Gaps and Dummies [Reprint 2010 ed.] 9783110889536, 9783110132762

Citation preview

Gaps and Dummies

Linguistic Models The publications in this series tackle crucial problems, both empirical and conceptual, within the context of progressive research programs. In particular, Linguistic Models will address the development of formal methods in the study of language with special reference to the interaction of grammatical components Series Editors: Teun Hoekstra Harry van der Hülst Michael Moortgat

Other books in this series: 1 Michael Moortgat, Harry van der Hülst and Teun Hoekstra (eds.) The Scope of Lexical Rules 2 Harry van der Hülst and Norval Smith (eds.) The Structure of Phonological Representations. Part I. 3 Harry van der Hülst and Norval Smith (eds.) The Structure of Phonological Representations. Part II. 4 Gerald Gazdar, Ewan Klein and Geoffrey K. Pullum (eds.) Order, Concord and Constituency 5 W. de Geest and . Putseys (eds.) Sentential Complementation 6 Teun Hoekstra Transitivity. Grammatical Relations in Government-Binding Theory 1 Harry van der Hülst and Norval Smith (eds.) Advances in Nonlinear Phonology 8 Harry van der Hülst Syllable Structure and Stress in Dutch

Gaps and Dummies Hans Bennis

1987 FORIS PUBLICATIONS Dordrecht - Holland/Providence - U.S.A.

Published by:

Foris Publications Holland P.O. Box 509 3300 AM Dordrecht, The Netherlands Sole distributor for the U.S.A. and Canada: Foris Publications USA, Inc. P.O. Box 5904 Providence U.S.A. CIP-DATA Bennis, Hans Gaps and dummies / Hans Bennis. - Dordrecht [etc.]: Foris. - (Linguistic Models; no 9) With index, ref. ISBN 90-6765-245-8 bound ISBN 90-6765-246-6 paper SISO 805.4 UDC801.56 Subject headings: empty categories; syntax / dummy pronouns; syntax.

First Printing Second Printing

1986 1987

ISBN 90 6765 245 8 (Bound) ISBN 90 6765 246 6 (Paper) © 1986 Foris Publications - Dordrecht No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission from the copyricht owner. Printed in The Netherlands by ICG Printing, Dordrecht.

for Annabertha

Table of Contents Acknowledgements Introduction 1 . Gaps and Parasitic Gaps Introduction 1 .1 Connectedness 1.2 Kayne's g-projection 1 .2.1 Two m o d i f i c a t i o n s of K a y n e ' s g - p r o j e c t i o n 1.2.2 Extraction from complement clauses 1.2.3 Complex cases of PPs 1.3 Van R i e m s d i j k ' s analysis of P-stranding 1.3.1 1 .3-2 A d j a c e n c y and P-stranding E x t r a c t i o n from complex PPs 1.3.3 Reanalysis in complex PPs 1.3.4 1.4 Parasitic gaps Parasitic gaps in Dutch 1.5 The d i s t r i b u t i o n of parasitic gaps 1.5.1 Parasitic gaps w i t h o u t real gaps 1 .5.2 1.6 Chains and f e a t u r e s An alternative approach to p a r a s i t i c gaps in Dutch 1.7 The analysis of Huybregts and Van R i e m s d i j k 1.7.1 1.7.2 An argument against our analysis Parallelism between across-the-board and 1.7-3 parasitic-gap c o n s t r u c t i o n s Notes

2. Het as a r e f e r e n t i a l expression 2.1 2.2 2.3 2.4

2.4.1

Introduction Weather-het R e f e r e n t i a l p r o p e r t i e s of the dummy pronoun net Het and sentential complements Het in object position

7 7 8 8 11 15 18 19 21 25 29 39 44 44 54 71 79 80 82 83 88 93 93 95 98 103 103

Table of Contents 2.4.2 2.5 2.6 2.7 2.8 2.9 2.10 2.10.1 2.10.2 2.10.3 2.11

Het in subject position Het-Raising Bisentential verbs Het in Small-Clause constructions Het and ergative verbs Psychological verbs Inversion phenomena Previous proposals Inversion and the Extended Projection P r i n c i p l e Het and Inversion Summary Notes

3. The adverbial pronoun er 3.1 Introduction 3.2 The d i s t r i b u t i o n of er 3.2.1 E x i s t e n t i a l / e x p l e t i v e er 3.2.2 Prepositional er 3.2.3 Quantitative er 3.2.4 Locative er 3.2.5 Co-occurrence of s y n t a c t i c f u n c t i o n s of er 3.2.6 An analysis of co-occurrence 3-3 Categorial status and derivation 3.3.1 The categorial status of prepositional R-pronouns 3-3-2 Movement of prepositional R - p r o n o u n s 3.3.3 The syntax of locative er 3.3.4 Quantitative er as a non-thematic argument 3.4 R-movement and the d o u b l e - R - c o n s t r a i n t 3.4.1 Van R i e m s d i j k ' s double-R-constraint 3.4.2 R-movement and a d j u n c t i o n 3-5 Expletive er: a dummy pronoun? 3.5.1 The d i s t r i b u t i o n of expletive er 3.5.2 Syntactic analyses of the construction with expletive er 3.5-3 Pragmatic m o t i v a t i o n of er-insertion 3 . 5 . 3 . 1 Pragmatic p r i n c i p l e s governing the application of a d j u n c t i o n 3 - 5 . 3 . 2 Pragmatic principles and expletive er

108 110 116 120 131 134 146 14? 151 160 163 165 171 171 174 174 176 177 177 178 181 188 188 195 197 198 205 206 207 211 212 217 221 222 224

Table of C o n t e n t s 3.6 3.6.1 3.6.1.1 3.6.1.2 3.6.1.3 3.6.2 3.7

Er and wh-movement Dat-_t phenomena Previous accounts Subject extraction and the Gap Condition Subject extraction and the EPC Subject e x t r a c t i o n in main clauses Conclusion Notes

228 229 229 239 243 247 248 250

4. Some related topics 4.1 Introduction 4.2 A d j u n c t i o n and Connection 4.3 The external argument and the s t r u c t u r e of Modern English 4.4 The s t r u c t u r e of Old English and its development into Modern English 4.4.1 The structure of Old English 4 . 4 . 1 . 1 P-stranding 4 . 4 . 1 . 2 That-_t configurations 4 . 4 . 1 . 3 Impersonal constructions 4.4.2 Development into Modern English: the INFL parameter 4.5 Some remarks on the s t r u c t u r e of the Romance languages 4.5.1 The postverbal subject in Spanish 4.5.2 Stylistic Inversion in French 4.5.3 The s t r u c t u r e of Italian and the syntax of ne 4.6 The pronoun es in German 4.7 It as a referential expression 4.8 Concluding remarks Notes

259 259 260 267

296 297 301 308 312 317 319

References

329

275 276 276 281 282 291 295

Acknowledgements So many people have -directly or indirectly- contributed by their presence, friendship and thinking to this book and more generally to my pleasure and interest in linguistics that I cannot even begin to try to acknowledge their support by mentioning their names here, for I am bound to forget one or the other. Instead, I shall single out the three people who have contributed most to this book. Heartfelt thanks go to Henk van Riemsdijk who has been a most influential person throughout my linguistic career, first as my teacher, then as the supervisor of my research project, as a short-time employer, and finally as the supervisor of this thesis. He has always been a good friend. Teun Hoekstra has had a profound influence on the contents of this book. For three years now we have worked together quite closely on a variety of projects. One concrete result of our collaberation is the first chapter of this book, and his influence is pervasive in the other chapters. I have benefited a great deal from discussions with him which have led to substantial revisions of earlier versions. Frits Beukema is the third person whose influence is very evident to the eye. I have profited much from his comments on earlier versions. I would also like to thank my parents, who have always encouraged me in my search for knowledge and insight. My final and deepest thanks go to Annabertha. Her love and friendship helped me through difficult periods and made it possible for me to finish this book. The dedication of this book to her is only a minor compensation for the lack of attention, the touchiness and the weariness she had to put up with during the past two years. That she has been able to cope with this, even in those times when our personal situation was under great stress and in distress, leaves me deeply in her debt.

Introduction In this study the syntactic properties are investigated of empty categories, i . e . " g a p s " , and what are sometimes r e f e r r e d to as "semi-empty categories", i.e."dummies". Whereas the existence of lexical categories is relatively t h e o r y - n e u t r a l , it is clear that the existence of gaps and dummies is to a large extent d e t e r m i n e d by theoretical considerations. One can easily imagine a theory without empty categories and dummy pronouns. The theoretical framework adopted here is the Government-Binding Theory, as developed in Chomsky ( 1 9 8 1 , 1 9 8 2 ) . Throughout this study I will assume f a m i l i a r i t y with the central concepts of this t h e o r y , such as the notions of Universal G r a m m a r , the Projection P r i n c i p l e . Government, the theories of B i n d i n g , Case and thematic roles, and so on. I shall not sketch these major theoretical assumptions h e r e . For readers unfamiliar with GB-theory a rapid overview would be i n s u f f i c i e n t , while it would be s u p e r f l u o u s for readers familiar with this theory. GB-theory differs quite substantially from early transformational theories such as Chomsky ( 1 9 5 7 , 1 9 6 5 ) . The main d i f f e r e n c e is that the centre of a t t e n t i o n has s h i f t e d from rules to principles. Specific t r a n s f o r m a t i o n a l rules have been replaced by the general rule schema "move α ". The a p p l i c a t i o n of this general rule is r e s t r i c t e d by a variety of c o n d i t i o n s , which are derived from the subtheories of Case, Government, 9-roles etc. Like transformational r u l e s , p h r a s e - s t r u c t u r e rules have also been replaced by general concepts and principles. This development was initiated in Chomsky ( 1 9 7 0 ) and has been developed f u r t h e r in Jackendoff ( 1 9 7 7 ) and Stowell ( 1 9 8 1 ) .

-1-

Introduction As a consequence of these two changes the explanatory power of the theory has increased considerably. At the same time, however, these changes have led to an increase of the complexity of linguistic research. It is now no longer possible to study a p a r t i c u l a r aspect of a particular language in isolation. We have to take into account the whole theory with all its subtheories and all the intricate interconnections. A minor change in one of the principles of a subtheory might have various far-reaching consequences for other subtheories. Although this can be seen as a positive development that indicates the cohesion of the theory, it makes it rather d i f f i c u l t to evaluate a specific proposal and its consequences for the theory. The main objective of this study is to demonstrate that it is not a defining p r o p e r t y of clauses that they consist of a subject and a p r e d i c a t e . The assumption that clauses must have a subject is implicitly or explicitly present in most, if not all, linguistic theories. In generative grammar this assumption has been incorporated in various ways. Until recently it has generally been accepted that there is a PS-rule of the form S - > - N P - VP. Although several theories have been developed to reduce the amount of stipulation in the base component of the grammar, the rule that introduces the subject of S has been very tenuous. In Chomsky (1982) the rule was replaced by an equally stipulative extension of the Projection P r i n c i p l e . As Chomsky argues "the requirement that a clause have a subject position is independent of the Projection Principle". But since this requirement is "conceptually quite closely related" to the Projection Principle, he introduced the Extended P r oj ect i on Principle, which combines the Projection Principle and the subject-stipulation. Stowell ( 1 9 8 1 , 1 9 8 3 ) argues quite convincingly that the notions 'clause' and ' s u b j e c t ' should be generalized across categories. His solution to the supposedly general condition that S contains a subject is to assume that only constituents which consist of a s u b j e c t - p r e d i c a t e configuration may be i n t e r p r e t e d as a clause at the level of Logical Form. This conclusion is to a large extent similar to that of Williams ( 1 9 8 0 , 1 9 8 1 , 1 9 8 4 ) . Williams argues against the relevance of grammatical relations in the t h e o r y . In his view the ' s u b j e c t ' of S must be present as a consequence of p r e d i c a t i o n . -2-

Introduction In this study a d i f f e r e n t approach to the presence of the subject of S is chosen. I agree with Williams that grammatical relations are r e d u n d a n t in the t h e o r y . In my view the presence of a ' s u b j e c t ' is entirely dependent on 9-theory or Case t h e o r y . What is generally referred to as the subject is either an external argument of the verb or an internal argument which has been moved in order to receive Case. I agree with Stowell that the external argument is a daughter of the maximal projection of its θ-assigning head. In my view the ensuing endocentric s u b j e c t conception does not require an additional predication rule at LF. It is a consequence of this theory that there are subjectless clauses if the verb does not select an external argument and if there is no internal NP-argument that has to be moved in order to receive Case. It will be shown that such subjectless sentences are abundantly present in Dutch in precisely the circumstances indicated above. Some relevant constructions are passive constructions in which verbs occur that do not subcategorize for an internal NP complement, but for a PP-complement, a sentential complement or no complement at all. Other instances can be found with raising and ergative verbs. Crucial in this discussion is the status of the so-called dummy pronouns. If the theory outlined above is correct, we do not expect the occurrence of dummy s u b j e c t s . The appearance of dummy s u b j e c t s is the strongest argument in favour of a subject stipulation in whatever f o r m . I argue that there are no dummy subjects in D u t c h . One of the arguments supporting this claim is that an analysis which incorporates a base-generated subject position has to assume the existence of various types of empty dummy p r o n o u n s . While the concept of empty dummy pronouns is undesirable in general, it is particularly unattractive in a language without pronoun-drop, as is D u t c h . There are two potential lexical dummy pronouns in D u t c h , i.e. het and er. It will be argued that het is not a dummy pronoun but a regular pronoun that needs a θ-role and Case. Er is not a dummy subject e i t h e r , but rather a PP which may have various f u n c t i o n s . If this view is correct, it follows that the central but unattractive stipulation concerning the presence of the s u b j e c t of S may be removed from the t h e o r y .

-3-

Introduction Chapter 1, which presents the results of work done j o i n t l y Teun Hoekstra and myself (Bennis & Hoekstra introduces a theory of empty categories. This elaboration of K a y n e ' s connectedness approach (Kayne conditions are proposed, the Gap Condition and the C o n d i t i o n , which play a central role in this study.

by

1984a,b,1985), theory is an 1 9 8 4 ) . Two new Theta-Referent It is argued

that these conditions enable us to explain the d i s t r i b u t i o n of empty categories, in particular the gaps in P-stranding constructions. In collaboration with the independently motivated rule of S-internal a d j u n c t i o n , these two conditions make precise and correct predictions w i t h respect to the occurrence of parasitic gaps as well. In Chapter 2 it is argued that the Dutch pronoun het is a referential expression in all its occurrences. This implies that het requires a thematic role as well as Case. As indicated above, the relevance of this claim lies in the non-dummy status of het. A variety of arguments supporting this proposal is p r o v i d e d . It is shown that ' d u m m y ' het can be the antecedent of P R O , reflexives and parasitic gaps. The non-dummy status of het is f u r t h e r motivated by an asymmetry in wh-movement from sentential complements. The proposed analysis of het requires an extensive discussion of a variety of constructions in which het appears, such as constructions with raising verbs, ergative verbs and 1 'psychological verbs, small-clause constructions and inversion constructions. Having established that het is not a dummy pronoun, I argue in chapter 3 that the other potential candidate for a dummy-subject s t a t u s , i.e. er, should not be analysed as a dummy subject e i t h e r . The R-pronoun er displays a variety of syntactic f u n c t i o n s . These are discussed in t u r n . It is argued that none of these different f u n c t i o n s j u s t i f i e s an analysis of er as a dummy subject. R-pronouns, including er, are PPs that may be arguments. It is proposed that in those cases in which er is generally taken to be a dummy subject er only has a s e m a n t i c / p r a g m a t i c f u n c t i o n . This analysis of er provides the means to explain the well-known problem of complementizer-trace phenomena in D u t c h . In chapter 4 some of the consequences of the theory introduced in the preceding chapters are investigated. The f u l l y endocentric

-4-

Introduction subject conception has far-reaching consequences for

the

analysis

of languages with an a p p a r e n t N P - I N F L - V P s t r u c t u r e , such as English, F r e n c h , Italian and Spanish. It is suggested that in these languages

the

preverbal

subject

occupies

D i f f e r e n c e s between Old English and Modern

a

derived

English

are

position. shown

to

support this analysis. It is claimed that the postverbal subject in Romance languages in so-called ( s t y l i s t i c ) inversion indicates the u n d e r l y i n g position of the s u b j e c t . W i t h

constructions respect

to

the occurrence of apparent dummy pronouns in other languages it

is

shown that German es and English it may be analysed in a way which is similar to het in Dutch in most respects.

It

will

be

evident

that the proposals in this chapter are r e l a t i v e l y t e n t a t i v e and that f u r t h e r detailed research remains to be done to fathom all the implications for the theory as a whole.

-5-

Chapter 1

Gaps and Parasitic Gaps 1 . 1 Introduction R e c e n t l y , the phenomenon of p a r a s i t i c gaps has become

one

of

the

most w i d e l y discussed topics in g e n e r a t i v e g r a m m a r . T h e i r relevance to linguistic t h e o r y derives on the one hand f r o m t h e i r d i s t r i b u t i o n , as pointed out by Engdahl ( 1 9 8 3 ) , and on hand from the f a c t that t h e i r a p p e a r a n c e is so p e r i p h e r a l is highly u n l i k e l y that new and invoked"

to

determine

their

independent

distribution

principles "or

that

systematic the other that

"it

need rules

p a r t i c u l a r grammars are involved" ( C h o m s k y 1 9 8 2 , 3 9 ) . In Concepts and Consequences ( C h o m s k y 1 9 8 2 ) , it is d e m o n s t r a t e d the d i s t r i b u t i o n of p a r a s i t i c gaps in English can in f a c t be

be of

Some that made

to f o l l o w f r o m i n d e p e n d e n t p r i n c i p l e s of G o v e r n m e n t and Binding T h e o r y , w i t h o u t a n y s t i p u l a t i o n s s p e c i f i c t o p a r a s i t i c gaps. In this c h a p t e r , the d i s t r i b u t i o n of p a r a s i t i c gaps in Dutch is i n v e s t i g a t e d . It will be d e m o n s t r a t e d t h a t this d i s t r i b u t i o n varies q u i t e s u b s t a n t i a l l y from the d i s t r i b u t i o n in E n g l i s h . Given the reasoning a b o v e , t h i s d i f f e r e n c e should f o l l o w from i n d e p e n d e n t l y e x i s t i n g d i f f e r e n c e s between the two languages. As we shall show, t h i s is in fact the case. * As pointed out in the introduction, this chapter contains work written jointly by Teun Hoekstra and myself. It is composed of the article Gaps and Parasitic Gaps (Bennis & Hoekstra 198^b) supplemented by a revised version of the final chapter of The Distribution and Interpretation of Gaps (Bennis & Hoekstra 1984a) and a part of our GLOW-paper A parametrized Gap Condition (Bennis & Hoekstra 1985).

Gaps and P a r a s i t i c Gaps This demonstration requires a discussion of the d i s t r i b u t i o n of gaps in general. In section 2, we provide an account of the d i s t r i b u t i o n of gaps in terms of a general p r i n c i p l e , which we call the Gap C o n d i t i o n . This c o n d i t i o n , which is modeled on K a y n e ' s ( 1 9 8 4 ) Connectedness C o n d i t i o n , replaces the standard ECP of Chomsky ( 1 9 8 1 ) . In section 3, it is demonstrated that the Gap Condition adequately accounts for d i f f e r e n c e s with respect to P-stranding in Dutch and English. Section 4 illustrates the a p p l i c a t i o n of the Gap C o n d i t i o n to parasitic gap constructions in English. These three sections provide the background for the discussion of parasitic gaps in Dutch in section 5. In 5.1 it will be seen how the Gap Condition explains the much more restricted d i s t r i b u t i o n of parasitic gaps in Dutch compared to English. In 5.2 we discuss the surprising phenomenon of gaps which have all the p r o p e r t i e s of parasitic gaps except for the apparent absence of a licensing gap. We shall argue that there is in fact a licensing gap and explain why this particular phenomenon can be found in an 0V language like D u t c h , but not in a VO language like English. The relevant d i f f e r e n c e will be seen to have various r a m i f i c a t i o n s , i n c l u d i n g a difference in transparency of Exceptional Case Marking constructions. 1.2 Connectedness 1.2.1 K a y n e ' s g-projection Much research in generative grammar has been d i r e c t e d towards restricting the class of possible t r a n s f o r m a t i o n a l operations. While these restrictions used to be formulated on the application of t r a n s f o r m a t i o n a l rules themselves, most of these r e s t r i c t i o n s are f o r m u l a t e d w i t h i n recent Government and Binding Theory as well-formedness c o n d i t i o n s on representations. Among these restrictions is the Empty Category P r i n c i p l e ( E C P ) , which limits the d i s t r i b u t i o n of traces l e f t by movement by r e q u i r i n g that the empty category r e s u l t i n g from movement be locally i d e n t i f i e d by a proper g o v e r n o r . The E C P , f o r m u l a t e d in ( 1 ) , d i s t i n g u i s h e s two d i f f e r e n t kinds of proper governors. -8-

Gaps and Parasitic Gaps (1) ECP

An empty category A must be properly governed by either a. a category [+N, +V] (or I N F L ) , or b. a local antecedent The p r i m a r y m o t i v a t i o n for the ECP lies in s u b j e c t / n o n - s u b j e c t asymmetries of the kind i l l u s t r a t e d by the examples in ( 2 ) . The contrast in g r a m m a t i c a l i t y follows from the ECP if it is assumed that the subject is governed by INFL and that INFL in English does not qualify as a proper governor. ( 2 ) a . Who do you think that Bill likes e_ b . * W h o do you think that £ likes Bill (3)

I wonder who e_ likes Bill

The empty category in ( 2 a ) is licensed by the ECP because it is governed by the lexical element like. The empty category in ( 3 ) , although it is not properly governed by a lexical category, is licensed by v i r t u e of being governed by a local a n t e c e d e n t , who in COMP. Many similar examples of s u b j e c t / n o n - s u b j e c t asymmetries, supporting the f o r m u l a t i o n of the ECP in ( 1 ) , can be found in the literature. The ECP thus has two separate ways of licensing an empty category, a local lexical governor a n d / o r a local a n t e c e d e n t . It has f r e q u e n t l y been noted that this is in fact an u n n a t u r a l c o m b i n a t i o n . On the one h a n d , the r e q u i r e m e n t for a trace to have an a n t e c e d e n t follows from its very nature of being a t r a c e . On the other h a n d , Kayne demonstrates that the notion of a local lexical governor makes the wrong predictions in cases like ( H ) , where the empty category has a local lexical governor in both cases, i.e. the preposition of. Hence, the ECP predicts that they should both be grammatical. ( 4 ) a . Which actress do you t h i n k that John likes p i c t u r e s of e_ b . * W h i c h actress do you t h i n k that pictures of e^ turn Bill on -9-

Gaps and P a r a s i t i c Gaps

The standard ECP thus meets w i t h both conceptual and empirical problems. Kayne t h e r e f o r e proposes to replace the ECP in ( 1 ) by a condition which collapses (a) and (b) in (1) in a c o n c e p t u a l l y more a t t r a c t i v e f a s h i o n . This condition is concerned w i t h the relation between the gap and the a n t e c e d e n t , which must c o n s t i t u t e a g-projection as defined in ( 5 ) . (5) Υ is a g - p r o j e c t i o n of X iff a. Y is a projection of X (in the usual sense of X-bar t h e o r y ) or of a g - p r o j e c t i o n of X or b. X is a s t r u c t u r a l governor and Υ immediately dominates W and Z , where Z is a maximal p r o j e c t i o n of a g-projection of X, and where W and Z are in a canonical government c o n f i g u r a t i o n The ECP is thus replaced by the r e q u i r e m e n t that the gap and the antecedent be connected by a g - p r o j e c t i o n . The d e f i n i t i o n of g-projection mentions the notion of canonical government c o n f i g u r a t i o n . This is a linear r e q u i r e m e n t on the relation between maximal projections on the path from the gap to the antecedent and its sister. The order of these is determined by the relative order of the verb and its o b j e c t , i.e. in a VO language like E n g l i s h , the canonical government configuration requires that maximal projections on the path be on a right b r a n c h . The canonical government requirement is not a local r e q u i r e m e n t , like the E C P , but takes the entire s t r u c t u r e between the gap and the antecedent as its scope. It t h e r e f o r e not only explains the contrast i n ( 2 ) , b u t also that i n ( 4 ) . Consider the contrast in ( 4 ) f i r s t . The governor of the empty category in both examples is the preposition of. The PP headed by of is t h e r e f o r e a g - p r o j e c t i o n of the g o v e r n o r , by v i r t u e of being an X - p r o j e c t i o n of it (in the sense of the X-bar t h e o r y ) . Since P is a proper governor in E n g l i s h , the g-projection can in p r i n c i p l e be c o n t i n u e d , if the PP and its sister stand in a canonical government c o n f i g u r a t i o n . This is the case in both ( 4 a ) and ( M b ) .

-10-

Gaps and P a r a s i t i c Gaps T h e r e f o r e , the NP pictures of e_ is a g - p r o j e c t i o n . A g a i n , the g-projection can be continued if the NP and its sister stand in a canonical government c o n f i g u r a t i o n . This is true only in ( 4 a ) , where the NP is on a right b r a n c h . In ( 4 b ) , h o w e v e r , the NP is on a l e f t b r a n c h . T h e r e f o r e , the g-projection stops at the level of N P , and no connection by means of a g-projection can be established between the gap and its a n t e c e d e n t . Let us now turn to the contrast in ( 2 ) . In ( 2 b ) , the empty category is not embedded in a l e f t b r a n c h , but is on a l e f t branch i t s e l f . Again, no g-projection can be built because the governor of the g a p , I N F L , is not a s t r u c t u r a l g o v e r n o r . Although both ( 2 b ) and ( M b ) are t h e r e f o r e u n g r a m m a t i c a l , the reasons are slightly d i f f e r e n t : in ( 2 b ) no g - p r o j e c t i o n can be built because the governor of the gap is not a proper governor, whereas in ( 4 b ) the g-projection stops because INFL and the NP subject do not stand in a canonical government c o n f i g u r a t i o n . The g-projection requirement thus constitutes an improvement on the standard E C P , because it eliminates the conceptual problem encountered by the standard ECP and because it has a wider empirical scope. In section 4 we shall see that its scope is wider still ( c f . Kayne 1 9 8 4 ) .

1 . 2 . 2 Two m o d i f i c a t i o n s of K a y n e ' s g-projection In spite of the u n i f y i n g e f f e c t that the g-projection has in comparison to the standard E C P , it still distinguishes between two d i f f e r e n t notions. Every g-projection consists of two p a r t s which can be distinguished by the d i f f e r e n t conditions that are being imposed on them in the d e f i n i t i o n in ( 5 ) . The f u l l g-projection starts with an X - p r o j e c t i o n , which is the maximal projection of the governor of the gap. The condition t h a t is placed on the gap and its sister is that its governing sister be a lexical c a t e g o r y . No such c o n d i t i o n is imposed on sisters of maximal p r o j e c t i o n s on the p a t h , i . e . on W in the d e f i n i t i o n in ( 5 ) . A l t e r n a t i v e l y , whereas the relation between maximal p r o j e c t i o n s and their sisters is subject to the linear canonical government c o n f i g u r a t i o n , no such l i n e a r i t y is imposed on the relation at the bottom of the -1 1-

Gaps and Parasitic Gaps g - p r o j e c t i o n , i.e. on the relation between the gap and its governor. This d i f f e r e n t i a t i o n lacks empirical support and is therefore conceptually u n d e s i r a b l e . We shall demonstrate that if identical conditions are imposed on all links on the p a t h , the notion of g-projection makes stronger p r e d i c t i o n s than the notion as d e f i n e d in ( 5 ) . These p r e d i c t i o n s appear to be c o r r e c t . Consider f i r s t the fact that the r e q u i r e m e n t of being lexical is not imposed on W in ( 5 ) . This would allow for extraction out of adjuncts, as in ( 6 ) , which would have a structure like ( 7 ) . ( 6 ) * W h o did you go home a f t e r seeing e

(7) Who

go PRO seeing e_ H e r e , the PP headed by after is a g - p r o j e c t i o n of the governor of e_, seeing.Since this PP is on a right b r a n c h , VP should also be a g - p r o j e c t i o n and we would t h e r e f o r e expect the s t r u c t u r e to yield a grammatical r e s u l t . If, h o w e v e r , a maximal p r o j e c t i o n must have a lexical sister and be on a right b r a n c h , the u n g r a m m a t i c a l i t y of (6) would be p r e d i c t e d . The e f f e c t of imposing a linear r e q u i r e m e n t on the r e l a t i o n between the gap and its governor cannot be observed in a head-initial VO language like English or in a h e a d - f i n a l 0V language, but it can be observed in a language with mixed b r a n c h i n g like D u t c h . D u t c h has OV-order. The canonical government

-12-

Gaps and Parasitic Gaps c o n f i g u r a t i o n r e q u i r e s maximal p r o j e c t i o n s to be on a l e f t

branch,

r a t h e r than on a right branch as in E n g l i s h . PPs may generally occur both p r e v e r b a l l y and p o s t v e r b a l l y . A d p o s i t i o n s are usually p r e p o s i t i o n a l , but postpositions

occur as

well.

Thus,

there

are

f o u r theoretical p o s s i b i l i t i e s , which are depicted in ( 8 ) .

(8)a.

V

b.

/ PP

V

Λ

PP

Λ

A

c. V

V

d.

Λ Λ Λ

Λ

V

PP

V

PP

If the canonical government r e q u i r e m e n t is imposed only on maximal p r o j e c t i o n s on the p a t h , we p r e d i c t that ( 8 c ) and ( 8 d ) are u n g r a m m a t i c a l , because the PP is on a right b r a n c h . No d i s t i n c t i o n is made between ( 8 a ) and ( 8 b ) , however. If we strengthen the definition also stand

of g - p r o j e c t i o n such that the gap and its in

a

canonical

government

1 9 8 4 ) , we predict that only ( 8 b ) is

governor

configuration

allowed.

This

must

(of.Koster

prediction

is

c o r r e c t , i . e . s t r a n d i n g of adpositions is only possible with postpositional PPs in preverbal position. P - s t r a n d i n g in Dutch is discussed in much greater detail in section 3- Postpositional PPs in Dutch usually have a so-called [ + R ] - c o m p l e m e n t . As the examples in ( 9 ) - ( 1 2 ) show, e x t r a c t i o n of these complements is possible only from preverbal P P s . (9) a. dat Jan naar het m e i s j e keek t h a t John at the girl looked b.*het m e i s j e dat Jan naar £ keek the girl who John at looked ( 1 0 ) a . dat Jan daar naar keek that John there at looked b. het huis waar Jan _e naar keek the house where John at looked ( 1 1 ) a . dat Jan keek naar het m e i s j e that John looked at the girl b . * h e t m e i s j e dat Jan keek n a a r £ the girl that John looked at

-13-

Gaps and P a r a s i t i c Gaps ( 1 2 ) a . dat Jan een boek h e e f t gekocht daar over that John a book has bought there about b.*het o n d e r w e r p waar Jan een boek h e e f t gekocht e_ over the subject where John a book has bought about It t u r n s o u t , t h e n , that the d e f i n i t i o n of g-projection can be strengthened and u n i f i e d in the f o l l o w i n g w a y . ( 1 3 ) Gap Condition A gap Z in a tree P is linked to its antecedent by a connected subtree of P which c o n s t i t u t e s a g-projection XP is a g-projection of the s t r u c t u r a l governor W of Z iff the head of XP c-governs Z or a g-projection of W X c(anonically)-governs Υ iff X precedes Υ in a VO language and follows Υ in an 0V language The r e q u i r e m e n t that the governor of a gap be a s t r u c t u r a l governor is motivated in Kayne ( 1 9 8 4 , 1 6 7 ) . Whether or not a specific category is a s t r u c t u r a l governor appears to some extent to be a language p a r t i c u l a r m a t t e r , e.g. English but not French prepositions q u a l i f y as s t r u c t u r a l governors. From the above discussion relating to the island character of a d j u n c t s , it will be clear that our notion of government as used in ( 1 3 ) d i f f e r s from the generally adopted d e f i n i t i o n in Aoun & Sportiche ( 1 9 8 3 ) . According to their proposal a lexical category governs all phrases within its maximal p r o j e c t i o n . We assume that government is restricted to the argument p r o j e c t i o n , as in Chomsky ( 1 9 7 9 ) . T h e r e f o r e , an a d j u n c t phrase is not g o v e r n e d , although it is in the maximal p r o j e c t i o n . Government is not a relational notion i t s e l f , but r a t h e r a domain r e s t r i c t i o n w i t h i n which certain relations h o l d . Among these relations are Case A s s i g n m e n t , Theta-role Assignment to i n t e r n a l arguments, and the UCC ( U n l i k e Category C o n d i t i o n ) , which is discussed in the next section.

Gaps and P a r a s i t i c Gaps

1 . 2 . 3 E x t r a c t i o n from complement clauses One reason to d i s t i n g u i s h (as K a y n e ' s d e f i n i t i o n in (5) d o e s ) between the X - p r o j e c t i o n part of a g-projection and the higher levels of the g - p r o j e c t i o n derives from the d i f f e r e n t

behavior

English of empty categories constituting a l e f t

branch

and empty categories embedded in a l e f t b r a n c h .

This

in

themselves differential

b e h a v i o r can be i l l u s t r a t e d by the examples in ( 1 4 ) . In ( I 4 a ) ,

the

subject of a clausal complement is extracted, whereas a part of the subject i s extracted i n ( I 4 b ) . ( I 4 ) a . Which r u n n e r do you believe e_ to have won the race b . * W h i c h book do you believe the f i r s t chapter of e_ to be f u l l of lies A c c o r d i n g to Kayne ( 1 9 8 4 , 1 6 9 ) , the empty category in ( I 4 a ) is licit since a g-projection of the verb believe contains the a n t e c e d e n t . This is so, since there is no r e q u i r e m e n t which i m m e d i a t e l y dominates both the

that

governor

there and

the

be

a

gap.

node The

immediate dominance r e q u i r e m e n t on higher levels of the g-projection explains t h e u n g r a m m a t i c a l i t y o f ( I 4 b ) , i.e. the s u b j e c t NP (the first chapter of e_) and believe do not stand in a canonical government c o n f i g u r a t i o n , because there is no category which immediately dominates b o t h . Our proposal makes the same d i s t i n c t i o n , h o w e v e r , without having recourse to a d i s t i n c t i o n between X - p r o j e c t i o n and g - p r o j e c t i o n . Consider the relevant of the tree s t r u c t u r e s of ( 1 4 ) given in ( 1 5 ) .

-15-

parts

Gaps and Parasitic Gaps

VP

I

believe

b.

NP

AVP

VP

believe NP

N

chapter

VP

PP

A

P

of

NP

e

The VP in both ( 1 5 a ) and ( 1 5 b ) is a g-projection of the governor of the gap (believe in ( 1 5 a ) and of in ( 1 5 b ) ) . For our p u r p o s e , the structure above this VP is i r r e l e v a n t . In ( 1 5 b ) , the PP and the NP are also g-projections. The distinguishing p r o p e r t y resides in S. In n e i t h e r c o n s t r u c t i o n is S a g - p r o j e c t i o n , since the head of S ( I N F L ) does not c-govern a g - p r o j e c t i o n of the governor of the gap. In the case of ( 1 5 b ) , this means that the maximal g-projection ( i . e . the S' which contains the a n t e c e d e n t ) does not constitute a connected subtree, or put d i f f e r e n t l y , the subtree from the governor of the gap to the antecedent is not a g-projection. In the case of ( 1 5 a ) , the fact that S is not a g-projection is irrelevant since it is not part of the relevant s u b t r e e , which starts with believe, the governor of the gap. It turns out that our Gap Condition in ( 1 3 ) handles extractions out of complement clauses ( i n c l u d i n g small c l a u s e s ) without having recourse to a separate notion of X - p r o j e c t i o n . A problem for K a y n e ' s approach as well as for ours arises in the case of W H - e x t r a c t i o n s out of sentential complement clauses in Dutch. Dutch is an 0V language. As a r e s u l t , the canonical government c o n f i g u r a t i o n is for the governor to occur on the right of the governee. H o w e v e r , although NP and AP complements have to precede the verb in D u t c h , f u l l sentential c o m p l e m e n t s , f i n i t e and

-16-

Gaps and P a r a s i t i c Gaps non-finite, have to follow the v e r b . T h e r e f o r e , we expect extraction out of sentential complements to be impossible in D u t c h , w h i c h , however, it is n o t . Consider ( 1 6 ) and its tree structure r e p r e s e n t a t i o n in ( 1 7 ) . We shall assume that S is the maximal projection of V and that S is governed from the right by INFL, which we take to be the head of S 1 (see Hoekstra 1984 for discussion). ( 1 6 ) Wat dacht je dat Piet t_ zag? What thought you that Peter saw?

(17)

S'

COMP wat

INFL' S(=VmaX)

Λ

NP

INFL

V

dacht INFL

Piet zag

In accordance with canonical g o v e r n m e n t , we can build a g-projection up to the embedded S ' : the V zag governs the gap to the l e f t , the V m a x is also governed from the l e f t by INFL. Hence S ' , being the maximal p r o j e c t i o n of I N F L , is a g-projection of V. The m a t r i x v m a x is not a g - p r o j e c t i o n of the embedded V, since dacht governs the S' in the wrong d i r e c t i o n . We shall assume that movement from the extraction site to the

-17-

Gaps and Parasitic Gaps matrix COMP does not take place in a single s t e p , but rather by successive movement via the embedded COMP. Since the embedded S' is a g-projection of the embedded V, movement to the embedded COMP is allowed. The second extraction leaves an empty category b e h i n d . This empty category in COMP is governed by the verb dacht, but not canonically. Essentially following a suggestion made by Van R i e m s d i j k ( p . c . ) , we shall assume that licensing requirements for categories in A ' - p o s i t i o n s are less strict than for categories in Α-positions. In this case, this means that we shall assume that the directionality of canonical government does not apply to the empty category in COMP. T h e r e f o r e , the matrix S' constitutes a g-projection of the governor of this empty category in COMP. Thus, the path from the empty category in Α-position to its u l t i m a t e antecedent in the matrix COMP consists of two separate subtrees, each of which constitutes a well-formed g - p r o j e c t i o n . This analysis makes an interesting p r e d i c t i o n . It not only explains why extraction from embedded sentential complements is possible in D u t c h , in spite of the apparent violation of the canonical government r e q u i r e m e n t , but it explains at the same time why the WH-island condition is so s t r i c t l y obeyed in D u t c h , much more so than e.g. in English. The reason is that e x t r a c t i o n in Dutch necessarily involves an intermediate step through COMP, whereas in English a g-projection can be built d i r e c t l y , i.e. without an intermediate step through COMP, given the f a c t that the embedded clause is canonically governed by the matrix v e r b . In section 5.1. we shall discuss a f u r t h e r interesting corollary of our analysis of extractions out of Dutch sentential complements.

1.3 Complex cases of PPs In the previous section it was shown how P-stranding in Dutch can be explained in terms of the Gap C o n d i t i o n . In this section we shall present a more elaborate discussion of the problems concerning P-stranding in general, and in Dutch in p a r t i c u l a r . The phenomenon of P-stranding is discussed at some length in Van R i e m s d i j k ( 1 9 7 8 a ) . The account given by Van R i e m s d i j k focuses

-18-

Gaps and Parasitic Gaps on the internal s t r u c t u r e of PP. In 3·2 we shall discuss those aspects of the syntactic context of the PP which are relevant to s t r a n d i n g , but f i r s t , in 3 - 1 > we shall go into Van R i e m s d i j k ' s a c c o u n t . The discussion in this section will prove to be relevant to the distribution of parasitic gaps in D u t c h , which will be discussed in section 5. 1 . 3 - 1 Van R i e m s d i j k ' s analysis of P - s t r a n d i n g D u t c h is mainly prepositional. As is indicated in ( I 8 b ) , extraction of the NP complement of a preposition is not allowed.

( I 8 ) a . Jan heeft op Marie/*Marie op gerekend John has on M a r y / Mary on counted b.*Wie heb je op t^ gerekend Who have you on counted If the complement of a p r e p o s i t i o n is a [-human] pronoun, it obligatorily appears in the so-called R - f o r m and precedes the P of which it is a c o m p l e m e n t . The notion R - f o r m r e f e r s to a set of pronominal elements that characteristically have a phoneme Irl which is not found with other pronouns. The phenomenon is illustrated by the examples in ( 1 9 ) . (19)a.*Jan heeft John has b.*Jan h e e f t John has c. Jan h e e f t John has d.*Jan heeft John has

op het on it op er on t h e r e er op there on het op it on

gerekend counted gerekend counted gerekend counted gerekend counted

In ( 1 9 c ) , the P is a p o s t p o s i t i o n and the R - p r o n o u n e x t r a c t e d , either by WH-movement or by regular R - m o v e m e n t : ( 2 0 ) a . Waar h e e f t Jan t_ op gerekend Where has John on c o u n t e d

-19-

may be

Gaps and Parasitic Gaps b. Jan heeft er vandaag niet t_ op gerekend John has there today not on counted The explanation of this set of facts given in Van R i e m s d i j k (1978a) can be summarized as follows. The impossibility of O8b) is explained by postulating the Head Constraint, which states that no material may be extracted from the domain of the h e a d . In order to account for the obligatory s h i f t of [-human] pronouns to R - f o r m s , Van R i e m s d i j k invokes an R-suppletion rule which changes the pronoun het in ( 1 9 a ) into er, as in ( 1 9 b ) . Then there is a f i l t e r that excludes an R-pronoun in the complement position to the right of the p r e p o s i t i o n . A rule of R-movement has the e f f e c t of relocating the R-pronoun to the l e f t of the preposition and outside of the P'-domain in order to escape the e f f e c t s of the Head Constraint. This movement is depicted in ( 2 1 ) . Apart from the Head Constraint, Van R i e m s d i j k also assumes that extraction from PP is restricted by s u b j a c e n c y , PP counting as a bounding node. Given the binding nature of PP, WH-extraction in ( 2 0 a ) cannot have taken place in a single step. T h e r e f o r e , the R-pronoun is f i r s t moved outside the PP to the position taken by er in ( 2 0 b ) , thus crossing only a single bounding category, and then moved to COMP, again crossing only a single bounding node ( S ) . (21)

COMP

waar NP [+R]

-20-

Gaps and P a r a s i t i c Gaps Although the R - p r o n o u n s in Van R i e m s d i j k ' s analysis o r i g i n a t e as N P s , it is i m p o r t a n t to note that an R-pronoun can only bind empty [ R ] - p o s i t i o n s , i.e. empty positions inside PP. This will become clear in our discussion of parasitic gaps in section 5. This is one of the reasons to assume that R-pronouns do not result from s u p p l e t i o n , but are base-generated in f r o n t of their P h e a d . The d i f f e r e n t o r i e n t a t i o n of R-complements and non-R-complements w i t h respect to P can be derived from Case Theory on the assumption that P assigns Case to the right and that R-forms inherently resist Case.Cl]. In the previous section, we demonstrated how the d i f f e r e n c e in extraction possibilities between p r e p o s i t i o n a l and postpositional PPs is a direct consequence of the requirement of canonical government ( c f . ( 8 ) - ( 1 2 ) ) . Although V a n R i e m s d i j k ' s account i s also capable of making this d i s t i n c t i o n , his analysis has nothing to say about the d i f f e r e n c e between postverbal and p r e v e r b a l P P s , i.e. his analysis does not p r e c l u d e extraction of R-pronouns from postverbal PPs. In this sense, t h e n , Van R i e m s d i j k ' s account relates the possibility of extraction entirely to PP internal aspects.

1 . 3 - 2 A d j a c e n c y a n d P-stranding It has previously been observed (see H o r n s t e i n & Weinberg 1 9 Ö 1 ) that the p o s s i b i l i t i e s of extraction from PP seem to depend in large measure on the context in which the PP appears. H o r n s t e i n & Weinberg p o i n t out that preposition stranding is possible if the P is reanalyzed to become part of a complex verb. This reanalysis rule applies only to material c o n t a i n e d in the domain of the verb. In this w a y , they can make a d i s t i n c t i o n between S-PPs and V P - P P s , only the l a t t e r of w h i c h allow P-stranding. In our terms, this d i s t i n c t i o n is an a u t o m a t i c consequence of the r e q u i r e m e n t that maximal p r o j e c t i o n s on the p a t h from the governor of the gap to the a n t e c e d e n t be canonically governed ( c f . ( 6 ) ) . There is t h e r e f o r e no need for a reanalysis r u l e . F u r t h e r m o r e , we shall follow Kayne ( 1 9 8 4 , ch. 5) in assuming that the d i f f e r e n c e between French and English w i t h respect to P-stranding is a consequence of the fact t h a t P is a s t r u c t u r a l governor in E n g l i s h , but not in F r e n c h .

-21-

Gaps and Parasitic Gaps Hornstein & Weinberg argue that the d i f f e r e n c e between S-PPs and VP-PPs is unexpected under Van R i e m s d i j k ' s approach. Now we want to draw a t t e n t i o n to a similar kind of p r o b l e m . In D u t c h , the position of a stranded preposition d i f f e r s quite strikingly from the possible positions of the corresponding f u l l PPs. This is illustrated in ( 2 2 ) . ( 2 2 ) a . dat ik over dat onderwerp met Piet sprak that I about that subject with Peter talked b. dat ik met Piet over dat onderwerp sprak that I with Peter about that subject talked c.*het onderwerp waar ik over met Piet sprak the subject where I about with Peter talked d. het onderwerp waar ik met Piet over sprak the subject where I with Peter about talked The stranded preposition over in ( 2 2 c ) cannot occupy the same relative position as the corresponding f u l l PP in ( 2 2 a ) , but has to occupy the position to the immediate l e f t of the v e r b . This descriptive generalization holds throughout: the stranded preposition must be left adjacent to the verb.[2]. This generalization follows from our Gap C o n d i t i o n , i.e. from the requirement that the PP itself is canonically governed and that the P must also canonically govern its complement. Let us consider the structure of ( 2 2 c ) . We shall assume, following Kayne ( 1 9 8 4 ) , that syntactic s t r u c t u r e is binary branching. Given the linear order of the constituents in ( 2 2 c ) , this means that the syntactic s t r u c t u r e of the verb p r o j e c t i o n of ( 2 2 c ) must contain something like ( 2 3 ) .

(23)

over

P

NP

sprak

met Piet

-22-

Gaps and P a r a s i t i c Gaps In this s t r u c t u r e

PPp

is

strictly

governed

(in

the

sense

of

m i n i m a l l y c-commanded) b y t h e lexical v e r b , unlike P P · « - T h e r e f o r e , if it is assumed that canonical government requires strict g o v e r n m e n t , the u n g r a m m a t i c a l i t y of ( 2 2 c ) is an automatic consequence. There appears to be some v a r i a t i o n between Dutch as to what counts as the government domain w i t h i n

and

which

government must h o l d . Consider ( 2 4 ) , which is the m i r r o r

English canonical image

of

( 2 2 c ) in the relevant r e s p e c t s . ( 2 4 ) Which subject did you talk with Peter about e_ H e r e , the about-PP, although belonging to the V - d o m a i n , s t r i c t l y governed by the v e r b , given the binary r e q u i r e m e n t . We have no e x p l a n a t i o n to

offer

for

this

cannot be branching variation

between Dutch and English. The r e q u i r e m e n t that a PP containing a s t r a n d e d preposition must be canonically governed predicts that no e x t r a c t i o n is possible from a PP in the complement of an N P , since PP complements to nouns follow the noun in D u t c h . This is i l l u s t r a t e d by the following examples. ( 2 5 ) a . * D e Stelling waar m i j n argument [_t tegen] op bezwaren stuitte The proposition where my argument against with o b j e c t i o n s met b . * H e t o n d e r w e r p waar ik een lezing [_t over] op die vergadering heb gehouden The s u b j e c t where I a. l e c t u r e about at that m e e t i n g have given c . * D i e a f f a i r e waar ik dat stuk [_t over] gisteren heb geschreven That m a t t e r where I that piece about yesterday have w r i t t e n The only NPs from which it seems possible to extract an R-pronoun from a PP complement are d i r e c t o b j e c t NPs which are adjacent to the v e r b .

-23-

Gaps and Parasitic Gaps ( 2 6 ) a . het onderwerp waar ik op die vergadering een lezing [t_ over] heb gehouden the subject where I on that meeting a lecture about have given b. die a f f a i r e waar ik gisteren dat stuk [J; over] geschreven heb that matter where I yesterday that piece about written have These facts are reminiscent of the discussion concerning PP extraction from NP ( c f . Bach & Horn 1 9 7 6 , Chomsky 1 9 7 7 ) . Chomsky ( 1 9 7 7 ) argues that extraction from NP as in ( 2 7 ) should be excluded, given subjacency with NP and S as bounding nodes, but proposes that the PP is reanalyzed as a separate constituent prior to movement into COMP. (27)

About which book did you w r i t e a review

We could adopt a similar line of reasoning and assume that in cases like ( 2 6 ) , where it seems that e x t r a c t i o n takes place from inside an N P , the PP is restructured out of NP f i r s t , a process which is needed independently to account for the examples in ( 2 8 ) . ( 2 8 ) a . dat ik een lezing op die vergadering over dat onderwerp heb gehouden that I a lecture on that meeting on that s u b j e c t have given b. dat ik over dat onderwerp op die vergadering een lezing heb gehouden that I on that subject on that meeting a lecture have given c. dat ik op die vergadering een lezing heb gehouden over dat onderwerp that I on that meeting a lecture have given on that subject A f t e r r e s t r u c t u r i n g , the PP is s t r i c t l y governed by the verb and hence a g - p r o j e c t i o n can be b u i l t if an R-pronoun is e x t r a c t e d .

-24-

Gaps and Parasitic Gaps

1.3.3 Extraction from complex PPs We now t u r n to more complex cases, where it will be demonstrated that our Gap Condition makes correct p r e d i c t i o n s . These cases involve adpositions which take a PP complement. Since a P may occur either as a preposition or postposition, we can distinguish between eight d i f f e r e n t situations involving a P w i t h a PP complement from which the NP complement is extracted. Four of these involve postverbal PPs. As we explained e a r l i e r , these constructions are ungrammatical because the matrix PP is not canonically governed. The four preverbal cases are given in ( 2 9 ) . (29)a.

V PP

b. PP

V

/\ Λ

A Λ

PP

P

Ρ

£

c.

£

V

/\V

A Λ

P

Ρ

d.

PP

PP

PP

P

V PP

A Λ PP

P

e_

e_

These four d i f f e r e n t combinations examples of ( 3 0 ) - ( 3 3 ) .

are

V

P

P

instantiated

by

( 3 0 ) a . dat hij de koekjes [voor [bij de k o f f i e ] ] koopt that he the cookies for with the c o f f e e buys b.*Ik vroeg wat hij de k o e k j e s voor bij t^ kocht I asked what he the cookies for with bought ( 3 1 ) a . dat hij de koekjes [voor [daar b i j ] ] koopt that he the cookies for there w i t h buys -25-

the

(b)

Gaps and Parasitic Gaps b . * I k vroeg waar hij de k o e k j e s voor t_ bij kocht I asked where he the cookies for with bought ( 3 2 ) a . Hij is net [[onder net hek] door] gekropen He is just under the fence through crawled b . * I k vroeg welk hek hij net onder _t door was gekropen I asked which fence he just under through was crawled ( 3 3 ) a - Hij is net [[daar onder] door] gekropen He is j u s t there u n d e r through crawled b. Ik vroeg waar hij net t_ onder door was gekropen I asked where he just under through was crawled Two factors are relevant in determining the possibility of extracting the complement of the embedded P: a. Is this complement canonically governed by P? b. Is the embedded PP canonically governed by the matrix P? If we inspect the tree representations in ( 2 9 ) , we see that in ( 2 9 a ) both the empty category and the embedded PP are not canonically governed; in ( 2 9 b ) the empty category is canonically governed, but the embedded PP is n o t ; in ( 2 9 c ) , the empty category is not canonically governed, although the embedded PP is; finally, both the empty category and the embedded PP are canonically governed in ( 2 9 d ) . The grammaticality of ( 3 3 b ) is therefore predicted by the Gap C o n d i t i o n , just like the u n g r a m m a t i c a l i t y of the other (b) examples. T h u s , we succeed in p r e d i c t i n g that only one out of eight possibilities yields a grammatical result w i t h o u t any statements specific to these constructions. The next question concerns the p o s s i b i l i t y of extracting the PP complement of an adposition i t s e l f . In this case, there are four situations, two with the PP in postverbal p o s i t i o n , which again we shall not discuss, and two with a preverbal PP. These two situations are depicted in ( 3 4 ) .

-26-

Gaps and Parasitic Gaps (34)a.

V

b.

/\V

/\V

PP

P

V PP

Λ

Λ

PP

PP

Ρ

These structures are i n s t a n t i a t e d by the (b) examples in ( 3 5 ) - ( 3 6 ) . It is predicted that only the s t r u c t u r e in ( 3 ^ b ) yields a grammatical outcome. In ( 3 4 a ) , the embedded PP is not canonically governed. The f a c t s are as p r e d i c t e d . ( 3 5 ) a . dat h i j de soep [voor [bij net d i n e r ] ] kookt that he the soup for w i t h the dinner boils b.*Ik vroeg [bij welke m a a l t i j d ] hij de soep voor _t kookte I asked with which meal he the soup for boiled ( 3 6 ) a . dat hij [[achter dat g o r d i j n ] vandaan] is gekomen that he behind that c u r t a i n from is come b. Ik vroeg [achter welk g o r d i j n ] hij _t_ vandaan is gekomen I asked behind which curtain he from is come It should

be

noted

that

Van R i e m s d i j k ' s

analysis

cannot

account for the contrast between ( 3 1 ) and ( 3 3 ) · Van R i e m s d i j k suggests that extraction from PPs is made possible via a base generated [+R]-position outside the head d o m a i n , i.e. on the P" level (see section 3 . 1 ) . Nothing would prevent the e x t r a c t i o n of the R-p'ronoun in ( 3 1 b ) in two successive steps as depicted in ( 3 7 ) .

-27-

Gaps and Parasitic Gaps

(37)

[+R]

P1

Van R i e m s d i j k appeals to the n o t i o n of bridge to account for the ungrammaticality of ( 3 1 b ) : the relevant prepositions would not q u a l i f y as suitable bridges, u n l i k e elements like vandaan and door, as in ( 3 3 ) and ( 3 6 ) , which he analyzes as postpositions ( c f . Van R i e m s d i j k 1978a , 299f f ) . If an appeal to the notion of bridge were on the right t r a c k , we would not expect the consistent preposition-postposition asymmetry that we f i n d in ( 3 1 ) - ( 3 3 ) and (35)-(36). Turning to comparable structures in E n g l i s h , we would expect extraction from a PP which is embedded in a PP to be possible, since this would constitute the exact m i r r o r image of ( 2 9 d ) . This expectation is borne out by the f o l l o w i n g examples. (38)a. b. (39)a. b.

Fido jumped from u n d e r the table Which table did Fido jump from under t_ Fido ran out into the meadow Which meadow did Fido run out into t

U n e x p e c t e d , h o w e v e r , is the fact that the embedded PP itself cannot

-28-

Gaps and P a r a s i t i c Gaps be e x t r a c t e d , u n l i k e the m i r r o r image c o u n t e r p a r t in Dutch in ( 3 6 ) . ( 4 0 ) a . * U n d e r which table did Fido jump from _t b . * I n t o which meadow did Fido run out t^ Sag ( 1 9 8 2 ) discusses these examples and argues for

an

in terms of a GPSG version of the A - o v e r - A p r i n c i p l e .

explanation Whereas

the

u n g r a m m a t i c a l i t y of ( 4 0 ) is unexpected under our a p p r o a c h , the grammaticality of ( 3 6 b ) c o n s t i t u t e s an anomaly for Sag's A-over-A account. We shall now demonstrate that the ungrammaticality of ( 4 0 ) is related to other phenomena, which are i n d e p e n d e n t of our Gap Condition.

1 . 3 - 4 R e a n a l y s i s in complex PPs Although the examples discussed in the p r e v i o u s section involve PPs in the complement of P , this c o n f i g u r a t i o n is usually not allowed. In general, P may not govern PP. This is in fact an instance of a much wider r e s t r i c t i o n , which states that no category may govern an element of the same category. So, N may not govern an N P , A may not take AP complements, V may not take V p r o j e c t i o n s as complements, and s i m i l a r l y , P cannot take PP complements. This r e s t r i c t i o n , called the U n l i k e Category Condition ( U C C ) , is amply discussed and illustrated

in

Hoekstra

(1984,

ch.

2).

To

illustrate

the

r e s t r i c t i o n on Ps, we can give examples like those in ( 4 1 ) . ( 4 l ) a . * W e s p r a k e n over in het huis We spoke about in the house b . * I k v e r t r o u w op onder het bed als een goede schuilplaats I t r u s t on u n d e r the bed as a good h i d i n g place c . * V o o r het opeten van de t a a r t r e k e n d e n we op t i j d e n s de l u n c h p a u z e For the eating of the cake counted we on during the lunch break The c o u n t e r e x a m p l e s to this r e s t r i c t i o n in the case of Ps concern only a few p r e p o s i t i o n s , which are s e m a n t i c a l l y similar for Dutch

-29-

Gaps and Parasitic Gaps and English. Some examples are given in ( 4 2 ) . ( 4 2 ) van voor de oorlog voor na the m a a l t i j d tot na die datum

from b e f o r e the war for a f t e r the meal until a f t e r that date

The main reason to consider these examples exceptional is that the configuration is subject to severe restrictions of a lexical n a t u r e . In order to m a i n t a i n the UCC as a general p r i n c i p l e , we shall assume that the constructions in ( 4 2 ) escape the e f f e c t of the UCC by being reanalyzed into a prepositional c o m p l e x . This situation closely resembles the case of V-raising complement structures in D u t c h , both in its m o t i v a t i o n and in its e f f e c t . In D u t c h , sentential complements do not occur in preverbal position, unlike APs and N P s , whereas PPs may in principle occur on either side of the v e r b . The reason that APs and NPs must occur preverbally can be accounted for by assuming that Case Assignment is directional.[3] · The fact that P P s , which do not require Case, may occur both preverbally and postverbally suggests that Theta-role Assignment is not d i r e c t i o n a l . The obligatory postverbal occurrence of sentential complements can be derived from the U C C , if we assume that the UCC is directional, just like Case Assignment, its direction being determined by the direction of Case Assignment. A sentential complement in preverbal position will be governed by V from the r i g h t , which is also the direction of Case Assignment. T h e r e f o r e , if a sentential complement is non-distinct from V in its f e a t u r e c o n t e n t , the preverbal occurrence is f o r b i d d e n by the U C C . Let us now t u r n to V-raising s t r u c t u r e s . These involve a verbal projection to the l e f t of the m a t r i x v e r b . Consider the s t r u c t u r e in ( 4 3 ) . This s t r u c t u r e is rejected by the U C C . T h e r e f o r e , the complement either has to occur in postverbal position, as in ( 4 4 a ) or V-raising has to a p p l y , giving rise to ( 4 4 b ) ( f o r discussion, see Evers 1 9 7 5 ) . (43)

*dat Jan [een a r t i k e l te s c h r i j v e n ] beloofde that John an article to w r i t e promised

-30-

Gaps and Parasitic Gaps ( 4 4 ) a . dat Jan beloofde [een a r t i k e l te s c h r i j v e n ] b. dat Jan [een a r t i k e l ] beloofde [te s c h r i j v e n ] The rule of V-raising, by which ( 4 4 b ) is d e r i v e d , reanalysis which applies to a s t r u c t u r e like ( 4 5 ) .

is

a

rule

of

(45)

The e f f e c t of this reanalysis, which is again severely restricted by lexical f a c t o r s , is that the resulting s t r u c t u r e no longer violates the U C C . In these respects, t h e n , the reanalysis operation of V-raising is identical to the reanalysis that we wish to invoke to account for the exceptions to the UCC in ( 4 2 ) . One might object that there is a clear d i f f e r e n c e between V-reanalysis (V-raising) and the proposed P-reanalysis, since V-raising has a linear e f f e c t of r e o r d e r i n g the matrix and the embedded v e r b . However, it has been established that V-raising applies in German without a similar linear e f f e c t . R e c e n t l y , Huybregts ( 1 9 8 3 ) and Haegeman & Van R i e m s d i j k ( 1 9 8 4 ) have argued that the process of V-raising must be broken up into two steps, one involving the actual reanalysis, which now looks even more similar to our P-reanalysis, and a second operation which realizes the linear shift in D u t c h . In f a c t , from this perspective the reanalysis operation does not d i f f e r in any important respect from the reanalysis rule proposed by Rizzi ( 1 9 8 2 , ch. 1) for I t a l i a n . We shall f o r m a l i z e the reanalysis operation by assigning a c o s u p e r s c r i p t to the head of the complement and its governor, in order to c a p t u r e the fact that no actual r e s t r u c t u r i n g takes place. This is evident from the fact that both heads can be separated by i n t e r v e n i n g m a t e r i a l , as is illustrated in ( 4 6 ) . The reanalysis operation in the case of constructions like those in

-31-

Gaps and Parasitic Gaps ( 4 2 ) is illustrated in ( 4 7 ) . ( 4 6 ) a . van vlak voor de oorlog from right before the war b. van daar voor from there b e f o r e c. van vlak daar voor from right there before

(47)

PP pi

pp

PJ

XP

The XP in ( 4 7 ) may be an NP as in the examples given in also an S ' . This is illustrated in ( 4 8 ) .

(42),

but

( 4 8 ) a . Dit model dateert nog van voor dat ik geboren werd This model dates still from before that I born was b. Wacht maar tot na dat we gegeten hebben Wait just until a f t e r that we eaten have c. We bewaren de bonbons voor tot na dat we k o f f i e gedronken hebben We keep the chocolates for u n t i l a f t e r that we c o f f e e drunk have This is in itself not s u r p r i s i n g , but what is unexpected is the fact that an i n f i n i t i v a l complement clause may not replace the f i n i t e complement clause. This is shown by the examples in ( 4 9 ) . ( 4 9 ) a . * W i j wilden wachten tot na gegeten te hebben We wanted wait u n t i l a f t e r eaten to have b . * H i j kan zieh dat nog h e r i n n e r e n van voor te z i j n geboren He can himself that still remember f r o m before to be born c.*Dit is een lotion voor na te hebben gedoucht This is a lotion for a f t e r to have showered

-32-

Gaps and P a r a s i t i c Gaps

C l e a r l y , the reason for

the

ungrammaticality

cannot be some semantic v i o l a t i o n , since it is m e a n t . Nor can an appeal be made to a the control of P R O , especially not in

the

of

these

examples

is e n t i r e l y clear

locality case

what

requirement of

(^9c),

on

which

would be an instance of P R O a r b . Before this p u z z l e can be solved, it one

further

observation,

which

also

is relevant to shows

introduce

an

asymmetric

d i s t r i b u t i o n of f i n i t e and n o n - f i n i t e clauses in the complement of p r e p o s i t i o n s , in spite of the fact that f i n i t e and n o n - f i n i t e clauses in general have the same d i s t r i b u t i o n

in

Dutch,

they have in English ( b a r r i n g lexical i d i o s y n c r a s i e s ) . clauses are found as complements to all m a j o r examples in ( 5 0 ) i l l u s t r a t e . It is a s t r i k i n g

just

as

Infinitival

categories, fact t h a t ,

as the whereas

with other category types i n f i n i t i v a l complements may optionally be i n t r o d u c e d by om (again b a r r i n g lexical i d i o s y n c r a s i e s ) , prepositional complementizer om is obligatorily absent in

the the

complement of P , whereas the f i n i t e complementizer dat is obligatorily present ( i . e . there is no rule comparable to English that-deletion in D u t c h , not with any c a t e g o r y ) . [ 4 ] . This is shown in ( 5 1 ) .

( 5 0 ) a . Hij dwong m i j ( o m ) weg te gaan He f o r c e d me ( f o r ) away to go b. De b e l o f t e ( o m ) die avond aanwezig te z i j n The promise ( f o r ) that night present to be c. Het is leuk ( o m ) hem weer te ontmoeten It is nice ( f o r ) him again to meet ( 5 1 ) a . Zonder [dat hij net boek gelezen h a d ] W i t h o u t that he the book read had b. Na [dat ik hem ontmoet h a d ] A f t e r that I him met had] c. Zonder ( * o m ) het boek gelezen te hebben W i t h o u t ( f o r ) the book read to have d. Na ( * o m ) hem ontmoet te hebben A f t e r ( f o r ) h i m m e t t o have

-33-

Gaps and Parasitic Gaps I n t u i t i v e l y , the two observations in

( 4 9 ) and

r e l a t e d : linearly speaking an i n f i n i t i v a l

( 5 1 ) are

clearly

clause is preceded by two

p r e p o s i t i o n s . In ( 4 9 ) , there is a combination of p r e p o s i t i o n s which normally allow reanalysis in order to avoid a violation of the U C C , whereas in ( 5 1 ) the i n f i n i t i v a l complement of a p r e p o s i t i o n may not be introduced

by

a

prepositional

complementizer.

Let

us

take

seriously the notion of p r e p o s i t i o n a l c o m p l e m e n t i z e r and assume om, which introduces i n f i n i t i v a l clauses to be a p r e p o s i t i o n the embedded COMP position. This assumption is itself other consequences. For i n s t a n c e , it be

excluded

from

the

subject

occupying

without

any

does not imply that PRO should

position

of

such

infinitival

c o m p l e m e n t s , since COMP is the s p e c i f i e r of the INFL p r o j e c t i o n . T h e r e f o r e , no government relation exists between P in COMP and the subject position of S, which is v max in Dutch. We shall follow Stowell ( 1 9 8 1 ) in assuming that in E n g l i s h , COMP is the head of S 1 , rather

than

INFL.

This

assumption

allows

us

to

d i f f e r e n c e between D u t c h and English in this r e s p e c t .

explain If

COMP

the is

the head of S' in English, for in COMP governs the subject position of S and assigns

Case

to

it

under

government.

Therefore,

subject of i n f i n i t i v a l clauses i n t r o d u c e d by for must

be

Conf igurationally , there is no d i f f e r e n c e between the

relation

the

lexical. of

COMP and the subject position on the one hand and the verb consider and the subject of its

small clause complement AP , as in ( 5 2 b ) .

( 5 2 ) a . for [John to leave] b. consider [John f o o l i s h ] In both cases, the relation is as in ( 5 3 ) . We shall assume that at least V and P in the p o s i t i o n of X govern the position W in ( 5 3 ) .

(53) „max

-34-

Gaps and P a r a s i t i c Gaps The assumption that om is a p r e p o s i t i o n in COMP p o s i t i o n allows us to explain the fact that i n f i n i t i v a l clauses i n t r o d u c e d by om do not occur in the complement of a p r e p o s i t i o n . The s t r u c t u r e of an example such as ( 5 1 c ) , with om p r e s e n t , would look like ( 5 4 ) .

(54) max

zonder

NP

V

The c o n f i g u r a t i o n a l relation between zonder and om is identical to the relation between X and W in ( 5 3 ) , i.e. zonder governs om, which is not allowed by the U C C . Therefore, the sentence is ungrammatical. It should incidentally be noted that om does not stand in a government relation to the NP subject of V m a x , as we stated above. We note h e r e t h a t , even if the combination of zonder and om f u l f i l l e d the lexical r e q u i r e m e n t s for reanalysis (which it happens not t o ) , reanalysis is i n a p p l i c a b l e in this c o n f i g u r a t i o n , since om is not the head of the complement of zonder. It is a characteristic p r o p e r t y of reanalysis rules that they apply to heads o n l y . If the COMP p o s i t i o n may be f i l l e d by the preposition om, other p r e p o s i t i o n s might in p r i n c i p l e occupy this position as well. T h e n , zonder in ( 5 1 c ) could also occupy this position if om is not p r e s e n t , thus y i e l d i n g a s t r u c t u r e like ( 5 5 ) instead of (54). INFL max

(55)

INFL' zonder

max

Suppose this is in fact c o r r e c t . Then we can account for the

-35-

Gaps and Parasitic Gaps contrast between ( 4 8 ) and ( 4 9 ) . This contrast is again by the pair in ( 5 6 ) .

illustrated

( 5 6 ) a . voor na dat je gedoucht hebt for a f t e r that you showered have b.*voor na te hebben gedoucht (=(49c)) for a f t e r to have showered Their respective tree s t r u c t u r e r e p r e s e n t a t i o n s are given in ( 5 7 a ) and ( 5 7 b ) , under the assumption that the preposition na p r o j e c t s a PP s t r u c t u r e of its o w n . H o w e v e r , u n d e r the assumption made above, that the preposition na may occupy the COMP position of the i n f i n i t i v a l clause, the s t r u c t u r e of ( 5 7 b ) would be as in ( 5 8 ) .

( 5 7 ) a.

PP

b.

PP

PP voor

PP

INFL max

P na

INFL1

COMP

dat

voor

V

na

INFLC+T]

je gedoucht hebt

(58) max

voor

P

COMP

PRO te hebben gedoucht

-36-

INFL max COMP

0

INFL'

V

max

INFL[-T]

PRO te hebben gedoucht

Gaps and P a r a s i t i c Gaps ( 5 7 a ) violates the UCC as it s t a n d s , but given the fact that lexical conditions are f u l f i l l e d , reanalysis of the two Ps can take place. If ( 5 7 b ) were the correct representation of ( 5 6 b ) , we would expect reanalysis to a p p l y in this case as well, and hence expect ( 5 6 b ) to be g r a m m a t i c a l . However, ( 5 8 ) is the s t r u c t u r e we are advocating in this section. Although lexical conditions are m e t , reanalysis cannot take place in this c o n f i g u r a t i o n , because the preposition na is not the head of the sister projection of the preposition voor. Since voor does govern na just as zonder governs om in ( 5 4 ) , (58) violates the UCC. The conclusion that the preposition i n t r o d u c i n g tenseless a d j u n c t clauses is in COMP, so that the construction as a whole is of the category S 1 ( = I N F L m a x ) , rather than P P , derives f u r t h e r support from the following o b s e r v a t i o n . PPs can have specifiers like vlak, drie uur, e t c . , as is shown in ( 5 9 ) . ( 5 9 a ) e x e m p l i f i e s this with P taking an NP complement and ( 5 9 b ) with P taking a tensed clause as complement. However, these specifiers are not allowed if the preposition introduces a tenseless clause. This follows from our analysis if it is assumed that the relevant specifiers only combine with a P p r o j e c t i o n , which we claim is absent in the constructions in ( 6 0 ) . [ 5 ] , ( 5 9 ) a . Vlak voor [,.„ de m a a l t i j d ] dronken we een glas sherry Right before the meal drank we a glass sherry Drie uur na [„ρ de w e d s t r i j d ] zaten de speiers nog te h i j g e n Three hours a f t e r the game sat the players still to to gasp b. Vlak voor [ „ , dat we gingen eten] dronken we een glas sherry Right b e f o r e that we went eat drank we a glass sherry Drie uur na [„, dat ze de w e d s t r i j d gespeeld h a d d e n ] zaten de speiers nog te h i j g e n Three hours a f t e r that they the game played had sat the players still to gasp

-37-

Gaps and Parasitic Gaps

( 6 0 ) a . * V l a k alvorens te gaan eten d r o n k e n we een glas sherry Right b e f o r e to go eat drank we a glass sherry b . * D r i e uur na de w e d s t r i j d gespeeld te hebben zaten de speiers nog te h i j g e n Three hours a f t e r the game played to have sat the players still to gasp One question remains to be a n s w e r e d , b e f o r e this analysis can be considered f u l l y s a t i s f a c t o r y : what makes ( 5 7 b ) unacceptable? Let us suppose that Dutch d i f f e r s from English in that

at

S-structure

COMP must be f i l l e d . This is evident in the case of tensed embedded clauses, as t h e r e

is

no

rule

in

Dutch

comparable

to

English

that-deletion. It is also clear in main clauses, where the sentence initial COMP must always be f i l l e d - by the f i n i t e verb in all cases and by a f r o n t e d c o n s t i t u e n t in clauses d i f f e r e n t from yes-no questions and imperatives. This assumption implies that there is an optional rule of om-deletion in the PF component of the grammar account for the o p t i o n a l i t y of om in cases like ( 5 0 ) . [ 6 ] . Given the r e q u i r e m e n t of COMP being f i l l e d at S - s t r u c t u r e , explain why ( 5 7 b ) is ungrammatical as it

we

to can

stands. The UCC applies at

S-structure. C o n s e q u e n t l y , since there cannot be an empty

COMP

at

S - s t r u c t u r e , ( 5 7 b ) can only have om in COMP p o s i t i o n , but then the s t r u c t u r e violates the UCC for the same reason as ( 5 4 ) , i.e. om is governed by na and no reanalysis can take place. Thus, there is no way in which ( 5 7 b ) can be a c c e p t e d . T h e r e f o r e , ( 5 8 ) is the only possible s t r u c t u r e , but this s t r u c t u r e also violates the U C C , as discussed. We are now in a position to r e t u r n to the problem that we l e f t unsolved at the end of the previous subsection. There we noted a contrast between Dutch and English with respect to the possibility of extracting the PP complement of P. T h u s , in Dutch a PP complement to a postposition could be e x t r a c t e d , as illustrated

by

the example in ( 3 6 b ) , r e p e a t e d h e r e , whereas the PP complement of a preposition in English cannot be e x t r a c t e d , as was shown by the examples in ( 4 0 ) , also r e p e a t e d h e r e . This contrast was u n e x p e c t e d , given the fact that their s t r u c t u r e s are m i r r o r images.

-38-

Gaps and P a r a s i t i c Gaps ( 3 6 ) b . Ik vroeg [achter welk gekomen I asked behind which come ( 4 0 ) a . * U n d e r which table did b . * I n t o which meadow did

g o r d i j n ] hij _t v a n d a a n was curtain he

from

was

Fido jump from _t Fido run out _t

From our p e r s p e c t i v e , it was the u n g r a m m a t i c a l i t y of ( 4 0 ) which was u n e x p e c t e d , since the Gap C o n d i t i o n would allow it, just as it allows ( 3 6 b ) . As we have argued in this section, PP complements to prepositions are in p r i n c i p l e excluded by the U C C . T h e r e f o r e , reanalysis must have a p p l i e d to the s t r u c t u r e s in ( 4 0 ) in order to escape the e f f e c t of the U C C . We assume that the f o r m a t i o n of a prepositional complex by reanalysis makes the complex inseparable. This accounts for the u n g r a m m a t i c a l i t y of ( 4 0 ) . We have argued that the UCC is a d i r e c t i o n a l p r i n c i p l e , the d i r e c t i o n in which it applies being the d i r e c t i o n in which Case is assigned. Since Ps in Dutch and English assign Case in the same d i r e c t i o n , the UCC does not apply to the s t r u c t u r e u n d e r l y i n g ( 3 6 b ) , since the m a t r i x P is p o s t p o s i t i o n a l . This accounts for the contrast between Dutch and English in this r e s p e c t .

1.4 Parasitic gaps A f t e r this digression on the d i s t r i b u t i o n of gaps in Dutch and the comparison w i t h that of E n g l i s h , we now t u r n to the d i s t r i b u t i o n of parasitic gaps. The p a r a s i t i c gap phenomenon may be i l l u s t r a t e d by the example in ( 6 1 ) . A p a r t from the gap created by W H - e x t r a c t i o n , indicated by _ t , t h e r e is a second g a p , i n d i c a t e d by e_, which is p a r a s i t i c on the first gap.[7]. ( 6 1 ) W h i c h a r t i c l e s did John f i l e _t w i t h o u t reading £ The reason for calling e^ p a r a s i t i c on the trace is that occurrence is indeed dependent on _ t , as can be d e m o n s t r a t e d by examples in ( 6 2 ) . ( 6 2 a ) shows that e_ is not licensed if

-39-

its the its

Gaps and Parasitic Gaps antecedent is in an argument position. ( 6 2 b ) shows that from the position of £ itself is not allowed.

extraction

( 6 2 ) a . * J o h n filed these reports without reading £ b . * W h i c h reports did John go home w i t h o u t reading t_ In ( 6 1 ) , both t^ and £ are bound by which articles, which is in a non-argument ( A 1 ) position. T h e r e f o r e , both empty categories a r e variables according to the functional definition of empty categories in Chomsky ( 1 9 8 2 , c h . 3 ) . This q u a l i f i c a t i o n explains a f u r t h e r restriction on parasitic gaps, which is illustrated by

(63). ( 6 3 ) *Who do you think _t_ went home without John having met £ The reason for the ungrammaticality of ( 6 3 ) follows from the required status of £ as a variable: ;t, the real gap, is coindexed with £ and c-commands £. T h e r e f o r e , £ has ;t rather than who as its local binder and is t h e r e f o r e locally Α-bound. Thus, the ungrammaticality of ( 6 3 ) reduces to the same principle that rules out ( 6 4 ) , v i z . the requirement that variables be locally A - f r e e . ( 6 U ) *Who.

did he.

expect John to see t^.

Similarly, the parasitic gap may not c-command the real gap, since in that case the parasitic gap would function as the local antecedent of the real g a p , so that the real gap is locally Α-bound. This explains the ungrammaticality of ( 6 5 ) . ( 6 5 ) *Which girl did you present £ to t_ ( c f . Which girl did you send a picture of £ to t) The descriptive statement of ( 1 9 8 2 : 6 6 ) is as in ( 6 6 ) .

a

parasitic

-40-

gap

given

by

Chomsky

Gaps and Parasitic Gaps ( 6 6 ) In the construction ( A ) , where order is irrelevant and we assume a_, t^, £ to be coindexed, the parasitic gap £ is licensed if and only if ( B ) : \η )

(Β)

· * * · 3 · · · Ό · · · 6

( i) ( ii) (iii) ( iv)

a_ c-commands t^ and £ _t does not c-command £ or conversely a_ does not head the chains (j*,_t) and (£,£) £ is governed (is not P R O ) and heads a chain with a theta-role

Here, ( i i i ) states in e f f e c t that the antecedent expression a_ must be in a non-argument position, whereas ( i v ) states that the parasitic gap is subject to the ECP, just like the real gap. An interesting aspect of the parasitic gap phenomenon is that the relation between the parasitic gap and the antecedent is not subject to the bounding condition, i . e . s u b j a c e n c y , which is what we expect given that no movement is involved, and that subjacency is a condition on extraction operations. This fact is illustrated by the contrast between ( 6 1 ) and ( 6 2 b ) . However, Kayne ( 1 9 8 U , c h . 8) observes that contrasts like those in ( 6 7 ) and ( 6 8 ) are unexpected. ( 6 ? ) a . The person that John described t^ without examining any pictures of £ b.*The person that John described _t without any pictures of £ being on file ( 6 8 ) a . The books you should read Jb before it becomes d i f f i c u l t to talk about £ b.*The books you should read t^ b e f o r e talking about £ becomes d i f f i c u l t These contrasts cannot be explained in terms of a subjacency violation since subjacency is irrelevant for parasitic gaps. Nor can they be explained by the E C P , although there is a s u b j e c t / n o n - s u b j e c t a s y m m e t r y , since the parasitic gap is governed by a preposition in all cases. The ungrammaticality of the (b) examples in ( 6 ? ) - ( 6 8 ) cannot be a t t r i b u t e d to the fact that the parasitic gap is embedded in a s u b j e c t , since cases like the one in ( 6 9 ) are grammatical.

-in-

Gaps and Parasitic Gaps

( 6 9 ) A person who close f r i e n d s of e_ admire t_ In section 2 . 1 , we saw that Kayne replaces the s t a n d a r d ECP in (1) by his g-projection r e q u i r e m e n t in ( 5 ) , which has a wider scope than s u b j e c t / n o n - s u b j e c t a s y m m e t r i e s . The contrasts between the (a) examples and the (b) examples of ( 6 7 ) and ( 6 8 ) are reminiscent of the contrast between the examples in ( 4 ) , repeated h e r e . This contrast was explained in terms of the g-projection r e q u i r e m e n t . ( 4 ) a . Which actress do you think that John likes p i c t u r e s of _t b . * W h i c h actress do you t h i n k that p i c t u r e s of t_ t u r n Bill on The contrast in ( 6 7 ) - ( 6 8 ) reduces to the g-projection requirement if it is assumed that both gaps must be related to their antecedent by a g-projection. The maximal g-projection of the governor of £ in the (b) examples is the embedded s u b j e c t N P , because this NP is not in a canonical government c o n f i g u r a t i o n . The maximal g-projection of the governors of both gaps in the (a) examples is the matrix S ' , which contains their antecedent. This simple extension of the a p p l i c a t i o n of the g-projection requirement does not explain the grammaticality of ( 6 9 ) . Let us consider the tree s t r u c t u r e of the relevant part of ( 6 9 ) .

(70)

The nodes numbered 1 represent the g-projection of the governor

-42-

of

Gaps and Parasitic Gaps £ and the nodes numbered 2 represent the g - p r o j e c t i o n of the governor of t_. Kayne proposes that the g-projection is extended to what he calls the Connectedness C o n d i t i o n , which requires that distinct g-projections that relate to the same antecedent be connected, such that the g-projections f o r m a connected s u b t r e e . This is the case in the s t r u c t u r e in ( 7 0 ) : the maximal g-projection of the governor of £ connects d i r e c t l y to the g-projection of the governor of _t without intervening nodes. This connection requirement can be built into our Gap Condition in ( 1 3 ) very easily by requiring that the relation between an antecedent and its gaps forms a connected subtree. Both in terms of the Connectedness Condition and our revised f o r m u l a t i o n of it as the Gap C o n d i t i o n , the contrast between the (a) and (b) examples of ( 6 7 ) - ( 6 8 ) is explained s t r a i g h t f o r w a r d l y , as can be seen in the tree representations in ( 7 1 ) . 12

without PRO

examining

any pictures

sale

With this background in the theory of parasitic gaps, discuss the distribution of parasitic gaps in D u t c h .

-43-

we

shall

Gaps and Parasitic Gaps

1.5 Parasitic Gaps in Dutch In the previous sections we have discussed the d i s t r i b u t i o n of normal gaps in Dutch and explained how this d i s t r i b u t i o n is a d e q u a t e l y accounted for in terms of the Gap C o n d i t i o n . In this section we examine the e f f e c t s of this condition on the d i s t r i b u t i o n of parasitic gaps. In 5.1 we shall concentrate on the normal cases of parasitic gaps, while 5 - 2 focuses on an apparent problem for the theory of parasitic gaps in constructions that do not appear to have a real gap to license the parasitic gap.

1.5.1 The d i s t r i b u t i o n of p a r a s i t i c gaps In this section we discuss the reasons why the d i s t r i b u t i o n of parasitic gaps in Dutch is much more limited than in other languages, specifically English. The f i r s t reason has to do with the fact that there are two types of gaps in D u t c h . As we showed in section 3, gaps inside PPs necessarily take a [ + R ] a n t e c e d e n t , while other gaps take a [-R] a n t e c e d e n t . This fact limits the possible combinations of two gaps, one r e a l , the other p a r a s i t i c , since either both of them are of the [ + R ] variety or both are of the [-R] v a r i e t y . Thus, there is no equivalent to an English example like ( 7 2 ) . ( 7 2 ) This is the guy that I suggested to £ that my brother could o f f e r t^ a job The reason is that e^ requires a [+R] a n t e c e d e n t , whereas t^ can only take a [-R] a n t e c e d e n t . So, both the option with the C - R ] relative pronoun die and the option with the [ + R ] r e l a t i v e pronoun waar are ungrammatical. ( 7 3 ) * D i t is de jongen d i e / w a a r ik e_ aan suggereerde dat m i j n broer t^ een baan kon aanbieden If t^ is replaced by a p r o n o u n , the sentence w i t h the [ + R ] relative

-44-

Gaps and P a r a s i t i c Gaps pronoun is g r a m m a t i c a l . ( ? 4 ) Dit is de jongen waar ik t_ aan suggereerde dat m i j n broer hem een baan kon aanbieden If both gaps are of the [ + R ] v a r i e t y , a construction of

this

type

is possible, as is shown by the example in ( 7 5 ) . ( 7 5 ) Dit is het artikel waar ik £ over zei dat H a r r y This is the a r t i c l e where I about said that Harry een reactie t_ op moest s c h r i j v e n a reaction to should w r i t e It should be noted that it is impossible to have a construction of this type w i t h two C - R ] gaps. The reason for this is that the C - R ] gap in the m a t r i x clause would c-command the gap in clause and hence f u n c t i o n as its local antecedent.

the embedded Consequently,

the embedded gap is not locally Α - f r e e . T h e r e f o r e , an example

like

( 7 6 ) is u n g r a m m a t i c a l , just as its English c o u n t e r p a r t is. ( 7 6 ) * D i t is de man die ik ^ v e r t e l d e dat m i j n broer This is the man who I

told

that my b r o t h e r

t_ zou bezoeken would visit In section 3i we explained why there can only be a single [ + R ] gap in a given clause, due to the r e q u i r e m e n t that the PP from which the R - p r o n o u n is e x t r a c t e d must itself be canonically governed. Since PPs in NPs follow the n o u n , only PPs which are governed by V allow e x t r a c t i o n . Since there is only one V per clause, there can only be one [ + R ] gap per clause. T h e r e f o r e , there are no Dutch c o u n t e r p a r t s t o constructions like ( 7 7 ) . ( 7 7 ) Who did you give a p i c t u r e of £ to t_ S i m i l a r l y , t h e r e can only be one [-R] gap per clause, since [-R] gaps cannot be couched in a PP and c o n s e q u e n t l y will always stand in a c-command r e l a t i o n s h i p . Summarizing the discussion so f a r , we

-45-

Gaps and Parasitic Gaps have seen two r e s t r i c t i o n s on parasitic gap c o n s t r u c t i o n s : a. the two gaps must be of the same type ( [ + R ] ) b. the gaps must be contained in d i f f e r e n t clauses These r e s t r i c t i o n s f o l l o w from the theory presented thus f a r . It is easy to see that both r e q u i r e m e n t s are f u l f i l l e d in constructions of the type e x e m p l i f i e d in ( 7 5 ) . The second clause in this type is a complement clause. There are two other conceivable combinations of two clauses: a) a m a t r i x clause and a relative clause, and b) a m a t r i x clause and an a d j u n c t clause. Both types allow parasitic gaps in English. We shall now examine these two types in D u t c h . Let us examine clauses embedded in NPs f i r s t . An example of this type of construction in English is given in ( 7 8 ) . ( 7 8 ) This is the book [that everyone [who reads e] becomes e n t h u s i a s t i c about J^] Constructions of this type are obviously impossible in D u t c h : the clause embedded in NP is not canonically g o v e r n e d , since clauses follow the noun in Dutch as well as in English. These constructions are impossible, both w i t h [-R] gaps, as in ( 7 9 ) , and w i t h [+R] gaps, as in ( 8 0 ) . ( 7 9 ) * D i t is net boek dat iedereen [die £ leest] t_ bewondert This is the book that everyone who reads admires ( 8 0 ) * D i t is een vraag waar iedereen [die £ over denkt] This is a question where everyone who about thinks een antwoord J; op weet an answer to knows Let us now t u r n to c o n s t r u c t i o n s with a d j u n c t clauses, which c o n s t i t u t e a very p r o d u c t i v e class of p a r a s i t i c gap c o n s t r u c t i o n s in other languages. We mentioned earlier that a d j u n c t clauses are islands for e x t r a c t i o n s , both in English and in D u t c h . The fact that they are islands in Dutch does not come as a s u r p r i s e , given the proposal we made in 2.5 c o n c e r n i n g e x t r a c t i o n from embedded clauses. C o n s i d e r ( 8 1 ) . -46-

Gaps and P a r a s i t i c Gaps

( 8 l ) * W a t is Jan [na [dat hij t_ zag]] v e r t r o k k e n What is John a f t e r that he saw l e f t The embedded S' is a g-projection

of the governor of the

this S' is on a right branch. So, the governor of S ' , the

PP

headed

by

maximal na,

is

gap,

but

of

the

projection not

a

g-projection.

C o n s e q u e n t l y , t h e p a t h breaks o f f a t S ' . In E n g l i s h , a d j u n c t clauses are islands for e x t r a c t i o n as w e l l , is i l l u s t r a t e d in ( 8 2 ) .

as

( 8 2 ) * W h a t did John leave a f t e r he had seen _t The ungrammaticality of ( 8 2 ) is predicted just as the u n g r a m m a t i c a l i t y of

(81)

is,

by

our

but

Gap C o n d i t i o n , the

reasons

are

slightly d i f f e r e n t . While in Dutch the g - p r o j e c t i o n of the governor of the gap breaks off at the level of the embedded

S',

it

breaks

off in English at the level of PP. The reason for this d i f f e r e n c e is that u n l i k e in D u t c h , the S' is canonically governed by the p r e p o s i t i o n in English. T h e r e f o r e , the PP is a g-projection of the governor of the gap. The reason why this g - p r o j e c t i o n cannot be c o n t i n u e d is that the PP itself is not governed by V. Hence, Vmax is not a g - p r o j e c t i o n . This d i f f e r e n c e allows us to make a precise p r e d i c t i o n . An English a d j u n c t clause may c o n t a i n a parasitic gap, if the a d j u n c t PP is connected to the p a t h of a real g a p , but a Dutch a d j u n c t clause should not be able to contain a p a r a s i t i c gap, since the PP itself is not a g - p r o j e c t i o n . This prediction turns out to be c o r r e c t , as is i l l u s t r a t e d by the contrast between ( 8 3 ) and their Dutch t r a n s l a t i o n s i n ( 8 4 ) . ( 8 3 ) a . Which book did you r e t u r n t^ b e f o r e you could read e_ b. A person that they spoke to t_ because they admired £ c. This is the kind of food you must cook t_ b e f o r e you eat £ d. The paper that we should d e s t r o y _t b e f o r e someone steals a copy of _e

-47-

Gaps and Parasitic Gaps ( 8 4 ) a . * W e l k boek moest je t_ terugbrengen voor dat je e^ kon lezen b.*Een persoon waar zij t_ tegen spraken om dat zij bewondering e^ voor hadden c.*Dit is het soort voedsel dat je _t moet koken voor dat je e^ eet d.*Het papier dat we t_ moesten vernietigen voor dat iemand £ steelt Let us illustrate why there is a difference in grammaticality between ( 8 3 ) and ( 8 4 ) by drawing the relevant tree structures of (83a) and ( 8 4 a ) . The point where the g-projection breaks off is boxed. It is clear that the boxed constituent connects up with the path of the real gap in the tree s t r u c t u r e of ( 8 3 a ) , but not in the tree structure of ( 8 4 a ) . (83a·) COMP

which books

return

t

before COMP 0

S V max

NP INFL you could

V V

read

-48-

NP

e

Gaps and Parasitic Gaps (84af) COMP welk boek

INFL

V

t_

terug-

voor COMP

INFL'

brengen

dat NP je

V'"ax

INFL

V

kon

Λ

NP V

e

lezen

These tree s t r u c t u r e s are in accordance with the assumptions about the head of S and S 1 that we made above. Although the prediction about the non-occurrence of parasitic gaps t u r n s out to be correct in tensed a d j u n c t clauses, parasitic gaps can be found in tenseless a d j u n c t clauses, as has been observed by Dutch (and German) linguists since the i n t r o d u c t i o n of the notion of parasitic gaps. Examples w i t h [-R] gaps are p r o v i d e d in ( 8 5 ) and examples with [+R] gaps in ( 8 6 ) . ( 8 5 ) a . Welke boeken heb je [zonder £ te b e k i j k e n ] _t_ Which books have you without to inspect b. Dit is de oom die ik [na j a r e n niet e_ gezien This is the uncle who I a f t e r years not seen gisteren weer _t tegenkwam y e s t e r d a y again met -49-

weggezet away put te hebben] to have

Gaps and Parasitic Gaps c. Dit is de oom die ik [na jaren niet e^ gezien te hebben] This is the u n c l e who I a f t e r years not seen to have hoopte dat ik in Kopenhagen _t zou ontmoeten hoped that I in Copenhagen would meet ( 8 6 ) a . Waar heb je [na twee jaar £ over nagedacht te hebben] Where have you a f t e r two years about thought to have een oplossing t^ voor gevonden a solution to found b. Ik hou niet van vragen waar ik [zonder eerst rüstig I love not of question where I without f i r s t q u i e t l y e_ over na te denken] een antwoord t_ op moet geven about to think an answer to must give c. Dit is het speelgoed waar ik [na jaren niet ^ naar This is the toy where I a f t e r years not to omgekeken te hebben] dacht dat ik Piet t_ mee zag speien looked-for to have thought that I Peter w i t h saw play The fact that these p a r a s i t i c gaps are allowed is surprising given our explanation of the impossibility of a parasitic gap in a tensed a d j u n c t c l a u s e . [ 8 ] . It will be r e c a l l e d , however, that we argued in section 3 - 4 that the p r e p o s i t i o n introducing an i n f i n i t i v a l clause occupies the COMP position of that clause, rather than being the head of a PP, taking this clause as its c o m p l e m e n t . As can be seen in the tree structure in ( 8 4 a ) , it is the PP which stands in the way of connecting the g - p r o j e c t i o n of the governor of the parasitic gap to the path of the real gap. However, in the case of i n f i n i t i v a l a d j u n c t s , this PP s t r u c t u r e would be absent, i.e. the tree s t r u c t u r e of the examples in ( 8 5 ) and ( 8 6 ) would not be as in ( 8 7 ) , but rather as in (88) ( c f . the tree structures in ( 5 5 ) and (58)).

-50-

Gaps and P a r a s i t i c Gaps r max

(87)

NP

COMP 0

INFL' Vmax

V

t

INFL

.max

(88)

COMP

It should be stressed t h a t the d i s t r i b u t i o n of

parasitic

gaps

in

Dutch is e n t i r e l y as p r e d i c t e d by the c o n s t r a i n t s t h a t are r e l e v a n t for normal

gaps.

Therefore,

it

would

be

counterproductive

to

p r o v i d e an a l t e r n a t i v e analysis to a c c o u n t for these second gaps if

-51-

Gaps and Parasitic Gaps this a l t e r n a t i v e were non-consequential for the d i s t r i b u t i o n of real gaps. In that case the theory would r e d u n d a n t l y allow for an analysis of these second gaps as parasitic gaps as w e l l . Before we t u r n to a discussion of the surprising phenomenon that we announced at the beginning of this section, we want to discuss an issue that is r e l e v a n t for the debate of the derivational status of parasitic gaps. In Chomsky ( 1 9 8 2 ) , parasitic gaps are assumed to be empty categories at D - s t r u c t u r e , their status of variables being d e t e r m i n e d at S-structure or LF by the functional definition of empty categories. Their status of variable, t h e n , does not result from movement of the content from the position of the parasitic gap to an A ' - p o s i t i o n , as is the case with normal variables, but rather from their local binding by an antecedent in an A ' - p o s i t i o n . More r e c e n t l y , this n o n - d e r i v a t i o n a l approach to parasitic gaps has been called into question ( C h o m s k y , class lectures where he r e f e r s to work by L o n g o b a r d i ) . Parasitic gap constructions appear to be sensitive to Complex Noun Phrase Constraint e f f e c t s , something which is unexpected under the approach taken in Chomsky ( 1 9 8 2 ) . It is t h e r e f o r e proposed that parasitic gaps do result from movement to A ' - p o s i t i o n , thus creating an operator-variable relation i n d e p e n d e n t l y . It t u r n s out that the latter approach conflicts with the situation found in Dutch. In section 2, we argued that extraction from sentential complements in Dutch is possible by successive cyclic movement only. T h e r e f o r e , it is p r e d i c t e d by the derivational approach to parasitic gaps that a p a r a s i t i c gap could be found in a complement clause w i t h i n an a d j u n c t clause. This situation is depicted in

(89).

-52-

Gaps and P a r a s i t i c Gaps

.max

(89)

COMP op.

INFL' V

INFL

NP

V

t

The most d e e p l y e m b e d d e d _t would be

the

extraction

site

of

the

parasitic g a p . The _t in COMP r e s u l t s from successive cyclic movement of the o p e r a t o r op. into the m a t r i x COMP, which connects to the path of the real g a p . Under the n o n - d e r i v a t i o n a l a p p r o a c h , there is no m o v e m e n t , hence no successive cyclic m o v e m e n t ,

and

we

therefore

predict

that

the

g - p r o j e c t i o n of the governor of a p a r a s i t i c gap c o n t a i n e d in a complement clause inside the a d j u n c t b r e a k s off at the level of this complement c l a u s e , as this clause i t s e l f is not c a n o n i c a l l y governed. Therefore, the derivational of the t y p e e x e m p l i f i e d in ( 9 0 ) to non-derivational approach adopted u n g r a m m a t i c a l . The l a t t e r p r e d i c t i o n

-53-

a p p r o a c h p r e d i c t s sentences be g r a m m a t i c a l , whereas the here p r e d i c t s them to be t u r n s out to be c o r r e c t .

Gaps and Parasitic Gaps

( 9 0 ) * W e l k e boeken heb je [zonder te weten [dat je e^ Which books have you w i t h o u t to know that you mocht b e k i j k e n ] _t doorgebladerd were allowed to inspect browsed-through This evidence thus favors the original approach to taken in Chomsky ( 1 9 8 2 ) .

parasitic

gaps

1.5.2 Parasitic gaps w i t h o u t real gaps Up to this p o i n t , the discussion of the d i f f e r e n t i a l d i s t r i b u t i o n of parasitic gaps in Dutch and English has been in accordance w i t h the methodological requirements that follow from the peripheral n a t u r e of the parasitic gap phenomenon. The d i f f e r e n c e s were all shown to f o l l o w from independently existing d i f f e r e n c e s between the two languages. T h e r e f o r e , no language specific statements are needed either in the grammar of Dutch or the grammar of English to account for the phenomenon. It was pointed out above, h o w e v e r , that Dutch appears to have gaps which look like p a r a s i t i c gaps, since they occur in positions from which no extraction seems possible, but which do not seem to be licensed by the presence of a real gap. Such gaps are found in the examples in ( 9 1 ) . ( 9 1 ) a . Jan h e e f t die boeken [zonder £ te b e k i j k e n ] weggelegd John has those books w i t h o u t to inspect away put b. Ik ben m i j n oom [na e_ jaren niet gezien te hebben] I have my uncle a f t e r years not seen to have gisteren weer tegen gekomen yesterday again against come c. Het is een schände dat de arts de p a t i e n t [zonder e_ It is a disgrace that the doctor the p a t i e n t w i t h o u t behandeld te hebben] v i j f t i g gulden liet betalen treated to have f i f t y guilders let pay

-54-

Gaps and P a r a s i t i c Gaps d. Ik heb deze s c r i p t i e [alvorens d e f i n i t i e f e_ te I have this term paper b e f o r e d e f i n i t i v e l y

to

b e o o r d e l e n ] eerst aan Jan voorgelegd j u d g e f i r s t to John shown The gaps are i n t e r p r e t e d as c o r e f e r e n t i a l

with

die

boeken,

mijn

oom, de patient, and deze scriptie, r e s p e c t i v e l y .

However,

if

no

movement is i n v o l v e d , these gaps are bound

category

in

an

by

a

Α - p o s i t i o n , which is generally e x c l u d e d as a consequence of the B i n d i n g Theory ( i . e . p r i n c i p l e C , which r e q u i r e s that v a r i a b l e s a r e locally Α - f r e e ) . The English c o u n t e r p a r t s of the examples

in

(91)

are indeed e x c l u d e d : ( 9 2 ) a . * J o h n put those books in the bookcase w i t h o u t looking into b.*I met my uncle y e s t e r d a y a f t e r not having seen for many years c.*It is a disgrace that the d o c t o r made the patient pay f i f t y guilders without having treated d . * I showed t h i s term p a p e r to John b e f o r e j u d g i n g definitively It would be c o u n t e r p r o d u c t i v e at this moment

to

assume

that

the

r e q u i r e m e n t that the p a r a s i t i c gap be Α - f r e e does not hold in D u t c h , since such a step would not only i n t r o d u c e language s p e c i f i c s t i p u l a t i o n s w i t h respect to the p a r a s i t i c gap phenomenon, but would also make the wrong p r e d i c t i o n s . T h u s , if it is assumed t h a t the a n t e c e d e n t of a p a r a s i t i c gap may be in an Α-position in D u t c h , the sentence in ( 9 3 ) would be p r e d i c t e d to be g r a m m a t i c a l , c o n t r a r y to f a c t . ( 9 3 ) Ik v e r t e l d e Jan dat ik [ z o n d e r hem/*e_ te w i l l e n I told John t h a t I w i t h o u t him to want beledigen] a n d e r s had besloten insult d i f f e r e n t l y h a d d e c i d e d Since the u n d e r s t o o d P R O - s u b j e c t of the a d j u n c t o b l i g a t o r i l y c o n t r o l l e d by the s u b j e c t of the embedded

-55-

clause clause

is ik,

Gaps and Parasitic Gaps the nearest antecedent for e_ would be Jan, which c-commands e^. N e v e r t h e l e s s , the sentence is u n g r a m m a t i c a l . It would be highly u n a t t r a c t i v e to assume that in D u t c h , u n l i k e in E n g l i s h , p a r a s i t i c gaps are s u b j e c t to some sort of locality r e q u i r e m e n t in order to account for the u n g r a m m a t i c a l i t y of ( 9 3 ) , since, as we have seen above, parasitic gaps in the examples that we discussed earlier are not subject to such a r e q u i r e m e n t . In the case of R-gaps, it can be demonstrated quite clearly that a real g a p , as well as a path with which the path of the parasitic gap can be connected, are required. The element er can have a number of d i s t i n c t f u n c t i o n s , some of which can coalesce in a given s t r u c t u r e . We can distinguish between a locative er as in ( 9 ^ a ) , a presentative er which appears in sentences with i n d e f i n i t e subjects as in ( 9 ^ b ) , a prepositional er, of which several examples have already been given ( c f . ( 9 ^ c ) ) and a q u a n t i t a t i v e er, which is in most respects similar to the q u a n t i t a t i v e use of ne in Italian and en in French ( c f . ( 9 ^ d ) ) . ( 9 O a . ik heb * ( e r ) j a r e n gewoond I have there years lived b. dat * ( e r ) een jongen loopt that there a boy walks c. ik heb * ( e r ) niet e^ over gesproken I have there with Peter about talked d. ik heb * ( e r ) gisteren twee e_ gezien I have there yesterday two seen

locative er presentative er prepositional er quantitative er

Even if in a given c o n s t r u c t i o n the c o n d i t i o n s for the appearance of er are met more than once, only a single er is present at surface s t r u c t u r e ( c f . c h . 3 for an analysis and q u a l i f i c a t i o n of this s t a t e m e n t ) , as is shown in the following e x a m p l e s . [ 9 ] . ( 9 5 ) a . quantitative and prepositional: Ik heb er twee e_ e_ van I have t h e r e two of b. p r e s e n t a t i v e and p r e p o s i t i o n a l ; Toen kwamen er verscheidene mensen e_ naar k i j k e n Then came there various people at look

-56-

Gaps and P a r a s i t i c Gaps c. preservative and l o c a t i v e : Er woont niemand There lives nobody d . p r e s e n t a t i v e , q u a n t i t a t i v e , and p r e p o s i t i o n a l : Hoe er d r i e £ £ van k u n n e n samenvallen How there three of can coincide If a parasitic R-gap is p r e s e n t , it cannot be licensed by a presentative er, as is shown in the examples in ( 9 6 a ) and ( 9 6 b ) . A parasitic R-gap can only be licensed by an R - p r o n o u n in the m a t r i x clause, if this R - p r o n o u n is extracted from a position more deeply embedded than the a d j u n c t clause containing the p a r a s i t i c gap. An example is given in ( 9 6 c ) ( 9 6 ) a . * D a a r o m z i j n er verscheidene mensen [na lang e_ T h e r e f o r e are there various people gewoond te hebben] v e r t r o k k e n

a f t e r long

lived to have left b . * D a a r o m z i j n er verscheidene mensen [zonder e_ T h e r e f o r e are there various people w i t h o u t naar te k i j k e n ] weggegaan at to look left c. Daarom z i j n er verscheidene mensen [zonder e^ T h e r e f o r e are t h e r e various people w i t h o u t naar te k i j k e n ] £ längs gelopen at to look along walked In ( 9 5 b ) and ( 9 6 b ) er should have both a presentative f u n c t i o n

and

an antecedent f u n c t i o n w i t h respect to the prepositional R-gap. The g r a m m a t i c a l i t y of ( 9 5 b ) shows that both f u n c t i o n s can be combined in one occurrence of er. T h e r e f o r e , the u n g r a m m a t i c a l i t y of ( 9 6 b ) must be related to the fact that the p r e p o s i t i o n a l R-gap is c o n t a i n e d in an a d j u n c t clause. Given the fact that the a d j u n c t clause is an island in this case as w e l l , the p r e p o s i t i o n a l R-gap in ( 9 6 b ) must be p a r a s i t i c . The g r a m m a t i c a l i t y of ( 9 6 c ) indicates that a parasitic R-gap in the a d j u n c t clause is indeed possible. The main d i f f e r e n c e between ( 9 6 b ) and ( 9 6 c ) is the fact t h a t , although in both sentences er occupies an A ' - p o s i t i o n , in ( 9 6 c ) er

-57-

Gaps and P a r a s i t i c Gaps binds an R-gap that is more deeply embedded than the a d j u n c t clause. T h e r e f o r e a path is created w i t h which the path of the parasitic R-gap can connect. In ( 9 6 b ) , on the other h a n d , there is no such path and consequently no connection can be made. These R-gap phenomena thus i l l u s t r a t e q u i t e nicely that not only an antecedent in A ' - p o s i t i o n , but also a path in the m a t r i x clause, is r e q u i r e d in order to license a parasitic gap. R e t u r n i n g to the examples in ( 9 1 ) , we observe that in these sentences the gap is contained in an island (i.e. the adjunct c l a u s e ) . T h e r e f o r e , these examples pose two r e l a t e d , though logically i n d e p e n d e n t , problems: a. the antecedent of the gap is in an A-position b. the gap is contained in an island and there does not seem to be a path w i t h which the g - p r o j e c t i o n of the governor of the gap can connect Both problems would be solved at the same time if the structure of the sentences in ( 9 1 ) were comparable to ( 9 6 c ) , rather than to ( 9 6 a ) or ( 9 6 b ) , i.e. if the antecedent NP originated in a position more deeply embedded than the a d j u n c t . This would lead us to postulate ( 9 7 ) as the tree r e p r e s e n t a t i o n of ( 9 1 a ) , with t_ i n d i c a t i n g the Α-position from which the a n t e c e d e n t NP is e x t r a c t e d . A similar analysis is i n d e p e n d e n t l y proposed in Felix ( 1 9 8 3 ) . The antecedent (die boeken) is attached to the V - p r o j e c t i o n by C h o m s k y - a d j u n c t i o n , and is consequently in a non-argument position. ( 9 1 ) a . Jan heeft die boeken [zonder e_ te b e k i j k e n ] weggelegd John has those books [without to inspect] away put

-58-

Gaps and P a r a s i t i c Gaps (97)

Jan

V

max

NP die boeken

S'

majv v.max

p zonder

A

INFL1

COMP

NP

INFL

_t

/ ^"\^^ NP V PRO

NP

V

heeft weggelegd

V in te

kijken This r e p r e s e n t a t i o n f u l f i l l s all the r e q u i r e m e n t s for the two gaps. Both gaps are canonically governed by their respective V-heads and the m a t r i x clause V - p r o j e c t i o n belongs to the path of t^ w i t h which the p a t h in the a d j u n c t S' is p r o p e r l y c o n n e c t e d . This S' is itself a g - p r o j e c t i o n of the governor of the parasitic gap. Both gaps are bound by a c-commanding NP in a non-argument position and neither gap c-commands the o t h e r . [ 1 0 ] . The r e l e v a n t question at this point is how the existence of the real gap in ( 9 7 ) can be m o t i v a t e d . The f i r s t p r i n c i p l e which gives us the position of _t as the base position of the direct object is the X ' - p r i n c i p l e according to which m o d i f i e r s are a t t a c h e d at a higher level of structure than complements ( c f . J a c k e n d o f f 1 9 7 7 ) . On the assumption that Dutch is 0V. this implies the order a d j u n c t - N P - V , rather than N P - a d j u n c t - V . A f u r t h e r p e r t i n e n t consideration is the a d j a c e n c y r e q u i r e m e n t on Case Assignment proposed in Stowell ( 1 9 8 1 ) . If such a d j a c e n c y is r e q u i r e d for the verb to assign Case to its d i r e c t o b j e c t , there must be an NP-position a d j a c e n t to the v e r b , as in ( 9 7 ) . In his discussion of the u n i v e r s a l i t y of the a d j a c e n c y r e q u i r e m e n t -59-

Gaps and P a r a s i t i c Gaps on

Case

apparent

Assignment, violations

Stowell into

develops

account.

two

The

hypotheses

first

order

of

NPs and

optional

t r a d i t i o n a l VP-domain is e s s e n t i a l l y

in

w i t h an o b j e c t NP and an o p t i o n a l a d v e r b i a l allowed:

The

within

in

PP,

is

Dutch.

adverbials

free. Thus,

take

hypothesis

s p e c i f i c a l l y designed to cope w i t h such violations relative

to

the

constructions

both

orders

are

( 9 8 ) a . dat Jan z i j n v r i e n d i n in Amsterdam ontmoette that John his girl f r i e n d in A m s t e r d a m met b. dat Jan in A m s t e r d a m z i j n v r i e n d i n ontmoette that John in A m s t e r d a m his girl f r i e n d met Given these e x a m p l e s , one would be i n c l i n e d to assume that Case assigned to the NP in the position in ( 9 8 b ) , u n d e r

adjacency

the v e r b . However, Stowell assumes that the VP in Dutch

is

is with

double

h e a d e d , i.e. has two head p o s i t i o n s , one of which is V P - i n i t i a l . He also assumes that

Case

is

assigned

by

the

V-head

in

position under a d j a c e n c y . His two p r i n c i p a l reasons assumptions are the f o l l o w i n g : a. in main clauses the verb o b l i g a t o r i l y a p p e a r s in

initial

for

these

VP-initial

p o s i t i o n , in this way m o t i v a t i n g the d o u b l e headed c h a r a c t e r of the Dutch V P ; b. the d i r e c t o b j e c t must appear in V P - i n i t i a l position Both assumptions are w r o n g . The verb-second position of the finite v e r b in D u t c h and German main clauses is not a p o s i t i o n w i t h i n VP. but rather in COMP (see Den Besten 1 9 8 3 ) . As the example in ( 9 8 b ) shows, direct o b j e c t s do not have to appear in V P - i n i t i a l p o s i t i o n . ( F o r more d e t a i l e d c r i t i c i s m of S t o w e l l ' s

proposal,

see

Hoekstra

1 9 8 4 , 1 0 9 f f . ) A p a r t from these general d e f i c i e n c i e s of Stowell's p r o p o s a l , we do not quite see how he would be able to account for those c o n s t r u c t i o n s in which the d i r e c t o b j e c t does not occupy the V P - i n i t i a l p o s i t i o n , but rather a V P - m e d i a l p o s i t i o n . A v a r i a n t of S t o w e l l ' s analysis is p r e s e n t e d in Koopman ( 1 9 8 4 ) . She does not assume that the Dutch VP is double h e a d e d , but n e v e r t h e l e s s agrees w i t h Stowell in assuming that Case is from V P - i n i t i a l p o s i t i o n u n d e r a d j a c e n c y . Case is assigned f r o m a V P - i n i t i a l c l i t i c

-60-

assigned

According to Koopman. position by the Case

Gaps and Parasitic Gaps f e a t u r e s of the v e r b . It will be evident that t h i s analysis meets w i t h the same problems as S t o w e l l ' s : a d j a c e n c y is not met at s u r f a c e s t r u c t u r e , n e i t h e r with the Case assigning position i n i t i a l l y in V P , nor with the Case assigning position at the end of V P . The main reason for Koopman to assume Case assignment from VP-initial position is her hypothesis that Case is always assigned in the same direction by the various categories in a particular language. Since Dutch is ( m a i n l y ) p r e p o s i t i o n a l , this harmony would not exist if verbs in Dutch assigned Case from right to l e f t . Although such a cross-categorial harmony, at least as far as the direction of Case Assignment of V and P is c o n c e r n e d , seems to be f a i r l y widespread among languages of the w o r l d , it is clearly not a necessary r e q u i r e m e n t . [ 1 1 ] . In our a n a l y s i s , Case is assigned by the verb under government from right to l e f t . With respect to a d j a c e n c y , we assume t h a t , rather than being subject to a linear adjacency r e q u i r e m e n t , Case assignment is subject to s t r i c t government. Under the additional assumption of b i n a r y b r a n c h i n g , this has the e f f e c t of linear a d j a c e n c y between the Case assigning category and its complement N P . In E n g l i s h , this is not only t r u e for o b j e c t NPs but also for s u b j e c t s of ( s m a l l ) clauses. For D u t c h , however, this hypothesis has a d i f f e r e n t e f f e c t on objects than on s u b j e c t s . This can be i l l u s t r a t e d by the tree s t r u c t u r e s for small clause complements in ( 9 9 ) , where ( 9 9 a ) represents English a n d ( 9 9 b ) Dutch constructions of this t y p e . (99)a.

V

b.

/\SC

V

/\

NP

X

V

,

NP

SC

X

The NP is s t r u c t u r a l l y governed by V in both cases, but while V and NP are s t r i n g a d j a c e n t in E n g l i s h , the NP is separated from V in Dutch by the p r e d i c a t i v e p a r t of the small clause c o m p l e m e n t . The same situation o b t a i n s in the case of s u b j e c t s of f u l l clauses u n d e r our a n a l y s i s : the s u b j e c t in Dutch is governed by I N F L , which appears on the right of the V m a x , whereas the s u b j e c t itself is the

-61-

Gaps and P a r a s i t i c Gaps leftmost constituent within Vmax. In conclusion, then, it that S t o w e l l ' s adjacency requirement can be government and b i n a r y b r a n c h i n g , i.e. theory.

it

made

to

turns out

follow

from

is not a p r i m i t i v e of

The cases in which the d i r e c t o b j e c t is not ( s t r u c t u r a l l y )

the

adjacent

to its Case assigner, as in ( 9 8 a ) , are then accounted for by optional m o v e m e n t . Since movement must be u p g r a d i n g , this has the linear e f f e c t of movement to the l e f t . The movement leaves behind a Case marked t r a c e , which is t h e r e f o r e a v a r i a b l e according

to

the

inherent d e f i n i t i o n of empty categories. The landing

of

the

can

be

movement operation must be a non-argument p o s i t i o n , created through a d j u n c t i o n

(see

Belletti

& Rizzi

site which 1981).

As

a

r e s u l t , the e x t r a c t i o n site is also a variable according to the f u n c t i o n a l d e f i n i t i o n of empty categories, since it is in an Α-position and locally bound by an a n t e c e d e n t in A ' - p o s i t i o n . It should be noted that the choice of a p p l y i n g this leftward NP-movement is determined by p r a g m a t i c rather than by s t r i c t l y syntactic c o n s i d e r a t i o n s : d e f i n i t e NPs are more l i k e l y to be moved than i n d e f i n i t e s , clitics are obligatorily moved. It is interesting to note that the operation is f o r m a l l y identical to in English. Since the n a t u r e of b r a n c h i n g is the D u t c h , this movement will always have a This explains why types of NPs that

are

mirror

rightward likely

Heavy-NP-Shift moving

to

be

image

of

effect. moved

by

Heavy-NP-Shift are precisely those which are not likely to be moved in D u t c h , whereas H e a v y - N P - S h i f t of c l i t i c s is essentially excluded - a g a i n , not for s y n t a c t i c but r a t h e r for p r a g m a t i c r e a s o n s . [ 1 2 ] . Let us now t u r n to S t o w e l l ' s second h y p o t h e s i s for dealing with a p p a r e n t c o u n t e r e x a m p l e s to the a d j a c e n c y c o n d i t i o n . This h y p o t h e s i s is invoked to account for cases of n o n - a d j a c e n c y at surface s t r u c t u r e in I t a l i a n , where the object may be separated from the v e r b by an i n t e r v e n i n g a d v e r b of m a n n e r . Stowell proposes that there may be a level of argument p r o j e c t i o n , at which the adverb does not a p p e a r . Then, a d j a c e n c y may be met e i t h e r at S-structure i t s e l f , r e s u l t i n g in s t r i c t a d j a c e n c y as in E n g l i s h , or at the level of the argument p r o j e c t i o n , the latter option being the m a r k e d one chosen in I t a l i a n . We shall not dwell on the m e r i t s of this p r o p o s a l for I t a l i a n , but shall c o n c e n t r a t e on a possible a p p l i c a t i o n of t h i s idea to a c c o u n t for cases of n o n - a d j a c e n c y in

-62-

Gaps and P a r a s i t i c Gaps D u t c h . When this idea is a p p l i e d

to

the

examples

S - s t r u c t u r e of each of the sentences might d i r e c t l y its s u r f a c e m a n i f e s t a t i o n , i.e. v e r b , now taken to occupy the

w i t h o u t traces. VP-final

cases be met at the level of argument

in

( 9 8 ) , the

correspond

Adjacency

to

of

the

position,

would

in

both

projection,

since

at

that

level the PP in Amsterdam would not a p p e a r . If this e x p l a n a t i o n is m a i n t a i n e d , we would be at a loss w i t h respect to the parasitic gap constructions in ( 9 1 ) , since under this proposal there is no need to assume the existence of a real gap which is r e q u i r e d in order to make the connection w i t h the g - p r o j e c t i o n of the

governor

of

parasitic gap. In f a c t , the n o t i o n of argument p r o j e c t i o n , as it

the is

used h e r e , obscures the hierarchical r e l a t i o n between the N P , the a d v e r b i a l , and the v e r b , just as it obscures the linear r e l a t i o n . E s s e n t i a l l y , these r e l a t i o n s appear to vanish into thin a i r , just as one can say that the r e q u i r e d a d j a c e n c y on Case

since,

Assignment

is met at the level of argument p r o j e c t i o n , it might be argued that the local r e l a t i o n between verb and o b j e c t which f o l l o w s from X ' - t h e o r y (or C h o m s k y ' s 1965 sister condition on strict s u b c a t e g o r i z a t i o n ) is d e f i n e d at the level of argument projection. The contrast

between

( 9 1 ) and

( 1 0 0 ) makes

it

crystal

clear,

however, that both the linear and the h i e r a r c h i c a l relation of

the

a d j u n c t and the NP are r e l e v a n t , since the reason for u n g r a m m a t i c a l i t y of the examples in ( 1 0 0 ) will u l t i m a t e l y be

the the

lack of c-command A ' -position.

of

the

parasitic

gap

by

an

antecedent

( 9 1 ) a. Jan h e e f t die boeken [zonder ze/e_ te b e k i j k e n ] _t weggelegd John has those books w i t h o u t them to inspect away put ( 1 0 0 ) a . Jan h e e f t [zonder ze/*e_ te b e k i j k e n ] die boeken weggelegd John has w i t h o u t them to inspect those books away put ( 9 1 ) b. Ik ben m i j n oom [na hem/e_ j a r e n niet gezien te I am my u n c l e a f t e r him years not seen to h e b b e n gisteren weer _t tegen gekomen have yesterday again against come

-63-

in

Gaps and P a r a s i t i c Gaps ( 1 0 0 ) b . Ik ben [na hem/*e_ j a r e n niet gezien te hebben] I am a f t e r him years not seen to have gisteren weer m i j n oom tegen gekomen y e s t e r d a y again my uncle against come ( 9 1 ) c. Het is een schände dat de arts de patient [zonder It is a disgrace that the doctor the patient without hem/£ behandeld te hebben] t^ v i j f t i g gulden him treated to have fifty guilders liet b e t a l e n let pay ( 1 0 0 ) c . Het is een schände dat de arts [zonder hem/*£ It is a disgrace that the doctor without him behandeld te h e b b e n ] de patient v i j f t i g gulden treated to have the patient f i f t y guilders liet betalen let pay ( 9 1 ) d. Ik heb deze scriptie [alvorens hem/e_ d e f i n i t i e f te I have this term paper b e f o r e it d e f i n i t i v e l y to b e o o r d e l e n ] eerst aan Jan t^ voorgelegd judge f i r s t to John submitted ( 1 0 0 ) d . Ik heb [alvorens hem/*e^ d e f i n i t i e f te beoordelen] I have b e f o r e it d e f i n i t i v e l y to judge eerst aan Jan deze scriptie voorgelegd f i r s t to John this term paper submitted It t u r n s out that the p r e d i c t i o n s made by our analysis are clear in this case, and what is m o r e , they appear to be c o r r e c t . Evidence for the c o r r e c t n e s s of the hypothesis that NPs may be moved upward to the l e f t in Dutch can also be derived f r o m Exceptional Case Marking c o n s t r u c t i o n s . These c o n s t r u c t i o n s are i l l u s t r a t e d by the following e x a m p l e s . [ 1 3 ] ( 1 0 1 ) a . dat ik [„ die man het e i l a n d a f t u r e n ] zag that I that man the island along peer saw b. dat de arts [„ de p a t i e n t v i j f t i g gulden b e t a l e n ] liet that the doctor the p a t i e n t f i f t y guilders pay let

-64-

Gaps and P a r a s i t i c Gaps We shall assume an analysis of these c o n s t r u c t i o n s which d i f f e r in any i m p o r t a n t respect f r o m the analysis constructions: the

representation

in

of

select

a

not

believe-type

( 1 0 1 ) assumes

( a l t e r n a t i v e l y , the r e l e v a n t m a t r i x v e r b s

does

S'-deletion Vmax,

rather

1

than a f u l l S ) , and Case is assigned to the subject of the embedded clause by the m a t r i x verb u n d e r g o v e r n m e n t . The relevant evidence for the u p g r a d i n g movement d e r i v e s f r o m the interaction w i t h the scope of a d v e r b i a l s . In ( 1 0 2 a ) , the person using the b i n o c u l a r s is the r e f e r e n t of the m a t r i x s u b j e c t . This is e x p e c t e d , of c o u r s e , since the a d v e r b i a l PP is c o n t a i n e d in the m a t r i x clause. Let us assume that the scope of an adverbial is to the clause w i t h i n which it p r o b l e m , since it subject or

the

can

be

referent

is c o n t a i n e d .

either of

the

the

Then

referent

matrix

restricted

(102b) of

subject

the

poses

a

embedded

who is

using

binoculars. N e v e r t h e l e s s , it would be u n t e n a b l e to assume that

the

embedded clause is t r a n s p a r e n t for the scope i n t e r p r e t a t i o n of a d v e r b i a l s , since in ( 1 0 2 c ) it can only be the r e f e r e n t of the embedded clause who is using b i n o c u l a r s . ( 1 0 2 ) a . Ik zag met een v e r r e k i j k e r dat iemand het eiland aftuurde I saw w i t h b i n o c u l a r s that someone the island along peered b. Ik zag iemand I saw someone c. Ik zag iemand I saw someone

met een v e r r e k i j k e r het eiland a f t u r e n with b i n o c u l a r s the island along peer het eiland met een v e r r e k i j k e r a f t u r e n the island w i t h b i n o c u l o r s along peer

How can we account for the a m b i g u i t y of ( 1 0 2 b ) ? It will be clear that no e x p l a n a t i o n is l i k e l y to emerge if the s t r u c t u r e s of ( 1 0 2 b ) and ( 1 0 2 c ) are as in ( 1 0 3 a ) - ( 1 0 3 b ) r e s p e c t i v e l y . ( 1 0 3 ) a . Ik zag [„ iemand met een v e r r e k i j k e r het eiland a f t u r e n ] Ο

b. Ik zag [„ iemand het eiland met een v e r r e k i j k e r a f t u r e n ] No appeal can be made to the o p a c i t y e f f e c t induced by the s u b j e c t , since the a d v e r b i a l is c o n t a i n e d in the domain of the s u b j e c t in both s t r u c t u r e s . The r e s o l u t i o n of the a m b i g u i t y could be sought by

-65-

Gaps and Parasitic Gaps appealing to a rule of raising to o b j e c t , but such a rule is inconsistent w i t h some of the basic assumptions of the G B - f r a m e w o r k , s p e c i f i c a l l y the t h e t a - c r i t e r i o n and the related P r o j e c t i o n P r i n c i p l e . However, nothing prevents a movement rule in Dutch which is in its e f f e c t q u i t e similar to raising to o b j e c t , if the preceding account of possible non-adjacency of verb and o b j e c t is correct. That rule d i f f e r s from raising to object in that the landing site is an A ' - p o s i t i o n , created via C h o m s k y - a d j u n c t i o n , rather than an Α-position. The ambiguity of ( 1 0 2 b ) can be accounted for by the S-structure representations in ( 1 0 4 ) . In ( 1 0 4 a ) , the adverbial is part of the embedded clause and can only be linked to the embedded s u b j e c t ; in ( 1 0 4 b ) , the embedded subject is attached to a higher level than the adverbial, which is now part of the m a t r i x clause, thus allowing linking to the m a t r i x s u b j e c t . ( 1 0 4 ) a . Ik zag [ and some e x t r a c t i o n s are ungrammatical for all speakers, as is shown by

( 1 2 1 ) a . Wie denk je dat er t^ gekomen is? Who think you that there come is b. Wie denk je dat t^ gekomen is? Who b'.Wie Who c.*Wie Who

think you that come is denk je dat _t dat zag? t h i n k you that that saw denk je dat _t komt? t h i n k you t h a t comes

R e u l a n d assumes t h a t traces should be properly governed and that INFL is not a proper governor. The u n q u e s t i o n e d a c c e p t a b i l i t y of ( 1 2 1 a ) is a c c o u n t e d for on the a s s u m p t i o n that the subject is moved from object p o s i t i o n , which is p r o p e r l y governed by the v e r b . The s u b j e c t is placed in o b j e c t p o s i t i o n as a consequence of er-insertion ( o f . Den Besten 1 9 8 1 ) . The fact that in Dutch B

-236-

The A d v e r b i a l Pronoun er ( 1 2 1 b , b ' , c ) are ungrammatical follows from an additional r e q u i r e m e n t in the grammar of Dutch B which imposes the condition that the s t r u c t u r a l subject must be either non-empty or governed by a coindexed non-empty e l e m e n t . If we call this p r i n c i p l e X, it is claimed by R e u l a n d that the sentences in ( 1 2 1 ) d i f f e r in such a way that ( 1 2 1 a ) is in accordance w i t h both the ECP and X, ( 1 2 1 b , b ' ) are in accordance with the ECP but v i o l a t e

X,

while

(121c)

violates

both the ECP and X. A p e r t i n e n t q u e s t i o n , t h e n , is how the ECP can be applied in such a way that ( 1 2 1 c ) , but not ( 1 2 1 b , b ' ) , violates it. Like H o e k s t r a , R e u l a n d argues that the verb is moved to I N F L . If verb and INFL merge, the r e s u l t i n g constituent is able to govern the trace in subject position p r o p e r l y . This explains the ! g r a m m a t i c a l i t y of ( 1 2 1 b , b ) as far as the ECP is c o n c e r n e d . It does not explain the u n g r a m m a t i c a l i t y of ( 1 2 1 c ) . R e u l a n d ' s claim is that the movement of the verb to INFL results either in a merged category in which the p r o p e r t i e s of the are preserved as properties of

the

two

resulting

constituent category

elements or

in

a

branching category as a result of a d j u n c t i o n of the verb to I N F L , in which case the verbal properties are not properties of the whole category. It

is clear that a merger of V and INFL is

required

for

proper government of the trace in subject p o s i t i o n . R e u l a n d argues that such a merger leads to a v i o l a t i o n of the t h e t a - c r i t e r i o n in case of ( 1 2 1 c ) , but not in ( 1 2 1 b , b ' ) . In ( 1 2 1 b ) the a u x i l i a r y is moved to I N F L . As epistemic a u x i l i a r i e s do not assign t h e t a - r o l e s , the s u b j e c t will be theta-marked only once, b e f o r e the

merger.

In

( 1 2 1 b * ) , the v e r b has merged with I N F L . The structure well-formed since there are two NPs and two θ-roles. However,

is in

( 1 2 1 c ) the subject will be t h e t a - m a r k e d t w i c e , once at D - s t r u c t u r e and once a f t e r merger of V and I N F L , since n o n - a u x i l i a r y verbs are able to assign a t h e m a t i c role. In ( 1 2 1 c ) , a d j u n c t i o n of the verb to

INFL

is

required,

which

makes

extraction

of

impossible as it would violate the E C P . In

addition

that

of

this

analysis

a s s u m p t i o n s , it of all,

requires

a

number

the to

weakly

subject the

motivated

leads to a number of observational problems.

the three-way d i s t i n c t i o n is hard

does seem to be a

tendency

according

a c c e p t a b l e than ( 1 2 1 b ) , while ( 1 2 1 b ) ( 1 2 1 c ) . H o w e v e r , the s e p a r a t i o n between

-237-

to

to

substantiate.

which

i s more what is

(121a)

fact

is

First There more

acceptable than grammatical and

The A d v e r b i a l Pronoun er what is ungrammatical and between D u t c h A and B is too black and f w h i t e . ( 1 2 1 b ) is u n q u e s t i o n a b l y acceptable to all speakers of D u t c h , although R e u l a n d p r e d i c t s that this sentence is grammatical only for speakers of Dutch A. F u r t h e r m o r e , ( 1 2 1 b ) is less acceptable for all speakers of Dutch than the corresponding sentence in which expletive er is p r e s e n t . Although for all speakers of Dutch ( 1 2 1 c ) seems to be relatively a w k w a r d , the verdict of u n g r a m m a t i c a l i t y seems to be far too strong, especially in comparison with ( 1 2 1 b ) . F i n a l l y , it is predicted that a sentence such as ( 1 2 2 ) should be ungrammatical, whereas it is perfectly acceptable and even b e t t e r than ( 1 2 1 b ) . ( 1 2 2 ) Wie denk jij dat t_ in de tuin liep? Who think you that in the garden walked Summarizing, I think that none of the proposals that I have discussed w i t h respect to long extraction of the subject is fully s a t i s f a c t o r y f r o m a theoretical and descriptive point of view. D e s c r i p t i v e l y , we can distinguish between proposals that allow all long subject extractions to take p l a c e ( H o e k s t r a , K o s t e r ) . proposals that do not allow subject e x t r a c t i o n in Dutch ( C h o m s k y & L a s n i k ) and proposals that distinguish between two variants of D u t c h , Dutch A and B ( M a l i n g & Z a e n e n , T a r a l d s e n , P e s e t s k y , B e n n i s , R e u l a n d ) . Below, I shall argue that all long extractions from subject position are g r a m m a t i c a l , but that not all of them lead to acceptable sentences. This u n a c c e p t a b i l i t y is to some extent s u b j e c t to i n d i v i d u a l v a r i a t i o n but is in no way consistent enough to j u s t i f y a d i s t i n c t i o n between Dutch A and Dutch B. T h e o r e t i c a l l y , all proposals create problems for the general theory of e x t r a c t i o n . I shall d e m o n s t r a t e that an approach based on the Gap Condition and the pragmatic constraints f o r m u l a t e d in section 3 - 5 provides a simple and consistent analysis of the Dutch problem.

-238-

The A d v e r b i a l P r o n o u n er 3 . 6 . 1 . 2 Subject e x t r a c t i o n a n d t h e G a p C o n d i t i o n In chapter 1 it

has been argued that s u b j e c t - o b j e c t a s y m m e t r i e s can

be accounted for q u i t e e l e g a n t l y by an a p p r o a c h

based

on

Kayne's

Connectedness C o n d i t i o n . The Gap C o n d i t i o n was i n t r o d u c e d as a more r e s t r i c t i v e v e r s i o n of K a y n e ' s c o n d i t i o n . Let us now investigate whether the Gap C o n d i t i o n allows an e x p l a n a t i o n of the COMP-_t phenomena and in p a r t i c u l a r of the p o s s i b i l i t y in D u t c h to

extract

a wh-phrase from s u b j e c t p o s i t i o n . A first assumption that has to made is that INFL is a governor that is in no way d i s t i n c t from other g o v e r n o r s . In most analyses INFL is considered to be a g o v e r n o r , but not a p r o p e r governor. This d i s t i n c t i o n is i n t r o d u c e d in order to be able to a c c o u n t for the fact that long e x t r a c t i o n of the s u b j e c t is severely c o n s t r a i n e d , whereas no such r e s t r i c t i o n s

are

extraction. N a t u r a l l y , a theory in

no

which

found

with

stipulated

object internal

division in p r o p e r and n o n - p r o p e r governors is made p r e f e r r e d . If INFL is a ( p r o p e r ) governor like other

is to be governors,

such as V, the s u b j e c t - o b j e c t a s y m m e t r y with respect to should receive a d i f f e r e n t e x p l a n a t i o n .

extraction

In f a c t , most of the COMP-^ phenomena f o l l o w d i r e c t l y from the Gap C o n d i t i o n . In languages like English and French the s u b j e c t precedes I N F L . This is g e n e r a l l y c o n s i d e r e d to be a r e f l e c t i o n of the base rule S —» NP INFL VP. In our a p p r o a c h one might say that INFL assigns Case to the l e f t , w h i c h i m p l i e s that e x t e r n a l arguments should appear to the l e f t of I N F L to s a t i s f y the Case r e q u i r e m e n t . On the other h a n d , the v e r b assigns Case to the right in these languages. It follows that canonical goverment is from l e f t to right in French and E n g l i s h . It also f o l l o w s t h a t the subject p o s i t i o n is governed by I N F L , but not c a n o n i c a l l y . The Gap Condition

requires

extraction

to

take

place

from

canonically

governed p o s i t i o n s . C o n s e q u e n t l y , e x t r a c t i o n f r o m s u b j e c t position would v i o l a t e the Gap C o n d i t i o n . Only by means of additional mechanisms,

such

as

the

que -*· qui-rule

or

deletion

of

c o m p l e m e n t i z e r can s u b j e c t s be e x t r a c t e d . The COMP-t^ e f f e c t thus be c o n s i d e r e d a d i r e c t c o n s e q u e n c e of the Gap C o n d i t i o n . This e x p l a n a t i o n of the COMP-t_ e f f e c t in resembles the explanation of the

-239-

the can

E n g l i s h and French impossibility of

The A d v e r b i a l Pronoun er p r e p o s i t i o n - s t r a n d i n g in D u t c h , which was extensively discussed c h a p t e r 1. In these two cases

the

direction

of

in

Case

assignment

d i f f e r s from the d i r e c t i o n of canonical government. In d i r e c t i o n of canonical government is from right

Dutch the to left.

Prepositions assign case from l e f t to r i g h t . Only in the case of postpositions does P - s t r a n d i n g appear to be possible. It follows that extraction of the subject in Dutch

can

be

predicted

possible. As indicated b e f o r e , I shall assume that

INFL

to

in

be

Dutch

appears in a position following S. Arguments in favour of this claim can be found in Hoekstra ( 1 9 8 4 ) . INFL governs an NP w i t h i n S, if it

appears outside the argument p r o j e c t i o n of V. Given the

fact

that INFL governs from right to l e f t , it canonically governs that NP. It thus follows that extraction of the subject is allowed as far

as

the

Gap Condition

is

concerned.

This

situation

is

illustrated i n ( 1 2 3 ) . INFLmax(S')

(123)

INFL

V

No merger between V and INFL ( H o e k s t r a , R e u l a n d ) or variation domain for the application of the NIC (Koster) is necessary a r r i v e at an analysis in which long extraction of the subject Dutch is a l l o w e d . With respect to I t a l i a n , it follows

that

extraction

of

in to in the

s u b j e c t is possible o n l y from postverbal p o s i t i o n . Italian is an VO language, just like French and English. It is therefore predicted that a n o m i n a t i v e NP can be e x t r a c t e d only if it follows I N F L . In preverbal p o s i t i o n the NP precedes INFL and extraction would violate the Gap C o n d i t i o n . In p o s t v e r b a l position INFL governs the n o m i n a t i v e NP c a n o n i c a l l y , in

which

case

the

Gap C o n d i t i o n

is

s a t i s f i e d . N o t e t h a t this analysis of Italian subject extraction d i f f e r s c o n s i d e r a b l y from the analyses presented in R i z z i ( 1 9 8 2 ) and -240-

The A d v e r b i a l P r o n o u n er C h o m s k y ( 1 9 8 1 ) . They argue that e x t r a c t i o n is possible since the postverbal NP is

f r o m postverbal governed

by

position

V.

This

necessary in their approach b e c a u s e INFL is not a proper

is

governor.

Only if the subject is governed by V is the ECP s a t i s f i e d . Such

an

analysis requires a government d e f i n i t i o n in which the a d j o i n e d subject is part of the g o v e r n m e n t domain of the v e r b . Such a definition

is p r o v i d e d by Aoun & S p o r t i c h e ( 1 9 8 3 ) ,

who

claim

the upper b o u n d a r y of government is the maximal p r o j e c t i o n

of

that the

governor. In chapters 1&2 it has been argued that government should be r e s t r i c t e d to the argument p r o j e c t i o n of the g o v e r n o r , i.e. X'. In our approach the postverbal s u b j e c t is not governed by the v e r b . Instead, it is governed by I N F L , similar to the way in which the nominative NP in Dutch is governed by I N F L . In f a c t , the r e s u l t i n g s t r u c t u r e appears to be the m i r r o r image of ( 1 2 3 ) . In postverbal position the s u b j e c t is c a n o n i c a l l y governed by INFL and e x t r a c t i o n is allowed by

the

Gap C o n d i t i o n

(cf.

ch.4.5.3

for

a

further

elaboration of this analysis w i t h respect to the o c c u r r e n c e of n e ) . The fact that the COMP-_t e f f e c t seems to be present in I t a l i a n t h e r e f o r e an interesting c o n f i r m a t i o n of the theory p r e s e n t e d

is in

chapter 1 . If the COMP^t e f f e c t is considered to be a consequence of the Gap C o n d i t i o n , it follows that we expect it to be absent in those languages in which the linear o r d e r of INFL and NPnom is similar to the order of V and 0. One of these languages is D u t c h , as discussed above. Other languages in which the two orders are similar are German and Old English. Long s u b j e c t e x t r a c t i o n is p e r m i t t e d in Old English, as is i l l u s t r a t e d in ( 1 2 4 ) ( c f . A l l e n 1 9 7 7 , 1 9 8 0 , Pesetsky 1982) . ( 1 2 4 ) Ac ic wolde w i t a n hu 3e S u h t e be 3aem monnum 3e but I would know how thee seemed the man that wit aer cwaedon 3aet unc 3uhte 3zt _t wzron we earlier said that us seemed that were w i l d d i o r u m g e l i c r a n Sonne monnu wild-beast like-er than men 'But I would like to know how it

seemed to you about

the men that we said e a r l i e r that were more like wild beast than m e n 1 ( B , X X V I I I . 5 p . 1 2 2 . 1 3 / A l l e n 1 9 8 0 , ( 3 ) )

-241-

The A d v e r b i a l Pronoun er A c c o r d i n g to Bayer ( 1 9 8 4 ) , COMP-J^ phenomena are not allowed in Standard G e r m a n , c o n t r a r y to what

we might

expect.

t u r n s out that long e x t r a c t i o n f r o m o b j e c t p o s i t i o n

However, is

it

impossible

as w e l l . This is shown in ( 1 2 5 ) . ( 1 2 5 ) a . * W e r glaubst du dass t_ Emma l i e b t ? Who t h i n k you that

Emma loves

b . * W e n g l a u b s t du dass Emma _t l i e b t ? Whom t h i n k you that Emma loves No s u b j e c t - o b j e c t a s y m m e t r y w i t h respect to long

extraction

shows

up. I n t e r e s t i n g l y , B a y e r ( 1 9 8 U ) shows that i n t h e Bavarian d i a l e c t of German b o t h long s u b j e c t and long o b j e c t e x t r a c t i o n are a l l o w e d , as in ( 1 2 6 ) . [ 2 5 ] . ( 1 2 6 ) a . Wer m o a n s t du dass _t d ' Emma mog? Who t h i n k you that Emma loves b. Weam moanst du dass d ' E m m a _t mog Whom t h i n k you that Emma loves The COMP-_t e f f e c t does not show up in t h i s

dialect.

Since

German

and Old English d i s p l a y b a s i c a l l y the same order of c o n s t i t u e n t s as D u t c h , this is no s u r p r i s e . I shall r e t u r n to German and Old English in the n e x t c h a p t e r . A more c o m p l i c a t e d s i t u a t i o n arises in Icelandic. As d e m o n s t r a t e d by M a l i n g & Z a e n e n ( 1 9 7 8 ) , I c e l a n d i c allows COMP-_t c o n f i g u r a t i o n s , as in ( 1 2 7 ) . ( 1 2 7 ) a . Hver sag5ir 3u a3 _t vseri k o m i n n til

Reykjavikur?

who said you t h a t was come to R e y k j a v i k b. Hver sagSir u a5 _t h e f ö i boröaö Setta epli who said you that had eaten this apple At f i r s t sight I c e l a n d i c seems to be a language w i t h the order N P - I N F L - C y p V - 0 ] , j u s t like E n g l i s h , F r e n c h . Norwegian etc. I f this is c o r r e c t , we e x p e c t t h a t e x t r a c t i o n of the s u b j e c t violates the Gap C o n d i t i o n . H o w e v e r , Icelandic observes a verb-second c o n s t r a i n t both in m a i n and

subordinate

clauses.

-242-

This

is

demonstrated

in

The A d v e r b i a l Pronoun er (128). ( 1 2 8 ) a . I dag var O l a f u r f a r i n n til

Islands

today was Olaf gone to Iceland b. Hun sagSi a3 i dag vsri O l a f u r f a r i n n she said that today was Olaf gone til Islands to Iceland These f a c t s i n d i c a t e that the basic order in I c e l a n d i c might be something like X P - I N F L - [ „ N P [ V , V - 0 ] ] . If this is t r u e , the s u b j e c t NP is canonically g o v e r n e d by INFL in its original position. E x t r a c t i o n of the s u b j e c t NP would then proceed in a c c o r d a n c e w i t h the Gap C o n d i t i o n . [26].

3 . 6 . 1 . 3 Subject e x t r a c t i o n and the EPC In the preceding section it has been argued t h a t as far as syntax is c o n c e r n e d long e x t r a c t i o n of the s u b j e c t is a l l o w e d . We may then wonder what causes the low degree of a c c e p t a b i l i t y of ( 1 2 9 a ) , in p a r t i c u l a r when we compare this sentence to ( 1 2 9 b ) . (129)a.??Wie Who b. Wie Who

denk je dat t körnt? t h i n k you that comes d e n k je dat er t_ körnt? t h i n k you t h a t there comes

As it stands, this c o n t r a s t f o l l o w s d i r e c t l y f r o m the E m p t y P r e s u p p o s i t i o n C o n d i t i o n , i n t r o d u c e d i n section 5 . 3 , ( 1 0 6 ) . I f at least one c o n s t i t u e n t is to belong to p r e s u p p o s i t i o n , the trace or r a t h e r the f r o n t e d w h - p h r a s e should be i n t e r p r e t e d in t h i s w a y . This is in clear c o n f l i c t w i t h the n a t u r e of w h - c o n s t i t u e n t s in q u e s t i o n s . As they s h o u l d be i n t e r p r e t e d as p a r t of f o c u s , the EPC is v i o l a t e d in ( 1 2 9 a ) . The i n t r o d u c t i o n of e x p l e t i v e er in ( 1 2 9 ) as a m i n i m a l f i l l e r of the p r e s u p p o s i t i o n makes the s e n t e n c e p e r f e c t l y acceptable.

-243-

The Adverbial Pronoun er The data r e l a t i n g to these subject e x t r a c t i o n s in the case of questions resemble the data on impersonal passives very closely. It is clear that the f r o n t e d wh-phrase cannot be i n t e r p r e t e d as part of p r e s u p p o s i t i o n , while in impersonal passives there is no subject at all. In both cases something else is r e q u i r e d to satisfy the E P C . I shall f i r s t demonstrate that the same acceptability judgements hold in this construction as in impersonal passives. It will then be shown that no such restrictions can be found if the extracted constituent can be interpreted as belonging to presupposition, which is the case in relative clauses. If the sentence contains -in addition to the extracted subject- only a verb and a P P , the PP must appear p r e v e r b a l l y . Consider the sentences in ( 1 3 0 ) . ( 1 3 0 ) a . ? ? W i e denk je dat J; rekent op salarisverhoging? Who think you that counts on rise in salary b. Wie denk je dat _t op salarisverhoging r e k e n t ? Who think you that on rise in salary counts c. Wie denk je dat er t^ op salarisverhoging Who t h i n k you that there on rise in salary r e k e n t / r e k e n t op salarisverhoging? c o u n t s / c o u n t s on rise in salary The contrast is brought about by the fact that the PP in ( 1 3 0 a ) must be i n t e r p r e t e d as part of focus, as a consequence of its postverbal occurrence. In ( 1 3 0 b ) the PP can be interpreted as part of p r e s u p p o s i t i o n . If the verb is t r a n s i t i v e , a reverse e f f e c t can be observed. The more c l e a r l y the object belongs to the p r e s u p p o s i t i o n , the less acceptable it is to insert expletive er. If the o b j e c t belongs to focus q u i t e c l e a r l y , there exists a strong tendency to insert er. (131)a.

Wie Who a'.??Wie Who b. ?Wie Who

denk je dat _t het leest? t h i n k you that it reads denk je dat er t_ het leest? think you that there it reads denk je dat _t een boek van Reve leest? t h i n k you that a book of Reve reads

-244-

The A d v e r b i a l P r o n o u n er b1.

Wie denk je dat er _t een boek van Reve leest? Who t h i n k you that t h e r e a book of Reve reads

These facts f o l l o w from the EPC and e x p l e t i v e er is considered to be a

from dummy

the

analysis

pronoun

empty p r e s u p p o s i t i o n . In ( 1 3 1 a ) the object is a

in

that

which

fills

pronoun.

an

Pronouns

belong to p r e s u p p o s i t i o n . It f o l l o w s that the i n s e r t i o n of er leads to a c o n t r a d i c t i o n , since i n s e r t i o n of er e n t a i l s that the

pronoun

het is not part of p r e s u p p o s i t i o n . In ( 1 3 1 b ) the i n d e f i n i t e

object

belongs to focus in the unmarked case. The sentence with er f i l l i n g the p r e s u p p o s i t i o n is u n m a r k e d . A very s t r i k i n g argument against

an

analysis

in

which

the

occurrence of e x p l e t i v e er is r e l a t e d to the p o s s i b i l i t y of e x t r a c t i o n is provided by the fact that t h e r e is a clear d i f f e r e n c e between d i f f e r e n t i n s t a n t i a t i o n s of w h - m o v e m e n t . If the s u b j e c t

is

e x t r a c t e d i n T o p i c a l i z a t i o n , R e l a t i v e Clause f o r m a t i o n a n d C l e f t i n g s t r u c t u r e s , e x p l e t i v e er

does

not

show

up.

Examples

of

these

c o n s t r u c t i o n s can be f o u n d in B r a c h i n ( 1 9 7 3 ) and P a a r d e k o o p e r ( 1 9 7 1 ) . ( 1 3 2 ) a . Die man denk ik niet dat _t_ veel haast zal maken That man t h i n k I not that much haste w i l l make ' T h a t man I d o n ' t t h i n k w i l l h u r r y much b. v r i e n d e n die h i j weet dat _t met genoemde f r i e n d s who he knows that w i t h a f o r e m e n t i o n e d f i r m a zaken doen f i r m business do ' f r i e n d s who he knows do business w i t h the aforementioned f i r m ' c. Het is Piet die ze zeggen dat t_ zou komen it is Pete who they say that would come In those cases e x p l e t i v e er is not i n s e r t e d , whatever of the sentence from

which

the

subject

is

d e m o n s t r a t e d i n ( 1 3 3 ) . [27]. (133)a.

de jongen die ik denk dat _t komt the boy who I t h i n k t h a t comes

-245-

the

extracted.

content This

is

The A d v e r b i a l Pronoun er a ' . ? ? d e jongen die the boy who I b. de j o n g e n die the boy who I b ' . ? ? d e jongen die the boy who I c. de jongen die the boy who I

ik denk dat er _t komt t h i n k that there comes ik denk dat t_ naar huis gaat t h i n k that home goes ik denk dat er _t naar huis gaat think that there home goes ik denk dat _t op school t h i n k that at school

een boek leest a book reads c ' . ? ? d e jongen die ik denk dat er _t op school the boy who I t h i n k that there at school een boek leest a book reads In the c o r r e s p o n d i n g questions in (13Ό the appearance of expletive er is r e q u i r e d ( 1 3 ^ a ) or p r e f e r r e d ( 1 3 4 b , c ) . ( 1 3 ^ ) a . Wie Who b. Wie Who c . Wie Who

d e n k je dat er t h i n k you that d e n k je dat er t h i n k you that denk je dat er t h i n k you that

t_ komt? there comes ;t naar h u i s gaat? there home goes t_ op school een boek leest? there at school a book reads

The f a c t s in ( 1 3 2 ) and ( 1 3 3 ) f o l l o w immediately from our theory. The s u b j e c t NP moved by wh-movement is a d e f i n i t e p r o n o u n . Since d e f i n i t e p r o n o u n s belong to p r e s u p p o s i t i o n , er does not appear in its expletive u s e . In those t h e o r i e s in which e x t r a c t i o n of the s u b j e c t and the a p p e a r a n c e of dummy p r o n o u n s are c o r r e l a t e d , these f a c t s are v e r y p r o b l e m a t i c a l , since q u e s t i o n formation and r e l a t i v i z a t i o n are i n s t a n t i a t i o n s of the same r u l e schema. If this was to be given up, it would a f f e c t the move alpha approach very seriously. It t u r n s out t h a t a s e p a r a t i o n of the s y n t a c t i c conditions that allow e x t r a c t i o n of the s u b j e c t and the conditions that govern the a p p e a r a n c e of e x p l e t i v e er is strongly suggested. The more so since the two relevant c o n d i t i o n s , i.e. the Gap Condition and the -246-

The A d v e r b i a l P r o n o u n er E P C , can be supported i n d e p e n d e n t l y of the COMP-_t p h e n o m e n a . 3 - 6 . 2 S u b j e c t e x t r a c t i o n i n main clauses If the analysis presented in the last section is c o r r e c t , we expect the occurrence of e x p l e t i v e er in case of short e x t r a c t i o n of the subject under exactly the same conditions as w i t h long e x t r a c t i o n . This seems indeed to be the case, as is i l l u s t r a t e d in ( 1 3 5 ) . ( 135)a.??Wie Who a ' Wie Who b. ?Wie Who 1 b Wie Who

komt t_? comes komt er t_? comes there komt _t naar h u i s ? comes home komt er t_ naar h u i s ? comes t h e r e home

c. ?Wie koopt t_ een boek van Reve? Who c ' Wie Who d. Wie Who d ' ?Wie Who e. Wie Who e'??Wie Who

buys a book of Reve koopt er _t een boek van Reve? buys t h e r e a book of Reve koopt _t dat boek van Reve? buys that book of Reve koopt er t_ dat boek van Reve? buys t h e r e that book of Reve koopt _t het? buys it koopt er t_ het? buys t h e r e it

These o b s e r v a t i o n s i n d i c a t e once more that the r e s t r i c t i o n s on the a p p e a r a n c e of e x p l e t i v e er are c o m p l e t e l y u n r e l a t e d to c o n d i t i o n s on e x t r a c t i o n . In a s t a n d a r d approach these simple e x t r a c t i o n s of the s u b j e c t are u n p r o b l e m a t i c a l . The f a c t t h a t the a p p e a r a n c e of er in these s i m p l e q u e s t i o n s m a t c h e s the p a t t e r n of long e x t r a c t i o n c o n s t i t u t e s an a r g u m e n t in f a v o u r of the a p p r o a c h a d o p t e d h e r e , in which the a p p e a r a n c e of e x p l e t i v e er is p r a g m a t i c a l l y governed. [28],[29]-247-

The Adverbial Pronoun er

3.7 Conclusion The main objective of this chapter has been to show that er is

not

a dummy pronoun that fills or binds the subject position for syntactic reasons. In doing so, we have had to discuss a number of rather complicated questions that are associated with the d i f f e r e n t f u n c t i o n s of er. Although not all the problems with respect to er have been solved, it has been shown that on the basis of a f a i r l y small number of i n d e p e n d e n t l y motivated assumptions we can present a coherent explanation of a large number of problems. The central hypothesis put forward in this chapter is that R-pronouns can be characterized as elements that are specified as [ p p + R , + θ , + A ] . The p r o p e r t i e s associated with the f o u r different f e a t u r e combinations correspond quite nicely to the four different types of er that are t r a d i t i o n a l l y d i s t i n g u i s h e d . This implies that we do not have to distinguish between f o u r separate pronouns er, each w i t h its own properties. The characterization predicts four d i f f e r e n t types to be possible, the p r o p e r t i e s of which follow from t h e d i f f e r e n t f e a t u r e compositions. A second important hypothesis is that the fact that a p p a r e n t l y one occurrence of er may have more than one f u n c t i o n can be considered to be a consequence of a phonologically motivated deletion r u l e . It has been argued that this rule applies at a level at which it does not a f f e c t the i n t e r p r e t a t i o n . This implies that at the level of i n t e r p r e t a t i o n several adjacent occurrences of er are possible. T h i r d l y , it has been argued that no rule of R-movement nor base-generated R-positions are required to account for the surface position of R - p r o n o u n s . The i n d e p e n d e n t l y m o t i v a t e d rule of S-internal a d j u n c t i o n is s u f f i c i e n t . As a consequence of the central objective of this c h a p t e r , a great deal of a t t e n t i o n has been paid to the expletive f u n c t i o n of er. Given its n o n - t h e m a t i c , non-argument status it is a dummy pronoun with a p r a g m a t i c f u n c t i o n . It f i l l s an empty presupposition in order to p r e v e n t a violation of the Empty Presupposition Condition ( E P C ) . From this analysis it follows that expletive er cannot be considered to f i l l empty s u b j e c t positions. This analysis

-248-

The A d v e r b i a l P r o n o u n er of expletive er has been shown to have several i n t e r e s t i n g consequences, in p a r t i c u l a r when compared to existing s y n t a c t i c analyses of this type of er. F i n a l l y , it

has been demonstrated that a

Gap C o n d i t i o n , introduced in chapter 1, and account for the long s t a n d i n g Dutch.

problem

-249-

of

combination the

the

EPC is COMP-t^

of

the

able

to

effect

in

The A d v e r b i a l Pronoun er

NOTES 1 . The asterisks and question marks in the examples in this chapter do not imply that the relevant examples are ungrammatical or less grammatical for syntactic reasons. They refer to unacceptable or less acceptable sentences. In semantic terms the unacceptability of the sentences in (3) is due to a tautology. In the sections 3· 5 and 3-6 the asterisks and question marks often indicate that the relevant examples are unacceptable for pragmatic reasons. 2. Coppen(1985) argues that quantitative er is a determiner of N 1 ' , which is moved from NP. He provides independent evidence in favour of his claim by assuming that the following two sentences are grammatical. (i) Boeken hebben we slechts een paar gelezen (ii) Mooie boeken heb ik maar drie gezien I have no doubt at all that these sentences are fully ungrammatical. 3. As indicated above, the first er, the quantitative er, cannot be replaced by other R-pronouns, since it lacks a paradigm, for reasons to be discussed below . 4. Schwa consonant deletion is a rather marked proces in agentive formation. Some verbs such as kliederen(to make a mess), in addition to the 'normal' agentive form kliederaar , which is formed by schwa-strengthening, allow a variant formed by schwa-consonant deletion, klieder. The same variation can be seen with the agentive formation of the verb babbelen(to chatter). Both babbelaar and babbel are possible variants with semantic specialization. 5. Den Besten (1983) observes that there is a phenomenon similar to contraction in Afrikaans. It involves contraction of two adjacent nie's.

er-er

6. Van Riemsdijk assumes that the ungrammaticality of (5^b) follows from the stipulation that absolute met does not allow R-movement, i.e. does not select a [+R] position, on a par with prepositions such as zonder. It is clear that this assumption is unattractive given that met allows R-movement in other constructions, i.e. in the case of intrumental or comitative met. Moreover, it turns out that absolute met allows R-movement if met is followed by a single NP argument. Van Riemsdijk argues that met in (ia) is absolute. It then follows that according to his analysis (ib) should be just as ungrammatical as (55b), contrary to fact. ( i ) a . [Met dit siechte weer] kun je beter thuis blijven With this bad weather can you better home stay b. Dit is het soort weer waarmee je beter thuis kunt blijven This is the type of weather where-with you better home can stay For a discussion on this and related issues I refer to above .

the

articles mentioned

7. The claim that prepositional R-pronouns are PPs is a contradictio in terminis. It would be more correct to label them pro-PP. For the sake of convenience I will stick to the usual term R-pronoun.

-250-

The A d v e r b i a l Pronoun er

8. There are two exceptions to this claim. If the empty head is preceded by an adjective or immediately followed by a partitive PP quantitative er does not appear, as in (i): (i)a. Ik heb een rode gekocht I have a red bought b. Ik heb twee van deze boeken gekocht I have two of these books bought c. Van deze boeken heb ik *(er) twee gekocht Of these books have I there two bought For a more detailed discussion of the distribution of quantitative er, to Blom(1977), Bennis(1977,1979).

I

refer

9. The ungrammaticality of (60b) with er holds for quantitative er. Er can of course be interpreted as a locative R-pronoun in this sentence. 10. As far as I know, all proposals concerning quantitative er explicitly or implicitly consider quantitative er to be homonymous with the weak form of the R-paradigm which we find with prepositional and locative er. Even Kirsner(1979), whose theoretical framework is based on the direct correspondence between form and content, acknowledges that quantitative er might be an accidental homonym. One of the arguments is that historically quantitative er developed from a different source, a genitive form of the third person singular, iro. In spite of this, I shall try to derive the properties of quantitative er from the assumption that quantitative er is an R-pronoun with basically the same properties as the other instances of R-pronouns. 11. Note that this analysis crucially differs from the analysis of quantitative ne presented by Belletti and Rizzi(198l). They argue that a quantified NP may contain PRO if the NP is not governed. If it is governed, ne should be moved from NP to clitic position. Their approach does not fit the approach adopted here for several reasons, such as their definition of government which is comparable to the one proposed in Aoun & SporticheC1983). I shall present a detailed alternative to the Italian quantitative construction below (ch.4.5.3). I shall argue that PRO within a quantified NP must be bound if the NP is canonically governed. 12. Van Riemsdijk's Subjacency approach to the double-R-constraint is further supported by the following three arguments: - if a locative and a prepositional er occur in one sentence only one er may appear and it must be interpreted as the object of the preposition, i.e.the locative reading is not available. (i) *Zij heeft er. er. vaak _t. [t_. over] gesproken She has there theVe often abcJut talked This fact indicates that only one R-pronoun can appear in the R-position. If one er is left out, the R-pronoun must be interpreted as the obligatory prepositional R-pronoun and cannot appear as an optional locative R-pronoun. - in case of multiple wh-questions in which both wh-pronouns are R-pronouns, the R-pronoun in COMP must be the locative R-pronoun, again for reasons of Subjacency.

-251-

The Adverbial Pronoun er

(ii)a. Ik vraag mij af waar. zij waar.j vaak t^. [t_. over] gepraat heeft I wonder where she where often about talked about b.*Ik vraag mij af waar. zij waar. vaak _t. [t_. over] gesproken heeft - two prepositional R-pronouns in the same domain are excluded, no matter how they are associated with the two prepositional phrases (iii)a. Dit is het boek waar. ik gisteren [Jt. voor] naar de bibliotheek gegaan ben This is the book where I yesterday for to the library gone have b. Dit is de bibliotheek waar. ik gisteren voor dat boek [_t.J naar toe] gegaan ben ^ This is the library where I yesterday for that book to gone have c.*Dit is het boek waar. ik er. gisteren [_t. voor] [t_. naar toe] J J gegaan ben d.*Dit is de bibliotheek waar. ik er. gisteren [_t. voor] [_t. naar toe] gegaan ben J The facts in (iii) follow if it is assumed that there is only one R-position within S. In (iiic.d) the R-position is occupied by er, which implies that movement to COMP would have to proceed in one step, in violation of Subjacency. The first and the third argument follow from our theory as well. The ungrammaticality of (i) is due to the filter on two adjacent er's (section 2.6). The ungrammaticality of (iiic,d) follows from the Gap Condition, since it is impossible for the verb to govern two PPs canonically. The second argument will be discussed below. 13. Arguments against the double-R-constraint that are in several respects similar to the ones presented here can be found in Bennis(1980a,fnt.11) and Van Bart & Kager (1984). Stowell(198l,ch.7) develops a different theory of preposition stranding, based on the idea that stranding is possible only if the preposition is reanalysed as part of a complex verb. Stowell argues that R-pronouns are NPs that are inherently Case-marked. He proposes the following general principle: 'If a preposition subcategorizes for an NP object, then it must always assign Case to that NP, if the NP appears in a position of Case assignment.'(p.468). Given the fact that in Dutch prepositions assign Case to the right, it follows that R-pronouns have to precede prepositions. His account of the double-R-constraint is rather unsatisfactory, since it requires a verb position as the first constituent of the VP. As has been argued in ch.1.5.2 and more extensively by Hoekstra(1984) and Koopman(1984), such an analysis is completely unmotivated and so is his alternative account of the double-R-constraint in terms of a condition on complex verbs. One element in his account may be worth discussing in more detail. Stowell argues that a problem for Van Riemsdijk's R-position approach arises in cases such as (i). (i) a. Waar. Where b.*Waar. Where

probeert zij op blote voeten _t^ in tries she in bare feet in probeert zij er op blote voeten _t. tries she there on bare feet in to

te klimmen to climb in te klimmen climb

The unacceptability of (ib) should be accounted for ungrammaticality of (6?c), i.e. as a regular case of the adposition in in (i) is a motional postposition, (ii), the R-position account does not provide non-wellformedness of (ib).

-252-

in the same way as the the double-R-constraint. If just as it seems to be in an explanation for the

The A d v e r b i a l Pronoun er

(ii) Zij probeert de boom in te klimmen She tries the tree into to climb However, it appears to be the case that there is no reason to consider in in (i) to be a motional postposition, as in in (ii). If the R-pronoun in (i) belongs to a regular PP, Van Riemsdijk's double-R-constraint can be maintained in this respect. The example provided by Stowell is somewhat unfortunate. If we replace the postpositional PP in (ii) by a prepositional one, as in (iii), the sentence can still be interpreted as expressing a motional aspect, (iii) is ambiguous between a motional and a locative reading. (iii) Zij probeert in de boom te klimmen She tries into the tree to climb Consequently, the sentences in (i) may equally well correspond to (iii), in which case Stowell's argument loses it force. Moreover, it is possible to provide examples in which only a postpositional PP may express a motional aspect, whereas the corresponding prepositional PP can be interpreted as locative only. Examples are given in (iv). (iv)a. Jan is de berg op gelopen John is the mountain up walked b.*Jan is op de berg gelopen John is on the mountain walked c. Jan heeft op de berg gelopen John has on the mountain walked d.*Jan heeft de berg op gelopen John has the mountain on walked The PP in (iva) can only be interpreted as motional, while the PP in (ivc) is locative. These sentences are interesting because the selection of the perfective auxiliary verb is dependent on the selection of a constituent expressing a motional aspect. For an extensive discussion of this matter, I refer to Hoekstra(1984,ch.3.2.5 & 4.4). Whatever the reason for this difference, it follows from (iv) that the appearance of a motional postposition correlates with the appearance of the perfective auxiliary zijn. This provides us with a diagnostic test for the appearance of R-pronouns in relation to motional postpositions. From the relative clauses corresponding to (iv) it follows that only prepositional PPs allow R-pronouns, while the relatives with motional postpositions are formed in a way analogous to direct object relatives, suggesting that these apparent postpositional PPs are in fact NPs followed by a complex verb. This is illustrated in (v). (v)a. De berg die Jan op gelopen is The mountain which John on walked b.*De berg waar Jan op gelopen is The mountain where John on walked c.*De berg die Jan op gelopen heeft The mountain which John on walked d. De berg waar Jan op gelopen heeft The mountain where John on walked

is is has has

These facts illustrate that it is indeed the case that motional postpositions do not allow R-pronoun complements. Consequently, the facts presented by Stowell do not constitute a problem for the double-R-constraint.

-253-

The A d v e r b i a l Pronoun er

11. Note that I have changed light-NP-shift into light-XP-shift here. The rule was labelled light-NP-shift or S-internal NP-adjunction in ch.1.5.2 because we were concerned with the movement of NP and because we considered this rule to be the Dutch counterpart of heavy-NP-shift in English. There is no reason to suppose that the rule of adjunction should be restricted to NP. To account for the relatively free word order in the Dutch 'Mittelfeld' rules of PP-adjunction and ADV-adjunction are required as well. 15. The construction with expletive er is essentially different from the English there-insertion construction. One of the main differences is that the occurrence of expletive er is not dependent on the choice of the verb. While in English the construction is restricted to sentences with the verb to be and a small class of similar 'existential' verbs, in Dutch this construction appears with all sorts of verbs. 16. I shall not be concerned here with precise definitions of definite/indefinite, which are without any doubt relevant to the construction under discussion. I shall merely assume that there exist semantic criteria that determine the definiteness of NPs. 17. To account for facts of the type in (83c) one might propose a filter that excludes configurations in which the complementizer immediately precedes the finite verb. Such a filter is proposed in Roelfszema (1983). If the filter does not take into account intervening traces, it applies equally to that-t^ configurations in English and in Dutch. In fact, Roelfszema's argumentation is based on that-t^ configurations exclusively. However, it is clear that only a subset of the relevant data fall under such an analysis. Given that such a filter is completely ad hoc and does not include related phenomena, I shall not discuss this proposal in any detail. 18. In fact, the correct generalization is that [-spec] indefinite NPs cannot be interpreted as part of the presupposition. In a sentence such as ( i ) , the indefinite NP is interpreted as [+specific]. (i) Jan zag dat een wielrenner viel John saw that a racing cyclist fell A similar observation applies to sentences of the type in (ii). (ii)a. De leraar wilde dat een jongen een meisje kuste The teacher wanted that a boy a girl kissed b. De leraar wilde dat er een jongen een meisje kuste The teacher wanted that there a boy a girl kissed In (iia) the first indefinite NP is interpreted as [+specific] indicating that a particular boy should kiss some girl. In (iib) the first NP is interpreted as [-specific], implying that some boy should kiss some girl. We can test the differential status of these NPs by coreference tests of the type discussed in Reinhart (1983). Definite pronouns may refer to [+spec] NPs but not to [-spec] NPs. This difference can be observed in (iii). (iii)a. De leraar wilde dat een jongen een meisje kuste, maar hij weigerde The teacher wanted that a boy a girl kissed but he refused b.*De leraar wilde dat er een jongen een meisje kuste, maar hij weigerde The teacher wanted that there a boy a girl kissed, but he refused

-254-

The A d v e r b i a l Pronoun er

The unacceptability of (iiib) indicates that the NP een jongen cannot be interpreted as [+spec]. A reverse effect is obtained in (iv). (iv)a.*De leraar wilde dat een jongen een meisje kuste, maar er bood zieh The teacher wanted that a boy a girl kissed, but there volunteered niemand aan nobody b. De leraar wilde dat er een jongen een meisje kuste, maar er The teacher wanted that there a boy a girl kissed, but there bood zieh niemand aan volunteered nobody The unacceptability of (iva) indicates that the NP een jongen cannot be interpreted as [-spec]. 19. Kirsner (1979) argues that expletive er appears solely to shift the participant-in-focus to the background. Although Kirsner's central hypothesis concerning expletive er seems correct to me, I disagree with him in most other respects. For a discussion of Kirsner(1979), see Bennis(1982). 20. Under this approach the unacceptability of (i) unacceptability of (iia).

is

similar to

the

(i)

*dat een jongen er werkte that a boy there worked (ii)a.*dat Jan gisteren ut zag that John yesterday it saw b. dat Jan ut gisteren zag 21. Haling and Zaenen eventually argue against (111a), since (112) does not follow from trace theory. The exact formulation of their account is irrelevant here. What is important is that they suppose that there is a correlation between dat-_t phenomena and the obligatoriness of dummy pronouns. 22. A similar distinction between Dutch A and B is adopted in Safir(1985), who argues that Dutch A is a NOM-drop language, implying that, just as in German and Italian, nominative Case need not be phonetically realized, whereas in Dutch B nominative Case must be phonetically realized where it is assigned. 23. (1l6b) is grammatical if the sentence is interpreted with a double negation, in which case nessuno has narrow scope. 24. Like Hoekstra and Koster, Koopman (1984) argues that long extraction of the subject is always possible in Dutch. Her theory differs from the other two proposals. She claims that the subject position is properly governed by COMP. COMP governs the subject position under coindexation. The index on COMP is the result of upward percolation of the index of the moved wh-phrase. She observes correctly that main clauses and embedded clauses show parallel behaviour insofar as the possibility of extraction of subjects is concerned. This leads her to a theory in which subject extraction is always possible. No account is given for the unacceptability of those sentences which led to the postulation of a dat-t^ effect in Dutch. 25. Bayer(1984) accounts for the difference between Standard German and Bavarian by assuming that a version of the doubly-filled-COMP filter is part of the grammar of Standard German, but not of the grammar of Bavarian. Although I shall

-255-

The Adverbial Pronoun er

not discuss such a proposal in detail, I would like to suggest that the difference might be accounted for in a different manner. In ch.1, it was pointed out that we should not expect extraction from sentential complements to be possible in Dutch at all. The reason for this is that the postverbal sentential complement is not canonically governed by the governing verb. In order to allow movement to be possible, we argued that movement from an A'-position (i.e.from COMP) might proceed without paying heed to canonical government and in violation of the Gap Condition. It might be the case that Standard German observes the Gap Condition more strictly than Dutch or Bavarian, in that it does not allow extraction from sentential complements as in (136). 26. This D-structure organization of Icelandic differs from what is generally taken to be the underlying structure of Icelandic (e.g.Thrainsson 1984, Platzack 1984), but see Thrainsson (1984,fnt.1). 27. O33a') is perfectly acceptable if er is interpreted as a locative expression. However, such an interpretation is not possible in (133b') since the sentence contains a motional PP that is difficult to combine with a locative expression. 28. A similar conclusion is reached in De Schutter(197^), Elffers(1977), Kirsner(1979) and Schermer-Vermeer(1985). However, their approach to the problem of the conditions on the appearance of (expletive) er differs considerably from the approach defended here. 29. An apparent problem for the theory presented here involves the contrast noted in Drewes et al.(1984): (i)a.

de manier waarop er wordt geantwoord the way where-in there is answered b.??de vraag waarop er wordt geantwoord the question where-on there is answered

The contrast in (i) is rather peculiar, since it does not appear if er is out or if there is no wh-movement, as in (ii).

left

(ii)a. dat (er) op die manier geantwoord wordt that there in that way answered is b. dat (er) op die vraag geantwoord wordt that there on that question answered is Drewes et al. provide a syntactic explanation for the contrast in (i) in terms of overlapping paths within the framework of Pesetsky (1982b). They argue that the following three factors are relevant: - wh-movement, to explain the absence of a contrast in (ii) - insertion of er - the status of the PP as a complement or an adjunct, which is the only syntactic difference between (ia) and (ib). They argue that the combination of wh-movement from complement position and the insertion of er leads to a violation of Pesetsky's Path Containment Condition. Although I think that the observations are basically correct and their analysis is an interesting attempt to apply Pesetsky's theory to a new domain of facts, their explanation is demonstrably wrong. If the analysis were correct, we would expect the same contrast to appear if the PP is questioned and not relativized. However, the contrast disappears completely if the PPs are questioned, as in (iii).

-256-

The A d v e r b i a l Pronoun er

(iii)a. Op welke In which b. Op welke On which

manier wordt er _t geantwoord? way is there answered vraag wordt er _t geantwoord? question is there answered

More striking, the absence of expletive er in questions produces unacceptable structures in these two cases. (iv)a.??0p On b.??0p On

welke which welke which

manier wordt t_ geantwoord way is answered vraag wordt _t geantwoord question is answered

The difference between (iii) and (iv) follows directly from the EPC, since no constituent in (iv) can be interpreted as belonging to presupposition. The difference in acceptability between (iv) and (ii) follows from the difference between questions and relative clauses in the matter of the pragmatic status of the moved constituent. What remains as a problem is the contrast in (i). (ib) seems to suggest that the relativized PP should be interpreted as presupposition, thus making the appearance of er superfluous. This unacceptability is of the same order as the unacceptability of the sentence in (v). (v) ??de jongens die er t_ komen the boys who there come If er is not interpreted as a purely locative expression -an interpretation that is available for (ib) as well- the relative clause in (v) is unacceptable because the moved wh-phrase is a definite pronoun which belongs to presupposition. If in a sentence corresponding to (ib) there is no movement and the PP contains a pronominal object, the same unacceptability can be observed, as is demonstrated in (vi). This indicates again that wh-movement is irrelevant for the explanation of the unacceptability of (ib). (vi) Hij hoopte dat (??er) daarop zou worden geantwoord He hoped that there there-on would be answered We have thus reduced the problem to the acceptability of (ia). It is clear that the contrast in (i) should be related to the status of the extracted PP. It is interesting to observe that daarop in (vi) cannot be interpreted as a manner adverbial adjunct, corresponding to op die manier in (iia), nor can a sentence like (vii) be interpreted as questioning the way in which an answer is given, but only as questioning the prepositional object of the verb. (vii) Waarop heeft hij geantwoord? Where-on has he answered Only if there is a local antecedent for the R-pronoun can it be interpreted as being part of a manner adverbial. This is not a general property of adjunct PPs, since locative and directional PP-adjuncts do not show the same behaviour. I have no insights to offer to account for this phenomenon. It is possibly related to other differences between manner adverbials and other constituents (cf.Huang 1982, Lasnik & Saito 1984, Aoun 1985, Koopman & Sportiche 1985). Although the answer to the problem with respect to (ia) has not been solved in a satisfactory way, it should at least be clear that a syntactic solution for

-257-

The A d v e r b i a l P r o n o u n er

the contrast in ( i ) , as proposed by Drewes et al., is inadequate as it cannot distinguish between wh-movement in questions and relative clauses. I shall leave the question as to how to account for the differences between manner adverbials and other constituents for further research.

-258-

Chapter 4

Some Related Topics ^. 1 Introduction In t h i s chapter I shall p u r s u e some

of

theory presented

chapters.

in

the

previous

the

implications In

sections I shall b a s i c a l l y be c o n c e r n e d w i t h issues. In the later sections I shall d i s c u s s the

the

of

the

first

few

two t h e o r e t i c a l i m p l i c a t i o n s of

the t h e o r y for several other languages. A l t h o u g h the two p a r t s of this c h a p t e r are closely r e l a t e d , it is less homogeneneous than the p r e c e d i n g ones. It w i l l also be r a t h e r s k e t c h y , since it is i m p o s s i b l e to discuss the d i s t r i b u t i o n and p r o p e r t i e s of gaps and dummies in other languages in j u s t as much d e t a i l as in D u t c h . I shall p r o v i d e some i n d i c a t i o n s in w h i c h d i r e c t i o n an e x p l a n a t i o n consistent w i t h the t h e o r y d e v e l o p e d above might be f o u n d . and

In section 2 I shall discuss the r u l e of a d j u n c t i o n once more argue that the t h e o r y of p a t h s s e v e r e l y c o n s t r a i n s the

p o s s i b i l i t y of a p p l i c a t i o n of a d j u n c t i o n . whereas no such limitations apply to substitution. Section 3 d e a l s w i t h the n o t i o n e x t e r n a l a r g u m e n t . The q u e s t i o n h e r e is what the e x t e r n a l a r g u m e n t is e x t e r n a l to. There are b a s i c a l l y two possible answers to this q u e s t i o n . E i t h e r it is e x t e r n a l to the m a x i m a l p r o j e c t i o n of the c a t e g o r y assigning the e x t e r n a l a r g u m e n t , or it is e x t e r n a l to the a r g u m e n t p r o j e c t i o n of the θ-assigning h e a d . The f i r s t o p t i o n is d e f e n d e d in r e c e n t work by W i l l i a m s , w h e r e a s the second v i e w is p r e s e n t in w o r k by Stowell and C h o m s k y . I shall d e f e n d t h e most r e s t r i c t i v e o p t i o n , i.e. that

Some Related Topics the external argument should not only be e x t e r n a l to

the

argument

p r o j e c t i o n but i n t e r n a l to the maximal p r o j e c t i o n of

the

category

assigning the external

-role as well.

relatively inconsequential

for

the

Although analysis

f a r - r e a c h i n g consequences with respect to and I t a l i a n .

such

of

a

Dutch,

languages

like

view it

is has

English

In what follows I shall discuss consequences and problems with respect to a number of languages. It will t u r n out that the theory is able to account for a number of problems in an interesting way. However, several problems cannot be satisfactorily accounted for. Given the rather t e n t a t i v e c h a r a c t e r i s t i c of this c h a p t e r , I shall leave these questions for f u r t h e r r e s e a r c h .

4.2 Adjunction and Connection In note 8 to chapter 1 it was observed that there is

an

asymmetry

in D u t c h in the case of parasitic gaps, d e p e n d i n g on the position of the a d j u n c t clause c o n t a i n i n g the p a r a s i t i c gap. If the a d j u n c t clause appears p r e v e r b a l l y , parasitic gaps are possible under conditions imposed by the Gap C o n d i t i o n , whereas the parasitic

the gap

appears to be impossible if the adjunct clause appears p o s t v e r b a l l y . This contrast is i l l u s t r a t e d in ( 2 ) . In (1) it is shown that no such contrast exists if no parasitic gap is present w i t h i n t h e a d j u n c t clause. ( 1 ) a . Jan h e e f t d e z e boeken [zonder ze te b e k i j k e n ] John has those books without them to inspect in de käst gezet in the bookcase put b. Jan h e e f t deze boeken in de käst gezet [zonder John has those books in the bookcase put without ze te b e k i j k e n them to i n s p e c t ( 2 ) a . Jan h e e f t deze boeken [zonder e_ te b e k i j k e n ] John has those books w i t h o u t to inspect in de käst gezet in the bookcase put

-260-

Some R e l a t e d Topics b . * J a n h e e f t deze boeken in de käst gezet [ z o n d e r John has those books in the bookcase put w i t h o u t e_ te b e k i j k e n to inspect In chapter 1 it was argued t h a t the d i f f e r e n c e between ( 2 a ) and ( 2 b ) is a r e l a t i v e m a t t e r , in s h a r p d i f f e r e n c e between p a r a s i t i c gaps in

in a c c e p t a b i l i t y contrast to the

finite

and

non-finite

a d j u n c t clauses. No e x p l a n a t i o n of the d i f f e r e n c e b e t w e e n ( 2 a )

and

( 2 b ) was provided. A f t e r d i f f e r e n c e between ( 2 a ) a n d

the All

reconsideration, I think that ( 2 b ) i s n o t r e l a t i v e a t all.

i n f o r m a n t s consistently judge sentences of type ( 2 b ) to be unacceptable in c o m p a r i s o n w i t h sentences of type ( 2 a ) . The theory proposed in the f i r s t chapter does not

provide

a

straightforward

explanation of this d i f f e r e n c e . As far as the Gap C o n d i t i o n is concerned both s t r u c t u r e s seem to be s i m i l a r . In each case the a d j u n c t clause contains a p a r a s i t i c gap the g - p r o j e c t i o n of which is p r o p e r l y connected w i t h the g - p r o j e c t i o n of the real is demonstrated in ( 3 ) .

(3)a. COMP

zonder

PRO NP e_

V

in de

heeft

kast

gezet

te

bekijken

-261-

gap.

This

Some R e l a t e d Topics (3)b.

INFL''(S') INFL1

COMP V1' NP

Jan

INFL V

NP

t

PP

V

P

in de

heeft

kast

gezet

NP

V

Λ

zonder PRO NP

V

te bekijken In each case, the path in the a d j u n c t clauses is d i r e c t l y c o n n e c t e d w i t h the path in the main clause. The only d i f f e r e n c e

between

two s t r u c t u r e s is the d i r e c t i o n of this c o n n e c t i o n . As it

the

stands,

the Gap C o n d i t i o n does not d i s t i n g u i s h between d i f f e r e n t d i r e c t i o n s o f c o n n e c t i o n . C o n c e p t u a l l y , t h e d i f f e r e n c e between ( 3 a )

and

(3b)

resembles the other l e f t - r i g h t a s y m m e t r i e s discussed in chapter

1,

which

of

were

argued

to

follow

from

the

directional

notion

canonical g o v e r n m e n t . Suppose t h a t we strengthen

the Gap C o n d i t i o n

, repeated here in ( 4 a ) , in

the

c o n n e c t i o n is

such

a

way

that

t a k e n into a c c o u n t . We may do so

by

direction

of

introducing

an

a d d i t i o n a l c o n d i t i o n such a s ( 4 b ) . ( 4 ) a . Gap Condition -A gap Z in a tree P is

linked to its

antecedent by a

c o n n e c t e d s u b t r e e of P which c o n s t i t u t e s a g - p r o j e c t i o n -XP is a g - p r o j e c t i o n of the s t r u c t u r a l governor W of Z the

head of XP c-governs Z or a g - p r o j e c t i o n of W

-X c ( a n o n i c a l l y ) - g o v e r n s Υ iff

X p r e c e d e s Υ in a

VO language and f o l l o w s Υ in an 0V language

-262-

iff

Some R e l a t e d Topics

( M ) b . A g-projection g-projection

X is p r o p e r l y c o n n e c t e d to a

Υ iff

X and Υ are

government c o n f i g u r a t i o n , i.e.

in a c a n o n i c a l X p r e c e d e s Υ in an

0V language and f o l l o w s Υ in a VO language Such an analysis would p r o v i d e an e x p l a n a t i o n of

the

contrast

in

( 2 ) . T h e u n g r a m m a t i c a l i t y o f ( 2 b ) w o u l d f o l l o w from t h e fact that the g - p r o j e c t i o n in the a d j u n c t clause is not p r o p e r l y c o n n e c t e d to the g-projection in the matrix clause. U n f o r t u n a t e l y , a c o n d i t i o n such as

(Ub)

cannot

be

correct,

since it would p r e d i c t sentences such as (5) to be u n g r a m m a t i c a l . (5)

A person who close f r i e n d s of e_ a d m i r e _t

In (5) the p a r a s i t i c gap is embedded in the s u b j e c t NP. The g - p r o j e c t i o n containing the p a r a s i t i c gap stops at the level of the s u b j e c t NP. The s u b j e c t NP is on a l e f t b r a n c h . T h u s , X precedes Υ in a VO language, c o n t r a r y to what is e x p e c t e d on the basis of ( 4 b ) . N e v e r t h e l e s s , I shall a r g u e below t h a t ( 4 b ) p r o v i d e s t h e correct explanation f o r t h e c o n t r a s t i n ( 2 ) . B e f o r e discussing

the

q u e s t i o n as to how the v i o l a t i o n of ( ^ b ) in (5) can be a c c c o u n t e d f o r , I shall f i r s t discuss a n o t h e r p r o b l e m that was l e f t unsolved in c h a p t e r 1. This p r o b l e m shows some r e s e m b l a n c e to the p r o b l e m discussed a b o v e . The rule of S - i n t e r n a l a d j u n c t i o n has been discussed e x t e n s i v e l y in the p r e c e d i n g c h a p t e r s . In c h a p t e r 1, note 12, the q u e s t i o n was raised why t h e r e is no a d j u n c t i o n to the right in Dutch and to the l e f t in E n g l i s h . No answer to t h i s q u e s t i o n was given t h e r e . Let us try to relate this q u e s t i o n to the p r o b l e m raised a b o v e . The g e n e r a l i z a t i o n seems to be that the l a n d i n g site of the a d j u n c t i o n is on the same side of the governor as the e x t r a c t i o n site i t s e l f . The p o s i t i o n of the e x t r a c t i o n site w i t h respect to its governor is d e t e r m i n e d by the Gap C o n d i t i o n . Suppose t h a t the l a n d i n g site of a d j u n c t i o n is d e t e r m i n e d by an c o n d i t i o n , such a s ( 6 ) .

-263-

additional

Some Related Topics (6) an antecedent X can be linked to a g-projection Υ iff X and Υ are in a canonical government c o n f i g u r a t i o n , i.e. X precedes Υ in an 0V language and follows Υ in a VO language Condition (6) implies that the antecedent X should precede a sister constituent belonging to the g-projection of the gap in an 0V language and follow it in an VO language. Condition (6) correctly predicts that a d j u n c t i o n to the right in Dutch and to the left in English is impossible. It is evident t h a t , if c o r r e c t , ( 4 b ) and (6) should be combined into a more general c o n d i t i o n . Within the theory of paths there is no explicit mention as to how the antecedent is related to the g-projection of the governor of the g a p , apart from the fact that the g-projection should reach the level of the a n t e c e d e n t . Suppose that we claim that the antecedent should be connected w i t h the g-projection in a way similar to the way in which a g-projection is connected w i t h another g - p r o j e c t i o n . Such a view implies that the antecedent c-commands the gap as a consequence of the connectedness theory. An intervening node blocks connection. No explicit c-command r e q u i r e m e n t on movement is r e q u i r e d . If we do so, ( 4 b ) and (6) can be combined quite easily, as in (7) . (7)

If X is an antecedent or a g-projection. X may be connected to a g-projection Υ iff X and Υ are in a canonical government c o n f i g u r a t i o n

Condition (7) is able to account for the contrast in (2) as well as for the u n i d i r e c t i o n a l i t y of a d j u n c t i o n . Again, however, this c o n d i t i o n cannot be c o r r e c t . It would imply that in all cases the landing site of a movement rule should precede the g - p r o j e c t i o n in an 0V language and follow it in a VO language. It is clear that in a simple passive sentence in English for e x a m p l e , condition (7) is not obeyed. E i t h e r we have to drop condition (7) or we have to restrict its a p p l i c a t i o n in some p r e f e r a b l y i n s i g h t f u l way. If we r e s t r i c t the a p p l i c a t i o n of (7) to a d j u n c t s and a d j u n c t i o n s o n l y , the correct results are d e r i v e d . In that case we would have to replace (7) by (8) .

-264-

Some R e l a t e d Topics

(8) For any n o n - d e s i g n a t e d p o s i t i o n X ( i . e .

a p o s i t i o n that is

not d e f i n e d by +Case, + Θ , + w h , or another f e a t u r e ) , if X contains an a n t e c e d e n t or a g - p r o j e c t i o n , X may be connected to a g - p r o j e c t i o n Υ iff X and Υ are in a canonical government c o n f i g u r a t i o n , i.e. X precedes Υ in an 0V languages and follows Υ in a VO language. As it

stands,

(8)

discriminates

sentences in (2) and the one in

(5).

appears in a subject NP,

might

which

between In

(5)

be

the the

parasitic-gap parasitic

considered

a

gap

position

d e f i n e d by +Case ( c f . n e x t s e c t i o n ) , while in (2) the p a r a s i t i c gap appears in an a d j u n c t c l a u s e . In the same way (8) discriminates between

the

adjunction

rule

under

discussion

and

passive

in

E n g l i s h , which might be considered movement to a +Case p o s i t i o n . It follows f r o m (8) that movement to a p o s i t i o n in which the canonical government c o n f i g u r a t i o n is not observed must be movement to a designated p o s i t i o n , i.e. s u b s t i t u t i o n . One of the consequences of t h i s is that wh-movement to COMP must be s u b s t i t u t i o n

in

VO-languages

whereas it can be a d j u n c t i o n in G e r m a n . It is then p r e d i c t e d

like

English

respect

mentioned

to

this

seems

French,

OV-languages such as Dutch and that a ' d o u b l y - f i l l e d COMP' is

possible in the l a t t e r case o n l y . W i t h above,

and

indeed

be

to the

the case,

languages as

is

illustrated in ( 9 ) & ( 1 0 ) . [ 1 ] . (9)

Dutch: Bavarian:

(10)English: French:

Ik vraag me af wat of dat hij _t zag I wonder what whether that he saw I woass ned wer dass _t des toa I know not who that this done *I d o n ' t know who that John saw *Je me demande qui que tu as vu

hod had _t _t

I wonder who that you have seen In chapter 2 it has been argued that passive and raising in Dutch are instances of a d j u n c t i o n . The d e r i v e d s t r u c t u r e is in accordance w i t h ( 8 ) . We may now go one step f u r t h e r and claim that if a s t r u c t u r e is in accordance w i t h (8) it is formed by a d j u n c t i o n

-265-

Some Related Topics and if it

is n o t , the

rule

involved

must

be

substitution.

i n t u i t i v e idea behind t h i s claim is that if the top of the c o n t a i n i n g a n t e c e d e n t and trace observes the

canonical

The

subtree

government

r e q u i r e m e n t , j u s t as all other levels of the r e l e v a n t s u b t r e e , structure

is

optimally

in

accordance

with

the

the

unidirectional

p a t h - t h e o r y . No a d d i t i o n a l provisions are r e q u i r e d to determine the w e l l - f o r m e d n e s s of the r e l a t i o n between a n t e c e d e n t and t r a c e . . The antecedent is v i s i b l e as a consequence of its ' c a n o n i c a l ' position w i t h respect to the p a t h . If the a n t e c e d e n t is not

connected

with

the path in a c c o r d a n c e with the canonical-government r e q u i r e m e n t , it can be visible only if the antecedent position is independently m o t i v a t e d . This i m p l i e s that the a p p l i c a t i o n of s u b s t i t u t i o n is r e s t r i c t e d and r e p r e s e n t s the marked case. This s i t u a t i o n resembles the s i t u a t i o n found

in

the

subject

e x t r a c t i o n cases discussed in c h . 3 - If long e x t r a c t i o n proceeds accordance w i t h the relevant p r i n c i p l e s , as in D u t c h , nothing need

be

said.

If,

however,

subject

extraction

violates

w e l l - f o r m e d n e s s constraint on the resulting s u b t r e e ,

it

can

in

else the only

take place by means of a d d i t i o n a l p r o v i s o s , as in English and French. If this is c o r r e c t , we can s c h e m a t i c a l l y represent a d j u n c t i o n and s u b s t i t u t i o n

in 0V and VO languages in the f o l l o w i n g way.

( 1 1 ) OV-languages a.adjunction

b.substitution

XP

XP

-266-

Some R e l a t e d Topics ( 1 2 ) VO-languages a.adjunction

b.substitution

XP

XP

In the following sections it will be shown that such a typology of r u l e s , based on a u n i f o r m a p p l i c a t i o n of the canonical government r e q u i r e m e n t consequences.

on

paths,

has

a

number

of

interesting

4.3 The external argument and the s t r u c t u r e of Modern English In the preceding c h a p t e r s it was argued that the e x t e r n a l a r g u m e n t is a c o n s t i t u e n t that is assigned an e x t e r n a l thematic role by a p r e d i c a t e . Whether or not a p a r t i c u l a r lexical category assigns an external argument is entirely dependent on the lexical s p e c i f i c a t i o n of that c a t e g o r y . F u r t h e r m o r e , it was argued t h a t there is no base-generated NP position to which the e x t e r n a l thematic role is assigned. In this section I shall raise the question to which position the external argument is assigned. This implies that we have to d e t e r m i n e what the external argument is external to and what the maximal u p p e r limit of the p r o j e c t i o n is w i t h i n which the external argument may be assigned. In the l i t e r a t u r e these questions have been discussed e x t e n s i v e l y w i t h respect to Small Clauses. There are basically three different proposals for the s t r u c t u r a l r e p r e s e n t a t i o n of Small Clauses. SCs may be constituents or n o t . In the p r e d i c a t i o n theory developed by W i l l i a m s C 1 9 7 8 ) the s u b j e c t - p r e d i c a t e r e l a t i o n is established by a rule of p r e d i c a t i o n that causes the s u b j e c t to be coindexed w i t h its p r e d i c a t e . In this view s u b j e c t and p r e d i c a t e do not c o n s t i t u t e a single c o n s t i t u e n t , since the e x t e r n a l argument is n e c e s s a r i l y external to the m a x i m a l p r o j e c t i o n of the p r e d i c a t e . If s u b j e c t and predicate are taken to f o r m a single c o n s t i t u e n t , t h e r e are two d i f f e r e n t a p p r o a c h e s . E i t h e r the e x t e r n a l a r g u m e n t belongs to the -267-

Some R e l a t e d Topics p r o j e c t i o n of the p r e d i c a t e , the endocentric a p p r o a c h , or the s u b j e c t is e x t e r n a l to the p r o j e c t i o n of the predicate but is part of a c o n s t i t u e n t consisting of s u b j e c t and predicate exclusively, the e x o c e n t r i c a p p r o a c h . The e n d o c e n t r i c approach is developed in Stowell ( 1 9 8 1 , 1 9 8 3 ) , whereas the exocentric approach can in C h o m s k y ( 1 9 8 D .[1]. S c h e m a t i c a l l y , the three different positions

be can

found be

represented as in ( 1 3 ) · ( 1 3 ) a . NP.^ ...

b.

XP i

SC

/\XP

NP

c.

XP(SC)

/XX '

NP

W i t h o u t going into much d e t a i l , I shall adopt S t o w e l l ' s proposal that the external argument belongs to the p r o j e c t i o n of the p r e d i c a t e . This implies that S t o w e l l ' s d e f i n i t i o n of SUBJECT (1*0 is taken o v e r . ( 1 4 ) The SUBJECT of a phrase XP is the argument of X or X ' which i s d i r e c t l y dominated b y X 1 ' (Stowell,1983,(21)) What is relevant to the discussion here is the s t r u c t u r a l position of the s u b j e c t of S. Stowell argues that the subject of S is covered b y t h e d e f i n i t i o n i n ( 1 4 ) : " I t (i.e.definition ( 1 4 ) , H . B . ) also a c c o u n t s for the s u b j e c t of S, provided that S is incorporated into the X-bar system as an X ' 1 p r o j e c t i o n of I N F L . Then if S = I " ' ( = I N F L ' ' ) , it must be that VP f u n c t i o n s as a s t r u c t u r a l complement of I N F L . so that there is a c o n s t i t u e n t I1 ( = I N F L ' ) corresponding to the traditional predicate phrase." ( S t o w e l l , 1 9 8 3 , p . 2 9 5 ) . This a s s u m p t i o n , i.e. that the subject of a sentence is the s u b j e c t of a predicate phrase INFL', is

-268-

Some R e l a t e d Topics problematical

in

several

respects.

The

relevant

configuration

proposed by Stowell is d e p i c t e d in ( 1 5 ) . (15)

INFL"(=S)

^\ INFL'

NP

INFL

(=Pred.P) VP

I N F L , the head of S, d i f f e r s

from

the

other

lexical

categories

p o t e n t i a l l y taking a s u b j e c t in that the presence or absence t h e m a t i c external argument is not d e p e n d e n t on

lexical

of

a

properties

of the h e a d , but is e n t i r e l y d e p e n d e n t on lexical p r o p e r t i e s of the complement of the h e a d , i.e. completely immaterial

in

V P . The lexical p r o p e r t i e s of INFL are this

respect.

This

implies

that

the

e n d o c e n t r i c approach is only a p p a r e n t . It is saved by claiming that the external argument is an argument of I N F L ' rather than of V P . The theory of e n d o c e n t r i c s u b j e c t s is u n d e r m i n e d since only in this case is the subject external to the maximal p r o j e c t i o n of category that d e t e r m i n e s the assignment of its t h e m a t i c role. A further incompatibility

in S t o w e l l ' s theory of

the

the

subject

of S is the fact that the VP that is a complement of INFL does not have a s u b j e c t . Stowell argues that C h o m s k y ' s claim that every S must contain a subject can be made to follow f r o m the r e q u i r e m e n t that every c o n s t i t u e n t that is i n t e r p r e t e d as a clause must contain both a subject and a p r e d i c a t e . Given the endocentricity requirement it follows that there can be no c o n s t i t u e n t s , i.e. maximal p r o j e c t i o n s , that are p r e d i c a t e s . It f u r t h e r follows that VP cannot be a p r e d i c a t e nor can it be a clause since it lacks a s u b j e c t . The question thus arises as to the status of VP. A related question involves the fact that VP may contain

a

s u b j e c t with p e r c e p t i o n v e r b complements.

a

If

the

VP contains

s u b j e c t in ( 1 6 ) , then why is it impossible in ( 1 7 ) ? Or to put d i f f e r e n t l y , why can sing a song be a p r e d i c a t e in ( 1 6 ) but not

(17)? ( 1 6 ) I heard [VP him [sing a song] ( 1 7 ) He INFL [VP sing a song]

-269-

it in

Some R e l a t e d Topics

We may s t i p u l a t e t h a t , j u s t as w i t h N P s , VPs need

not

contain

a

s u b j e c t . The p r o b l e m w i t h this assumption is

VPs w i t h o u t

a

that

s u b j e c t appear only if they are subcategorized by c o n f i g u r a t i o n they must appear w i t h o u t a s u b j e c t ,

I N F L . In that whereas VP in

other c o n f i g u r a t i o n s has to appear as a clause, i.e.

as

a

subject

and a p r e d i c a t e . A t h i r d i n d i c a t i o n that S t o w e l l ' s approach to the subject of S should be reconsidered can be

derived

argued in the preceding c h a p t e r s , belongs to the V - p r o j e c t i o n . contains both the argument

from

Dutch.

the external

As

argument

has

been

in

Dutch

[2]. The maximal projection of V projection of V and the external

a r g u m e n t . With respect to Dutch t h e n , the e n d o c e n t r i c approach be m a i n t a i n e d w i t h o u t p r o b l e m s . There is no problematic constituent and INFL is not

part

of

the

clause

but

rather

an

Case

if

operator w i t h the s y n t a c t i c f u n c t i o n to assign nominative it

is s p e c i f i e d for [+Tense]. If this analysis is c o r r e c t ,

wonder w h e t h e r Universal Grammar allows d e t e r m i n a t i o n of an external argument to such Dutch the e x t e r n a l argument is an

argument

variation an extent of

the

V

can VP

we may in the that in

projection

o n l y , w h e r e a s in English the external argument is an argument of INFL and VP t o g e t h e r . Such a v a r i a t i o n seems highly u n l i k e l y . Given that INFL does not c o n t r i b u t e to the selection of an external argument and that the status of the VP is p r o b l e m a t i c , it

seems

to

be the case that the e n d o c e n t r i c approach can be m a i n t a i n e d only if English is like D u t c h in t h i s r e s p e c t . This amounts to claiming that in English the maximal p r o j e c t i o n of V contains the external argument as w e l l . If such a claim cannot be motivated s a t i s f a c t o r i l y , we have to d r o p the a t t r a c t i v e idea of a f u l l y general e n d o c e n t r i c conception of the s u b j e c t - p r e d i c a t e relation, which reduces the n u m b e r of base rules c o n s i d e r a b l y . In that case we may be forced to t u r n to a f u l l y exocentric approach as d e f e n d e d by W i l l i a m s , who claims that both in the case of Small Clauses and S the subject is external to

the

constituent

containing the p r e d i c a t e . This a p p r o a c h seems less a t t r a c t i v e , for three reasons: it r e q u i r e s an a d d i t i o n a l rule of i n t e r p r e t a t i o n that r e l a t e s s u b j e c t and p r e d i c a t e ; it is less r e s t r i c t i v e w i t h respect to the r e l a t i v e position of subject and

-270-

predicate

and

it

Some R e l a t e d Topics r e q u i r e s the base r u l e s to p r o v i d e an e x t e r n a l a r g u m e n t p o s i t i o n . M o r e o v e r , such an a p p r o a c h would f o r c e us to adopt the existence of a VP in D u t c h , w h i c h is not i n d e p e n d e n t l y m o t i v a t e d and runs into t r o u b l e with the a d j u n c t i o n a n a l y s i s p r e s e n t e d in the preceding chapters. A t h i r d possibility is to adopt a mixed a p p r o a c h , as in Chomsky(1981). Some subject-predicate relations might be e x o c e n t r i c , such as the s u b j e c t - p r e d i c a t e r e l a t i o n in S in E n g l i s h , whereas others might be e n d o c e n t r i c , as in Small Clauses. A l t h o u g h we may e v e n t u a l l y have to t u r n to such an a p p r o a c h , it seems clear that this is the least a t t r a c t i v e approach to this p r o b l e m . Let us thus try to m o t i v a t e the e n d o c e n t r i c approach described above. [3]· The only p r o b l e m we have e n c o u n t e r e d thus far is the s u b j e c t - p r e d i c a t e relation in S in English. If the s u b j e c t belongs to the V p r o j e c t i o n in E n g l i s h , there is no c o n s t i t u e n t VP and INFL is external to V", as in ( 1 8 ) . ( I 8 ) a . INFL [ v , ,

NP, V

]

b. [ γ , ,

NP, V

] INFL

W i t h respect to English there seems to be no motivation whatsoever to assume that INFL should be clause f i n a l . Let us assume that ( I 8 a ) is the correct u n d e r l y i n g r e p r e s e n t a t i o n of English clauses. [ U ] . In order to derive a w e l l - f o r m e d simple sentence the s u b j e c t has to be moved to a position to the l e f t of I N F L . Why then is the s u b j e c t o b l i g a t o r i l y moved across INFL? Suppose the reason has to do with Case assignment, j u s t as in o t h e r instances of o b l i g a t o r y movement of N P , for example in passives. How can this be achieved? Given the a s s u m p t i o n that Case assignment is u n i d i r e c t i o n a l , which is extensively motivated in S t o w e l K 1 9 8 1 ) , H o e k s t r a ( 1 9 8 U ) and in chapter 1 of this b o o k , and that the d i r e c t i o n of Case assignment in a p a r t i c u l a r language is not u n i f o r m , c o n t r a r y to what is claimed in S t o w e l K 1 9 8 1 ) and K o o p m a n ( 1 9 8 4 ) , we may propose that in English INFL assigns n o m i n a t i v e Case to the l e f t . In t h a t case the e x t e r n a l argument has to be moved to a position to the l e f t of INFL in order to receive ( n o m i n a t i v e ) Case. If it is correct that lack of Case is responsible for the movement of the e x t e r n a l a r g u m e n t and consequently for its s u r f a c e p o s i t i o n , we may expect t h a t t h e r e are cases in w h i c h the s u b j e c t r e t a i n s its original p o s i t i o n if -271-

Some Related Topics there are other mechanisms to assign Case. That is exactly what can be observed in the case of p e r c e p t i o n - v e r b complements, as in ( 1 9 ) . [5]. ( 1 9 ) a. John saw [them leave] b. I heard [him sing] On this view, the complement of the perception verb is a bare S, or rather the maximal V p r o j e c t i o n including the external argument. The external argument can get Case since it is governed by the m a t r i x verb in the correct d i r e c t i o n , since V assigns Case to the r i g h t . The relevant part of the s t r u c t u r e of sentences of the type ( 1 9 ) , is given in ( 2 0 ) . (20)

Vmax ( = S ) NP

This s t r u c t u r e is only minimally d i f f e r e n t from the one proposed by Stowell. The d i f f e r e n c e resides in the fact that I shall assume that the external argument belongs to the V projection in u n d e r l y i n g s t r u c t u r e not only in this case but in all cases. This implies that the relevant part of the S-structure representation of a simple sentence such as ( 2 1 ) is as given in ( 2 2 ) . (21)

John saw me INFL1

(22)

max

John

saw me

-272-

Some R e l a t e d Topics

In order for ( 2 2 ) to be a w e l l - f o r m e d s t r u c t u r e the trace should be canonically g o v e r n e d , as r e q u i r e d by the Gap C o n d i t i o n . is not governed

by

V,

since

argument p r o j e c t i o n , i . e . V.

government H o w e v e r , the

is

The

restricted

trace

is

trace

to

the

governed

by

I N F L , since we do not assume m a x i m a l p r o j e c t i o n s to be b a r r i e r s to g o v e r n m e n t . Given the fact that English is a VO-language, INFL governs

the

trace

canonically.

Because

the

trace

is

not

Case-marked, the trace in s u b j e c t position is f o r m a l l y identical to a trace in passive c o n s t r u c t i o n s . Let us now t u r n to the site. If it is assumed t h a t INFL assigns Case to its l e f t , in position N P . gets n o m i n a t i v e C a s e . G i v e n the d i r e c t i o n assignment the position p r e c e d i n g INFL

is

defined

by

landing the NP of

Case

nominative

Case and thus constitutes a position which may serve as the landing site of a substitution r u l e .

This

analysis

has

the

interesting

consequence that it allows us to account for the fact that sentences must contain an NP s u b j e c t , whereas

such

a

seems to be lacking in D u t c h . If t h e r e is an NP p o s i t i o n

English

requirement in

front

of I N F L , this position has to be f i l l e d d u r i n g the d e r i v a t i o n . If it is n o t , there is a n o n - c a n o n i c a l l y - g o v e r n e d empty NP p o s i t i o n w i t h o u t a thematic role and w i t h o u t an a n t e c e d e n t . Such a position is clearly in v i o l a t i o n of several p r i n c i p l e s of the t h e o r y , such as the Gap C o n d i t i o n . We may thus d e r i v e the r e q u i r e m e n t that in English S must contain a s u b j e c t f r o m the f a c t that n o m i n a t i v e Case d e f i n e s an NP position in f r o n t of I N F L , i r r e s p e c t i v e of the thematic s t r u c t u r e of the sentence. Since OV-languages like D u t c h and German do not e n f o r c e such a p o s i t i o n for reasons of Case, it is expected that s u b j e c t l e s s sentences of the t y p e in ( 2 3 ) appear in Dutch and G e r m a n , but not in E n g l i s h . ( 2 3 ) a . In de t u i n wordt gedanst b. Im G a r t e n wurde getanzt c . * I n the g a r d e n is danced Before proceeding to the next section in which I shall try to make this analysis p l a u s i b l e from a d i a c h r o n i c point of v i e w , I shall once more discuss the phenomenon of e x c e p t i o n a l C a s e - m a r k i n g .

-273-

Some Related Topics In E n g l i s h t h e r e are two d i f f e r e n t types of e x c e p t i o n a l Case m a r k i n g c o n s t r u c t i o n s . One involves the p e r c e p t i o n v e r b complements discussed a b o v e . A n o t h e r ECM c o n s t r u c t i o n appears w i t h i n f i n i t i v a l complements of verbs like believe and expect, as in ( 2 4 ) . ( 2 4 ) a . I expect [ h i m to w i n ] b. I believe [ h i m to be a f o o l ] One of the d i f f e r e n c e s between ( 1 9 ) and ( 2 4 ) is the presence of to, which we may take to be a realization of INFL. [6]. In the same way temporal aspects may be expressed in c o n s t r u c t i o n s of the

type

in

( 2 4 ) but not in ( 1 9 ) . Given the presence of INFL in the i n f i n i t i v a l complements in ( 2 4 ) , the

maximal

projection

of

the

infinitival

11

clause is I N F L and the s u b j e c t is moved f r o m its p o s i t i o n w i t h i n Vmax to the NP position in f r o n t of I N F L . A p p a r e n t l y , both NP positions are a v a i l a b l e for exceptional C a s e - m a r k i n g . As

discussed

in c h a p t e r 1, COMP in English is the head of a COMP p r o j e c t i o n . One of the reasons adduced there was that for in COMP is able to govern the s u b j e c t of an i n f i n i t i v a l clause and assign Case to it,

as

in

s u b j e c t of can easily

an be

(25). ( 2 5 ) It was a p i t y [for [ John to lose]] We thus have t h r e e ways to assign Case to a i n f i n i t i v a l c l a u s e . These three ECM constructions accounted for if it

is

assumed

that

the

i n f i n i t i v a l clause is as d e p i c t e d in ( 2 6 ) . (26)

COMP11

ι COMP'

COMP

INFL11 ι INFL'

-274-

structure

of

a

full

Some R e l a t e d Topics A s s u m i n g t h a t an i n f i n i t i v a l INFL is not a r e l e v a n t will be P R O , unless COMP c o n t a i n s a c o n s t i t u e n t

governor,

that

is

able

govern NP.. and assign Case to it ( f o r ) . Let us assume that s e l e c t i o n of the m a t r i x v e r b d e t e r m i n e s w h e t h e r a v e r b COMP1' (the unmarked case), I N F L 1 ' or V 1 ' . If

INFL1'

NP..

is

lexical selects selected

(believe e t c . ) , the m a t r i x v e r b governs N P 1 and assigns Case to I f V ' i s selected ( p e r c e p t i o n v e r b s ) t h e m a t r i x v e r b and assigns Case to it.

governs

It is i n t e r e s t i n g to observe t h a t in

t h e r e are no exceptional Case-marking the English c o n s t r u c t i o n s

constructions

it. NP~

Dutch

analagous

w i t h believe-type v e r b s . The

to

to

reason

for

this is q u i t e s i m p l e . In D u t c h t h e r e is no NP p o s i t i o n o u t s i d e

V'

c o m p a r a b l e t o t h e N P - p o s i t i o n i n ( 2 6 ) . T h u s , t h e only case i n w h i c h a m a t r i x verb is able to assign Case to an embedded s u b j e c t is when it selects a c o n s t i t u e n t of the t y p e V ' 1 , which is the case w i t h p e r c e p t i o n - v e r b c o m p l e m e n t s . [?] . A f i n a l remark in this section concerns the f a c t t h a t if we accept ( 2 2 ) as the s t r u c t u r e for English

sentences,

longer a c o n s t i t u e n t

that

VP.

This

implies

we

there can

no

f o r m u l a t e rules like V P - d e l e t i o n and VP- f r o n t i n g . H o w e v e r ,

is

no

longer if

we

call these rules V ' - d e l e t i o n a n d V ' - f r o n t i n g , t h e same r e s u l t s might be o b t a i n e d . In order to get Case the s u b j e c t has to be moved f r o m V ' 1 t o NP.. a n d t h e r e s u l t i n g c o n s t i t u e n t c a n e i t h e r b e f r o n t e d or d e l e t e d .

U . 4 The s t r u c t u r e of Old English and its M o d e r n English In this section I shall suggest t h a t

d e v e l o p m e n t into

Old E n g l i s h

is

similar

to

Dutch in most r e l e v a n t r e s p e c t s . It w i l l then be proposed that many of the d i f f e r e n c e s b e t w e e n Old English and M o d e r n English may be d e r i v e d from one change in the grammar of E n g l i s h . This change involves the p o s i t i o n of I N F L . If we take Vmax to be the m a x i m a l p r o j e c t i o n of V t h a t c o n t a i n s all the a r g u m e n t s l e x i c a l l y selected by V , i n c l u d i n g t h e e x t e r n a l a r g u m e n t , I N F L

follows

Vmax

in

Old

E n g l i s h , j u s t as in D u t c h , whereas INFL p r e c e d e s Vmax in M o d e r n E n g l i s h . This d i f f e r e n c e would t h e n be r e s p o n s i b l e for a vast n u m b e r of d i f f e r e n c e s between Old English and M o d e r n E n g l i s h , such

-275-

Some Related Topics as the t r a n s i t i o n from 0V to VO, the obligatoriness

of

subject

English,

in

Modern

English,

but

not

in

Old

d i s a p p e a r a n c e of the impersonal c o n s t r u c t i o n , P-stranding in Modern English and c o n f i g u r a t i o n s in Old E n g l i s h . [ 8 ] .

the

the

a

lexical the

possibility

existence

of

of

that-t

If these r a t h e r t e n t a t i v e suggestions are in the correct d i r e c t i o n , this section serves two a d d i t i o n a l p u r p o s e s . F i r s t , it provides i n d e p e n d e n t evidence that the analysis of Dutch given the previous c h a p t e r s is c o r r e c t , since the same p r i n c i p l e s can applied to OE. Second,

it

provides

evidence

in

favour

of

in be the

proposal made in the last section concerning the basic s t r u c t u r e of Modern English.

4 . 4 . 1 The s t r u c t u r e of Old English 4 . 4 . 1 . 1 P-stranding W i t h o u t much discussion I shall OV-language. For discussion,

assume

see

that

Old

Canalei1978).[9] .

this to be c o r r e c t , it follows that canonical

English If

is

an

we assume

government

is

from

right to l e f t . T h u s , we do not expect preposition s t r a n d i n g in OE, but only p o s t p o s i t i o n s t r a n d i n g , as in D u t c h . This seems indeed be the case. For i n s t a n c e , only a passives can be found from b e f o r e

few the

to

examples of prepositional fifteenth c e n t u r y (Visser

1963, Van der Gaaf 1930, L i g h t f o o t 1 9 7 9 ) . As demonstrated by Wende ( 1 9 1 5 ) , there are three d i f f e r e n t cases of P - s t r a n d i n g in OE. A. preposition s t r a n d i n g b y R - p r o n o u n s In OE. p r o n o u n s like 5aer ( t h e r e ) and her ( h e r e )

may

function

a r g u m e n t s of P. In those cases they p r e c e d e P, as in ( 2 7 ) . ( 2 7 ) a . . . . 3aet crestene menn 5arto f a r a n magan t h a t C h r i s t i a n men thereto go may ' t h a t C h r i s t i a n men may go ( t o ) t h e r e 1 ( W , X V I I I . 3 5 / A l l e n 1980 ( 9 8 ) )

-276-

as

Some R e l a t e d Topics

b. . . . swa swa 3aet godspel her baef tan cvae3 as

the gospel

h e r e a f t e r relates

( A H T , I , 5 2 M , 2 1 / V a n K e m e n a d e 1984b ( 4 ) ) These R - p r o n o u n s can be s t r a n d e d , as is shown in ( 2 8 ) . ( 2 8 ) a . . . . 3zt hie 5aer m e h t e n betst fri b i n n a n habban that they t h e r e might best s e c u r i t y w i t h i n have " t h a t they might have the best s e c u r i t y in t h e r e ' ( Ο , ρ . 1 1 6 . 6 / A l l e n 1980 ( 1 0 2 ) ) b. . . . g i f 5aer gebedo aefter fylgeaa if t h e r e p r a y e r s a f t e r follow ' i f p r a y e r s follow a f t e r t h a t ' ( C P , 3 9 9 , 3 3 / V a t 1978 ( 3 5 ) ) Given the O V - s t r u c t u r e of OE and the Gap C o n d i t i o n , the p o s s i b i l i t y of s t r a n d i n g in ( 2 8 ) is e x a c t l y as e x p e c t e d . The fact that s t r a n d i n g is impossible with f u l l NP c o m p l e m e n t s of p r e p o s i t i o n s follows in the same w a y . B. p r e p o s i t i o n s t r a n d i n g by personal pronouns OE allows personal p r o n o u n s to appear o p t i o n a l l y in f r o n t of P, is demonstrated in ( 2 9 ) . [ 1 0 ] .

as

( 2 9 ) a . He ... h i r e mid gehaemde he her w i t h slept ' H e slept w i t h h e r ' ( C P . 4 1 5 , 1 7 / V a t 1978 ( 2 8 ) ) b. and hi ne dorsten him f o r e g e b i d d a n and they not dared him for pray ' a n d they d a r e d n o t p r a y f o r h i m * ( A H P , X I X , 2 2 6 / V a n K e m e n a d e 1984b ( 1 ) ) As e x p e c t e d , these personal p r o n o u n s can be moved from PP. the p r e p o s i t i o n s t r a n d e d . This is d e m o n s t r a t e d in ( 3 0 ) .

-277-

leaving

Some R e l a t e d Topics ( 3 0 ) a . Saet him mon symle 5aet tacn b e f o r a n baer that him someone o f t e n the ensign b e f o r e carried ' t h a t the ensign was often carried before h i m 1 ( B e d e , 1 4 6 , 2 / V a t 1978 ( 3 D ) b. Da wendon hi me heora base to Then t u r n e d they me their back to 'Then they turned their backs to m e ' ( B , I I , 8 , 1 2 / V a n Kemenade 1984b ( 2 ) ) Given that only R-pronouns and personal pronouns may appear in f r o n t of P as a r g u m e n t s , the observation that these two categories can be moved out of PP is an interesting confirmation of the idea that the d i r e c t i o n of government is relevant with respect to e x t r a c t i o n , as is implied by the Gap C o n d i t i o n . C. p r e p o s i t i o n s t r a n d i n g in relative clauses A more c o m p l i c a t e d situation emerges in case of relative clauses. There are several strategies in OE to form a relative clause. A r e l a t i v e clause in OE is i n t r o d u c e d either by a relative c o m p l e m e n t i z e r ( a e or sometimes aaet) , a relative pronoun C^e_), or a c o m b i n a t i o n of a relative pronoun and a r e l a t i v e c o m p l e m e n t i z e r (se 5 e ) . A p r e p o s i t i o n cannot be stranded by moving a relative pronoun of the se_ paradigm. In our terms the reason for this seems to be obvious given the fact that r e l a t i v e p r o n o u n s , just like demonstrative p r o n o u n s , do not appear in f r o n t of P. In those cases pied p i p i n g is o b l i g a t o r y . Examples are given in ( 3 1 ) . ( 3 1 )a. and net getimbrian medomlic h u s , on 5aet naenig and o r d e r e d to b u i l d small house in which no wer naefde ingang man not-had entrance ' a n d o r d e r e d a small house b u i l t , in which no man had admittance* ( M a r t . , p . 1 0 6 . 5 / A l l e n 1980 ( 2 3 ) ) b. Wa Sam men 5urh aone δ£ by mannes sunu belawed woe the man through whom that is m a n ' s son betrayed ' W o e the man through whom the son of man is b e t r a y e d ' ( S t . M a t . , 1 5 6 1 / A l l e n 1980 ( 2 8 ) )

-278-

Some R e l a t e d Topics

In ( 3 1 a ) and ( 3 1 b ) the

whole

PP

is

relativized.

In

(31b)

the

r e l a t i v i z e d PP is f o l l o w e d by the c o m p l e m e n t i z e r 5e. As e x p e c t e d , if the r e l a t i v i z e d p r o n o u n is an R - p r o n o u n , the P can be s t r a n d e d . This is shown in ( 3 2 ) . (32)

to u r u m e 5 e l e , 5aer we to gescaepene waeron to our land

where we for c r e a t e d were

' t o our l a n d , for w h i c h we were c r e a t e d ' ( Α Ε , Ι , 162, 19/ Vat 1978 ( 2 3 ) ) Somewhat s u r p r i s i n g l y , it is v e r y d i f f i c u l t to of Pied P i p i n g in this

construction.

Haling

(1978)

find

examples

claims

that

s t r a n d i n g is o b l i g a t o r y in 5a r - r e l a t i v e s . V a t ( 1 9 7 8 ) argues that it should be possible to f i n d instances of pied p i p i n g , a l t h o u g h Vat was u n a b l e t o f i n d a n y . A l l e n ( 1 9 8 0 ) p r o v i d e s only t h r e e e x a m p l e s of pied p i p i n g , f r o m w h i c h she c o n c l u d e s " t h a t it was an idiosyncracy of PP split that it was o b l i g a t o r y , or n e a r l y so, w i t h r e l a t i v e 5aer . " ( A l l e n 1 980 , f n t . 43 ) . One of the r e l e v a n t e x a m p l e s

is

given i n ( 3 3 ) · (33)

. . . h e waes on Simones huse 3aes lie r o w e r e s , Saerin he was in S i m o n ' s house the l e p e r ' s wherein geat daet wif 5a d e o r w y r an smerenesse on his h e a f o d p o u r e d the woman the p r e c i o u s o i n t m e n t on his head ' H e was in the house of Simon the l e p e r , w h e r e i n the woman poured the p r e c i o u s o i n t m e n t on his h e a d ' ( Β Η , ρ . 7 3 - 2 / A l l e n 1980 ( f n t . 4 3 ( i ) ) )

Let us suppose t h a t b o t h Vat and A l l e n are c o r r e c t in t h a t pied p i p i n g in 5aer-relatives is possible in p r i n c i p l e , but t h a t it is severely r e s t r i c t e d for some i d i o s y n c r a t i c r e a s o n .

If

this

is

c o r r e c t , the p a t t e r n of r e l a t i v i z a t i o n f r o m PP discussed t h u s f a r is completely similar to relativization in Dutch. Relative pronouns r e q u i r e p i e d p i p i n g , unless they a r e R - p r o n o u n s . OE and D u t c h d i f f e r w i t h r e s p e c t to the t h i r d s t r a t e g y of r e l a t i v i z a t i o n , the case in w h i c h the r e l a t i v e clause is i n t r o d u c e d by a lexical c o m p l e m e n t i z e r o n l y . This

-279-

pattern

appears

in

Dutch

Some Related Topics only in the case of i n f i n i t i v a l relatives ( o f . note 1 1 ) . In of OE this is called

the

5_£-pattern.

In

this

construction

o b j e c t of a p r e p o s i t i o n can be r e l a t i v i z e d , just as other the r e l a t i v i z e d c o n s t i t u e n t

studies the

NPs.

If

is the object of P, P must be s t r a n d e d .

An example is p r o v i d e d in (3*0 ( o f . ( 3 2 ) ) . ( 3 4 ) to urum esele . . . ,_δ£ we _t£ gescaepene wa»ron to our c o u n t r y that we for created were ' t o our c o u n t r y , for which we were c r e a t e d ' (AE I,

118, 29/ Vat 1978 ( 5 5 a ) )

This c o n s t r u c t i o n has been s u b j e c t

to

much

discussion

(cf.

1978, A l l e n 1980 and Van Kemenade 1 9 8 4 b ) . The relevant question

Vat is

whether this c o n s t r u c t i o n is the r e s u l t of movement (R-movement) followed by d e l e t i o n in COMP ( V a t ) , unbounded d e l e t i o n ( A l l e n ) or a resumptive pronoun strategy involving n e i t h e r movement nor deletion (Van K e m e n a d e ) . There are several arguments in favour of a movement approach. If t h e r e is movement we expect the Complex-NP-Constraint and the WH-island constraint to be o b e y e d . This is indeed the case. Both A l l e n and Van Kemenade argue that Subjacency should be considered a c o n d i t i o n on

coindexation

(as

proposed

( 1 9 7 8 ) ) , in o r d e r to account for the fact

by

Bresnan

that

S u b j a c e n c y . A second i n d i c a t i o n that movement is

&

Grimshaw

_ae_-relatives involved

obey

follows

f r o m the fact that -as Allen observes- "the surface position of the s t r a n d e d preposition was n e a r l y always d i r e c t l y b e f o r e the verb of the r e l a t i v e clause (or else b e f o r e ne ' n o t ' or to, which were inseparable f r o m the verb" ( A l l e n ( 1980) , p .267) . This which as far as I know is l e f t u n e x p l a i n e d in all

observation, analyses of

P - s t r a n d i n g in OE, is of course reminiscent of the condition on P - s t r a n d i n g in D u t c h , discussed in chapter 1. As argued t h e r e , this generalization follows from the Gap C o n d i t i o n , since the PP c o n t a i n i n g the gap should be c a n o n i c a l l y governed (by V). It is not i m m e d i a t e l y clear that the Gap C o n d i t i o n should apply to

unbounded

d e l e t i o n s or to the base-generated empty categories PRO and p r o . It does a p p l y to gaps r e s u l t i n g f r o m movement and to parasitic gaps, i.e. gaps r e q u i r i n g an a n t e c e d e n t in a n o n - t h e m a t i c p o s i t i o n . Since the gap in the PP in (3*0 is not p a r a s i t i c , a movement analysis is

-280-

Some R e l a t e d Topics suggested by the a d j a c e n c y of the s t r a n d e d P these arguments I shall a d o p t the

to

movement

the

analysis

verb.

Given

proposed

in

V a t ( 1 9 7 8 ) . [11]. What is relevant to

our

p a t t e r n of P-stranding in

discussion

OE observes

is

the

the

fact

left-right

that

the

asymmetry

expected of an OV-language.

4 . 4 . 1 . 2 That-_t c o n f i g u r a t i o n s Because the

external

argument

generally

precedes

the

internal

a r g u m e n t ( s ) in s u b o r d i n a t e clauses , we may conclude that OE is a SOV language. In p r i n c i p l e , INFL may appear b e f o r e or a f t e r the Vmax. There are several i n d i c a t i o n s , D u t c h , INFL should follow Vmax.

however,

First,

the

that, finite

just verb

as

in

appears

c l a u s e - f i n a l l y in s u b o r d i n a t e clauses, whereas it appears in second position in main clauses

(cf.Canale

1978,

Van Kemenade

1984a).

Examples are given in ( 3 5 ) . ( 3 5 ) a . On twam δ ingum haefde God 3aes mannes sawle gegodod w i t h two things had God

m a n ' s soul

endowed

( Α Η Τ , Ι , 1 , 1 8 3 / V a n Kemenade 1984a ( 1 c ) ) b. Saet ic 5as boc of L e d e n u m g e r e o r d e to Engilscre that I this book f r o m L a t i n language to English spraece awende tongue t r a n s l a t e ( A H T , I , p r e f . 6 / V a n Kemenade 1984a ( 4 a ) ) Just as in Dutch and G e r m a n , this phenomenon can

be

explained

by

assuming the basic position of the f i n i t e verb to be clause f i n a l . The f i n i t e verb p o s i t i o n in main clauses is then d e r i v e d by a r u l e of Verb Second ( c f . K o s t e r 1975, Den Besten 1 9 8 3 ) . A n o t h e r i n d i c a t i o n that INFL follows Vmax in OE is the fact that in OE long e x t r a c t i o n of the s u b j e c t is p e r m i t t e d , w i t h o u t y i e l d i n g a COMP-_t e f f e c t , as was shown in c h . 3 , ( 1 2 4 ) , r e p e a t e d here.

-281-

Some R e l a t e d Topics (36)

Ac ic w o l d e w i t a n hu 3e a u h t e be 5aem monnum 5e but I would know how thee seemed the man t h a t wit aer cwadon Saet unc S u h t e Saet _t Waeron we earlier said t h a t us seemed t h a t

were

w i l d d i o r u m g e l i c r a n δοηηβ monnu w i l d - b e a s t like-er than men ' B u t I would like to know how it

seemed to you about

the men t h a t we said earlier t h a t were more like w i l d beast t h a n m e n ' ( B , X X V I I I . 5 p . 1 2 2 . 1 3 / Allen 1 9 8 0 , ( 3 ) ) F o r discussion s e e A l l e n ( 1 9 7 7 , 1 9 8 0 ) , Pesetsky ( 1 9 8 2 a ) . If

INFL

follows

c a n o n i c a l l y , i.e.

Vmax

in

OE,

INFL

governs

f r o m right to l e f t , f r o m which

the

it

subject

follows

that

the subject can be e x t r a c t e d .

4 . 4 . 1 . 3 Impersonal c o n s t r u c t i o n s OE e x h i b i t s subject.

a

This

variety so-called

of

constructions impersonal

s u b j e c t of much r e s e a r c h . I m p o r t a n t

without

construction

descriptive

a

lexical NP

has

studies

been on

the this

c o n s t r u c t i o n a r e V a n d e r Gaaf ( 1 9 0 4 ) , Wahlen ( 1 9 2 5 ) , Elmer ( 1 9 8 1 ) . D i f f e r e n t analyses for the impersonal c o n s t r u c t i o n in OE and its d e v e l o p m e n t i n t o personal constructions

in Modern English have been

proposed b y J e s p e r s e n ( 1 9 2 7 ) , L i g h t f o o t ( 1 9 7 9 , 1 9 8 1 ) , T r i p p ( 1 9 7 8 ) a n d Fischer & Van der L e e k ( 1 9 8 3 ) . I shall not discuss the m e r i t s of these d i f f e r e n t proposals h e r e . I shall instead put f o r w a r d a d i f f e r e n t a n a l y s i s , which is in several respects s i m i l a r to the account of Fischer & Van der L e e k . Before discussing the relevant d a t a , I would like to propose to r e f e r to this c o n s t r u c t i o n as

the

'nominative-less construction', instead of impersonal or s u b j e c t l e s s c o n s t r u c t i o n . I p r e f e r this term because the label ' i m p e r s o n a l c o n s t r u c t i o n ' is g e n e r a l l y taken to refer to c o n s t r u c t i o n s in which the v e r b selects an NP argument which has no n o m i n a t i v e C a s e . I shall i n c l u d e c o n s t r u c t i o n s in which there is no NP at all. G i v e n the d i s c u s s i o n of the n o t i o n s u b j e c t above, the term ' s u b j e c t l e s s c o n s t r u c t i o n ' seems r e a l l y r a t h e r i n a p p r o p r i a t e .

-282-

Some R e l a t e d Topics The relevant g e n e r a l i z a t i o n involved

seems

to

be

that

constructions

are

in w h i c h there is no n o m i n a t i v e NP. If it is assumed

agreement of NP and verb is d e f i n e d on n o m i n a t i v e

Case

the lack of agreement is s i m p l y a consequence

the

of

that

(cf.ch.2), lack

of

nominative NP. Just as in D u t c h , n o m i n a t i v e - l e s s c o n s t r u c t i o n s appear passives of i n t r a n s i t i v e verbs or verbs t a k i n g a s e n t e n t i a l or

a in PP

complement . ( 3 7 ) a . Hu maeg aonne 5aer beon fram him gebeden How can then there be by him prayed ( C P , 3 3 6 , 5 / W a h l e n 1 9 2 5 , p. 12) b. be 6aem waes eac gecueden 5urh lohannis 3one godspellere about that was also spoken by John the evangelist ( C P , 4 M 5 . 18) c. Nu is gesene Sat ... Now is seen that . . . (C&S 228/ Visser 1963, I V , p . 2 1 0 9 ) A second n o m i n a t i v e - l e s s c o n s t r u c t i o n involves cases in the verb does not select an

external

argument

and

there

which is

no

i n t e r n a l argument NP , as in ( 3 8 ) . ( 3 8 ) a . sa gelamp then happened that . . . ( B H , 1 9 9 , 3 / Elmer 1981 , I ( 1 1 4 ) ) b. for3y is betere 3aet ... t h e r e f o r e is b e t t e r t h a t . . . ( B . 2 5 , 2 7 / Elmer 1981. I ( 1 1 ? ) ) The t h i r d type is the c o n s t r u c t i o n in w h i c h the verb does not select an e x t e r n a l a r g u m e n t and there is one i n h e r e n t l y C a s e - m a r k e d NP and a s e n t e n t i a l c o m p l e m e n t . These NPs take o b j e c t i v e Case, g e n e r a l l y D a t i v e , sometimes A c c u s a t i v e (sceamian, lystan and langian) . The lack of an e x t e r n a l argument can be l e x i c a l l y d e t e r m i n e d or is the consequence of passive. Examples are given in

(39)

and ( 4 0 ) .

-283-

Some R e l a t e d Topics ( 3 9 ) a . Me δ ϊ η ο δ nu Sast ... Me seems now that . . . ( B . 2 7 . 1 5 / Elmer 1981 , I ( 4 ) ) b. 3a o f S u h t e him Sat ... then rued him that ... ( 0 , 116, 14/ Elmer 1 9 8 1 , I ( 4 8 ) ) (40) Naes nanum men forgipen 3ast ... Not-was no o n e ( D ) forgiven that ... ( A H T , 4 8 , 3 / Visser 1963, I V . p . 2 1 0 9 ) In these three c o n s t r u c t i o n s OE is like D u t c h , as demonstrated with the following corresponding examples. ( 4 1 ) a . In de t u i n wordt gedanst In the garden is danced b. Daarover werd niet gesproken Thereabout was not spoken c. In de t u i n werd verteld dat ... In the garden was told that . . . d. Toen bleek dat ... Then a p p e a r e d that . . . e. Beter is dat ... Better is that .. . f . M i j bleek d a t . . . Me appeared that g. Hem i r r i t e e r t dat ... Him i r r i t a t e s that . . . h. Hem werd verteld dat ... Him was told that . . .

can

be

(37a) (37b) (37c) (38a) (38b) (39a) (39b) (40)

As in D u t c h , if the v e r b takes a propositional complement, it may sometimes be expressed by an NP (hit ( i t ) ) . If hit is present in these c o n s t r u c t i o n s it is an a r g u m e n t . It receives nominative Case if no o b j e c t i v e Case is a v a i l a b l e . Examples are given in ( 4 2 ) . ( 4 2 ) a . 3a gelamp hit ... Saet ... t h e n h a p p e n e d it that ... ( C h r . , 1 4 8 , 1 5 / Elmer 1981, I ( 1 2 6 )

-284-

Some R e l a t e d Topics b. hit ne gerist nanum r i c u m cynincge 3aet

...

it not becometh any rich king that . .. ( L i v e s , 3 8 2 , 2 6 0 / Elmer 1981, I ( 8 1 ) ) A f u r t h e r s i m i l a r i t y to D u t c h

is

that

with

some

of

the

allowing these nominative-less c o n s t r u c t i o n s the o b j e c t i v e sometimes become n o m i n a t i v e , t r i g g e r i n g agreement

with

verbs NP may

the

verb.

Dutch examples are given in ( 4 3 ) and OE in ( 4 4 ) . ( 4 3 ) a. Ons werd v e r z e c h t weg te gaan Us was requested to leave b. Wij werden v e r z e c h t weg te gaan We were r e q u e s t e d to leave ( 4 4 ) a. us sceamaS to secgenne ... us shames to say . . . •we feel ashamed to say' ( A H T , I , 3 7 0 , 1 0 0 / Elmer 1981, I ( 2 4 ) b. gif we Sonne scomiaS 5aet ... if we are ashamed that ( C P , 6 3 , 5 / Elmer 1981 , I ( 4 9 ) ) The situation becomes more complicated in those nominative-less constructions in which two NPs are i n v o l v e d . The class of verbs allowing a nominative-less c o n s t r u c t i o n w i t h a s e n t e n t i a l complement shows up in three d i f f e r e n t c o n s t r u c t i o n s if these verbs have two i n t e r n a l N P - a r g u m e n t s . Only one of these three is a nominative-less c o n s t r u c t i o n . 1. A subclass of the class of b i t r a n s i t i v e

verbs

allowing

the

N P o b j - V - S ' c o n s t r u c t i o n w i t h o u t external argument allows a nominative-less c o n s t r u c t i o n of the type N P o b j - N P g e n - V as w e l l . Examples a r e given i n ( 4 5 ) . ( 4 5 ) a . 3e scamode swelces gedwolan y o u ( D A T ) were-ashamed such a n e r r o r ( G E N ) ( B , 1 9 , 3 0 / Elmer 1981 , II

(3))

-285-

Some R e l a t e d Topics b. hine 5aes langode h i m ( D A T ) t h i s ( G E N ) longed ' H e longed f o r t h i s ' ( B H , 2 2 7 , 1 / Elmer 1 9 8 1 , I ( 7 ) ) Given that one of the NPs receives genitive Case, which is not Case assigned to d i r e c t o b j e c t s in t r a n s i t i v e assume

that

the

assignment

of

genitive

structures, Case

is

d e t e r m i n e d . A c c e p t i n g the B u r z i o g e n e r a l i z a t i o n , we do

the

we may lexically

not

expect

the verb to assign structural Case if it does not select an external a r g u m e n t . If it is assumed that the Burzio generalization only involves s t r u c t u r a l Case a s s i g n m e n t , the Case assigned by the verb to its object must be l e x i c a l . A f u r t h e r indication that genitive Case is not s t r u c t u r a l l y determined can

be

derived

from

sentences such as ( 4 5 c , d ) . ( 4 5 ) c . 3aet him w u r d e o f t o g e n Srymmes and wasda and pegnugna that him ( D A T ) was taken away glory and clothes and money and das anwalda and the p o w e r ( G E N ) ( B , 2 5 , 3 1 / Visser 1 9 6 3 , I V , p . 2 1 3 6 ) d. hi h i r e u n r i h t l i c e benumen waere he(NOM) h e r ( G E N ) unrightly deprived-of was 'She was taken away from h i m ' ( B e d e , 4 6 0 . 3 3 / Visser 1 9 6 3 , I V , p . 2 1 3 6 ) These c o n s t r u c t i o n s involve passives of verbs taking an external a r g u m e n t , an i n d i r e c t object and a genitive direct o b j e c t . In these passives the g e n i t i v e o b j e c t r e m a i n s in the g e n i t i v e , w h i l e the i n d i r e c t o b j e c t may retain its dative Case r e s u l t i n g in a n o m i n a t i v e - l e s s c o n s t r u c t i o n ( ^ 5 c ) or may become n o m i n a t i v e , as in ( 4 5 d ) . If g e n i t i v e Case had been s t r u c t u r a l l y assigned, we would have expected the d i r e c t object to lose its Case under p a s s i v i z a t i o n , which would result in c o n s t r u c t i o n s in which d i r e c t o b j e c t is moved in o r d e r to receive n o m i n a t i v e Case. The n o m i n a t i v e - l e s s

construction

in

v e r b s w i t h two i n t e r n a l a r g u m e n t s o n l y , Case-marked n o n - s t r u c t u r a l l y . Since there is

-286-

(45a,b)

thus

the

involves

both of which are no s u b j e c t position

Some R e l a t e d Topics and since t h e r e is no r e q u i r e m e n t that every subject,

these

structures

are

S

well-formed.

d e p i c t e d in ( 4 6 ) , w h e r e NP1 has o b j e c t i v e

Case

should

contain

The s t r u c t u r e and

a is

NP2 genitive

Case.

(46) V' nax

NP

INFL

1

There are two other double NP constructions associated with the set of verbs that allows nominative-less constructions of the type N P - V - S 1 . Both involve

constructions

with

a

nominative

E i t h e r NP1 or NP2 receives n o m i n a t i v e C a s e , while the Case other NP is p r e s e r v e d . 2. Let us start w i t h the c o n s t r u c t i o n in which NP2 has Case. Examples are given in ( 4 7 ) . ( 4 7 ) a . 5es sige gewear

of

NP. the

nominative

Punicum

this v i c t o r y ( N O M ) w o n t h e P u n i c i ( O B J ) ( B T / Elmer 1981 , II

(28))

b. aa o f a u h t e 5aet anum 3aes cyninges g e f e r a n then regretted t h i s ( N O M ) one of the k i n g ' s followers(OBJ) ( B T / Elmer 1981 , II

(27))

We may e x p l a i n the o c c u r r e n c e of this c o n s t r u c t i o n by assuming t h a t in these cases the verb is not able to assign lexical genitive Case. If this is so, the NP has to be moved to a position in w h i c h it may receive Case. Since n o m i n a t i v e Case is a v a i l a b l e , NP2 can be moved from the argument p r o j e c t i o n to a position governed and Case-marked by I N F L , as in ( 4 7 c ) .

-287-

in

which

it

is

Some R e l a t e d Topics

Vmax

NP

INFL

1

t2

v

Such an analysis is exactly parallel to the analysis of ergative verbs with an indirect object in D u t c h , discussed in chapter 2. Recall that these Dutch constructions were recognizable because they allowed I n v e r s i o n . If the analysis of this OE construction is c o r r e c t , we expect Inversion to occur. This is indeed the case, as is demonstrated by the following examples. ( 4 8 ) a . hu him se sige gelicade How h i m ( O B J ) this v i c t o r y ( N O M ) pleased ( 0 , 1 5 6 , 2 5 / Elmer 1 9 8 1 , II (3*0) b. gif 5am gifran ungemetlicu spraec ne eglde if the g r e e d y ( O B J ) the l o q u a c i t y ( N O M ) not grieved ( C P , 3 0 9 , 2 / Elmer 1 9 8 1 , I I ( 3 3 ) ) The appearance of Inversion in OE is natural in this construction given the fact that neither of the two NPs is an external argument and the a v a i l a b i l i t y of a d j u n c t i o n . [12]. Our analysis of this construction is based on the assumption that NP2 cannot receive Case as a consequence of lexical properties of a subclass of the verbs allowing n o m i n a t i v e - l e s s constructions of the type N P - V - S ' . If this is correct we expect a division within this class of v e r b s between verbs that assign genitive Case and verbs that do n o t . In general it seems to be true that verbs that are able to assign genitive Case do not appear in this c o n s t r u c t i o n . For e x a m p l e , both lician (to please) and lystan(to cause p l e a s u r e ) appear in the c o n s t r u c t i o n him lician/lystan 5at ..., but only lystan is able to assign genitive case. It then follows that if it takes two N P - a r g u m e n t s , lystan requires the p a t t e r n d e p i c t e d in ( 4 6 ) , i.e. N P o b j - N P g e n - V , while it cannot be

-288-

Some R e l a t e d T o p i c s found

in

however, former

the

pattern

NPnom-NPobj-V,

shown

in

(4?c}.

r e q u i r e s the l a t t e r p a t t e r n and cannot be one.

There

are

only

two

exceptions

found

to

the

observation that p a r t i c u l a r v e r b s enter in only one c o n s t r u c t i o n s as a consequence of t h e i r Case-assigning

Lician, in

the

general of these capacity.

The verbs hreowan(to r u e ) and o f 3 y n c a n ( t o r u e ) allow both p a t t e r n s . So both him aaes(GEN) hreowes and him Saet(NOM) hreowe5 can be f o u n d . O b v i o u s l y , the v e r b assigns Case o p t i o n a l l y to N P 2 . I n t e r e s t i n g l y , the two available c o n s t r u c t i o n s do not seem to c o r r e s p o n d w i t h two d i f f e r e n t semantic readings, as Although the correlated

possibility with

a

is

demonstrated

by

Elmer(1981).

of

assigning

genitive

Case

semantic

property,

such

the

as

might

be

degree

of

t r a n s i t i v i t y (Fischer & Van der Leek 1 9 8 3 ) , these exceptional

cases

demonstrate quite c l e a r l y that the e x p l a n a t i o n of the o c c u r r e n c e of these two c o n s t r u c t i o n s should be based on s y n t a c t i c p r o p e r t i e s . 3.

The

third

construction

involves

cases

in

which

NP1

is

n o m i n a t i v e and NP2 g e n i t i v e . Examples are given in ( 4 9 ) . ( 4 9 ) a . 5a se maessepreost Saes mannes o f h r e o w then t h e p r i e s t ( N O M ) that m a n ( G E N ) grieved ( L i v e s , I I , 1 4 2 , 2 6 2 / Elmer 1 9 8 1 , I I ( 4 0 ) ) b. nu b e h o f a S ure f r e o - d o m zfre godes f u l t u m e s now needeth our f r e e d o m ( N O M ) ever G o d ' s a s s i s t a n c e ( G E N ) ( L i v e s , I , 3 8 2 , 2 6 6 7 Elmer 1981 I I ( 4 4 ) ) I would like to argue that t h i s c o n s t r u c t i o n is d e r i v e d from the nominative-less c o n s t r u c t i o n by assigning n o m i n a t i v e Case to the O b j e c t i v e NP, which has been moved i n t o the government domain of I N F L . N o m i n a t i v e Case o v e r r u l e s O b j e c t i v e / D a t i v e Case as a m a r k e d p h e n o m e n o n . The s t r u c t u r e is as given in ( 5 0 ) . (50)

Vmax

INFL

-289-

Some R e l a t e d Topics

If this is correct we expect a) that the class of

verbs

appearing

in this c o n s t r u c t i o n is a proper subset of the class of verbs that a p p e a r i n t h e c o r r e s p o n d i n g nominative-less c o n s t r u c t i o n ( c f . ( 4 6 ) ) and b) that the class of verbs a p p e a r i n g in this c o n s t r u c t i o n is a proper subset of the class of verbs a p p e a r i n g in NPnom-V-S' discussed above (see ( 4 M b ) ) .

Both

the

claims

c o r r e c t . Each v e r b a p p e a r i n g in t h i s c o n s t r u c t i o n

construction seem

appears

to in

be the

nominative-less c o n s t r u c t i o n as w e l l . S i m i l a r l y , with the exception of one verb (hreowan), each verb a p p e a r i n g in this construction also shows up in the c o n s t r u c t i o n of the type in ( 4 4 ) . Given the relative exceptionality of both these c o n s t r u c t i o n s , I would like to

consider

the

fact

that

hreowan

is

not

attested

construction of the type in ( 4 4 ) an accidental gap. n e i t h e r this construction nor shows I n v e r s i o n . The order is

in

the

Interestingly,

the nominative-less construction N P o b j / n o m - N P g e n - V . As in D u t c h ,

Inversion appears only if the two NPs involved

are

both

internal

arguments. A f i n a l remark concerning this construction is that the d i f f e r e n c e between the n o m i n a t i v e - l e s s construction and this c o n s t r u c t i o n cannot be semantically m o t i v a t e d . We may thus consider this c o n s t r u c t i o n to be a s y n t a c t i c v a r i a n t of the nominative-less c o n s t r u c t i o n ( 4 6 ) . These t w o c o n s t r u c t i o n s d i f f e r only w i t h respect to the Case assigned to N P 1 . We may now draw the f o l l o w i n g t e n t a t i v e conclusions: - a n o m i n a t i v e NP is not r e q u i r e d in OE sentences; - if there is no n o m i n a t i v e N P , none of the NPs present

can

be

analysed as the s u b j e c t ; - the class of verbs that only select two i n t e r n a l arguments can be d i v i d e d into v e r b s assigning genitive Case and verbs that do n o t ; - if such a v e r b assigns genitive Case, the c o n s t r u c t i o n has no subject; - if the v e r b does not assign genitive C a s e , the Case-less NP has to be moved by a d j u n c t i o n in order to receive ( n o m i n a t i v e ) C a s e ; - N P - I n v e r s i o n a p p l i e s only if the verb does not i.e.

if

assign

one NP has to be moved for reasons of Case. Only

both NP orders s y n t a c t i c a l l y m o t i v a t e d ; - O b j e c t i v e Case (in most instances D a t i v e )

-290-

can

Genitive, then

exceptionally

are be

Some R e l a t e d Topics overruled by N o m i n a t i v e Case if t h e r e

is

no

other

NP

Nominative; W i t h the exception of the lexically d e t e r m i n e d option of

requiring assigning

Genitive Case to an i n t e r n a l a r g u m e n t , OE is similar to Dutch in all r e l e v a n t r e s p e c t s . In the next section I shall try to explain why and how the nominative-less c o n s t r u c t i o n d i s a p p e a r e d from the language.

4 . 4 . 2 Development into Modern English: the INFL parameter In this paragraph I shall p r o v i d e a t e n t a t i v e explanation of the d i f f e r e n c e s between OE and ME. Suppose OE was indeed basically similar to D u t c h , having a base s t r u c t u r e such as ( 5 1 ) . INFL m a x

(51) COMP

Wl r* IT

INFL

V

V and INFL assign s t r u c t u r a l Case to the l e f t , w h i l e P assigns Case to the r i g h t . Since V assigns Case to the l e f t , canonical government is from right to l e f t . The external argument can receive n o m i n a t i v e Case w i t h i n V m a x , since INFL ( c a n o n i c a l l y ) governs it. P-stranding of f u l l NPs is impossible since they appear in a n o n - c a n o n i c a l l y governed position for reasons of Case. The e x t e r n a l argument can be e x t r a c t e d since it is c a n o n i c a l l y governed by I N F L . Clauses w i t h o u t n o m i n a t i v e NP occur if the verb does not select an external argument and the i n t e r n a l N P - a r g u m e n t s , if p r e s e n t , may receive Case in t h e i r D - s t r u c t u r e p o s i t i o n . Now suppose that in the d e v e l o p m e n t of the language a change takes place in the position of INFL r e l a t i v e to V m a x , in such a way that I N F L precedes Vmax r a t h e r t h a n f o l l o w s it. The reason for this -291-

Some Related Topics change might be the g e n e r a l i z a t i o n

of

the

main

clause

which was derived by Verb Second. A f u r t h e r c h a n g e ,

pattern,

which

can

be

considered an immediate consequence of the change in position of I N F L , is the change from 0V to VO. This follows if we assume that V and INFL should be a d j a c e n t . The reason why we adopt a d j a c e n c y INFL and V is that we may consider the rule moving the verb more precisely one of the verbal elements) to INFL One argument in f a v o u r of the l o c a l i t y of

this

that the a p p l i c a t i o n of the r u l e is blocked if i n t e r v e n e s , as is the case in ( c f . n o t e 5). F u r t h e r m o r e , it

a

local

movement a

rule.

rule

lexical

Subject-AUX-Inversion

of (or is

subject

constructions

is clear that the rule moving

one of

the verbal elements to INFL or INFL to the V-position cannot be considered a m a j o r movement r u l e . The relevant V is not a c o n s t i t u e n t but r a t h e r the head of a c o n s t i t u e n t . M o r e o v e r , the relevant V might even be part of the V - n o d e . If ( c f . r u l e R ( C h o m s k y 1 9 8 1 ) ) , the rule

would

INFL is moved to

move

INFL

V

downwards,

leaving an ungoverned trace w i t h o u t a c-commanding

antecedent.

If

the rule relating INFL and

the

in

V is

a

local

rule,

change

position of INFL in the development of English requires a change in position of V as w e l l . [13]· One way to achieve this e f f e c t is to assume that the direction of s t r u c t u r a l Case assignment of V is reversed. Such a step would imply that the direction of canonical government changes from r i g h t - l e f t to l e f t - r i g h t . The change in P - s t r a n d i n g possibilities would be an i m m e d i a t e consequence of this s h i f t . The d i r e c t i o n of Case assignment of P remains the same. However, the object of a preposition can now be stranded in accordance with canonical g o v e r n m e n t , whereas postposition stranding with R-pronouns and personal pronouns v i o l a t e d the Gap C o n d i t i o n . This is exactly can be o b s e r v e d . A consequence of the change in position of INFL and the

what change

in the d i r e c t i o n of Case assignment of V is that the s u b j e c t w i t h i n Vmax is c a n o n i c a l l y governed by I N F L . H o w e v e r , if it is assumed that the d i r e c t i o n of Case assignment of INFL is not c h a n g e d , the s u b j e c t cannot receive Case from INFL d i r e c t l y . In order to receive Case the e x t e r n a l argument has to be moved to

a

position

to

the

l e f t of I N F L . This movement cannot be a d j u n c t i o n , according to the typology of movement rules discussed in section 2 of this c h a p t e r .

-292-

Some R e l a t e d Topics An NP position d e f i n e d by the presence of n o m i n a t i v e Case has to be created in order to move the e x t e r n a l argument as an instance of s u b s t i t u t i o n . If such a nominative position is c r e a t e d , the s t r u c t u r e will be as in ( 5 2 ) . (52)

INFL" INFL'

NPnom INFL

Given this s t r u c t u r e NP1 has to be moved to NPnom in order to receive Case. The movement from NP1 to NPnom is licit since position NP1 is canonically governed by I N F L . No extraction can take place from the NPnom position since this p o s i t i o n is not canonically g o v e r n e d . We thus expect the i n t r o d u c t i o n of the that-t e f f e c t as a consequence. Let us now t u r n to the demise of the nominative-less c o n s t r u c t i o n . There are two e f f e c t s that p l a y a decisive role in this d e v e l o p m e n t . The f i r s t e f f e c t concerns the loss of the possibility of assigning the l e x i c a l l y d e t e r m i n e d genitive Case to an internal argument of a verb that does not select an e x t e r n a l a r g u m e n t . This loss may simply follow from the fact that Case assignment is d i r e c t i o n a l . As English changed f r o m 0V to VO, we may assume that lexical Case, which was assigned from right to l e f t , is lost. It then follows that verbs can assign Case s t r u c t u r a l l y o n l y . The second i m p o r t a n t e f f e c t involves the i n t r o d u c t i o n of the position NPnom. This position cannot remain empty d u r i n g the derivation. These two factors d e t e r m i n e what is going to h a p p e n to the verbs which are involved here. Given the Burzio generalization it is p r e d i c t e d that verbs that do not select an e x t e r n a l argument do not assign s t r u c t u r a l Case. If genitive Case is lost, these verbs are not able to assign Case at all.

-293-

Some R e l a t e d Topics Let us f i r s t discuss the development of the c o n s t r u c t i o n a sentential c o m p l e m e n t and

an

inherently

Case-marked

c o n s t r u c t i o n v i o l a t e s the r e q u i r e m e n t that a n o m i n a t i v e

NP.

the

dative

NP becomes

This

NP should

be present as a consequence of N P n o m . There are two ways this p r o b l e m . E i t h e r

with

to

nominative

solve or

the

p r o p o s i t i o n a l a r g u m e n t is i n t r o d u c e d by an NP related to an a d j u n c t S'. The NP ( i t ) is moved to position N P n o m . These two c o n s t r u c t i o n s were a l r e a d y present in O E , as d e m o n s t r a t e d above. A l t h o u g h the n o m i n a t i v e - l e s s c o n s t r u c t i o n was p r e s e r v e d for a long time in exceptional cases like me seems that . . . , it

is

virtually

extinct

in the 16th c e n t u r y . Verbs like like, happen and long developed

by

moving the i n d i r e c t o b j e c t into the NPnom p o s i t i o n . Verbs like seem, grieve and become have taken the other d i r e c t i o n . W i t h respect to the double-NP c o n s t r u c t i o n , there seems to be only one s o l u t i o n at f i r s t s i g h t . The i n h e r e n t l y Case-marked NP keeps its

Case and the other NP is

moved

to

NPnom

in

receive Case. For a subclass of the relevant v e r b s , i.e.

order the

that were u n a b l e to assign g e n i t i v e , this was already the p o s s i b i l i t y in OE. This subclass i n c l u d e d verbs like lician. c o n s t r u c t i o n is strongly archaic in sentences of the t y p e it

to

verbs only This likes

me not and is found with verbs like happen and become. There a r e , h o w e v e r , two other strategies to solve the Case p r o b l e m . The f i r s t is to i n t r o d u c e a p r e p o s i t i o n to assign Case to the o b j e c t NP and to move the i n d i r e c t o b j e c t to N P n o m . This s t r a t e g y is f o u n d w i t h the verb long. The f i n a l strategy is that the v e r b assigns s t r u c t u r a l Case to the o b j e c t . Given the B u r z i o g e n e r a l i z a t i o n this implies that there should be an e x t e r n a l a r g u m e n t . This can be achieved by making the indirect object an external argument. This process of t r a n s i t i v i z a t i o n , w h i c h is of course accompanied by a change in m e a n i n g , can be observed w i t h v e r b s like like and rue. [ 1 4 ] . A l t h o u g h most OE n o m i n a t i v e - l e s s verbs have developed into one p a t t e r n or have d i s a p p e a r e d from the l a n g u a g e , it is i n t e r e s t i n g to observe t h a t in M i d d l e English most of these verbs still allow several s t r a t e g i e s to escape from the Case p r o b l e m . A l t h o u g h the r e m a i n i n g verbs are p r e s e r v e d in Modern English in a number of d i f f e r e n t w a y s , t h e y have in common that t h e i r d e v e l o p m e n t can be argued to be the result of the change in position of INFL w i t h

-294-

Some R e l a t e d T o p i c s respect to V m a x . The fact t h a t a v a r i e t y

of

changes

in

the

development

English can be a c c o u n t e d for by one change in the grammar

provides

evidence for the c o r r e c t n e s s of the h y p o t h e s i s t h a t in English external argument is base-generated w i t h i n section 3- The d i a c h r o n i c

analysis

Vmax,

proposed

as

here

the

proposed is

of

in

crucially

d e p e n d e n t on this e n d o c e n t r i c a p p r o a c h . [15].

4.5 Some remarks on the s t r u c t u r e of the Romance languages We have d e r i v e d

the

assumption

from

theory

that

the

external

argument of S s h o u l d be g e n e r a t e d as a c o n s t i t u e n t e x t e r n a l but i n t e r n a l to V m a x . This i m p l i e s t h a t we have to underlying structure of all from

English,

obvious

apparent

candidates

I t a l i a n , Spanish and F r e n c h . I

would

NP-INFL-VP are

Romance

like

to

to

reconsider

V the

languages.

Aside

languages

like

claim

that

these

languages are b a s i c a l l y languages of the t y p e [ I N F L [ [ V O ] S ] . The major a r g u m e n t in f a v o u r of t h i s a s s u m p t i o n is the f a c t that all three languages allow the external argument to appear sentence-finally in the so-called (stylistic) c o n s t r u c t i o n . Examples a r e p r o v i d e d i n ( 5 2 ) .

Inversion

( 5 2 ) a . F r . : A qui donnera ce cadeau ton f r e r e ? to whom w i l l give that p r e s e n t y o u r b r o t h e r b. S p . : T r a j o una c a r t a para mi el c r i a d o Brought a letter for me the s e r v a n t c. I t . : Scrive una lettera G i o v a n n i Writes a letter Giovanni On the assumption t h a t the e x t e r n a l a r g u m e n t is base-generated in postverbal p o s i t i o n , it f o l l o w s that a p r e v e r b a l s u b j e c t in these languages is moved f r o m p o s t v e r b a l p o s i t i o n .

-295-

Some Related Topics 4.5.1

The postverbal s u b j e c t in Spanish

For Spanish such an analysis

is

adopted

in

ContrerasC1983)

and

Groos & Bok-Bennema ( 1 9 8 5 ) . They argue that INFL is preceded by a p r a g m a t i c a l l y m o t i v a t e d XP p o s i t i o n , which is optionally filled by a c o n s t i t i t u e n t which is not necessarily the external a r g u m e n t . The optional preverbal XP position is motivated by the fact that in normal clauses only one consituent may precede the verb and that in questions the wh-word is moved into X P - p o s i t i o n , which prevents any other movement to that p o s i t i o n . This analysis is able

to

for the well-known word

(cf.Torrego

order

problem

in

questions

1984) w i t h o u t having to postulate a verb-preposing r u l e . If this is c o r r e c t , INFL is followed by V m a x , which

account

has

to

include the external a r g u m e n t . Contreras and Groos & Bok-Bennema argue that the basic ordering of Vmax is [[VO]S]. Although the surface order of c o n s t i t u e n t s following the verb

is

rather

free,

they argue that Spanish is a configurational language and that

the

surface order of a sentence such as ( 5 3 ) is derived by a d j u n c t i o n . (53)

T r a j o el c r i a d o una carta para mi brought the servant a letter for me

One of the arguments of Bok-Bennema & Groos in favour of a c o n f i g u r a t i o n a l S in Spanish involves a c e r t a i n type of parasitic gap. The contrast in ( 5 4 ) can be explained if we assume that the o b j e c t dos trajes is moved in ( 5 4 b ) by a d j u n c t i o n to the r i g h t . ( 5 4 ) a . *Ayer compre dos t r a j e s [sin probarme e_] Y e s t e r d a y I-bought two suits w i t h o u t trying-on b. Ayer compre _t [sin probarme e_] dos t r a j e s Given that the preverbal position in Spanish is clearly not a s u b j e c t position and that Spanish is a c o n f i g u r a t i o n a l language, it follows that in Spanish INFL assigns n o m i n a t i v e Case to the right to the external argument in its D - s t r u c t u r e p o s i t i o n . [16]. In this regard Spanish might be d i f f e r e n t from Italian and F r e n c h . Let us c o n c e n t r a t e on French f i r s t .

-296-

Some R e l a t e d Topics

4 . 5 . 2 Stylistic I n v e r s i o n in French In French the occurrence of a postverbal subject is severely l i m i t e d . There are basically three c o n s t r u c t i o n s in which the subject appears in final position. These constructions are w h - q u e s t i o n s , s u b j u n c t i v e clauses and p r e s e n t a t i v e clauses. [171· Examples of these three constructions are presented in ( 5 5 ) . (55)a. Quand partira ce garcon? When will-leave that boy b. Je veux que parte Paul I want that leave Paul c. II est arrive trois filles It is arrived three girls The presentative construction exemplified in ( 5 5 c ) d i f f e r s in several respects from the other two constructions. This construction requires the postverbal NP to be i n d e f i n i t e , it applies only to i n t r a n s i t i v e structures while neither the f i n i t e verb nor the participle agrees in p e r s o n , number and gender with t h e postverbal N P , o f . ( 5 6 ) . (56)a.*Il It b.*Il It c.*Il it

est arrive ce garcon is arrived this boy mangera cette t a r t e trois f i l l e s will-eat this pie three girls sont arrivees trois filles are arrived three girls

As B u r z i o ( 1 9 8 l ) observes, constructions of the type in ( 5 5 c ) appear only with s u b j e c t s that are D - s t r u c t u r e o b j e c t s . It shows up with ergative v e r b s , as in ( 5 5 c ) , which generally take the p e r f e c t i v e a u x i l i a r y etre(to be) and not avoir(to h a v e ) , w i t h passives, as in ( 5 7 a ) , and in constructions with "se-moyen", as shown in ( 5 7 b ) .

-297-

Some R e l a t e d Topics ( 5 7 ) a . II It b. II It

a ete mange p l u s i e u r t a r t e s has been eaten several pies se c o n s t r u i t b e a u c o u p d ' i m m e u b l e s dans cette ville "se" b u i l d s many b u i l d i n g s in this town

In these t h r e e

constructions

there

is

no

D-structure

external

argument as a consequence of lexical p r o p e r t i e s (ergative verbs), p a s s i v i z a t i o n or the i n t r o d u c t i o n of se. [18]. Given the g e n e r a l i z a t i o n that verbs in c o n s t r u c t i o n s w i t h o u t external argument are not able to assign Case, there is a Case p r o b l e m in these c o n s t r u c t i o n s . This problem can be solved by moving the o b j e c t into a position in which it may receive n o m i n a t i v e Case. In F r e n c h , j u s t as in E n g l i s h , this position in f r o n t of INFL is d e f i n e d by the presence of nominative Case. If we do so, the following results are o b t a i n e d . ( 5 8 ) a . Trois f i l l e s sont a r r i v e e s b. P l u s i e u r s t a r t e s ont ete mangees c. Beaucoup d ' i m m e u b l e s se c o n s t r u i s e n t dans cette ville The q u e s t i o n is how the Case problem is to be solved in the " i l - c o n s t r u c t i o n " . I would like to propose that the Case problem is solved by t r a n s i t i v i z a t i o n . II is an argument NP r e q u i r i n g Case. Suppose it d i f f e r s from other NPs in that it does not r e q u i r e a Q - r o l e . In this sense il in this c o n s t r u c t i o n is a kind of dummy p r o n o u n . If il is i n s e r t e d as the external argument in D - s t r u c t u r e , the s t r u c t u r e is s y n t a c t i c a l l y t r a n s i t i v e . The D - s t r u c t u r e o b j e c t r e m a i n s in its D - s t r u c t u r e position in which it may receive o b j e c t i v e Case. The e x t e r n a l , n o n - t h e m a t i c argument il is moved into p r e v e r b a l p o s i t i o n in order to receive nominative Case. Most of the r e s t r i c t i o n s on the occurrence of the "il-construction" f o l l o w i m m e d i a t e l y . The fact t h a t this c o n s t r u c t i o n appears only in case of i n t r a n s i t i v e s t r u c t u r e s w i t h o u t a l e x i c a l l y determined e x t e r n a l argument is a c o n s e q u e n c e of the p r o p e r t i e s of il. The f a c t t h a t t h e r e is no agreement between D - s t r u c t u r e o b j e c t and v e r b or p a r t i c i p l e f o l l o w s f r o m the f a c t t h a t agreement is d e f i n e d as a r e l a t i o n between v e r b and n o m i n a t i v e NP. In t h i s c o n s t r u c t i o n the v e r b agrees w i t h i l o n l y . [ 1 9 ] -298-

Some R e l a t e d Topics It is i n t e r e s t i n g to note that il not only a p p e a r s in the t h r e e c o n s t r u c t i o n s m e n t i o n e d a b o v e , but also in impersonal passives, as in ( 5 9 ) ( o f . Z r i b i - H e r t z 1 9 8 2 ) . ( 5 9 ) a . II a ete dormi ici r e c e m m e n t It has been slept here r e c e n t l y b. II a ete' d i s c u t e de c e t t e q u e s t i o n h i e r It has been discussed about this q u e s t i o n y e s t e r d a y This c o n s t r u c t i o n , which of course a p p e a r s only in case of non-ergative i n t r a n s i t i v e verbs (*il a ete' arrive' ici re'cemment) , demonstrates that il should be c o n s i d e r e d a n o n - t h e m a t i c n o m i n a t i v e NP in these impersonal c o n s t r u c t i o n s . [ 2 0 ] . An a d d i t i o n a l argument in f a v o u r of the a n a l y s i s in w h i c h the postverbal NP in ( 5 5 c ) and ( 5 7 ) occupies the o b j e c t p o s i t i o n in S - s t r u c t u r e concerns the fact that if the i n d e f i n i t e NP contains an empty h e a d , the clitic en is r e q u i r e d , just as with r e g u l a r objects. [21], ( 6 0 ) II * ( e n ) e s t a r r i v e trois It "en" is a r r i v e d three If it is a c c e p t e d that the " i l - c o n s t r u c t i o n " is not a c o n s t r u c t i o n with a p o s t v e r b a l s u b j e c t , o n l y two c o n s t r u c t i o n s are l e f t . Both ( 5 5 a ) a n d ( 5 5 b ) d i f f e r considerably from the " i l - c o n s t r u c t i o n " . The p o s t v e r b a l NP agrees w i t h the v e r b and p a r t i c i p l e , as in ( 6 l a ) , there is no d e f i n i t e n e s s c o n s t r a i n t , as shown in ( 6 t b ) , and the v e r b can be t r a n s i t i v e , as in ( 6 l c ) . ( 6 1 )a. Je veux que v i e n n e n t 35 personnes I want that come 35 p e o p l e b. A q u e l l e h e u r e sont arrive'es ces f i l l e s ? At what time are a r r i v e d these g i r l s c. Que v o u l a i t manger _t ce j e u n e homme? What wanted eat that young man These c o n s t r u c t i o n s seem to i n v o l v e clear i n s t a n c e s of p o s t v e r b a l s u b j e c t s . They d i f f e r f r o m all other c o n s t r u c t i o n s in that the -299-

Some R e l a t e d Topics preverbal N P - p o s i t i o n is not f i l l e d . I would like to claim that

in

these cases there is no preverbal n o m i n a t i v e NP position. The e x t e r n a l argument can receive Case d i r e c t l y from INFL without moving to a preverbal p o s i t i o n . This can be expressed by ( 6 2 ) . (62)

(62)

INFL assigns n o m i n a t i v e Case to the l e f t , unless it appears in the domain of WH or [ + S u b j u n c t i v e ]

expresses that if INFL is in the domain of WH or

subjunctive

Case can be assigned in both d i r e c t i o n s . One immediate advantage of this analysis is that it follows that ( 6 3 ) is ungrammatical. (63)

*Qu' il v o u l a i t manger ce jeune homme? What it will-eat that young man

In other analyses involving rightward movement of the subject and a dummy status of il, we expect it to be possible for the dummy pronoun to occupy the position vacated by the downward movement

of

the s u b j e c t . If INFL assigns Case to the right as a consequence

of

( 6 2 ) there is no NP position to the l e f t of I N F L , since such a position is m o t i v a t e d only in case n o m i n a t i v e Case is assigned to the l e f t . If INFL assigns Case to the l e f t , ( 6 3 ) is ungrammatical since there are two NPs and only one Case. A n o t h e r i n t e r e s t i n g argument in favour of such an analysis can be derived from an observation made by Kayne & P o l l o c k ( 1 9 7 8 ) . They point out that subjectless impersonal sentences appear in Inversion environments o n l y . Impersonal passives w i t h o u t il can be found both in s u b j u n c t i v e c o m p l e m e n t s , as in ( 6 4 a ) , and in wh-clauses, as in (64b). ( 6 M ) a . J ' e x i g e que soit mis fin ä ce c o n f l i t I r e q u i r e that be put end to this c o n f l i c t b . T Q u a n d sera mis fin au c o n f l i t ? When w i l l - b e put end to the c o n f l i c t These il-less

impersonals

are

completely

ungrammatical

complement is not s u b j u n c t i v e or if t h e r e is no wh-movement the r e l e v a n t c l a u s e .

-300-

if

the

within

Some R e l a t e d Topics

( 6 5 ) a . * E l l e d i t que sera mis fin ä ce c o n f l i t She says that w i l l - b e put end to this c o n f l i c t b.*Qui t ' a dit que sera mis fin ä ce c o n f l i t ? Who you has told that will-be put end to this c o n f l i c t The sentences in ( 6 5 ) r e q u i r e a p r e v e r b a l i l , whereas the occurrence of il is optional in sentences of the type in ( 6 4 ) . The correlation between the p o s s i b i l i t y of postverbal subjects and occurrence of s u b j e c t l e s s favour of

the

underlying

sentences postverbal

constitutes position

an of

argument the

in

external

argument. In the exceptional case that I N F L is able to assign to the r i g h t , no NP position has to be c r e a t e d . E x t e r n a l

the

Case

arguments

receive Case in their D - s t r u c t u r e position and Case-less internal arguments by a d j u n c t i o n to a position w i t h i n the government domain of

INFL.

If

there

is

no

external

argument

nor

an

internal

NP-argument r e q u i r i n g Case, the sentence may be subjectless w i t h o u t violating any p r i n c i p l e . The s i t u a t i o n is e x a c t l y the m i r r o r of D u t c h . If INFL assigns Case to the l e f t ,

which

is

the

image normal

case, nominative Case d e f i n e s a p o s i t i o n , which has to be f i l l e d d u r i n g the d e r i v a t i o n , t h e r e b y preventing subjectless sentences of the type in ( 6 5 ) .

4.5.3

The s t r u c t u r e of I t a l i a n and the syntax of ne

The relevant question is whether Italian is

like

French

Spanish. If it is like Spanish, the external argument

is

or

like

assigned

Case in its postverbal p o s i t i o n , from which it can o p t i o n a l l y be moved to a p r e v e r b a l p o s i t i o n . If Italian is like French, n o m i n a t i v e Case is assigned to the p r e v e r b a l NP p o s i t i o n and the postverbal NP has to be moved into that position or coindexed w i t h an empty category in preverbal position in order to receive Case. N e i t h e r of these two o p t i o n s is very a t t r a c t i v e . It is not clear at all that I t a l i a n has a p r e v e r b a l XP p o s i t i o n with a topic f u n c t i o n , as in Spanish. It r a t h e r seems to be the case that there is a preverbal NP p o s i t i o n . If t h i s is correct we have to INFL assigns Case to the l e f t in order to d e f i n e such

-301-

assume that a position.

Some R e l a t e d Topics On the other h a n d , if it

is

p r e v e r b a l NP p o s i t i o n , it

t u r n s out t h a t we have

there is a empty

dummy

assumed

pronoun

that

that

there

can

be

exists to

such

assume

inserted

a

that

in

that

position in order to t r a n s m i t Case to a p o s t v e r b a l lexical s u b j e c t . A t h i r d p o s s i b i l i t y is that n o m i n a t i v e Case can be assigned in both d i r e c t i o n s . If it is

assigned

to

the

left,

it

defines

p o s i t i o n . If it is assigned to the r i g h t , it assigns Case

an NP directly

to a postverbal N P . [ 2 2 ] . Let us see whether there are any arguments in f a v o u r of one of these a p p r o a c h e s . The o c c u r r e n c e of ne in r e l a t i o n to q u a n t i f i e d NPs w i t h an empty head might p r o v i d e some i n d i c a t i o n as to which o p t i o n should be p r e f e r r e d . As shown

by

Belletti

there is an asymmetry between NP-movement and

& Rizzi(198l) ,

wh-movement

in

the

ne-construction. This asymmetry is demonstrated in ( 6 6 ) . ( 6 6 ) a . Q u a n t i * ( n e ) h a letti G i a n n i ? How-many "ne" has read G i a n n i b. Tre ( * n e ) sono stati l e t t i da Gianni Three "ne" have been read by Gianni If the o b j e c t is moved by wh-movement ne is

obligatorily

present,

whereas ne m u s t be absent if the o b j e c t is moved by NP-movement. In chapter 1, it was argued that the position relevant to b i n d i n g r e l a t i o n s is the position in the chain to which Case If we assume t h a t the moved

object

is

assigned

is

Case

assigned. in

object

p o s i t i o n i n ( 6 6 a ) b u t n o t i n ( 6 6 b ) , t h e contrast between ( 6 6 a ) a n d ( 6 6 b ) can be e x p l a i n e d . In ( 6 6 a ) , but not in ( 6 6 b ) , the clitic ne c-commands the q u a n t i f i e d NP at the relevant level. A related a s y m m e t r y involves p o s t v e r b a l s u b j e c t s , as d e m o n s t r a t e d in ( 6 7 ) . ( 6 7 ) a . * ( n e ) sono passate t r e "ne" are elapsed t h r e e b . * ( * n e ) hanno p a r l a t o t r e "ne" have spoken three As has been argued by B u r z i o ( 1981 ) , the class of verbs r e q u i r i n g ne in this c o n s t r u c t i o n is the class of ergative v e r b s , which take the a s p e c t u a l a u x i l i a r y essere. If we want to base the

-302-

Some R e l a t e d Topics account of these a s y m m e t r i e s on c-command of the q u a n t i f i e d

NP by

ne at the level r e l e v a n t for b i n d i n g r e l a t i o n s ,

assume

that the p o s t v e r b a l NP in ( 6 ? a ) o c c u p i e s that

level,

as

is

assumed

R i z z i ( 1 9 8 l ) . This i m p l i e s

that

by

the

we have object

Burzio(198l)

nominative

and

Case

o b j e c t p o s i t i o n , where these ergative s u b j e c t s are

is

to

position

on

Belletti

&

assigned

in

base-generated.

An a l t e r n a t i v e a n a l y s i s in which the ergative s u b j e c t is adjoined to V in order to receive Case f r o m INFL d i r e c t l y -assuming that INFL assigns Case to the r i g h t - is i m p o s s i b l e , since we would expect the contrast in ( 6 7 ) to show u p . Since INFL is not govern an NP w i t h i n the a r g u m e n t p r o j e c t i o n of V

able

to

structurally,

we

are forced to adopt an a n a l y s i s in which the postverbal NP receives Case by c o i n d e x i n g . Let

us

thus

not

suppose

nominative that

there

exists an empty dummy pronoun in I t a l i a n , which is able to t r a n s m i t Case to a coindexed N P . Such an analysis is s i m i l a r to the analyses in B u r z i o ( 1 9 8 l ) and R i z z i ( 1 9 8 2 ) in most r e s p e c t s . It then follows that in I t a l i a n n o m i n a t i v e Case is assigned by INFL to the l e f t . In order to r e a l i z e n o m i n a t i v e Case and to f i l l the p r e v e r b a l NP p o s i t i o n , there are two s t r a t e g i e s . E i t h e r a p o s t v e r b a l NP which does not receive Case in its D - s t r u c t u r e position is moved to the preverbal position or the e m p t y dummy is i n s e r t e d in the p r e v e r b a l p o s i t i o n , assigning Case to a p o s t v e r b a l NP by c o i n d e x a t i o n . [ 2 3 ] . In c h a p t e r 2 it has been argued t h a t it is r a t h e r u n a t t r a c t i v e to adopt the existence of empty dummy p r o n o u n s . G i v e n the l i m i t e d a d o p t i o n of dummy pronouns in general and the p r o - d r o p s t a t u s of I t a l i a n , the u n a t t r a c t i v e n e s s of the a d o p t i o n of an empty dummy pronoun in I t a l i a n is somewhat m i t i g a t e d . H o w e v e r , t h e r e is an a l t e r n a t i v e a n a l y s i s in which an i n t e r n a l a r g u m e n t t h a t is not Case-marked by the verb may receive n o m i n a t i v e Case in its D - s t r u c t u r e p o s i t i o n . This involves the process of Chain-government proposed in

Den B e s t e n ( 1 9 8 2 ) .

Although

I

have

argued

against

Chain-government as a necessary mechanism to a c c o u n t for I n v e r s i o n c o n s t r u c t i o n s in D u t c h in c h a p t e r 2, it may p r o v i d e us w i t h a s o l u t i o n to the p r o b l e m discussed in this s e c t i o n , w i t h o u t having to adopt the e x i s t e n c e of an e m p t y dummy p r o n o u n . In such an approach t h e D - s t r u c t u r e o b j e c t receives n o m i n a t i v e Case e i t h e r f r o m INFL d i r e c t l y , in w h i c h case it has to be moved to p r e v e r b a l p o s i t i o n , or f r o m V by c h a i n - g o v e r n m e n t .

-303-

Some R e l a t e d Topics The contrast between NP-movement

and

wh-movement

illustrated

in

( 6 8 ) follows in b o t h analyses. ( 6 8 ) a. Tre ( * n e ) sono passate Three "ne" are elapsed b. Quanti * ( n e ) sono passate How many "ne" are elapsed The obligatory absence of ne in ( 6 8 a ) indicates that nominative Case is assigned in p r e v e r b a l p o s i t i o n . Since according to the T R - c o n d i t i o n ( c h . 1 ) the position in which an NP is

Case-marked

relevant for establishing b i n d i n g r e l a t i o n s , ne does not the q u a n t i f i e d NP at the

relevant

level.

In

both

is

c-command

theories

the

p r e v e r b a l NP must receive Case in its s u r f a c e position. In the ' d u m m y - p r o n o u n a p p r o a c h ' nominative Case can be assigned to a postverbal NP by c o i n d e x a t i o n . If a dummy pronoun is postverbal NP cannot be moved to

preverbal

position,

present, since

the this

position is occupied by d u m m y - p r o . In a 'chain-government a p p r o a c h ' a p r e v e r b a l NP position is present only if the INFL assigns Case to that position. If the postverbal NP is assigned Case by chain government no p r e v e r b a l NP position is present. The obligatory presence of ne in ( 6 8 b ) follows as w e l l . In both analyses the preverbal NP position is not canonically governed by I N F L . It then follows that wh-movement can take place f r o m postverbal position only. Both theories d i f f e r w i t h respect to the postverbal occurrence of external a r g u m e n t s . In the dummy approach it has to be assumed that the p o s t v e r b a l external argument receives Case under coindexing w i t h an empty dummy in preverbal p o s i t i o n . This analysis has the a d v a n t a g e that we can assume that INFL assigns Case u n i d i r e c t i o n a l l y . The chain-government approach implies that we should allow INFL to assign Case to both sides. I shall leave the choice between these d i f f e r e n t approaches f o r f u r t h e r research. W i t h r e s p e c t to the syntax of ne, we are now able to f o r m u l a t e the following p r i n c i p l e .

-304-

Some R e l a t e d Topics ( 6 9 ) In the context [ N p Q P R O ] PRO must be bound if

the NP

is c a n o n i c a l l y g o v e r n e d If the q u a n t i f i e d NP receives Case in o b j e c t p o s i t i o n ,

the

ne is able to bind P R O , in accordance w i t h ( 6 9 ) . If the NP is an external a r g u m e n t in postverbal p o s i t i o n , bound since the postverbal NP is

canonically

clitic

quantified

PRO should

governed

by

be

INFL.

Since ne cannot c-command a n y t h i n g o u t s i d e the argument

projection

of V, ne cannot bind P R O . This accounts for the data in

(6?b).

preverbal position the q u a n t i f i e d NP is not canonically which implies that PRO can be f r e e , as in ( 6 6 b ) .

governed,

Finally,

adverbial

quantified

NPs are

not

allowed

In

in

postverbal p o s i t i o n , as is shown in ( 7 0 ) . ( 7 0 ) Gianni * ( * n e ) e rimasto tre a M i l a n o Gianni "ne" is remained three in M i l a n o Just as postverbal e x t e r n a l a r g u m e n t s , these q u a n t i f i e d NPs are canonically governed by INFL and o u t s i d e the c-command domain of ne. C o n s e q u e n t l y , both options

are

excluded.

However,

adverbials are in a l e f t - d i s l o c a t e d p o s i t i o n , canonically g o v e r n e d . It f o l l o w s that u n b o u n d allowed. This is shown in ( 7 1 ) .

if

these

they are not PRO should be

( 7 1 ) (Di questi s e t t i m a n e ) Due e r i m a s t o a Milano Of these weeks

two he is r e m a i n e d in M i l a n o

A similar analysis of the q u a n t i t a t i v e c o n s t r u c t i o n based on ( 6 9 ) is possible for both French and D u t c h . In F r e n c h , objects and impersonal ' s u b j e c t s ' r e q u i r e the presence of en. ( 7 2 ) a . II * ( e n ) a vu deux He "en" has seen two b. II * ( e n ) est a r r i v e t r o i s It "en" is a r r i v e d t h r e e In p r e v e r b a l position en is not possible c - c o m m a n d . Only u n b o u n d PRO is p o s s i b l e .

-305-

because

of

lack

of

Some R e l a t e d Topics

( 7 3 ) a . Trois ( * e n ) sont a r r i v e s Three

"ne" are a r r i v e d

b. Le j o u r ou aucune n ' ( * e n ) est venue The day when no-one is come In Inversion s t r u c t u r e s the two options are not

allowed

(cf.Kayne

1979). ( 7 4 ) Le jour ou n ' * ( * e n ) est venue aucune The day when NEC "en" is come no-one In this c o n s t r u c t i o n , t h e r e is a m i n i m a l

contrast

between

French

and I t a l i a n . This contrast is d e m o n s t r a t e d in ( 7 5 ) . ( 7 5 ) a. *Le j o u r ou en sont venus deux b.

Ne sono v e n u t i due

The e r g a t i v e s u b j e c t in p o s t v e r b a l p o s i t i o n in I t a l i a n can be bound by ne, as in ( 7 5 b ) , but an ergative s u b j e c t bound by en ( c f . P o l l o c k assumptions

about

Case

1984).

This

assignment

in

contrast made

French

cannot

be

follows

from

the

Italian

an

above.

In

e r g a t i v e s u b j e c t may remain in object p o s i t i o n , receiving Case by c o i n d e x a t i o n or c h a i n - g o v e r n m e n t . In F r e n c h an ergative s u b j e c t has to be moved f r o m the argument p r o j e c t i o n of V in order to receive Case s t r u c t u r a l l y . This can be done e i t h e r by moving it into p r e v e r b a l p o s i t i o n or by a d j u n c t i o n to V . In either case the position in w h i c h Case is

assigned

to

the

ergative

subject

is

outside the c-command domain of en. The occurrence of unbound PRO in ( 7 4 ) is impossible since the NP is c a n o n i c a l l y governed by I N F L . As in I t a l i a n there is a contrast between NP-movement and wh-movement in F r e n c h , as d e m o n s t r a t e d in ( 7 6 ) . ( 7 6 ) a . Deux ( * e n ) ont ete' a c h e t e t_ Two "en" have been bought b. C o m b i e n * ( e n ) a-t-il achete _t_ How many "en" has he bought

-306-

Some R e l a t e d Topics In D u t c h the s i t u a t i o n

is

apparently

quite

different

Italian and F r e n c h . It appears to be the case that unbound

from

PRO is

never a l l o w e d . Some of the r e l e v a n t cases are given in ( 7 7 ) . [ 2 4 ] . ( 7 7 ) a . G i s t e r e n z i j n * ( e r ) twee gekomen Yesterday are "er"

two come

b . Gisteren h e b b e n * ( e r ) twee gewandeld Yesterday have "er"

two w a l k e d

c. Gisteren hebben * ( e r ) twee een boek gekocht Yesterday have "er"

two a book bought

d. G i s t e r e n heb ik * ( e r ) twee gekocht Y e s t e r d a y have I "er" two bought e. Van die vier weken ben ik * ( e r ) twee in M i l a a n gebleven Of those four weeks am I "er" two in M i l a n o remained f . G i s t e r e n z i j n * ( e r ) twee gekocht Yesterday are "er" two bought g . Hoeveel h e b j i j * ( e r ) gekocht? How-many have you "er" bought In all the c o n s t r u c t i o n s given above er is o b l i g a t o r i l y p r e s e n t . In ( 7 7 a ) the q u a n t i f i e d NP is an e r g a t i v e s u b j e c t ; in ( 7 7 b )

it

is

a

n o n - e r g a t i v e s u b j e c t of an i n t r a n s i t i v e v e r b ; in ( 7 7 c ) it is a s u b j e c t of a t r a n s i t i v e verb and in ( 7 7 d ) an o b j e c t . In ( 7 7 e ) er binds an a d j u n c t N P . ( 7 7 f ) is an instance of NP-movement and ( 7 7 g ) of wh-movement. H o w e v e r , given

the

structure

of

Dutch

and

the

p r o p e r t i e s of er, this is e x a c t l y what we e x p e c t . The q u a n t i f i e d NP is c a n o n i c a l l y governed in all cases, w h i c h excludes the a p p e a r a n c e of u n b o u n d P R O . From the f a c t that er is not a c l i t i c on V but r a t h e r a weak a d v e r b i a l t h a t is moved to the l e f t by a d j u n c t i o n , it follows t h a t er can c-command any NP w i t h i n S. [ 2 5 ] . What I have t r i e d to show in this section is t h a t it is i n d e e d possible to consider F r e n c h , Spanish and I t a l i a n to have the u n d e r l y i n g s t r u c t u r e INFL [ [ V O ] S ] , although a p p a r e n t l y they have t h e u n d e r l y i n g s t r u c t u r e N P INFL V P . O f

course,

many

potentially

relevant issues have been l e f t out of c o n s i d e r a t i o n , such as the d i s t r i b u t i o n of si in I t a l i a n , the A u x - t o - C O M P r u l e , the d i f f e r e n c e between ce and il in F r e n c h e t c . A f u l l d i s c u s s i o n of all issues in t h e s e languages would t a k e us too far a f i e l d .

-307-

relevant

Some R e l a t e d Topics

4.6 The pronoun es in German In this section I shall discuss the question as to whether there are dummy pronouns in German. I shall mainly concentrate on those aspects in which German d i f f e r s from D u t c h . In most r e s p e c t s , the d i s t r i b u t i o n of German es is similar to that of het in D u t c h . [26]. It shows up as a neuter pronoun in ( ? 8 a ) , as the external argument with weather-verbs in ( 7 8 b ) and as a resumptive pronoun, i.e. a pronoun in argument position related to a c-commanding S T in A ' - p o s i t i o n . This resumptive pronoun may remain in object position if Case can be assigned, as in ( 7 8 c ) . If n o t , it has to be moved to a position in which it may receive nominative Case, as shown by ( 7 8 d ) . ( ? 8 ) a . Das Kind war k r a n k . Es blieb deshalb zu hause The child was ill. It remained t h e r e f o r e home. b. Es regnet It rains c. V i e l e bedauern es, dass er nicht auch in Bonn die Many regret it that he not also in Bonn the erste Geige spielt f i r s t violin plays d. Mir ist es angenehm, dass er nur E i n f l u s s im Süden hat Me is it nice that he only i n f l u e n c e in the south has In all these cases the d i s t r i b u t i o n of es can be explained in the same way as the d i s t r i b u t i o n of Dutch het, discussed in c h . 2 . As p r e d i c t e d , es may be absent in constructions of the type i l l u s t r a t e d in ( ? 8 c , d ) , as demonstrated in ( 7 9 ) . ( 7 9 ) a . Viele bedauern dass er ... Many regret that he ... b. Mir ist angenehm dass er ... Me is nice that ... H o w e v e r , t h e r e is one class of c o n s t r u c t i o n s in which es which d i f f e r s q u i t e s u b s t a n t i a l l y from D u t c h . This

-308-

appears, concerns

Some R e l a t e d Topics sentences of the type i l l u s t r a t e d in ( 8 0 ) . ( 8 0 ) a . Es lebte hier e i n m a l ein Mann I t / T h e r e lived here once a man b. Es hat jemand a n g e r u f e n I t / T h e r e has someone called c. Es w u r d e h a r t g e a r b e i t e t I t / T h e r e is hard w o r k e d In the case of impersonal passives, as in an i n d e f i n i t e e x t e r n a l a r g u m e n t ,

as

in

( 8 0 c ) , or sentences with (80a,b).

es

may appear

s e n t e n c e - i n i t i a l l y . As discussed in c h . 2 & 3 , the c o r r e s p o n d i n g sentences in Dutch may take the p r a g m a t i c a l l y m o t i v a t e d dummy er, both in f i r s t position and w i t h i n S. The c o n s t r u c t i o n e x e m p l i f i e d in ( 8 0 ) does not only d i f f e r

from D u t c h , but also

c o n s t r u c t i o n s in

shows

which

es

up.

from

One i m p o r t a n t

between ( 8 0 ) and ( 7 8 ) is t h a t es in ( 8 0 ) is allowed

as

the

other

difference the

first

constituent in main clauses (in C O M P ) o n l y , w h e r e a s es in ( 7 8 ) may appear w i t h i n S as w e l l . This d i f f e r e n c e is i l l u s t r a t e d in ( 8 1 ) . ( 8 l ) a . M i r i s t ( e s ) a n g e n e h m dass . . . Me is it nice that . . . b. . . . d a s s ( e s ) mir angenehm ist dass that it me nice is that . . . c. Hier lebte (*es) einmal ein Mann Here lived it

...

once a man

d. ...dass (*es) hart gearbeitet wurde that it hard worked was W i t h respect to the c o n s t r u c t i o n in ( 8 0 ) t h e r e are two r e l a t e d logically i n d e p e n d e n t q u e s t i o n s : -why does es appear in sentences of t y p e ( 8 0 ) ? -why is es not allowed in ( 8 l c , d ) ? Suppose we take the o p t i m a l point of view that

es

is

but

basically

s i m i l a r in all the c o n s t r u c t i o n s in w h i c h it o c c u r s . This implies that in all cases es is a r e f e r e n t i a l expression bearing a t h e m a t i c role and r e q u i r i n g C a s e . If we do so, we are able to e x p l a i n occurrence of es in ( 7 8 ) in a way analogous to the e x p l a n a t i o n

-309-

the of

Some R e l a t e d Topics the occurrence of D u t c h het. At

the

same

time

account for the fact that es does not a p p e a r in

we

are

(8lc,d).

able

to

In

both

( 8 l c ) and ( 8 l d ) t h e r e is no θ-role available for es. [ 2 7 ] . It

then

f o l l o w s that the analysis which was developed for Dutch het is able to account for the c o m p l e t e d i s t r i b u t i o n of German es, with the e x c e p t i o n of the occurrence of es in f i r s t position in ( 8 0 ) . W i t h i n the l i t e r a t u r e t h e r e account

for

the

occurrence

are

of

es

two analyses

available

in

Haiman(1974) ,

(80).

In

to

B r e c k e n r i d g e ( 1 9 7 5 ) , M c C r a y ( 1 9 8 D and S a f i r ( 1 9 8 M ) it is assumed that es is inserted in f i r s t position in COMP, as a dummy p r o n o u n that is necessary to s a t i s f y the

V2 r e q u i r e m e n t

in

main

clauses.

problem w i t h these accounts is that they consider es in ( 7 8 b - c )

A to

be a dummy p r o n o u n as w e l l . Since es in c o n s t r u c t i o n s of the type in ( 7 8 b - c ) does not necessarily appear in f i r s t p o s i t i o n , it has to be assumed that dummy es can be inserted in several

positions. We

then need a kind of f i l t e r to exclude the appearance of es w i t h i n S in s p e c i f i c c o n f i g u r a t i o n s . An a l t e r n a t i v e approach is presented in Den Besten ( 1 9 8 3 ) . He claims t h a t es is generated as a ( d u m m y ) pronoun in s u b j e c t p o s i t i o n . From this position it can be moved to initial position by the general rule of C o n s t i t u e n t P r e p o s i n g . F u r t h e r m o r e , there is r u l e of ES-deletion, which may delete es if it

immediately

C O M P . One problem w i t h this analysis is the fact that the

a

follows rule

of

^

es-deletion must be

either

obligatory,

d e p e n d i n g on the c o n s t r u c t i o n demonstrated below.

in

optional

which

es

or

impossible,

appears,

as

is

( 8 2 ) a . dass ( * e s ) g e t a n z t w u r d e t h a t "es" danced was b. dass ( e s ) ihm angenehm ist t h a t "es" him nice is

...

c. dass * ( e s ) regnet t h a t it rains In o r d e r to account for the d i s t r i b u t i o n of es, I shall assume that es is always an NP r e q u i r i n g a t h e m a t i c r o l e . A f u r t h e r a s s u m p t i o n is t h a t a t h e m a t i c r o l e is assigned to the Topic p o s i t i o n , i.e. the f i r s t p o s i t i o n in main clauses. This t h e m a t i c

-310-

Some R e l a t e d Topics role is an a d j u n c t role if a c o n s t i t u e n t that already

possesses

a

t h e m a t i c role is moved to the f i r s t position ( c f . Z u b i z a r r e t a 1 9 8 2 ) . If n o t h i n g is moved into the f i r s t p o s i t i o n , es can be i n s e r t e d Topic p o s i t i o n , r e c e i v i n g t h i s t h e m a t i c r o l e .

From

the u n g r a m m a t i c a l l y of es in ( 8 2 a )

since

follows,

this

is able to account for

the

appearance

of

es,

it

analysis

there

available t h e m a t i c role to assign to es. A l t h o u g h such an

in

is

no

analysis

predicts

the

a c c e p t a b i l i t y of sentences of the type in ( 8 3 ) . (83)

*Er wurde getanzt He was danced

In this sentence er may receive the t h e m a t i c role ' t o p i c ' , j u s t as es. The main d i f f e r e n c e between ( 8 0 ) and ( 8 3 ) is that the NP in topic position is r e f e r e n t i a l in ( 8 3 ) but not in ( 8 0 ) .

In

1, it was proposed that the r e f e r e n t i a l i t y of

related

Case. No Case is differs

f r o m all

assigned

to

Topic

other NPs in that it

NPs is

position.

Suppose may not

r e f e r e n t i a l l y . If this is c o r r e c t , it f o l l o w s t h a t es non-referential

thematic

NP,

to

that

does not r e q u i r e C a s e ,

corresponds w i t h the o b s e r v a t i o n that es may or grammatical as a

chapter

be

es

which used

in

(80)

is

whereas

(83)

is

u n g r a m m a t i c a l since the NP generated in Topic p o s i t i o n is not assigned C a s e . One might o b j e c t that in ( 8 3 ) n o m i n a t i v e Case is a v a i l a b l e for the NP in i n i t i a l p o s i t i o n . H o w e v e r , if that NP is assigned n o m i n a t i v e Case, it

must have been moved f r o m

a

position

w i t h i n S, to which Case is assigned s t r u c t u r a l l y . Such a d e r i v a t i o n is impossible since no such p o s i t i o n is p r o v i d e d by the base r u l e s . Since passive verbs do not assign an e x t e r n a l t h e m a t i c r o l e , no NP can be generated as e x t e r n a l a r g u m e n t . Base g e n e r a t i o n of the NP in ( 8 3 ) w i t h i n S is e x c l u d e d for t h e m a t i c reasons while generation in Topic position is impossible for reasons of o c c u r r e n c e and d i s t r i b u t i o n of es can thus be the f o l l o w i n g two s t a t e m e n t s . [28].

C a s e / r e f e r e n c e . The accounted for w i t h

( 8 4 ) a. es is an NP, c h a r a c t e r i z e d by [ + Θ , +C] b. the Topic p o s i t i o n is assigned an ( a d j u n c t ) t h e m a t i c role

-311-

Some Related Topics Although we have provided an account of the occurrence and the d i s t r i b u t i o n of es, we may still wonder why German d i f f e r s from Dutch with respect to the construction e x e m p l i f i e d in ( 8 0 ) . Put d i f f e r e n t l y , why is there no pronoun similar to Dutch er in German? To some e x t e n t , German appears to have a pronominal element that is similar to Dutch er. This is the adverbial pronoun da. Da appears as a pure locative p r o n o m i n a l , as in ( 8 5 a ) , and as an argument of P, as in ( 8 5 b ) . (85)a. Ich habe dich da gesehen I have you there seen b. Ich habe nicht damit gerechnet I have not there on counted As the R-pronouns in D u t c h , da precedes P , whereas NPs follow P. We may thus try to develop an analysis along the lines of the analysis given in c h . 3 for er. The relevant German pronouns d i f f e r from Dutch in that none of them can be used in a way similar to the Dutch expletive and q u a n t i t a t i v e use of er. This suggests that the class of adverbial pronouns in German cannot be specified as n o n - t h e m a t i c , which is the common property of q u a n t i t a t i v e and e x p l e t i v e er. As indicated in chapter 3, the ultimate selection of a p a r t i c u l a r item as non-thematic is p u r e l y a language s p e c i f i c p r o p e r t y . It is thus in no way surprising to f i n d a German paradigm of adverbial pronouns which is similar to the Dutch paradigm with the exception that they are not selected as non-thematic c o n s t i t u e n t s . An advantage of the theory developed in chapter 3 is that it allows us to express the d i f f e r e n c e between Dutch and German with respect to the occurrence and d i s t r i b u t i o n of adverbial pronouns by one parameter. [29].

4.7 It as a r e f e r e n t i a l expression On the basis of what has been argued so far it should be possible to provide a consistent analysis of English it. In this section I shall m a i n t a i n that it is similar to Dutch het in all relevant respects. D i s t r i b u t i o n a l d i f f e r e n c e s follow from independent

-312-

Some R e l a t e d Topics d i f f e r e n c e s between Dutch and E n g l i s h . We f i n d it in all the English

counterparts

to

the

constructions discussed in c h a p t e r 2. It can be the subject of w e a t h e r - v e r b s , it may be a regular non-human pronoun, it ( o p t i o n a l l y ) appears as an object w i t h verbs taking a propositional c o m p l e m e n t , it appears as the subject of raising verbs t a k i n g f i n i t e complements, etc. Examples a r e p r o v i d e d i n ( 8 6 ) . ( 8 6 ) a . It rains

b. c. d. e. f. g.

I see it I regret it that ... It seems that ... I consider it u n l i k e l y that It is u n l i k e l y that ... It pleases me t h a t ...

...

In g e n e r a l , the analysis d e f e n d e d in chapter 2 for Dutch het carries over to English it. H o w e v e r , there are two i m p o r t a n t d i f f e r e n c e s between D u t c h and English. The appearance of it is optional only if it is the direct o b j e c t of the v e r b , i.e. in sentences of the type in ( 8 6 c ) . In chapter 2 we saw that the occurrence of het is optional in other constructions as w e l l . A second d i f f e r e n c e between Dutch and English involves the fact that in English w h - e x t r a c t i o n from S' is possible even if it is p r e s e n t . We have seen that the a v a i l a b i l i t y of e x t r a c t i o n in Dutch is generally confined to constructions w i t h o u t het. P o t e n t i a l l y , these two d i f f e r e n c e s are serious problems for the theory d e f e n d e d h e r e . If the p o s s i b i l i t y of e x t r a c t i o n from S 1 forces us to assume that S' in sentences of the type in ( 8 6 d , f , g ) is a d i r e c t argument of the v e r b , we have to assume that it is generated as a n o n - t h e m a t i c s u b j e c t , i.e. a dummy pronoun f i l l i n g the s u b j e c t p o s i t i o n . One argument in favour of such an analysis could be the obligatory presence of it in these constructions. Let us start w i t h a discussion of w h - e x t r a c t i o n from S 1 . The relevant q u e s t i o n is w h e t h e r the e x t r a c t i o n in ( 8 ? ) is e x t r a c t i o n f r o m an argument S T or an a d j u n c t S ' .

-313-

Some R e l a t e d Topics (8?)

Who is it

essential [that John will visit t_] ?

In our analysis it is a r e f e r e n t i a l expression w i t h a t h e m a t i c role and Case. At the level of D - s t r u c t u r e it

is the s u b j e c t of a

clause. It is moved to s u b j e c t position in order to From this analysis it

follows that

the

embedded

a d j u n c t c l a u s e . In a dummy p r o n o u n a p p r o a c h , it

receive S'

might

must be

small Case. be

a

dummy

pronoun which has been inserted in s u b j e c t position and embedded S' can be an argument d i r e c t l y . In order to show correctness of our hypothesis concerning it,

I shall have to

an the the

prove

that the e x t r a c t i o n in ( 8 ? ) involves e x t r a c t i o n from an a d j u n c t S 1 . It is i n t e r e s t i n g to observe that e x t r a c t i o n from a d j u n c t clauses in English is possible to a certain e x t e n t . Chomsky argues that a d j u n c t s are not

islands

to

extraction.

(1985)

A relevant

example is given in ( 8 8 ) . (88)

?he is the person who they l e f t [ b e f o r e meeting _t]

I will not be concerned here w i t h the exact analysis of these e x t r a c t i o n s . What is important for our analysis is that e x t r a c t i o n from a d j u n c t s seems to be possible in E n g l i s h , but not in D u t c h , as is demonstrated by the corresponding D u t c h sentence in ( 8 9 ) . ( 8 9 ) * h i j is de man die zij

weggingen a l v o r e n s _t te ontmoeten

he is the man who they l e f t b e f o r e to meet As Chomsky o b s e r v e s , there is a contrast between extraction a r g u m e n t s and a d j u n c t s from a d j u n c t clauses. This contrast

of is

i l l u s t r a t e d in ( 9 0 ) . ( 9 0 ) a . ? W h o did you leave [ b e f o r e meeting _t] ? b . * H o w did you leave [ b e f o r e f i x i n g the car t_] ? O b v i o u s l y , e x t r a c t i o n of an

adjunct

i m p o s s i b l e . This asymmetry does not

from show

an up

argument clauses, as is i l l u s t r a t e d in ( 9 1 ) . ( 9 1 ) a . Who do you t h i n k [that John saw t_]?

-314-

adjunct in

clause

extraction

is from

Some R e l a t e d Topics b. How do you t h i n k [ t h a t John w i l l fix the car t_] ? If these observations are c o r r e c t , it p r o v i d e s us w i t h a

criterion

to d e t e r m i n e w h e t h e r an S' is an a d j u n c t or n o t . If the S' t h a t co-occurs w i t h it is an a r g u m e n t , we expect a d j u n c t e x t r a c t i o n f r o m S' to be possible, w h i l e we expect it

to be impossible if the S'

is

an a d j u n c t c l a u s e . It a p p e a r s to be the case that a contrast of the t y p e in ( 9 0 ) shows up in it ... S 1 discusses the following c o n t r a s t : ( 9 2 ) a . Who is it b . * H o w is it

constructions.

Chomsky

(1985)

time [ ( f o r J o h n ) to v i s i t _t]? t i m e [ ( f o r J o h n ) to fix the car _t ] ?

A similar contrast seems to a p p e a r in the

sentences

in

( 9 3 ) and

( 9 Ό . [301. ( 9 3 ) a . Who is it clear [that John w i l l visit t.]? b . ? ? H o w is it clear [that John will fix the car _t] (9^)a.

Who does it

b . ? ? H o w does it

please you [that John w i l l visit _t] please you [that John will fix the car t_]

If these o b s e r v a t i o n s are c o r r e c t , they d e m o n s t r a t e that e x t r a c t i o n from S' in ( 9 2 ) - ( 9 O involves e x t r a c t i o n f r o m an a d j u n c t clause. It then f o l l o w s that in these sentences it must be an a r g u m e n t receiving a t h e m a t i c role d i r e c t l y . Thus it t u r n s out that an a p p a r e n t p r o b l e m for our h y p o t h e s i s c o n c e r n i n g it can be t u r n e d into an argument in f a v o u r of t h i s h y p o t h e s i s . The second a p p a r e n t p r o b l e m for the r e f e r e n t i a l status is the obligatoriness of it was shown that in D u t c h

the

in s u b j e c t p o s i t i o n . In occurrence

of

het

of

it

2

it

optional

in

chapter is

s u b j e c t and o b j e c t p o s i t i o n . This is i l l u s t r a t e d once more in ( 9 5 ) . ( 9 5 ) a . dat ik ( h e t ) betreur dat

...

t h a t I it regret t h a t ... b . d a t ( h e t ) door m i j b e t r e u r d w o r d t d a t that it by me r e g r e t t e d is that . . .

...

In the c o r r e s p o n d i n g sentences in E n g l i s h in ( 9 6 ) it

-315-

is o p t i o n a l in

Some R e l a t e d Topics object position but o b l i g a t o r y in subject p o s i t i o n . ( 9 6 ) a . that I regret ( i t ) that

...

b. that * ( i t ) was regretted by me that

...

We have to explain why it must be present in ( 9 6 b ) . In view of what has been said in section 3 of this c h a p t e r , the obligatoriness of it follows d i r e c t l y from the assumption that nominative Case d e f i n e s a ( N P ) position to the l e f t of I N F L . This p o s i t i o n must f i l l e d at S - s t r u c t u r e . Such an analysis is c o n f i r m e d that s u b j e c t l e s s sentences of the

type

in

by

( 9 6 b ) are

the

p a r t i c u l a r verb while it has to appear in the

the

object

passive

fact

abundantly

present in Old English, as demonstrated in section 4 . One p o t e n t i a l argument against such an analysis is that there are cases in which it cannot appear as an

be

fact of

a

counterpart

( c f . W i l l i a m s 1 9 8 1 ) . This is shown in ( 9 7 ) . ( 9 7 ) a . I reasoned ( * i t ) that b.*I reasoned it

...

c. * ( I t ) was reasoned that

...

There is a d i f f e r e n c e between the u n g r a m m a t i c a l i t y of ( 9 7 a , b ) where it is present and the u n g r a m m a t i c a l i t y caused the absence of it

in

( 9 7 c ) . The u n g r a m m a t i c a l i t y of ( 9 7 a ) w i t h it is solely d e p e n d e n t on p r o p e r t i e s of the p a r t i c u l a r v e r b , while the u n g r a m m a t i c a l i t y of ( 9 7 c ) without it is a s t r u c t u r a l p r o p e r t y of English. To account for these f a c t s , I shall assume that the selection of it or S' as the i n t e r n a l argument of V is s t r u c t u r a l l y f r e e . The actual selection of it or S 1 is d e p e n d e n t on lexically d e t e r m i n e d factors such as f a c t i v i t y ( c f . K i p a r s k y & K i p a r s k y 1 9 7 0 ) . If the verb is p a s s i v i z e d , the choice between S' or it is d e t e r m i n e d by s y n t a c t i c principles.

If

S'

is

selected,

the

sentence

u n g r a m m a t i c a l because the NPnom position will remain has to be selected for s y n t a c t i c reasons.

-316-

will

become

unfilled.

It

Some R e l a t e d Topics 4.8 Concluding remarks In this chapter we have p r o v i d e d f a i r l y t e n t a t i v e analyses of a variety of issues. The common d e n o m i n a t o r in these s u p e r f i c i a l l y unrelated topics is t h e i r r e l a t i o n to c e n t r a l issues in the preceding chapters. In the f i r s t few sections of this chapter we were concerned with a f u r t h e r elaboration of two central theoretical topics: the concept of connectedness and the t h e o r y of endocentric s u b j e c t s . With respect to connectedness it has been argued that a f u r t h e r extension of this concept can be achieved if we assume that the upper level of a subtree is subject to conditions on connection as w e l l . It was d e m o n s t r a t e d that the c-command condition can be seen as an instance of the connection r e q u i r e m e n t s and that d i f f e r e n c e s between a d j u n c t i o n and s u b s t i t u t i o n can be made to follow from c o n f i g u r a t i o n a l p r o p e r t i e s with respect to the d i r e c t i o n of connection of the a n t e c e d e n t . In section 3 it was argued that the general concept of small clauses should be extended in such a way that the s u b j e c t of S is base-generated as an e x t e r n a l argument w i t h i n the V - p r o j e c t i o n . This generalization of the e n d o c e n t r i c subject theory has i m p o r t a n t consequences for the analysis of languages with an a p p a r e n t N P - I N F L - V P o r d e r . It was shown that the r e q u i r e m e n t that English sentences have a subject should not f o l l o w from a u n i v e r s a l p r i n c i p l e , as proposed in C h o m s k y C 1 9 8 2 ) , but rather from the fact that in English n o m i n a t i v e Case d e f i n e s a p o s i t i o n e x t e r n a l to the V - p r o j e c t i o n . This position must be f i l l e d d u r i n g the d e r i v a t i o n by moving a non-Case-marked NP. E m p i r i c a l evidence s u p p o r t i n g these t h e o r e t i c a l claims has been provided in the following sections. A discussion of Old English has made it clear that a v a r i e t y of d i f f e r e n c e s between Old English and Modern English can be e x p l a i n e d by the theory developed h e r e . The occurrence of various types of s u b j e c t l e s s sentences in Old English has been p a r t i c u l a r l y r e l e v a n t in this c o n n e c t i o n . It would be completely ad hoc to assume that there exists a v a r i e t y of empty dummy pronouns in Old English. This assumption would be necessary to m a i n t a i n the E x t e n d e d P r o j e c t i o n P r i n c i p l e as a p r i n c i p l e belonging to U G . In our t h e o r y the o c c u r r e n c e of

-317-

Some R e l a t e d Topics s u b j e c t l e s s sentences in OE is not s u r p r i s i n g at all. INFL

is

able

to

assign

Case

to

an

external

As in

argument

Dutch, or

non-Case-marked i n t e r n a l argument d i r e c t l y u n d e r g o v e r n m e n t . If

a no

such NP is p r e s e n t , no p r i n c i p l e is v i o l a t e d . The d i f f e r e n c e between Old English and Modern English lies m a i n l y in the position of INFL w i t h respect to V m a x . A discussion of several Romance languages has been i n c l u d e d to d e m o n s t r a t e that these languages can be argued to have u n d e r l y i n g p o s t v e r b a l s u b j e c t . The ' n o r m a l ' p r e v e r b a l position the s u b j e c t is either a position d e f i n e d by ( F r e n c h , I t a l i a n ) or a t o p i c position ( S p a n i s h ) . In the f i n a l two s e c t i o n s , it

has been shown

nominative that

German

an of Case es

and English it are not dummy p r o n o u n s but r a t h e r referential expressions which are in most respects similar to Dutch het.

-318-

Some R e l a t e d Topics

NOTES

1. Chomsky(198l) argues that subcategorized SCs are endocentric, whereas adjunct SCs are exocentric. One reason for this distinction is that adjunct SCs may have PRO subjects. Given the fact that Chomsky adopts the Aoun-Sportiche definition of government the subject of an endocentric SC would be governed by the head of the SC. This implies that PRO can only appear whenever the subject is not part of the maximal projection of the predicate, since PRO would be governed otherwise. 2. One argument in favour of a VP constituent in 0V languages is presented in Haegeman & Van Riemsdijk (1984). They argue that there is a VP-external NP position in West-Flemish and Zurituutsch in order to account for the fact that apparently the subject cannot undergo Verb Projection Raising (VPR). As far as I can see, however, their argument is not conclusive. The fact that a subject of a complement to a caussative or perception verb cannot be moved via VPR to a position to the right of the perception verb can be made to follow from the fact that the perception verb assigns Case to the left. Such an account would be similar to the account presented in HoekstraC1984) of the impossibility of postverbal perception-verb complements in Dutch. 3. In general, this approach severely delimits the possible underlying structures of languages. Although in the endocentric approach no language is supposed to have a VP as a separate constituent, the notions of argument projection and endocentric subject-predicate relation imply that there are eight possible configurations for finite clauses, abbreviated in ( i ) . (i) [INFL,[Subj, [V,0]]J A further investigation of the consequences of research.

(i)

4. A similar underlying structure is proposed in Rigter(1985).

will be left

for

future

Rigter & Beukema(1985) and

5. Another construction for which it might be claimed that the external argument receives Case in its D-structure position involves the subject-Aux-inversion construction (SAI), illustrated in ( i ) . (i) When did you see him Given the fact that we have claimed that in 'normal' cases subject-AUX-inversion takes place to the extent that the subject is moved to a position to the left of INFL, (i) might be the result of the absence of movement of the subject. If this is true, INFL must be able to assign Case to the right. It might be the case that INFL is able to assign Case to the right in main clause questions only. Such an account would be comparable to the account of Stylistic Inversion in French, discussed in section 5.3. This assumption, together with the assumptions that INFL/AUX may not remain empty and that movement of V to INFL is blocked by the presence of an intervening lexical subject, allows us to account for the SAI-construction and the phenomenon of do-support. 6. The assumption that te(to) in Dutch is also part of INFL is rather problematical. One major problem is that in raising complements te is

-319-

Some R e l a t e d Topics

obligatorily present. In the account given here this is unexpected, since raising should be allowed in Dutch only if the complement is Vmax. The problem regarding the presence or absence of te in Dutch is quite complicated. In postverbal position an infinitival complement always requires the presence of te. However, in preverbal position a variety of factors are relevant for the presence of te. I shall leave these problems out of consideration here. 7. There is a further difference between Dutch and English that is relevant here. In Dutch, COMP is a specifier of the INFL projection (cf. chapter 1). This implies that a lexical complementizer, such as the prepositional complementizer om, is not able to assign Case to an embedded subject. Thus, of the three possible ECM constructions in English only one is present in Dutch. 8. The results presented in section 4 on Old English and the development into Modern English are rather tentative and incomplete. More detailed research in this direction is required, in particular with respect to the diachronic development . The abbreviations under the OE examples refer to OE texts. A list of these texts is provided below. 9. This assumption is adopted in most recent analyses of OE, see e.g. Lightfoot (1979,1981), Fischer & Van der Leek (1983), Van Kemenade (1984a). 10. The fact that pronouns may appear in front of P is somewhat problematical within the framework defended here, since Case is assumed to be assigned directionally . Given the fact that full NPs must and personal pronouns may appear after the preposition, we have to assume that in OE prepositions assign Case to the right. If we assumed that personal pronouns may be moved to the left of P, we would have a violation of the Gap Condition, since the trace would not be canonically governed. Such a movement approach is assumed in most analyses, such as Allen(1980), Vat (1978) and Van Kemenade ( 1984b) . If personal pronouns were base-generated in front of P, no Case could be assigned to them directly. We must therefore assume that personal pronouns preceding P are clitic-like elements that are base-generated in front of P and receive Case inherently or by coindexation with the empty NP(pro) that follows P. A clitic analysis of these personal pronouns is proposed in Van Kemenade ( 1 1 . There are several arguments against an analysis in which an R-pronoun or a personal pronoun is moved from PP to COMP. Relative pronouns are similar to demonstrative pronouns in that they cannot occur in front of P. The assumption that in ae-relatives personal pronouns are moved into COMP and subsequently deleted would be completely ad hoc. A movement analysis would then have to assume that the relative R-pronoun 5ar is moved from PP to COMP and subsequently deleted. Such an analysis would be in line with the theory of extraction proposed in chapter 1 . One problem with this analysis is the fact that there are no agr-relatives with animate heads, while there are many instances of ^-relatives with animate heads. If this is a general phenomenon, it implies that deletion in COMP would be obligatory if the antecedent is animate, while it is optional with inanimate antecedents. A potential solution to this problem is to assume that there is a resumptive pronoun strategy for jte-relativization in OE, as proposed by Van Kemenade (1984b). In that case the PP contains a base-generated empty resumptive pronoun, locally A '-bound by an (empty) operator in COMP. It seems reasonable to assume that the Gap Condition applies to resumptive pronouns in the same way in which it applies to traces, since the two categories require a local antecedent in a non-thematic position. Thus, the adjacency between P and V in this

-320-

Some R e l a t e d Topics

construction can be made to follow from the Gap Condition. The Subjacency effect follows from the fact that the resumptive pronoun should be locally A'-bound. Movement of the operator proceeds in accordance with conditions on movement. If such an analysis of 3£-relativization is correct, the resumptive pronoun must be generated in front of P, in order to satisfy the Gap Condition. Given that in OE personal pronouns may optionally precede P and that resumptive pronouns generally belong to the class of personal pronouns, the empty resumptive pronoun can be generated in front of P. It is clear that this analysis may proceed within the general framework outlined in chapter 1. However, a major problem with this theory is that we have to adopt the existence of an empty resumptive pronoun and an empty operator in COMP. Both claims are not independently motivated. It seems better to stick to an R-movement analysis. The obligatory deletion of the R-pronoun in COMP if the antecedent is [+animate] is reminiscent of the normative pattern in Dutch. As observed in chapter 3, R-pronouns cannot be [+human] according to normative grammar. According to this view (ia) is ungrammatical, and should be replaced by (ib). (i)a. de jongen waar ik mee gewerkt heb the boy where I with worked have b. de jongen met wie ik gewerkt heb the boy with whom I worked have In infinitival relatives the relative pronoun is obligatorily deleted and pied piping is impossible. Neverthess, P-stranding is possible from a normative point of view, even if the antedent is [-t-human]. This is shown in (ii). (ii) een leuk meisje om mee te werken a nice girl for with to work There can be no doubt that (ii) is derived by means of wh-movement of an R-pronoun, followed by obligatory deletion. All the conditions and restrictions on R-movement can be found in these infinitival relatives. For instance, R-pronouns are not allowed with certain prepositions, such as zonder(without). If the preposition met in (ii) is replaced by zonder, the sentence will become ungrammatical. Apparently, the lexical presence of an R-pronoun blocks a [+human] interpretation in normative grammar, while this interpretation is available if the R-pronoun is deleted. It turns out then that the situation in 'normative Dutch' is exactly parallel to the situation found in Old English. Given the fact that in 'normative Dutch' there is no restriction on [+human] R-pronouns (cf.(ii)) but only on lexical [+human] R-pronouns, as in (ia), the non-occurrence of lexical R-relatives with [+human] antecedents in Old English might be an instance of the same phenomenon. 12. The order NPobj NPnom can be derived by other rules as well. Particularly in root clauses, Topicalization of the NPobj to the position in front of the finite verb may cause this effect. However, the sentences in (U8) do not show the verb-second effect of root clauses. The two NPs precede the verb. A sentence like (i) cannot be argued to be the result of Inversion, but seems derived by Topicalization. (i) Hem behofea Cristes helpe Them(OBJ) needs Christ's help(NOM) (Chr.,262.18/ Elmer 1981, II (35))

-321-

Some Related Topics

13. This step in particular has to be motivated more extensively. For instance, the locality requirement on the movement of V to INFL applies only if INFL and V are separated by a lexical NP, as in the case of Subject-AUX-Inversion constructions, or not. In the framework adopted here, we have to assume that the trace of a moved subject and adverbials are irrelevant. If this analysis turns out to be incorrect, we have to assume that the two important changes in the development of English, i.e. the position of INFL with respect to Vmax and the position of V with respect to its objects, are unrelated. 1*1. Note that it follows from the Burzio generalization that 'transitivization' and 'personalisation' are necessarily combined in these constructions. 15. Although the development from Middle Dutch into Modern Dutch does not involve such drastic changes as the development of English, one change deserves to be mentioned in this chapter. The Case system which was present in Middle Dutch is almost completely lost in Modern Dutch. One of the consequences is that Middle Dutch shows a wider range of nominative-less constructions than Modern Dutch. The range of nominative-less constructions in Middle Dutch is discussed in Van den Berg (1985). As in Old English, several verbs were able to assign a lexically determined genitive Case to an internal argument. If such a verb does not select an external argument, a nominative-less construction of the type (i) is possible. (i) mi wondert des me(OBJ) wonders that(GEN) The general pattern of these Middle Dutch constructions is exactly similar to the Old English pattern discussed in 4.1.3. Given this similarity I shall not discuss this construction and its development here. An interesting and problematic construction discussed by Van den Berg involves nominative-less sentences with an accusative argument. Examples are provided in (ii). (ii)a. daer is den brief there is the letter(OBJ) b. te Wye stont bi die brugge enen torre in Wye was-standing at the bridge a tower(OBJ) c. groten strijt was daer gevochten big battle(OBJ) was there fought This construction seems to be rather common in Middle Dutch with verbs like zijn (to be) and staen (to stand). Van den Berg argues that this construction shows up only with non-agentive 'subjects', which are base-generated in object position. He claims that these verbs optionally assign structural objective Case. The problem with this analysis is the fact that these verbs do not assign an external argument. Given the Burzio generalization, we do not expect these verbs to assign structural Case. In order to maintain the generalization we might assume that verbs like zijn and staen optionally assign lexical accusative Case. In particular, the passive sentence in (iic) seems to point in this direction. The appearance of an accusative NP, thematically the direct object of the passivized verb, is in direct violation of the theory of passivization. In order to maintain this theory we have to appeal to an ad-hoc mechanism. The fact that the verb zijn is able to assign lexical accusative Case in (iia) as well as in (iic) is to some extent an argument in favour of such an ad-hoc mechanism of Case assignment.

-322-

Some R e l a t e d Topics

16. An argument against the postverbal base position of the external argument is given in Jaeggli (1982, ch.1!). He argues that the intonation pattern of (i) indicates that (ia) is the base-generated structure. (i)a. Juan compro una casa ayer Juan bought a house yesterday b. Compro una casa ayer Juan He claims that (ia) "has the typical intonation pattern of a sentence in what might be called canonical word order"(Jaeggli ( 1 9 8 2 ) , p . l H l ) . However, it is not clear to me in what way the intonation pattern of a sentence can provide information concerning the underlying structure of a sentence. As long as there are no theories on the relation between intonation patterns and basic sentence structure that settle the issue, I consider this argument to be non-valid. 17- I shall not discuss the phenomenon of subject clitics. These may also appear after a finite verb. The phenomenon of inversion of verb and clitic (subject-clitic inversion (SCI)) has properties that are quite different from the 'stylistic inversion' construction (SI) under discussion, as has been demonstrated in Kayne (1972). Some of these differences are: SCI, but not SI, is applicable in root clauses only; SCI, but not SI, appears in yes-no questions; SCI, but not SI, moves the subject immediately to the right of the leftmost verb. Further discussion of SCI can be found in Emonds (1976), Den Besten (1983) and Jaeggli (1982), etc. 18. I shall assume that se absorbs objective Case, with the external θ-role as a consequence , cf.note 19.

absorption of

the

19- At first sight, the ungrammaticality of (i) seems to be problematical. (i) *I1 se mange bien dans ce restaurant It "se" eats well in this restaurant The derivation of (i) should proceed as follows: se absorbs the external thematic role and il is inserted as the non-thematic external argument in order to fill up the nominative NP-position. However, as indicated in note 18, I shall assume that se absorbs objective Case, in a way similar to zieh in Dutch. The absorption of the external argument is merely a natural consequence, in accordance with the Burzio generalization. Consequently, se cannot be inserted with verbs that do not take an internal NP-argument. Similarly, se cannot be inserted with verbs that do not assign objective Case to their D-structure internal argument as a consequence of lexical properties (ergative verbs), as shown in (ii), or passivization. as shown in (iii). In these three cases no Case can be absorbed. (ii)

*I1 s'a ete invite It "se" has been invited (iii) *I1 s'est arrive It "se" is arrived Burzio (1981) claims that the non-occurrence of (i)-(iii) derives from the fact that il must bind a nominative NP. Although such an analysis is possible for the construction under discussion, it does not extend to other constructions in which il is used as a non-thematic argument, as, for instance in passives of intransitive verbs.

-323-

Some R e l a t e d Topics

20. In this respect French differs from Dutch. As pointed out in ch.3, impersonal passives in Dutch are allowed without the insertion of a dummy pronoun. This possibility is not available for French, since French requires the preverbal NP position to be filled. However, as will be demonstrated below, there are constructions in which there is no such preverbal NP-position. In those cases the introduction of the dummy pronoun il is indeed optional. 21. A similar analysis may be available for the "il y a-construction". The postverbal NP does not agree with the verb and there is a definiteness effect, as is shown in (i). (i) II y a deux/*ces livres sur la table

It there has two/these books on the table If it is assumed that the phrase [deux livres sur la table] is a Snail Clause with an NP subject, deux livres should receive Case from the matrix verb. If it receives Case from avoir, the verb must be a normal transitive verb. Again, il serves as an external argument to satisfy this requirement syntactically. It is interesting to observe that different verbs are chosen in (i) and the corresponding (ii). (ii) Ces livres sont/*ont sur la table These books are on the table 22. This option is in fact comparable to French, with the extension that assignment to the right is not restricted to specific environments.

Case

23. It follows from this analysis that we have to consider Inversion and null-subjects to be separate properties. There seems to be no way to predict the non-existence of a) languages with obligatory lexical subjects that have a lexical dummy pronoun with the same function as the empty dummy in Italian, b) languages that allow null-subjects but do not have such a dummy pronoun. English might be an instance of a language of type a), although the phenomenon of postverbal subjects is greatly restricted by lexical factors. With respect to a language of type b), Rizzi(1982, ch.U, fnt.20) notes that Brazilian Portuguese might be an instance of a language allowing null-subjects but no free application of Inversion. He refers to an unpublished paper by Chao(1980). 24. (77g) is not ungrammatical without er, but in that interpretation hoeveel is an NP rather than a QP followed by an empty noun. A similar contrast appears in (i). (i)a. Ik heb veel gekocht I have much bought Ί have bought much' b. Ik heb er veel gekocht I have "er" many bought Ί have bought many' 25. The sentences without er seem to improve somewhat if the topicalized, as in ( i ) . (i) (Wat betreft deze boeken) Twee heb ik ??(er) gekocht (Concerning these books) Two have I "er" bought

-324-

quantified

NP is

Some R e l a t e d Topics

Particularly in listing the construction without er appears to (ii).

improve,

as

in

(ii) (wat betreft deze jongens) Twee lopen ?(er) in de tuin, (concerning these boys) Two walk "er" in the garden drie zijn ?(er) weggegaan en vier zitten ?(er) binnen three are "er" away gone and four sit "er" inside I have no genuine explanation for this decrease in ungrammaticality. There are two factors which might be involved. First, contrary to all S-internal positions, the first position in main clauses is not canonically governed. It might be the case that somehow the derived position is taken to be the position at which binding relations are established. Secondly, it is a well-known phenomenon that empty categories are possible in first position in main clauses that are not possible within S. Substandard sentences typical of spoken language show this quite clearly. Examples are provided in (iii). (iii) a.

?(Dat) weet ik That know I Ί know (it)' a 1 . Ik weet *(dat) b. ?(Ik) ben in Amsterdam geweest I am in Amsterdam been '(I) have been in Amsterdam' b ' . In Amsterdam ben *(ik) geweest

It might be the case that by way of exception empty categories in first position can be contextually bound. For (i) and (ii) this would mean that er is not necessarily present to bind PRO, since PRO in that position might be bound otherwise. 26. The distribution of es is extensively described in Putz(1975). Most of the examples given in this section are taken from Putz. A discussion of the similarities and differences between Dutch er and het and German es, within a framework similar to the one proposed here, is presented in Van Leeuwen(1985). Other articles relevant to this issue are Den Besten(1983) and Safir(198M). 27. Note that we cannot relate the ungrammaticality of (8lc,d) to the requirement that NPs .including es, should receive Case. Such an explanation would correctly exclude es in ( 8 l c ) , but not in ( 8 l d ) , since nominative Case is available in ( 8 l d ) . 28. This analysis predicts that non-referential es may always appear as the first constituent in main clauses. Obviously, this is incorrect, since this is generally not allowed in case of definite external arguments. I shall assume that the actual appearance of es is determined by pragmatic/semantic principles, in a way similar to the appearance of dummy er in Dutch. Relevant factors appear to be the thematic function of es, which might be described as presentative, and the semantic/pragmatic properties of the constituents within S. 29. There is one further relevant difference between Dutch and German. With the exception of some northern dialects (cf.Van Riemsdijk 1978a) it is impossible to strand prepositions. The Gap Condition predicts stranding of prepositions to be possible in German if the argument of P precedes P and if PP is (canonically) governed by P. It turns out, however, that in those circumstances stranding is impossible, as is illustrated in ( i ) .

-325-

Some R e l a t e d Topics

(i) *Da habe ich nicht [_t mit ] gerechnet There have I not on counted Logically, there are two conceivable structural reasons for the ungrammaticality of (i). Either P is not able to govern the trace correctly or V does not govern PP correctly. Given that, even PP-internally, adjunction turns out to be impossible, as demonstrated in (ii), it follows that the first option has to be selected. (ii) a. Ich habe [genau davor] gestanden I have exactly there for stood b.*Ich habe [da genau t_ vor] gestanden Obviously, German differs from Dutch to the extent that P is a structural governor in Dutch but not in German. This difference is similar to the difference between English and French concerning P-stranding (cf.Kayne 1984). It thus follows that P-stranding is possible only if a) P is a structural governor (this holds for English and Dutch, but not for French and German) and b) P governs its argument canonically (to the left in Dutch and to the right in English). These conditions on P-stranding are part of the Gap Condition, as formulated in ch.1. 30. The possibility of extraction from adjunct clauses is clearly related to extraction from adjunct PPs. As discussed in Van Riemsdijk (1978a) and Hornstein & Weinberg (1981) extraction from adjunct PPs is possible, as in (i). (i) Who did John's mother travel with From the acceptability of (i) it follows that we have to assume that the adjunct PP is governed by V. If not, (i) would constitute a violation of the Gap Condition. If the government domain of V in English can be extended upwards to include adjunct PPs, it is expected that the same analysis applies to adjunct clauses. It has been observed by Hornstein & Weinberg that not all adjunct PPs can be stranded. In particular temporal adjunct PPs appear to be islands, as is illustrated in (ii). (ii)*Which meal did John speak to Bill after? It is assumed that the difference between (i) and (ii) is caused by a difference in the level of adjunction. It is likely that the difference in acceptability between (90a) and (93a) is caused by the same factor.

List of OE Texts: Aelfric's Lives of Saints (Lives), ed. W.Skeat, EETS 76,82,94 & University Press, Oxford, 1881-1900 The Homilies of the Anglo-Saxon Church (AKT) ed. B.Thorpe, 2 vols., Aelfric Society, 1844, rpt. Johnson Reprint Corp., New York 1971 Homilies of Aelfric: a Supplementary Collection (ΑΗΡ), ed. J.Pope, EETS 259 & 260, University Press, Oxford, 1967-1968

-326-

Some Related Topics

King Alfred's Old English Version of Boethius (Β), ed. W.Sedgefield, 1899 rpt. Wissenschaftliche Buchgesellschaft, Darmstadt, 1968. King Alfred's Orosius (0), ed.H.Sweet, EETS 79, N.Trubner, London, 1883. King Alfred's West-Saxon Version of Gregory's Pastoral Care (CP), ed.H.Sweet, EETS 45 & 50, N.Trubner, London, 1871-1ΒΫ27 Old English Version of Bede's Ecclesiastical History £f the English People (Bede), ed. T.Miller, EETS 95 & 96, N.Trubner, London 1890-1891. The B ckling Homilies of the Tenth Century (BH), ed. R.Morris EETS 58,63 & 73, N.Trubner, Oxford 1880. An Old English Martyrology (Mart.), ed. J.Herzfield, EETS 116 Kegan Paul, Trench, Trubner, and Co., London 1900. Two of the Saxon Chronicles (Chr.), ed. C.Plummer, Clarendon Press, Oxford 19214. The Homilies of Wulfstan(W), ed. D.Bethurum, Clarendon Press, Oxford 1957. 'Gospel of St.Matthew', in: The West-Saxon Gospels (St.Mat.) ed. M.Grunberg, Schelterna & Holkema, Amsterdam 1967. Bosworth-Toller, An Anglo-Saxon Dictionary (BT), Oxford.

-327-

References Allen, C.L. 1977 Topics in diachronic English syntax, diss. U.Mass, Amherst. 1980 "Movement and deletion in Old English", Linguistic Inquiry 11, 261-323. Aoun, J. 1981 The formal nature of anaphoric relations, diss.MIT. 1985 "A typology of wh-interrogatives", paper read at the GLOW conference, Brussels, April 1985. Aoun, J. and D.Sportiche 1983 "On the formal theory of government", The Linguistic Review 2, 211-236. Bach, E. and G.Horn 1976 "Remarks on 'Conditions on transformations'", Linguistic Inquiry 7, 265-300. Bart, P.van & R.Kager 1984 "Er is hier - een alternatief voor een diagnose", in G.J.de Haan, M.Trommelen & W.Zonneveld (eds) Van periferie naar kern, Foris Publications, Dordrecht, 1-11. Barwise, J. & R.Cooper 1980 "Generalized quantifiers and natural language", Linguistics and Philosophy 4, 159-219. Bayer, J. 1984 "COMP in Bavarian syntax", The Linguistic Review 3, 209-274. Been, G. 1952 Ueber das niederländische Adverbialpronomen er, Travaux du cercle linguistique de Copenhague 8. Belletti, A. 1980 "'Morphological' passive and Pro-drop. A note on the impersonal construction in Italian", unpubl.ms. MIT. Belletti, A. and L.Rizzi 1981 "The syntax of ne: some theoretical implications", The Linguistic Review 1, 117-154. Bennis, H. 1977 "Het kwantitatieve e_r_ in komparatief konstrukties", Spektator 6, 384-387. 1979 "Appositie en de interne struktuur van NP", Spektator 8, 209-228. 1980a "Er_-deletion in a modular grammar", in S.Daalder & M.Gerritsen (eds) Linguistics in the Netherlands 1980, North-Holland, Amsterdam, 58-68. 1980b "Coindexing and complementizer-trace phenomena", paper read at the GLOW conference, Nijmegen, April 1980. 1982 "Boekbespreking van: R.S.Kirsner 'The problem of presentative sentences in modern Dutch'", GLOT 5, 1-13-

References

1983 1985

"De PRO-drop parameter en subjectloze zinnen in het Nederlands", Spektator 12, 409-427. "Het-Raising", in H.Bennis & F.Beukema (eds) Linguistics in the Netherlands 1985, Foris Publications, Dordrecht, 11-20.

Bennis. H. & L.Haegeman 1984 "On the status of agreement and relative clauses in West-Flemish", in W.de Geest & Y.Putseys (eds) Sentential complementation, Foris Publications, Dordrecht, 33-55. Bennis, H. & T.Hoekstra 1984a "The distribution and interpretation of gaps", ms. Univ. of Leiden. 1984b "Gaps and parasitic gaps", The Linguistic Review 4, 29-87. 1985 "A parametrized Gap Condition", paper read at the GLOW conference, Brussels, April 1985. Bennis, H. & A.Hulk 1979 "A comparative study of the quantitative construction", paper read at the conference Levels of Syntactic Representation, Paris, Dec. 1979. Berg, E.van den 1985 "Onpersoonlijke konstrukties, Subjekten en nominatieve NP's in het Middelnederlands", De nieuwe taalgids 78, 201-213Besten, H.den 1981 "Government, syntaktische Struktur und Kasus", in M.Kohrt & J.Lenerz (eds) Sprache: Formen und Strukturen, Niemeyer Verlag, Tübingen, 97-107. 1982 "Some remarks on the ergative hypothesis", Groninger Arbeiten zur Germanistischen Linguistik 21, 61-82. 1983 "On the interaction of root transformations and lexical deletive rules" in W.Abraham (ed.), On the formal syntax of the Westgermania, John Benjamins, Amsterdam, 47-131. Beukema, F. & T.Hoekstra 1983 "Met met PRO of met zonder PRO", De nieuwe taalgids 76, 532-548. Blom, A.

1977

"Het kwantitative er", Spektator 6, 387-395.

Blom, A. & S.Daalder 1977 Syntaktische theorie en taalbeschrijving, Muiderberg. Booij, G.E. 1977 Dutch morphology: a study of word formation in generative grammar, Foris Publications, Dordrecht. Brachin, P. 1973/74 "Een in dubbel opzicht 'vreemde' constructie", De nieuwe taalgids 66, 59-60; 67, 177-202. Breckenridge, J. 1975 "The post-cyclicity of es-insertion in German", in R.Grossman, J.San & T.Vance (eds) Papers from the eleventh regional meeting of the Chicago Linguistic Society, Chicago, 81-91.

-330-

References

Bresnan, J. & J.Grimshaw 1978 "The syntax of free relatives in English", Linguistic Inquiry 9 331-391. Burzio, L. 1981 Intransitive verbs and Italian auxiliaries, diss.MIT. Canale, M. 1978 Word order change in Old English: base reanalysis in generative granniar, diss. McGill University. Chomsky, N. 1965 Aspects of the theory of syntax, MIT Press, Cambridge, Mass. 1970 "Remarks on nominalization", in R.A.Jacobs & P.Rosenbaum (eds) Readings in English transformational grammar, Ginn & Comp., New York. 1977 "On WH-movement", in P.Culicover, T.Wasow and A.Akmajian (eds) Formal syntax, Academic Press, New York, 71-132. 1979 "Markedness and core grammar", in A.Belletti et al. (eds) Theory of markedness in generative grammar, Scuola Normale Superiore, Pisa. 1980 "On binding", Linguistic Inquiry 1 1 , 1-46. 1981 Lectures on government and binding, Foris Publications, Dordrecht. 1982 Some concepts and consequences of the theory of government and binding, MIT Press, Cambridge, Mass. 1985 "Barriers", ms. MIT. Chomsky, N, & H.Lasnik 1977 "Filters and control", Linguistic Inquiry 8, 425-504. Contreras, . 1983 "Small Clauses in Spanish", to appear in A.Hurtado (ed.) Spanish syntax and linguistic theory, Foris Publications, Dordrecht. Coppen, P.A. 1985 "De aard van het quantitatieve er", De nieuwe taalgids 78, 149-163. Coppen, P . A . , L.Huiskens, M.Klein and P.Wever 1983 "Over de complementstructuur van perceptiewerkwoorden", De nieuwe taalgids 76, 1-10. Drewes, H . J . , E.Jacobs, F.v.d.Maagdenburg, J.Veld & M.de Wolff 1984 "Crossing R-graphs", in G.J.de Haan, M.Trommelen & W.Zonneveld (eds) Van periferie naar kern, Foris Publications, Dordrecht, 29-39. Elffers, E. 1977 "Er-verkenningen", Spektator 6, 417-422. Elmer, W. 1981 Diachronie grammar: the history of Old and Middle English subjectless constructions, Max Niemeyer Verlag, Tübingen. Emonds, J. 1970 Root and structure-preserving transformations, diss. MIT. 1976 A transformational approach to English syntax, Academic Press. New York.

-331-

References

Engdahl, E. 1983 "Parasitic gaps", Linguistics and Philosophy 6, 5-34. Everaert, M. 1982 "A syntactic passive in Dutch", Utrecht Working Papers in Linguistics 11, 37-73. Evers, A. 1975 The transformational cycle in Dutch and German, distributed by Indiana University Linguistics Club. Felix, S. 1983 "Parasitic gaps in German", Groningen Arbeiten zur Germanistischen Linguistik 22, 1-46. Fischer, 0. & F.van der Leek 1983 "The demise of the Old English impersonal construction", Journal of Linguistics 19, 337-368. Gaaf, W.van der 1904 The transition from the impersonal to the personal construction in Middle English, Carl Winter, Heidelberg. 1930 "The passive of a verb accompanied by a preposition", English Studies 12, 1-24. Geijn-Brouwers, M.van 1982 Parasitic gaps een Nederlands fenomeen?, MA thesis, University of Amsterdam. Groos, A. 1983 "Sentential complements and theta-marking", paper read at CILS, Barcelona. Groos, A. & R.Bok-Bennema 1985 "The structure of the sentence in Spanish", ms. Gueron, J. 1976 "Interpretation of PP complements: a semantic filter for PP extraposition", in H.C.van Riemsdijk (ed.) Green ideas blown up Publikaties van net instituut voor Algemene Taalwetenschap UvA, 131980 "On the syntax and semantics of PP extraposition", Linguistic Inquiry 11, 637-678. Haaften, T.van 1983 "Over gaten in zinnen", Spektator 12, 428-451. Haan, G.de 1979 Conditions on rules, Foris Publications, Dordrecht. Haan, G.de, G.A.T.Koefoed & A.L.Tombe 1974 Basiskursus algemene taalwetenschap. Van Gorcum, Assen. Haegeman, L. and H.van Riemsdijk 1984 Verb projection raising, scope and the typology of verb movement rules, Tilburg Papers in Language and Literature 64.

-332-

References

Haider, H. 1984 "Topic, focus and verb-second in German", Groninger Arbeiten zur Germanistischen Linguistik 25, 72-120. Haiman, J. 1974 Targets and syntactic change, Mouton, The Hague. Hoek, Th.van den 1970 "Opmerkingen over zinscomplementatie", Studia Neerlandica 7, 189-215. Hoekstra, T. 1983 "The distribution of sentential complements", in H.Bennis & W.U.S.van Lessen Kloeke(eds) Linguistics in the Netherlands 1983, Foris Publications, Dordrecht, p.93-1031984 Transitivity: grammatical relations in government and binding theory, Foris Publications, Dordrecht. Hornstein, N. and A.Weinberg 1981 "Case theory and preposition stranding", Linguistic Inquiry 12, 55-94. Huang, J. 1982 Logical relations in Chinese and the theory of grammar, diss. MIT. Huybregts, M.A.C. 1983 "Cross-serial dependencies and the form of syntactic structure", paper read at the Brussels Conference on Sentential Complementation, June 1983. Huybregts, M.A.C. & H.van Riemsdijk 1985 Parasitic gaps and ATB, Tilburg Papers in Language and Literature 76. Hulk, A. 1982 Het clitisch pronomen en, diss. RUU. Jackendoff, R. 1977 X'-syntax, MIT Press, Cambridge, Mass. Jaeggli, 0. 1982 Topics in Romance syntax, Foris Publications, Dordrecht. Jespersen, 0. 1927 A modern English grammar on historical principles. Allen & Unwin, London. Kayne, R. 1972 "Subject inversion in French interrogatives", in J.Casagrande & B.Saciuk (eds) Generative studies in Romance Languages, Newbury House, Rowley, Mass., 70-126. 1975 French syntax: the transformational cycle, MIT-Press, Cambridge, Mass. 1979 "Rightward NP movement in French and English", Linguistic Inquiry 10 710-719. 1981 "Two notes on the NIC", in A.Belletti, L.Brandi & L.Rizzi (eds) A theory of markedness in generative grammar Proceedings of the 1979 GLOW conference, Scuola Normale Superiore, Pisa. 1984 Connectedness and binary branching, Foris Publications, Dordrecht.

-333-

References

Kayne, R. & J.Υ.Pollock 1978 "Stylistic inversion, successive cyclicity and move NP in French", Linguistic Inquiry 9, 595-621. Kemenade, A.van 1984a "Verb second and clitics in Old English", in H.Bennis & W.U.S.van Lessen Kloeke (eds) Linguistics in the Netherlands 1984, Foris Publications, Dordrecht, 101-109. 1984b "Preposition stranding in Old English", in G.de Haan, M.Trommelen & W.Zonneveld (eds) Van periferie naar kern, Foris Publications, Dordrecht, 129-139. Kerstens, J. 1978 "Over WH-verplaatsing en Cl-verplaatsing in het Nederlands", Spektator 8, 340-359. Kiparsky, P.& C.Kiparsky 1970 "Fact" , in Steinberg & Jacobovits (eds) Semantics, Cambridge University Press, Cambridge. Kirsner, R.S. 1979 The problem of presentative sentences in modern Dutch, North-Holland. Klein. M. 1983 "Over de zgn. absolute met-constructie", De nieuwe taalgids 76, 151-164. Koopman, H. 1984 The syntax of verbs: from verb movement rules in Kru languages to Universal Grammar, Foris Publications, Dordrecht. Koopman, H. and D.Sportiche 1981 "Variables and the bisection principle", The Linguistic Review 2, 139-160. 1985 "θ-theory and extraction", paper read at the GLOW conference, Brussels, April 1985. Koster, J. 1975 "Dutch as an SOV language", Linguistic Analysis 1, 111-136. 1978a Locality principles in syntax, Foris Publications, Dordrecht. 1978b "Why subject sentences don't exist", in S.J.Keyser (ed.) Recent transformational studies in European languages, MIT-Press, 53-65. 1983 "Subjectsporen in het Nederlands", Spektator 13, 156-166. 1984 Global harmony, Tilburg Papers in Language and Literature 61. Lasnik, H. & M.Saito 1984 "On the nature of proper government". Linguistic Inquiry 15, 235-289. Leeuwen, A.van 1985 "Es, er_ en het; een taalvergelijkende Studie naar de status en distributie van zogenaamde dummy-pronomina", MA thesis, Leiden. Lenerz, J. 1977

Zur abfolge nominaler Satzglieder im Deutschen,

TEL Verlag Gunter Narr, T bingen. -334-

References

Lightfoot, D.W. 1979 Principles of diachronic syntax, Cambridge University Press, Cambridge. 1981 "The history of Noun Phrase movement", in C.L.Baker & J.McCarthy (eds) The logical problem of language acquisition, MIT-Press, 86-119Lubbe, H.F.A.van der 1985 "De structuur van de zgn. absolute met-constructie", De nieuwe taalgids 78, 6-17. Maling, J. 1978 "An asymmetry with respect to WH-islands", Linguistic Inquiry 9, 75-89. Maling, J. & A.Zaenen 1978 "The non-universality of a surface filter", Linguistic Inquiry 9, 475-497. Manzini, M.-R. 1983 "On control and control theory", Linguistic Inquiry 14, 421-446. May, R. 1977

The granmar of quantification, diss. MIT.

McCray, A. 1981 "Clause initial elements in German", in R.Hendrick, C.Masek & M.Miller (eds) Papers from the seventeenth regional meeting of the Chicago Linguistic Society, Chicago, 205-220. Paardekooper, P.C. 1971 Beknopte ABN-syntaxis, Malmberg, Den Bosch. Perlmutter, D. 1971 Deep and surface structure constraints in syntax, Holt, Rinehart and Winston, New York. 1978 "Impersonal passives and the unaccusative hypothesis", Papers from the annual meeting of the Berkeley Linguistic Society 4, 157-189. Pesetsky, D. 1982a "Complementizer-trace phenomena and the Nominative Island Condition", in The Linguistic Review 1, 297-344. 1982b Paths and categories, diss. MIT. Platzack, C. 1984 "The position of the finite verb in Icelandic", in W.de Geest & Y.Putseys (eds) Sentential complementation, Foris Publications, Dordrecht, 195-205. Pollock, J.Y. 1984 "Sur la syntaxe de en et le parametre du sujet nul", ms. Paris. Putz, H. 1975 Ueber die Syntax der Pronominalform es^ im modernen Deutsch, Verlag Gunter Narr, Tübingen.

-335-

References

Reinhart, T. 1983 "Coreference and bound anaphora", Linguistics and Philosophy 6, 47-88, Reuland, . 1983 "Movement versus merger: relations between inflection and verb", NELS XIII.

Riemsdijk, H.van 1978a A case study in syntactic markedness, Foris Publications, Dordrecht. 19?8b "On the diagnosis of WH-movement", in S.J.Keyser (ed) Recent transformational studies in European Languages, MIT-Press, 189-206. Riemsdijk, H.van and E.Williams 1981 "NP-structure", The Linguistic Review 1, 171-219. Rigter, B. 1985 "TO and PRO", in H.Bennis & F.Beukema (eds) Linguistics in the Netherlands 1985, Foris Publications, Dordrecht, 157-166. Rigter, B. & F.Beukema 1985 A government and binding approach to English sentence structure, Van Walraven, Apeldoorn. Rizzi, L. 1982 Issues in Italian syntax, Foris Publications, Dordrecht. 1983 "On chain formation", paper read at the GLOW conference, York, April 1983. Roelfszema, J. 1983 "Over that-trace-verschijnselen", TABU 13, 37-47Ross, J.R. 1967 Constraints on variables in syntax, diss. MIT. 1973 "The same side filter", in C.Corum, T.Smith-Stark and A.Weiser (eds) Papers from the ninth regional meeting of the Chicago Linguistic Society, Univ.of Chicago. Safir, K. 1984 "Missing subjects in German", to appear in J.Toman (ed.) Linguistic theory and the grammar of German, Foris Publications, Dordrecht. 1985 "Binding in relatives and LF", paper read at the GLOW conference, Brussels, April 1985.

Sag, I. 1982 "Coordination, extraction, and Generalized Phrase Structure Grammar", Linguistic Inquiry 13, 329-336. Schermer-Vermeer, E.G. 1985 "De enthüllende status van er_ in de generatieve grammatica", Spektator 15, 65-84. Scherpenisse, W. 1985 "The final field in German: extraposition and frozen positions", Groninger Arbeiten zur Germanistischen Linguistik 26, 80-99-

-336-

References

Schutter, G.de 1974 De Nederlandse zin. Poging tot beschrijving van zijn struktuur, Brügge. Smith, N.S.H. 1976 "-AAR", Leuvense bijdragen 65, 485-496. Smits, R. & J.Vat 1985 "Met jouw tanden in mijn bek, een onderzoek naar met-constructies", Spektator 14, 445-470. Stoweil, T. 1981 Origins of phrase structure, diss. MIT. 1983 "Subjects across categories", in The Linguistic Review 2, 285-312. Taraldsen, T. 1980 On the Nominative Island Condition, vacuous application and the that-t filter, Indiana University Linguistics Club. Thrainsson, . 1984 "Different types of infinitival complements in Icelandic", in W.de Geest & Y.Putseys (eds) Sentential complementation, Foris Publications, Dordrecht, 247-257. Torrego, E. 1984 "On inversion in Spanish and some of its Linguistic Inquiry 15, 1.

effects",

Tripp, R.P. 1978 "The psychology of impersonal constructions", Glossa 12, 177-189. Vat, J. 1978

"On footnote 2: evidence for the pronominal status of 5aer in Old English relatives", Linguistic Inquiry 9, 695-716.

Verhagen, A. 1979 "Fokusbepalingen en graramatikale theorie", Spektator 8, 372-402. Visser, F.Th. 1963 A historical syntax of the English language, Brill, Leiden. Wahlen, N. 1925 The Old English Impersonalia, part I, Göteborg. Wende, F.

1915

Ueber die nachgestellten Praepositionen im Angelsachsischen, Mayer & Muller, Berlin.

Williams, E. 1978 "Predication", Linguistic Inquiry 11, 203-237. 1981 "Argument structure and morphology", The Linguistic Review 1, 81-114. 1984 "Grammatical relations", Linguistic Inquiry 15, 639-673.

-337-

References

Zribi-Hertz, A. 1982 "La morphologie verbale passive en Francais: essai d'explication", in J.Gueron & T.Sowley (eds) Grammaire transfortnationnelle; theorie et methodologies, Paris. Zubizarreta, M.L. 1982 On the relationship of the lexicon to syntax, diss. MIT. Zwarts, F. 1981 "Negatief polaire uitdrukkingen I", GLOT 4, 35-132.

-338-