Essentials of functional grammar a structure neutral theory of movement, control, and anaphora

Essentials of Functional GrammarTrends in Linguistics Studies and Monographs 38EditorWerner WinterMouton de Gruyter Berlin ¡ New York ¡ AmsterdamEssentials of…
of 417
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Essentials of Functional GrammarTrends in Linguistics Studies and Monographs 38EditorWerner WinterMouton de Gruyter Berlin ¡ New York ¡ AmsterdamEssentials of Functional Grammar A Structure-Neutral Theory of Movement, Control, and Anaphora byGeorge M. HornMouton de Gruyter Berlin ¡ New York ¡ Amsterdam 1988Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.Library of Congress Cataloging-in-Publication Data Horn, George M. Essentials of functional grammar. (Trends in linguistics. Studies and monographs ; 38) Bibliography: p. Includes index. 1. Grammar, Comparative and general. 2. Anaphora (Linguistics) I. Title. II. Series. P151.H57 1988 415 88-18947 ISBN 0-8992-5348-2 (alk. paper)Deutsche Bibliothek Cataloging-in-Publication Data Horn, George M.: Essentials of functional grammar : a structure-neutral theory of movement, control, and anaphora / by George M. Horn. Berlin ; New York ; Amsterdam : Mouton de Gruyter, 1988 (Trends in linguistics : Studies and monographs ; 38) ISBN 3-11-011286-8 NE: Trends in linguistics / Studies and monographsPrinted on acid free paper.Š Copyright 1988 by Walter de Gruyter & Co., Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form - by photoprint, microfilm, or any other means - nor transmitted, nor translated into a machine language without written permission from Mouton de Gruyter, a Division of Walter de Gruyter & Co., Berlin. Printing: Ratzlow-Druck, Berlin - Binding: LĂźderitz & Bauer, Berlin. - Printed in Germany.AcknowledgmentsI am grateful to the following people for their time and patience in assisting me with the data: Ola Horn (Polish), Anita van der Wal (Dutch), Melinda Meredith (Samoan), Nobuaki Nishioka (Japanese), and Kiyoharu Ono (Japanese). I, of course, take full responsibility for the accuracy of the data and the conclusions reached. I also wish to thank Geraldine MacNeill and Peter Peterson for their help in editing and proofreading. Special thanks are due to Joyce Bennett, who worked long hours typing the manuscript and assisting in the editing and revision. Her services have been invaluable.INTRODUCTIONLexical-Functional Grammar, Horn (1983a), is essentially a collection of analyses of various bodies of data which are formulated in a functional framework related to that of Bresnan et al., yet differing from it in certain significant respects. This basic approach, I believe, is correct, but the emphasis is on the mechanical properties of the rules of the model, and relatively little attention is given to higher-level generalizations. Here, I develop a coherent theory of universal grammar (UG) along the same general lines. In this proposal, UG is composed of general rule schemata and related principles and parameters which together define, in effect, the notion "possible grammatical process." The major grammatical processes which I consider are what are traditionally labelled movement, control, and anaphora. Cross-language variation in each of these areas is a consequence of variation of the values of the relevant parameters. On this level, the proposed theory can be more meaningfully compared with the Government and Binding (GB)Theory and the other variant of Lexical-Functional Theory in the context of the debate over the configurational/non-configurational distinction and the nature of the rules and principles which provide the best explanation of the behavior of languages of these two types. The most important specific innovations of this proposal are the following: (a) The operations on F-structures discussed in Horn (1983a) are shown to be subcases of two generalized rule systems: the Argument Reduction System and the Generalized Co-Indexing System. The first includes the passivization process, certain other lexical and non-lexical processes which were discussed briefly, and object deletion. The Argument Reduction rules are roughly analogous to, but more general than, the Move NP subcase of the Move-Îą rule in the GB Theory. The remaining operations, the co-indexing rule, the Raqui rule, the reflexive/reciprocal rule, and the Gerundive/Infinitive Phrase Subject Interpretation Rule, are subcases of a single General Co-Indexing Rule,which is the analog of the control and anaphora rules of the GB theory, but, which again, applies to a rather wider domain. (b) An S-structure/F-structure Interface Component has been added to UG. This contains rales and principles which mediate between the S-structure and F-structure levels. The role of grammatical functions, as well as the relationship between case and grammatical function is more clearly developed, and the Grammatical Relations Deletion Rule, whose sole function was a house-keeping one, is eliminated. (c) The four universal F-structure types are replaced by a single canonical form. This, along with a small set of parameters, provides a definition of the notion "possible predicate." (d) The constraints on the EA rule, which is the analog of whmovement in the GB Theory, are generalized. The Noun Phrase Constraint (NPC) is shown to be a subcase of a more general Predicate Argument Constraint, which defines the conditions under which both NPs and APs are islands as well as those under which they are not. The NPGOC is eliminated. A set of parameters associated with the remaining two constraints defines the range of cross-language variation of the patterns of wh-movement. Some knowledge of the material in Horn (1983a) will be helpful, but not essential to understanding this proposal.TABLE OF CONTENTSIntroductioniChapter I: Preliminary Remarks11.Universal Grammar2. 3. 4.Configurational and Non-configurational Languages Structural and Functional Theories of UG Hie Proposed TheoryChapter Î : Basic Concepts 1. Overview 2. Basic F-structure Components: Predicates 3. Basic F-structure Components: Arguments 4. Building Complex F-structures 5. The S-structure/F-structure Interface 6. Summary and Concluding Remarks Notes Chapter III: The Co-Indexing System: A Reanalysis of Control and Bound Anaphora 1. Introduction 2. The Short-range Control Relation 3. Bound Anaphora 4. The Long-range Control Relation 5. Unification 6. The Generalized Co-Indexing System Notes Chapter IV: The Argument Reduction System: A Reanalysis of Bounded Movement and Deletion 1.Introduction1 4 9 31 35 35 40 60 87 99 125 142165 165 167 185 199 215 222 229243 2432. The Argument Reduction Relation 3. Extensions: The Generalized Argument Reduction System 4. Constraints: Defining the Notion "Possible Asymmetric FS-pair" 5. Summary and Conclusions Notes Chapter V: The EA System: A Reanalysis of Wh-movement 1. Introduction 2. The Ε-Anaphora Rule 3. The Predicate Argument Constraint 4. The Single Gap Constraint 5. Variation 6. The Generalized Ε A System Notes246 270 297 309 317329 329 330 338 356 368 381 384Bibliography394Index401Chapter I Preliminary Remarks1. Universal Grammar In formal terms, universal grammar (UG) consists of general rules and principles from which particular rules of grammar for particular languages are derivable by fixing the values of relevant parameters in permissible ways. An adequate theory of UG must be based on principles which restrict the class of grammars of natural languages in some meaningful way and parameters whose values vary within definable limits. In such a theory, the significant universal grammatical patterns and the observed range of variation within those patterns are a consequence of the interaction of these principles. Therefore, an adequate theory of UG, in effect, defines the notion "possible grammatical process." On the syntactic level, the major grammatical processes fall into three categories: movement, control, and anaphora. We may think of movement as the process which relates sentence pairs, or S-structure pairs, S j ^ , in which some constituent X occurs in position A in Sj, and in position Î’ in S 2 . One example is the active/passive pair shown in (1.1):(1.1) a. Brunhilda hated Smyth b. Smyth was hated by Brunhilda Here, the constituent X is either Smyth or Brunhilda. In the former case, position A is object position and position Î’ is subject position. Alternatively, we may think of movement as the process which relates pairs of successive phrase markers, PM1/PM2, in which some constituent X occurs in position A in PMj and in position Î’ in PM 2 . This is illustrated below:2 1. Preliminary Remarks (1.2) a. Brunhilda seemed to hate Smyth b.S NP[e]VPseemedc.^ p [Brunhilda] to hate SmyuiS NPVPBrunhilda Here, Brunhilda is constituent X, position A is the complement subject position in Structure b, and position Î’ is the matrix subject position in Structure c. Equivalently, we may define movement as the relation between Si/S 2 pairs or PM1/PM2 pairs of this sort. The control relation is the relation between a noun phrase (NP) antecedent in a phrase marker and an understood complement subject position, which contains no lexical material, elsewhere in the structure. This is illustrated by the following examples: (1.3) a. Hortense tried to eat termites b. Hortense forced her dog to eat termites In example (1.3a), Hortense, the subject of the verb try, is interpreted as the subject of the embedded verb eat also. Hortense is the controller of the phonologically empty subject position of eat. In some analyses, this position is not represented at any level of syntactic structure (see Brame 1976a,b and Bresnan 1978, for example), and in other analyses, it is represented at the D-structure and S-structure levels as PRO or its equivalent (see Chomsky 1981). Example (1.3b) is analyzed in the sameUniversal Grammar3way except that the object, her dog, is the controller. The anaphora relation is the relation between an overt element, or anaphor, which is a member of a set that is designated in each language, and an NP antecedent elsewhere in the structure that it is obligatorily related to, as shown below: (1.4) Randolf saw himself in the mirror Here, himself is the overt anaphor and Randolf is the antecedent The issue which I raise here does not concern the empirical adequacy of any of the current major theories, but rather the nature of the principles upon which an adequate theory of UG should be based. It is impossible to argue for or against a theory on purely empirical grounds. Any theory can be augmented to account for additional data. The criterion for evaluating theories is the rather more subtle one of relative plausibility. There are two basic approaches to the construction of a theory of UG. The first is what we may call the structural approach and the second is what we may call the functional approach. The first approach begins with the assumption that structural configurations are primitive, and that the principles of UG are formulated, or defined, in terms of structural configurations. GB theory is built on this assumption. Although this theory has appeared in several significantly different and increasingly sophisticated versions over the last fifteen or so years, all of the versions share this characteristic. The functional approach, in contrast, begins with the assumption that grammatical relations, or grammatical functions, are primitive, and are not derived from, or defined in terms of, syntactic configurations. The principles which interact to explain the patterns of movement, control, and anaphora are formulated in terms of grammatical function; that is, in terms of functions like subject and object. In this approach, structural configurations are also primitive, but the relation between structural configurations and grammatical relations or functions is quite different from that of the structural approach. Lexical Functional Grammar (LFG) is based on this assumption, and perhaps represents the extreme version of the functional approach. In the next section, I look at the fundamental body of data which any theory of UG must explain, and in Section 3,1 suggest that the functional4 1. Preliminary Remarks approach is the more fruitful of the two for constructing a theory of UG. I close the chapter with a broad outline of the basic characteristics of the proposed theory, which is based on this approach and the major differences between it and the two most important current theories: GB theory and LFG. 2. Configurational and Non-Configurational Languages Hale (1981) suggested that there are two major, structurally distinct, language types: configurational and non-configurational. Non-configurational languages are ones which have a certain set of properties which include relatively free word order, extensive case-marking systems, and discontinuous elements. He also claims that there is little or no evidence in such languages of NP movement. Most of the properties associated with the configurational/non-configurational distinction are relative at best. For example, there are probably no languages with completely free word order, or languages with no intra-clausal constituent structure. There are probably degrees of non-configurationality. There is also some disagreement over the configurationality or non-configurationality of a number of particular languages. These issues, however, are matters of detail, which need not concern us here. Some examples of non-configurational languages are Warlpiri (Hale 1981 and 1983), Guugu Yimidhirr, Japanese (Hale 1981), Icelandic (Andrews 1982), and Malayalam (Mohanan 1982.) Sentences in each of these languages, accompanied by their surface structures, are shown below: (2.1) a. Wawirri kapi-rna panti-rni yalumpu kangaroo AUX spear-NONPAST that I will spear that kangaroo Sb.wawirrikapi-mapanti-rniyalumpuConfiguration Vs Non-configuration 5 (2.2) a. Yarrga-ngun nambal dudaay-mani boy (ERG) rock (ABS) roll away-CAUSE-past The boy rolled the rock away b.yarrga-ngun nambaldudaay-mani(2.3) a. Hanako ga piano Îż hii-ta Hanako (NOM) piano (ACC) play-PAST Hanako played the piano(2.4) a. Stulkan kyssti drengina the-girl (NOM) kiss-PAST-3rd person-sg the-boys (ACC) The girl kissed the boys(2.5) a. Kufti aanaye kaatu child (NOM) elephant (ACC) see-PAST The child saw the elephant61. Preliminary RemarksAlthough the differences among these languages are obvious even from these few examples, the characteristic which they all share, and which distinguishes them from configurational languages like English, is the absence of various syntactic categories, in particular, the VP category, at the surface structure level. It is claimed that this lack exists at all levels of structure, and that their clausal structures are produced by rules of the following type: (2.6)S ~> Χ* V X*In this rule schema, X* represents a string of zero or more [-V] lexical categories or maximal projections, and S may be considered to be V . The relative position of V in the string may vary from language to language, and there may be other constraints on order. A comparison of these examples and the English example in (2.7), shown with its configurational surface structure, illustrates this contrast: (2.7) a. Matthew saw the elephant b.SMatthewVNPsawthe elephantOf course the same sorts of variation in preferred word order, constituent categories, and so forth are observed within the configurational language type. The crucial difference between structures which are variants of the type in (2.7) and structures like (2.1b) to (2.5b) is the structural subject/object asymmetry of the former structures and theConfiguration Vs Non-Configuration 7 absence of this asymmetry in the latter. These structural differences notwithstanding, essentially the same major grammatical patterns occur in both language types. Firstly, those patterns which are attributed to some movement process in configurational languages occur in non-configurational languages as well. Secondly, embedded complements of designated types, whose subject positions (however formally represented) are externally controlled, occur in both configurational and non-configurational languages. Finally, in both language types, anaphors in non-subject NPs are bound to subject NP antecedents, but anaphors in subject NPs cannot be bound to non-subject NP antecedents within the same clause. To illustrate the first point, consider the Japanese and Malayalam active/passive sentence pairs in (2.8) and (2.9), and the Icelandic sentences in (2.10): (2.8)a. Naomi ga Seiji Îż ut-ta Naomi (NOM) Seiji (ACC) hit-PAST Naomi hit Seiji b. Seiji ga ut-are-ta Seiji (NOM) hit-PASSIVE-PAST Seiji was hit (2.9)a.Kufti aanaye aafaatfhiccu child (NOM) elephant (ACC) worship-PAST The child worshipped the elephantb. Aana aafaadhikkappeftu elephant (NOM) worship-PASSIVE-PAST The elephant was worshipped (2.10)a. Hann verĂśist $[elska hana] He (NOM) seems [to-love her (ACC)] He seems to love her b. Skipstjorinn reyndist s [vera fifl] the-captain (NOM) proved [to be a fool] The captain proved to be a fool81. Preliminary RemarksIn Examples (2.8) and (2.9), the NPs which correspond to the subjects of the active sentences (in a.) are optional in the passive counterparts (in b). The location of these NPs when they occur is not important to this discussion. In each case, the object NP in the active member of the pair corresponds to the subject NP in the passive member of the pair. This relation is identical to that between the members of English active/ passive pairs like (1.1) in this respect. The sentences in (2.10) contain verbs that are members of the analog of the English seem class. The S-structure subjects of these verbs are interpreted as the subjects of the complements. This pattern is identical to the pattern created by the rule of raising to subject position (a subcase of the Move-α rule in the standard theory) in English, illustrated in (1.2). The following Icelandic and Malayalam sentences contain obligatorily controlled complements: (2.1 l)a.fig vonast til $[aö maslast vel ί kirkjunni] I(NOM) hope to [to speak well in the church] I hope to speak well in churchb. "peir akväöu $[aö vitja Ölais] they (NOM) decided [to visit Olaf] They decided to visit Olaf c. Hun skipaöi honum $[aö fara] She (NOM) ordered him (DAT) [to go] She ordered him to go (2.12)Ellaawarkkum [faawile kulik'k'unnate] i$t£maane All (DAT) [morning bathe-PRES-it] liking-be-PRES Everyone likes bathing in the morningThese sentences contain control verbs which are the analogs of the English try/force class. In each case, an NP argument of the matrix predicate is also interpreted as the subject of the embedded predicate. This pattern is identical to the English control pattern in (1.3), which is attributable to PRO indexing in the standard theory. The following Malayalam examples illustrate the bound anaphoraStructural and Functional Theories9pattern in that language: (2.13)a. Raajaaw-a swantam bhaafyaye nulli king (NOM) self "(POSS) wife (ACC) pinch-PAST The king pinched his own wife b. *Raajaawine swantam bhaafya nulli king (ACC) self (POSS) wife (NOM) pinch-PAST •His own wife pinched the king In example (2.13a), the reflexive form swantam occurs in a nonsubject NP and refers to the subject NP, raajaawe. In the ungrammatical example (2.13b), the reflexive form occurs in the subject NP swantam bhaafya and its antecedent is the object NP, raajaawine. This is functionally identical to the English pattern, which is shown in (2.14): (2.14)a. The king pinched himself b. *Himself pinched the king An adequate theory of UG must accommodate the configurational/ non-configurational distinction in a natural way, and explain the differences between languages of these two types, which I have only mentioned in passing. More importantly, however, it must explain the occurrence of essentially the same major grammatical patterns of movement (defined as a relation betwee
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!