (18) and (19): (18)
a. b.
Jani zag Karel j mij de slaven voor hemi/j Jani zag Karel j mij de slaven voor heni+ j
bij elkaar laten drijven bij elkaarlaten drijven
86
(19)
Martin Everaert
a. b.
Jan saw Karel me the slaves for him/them together make drive 'Jan saw that Karel made me drive together the slaves' neerleggen Jani zag Pietj de spullen naast zichi/j *Jani zag Pietj de spullen naast zichi + j neerleggen next to himself/themselves put Jan saw Piet the gear 'Jan saw Piet put the gear next to him/himself/them'
The ungrammaticality of the split-antecedent reading (19b) shows that zick behaves as an anaphor in this respect. It is a well-known fact that pronouns with a c-commanding definite antecedent allow both a bound-variable and a pragmatic-coreference interpretation (Reinhart (1983a, 1986)), cf. (20, 21), while anaphors allow a bound-variable interpretation only. In (22, 23) the readings with zick are given: (20)
(21)
Jani liet mij voor hemi werken en Karelj jou ook Jan made me work for him and Karel [made] you [work for him] too a. bound-variable reading [= Karelj liet jou voor hemj werken] b. pragmatic-coreference reading [= Karelj Het jou voor hemi werken] Jani zag de sneeuwbal op 'mi afkomen en Pietj ook Jan saw the snowball come to him and Piet [saw the snowball come to him] too
a. bound-variable reading [= Pietj zag de sneeuwbal op 'mj afkomen] b. pragmatic-coreference reading [= Pietj zag de sneeuwbal op 'mi afkomen] (22) Jani Het mij voor zichi werken en Pietj jou ook Jan made me work for him and Piet [made] you [work for him] too
(23)
a. bound-variable reading [= Pietj liet jou voor zich j werken] b. no pragmatic-coreference reading [:;6 Pietj liet jou voor zich i werken] Jani zag de sneeuwbal op zichi afkomen en Pietj ook Jan saw the snowball come to him and Piet [saw the snowball come to him] too a. bound-variable reading [= Pietj zag de sneeuwbal op zich j afkomen]
Contextual determination o/the anaphorlpronominal distinction
87
b. no pragmatic-coreference reading [* Pietj zag de sneeuwbal op zichi afkomen] On the basis of (18-23) we can conclude that, although the distributional properties of zich partly overlap with those of pronominals, zich really is an anaphor. In section 6 we will show how the complex distributional properties of reflexive zich might be explained without making use of an SD-anaphor/LD-anaphor distinction. We will amend the pronominal anaphor hypothesis in such a way that the observation that zich and zichzelJ are fundamentally identical, both being anaphors, will remain intact.
5 Binding theory reconsidered 5.1 The anaphor/pronominal distinction in BT In standard binding theory lexical elements are partitioned by means of two features, (±anaphor) and (±pronominal), resulting in the cross-classification for lexical elements presented in (24a). (24b) presents possible candidates from the English lexicon. (24)
a.
(+A, -P) (+A, +p)
b. himself, each other
(-A, -p)
John, the woman she, him
(-A, +p)
The question arises how the feature specifications of the lexical elements in (24b) are related to their behaviour with respect to the BT, formulated as in (25): (25) a. Anaphors have to be bound in a domain X. b. Pronominals have to be free in a domain X. c. R-expressions have to be free everywhere. Let us, as a preliminary conjecture, assume that the notions free and bound are directly related to the feature specification of a lexical element. We may, for instance, interpret ( + A) and ( - A) as 'need to be bound' and 'does not need to be bound' respectively. Likewise, ( + P) and ( - P) may be interpreted as 'needs to be free' and 'does not need to be free' respectively. For anaphors, marked (+A, -P), this will lead to the correct statement 'need to be bound and do not need to be free in a certain domain'. Likewise pronominals (-A, + P) 'need to be free and do not need to be bound in a certain domain'. Lexical pronominal anaphors, marked ( + A, + P), 'need to be bound and need to be free in the same domain', hence need
88
Martin Everaert
to be ungoverned and are thus not Case marked. Only the non-lexical variant PRO is possible. For R-expressions, however, the result would be 'do not need to be bound and do not need to be free', which implies that R-expressions may in fact be bound, which we know they may not. Suppose, therefore, that we interpret the features as in (26): (26)
( +A) --!> subject to binding condition (25a) (-A) --!> not subject to binding condition (25a) ( +P) --!> subject to binding condition (25b) ( - P) --!> not subject to binding condition (25b)
In such an approach the minus-features do not give any positive information. In the case of anaphors and pronominals this means that the minus-feature specificationcould be left out without any risk. Only for R-expressions, marked (-P, -A), and pronominal anaphors, marked (+ P, + A), the double specification is not redundant. With respect to R-expressions (26) has little to say. All we can deduce from the feature specification (-A, -P) is that R-expressions are subject neither to condition (25a) nor to condition (25b). We may, therefore, argue that the distribution of R-expressions falls outside the scope of BT proper (cf. Chomsky (1982)). With respect to the empty element PRO, the only possible manifestation of the feature combination (+ A, + P), it has been argued that its distribution (and its referential properties) can be accounted for without the assumption that it is a pronominal anaphor (Bouchard (1984), Kostet (1984), Sportiche (1983) and others). If such an approach were tenable, the need for a double feature combination in this case would become superfluous. But then we might consider the possibility of reducing the double feature specification (±A) and (±P) to a single feature specification, for instance (±A). Another possibility would be to redefine the features (±A) and (±P), and we will do so below in section 5.3.
5.2 The notion governing category The BT as formulated in (25) requires that an anaphor or a pronominal be bound or free within a specific domain, its governing category, defined as in (27): (27)
T is the governing category of IX iff T is the minimal maximal projection containing IX, the governor of IX and a subject accessible to IX
In the approach in Lectures on Government and Binding (Chomsky (1981)), both anaphors and pronominals have a unique governing category. We will illustrate this with an example from Dutch:
Contextual determination of the anaphorlpronominal distinction (28)
89
[T3 Evak denkt dat [T2 Jani [TJ henj voor zich*j/iI*k werken] laat]] zichzelfjl*iI*k Eva thinks that Jan them for themselves/himself/herselfwork makes 'Eva thinks that Jan makes them work for themselves/him/her'
In (28) 'T3 is never a potential governing category for an anaphor in the embedded prepositional object position. 'T2 is the governing category for the anaphor zich while 'TJ functions as such for zichzelJ This is not surprising if we assume that these anaphors are fundamentally different. On this assumption, each will have its unique governing category. However, if, as we have argued above, complex and non-complex reflexives are not fundamentally different, such a position would be unwarranted. If we want to keep the uniqueness requirement for governing categories, the anaphors in (28) must have the same governing category, 'T2 in this case, and it must somehow be stipulated that the complex reflexive takes the closest available subject as its antecedent. Another possibility is to drop the uniqueness requirement and to take both 'TJ and 'T2 as the governing categories for the anaphors in (28). We will explore the latter possibility. That is, an anaphor or pronominal in a position X has a set of governing categories ~, ~ = ('Tl>' •• ,'Tn).9 In the case of (28) this set should include 'TJ and 'T2 but not 'T3' Languages differ in their choice of potential governing categories. In a configuration [T2 ... [TJ ... anaphor/pronominal ... ] ... ], both 'TJ and 'T2 are potential governing categories only: - if 'TJ is a small clause complement for Dutch; - if'TJ is a small clause or infinitival complement for Norwegian; - if'TJ is a small clause, infinitival or subjunctive complement for Icelandic. We will not be concerned here with the question of how the choice for a governing category is made and why languages differ in their choices. We simply assume that it is possible to get the differences in the range of potential governing categories without making reference to the nature of the anaphoric or pronominal element, as I have shown in Everaert (1986a). We are now in a position to present our revision of the BT.
5.3 A revised binding theory Suppose we limit the class of elements subject to the BT to those lexical elements that do not denote by virtue of their inherent properties, i.e. only to anaphors and pronominals and not R-expressions. JO These elements, for which we will use the term pronoun, will be specified for <±A). Positively or negatively specified, this feature indicates that an element has a governing category - we could call it an A-
90
Martin E veraert
governing category - and therefore that it is a pronoun. In addition, pronouns are marked for the feature (±P). The feature (±P) indicates whether an element has a specific domain of interpretation, a P-governing category. In this way, each feature is directly linked to its own domain of interpretation. Furthermore, we will assume that the values of the (A)/(P)-features are directly linked to the notions bound and not bound: '+' ~ bound, '-' ~ not bound. It will be clear that the distinction between A-governing category and P-governing category is reminiscent of Chomsky's original Pisa approach to BT, where a distinction was made between the governing category and the minimal governing category of a lexical item. In our view, the pronominal feature will be associated with the notion minimal governing category of an element, the P-governing category in the terminology used above. The anaphoric feature is associated with the notion governing category, the A-governing category. From now on we will use the notion minimal governing category (MGC) and governing category (GC) as defined in (29): (29)
For 'I'b 'I'i member of k = ('I'I> ... ,'I'b ... ,Tn) (l:s; i:s; n), (i) 'I'i is a GC of ct iff'l'i is a maximal projection containing ct, and 'I'i is a 6domain; (ii) 'I'i is the MGC of ct iff'l'i a GC of ct and there is nOT;, '1'; is a GC of ct, such that 'I'i dominates '1';.
The notion 6-domain is reminiscent of Chomsky's (1986a) notion of complete functional complex and is defined as in (30):11 (30)
A maximal projection 'I' is a 6-domain iff (i) and (ii): (i) for every 6-defining head (J dominated by '1', 'I' dominates every argument 13 of (J; (ii) for every argument 13 that is dominated by '1', 'I' dominates the 6defining head (J of 13.
The above is summarized by the following binding principles: (31)
a. b. c. d.
A lexical element marked A lexical element marked A lexical element marked A lexical element marked
(+A) is bound in some GC. (-A) is free in some GC. (+ P) is bound in its MGC. (- P) is free in its MGC.
For convenience's sake I have formulated these binding principles in (32) in such a way as to resemble the standard BT: (32)
a. A lexical element marked (+A, -P) is bound in some GC and not bound in its MGC.
Contextual determination of the anaphorlpronominal distinction
91
b. A lexical element marked (+ A, + P) is bound in some GC and bound in itsMGC. c. A lexical element marked (-A, -PI is not bound in some GC and not bound in its MGC. d. A lexical element marked (-A, +p) is not bound in some GC and bound in its MGC. As was suggested above, the BT should be limited to 'pronouns', a collective term for anaphors and pronominals (to use the LGB-terminology). This revised BT distinguishes four types of pronouns and makes a primary distinction between anaphors, marked (+A), and pronominals, marked (-A). That is, we have a (+A, -P)-anaphor and a (+A, +P)-anaphor, and a (-A, +P)-pronominal and a (-A, -PI-pronominal. We are now in a position to formulate the opposition between SD-anaphors and LD-anaphors, notions that we have used purely descriptively so far, in terms offeature specifications. An SD-anaphor is a ( +A, + P)anaphor (cf. (3Zb)) and an LD-anaphor is a (+A, -P)-anaphor (cf. 3Za))Y In the next section we will illustrate the effects of the BT outlined above on the Dutch pronoun system.
5.4 Dutch anaphors We will assume that the anaphor zich and its complex counterpart zichzelf are both marked (+A). This characterizes the fact that both behave as true anaphors, as has been observed in section 4. The difference between these anaphors must therefore lie in their (±P) -specification. Let us, for the moment, simply stipulate that zich is marked for ( - P) and zichzelf for ( + P). The distribution of zich and zichzelf as exemplified in (14a-d) above will follow straightforwardly. These examples are repeated here as (33-36a), (33-36b) present their partial D-structures with relevant bracketing added: (33)
b. (34) a.
(35)
werken zichi *zichzelfj 'Jan made me work for him' [cp [IP! Janj [vp [IP2 mij [vp voor zich/zichzelfj werken]] liet]]] Janj zette de vaas voor zich j neer ??zichzelfj 'Jan put the vase in front of himself' [cP bp Janj [vp de vaas [pp voor zich/zichzelfjl neerzette]]] Janj verraste *zichj zichzelfj 'Jan surprised himself'
a. Jani liet mij voor
b. a.
92
(36)
Martin Everaert b. [cp hp Jani [vp zich/zichzelfi verraste]]] a. Jani mopperde op *zichi zichzelfi 'Jan grumbled at himself' b. [cp hp Jan; [vp [pp op zich/zichzelfi] mopperde]]]
According to (32a), zieh should not be bound in its MGC, and in (35) and (36) IP is the MGC of zieh. That is to say, IP is the minimal maximal projection which is a a-domain in (35,36). Within this MGC zieh is bound by the subject and the zieh readings of (35, 36) and therefore excluded. The anaphor ziehzelJ is subject to binding condition (32b), implying that it must be bound in one of its GCs and its MGC, which may very well be identical. In (35, 36), it is bound in IP, a GC and also its MGC, and thus predicted to be legitimately bound by the subject. Note that the VP in (35, 36) or the PP in (36) are not potential GCs for zieh and ziehzelJ since these maximal projections are not a-domains. The VP-node in (35b, 36b) dominates a a-defining head va (verraste, mopperde) but not all the arguments that are a-assigned by va, i.e. the subjects (Jan), so that these VP-nodes do not qualify as a-domains. The PP-node in (36b) dominates an argument (zieh, ziehzelj) but not the a-defining head, which we assume to be the va (mopperde), and is, therefore, not a a-domain. In (33) both the embedded IP and the matrix IP are potential GCs. Zieh is free in IPz, its MGC, and bound in IP] (note that one is free to choose both IP] and IPz as GC), as required. In (33) ziehzelJis free within the embedded IP and bound in the matrix IP and consequently excluded. The grammaticality of the zieh reading of (34) will follow if we assume that the locative preposition voor is a a-defining head and thus heads a a-domain following the definition in (30). In this view locative/directional PPs are a-domains although they are not clauses in the sense that a subject position is defined (cf. Hestvik (1989a)). The anaphor is free in its MGC, the PP, and of course bound in one of its GCs, the IP in (34b). For zichzelJ the reverse situation holds. Although it is bound in IP, it is not bound in its MGC. The distribution of the (-A, - P)-pronominal hem in (l7a-d), repeated in (37) with the relevant bracketing, follows without any problem. (37)
[IPI Jani Het hpz mij voor hem; werken]] 'Jan made me work for him' b. ?hp Jani zette de vaas [pp voor 'mil neer] 'Jan put the vase in front of him' c. *hp Jani verraste hemi] 'Jan surprised him' a.
Contextual determination of the anaphorlpronominal distinction
93
d. *[rp Jani mopperde op hemi] 'Jan grumbled at him' Like zich, the pronominal must be free in its MGC, IPz in (37a), the PP in (37b) and IP in (37c, d). In (37c, d) the pronominal hem is bound .in this MGC and therefore rightly excluded. In (37a) and (37b) hem is free in its MGC, but it must also be free in a given GC. In both cases there are two GCs, IP 1 and IPz in (37a) and IP and PP in (37b), and again one is free to choose. If one takes IP2 and PP to be the GCs, hem in these cases is free as required, hence the grammaticality of (37a, b). Up to this point the facts follow straightforwardly. However, we still have to account for the inherently reflexive cases in (38): (38)
a.
b.
af J ani droogde zichi zichzelfi 'Jan dried himself' Jan schaamde zichi *zichzelf 'Jan was ashamed'
In these cases zich does not behave as a <+ A, - P) -anaphor but, in fact, as a <+ A, +P)-anaphor, on a par with zichzelf In section 6.4 I will show that this can be brought about without stipulation. A discussion of the Frisian pronoun system will indicate how.
6 Unidentified pronouns In this section we will focus on the distributional behaviour of a class of pronouns from Frisian whose anaphoric properties seem to be indeterminate. Their phonological/morphological appearance gives no indication of their behaviour with respect to BT, just like the Dutch anaphor zich. 13 6.1 The Frisian pronoun system At first sight, the Frisian pronoun system seems to be on a par with the English pronoun system: it has a complex anaphor himsels, comparable with English himself, and a pronominal him. 14 HYi droege himselsi Of 'He dried himself' b. *HYi tinkt dat Marie himselsi Ofdroeget 'He thinks Marie dries himself'
(39) a.
94
Martin Everaert
c. *Jansi mem droeget himselsi Of 'Jan's mother dries himself' (40) a. HYi droege himj Of 'He dried him' b. HYi rette himj ta foar syn opkommen 'He prepared him for his performance'
In some cases, however, him can or must be used 'as an anaphor', i.e. this pronoun does not seem to be intrinsically marked as either anaphor or pronominal. (41) a. HYi skammet himil'j 'He is ashamed' b. HYi stoarret foar himil'j lit 'He is staring' (42) a. HYi droege himilj Of 'He dried himself/him' b. HYi rette himilj ta foar syn opkommen 'He prepared himself/him for his performance'
In (42), but not in (41), the pronominal can be replaced by the anaphor himsels: (43)
a. *HYi skammet himselsi 'He is ashamed' b. *HYi stoarret foar himselsi lit 'He is staring' (44) a. HYi droege himsels i Of 'He dried himself' b. HYi rette himselsi ta foar syn opkommen 'He prepared himself for his performance'
In examples (45, 46) him cannot be used anaphorically; in these cases only the anaphor himsels can be used. For (47), the opposite is true: (45)
a.
b.
c.
beneamd te wurden Op dy wize behindere eri him'ilj himselsi in this way prevents he him/himself appointed to be 'In this way, he prevents him/himself from being appointed' HYi joech him'ilj it boek himselsi 'He gave him/himself the book' HYi beoardielet him'ilj himselsi
Contextual determination of the anaphorlpronominal distinction
(46)
a.
b.
c.
(47)
a.
b.
95
'He judges him/himself' HYi soarget foar him*ilj himselsi 'He takes care of him/himself' HYi tinkt allinnich oan him'ilj himselsi 'He thinks only about him/himself' HYi prate oer him*ilj himsels i 'He talked about him/himself' Pyti lei it boek nest himilj del ?*himselsi 'Piet put the book next to him' HYi sette de faas foar himilj *himselsi 'He put the vase in front of him'
Again we have a case of a lexical item, i.e. him, which behaves both as an anaphor and as a pronominal. But here it does not do so with respect to the domain in which it can find its antecedent, as is the case with zich, but for a more fundamental reason: its anaphoric properties are indeterminate. In (41) him behaves as an anaphor exclusively, while it is used both as an anaphor and a pronominal in (42), and in (45-7) as a pronominal exclusively. This distribution excludes the obvious view that Frisian has in fact two homophonous pronouns, a pronominal and an anaphoric him (cf. Fanselow (1988)). It leaves open the possibility that him is either a pronominal or an anaphor, in which case the domains of interpretation in Frisian must be accommodated accordingly. We would like to determine whether him is a real anaphor in cases like (41, 42). For the inherent reflexive use of him in (41) the two anaphor tests used above (cf. 18-23) give a straightforward result. Him behaves as an anaphor in these cases: (48)
(49)
Jani skammet himi en Piterj ek 'Jan is ashamed and Piter [is ashamed] too' a. bound-variable reading [ = Piterj skammet himj] b. no pragmatic-coreference reading [=1= Piterj skammet himi] *J ani seach Piterj foar harreni + j lit stoarjen 'Jan saw Piter stare'
96
Martin Everaert
In (42) these tests fail because, in these cases, it is impossible to discriminate between the anaphoric and the pronominal uses of him. Still, there seems to be some evidence indicating that him does not behave as a pronominal in these examples. The argument involves the possibility of avoiding disjoint reference effects with pronominals. Reinhart (1983a) argues that binding condition B effects are limited to the bound-variable interpretation of pronominals. She argues that the ungrammaticality of (50a) as opposed to the grammaticality of (50b) must be explained in terms of a pragmatic strategy (only partly given in (51), Reinhart (1983a: 167»: (50) a. (*)Billi adores himi b. Billi adores himselfi (51) If the speaker avoids the bound-anaphora options provided by the structure he is using, then, unless he has reasons to avoid bound anaphora, he did not intend his expressions to corefer (hearer's strategy). Following (51) a hearer will interpret him in (50a) as not referring to Bill. However, given appropriate pragmatics, Bill adores him can be interpreted as 'Bill adores himself', that is, him can be made to corefer with Bill accidentally: (52) I know what Bill and Mary have in common. Mary adores Bill and Bill adores him too However, if Bill in (50a) is replaced by a quantified NP, him is necessarily interpreted as a bound variable and (51) cannot be operative. Only a reflexive can be used to indicate coreference (cf. Montalbetti & Wexler (1985), Reinhart (1983a, 1986), Sportiche (1986». (53)
a. *No onei adores himi b. No onei adores himselfi
Let us now see what happens in the case of Frisian him (cf. 41,42): (54) a. Iderieni skammet himi 'Everybody is ashamed' b. Iderieni droege himi Of 'Everybody dries himself' The grammaticality of the examples in (54) indicates that, in these cases, him is not a pronominal but an anaphor. On closer observation it will become immediately clear that the Frisian reflexivization facts in (39-46) show a striking parallel with the Dutch facts from section 4. Frisian himsels behaves on a par with Dutch zichzelf, while anaphorically used him
Contextual determination of the anaphorlpronominal distinction
97
in Frisian behaves on a par with Dutch zich, calling for an account in which their distribution follows from the same explanatory principles. In the next section we will discuss what these principles may be and we address the problem of the dual status of Frisian pronouns. 6.2 Contextual binding principles We want to argue that, apart from the regular binding conditions which determine the specific linking possibilities of anaphors and pronominals, BT also incorporates certain contextual definitions. These contextual binding principles, as I will call them, determine what the anaphoric status of some pronouns is, such as that of Frisian him discussed above. For this class of unidentified pronouns (UPOs), the contextual binding principles determine how these pronouns must be marked, and consequently how they behave with respect to the binding theory. By 'unidentified' we mean that these pronouns are not intrinsically marked for a specific value of (±A) or (±P). In our view, the contextual binding principles determine what the specific values of these features must be. Let us formulate the relevant principles as in (55): (55)
Contextual binding principles (CBPs) a. Assign to a pronoun D, D unmarked for (P), (i) ( - P) iff 8-marked, (ii) ( + P) iff 8-linked; b. Assign to a pronoun D, D unmarked for (A), (i) ( - A) iff 8-marked, (ii) ( + A) iff 8-linked; unless the resultant marking would lead to a feature specification of D, such that D cannot fall under the BT.
If a pronoun which is unmarked for either the feature (P) or the feature (A) is 8marked, it is interpreted as pronominal in nature, i.e. not necessarily referentially dependent upon a linguistic antecedent. Such a pronoun is marked negatively for the pronominal feature and for the anaphoric feature and thus free in a certain domain. The reverse feature specification is assigned if these pronouns are 8linked, i.e. if these pronouns are quasi-arguments (we will define the notion 8linking below). In these cases the pronoun is interpreted as anaphoric in nature, and therefore referentially dependent upon a linguistic antecedent. By contradictory marking we mean that no lexical pronoun may be marked in such a way that the result would be filtered out by the binding theory. As it stands, this excludes the possibility that a pronoun is marked ( - A, +p) . More has to be said about the notion 8-markingl8-linking as it is used in the
98
Martin Everaert
CBPs. Limitations of space allow only a general outline of the theta theory I have in mind. It has been observed that in some cases specific O-roles of a governor are not necessarily projected onto syntactic positions, although these thematic roles are still part of the O-grid of that governor. Consider, for example, (56) from Rizzi (1986a): (56)
a. This sign cautions people against avalanches b. This sign cautions against avalanches
(56b) means that some arbitrary person is cautioned against avalanches by the sign although in contrast to (56a) there is no syntactic position available to signal this. Rizzi (1986a) argues that in (56b) the lexically governed rule (57) has applied: (57) Assign arb to the direct O-role. For English, (57) applies in the lexicon. That means that in cases like (56b) the direct-object O-role of the verb is lexically saturated and no syntactic position is realized. In Italian, (57) applies in the syntax and, thus, a syntactic position is projected, filled by the empty category pro. The rule might also apply in syntax for English if only English had a licit filler for the object position; it cannot be pro because English, unlike Italian, does not license this empty category. However, as Rizzi observes, in some cases an internal O-role is not lexically saturated and still no syntactic position is available. In (58b) not John but an unspecified person is ordered to go away: (58)
a. John shouted the order to Bill to go away b. John shouted the order to go away
The (secondary) object of shout is syntactically visible for the control relation in both (58a) and (58b) even though no syntactic position is projected in the latter case. IS I will suggest here a specific formalization of O-assignment which will allow me to extend an analysis along the lines of Rizzi (1986a) to inherently reflexive verbs in section 6.3. Let us assume that O-marking means that the referential index of a phrase governed by a head X is projected onto a position in the O-grid of X (cf. Stowell (1981) and Williams (1987». (59)
V NPj -0>
(0)
V NPj
(OJ)
On this view, the O-criterion requires that all O-roles in a O-grid must be assigned an index. This also means that reference is attached directly to the O-role as an
Contextual determination of the anaphorlpronominal distinction
99
index (cf. Williams (1985)), and, therefore, it comes as no surprise that a a-role can also be independently indexed, for instance by a rule like (57). This raises the question whether or not a syntactic position can be projected in these cases. Rizzi (1986a) argues that a syntactic position cannot be projected if its corresponding arole is lexically saturated, i.e. syntactically invisible. The reverse, though, does not hold: syntactically visible a-roles are not necessarily projected (cf. 58b). Now suppose, contrary to Rizzi (1986a), that a syntactic position can be projected if the verb is specified to do so, even though the a-role is 'inherently' indexed. However, because the a-role is already lexically indexed, the lexical elements occupying the projected syntactic positions may not be referential. These lexical elements would then derive their (quasi-)referential status from the inherently indexed a-role. Possible candidates are quasi-arguments like idiomatic phrases (cf. Coopmans & Everaert 1988), expletives or, for that matter, (clitic) reflexive pronouns. I will use the term 'a-linking' for these cases of a-assignment. Now that we have clarified the general outline of our theory of a-government, we are in a position to define the notions a-marking and a-linking as they are used in the CBPs. I will define a-government as in (60) and a-markingla-linking as in (61) (cf. Coopmans & Everaert (1988)): (60)
(61)
a a-governs 13 iff a is a zero-level category that (i) a-marks 13 or (ii) a-links 13. a. 13 is a-marked by a iff a position in the a-grid of a is assigned an index by 13. b. 13 is a-linked to a iff a, 13 are sisters and a position in the a-grid of a is coindexed with 13, and a and 13 are dominated by the same lexical projections.
6.3 (Inherently) reflexive verbs With the theory of a-government in mind, we are able to say something more about the class of inherently reflexive verbs. In our view, inherently reflexive verbs are lexically specified as having an internal thematic role coindexed with the subject thematic role, while the verb is, at the same time, specified to project a syntactic position. As a result, the object position is necessarily occupied by an anaphor, zich in Dutch, the only phrase that can be a-linked to the lexically saturated a-role in these cases: (62) a. schamen: (ai> ai) b. Jani schaamt zichi 'Jan is ashamed'
100
Martin Everaert
A similar process might be at work in the English examples in (63) but in these cases no syntactic argument is projected (cf. Zubizarreta (1987)). (63) a. wash, shave, etc.:
a.
b.
(65)
a.
b.
J ani wast zichi Karel 'Jan washes himself/Karel' Hij; bezeert zichi Karel 'He hurts himself/Karel' Zij; verrast *zichi Karel 'She surprises herself/Karel' Jani geeft *zichi het boek Karel 'Jan gives himself/Karel the book'
6.4 Frisian him and Dutch zich as unidentified pronouns Let us turn to the pronominal system of Frisian as outlined in section 6.1. Frisian pronouns like him are clearly unidentified in the sense discussed above. Suppose they are marked for neither <±A) nor <±P). As such these pronouns will be subject to the CBPs. Observe, again, the examples in (45-7), partly repeated in (66-8) (below,
Contextual determination of the anaphorlpronominal distinction
101
a-marking is indicated by an arrow from NP to VIP, a-linking by an arrow from VtoNP): (66)
a.
b.
HYi beoardielet *himi himselsi 'He judges him/himself' [vp himi V ]
~)
(67)
a.
b.
HYi prate oer *himi himselsi 'He talked about him/himself' [vp [pp prep himi] V ] lji)
(68)
a.
b.
HYi sette de faas foar himi *himselsi 'He put the vase in front of him' [vp NP [pp prep himi] V]
In (66) the pronoun him is a-marked by VO and thus marked <-P) and <-A) by (55). In the case of prepositional objects as in (67), the a-role of the object NP is part of the a-grid of the verb which acts as the a-marker. Again him is assigned <- P) and <- A). This means that him behaves as a true pronominal and that it must be free in its MGC and a GC. Hence the ungrammaticality of (66,67). In (68) the pronoun is a-marked by the a-defining head po and, again, marked <- P, -A). The PP-node is a GC and a MGC for him. It is free in its (M)GC, as required, and legitimately bound by the subject hy. The distribution of the complex reflexive himself in Frisian would follow straightforwardly if one simply stipulated that it is inherently marked as a <+A, +P) anaphor. However, just as in the case ofthe Dutch zichlzichzelf pair, we will do justice to the morphological similarity between himsels and him. That is, we will take the structure of himsels to be as in (69b), similar to the structure of zichzelfin (16), here repeated as (69a): (69) a. b.
[NP' [NP [NP' [NP
zich] zel£] him] sels]
Within this structure, him would be unmarked but the question is whether sels is.
102
Martin Everaert
Observe that Dutch zelf, and the same holds for Frisian sels, can be used as an emphatic adverbial (pronominal) adjunct in sentences like (70): (70) J ani gaf hem het boek zelfi 'Jan gave him the book himself' In these cases zelJ has an interesting property. It always 'refers' to the closest subject, i.e. it never violates the specified subject constraint. This means that zelf (sels) in (70) may be characterized as a (+ P) -pronounY As I have argued in Everaert (1986a), there is no reason to distinguish the use of zelf in (70) from the use of zelfin (69a). Consequently, zelf/sels in zichzelf/himsels will bear a (+P)feature. I will assume that the (+P)-feature of sels in (69b) is transmitted to him under local coindexing. From this perspective him in himsels is marked for (+ P), contrary to bare him. We can now return to the discussion of the distribution of himsels in (66-8). Himsels is subject to the CBPs but only with respect to its (A)-specification. Since it is a-marked in (66-8), himsels should be marked (-A) by the CBPs. However, this would result in a (-A, +P) marking requiring that himsels be free in a GC, being specified (- A), and bound in its MGC, being specified ( + P). But that will lead to a contradiction since a lexical element cannot at the same time be bound in its MGC and free in a GC where the latter is either equal to the MGC or properly contains the MGC (but see also note 20). But, as stated, the application of the CBPs may not result in 'contradictory' marking and, therefore, the complex NP himsels is not marked for (A) but for (+P) only and should behave as an SDanaphor. 18 In (66, 67) himsels is, in fact, locally bound, but in (68) it is bound outside its MGC and therefore excluded. We are left with the distribution of him in the inherently reflexive constructions (41,42) above. I take it that, in these cases, the internal thematic role is lexically saturated. Consequently, only the lexically designated lexical item him may occupy the object position. In (71) him is a-linked and thus marked (+ A, + P) by the CBP, making it an SD-anaphor: (71)
a.
b.
HYi skammet himi *himselsi 'He is ashamed' [vp himi V
L;i)
I will assume that the internal thematic role of ofdroegje is optionally saturated, just like the case of Dutch wassen discussed above. In the case of a saturated a-role,
Contextual determination of the anaphorlpronominal distinction
103
only him can occupy the argument position, just as in the case of skamje (cf. 72); in the himsels variant the object is a-marked by the verb (cf. 73): (72) a. HYi droege himi Of 'He dried himself' b. [vp himi V ] Ui) (73) a. HYi droege himselsi Of 'He dried himself' b. [vp himselsi V ] Lji) The CBPs as presented in (55) thus give a straightforward explanation for the distribution of the Frisian unidentified pronoun him. a-government properties of governors and lexical properties of the complex anaphor determine whether him is interpreted as a pronominal or as an anaphor. We will now show that the same principles can account for the distribution of the Dutch anaphor zich. We have already observed that the anaphoric properties of zich are, in a sense, indeterminate. Suppose that this is due to the fact that zich, although positively marked for the anaphoric feature, i.e. (+ A) , is unmarked for the pronominal feature. Because of this, zich will be subject to part (a) of the CBPs. Let us, again, look at the relevant examples: (74)
a.
b.
c.
d.
(75)
a.
Jani moppert op *zichi zichzelfi 'Jan grumbles at himself' werken J ani liet mij voor zichi *zichzelfi 'Jan made me work for him/himself' Jani zette de vaas voor zich i neer ??zichzelfi 'Jan put the vase in front of him/himself' Jani verrast *zichi zichzelfi 'Jan surprises himself' Jani droogde zichi af zichzelfi 'Jan dried himself'
104
Martin Everaert b.
Jani schaamde zichi *zichzelfi 'Jan was ashamed'
The relevant a-government relations are presented in (76): (76)
a.
VP
b.
VP
A
A
NPi V zich(zelf) (ai)
I
t
PP
/\
c. V ( ai )
N~
prep
VP
/\
NPi V zich (ai)
LJ
zich(zelf) In (74) the pronoun zich is a-marked (cf. 76a, 76b) and thus marked (-P) by the CBPs. In (75) zich is a-linked (cf. 76c) and thus marked (+P) by the CBPs. In the case of zichzelf, the CBPs will not apply because it is already fully specified, ( + A) because of the zich part and (+P) because of the zelf part. Therefore, being marked (+A, -P) and (+A, +P) respectively, the distribution of zich and zichzelffollows, as has been described in section 4. 6.5 Reciprocals Let us now turn to the Dutch reciprocal elkaar, whose distributional properties coincide with that of reflexive zich in zichzelf, in accordance with the observation from section 2 that it is an SD-anaphor: (77) *Ziji lieten mij voor elkaari werken
'They had me work for each other' By taking elkaar to be a ( + A, + P) -anaphor, the ungrammaticality of (77) would straightforwardly be predicted, but is it possible to derive this result without stipulating it? Suppose one followed the approach taken for Dutch zichzelf elkaar is intrinsically marked for (+A) only. It would then be subject to the contextual binding principles for the (P) specification, just like the reflexive anaphor zich. In (77) elkaar is a-marked - just as zich would be in that position - and consequently marked (-P) by the CBPs, giving the wrong result that elkaar would be an LDanaphor, just like zich. An alternative approach in which the SD-properties of elkaar are derived from the semantics of reciprocity will be suggested below. Consider the examples in (78), discussed in Higginbotham (1982a): (78)
a. TheYi saw themselves shoot themselvesi b. ?*TheYi saw each otheri shoot themselvesi
Contextual detennination of the anaphorlpronominal distinction
105
Higginbotham (19S2a) attributes the unacceptability of (7Sb) to the phenomenon of antecedent clash, where antecedent clash is formulated as follows: in a maximal c-command chain, the proximate and evaluative antecedent of an NP may not be instantiated by different variables. I will not be concerned here with a detailed reconstruction of Higginbotham's argumentation, but the basic line of reasoning is important for the present discussion. In Higginbotham's view, the following c-command chains are formed from (7S): (79)
a. [They;, themselvesi, themselvesi] b. [They;, each other;, themselvesi]
In these c-command chains they is taken as the evaluative antecedent of the rightmost reflexives in (79); the reciprocal in (79b) and the leftmost reflexive in (79a) are taken as their proximate antecedents. According to Higginbotham, (SOb) is an approximation of the LF-representation of the simple reciprocal sentence (SOa): (SO) a. [a & b] pred. [reciprocal] b. [a] pred. [b] & [b] pred. [a] For the cases at hand, this will give the following LF-representations: (81)
a. b.
[[a & b] saw [a & b] shoot [a & b]] = (7Sa) [[a saw b shoot b] & [b saw a shoot a]] = (7Sb)
It is clear that in (SIb), but not in (81a), the evaluative antecedent and the proximate antecedent of the rightmost reflexive are differently instantiated, resulting in an antecedent clash. Crucial in the argumentation is that in the reciprocal-antecedent relation, as represented in (SO), the aritecedent and the reciprocal are not instantiated by the same variable at LF. Although the reciprocal is referentially dependent on its antecedent, it is not a simple case of 'coreference'. Suppose that this lack of coreference in the case of reciprocals is reflected in their (A)-specification, the feature that indicates whether or not a pronoun is anaphoric. A positive specification of the anaphor feature would imply the possibility of coreference. A reciprocal cannot therefore not be marked (+A) and, as we will assume, is automatically negatively marked for this feature, i.e. ( - A) .19 On the other hand, it is clear from the semantics of reciprocity that the reciprocal is dependent on a linguistic antecedent for its proper interpretation. This can only be the case if the reciprocal is marked positively for its (P)-feature. That is to say, reciprocals should be marked (-A, + P). But we have explained earlier in connection with himselslzichzelfthat such a feature specification is not allowed for governed pronouns. Now let us assume that, in the case of reciprocals, the features are not 'visible' simultaneously.
106
Martin Everaert
Suppose the <- A) -feature is ignored in syntax and only operative at LF, at which level the <+P)-feature is ignored (cf. Finer (1984) for a similar conclusion). On the assumption that BT (also) holds at S-structure, reciprocals would just be marked <+P) as far as BT is concerned and, therefore, behave as SD-anaphors. The 'pronominal character' of reciprocals would be visible at LF only (cf. 80b).
7 Some problems and consequences In the first two parts of this last section I will discuss two outstanding observational problems related to the pronoun distribution in Dutch discussed in section 6. The first one (discussed in 7.1) is as yet unsolvable in any BT-approach I know of. The second one (discussed in 7.2) raises more specific problems for the BT as developed above. In both cases we will offer some tentative solutions. In 7.3 and 7.4 I will briefly discuss some consequences of our approach to BT for the partitioning of empty elements and the acquisition of anaphors.
7.1 AcI-passives Consider the examples in (82): (82)
a.
b.
Jani laat het boek aan zichi overhandigen (door miD zichzelfi *hemi 'Jan lets the book be handed to himself/him (by me)' Jani liet een huis voor zichi bouwen (door miD zichzelfi *hemi 'Jan let a house be built for himself/him (by me)'
(82) is an example of a so-called Ad-passive. The complement of the causative laten has some of the properties of a passive; the subject is not syntactically projected and, optionally, a by-phrase can be added. In these Ad-passives, zich and zichzelJ are not in complementary distribution, unlike in 'normal' small clause complements (cf. (83)) or simple sentences (cf. (84)): (83)
a.
Jani laat mij het boek aan zichi overhandigen *zichzelfi hemi 'Jan has me hand the book to him/himself'
Contextual detennination of the anaphorlpronominal distinction b.
(84)
a.
b.
107
Jani liet mij een huis voor zichi bouwen *zichzelfi hemi 'Jan had me build a house for him/himself' Jani overhandigt het boek aan *zichi zichzelfi *hemi 'Jan hands the book to him/himself' Jani bouwt een huis voor *zichi zichzelfi *hemi 'Jan builds a house for him/himself'
In order to be able to say something more about the distribution of pronouns in these cases, we will have to take a closer look at the structural analysis of Adpassives. In the Dutch syntactic literature several suggestions have been made as to how these constructions should be dealt with. Hoekstra & Moortgat (1979) argue that causative laten in (82) takes a VP-complement, while Everaert (1986a) defends a (subjectless) IP-complement hypothesis. Coopmans (1985) takes a more radical position; he argues for a straightforward monoclausal analysis in which laten is interpreted as an affix forming a complex verb with the complement verb. For the facts under discussion, the VP- and the IP-hypothesis amount to the same, and I will compare the VP-hypothesis (cf. (85a)) with the complex predicate hypothesis (85b): (85)
a. b.
prep X] V] laten]] prep X] [V + laten]]]
[IP •.. [VP2 [VPl .•• [pp [IP •.• [VPl ••• [pp
With respect to the binding theory both hypotheses make different, but wrong, predictions. In the complex predicate analysis, where laten is interpreted as an affix forming a complex verb together with the embedded verb, (82) and (84) are structurally identical, i.e. monoclausal (cf. (85b)). In these cases, the IP is the only potential GC and automatically the MGC for a pronoun in position X in (85b). Contrary to the facts, both zich and hem are predicted to be ungrammatical in the indicated coindexing in (82), just as in (84). In the VP-hypothesis (85a), IP is a GC but so is VP 1. One might argue that the VP1-node is a a-domain because, the subject a-role being saturated, all the unsaturated a-roles of V are assigned within VP 1 in (85b). If both IP and VP 1 are GCs, VP 1 is the MGC. As a result, the VPhypothesis makes the opposite prediction from the complex predicate hypothesis; the grammaticality judgements in (82) should be reversed with respect to (84) and
108
Martin Everaert
identical to those in (83). Again, they are not. There are perhaps ways to account for the facts in a VP-analysis but we will not develop these here. 20 Instead we will further develop the complex predicate analysis. In a complex predicate analysis, a completely different approach for the examples in (82) has to be pursued. A discussion of the examples in (86) will make clear how. (86)
a.
b.
Hih laat op zichi wachten *zichzelfi he lets for himself wait 'He is giving us a long wait' Hih laat niets van zichi horen ?*zichzelfi he lets nothing from himself hear 'He never calls'
The sentences in (86) are examples of a limited group of prepositional object verbs which have an inherently reflexive variant when combined with laten. 21 As discussed in Everaert (1986a), these examples are cases of true inherently reflexive verbs in that the complex reflexive, or any other NP for that matter, is excluded. Without laten these verbs are non-inherently reflexive with the subsequent reflexive distributionY (87)
a.
b.
Hiji wacht op *zichi zichzelfi 'He waits for himself' Hiji hoort niets van *zichi zichzelfi 'He heard nothing from himself'
On a complex predicate account this could be explained if we assume that the internal thematic roles of the non-causative verbs in (86) are lexically saturated as a result of the affixation of laten. In that case only zieh can occupy the prepositional object position, as in the case of inherently reflexive verbs (cf. (62». Now observe that the pronoun distribution in the Ad-passives in (82), i.e. ziehl ziehzelf/*hem, is reminiscent of the pronoun distribution with verbs like wassen, discussed in section 6.4 and repeated here in (88): (88)
Jani wast zichi zichzelfi *hemi
Contextual determination of the anaphor/pronominal distinction
109
'Jan washes himself/him' We have argued that zich in Jan wast zich is linked to the optionally saturated internal argument of the verb. Suppose a similar situation arises in Ad-passives. One ofthe internal thematic roles of the complement verbs is (optionally) saturated allowing a zich-variant: (89)
a.
b.
dat Jani het boek aan zichi overhandigen laat
t
(6i)
dat Jani het boek aan zichzelfi overhandigen laat
( 6i)
t Just like in the wassen-case, the two variants (89a) and (89b) are identical to the extent that their thematic structures are identical. 23 For the examples in (90) a complex predicate analysis has been firmly established (Coopmans (1985), Coopmans & Everaert (1988)): (90)
a.
b.
c.
Jan liet het boek aan Karel zien Jan let the book to Karel see 'Jan showed Karel the book' Jan laat zich aan Karel niets gelegen liggen Jan lets himself for Karel nothing care 'Jan doesn't care for Karel' Jan liet het oogop Karel vallen Jan let the eye on Karel drop 'Jan showed interest in Karel'
Replacing Karel in (90) by one of the pronouns under discussion results in the following: (91)
a.
b.
c.
Jani laat het boek aan *zichi zien zichzelfi *hemi Jani laat zich aan *zichi niets gelegen liggen ?zichzelfi *hemi vallen Jani laat het oog op *zichi ?zichzelfi *hemi
110
Martin Everaert
Although judgements are delicate in (91b) and (91c), the conclusion seems to be clear; the pronoun distribution is as in simple sentences. This means that, somehow, the saturation of the internal thematic role in a complex predicate is dependent on the specific affixal status of laten. Only if laten blocks the external a-role of the verb it is attached to, as in Ad-passives, is saturation possible.
7.2 Disjoint reference Consider the examples in (92): (92)
a. Jani hoorde zichzelfi over Piet praten (op het bandje) b. Jani hoorde zichzelfi over zichzelfi praten (op het bandje) 'Jan heard himself talk about Pietlhimself (on the tape)'
Given appropriate pragmatics, the anaphor zichzelJ can be used as the subject of the clause headed by praten 'to talk', cf. (92a). (92b) shows that such an anaphoric subject can itself be a potential antecedent for another anaphor (note that the matrix subject Jan is not a potential antecedent for the rightmost zichzelJ, see section 6). Replacing the second instance of zichzelJ in (92b) by the pronoun hem leads to ungrammaticality (cf. (93a)). This is due to a violation of principle B, on a par with (93b): (93)
a. *Jani hoorde zichzelfi over hemi praten (op het bandje) 'Jan heard himself talk about him (on the tape)' b. *Ik liet Jani voor hemi werken 'I made Jan work for him'
If we replace hem in (93a) by zich, grammaticality judgements change. Although the judgements are subtle, I have the impression that (94) is considerably better than (93a): (94) ?Jani hoorde zichzelfi over zich i praten (op het bandje) 'Jan heard himself talk about himself (on the tape)'
If zich is a (+A, +P)-anaphor, one expects (94) to be ungrammatical, just like (93a), because it is bound in its MGC by zichzelf The difference in grammaticality between (93b) and (94) might be explained in terms of linking theory, as developed in Higginbotham (1985). In this theory (95) and (96) are the only potentiallinkings available for (93a) and (94), respectively:
Contextual determination oj the anaphorlpronominal distinction I
I
(95)
Jani hoorde [zichzelfi over hemi praten] I (+A, +P) (-A" -P)
(96)
Jani hoorde [zichzelfi over zich i praten] (+A, +P) (+A( -P) \
I
111
I
These structures are similar to (97) (from Higginbotham 1985) with respect to their linking properties. I I (97) J ohni said [hei saw himi] I
I
For Higginbotham, (97) is excluded by his binding condition B (98a) and the obviativity condition (98b): (98)
a. A pronominal is locally obviative. b. If X and Yare obviative, then they cannot be determined by the structure in which they occur to share a value.
(98) also excludes (95) and (96); both zich and hem must be obviative from zichzelJ because of their ( - P) -specification and, given the linking indicated, they share a value, violating (98b). Note, however, that in our BT hem is obviative from zichzelJ in (95) because of both the ( - A) -specification and the ( - P) -specification, while zich is obviative from zichzelJin (96) only because ofthe (-P)-specification. The difference between (95) and (96) might then follow if we reformulate (98b) as in (99): (99)
If X and Yare obviative for a feature ex, then they cannot be determined by the structure in which they occur to share a value for ex.
In (96) zich and zichzelJ must share a value for their (A) -specification, while at the same time they are obviative because of their (P)-specification and, therefore, they do not violate (99). The linking in (95) always violates (99) because, although zich and hem can share a value for their(A)-specification or (P)-specification (taking the embedded IP as (M)GC), they are obviative because of both their (A)-specification and their (P)-specification. 7.3 The classification of empty elements Contextual definitions were originally introduced to account for the distribution of empty categories (cf. Chomsky (1981, 1982». We have proposed to use a similar
112
Martin Everaert
approach for a special class of lexical categories, unidentified pronouns (UPOs). This immediately raises the question whether it is possible to generalize over empty categories and lexical categories in this case. As they stand, the contextual binding principles (CBPs) will give the wrong result for empty categories. But, of course, this follows as a matter of principle. It is very well conceivable that a specific formulation of the CBPs allows one to account for both the distribution of UPOs and empty categories. However, this would lead us beyond the scope of this chapter. In the following I will only show that the specific partitioning of lexical pronouns that has been advocated here may very well be reflected in the partitioning of empty categories, and that an approach in which the CBPs apply not only to UPOs but also to empty categories might in principle be possible. It is well known that NP-traces are always close bound. There are no instances of long-distance bound NP-traces (cf. Everaert 1986a). That means that NP-traces seem to be the non-lexical equivalents of the lexical (+ A, + P) -anaphor zichzelf The empty category pro behaves as a real pronominal and is, therefore, a likely candidate for a non-lexical (-A, -P)-pronoun. This leaves us with PRO, whtrace being excluded because it is the equivalent of an R-expression, and, therefore, falls outside the scope of BT as I have formulated it here. A controlled PRO is always free in the first dominating a-domain while at the same time bound in the next higher domain, essentially the properties of anaphoric zich in non-inherently reflexive constructions. Suppose we follow Bouchard (1984), Koster (1984), Sportiche (1983) and others in the assumption that controlled PRO is governed and, hence, has a GC. One could then argue that PRO is a ( + A, - P) -pronominal. It is free in its MGC and bound in a GC, just like zich. (100)
[IP
NPi
... [IP
PROi
t
t
GC
MGC
... ]]
Let us now look at the fourth potential feature combination, i.e. (-A, + P). As we have already discussed in section 6.4, such an element must be free in a GC, being specified ( - A), and bound in its MGC, being specified ( + P). But that will lead to a contradiction as an element cannot at the same time be bound in its MGC and free in a GC where the latter is either equal to the MGC or properly contains the MGC, unless such an element is ungoverned. It has been argued that PRO arb is such an element. In sum this will give us the following partitioning of empty categories: (101)
(+A, +P) NP-trace Izichzelf
Contextual detennination of the anaphorlpronominal distinction (+A, -P) PRO (-A, +P) PRO arb (-A, -P) pro
113
Izich 1Ihemlhaar etc.
7.4 Acquisition of anaphors It is generally assumed that, as far as BT is concerned, children only have to learn which lexical elements are anaphors and which are pronominals. So, for instance, once a child has learned that each other is an anaphor the distribution of each other should follow from the innate principles of BT. A somewhat different perspective on this issue emerges from the revised BT of section 6. Suppose that children are able to decide whether a lexical item is referential or not, i.e. whether it is a D-element or not (cf. Wexler & Manzini (1987)). Then all that is left to determine is what specific kind of non-D-element a lexical element is - in our approach a choice between four potential pronominal elements. It seems to me that the minimal assumption would be that pronouns are always 'unidentified' in the initial stage of language acquisition. That means that some NPs are identified by the child as non-D-elements but that their specific feature specification must be learned (cf. Lust, Mazuka, Martohardjono & Yoon (1989) for similar ideas). Some well-known facts from studies on the acquisition of anaphors indicate that this might be the case. It appears that there is a stage in language acquisition in which children treat pronouns as if they were anaphors (cf. Deutsch, Koster & Koster (1986) and Corver (1985) for Dutch, and Jakubowicz (1984) and Wexler & Chien (1986) for English). So a sentence like John washes him is often interpreted by very young children as meaning John washes himself. 24 According to Jakubowicz (1984), this might be explained with the help of the subset property as developed in Berwick (1985). The subset property says that in the process of parameter setting the parameter value that corresponds to the narrowest language compatible with the evidence is chosen. Jakubowicz argues that, with respect to the acquisition of pronouns, the subset property predicts that the most restrictive, unmarked, choice would be one where him is interpreted as an anaphor. Wexler & Manzini (1987) argue that it cannot be the case that children first treat pronouns as anaphors because a child's first use of pronouns seems to be deictic (cf. also Corver (1985)). Manzini & Wexler, thus, seem to claim that once a pronoun is used deictically, it must always be used as such. But, of course, there is no principled reason why this should be the case. The fact that a pronoun is treated as a pronominal (or as .an anaphor) in some cases does not necessarily imply that it is always treated as a pronominal (or anaphor). Compare (102a) and (102b):
114
Martin Everaert
(10Z) a. His mother hates me b. John lost his appetite In (lOZa) his is used deictically, which is not possible in (lOZb). We would like to suggest that the same holds for the acquisition facts under discussion. If pronouns are, in the initial stage of language acquisition, (partly) unidentified, it is to be expected that they are treated as anaphors in some cases and as pronominals in other cases. It seems worthwhile to look again at the test results with the specific BT-approach put forward here in mind.
8 Conclusion In this chapter I have investigated the consequences of the claim that the phenomenon of long-distance refiexivization as opposed to 'regular' short-distance refiexivization should not be explained by distinguishing two types ofBTs or two types of anaphors, each subject to its own specific conditions. The different behaviour of anaphors with respect to their domain of interpretation can be made to follow from their 'intrinsic' properties, such as, for instance, their morphological structure, their semantics and their syntactic behaviour outside BT. In pursuing this line of research it was necessary to readjust the binding theory, reintroducing Chomsky's original Pisa approach to BT, where a distinction was made between the governing category (or categories) and the minimal governing category of a lexical item. Furthermore, I have argued that, apart from the regular binding conditions, we need contextual binding principles for determining what anaphoric status must be assigned to pronouns that are not fully specified for the features (±A) or (±P), i.e. pronouns that are not lexically marked as 'anaphor' or 'pronominal'. I have shown that the Frisian pronoun him is such an unspecified pronoun and that the same holds for the apparent 'anaphor' zich in Dutch. Taken together, the binding conditions and the contextual binding principles give a straightforward explanation of why Dutch zich and Frisian him sometimes behave as SD-anaphors and on other occasions as LD-anaphors, while the complex variants zichzelJ and himsels behave like SD-anaphors throughout.
Acknowledgement I would like to thank the following people for helpful discussions and comments: Peter Coopmans, Ger de Haan, Emmy Jacobs, Hans van de Koot, Charlotte Koster, George Rebuschi, Tanya Reinhart, Eddy Ruys, Eric Reuland and Anne Zribi-Hertz. I am grateful to Ger de Haan for his help in gathering data from native-speaker informants of Frisian. This chapter was first presented at the workshop on Long-Distance Anaphora, Groningen Univer-
Contextual detennination of the anaphorlpronominal distinction
us
sity, 18-20 June 1987. Parts of this chapter were also presented during seminars at Tilburg University, 17 September 1987; and the University of Utrecht, 29 January 1988. I thank the respective audiences for critical and insightful remarks. This research has been made possible by a grant from the Niels Stensen Foundation and a fellowship of the Royal Netherlands Academy of Arts and Sciences.
Notes 1. Hereafter I will use the more neutral abbreviations L(ong) D(istance binding) and S(hort) D(istance binding), respectively. 2. Discussing German data from Grewendorf (1984), Sternefeld (1985) imposes a thematic hierarchy restriction on SD (see also Kiss, chapter 12). Giorgi (1984) argues that such a restriction is typical for LD-phenomena. 3. Cf. Zwicky (1977) on the distinction between special clitics and simple clitics. 4. This class is actually more complex. In some languages (French, Italian) these reciprocals allow a clitic reflexive to be added (cf. Van Riemsdijk (1985)). 5. As Safir (1989) observes, the notion 'anaphor' disguises fundamental differences between reflexives and reciprocals that playa role in their distribution. 6. One might well question Yang's observation. It has been observed that there are (marginal) cases where English allows long-distance binding of reciprocals. Acceptability decreases when a reflexive is used, contrary to what we might expect, cf. Chomsky (1981: 214), Kuno (1987: 95): (i) a. b. (ii) a. b.
?TheYi think it bothered each otheri that .. . *TheYi think it bothered themselvesi that .. . ?TheYi made sure it was clear to each otheri that .. . *TheYi made sure it was clear to themselvesi that .. .
Also in Dutch, some cases of long-distance binding of reciprocals seem to be marginally acceptable for some speakers compared with long-distance binding of complex reflexives: (iii) a.
??Zih hoorden mij over elkaari roddelen they heard me about each other gossip b. *Zih hoorden mij over zichzelfi roddelen they heard me about themselves gossip
7. Cf. Everaert (1986a) and Koster (1987) for a more detailed description of the distribution of Dutch reflexives. 8. Sometimes the clitic pronominal variant 'm is somewhat better than the non-clitic variant
hem. 9. Observe that Chomsky's (1986a) revision of BT in fact gives a similar result. In his approach, the governing category of an anaphor/pronominal in a specific position is not fixed. One is free to choose the governing category in such a way that the anaphorl pronominal satisfies BT. 10. That is, non-D-elements (Chomsky (1986a)). With this limitation, Condition C and its
116
11. 12. 13.
14. 15.
Martin Everaert effect on the distribution of R-expressions would be outside the scope of BT, as has been suggested before (Chomsky (1982), Reinhart (1983a». Cf. Giorgi (1987), Graffi (1988) and Hestvik (1989a) for arguments that the notion complete functional complex may be defined as 9-domain. Note that, in this approach, the feature specification (+A, +P) does not mean 'pronominal anaphor', as is the case in 'standard BT'. It appears that the pronoun system of Afrikaans and the pronoun system of Dutch for first and second person pronouns show similar distributional properties. For reason of space I will not discuss this issue here, cf. Everaert (1986a). See also Bouchard (1984) and Fanselow (1988) for discussion of similar data in other languages. In the text I use him 'him' but of course the same holds for har 'her' or it 'it'. The subject interpretation of possessives in (i) has been analysed as another case where a 9-role is syntactically visible but not projected onto a syntactic position: (i) the enemy's destruction of the city
Zubizarreta (1987) and Grimshaw (1986) argue that the possessive in (i) is adjunct-linked to the lexically saturated subject 9-role of destruction. 16. Zubizarreta (1987) claims that only verbs with an affected object allow a zich-variant. I do not think this claim is sustained by the facts. The following verbs allow a zich-variant without being 'affected': (i) a.
Jani zag zich i in de spiegel zichzelfi 'John saw himself in the mirror' b. Hij onderschat zichi zichzelfi 'He underestimated himself'
Conversely, there are also verbs with affected objects which do not allow zich-variants (and I here follow Rizzi's (1986a) classification of verbs with 'affected' objects): (ii) a.
Kareli verbaast *zichi met die opmerking zichzelfi 'Karel surprises himself with that remark' b. Hiji helpt *zichi zichzelfi 'He helps himself' c. Hih overtuigde *zichi van zijn ongelijk zichzelfi 'He persuaded himself of his being in the wrong'
17. Here I follow insights which were first formulated in Hellan (1980) although I crucially differ from his view on the relation between (69) and (70). 18. Note that, in our view, pronouns are not necessarily fully specified for the (A, P)features. This is one of the reasons for the BT to be formulated as in (31), and not as in (32). 19. Note also the difference between (i) and (ii) from Safir (1989):
Contextual detennination of the anaphorlpronominal distinction
117
(i) #The menj are similar to themselvesj (ii) The menj are similar to each otherj
20.
21.
22. 23.
He argues that the predicate to be similar to implies distinct reference between its arguments. The reciprocal, but not the reflexive, satisfies this condition, making (ii) perfectly grammatical while (i) is semantically odd (marked #). It might be possible to reinterpret the BT in such a way that the VP-hypothesis accounts for the facts in (82). Suppose we make a distinction between a EI-domain and a EI/domain, where the latter is defined as a maximal projection that contains all potential grammatical functions compatible with its head. This would mean that the embedded VP-node in an Ad-passive (the VP\-node in (8Sa)) could never be a EI'-domain because it does not contain a subject position. Assume, furthermore, that GCs and MGCs are defined in terms of EI/-domains and EI-domains, respectively. As a result the MGC of a pronoun is not necessarily also a GC. For instance, it would mean that only the IP-node in (8Sa) is a GC for X while VP\ is the MGC for X. Now, if we also incorporated the 'BTcompatibility'-notion of Chomsky (1986a) in our binding theory, the grammaticality judgements in (82) could be accounted for. Zich and zichzelJ, both marked ( +A), must be bound in IP, while zich must be free in its MGC (VP\) and zichzelJmust be bound in its MGC (IP). Although the pronominal hem is free in VP\, its MGC, it is not free in its GC (IP), as its (- P)-specification requires. The fact that inherent reflexivity is generally not allowed with prepositional objects and limited to object positions (cf. Everaert (1986a)) follows from the 'sisterhood condition' which is incorporated in the definition of EI-linking. The zichzelfvariants are pragmatically a bit odd but perfectly grammatical. Although it is firmly established that prepositional objects 'require' an SD-anaphor (cf. (14d)) there are some curious counterexamples: (i) Janj stelde mij aan zichj voor 'John introduced me to himself' (ii) Hijj nam mij tegen zichj in bescherming 'He protected me against himself' (iii) Hij; wist [PROj mij voor zichj te winnen door zijn houding] He managed me for himself to win by his attitude 'He managed to win me over' (iv) Hij; probeerde [PROj mij aan zichj te binden] He tried me to himself to bind 'He tried to get me tied up with himself' (v) Hij; wist [PROj de klas aan zichj te onderwerpen] he managed the class to himself to subject 'He managed to subject the class' (vi) Hijj probeerde [PRO i voor zichi op te komen] 'He tried to stand up for himself' (i) and (ii) are pragmatically odd but, still, not as bad as one would expect them to be. For most speakers (iii-vi) become considerably worse if we take simple clauses: (vii) a. ??Hij; heeft mij voor zichj gewonnen door zijn houding b. ??(Door zijn houding) heeft hij; mij aan zichj gebonden
118
Martin Everaert c. *Hij; heeft de klas aan zichi onderworpen d. ??Hij; komt voor zichi op
It must be noted that in all these examples the verb phrase contains two internal arguments: [vp NP PPj. (vi) is no exception as it is an ergative construction. One might argue for a small clause analysis of these constructions but there is no clear motivation to do so. Alternatively, the analysis proposed in the text for (82) might hold in these cases as well. 24. As Grimshaw & Rosen (1988) note, the choice of verbs in the experiments in which condition B is violated is important; they include verbs like wash, dress, etc. These verbs can often be used intransitively in English with a reflexive interpretation (cf. Zubizarreta (1987)) and are precisely the class of verbs which, in Dutch, allow zich in object position.
5 On the interaction between antecedentgovernment and binding: the case of longdistance rejiexivization Pierre Pica
1 Introduction It is a well-known fact that the various proposals aiming at a universal treatment of reflexives across languages formulated in the framework of generative grammar, or related frameworks, face counterexamples. This is particularly striking in the case of a rich variety of languages where the anaphoric relation between the reflexive pronoun and its antecedent does not seem to obey the locality conditions which have been formulated to account for the behaviour of reflexives in languages such as English. This remains true whether these conditions are expressed in terms of specified subject - or its corresponding notion in the framework of Chomsky (1981) and Chomsky (1986a) - in terms of Tense (see the propositional island constraint of Chomsky (1977)), of Case (as in Chomsky (1980)), or in terms of generalized SUBJECT (a notion which expresses apparent similarities between subject NPs and the element Agr of the verbal inflection, as in Chomsky (1981)).1 This point can be illustrated by a sentence such as (1) in Icelandic, whose grammaticality contrasts with the ungrammaticality of its direct apparent translation in English since the reflexive sig refers to the matrix subjectJ6n, in apparent violation of all formulations of the binding theory:
(1)
J6ni sagoi peim [ao Maria elski (subj) sigi] J6n told them [that Maria love 3sg self]
The type of reflexive which undergoes long-distance reflexivization is moreover subject to a strict subject-orientation requirement, as illustrated by the Danish sentence (2): (2)
*jeg forta:ller Gertrudei om sigi I told Gertrude about 3sg self
119
120
Pierre Pica
The type of reflexive illustrated by sig in (1) and (2) has therefore been the subject of very extensive studies in the literature, where two main attitudes can be distinguished: (a) It is claimed that elements such as sig are not reflexives, but rather exhibit some pronominal properties, similar to those of so-called logophoric pronouns (see, among others, Maling (1982, 1984), Sells (1987), and (for a different point of view) Koopman & Sportiche (1989». (b) It is claimed that, although elements such as sig are reflexives, they do not obey the usual conditions on binding; that is, that (1) and (2) illustrate a marked phenomenon. Many variants of this last attitude can be again distinguished, and it is sometimes claimed that reflexives of the type illustrated in (1) and (2) (i)
do not fall within the scope of core grammar (cf., among others, Chomsky (1982», and, from a different point of view, Kuno (1987); (ii) must be treated in terms of a parametric approach (cf., among many others, Yang (1983), or Manzini & Wexler (1987»; (iii) provide evidence for the need of specific devices such as - a level of lexical structure (cf., among others, Mohanan (1982), Kiparsky (forthcoming» - a thematic hierarchy (cf. Giorgi (1984» - a notion of predicate complex (cf. Hellan (1986a) and the lexical-functional grammar (LFG) literature in general, where the very similar notion of sentence nucleus is developed). One should keep in mind, however, that the various analyses that have been proposed to account for the phenomenon illustrated in (1) and (2) have led to the paradoxical situation that the following fact has been forgotten [mentioned by Faltz (1977) - eds.]: We never find long-distance binding (across a specified subject) of a reciprocal, or of a clearly non-mono-morphemic reflexive (see note 2 below) - even in languages where long-distance binding of reflexives can overtly be observed (see Pica (1985, 1987), and the references mentioned there). This point can, for example, be illustrated by the contrast between (1) and (3) in Icelandic:
Antecedent-government and binding; the case of long-distance refiexivization 121 (3)
*jJeirj sagoi [ao Maria elski (subj) hvorn annanj] they said [that Maria love each other]
2 On the distinction between mono-morphemic and non-mono-morphemic anaphors2 Such a situation - and the fact that reciprocal expressions and non-mono-morphemic anaphors (in the sense of Pica (1985)) are never restricted to subject orientation 3 - suggests, in our opinion, that the binding principles do not vary as much as one could believe they do in the prima-facie violations of the specified subject constraint (SSC) exhibited by long-distance reflexives. This state of affairs is the main motivation of the analysis proposed in Pica (1987) (developing ideas formulated in Pica (1985)), where it is claimed that the kind of binding exhibited in (1) and (2) belongs to the core of the grammar. We proposed, more specifically, that in an example such as (1), the apparent violation of the SSC of the mono-morphemic reflexive - or its equivalent in terms of governing category or complete functional complex - is due to a cyclic movement (from lnfl to lnfl) at the level of logical form. The analysis we had in mind was based on the hypothesis according to which anaphors could be divided in two ways - whether they could be analysed as heads (like, for example, sig in Icelandic, or soi in French), or as maximal projections (like, for example, himselfin English). It was presupposed that all anaphors move - either at the level of logical form or at S-structure - but that only XO anaphors can escape the SSC by an LF cyclic head-to-head movement through Compo XP anaphors could not make use of such an escape hatch and were consequently subject to more severe locality conditions. This way of looking at things amounts to saying that the asymmetry between anaphors which undergo long-distance reflexivization and anaphors which are subject to stricter locality requirements can be reduced to the effect of Emonds's structure preserving constraint (cf. Pica (1987) note 9). The hypothesis according to which the structure preserving constraint applies at the level of logical form has recently been challenged, explicitly or implicitly (see, among others, Chomsky (1988a), Stowell (1987)). It seems indeed reasonable to assume that while the projection principle or the O-criterion holds at all levels of representation, X' theory does not hold at LF, which is neither directly nor indirectly projected from the lexicon (see also note 4 below). While our leading idea of a cyclic movement of the reflexive from lnft to lnfl has been shown to account for the behaviour of reflexives in various languages (see,
122
Pierre Pica
with reference to Chinese, and among others, Battistella (1987) and Cole, Hermon & Sung (forthcoming)), we intend to show in the present chapter that, while this main idea is on the right track, the asymmetry between long-distance reflexives and reflexives which do not allow long-distance reflexivization does not follow from the status of the reflexives with respect to X' theory. We claim that it follows instead from the well-known asymmetries between arguments, adjuncts and XO elements with respect to extraction, in a theory in which reflexives move in order to be governed by their antecedents, along the lines (of the leading idea) of Chomsky (1986a), according to which all anaphors are clitic-like elements at LF. In particular, it will be shown that the nature of longdistance reflexivization provides evidence for a theory according to which binding theory applies to the trace of the reflexive only (and not to the reflexive-antecedent relation per se), contrary to what is suggested in Pica (1987). The theory we develop contradicts recent proposals according to which binding theory overlaps with the theory of antecedent-government (developed in, among others, Huang (1982), Lasnik & Saito (1984) and Chomsky (1986b)), to which it can be reduced (see also Manzini in chapter 10). These proposals seem at first glance to be supported by sentences such as (4) in English, which involves NP-movement at S-structure (and where, in Chomsky's terms, NP-movement crosses at least two barriers (VP and CP) (cf. Chomsky (1986b)): (4) *John [vp seems [cp that [IP Mary considers [e to be intelligent]]]] We intend to show, however, that, paradoxically, long-distance reflexivization provides an argument for a theory in which binding facts are not entirely derivable from antecedent-government, but rather derive from the interaction of antecedentgovernment and binding theories. We shall also see that the subject orientation of certain reflexives derives in a natural way from the interaction of these two theories with the a-criterion. If our analysis is on the right track, the linguistic variation observed across languages can be reduced entirely to the respective lexical properties of the different reflexives and is not the result of the setting of different values of a parameterin the traditional sense of this notion.
3 On the argument-like character of long-distance reflexives Our analysis of long-distance reflexives, to be developed below, can be summarized in the following way:
Antecedent-government and binding; the case of long-distance rejlexivization ' 123 (a) Clitics can be divided into two types, whether they cliticize onto Infl at Sstructure or at LF. (b) Reflexives such as sig cliticize onto Infl at LF. (c) A reflexive clitic or - more precisely - its trace can escape the SSC (or whatever the precise formulation of the binding condition on anaphors turns out to be) through an escape hatch such as the C position ofComp (CP).4 (a) is partly reminiscent of Huang's analysis of some apparent contrasts between the behaviour of wh-words in Chinese and English. We shall see below that (c) is the result of quite complex mechanisms involving (among other mechanisms) raising ofInfl to C. We shall see also that the interaction of (b) with some general principles on chain formation enables us to derive the strict subject orientation of
sig. Our analysis of long-distance reflexivization in terms of cyclic movement amounts to saying that (1) in Icelandic exhibits at LF properties very similar to those of 'clitic climbing' that can be overtly observed in some languages. This last phenomenon is illustrated by (5) in Italian, where the clitic si in the main clause is 'associated with' the object position e of the verb vedere of the embedded infinitive clause: (5)
Gianni si vuole [vedere e] Gianni 3sg self (cl) wants [to see]
This is so because this manner of looking at things implies that the relationship of sig toJ6n in (1) is established through an LF cyclic movement from Infl to Infl, through C, as illustrated in (6), the representation of (1): (6)
Jon Infl sagoi (cp [C
t
It
[IP
Maria Infl elski sig]]]
It
I
We assume, following Pica (1987), that the C-position is only available for movement when it is not lexically filled. This is the case when the C-position is not filled at S-structure, or when the complementizer deletes at LF - according to a general principle of interpretation which asserts that only complementizers associated with tensed (indicative) inflections remain present at LF (see Pica (1987) and, for a different suggestion, section 6 below). We can then account for the fact that in languages which make use of lexical inflection and of lexical complementizers, long-distance reflexivization (of the type described above) cannot, in a very general way, take place when the reflexive is embedded in a clause whose inflection is indicative. This is illustrated by the contrast between (1) and (7) in Icelandic, where the embedded clause is in the indicative mood:
124
(7)
Pierre Pica
*Joni veit [aD Maria elskar (ind) sigi] Jon knows [that Maria loves 3sg self]
The essence of this hypothesis, in which the reflexive moves at LF, seems also to be supported by the following Icelandic sentence (adapted from Maling), where the reflexive contained in the adverbial clause cannot refer to the subject of the matrix clause to which it is attached, but can however refer to the subject of the higher clause: (8)
Joni Infl! sagoi [ao Gertrude; Inflz vreri (subj) glOd [ef Maria Inft3 Jon said [that Gertrude would be happy [if Maria bydi (subj) Serif';]] invited 3sg self (= Jon)]]
This shows that ser can only move from Infl to Inft if the most prominent Inft ccommands the lowest Infl - a general state of affairs in the case of movement. Our analysis is also supported by the treatment of overt long clitic movement (at S-structure) developed in Kayne (1989a), who - along the lines of note 11 of Pica (1987) - shows that overt clitic movement at S-structure is derived through cyclic head-to-head movement of an XO element, though not in the same way we advocate for clitic climbing at the level of logical form. Note that, as expected under the 'C-deletion hypothesis', clitic climbing at Sstructure is ungrammatical when the C-position of the embedded infinitive is lexically filled at that level, as illustrated by (9) in Italian, from Rizzi (1982): (9)
*non Ii so [se [fare e]] (I) neg them (cl) know [if [to doll
As noted by Kayne (1989a: (IS», the ungrammaticality of (9) in Italian parallels the ungrammaticality of (10), in which the mood of the embedded clause is subjunctive. This is expected under our own analysis, since the C-position of the subjunctive clause is, in our terms, still present at S-structure in (10), which, in our terms, contrasts with (1) in Icelandic: (10)
*Gianni li vuole [che Maria veda (subj) e] Gianni them (cl) wants [that Mary see]
4 Some differences between clitic climbing at S-structure and clitic climbing atLF We do of course find some differences between, in our terms, clitic movement at Sstructure and clitic movement at LF, but these differences seem reducible to the
Antecedent-government and binding: the case of long-distance rejlexivization 125 effect of general principles. We know, for example, that long overt clitic elimbing at S-structure is restricted to 'pro-drop languages', while what is, in our terms, elitic elimbing at LF is not. Kayne points out that the absence of overt elitic climbing at S-structure in non-pro-drop languages can be derived from the theory of barriers, developed in Chomsky (1986b), according to which all maximal projections are potential barriers for antecedent-government. This theory accounts for the ungrammaticality of (11) and (12) in French, where the trace of the object elitic is not bound within its first maximal projection (VP): (11) (12)
*Jean Jean *Jean Jean
a promis [de les bien faire e] has promised [to them (el) well do e] se veut [voir e] 3sg self (d) wants [see e]
Kayne suggests that the fact that VP does not act as a barrier in a pro-drop language can be reduced to the effect of the general principle, according to which barrierhood can be overcome by a process of L-marking (see Chomsky (1986b)). That is, VP is not a barrier in pro-drop languages, because the rich character of Inti enables that element to L-mark the VP. This hypothesis accounts for the contrast between (11) and (12) in French on the one hand, and (13) and (5) in Italian on the other, (13)
Gianni ha promesso [de farli bene e] Gianni has promised [to do them (el) well e]
where V adjoins to Inti after ditic movement, as indicated by the fact that V appears before the ditic (see Kayne (1989a)). The fact that VP is a barrier for ditic movement at S-structure dearly indicates that the part of the object NP which moves at this level is an XO - presumably a DO, in terms of Abney (1987). This is confirmed by the ungrammaticality of the following sentence in French (from Kayne (1989a)), and its equivalent (IS) in Italian, which show that overt ditic dimbing is blocked by the intervening head ne (see also, on the status of ne, Pollock (1989)): (14) (15)
*Jean l'a fait [ne pas manger e par l'enfant] e by the child] Jean it (d) has made [NEG eat *Gianni l'ha fatto [non mangiare dal bambino]
The ungrammaticality of (14) and (IS) shows indeed that the original trace of the XC-moved element cannot be ')I-marked since XO are not arguments (see Lasnik & Saito (1984), Chomsky (1986b), Chomsky (1988a)). Long-distance retiexivization is however dearly not sensitive to the presence of an intervening head, and is not
126
PietTe Pica
restricted to pro-drop languages, as illustrated by the French sentence (16), where the behaviour of soi is reminiscent of the behaviour of sig in Icelandic: (16)
Oni souhaite toujours [que les gens ne disent (subj) pas du mal de SOii] one wishes always [that people do not slander 3sg self]
This state of affairs can be reduced, in terms of the present framework, to the fact that reflexives of this kind are argument XPs (or, more precisely, that the part which moves is both an XP and an argument - presumably an NP - contra Pica (1987». We can say that soi or sig can consequently adjoin to VP (or to V) at LF'I-marking its original trace, which is then antecedent-governed - since intermediate traces delete in the case of movement of an argument (see Chomsky (1988a». As a consequence, XP argument reflexives are not sensitive to the presence of intervening heads. Note that the distinction between XO-clitics, which cliticize at S-structure, and XP clitics, which cliticize at LF (see however note 4), allows us furthermore to explain another distinction between overt clitic climbing at S-structure and, in our terms, clitic climbing at LF. It is well-known that all standard cases of overt clitic climbing at S-structure are cases of subject control. As noted by Kayne (1989a), this is consistent with the idea that, once the XO-clitic has adjoined to Infl, it is the whole complex (lnfl (cl Infl» which moves to C, and then to the higher Infl, in such a way that the matrix and embedded subjects end up being coindexed. Comparable Infl-to-(C-to)-Infl movement in object control structures, however, would coindex two Infl whose respective subjects are themselves not essentially coindexed. The fact that the XO-clitic cannot be extracted alone from the complex (Infl (cl Infl» follows from Baker's prohibition against traces dominated by a zero-level category (see Baker (1988». It is tempting (as suggested by Chomsky (class lectures» to reduce this constraint to a derived effect by antecedent-government, that is, to the fact that such a trace could not be properly antecedent-governed. This can be done if we assume that XC-elements are not arguments and cannot consequently 'I-mark their original traces. Intermediate traces cannot delete and are then subject to antecedent-government, which subsumes Baker's constraint on traces dominated by a zero-level element. The existence of long-distance reflexivization in object control structures and in structures involving subjunctive complements clearly indicates - in these terms that Infl does not move along with the long-distance XP-reflexive. That is, the whole complex (Infl (sig Infl» does not move to the upper Infl after sig has adjoined to Infl. We can reduce this fact (from which 'long-distance reflexivization' derives) to
Antecedent-government and binding: the case of long-distance rejiexivization 127 the argumenthood of reflexives such as sig: intermediate traces of argument XPs at LF can delete and are consequently not subject to Baker's constraint, nor to the head constraint, as illustrated by (16) above.
5 On the interaction between binding and antecedent-government The 'Infl-to-Infl' analysis developed in the preceding sections raises the following question: why is extraction of a reflexive such as sig impossible from a tensed sentence, as illustrated by (7) above? That is, why does (3) (above) contrast with (17), which is of course perfectly grammatical: (17) Who did he say that Mary loves e? We assumed in Pica (1987) that the ungrammaticality of (3) derives from a conspiracy between the SSC and the structure preserving constraint which forced movement of the long-distance reflexive through Comp, in a theory where the SSC applies also to the reflexive itself. Yet, we saw earlier that there are reasons to doubt that Emonds's constraint should apply at LF. It is furthermore dubious that the theory of binding applies to the anaphoric relation which holds between the reflexive and its antecedent if this relation is expressed in terms of government. In addition, our 1987 analysis leaves open the two following questions: (i) Point (c) seems to imply that the Comp position should be incorporated into the definition of the specified subject condition. Why should this be the case if the SSC is expressed in terms of governing category (as in Chomsky (1981)) or in terms of complete functional complex (as in Chomsky (1986b))? (ii) Why should it be impossible for a reflexive like sig or himself-whether Comp is incorporated into the definition of the SSC or not - to adjoin to an embedded IP or C (whatever the character of the embedded IP might be with respect to [±Tense]), and then undergo further movement to the upper clause? We want to suggest that the answers to these questions follow from the interaction of antecedent-government, which applies to all non-de1etable traces, and binding theory, which applies to the original traces of all types of reflexives which, as we shall see, all move at LF - since these traces are identified as anaphors. Our new proposal suggests that the C-position does not need to be incorporated into the definition of the SSC itself - contra Pica (1987). Let's first analyse the case of long-distance reflexivization.
128
Pierre Pica
5.1 On some similarities between long-distance reflexivization and switch reference Note that, in the case of a mono-morphemic reflexive like sig in Icelandic or soi in French, the reflexive could adjoin directly to IP and move further, but that, in our terms, the embedded original trace would not be bound by any of the intermediate traces which we assume must delete, since the reflexive is an argument, as discussed in the previous section. The corresponding sentence would consequently be ruled out by the SSC or its reformulation in terms of Complete Functional Complex. We want to suggest (contra Huang & Tang (chapter 13)) that a long-distance reflexive must consequently move to Infl (or, more precisely, to Agr and then to Infl, if the general structure assumed in Pollock (1989) is on the right track). We assume further that such a reflexive can transmit its index to Infl and (adopting in another context the leading idea of Borer (1989)) that the whole complex (Inti (Infl (Agr (Agr sig)))) moves to C at LF if that position is empty, as is the case with infinitives and subjunctive clauses, whose complementizers delete at this level of representation (or, alternatively, in the case of subjunctive clauses, that Agr moves to C, whose complementizer cliticizes on the main verb at LF, as suggested to me by J. Uriagereka). It is thus the trace of Infl which, according to this analysis, binds the original trace of the reflexive. Note incidentally that the hypothesis according to which the C-position of subjunctive clauses is empty at LF is supported by Pollock's analysis in which French embedded subjunctives do allow a small pro in subject position. This empty pronoun is, in our terms, licensed by Agr in C - as illustrated by the following contrast in French, where (19) is ruled out because the trace is not properly governed (see Pollock (1986)) and cannot be supplemented by a small pro: II faudrait que pro viennent (subj) plus de linguistes a nos reunions it would be better that come more linguists to our meetings (19) *11 dit que e viennent (ind) plus de linguistes a nos reunions more linguists to our meetings he says that come
(18)
We believe that the same hypothesis can account for the fact - observed in subjunctives - that the subject pronoun of the embedded clause cannot corefer with the subject of the main clause (see, among others, Ruwet (1984), Picallo (1985)), as illustrated in (20) in French. This fact is reminiscent of the switchreference phenomenon overtly marked in many other languages: 5 (20)
I1s i souhaitent qu'ils*ilj partent they wish that they leave
Antecedent-government and binding: the case of long-distance rejlexivization 129 We shall say, in the terms ofthe present analysis, that at LF, Agr moves in the C position of the embedded clause and extends the governing category of the embedded subject (which enters into disjoint reference with the main subject). Coming back to the analysis of long-distance reflexives, we see that movement of Infi to C (and thus of the reflexive to Infl) is necessary for the binding of the trace of the reflexive in its CFC when the reflexive is an argument. 6 The same point is also illustrated by the fact that in (8) - repeated as (21) - sig (which satisfies antecedentgovernment in the adverbial clause) could not move directly from Infl3 to Infli if the embedded clause were indicative. In other words, sig could not move from Infl3 to Infli if Agr of the adverbial clause could not move to C. (21)
Joni Infli sagai [aa Gertrude; Inflz Vleri (subj) glOd ref Maria Infl3 Jon said [that Gertrude would be happy [if Maria bydi (subj) Serif';]] invited 3sg self (= Jon)]]
We see that under this analysis both movement of sig to Infl3 and of Infl3 to Care two necessary conditions for long-distance reflexivization. These two conditions are both required by the SSC: the former, because direct adjunction of sig to IP, or of sig to a constituent of IP, would lead to a violation of SSC - assuming that reference to the C-position of Comp in the definition of the CFC is not possible, contra Pica (1987); the second condition, because it creates the necessary antecedent for the original trace of the argument reflexive. This analysis excludes direct movement of the reflexive in the upper Infl in sentences like (1). It entails that long-distance reflexivization is achieved through a cyclic movement of a reflexive, under the assumption that this reflexive is an argument and can consequently move alone, stranding the complex (I (Agr I)) in C.
One question remains unanswered: why is it impossible for a reflexive such as sig to refer to an object NP when it is in its adjoined position to VP or to V (see note 4 below)? We shall assume that the so-called 'subject orientation' of the reflexive follows from the fact that the coindexation of the reflexive with an object antecedent would indirectly imply coindexing of the object antecedent with the trace of the reflexive and the reflexive itself, resulting in a violation of Rizzi's chain algorithm, hence in a violation of the {l-criterion (see Rizzi (1985)).7 5.2 The case of clause-bound reflexives
5.2.1 The case of XO-rejlexives The analysis developed in the preceding sections can be extended in a straightfor-
130
Pierre Pica
ward way to the connected form of soi, se, which diticizes at S-structure in French. Se cannot be long-distance bound when it is embedded in a subjunctive dause, as illustrated in (22): (22)
*Paulj souhaite [que Jean sej photographie (subj) e] Paul wishes [that Jean 3sg self (d) photograph e]
This is natural under the assumption that se - which undergoes overt clitic movement - is an XO (which adjoins to V in a non-pro-drop language, along the lines of Kayne (1989a)) and is, as such, subject to the requirement that all (nondeletable) traces should be antecedent-governed. The same requirement rules out the S-structures (23) and (24) below, which are reminiscent of (11) and (12) above: (23)
*Paul se souhaite [que Jean photographie e] Paul 3sg self (d) wishes [that Jean photograph e] (24) *Paul se veut [photographier e] Paul 3sg self (d) wants [to photograph e] As expected under the present framework, the equivalent of (24), where the main subject is coindexed with the embedded subject, is nevertheless grammatical in a pro-drop language, as illustrated by the grammaticality of (5) above, repeated below as (25): (25)
Gianni si vuole [vedere e] Gianni 3sg self (d) wants [to see]
Note that our analysis predicts the ungrammaticality of the French sentence (26), where the diticization of se is blocked by the presence of an intervening head (ne), and which is reminiscent of (14) above: (26)
*Jean s'est fait ne pas dessiner par l'enfant Jean 3sg self (d) has made NEG drawn by the child
5.2.2 On the adjunct-like character of the so-called non-mono-morphemic reflexives Let us now turn to the more complex case of non-mono-morphemic reflexives of the himself-type in English. We daimed in Pica (1987) that non-mono-morphemic reflexives such as himself were XP, and could not, as such, move through C, as a result of the structure preserving constraint. We saw above that this line of argument is questionable if the structure preserving constraint does not apply at LF. Let us admit, slightly reformulating a hypothesis of Pica (1985), that it is the (possessive) specifier (in this case him), and not the whole 'reflexive' (in this case
Antecedent-government and binding; the case of long-distance rejlexivization 131 himseif),8 which is the element subject to movement. Let us furthermore assume that this specifier, which agrees in gender and number with its antecedent, is in the internal specifier position of the relational noun self This amounts to saying that him of himself is very similar to lui in sentences like (27) in French, (27)
11 lui casse la main he to-him (dat) breaks the hand
where we daim that lui originates in the specifier position of the relational noun main, from which it is assigned a a-role (see Pica (forthcoming)). In the case of a reflexive such as himself, however, we assume that the specifier does not get a a-role since it cannot a-bind the open position associated with the head noun self This noun blocks the coindexation internal to the NP (along the lines of Pica (1987) footnote (6)). This point is, in our terms, illustrated by the lack of morphological uniformity (accusative or, more likely, dative in the case of him (of himself); genitive in the case of my (of myself)). The anaphoric specifier - which we tentatively assume is an XP - acts as an adjunct XP since it does not get a a-role. Let us assume (contra Pica (1987)) that such specifiers do also diticize into Infl at the LF-Ievel interpretation - a process which is reminiscent of the determinant cliticization illustrated in (27) above, and which can also overtly be observed in Galician, as illustrated by (28) from Uriagereka (in preparation): (28)
Vimo-Io Kremlin we saw-the (d) Kremlin
Note that the hypothesis according to which all 'reflexives' are clitic-like elements at the level of logical form gives us a straightforward explanation of the absence of long-distance reflexivization in the case of non-mono-morphemic reflexives. The intermediate traces of the reflexive's specifier, which is an adjunct, cannot delete and are thus subject to Baker's constraint. This amounts to saying that the ungrammaticality of (29) (29) *Peterj asks that you love (subj) himselfj parallels the ungrammaticality of (22) and (23) above. It is well known that reflexives of the himself-type are not strictly subject oriented, as illustrated by (30): (30) I told Bentj about himselfj
132
Pierre Pica
This can now be reduced to the fact that him does not bear a a-role: the coindexation of the specifier and its trace at LF with the object Bent is consequently not ruled out by the a-criterion. Although many details still need to be worked out, it seems to us that the leading ideas outlined above enable us to derive in a natural way the basic binding facts as well as the well-known relationship between long-distance binding and subject orientation. It also enables us to account for the absence of long-distance reciprocals if we assume that reciprocity always involves movement of an adjunct (see Pica (forthcoming) for the detail of the analysis, as well as for a treatment of some apparent counterexamples).
6 Some brief remarks on the so-called long-distance reflexive in English The analysis sketched in the preceding sections, if it is on the right track, provides strong evidence for a level of logical form, to which the well-attested rule of move-a applies. We believe that this analysis is corroborated by 'long-distance binding' of the type illustrated by sentence (31) in English: (31)
theYj said [cp [c that [rp [NP pictures ofthemselvesj] are on sale]]]
Under a movement analysis, (31) contrasts with overt movement from a subject position (as illustrated in (32)), which is ruled out by subjacency (along the lines of Huang (1982)), since movement of the wh-word crosses at least two barriers (NP and IP): (32) *who did you say [cp [c that [rp
[NP
pictures of e] are on sale]]]?
As expected under the Barriers framework, which we have been assuming throughout, the equivalent of (32) is however possible at LF, as illustrated by (33), where the argument wh-word moves, presumably to the specifier ofNP, 'Y-marking the original trace: (33) who thinks [cp [that [NP pictures of whom] are on sale]]? But why should movement of an adjunct (such as the specifier of a non-monomorphemic reflexive) out of a subject be possible in (31)? Let's assume that the whole reflexive (in this case themselves) moves to the Spec of NP and, adopting in another context an analysis proposed for extraction from NP in Spanish by Torrego, that the whole reflexive moves to the specifier of CP, as illustrated in (34), where it is L-marked by a process of specifier-head agreement, along the lines of Chomsky (1986b: 26):
Antecedent-government and binding: the case of long-distance rejlexivization 133 (34) theYi said
CcP themselvesi Cc that hp pictures of e are on sale]]]
The adjunct anaphor them can now move out of NP and CP, which are both Lmarked, to a position in which it will govern its antecedent, as in the simpler sentence (35):9 (35) theYi like pictures of themselvesi Note that our analysis, which amounts to saying that the element which is a reflexive (in the traditional sense) and the element subject to movement do not always correspond one to the other, enables us to suggest that the apparent linguistic diversity observed across languages can in fact be entirely reduced to the respective lexical properties of the different linguistic entities. We believe that such a reduction of language variation to the effect of variation within the lexicon constitutes strong evidence in favour of the formulation of a universal grammar, and that it constitutes in itself a considerable support for the idea that the study of UG is part of the study of the human mind within the more general domain of cognitive sciences. The analysis developed in the text shows moreover that, if the effect of the theory of antecedent-government and the theory of binding strongly overlap in the case of XO and adjunct reflexives, the behaviour of long-distance reflexives shows that neither of these theories can be reduced to the other, supporting the general idea of a modular organization of language.
Acknowledgements This work was supported by the MIT Center for Cognitive Science under a grant from the A. P. Sloan Foundation's particular program in Cognitive Science. Thanks for their help to L. Burzio, S. Bromberger, N. Chomsky, J. Frampton, S. Gutmann, K. Hale, R. Kayne, H. Lasnik, S. Iatridou, R. Larson, A. Hestvik, J. Higginbotham, E. Torrego, J. Uriagereka, L. Yafei and B. Zhiming, with whom I have been able to discuss several points related to the theory developed in this chapter. Thanks also to all the participants of the Groningen round table on long-distance anaphora, where part of this work was presented and discussed.
Notes l. Where the SSC is integrated in the notion of governing category (Chomsky (1981)) or in the notion of complete functional complex (Chomsky (1986a)). 2. As we shall see below, the distinction between mono-morphemic/non-mono-morphemic reflexives - which is reminiscent of the distinction between argumental and non-argumental reflexives of Pica (1985) - is to a certain extent misleading.
134
Pierre Pica
We shall see below that the right generalization is more complex and distinguishes adjunct reflexives and XO-reflexives from argument reflexives. Most mono-morphemic reflexives are arguments, although not all of them are, as illustrated by the case of the elitic se in French, which is an X O, as is shown in the text. 3. See, however, Herlofsky (1986), who shows that what has been considered subject orientation is, in many cases, the result of a third person orientation. 4. The formulation of (a-c) is actually oversimplified and should not hide the fact that movement at S-structure and movement at LF interact closely - as illustrated by the contrast between XO -clitics in French and in Italian, where this type of clitic adjoins to V at S-structure and moves together with V to Infl in French, while the same type of element can reach Infl at S-structure in Italian, if the analysis developed in Kayne (1989a) is on the right track. If Kayne's analysis is correct, it might indeed indicate that movement proceeds as far as possible at S-structure, until it is blocked, and might then proceed again at LF if no particular constraint blocks the operation at this new level of representation. This line of approach is supported by the fact that XP-clitics like sig do undergo a kind of cliticization (called object shift) at S-structure, as illustrated by the following contrast in Swedish, from Holmberg (1986): (i) att Sara uppfi:irde sig ilia that Sara behaved herself badly (ii) Varfiir uppfiirde sig Sara ilia? why behaved herself Sara badly? where we assume, following Holmberg, that V raises to C in (ii) and would like to propose that the reflexive adjoins to IP (after a first adjunction to VP), as illustrated in the representation (iii): (iii) [cp varfiir [c uppfiirde [sigi [IP Sara INFL [vp ei ilia]]]]]? Note that if the analysis developed in the text is on the right track, sig would have to be reconstructed in its base position to move up to Infl at LF - a movement which is ruled out at S-structure by the structure preserving constraint. 5. Note that the analysis developed in the text is not compatible with Borer's hypothesis according to which, in infinitives, Agr in C license a small pro (not PRO). Our analysis would lead us to expect such a pronoun always to be disjoint in reference from its controller in sentences like 'John wants to win', if 'John wants [[pro to win)l' were their grammatical representation. We could claim that the element Agr of an infinitive sentence is in general not able to license a pro (although inflected infinitives in Portuguese, whose subject enters into disjoint reference with the main subject, do - as pointed out to me by K. Hale), or alternatively, that non-inflected infinitives lack Agr entirely although they have an inflection node, as clearly indicated by the fact that they accept clitics. 6. Movement of Infl to C is also supported by the existence of overtly inflected complementizers (see Bennis & Haegeman (1984». 7. See however, for an alternative analysis, Pica (forthcoming) where it is suggested that the contrast between subject-oriented reflexives and non-subject-oriented reflexives might derive from the interaction of the theory of antecedent-government with general principles of economy.
Antecedent-government and binding: the case of long-distance re/lexivization 135 We also suggest that obligatory movement of INFL to C observed in the case of longdistance reflexivization could be linked to the syntax of reflexives itself, hereby indirectly suggesting that binding theory might be entirely reducible to antecedent-government, an obvious alternative to the analysis developed in this chapter. 8. The whole reflexive himselfbeing, of course, the argument with respect to a-theory. 9. Note that an intervening negation does not block movement of an adjunct reflexive at LF, as illustrated by the grammaticality of (i), which contrasts with (14) and (15) in the text: (i) hei did not admire himselfi We shaH tentatively assume that the negation raises at LF and does not induce any minimality effect with respect to the reflexive. We wiII come back to this point in Pica (forthcoming).
6 Binding in Polish Ewa Reinders-Machowska
1 Introduction Recent research shows that anaphoric elements pattern in at least two different ways: some of them obey a strict locality condition, and others do not. In the first case an anaphor requires an antecedent in a local domain and in the second case the binding domain is extended. Local domains and their extensions may vary from language to language. The goal of this chapter is to present a description of binding domains for Polish anaphors. Polish is a Slavonic language. The group of Slavonic languages has hardly been investigated in the light of the binding theory developed in Chomsky (1981). Some of the facts to be discussed will be unexpected from the perspective of current theory. I trust that the conditions under which they can be accommodated will shed light on the nature of parametrization. Section 2 of this chapter presents Polish binding facts. Section 2.1 gives some general background about Polish, and 2.2 about the Polish anaphoric/pronominal system. Anaphoric relations within simplex clauses are described in 2.3 and anaphoric relations across clause boundaries in 2.4 (clauses with tenseless complements) and 2.5 (clauses with participial complements). The NP as a binding domain will be discussed in 2.6. Section 3 of this chapter discusses the binding theory for Polish. In 3.1 binding domains for anaphors are investigated, and in 3.2 domains for pronominals. In the last section (3.3) it will be shown how the binding theory applies to Polish. 2 Polish binding facts 2.1 Some general remarks about Polish Before introducing the Polish binding facts, some remarks about Polish must be made. Polish is an SVO-Ianguage with a rather free word-order. It is a pro-drop
137
138
Ewa Reinders-Machowska
language, i.e. there is a tendency to omit subject pronouns, although with different degrees of optionality. Polish has a rich Case structure (seven Cases). All these Cases differ morphologically. Adjectives agree with nouns, assuming their Case, gender and number. There are no articles in Polish.
2.2 Polish pronominals and anaphors This chapter focusses on reflexives and reciprocals. Polish has two reflexive pronouns (using the traditional term): the personal pronoun siebie and the possessive pronoun sw6j. Neither form shows a morphological reflex of grammatical person, number or gender. The Polish reflexive possessive sw6j functions as the specifier of an NP. It has the full seven Cases-three genders paradigm of modifiers, completely agreeing with the noun it is construed with. The reflexive anaphor siebie shows distinctions for one number and one gender only, but it can assume several Case forms. Both siebie and sw6j are used for all three persons and both numbers. With the antecedent in the same clause, siebie is obligatory for all three persons; sw6j is obligatory for the third person (both singular and plural) and optional for the first and second persons. Instead a pronominal may be used. Polish has three reciprocals: jeden drugiego (one another), siebie and sw6j. The expression jeden drugiego is rarely used in Polish. In order to express reciprocity, usually, the reflexive pronouns siebie and sw6j are used.
2.3 Clause-internal binding of pronominals and anaphors Before discussing complex sentences, we will first have a look at a simple case. Consider the following sentences with the reflexive anaphors siebie and sw6j: (1)
(2)
Jani opowiada Piotrowij 0 sobiei/*;lnim*i/j Jani-NOM talks PiotrrDAT about selfi/-/him*i/j 'Jan talks to Piotr about himself/him' a. Aniai uderzyta Kasi~j swoj~i/4jej*i/j lalk~ Annai-NOM hit KasiarACC self'si/4her*i/j doll-INSTR 'Anna hit Kasia with her doll'
Bindinffin Polish b.
139 10"
NP
10' 10 ( +tense) Agr
VP
V'
~ V NP]
I
uderzyta
I
Kasi~j
NP Spec
I
N
I
swojl).iI*j lalkl). jej*ilj
The only possible antecedent for both anaphors siebie and sw6j is the subject of the clause. For example, swojq can be bound by Ania, but not by Kasi~. The subject can never be an antecedent for a pronominal in the same clause. But the non-subject pronouns can be coindexed with the object NP. Assuming a binary branching structure, in which the direct object forms a V' with the verb, these facts are not surprising. We can easily see that they are consistent with principles A and B of the binding theory if one assumes that binding relations require strict c-command rather than m-command. In that case the pronouns jejlnim are free and the anaphors sobie/ swojq bound within their governing category as required. Notice that, if this structure is correct, pronominals in PPs or equivalent structures must generally be free from subjects in Polish, but may be coindexed with objects. The binding theory leads us to expect that pronominals and anaphors are in complementary distribution. This is consistent with the facts observed. So far, we have discussed the reflexive use of both anaphors. But as mentioned before, the same elements can also have reciprocal meaning: (3)
ChtopcYi rozmawiali ze sobl).i /nimi*i boysi-NOM talked with self/each otheri /them*i 'The boys talked with themselves/each other'
This sentence is ambiguous - it has two readings: the reflexive one and the reciprocal one. In this case both readings of siebie have the same antecedent. Later on we will see that this need not be the case. Binding of reciprocals will be discussed in more detail in section 3 of this chapter. 2.4 Pronominals and anaphors in infinitival clauses In order to check whether the Tensed S condition (TSC) and/or the specified subject condition (SSC) hold in Polish, one must investigate complex sentences. Because all Polish clauses have subjects (lexical, pro, PRO), the SSC would predict
140
Ewa Reinders-Machowska
that it is never possible for an anaphor to be bound from outside its clause. We will see that this prediction is incorrect. This shows that the subject cannot be an opacity factor in Polish. In this section I will discuss binding out of finite, infinitival and participial complements. The last two complement types lack the feature (+tense). It is perhaps useful to point out that the marking (-tense) does not indicate the absence of a Tense slot. With Stowell (1981) we assume that infinitival and participial clauses are not absolutely tenseless. (- tense) just indicates the absence of an independent Tense. Within simplex clauses anaphors can be bound by subject NPs. Pronominals can be coreferential with non-subject NPs, but are not bound within their clause. Let us investigate what happens after embedding: (4)
a. Mariai wie, ze Aniaj opowiadata Kasik 0 Mariai-NOM knows that AnnarNOM told Kasiak-DAT about sobie*i/j/*k Inieji/*j/k self*i/j*k Iheri/*j/k 'Maria knows that Anna told Kasia about herself/her' b. 10"
NP~ 10
VP
(+tense) Agr
/\ V
CP
C~II"
/j
NP ------ --------------- II' 11--------(+tense) Agr
----------- VP
V'
V~NP Mariai
wie ze Aniaj
PP PNP
I I sobie*i/j/*k I
I opowiadata Kasik
0
nieji/*j/k The only antecedent for the anaphor in the above example is the subject of the minimal tensed clause in which the anaphor occurs. Anaphors cannot be bound outside tensed clauses. The subject of the matrix clause cannot be an antecedent
Binding in Polish
141
for them. But it can be an antecedent for a pronominal. Notice that the complementary character of the anaphoric/pronominal relations is preserved. The question arises whether an anaphor must be bound within the minimal tensed clause, or just simply within the minimal clause irrespective of Tense. Consider therefore sentence (5), which has an infinitival complement. (5)
a.
Jani kazal Piotrowij zbudowac dom dla siebiev/niego?iI*j Jani-nom ordered Piotrj PRO j to-build house for selfil/him?iI*j 'Jan ordered Piotr to build a house for him' b. 10" NP 10' 10 VP <+tense) Agr V' V NP/Y-C II"
o
~
11'
NP
I~VP
<-tense)~ V'
~
kazal Piotrowij
PP
/\
V
NPP
NP
I
I I
I
PROj zbudowac dom dla siebieilj niego?i;*j
When an anaphor occurs in an infinitival complement, its antecedent can be the subject of the same clause (as in the case of finite complements). But now also the subject of the matrix clause can be its antecedent. The S-node is not a barrier for the anaphoric relation. The subject is not a barrier either, as the embedded clause in sentence (5) has a PRO-subject, controlled by Piotrowi. That means that in Polish, anaphors need not be bound within their minimal S. Notice that the complementary distribution between pronominals and anaphors breaks down here, since Jan is a possible antecedent for both siebie and niego in the embedded clause. So, pronominals in infinitival complements cannot take the subject of the infinitival clause, but can take the subject of the matrix clause as their antecedent
142
Ewa Reinders-Machowska
(although this is not the preferred interpretation). For pronominals occurring in a Tensed embedded clause, the matrix is unquestionably a possible antecedent. The difference between (4) and (5) can be characterized as follows: In (4) the complementizer position C is filled, the inflection 11 has the feature <+tense) and there is agreement (Agr) between the subject NP and the verb. In (5) the complementizer C is empty, the inflection 11 has the feature <-tense) and there is no agreement between the subject NP (here it is PRO) and the verb of the embedded clause. Since infinitival clauses differ from finite clauses in both Tense and agreement, the absence of either of those could be the relevant factor. In the next section evidence will be discussed which narrows down the options.
2.S Pronomina1s and anaphors in participial clauses Besides finite and infinitival clauses Polish has also participial clauses. Before investigating their status as binding domains, some general properties of participles will be sketched. The Polish verb has two participles: Participium Activum Praesentis and Participium Passivum Praeteriti. They differ morphologically, and arise from different base forms of the verb (imperfective versus perfective). They are formed from the present and past tense stem respectively. Relevant properties of participles are the following. The categorial status of participles is [+ V], as they share properties of verbs [-N,+V] and adjectives [+N,+V] (Van Riemsdijk (1983)). Adjectival properties of participles are: agreement with the noun they are construed with (with respect to Case, number and gender) and their use as attributive modifiers. They are verbal in that they govern their objects in the same manner as verbs (objects get Case from participles), they exhibit voice differences (active versus passive), and are formed on the same basic form (the stem ofthe verb). Consider now the examples in (6) and (7): Aniai widziafa Kasi"'j bawi~c~ si", w Aniai-NOM saw Kasiaj PRO j playing-ACC, PART PRES ACT in swoim?*ilvjejil*j pokoju self' S?*iI( LOC/her*I*j-GEN room -LOC 'Anna saw Kasia playing in her room' (7) a. Jani kupit ksi~zk"'j mowl~q Jani-NOM bought book-ACC PRO j talking-ACC, PART PRES ACT o sobie*il4nimi about self*il4himi-LOC 'Jan bought a book talking about him' (6)
Binding in Polish
143
kupit ksi~zk~;
PRO; m6wi~c~
In the above sentences the only unquestionable antecedent for the anaphor is the subject of the same clause (here it is PRO, which is controlled by the object of the matrix clause). All native speakers whom I consulted agreed on this point. Some of them found the matrix subject of (6) a possible antecedent as well (but never a preferred one). However, sentence (7) is completely impossible with the subject of the matrix clause Jan as an antecedent for the anaphor (for pragmatic reasons the subject of the participial complement in this example is a bit odd as an antecedent as it is not animate, but this is irrelevant). With respect to anaphoric relations, participial clauses are similar to Tensed clauses apparently. The relevant opacity factor remains to be determined. The following considerations will help. In general one would not expect a difference in opacity between active and passive sentences. And between active and passive finite sentences there is in fact no difference in this respect. This is illustrated in (8). The anaphor must be bound by the grammatical subject and the pronominal cannot: (8)
Kasiaj zostaia uderzona przez Ani~; swoj~i/-/jej*i/; lalk~ Kasiaj-NOM was hit by Anna;-ACC selfi/-/her*i/; doll-INSTR 'Kasia was hit by Anna with her doll'
But there is what looks like a difference between binding out of passive and active participial constructions. Binding out of an active participial construction is impossible, as we saw in (7). Binding of siebie by Maria in (9) is unproblematic, a coreferential pronominal impossible:
144 (9)
Ewa Reinders-MacllOwska Mariai kupita ksi~zk~ napisan~ Mariai-NOM bought book-ACC PRO j written-ACC, PART. PAST PASS przez siebiejni~*i by selfjher*i 'Mary bought a book written by herself'
Although unexpected, this contrast makes sense, from the following perspective. Binding domains are determined on the basis of a limited set of opacity factors, such as Subject, Comp, Tense, and Agr (see Koster (1987: ch. 6) for discussion). From the facts of infinitival clauses we know already that in Polish the subject is not an opacity factor. Assuming that infinitival clauses are not formally tenseless, the facts indicate that in Polish, Tense could at best be an opacity factor conditionally. Let's assume now that Tense is in fact the opacity factor, and see what the conditions could be. If finite clauses are opaque, this implies that Tense is at least an opacity factor when it is independent. Now we have to determine what property finite clauses share with present/active participial clauses and not with either infinitival or past participial clauses. Curiously enough, finite clauses do indeed share their Tense with present participial clauses in a way in which they do not with the other types. It seems that the Tense of present participial clauses, like that of finite clauses, is always evaluated directly with respect to the time of the utterance. So, it must also be considered independent. The Tense of infinitival and past participial clauses is generally evaluated with respect to the Tense of some superordinate clause. Hence it is dependent. This indicates that in Polish a clause is opaque for binding if and only if it has an independent Tense. There is some further evidence for this claim in the contrast between (6) and (7). It was noted that a number of speakers allow binding of swoim by Ania in (6), but in (7) there was no such variation. There is also a structural contrast between (6) and (7). Kasi~ bawiqcq si~ w swoim pokoju in (6) can be construed not only as an NP + modifier, but also as a small clause complement to the perception verb widziafa. Under the construal as a small clause, the character is that of an ordinary dependent clause. The independent Tense interpretation becomes less prominent, and a dependent interpretation becomes possible, with the concomitant loss of opacity. In (7) there is no such ambiguity, hence the contrast. 2.6 Pronominals and anaphors in NPs In this section we will investigate the binding possibilities of Polish anaphors in NPs; specifically, subject and object NPs.
2.6.1 Anaphors in subject NPs In all sentences discussed so far, the subject of the sentence was the antecedent. As
Binding in Polish
145
the subject is the highest NP in the domain of Tense, it can never be an anaphor. Consequently, the reflexive anaphor siebie has no form for nominative Case (subjects in Polish mostly bear nominative). A nominative form of the anaphor sw6j exists but it can only be used with arbitrary reference. The relevant configuration is illustrated in (10). (10)
Przyjacielei widzieli, ze *swoje/ichi matki ogt~daty telewizj~ Friendsi saw that self's/theiri mothers-NOM looked TV 'Friends saw that *self's/their mothers watched the TV'
Examples (11) and (12) illustrate NP-internal binding. (12) shows that an anaphor can be bound by the subject of the NP as one would expect. The impossibility for sobie in (11) to be bound by Jan in the matrix sentence is more interesting. It either shows that null articles in Polish count as subjects for determining the governing domain, or else it is evidence that the i-within-i principle is irrelevant in Polish for determining the governing category. This is what one would expect if Tense rather than Agreement is the opacity factor. (11)
(12)
Jani wie, ze artykul 0 *sobie/nimi ... Jani-NOM knows that article-NOM about *self/himi 'Jan knows that an article about him .. .' a. Dysydencii mysleli, ze ichj artykuty 0 sobie*iI/nichil*j ... dissidentsi thought that theirj articles about self*iI/themil*j b.
AT MIR
sP~0& NP"
I ichj artykuty
P
NP
0
sobie*ilj nichil*j
I
I
In sentence (12) an anaphor in the subject NP is bound by the subject of this NP, ich. In sentence (11) there is no possible antecedent for an anaphor within the embedded sentence and this is the reason that this sentence is ungrammatical. Thus, not only the subject of the clause can be an antecedent of anaphors, but also the subject of an NP. 2.6.2 Anaphors in object NPs The subject of a clause is always a possible antecedent for an anaphor in object
146
Ewa Reinders-Machowska
position. But an object anaphor admits other NPs in the clause as antecedents. This is illustrated in (l3). (l3)
a.
Piotrj czytaf Jankaj artykuf 0 sobieil/nim?iI*j Piotrj-NOM read Janek'sj article about himselfil/him?iI*j 'Piotr read Janek's article about himself/him'
In (l3) the anaphor sobie has two possible antecedents: the subject of the object NP and the subject of the sentence. The pronominal cannot be bound within the object NP here but it can (although somewhat marginally) be bound by the subject of the sentence. From this case we can conclude that the subject is apparently never an opacity factor in Polish, neither in clauses nor in NPs. The irrelevance of the accessibility condition with respect to anaphors contained in the subject can now be unequivocally attributed to the fact that Tense, rather than SUBJECT, defines opacity. However, the pattern of binding is different if both the anaphor and its antecedent are in specifier position: (14)
a.
Piotrj czytaf artykuf Jankaj 0 Piortrj-NOM read article-ACC Janek'srGEN about swojejil?-/jego?*ilj pracy selfil-/his?*ilj work-LOC 'Piotr read Janek's article about his work'
In this sentence, coreference between the anaphor and the antecedent Janek is questionable. Both are in specifier position. It turns out to be the case that generally an anaphor in the specifier position of an NP does not take an antecedent that functions as the specifier of a higher NP. Only a few native speakers found such relations acceptable. The explanation for this phenomenon will be given in section 3 of this chapter.
3 Binding theory for Polish 3.1 Binding domains for anaphors In order to define binding domains in Polish we have to account for the following facts. Polish anaphors can always be bound by the subject of a clause (below I will mark such a subject 'S-subject', to distinguish it from the subject of an NP, which I will mark 'NP-subject'; if just any subject is meant, then I will simply use the term 'subject'). The NP-subject is a possible antecedent as well. There are therefore sentences in which there is more than one possible antecedent for an anaphor:
Binding in Polish (15)
a.
147
ChtOPCYi czytali dziewcz~tj wspomnienia 0 sobieilj bOySi read of-girlsj memories about selfil/each other*ilj Inich?iI*j Ithem?iI*j 'The boys read the girls' memories about them'
b.
10" NP---------
10
VP
<+tense) Agr
NP~
V Spec
N' N--------PP
p~p chtOPCYi
czytali
dziewcz~tj
wspomnienia
I
0
I
sobieilj
(recipr) sobie*ilj nich?iI*j In (15) the reflexive anaphor siebie has two possible antecedents: the S-subject chfopcy and the NP-subject dziewczqt. Notice, however, that this latitude is restricted to siebie under its reflexive reading. As noted earlier, the same element also allows a reciprocal reading. However, under that reading siebie has only one possible antecedent, namely the NP-subject dziewczqt. So, the reflexive anaphor siebie can be locally bound within the minimal NP or bound within the extended domain of the first S. The [Spec, NP] dziewczqt is a local antecedent and the NP chfopcy a remote antecedent. Under the reciprocal reading siebie is a local anaphor and must be bound within the first binding domain. To summarize, there are at least two binding domains in Polish: (a) the minimal NP properly containing the anaphor; (b) the minimal S with independent Tense (properly) containing the anaphor. Polish anaphors can be bound in a local domain and bound in an extended domain. An anaphor is bound in its local domain by the subject of its first binding domain (either a or b). An anaphor is bound in its extended domain by the subject of its second domain (necessarily b). Reflexives must be bound in their extended domain. Reciprocals must be bound within their local domain. The situation is somewhat different if an anaphor itself occurs in the specifier position and is thus an NP-subject. The reflexive and reciprocal siebie never occur in a specifier position. The only Polish anaphor which occurs in a specifier position
148
Ewa Reinders-Machowska
is the reflexive possessive sw6j. In sentences ofthis kind, the anaphor sw6j can only have one antecedent, namely the S-subject. Consider the structures of sentences (13) and (14), given as (13b) and (14b) below. (13)
b.
10"
/"-.
NP
10'
I~VP Agr
~ V/ ~NP
"N'
Spec
N ---------- PP \
Piotrj (14)
b.
czytal
Jankaj
f/ 'Nr
I artykul
SOblei/j nimi/*j
0
10"
NP~IO' I~VP
o~ / ~
Agr
V
NP 1
Spec~NI~ N
P --------
Piotrj
czytal
J ankaj
artykul
PP,,--NP 2
~
Spec
N
swojeji/?*j jego?*i/j
pracy
I
0
I
The difference between these sentences is as follows. In (14) the first N max (here NP 2) dominating the anaphor does not contain a possible antecedent. Such an antecedent is found in the second N max (here NP 1) dominating the anaphor. In (13) an antecedent was found within the first N max (NP) dominating the anaphor. This means that the Polish possessive anaphor can be bound by the NP-subject of the first N max or by the S-subject. The subjects of other dominating NPs are not possible antecedents. As such, the phenomenon is unexpected; and I have no explanation for it. But as I mentioned earlier, there are some native speakers who regard the last possibility as 'marginally possible': for them the NP-subjectJanka in (14) may be construed as the antecedent for the anaphor sw6j. This type of judgement should be the norm under current theories. Apparently, for many speakers some unknown factor interferes.
Binding in Polish
149
3.2 Binding domains for pronominals As has already been pointed out, the complementarity principle concerning anaphoric and pronominal distribution holds in Polish to some extent. It is valid in any case within the local binding domain. A pronominal must be free from the Ssubject, which is here the only antecedent for anaphors. An object pronominal can be coindexed with an NP within a PP, but this is expected since neither ccommands the other. The situation is different in cases of non-local binding. Sentence (15) illustrates this. The anaphor has two potential antecedents here: the first is the subject of the local domain, the second is the subject of the extended domain. The pronominal cannot be bound by the subject of the local domain, but it can be bound by the subject of the extended domain (although this option is not very good for me personally, other native speakers find it acceptable). The conclusion we can draw is given in (16): (16) A pronominal must be free within its local binding domain. So, pronominals are in complementary distribution with anaphors within the local binding domain, but there is only a tendency for this restriction to hold for the higher domain. 3.3 Extension of binding domains So far, two binding domains for anaphoric and pronominal binding relations in Polish have been discussed, namely the NP, primarily relevant for pronominal binding, and the Tensed S, relevant for both pronominal and anaphoric binding. In general there is complementary distribution between anaphors and pronominals. Under some conditions it breaks down, as we have seen. Not completely, however, since binding of the pronominal remains marginal. For example, the subject of S is a possible antecedent for both anaphors and pronominals within NP. The same holds true when there is binding into an infinitival clause; an anaphor can be licitly bound, a pronominal admits binding marginally. Taking all this into account I propose the following description of binding domains for Polish. There are two binding domains: the local binding domain and the extended binding domain. The local domain of a is defined as the minimal category dominating a (and a governor of a) which contains a SUBJECT. The subject of some category b is the most prominent nominal element contained in b. Following Huang (1982) we will assume that the head of an NP may count as the SUBJECT of that NP. The extended domain of a is defined as the minimal category dominating a (and a governor of a) which contains independent Tense. Note that a Tensed Scan qualify both as a local and as an extended domain for some a it contains.
150
Ewa Reinders-Machowska
We can now formulate the following conditions: (17)
(i) An anaphor must be bound in its extended domain. Oi) A reciprocal must be bound in its local domain. (18) Pronominals and anaphors are in complementary distribution within their local domain. Using this notion of complementarity does more justice to the variability in the judgements and the uncertainty of many speakers concerning the binding of pronominals in the extended domain. Thus the facts of Polish provide support for the idea, defended by Bouchard (1984), that binding condition B is not really an independent principle of grammar (see also Reinhart (1983a) and Koster (1987)). This account leaves unresolved the problem involving specifier-specifier binding noted above. We will leave this issue for further research. l
Notes 1. Cases as illustrated in (i) present an apparent problem for the complementarity thesis. (i)
Aniai nie mogta nasj odwiedzic ze wzgll;du na swoj~iJ*;ljeh chorobl; Annai not could USj to-visit because-of self'siJ-/heri illness 'Anna could not visit us because of her illness'
The subject of this sentence, Ania, is a possible antecedent for both the anaphor sw6j and the pronounjej. This contrasts with the pattern in (2). Notice, however, that the adverb in (i) is a sentence adverb. If one assumes that such a because-phrase can be attached ambiguously, either under S, or higher up in the structure, under S', the pattern follows. Lower attachment licenses the binding of the anaphor, higher attachment licenses coindexing between Subject and pronoun. Notice that this should not qualify as binding. Hence bound-variable readings are expected to be impossible. A detailed discussion of bound-variable readings would be beyond the scope of this chapter. I will therefore leave this for the future.
7 Anaphors in binary trees: an analysis of Czech reflexives Jindfich Toman
1 Descriptive preliminaries 1.1 Reflexive pronouns One of the prominent distributional properties of Czech reflexives is that the reflexive pronoun is generalized to all persons, i.e. it is not restricted to the third person only, as is the case in a number of languages, including German and French: (1) a.
German ich sehe mich 1 see me 'I see myself'
wir sehen uns we see us 'we see ourselves'
du siehst dich you see you 'you see yourself'
ihr seht euch you see you 'you see yourselves'
er sieht sich 'he sees himself'
sie sehen sich 'they see themselves'
b. French je me vois 1 me see 'I see myself' tu te vois you you see 'you see yourself'
nous nous voyons see we us 'we see ourselves' vous vous voyez you you see 'you see yourselves'
lSI
152 Jindrich Toman il se voit he himself sees 'he sees himself' c. Czech vidim se see-l ps REFL 'I see myself'
voient ils se they themselves see 'they see themselves'
vidime se see-l ps REFL 'we see ourselves'
vidiS se see-2ps REFL 'you see yourself'
vidite se see-2ps REFL 'you see yourselves'
vidi se see-3ps REFL 'he/she/it sees him/her/itself'
vidi se see-3ps REFL 'they see themselves'
(Reflexive pronouns are shown in boldface throughout this chapter.) The reflexive pronoun shown in (lc) is not a compound like the English myself or Dutch zichzelJ, although one could view it as morphologically complex, distinguishing the stem s- and the case endings, since the form se contrasts with the form si, the former being the accusative, the latter the dative, as illustrated by the following sentence exemplifying a dative reflexive: (2)
Karel si neveri Karl to-himself does-not-believe 'Karl does not believe himself'
A further important characteristic of reflexives, one that also applies to nonreflexive pronouns, is a clear distinction between weak and strong forms. Reflexives in (lc) and (2) are weak forms, i.e. clitics. As such they are restricted in distribution and may generally appear only in what is often referred to as the Wackernagel position. This is defined in Czech as the position following the first major constituent of the clause, hence: (3)
[Ten pan] se neholi that gentleman himself does-not-shave 'That gentleman does not shave himself' 1
Strong forms of reflexives are generally 'longer', they can bear emphasis and, moreover, with the exception of the nominative and vocative, they show a full range of cases, whereas gender and number distinctions are neutralized:
Anaphars in binary trees: an analysis a/Czech reflexives weak forms
(4)
nominative genitive dative accusative instrumental
Sl
se
153
strong forms
sebe sabe sebe sebau
1.2 Possessive reflexives Further descriptive notes should draw attention to the 'second half' of reflexivization in Czech, namely the presence of a reflexive possessive in the grammar of Czech. (As all possessives in Czech are strictly speaking adjectives, it is in fact more appropriate to speak about 'reflexive possessive adjectives'.) Like the reflexive pronoun, the possessive reflexive is generalized to all persons: because of the adjectival nature of possessives, case, number and gender distinctions are visible: (5) Navstivim svou tetulsve tety I'll-visit my-REFL aunt/my-REFL aunts 'I shall visit my aunt/aunts' Navstivis svou tetulsve tety you'll visit ... NavStivi svou tetulsve tety he'll visit ... Navstivime svou tetulsve tety we'll visit ... Navstivite svou tetulsve tety you'll visit ... Navstivi svou tetu/sve tety they'll visit ... An account of svu.j (m), sva (f), sve (n) which would be at least descriptively adequate is much more difficult than a description of the simple reflexive pronoun. This is because in a number of instances the possessive reflexive does not appear obligatorily, that is, a non-reflexive possessive is equally good in numerous, although not in all, cases: (6)
a.
VYi jste otnlvil vasi/svoui kocku? you have poisoned youlyour-REFL cat?
154 Jindfich Toman b.
Kareli otnlvil *jeho/svoui kocku Karl poisoned his/his-REFL cat
Facts of this type will not be further discussed in this chapter. 2 1.3 Some implicational generalizations This brief survey of Czech data and the few remarks on contrasts between Czech and other languages provide a basis for a set of implicational generalizations of the familiar kind: (7) First and second person reflexive imply third person reflexive. (8) Reflexive possessive implies reflexive pronouns. The first statement is descriptively consistent with cross-linguistic contrasts given in (1); the optional nature of reflexive possessives in the first and second persons is also covered by the first implicational statement. And, finally, the situation in languages such as Old English, which is traditionally described as having no reflexives at all, is vacuously consistent with the two implicational statements above because neither of the statements requires that special reflexive morphemes be among the set of morphemes of a particular language. The existence of the second implicational generalization would seem to suggest a major division in the system and might therefore imply that the two subsystems will have different properties. This expectation is borne out only in part. It will be seen in section 2.4 that reflexive possessives are bound in the same way as full reflexive pronouns. Nevertheless, the fact that the system of pronominal reflexives is more 'mechanical' and lacks the 'subtlety' (cf. note 2) of the possessive system would seem to qualify it as a core property of the grammar after all, whereas the more complex nature of possessive reftexivization and its relative cross-linguistic scarcity would seem to render it an extension beyond the core.
2 The question of the antecedent The binding domain for reflexives in Czech never extends beyond the domain of an inflected clause, the notion 'inflected clause' covering finite clauses: (9) a. Kareli vi, ze mu/*sii je spatne (indicative) Karl knows that to-hirnlto-himself is sick 'Karl knows that he is sick' b. *Kareli chce, aby sei Petr oholil (subjunctive) Karl wants that himself Peter share as well as infinitival clauses, i.e. clauses with non-finite inflection: 3
Anaphors in binary trees: an analysis of Czech reflexives (10)
a. b.
*Kareli Karl *Kareli Karl
ISS
mis nutil, oholit sei us forced to-shave himself mis nutil, oholit sebei us forced to shave himself
Placing examples such as (10) into comparative perspective, we see that Czech, a West Slavic language, differs not only from East Slavic Russian: (11)
Oteci poprosil menjai pobrit' sebjai/j me to-shave him/myself father asked
but also from Polish, otherwise a closely related West Slavic language: (12)
Mariai kazafa Piotrowii zbudowac dom dla siebieili Maria ordered Peter to-build house for her-REFLIhim-REFL 'Mary ordered Peter to build a house for her/him'
(Example from Reinders-Machowska, chapter 6.) As can be seen, both in Russian and in Polish, a full referential reflexive pronoun can take an antecedent outside an infinitival clause. Nothing of this order is possible in Czech. The use of a strong reflexive, cf. (lOb), does not make a longdistance reading possible either. 4 The fact that reflexivization is constrained to clauses does not however imply that the antecedent and the reflexive must be coarguments: (13)
a. Petri Peter b. Petri Peter
nepomyslil did-not-think nepomyslil did-not-think
[na about [na about
utoky [proti sobe;]] attacks against himself misledky [MoW [proti sobei]]] consequences of-attacks against himself
If the notion of coargument is defined as member of the same a-frame, then the reflexive and its antecedent clearly belong to different a-frames in these examples.
2.1 Full reflexives and small clauses In a more theory dependent perspective the question of the domain in which the reflexive pronoun is bound is closely related to the choice of the antecedent. To see the type of data relevant in this context, consider the following examples: (14) a. Kareli narovnal destickYi na sebeili Karl stacked plates on himself/themselves b. Kouzelniki zkfizil tycei pres sebeili magician crossed bars across himself/themselves
156 Jindfich Toman c.
Pan Novak postval sousedy proti sobei Mr Novak incited neighbours against him/themselves
These examples all allow of the reading on which the reflexive is bound by the subject, and, irrespective of pragmatic questions, they also all allow of readings on which the reflexive is bound by the direct object. It would thus seem that there is no particular constraint as far as the type of antecedent is concerned: both subject and object obviously represent possible choices. It will be argued in the following discussion that this is only a descriptive generalization and that a theoretically relevant formulation is close to: (15) Reflexives are subject oriented. The possibility that instances of object orientation of reflexives could be interpreted as instances of subject orientation was considered as early as in Lees & Klima (1963) (see also Koster (1985». A more explicit account however requires certain specific assumptions. In the following discussion we shall take the position that among these assumptions is the idea that in certain cases the verbal projection is constructed in such a way that the direct object and the PP-complement form a single projection rather than being placed in a 'layered' verb phrase in the sense of the earlier formulations of the X-bar theory. Such an approach has been argued for by R. Kayne, who, among other things, expounds the idea that the direct object and the PP complement can form a 'small clause' in a well-defined set of cases (Kayne (1981». Without going into details at this point, we note that Kayne's postulation of this kind of small clause ties in with certain core principles of grammar, in particular with the notion of c-command and with the closely related assumption of a (complete) binarity of phrase structure. Taking these ideas as a point of departure, we shall then be dealing with small clauses of the following kind: (16)
a. b. c.
. .. [destieky na sebe] . . . cf. (14a) cf. (14b) . .. [tyee pres sebe] . . . . .. [sousedy proti sobe] ... cf. (14c)
Clearly, these structures make it possible to sustain the claim formulated in (15) since the reflexives in these examples can now be seen as subject-oriented within the small clause. Although many points concerning the nature of small clauses remain open, such analysis is consonant with a number of semantic intuitions. For instance, it seems to be particularly natural with certain verbs involving causation. Thus, with verbs denoting acts of placing or transfer, such as in (14a), we claim it makes sense to
Anaphors in binary trees: an analysis oJCzech reflexives
157
speak of 'the plates being at some location' as a result of a type of causation involved in 'placing'. By the same token, such structures can also be viewed as structures in which the predication relation holds; see Hellan (1982) for a similar idea expressed in a somewhat different framework. It is important to realize that this approach places a bound on combining NPs and XPs to small clauses. Not every accusative NP forms a small clause with a PP. For instance, in: (17)
Jana navstivila Karla kwli sobe J ana visited Karl for-sake-of herself 'Jane visited Karl for her own benefit'
there is a clear intuition that Karla kvftli sobe cannot be interpreted in a way comparable to the above instances of small clauses, and that consequently no small clause can be involved in (17). Interestingly, this intuitive argument is consonant with the fact that this clause has only one reading: the reflexive can only be bound by Jana. In other words, when an NP-PP sequence (or, in general, a sequence of the form NP-XP) cannot be analysed as a small clause, the XP containing the anaphor is obviously outside the domain of the object NP, and, not being ccommanded by it, it cannot take the object NP as its antecedent. Finally, it is important to note that, in explaining the above judgement, one cannot appeal to the idea that certain adverbials are closer to the verb than other adverbials for reasons of subcategorization. Note, for instance, that (14b) involves an adverbial phrase not subcategorized by the verb. The same is the case in: (18)
Janai zavrazdila Karlaj ve svemil*j byte Jana murdered Karl in her-REFL apartment
yet only (14b) has two readings. On the other hand, the following example involves a three place verb requiring an oblique object, besides an accusative object: (19)
UfadYi zbavily novimifej svYchil*j nephitel authorities deprived journalists of-their-REFL enemies
There is only one reading here, too, hence sub categorization of an argument phrase does not automatically result in a small clause structure. An interesting issue is whether small clauses of the above type can have 'nonaccusative subjects', i.e. whether structures V [NP-dat XP], or even V [PP XPj, can ever be considered in this context. As far as data involving reflexives are concerned, the answer seems to be no:
158
(20)
Jindfich Toman
a. b.
Jana; Jana J ana; Jana
pomahala Karlovij ve svemi/*j byte helped Karl dat in her-REFLIhis-REFL apartment cekala na Karlaj ve svemi/*j byte waited for Karl in her-REFLIhis-REFL apartment
Semantic intuitions suggest that the postverbal phrases cannot possibly form a small clause in these examples, hence the ungrammaticality of the reading indicated by the j-index. Of course, (20a, b) might merely be regarded as wrongly constructed examples. They thus receive their value only in conjunction with the fact that examples with anaphors relating to 'dative subjects' cannot be constructed. Small clauses with dative subject NPs simply do not seem to exist. Speculating about a principled account, one might regard the whole issue together with the question of why Exceptional Case Marking involves only accusative, i.e. a structural case. It seems that the answer may be found in the area of interaction of Case theory and a-theory. We shall assume that in small clauses Case marking and a-assignment proceed as for instance in English believe-structures: the a-role of the subject of the embedded clause is assigned within the embedded clause and Case is assigned by the verb: (21)
believe-structures: V [NP XP] small clause: V [NP XP]
This seems to be a necessary conclusion for small clauses in any case, since the XPs involved are incapable of assigning Case, and other potential Case assigners, such as Infl (or Agr), are absent. But if a a-role is assigned to some nominal without Case being assigned to this nominal, there remain only structural cases to be assigned because these are the only cases not intrinsically connected to particular a-roles. In other words, given a set of a-neutral cases and a set of cases intrinsically linked to a a-role, only a-neutral cases can be assigned to an independently a-marked nominal; otherwise the a-criterion is violated. In concrete terms, the candidates for the small clause subjects are nominative, accusative and adnominal genitive. The choice of the accusative appears to be obvious in the given structure. Returning to the main topic, we see that, whereas the object-oriented readings in (14) are accounted for, the subject-oriented readings have now become somewhat mysterious: how are they possible at all? Should not the direct object, that is, the subject of the small clause, block them? An approximate answer to the above question can be presented in the following manner. Firstly, statement (15) will be reformulated as: (22) Reflexives are SUBJECT oriented.
Anaphors in binary trees; an analysis o/Czech reflexives
159
The notion of an 'accessible' SUBJECT generalizes the traditional subject. It comprises a variety of 'prominent nominal elements' which can define a governing category (see chapter 1 for discussion). Given this notion, Czech data seem to suggest that there are several ways of binding a reflexive. One way is to bind it by a nominal that is an accessible SUBJECT for independent reasons, cf. a typical case of a nominative subject of a tensed clause. The other way is to coindex the anaphor with a nominal c-commanding the anaphor: (23)
[NP i NPanaphor,i] [NPi [P NP anaphor, i]]
If the c-commanding nominal is coindexed with a reflexive anaphor, it becomes an accessible SUBJECT and consequently converts the small clause to a governing category; if, on the other hand, this nominal is not coindexed with the anaphor, no governing category is created, no accessible SUBJECT results, and the anaphor must look for an appropriate antecedent elsewhere to meet the basic condition on its distribution. This reasoning is essentially a variant of the current theory, the difference consisting in emphasizing the fact that an accessible SUBJECT may be a matter of a free choice in structures in which other factors do not impose SUBJECT-hood for independent reasons. Such factors, in particular the presence of Agr, are absent in the type of small clauses discussed so far. The main point thus is that a nominal not chosen as antecedent, that is, a nominal whose potential for acting as an accessible SUBJECT was not 'activated', does not create a governing category. This then explains why subject-oriented readings are possible in the presence of direct objects in clauses such as those in (14). It is intriguing to speculate whether the situation described so far is paralleled by data of the following type: (24)
a. b.
Janai zahodila KarloVYi basne 0 sobei/i Jana threw-away Karl's poems about herself/himself Predsedai odsoudil vase mlmitky proti sobei/i chairman condemned your attacks against himself/yourself
Disregarding pragmatic questions again, we take these judgements as suggesting that the anaphor has two options: one antecedent being 'close' (like the SUBJECT in a small clause), the other being 'distant' (like the nominative subject in (14).) The main point again consists in the optional nature of the close antecedent: as in small clause structures, the close antecedent need not function as antecedent. An account of (24) that would parallel the account of reflexives in small clauses would say that in NPs such as these:
160
(25)
Jindfich Toman
a. b.
KarloVYj basne 0 sobej Karl-ADJ poems about himself vasej namitky proti sobej your objections against yourself
there is nothing that forces the forms Karel and vas to act as SUBJECTS. They thus remain mere 'prominent nominals' with a SUBJECT potential unless coindexing with the anaphor actually converts them into accessible SUBJECTS. The plausibility of this account seems to be based on the fact that unlike in English, these 'prominent nominals' are not in fact nouns. Although they have a referential potential comparable with nominals, they are morphologically and positionally adjectives. This means, among other things, that they do not receive Case, either by agreement or on structural grounds. Pursuing the above line of reasoning, we may then suggest that the difference between English and Czech, namely the contrast between: (26) a. *She hates Bill's remarks about herself b. Janaj nenavidi Karlovy anekdoty 0 sobej Jana hates Karl's anecdotes about herself ('Karl's' in (26b) corresponds to an adjective, not to an NP.) derives precisely from the fact that in English an NP is involved in the det-position which must be Case marked and hence must act as accessible SUBJECT. In Czech, on the other hand, no factor comparable with Case assignment forces possessive (referential) adjectives to act as accessible SUBJECT. 6 2.2 Clitic reflexives and small clauses A set of more complicated data is shown below: (27)
a.
b. c. d. (28)
a.
Sultan darovallnabidl otroka vezirovi sultan donated/offered slave to-vezier 'The sultan donated/offered the slave to the vezier' Sultanj darovallnabidl otrokaj sobei/j sultan donated/offered slave to-himself Sultanj sij darovallnabidl otroka sultan to-himself donated/offered slave *Sultan sij darovallnabidl otrokaj sultan to-self donated/offered slave Hrabe pronajal sluhy biskupovi count rented servants to-bishop 'The count rented servants to the bishop'
Anaphars in binary trees: an analysis a/Czech reflexives
161
b.
Hrabei pronajal sluhYi sobei/j count rented servants to-himself c. Hrabei Sii pronajal sluhy count to-himself rented servants pronajal sluhYi d. *Hrabe sii count to-themselves rented servants In (27) and (28), the (b) sentences show that a full form of the reflexive pronoun can have both subject and object antecedents. Examples with clitic reflexives contrast with the (b) sentences, however: whereas the (c) sentences show that the clitic reflexive can be subject oriented, the (d) examples indicate that, in contrast to full reflexives, a clitic reflexive cannot be object oriented. In other words, the nominative subject is the only choice with clitic reflexives, although a small clause structure can be assumed in the (d) sentences too, once such a structure is assumed elsewhere. 7 A similar situation has been noted in Dutch (Koster (1985)) and in Italian (Rizzi 1986b )), and attempts at an explanation have been advanced. Rizzi has proposed to account for this type of distribution of clitic reflexives by recourse to the theory of binding, the basic idea being that in Italian examples such as: (29)
*Sii affidero Giannii (I) to himself will entrust Gianni (Rizzi's (56b))
the clitic and the antecedent Gianni are seen as c-commanding each other (symmetric c-command), whereas in parallel examples with a full (emphatic) reflexive form this is not the case: (30) Affidero Giannii a se stessOi (Rizzi's (56a)) The assumption is that in (30) the antecedent c-commands the anaphor, but the anaphor does not c-command the antecedent. The c-command relation is thus asymmetric. If the c-command relation employed for the statement of binding conditions is specified in terms of asymmetric rather than symmetric c-command, the contrast follows. The force of the account consists in the claim that the notion of 'asymmetric c-command' need not be stipulated for the antecedent-anaphor relation. As Rizzi argues, asymmetry is independently imposed by the principles of binding. The contrast, Rizzi concludes, is ultimately deducible from the core principles of grammar. There is unfortunately no obvious way in which to adapt this account for Czech. Assuming that in the Italian examples the asymmetry results from the fact that the
162
Jindfich Toman
anaphor is in a PP, hence necessarily lower than the antecedent, we see that the Czech cases are structurally different. The anaphor is not in a PP; it forms an NP with oblique Case instead. It thus follows that other ways of describing the situation in Czech must be considered. This, no doubt, is unfortunate since the two languages display what one might plausibly believe to be the same phenomenon - a unified account would clearly be preferable. In attempting to find an answer, we shall assume small clause structures again, that is, structures such as: (31)
... [otroka si] ...
This is parallel to [otroka sobe], that is, to structures with full reflexives. As far as the non-clitic form is concerned, the relevant points have been discussed in the previous section: both a local and a non-local choice of the antecedent is possible. In the latter case, the accusative nominal does not act as an accessible SUBJECT and consequently does not create a governing category. The same antecedent choices should be available to clitic reflexives, and, indeed, it is hard to think in which way the theory of binding could make forms such as soM and si different. It can thus be concluded that the restriction on antecedents which is actually observed with si follows from that part of the theory which regulates the behaviour of clitics, rather than from the theory of binding. Let us assume that at least one of the following statements is true: (32) (33)
A clitic must be locally supported by Compo A clitic must be locally supported by Inf!.
If the first condition is true, its instantiation in Czech will result in right-adjunction of clitics to Compo If the second condition is true, the instantiation in Czech will require left-adjunction to Inflmax. Both instantiations amount to the 'Wackernagel effect' in terms of string adjacency, but each involves a different constituent structure. A decision between the two hypotheses does not seem to be crucial here. Assuming that small clauses of the kind discussed so far have neither Infl nor Comp, a clitic starting (or, in a representational framework, linked to a position) in a small clause must seek the next Comp or Infl available, that is, move out of the small clause, in order to satisfy conditions on clitic distribution. It seems consistent with the discussion in the previous section to say that this is not possible if the antecedent is chosen locally, i.e. within the small clause, because then the antecedent functions as an accessible SUBJECT and creates an opaque domain. One of the consequences, then, is that clidc placement outside the small clause is blocked. On the other hand, if the long-distance choice of the accessible SUBJECT is made,
Anaphors in binary trees: an analysis of Czech reflexives
163
no factor creating opacity is present and the clitic is free to move to the 'position of support'. It seems that the local choice of the antecedent, that is, the creation of an accessible SUBJECT by electing it as an antecedent, is the only case in which the opacity factor is effective. This is particularly important because all kinds of clitics can otherwise freely leave the small clause in the absence of an accessible SUBJECT: (34)
a. b. c. d.
dal [sto korun ei] Karel mii Karl to-me gave hundred crowns dal [ei ej] jej Karel mii to-me them gave Karl Karel jei dal [ei Petrovi] Karl them gave to Peter zaprodal [ei spatnym ideahlm] Karel sei to-wrong ideals Karl himself sold 'Karl subscribed to wrong ideals'
It thus seems that the relevant idea is that the reflexive clitic must be properly supported, a condition which cannot be met if the small clause becomes an opaque domain. We note in passing that this approach accounts for the Italian data once the same premises are made as here. Finally, as far as 'local support' is concerned, we follow the interpretation of clitics as zero-bar projections in terms of X-bar theory, that is, regard clitics as a kind of 'floating (phrasal) affixes' in terms of a word syntax (such as proposed in Toman (1983), for instance). Given this, a link between clitic placement and the theory of movement proposed in Chomsky's Barriers (Chomsky (1986b)) can finally be established: clitic movement is a movement of zero-bar segments into zero-bar positions on Infl (or Comp); see also Pica (1987) for a more detailed explication of the same point. 2.3 Observations on inflected small clauses There are data which show certain interesting complications as well as independent support for the approach followed here. Note that rather simple small clause structures have been considered so far. A more complex situation arises in small clauses arguably based on adjectives such as: (35)
Studenti nasli [profesora opileho] students found professor drunk
One particular property of this kind of small clause is to be noted, namely the
164 Jindfich Toman fact that the accusative NP and the adjective agree in Case, gender and number. Following Kayne (1989b), we shall interpret this fact as pointing to the presence of the Infl-node, or, more adequately, of Agradi' in these 'inflected small clauses'. Given this, we then must discuss whether by establishing the agreement relation in the small clause an opaque domain is created. In fact, one should expect it, and, indeed, this expectation is borne out: an indirect clitic object dependent on the adjective cannot leave the inflected small clause: (36)
*Tehdy mUi nasli [onoho filosofa neverneho ei] then to-him they-found that philosopher unfaithful
where mu, glossed as 'to-him', may be taken as referring to a masculine dative NP, say, idealu pravdy 'to the ideal of truth'. At the same time, however, the clitic cannot be placed within the small clause either, there thus being no grammatical output: (37)
a. b. c.
*Tehdy then *Tehdy then *Tehdy then
nasli they-found nasli they-found nasli they-found
[onoho filosofa mUi neverneho e;] that philosopher to-him unfaithful [onoho filosofa neverneho mUi] that philosopher unfaithful to-him [mui onoho filosofa neverneho ei] to-him that philosopher unfaithful
We note that it cannot simply be the accusative Case marking that creates opacity but the agreement relation; recall that in non-inflected small clauses of the type shown in (16), clitic placement outside the small clause was unimpeded. Given the blocking nature of Agradi' a grammatical output with a clitic reflexive taking a local antecedent, that is, the accusative NP, will of course not be expected in the inflected small clause: (38)
*Nasli Sii [filosofai neverneho e;] they-found to-himself philosopher unfaithful 'They found the philosopher unfaithful to himself'
The final question, then, is whether the agreement relation also functions as blocking the non-local choice of antecedent for full reflexives. Although the relevant examples are not stylistically satisfactory, they are acceptable and show that a non-local antecedent can be chosen in the presence of agreement and a referential expression: (39)
a.
Studentii nasli [profesora neverneho sobei] students found professor unfaithful to themselves
Anaphors in binary trees: an analysis of Czech reflexives b.
165
uCinila [komisi na sobej nezavislou] Vhidaj government made commission of-itself independent
On the whole, a familiar pattern repeats: a full anaphor can have a long-distance reading despite there being a 'prominent nominal' closer to the anaphor. This finding, then, means that the opacity factor created by Agradj only holds for clitic placement, not for anaphoric relations. Although it is not quite clear why this should be so, it is a fact. Since wh-movement has a number of properties in common with clitic movement, it might then be expected that wh-movement will be equally impeded in structures parallel to (36). Indeed, this is the case: (40)
a. b.
*?To this *?To this
je is je is
princip, kteremu nasli toho filosofa neverneho principle to-which they-found that philosopher unfaithful president, na kterem soudce uCinil komisi nezavislou president of-which judge made commission independent
It thus seems that there is a restriction on movements out of inflected clauses and that this restriction does not coincide with conditions on anaphora. 2.4 Possessive reflexives As indicated above, in addition to primary reflexives Czech also has a system of possessive reflexives. They can be distributed in NPs fulfilling a variety of clausal functions: (41)
a. b. c. d.
kocku] Vide! [svou he-saw his-REFL cat Nevefil [SyYffi oCim]dat he-didn't-believe his-REFL eyes Dotkl se [sveho ucha]gen he-touched himself of-his-POSS ear Zatrasl [svou penezenkou]jnstr he-shook with-his-POSS purse
Smj cannot however appear in the subject:
(42)
*Svilj byt je nejlepsi POSS-REFL apartment is best [Attempted reading: 'One's own apartment is the best']
In this latter point Czech again sharply differs from Polish (example adapted from Reinders-Machowska, chapter 6):
166
(43)
Jindfich Toman
Sw6j dom jest zawsze najmilszy one's-REFL house is always dearest 'One's own house is always the dearest'
and Russian: (44)
[Svoja komnata]nom lucse vsech one's-REFL room best of-all 'One's own room is best of all (rooms),
Consider further Russian examples: (45)
(46)
U kazdogoi bylasgfem [SVOjai tocka zrenia]nom,Sgfem to everybody was his-REFL point of-view 'Everybody had his own point of view' V Sibiri [svoi porjadki]nom in Siberia REFL-POSS orders 'In Siberia, there are rules of their own kind'
In all these examples, the Russian counterpart of the Czech possessive reflexive appears in the subject phrase, nevertheless the sentences are well formed. 8 Returning to Czech, obviously a language requiring a syntactic antecedent for the possessive reflexive, we now show that the antecedent of svdj cannot be outside an inflected clause: (47)
a.
b.
*Kareli mi rekl, ze mu Jana vzala jeho/*svtiji revolver Karl to-me said that to-him Jana took his/himself's gun 'Karl told me that Jana took his gun from him' si SVUJi revolver *Kareli ji donutil, vzit Karl her forced to-take to-her his-REFL gun 'Karl forced her to take his-REFL gun from him'
These examples parallel (9) and (10). The following examples further show that the antecedent can be either a surface subject or a surface direct object: (48)
Hrabei poslal hrabenkuj ke srymi/j rodicum count sent countess to his-REFLIher-REFL parents
Example (48) parallels (14). Finally, the following examples show that the depth of embedding of the NP hosting the reflexive possessive is arbitrary:
Anaphors in binary trees: an analysis of Czech reflexives (49)
167
a.
Karel videl [kopii [sveho obrazu]] Karl saw copy of-his-REFL picture b. Karel vide! [ram [kopie [sveho obrazu]]] Karl saw frame of-copy of-his-REFL picture c. Karel videl [obal [ramu [kopie [sveho obrazu]]]] Karl saw wrapping of-frame of-copy of-his-REFL picture d. Karel videl [dodad list [obalu [ramu [kopie Karl saw delivery slip of-wrapping of-frame of-copy [sveho obrazu]]]]] of-his-REFL picture
These examples will be treated like those in (13). As was observed in connection with examples such as (24), possessives occurring in NPs do not necessarily act as SUBJECTS and the long-distance reading of the anaphor is often the more prominent one. (50)
a. b.
Karel vide! [Petrovu kopii [sveho obrazu]] Karl saw Peter's copy of-his-REFL picture Kareli nesnasel [PetroVYj 6dy rna svehoi/j uCitele]] Karl did-not-stand Peter's odes on his-REFL teacher
Our judgement is that in (50a) the long-distance reading is quite dominant, whereas in (SOb) the prominency evaluation vacillates, but this is comparable with what has been said in connection with (24). Finally, the following examples parallel local and long-distance readings observed with full reflexives in inflected small clauses: (51)
a.
Studentii nasli [profesora neverneho SvYmi idealU.m] students found professor unfaithful to-their-REFL ideals b. Vladai uCinila [komisij nezavislou na svemi/j programu] government made commission independent of its-REFL programme
Again, the question about the prominence of certain readings arises in a comparable manner as with plain reflexives. It would thus seem that apart from questions relating to the choice between a possessive and reflexive possessive mentioned in connection with Kuno's 'grammar of empathy' (cf. (6) and note 2), reflexive possessives do not introduce any particularly different aspects into the description. This strongly suggests that the domain of reflexive anaphora in Czech is remarkably coherent - reflexive possessives are essentially a further strong reflexive form, the basic split in the system being formed by the presence of clitic reflexives.
168
Jindfich Toman
3 A summary of results Table 7.1 summarizes the basic configurations. Table 7.1 full reflexive
clitic reflexive Type of binding
non-local non-local non-local non-local Intervener (potential SUBJ) local not present present local not present present inflected causes small clauses inflected small clauses picture nouns
+ + + +
dna dna dna
+
+ + + +
dna
dna dna dna dna
+ dna
dna = does not apply.
It can be seen that, with the exception of inflected clauses, a full reflexive anaphor can always be related to a non-local antecedent across an intervening referential expression (NP, possessive adjective). In other words, an anaphor can be subject oriented in the presence of a direct object (small clauses of both types) and can take an antecedent outside an NP in the presence of a possessive. We regard this as a possible situation because in these instances, intervening referential expressions do not exercise their SUBJECT potential. As far as clitic anaphors are concerned, local antecedents are never possible in a small clause. We have interpreted this not as the inability of the clitic anaphor to be coindexed with a local antecedent but rather as a result of failed clitic placement. As far as the non-local option is concerned, clitics are on a par with non-clitics in inflected clauses, a situation attributable to the presence of sentential Infl. It would seem attractive to generalize this and to say that the presence of Infladj (that is, Agradj) is the reason why a clitic reflexive cannot take a non-local antecedent in inflected small clauses. This conclusion is however not possible - a non-clitic reflexive in an inflected small clause is able to take a non-local antecedent, all things remaining equal. The conclusion is that also in this case, the proper placement of the clitic is frustrated. It thus appears that the set of grammaticality judgements discussed in this chapter can be seen as resulting from the interaction of a small number of independent principles. The first group of these principles regulates the question of the antecedent choice by providing the SUBJECT notion and thus defining the governing category in particular cases. It is clearly visible that neither full reflexives,
Anaphors in binary trees; an analysis of Czech reflexives
169
nor ditic reflexives, nor possessive reflexives can be seen as inherently specialized for local or long-distance antecedents. It is the principles defining governing categories which determine the range of these anaphors. The other major principle, which interacts with the above principles, governs the distribution of ditics: (52) Clitics must be properly supported (by Comp or Inflverbal). Assuming this, those anaphors which are ditics must simply meet additional conditions on their distribution. All 'irregularities' in the antecedent choice which were encountered with ditic reflexives are thus viewed as violations of this requirement. Finally, we recall that the adoption of the small dause hypothesis makes it possible to maintain that: (53) A reflexive anaphor is SUBJECT oriented. As far as problems of detail are concerned, it appears that a 'prominent nominal', that is, an NP or a possessive adjective (the latter essentially being a referential expression), does not automatically count as an accessible SUBJECT. It seems that they function as SUBJECTS in two cases: if elected as local antecedents and if 'forced' to be SUBJECTS for an independent reason, essentially by entering an agreement relation. This finding is helpful in sharpening the notion accessible SUBJECT.
Notes 1. See Toman (1986) and the literature quoted therein for further details of c1iticization in Czech. 2. Literature on this phenomenon includes Danes & Hausenblas (1962); see also descriptions of a comparable phenomenon in Russian, studied by Yokoyama (1980) within the framework of Susumu Kuno's 'grammar of empathy' (see also Kuno (1987». 3. Instances of antecedents appearing apparently outside the clause, as in infinitival clauses.: (i)
sei Karel fillSi nutil oholit Karl us forced to-shave ourselves
are assumed to involve a syntactic antecedent internal to the infinitival clause at its abstract level(s) of representation. 4. As the situation in Russian is in many ways much more complex than in Czech, we shall not even attempt to outline a comparative description here; cf. specialized studies such as Ruzicka (1973) and Timberlake (1979). One can merely note that the reflexive morpheme appearing in verbal inflection in Russian cannot be long-distance bound: (ii)
a. *Oteci poprosil menja pobrit'sja; (cf. (ll»
170
Jindfich Toman b.
Otec poprosil menjai pobrit'sjai Father asked me to-shave myselfl*himself
5. It is occasionally said that the reflexives in these cases are reciprocals. Our impression is that it is more adequate to say that a regular reflexive pronoun in Czech may have a reciprocal reading rather than singling out a reciprocal subclass. Reciprocal reading is simply a function of the plurality of the antecedent. In cases in which the antecedent is singular, no reciprocal reading is available in sentences otherwise completely parallel to (14c), cf.: (iii)
Janai hOj postvala proti sobei/j J ana him incited against herselflhimself
6. There are several problems associated with examples such as (24). One unexplained problem is that in a number of instances it is difficult to obtain a natural reading of the anaphor for any choice of antecedent, close or distant. This problem will not be discussed here. A more interesting point in our context is the fact that the long-distance reading is often more prominent than the local reading. One of the reasons might be the fact that adjectives are involved, not NPs. Yet even if this were the relevant fact, the way in which a principled account of the 'weakness' of these possessive forms could explicitly be derived remains not quite clear. 7. Clearly, in a more extensive description, such examples may require additional comments since they often do not sound particularly natural. Pragmatic considerations are the major reason: acts of exchange such as denoted here are evidently not very common. Even so, judgements are reasonably sharp: clitic reflexives seem to be able to take only subject antecedents. 8. It should be noted, however, that there are certain restrictions on the type exemplified by (44) and (46). In general, these sentences have a semi-proverbial, or generic, flavour. It is not possible to say: Svoja komnata trebuet remonta 'One's own room needs a repair'. However, no comparable restrictions seem to exist in cases such as (45).
8 Latin long-distance anaphora Elena Benedicto
1 Introduction In this chapter, I would like to discuss some aspects of the phenomenon known as the anaphoric long-distance strategy in Classical Latin. Latin third person reflexive se has no specific features of gender and number, and lacks nominative Case.1 It behaves both as a 'strict' anaphor following principle A of the binding theory and as a 'long' anaphor, that is, with its antecedent outside its minimal clause. The aim of this chapter is twofold. First of all, I would like to determine by what means and under which conditions the minimal binding domain of an anaphor can be enlarged. Second, I would like to discuss what kind ofNP can be an appropriate antecedent for a long-distance anaphor.
2 Extending the domains
2.1 The data observed in Latin permit us to establish a first characterization of longdistance (LD) se. 2 Consider examples 0-3). (Here and throughout the chapter anaphors and their antecedents are indicated by boldface.) (1)
Ciceroj effecerat [s' ut Quintus Curius consilia COMP Quintus Curius-NOM designs-ACC Cicero-NOM had achieved Catilinae sibij proderet] Catilina-GEN REFL-DAT reveal-SUBJ 'Cicero had induced Quintus Curius to reveal Catiline's designs to him' (Sall., Cat., 26.3) (2) Ariovistusj exercitu suo praesente conclamauit [quid Ariovistus-NOM army-ABL his present-ABL exclaimed why
171
172
Elena Benedicto
ad sej uenirent] to REFL-ACC came-SUBJ 'Ariovistus exclaimed in the presence of his army, why they were coming to him' (Caes., B.G., 1.47.6) (3) a. Perfugaj el est pollicitus, [[s' si praemium if a reward-ACC a deserter-NOM him-DAT promised sibi proposuisset] sej eum ueneno necaturum]] REFL-DAT assure-SUBJ REFL-ACC him-ACC poison-ABL to kill-INF 'A deserter promised to him, if (Fabricius) would assure him of a reward, to kill him with poison' (Cic., Off., 3.86) b. Camillusi mihi scripsit [te secumi locutum] Camillus-NOM me-DAT wrote you-ACC REFL-ABL-with talk-INF 'Camillus wrote to me that you had talked to him' (Cic., Au., 11.23.1) In these examples, we find three different types of sentences (an ut-clause, an indirect question, and an infinitival clause, respectively) which are subordinate complement clauses subcategorized by the matrix verb. 3 The verb in these complements has a remarkable property: either it is in the subjunctive mood (cf. (1), (2», or it is an infinitive (cf. (3». In the latter case, the anaphor can occur in subject position (3a), due to the existence of the Latin construction (Accusatiuus cum Infinitiuo (AcI» that allows for the presence of a lexical subject in accusative Case. On the basis of these facts, we can formulate a first approximation of the LDstrategy in Latin based on the dynasty concept formulated by Koster (1987: ch. 6): (4) A dynasty is a chain of governors such that each governor governs the minimal domain containing the next governor. If we now look at examples (1-3), we can see that effecerat, the matrix verb in (1), governs the minimal domain containing the next governor, that is S', the subordinate subjunctive clause in which the anaphor appears. So a dynasty can be constructed of the two clauses, and the anaphor can find its way up to the appropriate antecedent. The same process takes place between the matrix verbs in (2), conclamauit, (3a), est pollicitus, and (3b), scripsit, and the corresponding subordinate clauses. So far, we have been talking mainly about complement clauses depending on socalled verbs of saying. The analysis is, however, also valid for verbs taking a complement clause headed by the complementizer quod: (5) Decima legioj... el gratias egit [s' quod de sei the tenth legion-NOM him-DAT thanked COMP about REFL-ABL
Latin long-distance anapJlOra
173
optimum iudicium fecisset] excellent opinion-ACC formed-SUBJ 'The tenth legion expressed thanks to him for the excellent opinion he had formed of itself' (Caes., B.G., 1.41.2) In (5), the verbal compound gratias agere governs the minimal domain containing the next governor, namely S'. As before, a dynasty can be formed, and the anaphor finds an antecedent in the appropriate binding domain. 4
2.2 It is well known that an anaphoric element in an adjunct clause cannot be coreferential with the subject of its immediate matrix clause. 5 This can be explained by the fact that, since the adjunct clause is not governed by the immediate matrix verb, a dynasty cannot be formed: (6)
Ligariusi in prouincia pacatlSSlma ita se Ligarius-NOM in province-ABL very-peaceable-ABL so REFL-ACC gessit [s' ut eii pacem esse expediret] bore COMP him-DAT peace-ACC be-INF be-convenient-SUBJ 'Ligarius ... in an utterly peaceable province so bore himself that peace was his highest interest' (Cic., Lig., 4)
In (6) no dynasty can be formed between gessit and the verb of the adjunct clause S' , so that the pronoun coindexed with the matrix subject is correctly predicted to be ei. However, if this adjunct clause is embedded in a complement clause to another clause, the anaphoric element can be bound by the higher subject: although the adjunct clause would not be directly governed by the main verb, it would be contained in a domain governed by this verb and, hence, a dynasty could be formed again. This is what seems to happen also in Latin: (7)
AriouistUSi... respondit ... [s* nos esse Imquos, [s' quod Ariovistus-NOM answered we-ACC be-INF unfair-ACC because in suo iure sei interpellaremus]] in his jurisdiction-ABL REFL-ACC obstruct-SUBJ 'Ariovistus ... replied ... that we were unjust in obstructing him in his own jurisdiction' (Caes, B.G., 1.44.8)
S' is a causal clause included in a domain S*, in which it is not governed, but which is governed by the matrix verb. Hence, the anaphor in S' will not be bound in S*, but in the higher clause (the main clause), as predicted. We can illustrate this discussion with the following schema:
174
Elena Benedicto
ex is the domain containing the anaphor Y. If ex is an S' -complement to xo, a dynasty is established, and the domain in which Y can be bound is enlarged up to /3, where it can find the appropriate antecedent. This is what we find in examples (1-3) discussed above. If ex is as shown in (9): ex= S'
(9)
/\
S
S' a
6 Sa' being an adjunct clause, then the binding domain for Y is again /3; ex (S') is not the appropriate domain, because the verb of S does not govern Sa' and therefore does not form a dynasty with its verb. 6 Let us now consider the next problem, namely relative clauses with a longdistance anaphor and the role of the subjunctive mood in these clauses. At first sight we would expect that the appearance of the subjunctive would trigger LD se, as in (10): (10)
Ariouistusi respondit [magnam Caesarem iniuriam facere Ariovistus-NOM answered big-ACC Caesar-ACC injury-ACC do-INF deteriora faceret]] [qui suo aduentu uectigalia sibii who-NOM his arrival-ABL income-ACC REFL-DAT worse-made-SUBJ 'Ariovistus answered that Caesar was doing him a serious injury, for his advance was damaging his revenues' (Caes., B.G., 1.36.4)
But upon closer scrutiny we can find not only cases of subjunctive relative clauses (RCs) without the expected LD se, but also cases of indicative RCs with an unexpected LD se. These cases are illustrated in (11) and (12), respectively:
Latin long-distance anaphora
175
(11)
Verresj Milesios nauem poposcit [quae eumj Verres-NOM Milesios-ACC ship-ACC asked that-NOM him-ACC praesidii causa Myndum prosequeretur] protection-GEN for myndum-ACC follow-SUBJ 'Verres asked the Milesians for a ship that should follow him till Myndum for his protection' (Cic., Verr., l.86) (12) Epaminondasj [NP ei [qui sibij ex lege Epaminondas-NOM him-DAT that-NOM REFL-DAT by law-ABL praetor successerat]] exercitum non tradidit praetor-NOM succeeded-IND army-ACC not transferred 'Epaminondas did not transfer the army to the one that succeeded him as a praetor according to the law' (Cic., Inv., l.55) As a first approximation, we will focus on the opposition restrictive vs. nonrestrictive RCs, and the role of the subjunctive in both of them. Cinque (1982) proposes two different structural analyses for non-restrictive RCs, on the basis of Italian, French and English cases. 7 The first analysis, [NP, S'], is common for restrictive RCs, and analyses RCs and their heads as daughters of an NP. The second analysis offers a parenthetical structure (cf. (13b)), in which the RC is not included in the same NP as its head: (13) a. [NP, S'] b. NP ... S' ... We will assume here that non-restrictive RCs in Latin always have the second structure. The fact that this second structure is not represented in more detail (in a tree diagram, for instance) reflects a problem concerning the level of attachment of the RC. We will assume here that non-restrictive RCs are attached at a high level in the sentence structure, for instance as a daughter of S. Let us now consider again the examples in (10), (11) and (12). 8 In (10) we have a non-restrictive RC (connected with the NP Caesarem). Following the analysis proposed in (13), the domain immediately containing the non-restrictive RC is the infinitival clause. No dynasty is formed since no government relation exists between the verb in the infinitival clause and the RC. But the verb of the infinitival clause itself does form a dynasty with the superordinate verb. So again, the appropriate binding domain for the anaphor sibi is not the clause containing the RC, that is, the infinitival clause, but the higher one, namely the main clause. Similarly, the RC in (11) is, like the previous one, a non-restrictive RC. Hence, it is not contained in the NP containing its antecedent, nauem. Therefore, the minimal domain containing the RC is the main clause itself. But we know that,
176
Elena Benedicto
since no dynasty can be formed because the matrix V does not govern the RC, no LD-anaphor in the position of eum can yield a well-formed structure. That is why we have a form of the pronoun is instead of the reflexive se. A third case can be distinguished, namely the one presented in (12). In this case we have a restrictive RC in the indicative mood. If we accept the proposed structure for restrictive RCs, the minimal domain containing the RC is the NP also containing the head ei. In this domain (the NP), the LD se cannot find its antecedent. The domain containing this NP, however, is governed by the matrix verb and will be the relevant domain, as it is in fact. On the basis of structures (8) and (9), we can now explain these facts by Koster's (1987) notion of governing subject (g-subject):9 (14)
Governing subject: A g-subject for an anaphor Y is the subject of an XO such that XO governs Y or a domain containing Y
(15)
NP
In (15) the LD-anaphor Y is contained in S'. S' is not governed by a (which is not an XO head). Therefore, no connection is established in a'. That is, a' is not a good domain for Y to be bound in. If we now take (14) into account, we can see that the NP in (15) is the subject of an XO that governs a domain (a') containing the anaphor Y, that is, the NP in (15) is the appropriate antecedent to bind the anaphor y.1O In (9) we considered this schema, thinking of a' as containing an adjunct clause. If we now think of a' as an NP containing not an adjunct clause but a restrictive RC, we will have the case in (12).11 If we think of a' as an infinitival clause with a non-restrictive RC hanging directly from it (following the analysis of non-restrictive RCs discussed above), we will have the case in (10). In (11) we have a simple case of a' not embedded, as in (6), and again it is not an appropriate domain in which to bind an anaphor.
Latin long-distance anaphora
177
A last consideration concerns the role of the subjunctive mood in RCs. Anderson (1983: 21, fn. 3) says that a reflexive may appear in an RC, provided that it is in the subjunctive. We have seen that this is not the case in Latin. Being in the subjunctive is not a sufficient condition (in (11) the presence of the subjunctive could not block the presence of the pronoun eum); nor is it a necessary condition (in (12), in contrast, we found the anaphor sibi together with an indicative). In restrictive RCs, the presence of the subjunctive has been explained on the basis of a feature [-specific] in the antecedent of the relative pronoun. 12 In the cases of non-restrictive RCs, it has been proposed that these subjunctive clauses are like a predication over the head NP.13 In any case, the presence of subjunctive in RCs does not seem to be relevant for the binding phenomena we have been discussing here. We have seen up to now how the statements in (4) and (14) could account for the behaviour of both adjunct and relative clauses. Basically, what we saw was the following opposition: (16)
a.
b.
S
~VP
NP
/~
s'
V
~
S
S'.
~
.-/'
Unfortunately, the situation is not that simple. I will now schematically point out some further, rather obscure facts about adjunct clauses. The problems mainly concern two facts. First, the subject of a dynasty containing an adjunct clause sometimes cannot bind an anaphor in that adjunct clause (S.) (that constitutes a violation of the generalization in (16b )). Second, the subject of a matrix clause can sometimes bind an anaphor in an adjunct clause directly contained in it (contrary to what is established in (16a)). I have grouped the examples according to the class of adjunct clause and, as far as possible, as minimal pairs. So, (17) offers two kinds of conditional clause. In (a) we are confronted with the first case: ei in the conditional clause is coreferential with the matrix subject is, but contrary to expectations, it is a pronoun and not an anaphor. This is especially interesting when compared with (3) (repeated here as (17b )), which represents a 'canonical' case:
178
Elena Benedicto
(17)
a.
b.
Namque iSi pollicitus est regi [sei eum he-NOM promised king-DAT REFL-ACC him-ACC rex permitteret ut ... ]] interfecturum [sa si eii kill-INF if him-DAT king-NOM allow-SUBJ that-COMP 'For he (Mithridates) had promised the king to kill him (Datames), provided the king would allow him to .. .' (N ep., 14.10.1) Perfugai el est pollicitus, [[Sa si praemium a deserter-NOM him-DAT promised if a reward-ACC sibii proposuisset], sei eum ueneno necaturum] REFL-DAT assure-SUBJ REFL-ACC him-ACC poison-ABL to kill-INF 'A deserter promised to him, if (Fabricius) would assure him of a reward, to kill him with poison' (Cic., Off., 3.86)
The contrast in the examples with causal clauses shows a violation of both (16a) and (16b), which Giorgi (1984) assumes to be general for all languages which have long-distance anaphora. In (18a), the empty subject proi is not a governing subject for the anaphor in the adjunct causal clause, and yet it is an appropriate antecedent for se. What we see in (lSb) is a violation of (16b). (18)
a.
et tamen ipsam prOi quoque ictu calcis occidit, [Sa quod kick-ABL kill because her-ACC too sei . . . conuiciis incesserat] REFL-ACC reproaches-ABL assail-IND ' ... yet he caused her death too by kicking her because she had scolded him' (Suet., Ner., 35.3) b. Ambiorixi ad hunc modumlocutus est: [sesei . . . plurimum Ambiorix-NOM in that way talked REFL-ACCvery much ei; confiteri deb ere [Sa quod eii ... et filius et him-DAT be indebted-INF because him-DAT son-NOM fratris filius. .. remissi essent]] brother-GEN son-NOM had been sent-SUBJ 'Ambiorix admitted that he was very greatly indebted to him (Caesar) for a son and a nephew had been sent back to him' (Caes., B.G., 5.27.2)
The so-called 'narrative' cum (used with subjunctive, imperfect and plusquamperfect) seems generally to allow coreference with the immediate higher subject, as shown in (19a); but we also can find examples such as (19b), in which this is not the case: (19)
a.
Haec propterea de me dixi [ut mihi Tuberoi that-ACC therefore about me (I) said so that me-DAT Tubero-NOM
Latin long-distance anaphora
179
[Sa cum de sei eadem dicerem] ignosceret] COMP about REFL-ABL the same-ACC say-SUBJ forgive-SUBJ 'And I have spoken thus about myself in order that Tubero might forgive me when I said the same thing about him' (Cic., Lig., 8) b. Quin etiam [Sa cum eii . . . orationem ... Lysias Indeed on COMP him-DAT speech-ACC Lysias-NOM prOi legit attulisset ... ] non inuitus bring-SUBJ not unwilling-NOM read 'Indeed on Lysias bringing him a (written) speech, ... he read it not unwillingly' (Cic., de Or., l.231)
3 (Non-subject) antecedents It has often been proposed (for instance, by Anderson (1983), in his consideration of Latin reflexives) that LD-anaphors can only be bound by subjects. I will show now that this generalization is not entirely accurate. In general, the presence of non-subject antecedents has been detected in a number of languages. It is assumed, anyway, that there are more possibilities for non-subject antecedents with anaphors following principle A of the binding theory ('strict' anaphors) than with LD-anaphors. Giorgi (1984) discusses some Italian LD facts in which (non-subject) experiencers happen to be antecedents. She shows that experiencers belonging to the a-grid of experiencing verbs can trigger coreference with an LD-anaphor, that is, they function 'like' subjects. On the basis of these facts, she proposes what she calls the thematic hierarchy:
(20) Thematic hierarchy l. Agent 2. Experiencer 3. Theme and others The a-role agent is usually represented by the subject. That is why subjects usually trigger coreference with an LD-anaphor. When there is no agent, the experiencer is the most prominent element in the thematic hierarchy, that is, the experiencer becomes the 'prominent' argument. The prominent argument of a a-grid is, according to Giorgi, the appropriate antecedent of an LD-anaphor in its p-domain (the domain of the prominent argument). 14 Let us now consider the following Latin examples: (21)
lam inde ab initio Faustu10i spes fuerat since beginning-ABL Faustulus-DAT hope-NOM be-PAST
180
Elena Benedicta
[regiam stirpem apud sej educari] royal-ACC stock-ACC next-to REFL-ACC be-educated-INF 'Since the beginning, Faustulus had hoped that (someone of) royal stock was being educated with him' (Liv., 1.5.5) (22) Annalij litterae pergratae fuerunt [s' quod because Annalis-DAT letter-NOM pleasant-NOM was sej diligenter] curares de to worry-SUBJ about REFL-ABL diligently 'Annalis was very pleased with your letter, because you worried very much about him' (Cic., Quint., 3.1.20) (23) Aratorisj interest ita sej frumenta habere, rut quam farmer-GEN interest so REFL-ACC crops-ACC have-INF COMP plurimo decumae uenire possint] the highest tithes-NOM be-sold-INF be-able-SUBJ 'To the farmer it is important to have crops so heavy that the tithes may fetch the highest prices' (Cic. Verr. 3.147) In (21) we have a possessive dative, in (22), a dative dependent on the adjective pergratae, and in (23), we have a verbal genitive depending on interest. If we look at them carefully, we can see that they all represent experiencers of O-grids whose prominent argument is precisely the experience, the domain of which (that is, the p-domain) includes a clause (Ad-clause in (21) and (23), and a quod-clause in (22)) containing the anaphor. So, as predicted, the LD-anaphors in examples (21-23) are correctly bound by the prominent argument, that is, they are P-bound (in the sense of Giorgi (1984)). We will now consider another group of non-subject antecedents. Contrary to what we saw in the previous cases, they are not at all prominent in the sense that we discussed before: (24)
Canumj tam fida custodia.. . quid significat aliud nisi dogs-GEN such a trusty watchfulness-NOM what mean else except [sej ad hominum commoditates esse generatos?] REFL-ACC for men-GEN comfort-ACC be created-INF 'The trusty watchfulness of the dogs, ... what else does it mean, except that they were created for human comfort?' (Cic. Nat. dear., 2.158) (25) A Caesarej ualde liberaliter inuitor [sibij ut By Caesar-ABL very generously (I) am invited REFL-DAT COMP sim legatus] be-SUBJ legate-NOM
Latin long-distance anaphora
181
'Caesar most liberally invites me to take a place on his personal staff' (Cic., Att., 2.18.3) In (24) we have a genitive depending on an NP. In (25) we have an ablative agent phrase in a passive sentence. This case is even stranger, because agent phrases in passive contexts are said to be always unable to trigger coreference in Latin. In fact we have cases in which an agent phrase does not trigger coreference: (26)
a Curionej mihi nuntiatum est [eumj ad me uenire] by Curio-ABL me-DAT was announced he-ACC to me-ACC come-INF 'It was announced to me by Curio that he was coming to me' (Cic., Att., 1004.7
What I would like to propose here for (24), (25) and similar cases is that the phrase that triggers coreference with the anaphor is not in its 'normal' position, but that it is syntactically 'prominent', that it is in the Topic position in the sentence. IS Let us consider this proposal more carefully. It goes against the generally accepted statement that anaphors must have an antecedent in an A-position. Aoun (1986), however, distinguishes two kinds of anaphoric relations: A- and A'-anaphora. Binding of an LD-anaphor by an element in Topic means that the anaphor will be bound from an A' -position (Topic), that is, A'-bound. 16 But the dependency relation between binder and bindee will still be the same as in the other cases: a transfer of referential content. I? Aoun (1986) also establishes that what the BT has to determine is the domain in which an anaphor must be bound. As we saw in the previous sections, an LD anaphoric relation requires the existence of an argument-predicate structure in which the argument is a subject in the usual sense or the prominent argument of a given a-grid. In that sense, and following Williams (1980) and Koster (1987), NP is the 'subject' in the argument-predicate configuration [NP XP], where XP is any maximal projection, and where NP gets its a-role either from indirect assignment ('normal' subjects), or by binding an element in the predicate XP. The latter situation is what we see in the case of Topic. What is interesting in an LD anaphoric relation with the antecedent in Topic is that the two main requirements present in the other cases also hold: the existence of a predication relation between the antecedent (the Topic) and the domain containing the anaphor; and the binding of the anaphor by an argument, since the NP in Topic is itself bound to an argument in XP. And, of course, the general properties of an anaphoric relation also hold: obligatoriness, uniqueness of the antecedent, and locality.
182
Elena Benedicta
4 Conclusions In this chapter, we established the basic requirements and conditions about the domain in which an anaphor has to be bound. We adopted the notion of dynasty as the basic mechanism to enlarge the minimal domains, which permitted us to avoid direct reference to the notion 'indirect speech', which is so common in discussions of this topic. We saw that the concept of a dynasty is more comprehensive and, therefore, accounts for the indirect speech facts, as well as for (1) and similar cases (in which no indirect speech is involved). We also provided a structural account of relative and adjunct clauses, which showed some similarities between them. We paid some attention to the role of the subjunctive mood. The facts about RCs, in particular, showed that subjunctive itself is not a relevant concept, as far as domain enlarging is concerned. Like the infinitive, it might of course be the way of representing a [+dependent] verbal inflexion in sentential complements, but, as we demonstrated, it did not have any effect in triggering LD-anaphora (cf. (11) for RCs, and (17) for an adjunct case). As far as conditions on antecedents are concerned, the main requirement was the existence of an argument-predicate structure. We saw that the appropriate antecedent for an LD-anaphor was the 'prominent' argument of the extended domain (considering 'prominent' the agent subject and, otherwise, the experiencer or the one marked 'prominent' in the corresponding 8-grid). Finally, we extended the notion of thematic 'prominence' to 'Topic prominence' , to account for some cases of non-subject antecedents. It should be noted that our analysis of Topic (establishing a predication relation) is compatible with the notion of complete functional complex, proposed in Chomsky (1986), since [s' NP XPJ is the domain in which all the 8-roles pertaining to a lexical head are assigned, and in which all the grammatical functions of the same head are realized. It is also compatible with the view expressed in Giorgi (chapter 9), that predication is a necessary but not sufficient condition for LD-anaphora, but that also a (relevant) thematic relation is needed. The thematic relation is given through the relation of the NP in Topic with an argument in XP; it becomes relevant precisely because of the fact that this NP is in Topic position. Finally, it also obeys the principle of independent targeting of Hellan (chapter 2), the relevant R relation being 'predication command', together or combined with some kind of thematic involvement.
Latin long-distance anaphora
183
Notes 1. I will not discuss possessive suus, -a, -um, which has properties slightly different from reflexive se. 2. Latin grammars typically associate long-distance se (indirect reflexive) with indirect discourse. Although this is true most of the time, it is not a necessary condition. Indirect discourse enters into the more general proposal to be made below. 3. Some Latin grammars and students of Latin (see Bertocchi & Casadio (1980» consider lit-clauses of this type to be final clauses (or sometimes final completive clauses). Since these clauses are obligatory complements of the matrix verb, some of their syntactic features are determined by this verb and they are always associated with a thematic role of its a-grid, I will analyse the clauses in question as objects of the matrix verb. 4. Sentences of this type have also been considered causal clauses (see note 3). However, further research seems to be in order as to the a-grid of the relevant verbs and the (non-)exclusive presence of a quod-complementizer. I am not referring in this paragraph to the dico-qllod type, which is a late Latin construction. The few examples found in Classical Latin are not relevant to our present purposes. 5. See, among others, Anderson (1983) for Icelandic and Giorgi (1984) for Italian. 6. Note that in the previous case (ex = S', S' a sentential complement, as in (1)-(3», ex was an optional domain for the anaphor to be bound in. In that case, the anaphor would be a 'strict' anaphor and conform to principle A of the binding theory. This is always a real possibility for the reflexive se. 7. Since the very beginning of transformational grammar, attempts have been made to give a structural account for the difference between restrictive and non-restrictive RCs, either by proposing two different levels of attachment within the NP, or by proposing a coordination-based analysis for non-restrictive RCs. See Jackendoff (1977) for a general overview. 8. We will ignore some problems posed by linear order in certain Latin examples, particularly problems posed by the relative order between the head of the RC and the RC itself. 9. In order to explain the difference between zich and zichzelfin Dutch, Koster also uses the notion of d-subject, i.e. a governing subject such that XO is a member ofa dynasty. It seems to me that the notion g-subject suffices for Latin. 10. Although antecedents will be discussed later, I would like to briefly point out some peculiarities about intermediate subjects in long (three- or four-member) dynasties. Examples of those cases (cf. (i), (ii) and (iii) below) show: (a) that intermediate subjects (like cilles principes in (i», in spite of being g-subjects (cf. (14», cannot bind an LD-anaphor in their domain; and (b) that, in case they themselves are anaphors (accusative subjects in Ad-clauses), they can only be bound by the immediately dominating subject (Lentllillm in (ii», and not by the highest one (Proi in (iii». (i) [cum ciues principesj prOi animaduertisset [ej timere COMP citizens-ACC leading-ACe perceive-SUBJ-PAST fear-INF [ne propter sei bellum iis j Lacedaemonii et COMP-neg because-of REFL-ACC war-ACC them-DAT Lacedaemon and Athenienses indicerent]]] ad Admetum confugit prOi Athens-NOM declare-SUBJ towards Admetum-ACC escaped
184
Elena Benedicto 'When he perceived that the leading citizens were fearful that the Lacedaemonians and Athenians would declare war upon them because of his presence, he took refuge with Admetus' (Nep., 2.8.3) (ii) Galli;... dixerunt ... [Lentulum; sibi; confirmasse Gauls-NOM said Lentulus-ACC REFL-DAT assure-INF-PAST [se; esse tertium illum Cornelium Jl REFL-ACC be-INF third-ACC that-ACC Cornelius-ACC 'The Gauls ... said ... that Lentulus had assured them that he was that third Cornelius' (Cic., Cat. 3.9) (iii) [ ... Allobrogibus; sese; persuasuros] existimabant pro; [ut Allobroges-DAT REFL-ACC persuade-INF consider COMP [per suos fines eosi/'; ire] paterentur pro;] through their borders-ACC they-ACC go-INF suffer-SUBJ 'These supposed that they would persuade the Allobroges to suffer a passage through their borders' (Caes, B.G., 1.6.3)
11.
12. ·13. 14.
15. 16.
17.
What those facts - together with the consideration that the minimal domain (containing a subject) in which the i-subject could be bound is precisely the clause immediately containing it - suggest is that i-subjects are subject to the 'local' binding strategy: they are local binders and local bindees. What this implies must be left open for future research. We cannot say, following Koster (1987), that Epaminondas in (12) is a d-subject, since 'the head of an [object] NP is not dependent on a V' (1987: 337), in the sense that dynasties for anaphors are only formed by dependent heads, such as dependent verbs (infinitives and subjunctives) and the prepositions of prepositional objects. This was already observed for Romance, for instance by Rivero (1977) for Spanish and by Rigau (1981) for Catalan. For Latin, see Vester (1985). See Lavency (1981). Giorgi (1984) establishes that 'each lexical head that is a verb, an adjective or a noun, defines a thematic (8) domain, i.e. the set of arguments 8-marked by the head. One of the arguments of this thematic domain can be said to be 'prominent' with respect to the others, and the set of remaining ones, with all the material they dominate, can be called its P-domain' (1984: 317). For more details about Topic in Latin, see Benedicto (in preparation). It would be interesting to test whether this kind oflong-distance anaphora behaves in the same way as the other cases of long-distance A' -binding (for instance, long wh-movement). Curiously, the first kind of non-subject antecedent is also found in cases of 'strict' anaphora, while I have no evidence of a non-subject antecedent in Topic position in those cases. There is no transfer of lexical content, as in the case of variables. A clear distinction can be made between dependency relations involving a transfer of lexical content (as with empty categories) and dependency relations in which only a referential index is shared.
9 Prepositions, binding and 8-marking Alessandra Giorgi
1 Introduction
The aim of this chapter is twofold: on one hand I propose an account for the distribution of third person pronouns, clause-bound and long-distance anaphors within PPs; on the other, to explain such distribution, I individuate thematic properties peculiar to prepositions, which differentiate them from other lexical heads such as adjectives and nouns. In this way, some phenomena concerning the behaviour of PPs in small clause constructions, and in predicative contexts in general, will be captured. In section 2, I analyse the distribution of anaphors and pronouns. In section 3, I present the central hypothesis of this work, i.e. that the behaviour of such elements is related to the thematic properties of prepositions. In sections 4 and 5, I consider the distributions of PPs in small clauses and in predicative constructions. In section 6, I discuss some differences between locative and non-locative prepositions. Finally, I try to handle some potential counterexamples, suggesting that independent principles can account for certain apparent anomalies. This chapter is a portion of a larger work in which I will also consider cross-linguistic evidence (English, Dutch and French), which not only supports my view of the phenomena analysed here, but can also shed light on the peculiarities of the languages considered. In this chapter, for space reasons, I only examine data from Italian, briefly mentioning in a note the most salient aspects in which the other languages diverge.
2 Prepositions and long-distance anaphors (LDAs) It is well known (cf. Bouchard (1984), Chomsky (1981)) that, when embedded within PPs, anaphors and pronouns yield grammatical results, variable according to the choice of the antecedent. Consider the following examples:
185
186 (1)
Alessandra Giorgi a. b.
(2)
a. b.
(3)
a. b.
(4)
a. b. c.
Gianni i ha aizzato Maria contro di se/?contro di lui/*contro se stessoi Gianni turned Maria against self/against him/against himself Gianni ha aizzato Mariai *contro di se/*contro di lei/contro se stessai Gianni turned Maria against self/against her/against herself Giannii ha riconciliato Maria con se/?con lui/?*con se stesso Gianni reconciled Maria with self/with him/with himself Gianni ha riconciliato Mariai *con se/*con lei/con se stessai Gianni reconciled Maria with self/with her/with herself Gianni ha ripiegato la pasta sfogliai *su di se/*su di essa/su se stessai Gianni folded the dough on self/on it/on itself Giannii ha ripiegato la pasta sfoglia su di se/su di lui/*su se stessoi Gianni folded the dough on self/on him/on himself Giannij ha aizzato Mariai contro coloro che disprezzano il proprio*i/j figlio Gianni turned Maria against those who despise self's child Giannij ha riconciliato Mariai con coloro che amavano il proprio*iljfiglio Gianni reconciled Maria with those who loved self's child *Gianni ha ripiegato ogni pezzo di pasta sfogliai suI contenitore Gianni folded every piece of dough on the container che era adatto alla propriai cottura which was best for self's cooking
In these examples the distribution of pronouns and anaphors varies, at first sight, depending upon the grammatical function of the intended antecedent, Gianni or Maria. However, before analysing the phenomenon, let me clarify some general characteristics of the anaphoric system ofItalian. As I pointed out elsewhere (cf. Giorgi (1984)), the Italian non-clitic anaphoric system for third person includes a 'long-distance' possessive anaphor, proprio, and two non-possessive ones, se and se stesso. Proprio is morphologically an adjective and can co-occur with the article, like other possessive elements (cf. Giorgi & Longobardi (1987)). It can be clause bound and, in this case, both subject and object of the same clause can be possible antecedents, or it can be long-distance bound, and, if so, it is in most cases subject oriented, in a sense to be made precise in terms of O-theory. Se is a subject-oriented anaphor and is invariant for both numbers; se stesso is the clause-bound anaphor, corresponding to himself/herself, stesso is morphologically an adjective and the -0 (masc, sg) ending alternates with other ones (-a, -e, -i), according to the choice of the antecedent. N qtice also that se cannot be governed by a verb, at least in my dialect: (5)
Gianni ama solo se stesso/*se Gianni loves only himself
Prepositions, binding and 8-marking
187
Se must be governed by a preposition, whereas se stesso admits both governors, i.e. a verb, as in (5), and a preposition, as in (6b): (6)
a. b.
Giannii e molto contento di sei Gianni is very happy with self Una lunga terapia psicoanalitica riconcilio Mariai con se stessai a long psychoanalytic therapy reconciled Maria with herself
The binding options of proprio parallel the ones of se when long-distance bound, and those of se stesso when clause bound. I will not further discuss this matter here. Let us go back to examples (1-3): se can only refer to Gianni and never to Maria. Even if, when referring to Gianni, se is the preferred option, the pronoun lui is also grammatical, whereas the clause-bound anaphor is excluded, contrary to normal cases. When the object is the intended antecedent, abstracting away from the coreference possibilities of se, we can observe that the pronoun and the anaphor are again in complementary distribution, although reversing the judgements holding with respect to the case of a subject antecedent. We will come back to this point in a while. The data in (4) parallel the distribution of se in (1-3), since long-distance bound proprio can only refer to Gianni and not to an object. l However, consider the following examples: (7)
a. b. c.
Ho aizzato Mariai contro il proprioi figlio I turned Maria against self's child Ho riconciliato Mariai con il proprioi figlio I reconciled Maria with self's child Ho ripiegato ogni pezzo di pasta sfogliai sui proprio?i contenitore I folded every piece of dough on self's container
Examples (7a-c) minimally contrast with those in (4a-c), confirming that proprio, when clause bound, can refer also to an object, like se stesso. 2 The distribution of clause-bound anaphors and pronouns can be accounted for on the basis of the following hypothesis: the sequence NP-PP counts as the relevant domain for principles A and B of the binding theory, as if the NP were, in some sense, the 'subject' of the PP, able to make the domain opaque. Such a solution seems quite natural, given that in this way the exceptionality of the distribution disappears, being reduced to familiar notions, i.e. to a definition of the relevant domain almost identical to the one adopted in usual cases. Along the lines suggested by Chomsky (1986a), we can use the notion of complete functional complex (CFC) to identify the relevant domain for principles A and B to apply. Under the formulation proposed in Giorgi (1987), on the basis of independent evidence, a CFC can be defined as follows: 3
188
Alessandra Giorgi
(8) (Given a lexical head ex) 'Y is a CFC iff it meets at least one of the following requirements: a. it is the domain in which all the 6-roles pertaining to the head are realized b. it is the domain in which all the grammatical functions pertaining to the head are realized where 'grammatical functions' also include predication relations, essentially that of 'subject of' the X max projected by the head. In our cases, the NP in question can be defined the 'predication subject' of the structure, projected as a small clause. However, the situation immediately turns out to be less clear as soon as we consider the distribution of long-distance anaphors: if the relevant NP behaves as a subject, identifying the domain for the application of principles A and B, how is it that LDAs cannot refer to it? In fact, if such an NP has to be assimilated, under every relevant respect, to a subject, it should be predicted to be an acceptable antecedent for se and long-distance proprio. The puzzle becomes even more interesting if we contrast the examples given above with closely corresponding structures containing an adjective instead of a preposition. Consider the following examples: (9)
a.
b.
(10)
a.
b.
(11)
a. b.
Giannij ha visto il professorei contento degli studenti che Gianni saw the professor satisfied with the students who seguivano il proprioi/j corso followed self's class Giannij ha visto il professorei con gli studenti che seguivano the professor with the students who followed Gianni saw il proprio?*i/j corso self's course Giannij ha trovato il medicoi soddisfatto delle condizioni Gianni found the physician satisfied with the conditions del paziente affidato alle propriei/j cure care of the patient assigned to self's all'abitazione del Giannij ha trovato il medicoi davanti Gianni found the physician in front of the house of the paziente affidato alle proprie*i/j cure care patient assigned to self's Giannij ha visto il professorei contento dei proprii/j studenti Gianni saw the professor satisfied of self's students Giannij ha visto il professorei con i proprii/j studenti Gianni saw the professor with self's students
Prepositions, binding and 8-marking (12)
a. b.
Gianni j ha trovato Gianni found Gianni j ha trovato Gianni found
il the il the
medico; physician medico; physician
189 soddisfatto dei proprii/j pazienti satisfied with self's patients davanti ai proprii/j pazienti in front of self's patients
(9b) and (lOb) can be considered, with respect to the relevant structure, identical to the cases illustrated in (4). The contrast between (9b) and (9a) and (lOa) and (lOb) is rather sharp: in (9a) and (lOa) a long-distance proprio can refer to professore, or to medico, whereas it cannot in (9b) or (lOb), as in (4). If proprio is not longdistance bound, once again, the contrast disappears, so that (Ub) and (12b) are grammatical, as examples (7a-c) are. These data challenge the naive hypothesis one could be tempted to formulate, i.e. that, though creating an opaque domain with respect to principles A and B of the binding theory, the object remains an object and, therefore, is not a possible antecedent for a subject-oriented anaphor. If this view were correct, in fact, we should expect (9a) and (lOa) to be ungrammatical, on a par with (9b) and (lOb). It is fairly obvious, to the contrary, that the solution must be able to single out PPs against APs, in a principled way. A priori there are two possibilities: one could either try to show that the difference between (9a-b) and (lOa-b) is a structural one, or claim that the structure is the same, but that, due to the interaction of certain properties of prepositions with those of LDAs, the distribution given above is obtained. Given that there is no evidence suggesting a possible structural solution, the proposal I am going to develop here follows the second line of reasoning. I claim that the structures projected by both sentences in (9) and (10) are identical and that the differences are due to thematic properties of prepositions, which diverge from those of adjectives, as anticipated above. This hypothesis will turn out to predict other data in an independent domain, i.e. that of small clauses and predicative structures. Notice also that nouns pattern together with adjectives, as shown by the lack of contrast in (13). (13)
a.
b.
I dipendenti hanno eletto the dependent workers elected fondata dal proprioi padre founded by self's father I dipendenti hanno eletto the dependent workers elected
Gianni; presidente della ditta Gianni president of the firm
Gianni i presidente della propria; ditta Gianni president of self's firm
These data suggest once more that prepositions must be treated as exceptional.
190
Alessandra Giorgi
3 The hypothesis
In Giorgi (1984), I proposed an account of the distribution of LDAs. The crucial variables considered there were the thematic properties of the antecedent and the mood (subjunctive, or infinitive, vs. indicative) ofthe clauses intervening between the anaphor and the antecedent. Here we will abstract away from the mood, which is irrelevant to the present purpose, and will focus on O-role assignment. For an LDA to be bound, the following conditions must be met: (14) (15)
(16)
A long-distance anaphor is P-bound (X P-binds r3 iff r3 is coindexed with (X and r3 is in the P-domain of (x. P-domain of (X: Given a lexical head which assigns the O-roles kl ... k n to (Xl ... (Xm and kb prominent among kl ... k m assigned to (Xl (Xz ... (Xm together with their predicates and all the nodes they dominate, are the P-domain of (Xl
For a detailed discussion of these principles see Giorgi (1984), where a hierarchy of prominence for O-roles is also suggested. Let me briefly point out the empirical content: P-bound stands for 'bound by the Prominent argument': this means that an LDA can only take as antecedent one element in a certain thematic domain, i.e. the most prominent one, where agents and experiencers are usually prominent. Almost always, these O-roles are assigned to subjects and, therefore, LDAs turn out to be subject oriented. An approach in terms of O-relations permits us to distinguish structural subjects from O-prominent elements. Our hypothesis will develop this idea: in the examples given above, the relevant element, the apparent 'object' of the verb, is always the structural subject of the XP, forming with it an SC. In the case of a preposition, however, it is not thematically prominent, whereas it is so for an adjective or a noun. Let us formulate the most radical version of this principle: (17)
A preposition does not assign an external O-role
If (17) is correct, then we can predict the distribution of LDAs. The 'object' of the clause cannot be prominent in the O-domain identified by a preposition, given that the only element receiving a O-role from P is the internal argument of the latter. On the contrary, adjectives and nouns O-mark the external argument, with the consequence that it is a possible antecedent for an LDA. I attribute the following structures to the examples just discussed:
Prepositions, binding and e-marking (18)
191
a.
As illustrated by the arrows, the verb in both cases can satisfy the a-requirements ofNP j in a way which will be considered in more detail in the next section. The PP does not assign any a-role to NPj, whereas the AP and NP 3 do. 4 The consequence of this hypothesis is that the LDA dominated by an AP, or NP, node is contained in the P-domain of the prominent argument NP j and can be bound by it. An LDA dominated by the PP is not contained in the P-domain of NPj, given that such an NP is not a-marked at all by P and, if possible, must refer outside the SC, to NP z (which in turn is the prominent argument in the P-domain identified by the verb). The relation between the PP, on one side, and AP and NP on the other, with NPj, is in any case a predication relation (see Williams (1980), Rothstein (1983». This is sufficient to license the PP and the AP, according to the licensing conditions given in Chomsky (1986a). Notice that such a predication relation must be instantiated. A small clause representation, like the one in (17a) or (l7b), is the only one compatible with the distribution of (clause-bound) anaphors and pronouns (under the CFC definition given above) and a binary branching constraint on phrase structure (cf. Kayne (1984». From the viewpoint of the full interpretation principle proposed by Chomsky (1986a), the XP in these SCs is certainly not an argument, nor an operator, therefore it must be a predicate of NP j • 5 Before we proceed, it should be noted that, if the present analysis is correct, it constitutes a strong argument in favour of the idea that LDAs obey a thematic strategy like the one suggested in Giorgi (1984). Structural differences, in fact,
192
Alessandra Giorgi
seem not to be able to discriminate between APs-NPs and PPs, whereas the thematic approach to the distribution will be confirmed in the next sections by a full range of completely independent observations.
4 PPs in other small clause constructions In order to show that this hypothesis is correct, we will first discuss some additional data clearly suggesting that prepositions do not assign an external 8-role and then, in the following section, we will analyse copular constructions. As briefly mentioned above, according to Chomsky's (l986a) full interpretation principle, the NP subject of an SC must receive a 8-role in order to be licensed. My proposal is that with verbs like vedere 'see' or tirare 'pull' such 8-requirements can be satisfied in the following way: the verbs 8-mark the small clause, under sisterhood, and, as a lexical property of theirs, the 8-role can percolate down to the SC subject. The SC and its subject, therefore, end up sharing the same 8-role (e.g. theme). This hypothesis seems to be independently needed to capture the fact that the sentences in (19) logically imply the sentences in (20): (19)
a. b.
(20)
a. b.
Ho visto Gianni in cucina I saw Gianni in the kitchen Ho tirato il tappeto sotto al tavolo I pulled the carpet under the table Ho visto Gianni Gianni I saw Ho tirato il tappeto I pulled the carpet
In both cases, the relation between the verb and the NP seems to be the same and our hypothesis can provide an insight into this complex phenomenon (cf. also Higginbotham (1987)). A prediction following from the requirements imposed by 8-theory plus principle (17) is that, in structures where the only possible 8-assigner is the predicate of the SC, i.e. where 8-percolation is inhibited, a PP-predicate should result in ungrammaticality, due to lack of 8-role on the subject NP, contrasting with other XPs. To perform our test, we must look for verbs which can take an SC, but not a simple NP: rendere 'render' and ritenere 'believe' belong to this class: (21)
a.
Ho reso Gianni felice I rendered Gianni happy
Prepositions, binding and 8-marking
193
b. (22)
Ritenevo Gianni felice I believed Gianni happy a. *Ho reso Gianni I rendered Gianni 6 b. *Ritenevo Gianni I believed Gianni
Examples (21a-b) cannot imply (22a-b), and the latter indeed are sharply ungrammatical and uninterpretable, as if the NP were unlicensed. The impossibility of the NP subject of the SC receiving a a-role can explain these phenomena. If rendere and ritenere can a-mark the SC, but the a-role cannot percolate down to its subject, the NP by itself cannot be interpreted. My prediction concerning the distribution of PPs is actually borne out. Consider the following examples: (23)
(24)
(25)
a. ??Ritengo il giornale suI tavolo/qui I believe the newspaper on the table/here b. *Rendero il giornale suI tavolo/qui I will make the newspaper on the table/here a. Ritengo il giornale interessante I believe the newspaper interesting b. Rendero il giornale interessante I will make the newspaper interesting a. Ritenevo Gianni il medico pili bravo della I believed Gianni the best physician of the b. Un buon corso di specializzazione, rendera a good specialization will render il medico pili bravo della citta the best physician of the town?
The following structure can be attributed to (23-25): (26)
S
~
NPz
VP
A
V
sc
~ NP I
{AP NP 3
PP
citta town Gianni Gianni
194
Alessandra Giorgi
The verb, in these cases, does not assign a (I-role to NP 1 and the only way to satisfy the (I-criterion is through (I-marking by the predicate of the small clause. If such an XP is not able to satisfy this requirement, the structure is predicted to be ungrammatical. In fact, the PP is usually worse (to differing degrees) than the AP and the NP. Other verbs belonging to this class are: considerare 'consider' and stimare 'esteem': (27)
(28)
a.
Consideravo Gianni intelligente I considered Gianni intelligent b. Consideravo Gianni un buon medico I considered Gianni a good physician c. ??Consideravo Gianni a Parigi I considered Gianni in Paris a. Stimavo Gianni intelligente I esteemed Gianni intelligent b. Stimavo Gianni il medico pili bravo della citta I esteemed Gianni the best physician in town c. ??Stimavo Gianni a Parigi I esteemed Gianni in Paris 8
Under NP-movement, as expected, the thematic properties of the verb stay unchanged. Consider the following passive examples: (29)
Gianni e considerato un buon medico/intelligente Gianni is considered a good physician/intelligent b. ??Gianni e considerato a Parigi Gianni is considered in Paris a.
Analogously, the structure containing the PP is less acceptable even with raising verbs. (30)
(31)
a.
Gianni sembra intelligente Gianni seems intelligent b. Gianni sembra il medico pili bravo della citta town Gianni seems the best physician III c. ??Gianni sembra in salotto Gianni seems in the living-room a. Gianni risulto intelligente Gianni resulted intelligent b. (In quel concorso), Gianni risulto il medico migliore (in that competition), Gianni resulted the best physician
Prepositions, binding and 8-marking
195
b. ??Gianni risulto in salotto Gianni resulted in the living room With respect to a-marking there seems to be no difference between the examples in (27) and (28) and the ones given above. There is only a difference in Case marking: passives, sembrare 'seem' and risultare 'result' do not assign Case, so that movement of the subject of the SC to the subject position of the raising verb is obligatory. 9 Let me briefly summarize the results achieved until now. Verbs like vedere 'see', trovare 'find', eleggere 'elect', riconciliare 'reconcile', aizzare 'turn', etc. take a small clause complement and assign both Case and a-role, through percolation, to its subject. The result is that the predicate of the small clause can either be an AP, an NP, or a PP. If it is a PP, long-distance binding is not allowed; given our hypothesis in (14-16) (cf. Giorgi (1984)), this is a first piece of evidence suggesting that a preposition does not assign an external a-role. Verbs like rendere 'render', ritenere 'believe' and considerare 'consider' etc., plus sembrare and risultare, take a small clause complement, but the a-role does not percolate to its subject. As a consequence, if the predicate can a-mark it, the structure is grammatical, as in the case of APs, or NPs, otherwise it is ruled out by the full interpretation principle, as in the case of PPs. Moreover, raising Verbs such as sembrare 'seem' and risultare 'result' cannot even Case mark the subject of the SC, so that raising is obligatory.
5 More on predicative structures The brief discussion concerning the raising structures in the previous section is closely connected with copular constructions in general. Recall first that, according to Stowell (1978) and Burzio (1981, 1986), copular constructions are also derived via raising, by analogy with the cases just discussed: 10 (32)
a.
Maria
e [t bella]
L-J b.
Maria is handsome Maria e [t l'assassino]
c.
Maria is the murderer Maria e [t in cucina]
L-J ~ Maria is
in the kitchen
If essere/be does not assign a a-role to the subject of the SC, we expect (32a) and
196
Alessandra Giorgi
(32b) to differ in grammaticality from (32c), whereas they are all grammatical. Notice also that the ungrammatical or marginal structures given in the previous section considerably improve, if the verb be is inserted: ll (33)
a. ?Ritengo essere il giornale suI tavolo/qui I believe the newspaper to be on the table/here essere il giornale su tavolo b. ?Faro I will make the newspaper be on the table c. Il giornale sembra esser suI tavolo the newspaper seems to be on the table d. Il giornale risulta esser suI tavolo the newspaper results to be on the table
If the hypothesis we suggested above is correct, these facts can be straightforwardly accounted for in the following way: the verb be in (32c) and (33a-d) assigns a O-role to the subject, so that the full interpretation principle can be satisfied. Therefore, the structure be + PP in (32c) and (33a-d) is basically assimilated to the one projected by vedere 'see' or riconciliare 'reconcile' (in the passive), where the verb O-marks (without Case marking) the subject of the small clause. This thematic characteristic can be taken to be part of the lexical properties of be at least under its locative interpretation. Consider also that locative be is not isolated in the Italian lexicon, but takes the same semantic value as the verb stare 'to be'. Consider the following example:
(34)
Maria e/sta in cucina Maria is in the kitchen
Stare is an ergative verb like essere, subcategorizing for a PP, which in structures like (34) has a clear locative meaning, and can be substituted for be in all locative contexts: 12 (35)
a. (?)Ritenevo dover stare il giornale suI tavolo I believed must be the newspaper on the table 'I believed the newspaper to be on the table' b. Faro stare il giornale suI tavolo I will make the newspaper be on the table c. Gianni sembra stare in salotto Gianni seems to be in the living room d. Gianni risulto stare in salotto Gianni resulted to be in the living room
Prepositions, binding and a-marking
197
Of course, essere and stare, unlike other verbs considered above, only take an SC as their argument and not a full sentence. Now, with respect to the typology of possible verbs, we could look for a verb which is thematically like sembrare 'seem' but which only takes an SC as its internal argument, i.e. does not allow either for infinitival be, after raising, or for nonraising structures. In fact, in Italian we find the verb diventare 'become', which is an ergative one, compatible with APs (or NPs), but not with PPs: (36)
a.
Gianni Gianni b. *Gianni Gianni
diventera intelligente/un ingegnere will become intelligent/an engineer diventera a Parigi will become in Paris
We can now interpret the ungrammaticality of (36b) not as an idiosyncratic lexical gap, but as due to the properties of such a verb, filling the slot predicted by the proposed partitioning on the basis of a-marking, Case marking and complement structures. 13 Finally, in the cases illustrated in (37a-d), we also predict that the distribution of LDAs must not vary according to the presence of an adjective or a preposition: (37)
(38)
a.
Quel terroristai risulto essere soddisfatto dell'amnistia che with the amnesty which that terrorist resulted to be satisfied aveva ridotto la propriai pena had reduced self's sentence b. ?Gli avvocati ritenevano essere quel terroristai contento dell'amnistia the lawyers believed that terrorist to be happy of the amnesty che aveva ridotto la propriai pena which had reduced self's sentence a. Quel terroristai risultava ai servizi segreti essere ormai nello that terrorist resulted to the secret services to be by now in the stato che aveva dato asilo politico ai proprii genitori country which had given political refuge to self's parents b. ?I servizi segreti ritenevano esser quel terroristai nello stato che the secret services believe that terrorist to be in the country which aveva dato asilo politico ai proprii genitori had given political refuge to self's parents
No difference emerges between the sentences in (37) and (38). The structures in (37b) and (38b) are both marginal, due to the status of Aux-to-Comp, but do not contrast with each other, at least if compared with the contrasts in (9) and (10) above.
198
Alessandra Giorgi
6 Locative vs. non-locative prepositions
The claim made up to this point basically holds for what we may term 'locative' PPs, i.e. PPs specifying the location in time and space of their subject. In fact, prepositions with a clearly non-locative meaning may exhibit a different behaviour, in that they are more acceptable in those contexts where locative prepositions are ungrammatical: (39)
a. b. c. d. e.
Mario sembra in gran forma Mario seems in very good shape Mario risulta in gran forma Mario results in very good shape Consideravo Mario in gran forma I considered Mario in very good shape Ritenevo Mario in gran forma I believed Mario in very good shape ?Un lungo allenamento rese Mario in gran forma a long training rendered Mario in very good shape
The preposition in is stricto sensu a locative preposition, but, in spite of this, the phrase in gran Jonna has an adjectival interpretation and, in the examples above, also behaves like an adjective. According to our hypothesis, therefore, we can say that such PPs can assign an external a-role. On the other hand, the PP in question maintains the structural characteristics of the prepositions. It is well known, for instance, that prepositions cannot be dropped in coordinate structures, whereas all other major lexical categories can: (40)
a.
Ritenevo I believed b. *Ho messo I put c. *Ritenevo I believed
Teresa soddisfatta di noi e Luisa _ _ di VOl Teresa happy with us and Luisa _ _ with you il libro sotto al tavolo e la penna _ _ al quaderno the book under the table and the pen _ _ the note book Teresa in buona forma e Luisa _ _ eccellente forma Teresa in good shape and Luisa _ _ excellent shape
'Adjectival' and locative prepositions do not differ in this context, which means that there must always be some feature which unifies all types of preposition. My proposal is that the lexical categorization is the same, whereas thematic properties vary. Let us now consider other non-locative prepositions: (41)
a.
?Io sembro/risulto contro l'aborto I seem/result against abortion
Prepositions, binding and 8-marking
(42)
(43)
199
l'aborto b. ?Io sembro/risulto per in favour of (lit: for) abortion I seem/result c. ?Io sembro/risulto a favore degli ecologisti in favour of ecologists I seem/result d. ??Io sembro/risulto senza amicizie importanti without important relations seem/result I a. ?Ritenevo/consideravo/stimavo Mario contro l'aborto I believed/considered/esteemed Mario against abortion b. ?Ritenevo/consideravo/stimavo Mario per l'aborto I believed/considered/esteemed Mario in favour of (lit: for) abortion c. ?Ritenevo/consideravo/stimavo Mario a favore degli ecologisti I believed/considered/esteemed Mario in favour of ecologists d. ?Ritenevo/consideravo/stimavo Mario senza amlClZle Importanti I believed/considered/esteemed Mario without important friendships a. ??Una lunga militanza nel partito rese Mario contro l'aborto a long activity in the party rendered Mario against abortion b. ??Una lunga militanza nel partito rese Mario a long activity in the party rendered Mario per l'aborto in favour of (lit: for) abortion c. ??Una lunga militanza nel partito rese Mario a favore a long activity in the party rendered Mario in favour degli ecologisti of the ecologists d. ??L'improvvisa defezione dal partito rese Mario senza the sudden defection from the party rendered Mario without amicizie importanti important relations
The variation can be attributed to the degree in which the preposition is interpreted as 'adjectival', and therefore able to assign the relevant a-role. Other independent tests can be provided to show that these prepositions have a more adjectival interpretation. Consider, for instance, the compatibility with adjective intensifiers, such as particolarmente 'particularly', intensamente 'intensely', molto 'very' or poco 'little' (under the 'intensity' reading). Locative prepositions are not compatible with them, whereas adjectives are: 14 (44)
a.
Gianni e particolarmente bello handsome Gianni is particularly
200
Alessandra Giorgi
c. *Gianni e particolarmente a Parigi Gianni is particularly in Paris Such intensifiers are in various degrees compatible with the PPs illustrated above: (45)
Gianni e particolarmente in forma in good shape Gianni is particularly b. ?Gianni e particolarmente contro l'aborto Gianni is particularly against abortion c. ?Gianni e particolarmente a favore degli ecologisti Gianni is particularly in favour of ecologists d. ??Mario e particolarmente senza amlC1Z1e Importanti Mario is particularly without important relations a.
These facts can be informally interpreted in the following way: the more a PP has an 'adjectival' meaning, the more it can be modified by an intensifier; an adjective, in fact, specifies a quality which can be possessed in various degrees, whereas a location is usually a predicate of the type 'yes' or 'no' .15 Concluding these brief remarks, the principle in (17) must be rephrased as follows: 16,17 (46)
Locative prepositions do not assign an external O-role.
7 On some apparent counterexamples There are certain cases in which the distribution of anaphors and pronouns differs from the one we have just analysed. Here I will show how such cases are not exceptional, but simply constitute instantiations of different structures, without a small clause configuration. Consider the following examples: (47)
a.
b.
In sogno, Giannii riconciliava Maria con Tonino davanti in his dreams, Gianni reconciled Maria with Tonino in front of a se stesso/??a se/*a luii himself/in front of self/in front of him In sogno, Gianni riconciliava Mariai con Tonino davanti in his dreams, Gianni reconciled Maria with Tonino in front of a lei/*davanti a se stessai her/herself
In (47a-b), the distribution of anaphors and pronouns is exactly the opposite of the one considered in examples (1)-(3). Here, the preferred option for referring to the subject is the clause-bound anaphor, whereas the pronoun is ungrammatical and
Prepositions, binding and 8-marking
201
the subject-oriented anaphor is marginal. Conversely, the anaphor is starred if bound by the object, but the pronoun is acceptable. Consider also the following indexing: (48)
In sogno, Gianni riconciliava Maria con Toninoj davanti in his dreams, Gianni reconciled Maria with Tonino in front of a lui/*davanti a se stesSOj him/himself
If the prepositional object is the intended antecedent, the anaphor is starred and the pronoun is acceptable. This distribution can be easily explained by lack of ccommand: the prepositional object introduced by can 'with' does not permit ccommand by the embedded complement toward the PP. A similar explanation can be adopted for the previous cases too. It is reasonable, in fact, to attribute the following structure to the examples in (47):18 Infl'~
(49)
~NP
Infl'
--- ---- ---
PP
~
Infl
VP
A
V
sc
/ \PP
NP
The relevant PP is therefore interpreted as an adverbial and can probably be licensed by being a predicate of the subject NP, certainly not of the object. If (49) is the relevant configuration, the object NP cannot c-command into the adverbial, whereas the subject can. As a consequence, the anaphor is acceptable only if bound to the subject, and a pronoun only if disjoint from it. There is another piece of evidence which suggests that (49) is actually the structure for these sentences: as pointed out in Giorgi (1984), a subject-oriented anaphor embedded inside an adverbial cannot take the subject of the sentence immediately dominating the adverbial as a possible antecedent. In (47a) the occurrence of se is not fully grammatical, certainly contrasting with the distribution of se in examples (1-3) above. This fact could be explained by the configuration in (49), where se is embedded in an adverbial structure. This requirement can be found in many languages with LDAs, such as Icelandic (Maling (1982)), Norwegian (Hellan (1983)) and Japanese (Kuroda (1965)). The same happens in sentences such as (50):
202
Alessandra Giorgi
(50)
In sogno, Gianni mangiava delle mele davanti ??a set in his dreams, Gianni was eating apples in front of self/ davanti a se stesso/*davanti a luii in front of himself/in front of him
Finally, (51a-b) constitute other apparent counterexamples: (51)
a.
b.
Giannii recito
un PRO i discorso contro di se/contro se stesso/ *contro di luiiJ Gianni gave a speech against self/himself/him Giannii ascolto [NP un PRO j discorso contro di se/*se stesso/di luiil Gianni listened to a speech against self/himself/him [NP
Here, however, the PP is internal to the NP and the distribution of anaphors and pronouns can be attributed to coindexation with the PRO subject of this NP; see also Chomsky (1982, 1986a). Concluding this brief section, it seems that such cases cannot undermine my proposal, since they are likely to be instantiations of completely different configurations.
Notes 1. The distribution of anaphors and pronouns in similar constructions in English is slightly different. The most salient aspect is the variation, across speakers and lexical choices, of the acceptability of the anaphor, as exemplified by the following sentence: (i) Johnj pulled the blanket toward him/*-? himselfj where himselJfor some speakers is starred, like se stesso in Italian, but for others is almost acceptable. Dutch instead exhibits a distribution very similar to the Italian one, given that in such a language we also find an anaphor comparable for some aspects with Italian se (see Koster (1985, 1987), Everaert (1986a»: (ii) Janj trok de deken naar hem/zich/*zichzelfj Jan pulled the blanket toward him/self/himself French is apparently similar to Italian: (iii)
Victorj tire la couverture 11 lui/*lui-memej Victor pulls the blanket toward him/himself
But see Zribi-Herz (1981) for other, more problematic, cases. These phenomena require a detailed analysis of the anaphoric system in each language and some additional theoretical proposals which, for reasons of space, will be attempted in a larger version of this chapter.
Prepositions, binding and a-marking
203
2. Notice that an anaphor expressing the indirect object of a verb like restituire 'give back' can be bound both by the object and by the subject (see Giorgi (1986)): (i) a. b.
Una lunga terapia psicoanalitica ha restituito Mariaj a se stessaj a long psychoanalytic therapy restored Maria back to herself Giannij ha restituito Maria a se stessoj con il suo affetto paziente Gianni restored Maria back to himself with his patient love
In this chapter I will not consider verbs taking a direct and an indirect obj ect, since their thematic and structural properties are rather different from the phenomena we are considering here. The verb parlare 'talk' is another interesting case. To my knowledge, it is the only verb in Italian taking an indirect object and a complement introduced by di (which here means 'about'). The distribution of anaphors and pronouns, in my dialect, is as follows: (ii) a.
Gianni ha parlato a Mariaj di leij to Maria about her Gianni talked b. *Gianni ha parlato a Mariaj di se stessaj Gianni talked to Maria about herself c. Giannij ha parlato a Maria di se/se stessoj Gianni talked to Maria about self/himself d. *Giannij ha parlato a Maria di luij Gianni talked to Maria about him
Burzio (1987) also accepts (iib); in his paper he gives an interesting analysis ofthese and related phenomena. Under a slightly different perspective, a tentative proposal to account for the pattern in (ii) can be the following: (iii)
VP
V~PP
~~
V
I parlare
NP
P
I diI a Mariaj
NP
I
lei/*se stessaj
To predict the data in (ii) we must hypothesize: (1) that V' blocks c-command; (2) that a does not project a PP forming a relevant domain for c-command; (3) that the PP headed by di 'about' is not transparent to c-command. All of these claims seem rather plausible given that, with respect to the first, it has already been argued (Chomsky (1986a), Giorgi (1986)) that the notion of branching node makes more accurate predictions in binding phenomena than the notion of maximal projection; with respect to the second, it seems to me to be independently needed (cf. Giorgi (1986)), and the third is actually the 'unmarked' assumption. 3. In Giorgi (1987), the evidence analysed to define the notion of CFC comes from the distribution of anaphors and pronouns in two different domains: in NPs projected by nouns which have three arguments (possessor, agent and theme), as in (i), and in predicative constructions, as in (ii):
204
Alessandra Giorgi iI ritratto di se stessoil*; di Leonardoi di Ludovico iI Moro; the portrait of himself of Leonardo of Ludovico i1 Moro (ii) Gianni e i1 miglior giudice di se stesso Gianni is the best judge of himself (i)
The peculiarities of such constructions permit some theoretical insights concerning the nature of the domain relevant for principles A and B of the binding theory. For further discussion see Giorgi (1987). 4. In these structures containing a predicate AP, or NP, we are led to the conclusion that the same NP-position receives two a-roles, one originating from the verb and the other from the adjective or the noun. A similar situation is also found in the following structure (cf. also Chomsky (1986a)): (i)
Gianni e uscito nudo Gianni went out naked
On these structures, see Stowell (1987) and references cited there. At this point, one could ask which kind of a-role the AP and the NP are assigning to the subject of the SC. With respect to the AP in (9)-(12), we could say that their heads, contento 'happy' and soddisfatto 'satisfied', assign an experiencer a-role. The relationship between an NP such as presidente 'president' in (13) and its external maximal projection is less obvious. In certain cases the a-role of the predication subject is clearly agent. Consider for instance the following case: (ii)
Giannii e ti I'assassino di Mario Gianni is the murderer of Mario
Here Gianni is obviously the agent of the murder. This interpretation can be extended also to the following sentences: (iii)
Gianni e il presidente della Fiat Gianni is the president of Fiat
(iii), in fact, means that Gianni is the 'agent' who presides over Fiat. On predication see, among others, Williams (1980), Rothstein (1983), Higginbotham (1985). 5. The full interpretation principle proposed in Chomsky (class lectures, Fall 1986; 1986a) can be rephrased as follows: (i)
A maximal projection must be either an operator, or an argument, or a predicate.
Notice that in a small clause structure headed by a PP, we are hypothesizing a predication relation, without a-assignment. This situation is certainly not anomalous, given the existence of raising structures: (ii)
[s [NP Gianni] [vp sembra t essere partito]] Gianni seems to have left
The relation between NP and VP is barely a predication one, since the VP headed by the raising verb does not a-mark the subject NP. 6. Sentence (22a) is possible, even if pragmatically marked, when the verb rendere 'render' is interpreted as the homophonous 'give back':
Prepositions, binding and e-marking
205
(i) Ho reso il libro gave back the book 7. A stylistic variant of (25b) is the following sentence: (i)
Un buon corso di specializzazione fara di Gianni [NP il medico pill bravo the best physician a good specialization will make of Gianni della citta] in town ' ... will render Gianni the best physician in town'
The insertion of di is an interesting phenomenon to notice, even if, at the moment, we have no explanation for it. Notice also that APs are ungrammatical in this context: (ii) *Faro di te orgoglioso I will make of you proud
8. Considerare 'consider' and stimare 'esteem' do not admit a simple NP, exactly like ritenere 'believe' or rendere 'render': (i) a. b.
Consideravo Gianni *(stupido) I considered Gianni stupid Stimavo Gianni *(stupido) I esteemed Gianni stupid
Stimavo Gianni, by itself, can still be interpreted if the verb is taken to mean 'consider valuable'. An informal and tentative explanation for this fact can be the following: with stimare the predicate of the NP can either be overtly expressed or can be collapsed within the interpretation of the head V. If so, the semantic value of the predicate must be 'valuable' by default. Stimavo Gianni, therefore, can only mean that I have a positive opinion about him. Notice that verbs like supporre 'suppose' and credere 'believe' were almost grammatical even with an SC headed by a preposition: (ii) a. b.
(?) Supponevo Gianni a Parigi
I supposed Gianni in Paris Gianni a Parigi I believed Gianni in Paris
(?) Credevo
However, they do not admit a bare object, presumably they cannot 8-mark it: (iii)
a. *Supponevo Gianni I supposed Gianni b. *Credevo Gianni I believed Gianni
For a possible explanation of the data in (ii), see note 11 below. 9. Notice that also sembrare and risultare obligatorily require a predicate of the NP: (i) a. b.
Gianni Gianni Gianni Gianni
sembrava *(stupido) seemed stupid risultava *(stupido) resulted stupid
206
Alessandra Giorgi
If the predicate is not expressed, in fact, Gianni lacks a (J-role. 10. See also Longobardi (1983, 1985) and Higginbotham (1987) for discussion. Longobardi distinguishes two major copular structures: the predicational one and the identificational one, illustrated respectively by examples (i) and (ii). (i) 10 sono l'assassino I am the murderer (ii) L'assassino sono io the murderer am 'The murderer is me' These structures have different interpretative and syntactic properties. Here we are abstracting away from these differences, considering exclusively the predicational type, which is the one relevant to our purpose. 11. The potential presence of be is a very interesting topic, but one which we will not be able to consider here. The reader has presumably already noticed that an SC headed by a PP is not equally ungrammatical with all the verbs considered in the text. For instance, in (20) rendere 'render' is worse than ritenere 'believe', considerare 'consider', stimare 'esteem', etc. It is interesting to note that rendere is also incompatible with the insertion of be: (i) *Rendero essere illibro suI tavolo I will render the book to be on the table Such ungrammaticality is not just a matter of semantics, since, as we have seen in the text, fare 'make' has roughly the same interpretation, but is compatible with be. The following suggestion might connect these facts: in the intermediate cases we are marginally able to supply an empty copula. If the copula is not admitted, for independent reasons, as with rendere, we have strong ungrammaticality. The higher degree of grammaticality of small clauses headed by preposition with verbs like credere 'believe' and supporre 'suppose' could also be explained by hypothesizing the presence of a dummy be which can O-mark the subject. Cf. also note 8. 12. The verb stare is in some cases compatible with APs: (i)
II gatto sta buonoltranquillo/fermo the cat stays quiet/calm/still
However, this does not apply to every context: (ii) *I1 giornale sta interessante the newspaper stays interesting It seems that stare + AP is grammatical only if it can mean 'to stay in some place, in the state (usually quietness) specified by the adjective'. Apparently, the AP is a predicate of the subject, and the locative PP, subcategorized by the verb, is empty. Notice also that the 'location' can even be overtly specified. In this case the range of possible adjectives is wider: (iii)
a.
*Gianni sta malato Gianni is sick
Prepositions, binding and 8-marking
207
b. (?) Gianni sta malato all'ospedale Gianni is sick in the hospital 13. In the case of diverztare 'become', contrary to sembrare 'seem' and similar verbs, we cannot rely on independent evidence, such as the disappearance of the contrast when essere 'be' is inserted. Diverztare, in fact, like rendere 'render' does not admit it; in both cases the ungrammaticality is rather severe, much more than in (30c) or (31c). The data in (36b), therefore, could also be compatible with the idea that their ungrammaticality is due to subcategorization restrictions, which exclude PPs. This explanation, however, seems rather ad hoc, since it is obviously insufficient in the other cases discussed in the text; the best solution would be to state that diverztare too takes an SCcomplement and that the choice of the predicate is free, everything being equal. PPs would be excluded post hoc by the O-criterion. Notice also that from the semantic point of view, nothing is blocking an interpretation. Therefore, if it is accounted for just as a lexical idiosyncrasy, we should find variation among languages. This expectation, however, is not fulfilled: English: *John becomes in Paris French: *Jean devient a Paris German: *Hans wird in Paris
14. Malta 'very' or poco 'little' can also be used to modify the aspects of duration or frequency, which can be present also in a locative PP. Gianni i! malta a Parigi in questa periodo 'Gianni is very in Paris in this period', i.e. 'Gianni spends a lot of time in Paris', or 'He is often in Paris' etc. Cf. also Obenauer (1985) for the analysis of the distinction between intensive and non-intensive adverbs. 15. Notice also that some of these PPs have an adjectival counterpart: (i) a. b. (ii) a. b. (iii) a. b.
Gianni Gianni Gianni Gianni Gianni Gianni Gianni Gianni Gianni Gianni Gianni Gianni
e
contro I'aborto is against abortion e contrario all'aborto is unfavourable to abortion e a favore dell'aborto is in favour of abortion e favorevole all'aborto is favourable to abortion e senza amicizie importanti is without important relations e privo di amicizie importanti is lacking important relations
16. As expected, there is a certain improvement of the structures with diverztare 'become' as well, when there is such a non-locative preposition: (i) a. b.
(?)?Dopo la visita del Papa, Giannidivento improvvisamente contro l'aborto after the Pope's visit, Gianni became suddenly against abortion (?)?Dopo quel grande disastro in Cornovaglia, Gianni divento Gianni became after that great disaster in Cornwall,
208
Alessandra Giorgi improvvisamente a favore degli ecologisti suddenly in favour of ecologists
17. If our reasoning is correct, we should expect other differences between adjectives and locative prepositions (and between locative and non-locative ones) to emerge in various areas of grammar. This expectation is borne out in at least one other type of phenomenon. Longobardi (1983) noticed that in constructions with be and other copular verbs, a predicative NP can be pronominalized by means of the generalized pro-predicate 10 'it' , which is a third person masculine singular accusative clitic: (i) a. b.
10 I 10 I
sono/sembra/divento un assassino amlseemlbecome a murderer 10 sono/sembro/divento it-amlseemlbecome
Lo can also pronominalize VPs and APs:
10 sono uscito/io 10 sono have gone outiI it am 10 sono stancalio 10 sono I am tirediI it am
(ii) (iii)
Locative PPs in general cannot be pronominalized by means of 10: (iv) (v) a. b. c.
Gianni Gianni Gianni Gianni Gianni Gianni Gianni Gianni
e
a Parigil*Gianni 10 e is in Parisi Gianni it is sembra essere a Parigi seems to be in Paris sembra essere intelligente seems to be intelligent 10 sembra it seems (it = intelligentl*in Paris)
These facts too seem to correlate with our distinction: in PPs with an 'adjectival' interpretation, the structure with the elitic is more acceptable: (vi) a. 10 sono in buona formaiIo 10 sono I am in good shape/I it am b. Tu sembri contra l'aborto/?(Tu) 10 sembri you seem against abortion/you it seem c. ?Io sono diventato per l'abortol?(Io) 10 sono diventato I became for abortion/I it became d. ?Io sono a favore degli ecologistil? (10) 10 sono am in favour of ecologists/I it am The nature of these phenomena is not very clear; further research is needed to capture the relevant generalization and to express the correct principles. 18. This structure violates the binary branching constraint (Kayne (1984)) that I adopted in this chapter. However, the problem is more general than that, since it concerns the level of attachment of all kinds of adjuncts. Various solutions come to mind, but the question is too complex to be addressed within the limits of this chapter.
10 Locality, parameters and some issues in Italian syntax M. Rita Manzini
1 Introduction
The issues discussed in this chapter lie at the intersection of two lines of inquiry. The first line of inquiry can be characterized as a search for theories of locality under which all or some of the locality conditions can be given a unified formulation. One such theory, presented in Manzini (1988, 1989), is partially summed up in section 2. A crucial feature of the theory is that sentential subjects are generated in a non-VP-external position. This allows for the empty category principle (ECP) and binding theory to be unified under the same definition of locality, one that does not mention the notion of subject. The second strand of the discussion is related to recent theories of parametrization in binding theory. Section 2 briefly sums up the theory presented in Manzini & Wexler (1987) and in Wexler and Manzini (1987), according to which parametrization effects are due to the presence of a parameter, or multiple disjunction, in the definition of locality. The central proposal in this chapter is that the subject-based definition of locality in Chomsky (1981) and the non-subject-based definition under which ECP and binding theory are unified are not mutually exclusive; rather, the two definitions represent coexisting values of the locality parameter. Anaphors such as English himself, but also Italian se stesso, etc., are associated with the subject-based value. However, in sections 3 and 4, we will present evidence that the binding relations that characterize the reciprocal and inalienable possession constructions . in Italian are associated with the non-subject-based value. This, we argue, strongly supports our proposal, and indirectly the locality and parametrization theories on which it is based. The discussion will be concluded in section 5 by a brief examination of the theory in Pica (1987), which attempts to derive parametrization effects in locality without recourse to a locality parameter. We will show that under this theory the data of sections 3 and 4 are not derivable.
209
210
M. Rita Mallzini
2 Locality and parameters
In Manzini (1988,1989) it is argued that locality conditions, including subjacency, ECP and binding theory, can be given a unified formulation; and minimally that the notions of locality they refer to can be reduced to one notion. The first step in the argument concerns phrase structure theory. In particular, we propose that all subjects of propositions are adjoined to their predicates. Hence subjects of sentences are adjoined to their VPs, rather than being specifiers of IP as in Chomsky (1986b). If so, arguably, it becomes unnecessary to assume, as in Chomsky (1986b) again, that I is an independent head; instead we can assume that I is generated VPinternally, on V. We then obtain the basic sentential structure in (1): (1)
CP ~C'
---------
---------------
C
VP
NP
VP
---V' V+I
The crucial feature of (1) for our theory of locality is the non-VP-external position of the subject. In this respect the theory is reconcilable with other theories of phrase structure, such as that of Kitagawa (1986) or Sportiche (1988), that share this feature, though the non-VP-external position of the subject is assumed in those theories to coincide with the specifier of VP. It should be noticed that none of our proposals concerning locality is incompatible with a structure where I is a head, provided the non-VP-external subject hypothesis is maintained, as in (I'): (I')
CP
-----------
~C' C
IP ~,
VP
I
NP
VP
In particular, if we assume with Sportiche (1988) that the subject must systematically move from the non-VP-external position to the specifier of IP-position in order to receive Case, (1') re-creates the surface configuration in Chomsky (1986b). If so (I') also has the advantage of being immediately compatible with the theories concerning the position of adverbs etc. in Pollock (1989) or Chomsky (1988b); which is not the case for (1).
Locality, parameters and some issues in Italian syntax
211
Consider locality and, to begin with, ECP. Along the lines of Chomsky (1986b), we take ECP to require that, given any chain, the last element of the chain is head governed, or alternatively each trace in the chain is governed by its antecedent. We define a notion of g-marking, as in (2), so as to reconstruct the notion of government in Chomsky (1981); agreement, as referred to in (2), is taken to hold between a maximal projection, its head and its specifier as in Chomsky (1986b): (2) 13 g-marks ex iff 13 is a head and (i) 13 is a sister to ex; or (ii) 13 is a sister to 'Y and 'Y agrees with ex In terms of (2), we define a barrier for an element ex as a maximal projection that dominates ex and a g-marker for ex, or that dominates just ex if ex is not g-marked, as in (3). The second clause of (3) is necessary so as to take into account positions, such as adjuncts or derived adjunctions, that do not have g-markers, yet have locality domains: (3) 'Y is a barrier for ex iff 'Y is a maximal projection, 'Y dominates ex and (i) 'Y dominates a g-marker for ex; or (ii) ex is not g-marked Given (3), government is defined along the lines of Chomsky (1986b) as in (4): (4) 13 governs ex iff if'Y is a barrier for ex, 'Y does not exclude 13 If the definition of g-marking in (2) reconstructs government in the sense of Chomsky (1981), then (3), or at least (3i), reproduces the definition of governing category in Chomsky (1981), with the notion of maximal projection substituted for the notion of subject. Suppose then we formulate the binding conditions in terms of the notion of government in (4). Binding condition A will state that an anaphor ex must be both bound and governed by some 13; binding condition B that a pronominal ex cannot be both bound and governed by any 13. Crucially, if subjects of sentences are VP-external, binding conditions formulated in terms of government yield incorrect predictions. Consider for example an object ex. The first maximal projection that dominates ex and its g-marker, V, is VP. But VP excludes the subject. Hence the subject is predicted not to be a possible antecedent for ex if ex is anaphoric, and to be a possible antecedent for ex if ex is pronominal. Under a theory of phrase structure where subjects are not VP-external, however, the correct predictions follow. Given for instance (1), the barrier for an element ex in object position is again VP; but VP does not exclude the subject.
212
M. Rita Manzini
Hence the subject is correctly predicted to be a possible antecedent for ex if ex is anaphoric, and not to be if ex is pronominal. Consider then the subject, NP, in (1). In (1), C is a sister to VP. In turn, if! is finite, VP agrees with NP, its subject. Hence NP is g-marked by C. CP then is a barrier for NP in that it dominates it and its g-marker, C. It follows that NP cannot be bound from a CP-external position; but A'-binding of NP from the specifier of CP-position is allowed. In other words, we correctly predict that movement of a subject can take place through the specifier of CP and the specifier of CP only. On the other hand, there is no position in CP from which NP can be A-bound. Thus an A-bound anaphor is correctly predicted to be impossible in subject position. Consider then an ex in the specifier of CP-position. If CP is a sister to a head, say V, and CP agrees with its specifier ex, then V g-marks ex. It follows that CP is not a barrier for ex. From the specifier of CP-position, movement is then possible out of CP. In other words, the specifier of CP is correctly predicted to be an escape hatch for movement. From our discussion so far we conclude that the notion of government is shared by ECP and the binding conditions. In addition, the domain of application of binding condition A also subsumes the domain of application of ECP, if we accept a theory of empty categories, such as Aoun's (1985) or Brody's (1985), under which all empty categories, the domain of application of ECP, are anaphors. If so, binding condition A and the antecedent-government clause of ECP are seen to be essentially the same condition. Notice that binding conditions A and B can easily be expressed as one biconditional, stating that ex is an anaphor iff it is both bound and governed by some f3. The left-to-right reading straightforwardly corresponds to binding condition A; the right-to-Ieft reading corresponds to a statement empirically equivalent to B, by which if ex is not anaphoric, it cannot both be bound and governed. Hence the only disjunction in a unified formulation of binding theory and ECP remains the ECPinternal disjunction between the antecedent-government clause, now unified with the binding conditions, and the head-government clause. (5i) below provides the unified statement for the binding conditions and the antecedent-government clause of ECP; the other term of the disjunction, (5ii), is left open: (5) ECP/binding theory ex is an anaphor iff (i) there is some (ii) ...
f3 such that f3 binds and governs ex; or
(5ii) is not directly relevant for what follows. In Manzini (1988, 1989), (5i) is shown to subsume subjacency for those chains to which it applies. Thus the
Locality, parameters and some issues in Italian syntax
213
formulation of (5ii) that emerges is not just that certain chains, A' -chains whose last element is head governed, are not subject to (5i); but rather that these chains, and only these, are associated with subjacency, a weaker requirement. In Manzini (1988,1989) it is argued furthermore that subjacency is translatable into a government requirement; this effectively accomplishes the unification of all locality notions into one, the notion of government. So far our discussion has been based uniquely on sentence-internal positions, i.e. positions immediately dominated by either a projection of V or a projection of C, as in (1). The question then arises whether the results in (5) can be shown to hold for the other major group of positions, NP-internal positions. These are essentially 13 and a in (6), the subject and object position of a canonical English nominal: (6)
In the light of recent discussions of the internal structure of NP, notably English NP, as in Abney (1987), it is doubtful that (6) and not a more complex structure is to be postulated. We choose (6) for convenience; the results obtained for (6) should be extendable to a structure such as (6'), where D(et) is also a head: (6')
DP
-
------------D'
D
NP
-
~N' N
This is especially true if in (6') the position of the subject I?> is originally in the specifier of NP, with Case reasons impelling movement to the specifier of DP. If so, (6) represents not so much an alternative to (6') as a partial rendition of it. Consider the behaviour of an anaphor such as English himself in the position of a in (6). If the subject position 13 is realized, a must be bound within NP, as in (7): (7)
John loathes Peter's portraits of himself
If the subject position 13 is not realized, then if NP is itself in object position, a need not be bound within NP, but only in the immediately superordinate sentence, as in (8):
(8)
John loathes any portraits of himself
Data of the type shown in (7-8) provide an immediate counterexample to our
214
M. Rita Manzini
theory in (2-5). Under (3), NP is a barrier for himself in both (7) and (8), since it is a maximal projection that dominates himself and its g-marker, N. Under (5), then, himself is correctly predicted to be bound NP-internally by Peter in (7). (8), however, where himself has no NP-internal binder, is incorrectly predicted to be ill-formed. In fact, examples of the type shown in (7-8) strongly support a definition of locality incorporating the notion of subject, as in Chomsky (1981). Examples (7-8) would of course be accounted for if, while maintaining the notion of government based on (3) for movement, we reverted to a subject-based definition of locality for binding, thus giving up our unification of binding and at least the antecedent-government clause of ECP. Another possibility would be for us to seek a unified version of locality that incorporated the notion of subject, or some notion of opacity-creating element, and thus accounted for (7-8) in addition to the cases accounted for by (2-5). In what follows we will argue that the correct option is a third one which, contrary to the preceding two, leaves our discussion concerning (2-5) substantially unaltered. Binding theory, whether conceived of as unified with ECP or not, must be formulated so as to allow for parametrization. In this sense our theory in (2-5) is inadequate, quite independently of the problems in (7-8). Theories of parametrization in binding theory, including Yang (1984), Koster (1987), and Manzini & Wexler (1987), assume the correctness of the subject-based definition of locality in Chomsky (1981), and supplement it with a disjunction between the notion of subject and other opacity-creating elements. By analogy, we can add a disjunction between a number of opacity-creating elements to our definition of locality in (3), which mentions none. If this is the correct solution to the parametrization problem, the theory in (2-5) remains valid, though it is now seen to be part of a wider theory. What is important for our present purposes is that, given a parametrized conception of binding, our definition of locality in (3) and the subject-based definition need not absolutely exclude one another; rather, they can represent alternative values of the locality parameter. The theory that emerges then is one where a disjunction between a number of opacity-creating elements is added to the definition of locality in (3), which includes subjects. Now, let us assume that Manzini & Wexler (1987) and Wexler & Manzini (1987) are correct in hypothesizing that values of parameters are associated with lexical items. If so, we can assume that the English lexical item himself is associated with the value of the locality parameter involving the notion of subject. This immediately solves the problem in (7-8), while leaving the derivation of examples not involving himself unchanged in terms of the theory in (2-5). Notice that it is a fundamental feature of the theory in Manzini & Wexler (1987)
Locality, parameters and some issues in Italian syntax
215
and Wexler & Manzini (1987) that markedness features are assigned to values of parameters according to a principle, the subset principle, that takes into account the subset relations between the languages generated by the different values of the parameter. By this algorithm, the value of the locality parameter corresponding to our notion of barrier is less marked than the value corresponding to the subjectbased definition, at least for anaphors. To begin with, it is obvious that not all maximal projections have subjects. It follows that a maximal projection may be a locality domain for some element IX under our definition of barrier, but not under a subject-based definition. Vice versa every category with a subject will be a maximal projection. Hence if a given category is a locality domain for some element IX under a subject-based definition, it is also a barrier for IX in the sense of (3). It follows that whenever an antecedent-anaphor link satisfies government, i.e. it does not cross any barrier, it does not cross any subject-based locality domain either. However, an antecedent-anaphor link may cross a barrier and not cross a subject-based domain. Hence all sentences including anaphors that are grammatical under the definition of locality in (3) are also grammatical under the subjectbased definition, but not vice versa. Thus if markedness hierarchies are defined in terms of the subset principle, our notion of barrier represents a less marked value of the locality parameter than the subject-based value. Our notion of barrier represents then the unmarked value of the parameter altogether, since the subject value is unmarked with respect to the other values. Suppose we follow Manzini & Wexler (1987) and Wexler and Manzini (1987) in postulating that empty categories are associated with the unmarked value of a parameter. Empty categories are then predicted to fall under the value of the locality parameter represented by the definition of barrier in (3) without any need for further stipulation. Now, in what precedes we have assumed that the definition of locality includes a multiple disjunction, i.e. a parameter. We have furthermore assumed that the parameter includes at least two values, a value corresponding to our definition of barrier in (3) and a subject-based value. We have not however provided an actual formulation of the parameter. Consider first the notion of subject-based locality domain. In its simplest form, its definition can be rendered in our theory as in (9), if we accept that our notion of g-marking reconstructs the notion of government in Chomsky (1981) and the requirement that a barrier be a maximal projection is not parametrized: (9)
'Y is a locality domain for IX if 'Y is a maximal projection, 'Y dominates subject
IX
and a g-marker for
IX,
and 'Y has a
216
M. Rita Manzini
One difference between (9) and (3) is that (3) includes a disjunction that is not in (9). The disjunction itself can be easily reduced to a conditional statement that a gmarker for ex is required only if ex has one. As far as we can see the introduction in (9) of such a conditional statement or of the original disjunction in (3) is empiri-
cally void. The one difference then between (9) and (3) is that (9) mentions the notion of subject, where (3) does not. If so, the disjunction between (3) and (9) can be expressed in a parametrized definition of barrier as in (10). The obligatory core of (10) corresponds to (3); the presence of a subject is added to it as an optional requirement, to correspond to (9): (10)
'Y is a barrier for ex iff 'Y is a maximal projection, 'Y dominates ex,
if ex is g-marked, 'Y dominates a g-marker for ex (and (i) 'Y has a subject; or ... ) The disjunction left open in (10) will eventually accommodate other values of the parameter. In what follows we will argue in favour of the theory we have just presented on the basis of two sets of data involving reciprocals and inalienable possession constructions in Italian. The evidence that we will present will be directed at showing, first, that there are NP-internal anaphors which cannot be accounted for under the subject-based definition of locality that accounts for himself, se stesso in Italian, etc., but can be accounted for under our definition of locality in (3).
3 Reciprocals
The first set of data that we will take into consideration concerns Italian reciprocals. The Italian reciprocall'uno l' altro consists of two elements, l'uno (literally 'the one', 'each') and l'altro (literally 'the other'), which behave like separate lexical items. In particular I' altro is to be found in A-position; l'uno is to be found in A'position and behaves like a floating quantifier. L'altro can of course appear in all and only the usual A-positions. As for the positions in which l'uno appears, we can assume that they are independently determined for the class of floating quantifiers as a whole. However they are characterized, this need not concern us here. We can assume that like floating quantifiers in generall'uno is bound from Aposition and is then interpreted as quantifying over its antecedent. In other words, the configuration NP i ... l'unoi ('NP ... each'), where NP is plural, is interpreted roughly as l'uno di NP ('each of NP'). As for l'altro, we can assume that it is interpreted as having the (quantified) NP as its antecedent. In Sportiche (1988) a variant of the assumption that a floating quantifier is bound from A-position is
Locality, parameters and some issues in Italian syntax
217
suggested, under which the floating quantifier is left behind, together with a trace, by the movement of an NP to A-position; thus properly speaking it is the NP-trace which is A-bound. Another possibility is to assume that a floating quantifier is free at S-structure and only its movement at LF to take scope over NP, which is independently necessary, creates a trace and hence a binding relation. A system in any case emerges where two binding relations hold between an NP and the floating quantifier l'uno and between the same NP and the A-anaphor l'altro, though for the purposes of interpretation the floating quantifier in fact takes scope over the NP. In other words we have configurations of the type shown in (11): (11)
a.
b. l'unoj [NPj
I
I
RI ,
l'altroj I R2
••• tj ••• I
It is important to stress that our purpose is not to give an account of reciprocal constructions either universally or for the Italian type. What precedes is a set of working hypotheses that allow us to manipulate reciprocals within reasonably precise confines. Our purpose is to investigate what the locality constraints on the two anaphors l'uno and l'altro are. In particular we are interested in establishing the behaviour of l'uno and l'altro NP-internally. The question is whether they pattern with an anaphor like himself (or Italian se stesso) in requiring a subject-based definition of locality or whether they pattern as predicted by our definition of barrier. We will then start by considering examples of l'uno ... l'altro in NP-internal position directly. Thus we will avoid altogether problems involving sentential positions; for the difficulties involved see Belletti (1983). Consider an NP in the object position of a sentence. There are essentially four logical possibilities for NP-internall'uno ... l'altro, due first to the fact that l'uno can float either NP-externally or NP-internally, and second, to the fact that l'altro can be either the subject or the object of the nominal. If l'uno floats NP-externally, the sentence is well-formed with l'altro in either the subject or the object position of NP, provided NP is subjectless. Relevant examples are of the type in (12), where I' altro can be identified with either the object or the subject of ritratti: (12)
Quei pittori ammirano l'uno i ritratti dell'altro those painters admire each the portraits of the other
Example (12) obviously does not choose among locality domains for ['uno, which
218
M. Rita Manzini
is NP-external and presumably VP-internal. By our definition of barrier its locality domain is VP, and the sentential subject is correctly predicted to be a possible antecedent for it. The same prediction follows under a subject-based definition of governing category, in the obvious way. Consider l'altro. If its locality domain is subject-based, then this is identified by the subject of the sentence and l'altra is correctly predicted to be bound NPexternally, whether it is in object or subject position. Under our definition of barrier, however, at least if l'altra is in object position, its locality domain is NP, since NP dominates it and its g-marker N. Hence binding of l'altro by the subject of the sentence is predicted to be impossible. Thus examples of the type shown in (12) seem to argue in favour of a subject-based definition of locality for I' altra. If so, the prediction is that taking I' altra to be in object position in (12) and adding a subject to the nominal, an ungrammatical sentence is produced, since NP becomes the locality domain for l' altro under a subject-based definition as well. The prediction seems to be correct, as in (13). Examples of this type are provided in Belletti (1983); (13)
*Quei pittori ammirano l'uno i tuoi ritratti dell'altro those painters admire each your portraits of the other
Consider now the cases, crucial to the determination of the locality domain for l'uno, where this floats NP-internally. These are exemplified in (14); (14)
Quei pittori ammlrano 1 ritratti l'uno dell'altro those painters admire the portraits each of the other
Contrary to examples of the type in (12), which are generally judged well-formed, examples of the type in (14) give rise to contradictory judgements. According to Belletti (1983) they are altogether ill-formed. However, this does not in general correspond to the intuitions of native speakers. Our contention is that examples like (14) are in fact well-formed at least when l'altra is taken to be the subject ofthe nominal. In the same way examples of the type in (15) are well-formed, where the transitive noun ritratti is substituted by the intransitive noun colori, and l' altro must be in subject position, with l'uno again floating NP-internally; (15)
Quei pittori ammirano i colori l'uno dell'altro those painters admire the colours each of the other
In (14) and similar examples, furthermore, it is possible to interpret I'altro as the object; for if Cinque (1980) is correct, a genitive object of N can systematically move to become the subject of N, if the subject position is not otherwise filled. However, if l'altra is in an object position from which movement to the subject
Locality, parameters and some issues in Italian syntax
219
position is not possible, ungrammaticality results. This is the case with examples of the type in (16), where l'altro is a prepositional object, with l'uno still floating NP-internally: (16)
*Quei pittori amano Ie storie l'uno sull'altro those painters love stories each on the other
Thus if we are correct, l'uno can float NP-internally when l'altro is the subject of the nominal, but not when l'altro is the object ofthe nominal. This situation cannot be the result of some locality restriction on l'altro, since we have seen above that l'altro can occur in both positions when l'uno floats NP-externally. This is confirmed by the grammaticality not only of the counterparts to (IS), with l'uno floating NP-externally, as in (IS'), but also of the counterpart to (16), as in (16'): (IS') Quei those (16') Quei those
pittori painters pittori painters
ammirano l'uno i colori dell'altro admire each the colours of the other amano l'uno le storie sull'altro love each the stories on the other
Similarly, it is impossible to predict the fact as an effect of the locality constraints on l'uno if the locality domain is subject-based. For, under a subject-based definition, precisely in the case in which l'altro is in object position and there is no subject, the locality domain for l'uno is the whole sentence. Hence locality cannot block such cases. Suppose we take our notion of barrier as defining the relevant locality domain for l'uno. If l'altro is in object position, the relevant structure is as in (17), where we leave the attachment of l'uno open, assuming that it is nevertheless NP-internal; the preposition is omitted: (17)
Assuming that l'uno is in A' -position and it has no g-marker, the barrier for it under our definition in section 2 is of course NP, simply because it is a maximal projection that dominates it. L'uno, if anaphoric, must then have an antecedent within NP. Since in sentences of the type in (16), its antecedent, the sentential subject, is NP-external, we correctly predict ungrammaticality. Under the preceding account of the ungrammaticality of (16) it is also immediately predicted that an example where l'uno has as its antecedent the subject of the NP is well-formed. This prediction is confirmed by examples of the type in (18):
220
M. Rita Manzini
(18)
Ammiravo i loro ritratti l'uno dell'altro I admired their portraits each of the other
Consider then the case with l'uno floating NP-internally and l'altro in subject position. By the same reasoning as in (17), the barrier for l'uno is NP. Thus l'uno must have an NP-internal antecedent. This leads to the incorrect prediction that sentences of the type in (15) are ill-formed, given that the antecedent for l'uno is the sentential subject. Suppose we assume that the subject of an NP is either in the specifier of NP or coindexed with the specifier of NP. This means that in (15) the determiner is coindexed both with l'altro and with l'uno. In other words, we obtain structures of the type in (19); as usual we leave the attachment of l'uno unspecified, assuming that it is NP-internal: (19)
VP
NPj~VP V--------------NP
Det·~l'altro· ~
I
N
I
.. , l'UIlOj ...
Our binding theory correctly predicts the well-formedness of structures like (19), assuming that l'uno is sensitive to our notion of barrier. Of course l'uno is now bound NP-internally from the determiner position. On the other hand notice that NP is not a barrier for the determiner position. Indeed the determiner agrees with NP, of which it is the specifier, hence V g-marks it by agreement. VP then is the first barrier for the determiner, since it dominates it and its g-marker. Thus the theory correctly predicts that the determiner can be bound NP-externally. It follows that examples of the type in (15) are predicted to be well-formed with structures of the type in (19). In (16), finally, binding of ['uno through the determiner is presumably impossible because the determiner cannot be coindexed with l'altro or any other argument in object position. In fact sentences of the type in (14) admit of an interpretation with the reciprocal as the object of the nominal. This is the interpretation under which an implicit argument subject binds l'uno and l'altro; in other words, the portraying is reciprocal, not the admiring. Of course, this interpretation is irrelevant here, reducing essentially to that in (18), with an implicit argument substituted for the lexical pronoun. In summary, if we are correct, the contrast between sentences of the type shown in (15) and (16) in the subject and object interpretation of the reciprocal follows from binding theory, if the locality domain for l'uno is defined by our notion of
Locality, parameters and some issues in Italian syntax
221
barrier, and not by a subject-based notion. The additional assumptions that we have made are independently necessary assumptions concerning the reciprocal construction, and the assumption that the determiner is coindexed with the subject ofits NP. Notice that if the locality domain for l'uno is defined by our notion of barrier and not by a subject-based notion, another set of predictions follows. This is that l'uno cannot be embedded in a subject NP, even while l'altro is in subject position. Consider example (20), where for simplicity the intransitive noun colori is used, making only the relevant subject interpretation of the reciprocal possible: (20) *Quei pittori pensano che i colori l'uno dell'altro siano ammirevoli those painters think that the colours each of the other are admirable The ill-formedness of (20) is exactly as predicted under a barrier-based definition of locality. As before, we can assume that l'uno is bound by the determiner. In turn NP is not a barrier for its determiner. However, the immediately superordinate CP is a barrier for it. Thus the binding of the determiner by the matrix subject violates our barrier-based binding theory. Needless to say, if l'uno were associated with a subject-based definition of locality, we would predict incorrectly the well-formedness of (20). If, by contrast, l'altro is associated with a subject-based definition of locality, the counterpart to (20), with l'altro again internal to the subject NP and l'uno floating externally to the embedded CP, as in (21), is predicted to be well-formed: (21) Quei pittori pensano l'uno che i colori dell' altro siano ammirevoli those painters think each that the colours of the other are admirable Judgements as to the status of (21) are difficult. However, (21) does not give rise to intuitions of uninterpretability as (20) does. Finally, the data concerning l'uno reviewed so far could also be obtained on the basis of a subject-based definition of locality, if we were to assume that the determiner counts as a subject for locality purposes. If so, NP would of course be the locality domain for the object of the noun, and an anaphor in this position would only be well-formed if coindexed with the determiner itself. The problem with this is that under the same assumption we could not correctly derive the data concerning l'altro on the basis of a subject-based definition of locality. Nor could we derive the data involving himselfin English, as in (7-8), or similar data involving the reflexive se stesso in Italian.
222
M. Rita Manzini
4 Inalienable possession If our argument in section 3 is correct, the first element of the Italian reciprocal, l'uno, is associated with a locality domain defined by our notion of barrier and not by a subject-based notion. The same argument can be made with the binding relation that can be argued to characterize another Italian construction, the inalienable possession construction. If we are correct, this obeys locality constraints predictable on the basis of our notion of barrier, but not on the basis of a subjectbased notion of locality. Like the discontinuous reciprocal, the inalienable possession construction represents a complex subject of investigation in its own right. On the other hand, in this case the existence of the binding relation crucial to our argument cannot be simply postulated as in the case of the reciprocal; rather it needs itself to be established. Thus it is necessary for us to examine inalienable possession constructions first, to be able to proceed to our argument concerning locality. It must be stressed however, as in the case of the reciprocal construction, that our sole interest here is in locality, and in what does not concern locality we will aim simply at formulating reasonable working hypotheses. According to Gueron (1985), there are two types of inalienable possession constructions found in languages like French but not in English. The first type is exemplified with Italian in (22). In this example the subject of the sentence is interpreted as the possessor of an inalienable NP (a body-part) in object position: (22)
Maria alza la mano Maria raised the hand 'Maria raised her own hand'
In fact, in (22) the subject Maria need not be interpreted as the possessor of the body-part NP la mano; in principle a reading is also available under which Maria raises someone else's hand, not necessarily her own. This corresponds to the one interpretation that the literal translation of (22) admits of in English. The other inalienable possession construction found in French, and again in Italian, but not in English, is of the type in (23-4); this is the construction we will refer to as dative possession construction: (23)
(24)
Maria gli lava capelli Maria him washed the hair Maria washed the hair for him 'Maria washed his hair' Maria si lava i capelli Maria self washed the hair
Locality, parameters and some issues in Italian syntax
223
Maria washed the hair for herself 'Maria washed her (own) hair' As in the case of (22), a non-inalienable-possession reading is available in (23-4), under which the dative clitic is interpreted as a benefactive, and the possessor of the inalienable (body-part) NP is left undetermined. This interpretation is also available for the literal English translation of the sentences, where the benefactive is rendered by afar-phrase. Under the inalienable possession reading, on the other hand, the dative clitic in (23-4) is interpreted as the possessor of the inalienable NP. It is the construction in (23-4) that we will use for our locality argument. We will then try to establish the fundamental properties of this construction independently of the other. If we are correct, we do of course expect our explanation to extend to the construction in (22) as well. If dative possession constructions are in fact characterized by a dative clitic being interpreted in the determiner position of an NP, this can easily be formalized by saying that the dative clitic binds the determiner position and that the two form a chain, the determiner position supplying the a-role of the chain and the dative clitic the argument to which the a-role is ultimately assigned. In other words, examples of the type shown in (23-4) correspond to indexed structures of the form in (25-6): (25) (26)
Maria Maria Mariai Maria
glii him sii self
lavo washed lavo washed
ii the ii the
capelli hair capelli hair
For the chains in (25) and (26) to be well-formed it is necessary that their last element be in a-position. This requirement is satisfied by the determiner, presumably a position to which the subject a-role can be assigned. Conversely, it is also necessary that only the last position in the chain is a a-position. Thus our theory requires the dative clitic not to be in a-position. If therefore the dative Case is not associated with a a-role, this means that dative can function as a structural Case in Italian, in the sense of Chomsky (1986a). For this latter conclusion there is ample independent evidence. In particular, it is well known that in Italian the subject of a sentence can surface in the dative Case, when the sentence is embedded under a causative verb, as in (27); in this case, then, as in the dative possession one, the dative cannot be inherently associated with its a-role: (27)
Maria gli fece mangiare la torta Maria him-DAT made eat the cake 'Maria made him eat the cake'
224
M. Rita Manzini
The alternative interpretation of dative possession sentences as involving a benefactive can correspond on the other hand to the case in which the dative position is associated with a 8-role, in particular a benefactive 8-role. The fact that the dative possession interpretation is not available in English seems then to correlate with the fact that the dative Case is inherently associated with a 8-role. In other words, dative is not a structural Case in English. If we are correct so far, chains of the type in (25-6) contain only one 8-position, which is the last position in the chains. For the chain to be well-formed it is furthermore necessary that it contains only one Case-marked position, its initial position. This condition is obviously satisfied by both chains; the clitic is always Case marked, the determiner never is. Next, the chain must contain only one argument. As required, the only argument in the chains in (25-6) is the clitic; the determiner is clearly a non-argument. It is worth pointing out that while our examples all involve dative ditics there is nothing in our theory so far that predicts that the dative possession construction is possible with clitics only. This seems again to be correct. So (23-4) seem to have counterparts of the type shown in (28): (28)
Maria lava capelli a Piero Maria washed the hair to Piero 'Maria washed Piero's hair'
Notice that (28) is unambiguous; the non-clitic benefactive is expressed in Italian with a per (,for') phrase, and not with a dative, as in (29): (29)
Maria lava capelli per Piero Maria washed the hair for Piero
It is on the binding relation between the dative clitic and determiner that our argument concerning locality is based. Consider first the binding relation between the ditic and the determiner in (25) or (26). The determiner position is g-marked by V, since the determiner agrees with the NP it is the specifier of, and this is sister to a head, V. The first and crucial barrier for the determiner is then VP, since it dominates it and its g-marker. Under our binding theory this means that the determiner is required to have a binder not excluded by VP. This condition is trivially satisfied in (25-6); hence our theory correctly predicts (25-6) to be wellformed. The same prediction however can be achieved under a subject-based definition of locality domain. VP is the first category that contains the determiner and a subject; hence the binding of the determiner by the dative clitic VP-internally is predicted to be well-formed.
Locality, parameters and some issues in Italian syntax
225
It is possible however to construct sentences where the predictions of the two theories can be told apart. Consider for example (30-3): Maria mi ha sconvolto la vita the life Maria me has upset 'Maria upset my life' (31) *Maria mi ha sconvolto [la tranquillita [della vita]] Maria me has upset the quiet of the life 'Maria upset the quiet of my life' (32) Maria si e guastata la salute Maria self is undermined the health 'Maria undermined her own health' (33) *Maria si e guastata [la fonte [della salute]] Maria self is undermined the source of the health 'Maria undermined the source of her health' (30)
The examples in (30) and (32) parallel those in (23-4). Only the inalienable possession interpretation is indicated; the benefactive interpretation is in principle possible on syntactic grounds, but the examples have been chosen so as to make it practically impossible on pragmatic grounds. Like (23-4), (30) and (32) are wellformed; (31) and (33), however, where the inalienable possession NP is embedded in another NP, are not. Notice that if a possessive is substituted for the dative possession construction, (31) and (33) become well-formed, exactly like their English counterparts, as in (34-5): (34) (35)
Maria Maria Maria Maria
ha (has) ha (has)
sconvolto la tranquillita della mia vita upset the quiet of my life guastato la fonte della propria salute undermined the source of her health
If the definition of locality domain is subject based, the impossibility of the elitic binding the most embedded determiner in (31) and (33) is not explained. Neither of the two NPs in (31) and (33) has a subject, or at least a subject distinct from (not containing) the determiner; hence the only locality domain defined in (31) and (33) is the matrix VP. Under binding theory it should then be possible for the most embedded determiner to be bound by the dative elitic; and the dative possession interpretation should be possible, contrary to fact. Consider, however, our definition of barrier. The most embedded NP is not a barrier for its determiner, if the determiner is g-marked by the superordinate N. However, under the same assumption the superordinate NP is a barrier for it. Hence the binding of the
226
M. Rita Manzini
determiner by the dative clitic in (31) and (33) is correctly ruled out by our binding theory in that it crosses a barrier. In summary, the contrast between examples of the type of (30) and (32), and (31) and (33), can be explained only if the binding relation between the dative clitic and the determiner obeys our barrier-based theory of binding. Thus we have further support for the view that our definition of barrier and a standard subjectbased definition of locality represent two genuinely independent values of a parameter. Finally, if our theory of dative possession constructions is correct, we expect it to account also for the other inalienable possession construction typical of Italian, as exemplified in (22). As Gueron (1985) points out, this construction seems to be restricted to V-NP structures where the V and the NP are 'interpreted as a natural gesture'. Thus for instance the sentence in (36) does not admit of an inalienable possession interpretation, but is interpreted in Italian exactly as in English: (36)
Maria ha lavato la mano Maria has washed the hand
Since (36) is structurally identical to (22) and does indeed involve a body-part, the impossibility of the inalienable possession reading must depend on the verb. If, as under our theory, dative possession constructions involve chain formation between a dative and a determiner position, then inalienable possession constructions of the type in (22) must presumably involve the same type of chain formation between the subject of a sentence and a determiner. In other words, under the relevant interpretation, (22) must be associated with a structure of the type in (37): (37)
Mariai alzo lai mano Maria raised the (= her) hand
Locality here is no longer the issue. However, for the chain in (37) to be wellformed it is necessary to assume that Maria is not in 9-position; rather, the determiner is the 9-position of the chain. Another way of saying this is that alzare in (37) is really a verb 9-marking an object and not a subject; in other words, alzare shares the 9-marking properties of its unaccusative counterpart alzarsi. If we are correct in assuming chain formation as in (37), (37) should then be paraphrased by (38): (38)
La mano di Maria si alzo the hand of Maria itself raised ( = was raised)
If this or something like this is correct, no principled obstacle to a unified treatment of (37) and dative possession constructions would remain.
Locality, parameters and some issues in Italian syntax
227
Other requirements on the chain in (37) are trivially satisfied. The one Casemarked position is the chain-initial position; and there is only one argument, Maria. The restriction of inalienable possession constructions of this type to a certain class of verbs could finally depend on the independent availability of an unaccusative subcategorization frame. If so, our analysis of dative possession constructions could in fact be shown to advantageously extend to (37). If on the other hand the reason for the ungrammaticality of the English counterparts to (23) and (24) is the impossibility in English of structural datives, the reason for the impossibility of the English counterpart to (22) remains an open question. Certainly raise does not admit of an unaccusative construction, but the question is a more general one. Notice that the theory of inalienable possession that we are proposing here, based on the binding of a determiner by an NP in argument position, has at least one alternative, worked out in Gueron (1985) and Pica (1987). According to Pica (1987), in particular, inalienable possession constructions are characterized by NPs whose determiner does not saturate the head noun, in the sense of Higginbotham (1985). An unsaturated NP is then an anaphor by definition, and binding theory presumably applies to the NP as a whole. Our theory is perfectly compatible with the characterization of inalienable. possession constructions as lacking saturation of a noun by a determiner. What it is incompatible with is the assumption that the whole NP is anaphoric, and not the determiner. This is relevant for an argument that is made in both Gueron (1985) and Pica (1987) concerning the referential status of the NP. The argument is formulated in the following way: it is impossible for an inalienable noun to be associated with certain adjectives, say attributive or descriptive adjectives. This can be explained if the inalienable NP is non-referential, which is of course the case if it is anaphoric. Hence the data support the anaphoric status of NP. Notice that the data are far from straightforward. For example, (39), where the adjectives involved are of the relevant type, seems to be perfectly grammatical: (39) Le tagliarono i bei capelli neri 'They cut her beautiful black hair' Granting the data, however, we can attribute them directly to the fact that inalienable NPs are unsaturated, since this presumably means that they are referentially defective in some sense. Hence the data involved do not represent a principled problem for our theory. But suppose finally that following Gueron (1985) and Pica (1987) one takes the whole inalienably possessed NP to be an anaphor. IfNP is bound under a subjectbased definition of locality, then it correctly follows that binding must be sentence-
228
M. Rita Manzini
internal in simple examples of the type in (22-4). What does not follow, however, is that binding should not be possible when the inalienably possessed NP is embedded in another NP, as in our examples in (31) or (33). Hence even if the theory of inalienable possession constructions is modified in this way, our crucial argument, concerning locality, is left standing.
5 Parameters again If sections 3 and 4 are correct, reciprocal and inalienable possession constructions in Italian provide evidence that our definition of locality and the subject-based definition represent two values of a parameter. Now, the type of parametrization theory that we have been assuming so far, based on disjunctive statements of the type in (10), i.e. parameters proper, is not unchallenged. According to a proposal in Chomsky (l986a), lexical anaphors move at LF; binding theory, or ECP, holds of the (anaphor, trace) link, rather than of the anaphor and its antecedent. This proposal is expanded in Pica (1987) to take parametrization into account. According to Pica (1987), the long-distance binding effects with anaphors such as Icelandic sig, whose opacity-creating element in terms of a definition like (10) is an indicative Tense, follow from the fact that they are Xes and they move at LF from head to head. They can move out of a subjunctive sentence, but not out of an indicative sentence under the assumption that the C-position of a subjunctive, but not of an indicative, sentence can in general be empty at LF. This, according to Pica (1987), leaves us with anaphors of the type of himself, which are XPs and whose movement is strictly local. The strict locality of the movement follows then again from the fact that these anaphors move according to their categorial type. However, if XP-anaphors move according to their type, only two possibilities are open: either they move to A-position or they move to A'-position; and both options are highly problematic. If they move to A-position, it is indeed expected that they display strictly local binding effects; but in general there will be no A.position for them to move into. If they move to A'-position, then there is absolutely no reason why they couldn't move successive cyclically, thus producing longdistance binding effects once more. A similar problem characterizes the movement of XO anaphors. If it is necessary for an XO anaphor to move through a C that is empty at LF, how can it move through V and/or I positions that presumably are systematically filled? It seems that the theory of movement will have to be weakened to take these problems, which are quite fundamental ones, into account. Even so, the theory in Pica (1987) and the theory we are proposing here make different empirical predic-
Locality, parameters and some issues in Italian syntax
229
tions. Empirically Pica's (1987) theory is roughly equivalent to a locality parameter with only two values: a value under which subject is the opacity-creating element and a value under which the opacity-creating element is indicative Tense. We will leave aside the question whether intermediate opacity-creating elements must be postulated, as for example in Manzini & Wexler (1987). However, if our argument in sections 3 and 4 is correct, it is obvious that the theory in Pica (1987) becomes empirically inadequate for a new reason, since there are binding relations whose locality requirements are simply not explained under a subject-based definition of locality or a wider one. In general, a two-way opposition between head and phrasal movement cannot subsume a locality parameter which must be at least three-way, including, in addition to the two values on which Pica (1987) and Manzini & Wexler (1987) agree, the new value corresponding to our definition of barrier.
11 Long-distance binding in Finnish Marlies van Steenbergen
1 Summary In this chapter the anaphoric/pronominal system of Finnish will be discussed. A close look at Finnish anaphors and pronouns leads to the conclusion that the binding relations in Finnish can only be adequately accounted for if we assume that there are two kinds of binding domains, each with its own binding rules. Therefore, a theory of binding will be proposed which distinguishes between binding in a small domain, i.e. local binding, and binding in a larger domain, i.e. long-distance binding.
2 Introduction
I
In Chomsky (1981) a binding theory is proposed which states that an anaphor must be bound within the minimal maximal projection containing the anaphor, a governor for the anaphor and a subject or Agr accessible to the anaphor (called the governing category of the anaphor), while pronouns must be free within this domain. However, various languages, e.g. Icelandic and Italian, contain anaphors that can be bound by an antecedent outside this domain (cf. Yang (1984)). Although these anaphors obviously do not obey Chomsky's binding theory, they are very systematic in their behaviour, and hence have to be included in a general theory of binding. This raises the important question of how this is to be done: is long-distance binding (binding outside Chomsky's governing category) to be considered as local binding in a larger domain, or are there any fundamental differences between long-distance binding and local binding? An inspection of the Finnish anaphoriclpronominal system indicates the latter. The differences in binding possibilities between Finnish local anaphors and Finnish long-distance anaphors appear to suggest strongly that local binding is subject to other rules than
231
232
Marlies van Steenbergen
long-distance binding. These differences will have to be expressed in a general theory of binding.
3 The Finnish anaphoric/pronominal system 3.1 The possessive suffix Finnish possesses several anaphors: a reciprocal toiset 1 'each other'; two overt reflexives itse 'self' and han itse 'he self'; and an empty anaphor which I will refer to as e. The third person singular pronoun is hiin. All anaphors (and the genitive pronoun) occur with a possessive suffix, third person -ns[ii1 or -Vn, as illustrated in (1).2
(1)
a.
b.
c.
d.
e.
Hei rakastavat toisiansai they love each other-POSS 'They love each other' Pekkai naki itsensai Pekka saw self-POSS 'Pekka saw himself' Puhuin Pekallei hlinesta itsestaani spoke-Isg Pekka he self-POSS 'I spoke to Pekka about himself' Pekkai luki ei kirjaansai Pekka read e book-POSS 'Pekka read his book' Pekkai luki hlinenj kirjaansaj Pekka read he-GEN book-POSS 'Pekka read his book'
With the overt anaphors the suffix is attached to the anaphor, with the empty anaphor and the pronoun the suffix is attached to a following noun, a post-position or non-finite verb. The possessive suffix agrees in person and number with the antecedent. This suffix realizes the nominal inflection on these elements. As in Van Steenbergen (1987), we will assume that it spells out Agr, on a par with the Agr on tensed verbs, and structurally functions as the head. In order to bring out the parallel we will represent it as an I (with the associated projection). Of course this does not imply that nouns carry an Inft (marked for Tense, etc.) in the Chomsky (1981) sense. 3
Long-distance binding in Finnish (2)
a.
[NP hiinen kirjansa] he-GEN book-POSS 'his book'
233 b.
Pekka lauloi] Pekka-NOM sing-3sg-PAST 'Pekka sang'
[IP
I"
I"
~
~
I'
NP
~
I +Agr hiinen
I'
NP
~
N
I
\
-nsa
kirja
Pekka
I +Agr
VP
I
I laulaa
-i
The structure of the NP in (2a) is parallel to the structure of the tensed clause in (2b). The possessive suffix -nsa not only occurs on the heads of NPs, but also on the heads of PPs and the so-called clause reducts (IPs). Clause reducts are clauses containing an infinitive or participial verb and a 'subject' in the genitive Case in specifier position, just like the genitive subject within NPs (hiinen in (2a)). Clause reducts can substitute certain tensed clauses. These clause reducts have a structure parallel to that of NPs, and the possessive suffix can be attached to the non-finite verb: (3)
a.
Pekka niiki [IP hiinen tiskaavansa] he-GEN wash-up-PTC-GEN-POSS Pekka saw 'Pekka saw him washing the dishes'
b.
I"
~
I'
NP
~
I +Agr hiinen
I -nsa
VP \
tiskaavan
Notice the siInilarity between (3) and (2a). The suffix -nsa only occurs with a personal pronoun or the empty anaphor in specifier position, not with common nouns. I will assume that in the latter case Agr is not absent, but is realized as 0. Finnish exhibits limited pro-drop: first and second person pronouns in subject position of tensed clauses are generally omitted. The third person pronoun, how-
234
Marlies van Steenbergen
ever, cannot be left out. The same holds for the personal pronouns in specifier position of NP, PP and clause reduct. In general, first and second person pronouns are dropped while the third person pronoun is obligatory. 1 assume the presence of small pro, identified by Agr, when there is no overt pronoun. Occasionally, however, the third person pronoun must also be omitted. This is the case when it corefers with another c-commanding element in the sentence. 1 assume, on analogy with pro-drop, that in this case, too, an empty element occurs instead of the overt pronoun. (4)
a.
b.
Pekkaj luki hiinen*iI; kirjaansa Pekka read he-GEN book-POSS 'Pekka read his (another's) book' Pekkaj luki eil*; kirjaansa Pekka read e book-POSS 'Pekka read his (own) book'
If the book is Pekka's, the pronoun has to be omitted and instead an empty element occurs. As the third person pronoun is omitted if and only if it corefers with a c-commanding NP, the empty element in (4b) seems to be an anaphor: it must be bound according to strict rules, as we will see in the next section. 1 am not the first to posit an empty anaphor; the same has been proposed for Chamorro in Chung (1989). The empty anaphor must be considered an anaphoric counterpart to the pronominal small pro. 4 The empty anaphor e is in complementary distribution with the overt anaphor itse. Itse may occur in all argument positions except the specifier position oflP, NP or PP, while e can only occur in these specifier positions, being identified by the possessive suffix on the following head.
3.2 Binding of the Finnish anaphors We will now briefly discuss the Finnish anaphors to see how they must be bound. 5 We will start with the reciprocal lOiset. Toiset must be bound within the minimal maximal projection containing a subject, and may be bound by any c-commanding NP in this domain: (5)
a.
b.
Hej rakastavat toisiansaj they love each other-POSS 'They love each other' Kysyin ystiiviltiij toisistaanj asked-Isg friends each other-POSS 'I asked the friends about each other'
Long-distance binding in Finnish c.
d.
235
Ystavati nakivat [IP toistensai tiskaavan] friends saw each other-GEN wash-up-PTC-GEN 'The friends saw each other wash the dishes' Ystavati nakivat [IP tytti:ijenj katsovan toisiaan*ilj] friends saw girls-GEN watch-PTC-GEN each other-POSS 'The friends saw the girls watch each other'
In (sc) toistensa tiskaavan is a clause reduct containing Agr (zero in this particular case, cf. (3)). As toistensa can be bound by the matrix subject, it cannot be Agr that defines the binding domain for toiset. In (sd) toisiaan cannot be bound by the matrix subject, but has to be bound by the subject of the embedded tenseless clause reduct. (sc) and (sd) together indicate that it is the first c-commanding subject that defines the binding domain of toiset (tyttojen in (sd), ystiiviit in (sc)). I will call this domain the local domain. The binding possibilities of itse and e are identical. Both must be bound within the minimal maximal projection containing Tense, and they can only be bound by a subject. 6 (6)
a.
b.
c.
(7)
a.
b.
c.
Pekkai naki Matinj katsovan itseaanilj Pekka saw Matti-GEN watch-PTC-GEN self-POSS 'Pekka saw Matti watch himself' Pekkai naki etta Mattij katsoi itseaan*ilj Pekka saw that Matti-NOM watched self-POSS 'Pekka saw that Matti watched himself' *Puhuin Pekallei itsestaani spoke-lsg Pekka self-POSS 'I spoke to Pekka about himself' Pekkai naki Matinj lukevan [eilj kirjaansa] Pekka saw Matti-GEN read-PTC-GEN e book-POSS 'Pekka saw Matti read his book' Pekkai naki etta Mattij luki [e*ilj kirjaansa] Pekka saw that Matti-NOM read e book-POSS 'Pekka saw that Matti read his book' *Puhuin Pekallei lei kirjastaan] spoke-Isg Pekka e book-POSS 'I spoke to Pekka about his book'
In (6a) and (7a) itse and e can be bound by either the subject of the embedded tenseless clause or the matrix subject. In (6b) and (7b) the anaphors cannot be bound by the matrix subject because the embedded clause contains Tense (katsoi
236
Marlies van Steenbergen
and luki are third person past tense). Notice the difference between (Sd) on the One hand and (6a) and (7a) on the other hand: the binding domain of itse and e is larger than that of toiset. I will call the domain defined by Tense the long-distance domain. Notice also that in (Sb) toiset may be bound by a non-subject, while (6c) and (7c) illustrate that itse and e cannot be bound by a non-subject. Itse and e cannot always be bound by the first subject outside their local domain, as they are in (6a) and (7a). In (8) itse and e cannot be bound by the matrix subject Pekka. (8)
a.
b.
Pekkai lauloi Matinj puolustaessa itseiiiin*ilj Pekka sang Matti-GEN defend-INF-INESS self-POSS 'Pekka sang while Matti defended himself' Pekkai lauloi Matinj lukiessa e*ilj kirjaansa Pekka sang Matti-GEN read-INF-INESS e book-POSS 'Pekka sang while Matti read his book'
(8a) and (8b) are similar to (6a) and (7a) in that the anaphor occurs in object position of an embedded tenseless clause. However, whereas in (6a) and (7a) they can be bound by the matrix subject, this is not possible in (8a) and (8b). The relevant difference between the two sets of sentences appears to be that (6a) and (7a) contain a tenseless complement clause, while (8a) and (8b) contain a tenseless adverbial clause. Apparently, itse and e cannot be long-distance bound by the first subject outside their local domain when they are embedded within an adverbial clause. This distinction is also found in other languages containing long-distance anaphors. The generalization is that anaphors embedded within an adverbial clause cannot be long-distance bound by the subject of the minimal clause containing the adverbial clause (cf. Giorgi (1984), Maling (1984)). That this generalization only holds for long-distance binding and not for local binding is illustrated in (9) and (10): (9)
a.
b.
(10)
a.
Pekkai niiki ei tiskaavansa Pekka saw e wash-up-PTC-GEN-POSS 'Pekka saw himself wash the dishes' Pekkai lauloi ei tiskatessaan Pekka sang e wash-up-INF-INESS-POSS 'Pekka sang while he was washing the dishes' Ystiiviiti niikiviit toistensai tiskaavan each other-GEN-POSS wash-up-PTC-GEN friends saw 'The friends saw each other wash the dishes'
Long-distance binding in Finnish b.
237
Ystavatj lauloivat toistensaj tullessa maaliin friends sang each other-GEN-POSS come-INF-INESS finish 'The friends sang when each other finished'
In these sentences binding by the matrix subject is a case of local binding, as the matrix subject is the first c-commanding subject for e and toistensa respectively (the anaphors themselves occupy the subject position of the embedded clauses). (9a) and (lOa) contain a complement clause, (9b) and (lOb) an adverbial clause. In both cases local binding by the first subject outside the clauses is possible. Apparently, the adverbial-complement distinction is only relevant in cases of long-distance binding, not in cases of local binding. Like itse and e, han itse must also be bound within the minimal maximal projection containing Tense, i.e. the long-distance domain. It differs from itse and e, however, in that it can only be bound by a non-subject in the local domain (see (lIa, b)), while it can only be bound by a subject outside the local domain but inside the long-distance domain (see (lIc)): (11)
a. *Pekkaj puolusti hanta itseaanj Pekka defended he self-POSS 'Pekka defended himself' b. Puhuin Pekallej hanesta itsestaanj spoke-Isg Pekka he self-POSS 'I spoke to Pekka about himself' c. Pekkaj sanoi Jusillej Matink katsovan hanta itseaanil*j/*k Pekka said Jussi Matin-GEN watch-PTC-GEN he self-POSS 'Pekka said to Jussi that Matti watched himself' d. Pekkaj naki etta Mattij katsoi hanta itseaan*iI*j Pekka saw that Matti-NOM watched he self-POSS 'Pekka saw that Matti watched himself'
(lId) illustrates that hiin itse cannot be bound outside the long-distance domain. Notice that all Finnish anaphors that can be long-distance bound can only be longdistance bound by a subject, never by a non-subject. Binding by a non-subject is only possible with local binding (compare (lIb) and (Uc)). This phenomenon can be found in many other languages as well. The pronoun han, finally, must be free from any c-commanding NP within the minimal maximal projection containing a subject, i.e. the local domain: (12)
a. *Pekkaj katsoi hantaj Pekka watched him 'Pekka watched him'
238
Marlies van Steenbergen
b. *Kysyin Pekaltai hanestai asked-Isg Pekka him 'I asked Pekka about him' c. Pekkai naki Matinj katsovan hantai/*j Pekka saw Matti-GEN watch-PTC-GEN him 'Pekka saw Matti watch him' In (12c) hanta can refer to Pekka because Pekka is outside the local domain (Matin is the first c-commanding subject for hanta). Notice that (l2c) is similar to (6a): both itse and han in object position of an embedded tenseless complement clause can refer to the matrix subject. In other words, there is no complementary distribution between pronouns and long-distance anaphors. The domain in which itse, han itse and e can be bound is larger than the domain in which han must be free. Outside the local domain the occurrence possibilities of long-distance anaphors and pronouns overlap. This overlap between pronouns and long-distance anaphors, too, can be found in many other languages as well. 7 Summarizing, the Finnish anaphoric!pronominal system is as follows: (13)
anaphor toiset itse e han itse pronoun han
domain defined by subject Tense Tense { subject Tense domain defined by subject
antecedent c-commanding NP subject subject { non-subject subject not referring to c-commanding NP
From these Finnish anaphoric relations we can draw the following conclusions, conclusions supported by other languages (cf. Thrainsson (1976a), Anderson (1983), Giorgi (1984), Maling (1984) etc.): (14)
a. Long-distance anaphors must be bound by a subject; many local anaphors can be bound by any c-commanding NP. b. A long-distance anaphor embedded within an adverbial clause cannot be bound by the subject of the minimal clause containing the adverbial clause. • c. The negative domain of pronouns is local, and outside the local domain there is no complementary distribution between pronouns and longdistance anaphors.
Long-distance binding in Finnish
239
In the next section an attempt will be made to incorporate these generalizations into a general theory of binding.
4 A theory of long-distance binding Looking at the Finnish data we saw that in the anaphoric binding relations in Finnish two domains are involved, a domain defined by the subject and a domain defined by Tense. I called these the local domain and the long-distance domain respectively. Many languages distinguish between a local and a long-distance domain for binding. In these languages, too, we find a number of differences between binding in the one domain and binding in the other domain. In general, binding within the long-distance domain seems to be subject to more restrictions than binding within the local domain: while in the local domain antecedents need only c-command the anaphor they bind, in the long-distance domain the antecedent must in addition be a subject; while in the local domain no distinction is made between complements and adverbials as far as binding is concerned, in the long-distance domain such a distinction is made. Thus, in the local domain there are no restrictions but the requirement that the antecedent must c-command the anaphor, while in the long-distance domain the structural positions of antecedent and anaphor relative to each other are subject to more conditions. One way to express restrictions on the structural relation between two elements is to assume a chain of elements linking the two. The structural restrictions on antecedents could be expressed in terms of restrictions on possible chains between antecedent and anaphor. Such an approach is chosen in Everaert (1986a). There it is assumed that binding of an anaphor by a particular antecedent is only possible when a government chain (a chain consisting of successive governors, starting with the governor of the anaphor and ending with the governor of the antecedent) can be established between anaphor and antecedent. By positing various conditions on such government chains, Everaert can account for the fact that long-distance antecedents must be subjects (in fact he assumes that all antecedents must be subjects), that Tense defines the long-distance domain, and for the behaviour of long-distance anaphors in adverbial clauses. However, Everaert assumes that in all cases of binding, local and long-distance, a chain between antecedent and anaphor must be built. As a consequence, it will be difficult to account for the fact that long-distance binding distinguishes between binding into complement clauses and binding into adverbial clauses, while local binding does not. He accounts for the first fact by positing such conditions on chains that a chain starting in an adverbial clause cannot be extended into the next clause (it has to skip a clause). This works all right for long-distance binding. But if
240
Marlies van Steenbergen
local binding also happens by way of chain building, how will it be possible for a chain to be built out of an adverbial clause in cases of local binding if it is not possible for long-distance binding? In other words, how does Everaert account for the difference between (8b) and (9b) if in both sentences chain building is involved? Everaert (1986a) does not discuss these cases of local binding, but it is difficult to perceive how he would solve this dichotomy in an elegant way. Koster (1987) also includes chains in his theory of binding, his so-called dynasties (also chains of governors). However, he does not use dynasties to account for the difference made between complement clauses and adverbial clauses by longdistance antecedents, but simply defines long-distance anaphors as anaphors that are bound by a non-minimal governing subject, where a governing subject for an anaphor is the subject of an XO such that XO governs the anaphor or a domain containing the anaphor. This definition does not really explain anything, nor does it relate the various generalizations concerning binding. Still, the introduction of chains of elements between antecedent and anaphor seems a very straightforward way to express structural restrictions on the two. However, as such structural restrictions occur only in cases of long-distance binding, it seems feasible to restrict a mechanism of chain building to long-distance binding only, in contrast to Everaert (1986a). In local binding there is no indication whatsoever that antecedent and anaphor must be linked by a chain. Therefore, I want to propose an Everaert-like mechanism of binding, but only for a subset of all binding cases. This leads to the following picture of binding in general: Local binding happens directly, without mediating chain, the anaphor being directly bound by a c-commanding antecedent. This is the unmarked case of binding. All languages contain local binding, thus local binding is presumably part of universal grammar. Long-distance binding, however, is another matter. Not all languages have the possibility of long-distance binding, hence this kind of binding seems to be more marked. I assume that it is an extra option for some anaphors to find an antecedent outside their local domain. As such an antecedent is at quite a distance from the anaphor, it cannot simply bind the anaphor directly, but has to be linked to the local domain of the anaphor. The local domain of the anaphor must, as it were, be made accessible to the antecedent outside. This can be done by building a chain of elements between local domain and antecedent. This chain must link the antecedent to the binding domain of the anaphor. Let's assume that such a chain consists of Inff-nodes, starting with the head of the local domain of the anaphor and ending with the Inff-node governing the antecedent. We could express the formal definitions of long-distance binding by way of chains as follows: 8
Long-distance binding in Finnish (15)
(16)
241
Long-distance binding An anaphor can be bound by an antecedent outside its local domain iff an 1chain can be built, the foot of which is the head of the local domain of the anaphor and the head of which contains the antecedent in its governing domain.
I-chain C = (10, ... , In) is an I -chain iff Ii strictly c-commands Ii + 1 (0 ~ i < n) and Ii (0 < i ~ n) is not [+(indicative) Tense].
If we adopt this rather simple mechanism of chain building the generalizations found in section 3 follow straightforwardly. The fact that only subjects can be long-distance anaphors is a consequence of the fact that only subjects are governed by Inff. As a chain consists of Inff-nodes only, and the antecedent must be governed by the head of the chain, only subjects can ever be long-distance antecedents. Only subjects are related closely enough to the chain to make use of it. The distinction made between complement clauses and adverbial clauses in cases of long-distance binding follows from the different positions in which the two kinds of clauses are attached. I assume that complement clauses are attached under the VP and hence the head of these clauses can be connected to the Inff of the matrix clause to form a chain. Adverbial clauses, however, are attached under I', and their head is not c-commanded by the head of the matrix clause, hence no chain can be built (cf. Everaert (1986a)): (17)
a.
Pekkai naki Matinj katsovan itseaanilj Pekka saw Matti-GEN watch-PTC-GEN self-POSS 'Pekka saw Matti watch himself'
b.
(18)
a.
Pekkai lauloi Matinj puolustaessa itseaan*ilJ Pekka sang Matti-GEN defend-INF-INESS self-POSS 'Pekka sang while Matti defended himself'
242
Matlies van Steenbergen
lauloi
Matinj
puolustaessa
In both (17) and (18) the local domain for itse is the embedded clause, Matin katsovan itseiiiin and Matin puolustaessa itseiiiin respectively. Within the local domain itse can be directly bound. Thus in both (17) and (18) itse can be bound by Matin. Binding by Pekka, however, would be a case of long-distance binding and requires a chain between the local domain of itse and Pekka. In (17) such a chain of Infl-nodes can be built, as 10 strictly c-commands II> the head of the local domain. Because Pekka is in the governing domain of 10 , the head of the chain, Pekka can bind itse. In (18), however, 10 does not strictly c-command 11. Hence, no chain can be built and there is no way in which Pekka can bind itse. In local binding no distinction between adverbial clauses and complement clauses is made, because in local binding no chain building is involved: local binding happens directly, without a mediating chain, as long as the antecedent ccommands the anaphor. Thus, the chain-building mechanism proposed here makes the properties particular to long-distance binding follow straightforwardly. Besides, the theory developed here also gives an explanation of the fact that the negative domain of pronouns is always local. We assumed that local binding is the unmarked case of binding, while long-distance binding is a marked option available to some anaphors to find an antecedent outside their local domain. The chain-building mechanism of long-distance binding is a kind of escape-hatch for anaphors to find a necessary antecedent. Pronouns, however, do not need an antecedent. Therefore, the mechanisms of long-distance binding are totally irrelevant to pronouns. It is a mechanism for anaphors, not for pronouns. Pronouns just need to be free within the binding domain defined in universal grammar, i.e. the local domain. Summarizing, by distinguishing between local binding and long-distance bind-
Long-distance binding in Finnish
243
ing and by assuming a chain-building mechanism for long-distance binding only, we accounted for the several generalizations found in section 3 in the following way: only subjects can be long-distance binders, because only subjects can be in the governing domain of chain members; long-distance binding of an anaphor in an adverbial clause by the first subject outside the clause is impossible because no chain can be built; and the negative domain of pronouns is local because longdistance binding has nothing to do with pronouns but is only relevant to anaphors.
5 Conclusion In this chapter the anaphoric/pronominal system of Finnish has been discussed briefly. The data considered clearly pointed to a binding theory distinguishing between local binding and long-distance binding. A sketch of what such a theory might look like has been given: local binding happens directly, long-distance binding involves the linking of antecedent and local domain by a chain of Inflnodes. We saw that such a theory of binding could account for the fact that longdistance antecedents must be subjects, for the distinction made between complement clauses and adverbial clauses in cases of long-distance binding, and for the fact that the negative domain of pronouns is local.
Notes -1. In fact, there are three reciprocals: toiset 'others', toinen toiset 'other others' and toinen toinen 'other other'. As all three are subject to the same binding rules I will only discuss the form lOiset. 2. I will use the following abbreviations in the glosses:
GEN INESS INF NOM POSS PTC
genitive Case inessive Case infinitive nominative Case possessive suffix participle
3. This approach differs from that ofVainikka (1985) in which it is assumed that the suffix -nsa is the anaphor rather than the element to which the anaphor is attached. However, if -nsa is an anaphor instead of agreement it is difficult to account for the fact that its binding possibilities depend on the element it is attached to. If it occurs on lOiset, its binding possibilities are not the same as when it is attached to itse. It seems to me that lOiset and itse are the anaphors and that -nsa is nothing more than the spelling out of Agr. 4. If one is prepared to accept the idea that, like the subject position of tensed clauses, the specifier position of NPs, PPs and IPs containing Agr must be filled, either by an overt element or by an empty element, one must also assume that the empty element in (4b) is an anaphor. It must have an antecedent and it cannot have a split antecedent:
244 (i)
Marlies van Steenbergen a. *Pekkai kiiski Matin; antaa ei +; kirjansa Liisalle Pekka ordered Matti-GEN give-INF e book-POSS Liisa 'Pekka ordered Matti to give their book to Liisa' b. Pekkai kiiski Matin; antaa heidiini +; kirjansa Liisalle Pekka ordered Matti-GEN give-INF they-GEN book-POSS Liisa 'Pekka ordered Matti to give their book to Liisa'
Thus the empty element seems to be anaphoric. However, it is not likely to be PRO, as it occurs in a governed position and never receives arbitrary interpretation. Moreover, it has its own independent a-role, hence it cannot be an NP-trace either. The only possible conclusion is that it is a non-overt full-ftedged anaphor. 5. For a more extensive discussion of the Finnish anaphors and pronouns see Van Steenbergen (1987). 6. There seem to be some exceptions to this rule: (i) a.
b.
Psykoanalyytikkoi totutti potilaan; itseensiiil; Psychoanalyst accustomed patient self-POSS 'The psychoanalyst accustomed the patient to himself' Liihetimme rikolliseni ei kotimaahansa sent-lpl criminal e mother country-POSS 'We sent the criminal to his mother country'
However, I assume that these sentences contain small clauses and that as a consequence the antecedents of itse and e are in fact subjects (cf. Vainikka (1985)). 7. There seems to be an exception to the rule that han must be free from any c-commanding NP. When han occurs in specifier position, it need only be free from the subject: Vapautan Pekani hiineni iiidistiiiin free-lsg Pekka he-GEN mother-POSS 'I free Pekka from his mother' I want to argue, however, that this sentence does not contain the pronoun hanen, but the anaphor hanen e. In other words, han itse has a counterpart han e, which, like e, occurs in specifier position, whereas han itse, lik itse, occurs in other positions. 8. The mechanism of chain building presented here has much in common with the mechanism presented in Everaert (1986a). The main difference is the reduction of the chains to Inft-nodes only. By restricting the chains to Inft-nodes, the domain-defining property of Tense (which is situated under Inft) and the necessity of a subject as antecedent follow automatically. All restrictions reside in Inft, because the chain only consists of Inft-nodes. Also, the chains presented here start with the head of the local domain of the anaphor, not with the governor of the anaphor. Thus local and long-distance binding are linked. The chains make it possible for the antecedent to reach into the binding domain of the anaphor. For more examples of how the definitions given here work for the various binding cases see Van Steenbergen (1987).
12 The primacy condition of anaphora and pronominal variab Ie binding E. Kiss Katalin
1 Introduction It has been known for some time that the primacy condition of anaphora assumed in standard binding theory, requiring c-command of the anaphor by the antecedent, is insufficient in non-configurational languages, in which coarguments mutually c-command each other (cf. E. Kiss (1981), Hale (1983), Mohanan (19834), Manicz (1986)). The c-command condition of anaphora is also inadequate in certain constructions of configurational languages, e.g. English. Among other inadequacies, it is too permissive in licensing anaphoric relations between coarguments occupying hierarchically equally prominent positions. As this chapter will point out, the standard c-command condition on binding is also insufficient in licensing pronominal variable binding. Its failure is, again, most extensive in non-configurational languages, but the constraint also breaks down in various constructions of configurational languages. This chapter proposes a primacy condition of binding which adequately constrains both antecedent-anaphor relations and operator-pronominal relations in configurational and non-configurational languages alike. The proposed primacy condition involves three interacting primacy principles: S-structure c-command, S-structure precedence, and precedence in a thematically motivated lexical argument hierarchy.
2 Anaphor binding Principle A of the binding theory formulates the following requirement for anaphors: (1) An anaphor is bound in its governing category.
245
246
E. Kiss Katalin
(See Chomsky (1981), and, for a more complex formulation of the locality condition of anaphora, Chomsky (l986a).) The primacy condition of anaphora is specified in the definition of the term 'bound': (2)
a is bound by b iff a and bare coindexed, and be-commands a. 1
The above formulation of binding principle A and of the notion 'bound' predicts that in a fiat, non-configurational, VP-Iess sentence structure, in which the arguments mutually c-command each other, any argument can bind any of its coarguments. If this prediction were borne out, in non-configurational Hungarian, for example, then both members of the following sentence pairs would be grammatical: (3)
a.
b.
(4)
a.
b. (5)
a.
b. (6)
a.
b. (7)
a.
b.
A hinyok ismerik egymast the girls-NOM know each-other-ACC 'The girls know each other' *A lanyokat ismeri egymas the girls-ACC know each-other-NOM 'Each other know the girls' A lanyokat kibekitettem egymassal the girls-ACC conciliated-I each-other-with 'I reconciled the girls with each other' *A hinyokkal kibekitettem egymast 'I reconciled each other with the girls' A lanyokat meg6vtam egymast61 the girls-ACC protected-I each-other-from 'I protected the girls from each other' *A hlnyokt61 meg6vtam egymast 'I protected each other from the girls' A lanyokkal vitatkoztam egymasr61 the girls-with argued-I each-other-about 'I argued with the girls about each other' *A hinyokr61 vitatkoztam egymassal 'I argued about the girls with each other' A lanyokt61 reg nem hallottam egymasr61 the girls-from long not heard-I each-other-about 'I have not heard from the girls about each other for a long time' *A lanyokr61 reg nem hallottam egymast61 'I have not heard about the girls from each other for a long time'
The primacy condition of anaphora and pronominal variable binding
247
As the ungrammaticality of the (b) sentences (3-7) indicates, binding principle A makes the wrong prediction: anaphoric relations between arguments mutually ccommanding each other are not 'bidirectional'. There is only one exception: not only can an accusative NP bind a dative NP, but a dative NP can also bind an accusative NP, as shown in (8). (8)
a.
b.
A hinyokat megmutattam egymasnak the girls-ACC showed-I each-other-DAT 'I showed the girls to each other' A lanyoknak megmutattam egymast the girls-DAT showed-I each-other-ACC 'I showed the girls each other'
Binding principle A fails to rule out all ungrammatical anaphoric relations in English, too, if the two arguments involved in the anaphora are structurally equally prominent, i.e. if they are both VP-internal PPs or NPs. Thus it cannot predict the distribution of grammaticality in (9-11), since, under the assumption of V + P reanalysis, the antecedent and the anaphor satisfy both the c-command requirement and the locality requirement both in the grammatical (a) sentences and in the ungrammatical (b) sentences. (9) a. b. (10) a. b. (II) a. b.
1 argued with the girls about each other *1 argued about the girls with each other 1 have not heard from the girls about each other for a long time *1 have not heard about the girls from each other for a long time 1 showed the girls each other *1 showed each other the girls
The fact that in English a direct object cannot bind an indirect object contrasts with the observation that in Hungarian the anaphoric relation between an accusative NP and a dative NP is bidirectional. Our account of binding will explain this difference - see the discussion of (25) below. The inadequacy of binding principle A in non-configurational languages has triggered explanations along two lines. According to Maracz (1986) and others, the locus of anaphora interpretation in non-configurational languages is an abstract, 'virtual', hierarchical level of representation called lexical structure, the articulation of which is similar to the articulation of phrase structure in configurational languages. The advocates of this view seem to ignore the fact that the c-command condition of anaphora is also problematic when interpreted on a configurational sentence structure. 2 E. Kiss (1981, 1987a) suggests that in the case of anaphoric relations between
248
E.
Kiss Katalin
arguments mutually c-commanding each other, the primacy condition of c-command is to be supplemented with the auxiliary device of a Case hierarchy, too. The proposed Case hierarchy establishes the following order among coarguments: NPacc} 3 (12) NP nom> { NP dat > NPinstr > NP delat The case hierarchy is, in all probability, more articulated than is represented in (12). The ablative NP, for instance, certainly also represents a separate level of the hierarchy. The sentences in (5) and (7) indicate that it is located between the accusative NP and the delative NP. Its position relative to the instrumental NP, however, is unclear. In E. Kiss (1987a) the definition of 'bound' is extended as follows: (13) a is bound by b iff a and bare coindexed, and b is more prominent than a (14) b is more prominent than a iff (i) b c-commands a, and (ii) if a also c-commands b, b precedes a in the Case hierarchy The problem with the Case hierarchy in (12) is that the argument hierarchy it establishes in languages like English, which distinguish only two or three Cases, is not more articulated than the argument hierarchy established by the primacy notion of c-command. The English anaphoric constructions that binding principle A cannot handle have evoked somewhat different explanations in the literature. Barss & Lasnik (1986), realizing the failure of the c-command condition in the licensing of anaphora in double object constructions, propose also to include 'precedence' in the primacy condition of binding. They call the notion combining c-command and precedence 'domain of', which they define as follows: (15) Yis in the domain ofx iff x c-commands y and x precedes y The primacy notion 'domain of', however, can only predict the distribution of grammaticality in English double object constructions. It cannot account either for the Hungarian facts of anaphora, e.g. the data in (3-8), or the types of English anaphora involving two PPs, e.g. the data in (9-10), since the antecedent precedes the anaphor both in the grammatical and in the ungrammatical versions of these constructions. Nor would the - conceptually rather unattractive - assumption that precedence is to be checked at D-structure save the role of the notion 'domain of' in licensing anaphora. There is no independent evidence on the basis of which the arguments in the Hungarian sentence, or the VP-internal PPs in the English sentence, could
The primacy condition of anaphora and pronominal variable binding
249
be attributed a fixed order in D-structure; consequently, there is no independent evidence on the basis of which, for example, the anaphors in the grammatical (a) sentences of (3-11) could be linked to traces following their antecedents. According to Jackendoff (1972) and others, the primacy condition ofanaphora is the prominence of the antecedent over the anaphor in the following hierarchy of a-roles: (16) agent> source, location, goal> theme 4 There are, however, considerations which seem to argue against this view. For instance, the arguments involved in the anaphoric relation may have the same aroles in both members of the sentence pairs (17) and (1S) (beneficiary and theme in (17a) and (l7b), and source and goal in (lSa) and (lSb)); nevertheless, the argument functioning as the binder in the (a) sentences functions as the bound element in the (b) sentences. (17) a. b. (IS) a. b.
John showed the child himself in the mirror ?John showed the child to himself in the mirrors John and Mary gave presents to each other John and Mary got presents from each other
The primacy condition of anaphora to be proposed in this chapter also makes reference to a thematic hierarchy of arguments. This hierarchy, however, is not based on an absolute, inherent hierarchy of a-roles, independent of lexical items, but is the hierarchy of the a-roles associated with a given head in the lexicon. We assume that the a-roles in the a-grid of a head do not form an unordered set, but are ordered - on the basis of a conceptually motivated principle. This principle, in intuitive terms, puts the argument viewed as the most active participant of the action or state to the top of the hierarchy, and the argument viewed as the least active participant of the action or state to the bottom of the hierarchy. The most prominent argument of a verb, denoting the most active participant of the action or state, is the argument marked for external Case assignment in English, and for nominative assignment in Hungarian. It is most typically represented by the agent or the experiencer - if the a-grid of the verb has any. The second most prominent argument is the direct argument of the verb: most typically a theme. (Of course, if the theme represents the most active participant of the given action or state, it functions as the most prominent argument of the verb.) A locative or temporal argument, denoting a completely passive element of the action or state, typically occupies the lowest position in the argument hierarchy. Of a source argument and a goal argument, either one can be viewed as more active; thus, for example, in the a-grid of the verb give, the source is the most prominent argument,
250
E. Kiss Katalin
and the goal is the least prominent one, while in the case of the verb get, it is the other way round. We suggest that the hierarchy of coarguments be encoded in the lexicon - simply by the linear order of the arguments in the a-grid. It is generally assumed that the thematically most prominent argument of a predicate, the argument to be realized external to the maximal projection of the head in configurational languages, is distinguished in the a-grid, e.g. by being underlined. What we propose is that this distinguished, external argument be ordered first in the a-grid, and the order of the internal arguments be fixed, too. Consider, for instance, the a-grid that we assign to the verb inquire or its Hungarian equivalent megkerdez. (The lexical entry associated with the verb is formulated along the lines suggested in Higginbotham (1985).) (19)
inqUire} { megkerdez
,+V-N,(1,2,3,E)agent(1),theme(2),source(3)6
The following argument hierarchy can be read off from the a-grid: (20)
agent> theme> source
A lexical entry also contains information about the categorial realization of its elements: (21)
inquire, +V -N, ( 1, 2,
3,
E)
116
NP NP about NP
megkerdez, +V -N, (
r r l'
E)
NPnom NP ace NPdela! Consequently, the argument hierarchy represented by the linear order of the elements in the a-grid of e.g. megkerdez can also be translated into the following hierarchy: (22)
NPnom> NPace> NPdela!
The lexical hierarchy of arguments, encoded in the a-grid, is both functionally and structurally independent of the phrase structure argument hierarchy expressible in terms of c-command. While the phrase structure argument hierarchy is the expression of head-complement and specifier-head relations (cf. Speas (1986)),
The primacy condition of anaphora and pronominal variable binding
251
the lexical argument hierarchy reflects the relative prominence of coarguments. The fact that in the case of inquire, for example, the D-structure hierarchy of the three arguments is identical to their lexical hierarchy is a mere accident, which may not happen if two of the arguments are VP-internal arguments of the same category. A VP containing two NPs simply cannot be assigned a phrase structure in which the argument closer to the verb is hierarchically more prominent than the argument farther from the verb - as noted by Barss & Lasnik (1986). In the case of two PPs, the prominence of the PP next to the verb can only be achieved if its preposition is reanalysed as part of the verb. We claim that the licensing condition of anaphora has to make reference to the lexical hierarchy of arguments. The Hungarian data in (3-8) could, in fact, be accounted for on the basis of the lexical hierarchy of arguments alone, by means of the following simple primacy condition: (23) An anaphor must have an antecedent that precedes it in the lexical argument hierarchy. The bidirectionality of the anaphoric relation between an accusative NP and a dative NP could be derived from the - intuitively plausible - assumption that in a a-grid containing both a beneficiary argument, realized as NP dat> and a theme argument, realized as NPacc , either the beneficiary or the theme can be viewed as the more active participant of the action. Under this view, a verb with a beneficiary and a theme argument can be associated with two a-grids, which order the same arguments differently, as follows:
(24) megmutat, +V - N (i) (1, 2, 3, E) agent (1), theme (2), beneficiary (3) (ii) (1, 2, 3, E) agent (1), beneficiary (2), theme (3) In (8a), in which the theme argument binds the beneficiary argument, the primacy condition of anaphora is interpreted on a-grid (i), while in (8b), in which the beneficiary argument binds the theme argument, the primacy condition of anaphora is interpreted on a-grid (ii). In English, the two a-grids are realized categorially in different ways: (25)
show, + V - N (i) (1,2, 3, E) agent (1), theme (2), beneficiary (3)
II
~
NP NP toNP
252
E. Kiss Katalin (ii)
(1, 2,
I
3,
E) agent (1), beneficiary (2), theme (3)
I
NP NP NP If the a-grid in (i) is mapped onto syntactic structure, the primacy condition in (23) allows the theme argument to bind the beneficiary argument, as in (lla). If the agrid in (ii) is mapped onto a syntactic structure, the primacy condition in (23) allows the beneficiary argument to bind the theme argument, as in (lIb). The proposed primacy condition of anaphora could also predict the ungrammaticality of the anaphoric relation between a passive subject and a by-phrase:
(26) *John was shot by himself It is also possible, however, that in (26) the locality condition of anaphora is violated, the subject and the by-phrase not being coarguments (cf. Jaeggli (1986)). The tentative version of the primacy condition of anaphora in (23) also accounts for the distribution of grammaticality in the English examples in (9-11). The facts of English anaphora, however, are more complex than indicated in (9-11). The following range of data has to be explained: (27) a. b. c. d.
I *1 *1 *1
talked with the girls about each other talked about each other with the girls talked about the girls with each other talked with each other about the girls
The tentative primacy condition in (23), making reference only to the lexical prominence relation of the arguments involved, can only rule out sentences (27 c, d). The ungrammaticality of (27b) is the consequence either of an additional S-structure precedence requirement, or of an additional c-command requirement. The Hungarian data argue against an explanation in terms of S-structure precedence, since the Hungarian equivalent of (27b), with the anaphor preceding the antecedent, is grammatical: (28) . Egymasr61 beszelgettem a lanyokkal each-other-about talked-I the girls-with 'About each other, I talked with the girls' Since in (28) the anaphor has been focussed, i.e. it has been preposed into the specifier position of the VP (cf. E. Kiss (1987b)), the antecedent neither precedes the anaphor nor c-commands it at S-structure: (28')
[8 [VP Egymasr6li [" beszelgettem pro a lanyokkal tim
The primacy condition of anapllOra and pronominal variable binding
253
However, the antecedent does c-command the trace of the anaphor. Consequently, the primacy condition, which, together with condition (23), can predict both the grammaticality of the Hungarian (28) and the ungrammaticality of the English (27b), is the requirement that an anaphor or its trace be c-commanded by its antecedent. In the grammatical (27a), the NP the girls c-commands the anaphor because the preposition is reanalysed as part of the verb. In (27b) the c-command condition of anaphora cannot be satisfied because the preposition of the intended antecedent is not adjacent to the verb, therefore its reanalysis is impossible. That reanalysis is a precondition of anaphora in such constructions is demonstrated in the literature by the following contrast: (29) a. Who did you talk with about himself? b. *With whom did you talk about himself? In view of these observations, we replace the tentative primacy condition of anaphora proposed in (23) with the following - still tentative - condition: (30) An anaphor must have an antecedent that both c-commands it or its trace at S-structure, and precedes it in the lexical argument hierarchy. The primacy condition of anaphora in (30) is still only capable of handling anaphoric relations between coarguments. Its validity, however, can also be extended to non-coarguments if the term 'precedence in the lexical argument hierarchy' is given a broader interpretation, along the following lines: (31) a precedes b in the lexical argument hierarchy iff a is ordered prior to b or to the constituent containing b in a O-grid Consider the types of example that have to be accounted for: (32)
a. Janos es Mari nem ismerik egymas sziileit John and Mary not know each-other's parents-ACC 'John and Mary don't know each other's parents' b. ?Egymas sziileit nem ismeri Janos es Mad 'Each-others' parents, John and Mary don't know' c. ?Janost es Marit nem ismerik egymas sziilei 'John and Mary, each other's parents don't know' d. *Egymas sziilei nem ismerik Janost es Marit 'Each other's parents don't know John and Mary' (33) a. Elbeszelgettem Janossal es Marival egymas sziileir6l 'I conversed with John and Mary about each other's parents'
254
E. Kiss Katalin b. ?Elbeszelgettem egymas sziileir6l Janossal es Marival '1 conversed about each other's parents with John and Mary' c. ?Elbeszelgettem Janosr6l es Marir6l egymas sziileivel '1 conversed about John and Mary with each other's parents' d. *Elbeszelgettem egymas sziileivel Janosr6l es Marir6l 'I conversed with each other's parents about John and Mary'
In English, such anaphoric constructions involving two VP-internal arguments of the same category display a similar distribution of grammaticality except that the equivalent of (33b) is ungrammatical: (34) a. b. c. d.
I talked with John and Mary about each other's parents *1 talked about each other's parents with John and Mary ??I talked about John and Mary with each other's parents *1 talked with each other's parents about John and Mary
What the Hungarian data demonstrate is that the antecedent must be more prominent than the anaphor in terms of at least one of two primacy principles: the antecedent must precede the anaphor in the lexical argument hierarchy (as in (32a, b) and (33a, b», or, it must precede the anaphor at S-structure (as in (32a, c) and (33a, c». If neither of the two conditions is satisfied (as in (32d) and (33d», the sentence is out. As for English, the distribution of grammaticality in (34a, c, and d) also supports this generalization. The ungrammaticality of (34b) is a consequence of the violation of the c-command condition of anaphora. The analysis of the primacy factors influencing anaphora has led to the identification of three primacy principles: lexical prominence, S-structure c-command, and S-structure precedence. Their interaction can be summarized in the form of the following complex primacy condition of anaphora: (35) The primacy condition of anaphora An antecedent can bind an anaphor iff it c-commands the anaphor or its trace at S-structure, and at least one of (i) and (ii) also holds: (i) it precedes the anaphor in the lexical argument hierarchy (ii) it precedes the anaphor at S-structure The proposed primacy condition of anaphora is too powerful in so far as it also licenses sentences of the following type in Hungarian: (36) *Nem szereti egymas Janost es Marit not like each-other-NOM John-ACC and Mary-ACC 'Each other don't like John and Mary'
The primacy condition of anaphora and pronominal variable binding
255
The English equivalent of (36) would be ruled out as a violation of the c-command requirement of the primacy condition of anaphora. In the flat propositional component of the Hungarian sentence, in which all the arguments c-command each other, however, the antecedent in (36) both c-commands and precedes the anaphor. To account for (36), we cannot resort to binding principle C, because binding principle C cannot rule out (36) without also ruling out the grammatical (37): (37)
Nem szereti egymast Janos es Mari not like each-other-ACC John-NOM and Mary-NOM 'John and Mary don't like each other'
The application of binding principle C in the case of (37) and (36) is obviously blocked by a prohibition against referential circularity, which does not allow a constituent a to be the antecedent of a constituent b, and at the same time b to be the antecedent of a. Such a prohibition falls out from various theories of binding, e.g. from Higginbotham (1983), or Brody (1981). What we need in order to rule out (36) is a lexical structure equivalent of binding principle C, such as: (38)
An anaphor cannot precede its antecedent in the lexical argument hierarchy.
3 Pronominal variable binding The primacy notion of binding, requiring that the binder c-command the bound element, is also too powerful in licensing pronominal variable binding - especially in non-configurational languages. Consider, for instance, the Hungarian sentences in (39-40). The operator c-commands the pronominal in every example, as required; nevertheless, the sentences differ sharply in grammaticality. (39)
Mindenki; szereti az pro; anyjat7 everybody loves the mother-his-ACC 'Everybody loves his mother' mindenki; szereti b. Az pro; anyjat the mother-his-ACC everybody loves 'His mother, everybody loves' c. Mindenkit; szeret az pro; anyja everybody-ACC loves the mother-his-NOM 'Everybody, his mother loves' d. *Az pro; anyja mindenkit; szeret the mother-his-NOM everybody-ACC loves 'His mother loves everybody'
a.
256
E. Kiss Katalin
(40)
Kivel i beszelgettel a proi felesegerol? who-with talked-you the wife-his-about 'With whom did you talk about his wife?' kivel i beszelgettel? b. A proi felesegerol the wife-his-about who-with talked-you 'About his wife, with whom did you talk?' c. Kiroli beszelgettel a proi felesegevel? who-about talked-you the wife-his-with 'About whom did you talk with his wife?' d. *A Proi felesegevel kirolj beszelgettel? the wife-his-with who-about talked-you 'With his wife, about whom did you talk?' a.
If the c-command of the pronominal by the operator were the sole primacy condition of pronominal variable binding, as is generally assumed, it would remain unclear why (39d) and (40d) are ungrammatical. English sentences in which the antecedent and the pronominal are contained in arguments mutually c-commanding each other raise the same problem: (41) a. b. c. d.
I heard from each mani about his i wife ?I heard about hisj wife from each manj ?I heard about each mani from hisi wife *1 heard from his i wife about each manj
We suspect on the basis of (39-41) that the relative surface order of the operator and the anaphor is among the primacy factors that license binding. Therefore, in English we test pronominal variable binding in echo questions, in which the order of the wh-phrase and the pronominal is not invariant. (42)
a. b. c. d.
You talked with whomj about his i wife? ?You talked about hisj wife with whomj? ?You talked about whomj with hisj wife? *You talked with his i wife about whomi?
In English it is not only the ungrammaticality of (41d) and (42d) that has to be accounted for, but also the more or less (for some speakers more, for others less) marginal nature of (41b, c) and (42b, c). According to Barss & Lasnik (1986), the primacy condition of pronominal variable binding is that the pronominal be in the domain of the operator, i.e. it be both c-commanded and preceded by the operator. This solution not only rules out
The primacy condition of anaphora and pronominal variable binding
257
the ungrammatical (d) sentences in (39-42), but also excludes the grammatical (b) sentences. Besides, neither can it account for the somewhat marginal character of the English (41c) and (42c). Mankz (1986: 1l2), aiming to account for the distribution of grammaticality in (39), suggests the following rule of bound variable interpretation for Hungarian: (43)
(44)
A pronoun can be interpreted as a bound variable iff it does not both cyclicc-command its antecedent at lexical structure and precede it at phrase structure. a cyclic-c-commands b iff b. a c-commands b or b. if c is the minimal cyclic node (NP or S-bar) that dominates a but is not immediately dominated by another cyclic node, then c c-commands b.
In Mankz's interpretation, lexical structure is a representation of the sentence the articulation of which corresponds to the articulation of an English-type, configurational phrase structure. In this framework, the lexical structure assigned to (40a) and (40b) would be one of the three theoretically possible configurational constructions shown in (45a-c). (45)
a.
~
~
NP,
proj (45)
1 T6
beszelgettelj
kiveli
a Proi feleseger5l
c.
b.
A
r
NP 1
V
~
NP 1
VP
NP2
NP '
VP
vA NP2
NP 3
NP3 , a pro felesegerOl, c-commands NP2 , kivel, in each structure. Consequently, in (40b) pro not only precedes its binder at S-structure but also cyclic-c-commands it at lexical structure. The sentence is, nevertheless, grammatical, contrary to Mar-
258
E. Kiss Katalin
acz's prediction. If Maracz's primacy condition were extended to English, it would run into similar problems, e.g. in the constructions represented by (41) and (42). The primacy condition of pronominal variable interpretation to be proposed in this chapter retains Maracz's reference to S-structure precedence; however, it discards the reference to lexical structure c-command (since it cannot properly order two VP-internal NPs or PPs). It introduces the notion 'precedence in the lexical argument hierarchy' instead. According to the evidence of (39-42), the requirement of S-structure precedence and the requirement of precedence in the lexical argument hierarchy interact in the following way: The most felicitous cases of pronominal variable binding are those in which the operator precedes the pronominal both at S-structure and in the lexical argument hierarchy, i.e. the (a) sentences in (39-42). The examples in which only one of the two primacy principles is observed, i.e. the (b) sentences, in which the operator only precedes the pronominal in the lexical argument hierarchy, and the (c) sentences, in which the operator only precedes the pronominal at S-structure, are also grammatical, but - especially in English - are somewhat more marginal than the (a) sentences. If both primacy principles are violated, as in the (d) sentences, the bound variable interpretation is impossible. Apparently, English speakers may attribute different significance to the precedence and the lexical prominence conditions. Certain speakers find (41b) and (42b) perfectly grammatical, while barely accepting (41c) and (42c), or not accepting them at all. For these speakers S-structure precedence plays little or no role in licensing pronominal variable interpretation. There are also speakers for whom (41c) and (42c) are perfectly grammatical, and (41b) and (42b) are marginal. These speakers obviously value the S-structure precedence condition higher than the lexical prominence condition. In addition to S-structure precedence and precedence in the lexical argument hierarchy, S-structure c-command also seems to be involved in licensing pronominal variable binding. It is needed to prevent the bound variable interpretation of the pronominal in such sentences as: (46)
*1 discussed the rumours about each manj with hisj wife
The role of c-command, however, is not straightforward, as the operator phrase in (41) and (42), embedded under a PP, does not, strictly speaking, c-command the pronominal either. Although in the (a) and (c) sentences, the P could be assumed to be reanalysed as part of the verb, in which case the c-command condition would be satisfied, in the (b) sentences, which represent for most informants the same degree of grammaticality, reanalysis is impossible. It does not seem to be unreasonable, however, to assume that the feature 'universally quantified' in (41), or the
The primacy condition of anaphora and pronominal variable binding
259
wh-feature in (42), can percolate up to the PP-node containing the operator, yielding an operator phrase that c-commands the pronoun. The assumption that an operator feature can percolate up across one phrase boundary is, in all probability, independently needed in English to account for such cases as: (47) Everyone/s mother thinks hej is a genius Our tentative claim that the operator feature can percolate up across one maximal projection is based on the distribution of grammaticality in the following sentences: (48) (49) a. b. c.
*1 talked with the wife of each manj about hisj mother 1 talked with each man/s wife about hisj mother 1 talked about each man/s wife with hisj mother *1 talked about hisj mother with each manj's wife
Let us assume that the preposition immediately following the verb is reanalysed as part of the verb in (48) and (49). Let us, crucially, also assume that the feature 'universally quantified' percolates up from the NP minimally containing it to the next higher maximal projection. Then in (48) the resulting quantified PP (of each man) does not c-command the pronominal, while in (49a, b), the resulting quantified NP (each man's wife) does c-command it. In (49c) the preposition of the PP containing the operator cannot reanalyse, therefore the NP to which the operator feature has percolated up (each man's wife) does not c-command the pronominal. That is, the distribution of grammaticality in (48) and (49) falls out. The c-command requirement of pronominal variable binding has to be extended so as to allow the operator phrase to c-command, instead of the pronominal, the trace of the pronominal (or of the phrase containing the pronominal) - as the (a) sentences in (50) and (51) below indicate. (The extended c-command condition is satisfied also in the ungrammatical (SOb) and (51 b); they are ruled out because they observe neither the S-structure precedence requirement, nor the lexical prominence requirement of binding.) (50) a. b. (51) a. b.
?About hisj wifej, 1 talked with each manj tj for hours *With hisj wifej, 1 talked with each manj tj for hours ?About which ofhisj wife's friends j did you talk with each manj tj? *With which ofhisj wife's friends j did you talk about each manj tj?
The three primacy principles involved in licensing pronominal variable interpretation are not only identical to those involved in licensing anaphora, but they also interact in the same way. That is:
260
E. Kiss Katalin
(52) The primacy condition of pronominal variable binding A pronominal can be interpreted as a variable bound by an operator iff the operator phrase c-commands the pronominal or its trace at S-structure, and at least one of (i) and (ii) also holds: (i) the operator phrase precedes the pronominal in the lexical argument hierarchy (ii) the operator phrase precedes the pronominal at S-structure While the primacy condition of pronominal variable binding in (52) correctly predicts the binding possibilities in all the sentence types we have discussed, it seemingly breaks down in interrogative double object constructions in which the direct object functions as the wh-operator, and the indirect object contains the pronominal. Consider the following example, taken from Barss & Lasnik (1986): (53) *Which lioni did you show itsi trainer ti? The problem with (53) is that the intended but unavailable interpretation of the sentence, in which the wh-phrase binder of the pronoun is understood as the direct object, is predicted to be possible by principle (52). The wh-operator both ccommands the pronoun and precedes it; i.e. two of the three primacy requirements of pronominal variable interpretation are satisfied. Nevertheless, the intended interpretation does not arise. Notice, however, that the pronoun can be understood as a variable bound by the wh-operator in (53) - if the wh-phrase is interpreted as the indirect object, i.e. if the sentence is analysed as follows: (54) Which lioni did you show ti itsi trainer? In view of this fact, the question is why the former of the two theoretically
possible interpretations of the sentence ((53) and (54» never arises. We suggest the following explanation: In structure (54) all the three primacy requirements of bound pronominal interpretation are observed: the wh-trace c-commands the pronoun, the wh-phrase precedes the pronoun, and the wh-argument is also more prominent than the pronoun in the lexical argument hierarchy. In (53), on the other hand, only two elements of the primacy condition are satisfied - the c-command requirement, and the S-structure precedence requirement - so if the bound pronominal interpretation which is theoretically possible in (53) could be evoked, it would be somewhat marginal, representing the same degree of grammaticality as, for example, (41c) and (42c). Apparently, if English speakers have to choose between a marginal and a fully grammatical interpretation in the case of structural homonymy, they invariably choose the fully grammatical interpretation.
The primacy condition of anaphora and pronominal variable binding
261
4 Summary and conclusions
The analysis of anaphora and pronominal variable binding in English and Hungarian has revealed that anaphoric binding and pronominal variable binding are licensed by the same primacy condition. This condition requires the prominence of the binder over the bound element in terms of at least two ofthree primacy factors: S-structure c-command, lexical (thematic) prominence, and S-structure precedence. The proposed primacy condition of binding is inapplicable in the case of binding principles Band C, as the following examples show: (55)
(56)
a. *Beszeltem Janossal; rola; spoke-I John-with about-him 'I spoke with John about him' b. *Beszeltem J anosrol; vele; spoke-I John-about with-him 'I spoke about John with him' a. *Beszeltem Janos; anyjarol vele; spoke-I John's mother-about he-with 'I spoke about John's mother with him' b. *Beszeltem Janos; anyjaval rola; spoke-I John's mother-with about-him 'I spoke with John's mother about him'
If binding principles Band C were sensitive to lexical prominence, i.e. if disjoint reference only arose if the NP c-commanding the pronominal in (55) or the Rexpression Janos in (56) also preceded the pronominal or Janos in the lexical argument hierarchy, then (55b) and (56b) would not display any disjoint reference effects. The fact that anaphoric binding and pronominal variable binding are subject to the same primacy condition, while anaphoric binding and disjoint reference, i.e. binding principle A on the one hand, and binding principles Band C on the other hand, are subject to different primacy conditions, suggests that anaphoric binding and pronominal variable binding may form a more natural class than anaphoric binding and disjoint reference; that is, a reorganization of binding theory perhaps along the lines proposed in Reinhart (1983a) - might be desirable.
Notes 1. Barss (1986) proposes a version of binding principle A which combines the primacy
262
2.
3. 4.
5.
6. 7.
E. Kiss Katalin
requirement and the locality requirement of anaphora into a single notion: the notion of minimal chain-accessibility. According to Mohanan (1983-4), lexical structure is distinct from phrase structure (or configurational structure) in configurational and non-configurational languages alike. While lexical structure is the representation of grammatical functions, configurational structure is the representation of dominance and precedence relations. Mohanan claims that anaphora is interpreted at lexical structure in both language types. Delative Case represents a source; it may be rendered in English by prepositions such as of, from or about. A thematic hierarchy similar to (16) also plays a role in the anaphora theories of Hellan (1986b) and Kiparsky (1987), and also in the theory of long-distance anaphora of Giorgi (1984). It has also been claimed that while himself in (17a) is a beneficiary, to himself in (17b) is a goal. According to this view, the anaphor cannot represent the same, beneficiary, O-role in both cases because, if a beneficiary could alternatively be realized either as an NP or as a to-NP, then it could not be explained why the beneficiary in, for example, I gave him a headache has no to-NP variant. The morphological realization of O-roles, however, is known to involve a lot of idiosyncrasies, therefore it would not be illegitimate to include a constraint in the lexical entry associated with the idiom give somebody a headache to the effect that the beneficiary argument can only be realized as a bare NP. E corresponds to a 'hidden' argument place for events in the semantic theory of Higginbotham (1985). Hungarian is a pro-drop language. The subject pronoun also drops in the NP, where it is coindexed with an Agr-marker on the head noun.
13 The local nature of the long-distance reflexive in Chinese C.-T. James Huang and C.-C. Jane Tang
1 Introduction Chinese has two reflexive forms: the bare reflexive having the invariant form ziji 'self' and the compound reflexive having the form of a pronoun + ziji sequence, as in taziji 'himself/herself', niziji 'yourself', etc. As described in Y.-H. Huang (1984), Tang (1989), and Wang & Stillings (1984), these elements exhibit distributional and referential properties that are of considerable interest to linguistic theory. As far as their reference is concerned, one property of the reflexives is that the bare reflexive, though not the compound reflexive, exhibits possibilities of having a long-distance antecedent apparently outside of its governing category. Thus in (Ia) ziji may have either Zhangsan or Lisi as its antecedent, but in (Ib) taziji must be locally bound by Lisi: (1)
a.
b.
Zhangsani renwei [Lisij hai-Ie zijiilil Zhangsan think Lisi hurt-ASP self 'Zhangsani thought that Lisi j hurt himselfilj ' Zhangsani renwei [Lisi j hai-Ie ta-ziji*ilj] Zhangsan think Lisi hurt-ASP himself 'Zhangsani thought that Lisij hurt himself*ilj'
Long-distance binding with ziji is, however, restricted by a condition that requires the remote antecedent to agree in person and number features with all closer potential antecedents. In particular, a remote NP can antecede ziji only if it agrees with the local NP in the governing category of ziji. Thus, although ziji may have Zhangsan as its antecedent in (Ia), where it agrees with Lisi in person and number, long-distance binding is blocked in examples like (2), where the remote NP differs from the local NP either in person or in number, or both. In all these cases, ziji must be bound by the local NP: I
263
264
(2)
C.-T. James Huang and C.-C. Jane Tang
a.
b.
c.
d.
Zhangsani renwei [woi hai-Ie ziji*ili] Zhangsan think I hurt-ASP self 'Zhangsan thought that I hurt myself' Nii renwei [Zhangsani dui ziji*ili meiyou xinxin] you think Zhangsan to self not-have confidence 'You think that Zhangsan has no confidence in himself' WOi renwei [womeni yinggai dui ziji*ili you xinxin] should to self have confidence I think we 'I think we should have confidence in ourselves' Womeni renwei [tai dui ziji*ili meiyou xinxin] we think he to self not-have confidence 'We think that he has no confidence in himself'
In the following sentence, long-distance binding of ziji is also blocked, in spite of the fact that the remote NP agrees with the most local NP: (3)
Zhangsani shuo [woi zhidao [Lisi k chang piping ziji*iI*i/kll Zhangsan say I know Lisi often criticize self 'Zhangsan said that I feel that Lisi always criticized himself'
This is because of the intervening NP wo '1', which agrees with neither Zhangsan nor Lisi. If wo is replaced by Wangwu, as in (4), long-distance binding is again allowed: 2 (4) Zhangsani shuo [Wangwui zhidao [Lisik chang piping zijiifi/k]] 'Zhangsani said that Wangwui knew that Lisi k often criticized selfifi/k' The purpose of this chapter is to consider how these facts regarding the longdistance reflexive may be best explained. In particular, we will be concerned with (a) why only the bare reflexive may exhibit long-distance binding, and (b) why long-distance ziji is subject to the strict requirement of agreement just described. In section 2 we briefly indicate general conditions on what may qualify as a 'potential antecedent of the Chinese reflexives, and in section 3 we review two recent accounts of long-distance ziji. The discussion in these two sections will help to crystallize the nature of the problems we are dealing with, which will then lead to our proposal in section 4, followed by a brief conclusion in section 5. It will be our claim that the phenomena observed are best explained if we assume that the bare reflexive is an anaphor in two ways, since it lacks not only reference, but also intrinsic features normally associated with pronouns, and that as such it needs to receive two indices under binding theory, first at S-structure and again at LF. Other principles of grammar will combine to derive the facts to be explained.
The local nature of the long-distance reflexive in Chinese
265
2 Potential binders of the reflexive Before discussing the issue of long-distance binding, we must identify a few conditions on what, in general, may qualify as a binder, local or remote. First, the Chinese reflexive ziji can be bound only to a subject, but not in general to an object (see C.-T. J. Huang (1982». (5)
WO i gaosu Lisi j zijiiJ*j de fenshu 1 tell Lisi self's grade 'I told Lisi my own grade'
Second, only an animate, not an inanimate, NP can antecede ziji: (6)
Wo bu xiaoxin dapo-Ie ziji de yanjing 1 not careful break-ASP self's glasses 'Not being careful, 1 broke my own glasses' (7) *Yanjing diao-dao dishang, dapo-Ie Zl)l glasses drop-to floor break-ASP self 'The glasses dropped on the floor, and broke themselves' Generally, any c-commanding animate subject NP in the governing category of a reflexive may be its antecedent. There are two situations under which a non-ccommanding subject may be an antecedent. The first situation arises when a sentence contains a 'psychological' verb: (8) [Zijii de xiaohai mei de jiang de xiaoxi] shi Lisii hen nanguo self's child not get prize DE news make Lisi very sad 'The news that his own child did not get a prize made Lisi sad' The special status of psych-sentences is, of course, well known since Postal (197l), and recent work by Giorgi (1984) and others has revived an interest in such constructions (cf. Pesetsky (1987b), Belletti & Rizzi (1988), etc.). We will not discuss in this chapter how psych-sentences are to be analysed, but simply note that, in Chinese too, the experiencer argument may bind a reflexive even though the element is otherwise subject oriented. The other situation where an NP that is not a c-commanding animate subject may antecede ziji is when it appears as the 'most prominent' animate subject NP within an inanimate NP that c-commands ziji. This is illustrated in (9), to be compared with (10): (9)
[Woi de jiaoao]j hai-Ie Zl)liJ*j 1 's pride hurt-ASP self 'My pride hurt myself'
266
G.-T. James Huang and C.-C. Jane Tang
(10) [WOj de meimei]j hai-Ie ZlJhi/j I 's sister hurt-ASP self 'My sister hurt herself' Clearly, the fact that the non-c-commanding wo 'I' can be the antecedent of ziji is related to the fact that the c-commanding NP wo dejiaoao 'my pride' is inanimate, hence not a potential antecedent. To capture this intuition, Tang (1989) proposed the notion of 'sub-command': (11)
f3 sub-commands ex iff f3 is contained in an NP that c-commands ex or that sub-commands ex, and any argument containing f3 is in subject position.
The condition under which the c-command requirement may be relaxed is stated as in (12).3 (12)
A reflexive ex may take an NP f3 as its binder if a. f3 sub-commands ex, and b. there is no NP 'Y, 'Y a potential binder for ex, such that 'Y is closer to ex than f3 is.
For the notion of relative distance, assume that, other things being equal, a ccommander of ex is closer than a sub-commander is to ex, and a c-commander or sub-commander in the minimal clause containing ex is closer than one outside of the minimal clause, etc. In the case where a sub-commander is contained in a ccommander, the c-commander is closer to ex. A 'potential binder' is any NP that satisfies all conditions of being a binder of ex except that it is not yet coindexed with ex. Given (11-12), in (9) wo 'I' can antecede ziji, for the former sub-commands the latter and there is no potential binder closer to the reflexive. The only NP closer to ziji than wo 'I' is the NP wo de jiaoao 'my pride', which, not being animate, is not a potential binder. In (10), on the other hand, wo 'I' cannot antecede ziji, because, though the former sub-commands the latter, it is contained in the animate NP wo de meimei 'my sister', which is a potential binder and is closer to ziji. (See Tang (1989) for more details and other facts captured by (11-12).) Summarizing, we have seen that the reflexive may take as its antecedent an animate NP that is (a) a c-commanding subject, (b) a sub-commanding subject, or (c) an experiencer. Without attempting an exact formulation ofthe term, we shall refer to coindexing with any of these NPs as an instance of binding (even though the standard definition of binding entails c-command). We have seen examples in which certain NPs are potential local binders. What about potential long-distance binders? There is reason to believe that the con-
The local nature of the long-distance reflexive in Chinese
267
ditions are essentially the same (except for the restrictions noted in section 1). For example, we have seen in section 1 that a c-commanding subject can be a longdistance binder (see (la) and (4)). Furthermore, an experiencer can be a longdistance binder, too, as illustrated below: (13) [Zhangsanj taoyan zijii/j de xiaoxi] shi Lisi j hen nangguo Zhangsan dislike self DE news make Lisi very sad 'The news that Zhangsanj disliked selfi/j made Lisi j very sad' The only exception is that a sub-commander cannot, in general, be a long-distance binder. (14) allows only local binding: (14)
Zhangsanj de xin biaoshi [Lisij hai-Ie ziji'i/j] Zhangsan 's letter indicate Lisi hurt-ASP self 'Zhangsan's letter indicates that Lisi hurt himself'
This may be attributed to the fact that a sub-commander is picked out as a (marked) antecedent only as a 'last resort', when there is no other more accessible NP that can bind a given reflexive (cf. condition (l2b)). In a situation where a subcommander occurs in a remote position, there is already an NP in a local position that is more accessible to the reflexive. Therefore, given (12b), a sub-commander cannot be a long-distance binder. 4
3 Long-distance ziji: previous analyses How does the 'long-distance' ziji fit into an optimal theory of grammar? From what we have seen so far, it is clear that ziji is really not something whose referential properties are unconstrained by principle A of binding theory, nor an element that is free from the locality restriction imposed by the notion of minimal governing categories. The restrictions we saw suggest that in order for ziji to be bound by an NP outside of its governing category, it must first be licensed by an NP in its governing category that agrees in person and number with the remote NP. And each further remote NP may be an antecedent only if ziji can be successively licensed by all lower potential antecedents in the same way. This indicates that there is a sense of strict locality involved here, and that so-called long-distance binding should be described in terms of successive steps of local binding. 'Longdistance' ziji, in other words, should not be admitted by parametrizing the notion of a governing category, as might be suggested along the lines of Yang (1983) and Manzini & Wexler (1987). It would also be inappropriate to simply define ziji as a non-anaphor, as a 'pronominal anaphor' (Wang & Stillings (1984)), or as a bound pronoun (e.g. Sportiche (1986)). For all these proposals simply relax the locality
268
G.-T. James Huang and C.-C. Jane Tang
conditions on the binding of ziji, thus failing to capture the strict locality requirements observed. To capture the relevant restrictions, Tang (1985, 1989) proposed that the bare ziji originates as pro-ziji, a compound reflexive with an empty pro, and that the limited cases of long-distance ziji are derived from the optional feature-copying rule (15) and the iterative reindexing rule (16): (15) Feature-copying rule The pro in a pro-ziji reflexive may transfer its features (such as person and number) to -ziji after the application of binding theory, thus turning ziji into a 'long-distance' reflexive. (16) Reindexing rule Reindex the long-distance reflexive with the potential antecedent of the next-higher governing category. Given (15) and (16), the restrictions noted in section 1 are accounted for. Consider (la) and (lb), above. Since the feature-copying rule applies only to proziji, the compound reflexive in (lb) is unaffected and remains a local reflexive. In (la), on the other hand, pro-ziji can optionally undergo rule (15). If (15) does not take place, the pro-ziji is still a local reflexive and is bound to the local Lisi. If (15) applies, ziji becomes a long-distance reflexive, carrying the person and number features of Lisi (after binding theory has applied). At this time, ziji is reindexed with Zhangsan, under (16). Since ziji agrees with Zhangsan in person and number features, reindexing is allowed, and ziji comes to be bound by a remote antecedent. The blocking effects indicated above also follow straightforwardly. Consider the sentences in (2) and (3) above. When binding theory applies at S-structure, ziji in each of these sentences is bound by its local antecedent. So if the copying rule applies in (2a), ziji must carry the features [1st person, singular] as it is turned into a long-distance reflexive. But this prevents it from taking Zhangsan as its remote antecedent under reindexing. Therefore ziji cannot be turned into a long-distance anaphor. Similarly, since reindexing is required to be successive-cydic, longdistance binding by Zhangsan is also blocked in (3). Although the proposal embodying (15-16) accounts for the relevant facts, it also leaves a number of questions unanswered. For example, it does not explain why the reindexing rule (16) applies only to ziji, nor does it explain why the copying rule can change a local reflexive into a long-distance one. Furthermore, as Battistella (1987) points out correctly, it is not clear why copying should trigger reindexing, nor why reindexing should mimic the effects of binding. In recent years, a new analysis of the reflexive anaphor has aroused considerable
The local nature of the long-distance reflexive in Chinese
269
interest. Inspired by the work of Lebeaux (1983) and Chomsky (1986a), a number of writers (Pica (1987), Battistella (1987), among others) have suggested that certain reflexives may be raised in LF into Infl, in a way analogous to clitic movement in syntax, thereby accounting for their subject orientation. Furthermore, certain such reflexives may move from Infl to Infl (an instance of head-tohead movement, in a way analogous to the phenomenon of 'clitic climbing'), thereby accounting for their long-distance binding possibilities. Under Battistella's approach, for example, the compound reflexive is a full NP, whereas the bare ziji is an NO. Given that Infl-to-Infl movement is a head-to-head movement, he argues that only ziji may undergo this Infl-to-Infl movement. Assuming that no successive-cyclic movement is otherwise available for the compound reflexive, he accounts for the fact that only ziji exhibits long-distance binding. To account for the blocking effects of ziji, he proposes (a) that Inflmovement must go successive-cyclically, and (b) that each trace left by ziji in Infl, as well as the moved ziji itself, must agree in grammatical features with its own local subject, as a general requirement of subject-Infl agreement, construed abstractly in Chinese. Since all traces must be coindexed with the moved ziji, it follows that all local and non-local subjects must agree in person and number, in cases where ziji has a remote antecedent. Attractive as it is, the Infl-movement theory is faced with important difficulties. For one thing, Battistella does not explain why the compound reflexive taziji cannot adjoin successive-cyclically in LF, giving rise also to long-distance binding. (Note incidentally that the Lebeaux-Chomsky proposal in fact assumes that the compound reflexive himself is moved to Infl.) A more serious problem concerns the blocking effects. According to Battistella, the blocking effects follow partially from the fact that Infl is the locus of agreement. Notice, however, that the potential blockers oflong-distance ziji include not only c-commanding local subjects (as we have seen in all the relevant examples), but also local sub-commanders and experiencer non-subjects. For example, in the following sentences, ziji must be bound by its local sub-commander, but not by the matrix subject: (17)
(18)
Zhangsani shuo [WOj de jiaoao hai-Ie ziji*iliJ Zhangsan say I 's pride hurt-ASP self 'Zhangsani said that mYj pride hurt myself*ilj' Zhangsanj shuo [[nij zheyang zuo] dui ziji*ilj bu Ii] Zhangsan say you thus do to self not advantage 'Zhangsanj said that yourj doing this will do yourself*ilj no good'
270
G.-T. James Huang and C.-C. Jane Tang
(19)
Zhangsani shuo [[nij zuo shi de taidu] du ziji*i/j bu hao] Zhangsan say you do work REL attitude to self not good 'Zhangsani said that the attitude with which yOUj work is not good for yourself*i/j'
Battistella suggests that in these cases, the verb of the embedded clause containing
ziji in effect agrees with the sub-commanding NP. However, this way of looking at subject-Inft agreement does not seem seriously entertainable. There is little reason, other than to derive the blocking effects, to say that a matrix verb agrees not with its own subject, but with the subject of its sentential subject (as in (18)) or ofits complex NP subject (as in (19)).5 Furthermore, (20) shows that an experiencer non-subject may block long-distance ziji. (20)
[[[Zhangsani dui zijii/*j/*k mei xinxin de shi] shi WOj hen nanguo Zhangsan to self no confidence's fact make me very sad de xiaoxi] shi Lisi k hen yiwai] DE news make Lisi very surprised 'The news that I was saddened by the fact that Zhangsan had no confidence in himself surprised Lisi'
In this sentence, the matrix predicate shi Lisi henyiwai 'make Lisi very surprised' takes a complex NP subject meaning 'the news that the fact that Zhangsan has no confidence in himself saddened me'. Long-distance binding of ziji by the outermost experiencer Lisi is blocked by the inner experiencer wo 'me'. Here the blocker is not a subject, and it thus looks even more unlikely that the blocking effects have to do with subject-Inft agreement. Rather, the emerging generalization we want to capture is the following; (21) The set of potential blockers of long-distance ziji is exactly the set of its potential local, or less remote, binders. This generalization suggests that the blocking effects should not be treated as an effect of agreement, but as a property of binding. Finally, since in the Inft-movement theory the movement of ziji in LF is a case of head movement, the traces left over by ziji are subject to antecedent-government (see Chomsky (1986b, 1988b)). While this has the consequence that the movement must be successively cyclic, it incorrectly rules out certain acceptable cases of longdistance ziji. In particular, we know from independent evidence that movement of a phrase whose trace needs to be antecedent-governed cannot cross any singular
The local nature of the long-distance reflexive in Chinese
271
barrier. Thus, adjuncts located in adverbial clauses and relative clauses cannot be wh-moved out of these islands. This is true both in the syntax (22) and in LF (23): (22) a. *WhYi did you go home [before John bought the book ti]? b. *WhYi did you like [the man who kicked Bill ti]? (23) a. *Suiran Lisi weishenme mei lai, ni haishi bu shengqi? though Lisi why not come you still not angry '*Though Lisi didn't come why, you weren't angry?' b. *Ni zui zihuan [ta weishenme mai de shu]? you most like he why buy REL book '*You like the book that he bought why?' Furthermore, Huang (1982) argues that A-not-A questions in Chinese exhibit ECP effects. The A-not-A element is an element in Infl, and Infl-movement cannot cross barriers: (24)
a. *Ruguo ta lai-bu-Iai, ni jiu hui shengqi? if he come-not-come you then will angry '*If he comes or not, then you will be angry?' a. *Ni zui xihuan ta mai-bu-mai de shu? you most like he buy-not-buy REL book '*You like the books that he will buy or will not buy?'
These facts lead one to expect that, if ziji undergoes head-to-head movement in LF, no long-distance binding is possible across adjunct clauses or complex NPs. But this prediction is incorrect. In the following sentences, long-distance binding is fully acceptable, suggesting that LF-traces of ziji are not subject to antecedent-government. (25)
(26)
Zhangsani shuo [ruguo Lisi j piping zijii/j], ta jiu bu qu Lisi criticize self he then not go Zhangsan say if 'Zhangsani said that if Lisij criticized himselfi/j, then he won't go' renj] Zhangsani bu xihuan [neixie piping zijii/j de Zhangsan not like those criticize self REL person 'Zhangsani does not like those peoplej who criticize selfi/j'
In view of these problems, we must now look elsewhere for an explanation oflongdistance ziji. 6
272
G.-T. James Huang and C.-C. Jane Tang
4 The locality of long-distance ziji
Although the Infl-movement theory cannot account for the locality restrictions of long-distance ziji in a proper way, one property of the theory that seems to us to be correct is the idea that the locality restrictions are to be expressed by successivecyclic movement of ziii in LF. We shall pursue an explanation along this line, but attempt to derive the locality requirements from other sources. To see how this may be done, consider the following sentences: (27)
(28)
a. b. c. a. b. c.
John knows that Bill likes pictures of himself John knows that, pictures of himself, Bill likes Pictures of himself, John knows that Bill likes John knows that Bill likes these pictures of himself John knows which pictures of himself Bill likes. Which pictures of himself does John think that Bill likes?
These sentences exemplify the so-called 'reconstruction problem' or 'connectivity effect', well known since Higgins (1973) and more recently Barss (1986). In each (a) sentence above, the reflexive must have Bill as its antecedent, but not the remote John. In the (b) sentences, however, himself may have either John or Bill as its antecedent, though only John actually c-commands the reflexive. And in the (c) sentences, either John or Bill may antecede himself, though neither c-commands the latter. The suppression of the c-command requirement in these sentences is dealt with in Barss (1986) in terms of a condition of 'chain accessibility' on binding theory as applied at S-structure. That is, by virtue of its relation to a trace in the minimal c-command domain of John or Bill, himself is defined as being 'chain bound' by John or Bill in the (b) and (c) sentences. What is important here is that in these cases a locality requirement is still maintained for the binding of himself to be possible. Thus, binding by John in the (b) sentences is possible only because himself has been moved to a position where its governing category contains John. In (29), where himself is in the most deeply embedded Comp, only Bill may antecede himself: (29) John knows that Bill wondered which pictures of himself I would buy Similarly, Barss shows that the (c) sentences allow John to antecede himself only because the NP (which) pictures of himself binds a trace in the intermediate Comp, creating a chain configuration in which John is allowed to bind himself as a 'minimally accessible' antecedent. (See Barss (1986: chapter 3) for more details.) From the point of view of D-structure (cf. (27a) and (28a)), then, we may say that a reflexive in its D-structure argument position has only a local antecedent, but may
The local nature of the long-distance reflexive in Chinese
273
pick up a 'long-distance' antecedent as a result of successive-cyclic movement. But from the point of view of S-structure, all 'long-distance' antecedents are in fact local ones, each being a minimally accessible 'chain binder' in the sense of Barss. What we would like to suggest is that the 'long-distance' ziji is essentially the same phenomenon as that illustrated in (27-9), except that it is a phenomenon that occurs in LF rather than in the syntax. That is, from the point of view of Sstructure, ziji in its S-structure argument position has only a local binder, but may pick up a remote antecedent as a result of successive-cyclic movement in LF. From the LF point of view, however, all 'remote antecedents' are local antecedents. This proposal is similar to the Infl-movement or the 'clitic-climbing' theory, but we claim that the LF movement involved is simply A'-movement, more specifically IP-adjunction, perhaps as a case of QR. 7 Thus, the ambiguous readings of (30) are unambiguously represented at LF as in (31), with ziji bound in each case by a local antecedent: (30)
Zhangsan manyuan Lisi chang shuo Wangwu bu xihuan ziji Zhangsan complain Lisi often say Wangwu not like self 'Zhangsan complained that Lisi often said that Wangwu does not like Wangwu/LisilZhangsan' (31) a. Zhangsank manyuan [Lisi j chang shuo [Wangwui bu xihuan zijii]] b. Zhangsank manyuan [Lisij chang shuo [ziji j [Wangwui bu xihuan til]] c. Zhangsank manyuan [zijik [Lisi j chang shuo [tk [Wangwui bu xihuan tk]]]]
Our proposal thus assumes that the reference of ziji can be determined by binding theory applying at LF. This by itself is not a problematic assumption, given the arguments of Aoun (1985) and the discussion in Chomsky (1982: note 11). However, there is well-known evidence that binding theory must also apply at S-structure (see Chomsky (1981, 1982), Barss (1986), among others). For example, the binding possibilities of himself in (27-9), or in (32-3) below, must be determined at S-structure: (32)
a. b. (33) a. b.
John said that Bill criticized himself John said that, himself, Bill criticized John said that Bill likes every picture of himself John said that, every picture of himself, Bill likes
In these cases, himself can take John as its antecedent only as a result of its movement in the syntax. In LF, IP-adjunction of every picture of himself or himself may take place (under QR or as instances of Move u), but this movement does not alter binding possibilities of himself as does movement in the syntax. The same
274
C.-T. James Huang and C.-C. Jane Tang
point holds in Chinese with sentences containing pronoun + ziji. The compound reflexive taziji can have Zhangsan as its antecedent only in (34b): (34) a. Zhangsan shuo Lisi chang piping taziji Zhangsan say Lisi often criticize himself 'Zhangsan said that Lisi often criticized himself' b. Zhangsan shuo, taziji, Lisi chang piping 'Zhangsan said that himself, Lisi often criticized' Given that taziji may be IP-adjoined in the syntax (as in (34b)), nothing seems to prevent the same element in (34a) from being IP-adjoined in LF. The fact that (34a) does not allow Zhangsan to antecede taziji shows that LF-movement does not alter the binding possibilities of taziji. The index of taziji that is licensed (or "1marked, extending Lasnik & Saito's (1984) terminology) by principle A at Sstructure remains in LF wherever taziji goes. Therefore, not only must binding theory apply at S-structure, but the following must also hold, in both Chinese and English: (35) The indices licensed by the binding theory at S-structure cannot be undone inLF. This means that binding theory, if it applies in LF, can affect only NPs whose indices are not already licensed at S-structure with respect to specific binding principles. These considerations, however, contradict the hypothesis that longdistance binding of ziji arises as a result of LF-movement. A sentence like (30) is assigned the reading (31a) at S-structure, and, given (35), LF-movement should not be expected to derive representations like (31b) or (31c). Note that this is a problem not only for the hypothesis we are entertaining, but also for the Inflmovement or clitic-climbing account discussed in the preceding section. Clearly, the difference in referential behaviour between the bare reflexive in Chinese and the compound reflexive in Chinese and English must be tied to their difference in form. Note that the bare reflexive is more 'anaphoric' than the compound reflexive, in that it not only lacks inherent reference as the compound reflexive does, but also contains less 'sense' than the latter: it does not contain '
The local nature of the long-distance reflexive in Chinese
275
Zhangsan shuo Lisi chang piping ziji Zhangsan say Lisi often criticize self 'Zhangsan said that Lisi often criticized LisilZhangsan'
Let
Zhangsan(
On the other hand, Zhangsan and Lisi each have inherent
Zhangsan(
In LF, the
[ZhangsanCi,3) shuo [Lisi(i,2) chang piping ziji Ci ,2)]]
If ziji(i,o) is IP-adjoined in LF, then the LF-structure of (38) is either (40) or (41): (40) (41)
[ZhangsanCi,3) shuo bp zijiCi,o) [IP Lisi Ci ,2) chang piping t(i,O)]]] bp zijiCi,o) [IP ZhangsanCi,3) shuo [IP tCi,O) [IP LisiCi,2) chang piping t(i,O)]]]]
276
G.-T. James Huang and C.-C. Jane Tang
At LF, when binding theory applies again, (40) can be licensed if ziji is assigned either the R-index of Zhangsan or that of Lisi, as either (i, 3) or (i, 2). In the former case, ziji in IP-adjoined position is bound in its governing category in accordance with principle A. In the latter case, it is 'chain-bound' by Lisi, in the terms of Barss (1986). Similarly, (41) may be licensed if ziji is R-indexed as either 3 or 2, as in both cases it is properly chain-bound by a 'minimally chain-accessible' antecedent. The binding possibilities of ziji as provided in (40) and (41) are thus on a par with those of the English reflexive in (27b, c) and (28b-c) above. The only difference is that, whereas the ambiguity arises in English only in the syntax, it arises in LF in Chinese, giving ziji, from the point of view of S-structure, the appearance of a long-distance anaphor. According to our proposal, then, a systematic difference between the bare reflexive and the compound reflexive is accounted for in a principled way. Since the compound reflexive (in both English and Chinese) has an inherent
(43)
a. b. c. a.
You remember John bought what (D-str) What do you remember that John bought t? (S-str) (S-str) You remember what John bought t shenme]]? (D- & S-str) [Ni jide [Zhangsan mai-Ie you remember Zhangsan buy-ASP what (LF) b. [Shenmei [ni jide [Zhangsan mai-Ie til]]? 'What do you remember that Zhangsan bought?' c. [Ni jide [shenmei [Zhangsan mai-Ie til]] (LF) 'You remember what Zhangsan bought'
This difference presumably concerns the level at which a [+wh] Comp is licensed by a [+wh] phrase that fills it. Likewise, the difference between the bare reflexive and the compound reflexive is accounted for as one between licensing an R-index at S-structure and doing that at LF.
The local nature of the long-distance reflexive in Chinese
277
It should be clear by now that, besides providing a principled account for the different properties of two kinds of reflexives, our proposal also resolves the potential contradiction just noted, between principle (35) and the idea that longdistance ziji arises as a result of LF-movement. The two conflicting ideas are jointly satisfied by ziji. The
a.
b.
c.
Ni shuo Zhangsan chang piping ziji you say Zhangsan often criticize self 'You said that Zhangsan often criticized himself' Wo juede women bu yinggai piping ziji not should criticize self I feel we 'I feel we shouldn't criticize ourselves' Ta juede women zongshi piping ziji he feel we always criticize self 'He said that we always criticize ourselves'
In (45a) the local subject and the matrix subject differ in person features, in (45b) they differ in number, and in (45c) they differ in both. On the other hand, since Chinese pronouns are not marked for their gender features (cf. note 1), we can
278
C.-T. James Huang and c.-c. Jane Tang
assume that gender does not playa role in defining a
Li Xiaojie shuo Zhangsan zongshi piping ziji Li Miss say Zhangsan always criticize self 'Miss Li said that Zhangsan always criticized himself/her'
We thus correctly account for the fact that a remote NP can antecede ziji only nit has the same
Zhangsan shuo wo juede Lisi zongshi piping ziji Zhangsan say I feel Lisi always criticize self 'Zhangsan said that I feel that Lisi always criticized himself'
As indicated in the introductory section of this chapter, in sentences like these, ziji also cannot be bound by the remote subject Zhangsan, in spite of the fact that Zhangsan agrees with the local subject Lisi in
The local nature of the long-distance reflexive in Chinese
279
We can now explain the blocking effect observed in (47). Given (48) and the earlier requirement (35), ziji must be directly bound at all times in LF. In order to be bound by the matrix subject, ziji must be IP-adjoined in LF. In accordance with the successive-cyclic requirement, ziji must first adjoin to the lowest IP before adjoining to the next-higher IP. This required derivation is blocked at the first adjunction site, however, since ziji (with the
(50)
Zhangsani shuo [zijii/*j de shu, [Lisi j zui xihuan]] Zhangsan said self's book Lisi most like 'Zhangsani said that, self'si/*j book, Lisi j likes most' Zhangsani shuo [tazijii/j de shu, [Lisij zui xihuan]] 'Zhangsani said that himself'si/j book, Lisij likes most'
In (49-50), a reflexive has been moved in the syntax. In (49) ziji can only take the matrix subject as its antecedent, though in (50) taziji may also take the embedded subject as its antecedent. The contrast shows that ziji can only be directly bound, thus providing independent evidence for (48). As for the requirement of successive-cyclicity, we assume that this comes from considerations of 'economy of derivation', which requires movement to take place in short steps whenever it can (see Chomsky (1988b) for discussion). Our theory predicts that, although the long-distance ziji exhibits successivecyclicity, it does not exhibit subjacency, CED or ECP effects. Consider (51-2) (see also (25-6)): (51)
(52)
ta zhidao [[suiran Lisi piping-Ie ziji], dajia haishi hen xihuan tal he know though Lisi criticize-ASP self all still very like him 'Rei knows that although Lisij criticized selfi/j' we still like him' Lisi bu xihuan [[piping ziji de] neige ren] criticize self REL that person Lisi not like 'Lisii does not like the personj [who criticized selfi/j]'
These sentences show that ziji may be bound by an NP outside an island (adjunct or complex NP). Under our analysis, the relevant readings are obtained after ziji is
280
C.-T. James Huang and C.-C. Jane Tang
IP-adjoined successive-cyclically in LF. The adjunction process must be allowed to cross the island barriers, given the well-known fact that subjacency and CED do not obtain in LF, though ECP does (Huang (1982), Lasnik & Saito (1984) and Chomsky (1986b)). 8 Furthermore, since ziJi occurs in an argument position where it is lexically governed, its LF-adjunction to IP (an instance of XP-movement) is not subject to antecedent-government. We thus correctly predict long-distance binding to be possible in (51-2). On the other hand, under the head-movement analysis, movement of ziJi leaves an XO trace that falls under the requirement of antecedent-government (or the head movement constraint, see Chomsky (1986b, 1988b) for discussion). This analysis thus makes the wrong prediction that in sentences like (51-2) ziJi cannot have a long-distance antecedent.
5 Conclusion Our proposal is thus superior to the IntI-movement theory in two essential ways. First, we derive the locality properties of the so-called long-distance ziJi from the idea that ziji undergoes A' -movement (not head-to-head movement), regulated by the requirement of successive-cyclicity but not subject to antecedent-government. And we correctly predict that the long-distance ziJi exhibits only successive-cyclicity effects, but no ECP effects. Secondly, we derive the blocking effects from the common assumption that binding theory applies at both S-structure and LF, and capture the important generalization that the set of potential blockers of longdistance ziji is exactly the set of its closer potential binders. Our proposal also differs from certain previous approaches to long-distance anaphora. Instead of parametrizing the notion of a governing category for ziJi, or defining it as a non-anaphor or 'semi-anaphor' of some sort, we treat it as the 'most anaphoric' of all anaphoric elements, and show that it in fact obeys very strict locality restrictions. 9
Notes Earlier versions of this paper were presented at the Groningen Workshop on Long-Distance Anaphora and at a colloquium at the Centre de Recherches Linguistiques sur l' Asie Orientale, Paris, in 1987, and at a colloquium at Cornell University and at NELS 19 in 1988. We are grateful to the audiences, especially Gennaro Chierchia, Wayne Harbert, Chillin Shih, Peter Cole, Lisa Cheng and John Whitman for comments and suggestions. 1. But the local NP and the remote NP need not agree in gender features, as shown in (i): (i)
Li Xiaojiei shuo [Zhangsanj zongshi piping ziji vj ]
The local nature of the long-distance reflexive in Chinese
281
Li Miss say Zhangsan always criticize self 'Miss Lii said that Zhangsanj always criticized her/himselfj' This fact is apparently related to the fact that pronouns in Chinese are unmarked for gender, as in ta 'himlher' and taziji 'himself/herself'. We assume that gender is not a grammatical feature in Chinese, and thus does not enter into the determination of grammatical binding of anaphors. 2. For some speakers binding of ziji by the intermediate subject is quite marginal. That is, zifi appears to be bindable by the minimal clause subject or the maximal clause subject, but not by any intermediate subjects. This appears, however, not to be an absolute condition. Following Y.-H. Huang (1984) and Battistella (1987) we shall assume this to be an extragrammatical effect. 3. This formulation differs slightly from that given in Tang (1989). 4. This explanation is confirmed by the fact that, if the local NP is inanimate and therefore not a potential binder, acceptability improves for a sub-commander being a long-distance binder: (i)
Zhangsani de xin biaoshi [neiben shu hai-Ie zijiil Zhangsan 's letter indicate that book hurt -ASP self 'Zhangsan/s letter indicated that the book hurt himself/
As Wayne Harbert has suggested to us, it may be that since both long-distance binding and binding by a sub-commander each represent a marked case, long-distance binding by a sub-commander would be doubly marked and thus not allowed. 5. In cases where an inanimate subject contains an animate subject, Battistella suggests that the whole inanimate subject carries the index of the animate subject it contains. This means that a sub-commander contained in a c-commander is treated as if it is the actual ccommander. But this makes an incorrect prediction about sentences like (i) with respect to principle B, that in (i) ta must be disjoint from Zhangsan, which would 'c-command' it in its governing category: (i)
[[Zhangsan nadao del chengjil shi ta hen nanguo DE grade make he very sad Zhangsan get 'The grade that Zhangsan got made him very sad'
Cases like (i) show that 'sub-command' is not required in the characterization of a governing category for disjoint reference, just as the presence of an accessible Subject is irrelevant for the domain of principle B (see C.-T. J. Huang (1983), Chomsky (1986a». This means that 'sub-command' must be clearly distinguished from c-command. 6. Cole et al. (forthcoming) have recently developed a version of Battistella's theory in which they attempt to answer some of our objections. We hope to return to their paper some time in the future. 7. We thus follow Chierchia (forthcoming) and claim that the long-distance reflexive has the semantics of an operator. Katada (1988) argues for the same treatment of Japanese zibun by showing that it exhibits extensive properties of an operator. 8. One might say that subjacency and CED do obtain in LF, but due to independent factors their effects are not visible. See Pesetsky (1987a), Nishigauchi (1986) and Fiengo et al. (1988) for somewhat different executions of this idea.
282
C.-T. James Huang and C.-C. Jane Tang
9. We have not addressed the problem of subject orientation. An important claim of the Inflmovement theory is that subject orientation follows from the fact that ziji is located in Infl. However, it is not entirely clear that subject orientation is a property of long-distance reflexives, or bare reflexives alone. The compound reflexive taziji, for example, also exhibits a strong tendency for subject orientation, as indicated below: (i)
Zhangsani gaosu Lisi j tazijiv*j de shenshi Zhangsan tell Lisi himself's life-story 'Zhangsani told Lisi j about hisv*j own life'
14 Anaphors and logophors: an argument structure perspecttve Tanya Reinhart and Eric Reuland
1 Introduction
It has been known for quite some time that the binding theory developed in Chomsky (1981) and subsequent work does not account for the full range of binding facts (cf. Maling (1982), Giorgi (1984), Chomsky (1986a), Everaert (1986a), and others): The scope of conditions A and B is limited to binding within the domain of the first accessible subject, the local domain. Therefore, non-clausebounded reflexives, which are commonly referred to as long-distance (LD-) anaphors in languages as diverse as Icelandic, Finnish, Polish, Latin, Chinese, Japanese, etc., cannot be captured (see the various contributions in this volume for discussion and references). Even within the domain to which conditions A and B apply, language does not fully live up to their predictions. In a number of environments pronouns can be found where condition B excludes them. Well-known examples are English sentences of the type John saw the snake near him, reflexive verbs in Dutch and French which allow locally bound first and second person pronouns in object position, and Frisian, where this latitude also obtains with third person object pronouns (see Everaert (chapter 4), and Bouchard (1984)). For years, a unified theory of the various anaphors across languages seemed somewhat unfeasible, in view of the massive differences reported concerning their distribution, particularly in the case of LD-anaphors. A major breakthrough, however, has been the discovery that a distinction is needed between logophoric processes and structural binding relations. The differences found by Anderson (1986), Hellan (chapter 2) and Thniinsson (chapter 3) between LD-binding into finite clauses (indicative or subjunctive) and LD-binding into infinitival clauses indicate that only the latter falls under the structural binding theory. The other occurrences of LD are logophoric. As such, their distribution is much freer, and they are governed by discourse, rather than by structural considerations. As we will see, also the so-called local anaphors (like himself) allow logophoric uses, a fact which has led to certain unclarities concerning the domains of conditions A and B.
283
284
Tanya Reinhart and Eric Reuland
This enables us to distinguish three domains of anaphors: local, medium-distance, and logophoric (see Reuland & Koster, chapter 1). The structural binding theory should capture only the first two (although the interaction between binding and logophoricity is a bit more complex, as we shall see). In traditional terms, local binding respects the specified subject condition (SSC), and medium-distance (MD-) binding violates the SSC, but respects the tensed S condition. 1 As observed by Pica (1985, 1987) and further discussed in chapter 5, complex anaphors must be bound in the local domain, whereas MD-anaphors are simplex. English himself, Dutch zichzelJ, Norwegian seg selv, Italian se stesso, Finnish han itse, etc. instantiate the former category (henceforth SELF-anaphors), whereas in the simplex class one finds, e.g., Latin se, Dutch zich, Norwegian seg, Italian se, Finnish itse (henceforth (simplex expression) SE-anaphors). 2 The observation that there are only two domains for structural binding, for which we will provide further support in this article, should enable a great simplification of the binding theory: It eliminates the need for relativized, or parametrized, binding domains (the problems related to such domains are summarized by Reuland & Koster (chapter 1». Nevertheless, there is no obvious way to capture the full range of the distribution of anaphors within the binding theory as presently stated. Below we give a summary of the major facts about the distribution of pronouns and anaphors across languages, which the binding theory should capture, including the one we have just discussed (for a detailed survey of points (a-c), see Reuland & Koster in chapter 1). (a) There are two structural binding domains: the domain of the first (accessible) subject (the local domain), and the domain of the first finite Infi (the MDdomain). SELF-anaphors are bound within the local domain, SE-anaphors in the MD-domain. (b) SE-anaphors are subject oriented. For SELF-anaphors, subject orientation depends on the constituting elements: anaphors of the form (pronominal) SELF are not subject oriented, but those of the form (SE) SELF are (see also Hellan in chapter 2). (c) Complementarity effects obtain between (non-Iogophoric) SELF-anaphors and pronominals and (generally) between SELF-anaphors and SE-anaphors, but not generally between SE-anaphors and pronouns. (d) Both SELF- and SE-anaphors also allow logophoric uses, outside the domains specified in (a). A logophoric SELF-anaphor is not in complementary distribution with pronouns. Although it seems possible to capture most of these facts by a revision of the feature system underlying the current binding theory, as proposed by Everaert
Anaphors and logophors: an argument structure perspective
285
(chapter 4), we will attempt a more explanatory line for deriving all of them. 3 In terms of an LF-movement analysis of anaphors, as proposed in Chomsky (1986a), our analysis entails that there are two landing sites for anaphors, rather than just 1°. While the SE-type adjoins, indeed, to 1°, the SELF-type adjoins to vo, yielding a semantically reflexive predicate. This explains the apparent SSC-effects of the second, and both the TSC-effects and the subject orientation of the first. (Although we will assume a movement line, in the case of SELF-anaphors, the same analysis can be stated also in familiar terms of binding filters, as needed, anyway, for condition B.) This difference in landing sites is directly related to the different inherent properties of the two anaphor types. The full range of complementarity effects requires a reformulation of condition B and (the movement equivalent of) condition A, in terms of argument (9-) structure. Their reformulation will also determine (indirectly) the environments allowing logophoric SELFanaphors, since these are possible only where the binding conditions do not apply. Our analysis covers only the binding of pronouns and anaphors. This differs from the original government binding view of binding conditions A and B as regulating also NP-movement (traces) and the ungoverned nature of PRO. However, given recent developments such as Chomsky (1986b), the first is no longer governed by the binding theory, anyway, and there are several approaches with the same result for PRO. For reciprocals, we assume the movement analysis in Heim, Lasnik & May (1988), which yields them, essentially, independent of the binding conditions, and we will not discuss them here. 4
2 Anaphoricity We assume a definition of anaphors along the following lines (see, for instance, Chomsky (1981), Bouchard (1984), Keenan (1987) and Thrainsson (chapter 2)): anaphors (of both kinds) are referentially defective NPs, which entails, for example, that they cannot be used deictically. We will view binding as the procedure assigning the content necessary for their referential interpretation. It is, first, necessary, however, to determine which linguistic properties make these expressions referentially defective. Specifically, SE- and SELF-type anaphors differ in their lexical structure, and hence in the type of content they lack, and in the procedures needed to supply it. With Higginbotham (1983) and Abney (1987) we will assume that the canonical structure of NPs is as indicated in (1), where the determiner is viewed as saturating, or discharging, the argument in the N-grid. (We will mostly ignore the internal structure of DPs. For convenience, we will continue to refer to nominal arguments as NPs).
286
Tanya Reinhart and Eric Reuland
[We will assume that pronominals occur in determiner position, although in the case of simplex pronominals nothing hinges on this. Yet they project as full NPs, as in (2). (N is projected on the basis of the (f)-features of the pronominal.) (2)
[NPIDP
Pron [N' ... e ... J]
We will turn now to the 'defective' anaphoric expressions under consideration. SE-anaphors like Norwegian seg, Dutch zich or Icelandic sig differ from pronouns in that they lack (a complete specification for) (f)-features (number, gender, person; see Chomsky (1981)) and hence do not project an argument that can be interpreted independently. We take this lack of (f)-features to be the property responsible for their defective nature. Crucial for our analysis is that they are in the head position of their projection. Although their precise structure does not directly concern us here, we will be assuming that they are structurally identical to pronouns (together this requires a DP- rather than an NP-analysis). (3)
[NP/DP
SE ... J
In the case of SELF-anaphors, SELF is a noun, rather than a determiner. Following the line in Pica (1987), we assume that SELF has the lexical structure of a relational noun, i.e. its grid has two arguments, as in (4). (4)
SELF(y,x)
Semantically, we view SELF as an identity relation (identifying x and y). Combining SELF with a pronoun determiner, one obtains the structure in (S). (S) a. himself b. [Nphim [N' SELF (y,x)J
I
I
The pronoun in (Sb) discharges one of the argument positions in the grid (in the case of Higginbotham (1983)). However, the NP still contains one unsaturated argument (the second argument of SELF). In more intuitive terms we may say that SELF in (Sb) expresses an identity relation between him and another argument which needs to be found. Under this view, it is this missing argument which is responsible for the defective nature of SELF-NPs, i.e. for their anaphoric status. To be interpreted, the second argument must be identified, in a way we turn to in the next section. However, the specific details assumed here for capturing the SELF-relation are not all crucial for our analysis, and other formulations with the same results are conceivable.
Anaphors and logophors: an argument structure perspective
287
We will argue that the distribution of the two types of anaphors, as described in section 1, is directly related to their defective properties. First, the lack of 'Pfeatures in SE-anaphors is responsible both for their subject orientation and for their Infl-Iocality constraints: to get their
3 Local binding: SELF-anaphors Problems with the current binding theory have led several scholars to propose that binding conditions A and B cannot be stated in purely configurational terms. Rather, a more direct connection between 8-theory and binding theory is needed than presently assumed. 7 Williams (1987), which is based on evidence from binding of implicit arguments, argues that the binding conditions operate (only) on argument (8-) structures. Williams's formulation of the binding conditions is essentially equivalent to the configurational formulation (in the case of explicit arguments). However, Chomsky (1986a) and Giorgi (1984, 1987) argue that the definition of the governing category in Chomsky (1981) must be replaced by one based on the notion of a complete functional complex (CFC), a projection in which all thematic roles of its head are realized. 8 While we follow the line that the binding theory applies to argument (thematic) structure (particularly, in the spirit of Chomsky and Giorgi), we believe that a more radical change in the view of conditions A and B is needed. In fact, the domain relevant for their operation is more local than currently assumed, and they
288
Tanya Reinhart and Eric Reuland
operate only on coarguments of the same a-grid (see Everaert (1986a) and also Hellan (1988)). In some sense, this means returning to the more traditional view of reflexivization. In earlier studies a much closer relation between anaphora and argument structure (avant la lettre) was assumed. Reflexivization used to be analysed as a property of predicates. A typical statement is that a reflexive pronoun is used in the object position (to avoid 'repetition of the subject') when the verb expresses a reflexive relation (e.g. Jespersen (1933), Gleason (1965)). That is, the heart of the phenomenon is taken to reside in the nature of the relation expressed by the verb. An obvious reason why this traditional view was rejected is that modern anaphora theory discovered many cases where the antecedent and the reflexive pronoun are not on the same grid, most notably in picture-type NPs and in ECMstructures. However, we will argue that some of these cases are reducible to binding within the same grid, and the others, which are not, are instances of logophoricity, rather than local anaphora. Since we essentially reduce the binding theory to an analysis of reflexive marking, let us start with the concept of reflexivity assumed. 3.1 Reflexive and non-reflexive contexts of SELF-anaphors A predicate (verb, noun) denotes a reflexive relation if and only if two of its arguments are identical, as in (6). (6)
AX(P( ... ,x, ... ,x, ... ))
We will refer to predicates of the form (6) as reflexive predicates. For example, the predicates in (7a) and (8a) are reflexive, since two of their arguments are identical. (7) a. b. (8) a. b.
Luciei adores herselfi *Luciei adores heri Lucie'si joke about herselfi *Lucie'si joke about heri
Given this definition, it is trivial to observe that SELF-anaphors do not always form reflexive predicates. Two famous contexts where they do not are illustrated in (9) and (10). (9)
a. Luciei saw a picture of her/herselfi in the paper b. MaXi enjoyed the jokes about him/himselfi (10) a. MaXi saw a snake near him/himselfi
Anaphors and [ogophors: an argument structure perspective
289
b. Maxi counted five foreigners in the room apart from him/himselfi The anaphor and its antecedent are not coarguments of the predicate here. In (9) the anaphor is embedded in an argument, and in (10) it is an adjunct PP which cannot be viewed as an argument required by the verb's grid. Our starting point in arguing that a revision is needed in the binding theory, so that it distinguishes these two contexts, is the well-known fact that in the contexts of (9) and (10) anaphors are not in complementary distribution with pronouns. This distinguishes them from the contexts of reflexive predicates, where complementarity is obtained. In these contexts, bound pronominals are not allowed ((7b) and (8b». Given that condition A is formulated to allow reflexives in precisely the same contexts that condition B disallows pronominals, this is a serious problem which cannot be ignored. We argue that the binding conditions govern only the reflexive use of anaphors, and, consequently, neither condition B nor condition A applies to (9) and (10). More generally, non-complementarit.y is found in environments not subject to either condition. This move may seem disturbing, since we are used to the view that the distribution of SELF-anaphors is fully governed by condition A, i.e. that they are interpretable only in the environments specified by this condition. However, there is independent evidence that this view is, anyway, too restrictive: contrary to the prevailing assumption, long-distance (logophoric) anaphora is quite common also with SELF-anaphors, and, specifically, in English. In section 6, where we examine the logophoric use of anaphors, we will see that when a SELF-anaphor occurs in an adjunct or embedded position, as above, it can be long-distance bound in a way that cannot conceivably be allowed under the current condition A. If this is the case, there is no reason to assume any difference between the contexts in (9) and (10) above and the other instances oflogophoric anaphora. That these contexts are, indeed, logophoric and should not fall under the binding theory is witnessed also by the fact that the anaphor in these environments is permitted even when not bound (c-commanded), as in (11a) and (12a) below, which cannot possibly be allowed by condition A. This contrasts, again, with the core cases in (11b) and (12b), where the anaphor is in an argument position of its predicate's grid ((12a) is an example from an actual discourse, quoted in Zribi-Hertz (1988) from Taylor). (11)
a. b. (12) a. b.
Max's eyes watched eagerly a new picture of himself in the paper *Max's eyes watched himself (in the mirror) with a new eagerness Bismarck's impulsiveness had, as so often, rebounded against himself *Bismarck's impulsiveness had, as so often, failed himself
290
Tanya Reinhart and Eric Reuland
Other contexts where picture NPs do not obey c-command are discussed in e.g. Lebeaux (1988). Let us turn, then, to the formulation of the binding theory that will capture these facts. 3.2 Reflexive marking A universal property of natural language seems to be that reflexivity must be licensed. A predicate can be reflexive, i.e. denote the relation in (6), repeated in (13), only if it is linguistically marked as reflexive. This generalization is stated in (14). (13)
AX(P( ... , x, ... , x, ... ))
(14) A predicate is reflexive only if it is reflexive-marked Predicates can be marked reflexive intrinsically, or extrinsically. The intrinsic ones have their grid specified as reflexive in the lexicon, and they can be used only reflexively. They may, or may not, be also overtly morphologically marked. Adopting (15a) as the general format of thematic grids (see Higginbotham (1985)), their grid can be represented as in (15b), where we take this notation to mean that the two arguments of the verb (or any other predicate) must be variables bound by the same operator. 9 (15) a. V (x,y, ... ) b. V (x,x, ... ) We turn to some cases of intrinsic reflexives (like the Dutch schamen 'to shame') in section 3.6. Crucially, however, a transitive predicate which is not intrinsically reflexive may turn into a reflexive predicate, if (and only if) it is lexically marked as such. Specifically, SELF-anaphors (self(English), zelf(Dutch), sjdlfur (Icelandic), etc.) function as such reflexivizers. As we saw, this happens whenever they occur on a grid position of the predicate. We turn to the role of clitic anaphors in reflexivization in section 3.6.2, but here we should stress that SE-anaphors clearly do not function as reflexivizers of this sort. Let us state this as the generalization in (16). (16) A SELF-anaphor or an argument (grid) position of a predicate P reflexivizes the grid of P. Our next question is what is meant by 'reflexivizing P', or how a reflexive interpretation is derived compositionally. A semantic analysis of SELF as a reflexivizer was proposed by Keenan (1987). He argues that SELF is interpreted as a function from binary relations to properties (i.e. sets). In the VP case it applies to a transitive verb yielding an intransitive reflexive verb (Max criticized himself will be
Anaphors and logophors: an argument structure perspective
291
analysed as Max SELF(criticized), where SELF(criticize) denotes the set {xix criticized x}). We would like to obtain an equivalent result in a way more isomorphic with the actual argument structure of both the predicate and the SELF -NP, preserving, e.g., the transitive a-structure of the verb and acknowledging the existence of the pronoun in the SELF-NP, which, ideally, should not just vanish in the interpretative stage. As mentioned in section 2, we assume that SELF denotes an identity relation between two arguments, as in (17a). When a SELF-NP occurs on a grid position, it is interpreted as a restriction on that grid, which, consequently, restricts the range ofthe predicate interpretation. For example, a VP like (17b), which contains a SELF-argument, is interpreted as in (l7c), where the SELF-conjunct specifies that the two arguments of the V grid must be identical variables (where identity of variables means, again, being bound by the same operator). (17)
a. SELF(y,x) b .... [V(y,x) .. , [... SELF(x,y)JJ c. SELF(x,y) & V(y,x) ...
The interpretative result, then, is equivalent to that of the intrinsically marked reflexive in (lSb). Alternatively, SELF can be viewed as an operator from Vn to Vn (rather than as a relation), which applies to the V and identifies two of its argument, i.e. actually assigns the V in (17b) the grid of (iSb). This will entail a different perspective on the internal argument structure of SELF than presented in section 2, and at present we see no reason to prefer this option. An LF of the form (17c) can be derived by adjoining the N° SELF to V by headto-head movement. The LF of (18a), then, will be (18b). (18)
a.
Lucie adores herself
IP
b.
~
LUd~v~ SELF(x,y)
ry'X)
N~
adores
herx
(e)
Recall that the pronoun-determiner in the reflexive NP saturates one of the SELFpositions (thus giving the NP the index of this position). So herx and Lucie y are the
292
Tanya Reinhart and Eric Reuland
two arguments required by the V grid, and the SELF-relation specifies that these two must be identical. On this view, then, the reflexivizing function of SELF-anaphors is obtained by their obligatory movement to VO (or any predicate pO), in the contexts specified in (16). The restrictions on this movement are completely standard, so, in principle, they need not be stipulated. However, we will introduce directly a modification in the contexts relevant for (16) which is not currently reducible to known principles of movement. For this reason we will keep (16), at least for the time being, as an underived generalization, which may, however, be read as (16'). (16')
SELF occurring on a grid position of a predicate P adjoins to P.
We are assuming standard constraints on head-to-head movement. Once SELF is attached to V (or another relevant head), it cannot leave it and travel further, which would violate the head-movement constraint of Chomsky (1986b). Nor can it, under the Barriers framework, skip the V positions and land directly in a higher position. Once attached to V it can travel with the V to I. But this movement no longer affects the interpretation, since a SELF attached to V is interpreted in association with this V, as reflexivizing its grid, regardless where V gets interpreted. This entails strict locality for SELF-anaphors as reflexivizers. 3.3 Conditions A and B While the generalizations regarding reflexivization may seem trivial, it is less trivial to argue, as we do now, that these, in fact, are binding conditions A and B. The conditions are repeated in (19). But note that A contains a modification of (16) that will be explained in section 3.5. Given our analysis above, A and B of (19) are not of identical status. While B is a filter condition on appropriate derivations, as in the standard binding theory, A regulates, in effect, obligatory movement and may be reducible in the future to independent constraints on movement. For convenience, however, we will keep referring to it as condition A. Note also that it is not crucial to analyse A as a condition on movement. It too can be formulated as a filter on derivations, and a formulation along such lines is given in Reinhart & Reuland (forthcoming). Condition B, as we state it here, is slightly simplified, and its further details (which are discussed in Reinhart & Reuland (forthcoming)) are summarized in note 22.
(19) A. B.
A SELF-anaphor on an argument (grid) position of a fully assigned predicate P reflexivizes the grid of P. A predicate is reflexive only if it is reflexive marked.
As in Chomsky (1986a) we assume that the standard hierarchical effects of the
Anaphors and logophors: an argument structure perspective
293
antecedent-anaphor relations (distinguishing, e.g., I expect Max to praise himself from *1 expect himself to praise Max) follow from the analysis in terms of movement. However, a constraint prohibiting nominative SELF-anaphors is still required, for reasons discussed in Everaert (1986a).lO Apart from the conditions in (19), we assume only the generalization, underlying the current binding theory, that all anaphoric relations are instances of syntactic binding, where a binds b if and only if a and bare coindexed and a c-commands b, and we use c-command and m-command interchangeably. (An exception to that is, of course, the case definite-NP coreference, which, we assume, requires a separate inference mechanism, as proposed in Reinhart (1986).) This entails that any anaphoric element (e.g. pronoun, trace, etc.) is interpreted as a bound variable only if syntactically bound (for further details, see Reinhart (1986), though none of these details are crucial for the present analysis). Let us see first how these conditions cover the core cases of the binding theory. It should be recalled that a predicate denotes a reflexive relation whenever two of its arguments are variables bound by the same operator. Suppose, now, that in (20) him is coindexed with Max. (20)
a. Maxi criticized himi b. Max (AX(X criticized x))
Since the pronoun is bound, it could, in principle, be translated as a bound variable yielding (20b). The result, however, is a reflexive predicate. Condition B of (19) allows such predicates only if they are reflexive marked. This is not met here, since there is neither a SELF-anaphor nor any intrinsic reflexive marking. Hence, the derivation is filtered out. The same is true if an SE-anaphor occupies the same position. SE are obligatory anaphors, and not defined as reflexivizers. So, the predicate still does not meet the requirement in A. This is illustrated by the examples in (21). (21)
a. b.
Joni John Jani John
foraktet despises veracht despises
seg selv/*seg/*hami (Norwegian) (Hellan (1988)) himself/SE/him zichzelf/*zich/*hemi (Dutch) himself/SE/him
It follows, then, that a pronoun, or an SE-anaphor, cannot be interpreted as identical with another argument of the predicate (unless the predicate is intrinsically marked, as in the cases we turn to in section 3.5).11 It also follows that pronouns (or SE-anaphors) should be complementary to SELF-anaphors only in reflexive contexts. Although a SELF-anaphor may occur in (9a) and (lOa), repeated in (22), no reflexive predicate is formed in their
294
Tanya Reinhart and Eric Reuland
interpretation, since the two identical variables are not co-arguments. Hence, if a pronoun occurs in the same position, it can be translated as a bound variable without violating condition B. (22) a. Luciei saw a picture of her/herselfi in the paper b. Maxi saw a snake near him/himselfi As we turn to condition A, consider (23) and (24). (23) a. Luciei criticized herselfi b. Lucie'si joke about herselfi (24) a. *Lucie criticized myself b. *Max liked Lucie's joke about himself c. *Max promised Lucie [PRO to respect herself] In both (23) and (24) a SELF-anaphor occurs on a grid position and condition A applies, requiring that SELF adjoins to the predicate, reflexivizing its grid (yielding, e.g., SELF criticize and SELF joke). In (23) interpretation can then proceed and form a reflexive predicate. An antecedent is available, and so two arguments of the grid can be identified. In (24), however, no matching antecedent can be found within the predicate's grid. Therefore we obtain a predicate which is marked as requiring the identification of two of its arguments (see section 3.2), and this requirement is not met. The sentence is therefore ruled out as a violation of the grid requirement. Condition A applies only in those cases where the anaphor occupies a grid position of the relevant predicate. In these cases it has precisely the same effect as the standard condition A of Chomsky (1981) - it filters out any occurrence of an anaphor free in that grid (Le. not bound by another argument on the same grid), as in (24). However, when the anaphor is not on the grid, the condition says nothing about it, so logophoric instances are allowed. (We turn to some obvious apparent problems directly.) Let us look, for illustration, at the contrast in (12), repeated in (25). (25) a'. Bismarck's impulsiveness had, as so often, rebounded against himself b. *Bismarck's impulsiveness had, as so often, failed himself In (25b), with an anaphor on a grid position, the predicate fail is marked as reflexive by condition A. Since the antecedent Bismarck is not an argument of the verb's grid, no argument can be identified within the grid, and the derivation is filtered out. In (25a), the anaphor occurs in a non-grid PP, so condition A does not apply. Since the predicate rebound is not reflexivized, no interpretative requirement is violated by this sentence. Due to its defective nature (see section 1), the
Anaphors and logophors: an argument structure perspective
295
anaphor still needs to find an antecedent, logophorically, but this does not require binding. The details of the formulation of the conditions in (19), and how they operate on the full range of local anaphora are discussed in Reinhart & Reuland (forthcoming). Here, however, we will look briefly at two obvious questions that are raised for the analysis we propose: ECM-structures and picture NPs. 3.4 ECM-structures The most obvious problem for our approach is SELF-anaphors occurring in the subject position of ECM-structures, as in (26), where the anaphor is not an argument of the matrix predicate and still must be bound by an argument of that predicate. (26) a. Max considers [himself/*him a genius] b. Max expects [himself/*him to pass the exam] We should note, first, that given the syntactic restrictions on movement, there is no problem in allowing SELF to adjoin to the matrix verb here. For the sake of explicitness, condition A should be formulated as below, to allow that. (27)
If a SELF-anaphor is L-marked by a fully assigned predicate P, SELF adjoins to P.
This is so, since L-marking (at least under its standard definition in Chomsky (1986b» has been designed, to begin with, to capture both the cases of a-arguments and ECM-subjects. The problem, however, is that construed this way the matrix predicate will be defined as reflexive, and since the anaphor's original position is not on its grid, no two arguments can be identified, so the derivation should be filtered out. (This problem does not arise if our condition A is analysed as a standard (filter) binding condition, as in Reinhart & Reuland (forthcoming).) We believe, however, that once the semantics of ECM-structures is fully developed this problem can be solved. More generally the issue to be addressed here is what, in fact, explains L-marking, i.e. why is it that ECM-subjects appear to share some properties with matrix objects. Semantically, the appropriate interpretation of ECM-structures appears to accord with the verb-complement analysis proposed in Chomsky (1955/1975: 505-35), where (what is now assumed to be) the complement IP-projection is not interpreted as a proposition. Rather, at one level the matrix predicate and the ECM-predicate are amalgamated, as illustrated, informally, in (28), for (26). (Future attention needs to be paid to the precise compositional interpretation of the complex predicates in (28).)
296
Tanya Reinhart and Eric Reuland
(28) a. Max [[considers a genius] [himself]] b. Max [[expects to pass the exam] [himself]] Among the advantages of an analysis along such lines is, first, the fact that it enables us to maintain the generalization that only a CP-projection is interpretable as a proposition. Next, it captures famous contrasts such as illustrated in (29). (29) a. Max saw the table collapse b. Max saw that the table collapsed While in the propositional case of (29b) the sentence can be true if Max has not witnessed the actual event of the table collapsing, but only its results, (29a) cannot be true in the same circumstances. This follows, since in (29a), a complex predicate is formed (roughly, see collapse), so see is modified as the seeing of the collapse. Additional motivation for assuming complex predicate formation for issues of scope is provided in Stowell (1987), though along different lines than exemplified in (28). Suppose, now, that to capture this line of interpretation, a structure with the properties of (28) should be formed at LF. It turns out, then, that the ECMsubject is, at that level, an argument of the complex predicate. A SELF-restriction occurring on that complex predicate reflexivizes, then, the full complex predicate, and as long as two of the arguments of this predicate can be identified, as in (28), the derivation is interpretable. Similarly, if a pronoun occurs in the embedded subject position of (28), condition B will filter it out as a case of unlicensed reflexivization. There are, however, several non-trivial details that should be addressed by an LF-analysis along these lines. Note that we have been assuming (crucially, in fact) that what gets amalgamated is the whole embedded VP (as in Chomsky (195511975)) and not just the head of the predicate (as in Stowell (1987)). Extraction of the VP in this case is unproblematic, since, given the transparent nature of the ECM IP, not only the subject, but also the predicate is governed by the matrix verb. But it is less clear what this VP can adjoin to, assuming that an XP can adjoin only to XP. Therefore, we leave these execution issues for future research. Here we should note, however, that to derive interpretations of the form (28), it is not crucially necessary to actually derive corresponding LFs. For independent reasons, the grammar must contain principles specifying how predicates are construed with predicates (cf. Reuland (1986)). These will be applicable whenever a complement predicate is governed by a matrix predicate. Technically, an indexing chain exists between the matrix verb and the lower VP, which, for interpretation's sake, is equivalent to complex predicate formation. 12 If this is so, condition A
Anaphors and logophors: an argument structure perspective
297
should only require the subject SELF to adjoin to the matrix V at LF. The interpretation procedure creates a complex reflexive predicate in the semantics, so the derivation is interpretable. Condition B is, in any case, viewed here as a filter on interpretations, so if such a complex reflexive predicate is formed with no reflexive marking, the derivation is filtered out.
3.5 Picture NPs The distribution of SELF-anaphors in N-complements seems problematic for an analysis in terms of argument structure. Consider (30). (30) a. Lucie likes pictures of herself b. You should hang a picture of myself on your wall We argued in section 3.1 that the anaphor in (30a) is logophoric rather than reflexive. In section 5 we will see that this is also witnessed by the fact that an anaphor in the same position can be free syntactically, as in (30b). A logophoric interpretation, however, is possible only when condition A does not apply. We assume that picture here is a grid-taking predicate. Although opinions on this issue may vary, the same facts are found with nominalization predicates that clearly have a grid, as in (31). So the question is why condition A does not apply here. (31) a. The city watched the destruction of itself on TV b. The city watched the BBC's broadcast of the destruction of itself This question is particularly striking, in view of the fact that in the contexts of (32) and (33), which contain a subject (controller), the same N-predicates are subject to condition A. (32) a. b. (33) a. b.
?1*Lucie likes Ben's pictures of herself ?1*You should hang Ben's picture of myself on your wall *The city watched the aliens' destruction of itself on TV *The city watched the American destruction of itself on TV
This contrast provided one of the strongest motivations for the standard formulation of conditions A and B (i.e. for defining the governing category as a unit containing a subject). However, even apart from the complementarity problem discussed in section 3.1, this formulation cannot explain why the same contrast shows up also in long-distance cases, as shown by (30b) versus (32b). Note also that the mere existence of a subject cannot be the relevant factor: (31b) is just as acceptable as (31a), although it contains a subject (BBG), which should define the NP as the GC of the anaphor, in which it is inappropriately free. Crucially, though, it is not the subject of the predicate (destruction) of which the anaphor is an argument.
298
Tanya Reinhart and Eric Reuland
We distinguish these two contexts by properties of the predicate's grid. Our analysis closely follows that of Chomsky (1986a) and Giorgi (1987). The main difference is that in our analysis condition A only applies to SELF-anaphors on grid positions. Condition A, repeated in (34), restricts the relevant predicates to those with a fully assigned grid. A domain is a complete functional complex if in it (1) all the a-roles pertaining to a lexical head are assigned, or (2) all the grammatical functions pertaining to that head are realized. Only slightly modified, the definition of full assignment we assume is given in (35). (34) A. (35)
A SELF-anaphor on an argument (grid) position of a fully assigned predicate P reflexivizes the grid of P. A predicate is fully assigned if and only if all of its (obligatory) roles are either syntactically realized, or syntactically controlled.
In (30) and (31) the agent a-role of the grid is not assigned syntactically. With Williams (1984, 1985) and Grimshaw (1986) (among others), we assume that the agent a-role is satisfied in these cases lexically (by the predicate), i.e. there is no syntactic realization of this position. Hence the predicate does not count as relevant for condition A, the anaphor need not be bound within the predicate's grid, and it can find its antecedent logophorically. In the cases of (32) and (33), the same agent role is fully assigned. One need not assume that the possessive specifier actually realizes the agent role or functions as a grammatical subject (though one may, if independently desired). Taking Williams's or Grimshaw's position, it functions as a controller of this role. By (35) a controlled role is defined as fully assigned. The intuition behind this definition is that syntactic satisfaction correlates with the assignment of an actual reference (value) to a role. Since in (32) and (33) there is no unassigned position, condition A applies. The anaphor is required to be bound on its grid, and the derivations are filtered out. Note that in (33b), the agent role is controlled by an adjective, rather than an NP, with the same effect on the predicate's grid. (For more details of this type of control, see, e.g., Grimshaw (1986).) Further discussion of how full assignment is relevant to condition A (e.g. in the case of passive) can be found in Reinhart & Reuland (forthcoming). 3.6 Intrinsically reflexive predicates
3.6.1 Everaert (1986a and chapter 4) has noted an interesting contrast in Frisian, which constitutes a real challenge to the standard binding theory. Like English, Frisian has SELF-anaphors. These are formed by attaching the element sels 'self' to a
Anaphors and logophors: an argument structure perspective
299
pronominaL In standard contexts, a pronominal cannot be bound on the grid, as we would expect on any version of the binding theory. This is illustrated in (36) with the verb haatsje 'hate'. (36)
Ik haatsje *my/mysels; hy hatet *'mI'msels I hate *me/myself; he hates *himlhimself
However, other verbs, like waskje in (37), allow both SELF-anaphors and pronouns as their arguments. (37)
du waskest dy/dysels; hy wasket 'mI'msels you wash you/yourself; he washes himlhimself
Even more surprisingly, we also find verbs with which a pronominal is preferred over the anaphor, as in (38), with the verb skamje 'to be ashamed'. (38)
Ik skamje my/??mysels; hy skammet 'mI??'msels I shame me/??myself; he shames himl??himself 'I am ashamed; he is ashamed'
Everaert shows that these differences follow from the inherent properties of the verbs. A verb like skamje is inherently reflexive, as witnessed by the fact that it cannot take any object distinct in reference from the subject. The generalization that he draws (couched in somewhat different terms) is that an intrinsically reflexive predicate can occur only with pronouns. As for the class of verbs allowing both a pronoun and an anaphor, such as 'wash' in (37), Everaert argues that they are doubly listed in the lexicon, both as reflexive and as non-reflexive. (So the first selects a pronoun, while the second selects a SELF-anaphor.) Independent support for this claim comes from the interpretation of nominalizations. When subject and object arguments are omitted, waskje readily admits a reflexive interpretation, whereas haatsje does not. This is illustrated by the contrast between (39a) and (39b). (39)
a.
waskjen is sun washing (oneself) is healthy
b.
haatsjen is net sun hating (only someone else) is unhealthy
Let us see now, how these facts follow from our analysis. The crucial question is why condition B does not block a derivation with a pronoun in the case of an intrinsically reflexive verb. Recall that we assumed all along that a predicate is reflexive marked if its grid is reflexive, in which case condition B allows reflexivity.
300
Tanya Reinhart and Eric Reuland
So far we have only considered grids that were reflexivized by a (moved) SELF. However, the grid of the verbs under consideration here is intrinsically reflexive to begin with (see section 3.2). In (38), or the relevant case of (37), then, condition B is not violated. The predicate is semantically reflexive, but it is, as required, reflexive marked. A different question is why a SELF-anaphor is not allowed in the intrinsically reflexive cases, since this does not follow directly from anything we said. We believe that this follows from principles of economy: the same property should not be marked twice. Vacuous application of a morphological process is generally prohibited (for instance, one does not find a combination of two passivization processes on one lexical item). 13 The occurrence of pronouns in the contexts examined here provides further support to our view that condition B does not regulate the occurrence of pronouns directly, but, rather, is a condition on reflexive interpretation. The universal generalization is, as we argued, that reflexivization must be morphologically licensed.
3.6.2 The intrinsic reflexives above are marked as such only in the lexicon, with no overt morphological marking. It is important to note, however, that languages may have morphological means to mark intrinsically reflexive verbs. In Russian, for instance, the anaphor s'eb'a is an extrinsic reflexive marker, whereas inherently reflexive verbs are characterized by the morpheme s'/s'a. In Swedish, a elass of intrinsically reflexive verbs is characterized by the morpheme s. Although we will not discuss them here, we believe that elitics (Italian si, French se and Czech se) should also be viewed as morphological markers of intrinsic reflexivity, rather than as anaphors of either the SE- or the SELF-type. Clitics are best analysed as XO projections, as in Kayne (1989a). As such, they cannot be viewed as syntactic arguments saturating a grid position. Their relations to V may be taken as forming one morphological unit, even if they are base-generated detached (analogously, perhaps, to the relation of V and I which form one morphological complex, see Borer (1984) for discussion of the morpho syntactic status of elitics). Since they lack the internal structure of SELF-anaphors they cannot also function as a relational restriction on the grid. Rather, they should be viewed as saturating a grid position morphologically, analogously to the lexical saturation assumed in the intrinsic cases above. If this is their function, their strict locality follows too. (For an analysis along such lines, in which reflexive elitics are analysed as operators on the verb, see Zubizarreta (1987). For a discussion of the differences between reflexive elitics and Germanic SE-anaphors, see Hellan (1988).) A generalization which emerges is that languages differ as to whether they allow
Anaphors and logophors: an argument structure perspective
301
the object (patient) role of their intrinsically reflexive predicates to be saturated syntactically, as in Frisian, above (and in Dutch), or whether they require that they are saturated only lexically. English is an instance of the latter case. Verbs like Dutch zich schamen 'to shame oneself' have been replaced by different forms, like to be ashamed, lacking an object position altogether, or in the case of transitive verbs, such as wash, the patient role is satisfied lexically (by the predicate). An intermediate case is that of ditic languages, which do not allow syntactic saturation in these contexts, but still require a morphological marking of lexical satisfaction. Our analysis entails that a language can allow bound pronouns (or SE-anaphors) in reflexive environments only if it has intrinsically reflexive predicates which also allow syntactic assignment of both positions, as in Frisian above. Neither in English, nor in clitic languages, then, can we find bound pronouns saturating an intrinsically reflexive predicate.
4 SE-anaphors 4.1 Association with Agr As we saw in the introduction and in section 2, SE-anaphors give rise to the following questions:
(i) Why are SE-anaphors subject oriented? (ii) Why is finite lnfl the opacity factor for this type of anaphor? We argued in section 2 that the defining property of SE-anaphors is that they lack
302
Tanya Reinhart and Eric Reuland
Agr, which is anaphoric in the former case. Anaphoric Agr bears 'P-features just like finite Agr, hence, it is equally suitable as a 'host' supplying these features to the SE-anaphor. 14 The result is summarized, for convenience, in (40). Though, as we just saw, (40) need not be stipulated as a condition, since it follows as the only option for providing SE with its lacking features. (40)
SE-heads adjoin to Agr at LF.
Since Agr is always coindexed with the subject and SE-anaphors always associate with Agr, it follows that SE-anaphors, in their grammatical (non-Iogophoric) use, are subject oriented. Let us see how the movement in (40) proceeds: to get to I, SE must first adjoin to V (which is just the standard procedure of head-to-head movement, see Pica (1987». In this respect it is just like SELF. However, while SELF, due to its internal structure, is interpreted as a restriction on grids, SE lacks this function, so it has no effect on the verb's grid. It just leaves the relevant a-position unsaturated, until it receives 'P-features. In the Barriers framework, from there on, its fate is tied to that ofthe V (the head-movement constraint) and it moves along with the V to I, until they arrive at a position where the necessary 'P-features are available from Agr. The next issue is why it isn't simply the first Agr, finite or not, which is the opacity factor. That is, why is the SE-anaphor able to skip the first available Agr and associate with higher ones? Since, being associated with its governing verb, the anaphor cannot move up independently without violating the head-movement constraint, the relevant relation can only be established by movement of the SE-marked verb-Inft complex as a whole. Thus the presence of non-local anaphora in some languages requires the availability of predicate-raising structures at the relevant level. Chomsky has argued independently that processes applying overtly in particular languages are generally available at LF. So, LF admits restructuring of infinitival clauses, just as there is overt verb raising in languages like Dutch. The SE-marked verb will thus first be raised to its local Agr, and then be eligible for further raising to the next V and Agr higher up, etc., under conditions as discussed in Koster (1987). By way of illustration, consider the structure in (4la), which is a case of object control. Agr is represented by the index on I; the SE-anaphor is 'bound' by the subject of the matrix clause, as, e.g., in the Norwegian sentence (4lb). That is, SE must derive its 'P-features from the I of the matrix clause (it is irrelevant whether the latter is finite or not).
Anaphors and logophors: all argument structure perspective
303
a.
(41)
b.
J onj bad oss snakke om segj Jonj asked us to talk about SEj
(42) is the structure after adjunction to the lower V has applied. (For convenience, we use adjunction to the left, throughout, to keep the correlation with S-structure order transparent.) (42)
IP N0I1
J~
Ij~ Vm
NPi
IP
~ PROi
~
Ii
[-Too,,]
VP
~ SE/Vs ...
tj
We now assume that first SEjVs moves to Ii> amalgamating in that position, then SE;lV,IIi moves up to Vm, and finally, Vm/SEjlV,IIi moves up to I j • The result is given in (43).
304
Tanya Reinhart and Eric Reuland
(43)
Here the SE-marked verb is amalgamated with Ii' and SE may pick up the latter's
Joni bad oss forsoke a fa deg til a snakke pent om segi Joni asked us to try to get you to talk nicely about SEi
As we know, the first finite Infl is the opacity factor for SE-anaphors (i.e. they observe the TSC). This is naturally accounted for: it reduces to the fact that finite clauses are opaque for 'verb raising'. 15 This restriction has to be assumed independently. 4.2 Apparent binding domains We argue that the full (grammatical, i.e. non-logophoric) range of SE-anaphors is captured by the mechanism of LF-movement described above. This appears to contrast with a previous consensus in the study of SE-anaphors, namely that there are at least three possible (non-local) binding domains across languages, and, perhaps, the choice of a domain is parametrized (as in Manzini & Wexler (1987)).
Anaphors and logophors: an argument structure perspective
305
The three domains, as best summarized in Everaert (1986a), vary as to whether an SE-anaphor can be bound out of: (a) (b) (c)
small clauses (causatives and complements of perception verbs), all infinitive clauses, subjunctive finite clauses.
The first, most restrictive, domain is found with Dutch zich; the second, which corresponds to the domain we specified above, is found in many European languages, such as Norwegian, Polish and Finnish; and the third has been most notably observed for Icelandic seg/sin and (according to Giorgi (1984)) Italian proprio. So, the question which arises is whether our reduction of all these domains to just the second, which gets a straightforward account in terms of LF-movement, can be maintained. In the case of the third, subjunctive, domain, it appears to be clear by now, that there is no need to assume that this is a grammatical distinction. Hellan (chapter 2) and Thr:iinsson (chapter 3) show convincingly that the Icelandic seg/sin can be equally bound out of indicative clauses, under the appropriate discourse conditions. What is required, in both the subjunctive and the indicative long-distance binding, is that the antecedent must be a perspective holder (which they describe as 'perspective command'). These are the typical properties oflogophoric uses, as we will see in detail in section 6. The subjunctive just appeared to be a grammatical factor since, in languages with the subjunctive mood, it is often used in contexts reporting perspective or point of view. So, more occurrences of a logophoric SE will be found in such contexts than in others. The same can be argued about Italian proprio, as in Sells (1987), and Hellan (chapter 2). A more crucial question, however, is whether the smaller domain of small clauses needs to be stipulated as a language-specific or a parametrized option. Since this has been argued to be necessary for Dutch (see e.g. Koster (1985)), let us look at the Dutch facts in some detail. The Dutch zich in te-infinitives cannot be bound from the outside, as in (45), whereas zich in causative or perception-verb complements can be, as in (46). (45)
(46)
Ik hoor dat Jani Piet gevraagd heeft een boek voor zich*i meete brengen to bring I hear that Jan Piet asked has a book for SE 'I hear that Jani has asked Piet to bring a book for him SE*i' J ani hoorde Marie een lied voor zich/hemi fiuiten J an heard Marie a song for SE/him whistle 'Jan heard Marie whistle a song for him'
306
Tanya Reinhart and Eric Reuland
However, it can be shown that the contrast between zich in te-infinitives and in bare infinitives in Dutch follows from an independent factor: te-infinitives must be extra posed, bare infinitives cannot be. The contrast between te-infinitives in Dutch and their equivalent in Scandinavian follows from a difference in the position of attachment: It is an independent fact that the branching directions of the V-and I -projections in Dutch differ from Scandinavian. In Dutch both are rightward branching. This means that the extraposed te-clause is to the right of the finite verb, therefore the clause must be attached higher than the finite verb, and the anaphor will not be able to reach a higher Infi. By way of illustration, consider again the sentence in (45). Its base structure is indicated in (47) (omitting the topmost clause ik hoor 'I hear'). CP
(47)
~
C
IP
I~ I'
dat NP
I
I
Jani
I
heeft
C
IP
\
PR~II
(om)
I
Vr-------I
~ t~ ~
NP
V'
eentek
PP
~
P \
voor
V
I
zichi mee brengen
Anaphors and logophors: an argument structure perspective
307
At S-structure, there is extraposition of the complement clause to the right of the auxiliary heeft, which is in the I-position. It follows that the extraposed complement must adjoin to IP (and not to VP). This is indicated in (48). (48)
/\IP
I~
dat
CP
IP
/\ A 1\1\ PRO,/\ A I'
NP
C
J,n,
IP
I~
(om)
NP
v'
II
Pietj th
heeft
V
gevraagd
VP
1\
NP
V'
m""
bren~n.
ttnL ~ )1P voor
Schematically, this result is as indicated in (49).
I
V
308
Tanya Reinhart and Eric Reuland
CP
(49)
~IP
dat
IP/
~I'
Jan
/'1
VP
heeft
/\
... tep V
CPinfinitive
Q.
Therefore, zich and the verb it SE-marks are not c-commanded by the higher VI complex. Hence, they cannot move and be bound by a higher Agr. If te-infinitives obligatorily undergo extraposition in Dutch, then, no appeal to a specific parametrization is needed to capture the narrower distribution of SEanaphors, under our analysis. For discussion of extraposition, see for instance Reuland (1982b). That there is always extraposition in these cases is argued in Den Besten et al. (1988).16
5 Language-specific and open problems While our analysis captures most of the facts observed for SE-anaphors, at least in European languages, there are still some problems in capturing the full range of complementarity facts, which we outline now. 5.1 Pronouns and SE-anaphors Our analysis entails that pronouns are in complementary distribution with SELFanaphors. It does not imply complementarity between pronouns and SE-anaphors. In many environments, there is indeed no such complementarity. But an issue arises in cases like the following. (a) Dutch, like Frisian, allows syntactic saturation of intrinsically reflexive predicates. Hence, it is exempt from condition B in these contexts. However, Dutch differs from Frisian, in that only zich can occur in this position, but not the pronoun hem, as in (50). (50)
hiji schaamt zich/?*hem/??zichzelfi he shames SEI?*himl??himself
Anaphors and logophors: an argument structure perspective
309
The fact that zich is allowed in (SO) is not problematic, since if pronouns are allowed here in Frisian, the same consideration should allow zich, too, since with respect to both conditions A and B they pattern alike. The question is why the pronoun is excluded. Another case where an SE-anaphor is allowed but a pronominal is not is the subject of causative and perception-verb complements, illustrated in (51) (cf. Everaert (1986a)). (51)
Jani hoorde zich/*hemi over zichzelfi praten Jan heard SE/him about himself talk
In other contexts allowing zich, a pronoun can occur as well, as illustrated with binding into adjunct PPs in (52). (52)
a. b.
Klaasi duwde de kar voor zich/hemi uit (adverb) Klass pushed the car before him out Jani had Cindy met hem/zichi meegenomen Jan had Cindy with him taken
The same pattern is found in Norwegian (see the extensive discussion in Hellan (1988)). (b) Icelandic is reported to display complementarity of SE-anaphors and pronominals in all contexts in which SE can be grammatically bound (i.e. nonlogophoric contexts: an anaphor bound by a subject across infinitival clause boundaries at the most, see Anderson (1986), and also Manzini & Wexler (1987)). These differences do not follow from the principles proposed here. For some, this may well be as it should be. In Dutch, for instance, the contrast between hem and zich is weaker than one would expect of a binding-theoretic fact. There is even a contrast between singular and plural; given the appropriate context, the bound third person plural pronoun hun seems often more acceptable than hem (although zich may still be preferred). There is dialectal variation (many dialects have/allow hem instead of zich). Finally, note, that in all of these languages, an SE-anaphor is distinct from a pronominal only in third person. It may well be the case that these problems do not completely fall under syntactic principles at allY A full discussion will require a thorough analysis of the differences in representation and interpretation between intrinsic and extrinsic reflexives. For the beginning of such a discussion from the present perspective, see Reuland (1989). 5.2 SE- and SELF-anaphors Our analysis predicts complementarity between SELF- and SE-anaphors, since the first are subject to condition A, and the second, as non-SELF-markers, to
310
Tanya Reinhart and Eric Reuland
condition B. This, indeed, is supported by a variety of languages where SE cannot be bound in a grid that is not intrinsically reflexive. Still, possible exceptions have been reported. In Polish, an anaphor of the SE-form is locally bound. This, however, can be explained within our analysis. As shown by Reinders-Machowska (chapter 6), siebie behaves both as a medium-distance anaphor and as a local anaphor. In addition, in local contexts it allows a reciprocal interpretation. Given independent properties of Polish, the availability of siebie in both contexts can be explained as a case of ambiguity. In order to account for its behaviour as a local anaphor one may assume a relational structure, as indicated in (53). (53)
[NP ... [N' siebie (x,y)]]
No element comparable to the standard him-determiner found in SELF-anaphors is visible. But notice that Polish NPs may lack an overt determiner. 18 Building on ideas in Saddy (1987), one may assume that the rich nominal morphology in Polish by itself is sufficient to license a determiner and associated variable. As a consequence, one of the options associated with siebie is that an empty determiner is, nevertheless, projected in (53), which, then, will behave precisely like himself In its use as an MD-anaphor, siebie should be a defective element. This can be obtained under the assumption that not only the projection of the (null) determiner, but also that of the first grid position (Xi) is optional. If neither is projected an SE-anaphor results, as in (54). (54)
[... [N' siebie(y)]]
Siebie must now be licensed by a c-commanding Agr, along the lines we sketched above. 19 While Polish presents a clear picture, and, moreover, does not seem to be a real problem, other cases are less clear. SE-anaphors in Icelandic have been extensively discussed, but its SELF-anaphors have received considerably less attention. From the literature it can be concluded that they are local (Thniinsson, chapter 2; Maling (1986); Henan (1988». According to Henan (1988) the choice of the SELFanaphor is governed by the same conditions as in Norwegian. According to Maling (1986), these conditions are less clearly syntactic than in Norwegian, however. This suggests that there is more latitude for local SE-anaphors. It is important, then, to investigate the precise conditions under which they are possible. So far, we have not been able to do this. Swedish, according to Henan, admits an SE-anaphor in certain cases where Norwegian has a SELF-anaphor, and we would indeed expect SELF to be required. Again, the precise extent of this phenomenon cannot be determined on
Anaphors and logophors: an argument structure perspective
311
the basis of the literature consulted, and requires further investigation. An ambiguity analysis will not be supported by any independent property of these languages, so we leave these issues for future research.
6 Logophoricity The reduction of the various anaphoric domains to just two domains rests, in part, on the claim that, in the absence of an analysis of logophoricity, the binding theory has attempted to capture also anaphoric relations that fall outside of sentence-level grammar. For English SELF-anaphors this is witnessed, for example, in the attempt to capture the full range of reflexives in picture NPs within the binding theory. For the SE-anaphors this led to the unclarity concerning the relevant domains (surveyed in chapters 1,2 and 3, and in Anderson (1986)). Let us, therefore, turn to the logophoric use of anaphors. Since this use turns out to be quite common with SELF-anaphors in English, we will use this case to illustrate first the concept of logophoricity and its interaction with the binding theory, before turning to the semantics oflogophoricity.
6.1 Logophoricity with SELF-anaphors 6.1.1 In the early seventies it was noted (by Ross (1970) and Cantrall (1974)) that first person reflexives (and to a lesser extent also second person reflexives) can occur without an antecedent in blatant violation of (what is now known as) condition A, as illustrated below ((55a) and (56) are from Ross, (57a) is quoted in Zribi-Hertz (1988) from Roth). (55) a. b. (56) a. b. (57) a. b. c.
This paper was written by (Ann and) myself Apart from myself only three members protested Physicists like yourself are a godsend A picture of myself would be nice on that wall She .. gave both Brenda and myself a dirty look The chairman invited my wife and myself for a drink Max and myself are having a great time in Lima
It would be wrong to conclude from such examples that first person reflexives are simply deictic expressions, finding their reference in the discourse situation and exempt, therefore, from the binding conditions, since in other cases they behave as standard anaphors. Compare, specifically, the minimal pairs in (57) and (58).
312
Tanya Reinhart and Eric Reuland
(58) a. *She gave myself a dirty look b. *The chairman invited myself for a drink Furthermore, precisely in the same environments allowing a first person reflexive to be free (or discourse bound), a third person reflexive can be long-distance bound, in apparent violation of condition A. This has been observed already by Ross and Cantrall, but see also Kuno (1987) and the references cited there. ZribiHertz (1988) (hereafter ZH) provides a list of about 130 examples of actual uses in (mainly literary) texts, four of which are included below. (59) a. John said that the paper was written by (Ann and) himself (Ross (1970» b. Veteran soldiers ... walking quietly to ballot boxes and voting for more wars to be fought by themselves (ZH, p. 57, quoted from Catton) c. Clara found time ... to check ... (that) apart from herself ... there was a man from the BBC ... (ZH, p. 62, quoted from Drabble) (60) a. It angered him that she should have the egotism to try to attract a man like himself (ZH, p. 56, quoted from Lewis) b. Lucie said that (you agreed that) a picture of herself would be nice on that wall (61) a. Lucie boasted that the chairman invited her husband and herself for a drink b. She felt that he was criticizing the room and herself (ZH, p. 59, quoted from Woolf) c. She wrote that Max and herself are having a great time in Lima Correspondingly, where a first person pronoun cannot be free, no long-distance binding is possible either, as in (62) (compared with (61) and (58». (62)
a. *Lucie boasted that the chairman invited herself for a drink b. *She felt that he criticized herself
6.1.2 We take both instances of an appropriately free SELF-anaphor (in the first and third person) to be logophoric anaphora, which is not subject to the binding conditions. The crucial issue, however, is that logophoric anaphora is not arbitrarily free. The (widely observed but unexplained) inconsistency we illustrated above in the degree of respect SELF-anaphors show to condition A of the standard binding theory is fully accounted for within the alternative we proposed in section
3. Note, first, that there is a striking similarity between the positions allowing a SELF-anaphor to be logophoric, and the environments where the complementar-
Anaphors and logophors: an argument structure perspective
313
ity requirement fails which we discussed in section 3.1. In the bad cases of (58) and (62) (as * ... the chairman invited myself/herself for a drink) the reflexive occupies a position on the a-grid of the verb. Logophoric anaphora is possible only where this is not the case: In the adjunct cases (55) and (59) (by . .. self, apart from . .. self), the position is not on the grid to begin with. In the other cases the reflexive is embedded in an NP in a grid position (picture of . .. self, people like . .. self, Max and . .. self), which means that it is not itself on that grid. Needless to say, in all these contexts the anaphor is never in complementary distribution with a pronoun. This follows directly from the way we formulated condition A, repeated in (63). (63) Condition A A SELF-anaphor on an argument (grid) position of a fully assigned predicate P reflexivizes the grid of P. (58) and (62) are ruled out by A, since the SELF-marked position is on a a-grid. Hence the predicate is marked reflexive, but no two arguments can be identified. However, A says nothing about all other cases mentioned above, where the SELFmarked position is not on the grid of a fully assigned predicate. This means then that the logophoric SELF-anaphors should be possible only in the cases where the binding conditions do not rule out an unbound occurrence. This division of SELFanaphors, into those which fall under binding condition A and those which do not, follows from the replacement of the GC with the a-grid as the relevant binding domain, which, we argued was needed independently. We may note that logophoric anaphora is sensitive also to the same 'Subject' constraints we discussed in section 3.4.1. Note, for example, the difference between (64a) and (64b). (64) a. Lucie made the doctor read a detailed analysis of herself/myself b. *Lucie made the doctor read Ben's analysis of herself/myself In the first, the N-predicate is not fully assigned, but rather, its agent role is lexically satisfied. Hence condition A does not apply to reflexivize this predicate. In the second, where the agent role is syntactically controlled, the predicate is a relevant domain for condition A, so the reflexive predicate which is formed is uninterpretable. It seems dear, therefore, that there is no reason to assume any difference between the logophoric instances above and cases (9) and (10) of section 3.1, which are currently considered part of core grammar, thus creating the complementarity problem. We argue that both fall outside the scope of condition A, as presently formulated. This contrasts with two recent works which take the alternative line of extending the binding theory so it applies to some of the cases of long-distance
314
Tanya Reinhart and Eric Reuland
SELF-anaphors discussed here. Safir (1989) offers an account for the 'comparison predicates' (56a), (60a) and several other structures. Keenan (1987) proposes that in case the reflexive is embedded in an NP, the containing NP is viewed as the anaphor (although he provides only the semantics for the conjoined-type NPs, such as (61)). Even in the case of reflexives embedded in NPs, each of these analyses can cover only a subset of the cases. 20 The crucial problem for both, in any case, are the reflexives in an adjunct position, as in (55), (59) and (12). The availability of LD-anaphora, as well as the lack of complementarity with pronouns in these cases, indicates that the problem under consideration cannot be reduced either to the semantics of complex NPs, or to issues of comparison structures. 6.1.3 Let us now look at the way condition B interacts with logophoric contexts. An interesting contrast in the restrictions on pronominal binding is illustrated by the minimally different (65) and (66). (65) a. b. (66) a. b.
Maxi befriends linguists like himi Maxi likes jokes about himi *Maxi criticized Lucie and himi *Maxi criticized two members in addition to himi (in our last meeting)
The NPs containing the pronoun here are prototypical environments allowing long-distance logophoric anaphora, many examples of which were examined above. A SELF-anaphor replacing the pronoun in both cases cannot be subject to condition A. Still, while (65) conforms with our expectations, showing no complementarity with pronouns, in (66), surprisingly, a pronoun cannot be bound. (The problem posed by (66) was pointed out to us by Hans Kamp.i l This problem arises, typically, only with conjoined-type NPs (including complex-determiner cases such as everyone but him). To see why it arises, let us look at what a sentence like (66a) will mean if we interpret the pronoun as a bound variable. (66') Max(Ax(x criticized Lucie & x criticized x)) We do not intend (66') as the formal interpretation (clearly not the LF) of (66a). However, it is clear that whatever interpretation this sentence gets, it will be equivalent to (66'). The formal compositional analysis of the conjoined NP, based on the properties of SELF, is provided in Keenan (1987) (also for the complexdeterminer cases). So, semantically, binding in (66a) yields a reflexive predicate (... x criticized x). Since reflexivity is not licensed by a SELF-marked argument,
Anaphors and logophors: an argument structure perspective
315
the interpretation is blocked by condition B. The same reasoning applies to (66b).22 Obviously the same type of NPs will not show complementarity of a logophoric SELF-anaphor and a pronoun, when the antecedent and the anaphor are not on the same grid, as in (67a). (67) a. Max; said that the boss criticized Lucie and himself/him; b. Max(Ax (x said that the boss criticized Lucie & the boss criticized x)) The interpretation of the binding in (67a) is equivalent to (67b), where no reflexive predicate is formed. Hence, neither condition applies to this sentence. 23 Examples like (66) suggest that condition B is viewed, as we presented it, as a general filter on interpretations, i.e. on the mapping from syntactic to semantic representations. 6.2 Logophoric SE-anaphors Our analysis entails a rather different distribution of the logophoric use of the SELF-anaphors, and of the SE-type anaphors like sig. SE-anaphors, the betterknown representatives of logophoric anaphora, have much freer use in the languages allowing them to be logophoric. In contrast to (62), repeated in (68), there is no problem with such anaphors in a grid position in (69). (As illustrated, e.g., in Hellan (chapter 2) and Thrainsson (chapter 3).) (68) a. *Lucie boasted that the chairman invited herself for a drink b. *She felt that he criticized herself (69) Jon sagoi ao eg hefOi svikio sig (Icelandic, Thrainsson) John said that I had betrayed himself This is so, since, unlike SELF-anaphors, SE-anaphors are not subject to condition A. The only requirement placed on them by the grammar is that they are assigned (f)-features for interpretation. The grammar only determines (independently) the conditions under which they can be associated with I, but no grammatical condition (analogous to A) prevents them from being free in any specific domain. The only requirement is that as anaphoric (defective) expressions they must find an antecedent, which they can do logophorically. In other words, SE-anaphors are subject only to condition B, but, in languages which allow their logophoric use, there are no further syntactic restrictions on their occurrence as such. 24 Another difference predicted between the (1-) bound and the logophoric occurrence of SE-anaphors is that only the first shows obligatory subject orientation, since the restriction only follows from the attachment to I. This prediction is borne out: languages allowing the logophoric use of SE-anaphors, such as Icelandic, also
316
Tanya Reinhart and Eric Reuland
allow them with a non-subject antecedent (or an antecedent that does not even c-command). Obviously, languages like Dutch or Polish which do not allow logophoric use of SE-anaphors will show a strict subject orientation. 6.3 Centre orientation and the semantics of logophoricity From the perspective of the grammar, all that needs to be said about the logophoric use of anaphors is that it is possible as long as no binding rule is violated and an antecedent can be found. Nevertheless, whether a language allows its anaphors to be used logophorically is still subject to variation. For example, the logophoric use of SELF-anaphors in Hebrew is much more restricted than in English (if possible at all). Dutch allows only its SELF -anaphors to be used logophorically, but not its SE-anaphor zich, while in Scandinavian it is the SE-anaphors which are used this way. We are not able, at the present, to explain these variations. While no further syntactic constraints are needed to account for logophoricity, there appear to exist discourse conditions affecting the choice of a logophoric anaphor over a pronoun, and most analyses of logophoricity are couched in terms of point-of-view, or perspective. Unlike approaches such as Sells (1987), we do not believe that all occurrences of the so-called logophoric uses are reducible to such considerations, and attempting such a reduction renders the concept of point-ofview somewhat vacuous. Nevertheless, it is clear that a major function of logophoric anaphors, especially in written texts, is related to this concept. Furthermore, a well-known puzzle is that even third person anaphors can be used logophorically with no antecedent in the sentence, and such occurrences are only explainable in terms of point-of-view. Let us therefore examine this function oflogophoricity. Following Fillmore's (1971) notation, we will assume that an utterance (or a sentence in context) is associated with a centre, which is a triple consisting of the participants (speaker and addressee), the time and the place of the utterance. The centre is what provides reference to deictic expressions. That first and third person reflexives should be able to pattern similarly in centre-oriented environments would come as no surprise to literary scholars, who have been assuming for quite a while that a text may assume the consciousness of one of the characters as the centre relative to which deixis is interpreted (see Reinhart (1979) and Sells (1987) for some of the literary references). Kuno (1972) and Cantrall (1974) were the first to argue that point-of-view manipulations are not restricted to literary texts. Following this tradition, we will assume that logophoricity is a relation between expressions and centres (whether actual, or reported). As such it is not a strictly sentence-level phenomenon. As observed in the studies cited above and by Hellan (chapter 2) and Thniinsson (chapter 3), a centre-oriented use of third person
Anaphors and logophors: an argument structure perspective
317
anaphors does not require an antecedent in the sentence. As with first person anaphors they can find their reference in some centre mentioned or assumed in a previous context. It has been attempted to incorporate this logophoric type into the system of variable binding in the grammar (Sells (1987), Chierchia (forthcoming)). Such attempts are problematic in view of the fact that the distribution of logophoric anaphors does not obey any of the standard requirements on variable binding. For example, even when a sentence antecedent is available it does not have to c-command the pronoun: (70) a. That we hang a picture of himself on every wall is one of the president's most outrageous demands b. Skooun J6nsi er [ao pli hafir svikio sigi] ... (Thrainsson) opinion John's is that you have betrayed self Rather than extending the binding mechanism of the grammar, we believe that logophoricity, like other centre-oriented phenomena, requires a separate treatment. Reinhart (1983), extending a proposal by Kaplan (1971), suggests that occurrences of a definite description or a pronoun are associated with the threeplace relation of representation: R (a, b, c) holds when (some form of) an expression a represents an entity b for a person c. c is what we called here the 'centre'. This relation was used to account for several phenomena of centre dependency. (The modification 'some form of' an expression a is intended to abstract away from issues of grammatical person in the case of pronouns.) The logophoric (unbound) use of anaphors should be viewed within this approach as identifying two elements of the triple in the representation relation the represented entity and the centre (person). Why are anaphors selected for this specific use? Given our analysis of anaphors, both types are expressions that ~annot refer independently. SELF-anaphors express a relation, or contain one unsaturated position. To be interpreted, the second argument of the relation must be found. When it cannot be found grammatically in the a-grid the expression may still be rescued (from uninterpretability) if it is associated with an available centre. Similarly, an SE-anaphor needs to get
Acknowledgements We would like to thank Martin Everaert and Eddie Ruis for extensive comments and
318
Tanya Reinhart and Eric Reuland
discussion of the ideas in this chapter. We are also indebted to Eric Hoekstra and Wim Kosmeijer for their advice.
Notes 1. For a survey of the facts in a range of languages, see Yang (1984). For an interesting recent discussion of such facts, and a survey including Japanese, Chinese and Korean, see Katada (1989). The facts she discusses seem entirely consistent with the present approach. 2. But note that elitics like Italian si, French se and Czech se are always local. See Napoli (1979), Lasnik (1987), Zubizarreta (1987) and Toman (chapter 7) for discussion. 3. Everaert (chapter 4) argues that (summarized briefly) SE-anaphors must be [+pronominal] with respect to condition B, [-anaphoric] with respect to condition A, and [+anaphoric] with respect to the condition on MD-binding. Although his analysis captures most of the facts that ours does, and certainly much more than the standard formulation of the binding theory, in terms of explanation one may still wonder why this is so. Why, e.g., is the GC of SELF-anaphors defined by the subject, and the GC of SE-anaphors by finite Inff? Why is it that SE-anaphors are subject oriented, and (pronominal) SELFanaphors are not? 4. The distribution of reciprocals is not identical to that of the local reflexive anaphors (see, e.g., Lebeaux (1983»; nor does it allow logophoric uses. These differences will follow if the analysis of Heim, Lasnik & May (1988) is assumed. 5. As noted by Huang & Tang (chapter 13), discussing a proposal by Battistella (1987), such an analysis cannot capture the full distribution of the anaphor ziji in Chinese. From the perspective to be developed here, we will have to say that the Chinese pattern follows only in part from the structural binding theory. For the other part, it should fall under the theory of logophoricity. We leave this issue for further investigation. 6. Everaert's analysis (e.g. chapter 4) captures this same fact successfully by marking both pronominals and SE-anaphors [+pronominal]. 7. In Chomsky (1981) the binding theory is quite independent of other principles of the grammar. The components themselves of the binding theory are independent as well, and the definition of governing category is essentially arbitrary from the perspective of the structure ofUG. 8. Among the many other studies leading to the conclusion that argument structure should be taken into account by the binding theory we may note E. Kiss (1987b) and chapter 12, who shows that in Hungarian the binding conditions crucially refer to a hierarchy of thematic roles, rather than just c-command. She makes a similar claim for argument binding in English PPs. Giorgi (chapter 9) shows how binding in Italian is affected by thematic considerations. The analysis of the Dutch system of anaphora in Everaert (chapter 4) crucially involves the way thematic roles are assigned. Notions like governing subject, or the designated class of prepositions P in Koster (1985), also contain implicit reference to thematic structure. 9. Although not explicitly given there, Higginbotham's footnote 14 suggests that he had in mind a similar structure for intrinsic reflexives. It is perhaps useful to point out that the notation V(x,x) is equivalent to V(1,2) with (1) and (2) linked.
Anaphors and logophors: an argument structure perspective
319
10. An issue we leave open here is whether the standard hierarchy analysis is sufficient to capture the full range of anaphor binding within the same grid. One of the most cited problems with this (noted already in Postal (1971» is the contrast between (ia) and (ib), which led several scholars to suggest that thematic hierarchy, rather than structural c-command alone, is involved (see, e.g., E. Kiss (1987b, c) and chapter 12). (i)
a. We spoke with Lucie about herself b. *We spoke about Lucie with herself
Once condition A is stated in terms of thematic grids, it should be possible to incorporate also thematic hierarchy (though it is not clear that this can be done within a movement analysis of condition A). However, we leave it for future study to determine whether this is indeed obligatory (or feasible). 11. In addition to cases where the anaphor is directly governed by the verb, SELF-marking also obtains in the case of prepositional objects, as in (i-iii). (i) Johni talks about himself/*himi (ii) J ani spreekt over zichzelf;l*zich/*hemi (iii) Jani snakke om segselv/*seg/*hami
12.
13.
14.
15.
16.
The prepositional object in such cases is thematically part of the grid of the predicate; hence conditions A and B apply just as in the cases discussed in the text. Given the analysis in Chomsky (1986b), I and V are coindexed, irrespective of movement in the syntax. In the specific case of ECMs, Reuland (1982b, 1983) argues in detail that they involve coindexing of the matrix verb and the inflection marker (to). Together, these yield the indexing chain of the matrix verb and the lower VP. In fact, the SELF-forms are not completely excluded here, but can be used to express emotional involvement (besides emphasis). This is consistent with the view that they are ruled out by economy, rather than by a grammatical principle. In order to allow subjects of small clauses as antecedents of SE-anaphors we have to assume that small clauses carry Agr too. Much of the current literature makes this assumption, although direct empirical arguments like those for anaphoric Agr in infinitival clauses are still lacking. Below we come back to this issue. The movement assumed here wiIllead to PRO being governed. There may thus be reasons to consider an analysis of V-movement along the lines set out in Baker (1988). Discussing verb incorporation in control structures in Chichewa, Baker proposes an analysis in which the VP is moved into [Spec, CPl. This movement does not affect the ungoverned status of the PRO-subject. In its derived position the VP is L-marked by the verb of the matrix clause. This enables the lower V to move up to the matrix V. For the present, we will leave this issue open. Eddy Ruis has drawn our attention to the fact that this analysis is not obviously reconciled with the fact that not all control complements appear to require extraposition. They also occur in verb-raising structures. An example is given in (i). (i)
*Jani heeft Piet dat boek voor zichi gevraagd mee te brengen Jan has Peter that book for him asked to bring
Yet, there is no contrast with respect to the binding theory between (48) and (i); in neither is it possible for zich to be bound by the higher subject Jan. This would be
320
Tanya Reinhart and Eric Reuland unexpected if the structure is as in (ii), with gevraagd mee te brengen a verbal cluster (Evers (1975». (ii) Jan heeft Piet [PRO dat boek voor zich tl [v gevraagd mee te brengenl However, it has been argued extensively in Reuland (1982a) that there are significant differences between verb raising with bare infinitives and verb raising with these teinfinitives in Dutch. With te-infinitives there is no restructuring, as indicated by the nonapplication of the infinitive-for-participle process, unlike what happens in bare infinitive complements. The absence of such restructuring could already be enough to block amalgamation of the SE-marked verb with a higher Agr. Recently, Den Besten et ai. (1988) have argued that the structure of sentences like (i) is much closer to (45/48). They argue that the complement clause is in the position of the remnant mee te brengen, rather than in the position indicated by dat boek and voor zich. That is, (i) is derived by first extraposing the full te-complement (yielding a structure like (48», followed by intraposition of the relevant constituents (the direct object and the prepositional object in this case). Consequently, the structure is not (ii), but rather (iii). (iii) *Jani heeft Pietj [dat boekl h [voor zichil k gevraagd [s PROj th
17.
18. 19.
20.
tk
mee te brengen)
Now, the binding possibilities of constituents of the complement will be determined with respect to their traces, rather than with respect to their actual position. As a consequence, the binding facts of (i) reduce to those of (45/48). In Everaert's analysis, the difference in (50) is captured, in the grammar, by the feature system. However, it seems to us that choices ofthis form would follow from an analysis of accessibility of anaphoric forms in discourse, as developed in Ariel (forthcoming). She argues that (when a selection is not enforced by the grammar), the choice among anaphoric forms signals the degree of accessibility of the antecedent. For example, how far away should the antecedent be sought (though accessibility is not defined in terms of distance alone)? Anaphors (of both types) signal in her analysis a higher degree of accessibility than pronouns. (She shows that many issues of anaphor resolution in several languages find an elegant explanation in this way.) Since in the environments under consideration, the antecedent is most highly accessible, it would be misleading to choose a pronoun over an SE-anaphor in this case. This problem does not arise in Frisian, which lacks SE-anaphors, so a pronoun is the only choice. For instance, a form like dom can be used to mean either a house, or the house. It is not the place here to develop a theory of reciprocals. However, intuitively, what happens seems to be the following. Siebie, not being marked for number, allows a plural as a syntactic antecedent, while at the same time allowing its variables to pick out individuals from the antecedent set. It is clear that if the variable freely ranges over individuals, the standard reciprocal interpretation is included. One would predict, however, that unlike in English there would be no disjointness effect. That is, if the predicate is compatible with both a reciprocal and a reflexive interpretation (like wash), assigning the reciprocal reading would not exclude reflexive instantiations. We have not determined whether this is borne out. There is no obvious way to extend the semantics proposed by Keenan to picture NPs or to the comparison NPs in (61a), so their occurrence in long-distance contexts remains
Anaphors and logophors: an argument structure perspective
321
unexplained. Safir's analysis, by contrast, does not extend to the simple conjoined NPs like (61) which do not involve a comparison. 21. As we mentioned, NPs of this sort are discussed in Keenan (1987) and Safir (1989), who attempt extensions of the binding theory to capture also what we here called logophoric uses (although Safir (draft 4) allows a role to pragmatic considerations). The contrast between (65) and (66) is a problem for both. Although Keenan does not discuss contexts like (65), extending his analysis of (66) to these cases would also entail complementarity. Safir (draft 4) gives a very detailed analysis of reflexives in the domain of particular predicates, but it cannot generalize over the (a) and the (b) cases. Also a range of other constructions is beyond its scope. 22. A question which may, appropriately, be raised is why the analogue of (61) with an anaphor, as in (i), is not ruled out as well, by the same condition B. (i) Max criticized Lucie and himself SELF here cannot adjoin to criticize, or more generally it is not subject to condition A. So it appears that the predicate is not reflexive-marked and condition B should block the derivation. However, we have not actually entered in this chapter the details of the definition of 'reflexive-marking' we assume in condition B. The condition as we actually state it in Reinhart & Reuland (forthcoming) is given in (ii). (li)
B. Def.
A predicate is reflexive only if it is reflexive-marked. A predicate is reflexive-marked iff either (a) its grid is reflexive or (b) one of its arguments is SELF-marked.
So, for condition B, the fact that an argument is a SELF-anaphor is sufficient to license a reflexive interpretation, even if the grid is not reflexivized, as is the case in (i). (Further motivation for this definition is provided in Reinhart & Reuland (forthcoming).) 23. Note that condition B is formulated as an only if rather than iff condition. For this reason it says nothing about cases where a SELF-anaphor not in a grid position syntactically, gets interpreted, in the semantics, as an argument of the predicate. This situation pertains, e.g. in (67), where, due to the semantics of conjoined NPs, the SELF-anaphor ends up as an argument. 24. This does not exclude that certain languages may impose stricter requirements on the logophoric use of anaphors.
References
Abney, S. (1987), 'The English Noun Phrase in its Sentential Aspect', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. Anderson, S. (1983), 'Types of Dependencies in Anaphors: Icelandic (and Other) Reflex· ives', Journal of Linguistic Research 2, 1-23. Anderson, S., and P. Kiparsky, eds. (1973), A Festschrift for Morris Halle, Holt, Rinehart and Winston, New York. Anderson, S. R. (1986), 'The Typology of Anaphoric Dependencies: Icelandic (and Other) Reflexives', in L. Hellan & K. Koch Christensen, eds. (1986b). Anward, J. (1974), 'Swedish Reflexivization', in Osten Dahl, ed., Papers from the First Scandinavian Conference of Linguistics, Department of Linguistics, Goteborg. Aoun, J. (1985), A Grammar of Anaphora, MIT Press, Cambridge, Massachusetts. (1986), Generalized Binding, Foris, DOl'drecht. Aoun, J., and Y.·H. A. Li (to appear), 'Minimal Disjointness', Journal of Linguistics. Ariel, M. (forthcoming), Anaphoric Antecedents, Croom Helm, London. Baker, M. (1988), Incorporation, University of Chicago Press, Chicago. Barnes, M. (1984), 'Reflexives in the Scandinavian Languages: A Review of Some New Ideas', unpublished manuscript, University College, London. (1986), 'Reflexivisation in Faroese. A Preliminary Survey', Arkiv for nordisk filologi 101, 95-126. Barss, A. (1986), 'Chains and Anaphoric Dependence', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. Barss, A., and H. Lasnik (1986), 'A Note on Anaphora and Double Objects', Linguistic Inquiry 17, 347-54. Battistella, E. (1987), 'Chinese Reflexivization', paper presented at the 2nd Harbin Confer· ence on Generative Grammar, Heilongjiang University, Harbin, People's Republic of China; unpublished manuscript, University of Alabama at Birmingham. Belletti, A. (1983), 'On the Anaphoric Status of the Reciprocal Construction in Italian', The Linguistic Review 2, 101-38. Belletti, A., and L. Rizzi (1988), 'Psych·Verbs and Theta Theory', Natural Language and Linguistic Theory 6, 291-352. Benedicto, E. (in preparation), 'Sentence Structure in Latin', unpublished manuscript, Universita Autonoma de Barcelona, Bellaterra.
323
324
References
Beninclt, P., ed. (1989), Dialect Variation on the Theory of Grammar, Foris, Dordrecht. Bennis, H., and L. Haegeman (1984), 'On the Status of Agreement and Relative Clauses in West-Flemisch', in W. de Geest & Y. Putseys, eds. (1984). Bertocchi, A., and C. Casadio (1980), 'Conditions on Anaphora: An Analysis of Reflexives in Latin', in G. Calboli, ed., Papers on Grammar, Bologna. Berwick, R. (1985), The Acquisition of Syntactic Knowledge, MIT Press, Cambridge, Massachusetts. Besten, J. den, J. Rutten, T. Veenstra en J. Veld (1988), 'Verb Raising, Extrapositie en de Derde Constructie', unpublished manuscript, University of Amsterdam. Bok-Bennema, R. (1985), 'On Pro-drop Pronominal Anaphors in Eskimo', in J. Gueron, H. G. Obenauer & J.-Y. Pollock, eds. (1985), 1-18. Borer, H. (1984), Parametric Syntax, Foris, Dordrecht. (1986a), 'I-Subjects', Linguistic Inquiry 17.3, 375-417. ed. (1986b), Syntax and Semantics, Vol. 19: The Grammar of Pronominal Clitics, Academic Press, New York. (1989), 'Anaphoric AGR', in O. Jaeggli & K. Safir, eds. (1989). Bouchard, D. (1984), On the Content of Empty Categories, Foris, Dordrecht. (1985), 'PRO, Pronominal or Anaphor', Linguistic Inquiry 16, 471-7. Bremen, K. von (1984), 'Anaphors: Reference, Binding and Domains', Linguistic Analysis 14, 191-229. Brody, M. (1981), 'On Circular Readings', Linguistic Research 1, 95-105. (1985), 'On the Complementary Distribution of Empty Categories', Linguistic Inquiry 16, 505-46. Brody, M., and M. R. Manzini (1987), 'On Implicit Arguments', unpublished manuscript, University College London, and (in press) in R. Kempson, ed., Mental Representations: The Intelface between Language and Reality, Cambridge University Press. Burzio, L. (1981), 'Intransitive Verbs and Italian Auxiliaries', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. (1986), Italian Syntax: A Government-Binding Approach, Reidel, Dordrecht. (1987), 'On Deriving the PRO Theorem', unpublished manuscript, Harvard University, Cambridge, Massachusetts. Cantrall, W. (1974), View Point, Reflexives and the Nature of Noun Phrases, Mouton, The Hague. Cardinaletti, A., G. Cinque and G. Giusti, eds. (1988), Constituent Structure, Foris, Dordrecht. Chierchia, G. (forthcoming), 'Anaphora and Attitudes de se', in R. Bartsch, J. van Benthem & P. van Emde Boas, eds., Language in Context, Foris, Dordrecht. Chomsky, N. (1955/1975), The Logical Structure of Linguistic Theory, Plenum Press, New York. (1973), 'Conditions on Transformations', in S. Anderson & P. Kiparsky, eds. (1973). (1977), 'On WH-movement', in P. Culicover, T. Wasow & A. Akmajian, eds., Formal Syntax, Academic Press, New York. (1980), 'On Binding', Linguistic Inquiry 11, 1-46. (1981), Lectures on Government and Binding, Foris, Dordrecht. (1982), Some Concepts and Consequences of the Theory of Government and Binding, MIT Press, Cambridge, Massachusetts.
References
325
(1986a), Knowledge of Language: Its Nature, Origin and Use, Praeger, New York. (1986b), Barriers, MIT Press, Cambridge, Massachusetts. (1988a), 'Linguistics and Cognitive Science: Problems and Mysteries. Prospects for the Study of Language and Mind', unpublished manuscript, MIT, Cambridge, Massachusetts. (1988b), 'Some Notes on Economy of Derivation and Representation', unpublished manuscript, MIT, Cambridge, Massachusetts. Chung, S. (1989), 'On the Notion "Null Anaphor" in Chamorro', in O. Jaeggli & K. Safir, eds. (1989). Cinque, G. (1980), 'On Extraposition from NP in Italian', Journal of Italian Linguistics 5. (1982), 'On the Theory of Relative Clauses and Markedness', The Linguistic Review 1, 247-96. Clark, R., and E. Keenan (1987), 'A Semantic Characterization of Binary Quantification in Natural Language', in M. Dalrymple et al., eds., Proceedings of the West Coast Conference on Formal Linguistics 5, Stanford Linguistics Association, Dept. of Linguistics, Stanford University, California. Clements, N. (1975), 'The Logophoric Pronoun in Ewe: Its Role in Discourse', Journal of West African Languages 10, 141-77. Cole, P., G. Hermon and L. Sung (forthcoming), 'Principles and Parameters of Long Distance Reflexives', unpublished manuscript, University of Delaware. Coopmans, P. (1985), Language types: Colltinua or Parameters? , doctoral dissertation, U nivershy of Utrecht. Coopmans, P., and M. Everaert (1988), 'The Simplex Structure of Complex Idioms: The Morphological Status of laten', in M. Everaert, A. Evers, M. A. C. Huybregts & M. Trommelen, eds. (1988), 75-104. Corver, N. (1985), 'Children's Interpretation of Reflexive and Non-reflexive Pronouns in Dutch', unpublished manuscript, Tilburg University. Dand, F., and K. Hausenblas (1962), 'Pl'ivlastnovaci zajmena osobnf a zvratna ve spisovne cestine [Personal and Reflexive Possessive Pronouns in Standard Czech]" S lavica Pragensia 4, 191-202. Deutsch, W., C. Koster and J. Koster (1986), 'What Can We Learn from Children's Errors in Understanding Anaphora?' Linguistics 24, 203-25. Ernout, A., and F. Thomas (1972), Syntaxe Latine, Editions Klincksieck, Paris. Everaert, M. (1986a), The Sylltax of Rejlexivization, Foris, Dordrecht. (1986b), 'Long Reflexivization and Obviation in the Romance Languages', in P. Coopmans,1. Bordelois & B. Dotson Smith, eds., Formal Parameters of Generative Grammar, II - Going Romance, Foris, Dordrecht, 51-72. (to appear), 'Nominative Anaphors in Icelandic: Morphology or Syntax?', in W. Abraham, W. Kosmeijer & E. J. Reuland, eds., Issues in Germanic Syntax, Mouton-De Gruyter, Berlin. Everaert, M., A. Evers, M. A. C. Huybregts and M. Trommelen, eds. (1988), Morphology and Modulan'ty, Foris, Dordrecht. Evers, A. (1975), The Transformational Cycle in Dutch and German, distributed by Indiana Linguistics Club. Falk, Y. N. (1984), 'Grammatical Configurations and Grammatical Relations', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts.
326
References
Faltz, L. M. (1977), 'Refiexivization: A Study in Universal Syntax', doctoral dissertation, University of California at Berkeley, reproduced by University Microfilm International, Ann Arbor/London. Fanselow, G. (1988), 'Konkurrenzphiinomene in der Syntax: eine nicht pragmatische Reduktion der Prinzipien Band C der Bindungstheorie', unpublished manuscript, University of Pass au. Farmer, A. (1980), 'On the Interaction of Morphology and Syntax', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. Filengo, R., J. Huang, H. Lasnik and T. Reinhart (1988), 'The Syntax of Wh-in-situ', Proceedings of the West Coast Conference on Formal Linguistics 7, University of California, Irvine, California, 81-98. Fillmore, C. (1971), Santa Cruz Lectures on Deixis, distributed by Indiana Linguistics Club. Finer, D. (1984), 'The Formal Grammar of Switch-Reference', unpublished doctoral dissertation, University of Massachusetts, Amherst, Massachusetts. Fretheim, T., and L. Hellan, eds. (1982), Papers from the Sixth Scandinavian Conference of Linguistics, Tapir, Trondheim. Geest, W. de, and Y. Putseys, eds. (1984), Sentential Complementation, Foris, Dordrecht. Giorgi, A. (1984), 'Towards a Theory of Long Distance Anaphors: a GB approach', The Linguistic Review 3, 307-59. (1986), 'The Proper Notion of C-command and the Binding Theory: Evidence from NPs', in S. Berman, J.-W. Choe & J. McDonough, eds., Proceedings of NELS 16, GLSA, University of Massachusetts, Amherst, Massachusetts. (1987), 'The Notion of Complete Functional Complex: Some Evidence from Italian', Linguistic Inquiry 18, 511-18. Giorgi, A., and G. Longobardi (1987), 'NP Parametrization and the Position of Subjects', unpublished manuscript, Istituto di Psicologia del CNR, Universita di Venezia. Gleason, H. (1965), Linguistics and English Grammar, Holt, Rinehart and Winston, New York. Graffi, G. (1988), 'Structural Subject and Thematic Subject', LingvistictE Investigationes 12, 387-414. Grewendorf, G. (1984), 'Refiexivierungsregeln im Deutschen', Deutsche Sprache 12, 14-20. Grimshaw, J. (1986), 'Nouns, Arguments and Adjuncts', unpublished manuscript, Brandeis University, Waltham, Massachusetts. Grimshaw, J., and A. Mester (1988), 'Light Verbs and Theta-marking', Linguistic Inquiry 19, 205-32. Grimshaw, J., and S. Rosen (1988), 'The Developmental Status of the Binding Theory or "Knowledge and Obedience"', unpublished manuscript, Brandeis University, Waltham, Massachusetts. Gueron, J. (1985), 'Inalienable Possession, PRO-Inclusion and Lexical Chains', in J. Gueron, H.-G. Obenauer & J.-Y. Pollock, eds. (1985). Gueron, J., H. G. Obenauer and J.-Y. Pollock, eds. (1985), Grammatical Representation, Foris, Dordrecht. Hale, K. (1983), 'Warlpiri, and the Grammar of Non-Configurational Languages', Natural Language and Linguistic Theory 1, 5-47. Hankamer, J., and I. Sag (1976), 'Deep and Surface Anaphora', Linguistic Inquiry 7, 391-428.
References
327
Harbert, W. (1981), 'Should Binding Refer to SUBJECT?', in J. Pustejovsky & P. Sells, eds., Proceedings of NELS 12, GLSA, University of Massachusetts, Amherst, Massachusetts, 116-32. (1982), 'In Defense of Tense' ,Linguistic Analysis 9, 1-18. (1983), 'Germanic Reflexives and the Implementation of Binding Conditions', in 1. Rauch & G. F. Carr, eds., Language Change, Indiana University Press, Bloomington, 89127. (1986), 'Binding, SUBJECT and Accessibility', unpublished manuscript, Cornell University, Ithaca, New York. Heim, 1., H. Lasnik and R. May (1988), 'Reciprocity and Plurality', unpublished manuscript, UCLNUniversity of Connecticut/Irvine. Hellan, L. (1980), 'On Anaphora in Norwegian', in J. Kreiman & A. Ojeda, eds., Papers from the Parasession on Pronouns and Anaphora, CLS 1980, 166-82. (1982), 'Semantic and Functional Government of Reflexives in Norwegian', in T. Fretheim & L. Hellan, eds. (1982). (1983), 'Anaphora in Norwegian and Theory of Binding', Working Papers in Scandinavian Syntax 5, University of Trondheim. (1986a), 'On Anaphora and Predication in Norwegian', in L. Hellan & K. Koch Christensen, eds. (1986b). (1986b), 'Reference to Thematic Roles in Rules of Anaphora', in O. Dahl & A. Holmberg, eds., Scandinavian Syntax, Institute of Linguistics, University of Stockholm. (1988), Anaphora in Norwegian and the Theory of Grammar, Foris Publications, Dordrecht. Hellan, L., and K. Koch Christensen (1986a), 'Introduction', in L. Hellan & K. Koch Christensen, eds. (1986b), 1-29. eds. (1986b), Topics in Scandinavian Syntax, Reidel, Dordrecht. Hellberg, S. (1984), 'Satsens subjekt och textens', Nysvenska studier 64, 29-82. Herlofsky, B. (1986), 'Tuki Reflexives: Subject Orientation in Locally Bound Anaphors', Journal of the College of Arts and Sciences 19, Chiba University. Hestvik, A. (1987), 'Norwegian Reflexive Verbs', unpublished manuscript, Brandeis University, Waltham, Massachusetts. (1989a), 'SUBJECT-less Governing Categories', unpublished manuscript, Brandeis University, Waltham, Massachusetts. (1989b), 'Norwegian Reflexive Verbs,' unpublished manuscript, Brandeis University, Waltham, Massachusetts. Higginbotham, J. (1982a), 'Reciprocal Interpretation', Journal of Linguistic Research, 97-117. (1982b), 'The Logic of Perceptual Reports', Report 21 , Center for Cognitive Science, MIT, Cambridge, Massachusetts. (1983), 'Logical Form, Binding and Nominals', Linguistic Inquiry 14., 395-420. (1985), 'On Semantics', Linguistic Inquiry 16, 547-94. (1987), 'Indefiniteness and Predication', in E. J. Reuland & A. G. B. ter Meulen, eds. (1987). Higgins, R. (1973), 'The Pseudo-Cleft Construction in English', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. Hoekstra, E. (1987), 'Psych-verbs and the Theory of Binding', unpublished manuscript, University of Groningen.
328
References
(1989), 'Binding, Objects and the Structure of the English VP', in H. Bennis & A. van Kemenade, eds., Linguistics in the Netherlands 1989, Foris, Dordrecht. Hoekstra, T., and M. Moortgat(1979), 'Passiefin het lexicon', Forum der Letteren 20,137-61. Holmberg, A. (1986), 'Word Order and Syntactic Features in the Scandinavian Languages and English', unpublished doctoral dissertation, University of Stockholm. Huang, C.-T. J. (1982), 'Logical Relations in Chinese and the Theory of Grammar', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. (1983), 'A Note on the Binding Theory', Linguistic Inquiry 14,554-61. Huang, Yun-Hua (1984), 'Reflexives in Chinese', Studies in English Literature and Linguistics 10, 163-88, National Taiwan Normal University, Taipei. Huybregts, M. A. C. (1979), 'On Bound Anaphora and the Theory of Government-Binding', paper presented at NELS 10, Ottawa. Inoue, K. (1976), 'Reflexivization: An Interpretive Approach', in M. Shibatani, ed., Syntax and Semantics, Vol. 5: Japanese Generative Grammar, Academic Press, New York, 117-91. Jackendoff, R. (1972), Semantic Interpretation in Generative Grammar, MIT Press, 'Cambridge, Massachusetts. (1977), X' -syntax: A Study of Phrase Structure, MIT Press, Cambridge, Massachusetts. Jacobs, R., and P. Rosenbaum, eds. (1970), Readings in English Transformational Grammar, Ginn and Co, Waltham, Massachusetts. Jaeggli, O. (1986), 'Passive', Linguistic Inquiry 17, 584-623. Jaeggli, 0., and K. Safir, eds. (1989), The Null Subject Parameter, Kluwer, Dordrecht. Jakubowicz, C. (1984), 'On Markedness and Binding Principles', in C. Jones and P. Sells, eds., Proceedings of NELS 14, GLSA, University of Massachusetts, Amherst, Massachusetts, 154--82. Jespersen, O. (1933/1983), Essentials of English Grammar, Allen and Unwin, London. Kaplan, D. (197l), 'Quantifying in', in L. Linsky, ed., Reference and Modality, Oxford University Press, Oxford. Karlsson, F. (1978), Finnish Grammar, Werner Soderstrom Osakeyhtio, Juva. Katada, F. (1988), 'LF-binding of Anaphors', Proceedings of the West Coast Conference on Formal Linguistics 7, University of California, Irvine, California, 17l-86. (1989), 'The LF-Representation of Anaphors and its Theoretical Implications', paper presented at the GLOW 1989 Syntax Workshop; unpublished manuscript, USC, Los Angeles, California. Kayne, R. (1981), 'Unambiguous Paths', in R. May & J. Koster, eds. (1981). (1984), Connectedness and Binary Branching, Foris, Dordrecht. (1989a), 'Null Subjects and Clitic Climbing', in O. Jaeggli & K. Safir, eds. (1989). (1989b), 'Facets of Romance Past Participle Agreement', in P. Beninca, ed. (1989). Keenan, E. (1987), 'On Semantics and the Binding Theory', to appear in J. Hawkins, ed., Explaining Language Universals, Blackwell, Oxford. Kenesei, I., and J. Szeged, eds. (1987), Approaches to Hungarian, Vol. II, Budapest. Kiparsky, P. (1987), paper presented at the XIVth International Congress of Linguists, Berlin. (forthcoming), 'Morphology and Grammatical Relations', unpublished manuscript, Stanford University, California.
References
329
Kiparsky, P., and C. Kiparsky (1970), 'Fact', in M. Bierwisch & K. Heidolph, eds., Progress in Linguistics, Mouton, The Hague, 143-73. E. Kiss, K. (1981), 'Move Alpha and C-command in a Non-Configurational Language', GLOW Newsletter 6,41-3. (1987a), Configurationality in Hungarian, Reidel, Dordrecht, and Akademiai Kiad6, Budapest. (1987b), 'Eliminating the Configurationality Parameter', unpublished manuscript, Linguistic Institute of the Hungarian Academy of Sciences, Budapest. (1987c), 'Is the VP Universal?', in 1. Kenesei & J. Szeged, eds. (1987) (in preparation), 'More on Double Objects', unpublished manuscript, Linguistic Institute of the Hungarian Academy of Sciences, Budapest. Kitagawa, Y. (1986), 'Subjects in Japanese and English', unpublished doctoral dissertation, University of Massachusetts, Amherst, Massachusetts. Koopman, H. (1984), The Syntax of Verbs, Foris, Dordrecht. Koopman, H., and D. Sportiche (1989), 'Pronouns, Logical Variables and Logophoricity in Abe', Linguistic Inquiry 20.4,555-89. Koster, J. (1984), 'On Binding and Control', Linguistic Inquiry 15,417-59. (1985), 'Reflexives in Dutch', in J. Gueron, H. G. Obenauer, & J.-Y. Pollock, eds. (1985). (1987), Domains and Dynasties: the Radical Autonomy of Syntax, Foris, Dordrecht. (1988), 'The Residual SOV Structure of English', unpublished manuscript, University of Groningen. Kiihner, R., and C. Stegman (1912), Ausfuhrliche Grammatik der Lateinischen Sprache. Satzlehre, 1. Teil, Hahnsche Buchhandlung, Hannover. Kuno, S. (1972), 'Pronominalization, Reflexivization and Direct Discourse', Linguistic Inquiry 3.2, 161-96. (1973), The Structure of the Japanese Language, MIT Press, Cambridge, Massachusetts. (1987), Functional Syntax: Anaphora, Discourse and Empathy, University of Chicago Press, Chicago. Kuno, S., and E. Kaburaki (1977), 'Empathy and Syntax', Linguistic Inquiry 8, 627-72. Kuroda, Yuki (1965), 'Generative Grammatical Studies in the Japanese Language', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. Lasnik, H. (1986), 'On the Necessity of Binding Conditions', unpublished manuscript, University of Connecticut, Storrs, Connecticut. (1987), Essays on Anaphora, Kluwer Academic Publishers, Dordrecht. Lasnik, H. and M. Saito (1984), 'On the Nature of Proper Government', Linguistic Inquiry 15.2,235-90. Lavency, M. (1981), 'La Proposition relative du Latin c1assique', L'Antiquite Classique, Tome L.1.2. Lebeaux, D. (1983), 'A Distributional Difference between Reciprocals and Reflexives', Linguistic Inquiry 14.4,723-30. (1985), 'Locality and Anaphoric Binding', The Linguistic Review 4,343-63. (1988), 'Language Acquisition and the Form of Grammar', unpublished doctoral dissertation, University of Massachusetts, Amherst, Massachusetts. Leek, F. C. van der (1980), 'Reflexive or Non-reflexive?', Dutch Quarterly Review 2,124-46. Lees, R., and E. Klima (1963), 'Rules for English Pronominalization', Language 39,17-28.
330
References
Longobardi, G. (1983), 'Le frasi copulari in italiano e la struttura della teoria sintattica', Annali della Scuola Normale Superiore di Pisa, Pisa. (1985), 'Su alcune proprieta della forma logica delle frasi copulari', in Franchi, De Bellis & Savoia, eds., Sintassi e morfologia della lingua italiana d'uso, Bulzoni, Roma. Lust, B., ed., (1986), Studies in the Acquisition of Anaphora, Vol. I, Reidel, Dordrecht. Lust, B., R. Mazuka, G. Martohardjono and J.-M. Yoon (1989), 'On Parameter Setting in First Language Acquisition: The Case of the Binding Theory', paper presented at GLOW 12, Utrecht, unpublished manuscript, Cornell University, Ithaca, New York. Maling, J. (1982), 'Non-Clause-Bounded Reflexives in Icelandic', in T. Fretheim & L. Hellan, eds. (1982). (1984), 'Non-Clause-Bounded Reflexives in Icelandic', Linguistics and Philosophy 7, 211-41. (1986), 'Clause-bounded Reflexives in Modern Icelandic', in L. Hellan & K. Koch Christensen, eds. (1986b). Manzini, R. (1983), 'On Control and Control Theory', Linguistic Inquiry 14,421-45. Manzini, M. R. (1988), 'Constituent Structure and Locality', in A. Cardinaletti, G. Cinque & G. Giusti, eds. (1988). (1989), 'Locality', unpublished manuscript, University College London. Manzini, R., and K. Wexler (1987), 'Parameters, Binding Theory and Learnability', Linguistic Inquiry 18,413-44. Maracz, L. (1986), 'On Coreferential and Bound Variable Interpretation in Non-Configurational Languages', Theoretical Linguistic Research 2, 85-171. (1989), 'Asymmetries in Hungarian', doctoral dissertation, University of Groningen. May, R. (1985), Logical Forni, MIT Press, Cambridge, Massachusetts. May, R., and J. Koster, eds. (1981), Levels of Syntactic Representation, Foris, Dordrecht. Mikkola, R. (1986), 'Versuch einer kontrastiven Darstellung der Reflexivierung im Deutschen und im Finnischen mit einem kleinen Exkurs in die Reflexivierung im Niederlandischen', Master's thesis, University of Groningen. Milner, J.-C. (1978), 'Le Systeme de reflechi en latin', Langages 50,73-86. Mohanan, K. P. (1982), 'Grammatical Relations and Anaphora in Malayalam', in A. Marantz & T. Stowell, eds., MIT Working Papers in Linguistics, Papers in Syntax, MIT, Cambridge, Massachusetts, 163-90. (1983-4), 'Lexical and Configurational Structures', The Linguistic Review 3,113-39. Montalbetti, M., and K. Wexler (1985), 'Binding is Linking', Proceedings of the West Coast Conference on Formal Linguistics 4,228-45. Muysken, P., and H. van Riemsdijk, eds. (1986), Features and Projections, Foris, Dordrecht. Napoli, D. J. (1973), 'The two si's ofItaHan', unpUblished doctoral dissertation, Harvard, Cambridge, Massachusetts. Napoli, D. (1979), 'Reflexivization across Clause Boundaries in Italian' ,Journal of Linguistics 15, 1-28. Nishigauchi, T. (1986), 'Quantification in Syntax', unpublished doctoral dissertation, University of Massachusetts, Amherst, Massachusetts. Obenauer, H. G. (1985), 'On the Identification of Empty Categories', The Linguistic Review, 4. Pesetsky, D. (1987a), 'Wh-in-situ: Movement and Unselective Binding', in E. J. Reuland & A. G. B. ter Meulen, eds. (1987).
References
331
(1987b), 'Binding Problems with Experiencer Verbs', Linguistic Inquiry 18, 126-40. Pica, P. (1984), 'On the Distinction Between Argumental and Non·argumental Anaphors', in W. de Geest & Y. Putseys, eds. (1984), 185-94. (1985), 'Subject, Tense and Truth: Towards a Modular Approach to Binding', in J. Gueron, H. G. Obenauer & J.·Y. Pollock, eds. (1985). (1987), 'On the Nature of the Reflexivization Cycle', in J. McDonough & B. Plunkett, eds., Proceedings of NELS 17, GLSA, University of Massachusetts, Amherst, 483-99. (forthcoming), 'R·Nouns', unpublished manuscript, Pennsylvania State University, Pennsylvania. Picallo, C. (1985), 'Opaque Domains', unpublished doctoral dissertation, City University of New York, New York. Pierrehumbert, J. (1980), 'The Finnish Possessive Suffixes', Language 56, 603-21. Pollock, J.·Y. (1986), 'Sur la syntaxe de en et Ie parametre du sujet nul', in M. Ronat & K. Couquaux, eds., La Grammaire Modulaire, Minuit, Paris. (1989), 'Verb Movement, Universal Grammar, and the Structure ofIP', Linguistic Inquiry 20, 365-424. Postal, P. (1971), Crossover Phenomena, Holt, Rinehart and Winston, New York. Rebuschi, G. (1988), 'A Propos de quelques "Universaux" de la Theorie du Liage', unpub· lished manuscript, Universite de Nancy II, Nancy. Reinhart, T. (1979), 'Reported Consciousness and Point of View', Poetics and the Theory of Literature 4, 63-75. (1983a), Anaphora and Semantic Interpretation, Croom Helm, London. (1983b), 'Point of View in Language - the Use of Parenthetica Is' ,in G. Rauh, ed., Essays on Deixis, Gunter Narr, Tiibingen. (1983c), 'Coreference and Bound Anaphora: A Restatement of the Anaphora Questions', Linguistics and Philosophy 6, 47-88. (1986), 'Center and Periphery in the Grammar of Anaphora', in B. Lust, ed. (1986), 123-50. (1988), 'Wh·in·situ: WHO vs. WHICH N', paper presented at the Groningen Linguistics Colloquium, unpublished manuscript, Utrecht/Tel Aviv. Reinhart, T., and E. J. Reuland (forthcoming), 'Anaphoric Territories', unpublished manu· script, University of Tel Aviv/University of Groningen. Reuland, E. J. (1982a), 'Why Count your Auxiliaries in Dutch?', in J. Pustejovsky & P. Sells, Proceedings of NELS 12, GLSA, University of Massachusetts, Amherst, Massachusetts. (1982b), 'On the Governing Properties of Infinitival Markers', in T. Fretheim & L. Hellan, eds. (1982). (1983), 'Governing -ing', Linguistic Inquiry 14, 101-36. (1986), 'A Feature System for the Set of Categorial Heads', in P. Muysken & H. van Riemsdijk, eds. (1986). (1988), 'Phrase Structure and the Theory of Levels', in A. Cardinaletti, G. Cinque & G. Giusti, eds. (1988). (1989), 'Anaphoric Relations and Thematic Structure', paper presented at NELS 20, Pittsburgh, unpublished manuscript, University of Groningen. Reuland, E. J., and W. Kosmeijer (1988), 'Projecting Inflected Verbs', in W. Abraham, ed., Groninger Arbeiten zltr Germanistischen Linguistik 29 , University of Groningen.
332
References
Reuland, E. J., and A. G. B. ter Meulen, eds. (1987), The Representation of(In)definiteness, MIT Press, Cambridge, Massachusetts. Riemsdijk, H. van (1983), 'The Case of German Adjectives', in F. Heny & B. Richards, eds., Linguistic Categories: Auxiliaries and Related Puzzles, Reidel, Dordrecht. (1985), 'Why Long Reciprocals Don't Exist', Theoretical Linguistic Research, 37-45. Rigau, G. (1981), Gramiitica del discurs, Universita Autonoma de Barcelona, Bellaterra. Rivero, M.-L. (1977), Estudios de Gramatica Generativa del Espanol, Catedra, Madrid. Rizzi, L. (1982), Issues in Italian Syntax, Foris, Dordrecht. (1985), 'Conditions de bonne formation sur les chaines', in M. Ronat & P. Pica, eds., Categories vides et explication en syntaxe, ModeJes Linguistiques VII. 1. (1986a), 'Null Objects in Italian and the Theory of pro', Linguistic Inquiry 17,501-57. (1986b), 'On Chain Formation', in H. Borer, ed. (1986b). (to appear), 'Relativized Minimality', unpublished manuscript, University of Geneva. Rizzi, L., and A. Belletti (1988), 'Psychological Verbs and Thematic Theory', Natural Language and Linguistic Theory 6.3., 291-352. Roeper, T., and E. Williams, eds. (1987), Parameter Setting, Reidel, Dordrecht. Riignvaldsson, E. (1986), 'Some Comments on Reflexivization in Icelandic', in L. Hellan & K. Koch Christensen, eds. (1986b). (1988), 'On Empty Subjects and Objects', unpublished manuscript, University ofIceiand, Reykjavik. Ross, J. R. (1970), 'On Declarative Sentences', in R. Jacobs & P. Rosenbaum, eds. (1970). Rothstein, S. (1983), 'The Syntactic Form of Predication', unpublished doctoral dissertation, MIT, Cambridge, Masachusetts. Rullmann, H. (1988), 'Zich and zichzelf, unpublished manuscript, University of Groningen. Ruwet, N. (1984), 'Je veux partir/*Je veux que je parte', in Cahiers de Grammaire 7, Universite de Toulouse, Le Mirai!. Ruzicka, R. (1973), 'Reflexive vs. Nonreflexive Pronominalization in Modern Russian and Other Slavic Languages', in F. Kiefer, ed., Generative Grammar in Europe, 445-81 [in German in Zeitschrift fur S lawistik 17: 636-779]. Saddy, D. (1987), 'On the Notion Maximal Projection', in J. McDonough & B. Plunkett, eds., Proceedings of NELS 17, GLSA, University of Massachusetts, Amherst, Massachusetts, 539-50. Safir, K. (1985), Syntactic Chains, Cambridge University Press, Cambridge. (1989), 'Implied Non-Coreference and the Pattern of Anaphora', unpublished manuscript (draft 4), Rutgers University, New Brunswick, New Jersey. Sells, P. (1986), 'Coreference and Bound Anaphora: A Restatement of the Facts', in S. Berman, J.-W. Choe & J. McDonough, eds., Proceedings of NELS 16, GLSA, University of Massachusetts, Amherst, Massachusetts, 434-46. (1987), 'Aspects of Logophoricity', Linguistic Inquiry 18.3, 445-81. Sells, P., A. Zaenen and D. Zec (1986), 'Reflexivization Variation: Relations between Syntax, Semantics and Lexical Structure', Studies in Grammatical Theory, Volume 1: Interactions of Morphology, Syntax and Discourse, CSLI, Stanford University, Stanford. Sigurosson, H. (1986a), 'Moods and (Long Distance) Reflexives in Icelandic', Working Papers in Scandinavian Syntax 25, Trondheim. (1986b), 'On the Semantics of Long Distance Reflexives and Moods in Icelandic', unpublished manuscript, University of Stockholm.
References
333
Speas, M. (1986), 'Adjunctions and Projections in Syntax', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. Sportiche, D. (1983), 'Structural Invariance and Symmetry in Syntax', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. (1986), 'Zibun', Linguistic Inquiry 17,369-74. (1988), 'A Theory of Floating Quantifiers and its Corollaries for Constituent Structure', Linguistic Inquiry 19,425-49. Steenbergen, M. E. van (1987), 'Binding Relations in Finnish', unpublished Master's thesis, University of Groningen. Sternefeld, W. (1985), 'Deutsch ohne grammatische Funktionen: Ein Beitrag zur Rektionsund Bindungstheorie', Linguistische Berichte 99, 394-439. Stowell, T. (1978), 'What Was There Before There Was There', in D. Farkas, W. Jacobsen and K. Todrys, eds., Proceedings ofCLS XIV, Chicago. (1981), 'Origins of Phrase Structure', unpublished doctoral dissertation, MIT, Cambridge, Massachusetts. (1987), 'Small Clause Restructuring', unpublished manuscript, UCLA, Los Angeles, California. Tang, C.-c. J. (1985), 'A Study of Reflexives in Chinese', unpublished Master's thesis, National Taiwan Normal University, Taipei. (1989), 'Chinese Reflexives' , Natural Language and Linguistic Theory 7.1, 93-122. Thrainsson, H. (1976a), 'Reflexives and Subjunctives in Icelandic', in Proceedings of NELS 6, GLSA, University of Massachusetts, Amherst, 225-39. (1976b), 'Some Arguments against the Interpretive Theory of Pronouns and Reflexives', in J. Hankamer & J. Aissen, eds., Harvard Studies in Syntax and Semantics, 573-624. (1976c), 'A Semantic Reflexive in Icelandic', unpublished manuscript, Harvard University [to appear in J. Maling & A. Zaenen, eds., Icelandic Syntax, Academic Press, New York]. (1979), On Complementation in Icelandic, Garland, New York. Timberlake, A. (1979), 'Reflexivization and the Cycle in Russian', Linguistic Inquiry 10, 109-41. Toman, J. (1983), Wortsyntax, Niemeyer, Tiibingen. (1986), 'Cliticization from NPs in Czech and Comparable Phenomena in French and Italian', in H. Borer, ed. (1986b), 123-45. Uriagereka, J. (1988), 'Subject-Oriented Anaphors', guest lecture given at Cornell University; unpublished manuscript, University of Connecticut, Storrs, Connecticut. (in preparation), doctoral dissertation, University of Connecticut, Storrs, Connecticut. Vainikka, A. (1985), 'Binding -nsa', unpublished manuscript, University of Massachusetts, Amherst, Massachusetts. Vat, J. (1980), 'Zich en Zichzelf, in S. Daalder and M. Gerritsen, eds., Linguistics in the Netherlands 1980, North-Holland, Amsterdam, 127-39. Vester, E. (1985), 'Latin Relative Clauses and the Notion of Specificity', in H. Bennis & F. Beukema, eds., Linguistics in the Netherlands 1985, Foris, Dordrecht. Vikner, S. (1985), 'Parameters of Binder and of Binding Category in Danish', in Working Papers in Scandinavian Syntax 23, University of Trondheim. Wali, K. (1979), 'Two Marathi Reflexives and the Causative Structure', Studies in Language 3,405-38.
334
References
Wang, J., and J. Stillings (1984), 'Chinese Reflexives', in C. Ning et al., eds., Proceedings of the First Harbin Conference on Generative Grammar, Heilongjiang University, Harbin, People's Republic of China. Wexler, K., and Y.-C. Chien (1986), 'The Development of Lexical Anaphors and Pronouns', Papers and Reports on Child Language Development 24, Stanford University, Stanford, California. Wexler, K., and M. R. Manzini (1987), 'Parameters and Learnability in Binding Theory', in T. Roeper & E. Williams, eds. (1987),41-76. Williams, E. (1980), 'Predication', Linguistic Inquiry 11.1,203-38. (1984), 'Grammatical Relations', Linguistic Inquiry 15.4,639-75. (1985), 'PRO and Subject of NP', Natural Language and Linguistic Theory 3, 297-316. (1987), 'Implicit Arguments, the Binding Theory, and Control', Natural Language and Linguistic Theory 5, 151-80. XU, L.-J. (forthcoming), 'The Binding of ziji', unpublished manuscript, University of Shanghai/University of Tilburg. Yang, D.-W. (1983), 'The Extended Binding Theory of Anaphors', Language Research 19, 169-92. (1984), 'The Extended Binding Theory of Anaphors', Theoretical Linguistic Research I, 195-218. Yokoyama, O. T. (1980), 'Studies in Russian Functional Syntax', Harvard Studies in Syntax and Semantics 3, 451-773. Zribi-Herz, A. (1981), 'Coreference et pronoms reflechis: notes sur Ie contraste luillui-meme en fran<;ais' , Linguisticae I nvestigationes, IV. 1. (1988), 'A-type Binding and Narrative Point of View: English Reflexive Pronouns in Sentence and Discourse', unpublished manuscript, University of Paris VIII, Paris [appeared (1989) as 'Anaphor Binding and Narrative Point of View: English Reflexive Pronouns in Sentence and Discourse', Language 65.4]. Zubizarreta, M.-L. (1985), 'The Relation between Morphophonology and Syntax: The Case of Romance Causatives', Linguistic Inquiry 16, 247-89. (1987), Levels of Representation in the Lexicon and in the Syntax, Foris, Dordrecht. Zwicky, A. (1977), 'On Clitics', Indiana University Linguistics Club.
Index
A-anaphora, and A'-anaphora 7, 21 accessibility, defined 24 accessible SUBJECT 2, 77,159-60,162-3, 169 Ad-passives 106-10 acquisition, of anaphora 113-14 adjunct clauses, anaphoric elements in Latin 173-9 Agr, association of SE-anaphora with 301-4 anaphor, distinguished from anaphoric element 27-8 anaphor binding 1, 3, 245-55 anaphor/pronominal distinction inBT 87-8 contextual determination of 77-118 anaphora a-related nature of 5 acquisition of 113-14 argument structure perspective 283-322 in binary trees 151-70 classification of 81 definition 285 distribution of within binding theory summarized 284 in the extended sense, common property of 59-63 long-distance, overview 1-25 mono-morphemic and non-monomorphemic 121-2 primacy condition on 245-62 referentially defective 285-7 SD vs. LD 80-7 anaphoric domains properties of 8-10 survey 10-20
three-way split 23-4 anaphoric element, distinguished from anaphor 27-8 anaphoric pronouns, logophoric use of 22 anaphoricity 285-7 antecedent-government, interaction with binding 127-32 antecedents in Czech reflexives 154-67 non-subject in Latin 179-81 APs, a-marking properties of 5,191,195, 197 argument anaphora 6 argument modules 27, 37 argument structure, and binding 3-6, 283-322 argument-predicate structure 181-2 barriers 7,211-12,215,302 binary trees, anaphora in 151-70 binding a-marking and prepositions 185-208 and argument structure 3-6 clause-internal of Polish pronominals and anaphora 138-9 of Finnish anaphora 234-9 interaction with antecedent-government 127-32 long-distance: and the containment! connectedness distinction 27-43; definition 241; in Finnish 9-10, 231-44; theory of 239-43 and movement 6-8; in Polish 137-50 binding across a subject, see mediumdistance binding
335
336
Index
binding conditions A and E, and reflexivization 292-5 binding domains 11-24 apparent 304-8 differences between languages in 2 differences within one language 2 extension of in Polish 149-50 Finnish 9,231-44 parametric variation in 49, 70 Polish 146-8 binding theory anaphor/pronominal distinction in 87-8 and argument structure 5 based on parametrization of opacity factors too restrictive 24 Chomsky (standard) model 1-2, 49,
50-1,77,231 distinction between local and longdistance binding 231-44 explanatory theory of 24 Huang's modification 2-3 and its parameters 1-3,77 reconsidered 87-93 revised 89-91: feature representation 66 (table) syntactically defined domain extended
51-4 bound/free notion 77 c-command 5,11,245 and containment conditions 3 and role-command 11-12 Case 119, 248 centre orientation, and the semantics of logophoricity 316-17 chain-building mechanism 240-3 Chinese 3, 7, 40-1 anaphora in adjunct clause 8, 10 anaphoric domains 19-20 (table) local nature of long-distance reflexives
263-82 clitic anaphora 168 clitic climbing 123-7, 273 clitic movement, of anaphora 8 clitic reflexives 6, 81-2,123-7,168-9 and small clauses, Czech 160-3 clitics, properties of 9 coindexing 6 command module 27, 37, 43 and containment conditions 47
complementarity between pronominals and anaphors 2-3, 4,10,77-118; conditions for 24 and binding 4,313 with respect to pronominals, of various languages 11-20 complete functional complex (Chomsky) 90,
182,287 complex anaphora, local character 6, 8, 10,
284 compound anaphora 7 compound reflexives 274-5, 276 configurational languages 245 connectedness anaphora 27-48 differences from containment anaphora
43-6 connectedness conditions 3, 27, 28, 30 connectedness relation 5, 28-9 connectivity effect 272 containment anaphora 27-48 differences from connectedness anaphora
43-6 containment conditions 3, 27, 28, 29, 43 and command module 47 containment relations 28 contextual binding principles (CEPs) 97-9,
114 and 6-marking 97-9 contextual determination, of the anaphorl pronominal distinction 77-118 cross-linguistic differences in binding conditions 70-1, 77 Czech anaphoric domains 16-17 (table) reflexives 8-9,151-70
Danish 52, 63-4, 68 discourse factors, and logophoricity 3-4,
22,74,316-17 disjoint reference 110-11, 261 'double anaphor' 275 Dutch 3, 91-3,301 anaphoric domains 14-15 (table) compared with Frisian 4 preposition domain 21 pronoun distribution 106-10 pronouns and SE-anaphora 308-9 SD vs. LD 84-7 dynasty concept (Koster) 172-3, 182,240
337
Index e (Finnish) 9, 236-8 ECM structures 295-7 elkaar (Dutch) 104 empty categories 215 andCBPs 112 empty category principle (ECP) 209, 211-12 empty elements, classification of 111-13 English 23 anaphoric domains 15 (table) anaphors which are logophors 10 exceptions to standard binding theory 22 LD-reflexives 132-3 and logophoricity 311-15 possessive pronouns 63 reflexives 61-2 reflexivization 79-80 variable binding and anaphora in 5 extended domain 8, 23,149-50 Latin 171-9 extinct languages 9 extraposition, in Dutch te-infinitives 306-8 Faroese, long-distance reflexives 55, 58, 59, 64, 70 feature-copying rule 268 finite Inflll, 21-2 Finnish 8 anaphoric domains 18-19 (table) anaphoric pronominal system 9-10, 232-9 binding of anaphora 234-9 long-distance binding in 231-44 free anaphora, see logophors French 125-6 Frisian 4, 283, 298-9, 301 anaphoric domains 15 (table) pronoun system 93-7 German, possessive pronouns 63 Germanic languages, reflexivization 78,80 Gothic 52 governing category 2,3,5,20,23,77,88-9, 211,245,287 definition 1 and domain specifications 11 minimal of a lexical item 90, 114 see also opacity factor governing subject (Koster) 176
government dynasty ofl72-3, 182,240 and locality conditions 6-7, 209-30 see also antecedent-government government chain 239, 240-3
hiin itse (Finnish) 9, 237-8 Hebrew 316 hierarchy (Giorgi) 3, 190 him 6,59-60 him (Frisian) 93-7,100-4 himself6, 23, 60, 209, 213-14, 273 himsels (Frisian) 93-5 Hungarian anaphoric domains 19 (table) variable binding and anaphora in 5, 246-7,252-7,261 Icelandic 3, 33-7, 52-3, 63-4, 315 anaphoric domains 12-13 (table), 23-4 logophors 23 long-distance reflexives 54-8, 59, 70, 79 pronouns and SE-anaphora 309 reflexives 62, 79 independent targeting, principle of (Hellan) 32-4,182 infinitival clauses, Polish pronominals and anaphora in 139-42 Infl-movement theory 269-71,280 Italian 3, 22-3, 64-5 anaphoric domains 17-18 (table) locality and parameters in syntax 209-30 long-distance reflexives 55, 79 reciprocals 7 itse (Finnish) 9, 236-8 Japanese 3, 39-40 language acquisition, and anaphora 113-14 Latin 9, 54,171-84 anaphoric domains 17 (table) lexical argument hierarchy 247-55 precedence in 258-61 lexical properties, of anaphoric character of NPs 4, 61-70 lexical structure 247 licensing of anaphora 7 licensing of reflexivity 290-2,300 local anaphora, and non-local 4-5, 6, 20-4
338
Index
local binding 23-4 argument binding via the a-grid 3, 10 and long-distance binding 10, 231, 239-43 SELF-anaphora 287-301 local domain 8, 23, 149,283 locality and LD-anaphora in Chinese 263-82 and parameters 210-16 in Italian syntax 209-30 logical form (LF), and X' theory 121-2 logophoric pronouns 58-9 logophoric SE-anaphora 315-16 logophoric use of anaphoric pronouns 22 logophoricity 4,9, ll, 24, 3ll-17 and discoursefactors 22, 74, 316-17 with SELF-anaphora 3ll-15 semantics of and centre orientation 316-17 and structural binding relations 4, 22-4, 283-321 logophors 23 argument structure perspective 283-322 long-distance anaphora (LD-anaphora! LDAs) defining characteristic 10 distinction from local 4-5, 20-4 properties of 10-11 ['uno I'altro (Italian) 7, 21, 82, 216-21
m (Frisian) 5 Malayalam 69, 70 Marathi 68, 69 markedness 215 medium-distance binding 8, 9, 10,24,284 mono-morphemic anaphora (simplex) 6, 121-2 Mood 22 movement binding and 6-8 theory of, and parameters 228-9 'mzelf(Dutch) 24-5 non-local anaphora 6, 8 Norwegian 3,31-3,310 anaphoric domains 13-14 (table) connectedness and containment anaphora 43-6 NPs a-requirements of 5,191-5
binary classification 50 binding of anaphors in 7 in Czech 9 new classification: features 65-70; schematic 63-5 picture 297-8 Polish pronominals and anaphora in 144-6 referential features, see phi-features referential properties of 4,61-70 standard binding theory classification of 50-9 towards a new classification of 63-70 typology of 49-76 null hypothesis 4
opacity factors and domain specifications 7, 20-1 explanation of 24 parametrizing of 2
parameters 228-9 and locality 210-16 in Italian syntax 209-30 parametrization 7,20,22, 137,209,214 of opacity factor 2 participial clauses, Polish pronominals and anaphora in 142-4 perspective oflogophoricity 28,316 theory of23 perspective command 12, 22 phi-features 7-8,10,274-9,287,301-2, 317 phrase structure theory 210 picture NPs 297-8 Polish 8, 22,137-8,165,310 anaphoric domains 16 (table) binding facts 137-46 binding theory for 146-50 pronominals and anaphora 138-46: clause-internal binding 138-9; in infinitival clauses 139-42; in NPs 144-6; in participial clauses 142-4 reciprocals in 20 possession, inalienable 222-8 possessive reflexives, Czech 153-4, 165-7 possessive suffix, Finnish, 232-4
Index PPs a-marking properties of 5 anaphora and pronouns 5, 185-208 asymmetries in English 5 in other small clauses 192-5 and role command requirement 12 precedence, in the lexical argument hierarchy 5, 245, 258-60 predication 28 minimal 21 predication command 12, 32-4, 182 predication relation 191, 314 predicative structures 195-7 prepositions binding and a-marking 185-208 and external a-role assignment 190,
192-202 and LD-anaphora 185-9 locative vs. non-locative 198-200 thematic properties 190-208 primacy condition on anaphora 245-62 of anaphora, summary 254 of pronominal variable binding 260-1 primacy principles 245 prominence requirements, of various languages 11-20 prominent argument 179-81, 182, 190 pronominal elements, classification of 77 pronominal epithets 67 pronominal variable binding 255-60 primacy condition of 260-1 pronominal/anaphor distinction in binding theory 87-8; contextual determination of 77-118 pronominals and binding theory 1 opacity factor for 2 Polish 138-46 pronouns Frisian system 93-7 logophoric 58-9 and SE-anaphora 308-9 unidentified 93-106: and empty categories 112; Frisian and Dutch
100-4 proposition, minimal 21 proprio (Italian) 2,5,22,23,37-9, 186-9 R-expressions 1,67
339 reciprocals 5, 11, 104-6 Italian 7, 216-21 reconstruction problem 272 reference disjoint 110-11,261 independent or deictic 61-3,65,70 referential index 7-8 reflexive clitics 300-1 reflexive marking 290-2 reflexive predicates 288-9, 298-301, 313 reflexive pronouns, Czech 151-3 reflexive verbs 99-100, 300-1 reflexives antecedents in Czech 154-67 clause-bound 129-32 cross-linguistic variation and lexical properties 6 Czech: anaphors in binary trees 151-70; full and small clauses 9, 155-60; summary of findings 168-9 (table) first person English 311-12 implicational generalizations 154 logophoric 49-50,59 long-distance 11, 49-75,119-36: argument-like character of 122-4; binding requirement not necessarily structural 4; bound and unbound 54-9; in English 132-3; local nature of Chinese 263-82 non-clause bounded (NCBR), in Icelandic
2,34-7 non-mono-morphemic, adjunct-like character of 130-2 potential binders of Chinese 265-7 successive cyclicity 6,279-80 see also clitic reflexives; possessive reflexives reflexivization clause-bounded 79-80 into non-tensed complements 79 into subjunctive complements 79-80 long-distance, and switch reference 128-9 as a property of predicates 288 types of 78-80 reindexing rule 268 relation modules 27 relative clauses restrictive vs. non-restrictive 174-7 subjunctive 174, 177, 182 representation relation 317
340
Index
role-command, and c-command 11-12 Romance reflexivization 78 root cause, and logophors 23 Russian 166, 300 Scandinavian languages 31-7, 306 reflexives in 4, 51-4 se (Czech) 8, 9 se (French) 81 se (Italian) 5, 64, 186-9 se (Latin) 9 se stesso (Italian) 5,186-9,209 SE-anaphora 284, 301-8 logophoric 315-16 pronouns and 308-9 and SELF-anaphora 309-11 sebe (Czech) 8, 9 seg (Icelandic) 22 seg (Norwegian) 31-3, 43-6 seg selv (Norwegian) 43-6 SELF-anaphora 284 and condition A 312-14 and condition B 314-15 local binding 287-301 logophoricity with 311-15 reflexive and non-reflexive contexts 288-90 and SE-anaphora 309-11 semantics oflogophoricity, centre orientation and the 316-17 si (Italian) 81 siebie (Polish) 8,83,310 sig (Danish) 119 sig (Icelandic) 6, 23, 33-7, 60-1, 119 sin (Icelandic) 22, 23 sin (Norwegian) 31-3, 43-6 Slavonic languages 137 small clauses 305 and clitic reflexives, Czech 160-3 in Czech 9, 155-65 inflected in Czech 163-5 ,PPs in 192-5 speaker construal, and containment and connectedness anaphora 46 specified subject constraint (SSC) 119, 121, 284,285 structural binding relations, and logophoricity 4, 22-4, 283-321 structural binding requirements, and LDreflexives 4, 49-76
structure preserving constraint (Emond) 121 SUBJECT, generalized 119 subject specified 119 varying use of notion 12 see also accessible SUBJECT subjunctives 22,174,177,182,305 and non-clause-bounded reflexivization 34-7,79-80 superordinate subject 72 svUj (Czech) 8 Swedish 300,310 switch reference, and long-distance reflexivization 128-9 sw6j (Polish) 8
taziji (Chinese) 7, 263, 274 Tense 2, 22, 119 O-domain 90 O-government 5, 98-9 O-marking binding and prepositions 185-208 and contextual binding principles 97-9 O-roles assignment 5,190-2,298 hierarchy of 3,179,249-52 O-theory and binding theory 287 and selection of anaphors 3, 97-104 third-person anaphora, used logophorically 316-17 toiset (Finnish) 9, 234-6 Topic prominence 181, 182 uniqueness requirement, and governing category 89 variable binding 317 and anaphora in Hungarian 5, 245-62 verbs, reflexive 99-100, 300-1 VP-anaphora, and containmentconnectedness distinction 44-6 West African logophoric pronouns 58
zibun (Japanese) 39-40 zich (Dutch) 4-5,21,24-5,41-3,84-7, 91-3,305-9 as unidentified pronoun 100-4 zichzelj(Dutch) 21,84-7,91-3 ziji (Chinese) 7, 40-1, 263-80 long-distance 267-71; locality of272-80
