3110146851_avax.pdf

The Tense System in English Relative Clauses

W DE

G

Topics in English Linguistics 16

Editor

Herman Wekker

Mouton de Gruyter Berlin · New York

The Tense System in English Relative Clauses A Corpus-Based Analysis

Ilse Depraetere

Mouton de Gruyter Berlin · New York 1996

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin.

@ Printed on acid-free paper which falls within the guidelines of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data

Depraetere, Ilse, 1964— The tense system in English relative clauses ; a corpus-

based analysis / Ilse Depraetere. p. cm. - (Topics in English linguistics ; 16)

Includes bibliographical references and index. ISBN 3-11-014685-1 (cloth ; acid-free paper) 1. English language - Relative clauses. 2. English

language - Tense. I. Title. II. Series. PE1385.D46 1995 425-dc20 95-42930

CIP

Die Deutsche Bibliothek — Cataloging-in-Publication Data

Depraetere, Ilse. The tense system in English relative clauses ; a corpus-based analysis / Ilse Depraetere - Berlin; New York : Mouton de Gruyter, 1996

(Topics in English linguistics ; 16) ISBN 3-11-014685-1

NE: GT

© Copyright 1995 by Walter de Gruyter & Co., D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mecha-nical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printing: Gerike GmbH, Berlin. Binding: Lüderitz & Bauer, Berlin. Printed in Germany.

To Luk

Perhaps if the future existed, concretely and in-dividually, as something that could be discerned by a better brain, the past would not be so seductive: its demands would be balanced by those of the fu-ture. Persons might then straddle the middle stretch of the seesaw when considering this or that object. It might be fun. But the future has no such reality (as the pictured past and the perceived present pos-sess); the future is but a figure of speech, a specter of thought. (Vladimir Nabokov, Transparent things)

Acknowledgements

First and foremost, I would like to thank Prof. R. Declerck: not only did he introduce me to the intriguing field of tense and aspect in En-glish, but his careful and rigorous comments on my text have con-tributed considerably to shaping it into its present form. He has set standards of professional excellence that will provide a continuous challenge to my work.

I am grateful to Prof. J. Buysschaert, Prof. S. Greenbaum and Prof. O. Leys and Prof E. Vorlat for their comments on various chapters of this book.

I am indebted to a large number of native speakers who were will-ing to offer their views on the acceptability of the verb forms in the many examples of the book. As a complete list of the often acciden-tal informants would be too long, I have to restrict myself to naming the most dedicated among them: T. Caldicott and E. Lee.

I would like to thank P. Bonte, for the exceptional patience he dis-played in drawing and redrawing the diagrams in this book.

The completion of this project would have been twice as hard without my colleagues at the university, whose good humour helped keep me sane.

It is my parents, my sister and my husband, though, who remain my most faithful supporters: their love is reflected in all that I do.

Contents

List of abbreviations xv

Introduction 1

Chapter 1 Definition of the concepts bounded-unbounded and telic-atelic

1.1. Definition 6 1.2. Tests to distinguish unboundedness from boundedness and

atelicity from telicity 17 1.2.1. Test 1 17 1.2.2. Test 2 20 1.2.3. Test 3 22 1.2.4. Test 4 23

1.3. Factors influencing (un)boundedness and (a)telicity 25 1.3.1.NPs 25 1.3.2. Directional PPs 27 1.3.3. Tense 28

1.3.3.1. Past tense 28 1.3.3.2. Present perfect 30 1.3.3.3. Present tense 31 1.3.3.4. Conclusion 32

1.3.4. Adverbials 33 1.3.4.1. Adverbials affect (a)telicity 34 1.3.4.2. Adverbials affect (un)boundedness 38 1.3.4.3. Effect of adverbials on implicature past tense 42

1.3.5. Progressive 47 1.3.6. Non-indicative mood and negation 49 1.3.7. Interaction of the different factors 50

1.4. Zero-boundedness and zero-telicity 50 1.5. Left boundedness and right boundedness 52 1.6. General conclusion 53

χ Contents

Chapter 2 Declerck (1991a): tense in discourse (part 1)

2.1. Declerck (1991a): basic concepts 5 5 2.1.1. Time-spheres 5 5 2.1.2. Absolute sectors, absolute tenses, temporal

subordination, relative tenses 55 2.1.3. Time of orientation (TO), situation-time of orien-

tation (sit-TO, TOsit), T02 57 2.1.4. Shift of domain 58 2.1.5. System of domain-internal tenses 60

2.1.5.1. Past sector 60 2.1.5.2. Pre-present sector 61 2.1.5.3. Present sector 64 2.1.5.4. Post-present sector 64

2.1.6. Syntactic subordination * temporal subordination 66 2.1.7. The principle of unmarked temporal interpretation 66 2.1.8. Direct and indirect binding 6 8

2.2. Methodology and aim 72

Chapter 3 The expression of temporal relations in past sector RCs

3.1. Introduction 77 3.2. (W-)anteriority in past sector (N)RRCs 84

3.2.1. Constraints on the Replacement of the Past Perfect (CRPP) 84

3.2.2. RRCs and NRRCs: different tense system? 106 3.3. Simultaneity in past sector (N)RRCs: relative and absolute

past tenses? 112 3.4. W-posteriority in past sector (N)RRCs 120

3.4.1. Introductory remarks 120 3.4.2. Shift of domain form not acceptable 123

3.4.2.1. Constraints on the Replacement of the Conditional Tense (CRCT) 123

3.4.2.2. Intensional domains 124 3.4.3. Temporal subordination unacceptable/relatively

unacceptable 126 3.4.4. RRCs vs. NRRCs: different tense system? 131

3.5. General conclusion

Contents xi

135

Chapter 4

The expression of (W-)anteriority in pre-present sector RCs

4.1. The use of tense in the pre-present sector RCs: Declerck (1991a) 137 4.2. Constraints on the expression of (W-)anteriority in

pre-present sector RCs 141 4.2.1 Temporal subordination not possible 141

4.2.1.1. Links with t0 141 4.2.1.2. Adverbials 145

4.2.2. Shift to past sector not possible 146 4.2.2.1. Constraints on the Replacement of the

Present Perfect (CRPrP) 146 4.2.2.2. Anchoring time missing 153 4.2.2.3. Links with t0 154 4.2.2.4. Adverbials 155

4.2.3. Shift within pre-present sector not possible 156 4.2.3.1. Links with past TO 156 4.2.3.2. Adverbials 157

4.3. The tense system in the pre-present sector: some modifications 159

4.4. RRCs and NRRCs: different tense system? 162 4.5. General conclusion 167

Chapter 5 The expression of (W-)anteriority and (W-)simultaneity in post-present sector RCs

5.1. The use of tense in the post-present sector: Declerck (1991a) 169

5.2. RRCs vs. NRRCs: different tense system? 171 5.3. Factors determining interpretation of verb forms in RRCs 176

5.3.1. General principle exemplified 176 5.3.1.1. Present tense in the RRC 178 5.3.1.2. Present perfect of the indefinite type

and past tense in the RRC 179

xii Contents

5.3.1.3. Present perfect of the continuative type in the RRC 184

5.3.2. Influencing factors 186 5.3.2.1. Aspect and (a)telicity 186 5.3.2.2. Definite vs. indefinite antecedent 190 5.3.2.3. Influence of lexical material 192 5.3.2.4. Influence of adverbials 194 5.3.2.5. Linear order of the SUPC and RC 205 5.3.2.6. Gnomic quality of the RC situation 218 5.3.2.7. FPS forms misinterpreted as PPS forms 220

5.4. General conclusion 221

Chapter 6 Difference in contextual effects between mutually substitutable verb forms 6.1. General procedure vs. implementation of general

procedure 226 6.1.1. Past sector 226 6.1.2. Pre-present sector 227

6.2. Prediction vs. fact 228 6.3. Explicit location in time - explicit indication of temporal

relation 231 6.3.1. Past sector 231 6.3.2. Pre-present sector 234 6.3.3. Post-present sector 236

6.4. Resultativeness 239 6.5. Gradual development vs. quick succession 242 6.6. Recent vs. distant situation 244

Chapter 7 The principle of unmarked temporal interpretation

7.1. Introduction 249 7.2. The PUTI (Declerck 1991a): theory 250 7.3. The PUTI in other works 252

Contents xiii

7.4. The PUTI for embedded clauses (Declerck 199la): examples 260 7.4.1. Bounded + bounded 260 7.4.2. Unbounded + unbounded 261 7.4.3. Unbounded + bounded or bounded + unbounded 264

7.5. The PUTI ((a)-part) and RCs 265 7.5.1. Question 1 and question 2 266 7.5.2. Question 3 273

7.5.2.1. Other works on foregrounding in RCs 273 7.5.2.2. Corpus evidence 275 7.5.2.3. Restriction on foregrounding in RRCs 276 7.5.2.4. Comparison RRC-NRRC 279 7.5.2.5. Restriction on foregrounding in RRCs

explained 281 7.5.3. Question 4 282 7.5.4. Question 5 285

7.5.4.1. Future tense 285 7.5.4.2. Past perfect 286 7.5.4.3. Present perfect 288

7.6. General conclusion 291

Chapter 8 Choice of binding TO in RCs

8.1. TO for (N)RRC situations 293 8.1.1. Possible candidates for T02 function 293

8.1.1.1. Finite SUPC binds SC situation 293 8.1.1.2. Nonfinite SUPC binds SC situation 294 8.1.1.3. T02 = implicit 294 8.1.1.4. T02 = noun 295 8.1.1.5. T02 = time indicated by adverbial 296 8.1.1.6. T02 to be found in extra-linguistic context 297 8.1.1.7. Indirect binding 299

8.1.2. Choice of binding TO in RCs: Declerck (1991a) 308 8.1.3. General conclusion 310

8.2. (In)direct binding in (N)RRCs 310 8.2.1.(In)direct binding in (N)RRCs: Declerck (1991a) 310

xiv Contents

8.2.2. RRCs vs NRRCs: different systems? 318 8.2.2.1. (In)direct binding in RRCs 318 8.2.2.2. Factors determining the choice between

direct and indirect binding 321

8.2.3. General conclusion 338

General conclusion 342

Appendix 349

Notes 365

References 404

Index 429

List of abbreviations

RC RRC NRRC SC SUPC to TO TO, T02

TO sit DB IB FPS

PPS

PUTI

CRPP CRCT

CRPrP W-anterior

W-posterior

W-simultaneous

relative clause restrictive relative clause non-restrictive relative clause subclause superordinate clause temporal zero-point time of orientation central TO binding TO in a complex tense situation-time of orientation direct binding indirect binding future perspective system (t0 functions as binding TO for the post-present sector situation) present perspective system (the post-present TO functions as pseudo-t0, to which the post-present sector situation is temporally subordinated) principle of unmarked temporal interpretation (principle determining the temporal interpretation of the situations when tenses, adverbials or prag-matic knowledge do not impose a certain order) constraint on the replacement of the past perfect constraint on the replacement of the conditional tense constraint on the replacement of the present perfect world-anterior (the anteriority relation between two situations is not linguistically realized by means of a particular verb form) world-posterior (the posteriority relation between two situations is not linguistically realized by means of a particular verb form) world-simultaneous (the simultaneity relation be-tween two situations is not linguistically realized by means of a particular verb form)

xv i List of abbreviations

? acceptability questionable ?? acceptability highly questionable * ungrammatical form 'verb' verb form changes chronological order between

two situations or changes the temporal location of the situation

"verb" verb form is mutually substitutable with other verb form

Introduction

When confronted with the vast amount of literature on tense and aspect, one may wonder whether there remains anything substantial to be said about this topic. The first book on the issue which I read was a rough copy of the text which was later to be published as Tense in English: Its structure and use in discourseAs I browsed through other works on tense and aspect, it soon became apparent that Declerck's approach opened up new perspectives, i.e. a possi-bility of an in-depth analysis of a question which seems self-evident and yet has received only scant attention: how are temporal relations2

expressed in different types of sentences? Three issues are raised by this self-evident question:

1. How does one express that a situation comes after, or before or is simultaneous with another one?

2. Is there any difference between independent and dependent clauses regarding the way temporal relations are expressed? Do different types of dependent clauses behave differently in this respect?

3. Does the tense system differ in past tense, future tense, present perfect tense and present tense sentences?

I would not be doing justice to the linguists who have written about tense if I claimed that these questions have not been dealt with at all. They have indeed received attention, though no systematic in-vestigation has been carried out so far. The following are only a few examples which illustrate the lack of comprehensiveness in preced-ing studies:

1. The observation that the past perfect is used to refer to a situation that comes before another past situation is a common one. However, hardly any linguist has considered the question how anteriority is expressed in a present perfect or a future tense sentence.

2. The difference between the present perfect and the past tense has caused much ink to flow. Yet no one has compared the tenses with a view to possible differences in the way temporal relations are in-dicated in present perfect and past tense sentences.

3. It is common knowledge that it is not always necessary to use the past perfect to refer to an anterior past situation:

2 Introduction

(1) He explained that he had forgotten (forgot) to switch off the oven when he left that morning.

However, the constraints on the interchangeability of the past and the past perfect have not been investigated in detail.

4. Linguists have been eager to point out that the future tense is sel-dom used to establish future time reference in some types of sub-clause:

(2) a. He will do it when he has (* will have) time. b. Susan will forget to send a card unless you tell (*

will tell) her once more this afternoon.

The question whether the future tense is ruled out altogether in any kind of subclause has received relatively little attention. So has the question whether there is any difference between subclauses which refer to anterior future time situations and those which refer to simultaneous future time situations.

Declerck's book provides a general framework within which the above issues can be dealt with satisfactorily. In the first part of his book, he makes a survey of how temporal relations can be expressed in present tense, past tense, future tense and present perfect tense sentences. In doing so, he does not aim at maximal generalization at the cost of reducing linguistic reality to a limited set of rules that are subject to a number of constraints but provides an apparatus in which the wealth of phenomena relating to tense and aspect find their place. Although it requires some effort to familiarize oneself with Declerck's terminology, having done so, one gets a comprehensive view of this complex aspect of the English language.

The aim of this book is to investigate in detail how temporal rela-tions are expressed in relative clauses. The distinction between restrictive and non-restrictive relative clauses, indeed, offers an additional challenge. It has been observed by (amongst others) Haegeman - Wekker (1984: 54), Quirk et al. (1985: 1070), Ruther-ford (1970), Sampson (1971) and Tregidgo (1979: 194-195) that many clauses may be given an "adjunct" and a "disjunct" reading: "The primary difference is that they [adjuncts and disjuncts] differ syntactically in that disjuncts are peripheral to the clause to which they are attached" (Quirk et al. 1985: 1070). Allen (1966: 177-179), Declerck (1991a: 53, 154), Gelhaus (1974: 121-122) and Ota (1963: 113) have pointed out that subclauses which are relatively dependent on the main clause should be set apart from those which are rela-

Introduction 3

tively independent of the main clause. It has also been suggested that the difference in syntactic status is reflected in the tense system that is used (especially when future time is indicated):

In other words, the difference caused by introducers can be explained by the fact that the choice of tense forms in D-clauses [adverb clauses (1963: 110)] is semantic and not bound by the grammatical rule of "sequence of tenses." Clauses indicating reason, manner, comparison, and result have more unmatched cases, because what these clauses indi-cate can be semantically independent of the time indicated by the verb of the main clause; (...) On the other hand, what is indicated by clauses of time or condition is semantically more bound to the time indicated by the predicate of the verb of the main clause. (Ota 1963: 112-113)

The most typical example of a "free" verb-cluster [e.g. future perfect] is a verb-cluster which functions as the main verb in a sentence. (...) A free cluster may also occur, of course, in a clause introduced by one of the co-ordinators: (...) But a free verb-cluster may also occur in an included clause if the clause is "free": that is, if it is not closely bound to the rest of the sentence. A clause introduced by one of the following includers is usually a free clause: although, for, such (that) as (= because), inasmuch as, though because, since (= because), while (= although) except (that), so (that). (Allen 1966: 177)

An adjectival (i.e., "relative") clause is usually bound, although it may be free. (Allen 1966: 178)

The includers listed on page 177 occasionally introduce bound clauses, (ibid.)

This reaffirmation is necessary because the kind of clause in which the future perfect can be used is the kind in which the domain can easily be shifted, viz. unembedded clauses and subclauses whose situation-TO need not be temporally dependent on the situation-TO of the matrix (e.g. relative clauses, subclauses of reason, concession, comparison, place, etc.). (Declerck 1991a: 53)

The FPS [future perspective system] will be used whenever the speaker deems it necessary to explicitly represent the domain he is referring to as a post-present domain. This is the case when there is no close temporal relation between the

4 Introduction

clause containing the relevant verb form and the clause(s) in the context referring to a post-present domain. Thus we will use the FPS rather than the PPS [present perspective system] in independent clauses (e.g. I will remember your name) and in subclauses that establish their own time reference (e.g. non-restrictive relative clauses: I will ask John, who will know the answer tonight). Subclauses whose situation is represented as temporally connected to that of the matrix (e.g. adverbial time clauses) use the PPS. (Declerck 1991a: 154)

I will put the claims made in the above quotations to the test by relating them to an analysis of relative clauses. Throughout the dis-cussion I will try to find out whether the syntactic difference between (the relatively independent) non-restrictive relative clauses and (the relatively dependent) restrictive relative clauses is indeed reflected in a different tense system, not only in future time reference relative clauses but also in present perfect and past tense relative clauses. In order to find an answer to this question, I will examine five issues:

1.How is anteriority/posteriority/simultaneity expressed in what Declerck calls "past sector"3 (non-)restrictive relative clauses?

2. How is anteriority expressed in "pre-present sector" (non-) restrictive relative clauses?

3. How is anteriority/simultaneity expressed in "post-present sec-tor" (non-)restrictive relative clauses?

4. Is there a difference in the temporal relations established by non-restrictive and restrictive relative clauses in which the tenses do not give information about the chronological order between the main clause situation and the subclause situation?

5. Which elements can function as "binding time of orientation" (Declerck 1991a: 20-21) for (non-)restrictive relative clauses? Do non-restrictive and restrictive relative clauses differ as far as "(in)direct binding" (Declerck 1991a: 62-66) is concerned?

The central hypothesis of this book is that the differences to be observed between future time restrictive and future time non-restric-tive relative clauses cannot be generalized to include just any issue relating to tense in relative clauses. I will illustrate the similarities between restrictive and non-restrictive relative clauses and show that the differences between them occur primarily in (a) future time con-texts and (b) "world-posterior indirectly bound" contexts. The distri-bution of the differences and similarities will raise the issue of the

Introduction 5

relative importance of the various findings, and it is on this hierarchy of relatively less/more important factors that the conclusion whether or not restrictive and non-restrictive relative clauses differ in the use of tense, will have to be based. My detailed investigation of relative clauses will allow me to reveal some of the underlying principles useful for explaining the use of tense in English in general and to comment on the differences in effect produced by "mutually substi-tutable" forms. My analysis also provides a means of putting Declerck's model to the test: if it contains any inaccuracies or short-comings, they are likely to emerge in the course of the discussion. It will be shown that (apart from a few minor inaccuracies) the predic-tions the model makes about the way in which temporal relations are expressed in English are borne out by the facts and that therefore, it constitutes a theory that is descriptively and explanatorily adequate.

I will proceed in the following way: in chapter 1,1 give a definition of the concepts (un)boundedness and (a)telicity, as these will prove extremely important for the discussion which follows. Chapter 2 gives an outline of the theoretical framework of this book, i.e. the descriptive theory of tense as developed in Declerck (1991a) and shows how it can be applied to relative clauses; a survey is given of the different options which the system allows. The first two chapters, then, provide the necessary tools to embark upon the analysis of the use of tense in relative clauses. In the next three chapters, some of the possibilities illustrated in chapter 2 are considered in detail. In chapter 3,1 examine how anteriority and posteriority are expressed in past sector relative clauses, some observations relating to the expres-sion of simultaneity in the past sector also being included. Chapter 4 deals with anteriority in pre-present sector relative clauses. Chapter 5 offers an analysis of anteriority and simultaneity in post-present sector relative clauses. This approach will result in a comparison between the three sectors and the two types of relative clauses. Hav-ing explained the constraints on, and possibilities of, expressing tem-poral relations by means of tense, I discuss the differences in effect between mutually substitutable forms in more detail (chapter 6). Chapter 7 is focussed on how relative clauses are temporally inter-preted when neither tense, adverbs nor pragmatic or contextual knowledge provide information about the temporal location of the situations. In the last chapter, I examine whether non-restrictive rela-tive clauses differ from restrictive relative clauses as far as the choice of binding time of orientation and (in)direct binding are concerned.

Chapter 1 Definition of the concepts bounded-unbounded and telic-atelic

The concepts (un)boundedness and (a)telicity will prove to be im-portant for explaining the use of tense. This chapter is therefore aimed at describing accurately which properties are referred to when the concepts are applied to sentences.

1.1. Definition

I agree with Declerck (1991a: 121) that (un)boundedness and (a)telicity are not synonymous. It is the latter concept which has re-ceived most attention so far, be it under different headings (Aktionsart, aspect). There is also a lot of variety in the labels used to indicate the classes in taxonomies of the type under discussion.4 One should also bear in mind that "the use of the name of a Vendler-class does not warrant the conclusion that the concept related to this name is a Vendler-concept" (Verkuyl 1989: 42). Although a comparison between the different labels and their respective definitions would be illuminating, a survey of this kind would not contribute directly to clarifying the concepts as they are used in this book. The discussion will therefore be focussed on a clear delineation of the concepts (a)telicity and (un)boundedness without aiming at a comprehensive comparison with other works about this topic.5

The definitions of (a)telicity and unboundedness hinge on the no-tions of endpoint or terminal point and temporal boundary. A situ-ation may be limited in time: for instance, a situation of sunbathing may last for half an hour; it reaches a temporal boundary once the person in question leaves the beach. If someone runs a marathon, the endpoint to this particular situation is when the runner reaches the finish. A deliberate attempt to stay five minutes under a cold shower reaches its endpoint once the five minutes are over. Although end-points of some kind are involved in each of the situations described, they are not completely similar. In the second case, unlike in the first, there is an endpoint inherent in the situation: no matter whether the runner manages to run 40 kilometres or not, the situation as such potentially implies a terminal point. The kind of endpoint which

Definition 7

characterizes the third example is similar: after five minutes, the sit-uation of staying under the water for five minutes is over. However, this does not necessarily mean that a terminal point is actually reached. If someone says He is running a marathon or John is al-ways boasting he can remain in ice-cold water. At the moment, he is staying under the cold water for five minutes the situations are not described as reaching their inherent/intended endpoint and hence they are not represented as having reached temporal boundaries. That is, even though the actual world situations must have a beginning and an end, there is no linguistic reference to these temporal boundaries. In the case of sunbathing, there is no inherent or intended endpoint to the situation as there is in the other two cases. Although it is over when the person in question leaves, the endpoint is not part of the semantics of sunbathing. The examples just given already show that a twofold distinction can be made:

(I) + inherent/intended endpoint (a) + endpoint reached: + temporal boundary (b) - endpoint reached: - temporal boundary

(II) - inherent/intended endpoint (a) + temporal boundary (b) - temporal boundary

(A)telicity has to do with whether or not a situation is described as having an inherent or intended endpoint; (un)boundedness relates to whether or not a situation is described as having reached a temporal boundary (cf. Declerck 1989: 277, 1991a: 121). So far, only right-hand boundaries have been mentioned. It will be clear though that a particular sentence may represent a situation as (not) having a left-hand boundary: in I waited for her from five to eleven, unlike in At six o'clock, I was still waiting for her, the adverbial indicates the starting point of the situation. In the former case, the situation is rep-resented as having a left-hand boundary; in the latter, it is not. In the case of (a)telicity, the focus is on endpoints; as far as (un)-boundedness is concerned, however, a distinction can be made between situations that are represented as bounded to the left and un-bounded to the right (Suddenly, he was fast asleep), situations rep-resented as bounded to the right and unbounded to the left (I appre-ciated his presence until he told me he hated Jews), and those repre-sented as bounded on both sides (/ worked in the garden from 2 till 5 o'clock) (cf. section 1.4.).

8 Definition of the concepts bounded-unbounded and telic-atelic

A clause is telic if the situation is described as having a natural endpoint6 (cf. (la), (lb)) or an intended endpoint (cf. (lc)) which has to be reached for the situation (as it is described in the clause) to be complete and beyond which it cannot continue. Otherwise it is atelic. (A)telicity is a matter of Aktionsart; it is a characteristic of a de-scription of a situation. Examples (la), (lb) and (lc) are telic, (Id) is atelic:

(1) a. The bullet hit the target. b. Sheila fainted. c. Sheila deliberately swam for two hours. d. Sheila is working in the garden.

There is no one-to-one correspondence between (a)telicity and the (absence) presence of particular syntactic constituents (e.g. telicity does not presuppose the presence of a direct object - cf. (lb)).

A sentence is bounded if it represents a situation as having reached a temporal boundary.7 Otherwise it is unbounded. The sentences in (2a) to (2d) are bounded, those in (2e) and (2f) are unbounded:

(2) a. Jeremy left at 5 o'clock. b. Judith played in the garden for an hour. c. Julian lived in Paris from February 1979 until May

1980. d. I have lived in Paris. e. She lives on the corner of Russell Square. f. She is writing a nursery rhyme.

In (2a), the nature of the situation is such that we know it takes up a limited amount of time. This example shows that there need not be an explicit indication that the situation is ending in order for a sen-tence to be bounded: the punctual character of the clause8 together with the use of a non-progressive tense determine the boundedness of the sentence. In (2b) and (2c), the adverbials impose temporal boundaries. In (2d), the tense used (the indefinite use of the present perfect)9 indicates that there are temporal boundaries to the situation; the bounded character of (2d) is the result of the indefinite reading of the perfect tense. Again, there is no one-to-one correspondence be-tween (un)boundedness and the (absence) presence of particular syntactic constituents (e.g. the absence of a direct object does not automatically imply that the sentence is unbounded. Cf. bounded sentences in (2a) to (2d)). In section 1.3., a survey will be given of

Definition 9

the factors which determine how a particular sentence is interpreted in terms of (un)boundedness and (a)telicity.

The phrase having reached is somewhat misleading in that it sug-gests that it is only situations referred to in the past tense that can be classified as either bounded or unbounded. Moreover, it also suggests that any past tense sentence is bounded as the situation it refers to lies in the past. It will become clear in the course of the discussion that these are observations are not correct. The reason why I use having reached rather than reaching is that the latter term wrongly gives the impression that the situation only needs to be tending to-wards a temporal boundary, which is not what boundedness refers to.

(Un)boundedness and (a)telicity differ in the following respects: a. A telic situation may be represented as bounded (as in (3)) or un-

bounded (as in (4)):

(3) a. John has already left, (telic bounded) b. He has pushed the cart into the barn, (telic bounded) c. He will eat an apple, (telic bounded)

(4) a. John is leaving, (telic unbounded) b. He was pushing the cart into the barn, (telic un-

bounded) c. He will be eating an apple, (telic unbounded)

b. (Un)boundedness is not to be equated with the aspectual opposi-tion progressive vs. non-progressive. The progressive form indeed establishes an unbounded reading in most cases (cf. (5b), (6b)), but this is not the only way in which an unbounded reading can be ar-rived at (cf. (5c), (6c)):

(5) a. I ate an apple, (bounded telic) b. I was eating an apple, (unbounded telic) c. John eats an apple every day. (unbounded atelic)

(6) a. John lived in London for a year, (bounded atelic) b. John is living in London, (unbounded atelic) c. John lives in London, (unbounded atelic)

Moreover, not all progressive sentences are unbounded:

(7) a. A: Why are your hands so dirty? B: I've been playing in the mud. (bounded atelic)

b. A: Her eyes are red. B: She's been crying, (bounded atelic)


In sentences like these, the effect of the progressive is overruled by the bounded reading established by the present perfect when it gets an indefinite interpretation. This means that clauses can be classified according to three parameters:

1. bounded or unbounded 2. telic or atelic (Aktionsart) 3. progressive or non-progressive (aspect)

The first of these three oppositions has not received sufficient atten-tion in discussions on tense and aspect.

c. The (a)telic character of a sentence, unlike (un)boundedness, is not affected by the progressive. The following examples are telic, irrespective of whether or not a progressive verb form is used:

(8) a. John opened the parcel, (telic bounded) b. John was opening the parcel, (telic unbounded)

(9) a. Ten firecrackers exploded, (telic bounded) b. Ten firecrackers were exploding, (telic unbounded)

d. Strictly speaking, (a)telicity is a characteristic of a description of a situation in the sense that a speaker is free to decide how he will refer to a particular situation he observes (cf. Dahl 1981: 83, 1985: 28, Declerck 1979: 764, 1991a: 120, De Vuyst 1983: 162, Hatav 1989: 498, Moens 1987: 59, Smith 1982: 169, 1983: 480). The situations in (10) might also have been referred to by means of the sentences in (11):

(10) a. Sheila fainted, (telic) (non-repetitive interpretation) (=( lb))

b. Sheila deliberately swam for 2 hours, (telic) (= (lc)) (11) a. Sheila fainted whenever she smelt incense, (atelic)

b. Sheila is swimming, (atelic)

e. The same comment applies to (un)boundedness; it is a matter of choice on the part of the speaker how he will represent the situation:

(12) a. Judith played in the garden for an hour, (bounded) (= (2b))

b. Julian lived in Paris from February 1979 until May 1980. (bounded) (= (2c))

Definition 11

(13) a. Judith was playing in the garden in the course of the afternoon, (unbounded)

b. Julian lived in Paris at that time, (unbounded)

The sentences in (12) may refer to the same situations as the corre-sponding sentences in (13): in the latter case, they are not represented as having ended, whereas in the former case they are. Similarly, if a child is painting, this situation may be referred to by means of the sentence Susan is painting a picture (telic unbounded) as well as Su-san is painting (atelic unbounded) or Susan has painted that picture all by herself (telic bounded). From that point of view, (un)boundedness as well as (a)telicity characterize descriptions of situations rather than offer a classification of situations. However, (a)telicity differs from (un)boundedness in the sense that having cho-sen a particular set of words to refer to the situation, the sentence's (a)telic character is not affected by the use of the progressive (cf. Declerck 1991a: 121). Moreover, as Smith (1991: 126-127) correctly points out, contextual information and mutual knowledge may im-pose restrictions on how a particular situation is referred to. In the discussion to follow, the phrases "the situation/sentence is (un)bounded/(a)telic" will sometimes be used instead of the more ac-curate phrase "the situation as it is represented in the sentence is (un)bounded/(a)telic".

f. An additional terminological issue which confuses matters is that some linguists (e.g. Nakhimovsky 1988: 31, Quirk et al. 1985: 189-197) use the term perfective to refer to the perfect tenses while (im)perfective is often used as a synonym of (non-)progressive. Still another approach is taken in Gabbay - Moravcsik (1980): "We pos-tulate two aspects; the continuous or progressive and the complete, or perfect" (Gabbay - Moravcsik 1980: 73). Eisterhold (1986) cor-rectly points out that "aspect is situation-internal and can be either perfective or imperfective. Perfective aspect is a way of viewing a situation as complete, a single unanalyzable whole. Imperfective as-pect views the situation internally with no explicit reference to either the beginning or end" (Eisterhold 1986: 93). She argues that for the past tenses the distinction perfective/imperfective coincides with the distinction non-progressive/progressive past tense. As she sees it, things differ for the present tense. Imperfective aspect is not neces-sarily realized by means of the progressive present tense. Sentences such as the following are also imperfective:

(14) a. I walk to school every day. (Eisterhold 1986: 96)


b. I know it. (ibid.)

In other words, the semantic distinction perfective/imperfective does not coincide with the morphological distinction non-progres-sive/progressive present tense. Unlike Eisterhold, I would argue that a habit may be represented perfectively (cf. (14a)) or imperfectively (cf. (14c)):

(14) c. I am walking to school these days.

I would claim that (14a) and (14b) are perfective and unbounded, whereas (14c) is imperfective and unbounded. In Eisterhold's ap-proach, the aspectual distinction perfective/imperfective and the dis-tinction bounded/unbounded are intertwined. The same observation applies to Comrie's (1976: 25) taxonomy:

Perfective Imperfective

Habitual Continuous

Nonprogressive Progressive

As I have just pointed out, I do not support the idea that a habit is in-herently imperfective. It is usually unbounded, but it may be rep-resented as "a single whole" (Comrie 1976: 16) (perfective) or as viewed "from within" (Comrie 1976: 24) (imperfective). Hirtle (1987) argues that "whenever an achievement, an accomplishment or an activity is thought in -is represented by means of- the simple form, we are obliged to see it from the beginning to the end of its du-ration" (Hirtle 1987: 92), i.e. perfectively. This does not apply to states, as these "differ from the other types of event in excluding any development, change or successive realization" (Hirtle 1987: 93). The latter observation (mistakenly, as will be clear from the remarks made above,) suggests that states cannot be represented perfectively. In the discussion to follow, (im)perfective will be used as a synonym of (non-)progressive.

Definition 13

Most discussions about taxonomies of the type under discussion centre around the question of (a)telicity (Aktionsart). (Un)boundedness as we have defined it has not received much atten-tion in English linguistics. The following factors often make it diffi-cult to determine which concept is being discussed:

a. The examples that are given to illustrate the different classes are usually bounded telic or unbounded atelic;

b. The elements that are said to determine the classification (NPs, PPs,...) very often affect (un)boundedness as well as (a)telicity;

c. There is a divergence in the nature of the tests used to distinguish between, for instance, states, activities, accomplishments and achievements: some tests reveal something about the (a)telicity of the statement, others reveal something about the (un)boundedness of the statement, while still others apply to (a)telicity as well as (un)boundedness. As is clear from the examples given so far, (un)boundedness and (a)telicity indeed coincide very often. This probably explains why so many people have failed to notice the dis-tinction between these two concepts. Declerck (1989: 277-278, 1991a: 121, 1991b: 57-60) first drew attention to the difference be-tween (un)boundedness and (a)telicity.10 The common ground be-tween (a)telicity and (un)boundedness will also be become clearer in the following sections, which present a survey of the factors deter-mining (un)boundedness/(a)telicity (section 1.3.) and the tests used to find out whether a sentence is (un)bounded or (a)telic (section 1.2.). Before giving a survey, it is useful to consider in more detail some of the ideas put forward by Dahl (1981) and Moens (1987). It is particularly interesting to see to what extent the double distinction they make corresponds to what we call (un)boundedness and (a)telicity.

a. Dahl (1981)

At first sight, Dahl's (1981) distinction between sentences having the Ρ property and those having the Τ property appears ίο be related to the distinction between (un)boundedness and (a)telicity presented here: "A situation, process, action, etc. or the verb, verb phrase, sen-tence, etc. expressing this situation, etc. has the Τ property iff (...) it is directed toward attaining a goal or limit at which the action ex-hausts itself and passes into something else" (Dahl 1981: 81). "A sit-uation, process, action etc. has the Ρ property iff it has the Τ property and the goal, limit, or terminal point in question is or is claimed to be


actually reached" (Dahl 1981: 82). Dahl summarizes the possible combinations of the Ρ property and the Τ property as follows (Dahl 1981: 82):

not-T Τ

not-P I was writing I was writing a letter

Ρ (does not occur) I wrote a letter

However, some of Dahl's claims are not compatible with our ap-proach:

a. "The Ρ property entails the Τ property" (Dahl 1981: 82). If the Ρ property corresponded to (un)boundedness and the Τ property to (a)telicity, this would imply that all bounded sentences are telic, which is definitely not the case (cf. (6a), (7a) and (7b)). The line of reasoning behind this claim seems to be that if a situation does not have an inherent endpoint, no endpoint can be reached. We have ar-gued, however, that a particular situation may have actual temporal boundaries even if there is no inherent or intended endpoint to the situation.

b. "Verb phrases that do not have the Τ property are always imper-fective, whereas verb phrases that have the Τ property are perfective or imperfective according to whether they have the Ρ property or not" (Dahl 1981: 82). This observation suggests that the Ρ property stands for (im)perfectiveness. As perfectiveness-/imperfectiveness applied to a language such as English corresponds to the non-pro-gressive/progressive distinction (cf. supra), it follows that the Ρ property is not synonymous with (un)boundedness: as pointed out before, (un)boundedness should not be equated with the progres-sive/non-progressive opposition. These two observations make it sufficiently clear that our (un)boundedness is not synonymous with Dahl's Ρ property. Dahl presents some "problematic cases" for his approach and ends the ar-ticle with the pessimistic idea that "my personal opinion is that it may not be worthwhile to go on and try to refine the definitions of the properties we have been talking about in this chapter. Rather, a

Definition 15

better understanding of the problems will be obtained by developing an adequate semantics which takes account of aspectual properties" (Dahl 1981: 89).

b. Moens (1987)

Moens (1987: 57) proposes the following taxonomy:

EVENTS STATES

atomic extended

understand, love, know,

resemble

+conseq CULMINATION

recognize, spot win the race

CULMINATED PROCES

build a house, eat a sandwich understand,

love, know, resemble - conseq POINT

hiccough, tap, wink

PROCESS

run, swim, walk, play the piano

understand, love, know,

resemble

He writes: "Events will be referred to as "bounded" in that they are supposed to start and end at relatively precise points in time" (Moens 1987: 57-58); if they have consequences they are telic (Moens 1987: 58). "States are "unbounded" since (...) no reference is made to their start and end points" (Moens 1987: 58).

The following are certain respects in which our approaches differ: a. As bounded situations are "supposed to start and end at relatively

precise points in time" (i.e. there is no reference to the actual reach-ing of a temporal boundary), Moens' definition of boundedness seems to correspond with what we call telicity. However, in our ap-proach, not all of Moens' classes subsumed under the category of events (i.e. culmination, culminated process, point, process) are in-herently telic, which they are in Moens' opinion.

b. Even if Moens' boundedness corresponds with what we under-stand by the concept boundedness, our approaches are not compati-ble either as not all of Moens' classes subsumed under the category of events (i.e. culmination, culminated process, point, process) are inherently bounded, which they are in Moens' opinion."


c. Moens argues that the addition of a/or-adverbial to a process or state results in a culminated process (Moens 1987: 67, 70). As will be shown below, it is only if the adverbial forms part of the aim to be reached or if it refers to an inherent endpoint that the sentence with a /or-adverbial can be called a telic proposition. John played in the garden for two hours is telic only if it was John's intention to engage in this activity for a particular length of time (cf. section 1.3.4.1.).

d. As pointed out above, Moens (1987: 58) subdivides the event-categories into those that have consequences and those that do not. The former are telic, the latter atelic. The consequences can be re-ferred to by means of a perfect tense. Moens then draws the con-clusion that what he calls atelic events cannot be used in a perfect tense sentence:

Atelic extended events extend in time without such a final point. Because of this, they do not allow reference to consequences, as can be seen from the infelicity of I have just worked in the gar-den. Of course, / have worked in the garden can be used felici-tously, for example in the case where work in the garden de-scribes a particular job that had to be carried out. In this context, work in the garden has a particular terminal point associated with it; the perfect can then be used to refer to the consequences asso-ciated with this telic event (Moens 1987: 58).

If one adopts such an approach, it means that sentences such as I have lived in London, The guests have complained about the bad service should all be unacceptable, as they do not contain culminated processes.

e. It is difficult to determine accurately those respects in which my use of the terms (a)telic and (un)bounded differs from Moens'. The system he proposes is very dynamic: Moens believes that states can-not be bounded (in his sense) whereas processes (i.e. activities) are inherently bounded. States first have to be turned into processes be-fore they can combine with an adverbial (Moens 1987: 70). Sim-ilarly, "for a culminated process to occur in the progressive, it has to be turned into a process first" (Moens 1987: 91).

Although Dahl as well as Moens make a lot of interesting points in connection with taxonomies of the type under investigation, it will be shown in what follows that a distinction in terms of what I call (un)boundedness and (a)telicity is necessary in order to give an ade-quate account of a number of issues relating to tense.

Tests 17

1.2. Tests to distinguish unboundedness from boun-dedness and atelicity from telicity

There are a number of tests which are useful for finding out whether a sentence is (un)bounded or (a)telic. The following set of examples will be used to illustrate the tests:

(15) John drank beer, (atelic unbounded) (16) John drank a glass of beer, (telic bounded)

For each test, I will indicate whether it is useful for (a)telicity and/or for (un)boundedness, and point out possible problem cases.

1.2.1. Test 1

Bounded sentences can be used in answer to the question "How long does it take X to X?", unbounded sentences can be used in answer to the question "For how long?"

(15) a. For how long did John drink beer? John drank beer for half an hour, (original sentence is unbounded)

(16) a. How long did it take John to drink a glass of beer? John drank a glass of beer in half an hour, (original sentence is bounded)

It should be noted, however, that adding a for-adverbial indicating a specific length of time to an unbounded non-progressive sentence turns it into a bounded one. In other words, the answer in (15a) is a bounded statement (cf. Declerck 1991a: 265-266, Hirtle 1987: 90-91).

A non-progressive atelic sentence does not allow the use of an in-adverbial (cf. (17a)), whereas a non-progressive telic sentence does (cf. (17b)):

(17) a. I lived with her for two years/ * in a year, b. I did the exercise in 10 minutes.

The following points indicate there are limits on the applicability of the test:


a. Certain sentences can collocate with a /or-adverbial as well as with an /n-adverbial:

(18) a. I mowed the lawn in half an hour, (telic) b. I mowed the lawn/or half an hour, then my brother

took over.

In (18a), I mow the lawn is a telic proposition. The sentence in (18b) can be interpreted in two ways: for half an hour in (18b) can form part of the telic proposition itself, in which case the telic proposition is I mow the lawn for half an hour (i.e. once I have been mowing the lawn for half an hour, my aim is reached). If for half an hour is not the intended endpoint to the situation of mowing, I mow the lawn is understood as an atelic activity. Moreover, if there is reference to a habitual situation in (18b) (e.g. When I was a girl, I mowed the lawn for half an hour, then my brother took over) the sentence is also atelic. The examples in (18) prove that certain sentences allow a telic as well as an atelic interpretation, i.e. they are zero-telic (cf. section 1.4.).

b. The test can only be used for (a)telicity if the sentence is non-progressive, as progressive telic sentences do not collocate with an m-adverbial:

(19) a. ?? He was pushing the cart into the barn in 15 min-utes.

A for-adverbial is acceptable in this context, but it has the effect of turning a telic proposition into an atelic one (cf. example (18b)):

(19) b. He was pushing the cart into the barn for 15 min-utes.

c. Moreover, as punctual situations hardly take up any time at all, it is clear that one is likely to run into difficulties when applying this test to non-progressive telic punctual situations; the situation takes up so little time that the addition of an adverbial is superfluous (cf. e.g. Canavan 1983: 80, Stegu 1983: 322). If one wants to add an ad-verbial all the same, it will have to be one which refers to an ex-tremely short period of time, e.g. in a second. In (20b), the m-adver-bial is understood as referring to the final point in time of the period preceding the situation (cf. also (24a) and (24b)):

Tests 19

(20) a. The bomb exploded in 1 second.12

b. The train arrived in 5 minutes. (Moens 1987: 71) c. The mountaineer reached the top in less than three

hours, (ibid.)

I agree with Moens (1987: 71) that the time adverbial in (20c) refers to the period of time which precedes the actual arrival on top of the mountain (cf. also Van Voorst 1987: 264). However, unlike Moens, I do not believe that a similar line of reasoning is applicable to (20b). In this example the adverbial is understood as referring to a point in time, i.e. five minutes later (cf. e.g. Van Voorst 1987: 264, Smith 1991: 56, 157).

d. The test is not useful for sentences already containing a dura-tional adverbial:

(21) a. John was at home from two till four. b. ?? For two hours, John was at home from two till

four. c. ?? John was at home from two till four in two hours. d. For two years, John was at home from two till four.

For two years in (2Id) indicates the period during which John was in the habit of being at home from two till four. In other words, (2Id) only allows a habitual reading, whereas (21a) also allows a single event reading. Sentence (21a) is bounded on the non-repetitive read-ing but unbounded on the habitual reading. The addition of an adver-bial to a sentence which already contains an w-adverbial has the same effect:

(22) a. He reached home in less than 20 minutes. b. For two years, he reached home in less than 20 min-

utes. c. * He reached home in less than 20 minutes in 5

hours.

Like (2Id), (22b) only allows a habitual interpretation. Test 1 can therefore not be used for the classification of (22a) on its single event reading.

e. The following examples do not conform to the expected pattern because a bounded telic clause appears used in combination with a for- adverbial:


(23) a. He went to London for two weeks. (Dillon 1973: 274)

b. The lake froze for four months, (ibid.)

Dillon points out that these examples are exceptional because there is reference to the state which arises as a result of the event: "When ex-pressions of duration are used, the interpretation is 'duration of state after completion', (or: 'become and remain')" (Dillon 1973: 274). Schöpf (1984) makes the same observation in connection with the sentence I've opened the window for a second: "So ist das einfache Present Perfect einer punktuellen Veränderung (open the door z.B.) in der Regel ein resultatives Perfect und verzeitet in Verbindung mit einem Intervalladverb den Nachzustand und nicht die Veränderung selbst" [A punctual change (e.g. open the door) referred to by means of a non-progressive present perfect is usually a resultative perfect; when used in combination with an adverbial of duration it refers to the resulting state rather than to the change itself.] (Schöpf 1984: 23) (cf. also Mittwoch 1980: 204, 221, Moens 1987: 70, Mommer 1986: 27, 69-73, Zydatiss 1976: 100). Conversely, it also appears to be pos-sible for unbounded sentences to combine with an in-adverbial:

(24) a. John slept in an hour. (Moens 1987: 72) b. In less than a day, Bea knew the answer, (ibid.)

However, what happens here is that the addition of an adverbial changes the lexical meaning of what are without the adverbial atelic sentences (i.e. John slept, Bea knew the answer). Sentence (24a) refers to the situation of falling asleep, whereas (24b) refers to find-ing the answer. The adverbials indicate in the first place a point in time and not an interval of time (cf. Mittwoch 1980: 204, Verkuyl 1989: 51). Two conclusions can be drawn from these examples: (a) they suggest that particular sentences are ambiguous between a telic and atelic or bounded and unbounded reading (cf. section 1.4.), (b) adverbials may influence a sentence's classification as (un)bounded and (a)telic (cf. section 1.3.4.).

1.2.2. Test 2

Unbounded sentences refer to homogeneous situations, i.e. the situ-ation remains the same throughout the time it holds. This charac-teristic has also been called the subinterval property. Bounded sen-

Tests 21

tences refer to heterogeneous situations: if we focus on subintervals of the bounded sentence, the activity differs in the different stages of the situation, "they cannot be broken down into smaller entities that can be indicated with the same name" (Declerck 1991a: 264):13

(15) b. Even though John may not have actually been drinking throughout a period of thirty minutes, the activity (drinking beer) was the same at all the in-tervals during which he drank.

(16) b. The emptying of the glass progresses as John drinks. Therefore we can say that, although the activity re-mains the same (i.e. John drinks beer), the situation as a whole (John drinks a glass of beer) does not ap-ply to all the subintervals of the situation.

The fact that telic as well as atelic sentences may be bounded/unbounded implies that this test is useful for (a)telicity only if it is applied to the (a)telic proposition (e.g. I drink a glass of beer) and not to the actual progressive sentence in which it may be used (e.g. I am drinking a glass of beer).

Verkuyl (1989: 54) and Zydatiss (1976: 71) point out that one may run into difficulties with this test in case a plural NP is used:

(25) a. John played organ concertos.

They argue that although play organ concertos is atelic (and un-bounded), one cannot say that the situation is homogeneous at the moment when John is playing the first concerto: he can only play one concerto at a time. However, the test can be used for sentences such as (25a) as long as one knows that it was John's purpose to play more than one concerto. Still, the following related sentence causes prob-lems for the test:

(25) b. John played five organ concertos.

One could adopt a line of reasoning similar to that used for (25a): if one knows beforehand that John will play five concertos, it is pos-sible to defend the idea that the sentence refers to a homogeneous activity and should hence be classified as an atelic (and unbounded) sentence. However, it will be clear that it is bounded and telic if we know that playing five organ concertos is the intended limit to be at-tained. In other words, if we follow the test, sentence (25b) will be


classified as atelic (and unbounded) because it is homogeneous, whereas it should be interpreted as representing the situation reach-ing a terminal point, i.e. as bounded and telic.

1.2.3. Test 3

If the clause can be added to stop in the form of an ing-clause, the sentence is atelic. If it can be added to finish in the form of an ing-clause, the sentence is telic:

(15) c. John stopped drinking beer. * John finished drinking beer.

(16) c. * John stopped drinking a glass of beer. John finished drinking a glass of beer.

The test is not useful for the classification of sentences as bounded or unbounded. Moreover, it should be taken with some caution as far as (a)telicity is concerned, as the following set of examples prove:

(26) a. He stopped filling the tank, (because it was time to stop working)

b. He finished/completed filling the tank before it was time to stop working.

The fact that both constructions are possible indicates that some sentences allow a telic as well as an atelic interpretation, i.e. some sentences are zero-telic (cf. section 1.4.). Example (26a) refers to the cessation of the activity which precedes the actual attainment of the endpoint.

Furthermore, the test is useful only for telic activities (not for telic states (cf. (27)) and non-punctual telic (and not for punctual telic (cf. (28)) sentences:

a. He lived in London for 20 years can have a telic reading (i.e. living in London for a period of twenty years was the aim to be reached). It should therefore be possible to use this sentence after finish, which it is not:

(27) * He finished living in London for 20 years.

b. He opened the door is bounded telic. However, it cannot be used as a complement of finish (cf. e.g. Cochrane 1977: 96, Dowty 1986:

Tests 23

42, Heinämäki 1978: 7, Mommer 1986: 30, Smith 1991: 62, Zydatiss 1976: 89):

(28) ?? He finished opening the door.

Verkuyl (1989) certainly has a point when he argues that it is (voluntary) agentivity which ultimately determines whether a par-ticular sentence can be used as a complement of stop:

(29) a. ? Stop being ill, being eight feet long, being loved. (Verkuyl 1989: 48)

b. * She stopped loving him. (ibid.)

The sentence is unacceptable when love is interpreted as a stative verb. In (29c), be loved is interpreted as a dynamic verb and can ac-cordingly follow stop:

(29) c. She stopped being loved by her superiors a long time ago.

1.2.4. Test 4

A test which is useful for (a)telicity but not for (un)boundedness is the following: if the answer to "If one was verb-'mg but was inter-rupted while verb-ing, has one verb-edT' (Garey 1957: 105) is pos-itive, the sentence is atelic. A variant of this test is that whenever a progressive sentence entails its non-progressive counterpart, the clause is atelic:

(15) d. If John was drinking beer and he stopped drinking beer, has he drunk beer? Yes.

(16) d. If John was drinking a glass of beer and he stopped drinking a glass of beer, has he drunk a glass of beer? Not necessarily.

McCoard (1978: 161-162) points out that native speakers' judgments about the acceptability of the answers do not always coincide. One should therefore handle the results of this test with some caution.

The following problem cases should be pointed out: a. Zero-telic sentences (cf. section 1.4.) can again be used in both

constructions:


(26) c. If John was filling the tank and he stopped filling the tank, has he filled the tank? Yes.

d. If John was filling the tank and he stopped filling the tank, has he filled the tank? No.

In (26c), fill the tank is understood as an atelic activity: the complete filling of the tank is not something John aims to do. The question can also be answered negatively, as (26d) shows. On this (telic) reading, there is a natural endpoint to the situation, namely the completion of the filling of the tank.

b. It is difficult to use this test to assess sentences of the type John played piano concertos. If someone asks the relevant question in the course of the first concerto, the answer will be negative, in spite of the fact that the sentence is atelic.

c. Another problem is that punctual verbs, when used in the pro-gressive, tend to be given a repetitive reading. This implies that test 4 cannot be used to classify non-repetitive punctual situations (cf. Mommer 1986: 173, Smith 1991: 56):

(30) If someone is coughing (necessarily repetitive interpreta-tion) and he stops coughing, has he coughed?

Although the above tests are usually helpful, it is not always easy to classify sentences straightforwardly as bounded or unbounded and telic or atelic:

(31) a. The endless procession walked by the church. (Declerck 1979: 768)

b. We enjoyed camping until mosquitoes showed up. (Heinämäki 1978: 81)

c. John lived in Paris for some time. d. John was here yesterday. (The sentence refers to a

subpart of yesterday) e. I spoke to John at five o'clock. f. John wrote a letter.

Example (31a) has led Declerck (1979) to posit a category of zero-bounded statements, i.e. statements that obey the tests for unboun-dedness as well as those for boundedness. This issue will be taken up in section 1.4. The sentence in (31b) indicates that adverbials can af-fect (un)boundedness in that they may impose a right or left bound-ary on an unbounded situation. This example calls for the additional

Factors influencing (un)boundedness and (a)telicity 25

concepts 'bounded to the right/unbounded to the left" and "bounded to the left/unbounded to the right" (cf. section 1.5.). Examples (31c), (3 Id), (31e) and (3 If) raise the question of what it means for a situa-tion to be represented as having reached a terminal point. How ex-plicit should the indication of the boundaries be? The issue brought up by the latter examples will be discussed in the next section: a sur-vey will be given of the elements which determine whether a sen-tence is interpreted as unbounded or bounded and atelic or telic.

1.3. Factors which influence (un)boundedness and (a)-telicity

Every effort has been made to list the factors which influence the classification of a sentence as an accomplishment, achievement, ac-tivity or state.14 Again, some of these factors affect the (un)-boundedness of the clause as well as its (a)telicity. Mourelatos (1981) correctly observes that

In all cases, a total of six factors are involved: (a) the verb's in-herent meaning; (b) the nature of the verb's arguments, that is, of the subject and of the object(s), if any; (c) adverbials, if any; (d) aspect; (e) tense as phase (e.g., the perfect); ( f ) tense as time ref-erence to past, present or future. (Mourelatos 1981: 199)

The factors dealt with at length elsewhere will not be discussed in great detail. The focus will be on the influence of tense and adver-bials on (un)boundedness and (a)telicity, as a number of points have so far not received sufficient attention in this respect.

1.3.1. NPs

NPs affect (a)telicity. They affect (un)boundedness indirectly: if a NP has the effect of turning an atelic proposition into a telic one, and if the telic proposition is used in a non-progressive sentence, the lat-ter will be bounded:

(32) a. Petrol leaked out of the tank, (unbounded atelic) b. All the petrol leaked out of the tank, (bounded telic)


The use of a mass noun in an indefinite NP in (32a) makes the sen-tence atelic. The following examples also illustrate the effect of the NP:

(33) a. I ate several apples, (bounded telic) b. I ate an apple, (bounded telic) c. I ate apples, (unbounded atelic)

The sentence in (33a) raises an issue referred to earlier: how explicit should the indication of a boundary be in order for a sentence to be classified as bounded? The NP several apples implies, but does not explicitly indicate a limited number. However, as there is an impli-cation of limitation, sentence (33a) may be classified as bounded. The following sentence from Shi (1990) is an interesting one:

(34) Tourists drank the milk in an hour. (Shi 1990: 49)

Sentence (34) can be interpreted in several ways: a. There may be reference to a number of subsituations following

each other. Tourists drank milk one after the other and it took each tourist an hour to drink the milk. Once one tourist had drunk his milk, the next tourist started drinking. The subsituations are bounded (and telic), the whole situation is unbounded (and atelic), as the sub-ject NP does not indicate a limited number of tourists.

b. Another possible interpretation was that it took each tourist an hour to drink the milk but the subsituations occurred at the same time, for instance, the tourists all started drinking at 9 o'clock and finished drinking at 10 o'clock. A limited number of tourists is im-plied. The situation is represented as bounded and telic.

c. A third possibility is that the different tourists drank milk for e.g. five minutes. Once one tourist had finished drinking, the next one started drinking and the whole process took one hour of time. Again, there is an implicit limit on the number of tourists. The situation is represented as bounded and telic. As the effect of NPs has already been dealt with by many linguists, I will not discuss this issue in more detail.

It has already been pointed out that (a)telicity and (un)boundedness characterize descriptions of situations and that from that point of view a particular situation is not inherently (a)telic or (un)bounded. Conversely, the sentence in (34) shows that one sentence may repre-sent several situations. An additional observation in this respect is that the changes effected by a particular feature often result in the


sentence referring to a different extra-linguistic reality. When dis-cussing the factors that influence the categorization, linguists indeed seem to lose sight of this non-trivial proviso: although changing a singular NP into a plural NP, for instance, is likely to change a sen-tence's classification as bounded/unbounded or telic/atelic, it may also result in the sentence no longer capturing a particular situation. Although theoretically, the speaker can choose freely how to repre-sent a particular situation, the freedom is not complete because the situation imposes constraints on the linguistic material by which it can be represented. This observation applies to all the features possi-bly influencing (un)boundedness and (a)telicity. The following ex-ample may clarify the point made: if someone is reading a book, one might refer to that particular situation by means of either John is reading (atelic unbounded) or John is reading a book (telic un-bounded): the addition of the direct object coincides with a change from atelic to telic. Changing the singular NP into a plural NP turns the sentence into an atelic unbounded statement: John is reading books. However, if John is not reading several books at the same time, this statement is no longer a truthful presentation of the situa-tion, although it does illustrate that NPs affect (a)telicity.

1.3.2. Directional PPs

Directional PPs15 affect (a)telicity; their influence on (un)-boundedness is similar to that of NPs (cf. section 1.3.1.):

(35) a. John pushed the cart. (- directional PP) (unbounded atelic)

b. John pushed the cart into the barn. (+ directional PP) (bounded telic)

(36) a. She walked around in the station. (- directional PP) (36) (unbounded atelic)

b. She walked to the station. (+ directional PP) (bounded telic)

The effect of directional PPs is also an issue which has been exam-ined before by many people. Accordingly, no more attention will be devoted to it here.


1.3.3. Tense

Tense may affect the representation of a situation as bounded or un-bounded. As far as I know, this is a point which has not been ana-lysed in the articles and books on the issue under discussion. The following examples show that the use of a non-progressive perfect tense may turn an unbounded sentence into a bounded one:16

(37) a. John too loves Mary, (unbounded atelic) b. John too loved Mary, (unbounded atelic) c. John too will love Mary, (unbounded atelic) d. John too has loved Mary, (bounded atelic) e. At that time, it was clear that John too had loved

Mary, (bounded atelic) (38) a. John plays football, (unbounded atelic)

b. John played football, (unbounded atelic) c. John will play football, (unbounded atelic) d. John has played football, (bounded atelic) e. It was clear that John had played football before,

(bounded atelic)

The non-progressive perfect sentences in (37) and (38) differ from the other non-progressive tense examples in that they represent a sit-uation as bounded. As will become clear in chapters 3 (section 3.2.1., section 3.4.2.1.) and 4 (section 4.2.2.1.), this observation has important consequences for the expression of temporal relations in the past sector and pre-present sector. Before going into the effect of the perfect on (un)boundedness, some observations will be made about the past tense - (un)boundedness relationship.

1.3.3.1. Past tense

It is a well-known fact that a situation referred to by means of a non-progressive past tense has the implicature that the situation no longer holds (cf. e.g. Comrie 1985: 41-42, Declerck 1991a: 266, Riddle 1978: 78, 100-101):17

(39) a. He remained silent. b. He talked about his trip to London. c. He lived in London at the time.


From that point of view one might wonder whether all non-pro-gressive past tense sentences cannot be classified as bounded, as they are over at t0 (the temporal zero-point, i.e. the moment of speaking). In order to formulate an answer to this question, it is necessary to take into account the structure of the tenses. Declerck (1991a) sum-marizes the semantic structure of the past tense as follows:

TS18 simul TO 19

TO , before TO, Sit 1 TOsU belongs to the past time-sphere TO,™ belongs to the present time-sphere (Declerck 1991a: 295)

If one assesses the past time situations in (39a) and (39b) from the position of TOsjt, the sentences are unbounded:21 there is no explicit or implicit reference to boundaries. From that point of view, (39a) and (39b) do not differ from (39c), in which the TE (established time), "a time that is established by an adverbial or by the context" (Declerck 1991a: 251), is linguistically realized. However, if one looks at the situations referred to by means of a non-progressive past tense from the point of view of t0, they seem to be bounded: the im-plicature that the situation no longer holds at t0 implies that there is a boundary to any past tense situation. However, the latter approach is not very useful:

a. For one thing, it implies that the progressive counterpart of (39b) should also be labelled as bounded, as the progressive past tense also has the implicature that the situation no longer holds:

(39) d. He was talking about his trip to London.

This runs counter to a claim made earlier on (cf. section 1.1.), i.e. the use of a progressive tense usually results in an unbounded reading.

b. For another, on this approach (i.e. when assessing sentences from the point of view of t0), any future tense situation (which is not of the "from t„ onwards" type, e.g. From now on I will call you Johnnie) is to be classified as bounded to the left (cf. section 1.5.) and can never be unbounded to the left.

c. Moreover, as far as the importance of (un)boundedness for the use of tense is concerned, it is (un)boundedness with respect to TOsit which is important. In other words, in order to make a judgment on the (un)bounded na-ture of a past tense/future tense sentence, I will keep to the practice of doing so from the point of view of TOsit and not from the point of


view of t0 (for reasons which will become clearer in chapters 3 and 4).

1.3.3.2. Present perfect22

The non-progressive present perfect/past perfect, unlike the non-pro-gressive past tense, has a bounding effect on sentences:23

a. When the clause is telic, both the (non-progressive) indefinite present perfect/past perfect and the (non-progressive) past tense rep-resent a situation as bounded:

(40) a. John killed the suspect. b. John has killed the suspect.

b. The bounding effect of the indefinite perfect is particularly clear in atelic clauses:

(41) a. John played in the garden, (unbounded atelic) b. John has played in the garden, (bounded atelic)

(42) a. John lived near the Tower, (unbounded atelic) b. John has lived near the Tower, (bounded atelic)

Although it has been pointed out before that a situation referred to by means of a present perfect tense is an "accomplished fact" (Bauer 1970: 192) (cf. also Defroment 1973: 20, and Smith 1991: 148), this view is not shared by everyone. Binnick (1991), for instance, wrongly assumes that "English has no completive marking, since both its past and present perfect tenses are neutral in this regard. Neither Tom read the book nor Tom has read the book carries any implication in itself as to whether he finished reading it" (Binnick 1991: 59). The different behaviour of the past and the present perfect can be explained in terms of their temporal structures: the indefinite present perfect does not presuppose a past time in the way the past tense does. Declerck (1991a: 306) argues that the difference between the present perfect and the past tense lies in the fact that the former locates a situation in the present time-sphere, whereas the latter lo-cates a situation in the past time-sphere (cf. chapter 2, section 2.1.2.). The difference between locating a situation in the present time-sphere and locating a situation in the past time-sphere boils down to the fact that the past presupposes a past time R (E,R - S in Reichen-bachian terminology), whereas the present perfect does not (E - R,S


in Reichenbachian terminology). An indefinite present perfect sen-tence is evaluated from the point of view of t0 or of the post-present TO, which behaves as if it were t0 (cf. chapter 2, section 2.1.5.4. and example (43)); there is reference to a situation which does not lead up to tQ (or to the post-present binding TO). Accordingly, the sen-tence is bounded:

(43) He will say he has lived near the Tower.

Another question which deserves to be looked into is that of the classification of progressive perfect sentences and continuative pre-sent perfect sentences (cf. Allen 1966: 205):

a. Some sentences with a progressive present perfect which gets an indefinite interpretation are bounded and imperfective:

(44) a. I've been playing football. (That's why my shirt is so dirty.)

b. I've been drinking beer. (That's why I can no longer walk properly.)

b. Sentences with a present perfect tense which gets a continuative interpretation (no matter whether the verb form is progressive or non-progressive) are bounded as well. The left boundary is explicitly mentioned in the clause, the right boundary is implicit in the struc-ture of the continuative present perfect itself, i.e. a situation starts in the past and leads up to now:

(45) a. I have lived here since 1985. b. I have been waiting for you since 8 o'clock.

The situations which sentences of the type in (45) refer to are likely to continue after t0. However, what matters is how the situations are represented in the sentence and not the actual duration of the situa-tion itself. Therefore, one might say that in sentence with a continua-tive present perfect, the situation is represented as having t0 as right-hand boundary.

1.3.3.3. Present tense

If the tense-(un)boundedness/(a)telicity interaction has been touched upon at all, it relates to the effect of the present tense. The following


examples have been given to illustrate how this particular tense may change the Aktionsart of a sentence:

(46) a. John writes a good book. (Moens 1987: 54) b. John wrote a good book, (ibid.)

Moens comments: "The example in [46a] will be interpreted as ex-pressing a dispositional or habitual state of affairs, whereas [46b] will normally be interpreted as a description of a one-off event. It is clearly the difference in tense which accomplishes this aspectual dis-tinction" (Moens 1987: 54). Mommer (1986) and Vasudeva (1971) give similar examples:

(47) a. Nick crossed the Graffiti Bridge, (durational - cul-mination point) (Mommer 1986: 88)

b. Nick crosses the Graffiti Bridge, (series) (ibid.) (48) a. He arrived late, (momentaneous) (Vasudeva 1971:

128) b. He arrives late, (momentaneous or durative) (ibid.)

The use of a non-progressive present tense to refer to present time events is restricted to "instantaneous" situations (Leech 1971: 6-7). The present tense is usually reserved to refer to habits or states. As an event interpretation of a present tense sentence is marked, sen-tences of type given in (46a), (47b) and (48b) are likely to be given a habitual interpretation. In other words, it is because the present tense triggers a habitual reading that the sentence is classified as atelic and unbounded: although the separate occasions are bounded and telic, the situation as a whole is unbounded and atelic. From this it follows that any factor which induces a habitual reading will affect a sen-tence's classification in terms of situation type. This line of reasoning also implies that if a habitual reading is given to the sentences in (47a) and (48a), they will not be considered as bounded and telic ei-ther, but rather as unbounded and atelic.

1.3.3.4. Conclusion

The following conclusion can be formulated: a. Situations referred to by means of the present tense and present

perfect are considered from the point of view of t0 to be classified in terms of (un)boundedness.

Factors influencing (un)boundedness and (ajtelicity 33

b. Situations referred to by means of the past tense and future tense should be assessed from the past/future point of view of TOsit, in which case there are a number of interacting factors which determine whether they are bounded or unbounded.

1.3.4. Adverbials

In this section, I will examine the effect of temporal adverbials on (a)telicity and (un)boundedness, as this issue has not been dealt with extensively by other linguists, witness the following quote from Schöpf (1984):

Wir konnten sehen, dass man zunächst Klassenbildung im Bereich verbaler Lexeme vorgenommen hatte, später wurde die Komplementierung der Verben durch Nominalphrasen mit einbe-zogen. Schliesslich wurde die Quantorenmodification der Nomi-nalphrasen mitberücksichtigt, so dass man davon ausgehen kann, dass der einfache Satzbegriff bzw. häufig auch der Predikatsbe-griff den Bereich darstellt, in dem nach dem inhärenten Zeitbezug zu fragen ist. Ungeklärt erscheint bislang, wieweit Zeitadverbiale hier einzubeziehen sind. (Schöpf 1984: 79, italics mine) [It has been pointed out that first, the classification of situation types focussed on the level of the verb; later on, the effect of the NP complements to the verb were also taken into account. Ulti-mately, the effect of quantification bearing on the NP was also considered, so that we can conclude that it is the sentence or the predicate which constitutes the level of inherent time reference. It has as yet not been established, though, to which extent time ad-verbials should be taken into account.]

Mommer (1986: 85-87) is one of the rare linguists to discuss the ef-fect of adverbials to some extent (cf. also Smith 1991: 159). She gives the following examples:

(49) a. Brandy combed her doll's hair in 10 minutes. (durational - culmination point) (Mommer 1986: 85)

b. Brandy combed her doll's hair for 10 minutes, (durational) (ibid.)

(50) a. The snowman melted, (unstructured culmination point) (ibid.)


b. The snowman melted completely, (durational cul-mination point) (ibid.)

(51) a. Garfield blinked (once) with surprise, (punctual) (ibid.)

b. Garfield blinked for five minutes after he woke up from his nap. (series) (Mommer 1986: 86)

In the next sections, the focus will be on certain types of temporal adverbials.

1.3.4.1. Adverbials affect (a)telicity

Time-position adverbials (Quirk et al. 1985: 529) and adverbials of span and duration (Quirk et al. 1985: 533) may affect the status of a clause in terms of (a)telicity. However, they do not necessarily be-long to the (a)telic proposition. Let us first consider the effect of adding a temporal adverbial to a telic (punctual) situation:

(52) a. Helen dropped the teddy bear on the floor. b. Helen dropped the teddy bear on the floor for half

an hour.

The addition of a temporal adverbial to the original sentence induces a repetitive interpretation (cf. e.g. Moens 1987: 69, Schöpf 1984: 85). I will assume that drop indicates an unintentional action. On this interpretation it cannot be Helen's intention to drop the bear on the floor for half an hour. In other words, for half an hour cannot be con-sidered the intended endpoint to the activity of dropping the bear on the floor. This implies that although for half an hour triggers a repet-itive interpretation and in this way imposes an atelic reading, it is not part of the atelic proposition itself {Helen drop the teddy bear on the floor). The following test can be used to determine whether the ad-verbial is part of the (a)telic proposition or not. If the adverbial can fill position X in the pseudo-cleft construction "What + subject + form of do + X + is + nonfinite clause" without changing the mean-ing of the original sentence, it means that it is an optional constituent of the VP (cf. Declerck 1991b: 46-47) and hence is not part of the (a)telic proposition. As (52c) is an accurate paraphrase of (52b), it must be that the adverbial does not belong to the atelic proposition.

Factors influencing (unjboundedness and (a)telicity 35

(52) c. What Helen did for half an hour is drop the teddy-bear on the floor.

It should still be added that a repetitive situation can also be telic, i.e. when the number of times the situation holds is predetermined. John went 5 times to London is telic if it was John's aim to go there 5 times. However, John went to London five times a year will be clas-sified as atelic unless the period during which he had this habit is given in the context (cf. Declerck 1979: 770, 789).

The clause in (53a) is zero-telic:

(53) a. John trimmed the hedge, (telic (or atelic), i.e. zero-telic)

b. John was trimming the hedge, (atelic or telic, i.e. zero-telic)

The unmarked interpretation of example (53a) is that it indicates that John trimmed the hedge completely, i.e. the completion of the trim-ming is the built-in endpoint. The sentence is telic on this reading. However, (53a) might also be interpreted as referring to an activity, in which case it is atelic. It should immediately be added that an atelic interpretation is likely to result in a progressive form being used (cf. (53b)). Accordingly, (53b) may be classified as atelic al-though a telic interpretation is not ruled out either. On the telic read-ing, a natural endpoint is still implied, but the sentence does not rep-resent the situation as having reached this endpoint. (Non-punctual) time-position adverbials and adverbials of span and duration may af-fect the (a)telicity of a sentence in several ways:

(53) c. John trimmed the hedge when he was a boy. (atelic) d. John trimmed the hedge for two hours, (atelic or

telic) e. John trimmed the hedge in two hours, (telic)

a. Sentence (53c) refers to a habitual situation. The adverbial influ-ences the (a)telicity of the sentence: when it is not present, the sentence is zero-telic. When it is added to the sentence, the sentence becomes atelic, the atelic proposition being he trim the hedge (cf. (52b)).

b. As for (53d), John's intention may have been to trim part of the hedge (atelic), i.e. the completion of the trimming was not the end-point towards which he strove. Although the addition for two hours


is not part of the goal to be reached, it does affect the (a)telicity of the sentence. As a consequence of adding the adverbial, trim the hedge is considered an activity which turned out to have a duration of two hours. In this particular context, in which for two hours is not part of the (a)telic proposition, the adverbial affects the (a)telicity of the sentence all the same: it makes the sentence atelic. The atelic proposition is he trim the hedge.

c. It may also have been John's intention to trim the hedge for two hours (telic), i.e. trimming the hedge for a certain amount of time, namely two hours, was John's aim. On this reading, for two hours is part of the telic proposition, which is he trim the hedge for two hours. An accurate paraphrase of the sentence on this reading is the pseudo-cleft construction What John did is trim the hedge for two hours rather than What John did for two hours is trim the hedge. In other words, the test proves that the adverbial is an obligatory con-stituent and therefore, it belongs to the telic proposition. Example (53d) shows that not all the constituents which feature in the clause need be part of the (a)telic proposition. A fourfold dis-tinction can be made as regards the effect of adverbials on (a)telicity:

1. The addition of the adverbial changes the (a)telicity of the sentence. The adverbial belongs to the atelic proposition (e.g. possible interpretation of (53d))

2. The addition of the adverbial changes the (a)telicity of the sentence. The adverbial does not belong to the atelic propo-sition (e.g. (52b)).

3. The addition of an adverbial does not change the (a)telicity of the sentence. The adverbial belongs to the (a)telic propo-sition (e.g. He intentionally crossed the street (for ten min-utes).).

4. The addition of an adverbial does not change the (a)telicity of the sentence. The adverbial does not belong to the telic proposition (e.g. He forged the lock (in 20 minutes).).

The claim that adverbials may "change the (a)telicity of a sentence" means that the addition of an adverbial to the sentence changes the clause's classification in terms of (a)telicity or rules out one of two possible interpretations which the sentence allows when the adver-bial is not used. However, the fact that adverbials may affect the (a)telicity of a sentence does not automatically mean that they are part of the (a)telic proposition: adverbials can be part of the (a)telic


proposition only if there is an agent (actor) involved or when there is reference to an inherent capacity (e.g. The bulb lasts for 24 hours).

The following sentences also illustrate how adverbials may affect (a)telicity:

(54) a. The ball bounced for a couple of seconds, b. The ball bounced.

In (54a), bounce may be classified as atelic because the adverbial in-dicates repetition. On this interpretation, for a couple of seconds is not part of the telic proposition because it cannot be the ball's aim to bounce for a couple of seconds, the ball is theme (patient). However, there might also be reference to an inherent quality of the object: if the ball is designed to bounce for a couple of seconds, the ball bounce for a couple of seconds may be classified as telic. When there is no adverbial in the sentence (cf. (54b)), the sentence is zero-telic: it is telic if the ball bounced once, atelic if there was an iteration of bounces.

Example (55) will be telic or atelic depending on whether the ad-verbial was part of the subject's aim:

(55) John lived in London for 20 years.

If John intended to live in London for 20 years, the sentence is telic bounded (telic proposition: John live in London for 20 years). If his intention was to live in London, and he noticed afterwards that he did so for 20 years, the statement is bounded atelic (atelic proposition: John live in London):

Bach (1980) observes that the addition of a for-adverbial may change a process into an accomplishment:

(56) a. Max pushed a cart, (process) (Bach 1980: 28) b. Max pushed a cart for an hour, (accomplishment)

(ibid.)

The confusion between (a)telicity (Aktionsart) and (un)boundedness is again clear from this observation. Example (56a) is atelic un-bounded. Example (56b) is either bounded telic or bounded atelic, depending on whether it was Max's intention to push the cart for one hour or just push the cart. The adverbial renders (56a) in any case bounded, but does not necessarily change the Aktionsart of the sen-tence.


1.3.4.2. Adverbials affect (un)boundedness

The (un)boundedness of a sentence is also affected by adverbials. The type of adverbials taken into account in the discussion are punctual adverbials, adverbials of duration (Quirk et al. 1985: 533) and time-position adverbials (Quirk et al. 1985: 529). Declerck (1991a) makes a distinction between bounded and unbounded adver-bials: "A TE is bounded if the adverbial refers to its boundaries (beginning and end) rather than merely indicating duration" (Declerck 1991a: 268), "adverbials like yesterday, in 1983, etc. refer to an interval without drawing attention to its boundaries. For that reason they represent an interval as unbounded" (Declerck 1991a: 268).

a. Non-punctual temporal adverbials which provide given informa-tion

One must be careful not to understand bounded adverbial as adver-bial which has a bounding effect: the addition of an (un)bounded ad-verbial to a sentence does not automatically result in making the sentence (un)bounded. Although yesterday and in 1983 are un-bounded adverbials according to the definition, B's answer in the following example is a bounded sentence:

(57) A: Where did John live in 1983? B: John lived in London in 1983.

(58) A: Where was John yesterday? B: He was in the library yesterday.

B's reply in (57) probably implies that John lived in London throughout 1983.24 He may still live there, but this fact is not some-thing which the speaker is interested in.25 He picks out a certain pe-riod in 1983 and asks the hearer to make a claim about this limited length of time. B's reply is a bounded sentence. As far as (58) is con-cerned, pragmatic knowledge tells us that it is likely that John was in the library during the day, possibly at the time when speaker A was looking for him, but not for 24 hours. Still, it makes sense to say that B's reply in (58) gets a bounded interpretation: the speaker focuses on the implicitly bounded interval indicated by the adverbial, i.e. John was in the library for some time yesterday. The situation (of John's working in the library) as such need not have come to an end


(John may still be working in the library at t0), the situation as repre-sented in the sentence has.

Let us compare sentences of the type in (57) and (58) with pro-gressive sentences with a non-punctual adverbial:

(59) A: What was John doing this afternoon? B: He was working in the garden this afternoon.

The progressive refers to the middle of the situation and indicates that the state is temporary. The TE in (59), this afternoon, does not have the effect of making the sentence bounded. There is a certain degree of tension between the bounded period referred to by the ad-verbial and the fact that the progressive typically represents a sit-uation as going on. Person A is only interested in the limited period of time this afternoon and from that point of view there are temporal boundaries on the situation. Still, the situation is represented as un-bounded: the progressive overrides the bounding effect of the given adverbial.

The conclusion to be drawn so far is that temporal adverbials which indicate given information give rise to a bounded reading in non-progressive past tense sentences, but they do not cancel the un-bounded character brought about by the use of a progressive tense.

b. Punctual time adverbials which provide given information

If the punctual adverbial is given information, progressive and non-progressive sentences behave alike:

(60) A: Where was John at 5 o'clock? B: He was in the garden at 5 o'clock.

(61) A: What was John doing at 5 o'clock? B: John was playing in the garden at 5 o'clock.

Strictly speaking, B's reply in (61) merely indicates where John was at 5 o'clock. Even so, the fact that the progressive refers to the mid-dle of a situation implies that the situation extends further into the past and into the future.26 Accordingly, the sentence is unbounded. A similar argument applies to the non-progressive past sentence in (60): although, strictly speaking, person A inquires into where John was at the punctual moment in time 5 o'clock, the state character of the reply suggests that the sentence is unbounded.


The conclusion to be drawn is that a non-progressive sentence with a punctual adverbial which indicates given information (cf. (60)) dif-fers from a non-progressive sentence with a non-punctual time ad-verbial (which explicitly -in the case of adverbials of duration- or implicitly -in the case of time-position adverbials- imposes bound-aries) which indicates given information (cf. (57) and (58)), because the non-punctual adverbial has a bounding effect on the non-progres-sive sentence whereas a punctual adverbial does not.

given information punctual adverbial non-punctual adverbial

progressive UB UB non-progressive UB Β

c. Non-punctual time adverbials which provide new information

The adverbial may also provide new information:

(62) A: When did John live in London? B: He lived in London in 1983.

(63) A: When was John in the library? B: He was there yesterday.

If the speaker conforms to the rules of communication, he will indi-cate the complete period during which John lived in London/was in the library. If John still lives in London, or if he lived there in 1984 as well, the speaker will explicitly mention all the periods when John lived in London. In other words, the adverbial has a bounding effect in B's reply in (62): it is a bounded sentence.27 In (63), the time dur-ing which John was in the library is a subinterval of yesterday; the situation is represented as bounded through the presence of this time-position adverbial.

B's reply in (64) proves that a non-punctual time adverbial giving new information also has a bounding effect on a progressive sen-tence:

(64) A: When was John working in the garden? B: He was working in the garden in the afternoon.

The speaker asks for a past time indication. No matter whether John worked in the garden throughout the afternoon or not, the adverbial


has a bounding effect on the progressive sentence. The "fight" be-tween the bounding effect of the adverbial and the non-bounding ef-fect of the progressive is "won" by the adverbial, because it gives new information and in this way imposes temporal boundaries.

It follows from examples (59) and (64) that the given or new status of the adverbial affects the classification of a sentence in terms of (un)boundedness: in example (59), the bounding effect of the given adverbial is overruled by the unbounding effect of the progressive tense; the sentence is unbounded. In (64), on the other hand, the ad-verbial overrules the progressive tense: the sentence is bounded.

d. Punctual time adverbials which provide new information

Punctual adverbials which provide new information in combination with a (non-)progressive past tense do not have a bounding effect:

(65) A: When was John in the library? B: He was in the library at 5 o'clock.

(66) A: When was John working in the garden? B: He was working in the garden at 5 o'clock.

The presence of a punctual adverbial does not have the effect of making B's replies bounded. B's reply is an unbounded sentence in (65) and (66).28

new information punctual adverbial non-punctual adverbial

progressive UB Β non-progressive UB Β

The effect of past time adverbials on (un)boundedness can be summarized as follows:

1. non-punctual adverbials that indicate given information a. bounding effect on non-progressive past tense sentences (cf. (57), (58)) b. no bounding effect on progressive past tense sentences (cf. (59))

2. punctual adverbials that indicate given information a. no bounding effect on non-progressive past tense sen-tences (cf. (60))


b. no bounding effect on progressive past tense sentences (cf- (61))

3. non-punctual adverbials that indicate new information a. bounding effect on non-progressive past tense sentences (cf. (62), (63)) b. bounding effect on progressive past tense sentences (cf. (64))

4. punctual adverbials that indicate new information a. no bounding effect on non-progressive past tense sen-tences (cf. (65)) b. no bounding effect on progressive past tense sentences (cf. (66))

In some cases, the addition of a punctual adverbial to an unbounded sentence may have the effect of imposing a temporal boundary:

(67) a. I saw Yael. (Hatav 1989: 506) b. Suddenly, I saw Yael. (ibid.)

As a result of adding suddenly, saw is understood as the moment of perceiving and consequent seeing, i.e. the situation is bounded to the left (cf. section 1.5.).

Another point which it is useful to make is that non-punctual ad-verbials may impose a bounded reading on a sentence, irrespective of whether or not there is an explicit indication of the exact duration. There is no difference in boundedness between the following exam-ples:

(68)

(69)

a. John lived in Paris for five years, (bounded) b. John lived in Paris for some years, (bounded) c. John lived in Paris for years, (bounded) a. John worked on his paper from 2 until 5 o'clock,

(bounded) b. John worked on his paper in the afternoon,

(bounded)

1.3.4.3. Effect of adverbials on implicature past tense

A sentence in the past tense has the implicature that the situation it refers to no longer holds at t0 (cf. section 1.3.3.1.):


(70) a. John lived in Paris in 1958 (but he no longer does).

The implicature character is clear from the fact that it can be can-celled:

(70) b. John lived in Paris in 1958 and he still does.

The observations in 1.3.4.2. throw light on the implicature of the past tense sentence that the situation it refers to does not last up to t0:

a. First of all, discussions dealing with this issue (apart from Riddle 1978: 82-100) have not sufficiently stressed that the semantics of the clause together with pragmatic knowledge contribute considerably to the likelihood of the implicature that the situation is no longer the case at the time of speaking (not) being cancelled:

(71) a. Laura worked on her essays for many hours.

If the hearer knows that Laura is no longer a student, it will be self-evident to him that the past situation no longer holds. The sentence will be interpreted as referring to a past habit. The following example is similar:

(71) b. Laura played with her dolls in the garden whenever she had a day off.

If the hearer knows that Laura is now 20 years old, he will be in no doubt as to whether or not she still plays in the garden. Again, he will interpret the sentence as referring to a past situation which is over.29

The examples in (72) show that the pragmatics of the clauses (i.e. our general knowledge about extra-linguistic situations) also influences the likelihood of the implicature applying to a statement:

(72) a. Laura was in the kitchen two hours ago. b. Laura was in the kitchen a minute ago.

It is more likely that Laura is still in the kitchen at t0 in (72b) than in (72a) provided the adverbial represents old information. The neces-sity of adding that the adverbial should provide old information will be clear from the following: if (72b) is an answer to the question When was Laura in the kitchen? (i.e. the adverbial provides new in-formation), a speaker conforming to the rules of communciation will reply that she was there a minute ago only if he knows or suspects


that Laura is no longer there, which means that the situation no longer holds at t0.

b. If it is not clear from the semantic contents and pragmatics of the sentence whether or not the situation still holds at t0, there is a gra-dient in the probability that the implicature arises that the past situ-ation no longer holds at t0:

- to + t0 ( i ) (11) ( l i i ) (IV) ( ν )

Sentence type "-10" (the situation no longer holds at t0) is the type of clause which excludes the possibility of its situation leading up to t0, whereas sentences type "+ t0" (the situation may still hold at t0) con-stitutes the other end of the gradient: the past tense sentence still has the implicature that the situation it refers to does not last up to t0, though the likelihood of the implicature being cancelled is greater. The distinction between past tense bounded and unbounded situa-tions, telic and atelic situations and given and new information plays an important role in setting up the gradient:

(i) A telic bounded situation referred to in the past tense necessarily lies completely before t0 (cf. Declerck 1991a: 266):

(73) a. She killed her husband. b. She drove the car into the garage.

(ii) Past telic unbounded sentences, which are often progressive, exhibit a slightly different pattern:

(74) a. He was painting the house white when I visited them.

b. We watched the children while they were making a snowman.

The temporary character of the progressive sentence makes it un-likely that the situations they refer to will still be in progress at t0. Moreover, telic situations, unlike atelic situations, have an inherent or natural endpoint; it is part of their semantics that they must even-tually come to an end. Our knowledge of the world again plays an important role in deciding whether the situation is still likely to hold at t0. Making a snowman does not take up as much time as painting a house and is moreover limited to the wintertime, which implies that the implicature is more likely to be cancelled in (74a) than in (74b).


It will be clear that a lot depends on how much time has elapsed be-tween TE and t0.

(iii) A past atelic situation bounded by the addition of an adverbial providing new information strongly implicates that the situation does not lead up to t0:

(75) A Β

(76) A Β

When did John live in London? He lived in London in 1983. ?? In fact, he still does. When was John at Judith's place? He was there yesterday. ?? In fact, he is still there.

Indeed, an important stipulation is that the status of the adverbial in terms of given/new information has an influence on whether or not the situation can lead up to t0. If the bounding adverbial provides new information, and if the hearer assumes that the speaker conforms to Grice's maxim of Quantity,30 the situation that is bounded through the addition of an adverbial will be interpreted as being over at t0.

(iv) If, on the contrary, the adverbial provides given information, there are no restrictions of that kind. The relative unacceptability of B's reply in examples (75) and (76) (in which the adverbial repre-sents new information) as compared with the relative acceptability of B's reply in (77) and (78) (in which the adverbial represents given in-formation) demonstrates that the difference between sentence types (iii) and (iv) is justified:

(77) A Β

(78) A Β

Where did John live in 1983Ί John lived in London in 1983. In fact, he still does. Where was John yesterday? He stayed at Judith's place yesterday. In fact, I think he is still there.

Strictly speaking, bounded atelic examples of the type illustrated in (64), in which the adverbial overrides the unbounding effect of the progressive, should be classified under (iii). However, it appears that their behaviour is not fully similar to that of the bounded atelic ex-amples given in (75) and (76):

(79) A: When were you looking for June? Β: I was looking for her throughout the afternoon. In

fact, I still am.


Unlike in the case of (75) and (76), the cancellation of the impli-cative does not result in questionable acceptability. The difference may be explained as follows: the progressive indicates that a situa-tion is on-going and temporary. Accordingly, the first clause in B's reply does not inform us whether Β was successful in her attempt to find June. This makes it more likely that the past tense situation is still applicable to the moment of speaking. In other words, examples of this type should be classified with the examples described in (iv).

(v) An atelic unbounded situation not accompanied by an adverbial weakly implicates that the situation no longer holds at t0:

(80) a. Chris hated everything which made him think of his mother.

Chris may still hate everything that makes him think of his mother. However, the speaker decides to focus on the past part of the situa-tion. When an adverbial is used, it is slightly less likely that the sit-uation still holds at t0:

(80) b. At the time, Chris hated everything which made him think of his mother.

If there is no adverbial present, one of the major factors lying at the origin of the implicature (i.e. the adverbial, apart from the tense, picks out a past moment in time and therefore suggests that the sit-uation is no longer applicable at t0) disappears. This is probably why it is more likely that the situations referred to by sentences of the type in (80a) lead up to L. The above observations explain why I do not agree with Smith (1991), who argues that whereas a stative situ-ation referred to by means of a past tense may still hold at t0, an ac-tivity referred to in the past tense can no longer be the case at t0:

(81) a. ? Lily swam in the pond and she may still be swim-ming. (Smith 1991: 106)

b. Jennifer knew Turkish (a)... but she has forgotten it all. (b)... and she still knows it. (Smith 1991: 109-110)

As pointed out before, a lot depends on the type of situation that is referred to:


(82) When I met him in 1972 he was working on what he called the biggest achievement in his life,... and he's probably still working on it.

In the same way, it may be argued that knowing something can po-tentially last for a longer time (e.g. a year) than swimming can (one cannot swim continuously for a year). Although several linguists have pointed out that the use of the past tense does not necessarily mean that the situation no longer holds at t0 (cf. e.g. Comrie 1985: 24, 41-42, Declerck 1991a: 266, Mommer 1986: 74, Riddle 1978: 78, 100-101, Smith 1982: 177, 1983: 486, Woisetschläger 1976: 34-35),31 no one has ever explicitly pointed out that there is a gradient as regards the possibility of a past tense situation lasting up to t0:

a. telic bounded situations b. telic unbounded situations c. atelic situations accompanied by a bounding adver-

bial providing new information d. atelic situations accompanied by an adverbial pro-

viding old information. e. atelic unbounded situations

1.3.5. Progressive

Whereas some linguists believe that the progressive may affect the Aktionsart of a sentence, others have pointed out that this is not necessarily the case. To the latter group belongs Allen (1966), who uses the term bounded and unbounded to refer to what we call telic/atelic:

In the sentence They are playing a rubber of bridge, (...), the bounded Predication is not "whole" yet - and may never be com-pleted - (...) but the Predication play a rubber of bridge (...) may be called "bounded" (in the grammar), even though (...) the Predication (...) refer[s] to "incomplete" units (in the "practical world") (Allen 1966: 198). (Cf. also 1966: 202, 204)

In a similar way, Zydatiss (1976) classifies examples of this type (i.e. Grenzübertritte, Inchoativa, Accomplishments + Erweiterte Form) as "telisch/zielgerichtet aber nicht-terminiert" [telic/goal-oriented but not over] (Zydatiss 1976: 248) (cf. also 1976: 42-43). The use of the


progressive indeed establishes an unbounded reading in most cases. However, I do not follow Zydatiss in his claim that the progressive influences the Aktionsart of a sentence. The progressive presents a situation in a different way (i.e. as "ongoing") but it does not neces-sarily affect its nature in terms of (a)telicity. Binnick (1991) is also aware of the possible effect of a progressive verb on a telic sentence, as is clear from his definition of telicity:32

The appropriate definition of telicness involves neither the achievement of a goal, nor the potentiality of such an achieve-ment, but the inference of such a potentiality in the characteriza-tion of the situation (Binnick 1991: 192).

Binnick's definition is inspired by telic unbounded examples such as They were sailing around the world:

In the imperfective, the actualization of a culmination is not as-serted. To say that someone is sailing around the world is to say that they are engaged in an activity which potentially leads to an accomplishment (i.e. they are in the activity phase of an accom-plishment). They were sailing around the world is the imperfec-tive of a telic. A solution for the problems of telicness along such lines - in terms of phasic structure - depends of course on achieving an adequate semantic theory of phase. (Binnick 1991: 192)

None of the above people, however, make the generalization that three different types of classification can be made, i.e. one in terms of (un)boundedness, one in terms of (a)telicity, and one in terms of (im)perfectivity (i.e. (non)progressiveness). The use of the progres-sive is indeed a powerful means of turning a bounded sentence into an unbounded one:

(83) a. He walked to the station, (bounded) b. He was walking to the station, (unbounded)

(84) a. The bus stopped, (bounded) b. The bus is stopping, (unbounded)

However, there are exceptions to this general rule. The following bounded sentences have already appeared in the present discussion:


(85) a. I have been playing football. (That's why my shirt is so dirty)

b. I've been drinking beer. (That's why I can no longer walk properly) (cf. section 1.3.3.2.)

(86) I have been waiting for you since 8 o'clock, (cf. section 1.3.3.2.)

(87) A: When was John working in the garden? B: He was working in the garden this afternoon, (cf.

section 1.3.4.2.)

1.3.6. Non-indicative mood and negation

The presence of not or never may turn a bounded sentence into an unbounded one:33

(88) a. He opened the window, (bounded) b. He did not open the window, (unbounded)

(89) a. John will give in. (bounded) b. John will never give in. (unbounded)

In a similar way, sentences that are not in the indicative mood (i.e. sentences with modal auxiliaries (cf. Dowty 1986: 44), imperatives or sentences with a subjunctive form) are unbounded:

(90) a. We must not forget that she was the one who de-cided to join him on his trip to California.

b. Enter you occupation and your employer's name on the form.

c. The Lord be with us.

It should also be added that in sentences with an infinitive clause or participial clauses it is the situation in the main clause that deter-mines whether the sentence is bounded or unbounded and telic or atelic:

(91) a. He is too pigheaded to give in. b. I enjoy walking in the rain.

(92) a. They arranged for the party to be held on 1 April, b. He denied having stolen her bike.


The sentences in (91) are unbounded and atelic, those in (92) are zero-telic and zero-bounded.

Although the effect of the factors mentioned in this section is more intricate than has been suggested, the scope of the present discussion does not allow me to go more fully into these issues.

1.3.7. Interaction of the different factors

It will be clear that the factors influencing the (a)telicity and (un)boundedness of a sentence interact. The sentence in (93a) is telic and bounded. The addition of an adverbial induces a repetitive read-ing and turns the sentence into an atelic, unbounded clause (cf. (93b)). As a result of adding when he worked for IBM, the sentence becomes bounded, but it remains atelic (cf. (93c)):

(93) a. John arrived at 10 o'clock sharp. b. John arrived at 10 o'clock sharp on Mondays. c. When he worked for IBM, John arrived at 10

o'clock sharp on Mondays.

Similarly, the addition of an adverbial to (94a) coincides with a change from unbounded to bounded (cf. (94b)), which is overruled through the use of a progressive form (cf. (94c)), provided the adverbial is given information:

(94) a. He stands by the window. b. He stands by the window from 5 to 6 p.m. c. He is standing by the window from 5 to 6 p.m.

1.4. Zero-boundedness and zero-telicity

It has been pointed out before that certain sentences have charac-teristics typical of bounded/telic sentences and unbounded/atelic sentences. When we apply the tests mentioned in section 1.2. to the sentences in (95), it appears that they can be classified both as telic and atelic or as bounded and unbounded:

(95) a. The endless procession walked by the church. (Declerck 1979: 768) (zero-bounded/zero-telic)

Zero-boundedness and zero-telicity 51

b. The insect crawled through the tube, (ibid.) (zero-bounded/zero-telic)

c. John filled the tank with water, (ibid.) (zero-bounded/zero-telic)

d. Mary rejected the proposal, (zero-bounded/zero-telic)

e. He realized why she had left him. (zero-bounded/zero-telic)

Several linguists have pointed out that certain clauses allow a double interpretation.34 Declerck (1979) sets them apart in a separate class of zero-bounded sentences. The above examples are not only zero-bounded, they are also zero-telic. In the discussion which follows, the term purely telic will sometimes be used to make a clear distinc-tion between telic sentences which do not allow an atelic reading (purely telic) and those which do (zero-telic). Binnick (1991) cor-rectly observes that "any telic expression in the English past tense is ambiguous in this way, so long as it denotes a situation having an activity phase" (Binnick 1991: 190):

(96) a. John earned the gold medal for the first ten seconds of the sprint and then coasted. (Binnick 1991: 190)

b. John earned the gold medal in the first ten seconds of the sprint and then coasted, (ibid.)

c. Max wrote his speech all afternoon but finally gave up in disgust, (ibid.)

d. Max wrote his speech within ten minutes of learning the proposed topic, (ibid.)

In (96a) and (96c), it is the adverbial which induces the hearer to in-terpret the sentences as referring to atelic situations. Binnick's claim that accomplishments can all be considered as activities if we focus on the activity phase requires some further comment, because native speakers' judgments differ with respect to the possibility of non-pro-gressive sentences of the type under discussion being understood as on-going activities. Binnick (1991) writes:

In a language such as English it is possible to indicate whether an action is already completed, or is ongoing but not necessarily ever to be completed - she had read the book (when John ar-rived) versus she was reading the book (when John arrived). But this indication is optional, not obligatory, she read the book is


neutral in this regard: she read the book (in under an hour) versus she read the book (for a few minutes but soon tired of it). But note: She read the book when John arrived is not ambiguous. (Binnick 1991: 139)

However, most of the native speakers consulted are reluctant to con-sider a sentence such as She read the book as an activity. In any case, they all agree that that particular interpretation is a marked option, as a speaker conforming to the rules of communication is likely to use a progressive form if an activity reading is intended.

1.5. Left boundedness and right boundedness

So far, the concept (un)boundedness has been illustrated with exam-ples which are (un)bounded to the right and to the left. However, it is also possible for situations to be bounded to the left/right and un-bounded to the right/left in the sense that unbounded situations may get an inchoative/terminative reading, which means that they become bounded to the left/right. This inchoative/terminative reading may be brought about by adverbials (cf. the examples in (97)) or by the con-text (cf. (98)):

(97) a. Mary ran at 2.30. (Vlach 1981: 276) (bounded to the left)

b. I appreciated his presence until he told me he hated Jews, (bounded to the right)

c. Suddenly, Tom could afford to control himself, (bounded to the left). (Dry 1979: 66)

d. At 5 o'clock, he drove to the station (Schöpf 1984: 48) (bounded to the left)35

e. After a few moments, the big house loomed in front of them. (Allen 1966: 201) (bounded to the left)

(98) a. Mary entered the president's office. A copy of the budget was on the president's desk. The president's financial advisor stood beside it. The president sat regarding both admiringly. The advisor spoke. (Dowty 1986: 49) (bounded to the left) (italics mine)

b. Jameson entered the room, shut the door carefully and switched off the light. It was pitch-dark around him because the Venetian blinds were closed.


(Hinrichs 1986: 68) (bounded to the left) (italics mine)

The concept bounded to the left/right will prove important for the principle of unmarked temporal interpretation (chapter 7). The ex-amples in (98) differ from those discussed so far in that a bounded reading is inferred from the context rather than established by a par-ticular constituent.36


I have discussed in detail the (un)boundedness and (a)telicity dis-tinctions, as these are concepts which, in spite of the numerous arti-cles devoted to taxonomies of this type, were still in need of accurate definition taking into account the effect of adverbials and tense in particular. A definition has been given of the concepts (a)telicity and (un)boundedness as they will be used in the ensuing discussion: a situation is represented as bounded if it is represented as having reached a temporal boundary; otherwise it is unbounded. A situation is telic if it has a natural or intended endpoint which has to be reached for the situation (as described in the clause) to be complete (and beyond which it cannot continue). The bounding effect of temporal adverbials on (un)boundedness and (a)telicity has been considered in detail and it has been shown how tense, adverbials and given/new information interact. The existence of zero-bounded/zero-telic sentences and the fact that the perfect tense has a bounding effect on sentences will prove of major importance for the use of tense. The analysis has also shown that (un)boundedness, (a)telicity and the given/new information opposition determine the likelihood of the implicature associated with the past tense (i.e. a past tense situation no longer holds at tQ) applying.

Chapter 2 Declerck (1991a): tense in discourse (part 1)

Declerck's aim is to fill one of the "gaps" in the literature on tense:

The first gap is the absence of what could be called a 'descriptive theory' of tense. What I mean by this is a theory which functions as an intermediate level between the concrete data and the abstract, hypothetical treatments of tense in theoretical frameworks. It is my conviction that the reason why there has been a proliferation of widely divergent and often conflicting approaches to tense is the lack of such a descriptive theory. As I see it, a descriptive theory is a coherent set of generalizations, which function as ordering principles and render it possible to classify the observational data and hence to interpret them correctly. (Declerck 1991a: 1)

Declerck's book Tense in discourse: Its structure and use in English (1991a) consists of two relatively independent parts. In the first, Declerck concentrates on the use of tense in discourse; he offers a systematic and comprehensive survey of the options English allows for the use of tense. In the first two chapters, he presents his theory and the terminological apparatus required to account for the data. In chapter 3, the model is put to the test by applying it to the tense system in indirect speech and conditional sentences. In the second part of the book, the structure of the tenses is dealt with. These are more theoretical observations centering around the issue of the primitives required to describe the semantics of the separate tenses. As the main issue under investigation here is the practical question of how temporal relations are expressed in RCs, the attention given to the second part of Declerck's book will be rather limited. I will con-centrate on the descriptive theory presented in the first part of Declerck (1991a) and introduce the concepts and terminology ne-cessary for the analysis of the use of tense in RCs.

In this chapter, the model will be explained and illustrated with RRCs (cf. (b)-examples). However, this should not be taken as an indication that the theoretical premises are blindly accepted. This survey constitutes a first stage in which the possibilities generated by the tense system as it is described in Declerck (1991a) are explained.

Declerck (1991a): basic concepts 55

It has been a conscious choice to illustrate the theory with RRCs, as they allow a wider range of verb forms than NRRCs. In a second stage, when the different possibilities are considered in more detail (chapters 3 to 8), constraints will be formulated on the available options. It will become clear (a) that RRCs and NRRCs differ in the tense forms they allow and (b) that there a number of -minor-respects in which Declerck's model could be refined.

2.1. Declerck (1991a): basic concepts

2.1.1. Time-spheres

Declerck starts by pointing out that the tense system of a language makes a twofold distinction (past vs. nonpast) when locating situa-tions in time.37 In other words, that system does not treat past time, present time and future time as equally important. Therefore, Declerck makes a distinction between the past time-sphere, "an indefinite length of time which lies completely before t0 (and hence does not include t0)" (Declerck 1991a: 16), and the nonpast or pre-sent time-sphere, "an indefinite timespan including t0" (Declerck 1991a: 16). t0 is the temporal zero-point, "the ultimate 'origin' of all the temporal relations expressed in the sentence, i.e. the time to which all situations referred to in the sentence are directly or indi-rectly related, and which is not itself presented as dependent on any other (more basic) time" (Declerck 1991a: 10). t0 coincides in most cases with the time of utterance (TU).38

2.1.2. Absolute sectors, absolute tenses, temporal subordination, relative tenses

t0 divides the nonpast time-sphere into three sectors: the pre-present sector, the present sector and the future sector. The past time-sphere does not have an inherent dividing point similar to t · the past time-sphere constitutes one single sector, the past sector. The four sectors just mentioned are called absolute sectors because they "are defined in direct relation to t0" (Declerck 1991a: 19). The dotted line in the visual representation of the two time-spheres below indicates the break between the nonpast and the past time-sphere:

56 Declerck (1991a): tense in discourse (part 1)

1 past pre-present present post-present

sector sector sector sector

The limitation of the scheme above is that it suggests that there is a chronological order between situations located in the past sector and situations located in the pre-present sector: the former seem to be located at a greater distance from t0 than the latter. However, this is not always the case (cf. He was here a minute ago vs. I have been to India (said by a 35-year-old who went to India when he was 15)). There is no necessary difference between past sector situations and pre-present sector ones in terms of their "absolute" temporal distance with respect to t0. The past sector and the pre-present sector differ in that situations located in the former are represented as not having any links with t„ whereas those located in the latter do have links with t0. In order to locate a situation in the past, pre-present, present or post-present sector, English uses the past tense (cf. (1)), the present perfect (cf. (2)), the present tense (cf. (3)) and the future tense (cf. (4)) respectively:

(1) She was happy. (2) I have been here before. (3) She is happy. (4) I will do it tomorrow.

Other situations may be temporally related to a situation that has been located in a particular sector:

(5) She was happy because John was there, the problems had been solved and life would be easier from then onwards.

Was indicates that the situation in the first because-clause is simul-taneous with the main clause situation, had been indicates that the situation in the second because-clause is anterior to the main clause situation and would be indicates that the situation in the third because-clause is posterior to the main clause situation. "The time interval taken up by a situation [examples (1), (2), (3) and (4)] or by a number of situations [example (5)] which are temporally related to each other by means of special tense forms" (Declerck 1991a: 20) is called a temporal domain. The past domain in (5) thus comprises the


time of the four situations referred to. When tenses are used to establish a domain, they are called absolute tenses.39 The special verb forms used to indicate the temporal order between situations belonging to the same domain are called relative tenses.*0 Applying the terminology to (5), we can say that

a. was in the main clause is an absolute past tense which establishes a past domain,

b. was in the first because-clause is a relative past tense which indicates that its situation is simultaneous with the main clause situation,41

c. had been is a relative tense which indicates that its situation is anterior to the main clause situation,

d. would be is a relative tense which, indicates that its situation is posterior to the main clause situation.

The process whereby particular verb forms are used to express the temporal order between situations belonging to the same domain is referred to as temporal subordination.

2.1.3. Time of orientation (TO), situation-time of orientation (sit-TO, TOsit), T02

In examples (1) to (4), t0 functions as the time of orientation (TO) for the situation in the sentence, a time of orientation being "any time that functions as the origin of a temporal relation" (Declerck 1991a: 18). The notion time of orientation should be distinguished from situation-time of orientation (TOsit). "Roughly speaking, it [TOsit] is the time with which the situation is represented as simultaneous" (Declerck 1991a: 18). In John was in London on Tuesday, the speaker makes it clear that the situation (John's stay in London) took place on (a part of) Tuesday. The stretch of time during which the situation (as it is presented in the sentence) holds is the TO jt. This does not necessarily imply that John was in London on Tuesday only; he may have been there on the day(s) before or after Tuesday. The complete time of his stay is called the time of the full situation (Declerck 1991a: 18). The basic time of orientation (TOt) is "that TO in the structure of a tense from which the temporal relations expressed by the tense begin to be computed" (Declerck 1991a: 252). The "intermediate TO" in the structure of a complex tense is called


T02 . Applying the notion of time of orientation to example (5), we may say that

a. every situation is simultaneous with a TOsjt. b. t0 functions as basic TO (TO,) for the situation in the main

clause and for the situations in the subclauses. c. the situations referred to in the various because-clauses are

temporally related to the main clause situation. Therefore, the TOsjt of the main clause situation is said to bind the situations in the because-clauses. Put differently, the situ-ations in the because-clauses may be said to be bound by the main clause situation.

d. TOsit of the main clause functions as T0 2 for the past per-fect situation and the conditional tense situation.

The TOsi which establishes the domain and which acts as binding TO for tne next situation introduced into the domain is called the central TO (Declerck 1991a: 21). In (5), the main clause situation establishes the past domain and its TOsit acts as binding TO for the situations in the because-clause; it is therefore the central TO. Any temporally subordinated TOsjt which is not the central TO is called an ancillary TO (Declerck 199la: 21).

2.1.4. Shift of domain

Instead of incorporating a new situation into the existing domain, it is often also possible to relate it directly to tQ. This results in sequences such as the following:

(6) He left. He never came back. (7) He told us that he likes opera.

The situation in the second clause in (6) is not temporally subordin-ated to the situation in the first clause. If this had been the case, the speaker would have used the verb form would come, which would have indicated that the situation in the second clause is posterior to the situation in the first clause. Instead, the speaker has chosen to relate the situation in the second clause to the temporal zero-point. This procedure, whereby a number of situations are all related to t0, is called a shift of domain. Declerck explains shift of domain as follows:


In this example [(6) John had been in Venice before. He went there when he was only fourteen.] the situation-TO of the first clause is an ancillary TO anterior to the central TO. The same time (viz. the time of John visiting Venice) is referred to in the second clause, but this time the situation-TO is no longer treated as an ancillary TO: the situation-TO of he went there is the central TO of a new past domain. This shift of domain is marked by a shift of tense (from the past perfect to the preterit). It should be noted that such a shift of domain within the same time-sphere is not obligatory. Instead of (6), the speaker might also have uttered (7): (7) John had been in Venice before. He had been there

when he was only fourteen. In this sequence the situation-TO of the second clause is not represented as a new central TO but as an ancillary TO anterior to the central TO of the current domain. (Declerck 1991a: 22)

A domain may be shifted (a) within the same sector (cf. (6)) or (b) from one sector to another (cf. (7)). It should be stressed that when the domain is shifted within a sector, the tenses used do not provide any information about the chronological order between the situa-tions. As will be pointed out below, the temporal interpretation is in this case determined by the principle of unmarked temporal interpretation (cf. section 2.1.7.). When a domain is shifted from one sector to another, as in (7), the tenses do signal what the temporal order between the different situations is, but there is no temporal subordination. The terms W(orld)-simultaneous, W(orld)-anterior and W(orld)-posterior will be used to refer to "real world" relations in time; these need not be linguistically realized by means of a particular tense. The term chronological order indicates that the W-relations are not expressed by the tenses. Simultaneous, anterior and posterior refer to relations in time that are linguistically expressed by the use of tenses. If this is the case, the term temporal order will be used.42 In connection with the shift of domain to another sector, we can speak of the temporal order between the situations although there is no temporal subordination. When there is a shift within the sector, the order between the real-world situations cannot be derived from the tenses, hence the use of chronological order. As was pointed out in the introduction (cf. footnote 2), it is not always possible to use an adequate terminology to reflect the conceptual


distinction tense vs. time. It is basically in combination with the noun order that the distinction can be retained: chronological order (the order between the situations in the real world) and temporal order (the order between the situations in the real world expressed by special tenses). However, a number of expressions, e.g. temporal location (i.e. location in time), temporal interpretation of the RC situation (i.e. location in time of the RC situation by means of the verb used), temporal information (i.e. information about the location in time) do not have an idiomatic terminological counterpart with chronological: cf. ? chronological location, ? chronological interpretation of the RC situation, ? chronological information. Therefore, the term temporal does not necessarily indicate grammaticalization by means of tense when used in combination with the above-mentioned words.

2.1.5. System of domain-internal tenses

Each absolute sector has its own system of tenses to express the domain-internal relations created by the process of temporal subor-dination. For each of the options described, a RRC example will be given.

2.1.5.1. Past sector

Simultaneity is expressed by means of the past tense (cf. (8)), ante-riority is expressed by means of the past perfect (cf. (9)), posteriority is expressed by means of the conditional tense43 (cf. (10)), regardless of the location of the binding TO in the past domain:

(8) a. She said / had said / would say she felt very unhappy.

b. This was the way things actually did happen. (Scheurweghs 1961: 285)

(9) a. She said / had said / would say she had stolen the money.

b. He spoke largely of the men and the things that he had seen. (Curme 1931: 229)

(10) a. She said / had said / would say she would never (10) leave him again.


b. He picked out from the group all the doctors whose experience would be most useful in the emergency. (Mittins 1962: 94)

The past tense is used to express the domain-internal relation of simultaneity in the past sector. It can also be used as an absolute tense to establish a W-simultaneous domain. In other words, it is not possible to distinguish between the temporal subordination option and the shift of domain option on the basis of the verb form used.

2.1.5.2. Pre-present sector

Declerck (1991a) makes a distinction between the "indefinite" use of the perfect and the "continuative" use of the perfect.

a. The indefinite perfect44 is used to locate a situation at some indefinite time within a period lasting up to tQ. Once the situation has been introduced, the focus shifts from the fact that it took place to the time when it took place, which implies that the link with tp is disregarded. Declerck argues that to go on talking about the situation, the system used to express temporal relations in a past domain is resorted to:

(11) I have eaten lobster once, but I can't say I enjoyed it. (Declerck 1991a: 29)

This is an example of what Declerck calls a shift of temporal per-spective (Declerck 1991a: 24), i.e. the tense system typical of a particular sector (in this case the past sector) is used to indicate temporal relations in another sector (in this case the pre-present sector): "There is thus a shift of perspective from the pre-present sector to the past time-sphere. The situation functioning as central TO is referred to in the present perfect tense, but the other situations introduced into the domain are referred to by means of past time-sphere tenses" (Declerck 1991a: 29).45 When the domain is established by an indefinite perfect, simultaneity is expressed by means of the past tense (cf. (12)), anteriority is expressed by means of the past perfect (cf. (13)), and posteriority is expressed by means of the conditional tense (cf. (14)):

(12) a. I have seen the film, but I did not like it.


b. She has taken care of the children who were sad because their dog had died.

(13) a. I have already told you that she had not read the book.

b. He has beaten up a journalist who had just taken a picture of him. (heavy stress on has)

(14) a. I have never told you she would fail her exams. b. Have you ever written an unfavourable review of a

book that would turn out to be a best-seller?

b. When the situation referred by means of the perfect gets a con-tinuative interpretation, the domain it establishes includes tQ. "In that case there is no special system of tense forms to represent domain-internal relations. Since the domain includes tQ, all relations must start from t0, i.e. they must be "absolute relations" of the kind that establish domains in different sectors" (Declerck 1991a: 31). The present tense is used to create a W-simultaneous domain (cf. (15)), the past tense (cf. (16)) or a present perfect of the indefinite type (cf. (17)) indicates W-anteriority, the future tense is used to refer to W-posterior situations (cf. (18)):

(15) a. He has always maintained that he loves her. b. She has been working on an essay on the problems

which affect youngsters nowadays. (16) a. He has always maintained that he did not kill her.

b. She has been tearing to pieces all the documents which her lawyer wrote to support her.

(17) a. He has always maintained that he has not killed her. b. Ever since this morning she has been looking for the

pen she has lost. (18) a. He has always maintained that he will take revenge

one day. b. Ever since she got pregnant, she has been looking

for someone who will take care of her baby.

c. If the present perfect situation is repetitive, i.e. if it consists of different subsituations, there are two possibilities: the speaker may concentrate on the different subsituations, which all lie before TU, i.e. he may treat the present perfect as an indefinite perfect. Accordingly, he will use the past tense (cf. (19)), the past perfect (cf. (20)), and conditional tense (cf. (21)) to refer to a simultaneous, anterior and posterior situation respectively (cf. indefinite perfect):


(19) a. He has repeated several times that he didn 't want to marry her.

b. She has complained several times about the treat-ment she received at the chiropodist's.

(20) a. He has repeated several times that he had had no intention of marrying her.

b. On several occasions, she has explained to us the relevance of the various tests she had used with the applicants.

(21) a. He has repeated several times that he would not leave for Paris,

b. Whenever I have met him, he has tried to explain to me the approach which would solve all problems.

The fact that the situation as a whole leads up to t0 also allows a continuative perfect interpretation. In this case, the temporal rela-tions of W-simultaneity, W-anteriority, and W-posteriority are expressed by means of the present tense (cf. (22)), the indefinite present perfect (cf. (23)) or past tense (cf. (24)) and the future tense (cf. (25)) respectively:

(22) a. He has repeated several times that he doesn 't want to marry her.

b. Ever since last week, she has been complaining about the noise her neighbours make.

(23) a. He has repeated several times that he has not had any intention of marrying her.

b. On several occasions, she has explained to us the relevance of the various tests she has used with the applicants.

(24) a. He has repeated several times that he didn't have any intention of marrying her.

b. On several occasions, she has explained to us the relevance of the various tests she used with the applicants who came to see her on Monday.

(25) a. He has repeated several times that he will not leave for Paris.

b. Whenever I have met him, he has tried to explain to me the approach which will solve all problems.


2.1.5.3. Present sector

The only temporal relation that can be expressed without shifting the domain is that of simultaneity (cf. (26)). It will be clear that as soon as one locates a situation before or after a present time situation, one moves into another sector and thus shifts the domain (cf. (27) to (29)):

(26) a. I am reading while you are doing nothing. b. The second half of the book discusses some ways in

which a change (...) is being brought about. (Survey of English Usage - SEU)

(27) a. Susan is angry because she has failed her exams, b. She wants to have a discussion about some movie

she has seen. (28) a. I love the book you gave me for my birthday.

b. She is trying to get rid of the printer she bought in January 1991.

(29) a. He says that he will soon leave her. b. This is a figure which will almost certainly continue

to rise steadily in the years ahead. (SEU)

2.1.5.4. Post-present sector

Simultaneity is expressed by means of the present tense (cf. (30)), anteriority is expressed by means of the present perfect (cf. (31)) or the past tense (cf. (32)) and posteriority by means of the future tense (cf. (33)):

(30) a. b.

(31) a. b.

(32) a.

She will say she feels very unhappy. It is not impossible that some time or other a bomb will explode which is hidden in our building. (Declerck 1991a: 56) She will say she has stolen the money. And Filip will shoot a buck for food and hang it to a tree outside, so that they can go and cut out chunks of it, and one day, soon afterwards, they'll be taking up the floor to see what has died under it. (Buyssens 1968: 270) She will say she stole the money the day before.


b. One day, you will wake up and tell her the awful story you dreamed the night before.

(33) a. She will say she will never leave him again. b. There is no sort of guarantee that this method will

produce seeds which will give a good rose. (SEU)

This is another example of a shift of perspective. The time of the situation that establishes the future domain and binds the situations that are temporally related to it (i.e. the central TO) is treated as if it were t0. This explains why the system of tenses used to express the domain-internal relations in the post-present sector is the same as the system of tenses used to define the four absolute sectors. However, it must be stressed that although the tenses used to express anterior-ity/simultaneity/posteriority in the post-present sector can also be used as absolute tenses to establish domains (cf. (l)-(4)), they are not used as absolute tenses in (30) to (33). In the latter sentences they are used as relative tenses, to relate a situation to a future TO. To indicate the set of tenses used to temporally subordinate situations to a future TO, Declerck (1991a: 50) uses the label present perspective system (PPS) because in this system the future TO behaves as if it were t„. When the domain is shifted in the post-present sector, the label future perspective system (FPS) is used, i.e. the newly in-troduced situations are not related to the post-present sector pseudo-t0, but to (the "real") t0:

(34) a. There will be a party next Tuesday. John and Mary will also be there,

b. And if Madge ever marries, I know it will be to some kind of Socialist in a tweed suit, who'// insist on being married in a Register Office. (Buyssens 1968:199)

(35) a. They will give her a painting. They will have bought it at Sotheby's.

b. Knowing aunt Beatrice, she will talk endlessly about the presents she will have received for her 60th birthday.

Will have bought in (35a) is called an absolute-relative tense because apart from establishing a post-present domain, which accounts for the label absolute, it also expresses anteriority with respect to the post-present "pseudo"-t0, which accounts for the label relative. The future perfect is not an absolute tense in the sense that it locates its


situation in one of the sectors and in doing so relates it directly to t0. Rather, it relates its situation as anterior to a post-present TO and in doing so leaves vague the temporal relation between the situation in question and t0. Theoretically, the situation referred to in the second sentence of (35a) could lie before, after or coincide with t0.

2.1.6. Syntactic subordination Φ temporal subordination

It is important not to confuse syntactic subordination and temporal subordination. In a syntactically subordinated clause, not only rela-tive tenses can be used (cf. (36)), in which case there is temporal subordination, but also absolute tenses (cf. (37)), in which case there is no temporal subordination:

(36) He said he felt ill. (37) He said he feels ill.

In other words, syntactic subordination does not always coincide with temporal subordination. Nor is it the case that a sequence of unembedded sentences always coincides with a sequence of newly established domains. In (38a), the domain is shifted in the second sentence, which implies the use of an absolute tense; in (38b), there is temporal subordination, which implies the use of a relative tense:

(38) a. He left. He never came back. b. He left. He would never come back.

Similarly, in (39), will cause establishes a post-present domain and the main clause situation is temporally subordinated to it (is):

(39) In the whole field of discrimination against women, the area which will cause the most active, immediate suffering is the economic insecurity of wives.

2.1.7. The principle of unmarked temporal interpretation

As has already been pointed out, situations are not always temporally subordinated: the speaker may choose to relate the new situation to t0 (= shift of domain) instead of to the TOsit of (one of) the preceding situation(s) (= temporal subordination). In this case, the tenses do not


express what the temporal relation between the different situations is. The following conversational implicature is said to regulate the temporal interpretation of consecutive newly established domains:

Principle of unmarked temporal interpretation: In a sequence of unembedded clauses where each clause establishes its own domain (within the same absolute sec-tor) and where there is no adverbial, contextual or prag-matic indication of temporal order, the unmarked temporal interpretation is as follows: (a) when the situations are represented as bounded, they are interpreted as following each other in the order in which they are reported; (b) when the situations are represented as unbounded, they are interpreted as simultaneous with each other; (c) when one situation is represented as bounded and the other as unbounded, the bounded situation is interpreted as temporally included in the unbounded one. (Declerck 1991a: 119)

The following examples illustrate (a), (b) and (c) respectively:

(40) a. He left. He never came back. b. They were crying. They felt very unhappy. c. He entered the room. He was laughing.

The principle of unmarked temporal interpretation for embedded clauses runs as follows:

Principle of unmarked temporal interpretation (for embed-ded sentences): If both the head clause and the subclause shifting the domain are bounded, the two situations tend to be inter-preted as holding in the order in which they are reported only if the subclause is one of the clauses that carry the action forward (i.e. that have a foregrounding function). (Declerck 1991a: 138) When the two clauses are unbounded, the unmarked inter-pretation is that of simultaneity. (Declerck 1991a: 139) When one clause is unbounded while the other is bounded, the unmarked interpretation is that the bounded situation is included in the unbounded one. This is the case irrespective


of whether the unbounded clause is the head clause or the subclause and irrespective of the surface order of the clauses, (ibid.)

Examples (41a) to (41c) illustrate the different possibilities:

(41) a. John hit Bill, who hit him. (Declerck 1991a: 138) b. John carried the suitcase because Bill pushed the

cart. (Declerck 1991a: 139) c. Bill, who seemed to know everything about the

stranger, told us who he was. (ibid.)

The principle of unmarked temporal interpretation will be dealt with in more detail in chapter 7.

2.1.8. Direct and indirect binding

The notions direct binding and indirect binding will also be used in the ensuing discussion. The sentences in (42) prove that a clause is not always bound by the TOsit of its superordinate clause (SUPC). The terminology used is that of Quirk et al. (1985: 988-991): a subclause is embedded in a superordinate clause (SUPC), which may also be called the matrix clause if it is the "ultimate" SUPC. If the SUPC is embedded in another clause, the latter will be called the matrix clause. All SUPCs are main clauses:

(42) a. Bill knew that John had often said that he was unhappy.

b. Mary knew that John had left because he had felt unhappy.

(42a) is an instance of direct binding. Each subclause is temporally subordinated to the TOsit of its SUPC: the situation of the first that-clause is represented as anterior to the matrix clause situation and the situation of the second that-clause is represented as simultaneous with the situation of its SUPC, i.e. the situation of the first that-clause. This implies that the chronological order between the situation of the matrix clause and that of the second that-clause is not made clear by means of a special tense form. Example (42b) is an instance of indirect binding: not only the situation of the first that-clause, but also that of the second ί/ιαί-clause is temporally


subordinated to the matrix clause situation. In this case, the chronological order between the situations in the two that-clauses is not linguistically expressed.46 The following schemes may give a better insight into the kind of choice that is involved in the direct binding - indirect binding distinction. In each of the cases, the speaker has to choose a binding TO for the RC situation from a number of available binding TOs. At this stage, no claims are made about the choice likely to be made; this particular issue will be dealt in chapter 8:

a. The RC situation is W-anterior to the SUPC situation; the RC situation and the SUPC situation are both W-posterior to another past sector situation:

X

X SUPC sit.

X RC sit.

Instead of indicating the relationship of W-anteriority between the RC and the SUPC situation (by using the past perfect), the speaker may choose to express the relation of W-posteriority between the RC situation and the matrix clause situation (i.e. use the conditional tense):

(43) a. He said he would be helped on Tuesday by the man who would kill the inspector on Monday, (indirect binding)

b. He said he would be helped on Tuesday by the man who had killed the inspector on Monday, (direct or indirect binding)

It is necessary to point out that in (43b) the RRC situation is either bound by its SUPC situation (direct binding) or by the TO which binds its SUPC situation (indirect binding). In the latter case, the killing is represented as anterior to the TO to which the helping is posterior rather than to the helping (the SUPC situation). In this reading, the temporal relations between the situations are not longer represented by the above scheme (i.e. the RC is W-anterior to the matrix clause situation).


b. The RC situation is W-anterior to the SUPC situation; the SUPC situation and RC situation are W-anterior to the matrix clause situation:

X

X S U P C sit.

X RC sit.

Theoretically, the speaker can choose between two options: he may express the anteriority relation between the RC situation and the SUPC situation (direct binding) or he may express the anteriority relation between the RC situation and the matrix clause situation (indirect binding). However, it is not possible to distinguish between the two options on the basis of the tense used, the past perfect being used in both cases:

(44) He said the present had been given to him by a friend who had spent his holiday in Jamaica, (direct or indirect bind-ing)

c. The RC situation is W-simultaneous with the SUPC situation; the RC situation and the SUPC situation are both W-posterior to a past sector situation:

X

X s U P C sit.

X RC sit.

In (45a), the relation of W-simultaneity between the SUPC and the RC situation is not expressed by means of the tense used. Both the SUPC situation and the RC situation are bound by he said\ there is indirect binding:47

(45) a. He said he would go to someone who would be willing to explain the issue to him. (indirect binding)

b. He said he would go to someone who was willing to explain the issue to him. (direct binding)


In example (45b), the verb expresses the W-simultaneity relation between the RC situation and the SUPC situation rather than the relation of W-posteriority between the RC situation and the matrix clause situation. Was willing can also be understood to express a relation of simultaneity between the RC situation and the SUPC sit-uation, in which case it is not interchangeable with the conditional tense: wheras the latter form captures a W-relation of posteriority between the matrix clause situation and the RC situation, the former represents the RC situation as simultaneous with the matrix clause situation.

d. The RC situation is W-simultaneous with the SUPC situation; the RC situation and the SUPC situation are both W-anterior to a past sector situation:

X

SUPC sit. X

RC sit. X

The speaker has the choice of expressing the relationship of W-simultaneity between the RC situation and the SUPC situation (cf. (46b)) or that of W-anteriority between the RC situation and the matrix clause situation (cf. (46a)):

(46) a. He said he had lived with people who had collabo-rated with the Germans, (direct or indirect binding)

b. He said he had lived with people who collaborated with the Germans, (direct binding)

The past perfect in (46a) is ambiguous between a directly bound form expressing anteriority with respect to the SUPC situation and an indirectly bound form expressing anteriority with respect to the situation of saying.

e. The RC situation is W-posterior to the SUPC situation; the RC situation and the SUPC situation are both W-anterior to another past sector situation:


X

SUPCsit. X

RC sit. X

The W-posteriority relation may be linguistically expressed, in which case the conditional tense will be used (cf. (47b)), or the W-anteriority relation between the RC situation and the binding TO of the SUPC situation may be linguistically expressed, in which case the past perfect will be used (cf. (47a)):

(47) a. And what was it which Wulf had recognized in Hypatia which had bowed the old warrior before her? (Poutsma 1929,1-2: 640) (indirect binding)

b. And what was it which Wulf had recognized in Hypatia which would bow the old warrior before her? (adapted from Poutsma 1929,1-2: 640) (direct binding)

2.2. Methodology and aim

It will be clear that an 18-page summary of a 143-page theory is bound not to do it full justice. However, the concepts which have been explained are sufficient to start the investigation into the use of tense in RCs. Additional concepts will be introduced and more elaborate summaries will be given in the course of the discussion whenever this is necessary.

My reasons for having chosen the theoretical framework offered in Declerck (1991a) to analyse the tense system in RCs relate to the comprehensiveness of the model. Although many people have made useful points in connection with the use of tense in English, none of them explores the matter fully. Declerck (1991a) is the only work which considers the question how temporal relations are expressed in the four sectors; it is the only comprehensive theory available today. Declerck's approach is also exceptional in that he provides a label for the wealth of phenomena which can be observed in the field of tense and hence puts at one's disposal a terminological apparatus which contributes to a better grasp of the facts. Again, many of these

Methodology and aim 73

phenomena have been observed by other linguists in a number of articles, though Declerck (1991a) outshines them in terms of comprehensiveness. For instance, King (1983) argues at length that it is necessary to make a distinction between the tense used to refer to a situation and the actual location in time of a situation. This is what Declerck calls a shift of perspective (Declerck 1991a: 24). Huddleston (1969: 797) points out that a tense expressing simul-taneity is sometimes used to refer to situations which actually follow each other. Declerck captures this phenomenon by means of the concept sloppy simultaneity (Declerck 1991a: 41).48 Lakoff (1970) makes a number of observations in connection with the at first sight deviant use of tense in sentences of the type That thing rustling in the bushes over there will no doubt be a chipmunk. Let's wait till it comes out. Declerck incorporates sentences of this type into the system through the concept of temporal focus (Declerck 1991a: 80-89). Lo Cascio - Rohrer (1986) deal with the question: "Under what conditions can an anaphoric tense skip its preceding anaphoric tense and go right up to the dominating deictic tense?" (Lo Cascio - Rohrer 1986: 235). Declerck provides us with the labels direct and indirect binding (Declerck 1991a: 62) to discuss matters of this kind. Haegeman - Wekker (1984) point out that there are differences in syntactic behaviour between conditional clauses in which will can be used and those in which it cannot. Declerck's approach allows the generalization that certain types of future time clauses use the present perspective system (Declerck 1991a: 50) to express temporal relations, while in others the future perspective system has to be used. The principle of R-progression (cf. e.g. Partee 1984) and the notions of foreground/background, which feature in discussions of the movement of narrative time (cf. e.g. Couper-Kuhlen 1989b), are fused into the principle of unmarked temporal interpretation (Declerck 1991a: 119, 138-139). I will refrain from comparing Declerck's theory with other works at this stage,49 as this would result in terminological quibbles which would not substantially enrich the discussion; indeed, it would complicate matters even further instead of elucidating them. When exploring the use of tense in RCs, references will be given of other works in which similar points have been made.

The aim of this study is twofold: first and foremost, the analysis will reveal whether RRCs and NRRCs differ in terms of the tense options they allow. Secondly, by applying Declerck's model to a corpus of RCs, we will find out whether the predictions made by the theory are substantiated by the data at our disposal.


RRCs differ from NRRCs in the number of information units the NP in which the RC is embedded consists of (cf. Huddleston 1971: 213, Bache - Jakobsen 1980: 256): whereas a RRC forms one information unit with its antecedent, the NRRC and its antecedent constitute separate information units. This distinction manifests itself prosodically, and affects the semantic interpretation and the syntactic structure of the NP. The semantic difference is reflected in the different functions performed by RRCs and NRRCs: RRCs offer relevant, restricting information which imposes constraints on the antecedent N. The additional, relevant information offered by a NRRC, on the other hand, does not help to determine the referent of the NP in which it is embedded. On the syntactic level, the bonds between the antecedent and the RC are much tighter in the case of a RRC than in the case of a NRRC. The syntactic differences between RRCs and NRRCs have received ample attention in transformational grammar. However, as the realization that there is a syntactic difference between RRCs and NRRCs is sufficient for the ensuing discussion, I will refrain from giving a survey of possible transformational grammar approaches to the RRC/NRRC distinction. I will concentrate on adnominal, headed RCs. This means that I will not deal with RCs of the following type:

(48) a. I would like to win a lot of money in a lottery, which would save me from the efforts of having to work, (sentential RC)

b. Your present is exactly what I have always been eager to buy. (headless RC)

The reason why I exclude sentential and headless RCs is that the former are all NRRCs, while the latter are all RRCs. Hence, a comparison between RRCs and NRRCs is not possible in these types of examples.

Having given a survey of the different options in the different sectors, I will now list the possibilities that will be focussed on in the discussion to follow. My answer to the question whether or not RRCs and NRRCs differ as far as the use of tense is concerned is based on an investigation of five issues:

1. How are W-anteriority and W-posteriority expressed in (N)RRCs referring to the past sector? The problem referred to in section 2.1.5.1., i.e. that of determining the status of the past tense in RRCs and NRRCs when it refers to a W-simultaneous situation, will also be dealt with, (chapter 3)

Methodology and aim 75

2. How is W-anteriority expressed in (N)RRCs locating their situa-tions in the pre-present sector? I will concentrate on the temporal relationship of W-anteriority; a more detailed analysis of the tem-poral relations of W-simultaneity and W-posteriority has revealed that the tendencies which emerge are similar to the restrictions on expressing W-anteriority in the pre-present sector and/or temporal relations in the past sector, (chapter 4)

3. How are W-anteriority and W-simultaneity expressed in (N)RRCs referring to the post-present sector? (chapter 5)

4. What is the unmarked temporal interpretation of shift of domain (N)RRCs in which neither the tenses nor adverbials impose a tem-poral order? Do NRRCs and RRCs differ considerably in this respect? How can possible differences be explained? (chapter 7)

5. Do NRRCs and RRCs in the past perfect tense differ in their choice of binding TO? What are the restrictions on direct and indi-rect binding in NRRCs and RRCs? (chapter 8)

The analysis is based on corpus examples. Most of the examples are taken from the Survey of English Usage (henceforth SEU), the Lancaster Oslo Bergen corpus (henceforth LOB), the Brown corpus (henceforth BR) and the Wall Street Journal corpus (henceforth WSJ). I have sometimes chosen to simplify the original examples when the intricacy of the sentences needlessly complicated a clear understanding of the points made. Although the corpus at my dis-posal consists of over 10,000 RCs, there were insufficient post-pre-sent sector and pre-present sector examples. In order to overcome this problem, while at the same time not basing the analysis on familiar and stereotyped made-up examples, I have in some cases chosen to transpose corpus examples to different sectors. The pos-sible drawback of using adapted corpus examples is compensated by the fact that such examples are rich in contents. Like the "real" corpus examples, they offer a wealth of data which the non-native mind could not possibly devise itself. It should also be stressed that all the examples have been presented to native speakers with a view to obtaining objective acceptability judgments and feedback for the (differences in) meaning of particular forms; the views expressed are the result of consultation and discussion and are in no way determined by preconceptions or prejudice.

Now that the theoretical framework has been introduced, we are fully equipped with the terminological and conceptual tools needed to analyse how tense works in RCs. Declerck believes that the pos-sibility of shifting the domain is determined by the kind of subclause used: "Some subclauses (e.g. concessive clauses, relative clauses,


because-clauses, locative clauses, etc.) easily allow the shift because their semantics is such that they need not represent their situations as temporally dependent on the head clause situation. Other subclauses (e.g. conditional clauses, time clauses) allow the shift only in exceptional cases" (Declerck 1991a: 134-135). On pages 55, 58 and 63 he lists a number of aspects (relating to the use of the FPS and PPS, the use of the conditional perfect and indirect binding respectively) in which the syntactic difference between the two types of RC is reflected in the tense system used. Where Rohrer (analysing French) considers the question "Warum lassen Relativsätze eine solch grosse Freiheit in der Wahl der Tempora zu?" (1977: 46) [Why do RCs allow great freedom in the use of tense?], I will address the question whether or not RCs in English also allow great freedom in the use of tense. I will formulate restrictions and possibilities, and point out possible differences between RRCs and NRRCs in this respect. Apart from bringing to light possible differences between RRCs and NRRCs, our investigation will put us in a position to check whether the model of the tense system presented in Declerck (1991a) is borne out by the data at our disposal.

Chapter 3 The expression of temporal relations in past sector RCs

3.1. Introduction In this chapter, an analysis will be given of past sector RCs whose situations are W-anterior or W-posterior to the SUPC situation.50 In particular, an answer will be formulated to the question whether temporal subordination is compulsory in RRCs and NRRCs or whether temporally subordinated and shift of domain verb forms are mutually substitutable. McCoard (1978) points out: "There exist various means of marking the order of sequenced events: there are conjunctions (before, as, after), there are adverbials {later, not long before), there are different embedding-verb classes (intend to vs. regret) and, of course there is word order. These devices are usually sufficient, so that the past perfect can be dispensed with entirely" (1978: 186) (italics mine). McCoard is right to point out that there are means other than tense that can help to safeguard a correct temporal interpretation. Of the factors he mentions, adverbials (cf. (1)) and lexical content (cf. (2a)) apply to RCs:51

(1) James Wilkinson became apprenticed to a celebrated London gunsmith called Henry Nock, who set up his business (...) in 1772. (SEU)52

(2) a. He remembered the man he had run {ran) into when he travelled in China.53

However, example (2b) shows that although remember is often followed by a situation that is W-anterior and from this point of view provides information in itself about the temporal location of situations (i.e. the explicit location in time by means of the past perfect seems superfluous), its presence does not automatically result in the substitutability of the past tense and past perfect:

(2) b. Whenever the club held a meeting there was always someone who had been before and would remember everyone who had been (? was) there. (Needham 1976: 317)

78 The expression of temporal relations in past sector RCs

Pragmatic knowledge or knowledge of the world can also be helpful in recovering the correct chronological order:

(3) The Paraclete was to assist the disciples in recalling all the things which Jesus had said and done (said and did). (SEU)

Apart from the fact that recall is like remember in that it is usually followed by a W-anterior situation, Christians are supposed to know that Jesus died before the Paraclete assisted the disciples. It follows that no matter which tense is used, the RC situation54 will be under-stood as referring to a W-anterior situation. However, when none of the above means are available, it may still be possible to use the past tense instead of the past perfect:

(4) I could not believe what she told me: Susan had actually fallen in love with the man who reported her to the police.

Although, theoretically speaking, Susan could have fallen in love with the man before or after he reported her to the police, the native speakers consulted on this point agree that the RC situation is most likely to be interpreted as referring to a W-anterior situation.

There is indeed nothing extraordinary about the observation that a W-anteriority relation between two situations does not necessarily have to be linguistically expressed by means of the past perfect; linguists have been eager (especially in connection with a/ter-clauses and noun clauses) to point out that "tense simplification" exists.55

The term tense simplification should be understood to mean using a tense which has a less intricate structure. In this context, it means using the past tense instead of the past perfect:56

(5) a. He said he saw (had seen) it in 1960. (Berland-Delepine 1971: 138 in Salkie 1989: 24)

b. The woman that had stolen / stole the book saw John. (Dowty 1982: 43)

c. Another ex-champion had told me it took him 20 years before he (had) no longer yearned to be out there on the horses, and I'd said thanks very much. (Fenn 1987: 219)

However, as far as I know, a systematic investigation into the question of when the past tense can be used instead of the past perfect still has to be carried out.57 The two forms are usually said to

Introduction 79

be interchangeable "if there is some other sentence-internal specification to the effect that the complement describes a state of affairs prior to the embedding verb" (Costa 1972: 43); "the shift generally occurs where time relationships are not adversely affected" (Berezovsky 1980: 91).58 Generally speaking, the reasons given do not extend beyond the claim that tense substitution is possible provided the chronological order between the situations is not blurred. Although statements of this kind have a certain explanatory value, they are vague in the sense that they do not explicitly name the factors that influence the choice of verb form. Bouscaren et al. (1982) is the only article I know of in which there is a more or less systematic inquiry into the (non-)substitutability of the past perfect and the past tense. Their treatment differs from mine in that their investigation is aimed at proving that the past perfect "est utilise pour autre chose que pour marquer l'anteriorite" (Bouscaren et al. 1982: 77) [is (can be) used to express something other than anteriority]. As they see it, the effect of the past perfect is "reintroduction du proces-sus, explication, bilan, reprise d'information, retour en arriere" [re-introduction of the process, explanation, survey, resumption of information, return to the past] (Bouscaren et al. 1982: 91). It will be clear, though, that retour en arriere cannot be anything but the expression of anteriority. Brecht (1974) argues that "forms expressing endophoric tense do not occur in English either in complement or in relative clauses" (Brecht 1974: 502). Judging from his definition, the term endophoric tense is the equivalent of what we call relative tense; it is "that tense which is specified in relation to the time of another action in a higher sentence or discourse" (Brecht 1974: 495). In other words, Brecht seems to be saying that relative tenses are never used in RCs, i.e. temporal subordination does not occur in RCs. The examples with a past perfect in the RCs given so far seem to contradict Brecht's observation. However, they cannot be adduced as evidence against his claim, as Brecht considers the perfect as an aspect (Brecht 1974: 501). Moreover, on the basis of a comparison between English and Russian, Brecht rejects the existence in English of what Declerck calls a relative past tense. To Brecht, all past tense forms are exophoric because they explicitly indicate that the situation referred to lies before t0. It is only present tense forms interpreted as referring to past/future time under the influence of past/future embedding verbs that are classified as rela-tive tense forms. Such an approach implies that a past tense which refers to a W-simultaneous situation is never a relative tense. So it appears that what at first sight seemed to be a major challenge to the


claims we will be making (i.e. a RC situation may or may not be temporally subordinated to another situation) results from (a) a different concept of relative tense -although the definition is similar to that of Declerck (1991a) (cf. Brecht's definition given above)- and (b) the fact that the perfect is considered to be an aspect.

In the case of W-anterior situations, the RC situation may lie completely before the SUPC situation (cf. (6a)) or it may lead up to it (cf.(6b)):59

(6) a. The Govt, are prone to spring decisions on dele-gations: they announced the 70 m.p.h. limit to a delegation of chief constables who had come to the Home Office to discuss the breathalyser. (SEU)

b. We talked of London, which Bernard had never visited, and of Paris, where he had studied for some time in a sculptor's atelier. (Isherwood, Goodbye to Berlin)

The same can be said of W-posterior RC situations: they may lie completely after the SUPC situation (cf. (7a)) or begin from the time indicated by the SUPC situation onwards (cf. (7b)):

(7) a. Young Margarita Cansino is seen being brought to Hollywood by 40-year-old Ed Judson ... the used-car dealer who would become her first husband. (O'Connor 1983 in Tottie - Övergaard 1984: 162)

b. Julian hit John, who from that moment onwards would never talk to Julian again.

It is the type of anteriority/posteriority relation illustrated in (6a) and (7a) which will be discussed in this chapter. I will try to find out if the chronological order between the situations is made clear by means of special tenses (i.e. there is temporal subordination) or not (i.e. the temporal domain is shifted). I will only consider RCs with an indefinite past perfect in the discussion (in more accurate terms, a past perfect which gets an indefinite interpretation). Replacing a continuative past perfect (i.e. a past perfect which gets a continuative interpretation) with a past tense in RCs obviously affects the interpretation. As Declerck (1991b) points out, "the past tense cannot be used [in the place of the past perfect] if (...) continuative meaning is to be expressed" (Declerck 1991b: 119):

Introduction 81

(8) For domestic consumption the agenda was demonstrably produced by a Reagan Administration and not by a Mondale opposition which had been canvassing ('was canvassing') the maintenance of contacts with the Kremlin in every state primary, (adapted from SEU)60

(9) He complained about his secretary, who had made (* made) one mistake after the other ever since she started working in his department.

RCs in which there is a shift to another sector, as in (10) and (11), will not be dealt with, because the tenses used in RCs of this type automatically reveal the temporal relation between the SUPC and the RC situation:

(10) His call for action brought a swift response from South Birmingham women councillors who chair the appropriate committees. (SEU)

(11) Signor Leone, who formed his caretaker Government when the Socialists, after the May elections, refused to enter into a new centre-Left coalition, is now faced by a revolt not only among the Left-wing of his own party but also by an anti-coalition movement by Left-wing Socialists. (SEU)

The reason for the exclusive consideration of sentences with a sequence of events is the following: if there is a relationship of W-simultaneity between the SUPC and the RC situation and if the RC situation is temporally subordinated to the SUPC situation, the past tense is the relative tense used to express this relationship, as in (12a).61 If there is a relationship of W-simultaneity between the SUPC and the RC situation and if the latter is not temporally subordinated to the former, the past tense will also be used, as in (12b):

(12) a. He was afraid for his daughter who was in Russia at that time.

b. He was afraid for his daughter, who was in Russia at that time.

This implies that in the case of W-simultaneity, the tenses as such do not signal whether or not there is temporal subordination, as the past tense is used in both cases. On the other hand, the tense expressing a domain-internal relationship of anteriority (i.e. the past perfect (cf.


(13a)) and posteriority (i.e. the conditional tense (cf. (14a)) differs from that used when there is no temporal subordination (i.e. the past tense (cf. (13b) and (14b)):62

(13) a. The Govt, are prone to spring decisions on dele-gations: they announced the 70 m.p.h. limit to a delegation of chief constables who had come to the Home Office to discuss the breathalyser. (SEU)

b. The Govt, are prone to spring decisions on dele-gations: they announced the 70 m.p.h. limit to a delegation of chief constables who came to the Home Office to discuss the breathalyser.

(14) a. He blamed a knee injury in [2222] for this although he did not have any pain until [2222] for which he would later be referred to St. [Name]'s Hospital and given a course of physiotherapy,

b. He blamed a knee injury in [2222] for this although he did not have any pain until [2222] for which he was later referred to St. [Name]'s Hospital and given a course of physiotherapy. (SEU)63

In other words, when the RC situation is W-anterior or W-posterior to the SUPC situation, the choice of tenses as such enables us to formulate conclusions as to whether or not there is temporal subordination. We do not need to call in the help of tests that distinguish relative past tenses from absolute pasts (cf. section 3.3.), which are definitely needed if we concentrate on examples in which there is a relationship of W-simultaneity between the SUPC and the RC situation. The basic question addressed in this section is thus the following: does the existence of a relationship of W-anteriority/W-posteriority between the SUPC situation and the RC situation necessitate the use of the past perfect/the conditional tense in RRCs and in NRRCs or not?

Before examining the principle which dictates the (non-) substitutability of the past tense and the past perfect, a few remarks need be made about the terminology used. The term superordinate clause (SUPC) is used instead of main clause to refer to the clause which dominates the RC, because a RC may be embedded in a clause which is not the ultimate main clause (i.e. matrix clause). In this way, different labels can be used to distinguish between the RC, the clause by which it is dominated (SUPC) and the clause which dominates the latter (matrix) (cf. Quirk et al. 1985: 987-991).

Introduction 83

It is important to point out that throughout the discussion, the term mutually substitutable will be used to indicate, for example, that both the past and the past perfect can be used to represent the RC situation as W-anterior to the SUPC situation. The term does not mean that there is no difference in meaning at all between the two forms in a particular context. The label not mutually substitutable - as opposed to ungrammatical, which signals that a form violates a rule of English grammar - indicates that one verb form cannot be used in place of another without suggesting a different temporal inter-pretation. The following examples illustrate the terms just introduced:

(15) * I have seen her yesterday. (16) She was talking to the man who had won / "won" the prize. (17) She was talking to the man who had lived / 'lived with

Marguerite Duras.

The present perfect is ungrammatical in (15), the past perfect and the past tense are mutually substitutable in (16), the past perfect and past tense are not mutually substitutable in (17).

The following typographical conventions will be used: 'verb' indicates that the verb cannot be used instead of the original verb form because it either changes the temporal location of the situation or the chronological order between the RC and the SUPC situation (i.e. the verb forms are not mutually substitutable). "Verb" indicates that the verb can be used instead of the original verb form; there are differences in contextual effects between the verb forms which do not relate to the temporal location of the situation or the chronological order between the RC and the SUPC situation. * Verb indicates that the verb form is ungrammatical. ?? Verb means that it is highly questionable whether the form is acceptable; ? verb indicates that the acceptability of the verb form is questionable.

The concepts contextual effect, processing effort and mutually manifest are from Sperber - Wilson (1986). In Relevance, they argue that Grice's maxims of conversation can be reduced to a single principle, the principle of Relevance: "Every act of ostensive communication communicates the presumption of its own optimal relevance" (Sperber - Wilson 1986: 158). Simply stated, the principle of Relevance means: "I have information to communicate which is relevant enough to be worth your attention". In saying something, the speaker makes it clear that what he has said is relevant enough to be processed; he guarantees that the effort needed to process the


stimulus will be offset by the contextual effects the processing will give rise to. It is not possible to measure relevance in absolute terms; it is a comparative notion that is a function of (a) processing effort and (b) contextual effects. In other words, an utterance is relevant to the extent that the effort required to process it optimally is small (= (a)); it is also relevant to the extent that the contextual effects achieved when it is optimally processed are large (= (b)). I will not go any further into Relevance theory as the above explanation suffices to clarify the concepts contextual effect and processing effort. Applied to the issue under discussion, we will have to try and find out whether the temporally subordinated form requires more processing effort than the shift of domain tense form and whether or not the possible extra processing effort is offset by particular contextual effects which the other form does not have. When the chronological order between situations is clear to both speaker and hearer, it is said to be mutually manifest (Sperber - Wilson 1986: 42). In this case, the use of tenses conveying temporal information will be less vital.

3.2. (W-)anteriority in past sector (N)RRCs

3.2.1. Constraints on the Replacement of the Past Perfect (CRPP)

The past tense and past perfect are not always mutually substitutable, i.e. they cannot always both be used to express a relationship of W-anteriority between two situations. To formulate the underlying principle, it is necessary to reconsider the effect brought about by the perfect tenses. The use of a perfect tense in an unbounded atelic past tense sentence has a bounding effect (cf. chapter 1, section 1.3.3.2.). From this observation it follows that changing a past perfect into a past tense will also affect the boundedness of a sentence: the use of the past tense instead of the past perfect has the effect of changing a bounded atelic situation (which is not accompanied by an adverbial with a bounding effect) into an unbounded one. As a result of this shift, the RC situation is no longer felt to be W-anterior to the SUPC situation, but simultaneous with it:

(18) a. It was with an insight into the inspired Scriptures of the Old Testament that He accepted the sacrificial

(W)-anteriority in past sector (N)RRCs 85

vocation of a servant and suffering Messiah which Israel as a whole had rejected. (SEU)

b. It was with an insight into the inspired Scriptures of the Old Testament that He accepted the sacrificial vocation of a servant and suffering Messiah which Israel as a whole 'rejected.

The past tense in (18b) suggests that Israel is in a state of rejecting the sacrificial vocation of a servant and suffering Messiah. There is consequently no longer a relationship of W-anteriority between the RC and the SUPC situation. Example (19) is similar:

(19) a. The Vice-Chancellor, (...) informed all those to whom the Memorandum had been circulated of the general approval which the Senate had given to it and invited all Teachers of the University to submit comments. (SEU)

b. The Vice-Chancellor, (...) informed all those to whom the Memorandum had been circulated of the general approval which the Senate 1gave' to it and invited all Teachers of the University to submit comments.

The past perfect refers to the event of giving approval. Although this implies that there is a state of approval with respect to the Memorandum, this is not the primary meaning conveyed by the past perfect sentence. The unmarked reading of the past tense sentence is that of a state which is simultaneous with the SUPC situation. The same principle is again at work in the following example:

(20) a. He suddenly knew envy for the easy happiness with which she had returned to life, while his own return was so lonely and uneasy. (SEU)

b. He suddenly knew envy for the easy happiness with which she 'returned to life, while his own return was so lonely and uneasy.

When the past tense is used, the process of returning to life is represented as going on at the time of the SUPC situation. The past perfect, on the contrary, suggests that the process is completely over: "she is back in life". Example (21) is from Bouscaren et al. (1982):


(21) a. By the late Summer, the news of what had hap-pened on Animal Farm had spread across half the country. Every day Snowball... (ä cet instant du recit, nous savons que les animaux ont pris seuls, le contröle de la ferme; et c'est de cette nouvelle qu'il s'agit) [At this moment in the story, we know that the animals have taken control of the farm; this is the news that is referred to.]. (- Animal Farm - op. cit. -) (Bouscaren et al. 1982: 85)

b. By the late Summer, the news of what 'happened on Animal Farm had spread across half the country. Every day Snowball...

They correctly comment: "By the late summer the news of what happened (PAST) serait compris comme "ce qui se passait ä ce moment-la"" ['By the late summer the news of what happened (PAST) would be understood as "what was going on at that very moment""] (Bouscaren et al. 1982: 85). The following example is also from Bouscaren et al.'s (1982) discussion of the past perfect:

(22) a. I followed behind my mother, holding tightly to my brother's hand. He stopped at the fallen tree and bent to inspect a colony of ants - that had made a home in the earth beneath the torn roots. (- In the be-ginning - op. cit. -) (Bouscaren et al. 1982: 87-88)

b. I followed behind my mother, holding tightly to my brother's hand. He stopped at the fallen tree and bent to inspect a colony of ants - that'made' a home in the earth beneath the torn roots. (- In the beginning -op. cit. -)

Bouscaren et al. (1982: 88) point out that the expression of simul-taneity requires the use of the past progressive. One would indeed probably be more inclined to use a past progressive form if one wanted to represent the RC situation as a past on-going situation. Ladusaw (1977) makes a similar remark when he points out that Mary saw the unicorn that was walking only allows a temporal interpretation of simultaneity, whereas in Mary saw the unicorn that walked "the event of walking may have taken place either before or after Mary saw this particular individual unicorn" (Ladusaw 1977: 96). Note that in the latter sentence the unmarked interpretation is also one of simultaneity. However, if this is the information the


speaker wants to convey unambiguously, he will choose to use a progressive form. It follows from this observation that a progressive form is the unmarked option to indicate a relationship of simultaneity between the SUPC situation and the RC situation in all the examples in which there is reference to an activity or in which there is a situation with an activity phase (cf. examples in (20) to (23)). However, this is not to say that non-progressive past tense form is unacceptable in examples of this type on the simultaneity reading. In any case, replacing the past perfect by the non-progressive past tense may change the interpretation of the sentence. Example (23) illus-trates the same tendency:

(23) a. One day, all those who had taken part in the plot were going to be brought to justice. (Needham 1976: 317)

b. One day, all those who 'took' part in the plot were going to be brought to justice.

The past tense suggests reference to a group of people who are participating in a plot at the time indicated by one day. The past perfect refers to people who have taken part in a plot before the TO indicated by one day. Example (24) is similar:

(24) a. ... the spider web of French interests that overlay all Western Europe and that had been so well and closely spun. (Olofsson 1981: 63)

b. ... the spider web of French interests that overlay all Western Europe and that 'was' so well and closely 'spun'.

This example involves a statal passive (Quirk et al. 1985: 170): had been spun refers to the activity of spinning, which results in the state of being spun (was spun). In other words, the meaning of the verb may change from activity (past perfect) to state (past tense). Was spun could theoretically also be understood as an activity (i.e. was being spun), in which case it would represent the activity of spinning as simultaneous with rather than W-anterior to overlay. However, in this context, this reading will not arise. There is a similar shift in (25), (26), (27) and (28):

(25) "You've seen these Abrey people?"


"James has - he did the report," said Gordon, who had insisted ('insisted) on taking the fold-down seat and was facing us both. (Fenn 1987: 177)

(26) Liberty was a word, like sadness in books of poetry, or the language of romance in the novels which her mother had read ('read) in the late afternoons (...) Perhaps the novels had told her of unexperienced feelings. (Thomas Flanagan, The year of the French) (Salkie 1989: 28)

(27) Now that in her blindness she was perpetually alone she lost all the inhibitions which had confined ('confined) the poetry in her soul (...). (SEU)

(28) The [Name] authorities do not supply confidential reports on expatriate lecturers, but our [Word Word] has commented that Mr [Name] was a very useful and sensible man, whose wide experience in the field of English language teaching had stood ('stood) him in a good stead at [Name]. (SEU)

The past perfect refers to a period of time that lies completely before the SUPC situation; the past tense represents the RC situation as an unbounded state that is simultaneous with the SUPC situation. In accordance with the general principle emerging from the examples given so far, the use of the past tense instead of the past perfect sometimes suggests that a repetitive situation is unbounded rather than bounded:

(29) a. Scientists from the HSE were this week extending their investigations under-ground and were planning to drain a water tunnel from the River Lüne to the Abbeystead plant. They located the source of methane gas which had caused the explosions, (adapted from SEU)

b. Scientists from the HSE were this week extending their investigations under-ground and were planning to drain a water tunnel from the River Lüne to the Abbeystead plant. They located the source of methane gas which 'caused the explosions.

The unmarked interpretation of the past tense in (29b) is that there were several explosions at different moments in time. The period referred to is interpreted as simultaneous with the SUPC situation. The past perfect suggests either that there were several explosions at


one specific moment in time or that there were explosions at different moments in time during a period that lies completely before the SUPC situation. When the past tense is interpreted as a shift of domain, these two interpretations are also possible. In other words, this example again demonstrates the same tendency: unlike the past perfect, the past tense refers to a repetitive situation which is simultaneous with the SUPC situation rather than to one particular event or a series of events (i.e. a state) which lies completely before the SUPC situation. Notice that when we replace explosions by explosion, no confusion arises:

(30) a. Scientists from the HSE were this week extending their investigations under-ground and were planning to drain a water tunnel from the River Lüne to the Abbeystead plant. They located the source of methane gas which had caused the explosion, (adapted from SEU)

b. Scientists from the HSE were this week extending their investigations under-ground and were planning to drain a water tunnel from the River Lüne to the Abbeystead plant. They located the source of methane gas which "caused" the explosion.

The past perfect will preferably be used in this context because the explicit location in time of the restricting information is favoured in this sentence (cf. chapter 6, section 6.3.). In any case, it seems that the grammatical feature of the number of the NP, which may induce a repetitive reading, plays a role as far as the use of tense is concerned.

As will be clear from the above examples, replacing a past perfect by a past tense verb coincides with a change from a bounded to an unbounded sentence in the following cases:

I. clauses which refer to zero-telic situations that are bounded through the use of an indefinite perfect: e.g. (18) reject a vocation, (19) give one's approval to something, (20) return to life, (21) happen, (22) make a home

II. clauses which refer to atelic situations that are bounded through the use of an indefinite perfect: a. activities: (23) take part, (24) be spun, (25) insist,

(26)read


b. states: (27) confine, (28) stand, (29) cause explo-sions

In other words, Costa's claim (in connection with noun clauses) that in complements containing a punctual past tense verb "have may be simply omitted" (Costa 1972: 43), is not quite accurate.64 However, she is right when she claims that ambiguity can occur if the complement is durative, i.e. "if it contains an inherently durative verb, progressive aspect or a generic, such that it describes a state of affairs that can be reported while it continues to be true" (Costa 1972: 43). The first generalization to be made is the following:

(A) Constraint on the replacement of the past perfect (CRPP) (1st part) Unless the pragmatics of the situation or adverbials reveal the temporal relations clearly, the replacement of a past perfect that refers to a bounded atelic situation by a past tense will have the effect of representing the situation as unbounded provided the sentence does not contain an adverbial with a bounding effect. As a consequence, the unbounded past tense RC state/activity will normally be interpreted as being simultaneous with the SUPC situation. The past tense is therefore not substitutable for the past perfect.

The sentence should not contain a bounding adverbial because such sentences remain bounded no matter whether the past perfect or the past tense is used:

(31) a. He had lived in London for five years, (bounded) b. He lived in London for five years, (bounded) c. He had lived in London, (bounded) d. He lived in London, (unbounded)

In other words, the constraint formulated applies to sentences in which the perfect tense establishes the bounded reading. It needs to be added that bounded clauses with zero-telic verbs allow an unbounded reading when the past perfect is replaced by the past tense (cf. examples (18)-(22)), while purely telic situations referred to by means of a sentence in the past tense do not result in a different chronology (cf. examples (30), (32), (33)):


(32) a. She was angry with the man who had killed her cat. b. She was angry with the man who "killed" her cat.

(33) a. They were admiring the vase he had bought at Debenhams.

b. They were admiring the vase he "bought" at Debenhams.

The generalization that can be made in this respect can be formulated as follows:

(B) Constraint on the replacement of the past perfect (CRPP) (2nd part) Zero-telic verbs, i.e. telic verbs that lexically imply the possibility of referring to a resultant atelic state or activity which is unbounded usually result in a different temporal interpretation when used in the past tense without any other temporal specification. The past perfect sentence is interpreted as a bounded telic situation which is anterior to the SUPC situation; the past tense situation is interpreted as an unbounded atelic situation which expresses simultaneity with the SUPC situation.65

It is still necessary to somewhat modify the claim that CRPP does not apply if the sentence contains an adverbial with a bounding effect. The examples in (34) and (35) show indeed that the past perfect and past tense are substitutable if there is a temporal adverbial that imposes temporal boundaries on the situation, as changing the past perfect into a past tense does not have the effect of making the situation unbounded. In the examples in (34) and (35), the situations are bounded through a durational adverbial (Quirk et al 1985: 540); those in (34) are activities, those in (35) are states:

(34) a. He talked about the people he had worked ("worked") with from 1953 till 1957.

b. In its ranks were men steeled in the struggle, who for many years had worked (" worked") without stint to the best of their ability and knowledge for the working people. (LOB)

(35) a. He talked about the situation which had existed ("existed") from 1953 till 1957.

b. He knew that certain girls who had been ("were") in her employ during the war and had chatted too


much to the Japanese had simply disappeared. (LOB)

As the examples in (36) also show, the past tense can also be used instead of the past perfect in a state sentence with a time-position non-durational adverbial (Quirk et al. 1985: 530) that is anchored (Smith 1978: 45-46) to the main clause situation. Although strictly speaking, the adverbial does not have a bounding effect, its presence safeguards a correct temporal interpretation:

(36) a. When a member of man's family died, the body, which only an hour before had been ("was") warm, talked and breathed, suddenly was inert and cold. (LOB)

b. He knew the significance of night-shift workers pouring into factories where, only months previ-ously, there had been ("was") hardly enough work for the day shifts. (Dubos 1985/86: 149)

c. There were three bedrooms, Bertie's, the one where she slept and which had formerly belonged ("belonged") to Uncle Greg, and a tiny spare room, kept sacred for the infrequent visitor. (LOB)

d. Miss Moberly was, in fact, the seventh child of a bishop of Salisbury who had previously been ("was") headmaster of Winchester. (LOB)

The acceptability of the past tense in (36a) and (36b) may have to do with the fact that the temporal adverbial precedes the verb. It should still be added that the informants consulted agree that the past perfect is the unmarked verb form in examples of this type. This may be related to the tendency for verbs to copy the structure of the adverbial: the adverbial is anchored to a past time with respect to which it is anterior; accordingly, the verb form used is likely to be one which has the same structure, i.e. one expressing anteriority with respect to a past TO. In the examples in (37), the time adverbial is a time-position adverbial not anchored to the main clause situation:

(37) a. She was looking a bit scornful (as she always said herself), not so friendly as the other girl, Mrs. Waverton, who had seemed (? seemed) so cool and superior at first. But then Gladys was probably not


judging the case but merely happened to be out of patience with him. (Buyssens 1968: 57)

b. For there was his garage hand, Dai Pugh, taking a tearful farewell of Morfydd Uwen. Morfydd, who had until recently been (? was) Ifor's typist, but now was returning to her home as a fully-fledged schoolteacher. (LOB)

c. She later learnt that the women were of 'easy virtue', owing to which social stigma she could not even claim acquaintance with certain of their male companions who had once been (were) her suitors. (LOB)

d. When placidity was restored, she started to look for the book she had had (had) in her hand before the telephone rang. (Nabokov, Transparent things)

The effect of the past tense in (37a) is that the RC situation is understood as having held several times during the time indicated, i.e. there were several occasions on which she behaved in a superior way. In (37c) and (37d), but especially in (37b), in which the RC precedes the SUPC, the use of the past perfect appears to be pre-ferred because it reinforces the reference to a W-anterior time established by the adverbial.

Although the examples in (34) to (37) show that the presence of adverbials may safeguard a correct temporal interpretation, the following examples show the CRPP is not always overruled as soon as the RRC contains an adverbial that specifies the temporal location of the situation. In these examples, there is reference to an activity accompanied by a non-durational adverbial indicating the time of the activity. The past tense is not always fully acceptable, no matter whether the adverbial is unanchored (cf. examples in (38)) or anchored to some past time (cf. examples in (39)), the reason being that the presence of the adverbial does not counteract the change in temporal interpretation (from W-anteriority to simultaneity) brought about by the past tense:

(38) a. For so long, there had been only the tins of green water, stinking of chlorine, to drink, and the grey scum in which someone else had washed (? washed) in the morning and the foul water at the bottom of shell holes, before the sun of June and July had come to dry them out. (SEU)


It is difficult to replace the past perfect by a past tense in (38a) although there is a time adverbial in the RC. The past tense rep-resents the situation as an atelic unbounded habit whereas the past perfect either represents the situation as an atelic bounded habit or as a single event. The past perfect is not interchangeable with the past tense on the single event reading. Moreover, the idea of result is strongly present: it is because someone has washed in the water before him that there is scum on it. This factor probably also contributes to the preference for the past perfect, a tense which is apt to express resultative implicatures.66 Example (38b) is similar: although an adverbial locates the RC situation before the SUPC situation, the past perfect will preferably be used. There is a clash between the simultaneous reading suggested by the past tense and the W-anterior reading suggested by the adverbial which follows:

(38) b. She was a plain, motherly kind of woman, who had worked (? worked) hard in her youth, and now thought herself entitled to the occasional holiday of a tea-visit. (Emma, p.9 in Dry 1979: 61)

Still, this is a tendency rather than a strict rule resulting in plain unacceptability. Native speakers admit that the judgments are delicate. The following example proves that the past tense and past perfect may be equally acceptable in contexts of the type under discussion, although the native speakers agree that the past perfect "sounds better":

(38) c. And suddenly, I remembered the story about her that had eluded {eluded) me that night at Daisy's... (The Great Gatsby, p. 19 in Dry 1979: 63)

The following examples contain an anchored adverbial in combi-nation with an activity:

(39) a. It occurred to me now that I had seen (? saw) her, or a picture of her, somewhere before. (The Great Gatsby, p. 19 in Dry 1979: 63)

b. On Tuesday, he discussed the match they had played (? played) the week before.

c. He approached the bridge with some caution, not entirely convinced of its safety, even though


engineers had tested (? tested) it thoroughly only two days before. (Diver 1963: 150)

d. He talked about the people he had worked (? worked)61 with only two days before.

e. ... watching a Wimpey scarcely a wing-tip away disintegrate into rather gaudy green and orange flames, knowing that the men inside, with whom one had been drinking (? was drinking) a few hours ago, were being fried in their own fat like bacon. (Defroment 1973: 83)

The temporal adverbials indicate the chronological order between the situations. Still, the past tense is not fully acceptable in such contexts. Apart from the fact that there is a clash between the chronological relations suggested by the adverbial (anteriority) and the RC verb (simultaneity), the tendency for verbs to copy the time relationship indicated by adverbials68 may also contribute to the low acceptability of the past tense.

Another striking point is that even in examples with telic situations, the past perfect is not always mutually substitutable with the past tense: if the context allows it, the effect of the past tense in combination with an adverbial will be that the situation is no longer understood as a single event, but rather as an atelic unbounded series of events:

(40) a. One of the first things he did in Italy was to meet the King- the first sovereign he had ever met. He found the King of Italy in a drawing room very much like that in which he had met ('met') General Joffre a few days before. (LOB)

b. Near at hand was Jimmy's small rowing boat to which he had recently attached (attached) an outboard motor. (LOB)

c. We seemed to be interested only in things that could not be bought or sold, of no interest to the outside world. We argued about latitudes, or about an event that had happened ('happened) seven hundred years earlier. (Ondaatje, The English patient)

A clause with the verb happen can be zero-telic (cf. (21)). However, in (40c), in which it combines with a singular subject NP, it is telic.


The effect of using a past tense is that the sentence may be understood as referring to the repetition of the event in question.

The examples just given show that the CRPP is a strong principle; it may be said to come first in the hierarchy of factors that influence the (non-)substitutability of the past and past perfect: the examples with temporal adverbials in (38) to (40) show the use of the past tense is questionable even though there is another factor providing information about the chronological order between the different situations. Even in sentences in which pragmatic knowledge and lexical verbs suggest a chronological order, it is not always possible to replace the past perfect by a past tense, although, strictly speaking, a verb form is not necessary to make the chronological order explicit. The example in (41) indicates that the CRPP may overrule the presence of a lexical verb which conveys information about the chronological order between the situations:

(41) Whenever the club held a meeting there was always someone who had been before and would remember everyone who had been (? was) there. (Needham 1976: 317)

This type of construction with remember {remember someone) suggests that the RC refers to a W-anterior situation and from that point of view it is not necessary to use a verb form expressing anteriority. However, the past perfect is likely to be used because the use of a past tense results in a clash between the chronological relation of W-anteriority suggested by remember and the simultaneity interpretation suggested by the past tense.

Pushdown relative clauses (Quirk et al. 1985: 1298-1300) which are introduced by a verb of propositional attitude (i.e. a verb of saying, belief or feeling) constitute another piece of evidence for the claim that the CRPP is a strong principle. In pushdown relative clauses, the relative pronoun occurs in the matrix clause whereas the relative clause has been "pushed down" to a lower syntactic level: matrix - relative pronoun - SUPC - RC. Clauses of this type embedded in a clause with a verb of propositional attitude are often temporally subordinated:69·70

(42) a. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica attacks which she said had come on fairly suddenly, (adapted from SEU)


b. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica attacks which she said ? came on fairly suddenly.

However, we may ask ourselves whether it is the presence of a verb of the type just mentioned which leads to the requirement of temporal subordination. The past tense cannot be used in (43), (44) and (45) because, unlike the past perfect, it is understood as an unbounded situation which is simultaneous with the SUPC situa-tion.71 In other words, it is rather the CRPP principle which determines whether it is possible to shift the domain:

(43) a. There were those who, as followers of Mendel, believed that evolution was gradual (...) and those, who, impressed by the abrupt mutations which de Vries thought he had discovered in the evening primrose, were convinced that evolution was jerky. (SEU) (zero-telic)

b. There were those who, as followers of Mendel, believed that evolution was gradual (...) and those, who, impressed by the abrupt mutations which de Vries thought he 'discovered in the evening primrose, were convinced that evolution was jerky.

(44) a. I vividly remember a meal-time conversation with the Pastor of an Eastern European Church who told what a great surprise it had been to him to discover that Churches in one of the Western European countries had any interest or played an active part in relation to the social problems of the community. (LOB) (state)

b. I vividly remember a meal-time conversation with the Pastor of an Eastern European Church who told what a great surprise it 'was' to him to discover that Churches in one of the Western European countries had any interest or played an active part in relation to the social problems of the community.

(45) a. They rejected the proposal which they thought Jonathan Mahy had defended, (activity)

b. They rejected the proposal which they thought Jonathan Mahy 'defended.


This implies that the CRPP principle ranks higher in the hierarchy of factors determining the use of tense than the incorporation into a domain established by a verb of propositional attitude.

The examples in (37) to (45) demonstrate that the CRPP is a strong principle; when it is not observed, the past tense form may retain a touch of unacceptability or bring about a clash between the interpretation it suggests and that suggested by another element in the sentence which conveys information about the chronological order between the situations.

Digression: intensional domains

Clauses introduced by verbs which create another possible world have been called intensional contexts.72 Declerck remarks: "It appears that an intensional context functions as a temporal domain: any clause that is to be interpreted as belonging to the 'intensional domain' must be incorporated into the temporal domain containing the clause establishing the opaque context" (Declerck 1991a: 105).

There seems to be a gradient as regards the likelihood with which a verb of propositional attitude leads to temporal subordination in a pushdown relative clause:

(i) Pushdown relative clauses introduced by a verb of saying or belief are temporally subordinated unless a time adverbial locates the situation in time:

(46) a. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which she said had come on fairly suddenly.

b. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which she said ? came on fairly suddenly.

c. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which she said had come on fairly suddenly after a fall eight weeks previously. (SEU)

d. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which she said came on fairly suddenly after a fall eight weeks previously.


(47) a. I was slight of build, always short for my age, and often ill from the injury our family doctor said had not been done to me when my mother tripped and fell with me in her arms. The injury that had not been done to me by that fall is called a deviated Septum ... (In the Beginning - opus cit. -) (Bouscaren et al. 1982: 95)

b. I was slight of build, always short for my age, and often ill from the injury our family doctor said was not done to me when my mother tripped and fell with me in her arms. The injury that was not done to me by that fall is called a deviated Septum ... (In the Beginning - opus cit. -)

(48) a. He was talking about the vase which he believed Mary had broken.

b. He was talking about the vase which he believed Mary ? broke.

c. He was talking about the vase which he believed Mary had broken when stumbling over the cat.

d. He was talking about the vase which he believed Mary broke when stumbling over the cat.

(ii) Pushdown relative clauses introduced by a verb expressing doubt or conjecture are mostly temporally subordinated even if a time adverbial locates the situation in time:

(49) a. He carefully studied the gun, which he doubted had been bought in an ordinary weapon shop.

b. He carefully studied the gun, which he doubted ?? was bought in an ordinary weapon shop.

c. He carefully studied the gun, which he doubted had been bought in an ordinary weapon shop two weeks before the crime.

d. He carefully studied the gun, which he doubted ?? was bought in an ordinary weapon shop two weeks before the crime.

(iii) There is always temporal subordination in Represented Speech and Thought (RST):73 the narrator assumes the point of view of the characters and describes what is passing through their minds and how they apprehend reality without explicitly indicating that he is present. Although the introductory verb has been erased, we are


clearly confronted with the ideas of one of the characters. This technique has been called "style indirect libre". It is the use of the past perfect which signals that the narrator is still present.74 Example (50) is taken from Ehrlich:

(50) a. And it was still going on, Mrs Ramsay mused, gliding like a ghost among the chairs and tables of that drawing room on the bank of the Thames where she had been so very, very cold 20 years ago. (To the Lighthouse, p. 101) (Ehrlich 1990: 65)

b. And it was still going on, Mrs Ramsay mused, gliding like a ghost among the chairs and tables of that drawing room on the bank of the Thames where she "was" so very, very cold 20 years ago.

Had been indicates that it is Mrs Ramsay's thoughts the hearer is reading. When was is used it is the implied author who is making a comment. Example (51) is also from Ehrlich:

(51) a. All this bother about a brooch really didn't do at all, Andrew thought... The tide was coming in fast. The sea would cover the place where they had sat in a minute. (To the Lighthouse, p.89) (Ehrlich 1990: 66)

b. All this bother about a brooch really didn't do at all, Andrew thought... The tide was coming in fast. The sea would cover the place where they * sat in a minute.

The direct idea which Andrew has in (51a) is The sea will cover the place where we have sat. In this sentence there is a shift of perspective in the RC: the main clause future TO is treated as if it were t0 and the RC is temporally subordinated to it. The present perfect indicates W-anteriority. In this example the past perfect cannot be replaced by the past tense. Unlike the RC verb, the ad-verbial in the stretch of RST is anchored to the time of Andrew's thought. The following schematic representation can be given of the sentence:75


The following example is similar:

(52) a. It was 5 p.m. It wouldn't take long before the others would arrive now. They would probably bring flowers they had bought at half-price just before the flowershop closed.

b. It was 5 p.m. It wouldn't take long before the others would arrive now. They would probably bring flowers they ? bought at half-price just before the flowershop closed.

The RC in (52b) is interpreted as referring to a habitual situation and is not a thought passing through the character's mind.

An observation which transcends the gradient is that: a. The CRPP must be taken into account. b. A first-person subject is more likely to result in temporal sub-

ordination than a second- or third-person subject. The reason is obvious: if the speaker talks about his own experiences, he will represent the situation as he perceived it at that past moment in time. When the introductory subject is another person, the speaker is free to represent what that person said from the point of view of the moment of speaking.

(53) a. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which I believed had caused nausea on several occasions, (adapted from SEU)

b. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which I believed ? caused nausea on several occasions.


c. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which she believed had caused nausea on several occasions.

d. I first saw the above on 8.11.22 with a history very much resembling polymyalgia rheumatica which she believed caused nausea on several occasions.

(54) a. I went back to the shop and asked for my purse, which I explained I had left on the counter when I paid for the flowers.

b. I went back to the shop and asked for my purse, which I explained I ? left on the counter when I paid for the flowers.

c. She went back to the shop and asked for her purse, which she explained she had left on the counter when she paid for the flowers.

d. She went back to the shop and asked for her purse, which she explained she left on the counter when she paid for the flowers.

It seems reasonable to posit the hypothesis that the CRPP does not only apply to the expression of anteriority in RCs; it should be possible to illustrate the CRPP with other types of clause as well. Interesting though such an investigation may be, it falls outside the scope of the present discussion. Therefore, I will limit myself to some of the examples in Bouscaren et al. (1982) and comment on how they explain the (un)acceptability of the past tense.

a. They do not explain the unacceptability of the past tense in the following example:

(55) It was only after she had been performing (? was per-forming) for an hour that she realized that practically every other man in the hall was snoring too. At which point the game changed slightly. (Coronation Street - B.B.C./T.V. Series) (Bouscaren et al. 1982: 78)

The perfect tense cannot be replaced by a past tense: the presence of for an hour requires the use of a perfect which is given a continuative interpretation. A similar comment applies to the following example:

(56) He died after he had been (? was) ill for a long time. (Bouscaren et al. 1982: 77)


In Bouscaren et al.'s opinion, the past tense is unacceptable because there are tight semantic links between illness and death. If "tight semantic links" stands for "a continuative reading", their explanation is correct; however, they do not specify how the term should be understood.

b. Bouscaren et al. (1982) use the following examples to prove that the event indicated by the past perfect is not necessarily the one "le plus eloigne dans le passe" [that is the furthest removed from the present] (Bouscaren et al. 1982: 79):

(57) Now England stood alone in her "finest hour". With an army half of whose arms and equipment had had to be left on the beaches of France she awaited invasion, and with a handful of fighter planes, Spitfires and Hurricanes, she challenged - and by the end of September had broken - the power of Germany's Luftwaffe. (Churchill - The man of the century) (Bouscaren et al. 1982: 79)

Bouscaren et al. are right to point out that the past perfect does not necessarily imply reference to a very distant situation. However, the speaker has chosen to use the past perfect rather than the past tense because the use of the former tense indicates that the state at the end of September is the result of a gradual process taking place before September and accomplished by the end of September. By the end of September specifies the time which functions as T02 . The past perfect is used because it is the tense par excellence to indicate that the result brought about by the situation still applies at T02 . A past tense in this context would suggest that breaking the power of the Germans takes place at the end of September. The following is another example they give:

(58) The Houdinis were riding the crest of a high wave which, by late fall, gave no evidence of breaking. Bessie had long ago forgotten the fear that tortured her in London - the fear that the Alhambra engagement, like their one week at Tony Pastor's and their one brief tour of the Keith circuit, would lift them up only to drop them again with a sickening thud. (B. Williams and S. Epstein - The great Houdini -Scholastic Book Service - 1970) (Bouscaren et al. 1982: 79)

As in the previous example, the past perfect (had forgotten) is probably used in (58) because it stresses the idea of result: she never


thought about the fears she had in London. The past tense (tortured) is used in the RC because the restricting information need not be explicitly located in time (cf. chapter 6, section 6.3.); the place ad-verbial in London, which coincides with a certain time, fulfils that function. The CRPP is overruled by pragmatic factors.

c. As the following examples and the comment that goes with them show, Bouscaren et al. take great care not to explicitly use the term anteriorite when characterizing the uses of the past perfect:

(59) a. In Blackburn, England, in the fall of 1902, Houdini attempted to release himself from fetters supplied by a Mr. Hodgson, the principal of a school of physical training. It appears that Houdini had not, in this case, seen the irons beforehand, and when he examined them on the evening of the test.. He claimed that they had been tampered with. (- The great Houdini - op. cit. -) "Avec certains proces ä la forme passive, il est des cas oü la mise en relief du processus par une forme had-en est la seule fafon de donner un sens a l'enonce" (Bouscaren et al. 1982: 86). [With certain processes in the passive form, the process needs to be focussed on by the past perfect, as this is the only way to make the utterance meaningful.]

b. My mother fell on the ice outside our supermarket and broke her hip. They hadn't even bothered to put salt down. Can we sue for damage? (Daily Express) "L'enonciateur, d'apres l'ensemble de son enonce, veut rendre compte d'un moment du passe que le past serait impossible ä decrire. ... La forme had-en permet de rompre le recit pour revenir sur une negli-gence fächeuse qui explique la chute" (Bouscaren et al. 1982: 87). [Judging from this utterance, the speaker wants to refer to a past moment which it would be impossible to describe with a past tense ... The past perfect enables one to break the story line and focus once more on the gross negligence which explains the fall.]

c. He was walking down a London street one day in January 1901 - a street draped in heavy mourning


for the beloved Queen Victoria who had died several days before - when he saw in a shop window the model of a gown that had been designed for Her Late Majesty. (- The Great Houdini - op. cit.-) "Le fait de completer un nom par une relative peut s'analyser comme une attitude de COM-MENT AIRE. (...) Le COMMENT AIRE par une relative comporte done toujours un element de reprise (le pronom relatif) en meme temps qu'il permet d'accroitre la determination de l'antecedent" (Bouscaren et al. 1982: 96-97). [The fact that a relative clause is added to a noun may be analysed as an attitude of comment (...) Accordingly, a comment in the form of a relative clause always contains an element of repetition (the relative pronoun) and at the same time contributes to determining the reference of the antecedent.]

d. There was a lion in his cage. He was lying very still in a corner, his eyes half closed. But Ehrich had read about lions. He knew how wild they were. (-The great Houdini - op. cit.-) "but oppose l'etat present constate par Ehrich lying very still, his eyes half closed ä une connaissance anterieure. Ce type de proposition peut egalement constituer un commentaire" (Bouscaren et al. 1982: 98). [But establishes a contrast between the present state observed by Ehrich lying very still, his eyes half closed and a former knowledge. This kind of proposition may also be a comment.]

It is difficult to understand the characterizing labels, which are insufficiently explained.76 It is much clearer and relevant to say that the past perfect indicates anteriority in the quoted examples. Moreover, the past perfect implicates that the result still holds at T 0 2 (cf. Declerck 1991a: 343). The latter are tangible labels and have the additional advantage of capturing the basic meaning of the form.


3.2.2. RRCs and NRRCs: different tense system?

3.2.2.1. Our investigation into the expression of W-anteriority has revealed that it is not the case that NRRCs only use shift of domain forms and RRCs only temporally subordinated verb forms. In other words, statements of the kind that verb forms which can be used in a RRC cannot be used in a NRRC (cf. the authors quoted in the introductory chapter) need to be specified further: what is not possible in one sector (e.g. the post-present sector) is not necessarily impossible in another (e.g. the past sector).

The examples illustrating the CRPP include RRCs as well as NRRCs. The effect of replacing the past perfect by the past tense when the past perfect sentence is bounded atelic or zero-telic (and there are no adverbials or pragmatic information which reveal the chronological order between the situations) is the same in RRCs and NRRCs. Turning examples into their (non-)restrictive counterparts does not affect the unmarked temporal interpretation, as the following examples show:

(60) a. We liked the students we had entertained ('entertained) in London, (atelic, activity) (RRC)

b. We liked Phil, Jonathan and Alice, whom we had entertained ('entertained) in London. (NRRC)

(61) a. The [Name] authorities do not supply confidential reports on expatriate lecturers, but our [Word Word] has commented that Mr [Name] was a very useful and sensible man whose wide experience in the field of English language had stood ('stood) him in a good stead at [Name]. (SEU) (atelic, state) (RRC)

b. The [Name] authorities do not supply confidential reports on expatriate lecturers, but our [Word Word] has commented that Mr [Name] was a very useful and sensible man, whose wide experience in the field of English language had stood ('stood') him in a good stead at [Name]. (SEU) (atelic, state) (NRRC)

(62) a. It was with an insight into the inspired scriptures of the Old Testament that He accepted the sacrificial vocation which Israel as a whole had rejected ('rejected). (SEU) (zero-telic) (RRC)

b. It was with an insight into the inspired scriptures of the Old Testament that He accepted the sacrificial


vocation of a servant and suffering Messiah, which Israel as a whole had rejected ('rejected). (NRRC)

As far as the CRPP for atelic and zero-telic verbs is concerned, RRCs and NRRCs behave alike: the unmarked interpretation of the past tense, unlike that of the past perfect, is that it indicates a relationship of simultaneity between the SUPC situation and the RC situation (whenever there is no sentence-internal, contextual or pragmatic information which imposes a chronological order). However, the chronological order established by shift of domain purely telic verbs differs in RRCs and NRRCs (cf. 3.2.2.2.)·

3.2.2.2. RRCs and NRRCs may indeed differ in the way they locate a situation in time by means of a particular verb form: if (a) the situation is telic bounded77 and not temporally subordinated to the main clause and if (b) no linguistic or contextual (pragmatic) clues indicate that the RC situation is W-anterior to the SUPC clause situa-tion and if (c) the situation occurs in a NRRC, then the NRRC situation will be interpreted as referring to a W-posterior situation:

(63) a. James Wilkinson became apprenticed to a London gunsmith who had set up ("set up") his business in Ludgate Street. (RRC)

b. James Wilkinson became apprenticed to a cele-brated London gunsmith called Henry Nock, who had set up ('set up') his business in Ludgate Street, (adapted from SEU) (NRRC)

The past tense will probably not be. used in example (63b) if the speaker wants to indicate that Henry Nock set up his business before James Wilkinson became apprenticed to him. In other words, the past perfect is the tense likely to be used if the NRRC bounded telic situa-tion is W-anterior to the SUPC situation. If the past tense is used, the NRRC situation will be interpreted as a continuative NRRC (Jespersen 1961, III: 105), whose situation is W-posterior to the SUPC situation (cf. chapter 7), although the W-posterior reading may be overruled by contextual knowledge. The following example exemplifies the same tendency:

(64) a. On this night of the disintegration of United's defence Astle struck again in the seventy-fifth minute. They headed in a pass from Lovett, who had


come (? came) on as a substitute for Rees, (adapted from SEU) (NRRC)

b. On this night of the disintegration of United's defence Astle struck again in the seventy-fifth minute. They headed in a pass from the Astle player who had come ("came") on as a substitute for Rees. (RRC)

There seems to be some sort of clash in (64a) between the inter-pretation suggested by the past tense and the context. With the past tense, we get the impression that Lovett came on as a substitute after the goal had been scored. This is not compatible with the fact that it was Lovett who delivered the pass from which the goal was scored. So this means that the hearer, guided by his knowledge of the world, will adjust the interpretation suggested by the verb. This example differs from (63) in that in the latter example any chronological order is possible, i.e. Henry Nock might have set up his business before or after Wilkinson became his apprentice. In (64) on the other hand, only one interpretation is pragmatically plausible (1. come on as a substitute - 2. score). No confusion arises in (64b): the past tense as well as the past perfect represents the RC situation as W-anterior to the SUPC situation. It follows that the order in which the NRRC and the SUPC are reported is important if there is reference to a purely telic situation. The same tendency can be observed in (65):

(65) a. Mr Wall, who was smuggled ("had been smuggled") into the debating chamber of the students' union, was greeted with shouts of 'Sieg Heil' and standing students giving the Nazi Salute. (SEU) (NRRC)

This example clearly shows the effect of passivization and the influence of the position of the RC in the sentence. In example (65a), the result (Mr Wall was, against all expectations, present as a result of the students' smuggling him into the room) is conveyed by the past tense as well as by the past perfect sentence. The past tense suggests a quick succession of two events: smuggling Mr Wall into the debating room followed by his being greeted there. If the SUPC is made active, which has the effect of putting the verb of the SUPC before the RC (cf (60b)), the past tense changes the meaning of the sentence:


(65) b. The students greeted Mr Wall, whom they had smuggled (? smuggled) into the debating chamber. (NRRC)78

The chronological order suggested by the past tense is different in (65b): the past tense suggests that the act of smuggling Mr Wall into the debating room follows his being greeted. If this particular order is not compatible with the context, the hearer will have to adjust the chronology guided by his knowledge of the world. In (65c), in which the RC is active but precedes the SUPC verb, both the past tense and the past perfect are acceptable:

(65) c. Mr Wall, whom the students smuggled ("had smuggled") into the debating room, was greeted with 'Sieg Heil'. (NRRC)

However, the past tense gives rise to a slightly different interpre-tation: the act of smuggling is represented as something that gives relevant information about Mr Wall; the focus is no longer on the laborious process of smuggling which preceded the greeting. The following example also illustrates the importance of the order in which a purely telic NRRC and its SUPC are reported:

(66) George McLean took a free kick 25 yards out, Reaney headed the ball away, but Bobby Wilson, who had appeared ("appeared") from nowhere, so to say, forced it past Harvey. (SEU) (NRRC)

(67) a. On this night of the disintegration of United's defence Astle struck again in the seventy-fifth minute. They headed in a pass from Lovett, who had come (? came) on as a substitute for Rees. (NRRC) (= (64a))

Although example (66) is similar to (67a), the verb of the SUPC follows the NRRC; therefore, the past tense can be used. As can be gathered from (67b), the reversal of the RC and the SUPC results in an acceptable past tense RC:

(67) b. Lovett, who had come on ("came on") as a substitute for Rees, headed in a pass. (NRRC)


If we change the NRRC examples in 3.2.2.2. into RRCs, the interpretation of the situation referred to by means of the past tense is likely to be different. In the RRC, unlike in the NRRC, the unmarked interpretation of the past tense situation is that it refers to a W-anterior situation. Sentences of this type will also be discussed in chapter 7.

3.2.2.3. Occasional differences in acceptability of temporally sub-ordinated verb forms can be observed between RRCs and NRRCs. Unlike the examples in section 3.2.2.2., in which the temporal location of the past tense situation differs in RRCs and NRRCs, the unacceptability of the past perfect has nothing to do with a possible difference in the chronological order established between, on the one hand, the RRC and the SUPC situation and, on the other, the NRRC and the SUPC situation:

(68) a. This latest success was most recently acknowledged by the organization who selected the Bonded shaving system for its award for technological innovation and excellence in design, (adapted from SEU) (RRC)

b. This latest success was most recently acknowledged by the organization who had selected the Bonded shaving system for its award for technological innovation and excellence in design. (RRC)

The difference in acceptability between RRCs and NRRCs in this respect can be explained in terms of the different functions performed by the two types of clause. In (68a), the links between the RRC and the SUPC are tighter than in (68b) because, apart from defining which organization is being talked about, the RC also indicates how the relevant organization showed its acknowledgment, namely by choosing the Bonded shaving system for its award for technological innovation. When the past perfect is used in the RRC, the organization is merely defined, i.e. by referring to the fact that some time before that of the SUPC situation it once selected the Bonded shaving system for its Award for technological innovation and excellence in design. The RC does not tell us how the organization showed its acknowledgment of the latest success. In (68c) and (68d), the RRC is made non-restrictive:


(68) c. This latest success was most recently acknowledged by the Design Council, who selected the Bonded shaving system for its award for technological innovation and excellence in design. (NRRC)

d. This latest success was most recently acknowledged by the Design Council, who ?? had selected the Bonded shaving system for its award for technological innovation and excellence in design. (NRRC)

It is not felicitous to use the past perfect in the corresponding NRRC. We know which council is being talked about without the RC information. If the RC situation is supposed to show how the council showed its acknowledgment, the RC situation should relate to the same past moment in time as the SUPC situation. It seems more reasonable to use the past tense, because this tense is interpreted as referring to a situation that held at more or less the same time as the SUPC situation; there is a quick succession of the two situations. The past perfect, on the other hand, indicates that at some moment in time preceding the SUPC situation, the council selected the Bonded shaving system for its award for technological innovation; there seems to be a greater lapse of time between the SUPC and the RC situation. The causal link between their doing so as a means of acknowledging the latest success is barely apparent when the past perfect is used.

The difference in acceptability of the past perfect in the RRC and the NRRC should not be ascribed to a different location in time of the RRC/NRRC situation. The defining function of the RRC allows more "temporal freedom" than that of giving relevant information (NRRC), which imposes greater temporal restrictions.

3.2.2.4. The conclusion to be drawn is that the temporal interpre-tation of a RRC differs from that communicated by its non-restrictive counterpart only when the RC situation is bounded telic. As pointed out in section 3.2.2.2., a telic bounded situation in a NRRC in which the domain is shifted allows a W-posterior reading. The NRRC situation is likely to be interpreted as belonging to the foreground (cf. chapter 7, section 7.3.) and the chronological order between the situations coincides with the order in which the clauses are reported. A W-posterior reading is not the unmarked one in the corresponding telic bounded RRC; RRCs do not belong to the foreground unless a W-posterior order is implied in the semantics of the situations


referred to (cf. chapter 7, section 7.5.2.). If the RC is not of the type mentioned so far, there is no difference between the unmarked temporal interpretation of the past perfect/past tense forms in RRCs and NRRCs. The difference in acceptability of the temporally subor-dinated verb form referred to in 3.2.2.3. is due to the difference in function performed by RRCs and NRRCs; it is not related to a different location in time by a particular verb form. In any case, as far as temporal subordination is concerned, RRCs and NRRCs are similar in the sense that they both allow the use of the past perfect to refer to an anterior situation. As will become clear in the course of the discussion, this is the basic difference between past sector (N)RRCs and post-present sector (N)RRCs, as post-present NRRCs are never temporally subordinated to their SUPC situation.

3.3. Simultaneity in past sector (N)RRCs: relative and absolute past tenses?

Having demonstrated that there is a considerable degree of similarity between RRCs and NRRCs as regards the expression of W-anteriority in the past sector, one might wonder whether the status of the past tense referring to an unbounded W-simultaneous situation is different in RRCs and NRRCs, a line of reasoning implicit in De-clerck (1991a). Although intuitively, the idea that the syntactic difference between the RRC and the NRRC is reflected in the status of the verb form seems to makes sense, there appears to be little positive evidence in favour of the claim that the past tense in the NRRC in (69b) is an absolute tense, whereas the past tense in the RRC in (69a) is a relative tense:

(69) a. I talked to the man who kept Susan hidden in his flat. (RRC)

b. I talked to John, who kept Susan hidden in his flat. (NRRC)

The difference referred to can be maintained only if (a) one takes it as an a priori axiom that the syntactic difference between RRCs and NRRCs must be reflected in the tense used or if (b) one draws conclusions about the status of the past tense in (69a) and (69b) on the basis of a comparison with their post-present counterparts (cf. section e).

Simultaneity in past sector (N)RRCs 113

a. The first argument is a purely formal79 one: the past perfect is the relative tense used to express anteriority both in RRCs and NRRCs. It therefore makes sense to put forward the hypothesis that the past tense is the relative tense used to express simultaneity in past sector RRCs and NRRCs alike; if a relative tense can be used in a NRRC to refer to an anterior situation, it should also be possible to use a relative tense to refer to a simultaneous situation in a NRRC.

b. Declerck uses two tests to distinguish relative pasts from absolute pasts. The observations made in 3.2.1. allow us to explain the principles which underlie the tests. In my opinion, the tests do not give conclusive evidence about the status of the past tense in RRCs and NRRCs (and independent clauses in general).

Declerck writes that if the present perfect in Dutch is inter-changeable with the past tense, the latter is an absolute tense:

(70) Hij is weggegaan toen hij wist dat ik boos was/* ben geweest. (He has gone away when he knew that I was/* have been cross) (Declerck 1991a: 102)

(71) Betty zei dat John geloofde dat de erfgenaam een meisje was/* is geweest. (Betty said that John believed that the heir was/* has been a girl.) (ibid.)

In the examples in (70) and (71), the past tense in the SC is a relative tense expressing simultaneity with its SUPC. Therefore, the present perfect cannot be used in Dutch.

In what follows, it will be shown that it is possible to formulate the constraints resulting in the (non-)interchangeability of the present perfect and the past tense in terms other than of relative and absolute tense. The semantics of the present perfect in Dutch predicts that the past tense referring to an unbounded situation will never be substitutable with the present perfect, no matter whether the past tense is a relative or an absolute tense, because the Dutch present perfect is always of the indefinite type.80 In other words, the Dutch present perfect always refers to bounded situations.81 From this it automatically follows that the present perfect cannot replace any past tense which refers to an unbounded state without establishing a change in meaning:82

(72) a. Jan studeerde van 1989 tot 1990 in London. Mieke woonde toen bij hem.


(John studied from 1989 till 1990 in London. Mieke lived with him at that time.)

b. Jan heeft van 1989 tot 1990 in London gestudeerd. ? Mieke heeft toen bij hem gewoond. (John has studied in London from 1989 till 1990. ? Mieke has lived with him at that time.)

(73) a. Jan studeerde met een meisje dat in Engeland woonde. (John studied with a girl who lived in England.)

b. Jan studeerde met een meisje dat in Engeland gewoond heeft. (John has studied with a girl who has lived in England.)

(74) a. Ik was verliefd op Jan in die tijd. (I was in love with John at that time.)

b. Ik ben verliefd geweest op Jan in die tijd. (I have been in love with John at that time.)

In all the examples given, the unmarked reading of the present perfect sentence differs from the unmarked reading of its past tense counterpart, no matter whether the present perfect replaces an absolute (main clause atelic unbounded) past tense (cf. (72) and (74)) or a relative (subclause atelic unbounded) past tense (cf. (73)). In (72) and (74), the present perfect represents the situation as holding during a subpart of the time indicated by the adverbial, whereas the past tense suggests that its situation held throughout the time indicated by the adverbial. In (73), the unmarked interpretation of the present perfect situation is W-anteriority with respect to the SUPC situation, whereas the past tense indicates simultaneity with the SUPC situation.

The reason why many English relative past tenses cannot be translated by a Dutch present perfect is the following: the unmarked temporal relation which the unbounded past tense situation establishes is one of simultaneity. It follows that English past tense forms of that type will never be translated into Dutch by means of a present perfect because the latter will change the unbounded situation into a bounded one, which in turn will affect the temporal relations between the situations. The above observations imply that whenever the Dutch present perfect can translate an English past tense, there will always be reference to a bounded situation. Bounded situations do not often express simultaneity in a past domain (they are mostly W-anterior to a state or another bounded situation).


Accordingly, they can mostly be translated by a present perfect, as is clear from the examples in (75):

(75) a. Jan zei dat de erfgenaam het grote lot wont gewonnen heeft in 1985. (John said that the heir won/has won the lottery in 1985.)

b. Jan zei dat de erfgenaam een huis op Jamaica kocht/gekocht heeft in 1985. (John said that the heir bought/has bought a house in Jamaica in 1985.)

c. Jan zei dat de erfgenaam van 1975 tot 1980 in London woonde/gewoond heeft. (John said that the heir from 1975 till 1980 in London lived/lived has.)

d. Jan zei dat de erfgenaam gedurende vijf jaar in London ? woonde/gewoond heeft. (John said that the heir ? lived/has lived in London for five years.)

It is crucial to remark, however, that the English past tenses in (75) are absolute tenses and cannot be misunderstood as referring to a W-simultaneous situation. The necessity of having a test to determine the relative/absolute status of the past tense does not arise in sentences of this type. Still, the example in (75d) shows that even if there is an adverbial with a bounding effect on the clause, the present perfect may still be preferred. The adverbial suggests a limited amount of time; it imposes boundaries. However, it seems to need the perfect tense to reinforce the boundedness of the situation. The questionable status of the past tense is due to the tension between the boundedness inherent in the adverbial and the unboundedness inherent in the past tense state woonde.

So far, I have argued that the present perfect is substitutable with the past tense in Dutch provided the latter refers to a bounded situation. However, the following point must be added: if the past tense refers to a bounded situation that is W-simultaneous with another situation (i.e. two punctual situations are (sloppily) sim-ultaneous (cf. footnote 62)), the perfect tense cannot be used instead of the past tense. In the following example, the SC past tense situation is bounded and W-simultaneous with the SUPC situation. The present perfect cannot be used:


(76) Toen heeft Jan gehoord dat een schot weerklonk (* weerklonken heeft). (Then, John has heard that a shot resounded (* has resounded).)

Even though the situation of resounding is punctual, the fact that it is simultaneous with John's hearing something requires the use of a past tense.

The above observations clearly demonstrate that an unbounded past tense sentence cannot be translated using a Dutch present perfect, no matter whether the verb occurs in a SC or main clause. This makes it more difficult to draw firm conclusions about the status (relative or absolute) of the English past tense form in a RC that cannot be translated by a Dutch present perfect form. When we apply the test to the RC examples in (69) it seems that the effect of the present perfect is the same in the RC and the NRRC:

(77) a. Ik praatte met de man die Suzanne verborgen hieldf gehouden heeft'.

b. Ik praatte met Jan, die Suzanne verborgen hie Id/' gehouden heeft'.

In both (77a) and (77b), the unmarked interpretation of the past tense is W-simultaneity, the present perfect indicating W-anteriority. Moreover, as Mommer (1986) points out, one has to be careful when drawing conclusions for one language on the basis of another language: "Naturally no conclusion can be based solely on transla-tions into a second language. To do this would be to impose the conceptual system of the second language onto the first. We cannot assume that just because a temporal construction in one language is usually translated by a particular construction in another language, that those two constructions are semantically and pragmatically equivalent in every way. However, the constructions may very well share some semantic and pragmatic characteristics" (Mommer 1986: 3). I subscribe to the view that, although English and Dutch are both Germanic languages, one must be cautious when drawing conclusion about English on the basis of the behaviour of tenses in Dutch, the more so as the use of the present perfect is precisely an area in which Dutch and English differ.

c. The other test relates to indirect speech. Declerck (1991a: 100-101) argues that relative tenses cannot be backshifted, whereas absolute tenses can (but need not always be):


(78) a. Bill believed it was five o'clock. (Declerck 1991a: 100)

b. John said that Bill (had) believed it was/* had been 5 o'clock.

Believe in (78a) establishes the domain and is an absolute tense. Accordingly, it allows the backshift. Was is a relative tense expressing simultaneity with the main clause and can therefore not be backshifted. This leads Declerck to posit the claim that past tenses in when-clauses are relative tenses, because they are "not normally backshifted" (1991a: 101):

(79) a. Bill was in London when Jenny was in Paris. (Declerck 1991a: 101)

b. I explained that Bill had been in Paris when Jenny was (* had been) in Paris.

In a footnote Declerck adds: "I am adding the proviso 'normally' because there are types of subclause in which English allows indirect binding. Perhaps some speakers do accept sentences of the kind ? Bill said he had been unhappy when he had been young. However, it is clear that the use of the preterit is strongly preferred in the when-clause of such examples" (1991a: 101). As a matter of fact, there are far more examples with indirect binding than Declerck believes there are (cf. Depraetere 1990: 43, chapter 8):

(80) a. She said: Ί felt sick when I saw the accident.' b. She said she had felt sick when she saw/had seen

the accident. (81) a. She said: 'He was even less keen when the minister

hinted that he might lose his job.' b. She said that he had been even less keen when the

minister had hinted/hinted that he might lose his job. (82) a. She said: 'Hugh was 8 and he 10 when they first

realized there was something different about them.' b. She said that Hugh had been 8 and he 10 when they

first realized/had first realized there was something different about them.

(83) a. She said: Ί hit him because I was angry with him.' b. She said she had hit him because she was/had been

angry with him.


(84) a. She said: Ί had an argument with Mary, who was very angry.'

b. She said she had had an argument with Mary, who was!had been very angry.

This means that the test faces the following problem: how can we know whether the past perfect can be used as a result of indirect binding (which implies that the original past tense is either relative or absolute) or because the original verb is an absolute tense? In other words, we are in need of another test enabling us to make that distinction. I cannot solve this problem. Even so, the large number of indirectly bound examples makes it difficult to use this test as a reliable means of determining the status of the past tense in RRCs/NRRCs:

(85) a. She said: Ί talked to the man who kept Susan hidden in his flat.'

b. She said she had talked to the man who kept ('had kept') Susan hidden in his flat.

c. She said: Ί talked to John, who kept Susan hidden in his flat.'

d. She said she had talked to John, who kept ('had kept') Susan hidden in his flat.

No matter whether the RC is restrictive or non-restrictive, the unmarked temporal interpretation conveyed by the past/past perfect is the same (W-simultaneity vs. W-anteriority). It is therefore difficult to draw conclusions about the status of the past tense in the RCs in (85a) and (85b) (= (69a) and (69b)) on the basis of the indirect speech test.

d. I hope to have shown that it is difficult to rely on the above two tests to determine the status of the past tense used to express simultaneity in a (N)RRC. However, so far, no tests have been applied to show that the unbounded past tense referring to a W-simultaneous situation is relative, no matter whether it occurs in a RRC or NRRC. Nonetheless, when one wants to indicate that the (N)RRC situation is W-simultaneous with the SUPC situation, the same form, i.e. the past tense, is used. In my opinion, this is a point in favour of a unified approach to the past tense in both RRCs and NRRCs: the tense used is the same (the past tense), the temporal information is the same (in both cases, there is a relationship of W-simultaneity between the SUPC situation and the RC situation),


hence one should allow for the possibility that the status of the past tense be the same.

e. However, when example (69) is transposed to the post-present sector, the verb forms used differ in RRCs and NRRCs:

(86) a. They will threaten the girl who will hide / hides John.

b. They will threaten Mary, who will hide / 'hides' John.

As will be pointed out in chapter 5, the NRRC situation, unlike the RRC situation, cannot be temporally subordinated to the SUPC situation. Absolute tenses have to be used in the NRRC. This might lead one to conclude that the past tense in a W-simultaneous past sector NRRC must be an absolute tense as well. However, this argu-ment may not be entirely watertight. It will become clear in the course of the discussion that RRCs and NRRCs basically differ in indirectly bound W-posterior and post-present sector contexts. Therefore, any comparison with sentences of the latter type is bound to result in a claim that NRRCs and RRCs differ. However, this conclusion is not supported by the results of any other type of test.

To end this section, it might be interesting to consider the fol-lowing example from Ladusaw (1977), which Declerck quotes to illustrate the shift of domain in the past sector:

(87) Mary saw the unicorn that walked. (Declerck 1991a: 48)

Ladusaw observes that the sentence is true irrespective of whether the walking took place before, after or at the same time as the seeing. Declerck claims that this is the result of the use of a shift of domain verb form: both verbs are absolute tenses, therefore, the temporal relation between the situations is left vague. Again, the problem of determining the status of the past tense emerges. Three remarks are in order:

1. It is questionable whether walked should be classified as an absolute tense if the RC is interpreted as referring to a W-simul-taneous situation. The backshifting test again does not solve the problem of determining the status of the past tense:

(88) He said that Mary had seen the unicorn that walked (had walked).


When we backshift the statement, had walked changes the chronological order between the two situations: the unmarked interpretation is that the RC situation is anterior to the SUPC situation. If the test is correct, this might be taken as an indication that walked is a relative tense. An interpretation in terms of indirect binding, in which case the past perfect RC situation is W-simulta-neous with the SUPC situation is not ruled out, but is not the unmarked interpretation.

2. Even if walked is an absolute tense form, the unmarked temporal interpretation is one of W-simultaneity. In other words, there is an unmarked interpretation associated with the sentence, which may of course be overruled by contextual factors.

3. Walk may mean can walk. This may explain why there is doubt as to when the unicorn walks. If there is reference to an ability, it is very likely that walked will be interpreted as referring to a W-simultaneous situation. Moreover, see indicates direct perception, which also contributes to the preferred reading of W-simultaneity.

3.4. (W-)posteriority in past sector (N)RRCs

3.4.1. Introductory remarks

The conditional tense83 is the relative tense used to express pos-teriority in a past domain:84

(89) a. He picked out from the group all the doctors whose experience would be most useful in the emergency. (Mittins 1962: 94) (RRC)

b. He picked out from the group all the doctors, whose experience would be very useful in the emergency. (NRRC)

It is important to point out that the use of the conditional tense does not necessarily imply that the situation lies before t0 (cf. Van Voorst 1987: 269). In the following examples, the conditional tense may refer to a post-present sector situation:

(90) a. Myrna reported that the bomb would go off next week.

b. Myrna reported that the bomb will go off next week. (Smith 1977: 166)

( W- posteriority in past sector (N)RR Cs 121

(91) a. Pete said that the critics would see that you are (91) nervous. (Rigter 1986: 127)

b. Pete said that the critics will see that you are ner-vous. (ibid.)

(92) a. A child was born who would be king. (Dowty 1982: 30)

b. A child was born who will be king, (ibid.) (93) a. Bill sought a man who would be leaving. (Dowty

1982: 42) b. Bill sought a man who will be leaving, (ibid.)

(94) a. James bought a stamp that would be priceless. (Richards 1982: 83)

b. James bought a stamp that will be priceless, (ibid.)

From the point of view of the (a-part of) Grice's Maxim of Quan-tity,85 a speaker wishing to comply with the conversational principles should use the future tense to refer to a situation that lies after tQ. When the RC situation lies after t0, the speaker is likely to decide to use the future tense in order to be more informative:

(95) They were discussing the terms of payment of the goods that will be delivered later on. (i.e. after the moment of speaking)

(96) She was angry because he had invited the girl who will take part in the beauty contest, (i.e. after the moment of speaking)

When I say that the speaker is "more informative" when he uses a future tense to refer to a post-present sector situation, I mean that there is no need to bind a post-present sector situation to a TO other than t0 if there is no particular reason to do so. If the extra effort required to process a conditional tense (referring to a post-present sector situation) is not offset by extra contextual effects, the speaker will not be considered to be optimally relevant. However, even in cases like these, he may nevertheless decide to use the conditional to attain certain effects (cf. section 3.4.3.).

There are not that many examples in which the conditional is used in RCs that are W-posterior to a past tense SUPC situation in the corpus at my disposal. This is basically due to the following fact: when the RC situation lies before t0, which implies that it is a past fact, the speaker will prefer to represent the situation as a past fact (past tense) rather than a past prediction (conditional tense), except


in narrative contexts, where the conditional tense can also be used to describe past facts:

(97) a. He blamed a knee injury in [2222] for this although he did not have any pain until [2222] for which he was later referred to St. [Name]'s Hospital and given a course of physiotherapy. (SEU)

b. He blamed a knee injury in [2222] for this although he did not have any pain until [2222] for which he would later be referred to St. [Name]'s Hospital and given a course of physiotherapy.

RRC and NRRC examples can be found in which the conditional has to be used; a shift of domain form (in this case, a past tense, which is the result of a shift of domain within the past sector) cannot be used, either because it results in an ungrammatical sentence (cf. (98)) or because it changes the temporal order between the SUPC and the RC situation (cf. (99)).86 In other RRCs and NRRCs (cf. (100) and (101)), the conditional tense is substitutable with the past tense:

(98) Yesterday the President of the Board of Trade, Mr Crosland, said in a written answer to a question by Mr James Johnson (...) that he had chartered the trawler Orsino, which would sail (* sailed) on about November 27 and remain stationed off of Iceland until April 30, 1969. (SEU) (NRRC)

(99) Davis's original 2.9% investment in NWA resulted from a study of the airline industry that suggested that a small group of airlines had emerged from deregulation whose results would be ('were') steadier, with less fare-cutting, than in the past decade (WSJ) (RRC)

(100) He blamed a knee injury in [2222] for this although he did not have any pain until [2222] for which he was later referred ("would be referred") to St. [Name]'s Hospital and given a course of physiotherapy. (SEU) (NRRC)

(101) For example, we (...) recommended improvements which led ("would lead ) to many building societies changing their practices. (SEU) (RRC)

This means (a) that temporal subordination is not restricted to either RRCs or NRRCs and (b) that there are constraints on the use of a shift of domain form in both RRCs and NRRCs.

(W-)posteriority in past sector (N)RRCs 123

3.4.2. Shift of domain form not acceptable

3.4.2.1. Constraints on the Replacement of the Conditional Tense (CRCT)

a. The use of the past tense sometimes blurs the temporal relation of W-posteriority between the SUPC and the RC situation. In the following examples, the past tense suggests that the RC situation is simultaneous with the SUPC situation rather than W-posterior to it:

(102) a. But then Judge O'Kicki often behaved like a man who would be king — and, some say, an arrogant and abusive one. (WSJ) (RRC)

b. But then Judge O'Kicki often behaved like a man who 'was' king — and, some say, an arrogant and abusive one.

(103) a. The breakthrough year was 1954 when (...) Francis Crick and James Watson proposed a model which would account both for the coding and the duplication. (SEU) (RRC)

b. The breakthrough year was 1954 when (...) Francis Crick and James Watson proposed a model which 'accounted both for the coding and the duplication.

(104) a. Meanwhile work was going on which would put all these discoveries in the shade, while giving more precision to the notion of a gene. (SEU) (RRC)

b. Meanwhile work was going on which 'put' all these discoveries in the shade.

(105) a. We liked Phil, Jonathan and Michael, whom the three of us would entertain in London. (NRRC)

b. We liked Phil, Jonathan and Michael, whom the three of us 'entertained in London.

It seems that the class of verbs bringing about a change in temporal interpretation is the same as that which does not allow the replacement of the past perfect by a past tense form, i.e. atelic verbs (states (cf. (102), (103)) and activities (cf. (105)) and zero-telic verbs (cf. (104)).87

b. The shift of domain in RRCs containing purely telic verbs is also constrained: the past tense RRC situation is likely to be interpreted as referring to a W-anterior event when the past tense is used (provided the pragmatics of the situation and the lexical content do not impose


a W-posterior chronological order) (cf. (106a), (107a)). This con-straint does not apply to NRRCs, whose situations are usually inter-preted as forming part of a sequence (cf. (106c), (107c)):

(106) a. John married the girl who 'got' pregnant. (RRC) b. John married the girl who would get pregnant. (RRC) c. John married June, who "got" pregnant. (NRRC) d. John married June, who would get pregnant. (NRRC)

(107) a. He gave the letter to the clerk who 'copied it. (RRC) b. He gave the letter to the clerk who would copy it.

(RRC) c. He gave the letter to the clerk, who "copied" it.

(NRRC) d. He gave the letter to the clerk, who would copy it.

(NRRC)

As pointed out before, a self-evident, though important constraint on the tendency mentioned is that it does not apply to the same degree whenever adverbials, the lexical content of the verbs or the prag-matics of the situation indicate that the RRC situation is W-posterior to the SUPC situation.

c. The conclusion is that atelic and zero-telic RRCs and NRRCs behave alike with respect to the CRCT: the unmarked temporal in-terpretation of the past tense is one of simultaneity with the SUPC situation in both RRCs and NRRCs (provided the lexical content, contextual or general pragmatic information do not impose a chrono-logical order).

3.4.2.2. Intensional domains

The use of the past tense is often ungrammatical in sentences in which an intensional domain is created, i.e. a domain created by a verb of propositional attitude (cf. section 3.2.1.). The stipulation that there is temporal subordination in an intensional domain referring to a W-posterior event applies rather strictly.88 It is not difficult to see why: no matter whether the subject of the embedding verb is the 1st, 2nd or 3rd person, at the time the RC situation is said, doubted or thought about, one does not know its outcome:

(108) I put her on Naprosyn 250 mgm., q.d.s., which I hoped would relieve (* relieved) at least some of her symptoms


until I see her again in one month's time, (adapted from SEU)

(109) Yesterday the President of the Board of Trade, Mr Crosland, said in a written answer to a question by Mr James Johnson (...) that he had chartered the trawler Orsino, which would sail (* sailed) on about November 27 and remain stationed off of Iceland until April 30, 1969. (SEU)

(110) He then sold her some capsules that he asserted would take care (* took care) of the tumors and cysts until she could collect the money for buying his machine. (Bree 1985: 25)

Sentences containing represented speech and thought, in which the narrator looks at things from the point of view of one of the charac-ters without indicating his own presence (cf. section 3.2.1.), also re-quire the use of temporally subordinated verb forms:

(111) She looked at him thinking to find this shown in his face; he would be looking ('was looking') magnificent... But not in the least! He was screwing his face up, he was scowling and frowning and flushing with anger. (To the lighthouse, 110) (Ehrlich 1990: 75)

(112) Sir William explained to her the state of the case. He [Septimus] had threatened to kill himself. There was no alternative. It was a question of law. He would lie in bed in a beautiful house in the country. The nurses were ad-mirable. Sir William would visit ('visited) him once a week. (Mrs Dalloway, 107) (Ehrlich 1990: 75)

The past tense is not ungrammatical in (111) and (112) but it changes the temporal location of the situation.

In past tense intensional contexts, there is a gradient in the likeli-hood of a temporally subordinated form having to be used to refer to a W-anterior RC situation (cf. section 3.2.1.). This gradient does not apply to past tense intensional context sentences which refer to a W-posterior situation; no matter whether the introductory verb is one of saying or thinking or indicates doubt, the conditional tense has to be used. At first sight, it seems that the domain can be shifted when the RC indicates a "timetable future":


(113) He sold her tickets for the football match between the 49ers and the Seahawks which he said started at 5 p.m.

However, this is only apparently so. The direct speech sentence con-tains a present tense which is the result of a shift of perspective The match starts at 5 p.m. In (113), the original tense form is backshifted. In other words, the past tense is not the result of a shift of domain. Moreover, when the embedding verb indicates doubt or uncertainty, the past tense is unacceptable, even if there is reference to a "timetable future":89

(114) He sold her tickets for the football match between the 49ers and the Seahawks which he wasn't sure / doubted would start (* started) at 5 p.m.

3.4.3. Temporal subordination unacceptable/relatively unac-ceptable90

It has already been pointed out that the RC W-posterior situation of-ten lies before the moment of speaking. In other words, there is ref-erence to two past facts. Replacing the past tense by the conditional has the effect of changing the "facts" into "predictions". Put differently, using the conditional instead of the past tense means that there is no looking back from the present to two past situations (= shift of domain); rather, the situation which comes second in time is considered from a past point of orientation. Since we can never be absolutely sure that statements about the future will turn out the way they are expected or predicted to do, the W-posterior situation gets the connotation of "prediction, expectation" in this case.91 The use of would in examples of this kind often gives an overtone of "destiny known to an omniscient author describing the past" (Tottie - Över-gaard 1984: 152). Depending on the context, there may a preference for either the past tense or the conditional tense. In extreme cases, the conditional will be ruled out altogether:

a. There is an unspoken consensus that a football match consists of a series of (past) events following each other quickly in time. The use of the conditional, although not excluded, is therefore rather strange in football reports. Moreover, the conditional is often associated with a narrative context and is therefore not likely to be used in the following examples:


(115) Leeds eventually changed into top gear, and after a bewildering movement, in which everyone except Harvey appeared to take part, Gray delivered a splendid shot, which brought (? "would bring") a save no less worthy from Donaldson. (SEU) (NRRC)

(116) A cluster of players from both sides lunged at the ball and it ran free again to Book who made (? "would make") no mistake with his drive. (SEU) (NRRC)

(117) A corner from the right came to Doyle, whose header from in front of goal struck (? "would strike") the crossbar so solidly that the top netting shook, giving the fleeting impression that he had scored. (SEU) (NRRC)

(118) Dougan flicked the ball, first time, back to Kenning, whose shot thudded (? "would thud") into the net. (SEU) (NRRC)

b. When talking about one's professional or educational back-ground, one can either present the different stages from the point of view of t0 or shift the focus to a past moment in time as soon as one particular school has been mentioned. In the latter case, situations happening later will be represented as predictions, i.e. events that are W-posterior to one's arrival at the school. This technique makes the narration more vivid. It is the situation one is in (letter of application, application interview, informal talk with friends) which determines whether the use of the past tense/conditional is appropriate or not:

(119) After a normal primary school education at [Name] (...) Infants & Boys Junior Schools, I entered [Name] Grammar School at which I gained {"would gain") 9 Ό' Level passes in June [2222] and 3 Ά' Level passes (...) in June [2222], (SEU)

(120) I enclose a curriculum vitae, from which you will see that I spent [word] years at [Name] universities, working for my Masters and Doctorate; after which I taught {"would teach") for a year at [Name] University (...) and for a further year at the University of [Name] (...). (SEU)

c. The same can be said about medical reports: one can represent the different stages of the patient's evolution from the temporal standpoint of t0, or shift the focus to one particular event and present all/some of the subsequent situations as W-posterior to that particular


event. As stylistic variation is probably not always appreciated in prose of this type, it seems more natural to use the past tense:

(121) I gave her a short prescription for Ponstan Forte which helped {"would help") a bit. (adapted from SEU)

At the stage when the speaker prescribed Ponstan Forte, he did not know for sure whether the medicine would be effective or not. The conditional represents the RC situation as a prediction that has come true. The RC may also be given a RST reading: which (I argued) would help a bit. Examples (122) and (123) show the effect of the person of the subject (cf. section 3.2.1.):

(122) He blamed a knee injury in [2222] for this although he did not have any pain until [2222] for which he was {"would be referred") later referred to St. Name's Hospital and given a course of physiotherapy. (SEU)

(123) (...) She has some medial ligament tenderness which I have injected and she also developed a troublesome left tennis elbow which I also treated (? "would treat") with local steroid and I have arranged to see her again, (adapted from SEU)

In this type of text one will be more inclined to use the past tense when the subject is I or we, as one is talking about one's own past. When talking about someone else, one is at a greater distance from the facts; this allows a freer choice whether or not to represent situations as facts. Accordingly, the use of the conditional in (123) may be questionable because the speaker is talking about himself. It is strange, when giving a description of a patient's evolution in which facts are important, to present one's own initiatives as predictions which have come true. It is easier to do so when talking about someone else's initiatives or when it is not specified who is at the origin of the event (e.g. by using the passive (122)).

d. In the following context, the situations follow each other closely in time, which probably even leads to the ungrammaticality of the conditional:

(124) He had been listening to the exchange with a terrifying frown of concentration, and on hearing his name men-tioned he uttered a low "Ur" of approval and pleasure which made (* would make) little Miss Umbrage hold her

(W-posteriority in past sector (N)RRCs 129

whalebone umbrella in a tighter grip, and made the honest inspector blanch. (SEU) (RRC)

The use of a conditional tense requires there to be a certain lapse of time between the past tense situation and the conditional tense situation. This condition is not met in (124). The following examples contain continuative NRRCs (Jespersen 1961, III: 105):

(125) Sue wasn't sure whether to go or not - however, Peter arrived with a ticket which he gave (* would give) her & she didn't like to <- /refuse-> it, so I hope they'll enjoy it. (SEU) (NRRC)

(126) He half-rose and gave an expansive gesture with his hand, which overturned (* would overturn) the teapot, pouring its contents neatly into Emma's lap. (SEU) (RRC)

(127) Farmer-trainer Marsh called a vet who found (" would find") both horses fit. (SEU)

(128) Dr. [Name] approached Mr. [Name], who also left {"would also leave") to join your company. (SEU)

A continuative NRRC can be paraphrased by means of "and he": there is reference to a series of events. In order to give the relevant information required from a NRRC, it is not necessary to explicitly represent the RC situations as being W-posterior to the respective SUPC situations. It is far more relevant to represent the RC situations as factual. The use of the conditional suggests the existence of an omniscient narrator who knows the outcome of the story but prefers to predict rather than present situations as plain facts. At the same time, the speaker gives the guarantee that the predictions will come true. If little time elapses between the two situations, the conditional tense can hardly be used at all (cf. (125) and (126)).

e. If, however, there is a (reasonably) long stretch of time between the SUPC and RC situation or if the RC indicates a gradual change or evolution, the conditional can more easily be used. The difference in contextual effect between the verb forms is again that the conditional predicts whereas the past tense represents the situation as a fact:

(129) So both sides accepted the compromise, which would lead {"led") to the first lifting of the minimum wage since a four-year law was enacted in 1977, raising the wage to $3. (WSJ) (NRRC)


(130) For example, we (...) recommended improvements which led {"would lead") to many building societies changing their practices. (SEU) (RRC)

Apart from the "historical narrative" option, a possible interpretation of (130) is that the RC is interpreted as a stretch of RST (cf. example (121)). The RC should then be read as which (we claimed) would lead to many building societies changing their practices. In other words, in this reading the RC situation is a prediction used as an argument at the time when the recommendation is made.

f. When the RC situation indicates a permanent, characteristic action of the NP referent, it is less likely that the conditional will be used:

(131) He selected the FROM BLOCK option which inserted (?? would insert) the text which he had saved to buffer using the CUT PASTE option, (adapted from SEU)

The past tense RC in (131) can be interpreted as referring to a "permanent characteristic". The past tense could also be said to express sloppy simultaneity (Declerck 1991a: 42), which would explain why it is not substitutable with the conditional tense.

g. The conditional may be used for stylistic effects in historical narrative. The conditional is often used to refer to facts we are assumed to be familiar with, i.e. we all know they happened before the moment of speech:

(132) General Peltz, the commander who would conduct the "Little Blitz" on London in 1944... (Brown: F 02071) (Tottie - Övergaard 1984: 155)

(133) ... such tribes as ... Zaporogian Cossacks (in whose islands along the lower Dnieper the Polish novelist Sienkiewicz would one day place With Fire and Sword) were just elements. (Tottie - Övergaard 1984: 155)

The effect of the conditional in (132) and (133) is slightly different from that in examples of the type given in (134) and (135); the former relate to events belonging to common knowledge and not to the personal history of an unknown individual. The conditional tense adds an element of vividness to the narration:


(134) Comments were invited from all Schools and Teachers of the University and, in the light of these, the Committee prepared their final report which was submitted {"would be submitted") to the Academic Council in May of this year. (SEU)

(135) In the afternoon we had a talk on Advertising from Mr Fleming, who came ("would come") under a lot of heavy fire in the ensuing discussion. (SEU)

The following (non-RC) example is an interesting one:

(136) Now England stood alone in her 'finest hour1. With an army half of whose arms and equipment had had to be left on the beaches of France she awaited invasion, and with a handful of fighter planes, Spitfires and Hurricanes, she challenged - and by the end of September had broken ("would have broken")- the power of Germany's Luftwaffe (W. Churchill, The man of the century) (Salkie 1989: 29)

The past perfect is the preferred verb form probably because of its factual character. As Salkie puts it: "The pluperfect emphasizes the termination and completeness of the breaking in a way that the conditional perfect does not" (Salkie 1989: 29).

3.4.4. RRCs vs. NRRCs: different tense system?

The conditional tense of an atelic and a zero-telic verb is not substitutable with the past tense in RRCs and NRRCs unless a W-posterior interpretation is dictated by the semantics of the sentence or pragmatic knowledge. When the past tense is used, the unmarked interpretation is one of simultaneity, both in RRCs and in NRRCs:

(137) a. All he had withheld from Bismarck was information and plans which would be of value to the Chancellor. (LOB) (RRC)

b. All he had withheld from Bismarck was information and plans which 'were' of value to the Chancellor. (RRC)

c. The remaining $30 billion would be raised over two years through industry bonds, which would be counted as part of the budget. (WSJ) (NRRC)


d. The remaining $30 billion would be raised over two years through industry bonds, which 'were counted as part of the budget. (NRRC)

(138) a. She greeted the man who would be nominated for the prize.92 (RRC)

b. She greeted the man who 'was nominated for the prize. (RRC)

c. She greeted John, who would be nominated for the prize. (NRRC)

d. She greeted John, who 'was nominated for the prize. (NRRC)

The following example indicates that the semantics of the sentence may overrule the CRCT:

(139) a. Mr. Guttman told one person familiar with the New York exchanges during the search for a replacement that he was looking for a president who would be "responsive to the needs of the membership and the board". (WSJ)

b. Mr. Guttman told one person familiar with the New York exchanges during the search for a replacement that he was looking for a president who "was" "responsive to the needs of the membership and the board".

The context indicates that the person in question has not been found yet; therefore, he cannot be responsive to the needs of the member-ship and the board yet. Accordingly, substituting the conditional for a past tense does not affect the temporal relations expressed.

RRCs and NRRCs differ when the RC situation is telic. Shift of domain NRRCs of this type are interpreted as forming part of a se-quence of situations (cf. (140d)). The corresponding RRC is more likely to be interpreted as referring to a W-anterior situation (cf. (140b)) (cf. chapter 7):

(140) a. She felt attracted to the man who would report her to the police. (RRC)

b. She felt attracted to the man who 'reported her to the police. (RRC)

c. She felt attracted to John, who would report her to the police. (NRRC)

( W-)posteriority in past sector (N)RRCs 133

d. She felt attracted to John, who "reported" her to the police. (NRRC)

However, apart from the difference just illustrated, there are no re-strictions of the kind that one option (i.e. temporal subordination or shift of domain) is exclusively used in RRCs or NRRCs only. The conditional tense has to be used in RRCs and NRRCs alike in inten-sional domains. In continuative NRRCs, the conditional can be used if it is compatible with the meaning associated with the conditional, i.e. narrative prediction.

To end this section, Hornstein's (1990) ideas about the use of tense in RCs will be considered in more detail. He argues that the SOT (Sequence of Tense) rule (i.e. temporal subordination) cannot apply between the SUPC and the RRC93 as the SUPC and the RC "are not neighbors", "the finite relative clause is not adjacent to the matrix, given that it is embedded within a noun phrase" (Hornstein 1990: 138). Accordingly, in Hornstein's opinion, the chronological order between the SUPC and the RC situation is indeterminate in (141):

(141) We spoke to the man who was crying. (Hornstein 1990: 138)

Even so, Hornstein adds in a footnote: "This is not to say that prag-matic effects will strongly prefer one reading over others. However, it is not necessary to interpret the relative clause as contemporane-ous" (Hornstein 1990: 221). In the same way, Hornstein goes on to say that what we call temporal subordination is occasionally pos-sible: "This does not preclude the possibility that linking can option-ally take place and does so in those cases in which the temporal order of the two events is fixed" (Hornstein 1990: 139). In other words, a present tense might be argued to be embedded in the SUPC past tense in (141). However, Hornstein argues that one is likely to run into difficulties if one adopts the approach that a RC may occasion-ally exhibit SOT (especially in sentences with adverbials), witness the fact that (142) is unacceptable:94

(142) * John has spoken to the man who would win. (Hornstein 1990: 139)

Hornstein's example (143) constitutes an apparent counterexample, as the RC situation is temporally subordinated to the main clause sit-uation:

134 The expression of temporal relations in past sector (N)RRCs

(143) A man who would be king was born. (Hornstein 1990: 141)

However, the author argues that SOT is not really involved in (143), as the RC situation lies before t0. In Hornstein's opinion, examples with this kind of W-posterior situation should not be taken into consideration. If the RC situation lies after t0, the conditional tense cannot be used:

(144) ?? A woman who Fred would marry in a week arrived yesterday. (Hornstein 1990: 142)

However, it is not difficult to find counterexamples to Hornstein's claim, i.e. RCs with would that refer to situations that lie after tQ:

(145) a. He mentioned something he would do tomorrow, but I don't remember what it was.

b. John talked to someone who would be in New York tomorrow.

c. Bill sought a man who would be leaving. (Dowty 1986: 42)

As pointed out before, the conditional tense is not often used to refer to a situation which lies after t0. However, this probably follows from the conversational principle that a speaker conforming to the rules of communication refers to situations of that type by means of the fu-ture tense, which explicitly indicates that the situation is located after t0. The observations above together with the previous sections war-rant the conclusion that Hornstein's claim is based on too few exam-ples and cannot be used as a trustworthy hypothesis. In his review of Hornstein (1990), Dahl (1992) arrives at a similar conclusion: "This [the idea that temporal subordination is not possible in W-posterior RCs] is not backed up by any large amount of supporting evidence, and as far as I can see it is contradicted by examples such as John said that he was looking for someone who would help him, where would in the relative clause must have an R that is given by the Ε of the matrix verb said" (1992: 649).



The investigation into the expression of W-anteriority and W-poster-iority in past domain RCs has revealed the following facts:

a. The possibilities which RRCs and NRRCs allow do not differ: temporal subordination is not restricted to RRCs and the possibility of shifting the domain is not restricted to NRRCs.

b. When the pragmatics of the situation, the lexical contents of the clauses, or adverbials do not convey information about the chrono-logical order between the SUPC situation and the RC situation, it is still possible to use an absolute past tense to refer to a W-anterior event provided (a) the RC is restrictive and (b) the past tense refers to a telic bounded situation that cannot be interpreted as an un-bounded situation (i.e. the past tense situation is purely telic). NRRCs of the type just mentioned differ from their restrictive coun-terparts in that they are likely to be interpreted as referring to a W-posterior situation. RRCs and NRRCs which are not purely telic are subject to the same constraints. The difference in effect between mutually substitutable verb forms will be dealt with in chapter 6.

c. When the pragmatics of the situation, the lexical contents of the clauses, or adverbials do not indicate the chronological order be-tween the SUPC situation and the RC situation, it is still possible to use a past tense to refer to a W-posterior event provided (a) the RC is non-restrictive and (b) the past tense refers to a telic bounded situa-tion that cannot be interpreted as an unbounded situation (i.e. the past tense situation is purely telic). RRCs of the type just mentioned differ from their non-restrictive counterparts in that they are more likely to be interpreted as referring to a W-anterior situation. RRCs and NRRCs which are not purely telic are subject to the same constraints.

d. As the "function" performed (i.e. the temporal relation that is in-dicated) by a past tense referring to a W-simultaneous RRC situation is no different from that of a past tense in a W-simultaneous NRRC, a differential approach to the status of the past tense in NRRCs and RRCs should be questioned; the evidence for the claim that the past tense is an absolute tense in a NRRC whereas it is a relative tense in a RRC is not completely watertight, (section 3.3.)

e. Because of the difference in function of RRCs and NRRCs, verb forms may differ as regards their acceptability, even when there is no difference in the chronological order established between the (N)RRC situation and the SUPC situation (cf. section 3.2.2.3.).

f. When the pragmatics of the situation, the lexical content of the clauses or adverbials indicate the chronological order between the

136 The expression of temporal relations in past sector (N)RRCs

situations, it is possible to replace the past perfect by the past tense to refer to a W-anterior situation, both in RRCs and in NRRCs, al-though the CRPP occasionally overrules the effect of the above-mentioned factors conveying information about the chronological order between the situations.

Chapter 4 The expression of (W-)anteriority in pre-present sector RCs

Considerable attention has been paid to the present perfect in English linguistics; the fact that the English present perfect, unlike the present perfect in many other languages, is not compatible with a definite time expression, has resulted in a genuine quest for the true semantic nature of the present perfect.95 However, -apart from a few scattered remarks-961 have come across no books or articles touching upon the question of how temporal relations are linguistically realized in present perfect sentences. "How does one express that a situation is W-anterior to, W-simultaneous with or W-posterior to a present perfect situation?", a seemingly self-evident question, has not been considered in detail so far by linguists other than Declerck (1991a).

In section 4.1. below, Declerck's ideas on the expression of tem-poral relations in present perfect sentences are recapitulated. A detailed analysis of how W-anteriority is expressed will provide a basis for formulating possible constraints on the different options (i.e. temporal subordination, shift within the sector, shift to the past sector) (section 4.2.). As pointed out in chapter 2 (section 2.2.), the focus will mainly be on W-anteriority because an analysis of W-posteriority and W-simultaneity would not substantially enrich the findings. Apart from calling for a number of refinements to the pre-present tense system (section 4.3.), the investigation will make it possible to answer the question whether RRCs and NRRCs differ in respect of the use of tense (section 4.4.).

4.1. The use of tense in the pre-present sector: Declerck (1991a)

In order to discuss the tense system used in pre-present sector RCs, a distinction must be made between a domain established by a perfect which gets an indefinite interpretation and one established by a perfect which gets a continuative interpretation.

In the former case, a situation which does not last up to t0 is located in the pre-present sector. When other situations are temporally

138 The expression of (W-)anteriority in pre-present sector RCs

subordinated to it, there is a shift of perspective to the past sector,97

i.e. the tenses used to express temporal relations are the same as those used in a past domain (cf. Declerck 1991a: 28-29). This means that the past perfect is used to indicate that a situation is anterior to the pre-present sector situation, as in (1), the past tense is used to indicate that a situation is simultaneous with the pre-present sector situation, as in (2), and the conditional tense is used to indicate that the situation is posterior to the pre-present sector situation, as in (3):

(1) a. I have never denied that I had used that money. (Declerck 1991a: 29)

b. The sequence of events has shaken the confidence of stock traders who had assumed Sir James has some strategy up his sleeve for coping with the insurance problem. (WSJ)

(2) The Egyptian leader has died at a most awkward moment for the Russians, who were trying to consolidate their position in the area. (McCoard 1978: 82 in Declerck 1991a: 29)

(3) a. I have never promised that I would help you. (Declerck 1991a: 29)

b. They have arrested the pop group that would repre-sent Belgium in the Eurovision song contest.

Apart from temporal subordination, it is possible to shift the domain in the following cases:

a. The new situation is W-anterior to the SUPC binding TO but is not temporally subordinated to it; an indefinite present perfect or an absolute past tense is used, as in examples (4) and (5) respectively. In the former case, there is a shift of domain within the pre-present sector: "the speaker may wish to represent a relation of some kind between the SC situation and t0. He may, for example, wish to rep-resent the SC situation as still relevant to the structure of the world at t0. In that case he does not relate the SC situation to the head clause situation-TO but rather relates it to t0" (Declerck 1991a: 30).98 In the latter case, there is a shift of domain to the past sector:

(4) The doctor has already confirmed that Bill has sprained his ankle. (Declerck 1991a: 30)

(5) The doctor has already confirmed that Bill sprained his ankle yesterday. (Declerck 1991a: 31)

The use of tense in the pre-present sector: Declerck (1991a) 139

The shift may also be to the present sector, as in (6), or the post-present sector, as in (7):

(6) There have been times in my life when I required soothing, and then I have felt that a whiff of tobacco stills and softens one like a kiss of a little child. (Poutsma 1926: 259 in Declerck 1991a: 31)

(7) The doctor has already confirmed that Bill will be unable to walk to school for some time because he has sprained his ankle. (Declerck 1991a: 31)

b. If the indefinite perfect situation establishing the domain consists of several subsituations, the speaker may look upon the situation as a whole as leading up to t0 or he may concentrate on the different subsituations which lie completely before t0. If he apprehends the sit-uation in the former way, he will use the present perfect rather than the past tense to refer to a W-simultaneous situation:99

(8) a. I have often cried when I have felt / was feeling lonely. (Declerck 1991a: 30)

b. Mr. Baird hasn't always been well-liked by defense lawyers with whom he has dealt {dealt). (WSJ)

c. When discussing the semantics of the present perfect, Declerck draws attention to examples of the following type:100

(9) a. Every day, when I arrive at 12, Mary has been practising the piano since nine o'clock. (Riviere 1980: 127 in Declerck 1991a: 350)

b. Every time I have seen John, he has been ill. (Declerck 1991a: 350)

c. "Each time they have been refused by people who have never been to Warwick," he added. (LOB)

There is a shift of perspective, i.e. the time clause functions as if it were t0: "Sentences like [9a-b] differ from other perfect sentences in that they involve an intermediate type of TO, in their temporal schema: whereas the present perfect normally represents TOsit as an indefinite interval within or throughout a TE-up-to-L, the present perfect in [9a-b] expresses that there are a number of TOsits which are each indefinite intervals within or throughout a TE reaching up to an


(each time different) TO,, and that the latter TO.s are themselves indefinite intervals in an implicit TE-up-to-t0" (Declerck 1991a: 351).

The tense system in pre-present domains established by a continu-ative present perfect will not be dealt with; in examples of that type, one does not have the choice of shifting the domain or not. The domain includes t0 and the tenses used to express temporal relations cannot be but absolute, i.e. t0 functions as binding TO;101 the present tense (cf. (10a)) and the continuative present perfect (cf. (10b)) are the absolute tenses establishing a W-simultaneous domain. When the newly introduced situation is W-anterior to a continuative perfect situation, there is a shift of domain within the pre-present sector (i.e. an indefinite present perfect is used), as in (11), or a shift to the past sector, as in (12). A W-posterior situation is referred to by means of the future tense (i.e. there is a shift to the post-present sector), as in (13):

(10) a. Up to now he has always maintained that he works in a bank. (Declerck 1991a: 32)

b. Since then I've felt I've missed some vital experi-ence. (Vermant 1983: 73 in Declerck 1991a: 32)

(11) I have known for some time that Jim has sprained his ankle. (Declerck 1991a: 32)

(12) I have known for some time that he did not do it. (Declerck 1991a: 32)

(13) Ever since this morning he has repeated that he will move to London. (Declerck 1991a: 33)

Sentences (4) to (8) exemplify the shift of domain options the pre-present sector allows. In the sections below, the following issues will be discussed:

a. Is there a free choice whether or not to temporally subordinate a RC situation which is W-anterior102 to the TO established by the indefinite present perfect? If not, what are the constraints on the choice of tense (section 4.2.)? The results of this investigation will indicate that the tense system in the pre-present sector as presented by Declerck (1991a) should be slightly modified (section 4.3.).

b. Is there a difference between pre-present sector RRCs and pre-present sector NRRCs as regards the tense system used (section 4.4.)?

Constraints 141

4.2. Constraints on the expression of (W-)anteriority in pre-present sector RCs

4.2.1. Temporal subordination not possible

The pre-present sector differs from the past sector in that the so-called unmarked form, i.e. the temporally subordinated form, is not always acceptable:

4.2.1.1. Links with t0

The RC situation may be mentioned because of its links with the present rather than because it is W-anterior to the SUPC situation:

(14) a. The Indian planners have been criticized for the rigidity of their plans and the emphasis which has been given to Government investment in urban and already industrialised sectors. (LOB) (RRC)

b. The Indian planners have been criticized for the rigidity of their plans and the emphasis which 'had been given' to Government investment in urban and already industrialised sectors.

It is no longer certain that the Indian planners still emphasize gov-ernment investment in industrialised sectors when the past perfect is used in the RC, whereas the present perfect implicates103 that they still do. This difference can also be observed in (15):

(15) a. Talks have already been held between management and unions to halt the dispute, which has stopped production. (SEU) (NRRC)

b. Talks have already been held between management and unions to halt the dispute, which ?? had stopped production.

In (15), the present perfect in the RC implicates that the employees are no longer at work. It explains why talks have been held. The past perfect strongly suggests that the people have already gone back to work. This interpretation will probably be ruled out in (15b) because it is pragmatically unsound; it seems strange to try to halt a dispute, the effects of which are no longer visible (as people may have gone


back to work). The speaker chooses to mention the fact that talks have been held, which is likely to imply that no solution has been found. If this had been the case, the speaker would not have observed the Maxim of Quantity104 because the news that a solution has been found is more informative than the fact that talks have been held. If successful were added to the sentence, the past perfect would be acceptable:

(15) c. Successful talks have already been held between management and unions to halt the dispute, which had stopped production.

Example (16) is similar:

(16) a. But he has criticized the electoral system which has left the Alliance trailing without a seat despite a 20% share of the vote. (SEU) (NRRC)

b. But he has criticized the electoral system which ?? 'had left' the Alliance trailing without a seat despite a 20% share of the vote.

The present perfect, unlike the past perfect, implicates that the Alliance does not have a seat at t0. The past perfect could be argued to make sense because one can criticize something (the electoral system) for the effects it had in the past (it left the Alliance trailing without a seat). On this reading, there is a link not between the present criticism and a present situation but between the criticism and a past situation. However, there is no reason to represent the RC situation as W-anterior to the SUPC situation instead of W-anterior to t0; it is because of a present situation that the criticism arises. There is a similar difference between the present perfect and the past perfect in (17):

(17) a. They have explained to me the rationale which this work has followed, (adapted from SEU)105 (RRC)

b. They have explained to me the rationale which this work ?? had followed.

If someone follows a particular rationale when writing a book, it will be possible to recover it at any time, as it is inalienable. The use of this (this work) implies that the speaker is concerned with the here-and-now and makes it impossible to use the past perfect.

Constraints 143

It may also be the SUPC's strong links with t0 which block a shift of perspective to the past sector, as in (18):

(18) a. I made a note of the address to which to write, but unfortunately I have now mislaid the copy of the T.H.E.S. in which the advertisement has appeared. (SEU) (RRC)

b. I made a note of the address to which to write, but unfortunately I have now mislaid the copy of the T.H.E.S. in which the advertisement * had appeared.

The speaker is concerned with now in the SUPC. The lack of a plausible T02 shows that there is no shift of perspective and explains why the past perfect cannot be used in the RC.

In the following example as well, the present perfect is preferred to the past perfect:

(19) a. Reconstruction work on the Treasury offices in Whitehall, which has been going on for some months behind masses of scaffolding on the street side and high wooden fences on Horse Guards Parade, has surprised the Ministry of Works by the amount of Tudor brickwork it has revealed. (LOB) (RRC)

b. Reconstruction work on the Treasury offices in Whitehall, which has been going on for some months behind masses of scaffolding on the street side and high wooden fences on Horse Guards Parade, has surprised the Ministry of Works by the amount of Tudor brickwork it ?? 'had revealed

Whenever a past perfect is used, one looks for a situation that can function as T02, to which the past perfect situation is anterior. The then (T02) we are looking for cannot be the moment when the ministry were surprised but must lie before that moment. Accord-ingly, the meaning of the sentence changes: it is as though the hidden layer of Tudor brickwork which was revealed the moment the workers started taking away the surface of the building has been covered again and can no longer be seen.

It is sometimes difficult to determine clearly whether it is the links between the SUPC situation and t0 (through the use of the present


perfect) which require the RC situation to take t0 as TO, as in (18), rather than the links between the RC situation and t0, as in (14) up (17). In (19), for instance, it is much more difficult to indicate accurately which of these two factors leads to the preference for the present perfect.

If the SUPC situation consists of several subsituations, the speaker can choose between two alternatives: (a) he can focus on the subsituations or (b) he can look upon the whole situation as leading up to t0 (cf. Declerck 1991a: 29-30):

(20) a. In the course of considering the reactions of Schools and individual Teachers to their Memorandum on the Reorganisation of the University, the Steering Committee have been impressed by the frequency with which attention has been drawn to the im-portance of providing opportunities to enable a greater proportion of the members of the University to participate in its government. (SEU) (RRC)

b. In the course of considering the reactions of Schools and individual Teachers to their Memorandum on the Reorganisation of the University, the Steering Committee have been impressed by the frequency with which attention ? had been drawn to the importance of providing opportunities to enable a greater proportion of the members of the University to participate in its government.

The subsituations making up the RC situation lead up to t0 or to the SUPC situation when the present perfect is used. In other words, the steering committee have been impressed at a certain moment in time by a phenomenon which possibly continues -after their being impressed- until t0. One tends to think of the many occasions on which the RC situation held.106 Although the theory predicts that the (unmarked) tense to be used is the past perfect, a shift of perspective is not likely to take place. The SUPC's relatively strong links with t0 result in there being no TO which can function as T02 for the past perfect. This explains why the acceptability of the past perfect is questionable.

Constraints 145

4.2.1.2. Adverbials

The use of the past perfect is sometimes ruled out because it is incompatible with certain adverbs used in the sentence:

(21) a. Have rung Ma who got back from Painswick today. (adapted from SEU) (NRRC)

b. Have rung Ma who * had got back from Painswick today.

In (21), the presence of today leads to the rejection of had got. In (22), the adverbial formerly can theoretically be anchored to t0 or to a past TO. The fact that the past perfect can hardly be used indicates that the adverbial is anchored to t0 and consequently that the shift of perspective in the main clause which should provide a past sector binding TO has not taken place:

(22) a. Toshiba Corp. said its new French marketing con-cern has started operating under the aegis of the company's West German subsidiary, which formerly handled all sales of Toshiba electronic products in France. (WSJ)

b. Toshiba Corp. said its new French marketing con-cern has started operating under the aegis of the company's West German subsidiary, which ? had formerly handled all sales of Toshiba electronic products in France.

Example (23) brings up a new element:

(23) a. You have claimed for all necessary travelling expenses which you have had to pay in the course of your duties, (adapted from SEU) (RRC)

b. You have claimed for all necessary travelling expenses which you ?? had had to pay in the course of your duties.

It is probably not the adverbial as such which leads to the rejection of the past perfect. The presence of the second person singular, i.e. someone directly involved in the conversation, establishes strong links with t0. Consequently, t0 will function as binding TO for the RC


situation. Accordingly, in (23c), in which the third person singular is used, the past perfect is an acceptable verb form:107

(23) c. She has claimed for all necessary travelling expenses which she has had (had had) to pay in the course of her duties.

The common feature of the examples in which the past perfect cannot be used is that there is no T02 available. In other words, the shift of perspective (from the fact that something happened to the time when it happened), which should provide a past sector T02, does not always take place. As a result, temporal subordination is not always possible in the pre-present sector. It is in this respect that the tense system as presented by Declerck (1991a) will have to be modified (section 4.3.).

4.2.2. Shift to past sector not possible

The examples analysed in this section do not allow a shift to the past sector to refer to a W-anterior situation, whereas a shift within the pre-present sector is possible.

4.2.2.1. Constraints on the Replacement of the Present Perfect (CRPrP)

The past tense may not be acceptable in a particular context because it leads to an interpretation in terms of W-simultaneity instead of W-anteriority. The group of verbs resulting in such a different interpretation is the same as that identified in section 3.2.1. of chapter 3 in connection with the (non-)substitutability of the past tense and the past perfect.

Constraints on the replacement of the present perfect (CRPrP) (I) The replacement of a non-progressive present per-

fect that refers to a bounded atelic situation by a past tense has the effect of representing the situation as unbounded when used in the past tense (provided there is no adverbial with a bounding effect) (cf. examples (24) to (28))

Constraints 147

(II) Bounded non-progressive perfect zero-telic situa-tions allow two interpretations when they are put in the past tense (cf. examples (33) and (34)): (a) a bounded telic interpretation, (b) an unbounded atelic interpretation. These non-progressive present perfect situations usually do not allow the shift to the past sector, because the use of the past tense changes the chronological order between the situations.

It should immediately be added that the CRPrP does not apply as strictly as the CRPP, the reason being that the SUPC situation in the present perfect does not automatically function as binding TO for the RC situation. As has been established in the previous section, the shift of perspective from the pre-present sector to the past sector which in Declerck's model is said to be typical of indefinite present perfect sentences does not always take place. From this it follows that a past tense RC which refers to an unbounded situation will not automatically be interpreted as bound by the SUPC situation; it can also be understood as an absolute tense or as bound by another TO. In the latter cases, the temporal relation of W-anteriority between the SUPC situation and the RC situation will not be blurred. These observations imply that confusion will arise only when the SUPC situation can function as binding TO for the RC situation.

The following examples illustrate the CRPrP: a. Atelic situations:

(24) a. We have been struck by the extent to which working class people have improved their lot, acquired more power and more possessions; we are likely to be even more impressed by the degree to which they no longer feel themselves members of the 'lower orders', with a sense of other classes, each above them and each superior in the way the world judges, (adapted from SEU) (RRC)108

b. We have been struck by the extent to which working class people 'improved their lot, acquired more power and more possessions; we are likely to be even more impressed by the degree to which they no longer feel themselves members of the 'lower orders', with a sense of other classes, each above them and each superior in the way the world judges.


In (24), the past tense, unlike the present perfect, represents the RC situation as an activity that is simultaneous with the SUPC situation. When the past tense is used in the RC, the activity of improving one's lot is going on at the time when the SUPC subject is struck by it. It is crucial to remark that the above interpretation is the unmarked one arising out of context. The past tense is substitutable with the present perfect if the context indicates that improved refers to a W-anterior situation. The following present perfect RCs also describe bounded states (cf. (25) and ((26)) and activities (cf. (27)). Changing the present perfect in the RC into a past tense makes the situation unbounded and results in a change in temporal interpretation:

(25) a. Jane too has studied the attitude of people who have belonged to a rightist organization. (RRC)

b. Jane too has studied the attitude of people who 'belonged to a rightist organization.

(26) a. New Jersey's own highest court remains a liberal bulwark against major restrictions on abortion, but the U.S. Supreme Court ruling, Webster vs. Mis-souri, has engaged voters across the nation who have been insulated from the issue, (adapted from WSJ) (RRC)

b. New Jersey's own highest court remains a liberal bulwark against major restrictions on abortion, but the U.S. Supreme Court ruling, Webster vs. Mis-souri, has engaged voters across the nation who 'were insulated from the issue.

(27) a. Pinturischio has impressed so much by the way in which he has done his work rather than what he has done in it, that I fully expect him to lay a solid foundation to his claims to be considered the Derby winner by scoring on the first occasion he faces the racecourse's acid test. (LOB) (RRC)

b. Pinturischio has impressed so much by the way in which he 'did his work rather than what he did in it, that I fully expect him to lay a solid foundation to his claims to be considered the Derby winner by scoring on the first occasion he faces the race-course's acid test.

The RC situation in the three examples just given is represented as W-anterior to the SUPC situation when a perfect tense is used and as

Constraints 149

a simultaneous state/activity when the past tense is used. However, in (27b) a past progressive tense form will preferably be used to indi-cate simultaneity. Indeed, in some examples the past tense is not acceptable for two reasons: first, it changes the original relationship of W-anteriority into one of W-simultaneity; second, even if this temporal order were compatible with the context, a progressive form would normally have to be used.

As pointed out when formulating the general principle, the CRPrP does not apply as strictly as the CRPP, because the SUPC situation does not always function as binding TO, which results in the past tense in the RC being interpreted as an absolute tense that refers to a W-anterior situation. The following examples illustrate the point made:

(28) a. The sequence of events has shaken the confidence of stock traders who had assumed Sir James has some strategy up his sleeve for coping with the insurance problem. (WSJ)

b. The sequence of events has shaken the confidence of stock traders who "assumed" Sir James has some strategy up his sleeve for coping with the insurance problem.

Had assumed clearly communicates they the stock traders gave up their belief in Sir James as a result of what happened. Theoretically speaking, assumed could be bound by the main clause situation. However, the semantics of the sentence are likely to safeguard a cor-rect interpretation: as it is explicitly indicated they no longer believe in Sir James, the RC situation will not be understood as simultaneous with the main clause situation, but rather as containing an absolute past tense that refers to a situation that is W-anterior to the main clause situation. The following example is another case in point:

(29) a. He has destroyed the Watson theory, which has been promoted by a rival, (adapted from SEU) (NRRC)

b. He has destroyed the Watson theory, which "was promoted" by a rival.

The present perfect in the RC in (29a) should probably be understood as referring to a period of time which is W-anterior to the SUPC situation, i.e. as an indefinite perfect. A continuative present perfect


reading seems to be ruled out for reasons of pragmatic feasibility: it seems difficult to refer to a theory as being destroyed at an indefinite past moment (SUPC situation) and being promoted until then (or until t0) at the same time (RC situation), which is what would be suggested by the continuative interpretation of the present perfect RC tense. Moreover, a non-progressive present perfect is not usually interpreted as continuative unless there is a time duration adverbial present (cf. Declerck 1991b: 103). In any case, in the indefinite reading, the present perfect is interpreted as referring to a situation which is W-anterior to the SUPC situation, whereas the past tense may be said to represent the RC situation as W-simultaneous with the SUPC situation. The latter reading might be argued to be pragmatically unacceptable as well for reasons similar to those explaining the rejection of a continuative reading of the present perfect in the RC. The CRPrP predicts that the past tense modifies the chronological order between the situations. However, the form remains acceptable; the fact that the theory was promoted by a rival may be considered to be relevant information in itself. In this case, the speaker does not consider the exact temporal location of the RC situation (i.e. as W-anterior to the SUPC situation) particularly rele-vant and may therefore choose to use the past tense (cf. chapter 6, section 6.3.). Another explanation for the relative acceptability of the past tense runs along the following lines: as the strict simultaneity reading is pragmatically ruled out, it is rejected and the past tense is reinterpreted as an absolute past referring to a situation lying completely before the SUPC situation. In this interpretation, the use of the present perfect in the main clause does not result in a shift of perspective. From this it follows that there is no TO available with which the RC past tense can express simultaneity. This may be the main reason why the past tense is reinterpreted as establishing a W-anterior domain of its own. Similarly, in (30), the SUPC situation does not function as binding TO for the past tense in the RC:

(30) a. The drug has a three-dimensional structure similar to that of DES, the anti-miscarriage drug that has been linked to cervical and vaginal cancer in some of the daughters of the women who have taken it.

b. The drug has a three-dimensional structure similar to that of DES, the anti-miscarriage drug that has been linked to cervical and vaginal cancer in some of the daughters of the women who "took" it. (WSJ)

Constraints 151

The past tense in (30b) is interpreted as an absolute tense referring to a period of time that lies before the main clause situation.

In some cases, the past tense is not unacceptable because the simultaneity relation it expresses is only sloppy (especially if the RC has conditional connotations), i.e. a tense which usually expresses simultaneity is used to refer to a situation that is W-anterior to the binding TO (cf. footnote 62):

(31) a. Grants have only been issued to students who have produced valid College identity cards, (adapted from SEU) (RRC)

b. Grants have only been issued to students who "produced" valid College identity cards.

The acceptability of the past tense in this example (with an atelic verb) is partially due to the fact that the RC defines a class of people by ascribing to them a certain characteristic which is not precisely located in time. Moreover, the above RC expresses a condition, which often results in the use of a verb form expressing sloppy simultaneity. A similar remark applies to (32):

(32) a. The technique of historical criticism has been applied to anything that has occurred to which any importance was attached, (adapted from SEU) (RRC)

b. The technique of historical criticism has been applied to anything that "occurred" to which any importance was attached.

The past tense might be argued to be unacceptable because it changes the W-anteriority interpretation of the present perfect situ-ation into a simultaneity interpretation. However, the use of the past tense in the RC may also be interpreted in terms of sloppy simultaneity; although, strictly speaking, the RC situation is W-anterior to the SUPC situation, it is represented as W-simultaneous. Moreover, in this example, the RC situation makes it clear that the SUPC situation consists of several subsituations. In this case, the speaker can either express W-anteriority by means of the present perfect or he may express sloppy simultaneity with each of the sub-situations. However, one native speaker pointed out that exactly because of the repetitive reading, he preferred the use of the present perfect tense.


b. Zero-telic situations

The situations in the RCs in (33) and (34) are represented as zero-telic:

(33) a. They have attacked the trade unions, whose restrictive practices have prevented a price reduc-tion, and are deceiving the public by saying this Bill will reduce the cost of living, (adapted from SEU) (NRRC)

b. They have attacked the trade unions, whose restrictive practices 'prevented a price reduction, and are deceiving the public by saying this Bill will reduce the cost of living.

The present perfect in the RC in (33a) (have prevented a price reduction) may be interpreted as an indefinite or continuative per-fect; in the former reading, the restrictive practices no longer hold, have prevented suggesting a relationship of W-anteriority between the RC situation and the SUPC situation. In the latter reading, they continue until the time of the SUPC situation or t0 and still prevent a price reduction. The unmarked interpretation of the past tense is that the RC situation is atelic unbounded; it suggests that the restrictive practices held at the moment when the trade unions were attacked (simultaneity with the SUPC TO). It does not tell us whether they still hold or not. The following example is similar:

(34) a. In addition, several clinics have begun treating patients whose corneas have been distorted by infections, accidents, glaucoma or other pathological conditions. (WSJ) (RRC)

b. In addition, several clinics have begun treating patients whose corneas 'were distorted by infec-tions, accidents, glaucoma or other pathological conditions.

Have been distorted in (34a) refers to the telic bounded event of getting distorted. The passive form in (34b) allows two readings: a telic bounded one, as in (34a), or an atelic unbounded one, in which case the RC situation refers to the state of being distorted. In the latter case, the RC state (whose corneas were distorted at that time) is understood as W-simultaneous with the SUPC situation.

Constraints 153

c. Purely telic situations

As in the case of the CRPP, changing a present perfect into a past tense does not always result in a different temporal location of the RC situation. The CRPrP applies only to indefinite non-progressive present perfect forms referring to atelic and zero-telic situations (not accompanied by an adverbial which indicates the chronological order) and not to purely telic ones. This explains why the past tense and the present perfect are mutually substitutable in (35) and (36):109

(35) a. She has gone out with Jonathan Clark, who has killed his former mistress,

b. She has gone out with Jonathan Clark, who "killed" his former mistress.

(36) a. She has discussed the problem which Mr. Jones has alluded to.

b. She has discussed the problem which Mr. Jones "alluded to".

4.2.2.2. Anchoring time missing

The past tense can sometimes not be used to refer to a W-anterior situation because there is no anchoring time present:

(37) a. We have focussed in particular on the complex of technological changes which has been termed the Green Revolution, (adapted from SEU) (RRC)

b. We have focussed in particular on the complex of technological changes which ?? was termed the Green Revolution.

The sentence with a past tense is odd because the SUPC does not provide a past anchoring time. If we add an adverb which can per-form that function, the sentence is acceptable:

(37) c. We have focused in particular on the complex of technological changes that has occurred since the late 1960s, which at the time was termed the Green Revolution. (SEU)

The following example is similar:


(38) a. He has recovered some of the estate duty which has been paid, (adapted from SEU) (RRC)

b. He has recovered some of the estate duty which ?? was paid.

c. He has recovered some of the estate duty which was paid in May of this year.

It is not impossible to use the past tense in (38b) (i.e. it is not ungrammatical), but one seems to be waiting for the definite past time specification, which is not given.110 If the payment of the estate duty has already been mentioned in the context which precedes (38b), which implies that there is a binding TO available, the sentence with the past tense is acceptable.

In most examples of this type, the shift of domain to the past sector is the least preferable option except for (a) the RCs which simply give characterizing information without bothering to locate the situation exactly in time, as in (39), and (b) the RCs in which a definite time specification is given, as in (37c) and (38c):

(39) a. They have exhibited items which have been selected for their value and attractiveness, (adapted from SEU)

b. They have exhibited items which "were selected" for their value and attractiveness.

4.2.2.3. Links with t0

The past tense is not likely to be used in the following example because the links between the RC situation and t0 are so strong that a tense must be used which indicates that the RC situation has effects on the present:111

(40) a. But alongside Mr. Tosh's article the journal has published the experience of authorities who have lost all faith in corporal punishment, (adapted from SEU)

b. But alongside Mr. Tosh's article the journal has published the experience of authorities who lost all faith in corporal punishment.

Constraints 155

The present perfect RC implicates that the authorities do not believe in corporal punishment at t0. If the speaker attaches a lot of importance to this element of his message, he will not use the past tense, which leaves it vague how the authorities feel about corporal punishment now. In other words, the unacceptability is relative, by which I mean that it depends on the speaker's intention. If the speaker merely wants to indicate which kind of authority he is talking about, he will be less concerned to establish an exact location in time. The difference in meaning between the two forms remains, but the past form will no longer be unacceptable for the speaker, i.e. it does not express his message less adequately in this situation. However, even in this case, there may still be a preference for the present perfect because there is no anchoring time available for the past tense and because the verb is zero-telic and could accordingly be interpreted as referring to a W-simultaneous state.

4.2.2.4. Adverbials

The past tense is sometimes ungrammatical because it is incompati-ble with the adverbials used in the sentence:

(41) a. The committee has continued the process of con-sultation which it had already initiated, (adapted from SEU) (RRC)

b. The committee has continued the process of con-sultation which it * already initiated.

(42) a. We have relieved him from the pain in his right lower limb which he had had since [2222]. (adapted from SEU) (NRRC)112

b. We have relieved him from the pain in his right lower limb which he * had since [2222].

(43) a. We have focused in particular on the complex of technological changes that has occurred since the late 1960s, and which has been termed the Green Revolution, (adapted from SEU) (RRC)

b. We have focused in particular on the complex of technological changes that * occurred since the late 1960s, and which has been termed the Green Rev-olution.


4.2.3. Shift within pre-present sector not possible

In this section, I will concentrate on examples in which the past perfect is not substitutable with the present perfect.

4.2.3.1. Links with past TO

The present perfect cannot be used when the RC situation implicates resultativeness with respect to a past moment in time (i.e. the time indicated by the SUPC present perfect situation) and not with respect tot0:

(44) a. Men have departed for the front and returned from it. He has seen those who had come with them to say goodbye, and those who walked in agitation up and down the platforms, (adapted from SEU) (RRC)

b. Men have departed for the front and returned from it. He has seen those who * have come with them to say goodbye, and those who walked in agitation up and down the platforms.

In this example there is clearly a shift of perspective to the past sector; the speaker shifts his attention from the fact that a situation took place to the time when it took place and the people who were present on that occasion. The fact that the relevant persons accom-panied the men to the platforms cannot be conveyed by the use of the present perfect, which means they are present. In order to communicate the intended message (they were present), the past perfect has to be used. In the following example as well, the past perfect cannot be substituted for the present perfect:

(45) a. The Swedish economy has been unexpectedly buoyant. This has confounded most economic fore-casters, who had warned of a first-half downturn, (adapted from WSJ)

b. The Swedish economy has been unexpectedly buoyant. This has confounded most economic fore-casters, who ? 'have warned of a first-half down-turn.

Constraints 157

The use of the present perfect changes the meaning of the sentence: there is a certain degree of incompatibility between the meaning conveyed by the present perfect in the RC and the SUPC situation in the sense that since there has been an economic upsurge rather than a downward trend, the warning by the forecasters has already proved incorrect. Accordingly, it is not appropriate to use the present perfect because this verb form implicates that the economic forecasters still cling to the erroneous view that the economy is not doing very well, which is not very likely to be the case once they notice that the economy is successful. The past perfect conveys the meaning intended, i.e. the economic forecasters have had to abandon their view because, contrary to what had been predicted, the economy is buoyant.

It should be pointed out that the shift of perspective frequently occurs in present perfect sentences with heavy accent on have:

(46) He HAS attacked the journalist called John White, who had just taken (?? has taken) a picture of him.

4.2.3.2. Adverbials

The present perfect cannot be used sometimes, because it is incom-patible with certain adverbials used in the sentence:

(47) a. She has made a surprisingly good response, either to the Indomethacin which I had prescribed at her first visit, or to the fact that she has now started at a Special School and has been released from the stresses and strains associated with her previous school-phobia, (adapted from SEU) (NRRC)

b. She has made a surprisingly good response, either to the Indomethacin which I * have prescribed at her first visit, or to the fact that she has now started at a Special School and has been released from the stresses and strains associated with her previous school-phobia.

Have prescribed is ungrammatical in (47b) because the present per-fect cannot be used in combination with adverbials with definite past time reference.113 In (48), the present perfect is not compatible with the meaning conveyed by already:


(48) a. The committee has continued the process of con-sultation which it had already initiated, (adapted from SEU) (RRC)

b. The committee has continued the process of con-sultation which it * has already initiated.

The only meaning compatible with already in this sentence is at the time of the SUPC situation, they had already initiated the process of consultation and not at tff they have already initiated the process of consultation. If already were not included, the present perfect would be acceptable in the sentence. The same observation applies to the following examples:

(49) a. This private enterprise Emperor has inevitably fol-lowed the same primrose path that had already been blazed by other British long-range airliners designed since the war. (adapted from SEU) (RRC)

b. This private enterprise Emperor has inevitably fol-lowed the same primrose path that * has already been blazed by other British long-range airliners designed since the war.

(50) a. Thus, if devaluation has extricated the U.K. from the economic straitjacket that had previously encased it, the result has not been to make Britain's economic future look as promising as that of some of the more growth-conscious of the overseas £ area countries, (adapted from SEU) (RRC)

b. Thus, if devaluation has extricated the U.K. from the economic straitjacket that * has previously encased it, the result has not been to make Britain's economic future look as promising as that of some of the more growth-conscious of the overseas £ area countries.

It has become clear in section 4.2. that the domain cannot always be shifted within the pre-present sector or to the past sector to refer to a W-anterior situation. Therefore, constraints have had to be formulated on the use of the past tense (section 4.2.2.) and on the use of the present perfect (section 4.2.3.). Moreover, the pre-present sector differs from the past sector in that temporal subordination is not always possible, hence the constraints on the use of the past perfect (section 4.2.1.). Before answering the question whether or not

The tense system in the pre-present sector: some modifications 159

RRCs and NRRCs differ as far as the expression of W-anteriority in the pre-present sector is concerned (section 4.4.), the impossibility of temporal subordination in some pre-present sector sentences will be considered in more detail.

4.3. The tense system in the pre-present sector: some modifications

It is clear from the above discussion that the present perfect has to be used (i.e. temporal subordination is excluded) to refer to a W-anterior situation when the RC situation or the SUPC situation has clear links with tQ. The perspective is not shifted and tQ functions as binding TO for the RC situation. This happens especially when the indefinite perfect is not of the repetitive type or if have is not intonationally prominent. Indeed, the examples with temporal subordination often contain an indefinite perfect with heavy stress on have, as in (51a), or a form of have that can be paraphrased as it has happened that, as in (51b). Temporal subordination also occurs when the SUPC situation consists of several subsituations, as in (51c). It is more difficult to make generalizations about domains established by a perfect tense which gets an indefinite interpretation but which is not of the type exemplified in (51a) to (51c): the more elements there are (either in the RC or in the SUPC) which establish links with t0, the less likely it is that there will be temporal subordination (cf. (5Id) and (51e)). When the speaker does not concentrate on the links between the SUPC situation or the SC situation and t0, it is more likely that there will be temporal subordination (cf. (5If)):

(51) a. They HAVE exhibited items which had been selected for their value and attractiveness, (adapted from SEU).

b. We went into a fight like it was all a joke, but I've seen Fat beside a guy that'd been bumped off and cry like it was his own brother, (adapted from Defroment 1973: 77)

c. On different occasions, she has talked about the way in which he had treated her.

d. They have explained the rationale which this work has followed (?? had followed), (adapted from SEU)

e. I made a note of the address to which to write, but unfortunately I have now mislaid the copy of the

160 The expression of(W-)anteriority in pre-present sector RCs

T.H.E.S. in which the advertisement has appeared (?? had appeared). (SEU)

f. I have, in addition, read two papers which he has submitted (had submitted) for publication - one on [Name] and one on [Name]. (SEU)

In any case, one of the major points emerging from the discussion so far is that in the pre-present sector it is not always possible to temporally subordinate a RRC situation that is W-anterior to the SUPC situation, i.e. temporal subordination is not the unmarked option. Two possible conclusions can be drawn from this observa-tion:

Alternative 1: There is no "cross-sectorial" similarity as regards the expression of W-anteriority, i.e. the systems used in the past, post-present and pre-present sectors are not similar. Whereas temporal subordination is the unmarked option in the past sector (in the sense that it is always possible to use it), it is not in the pre-present sector. Alternative 2: Temporal subordination is the unmarked option as regards the expression of W-anteriority in the three (past, post-present and pre-present) sectors. This implies that Declerck's system of how temporal relations are expressed in indefinite present perfect sentences should be refined so that the cases in which the past perfect cannot be used are not exceptions114 but the consequence of a more intricate system dictating the use of tense in the pre-present sector.

a. On the one hand, it may seem more reasonable to take the second approach for the following reason: so far, it has been accepted that temporal subordination is the unmarked form expressing a relationship of W-anteriority. It is the unmarked option in the sense that it is always possible to temporally subordinate a certain situation. Occasionally, the domain can be shifted, though this possibility is constrained. In the discussion of the past sector and pre-present sector, I have tried to reveal the constraints underlying the possibility of shifting the domain. However, if we allow constraints on the use of the unmarked form, the theoretical construc-tion seems to be weakened: use of the terms marked and unmarked becomes dubious if there are constraints on both alternatives. I

The tense system in the pre-present sector: some modifications 161

therefore feel it would be wrong to weaken Declerck's claim that temporal subordination is always the unmarked option. Though it would account for RCs requiring the use of a shift of domain form, this solution has the drawback that the general rule "unmarked option - temporal subordination" vs "marked option - shift of domain" cannot be maintained. That is why it may seem more reasonable to refine the system by incorporating the following additional principles: * If the RC situation is of the "before and up to the SUPC situation" type or if the W-anterior RC situation (of the "completely before" type) has strong links with t0, the present perfect is usually used. It stresses the bonds between the RC situation and t0 (rather than those between the RC situation and the SUPC situation); temporal subordination is not possible. This is not just an option the speaker has (cf. Declerck 1991a: 30); he must use the present perfect if he wants to communicate the intended meaning. * If the SUPC situation has strong links with the present sector, the perspective is not often shifted to the past sector, i.e. there is no shift from "the fact that the situation held" to "the time when the situation held". In such cases, the present perfect is usually used in the RC.

b. On the other hand, it seems that there is indeed a fundamental difference between a shift of domain (within the sector) in the past sector and a shift of domain (within the sector) in the pre-present sector: * As far as the past sector is concerned, there are no links with t0 no matter whether the past perfect (temporal subordination) or the absolute past tense (shift of domain) is used in the RC; there is a clear break between the past and present sector. * Shifting the domain within the pre-present sector (choosing t0

rather than the pre-present TO as binding TO) may affect the inter-pretation of the sentence because there is a fundamental difference between the past perfect (temporal subordination) and the present perfect (shift of domain) as regards their relation with t0. When a situation is temporally subordinated, no links are established between the situation and t0. When the domain is shifted within the sector (i.e. when the present perfect is used), there are links between the situation and t0: there is no break between the present sector and the pre-present sector of the type that exists between the present and the past sector. From this point of view, the two alternatives can be fused into one; they are complementary in that it is the nature of the pre-present sector which accounts for the -at first sight "deviant"- use of tense in


sentences of this type. Just as there is no temporal subordination in domains established by a continuative perfect (because they include L and therefore all the temporal relations expressed take t0 as TO), there are certain environments with an indefinite present perfect in which the force of attraction of t0 is so strong that newly introduced situations take t0 as TO. In other words, it is important to bear in mind that the pre-present sector differs from the past sector in that the unmarked option (temporal subordination) to express anteriority cannot always be used in the pre-present sector.

4.4. RRCs and NRRCs: different tense system?

In order to formulate an answer to this question, we have to check whether any of the constraints formulated in section 4.2. apply to only one type of RC.

4.4.1. As was the case in connection with past sector RCs, the con-straints on the use of tense in the pre-present sector apply to RRCs and NRRCs alike, except for the examples in which the RC refers to a purely telic situation (cf. 4.4.2.). This is why the principles have been illustrated with both RRCs and NRRCs. Indeed, changing the syntactic status of a RC from relatively dependent (RRC) to relatively independent (NRRC) does not affect the tense system used. The three alternatives (past tense, present perfect, past perfect) are acceptable in the following pairs of RRCs and NRRCs. Any possible difference in effect between the verb forms in the RRCs is the same in the NRRCs:

(52) a. He has destroyed a theory that has been promoted / had been promoted / was promoted by a rival, (adapted from SEU) (RRC)

b. He has destroyed the Watson theory, which has been promoted / had been promoted / was promoted by a rival. (NRRC)

(53) a. Grants have only been issued to students who have produced / produced / had produced valid College identity cards, (adapted from SEU) (RRC)

b. Grants have only been issued to last-year students, who have produced / produced / had produced valid College identity cards. (NRRC)

RRCs and NRRCs: different tense system ? 163

4.4.2. Shift of domain (within the sector, i.e. present perfect tense) NRRCs which are telic and bounded (cf. (54b)) may be interpreted as referring to a W-posterior or W-anterior situation and therefore differ from their past time RRC counterparts:

(54) a. John married Mary, who got pregnant. (NRRC sit-uation W-posterior to SUPC situation)

b. John has married Mary, who has got pregnant. (NRRC situation W-anterior or W-posterior to SUPC situation)

c. John has married the girl who has got pregnant. (RRC situation W-anterior to SUPC situation)

The RC situation in (54b) is less likely to be interpreted as referring to a W-posterior situation than the RC situation in (54a). The present perfect NRRC differs from its restrictive counterpart in (54c) in that the unmarked interpretation of the latter is that it refers to a W-anter-ior situation. In chapter 7, a more in-depth analysis will be given of examples of this type (cf. section 7.5.4.3.).

4.4.3. As was the case in the past sector, the difference in function between RRCs and NRRCs may sometimes lead to a difference in acceptability of a particular verb form. The possible differences between RRCs and NRRCs cannot be explained in terms of different principles determining the use of tense in the two types of relative clauses; they are not due to the fact that the temporal location of a RC situation by means of the verb form is different for RRCs and NRRCs. It is the change from a fully identified antecedent (NRRCs) into one that depends on the RC for restricting information (RRCs) which lies at the origin of the difference in acceptability of certain verb forms, i.e. RRCs seem to have more "freedom" as regards the tenses used. A few examples will suffice to illustrate the effect of making a RRC non-restrictive and vice versa:

(55) a. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen, who have used the jungles as cover, (adapted from SEU) (NRRC)

b. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen, who ?? had used the jungles as cover. (NRRC)


c. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen, who used the jungles as cover. (NRRC)

The antecedent of the NRRC is fully identified: the speaker knows which tribes are being talked about and probably associates certain characteristics with them (i.e. they are cunning, violent, ...). The RC situation establishes a W-simultaneous domain by means of the present perfect and explains why so many people got killed: it is because the tribesman have used the jungle as cover. It does not make much sense to use had used because this tense expresses ante-riority: it establishes a link between a past cause and a past effect. What is needed is a RC which lies at the origin of a present state (i.e. many people are dead). The present perfect is the only tense that can establish links between a past situation and the present time, because it takes t0 as binding TO. The past tense is acceptable but it does not have the strong explanatory resultative implicatures the present perfect has. A possible reading of the past tense is that it was habitual for the Indian troops to use the jungles as cover. By contrast, the present perfect merely suggests that on this particular occasion they have done so. Examples (55d)-(55f) are the restrictive counterparts of (55a)-(55c):

(55) d. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen who have used the jungles as cover. (RRC)

e. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen who had used the jungles as cover. (RRC)

f. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen who used the jungles as cover. (RRC)

It may be argued that the past perfect is acceptable in the RRC. Unlike the antecedent of the NRRC, the tribesmen does not enable the hearer to find out which people are talked about. The restricting information which is needed is provided by the RC: there is refer-ence to tribesmen, who, at a past moment in time -not the time of the SUPC situation-, used the jungles as cover. The past perfect is acceptable, but in this case, the RC merely assigns reference, it does not explain why so many people got killed. When the present perfect is used, the RC may be said to not only establish reference (i.e. it is

RRCs and NRRCs: different tense system ? 165

not the tribesmen who have used self-made wooden barracks as cover against which Indian troops have fought, but those who have used the jungles as cover (= definition of opponents)). The present perfect RRC, just like the present perfect NRRC, also tells the hearer why so many people got killed. The past tense has a similar effect in both the RRC and NRRC. It is necessary to point out that observations like these are highly context-dependent. In other words, it is not possible to make generalizations of the kind, for instance, that the past perfect can never be used in the NRRC in examples of this type, or that the past perfect only specifies the referent of the RRC and does not explain the SUPC situation. A different lexical content in the RRC allows the use of the past perfect to explain the SUPC situation:

(55) g. Several thousand have been killed on both sides as Indian troops have fought with the tribesmen who had stolen a lot of arms from a barracks.

There has to be a relationship of W-simultaneity between the fight and the factor that explains the number of casualties. If that factor is the result of a W-anterior situation, the past perfect can be used more easily, as will be clear from (55g).

Changing the NRRC into a RRC also affects the use of tense in the following example:

(56) a. We have recovered "Agapenor" and "Melampus", which ?? have been trapped in the Suez Canal - a harsh penalty for maintaining schedules by allowing our average frequency of almost a daily transit to run unchecked before the Canal was closed, (adapted from SEU) (NRRC)

b. We have recovered "Agapenor" and "Melampus", which had been trapped in the Suez Canal - a harsh penalty for maintaining schedules by allowing our average frequency of almost a daily transit to run unchecked before the Canal was closed. (NRRC)

c. We have recovered "Agapenor" and "Melampus", which were trapped in the Suez Canal - a harsh penalty for maintaining schedules by allowing our average frequency of almost a daily transit to run unchecked before the Canal was closed. (NRRC)


In general, NRRCs give additional, relevant information. If the speaker wants to be optimally relevant in (56), he will indicate in the NRRC where the boats have been recovered from. From this point of view, it seems reasonable to expect that the boats have been recovered from the Suez canal. The use of the present perfect in the RC (have been trapped) is not completely acceptable for the following reasons: (a) if it is interpreted as a continuative present perfect, it is not acceptable because something cannot be at the same time in a state of being trapped (RC situation) and having been recovered (SUPC situation), (b) if the present perfect is interpreted as an indefinite present perfect, its use is still not completely felicitous. The present perfect suggests that in a period leading up to now, the boats have been trapped from the Suez Canal (indefinite perfect with heavy stress on have). There is no link between their having been in the Suez canal and their just having been recovered from that place. The information that it is from the Suez canal that they have been recovered is much more explicitly conveyed when the SUPC situation instead of t0 is taken as binding TO, i.e. when the past perfect is used. The past tense gives background information which is not explicitly located in time. When the RC is made restrictive, the present perfect may be said to be not entirely acceptable for the same reasons:

(56) d. We have recovered the boats which ? have been trapped in the Suez Canal - a harsh penalty for maintaining schedules by allowing our average fre-quency of almost a daily transit to run unchecked before the Canal was closed. (RRC)

e. We have recovered the boats which had been trapped in the Suez Canal - a harsh penalty for maintaining schedules by allowing our average fre-quency of almost a daily transit to run unchecked before the Canal was closed. (RRC)

f. We have recovered the boats which were trapped in the Suez Canal - a harsh penalty for maintaining schedules by allowing our average frequency of almost a daily transit to run unchecked before the Canal was closed. (RRC)

If the RC situation is understood as leading up to t0 (i.e. the boats are still in the Suez Canal), it is not possible to interpret the RC, as one could in (55e), as merely identifying and not expressing any other


logical link between the RC situation and the SUPC situation because the definition would imply that the boats are still in the Suez canal. The definition itself is not compatible with the SUPC situation; the boats have been recovered, consequently, they can no longer be in the Suez Canal. If the present perfect in the RC is given an indefinite reading (at some time in the past, the boats were trapped), the present perfect might be argued to be acceptable. In this reading, the RRC defines which boats are being talked about (i.e. those that at some time in the past had been trapped in the Suez canal) but does not tell us where they have been recovered from now; the boats have not been recovered from the Suez Canal, but from some other place where they were trapped. The past perfect RRC (56e) defines which boats are being talked about by referring to the place they have been recovered from. In (56f), the restricting information is not explicitly located in time.


There is no difference in the tense system used in RRCs and NRRCs in the sense that one option (e.g. temporal subordination, shift to the past sector, shift within the pre-present sector) is exclusively reserved for either RRCs or NRRCs. In section 4.4.3., it has been shown that RRCs and NRRCs may differ as regards the acceptability of certain verb forms. The fact that RRCs have a restricting function explains why certain verb forms, rejected in the NRRC counterparts, are acceptable. The temporal relations expressed by means of a particular verb form may also be different in RRCs and NRRCs if the situation is purely telic (cf. section 4.4.2.). Our investigation into the use of tense in pre-present sector RCs has also revealed that the system presented in Declerck (1991a) should be slightly modified because the unmarked option of temporal subordination cannot al-ways be used. In chapter 6, some of the possible differences in effect between mutually substitutable verb forms will be exemplified.

Chapter 5 The expression of (W-)anteriority and (W-) simultaneity in post-present sector RCs

5.1. The use of tense in the post-present sector: Declerck (1991a)

In order to locate RC situations in a post-present domain, one of the following tense systems may be used: the present perspective system (PPS) or the future perspective system (FPS) (cf. chapter 2, section 2.1.5.4.).

In the former case, the RC situations are temporally subordinated to the post-present TO which establishes the domain. There is a shift of perspective to the present: the post-present TO is treated as if it were t0 and all the temporal relations are defined with respect to this "pseudo-t0". From this it follows that the tenses used are the same as those used to establish a domain in one of the four absolute sectors. The present tense115 is used to indicate that a situation is simultane-ous with the post-present TO, as in (1). The continuative present per-fect is used when the newly introduced situation starts before the post-present pseudo-t0 and leads up to it, as in (2). The past tense116

(cf. (3)) or the indefinite present perfect117 (cf. (4)) indicates anteri-ority in the post-present sector. The future tense is used to indicate that the newly introduced situation is posterior to the post-present TO, as in (5):118

(1) They will believe that Jack is back in town. (Declerck 1991a: 38)

(2) He will soon find out that you have been following him. (Declerck 1991a: 37)

(3) (said when planning someone's murder) The police will think that he was killed tonight. (Declerck 1991a: 36)

(4) What will happen when the others have left? (Declerck 1991a: 37)

(5) He will say that he will never leave her. (Declerck 1991a: 39)

The PPS enables Declerck to explain why will cannot normally be used in temporal and conditional clauses:119 it is because these

170 (W-)anteriority and (W-)simultaneity in post-present sector RCs

It appears, then, that the use of present verb forms (to show "same time") and of present perfect verb forms (to show "earlier time") in clauses included in sentences (or other clauses) referring to future time is actually what one should expect and not a violation of the rules of logic. The use of will in an included clause in such sentences would not be "more correct", as Reichenbach suggests, nor are present verb forms used in such included clauses because "futurity is sufficiently indicated in the main verb", as Jespersen states. (Allen 1966: 168)

In the FPS, the newly introduced situations are not temporally subordinated to the TO establishing the post-present domain. (The present sector) t0 functions as binding TO; the domain is shifted within the sector:

(6) The whole family will assemble next year. Even John and Susan will come. They will have been married for forty years. (Declerck 1991a: 52)

The future tense will come is said to re-establish (Declerck 1991a: 52) the domain, i.e. it explicitly indicates that the situation it in-troduces is located in the post-present sector. The tenses as such do not reveal whether the situation is anterior to (cf. (7a)), simultaneous with (cf. (7b)) or posterior (cf. (7c)) to the situation establishing the post-present domain. Pragmatic and contextual knowledge or adverbials will contribute to revealing the temporal relations between the different situations, as the tenses as such do not fulfil this func-tion (cf. Lo Cascio - Rohrer 1986: 247-248):

(7) a. John will be caught because he will set off the alarm. (Sampson 1971: 588 in Declerck 1991a: 56)

b. I will ask John, who will know the answer tomorrow. (Declerck 1991a: 55)

c. I will ask John if Mr. and Mrs. Jones will also be present at the party.

It is likely that will be in (7c) is interpreted as a temporally sub-ordinated form, i.e. as expressing posteriority with respect to the post-present sector situation which establishes the domain. The future tense is like the past tense in that it is sometimes difficult to say whether it is used as a relative or as an absolute verb form, as

RRCs vi. NRRCs: different tense system ? 171

there is no morphological distinction between the temporally subordinated and the shift of domain verb form to refer to a W-posterior situation in the post-present sector (just as there is no difference between the relative verb used to refer to a simultaneous situation in a past domain and its absolute alternative) (cf. e.g. Ladusaw 1977: 95, Lo Cascio - Rohrer 1986: 238-239, Richards 1982: 85). The future perfect also belongs to the FPS. This tense may be called an absolute-relative tense (Declerck 1991a: 52) because, apart from indicating that the situation is anterior to a TO (which accounts for the label relative), it makes explicitly clear that there is reference to a post-present domain (which accounts for the label absolute):

(8) If it rains tomorrow, we will have worked in vain yesterday. (Comrie 1985: 73)

The future perfect as such does not indicate whether the situation lies before tg, coincides with it or lies after t0.121 That is why the adverbials today and tomorrow could also be used in (8) (cf. section 5.3.1.2.).

The amount of attention given to how temporal relations are expressed in post-present domains (let alone in post-present RCs) is fairly limited. As in the case of the present perfect, a major amount of the (considerable) literature on the future centres around a limited number of issues, the most crucial of which are: (a) Does English have a future tense?,122 (b) What is the difference in meaning between the different forms that establish future time reference (be going to, be to, will/shall, ...)? Sporadic references are made to the difference between (what is here called) a FPS verb form and a PPS verb form in a particular context, although no attempt is made (apart from Allen 1966) at a comprehensive systematization, i.e. how does one express (1) simultaneity, (2) anteriority and (3) posteriority in the post-present sector?

5.2. RRCs vs. NRRCs: different tense system?

The choice between using the FPS rather than the PPS to locate RRC and NRRC situations in time is not a free one. As Declerck (1991a) points out, there appears to be a tendency for NRRCs to use the FPS exclusively, whereas in RRCs, either system can be used, though not without constraints: "In subclauses whose situation is not represented


as temporally dependent on that of the matrix, the Future Perspective System is used. Nonrestrictive relative clauses are always of this type. (...) In some types of subclause (e.g. restrictive relative clauses) the situation may or may not be represented as temporally dependent on the situation of the matrix" (Declerck 1991a: 55):123

(9) a. I will ask John, who will know the answer tomor-row. (* knows) (Declerck 1991a: 55)

b. I will ask John, who will have heard the news by then. (* has heard) (ibid.)

(10) When she gets paid next month she will tell us that she will soon buy a new hat with the money she has saved / will have saved from that paycheck, (ibid.)

Wekker (1976) also sets apart a number of clause types in which what we call PPS forms are likely to be used and a number in which FPS forms are likely to occur: in independent clauses and non-restrictive clauses, "there seems to be no future subordination (...). This is not always the case with restrictive attributive clauses" (Wekker 1976: 88).

Before a survey is presented of the factors making it impossible to use the FPS or the PPS in RRCs (section 5.3.2.), an explanation will be offered for the fact that the FPS is indeed the only tense system that can be used in NRRCs. Examples (11), (12), (15) and (16) show that the forms which could be interpreted as belonging to the PPS are never understood in this way in NRRCs, i.e. they never establish post-present time reference in NRRCs:

(11) a. From the collection centers, toys will be taken to a warehouse at 198 Second street, where they will be repaired and made ready for distribution. (BR)

b. From the collection centers, toys will be taken to a warehouse at 198 Second street, where they 'are repaired and made ready for distribution.

The only interpretation which are repaired allows is a present sector one: the habit of repairing toys in a warehouse at 198 Second Street exists at the moment of speaking. Future time reference requires the use of a FPS form: will be repaired explicitly communicates that the toys are not as yet taken to the warehouse to be made ready for distribution. The present tense is also interpreted as indicating present time in example (12):

RRCs vi. NRRCs: different tense system ? 173

(12) a. All will leave by 8.45 except Donald Payne, who stays a little later, (adapted from SEU)124

b. All will leave by 8.45 except Donald Payne, who will stay a little later.

The present tense refers to an existing habit. Stays could also be interpreted as an absolute shift of perspective tense form (a present tense is used to establish future reference), provided this sentence is said by someone (e.g. a general in the army, director of a school) with the authority to decide when the people present have to leave. When the future tense is used, the speaker is explicitly making a prediction.

In some cases, the present tense establishes future time reference. However, these examples do not constitute a challenge to the principle formulated above; in this case the verb forms are the result of a shift of perspective, which implies that the present tense is an absolute tense form:

(13) a. Leeds will eventually change into top gear, and after a bewildering movement, in which everyone except Harvey will appear to take part, Gray will deliver a splendid shot, which brings a save no less worthy from Donaldson, (adapted from SEU)

b. Leeds will eventually change into top gear, and after a bewildering movement, in which everyone except Harvey will appear to take part, Gray will deliver a splendid shot, which will bring a save no less wor-thy from Donaldson.

The switch from the future tense to the present tense at the climactic moment "shows good style", according to one native speaker, because it adds dramatic force to the statement. In example (14), there is reference to a planned future, which can be expressed by a present tense accompanied by a future time adverbial:

(14) a. Sir Peter will succeed Sir John Milne, 65, who retires as Blue Circle nonexecutive chairman on June 1. (WSJ)

b. Sir Peter will succeed Sir John Milne, 65, who will retire as Blue Circle nonexecutive chairman on June 1.


The present perfect and past tense in the NRRC are not interpreted as PPS forms in the following examples either; they are understood as absolute tenses that locate the NRRC situation in the pre-present and past sector, respectively:

(15) Rather, the brunt of the slowdown will be felt in the fast-growing private and semi-private "township" enterprises, which have fallen into disfavor as China's leaders re-emphasize an orthodox Marxist preference for public ownership. (WSJ)

(16) They will soon extinguish the candle, which was lit at 5 o'clock this afternoon.

As I see it, the impossibility of using the PPS in NRRCs is due to (a) the nature of the tenses of the PPS and (b) the function of NRRCs. In section 4.3. of chapter 4 it was pointed out that the occasional impossibility of using temporally subordinated forms in pre-present sector RCs should be ascribed to the nature of the tenses involved (temporal subordination: past perfect vs. shift of domain within sector: present perfect). The present perfect tense establishes links with t0 (present time-sphere), the past perfect does not (past time-sphere). There is no similar difference between the temporally subordinated and the shift of domain verb forms in the past sector: the use of the past perfect (temporal subordination) instead of the past tense (shift of domain) can never result in the location in a sector other than the past. The nature of the tenses involved in the choice between temporal subordination and a shift of domain is also important for the use of tense in the post-present sector. The temporally subordinated forms (PPS), e.g. the present tense, can be misinterpreted as referring to another, e.g. the present, sector: the unmarked interpretation of a present tense form is that it refers to present time. NRRCs give additional, relevant information. These two elements, together with the fact that "other things being equal statements about the present moment are more relevant than those about other times" (Comrie 1985: 41-42) (cf. also Andersson 1991: 2, Declerck 1991a: 69, Fleischman 1989: 3-4) (i.e. facts are more important than predictions), explain why present tense NRRCs are interpreted as having present time reference whenever possible. A similar line of reasoning applies to past tense and present perfect NRRCs. In this case, it is the preference for a situation with a factual character (because it is anterior to t0) rather than a prediction (because the situation is anterior to the pseudo-t0) which lies at the

RRCs vs. NRRCs: different tense system? 175

origin of the past/pre-present sector interpretation of the verb forms. Any additional future time information given in a NRRC has to be explicitly located in time by means of FPS forms.

However, the ultimate question to be answered is whether the principles determining the use of tense in post-present RRCs and NRRCs are different. My answer is that forms that could be interpreted as PPS forms are likely to be interpreted as having present/past/pre-present time reference whenever possible, no matter whether they occur in a RRC or in a NRRC. From this point of view, RRCs and NRRCs do not differ. For instance, when we change the NRRCs given in Declerck (1991a) (cf. examples in (9)) into RRCs, it becomes clear that the latter type of clause does not always allow temporal subordination either:

(17) a. I will ask the man who will know (* knows) the answer tomorrow.

The ungrammatically of the PPS forms is due to a certain degree of incompatibility of two elements: a present tense accompanied by a future time adverbial can be interpreted as a shift of perspective provided there is reference to a situation which is predictable (e.g. The train leaves at 5 o'clock). Know the answer is, on its unmarked interpretation, not predictable. There is therefore a clash between the situation referred to and the type of situation needed to establish future time reference by means of a present tense. It is possible, though, to devise an "exceptional" context in which the man who knows the answer tomorrow becomes acceptable. If a contest is to take place tomorrow, and it is certain now that there will be one man who will know the answer to the question and another man who will fail to answer the question, it is possible to say: / will ask the man who knows the answer.125 However, a shift of perspective tense is an absolute, not a relative tense, which means that even if knows is acceptable, it has to be classified as an absolute tense and not as belonging to the PPS. By tomorrow in (17b) requires the use of an absolute-relative tense, i.e. the future perfect:

(17) b. I will ask the man who will have heard (* has heard) the answer by tomorrow.

Even if there are no adverbials present, knows and has heard do not get a post-present interpretation:


(17) c. I will ask the man who will know ('knows') the answer.

d. I will ask the man who will have heard ('has heard) the answer.

However, RRCs and NRRCs differ in the sense that the likelihood of verbs belonging to the PPS being "misinterpreted" is far greater in NRRCs than in RRCs: "misinterpretation" is the rule in NRRCs, whereas it is merely possible in RRCs. This has to do with the func-tion of NRRCs, which is reflected in their independent syntactic status. Indeed, the importance of the tight syntactic links between the main clause and the RRC cannot be denied. It is basically because the syntactic ties between RRCs and their (future tense) SUPCs are close that present tense forms, for example, can be interpreted as having future reference. From that point of view, the tense system in post-present RCs is determined by the different syntactic status of RRCs and NRRCs. Even so, the fact that there is similarity in some respects prohibits a generalization only highlighting the distinction between RRCs and NRRCs as regards the use of tense. We cannot conclude on the basis of the way in which temporal relations are expressed in some post-present sector RCs that the syntactic status (relatively dependent vs. relatively independent) of the RRC and the NRRC is important for the use of tense in the other sectors as well, especially if there is no independent evidence for this hypothesis in the other sectors (i.e. if the principles determining the use of tense are to a considerable degree the same for both RRCs and NRRCs in the pre-present and past sector).

To summarize, forms which can be interpreted as PPS forms are always "misinterpreted" in NRRCs and may be "misinterpreted" in RRCs as well. The next section presents a survey of the factors determining the interpretation of verb forms in RRCs.

5.3. Factors determining interpretation of verb forms in RRCs

5.3.1. General principle exemplified

The use of tense in RRCs appears to be dictated by the interaction of and the tension between two principles. The first is that whenever verb forms which can be interpreted as PPS forms (i.e. the pseudo-t0 functions as binding TO) can also be interpreted as absolute tense

Factors determining interpretation of verb forms in RRCs 177

forms, the second reading is likely to be chosen for reasons given in the previous section. The other principle which plays a role is that less processing effort seems to be required if the RRC is incorporated into the post-present domain established by the SUPC; shifting the domain requires some kind of re-orientation.126 In other words, PPS forms are processed more easily than FPS forms. From this it follows that whenever the "correct" (in this context, post-present sector) tem-poral information (e.g. the RC situation is simultaneous with or ante-rior to the post-present SUPC situation) is conveyed by the PPS, the latter system is likely to be used. As is explained in Declerck -Depraetere (1995), FPS forms create a kind of intensional domain. PPS forms will be interpreted as belonging to an existing post-pre-sent intensional domain, provided a post-present domain has been created in the preceding context, either by a FPS form, an imperative or some other device which indicates that there is reference to a fu-ture time. Whether or not a verb form allowing a PPS interpretation will be understood in that way depends on the number of semantic and/or pragmatic links there are between the post-present TO and the RC situation. The FPS will be used (a) if there are factors which pre-vent the intended interpretation when the PPS is used or (b) if the FPS conveys contextual effects which the PPS does not have. The examples in this section will show that PPS forms are sometimes subject to "misinterpretation". A second set of examples (section 5.3.2.) will serve to illustrate the factors influencing the interpreta-tion of verb forms in RRCs.

I will concentrate on examples in which the RC situation is either simultaneous with or anterior to the SUPC situation and leave out RC situations which are posterior to the SUPC situation, because in the latter case the tenses as such do not signal whether the PPS or the FPS is used: the future tense is used no matter whether there is tem-poral subordination or not (cf. Lo Cascio - Rohrer 1986: 238-239) (cf. examples in (18) and (19)), whereas in the case of simultaneity or anteriority, the tenses do tell us whether or not there is temporal subordination (i.e. the present tense (PPS) vs. the future tense (FPS) for simultaneity, the indefinite present perfect/past tense (PPS) vs. the future perfect (FPS) for anteriority):

(18) At the old nest, the queen will in the early fall cease to lay the fertilized eggs that will produce females. (BR)

(19) It is to be hoped that both management and workers will be able to put forward constructive ideas which will help to


push further into the background the dreaded threat of unemployment. (LOB)

5.3.1.1. Present tense in the RRC

When the RRC refers to a situation which might just as well hold at t0, it is likely that the present tense form will be interpreted in terms of a shift of domain to the present sector.127 In order to make explicitly clear that the situation does not hold at t0, the speaker will have to use the FPS in examples of this type; to prevent the hearer from processing the non-progressive present tense as an absolute tense, the speaker will re-establish the domain by means of the future tense:

(20) a. This year, Americans will discover previously unheard of refinements in trailers that will be exhibited in about one hundred of our nation's national, regional and local boat shows. (BR)

b. This year, Americans will discover previously unheard of refinements in trailers that 'are exhibited in about one hundred of our nation's national, regional and local boat shows.

When the present tense is used in the RRC, the meaning conveyed is that the boats are already on show, i.e. the present tense is interpreted as an absolute tense. The future tense in the RC indicates that Americans will visit shows that still need to be organized. Example (21) illustrates the same tendency:

(21) a. He will jump a long queue of coloured men who 'are' in line for this job. (adapted from SEU)

b. He will jump a long queue of coloured men who will be in line for this job.

In this example, the RC present tense allows two interpretations, i.e. simultaneity in the post-present or shift of domain to the present sector. The presence of the demonstrative adjective this, usually associated with reference to something close to the speaker, i.e. to his here-and-now, tips the scales in favour of the present sector interpretation of the present tense. The choice of verb form depends


to a large extent on how sure the speaker is the hearer will not misinterpret his utterance.

The following examples show that if there is a future time, intensional domain into which the RRC situation can be incorpo-rated, the present tense is interpreted as a PPS form:

(22) When they lack zest, belief and knowledge, they will mutilate opportunities to make sales. When they possess enthusiasm, belief and knowledge, they will create opportunities as well as accept those which are presented to them. (LOB)

(23) "Hang this around your neck or attach it to other parts of your anatomy, and its rays will cure any disease you have", said the company. (BR)

In both (22) and (23), the context indicates that there is reference to a possible future world in which certain things will be the case, i.e. the future time is presupposed and the existence of the RC situations are asserted within this future world (cf. Declerck and Depraetere 1995).

In (24), no confusion arises either. However, in this example, the present tense is interpreted as an absolute tense form which is the result of a shift of perspective:

(24) a. Managers who retire Dec. 30 will have an additional 15% added to their monthly pension for as long as five years or age 65, whichever comes earlier. (WSJ)

b. Managers who will retire Dec. 30 will have an additional 15% added to their monthly pension for as long as five years or age 65, whichever comes earlier.

The present tense can only be interpreted as an absolute tense resulting from a shift of perspective, i.e. a present tense is used to locate a situation in the post-present sector.

5.3.1.2. Present perfect of the indefinite type and past tense in the RRC

So far, only present tense examples have been given. It will be clear that the other tenses of the PPS also bring about the potential danger


of being interpreted as absolute tenses establishing (in the case of anteriority) a past domain (when the past tense is used) or a pre-pre-sent domain (when the indefinite present perfect is used) (cf. Lo Cascio - Rohrer 1986: 240-241, Partee 1973: 608). The following example shows that the future perfect may have to be used to prevent the hearer from processing the RC situation as a pre-present sector situation:

(25) a. He will suddenly know envy for the easy happiness with which she 'has returned to life, while his own return 'has been' so lonely and uneasy, (adapted from SEU)

b. He will suddenly know envy for the easy happiness with which she will have returned to life, while his own return will have been so lonely and uneasy.

The present perfect may be misunderstood as referring to an event that took place before t0 and is therefore not likely to be used if post-present sector time reference is intended. Unambiguous post-present sector location of the W-anterior situation return to life requires the use of the future perfect.

Theoretically, a situation referred to by the future perfect can also lie before t0. Comrie gives the following example: If it rains tomorrow, we will have worked in vain yesterday (Comrie 1985: 73). However, this use of the future perfect is marked. If the situation lies before t0, and the speaker decides to use the future perfect in spite of this, it must be that this tense communicates effects which the past tense does not (cf. Binnick 1991: 279). If, for instance, it is not sure that it will rain, the future perfect is suitably used because it also conveys an element of expectation, which matches the conditional context in which it is couched.128 Still, the unmarked option is to use an absolute past tense/present perfect since there is reference to a past sector/pre-present sector fact:129

(26) a. Until then, it'll be hard to coax participants who were burned by the dollar's dramatic fall Friday to re-enter the market. (WSJ)

b. Until then, it'll be hard to coax participants who ? will have been burned by the dollar's dramatic fall Friday to re-enter the market.


(27) a. I do not know at all. It will depend upon the instructions Mr Tsarapkin (the Russian delegate) has brought back with him from Moscow. (LOB)

b. I do not know at all. It will depend upon the instructions Mr Tsarapkin (the Russian delegate) will have brought back with him from Moscow.

It is particularly clear from example (26) that it is odd to use the future perfect if there is reference to a past sector situation. Similarly, the unmarked interpretation of the future perfect RC in (27) is that Mr Tsarapkin has not yet returned from Moscow. The present perfect is interpreted as an absolute tense referring to a situation that lies before t0. Moreover, in (27b), the speaker predicts that Russian delegate will have been given instructions in Moscow, whereas in (27a), it is sort of taken for granted he has; the speaker does not explicitly predict that it is his conviction that this will be the case. These examples show in any case that the future perfect is seldom used when there is reference to a past/pre-present sector situation (except for the context mentioned in footnote 128). Unless pragmatic knowledge or an accompanying adverb indicates the contrary (as in Comrie's example), a future perfect will be interpreted as locating the situation after t„. From this it follows that if the speaker wants to make completely sure that the RC will not be misunderstood as effecting a shift of domain to the past sector, he will use the future perfect to refer to a situation which is anterior to the future TO. How-ever, as I have found no examples of this type in the corpus of examples at my disposal, it must be that in these cases as well, the future perfect is not often used, a PPS form being the preferred option. This observation also implies that since the speaker uses PPS forms rather than FPS forms in the case of anteriority in the post-present sector, he will make sure that the RC situation can only understood as belonging to the post-present intensional context and cannot be processed as an absolute tense. The reason for the preferred use of PPS forms may also be related to the following: the element of personal prediction of future factuality on the part of the speaker is very strongly present when the future perfect is used (cf. example (27)), which may be why this verb form is not often chosen if the speaker wants to present situations as "mere facts" anterior to the post-present pseudo-t0:


(28) a. "You are right; all this unhappiness will pass. It will seem like some vanished dream from which we have awakened to reality once more. (LOB)

b. "You are right; all this unhappiness will pass. It will seem like some vanished dream from which we will have awakened to reality once more.

(29) a. You will find a chart giving the various speed ratios available with your particular drill press somewhere in the instruction booklet that came with the tool. (BR)

b. You will find a chart giving the various speed ratios available with your particular drill press somewhere in the instruction booklet that will have come with the tool.

(30) a. Also, under the agenda of the anti-choice move-ment, there will be absolutely nothing that the adoption option will be able to do for the women who wanted an abortion, couldn't have one and died from continuing the pregnancy. (WSJ)

b. Also, under the agenda of the anti-choice move-ment, there will be absolutely nothing that the adoption option will be able to do for the women who will have wanted an abortion, won't have been able to have one and will have died from continuing the pregnancy.

In each of the above examples, the use of the future perfect explicitly communicates that the speaker is making a prediction about what he thinks will happen and at the same time asserts that the situations will actually hold:

(28c): We will awake from the sad reality. (29c): An instruction booklet will come with the tool. (30c): There will be women who will want an abortion.

PPS forms can be used in (28) to (30) as it is sufficiently clear that the situations belong to the post-present, intensional domain created by the verb form and the context. Example (31) too shows that PPS forms are not always misinterpreted:

(31) a. Mr Wilson might, for example, be defeated - in which case Mr. Heath will have no place at all for a


land commission or any other legislation that a Labour government has left behind, (adapted from SEU)

b. Mr Wilson might, for example, be defeated - in which case Mr. Heath will have no place at all for a land commission or any other legislation that a Labour government will have left behind.

The context makes it clear that there is reference to future election results. As the votes have not yet been cast, the Labour government cannot have left behind any commissions or legislation at the time when the statement is made. The PPS form will therefore be inter-preted correctly: the present perfect expresses anteriority. If it is true that the PPS requires less processing effort, a FPS form is less likely to be used in this sentence, although it is acceptable. Unlike the PPS form, it adds a touch of prediction to the sentence. The indefinite per-fect in the RC in (32) will not be misinterpreted either:

(32) "If you go on making Fay's life a hell for her you'll find yourself with another wife who's deserted you." (Kirsten 1988: 26)

The present perfect unambiguously establishes post-present time reference because the main clause situation depends on a condition which has not yet been fulfilled.

Another problem is that it is not always possible to determine the status of the present perfect, i.e. whether it is an absolute tense or a relative PPS form:

(33) By seeing such varied places, both interesting and beautiful, you will become aware of the many different civilizations Rome has lived through, and in particular, get a feel of Renaissance Rome. (BR)

Has lived could be bound by the post-present TO established in the SUPC, in which case it is a PPS form, or by t0, in which case it is an absolute tense.

The conclusion to be drawn from the examples in this section is that situations that lie before t0 are not likely to be referred to by means of the future perfect. However, when the present perfect and the past tense are used, it is often not possible to determine whether they are absolute tenses or PPS forms bound by the post-present TO


established by the SUPC situation. If the situation lies after t0, and if there is an intensional, post-present domain into which the RC situation can be incorporated, PPS forms are likely to be used, not in the least because the overtone of personal prediction of future factuality by the speaker, which is characteristic of the future perfect, is often incompatible with the message to be conveyed.

5.3.1.3. Present perfect of the continuative type in the RRC

The continuative present perfect is not very likely to be misinter-preted as referring to the pre-present sector. This may be due to the close semantic link between the present perfect situation and the future SUPC situation:

(34) a. They will encourage groups of miners in all parts of the country, who will be fed up with a strike in which they have been denied a say, and who will dearly want to create the opportunity for a return to work, (adapted from SEU)

b. They will encourage groups of miners in all parts of the country, who will be fed up with a strike in which they "will have been denied" a say, and who will dearly want to create the opportunity for a return to work.

c. They will encourage groups of miners in all parts of the country, who will be fed up with a strike in which they "will be denied" a say, and who will dearly want to create the opportunity for a return to work.

The present perfect will not be interpreted as an absolute tense because the sentence would in that case be rejected as pragmatically anomalous; it would not make sense to claim that at some future stage, people will be fed up with a strike (in which they have been denied a say) which started before, and reached up to, t0. In order for the causal link between the SUPC and the RC (they are fed up because they are and have been engaged in a strike in which they do not have a say) to make sense, there cannot be a lapse of time between the two situations. The example just given points to an im-portant difference between, on the one hand, past tense and indefinite perfect sentences and, on the other, continuative perfect sentences. A


situation located in the past sector may be considered from the point of view of t0 or from the point of view of the post-present pseudo-t0. The choice of TO does not affect the actual duration of the situation: the situation lies completely before t0 in both cases. The same line of reasoning applies to present perfect sentences which are given an indefinite interpretation. However, things differ for present perfect sentences of the continuative type. Depending on whether the continuative present perfect is understood as bound by the post-present pseudo-t0 or by the present sector t0, the situation referred to will last up to the post-present t0 or to t0. Only one of these two options is a valid representation of the real-world situation. The continuative present perfect will not be misinterpreted in (35 a) either:

(35) a. Mr. [Name] will be with us only for the academic year [2222/22], when he will have to resign because of family difficulties which have affected him throughout his time with us. (adapted from SEU)

b. Mr. [Name] will be with us only for the academic year [2222/22], when he will have to resign because of family difficulties which "will have affected" him throughout his time with us.

c. Mr. [Name] will be with us only for the academic year [2222/22], when he will have to resign because of family difficulties which "will affect" him throughout his time with us.

Both the FPS and the PPS can be used in this example. The first sentence establishes future reference: it indicates that Mr. [Name] is not with us yet. Throughout his time with us in the RC therefore refers to a period yet to come. Accordingly, the PPS forms will not be misinterpreted. Even if the adverb were not present in the RC, no confusion would arise. The present perfect situation cannot be understood as lying before t0: the affecting cannot have started before t0 because it must be W-simultaneous with the time he will spend with us, which has not yet begun. In other words, the cause-effect relationship between the RC and the SUPC would prevent the pre-sent perfect from being interpreted as an absolute tense. The future and future perfect explicitly communicate that Mr. [Name] is not working for the institution in question at t0. The future perfect suggests a higher degree of factuality than the future tense.


However, not all continuative present perfect tenses unam-biguously refer to the post-present sector, as is clear from the fol-lowing example:

(36) I has a great many assets to recommend it and if you haven't made avocado a part of your diet yet, you really should. You will find that avocado is unlike any other fruit you have ever tasted. (BR)

Depending on whether ever means until now or until then, the verb form should be classified as an absolute or relative verb form.

The following conclusion can be drawn so far: pragmatic, espe-cially contextual, knowledge determines whether it is possible to use PPS forms (cf. Adelaar - Lo Cascio 1986: 267-268, Declerck 1991a: 56, Richards 1987: 381). This implies that it is not possible to deter-mine in advance when the PPS forms will be misinterpreted as not referring to the post-present sector; what is mutually manifest130 in one context (i.e. post-present sector reference) is not necessarily so in another. The possible difference in effect between mutually substi-tutable verb forms will be discussed in chapter 6.

5.3.2. Influencing factors

As it is pragmatic knowledge which plays an important role in the interpretation of PPS verb forms it is not possible to make more ex-plicit claims about the way in which this happens, as the context changes with every utterance. However, the following elements de-termining the likelihood of a PPS form (not) being "misinterpreted" can be derived from the examples:

5.3.2.1. Aspect and (a)telicity

The following principle no doubt plays a role in the interpretation of verb forms in RRCs: if the RC verb allows it, the speaker will use a progressive tense if he wants to indicate that the situation is going on at t0. This means that the non-progressive present tense is less likely to be interpreted as having present time reference than its progressive counterpart:


(37) a. He will be judged on the way in which he is behaving.

b. He will be judged on the way in which he behaves. c. He will be judged on the way in which he will

behave. (38) a. "It will become more important to provide incen-

tives to companies who are considering all the alternatives," says Stanley E. (WSJ)

b. "It will become more important to provide incen-tives to companies who consider all the alterna-tives," says Stanley E.

c. "It will become more important to provide incen-tives to companies who will consider all the alter-natives," says Stanley E.

(39) a. They will talk, and even write, about their favourite books. They will sometimes explicitly tell us, and more often unintentionally reveal, the sort of pleasure they are taking in them and the sort of reading it implies.

b. They will sometimes explicitly tell us, and more often unintentionally reveal, the sort of pleasure they take in them and the sort of reading it implies. (LOB)

c. They will sometimes explicitly tell us, and more often unintentionally reveal, the sort of pleasure they 'will take' in them and the sort of reading it implies.

The native speakers consulted agree that the judgments are delicate. However, they believe that the progressive and non-progressive forms differ indeed in the likelihood of being interpreted as PPS forms. Although it is not impossible to interpret the RCs with progressive forms in (37a) to (39a) as referring to future time activities, this is not their unmarked interpretation. Will take is not likely to be used in (39c) because it is interpreted as a relative tense expressing posteriority with respect to the SUPC situation.

A state (cf. the examples in (40) and (41)) cannot usually be represented by the progressive form. Accordingly, the chances of a non-progressive present tense form being interpreted as an absolute tense are higher than that of a non-progressive present tense form referring to an activity:


(40) a. I will be Head of the Department in which she 'iV (* is being) research assistant, so I will know her quite well, (adapted from SEU)

b. I will be Head of the Department in which she will be research assistant, so I will know her quite well.

(41) a. In a short, poignant article, "Two Bus Lines to Bethlehem," he ponders how Jews and Arabs will coexist in a country in which even public transport 'is segregated (* is being segregated). (WSJ)

b. In a short, poignant article, "Two Bus Lines to Bethlehem," he ponders how Jews and Arabs will coexist in a country in which even public transport will be segregated.

The progressive form is unacceptable in the above examples. Accordingly, the non-progressive present tense is more likely to be interpreted as referring to the present sector than in the examples with activities and accomplishments, which have the possibility of using a progressive form to communicate that present sector reference is intended. In (40) and (41), a FPS form has to be used to locate the RC situations in the post-present sector.

As in the case of all the principles formulated in section 5.3.2., the tendency just illustrated does not apply strictly. In most of the cases, if post-present sector reference is intended, there will be other factors which contribute to incorporating the RC situation into the post-present intensional domain and in this way safeguard a post-present sector interpretation, which means that the effect of aspect will be overruled:

(42) To me you'll still be my beautiful sweetheart who's waiting for me. (LOB)

In other cases, no matter whether the present progressive or present non-progressive is used, the RC situation is not incorporated into a post-present domain, because an adverbial or the context imposes a present sector reading:

(43) a. Luthuli cannot speak freely to us. But he must draw encouragement from the reception that he will receive here and elsewhere on his journey. (LOB)


b. Luthuli cannot speak freely to us. But he must draw encouragement from the reception that he 'is receiving' here and elsewhere on his journey.

c. Luthuli cannot speak freely to us. But he must draw encouragement from the reception that he 'receives' here and elsewhere on his journey.

There is no post-present sector domain available into which the RC situation can be incorporated: both the progressive and the non-progressive present tense are interpreted as indicating present time. The next example is somewhat like the state-examples in (40) and (41) because the progressive form results in an unacceptable sentence: it suggests that the new gas will be promoted to people who, at the very moment of speaking, are busy filling up their car:

(44) a. Though Arco won't describe the chemical properties of its new unleaded gasoline, it has hinted to some distributors that it will be promoted to customers who fill up with leaded as well as to owners of newer models, (adapted from WSJ)

b. Though Arco won't describe the chemical properties of its new unleaded gasoline, it has hinted to some distributors that it will be promoted to customers who ?? are filling up with leaded as well as to owners of newer models.

c. Though Arco won't describe the chemical properties of its new unleaded gasoline, it has hinted to some distributors that it will be promoted to customers who will fill up with leaded as well as to owners of newer models.

As the progressive present tense cannot be used as an alternative to the non-progressive form, the expectation is that the latter is likely to be interpreted as having present time reference. However, this is not the case because, when the FPS alternative is used, the sentence is understood to mean that there are no people at all, who, at t0, are in the habit of using leaded gas. Consequently, the non-progressive present tense can be used: in this case, there is reference to an existing category of people, but because of the post-present domain that is established in the SUPC, the RC is interpreted as a future time actualization of the existing category. In the next example as well, the present tense is not likely to be interpreted as having present


sector time reference because the sentence only makes sense if the category of people referred to is taken as a post-present sector in-stantiation of an existing group:

(45) a. We will need partners who want a good return but are willing to take a long-range view. (WSJ)

b. We will need partners who will want a good return but are willing to take a long-range view.

5.3.2.2. Definite vs. indefinite antecedent

The (in)definiteness of the antecedent plays a role in the temporal location of the RC situation.131 The following example illustrates this observation:

(46) a. You will meet a man who is wearing a blue coat, b. You will meet the man who is wearing a blue coat.

Out of context, one is much more inclined to interpret the present tense as referring to the present sector in (46b) than in (46a). This preference can only be ascribed to the difference in definiteness of the antecedent. The (in)definiteness of the article undoubtedly influences the temporal interpretation of verb forms in future time sentences. However, when couched in a future time context (e.g. if it has already been mentioned that at a particular moment, a man will appear wearing a blue coat), it is likely that the definite NP will be interpreted as having future time reference as well.

A definite non-generic NP communicates that the speaker believes that the hearer is in a position to identify the referent. If the definite NP does not contain a modifier (e.g. the man) giving the required identifying information (e.g. the man you see over there, the black man, the man who is wearing a red coat), the speaker will look for this information in the immediate context, i.e. the situation of speaking; being guided by the principle of achieving maximal effect at minimal cost, he will first look around to see whether he can find an adequate referent. If he cannot establish reference in this way, he will take into account the preceding linguistic context; if this also fails, he will make use of his general pragmatic knowledge of the world. If the referent can be found in the situation of speaking or by making use of his knowledge of the world, it implies that the referent exists at the moment of speech (at t0). If the referent can be found in


the linguistic context (i.e. in the NP to which the RC belongs), the tense of the RRC may decide when the referent exists (e.g. the bridge which will be built i.e. which does not exist yet), although the kind of situation also influences the temporal location of the situation (cf. the bridge which will be built vs. the bridge which will be repaired). As the unmarked interpretation of the present tense is present time (the man who is wearing a blue raincoat), a RC with a definite antecedent will more readily be interpreted as referring to the present sector, i.e. eager to assign reference, the hearer will carry out his task with as little effort as possible. This explains why it is rather likely that the present tense will be interpreted as indicating present time if the Ν has not been mentioned before in a future time context: it is the most easily accessible interpretation. The corresponding RC with an indefinite antecedent does not indicate that the speaker believes that the hearer can identify the referent and therefore there will be no similar readiness to interpret a present tense in an indefinite NP as referring to the present sector.

The following example also illustrates the influence of definiteness on the temporal interpretation of the RC:

(47) a. A left-wing student will seize the microphone but will be quickly cut off by an official who cuts the cord, (adapted from SEU)

b. A left-wing student will seize the microphone but will be quickly cut off by the official who cuts the cord.

When the definite article is used in (47b), one gets the impression that there is reference to a person whose task it is to cut the cord whenever something irregular happens; the NP refers to a habitual situation. When the indefinite article is used, this interpretation is less likely, the unmarked interpretation being that there is reference to a "post-present" official who, forced by the particular situation, happens to cut the cord. Example (48) also shows that the definite article may favour a present sector interpretation of a present tense RC:

(48) a. Lady Foxglove will look up with a busy smile at the enormous figure who is clearing (will be clearing) his throat importantly, (adapted from SEU)

192 (W-)anteriority and (W-Simultaneity in post-present sector RCs

b. Lady Foxglove will look up with a busy smile at an enormous figure who is clearing (will be clearing) his throat importantly.

Out of context, the present tense in (48a) is more likely to have present sector reference than that in (48b). The following examples illustrate the same tendency:

(49) a. They will be assisted by the committee which gives financial advice, (adapted from SEU)

b. They will be assisted by a committee which gives financial advice.

(50) a. Correct, ducky! You just wait till you meet the girl who thinks you're a god. (LOB)

b. Correct, ducky! You just wait till you meet a girl who thinks you're a god.

However, the above observations do not necessarily imply that a present tense form in a definite NP is never interpreted as a PPS form. If it is mutually manifest that the RRC situation belongs to the post-present domain established in the SUPC, no confusion will arise about the temporal reference of the SC verb form:

(51) Consultants saw that as significant because Siemens' PBXs offer ISDN capabilities, so the IBM move in that direction makes it more plausible that the companies can merge their PBX technology and minimize disruption for customers. The new software will also let IBM customers' telephones automatically identify the number from which an incoming call originated. (WSJ)

The conclusion to this section is that the definiteness of the NP is an important factor which may hamper a future time interpretation of PPS forms.

5.3.2.3. Influence of lexical material

Lexical material is to be understood here as nouns, adjectives and verbs; it does not include temporal adverbs. The presence of certain nouns referring to entities which already exist at the moment of speech may influence the interpretation of the verbs:


She will retain in the vitality of her spirit, in the vigour of her language, in the occasional peasant quality of humour, a strength which her children have not' and towards which they 'have' at times something of a sophisticated and urbanized 'neshness' (soft squeamishness). (adapted from SEU) She will retain in the vitality of her spirit, in the vigour of her language, in the occasional peasant quality of humour, a strength which her children will not have and towards which they will have at times something of a sophisticated and urbanized 'neshness' (soft squeamishness).

The following factors contribute to establishing a present sector interpretation of have', (a) there is reference to her children, who already exist at t0, and who therefore already possess strength; (b) one can only retain something if one possesses it already. In other words, retain suggests the idea of an existing (present time) charac-teristic. When we replace children by future children, the present tense cannot establish a post-present sector interpretation:

(52) c. She will retain a strength which her future children will not have and towards which they will have at times something of a sophisticated and urbanized 'neshness' (soft squeamishness).

d. She will retain a strength which her future children * have not and towards which they * have at times something of a sophisticated and urbanized 'neshness' (soft squeamishness).

The use of the future tense in both the RC and the SUPC may be required because the element of prediction is strongly present: (a) she will have children, (b) her future children will behave in a particular way. The point made is corroborated by Lakoff (1971), who claims that (53a) is odd because the relative pronoun who requires "that the person referred to either be presupposed to be alive at the time referred to in the relative clause, or thought of as a human being" (Lakoff 1971: 331):

(53) a. * We have just found a good name for our child, who we hope will be conceived tonight. (Lakoff 1971: 331)

(52) a.

b.


b. We have just found a good name for our child, who we hope will grow up to be a good citizen after he is born, (ibid.)

What we notice in (52) is also comparable to what Riddle (1978: 13) remarks in connection with the following example:

(54) ?? Jane declared that her husband is an inveterate gambler and that's why she divorced him. (Riddle 1978: 13)

Riddle comments: "The present tense in [54] is unacceptable because the definite description her husband describes a relationship which does not exist at the time of the speech act. That is, the time that the definite description was valid is inconsistent with the time reference of the verb. If ex-husband were the definite description, then the present tense would be acceptable, because this description would accurately reflect that current state of affairs" (Riddle 1978: 13).

The following example also shows that the lexical content of the RC may contribute to a correct interpretation of the PPS forms:

(55) If you wish to change this default size, press the SET STD key and the current settings will be saved as STANDARD and used for all future documents you create. (SEU)

Create will not be misinterpreted as referring to the present sector because there is reference to future documents.

5.3.2.4. Influence of adverbials

The issue to be considered in this section is whether the presence of an adverbial is sufficient to safeguard a future time interpretation of the RC verb form. Adverbials may be of three types:

(a) Ungebundene Adverbiale (Schöpf 1984: 125) [unanchored adverbials]: Some adverbials are not anchored to a TO, they either locate a situation in the past sector (e.g. in 1983), in the post-prese nt sector (e.g. in 1998), or they are ambiguous (out of context) between a past sector or a post-present sector reference (e.g. soon, before)


(b) Referenzzeitgebundene Adverbiale (Schöpf 1984: 125) [adverbials anchored to a reference time]: This class is constituted by adverbials anchored to a TO which is not the time of speaking (e.g. two weeks before, the previous week,...).

(c) Sprechzeitgebundene Adverbiale (Schöpf 1984: 125) [adverbials anchored to the time of speaking]: Adverbials of this type are anchored to t0 (e.g. two weeks ago, next week).

In this section, the focus will be on adverbials that are unanchored (section (5.3.2.4.1.)) and on adverbials that are anchored to the post-present TO established in the SUPC (section (5.3.2.4.2.)).

5.3.2.4.1. The adverbial in the RC is unanchored

a. If the adverbial refers to a time that lies in the past sector, the verb in the RC will be interpreted as an absolute tense:

(56) Courts have ruled that taxpayers must submit to TCMP audits, but the IRS will excuse from the full-scale rigors anyone who was audited without change in either 1986 or 1987. (adapted from WSJ)

b. If the RC situation can be interpreted as lying either before t0 or before the post-present pseudo-tp but after t0, the (im)possibility of using the PPS to indicate a relationship of anteriority132 will depend to a great extent on the pragmatico-semantic links between the SUPC (the matrix) situation and the RC situation. If, for instance, there is a causal link between the RC situation and the SUPC (or matrix) situation, it will be possible to use the PPS without giving rise to misinterpretations, i.e. the PPS forms will not be misinterpreted as absolute tenses. In (57a), the pragmatico-semantic link between getting cut and bleeding is tight:

(57) a. The bleeding will be caused by the cut she got when he attacked her.

b. The bleeding will be caused by the cut she will get when he attacks her.

c. The bleeding will be caused by the cut she gets when he attacks her.


The situation referred to by means of a past tense {she got a cut) in (57a) is likely to be interpreted as resulting from a post-present sector attack. When the present tense is used (cf. (57c)), the temporal interpretation is one of sloppy simultaneity: a situation which, strictly speaking, happens before another situation, is represented as simultaneous with it. The present tense cannot be misinterpreted because the adverbial (in its non-habitual reading) can only refer to future time. In (58), the RC situation also lies at the origin of the SUPC situation. However, the links are looser than those in (57a). This explains why the RC sentence in (58) allows a past sector as well as a post-present sector interpretation:

(58) She will blame her upset stomach on the medicine she took after she had her meal.

Example (59) is similar to (57b) and (57c): although the adverbial could also be used to refer to a present time habit, the context in (59) results in a future time reading being assigned to the adverbial. The present tense does not give rise to confusion; it is interpreted as a PPS form:

(59) a. There will be losses caused by emergencies that arise while he is away at his off-farm job. (BR)

b. There will be losses caused by emergencies that will arise while he is away at his off-farm job.

However, the next example shows that even if there is an adverbial establishing future time reference, PPS forms may still be ruled out:

(60) a. She will create a very solid foundation on which you will be able to build when you come here, (adapted from SEU)

b. She will create a very solid foundation on which you * are able to build when you come here.

The PPS cannot be used. The RC gives further information about the RC in the sense that it highlights one particular aspect of the antecedent N. Although punctuation does not provide any indication as to whether the RC is a RRC or a NRRC, the latter option is likely to be the one intended, which may explain why a FPS form has to used.


When the adverbial either refers to a time lying before t0 or after it, the interpretation of the past tense as either a PPS form referring to a situation that is anterior to the post-present TO or an absolute tense referring to a situation located in the past sector133 also depends on the contextual knowledge available to the speaker and hearer:

(61) a. A few scuffles will break out among the 600 stu-dents who were allowed into the hall before uni-versity ushers locked the door, (adapted from SEU)

b. A few scuffles will break out among the 600 stu-dents who will have been allowed into the hall before university ushers locked the door.

If it is mutually manifest to both speaker and hearer that the students were not allowed into the hall before t0, the past tense (were) will not be "misinterpreted" as an absolute tense. The FPS can be used as well. The two systems differ in that the PPS represents the anterior situation (the students were allowed into the hall) as a fact, whereas the FPS represents the situation as a prediction. Although any situation that is located in the post-present sector is a prediction,134

the speaker is not morphologically reminded of this when the PPS is used. The claim that the element of prediction is more conspicuously present when the FPS is used can be further explained by comparing the expression of anteriority in the post-present sector with the expression of posteriority in a past domain. When the situation that is W-posterior to the relevant past TO lies before t0, it is sometimes not referred to by means of the conditional tense, because in this way the situation loses its factual character.135 It is no longer represented as a fact (which it definitely is, since it lies before t0), but as a prediction on the part of the speaker (cf. chapter 3, section 3.4.3.):

(62) a. He half-rose and gave an expansive gesture with his hand, which overturned the teapot, pouring its contents neatly into Emma's lap. (SEU)

b. He half-rose and gave an expansive gesture with his hand, which ?? would overturn the teapot, pouring its contents neatly into Emma's lap.

(63) a. Dougan flicked the ball, first time, back to Kenning, whose shot thudded into the net. (SEU)

b. Dougan flicked the ball, first time, back to Kenning, whose shot ?? would thud into the net.


(64) a. The sixth time, they had sent a runner who was hit (...). (SEU)

b. The sixth time, they had sent a runner who ?? would be hit.

Having made this point, a comparison can be made between W-posteriority in the past sector and W-anteriority in the post-present sector. In example (61), there is reference to something that will happen in the future: the students will be allowed into the hall and scuffles will break out among them. When the PPS is used, the situation is considered from a future point of orientation, and although it lies after t0, it acquires a certain degree of factuality. If, however, the FPS is used, it is explicitly made clear that the situation lies after t„; in this case, the element of prediction prevails. The following diagrams illustrate the points made:

past sector post-present sector

fact

χ χ

shift of domain

temporal subordination

PPS

FPS prediction

The next example also shows that the future perfect may have to be used if the SC situation is to be understood as located in the post-present sector, as the past tense is interpreted as an absolute tense:

(65) a. The expected growth in patient population, coupled with recent production economies, will reduce somewhat our financial risk and will remove some of the uncertainties which existed when this drug was first introduced. (WSJ)


b. The expected growth in patient population, coupled with recent production economies, will reduce somewhat our financial risk and will remove some of the uncertainties which will have existed when this drug was first introduced.

Depending on whether ever is understood as until then or until now in (66a) and (67a), the present perfect tense will be processed as a PPS form/absolute tense respectively:

(66) a. You will find that avocado is unlike any other fruit you have ever tasted. (BR)

b. You will find that avocado is unlike any other fruit you will ever have tasted.

(67) a. These will represent one of the biggest reforms in technical education that we have ever made. (LOB)

b. These will represent one of the biggest reforms in technical education that we will ever have made.

c. The next set of examples contain unanchored adverbials which clearly refer to the post-present sector. In (68), until then refers to a period which starts before and leads up to the main clause situation:

(68) a. Its report will be welcomed for the constructive solutions it will provide to the main problems which until then will have hindered full co-operation in the joint use of the research facilities of these institutes, (adapted from SEU)

b. Its report will be welcomed for the constructive solutions it will provide to the main problems which until then ? have hindered full co-operation in the joint use of the research facilities of these institutes.

The acceptability of the present perfect tense is questionable; the future perfect must be used. The PPS also has to be used in the RRC when an adverbial of the type by then occurs in the RC and the perfect is of the indefinite type:

(69) a. I will call up Jimmie, who * has heard the news by then.

b. I will call up Jimmie, who will have heard the news by then.


c. I will call up the man who * has heard the news by then.

d. I will call up the man who will have heard the news by then.

Allen (1966) points out that in Mr. Deχ will go home at five-thirty because he will have finished his work by then (Allen 1966: 178) the use of the future perfect is obligatory, even though "futurity is sufficiently indicated in the main verb" (Allen 1966: 178). RCs and because-clauses with by then or until then differ from ^/-clauses in that the latter require the use of a PPS form (cf. Mr. Dex will go home at five-thirty if he has finished his work by then (Allen 1966: 178)), whereas the former do not. It is strange that the present perfect cannot be used in because-clmses and RCs, as temporal subordination is usually possible in these contexts. When until then is put at the end of the clause, the use of the present perfect in example (68b) seems to be ruled out altogether:

(68) c. Its report will be welcomed for the constructive solutions it will provide to the main problems which ?? have hindered full co-operation in the joint use of the research facilities of these institutes until then.

The use of PPS forms in RRCs with by then is not always impos-sible, witness the following example:

(70) The students who have finished their homework by then will be allowed to go home.

It is probably the conditional meaning implicit in the RC which explains why the PPS form can be used.

The following conclusions can be drawn from the above examples with unanchored adverbials:

a. If the RC contains an adverbial of the until then or by then type, the acceptability of the PPS forms is questionable (cf (68), (69)) unless the RC has conditional connotations.

b. If there is reference to a W-anterior situation and the temporal adverb indicates a time which might just as well come before t0 as after t0, the semantic links between the SUPC situation and the RC situation will help to interpret the RC verb form. If the links are tight, the RC verb will most likely be interpreted as temporally subordinated to the future pseudo-t0 (cf. (57a)); if not, it will be


possible to interpret the verb as establishing its own domain (cf. (58)). If there is ambiguity, the domain will have to be re-established.

5.3.2.4.2. The adverbial in the RC is anchored to the time indicated by the SUPC situation

It is obvious that the PPS forms are more likely to be interpreted correctly if the adverbial is clearly anchored to a future TO. If an adverbial is anchored to a future TO (e.g. two days before) and if the situation lies before t0 (Tomorrow, John will take out the girl he met ?? two days before/yesterday), the speaker does not conform to the communicative rules in the sense that he chooses to use a form requiring more processing effort; the use of an adverbial anchored to t0 would require less processing effort in this case.136 Therefore, a cooperative speaker who uses an adverbial that is anchored to a future TO will only do so if the situation lies after t0 (cf. section 5.3.1.2.). This explains why PPS verb forms in a sentence with an adverbial of this type will not be interpreted as shifts of domain:137

(71) a. Students of the engineering faculty will continue a sit-in protest which will have begun the day before. (adapted from SEU)

b. Students of the engineering faculty will continue a sit-in protest which will begin the day before.

c. Students of the engineering faculty will continue a sit-in protest which began the day before.

d. Students of the engineering faculty will continue a sit-in protest which ? begins the day before.

The past tense cannot be understood as taking t0 as TO, because there is an adverbial which explicitly takes the future TO as reference point. Will begin is less likely to be used precisely because an adverbial expressing anteriority is used in combination with a verb form indicating posteriority with respect to t0. The same observation explains why the following example from Comrie is only marginally acceptable: the combination of an element indicating posteriority (the t0-anchored adverbial indicating future time) and an element indicating anteriority (the PPS past tense) results in an unacceptable sentence:


(72) ? On Friday, Oswald will say that he arrived tomorrow. (Comrie 1986a: 287)

Begins in (7Id) is possible only if the speaker is relatively sure about the time when the protest is to begin. In this case, the present tense is the result of a shift of perspective: a tense normally establishing present sector time reference is used to locate a situation in the post-present sector. This option is probably less likely to be chosen for reasons similar to that explaining the questionable acceptability of the future tense. The future perfect, unlike the past tense, emphasizes the element of prediction; like the past tense, it conveys a high degree of factuality (i.e. higher than that of the future tense). The past tense cannot be misinterpreted in the following example either:

(73) a. The building of Brasilia will inspire the first long forest road - the 2,123 km from Belem to Brasilia -which will bring into existence 110 towns in the next decade, a jump in population from 100,000 to two million and some five million cattle which (??) will not have existed before, (adapted from SEU)

b. The building of Brasilia will inspire the first long forest road - the 2,123 km from Belem to Brasilia -which will bring into existence 110 towns in the next decade, a jump in population from 100,000 to two million and some five million cattle which ?? will not exist before.

c. The building of Brasilia will inspire the first long forest road - the 2,123 km from Belem to Brasilia -which will bring into existence 110 towns in the next decade, a jump in population from 100,000 to two million and some five million cattle which did not exist before.

The (im)possibility of using the FPS, more specifically, the future perfect in examples of this type, depends on where the RC situation is located in absolute terms. This explains why the question marks indicating questionable acceptability have been put in brackets in (73 a): if the situation lies after t0, the future perfect is an acceptable form, if the situation lies before t0, the use of the future perfect will be questionable. The latter observation does not contradict Comrie's claim that the future perfect can be used even if the situation lies before t0, because in his example (If it rains tomorrow, we ΊI have


worked in vain yesterday (Comrie 1985: 73)) an "absolute" time adverbial is used, i.e. one that is anchored to t0. If the speaker uses the future perfect together with an adverb anchored to a future TO to refer to a situation lying before t0, he chooses to use a form requiring more processing effort; if a situation is located before t0, the unmarked possibility is that the adverbial specifying that time is anchored to t0 and not to some future TO. It will be clear that once again it is the principle of maximal contextual effects at minimal cost that underlies this observation, unless the additional processing effort is offset by extra contextual effects. In (74), the future perfect is acceptable if forty-five years before does not indicate a point in time before t0:

(74) a. The governor will be an elderly man, and he will have the clearest possible recollection of the British who (??) will have come with Colonel Y o u n g h u s b a n d y e a r s before, (adapted from SEU)

b. The governor will be an elderly man, and he will have the clearest possible recollection of the British who ?? will come with Colonel Younghusband forty-five years before.

c. The governor will be an elderly man, and he will have the clearest possible recollection of the British who came with Colonel Younghusband forty-five years before.

The low acceptability of the future tense is due to the combination of a verb form expressing posterior time and an adverbial expressing anteriority (cf. (71) and (72)). The next example also contains an adverbial that is anchored to the SUPC TO:

(75) a. I feel it will bring the advantages of bulk to many who have previously been deterred by the thought of having to spend +4,000, because one can literally start off with spending only about +118 for one of these bins ex works. (LOB)

b. I feel it will bring the advantages of bulk to many who will previously have been deterred by the thought of having to spend +4,000, because one can literally start off with spending only about +118 for one of these bins ex works.


The present perfect is understood as a PPS form. When the future perfect is used, the speaker puts extra emphasis on the predictive force of his statement.

Two more observations must be made: a. When the RRC has conditional meaning, it is seldom possible to

use the FPS, even if there is an adverbial with future time reference in the RC (cf. (76b)). This follows from a more general constraint on conditional clauses. The following examples illustrate the point made:

(76) a. We will make use of every new method that works (* will work) to get the houses up and keep the prices down. (SEU)138

b. It is now 8.45 a.m. You are to start work at 9 a.m. Any man who has finished (* will have finished) his work before mid-day will get a bonus. (Penhalluriack 1981: 222)

The examples in (77) are exceptions to the rule that it is impossible to use a FPS form in a RC with conditional connotations:

(77) a. If you hold or are applying for a scholarship or other award .... Those who will not be in receipt of a scholarship or those whose award will be inadequate to meet the full fees and expenses, please state how you propose to meet those fees and expenses... (adapted from Haegeman 1983: 153)

b. Those who will be amused by it, will be told a joke, (adapted from Comrie 1982: 150)

c. Those who will be untidy will be punished, (adapted from Declerck 1991a: 212)

RCs of this type will not be dealt with in more detail; whenever the conditional clause allows the use of the future tense, the cor-responding RC with conditional connotations does so as well.139

b. The second point relates to the order in which the RC and SUPC are given. If the RC precedes the SUPC, the domain in the RC will often have to be re-established by a verb form from the FPS, even if there is an adverb with future time reference in the RC:

(78) Meanwhile their rivals who by then will have (* have) huge populations of fruit-flies which they are subjecting to stress


from heat, chemicals and anything else they can think of, will be beginning to map on the chromosomes the actual positions of some of the mutated genes, (adapted from SEU)

By then again requires the use of a future tense (cf. (69)). The present tense can be used if it is the result of a shift of perspective (cf. (79)), in which case it is not a PPS form, or if a certain class of people or type of person is defined (cf. (79b)):

(79) a. The patient who arrives on Monday will be examined by Dr. Smith,

b. The patients, if any, who arrive on Monday, will be examined by Dr. Smith.

In (79a), arrives is an absolute present tense used to refer to the post-present sector. In (79b), a certain class of patients is defined, i.e. those who arrive on Monday.

5.3.2.5. Linear order of the SUPC and RC

5.3.2.5.1. As has just been pointed out, the order in which the SUPC and RC are reported also influences the choice of the verb form. It seems reasonable to expect that whenever a RC precedes the SUPC, the present tense will automatically be interpreted as an absolute tense establishing a present domain. The following examples seem to confirm this hypothesis:

(80) a. This executive says that business should hold up even during a recession. And with fewer airlines remaining, he concludes, "those of us who are left will make a lot of money". (WSJ)

b. This executive says that business should hold up even during a recession. And with fewer airlines remaining, he concludes, "those of us who will be left will make a lot of money".

If the executive wants to refer to the airlines who will be making a lot of money because competitors will go into failure due to the impending recession, he will have to use a future tense, as the only


interpretation the present tense allows is one in terms of present time reference.

The following examples all have the structure: "abstract definition + to be + concrete example"; they are specificational sentences, a specificational sentence being a sentence "whose semantic function is to specify a value for a variable" (Declerck 1988: 2). There is a relationship of W-simultaneity between the RC and the SUPC:

(81) a. In the whole field of discrimination against women, the area which will cause the most active, immediate suffering, will be the economic insecurity of wives, (adapted from SEU)

b. In the whole field of discrimination against women, the area which causes the most active, immediate suffering, will be the economic insecurity of wives.

(82) a. One aspect which will concern the Falklanders will be Lord Chalfont's - and the British Government's -assumption that they suffer great inconvenience through Argentina's cold-shouldering of them, (adapted from SEU)

b. One aspect which concerns the Falklanders will be Lord Chalfont's - and the British Government's -assumption that they suffer great inconvenience through Argentina's cold-shouldering of them.

(83) a. One of the major problems to which constant reference will be made by Schools will be the need to alter the composition of the Senate so as to give a greater degree of institutional representation, (adapted from SEU)

b. One of the major problems to which constant reference is made by Schools will be the need to alter the composition of the Senate so as to give a greater degree of institutional representation.

(84) a. Perhaps the Pirate who will be the unhappiest over the news that the Musical probably will sit out most of the series is Bob Friend, who was beaten by The Man twice last season on dramatic home runs. (BR)

b. Perhaps the Pirate who is the unhappiest over the news that the Musical probably will sit out most of the series will be Bob Friend, who was beaten by The Man twice last season on dramatic home runs.


When the RC precedes the main clause, the FPS must be used; if not, the situations are interpreted as holding at t0. When the present tense is used, the Maxim of Manner140 is flouted; the hearer initially interprets the RC in terms of a temporal relation with t0 (which is the unmarked interpretation), but on hearing the SUPC he is forced to give up this interpretation. If we reverse the order of the SUPC and the RC, it turns out that the present tense can be used as a relative tense expressing simultaneity in a post-present domain. In this case, it seems most natural to use the PPS in the RC:

(81) c. The economic insecurity of wives will be the area which will cause the most active, immediate suf-fering.

d. The economic insecurity of wives will be the area which causes the most active, immediate suffering.

(82) c. Lord Chalfont's - and the British Government's -assumption that they suffer great inconvenience through Argentina's cold-shouldering of them will be one aspect which will concern the Falklanders,

d. Lord Chalfont's - and the British Government's -assumption that they suffer great inconvenience through Argentina's cold-shouldering of them will be one aspect which concerns the Falklanders.

(83) c. The need to alter the composition of the Senate so as to give a greater degree of institutional rep-resentation will be one of the major problems to which constant reference will be made.

d. The need to alter the composition of the Senate so as to give a greater degree of institutional rep-resentation will be one of the major problems to which constant reference is made.

(84) c. Bob Friend will be the Pirate who will be the unhappiest over the news that the Musical probably will sit out most of the series,

d. Bob Friend will be the Pirate who is the unhappiest over the news that the Musical probably will sit out most of the series.

Example (85) is slightly different:


(85) a. Cases which will be relevant will be [Name] Ltd. v. [Name Name Word] Ltd [2222] (...). (adapted from SEU)

b. Cases which are relevant will be [Name] Ltd. v. [Name Name Word] Ltd [2222] (...).

c. [Name] Ltd. v. [Name Name Word] Ltd [2222] and [Word & Word] Ltd v. [Name] [2222] will be cases which are relevant.

d. [Name] Ltd. v. [Name Name Word] Ltd [2222] and [Word & Word] Ltd v. [Name] [2222] will be cases which will be relevant.

In (85), the subject NP is indefinite plural and defines a certain type of cases, i.e. relevant ones. This type of case is not restricted in time:141 relevant cases have always existed and will continue to do so. The fact that future time reference is established in the SUPC suffices to yield a "correct", post-present interpretation of the subject (cf. examples in (93)). Moreover, the SUPC future tense follows the present tense in the RC closely in (85b), which also contributes to establishing a post-present sector reading of the present tense verb.

It is more difficult to answer the question whether or not a SUPC can be temporally subordinated to a preceding future tense RC:

(81) e. In the whole field of discrimination against women, the area which will cause the most active, immediate suffering, is (will be) the economic insecurity of wives.

(82) e. One aspect which will concern the Falklanders is (will be) Lord Chalfont's - and the British Gov-ernment's - assumption that they suffer great inconvenience through Argentina's cold-shouldering of them.

(83) e. One of the major problems to which constant reference will be made by Schools is (will be) the need to alter the composition of the Senate so as to give a greater degree of institutional representation.

(84) e. Perhaps the Pirate who will be the unhappiest over the news that the Musical probably will sit out most of the series is (will be) Bob Friend, who was beaten by The Man twice last season on dramatic home runs.


(85) e. Cases which will be relevant to this consideration are (will be) [Name] Ltd. v. [Name Name Word] Ltd [2222], (...).

The future tense has to be used in the SUPC, unless one is com-pletely sure that the economic insecurity of wives will indeed be the area which causes the most suffering (cf. (81)), that Lord Chalfont's assumption will definitely be an aspect that concerns the Falklanders (cf. (82)), that the need to alter the composition of the senate will be the major problem that will be referred to (cf. (83)), that Bob Friend will be the unhappiest Pirate (cf. (84)), that the cited cases will be the ones which will be relevant (cf. (85)). Future time reference is established in the RC. When the present tense is used in the main clause, only one prediction is made (RC situation) and one fact is given (SUPC situation). When the future tense is used, two predictions are made (cf. chapter 6, section 6.2.). Declerck (1988) also touches upon the question whether or not PPS forms or FPS forms are used in specificiational sentences. He writes that

when the variable is expressed in the form of a WH-clause, the rule that the tense of the copula must agree with that of the WH-clause is often obscured by the fact that 'tense simplification' is then often possible and sometimes even more or less obligatory. Compare: (...) (29) (a) The one who will win is / ?? will be Fred.

(b) Fred will be the one who wins / ? will win. (30) (a) ?? It will be Fred who will win.

(b) It is Fred who will win. (c) It will be Fred who wins. (Declerck 1988: 82)

As I see it, the future tense can be used in the main clauses in (29) and (30a), the difference between the future and the present tense being comparable to that described in connection with examples (81) to (85): when the present tense is used, one prediction is made, when the future tense is used, two predictions are made. If the sentence is not specificational, as in (86), the future tense has to be used in the SUPC which follows the future tense RC:

(86) a. For instance, if - as in what follows in this chapter -it is necessary to talk at length about the resem-blances between two things (...) then the sheer


frequency with which the word 'resemblance' (unqualified) will appear on the page ?? creates {will create) a disposition to think that this is the most important aspect of the relation between the two things brought together in a metaphor, (adapted from SEU)

b. ... will be created by the sheer frequency with which the word 'resemblance' appears (will appear) on the page.

5.3.2.5.2. The sequential order RC - SUPC does not always require the use of the FPS in the RC. In the following cases PPS forms are acceptable:

a. When the SUPC future tense immediately follows the RC pre-sent tense, the PPS forms can be used more easily in the RC:

(87) The way in which she behaves will reveal that she does not feel at ease.

b. Example (88) shows that when the RC preceding the SUPC is couched in a post-present context, the use of FPS forms in the RC may not be necessary:

(88) a. Everybody will look for the body. The one who finds (has found) it, will receive a reward later,

b. Everybody will look for the body. The one who will find / ? will have found it, will receive a reward later.

A similar comment applies to the example in (89):

(89) In addition to the valuable life assurance that this plan provides, once you reach age 60 you will receive a con-siderable cash sum, amounting on current bonus rates to at least all the premiums you have paid - and maybe much more. The cheque you receive will be entirely tax-free, un-der current legislation. (SEU)

It is not merely because will be immediately follows receive that the PPS can be used in the RC. The context preceding receive clearly indicates that there is reference to a cheque which will be received in


the future; as the hearer does not have an assurance yet, he cannot have paid any premiums at the time when the statement is made. The temporal adverbial once you reach age 60 also contributes to establishing a post-present sector interpretation. Example (90) is again similar:

(90) Richard Dimbleby will interview people in each island from a point in Guernsey, and the musical records they ask for will be played from a London studio. (Buyssens 1968: 250)

However, it is not because PPS forms do not give rise to misin-terpretations in examples of this type that they are always the verb forms that are used. The following corpus examples show that the speaker may all the same decide to use FPS forms:

(91) The details of the suburban concerts next season, and the centers in which they will be given, will be announced later, Mr Toobin said. (BR)

(92) Whether such an apparatus can be incorporated in a reactor circuit in a manner that will satisfy safety requirements will need further study. (LOB)

c. When there is reference to a particular type or class of people or things the present tense can mostly be used, even when the RC precedes the SUPC. It will be clear that a future tense must be used in the SUPC:

(93) a. The human beings and events which emerge (will emerge) from rigorous historical enquiry and after severe historical criticism will often be stranger, more idiosyncratic and in reality more interesting than what is normally supplied by the common-places of romanticism, or of political partisanship, (adapted from SEU)

b. 'The brick lining which stores (will store) the heat will be nice and solid, and opening the doors has virtually no effect on the temperature,' said Mr Coombs, (adapted from SEU)

c. The lesser persons who appear (will appear) will also be described vividly; the old hermit to whom the Red Cross Knight goes, with his beautiful white hair like spangles of frost; the wandering old Ignaro;


small detailed touches of people in a crowd, like the mother whose child touched the dead dragon; and, particularly vivid, the pageant of sins in Bk I, surely one of the cleverest portrayals of them ever, (adapted from SEU)

d. The people who are promoting (will be promoting) it will be reasonably cultured & 'right' - 1 hope, (adapted from SEU)

e. The people who are showing (will be showing) signs of enthusiasm will be looking at those modern problems which centre on the affluent society and at groups with special needs, (adapted from SEU)

f. People who behave (will behave) worse than ani-mals will not in the least be deterred by them." (adapted from SEU)

g. It is salutary to remember that even in the case of Jesus Christ, although He is the incarnate Son of God, and filled with the Holy Spirit, not everyone who sees (will see) Him will recognize Him as the Son of the Most High, (adapted from SEU)

h. You will wish to know and be able to verify my company's identity and corporate substance, for the most important factors that determine (will determine) a security purchase decision will be that the products source is reputable and stable; that the product is effective, reliable and guaranteed; and supported by advice and service throughout its life, (adapted from SEU)

In example (93d) and (93e), a continuous tense is used. As pointed out before, a verb in the progressive form is more likely to be interpreted as indicating present time than a non-progressive verb (cf. section 5.3.2.1.). In other words, the RC will probably be understood as referring to the present. Another point is that RCs in the present tense depending on a definite antecedent (cf. (93a)-(93e), (93h)) differ from those depending on an indefinite antecedent (cf. (930, (93 g)) in that the likelihood of the former being understood as indicating present time is greater (cf. section 5.3.2.2.). It must also be added that the presence of deictic elements or other words explicitly referring to the speaker's here-and-now may hamper a "general class interpretation". The presence of the word this in (94) and these in


(95) is sufficient reason for some people to reject the use of the pre-sent tense as a means of indicating future time:

(94) Those Labour members who use (will use) this issue as a stick to beat the Government will be as much in the wrong, or almost as much so, as those tireless Tories who will seek to use it in their selfish battle against nationalization and their self-interested attack on the Civil Service, (adapted from SEU)

(95) It is therefore the writer's opinion that general purpose computers have no long term significance so far as process or machine tool control is concerned. Rather the data loggers and computers which will be used (are used) in these circumstances will be small black boxes, designed to do specific jobs. (LOB)

It may well be that the above-mentioned class of RRCs (which form part of an NP referring to a type or a class) should be extended to include any RRC used in a NP that is attributive in Donnellan's sense (Donnellan 1966: 285):142

(96) The one who is reluctant to carry out his plans his plan will be executed.

(97) The instructions that are hidden in the dustbin will tell you what to do next.

If the NPs are interpreted attributively (i.e. a specific person or set of instructions is intended, though not yet clearly identifiable), the NP will not be understood as being located at a particular time. It is not until the SUPC is mentioned that some additional temporal in-formation is given about the NP: there is reference to someone who may be reluctant to carry out the plan in the future (96) or instructions which will be hidden in the dustbin in the future (97). However, one must be extremely cautious when trying to make generalizations, as the temporal interpretation of attributive NPs de-pends to a great extent on the context in which they are used. If, for instance, (96) is used in a context in which A and Β are at a meeting of the mafia and A mentions to Β that there is someone (C) present who is reluctant to carry out the relevant person's plans, it may be that Β does not know exactly who A is talking about. However, if Β goes on to talk about C and utters (96), A will know that C is present, although he cannot actually indicate him. However, if A is talking

214 (W-)anteriority and (W-Simultaneity in post-present sector RCs

about people in general and C has not been explicitly mentioned before as being present now, the present tense is more likely to be interpreted as establishing post-present reference. A similar line of reasoning applies to (97).

Although the different factors influencing the interpretation of the verb form in the RC are treated separately in different sections, there are usually a number of factors which interact and establish a particular interpretation:

(98) There is still one glittering prize to be grasped. The man who captures it will go down in history as one of the greatest of mortals. (LOB)

The NP in which the RRC is used is attributive. The sentence preceding the RC indicates that there is reference to a future event and the RC has conditional connotations. These three elements, together with the fact that a non-progressive verb form is used, contribute to the post-present sector interpretation of captures. The following example also shows that the attributive character of the NP is usually not the only factor which safeguards a post-present sector interpretation of the verb form:

(99) "The market is so large that no company will dominate it, but there's no question that the one who gets there first will have a significant advantage," says, Roger Guidi, vice president of marketing for Visx in Synnyvale, Calif. (WSJ)

Apart from the fact that the NP is attributive, the conditional connotations of the RRC and the context result in gets being pro-cessed as a PPS form.

d. When the semantic content of the RC and pragmatic knowledge suggest that the situations or referents still have to be realized, the present tense can be used in the RC:

(100) a. 3. The objects for which the company is established will be: (A) To acquire the freehold property known as 18 South Villas in the London Borough of Camden and to hold the same as an investment for the benefit of the lessees of the flats comprised therein, (adapted from SEU)

b. 3. The objects for which the company will be established are : (A) To acquire the freehold property


known as 18 South Villas in the London Borough of Camden and to hold the same as an investment for the benefit of the lessees of the flats comprised therein.

c. To acquire the freehold property known as 18 South Villas in the London Borough of Camden and to hold the same as an investment for the benefit of the lessees of the flats comprised therein will be the objects for which the company is established.

d. To acquire the freehold property known as 18 South Villas in the London Borough of Camden and to hold the same as an investment for the benefit of the lessees of the flats comprised therein will be the objects for which the company will be established.

Although the verb combinations will be established-are (100b) and is established-will be (100a) do not result in a different truth value of the specificational sentence, they are not completely interchangeable. When the former set of verbs is used, the speaker knows why the company will be established. The latter set indicates that the speaker is sure that the company will be established, but he makes a prediction about the motives for which this will happen. The verb establish automatically suggests reference to something which does not yet exist, something yet to come.143 Example (101) also contains lexical material which establishes future time reference:

(101) a. Newspapers will give it splash treatment, but the arguments that follow will sound exactly like those I heard when I was living there 10 years earlier, (adapted from SEU)

b. Newspapers will give it splash treatment, but the arguments that will follow will sound exactly like those I heard when I was living there 10 years earlier.

In (101a), it may not be the presence of the verb follow as such which safeguards a post-present sector interpretation of the RC. The use of a future tense before the RC no doubt also contributes to the correct interpretation of the present tense RC.

However, the following example shows that the generalization does not apply indiscriminately in the sense that the speaker always chooses to use PPS forms in case the semantics of some of the words entail future time reference:


(102) a. The aim of this chapter has been to paint the background and landscape of the Madrid picture. The subsequent pattern that will emerge will not be the comparatively regular one of the pueblo but, rather, a jigsaw of interlocking social relationships which merge their various forms and colours. (LOB)

b. The aim of this chapter has been to paint the background and landscape of the Madrid picture. The subsequent pattern that ? emerges will not be the comparatively regular one of the pueblo but, rather, a jigsaw of interlocking social relationships which merge their various forms and colours.

In spite of the presence of a future tense immediately after emerges and the future time reference inherent in the words subsequent and emerges, the acceptability of the PPS form is questionable.

e. When a time adverbial with future time reference precedes the RC, the PPS can be used in the RC:

(103) a. When this happens, the gene which they invade will be incapacitated and very often the next gene in sequence will also be affected, (adapted from SEU)

b. When this happens, the gene which they will invade will be incapacitated and very often the next gene in sequence will also be affected.

In this example the adverbial establishes future time reference; it specifies the TE which functions as post-present TO. The RC indicates a type of gene (one that is invaded). This, together with the use of the future tense in the SUPC safeguards a correct temporal interpretation. Example (104) is similar:

(104) a. When the invading armies of the Allies reach Belsen, the troops who witness the horrors of the camp for the first time will be so affected by the sights and the smells and the testimony to human depravity that many of them will be taken ill and will have to be sent home on leave, (adapted from SEU)

b. When the invading armies of the Allies reach Belsen, the troops who will witness the horrors of the camp for the first time will be so affected by the sights and the smells and the testimony to human depravity that many


of them will be taken ill and will have to be sent home on leave.

There is reference to a group of people who will be involved in a certain future situation, the future time reference being established by the adverbial. The following example also illustrates the same point:

(105) Under the bylaws change, shareholders who leave the firm will have to wait significantly longer to get paid for all their Drexel holdings than they used to - about 5. (WSJ)

f. The presence of an adverbial in the RRC does not always imply that PPS forms can be used (cf. example (60)):

(106) These findings, and others which will in time be developed (?? are developed), will affect the method of glot-tochronological inquiry. (BR)

When the RC has conditional implications, the PPS system can be used:

(107) Belgium decided that investors who demand the delivery of their securities when they buy shares or domestic bonds will have to pay an additional 100 Belgian francs. (WSJ)

g. It is sometimes difficult to interpret the non-progressive present tense as referring to the present sector because the present continuous would be the "normal" form used to locate the situation in the present sector (cf. section 5.3.2.1.). This, together with the presence of a future tense verb form in the SUPC, makes it more likely that the hearer will interpret the RC as referring to the future. This observation applies to the single event reading of a situation containing a dynamic verb:

(108) a. The blood tests which we do for the bruising which mother will report, but which I will never actually see, will be quite normal, (adapted from SEU)

b. The blood tests which we will do for the bruising which mother will report, but which I will never actually see, will be quite normal.

(109) a. The manservant who opens the door to Gerald will be as old as his master, (adapted from SEU)


b. The manservant who will open the door to Gerald will be as old as his master.

(110) a. I don't think they have even heard of C. Day Lewis, and the poor man who reads from Louis MacNeice will get a poor reception, (adapted from SEU)

b. I don't think they have even heard of C. Day Lewis, and the poor man who will read from Louis MacNeice will get a poor reception.

By way of conclusion to this section on the influence of the order in which clauses are reported, I will comment on the following example from Leech, which he gives to illustrate the claim that "the Simple Present is also used in some that-clauses and relative clauses of future reference" (Leech 1971: 65):

(111) The man she marries will have to be rich. (Leech 1971: 65)

The RC precedes the SUPC and is embedded in a definite NP. The NP will first be interpreted as referring to a well-identified man who exists at t0 (referential use of a definite NP). However, as we read on, this option has to be abandoned; it becomes clear that the speaker is defining a particular kind of man, i.e. one she will want to marry: a specific, but not yet found exemplar (attributive use of the definite NP). The present tense can be used because the SUPC future tense immediately follows the RC present tense. Moreover, the RC also has a conditional connotation: if he is to marry her, he will have to be rich.

5.3.2.6. Gnomic quality of the RC situation

The use of PPS forms in RRC gnomic (Declerck 1991a: 282) statements, relating to irreversible, unalterable situations, will not be interpreted as referring to the post-present sector:

(112) a. Flooding will result in iron, aluminium, manganese and calcium becoming more soluble, and as a result these elements chemically will release the phosphorus which under aerated, dry-soil conditions is chemically bound up with them, (adapted from SEU)


b. Flooding will result in iron, aluminium, manganese and calcium becoming more soluble, and as a result these elements chemically will release the phosphorus which under aerated, dry-soil conditions will be chemically bound up with them.

The w/nc/z-clause is gnomic, i.e. it refers to a permanent charac-teristic of the relevant elements. The present tense can only be interpreted as referring to present time; there is reference to a transformation that is known (cf. Malcolm 1987: 38). The future tense must be used if the transformation is a nonexistent but expected one. However, the fact that there is reference to an irreversible law implies that the characteristic or situation will still hold in the post-present sector. Although the SUPC does not have future reference (unlike the other examples discussed in this chapter) in the following examples, the difference in effect between the present tense in (113a) and the future tense in (113b) in the following Dutch sentences is similar to that in example (112):

(113) a. Dit is een drankje waar je van opknapt. (Brisau 1977:57) [This is a potion which fixes you up.]

b. Dit is een drankje waar je zal van opknappen. (Brisau 1977: 58) [This is a potion which will fix you up.]

"[113a] suggests a context in which the potion is shown and its properties explained; [113b] (...) suggests] that it is shown to a sick person with the assurance that it will fix him up if he drinks it" (Brisau 1977: 58). Although the potion's characteristics are referred to in the present tense in (113a), its healing qualities will still be effective in the future. Examples of the type in (112) are somewhat like the "class-examples" discussed in section 5.3.2.5.2. in that the presence of a future time domain is likely to induce the hearer to interpret the RC situation as a future time instantiation of the general rule:

(114) With the cell electrodes consisting of sodium with oxygen at different activities a voltage will be developed that is a function of the difference in the oxygen potential at the two electrodes. (LOB)


(115) If any adaptation occurs in these cells as a result of prolonged firing, when a T-contour in a slightly different orientation to the I line is exposed on the same part of the retina, the cells fired maximally by it will be ones which are normally maximally responsive to contours in ori-entations lying further away from the orientation of the I-figure. (LOB)

It should still be added that in some contexts, the future tense is used to represent a present time gnomic quality as one which will be evaluated at a future moment in time (e.g. Oil will float on water. (Declerck 1991a: 87)).

5.3.2.7. FPS forms misinterpreted as PPS forms

It is not only the use of certain PPS forms (the present tense/present perfect/past tense) which is sometimes avoided because it results in a "faulty", i.e. non-post-present interpretation; the future tense as well can sometimes not be used for similar reasons. In some examples, an absolute future tense establishing for instance a W-simultaneous do-main is (or can be) misinterpreted as a relative future tense referring to a posterior post-present situation. The latter interpretation is possi-ble because apart from being used as an absolute tense to establish a W-simultaneous post-present domain, the future tense is also the rel-ative tense which indicates that a situation is posterior to a post-pre-sent TO (cf. section 5.1.). In examples of this type, the PPS must be used to avoid misunderstandings:

(116) a. No religious group, he declared in an interview, will receive Peace Corps funds unless it forswears all proselytizing on the project it proposes. (BR)

b. No religious group, he declared in an interview, will receive Peace Corps funds unless it forswears all proselytizing on the project it will propose.

(117) a. He will not curb his instinctual desires but release the energy within him that makes him truly and fully alive, even if it is only for this brief moment before the apocalypse of annihilation explodes on earth,

b. He will not curb his instinctual desires but release the energy within him that will make him truly and fully


alive, even if it is only for this brief moment before the apocalypse of annihilation explodes on earth.

The future tense in (116b) and (117b) unequivocally refer to a situation that is posterior to the SUPC situation. Example (118) is similar:

(118) a. For </T/> very small, this means that the energy </E/> will take as low a value as possible, for </T/> very large, the equilibrium state will be the one for which </S/>, the entropy, is a maximum - that is the state which can be realised in the greatest number of ways. (SEU)

b. For </T/> very small, this means that the energy </E/> will take as low a value as possible, for </T/> very large, the equilibrium state will be the one for which </S/>, the entropy, will be a maximum - that is the state which can be realised in the greatest number of ways.

Although it might be argued that the future tense is not completely unacceptable as a means of establishing a W-simultaneous domain in (118b), its unmarked interpretation seems to be that the entropy will be a maximum at a moment in time after the equilibrium state has been established. It is obvious that such an interpretation does not make sense; therefore, the present tense will be used in preference.


The above discussion has revealed that the exclusive use of the FPS in NRRCs should be ascribed to (a) the function and syntactic status of NRRCs and (b) the intrinsic nature of the PPS forms. PPS forms may be mistaken for absolute tenses; in NRRCs, they are always interpreted as referring to sectors other than the post-present, whereas in RRCs, they may establish post-present time reference. The discussion of the factors which influence the interpretation of verb forms in RRCs shows that it is rather difficult to make generalizations; the interaction of a number of factors (adverbials, the definite article, possible conditional connotations, the order of RC and SUPC, the gnomic quality of the RC, aspect) determines whether or not the relevant verb forms are likely to be "misinterpreted". A number of tendencies can be observed, but


pragmatic, especially contextual, knowledge is probably the most important factor in this respect.

It appears that a similar difference in the tense system occurs in other types of clause allowing an adjunct and a disjunct (Quirk et al. 1985: 1070)144 reading. Haegeman - Wekker (1984) show that there are differences in the syntactic behaviour between //"-clauses in which will can be used and those in which it cannot. Clauses of the latter type may be called central, those of the former type peripheral (Haegeman - Wekker 1984: 54). The distinction is equivalent to that between bound and free clauses which Allen (1966) makes when discussing the difference between the clauses in which the future tense is used to establish future time reference (free clauses) and those in which it is not (bound clauses). A free clause is

not closely bound to the rest of its sentence. A clause introduced by one of the following includers is usually a free clause: although for such (that) as (= because) inasmuch as though because since (= because) while (= although) except (that) so (that). (Allen 1966: 177)

However, he stresses the fact that the typically free clauses he mentions are occasionally bound:

(119) a. You think that I won't visit Mrs. Fustle when I'm in Washington next week because I won't have time. (Allen 1966: 179)

b. I won't visit Mrs. Fustle because I don't have time -1 won't visit her because I can't stand the woman, (ibid.)

RRCs are not necessarily bound either:

(120) a. The man who will speak at [sic] assembly tomorrow will be here for most of the day. (free) (Allen 1966: 178)

b. I'll try to get the autograph of the man who speaks at [sic] assembly tomorrow, (bound) (ibid.)

c. I'll try to get the autograph of the man who will speak at [sic] assembly tomorrow, (free) (ibid.)


In other words, the distinction Allen makes closely corresponds to Declerck's observations about the use of the FPS and the PPS in RRCs and NRRCs.

The claim that temporal clauses, concessive clauses, clauses of comparison and clauses of reason either get an adjunct or a disjunct reading is not new (cf. e.g. Hirtle 1981: 221, Quirk et al. 1985: 1070, Rutherford 1970, Sampson 1971, Tregidgo 1979: 194-195):

(121) a. While John's wife was in hospital he did all the housework.

b. While John's wife studied literature, John will study linguistics. (Haegeman - Wekker 1984: 54)

(122) a. When John's wife was ill he did all the housework. b. He always walks to work, when really he could afford

to travel by taxi. (Haegeman - Wekker 1984: 54) (123) a. Stir the milk continuously so that it doesn't stick.

b. He has stirred the mixture too much, so that it won't set now. (Haegeman - Wekker 1984: 54)

In (121) up to (123), the meaning of the subclause changes depending on whether it gets an adjunct (a-examples) or a disjunct reading (b-examples): from temporal to concessive in (121) and (122) and from a purpose clause to a result clause in (123). The following examples from Rutherford (1970: 97) show that a change from adjunct to disjunct does not necessarily coincide with a change in the type of subclause:

(124) a. He's not coming to class because he's sick. b. He's not coming to class, because he just called from

San Diego. (125) a. She loves her husband (even) though he beats her.

b. She loves her husband, (al)though (I know) he beats her.

(126) a. Mary will marry John unless the fortune teller is too pessimistic.

b. Mary will marry John, unless the fortune teller is too pessimistic.

(127) a. He'll take his umbrella in case it rains. b. He'll take his umbrella, in case you're wondering.

(128) a. Mary will marry John whether the fortune teller predicts it or not.


b. Mary will marry John, whether the fortune teller predicts it or not.

It appears that when the subclauses have future reference, the adjunct type allows the use of a PPS form or a FPS form. The disjunct counterparts require the use of a FPS form if reference to the post-present sector is to be established:

(129) a. You will dress as he tells you. b. You will dress as he will tell you. c. You will dress, as he will tell you.

Sentence (129c), unlike (129a) and (129b), is a sentential RC. (129a) and (129b) can be paraphrased as You will dress in the way he tells you (now) (129a) / will tell you (then) to dress ((129a) and (129b)). Sentence (129c) means You will have to put on clothes, as he will tell you, i.e. this is what he will tell you to do.

As in the case of RRCs, misunderstandings may arise as to whether the present tense in an adjunct clause refers to the present or to the post-present sector:

(130) a. He will not be present at the ceremony because he will be (is) on a business trip,

b. He will not be present at the ceremony, because he will be on a business trip.

The present tense version of (130a) can be paraphrased as It is because he is on a business trips/he will be on a business trip that he will not be present. The present tense either has present sector of post-present sector reference. Example (131) is similar:

(131) a. He will not beat her because she will be pregnant / is pregnant.

b. He will not beat her, because she will be pregnant, (cf. Tregidgo 1979: 194-195)

The sentence in (131a) allows two interpretations (no matter which verb form is used): 1. It is not because she is pregnant that he beats her, but for some other reason. 2. He doesn't beat her because she is pregnant. The first interpretation requires the use of a fall-rise on pregnant. The present tense may again have present time or post-present time reference. Example (131b) only allows the


interpretation that she will not get beaten up as she will be pregnant. It is more difficult to find a post-present sector although-clause with an adjunct reading. It seems that this clause is always of the disjunct type:

(132) She will sell the house, although he will tell her not to.

The above examples corroborate the view that the FPS has to be used in the disjunct version of any kind of clause. The FPS or the PPS can be used in the adjunct variant, although possible misinterpretations of the PPS forms may necessitate the use of the FPS. The syntactic tightness between the SUPC and the adjunct subclause lies at the origin of the "correct" (i.e. post-present sector) interpretation of the PPS forms; however, it does not safeguard a post-present interpretation in the sense that not any form which can be interpreted as a PPS form gets a post-present time interpretation in an adjunct clause.

Chapter 6 Difference in contextual effects between mutually substitutable verb forms

The commonly accepted belief that two mutually substitutable forms are pragmatically never completely synonymous implies that choosing, where possible, between a temporally subordinated verb form and a shift of domain verb form will result in (possibly vefy subtle) differences in meaning being expressed. This chapter· lists certain factors which may induce a speaker to use a shift, of domain tense/temporally subordinated verb form; it deals with the difference in contextual effect between mutually substitutable verb forms (i.e. verb forms conveying the same temporal information).

The following differences in shades of meaning between two alternatives can be observed. Some tendencies are clear from more than one sector:

6.1. General procedure vs. implementation of general procedure

6.1.1. Past sector

In the case of general statements, replacing the past tense· by the past perfect has the effect of particularizing the situation to a specific event:

(1) a. The rivers gained the majority of their supply from snow which fell and melted in the spring, (adapted from SEU)145

b. The rivers gained the majority of their supply from snow which "had fallen and melted" in the spring.

When the past tense is used in the RC, the clause is interpreted as referring to a habitual situation: it explains in general where the rivers get their water from. The reason why a habitual reading is assigned to the past tense RC is that the speaker would presumably have used a progressive form if he had wanted to indicate a particular occasion. The past perfect, on the contrary, refers to one instance

General procedure vs. implementation 227

rather than the habitual pattern. Put differently, when the past perfect is used, we could add an adverbial such as in 1982·, if asked to add a time adverb to the past tense sentence, we would use an adverbial not limited to a single occasion (e.g. in those times). The same observa-tions apply to the following example:

(2) a. Robinson Crusoe ate the food he found in the woods.

b. Robinson Crusoe ate the food he "hadfound" in the woods.

A suitable adverbial for the past perfect sentence is on that particular occasion, whereas the past tense goes with usually, when he lived on the island.

6.1.2. Pre-present sector

The past perfect may perform a similar function in the pre-present sector. The present perfect has the effect of representing a situation as a general procedure, whereas the past perfect refers to an instanti-ation of the general procedure:

(3) a. Applications have been accepted from men and women in any profession or field of study who (...) have completed a minimum of four years' secondary schooling as well as the major part of their further education, (adapted from SEU)

b. Applications have been accepted from men and women in any profession or field of study who (...) "had completed" a minimum of four years' secondary schooling as well as the major part of their further education.

The RC has conditional connotations. The present perfect will be used in the RC to explain the general procedure, i.e. the procedure used in former times as well as now. The past perfect explains how they went about things in the past. When the past perfect is used, one tends to think of the particular persons, the individuals who met the requirement. The present perfect does not represent the situation as being embedded in reality to the same extent, by which I mean that it could be used even if no one had been accepted (because no one met

228 Difference in contextual effect between mutually substitutable forms

the requirements). The past perfect would not be acceptable in that case.

6.2. Prediction vs. fact

An aspect inherent in the meaning of the future tenses (and therefore of the FPS) is that they always convey an element of expectation, as it is impossible to predict with full certainty that the future tense situ-ations will be realized. This element of prediction is no longer ex-plicitly present when the PPS is used. Although the present tense (like the other tenses from the PPS) is temporally subordinated to a future tense clause, and in a way inherits the connotation of expecta-tion and prediction, this aspect of meaning is no longer explicitly conveyed by the verb form as such. This difference sometimes gives rise to (slightly) different interpretations between two alternative forms. Put differently, we might say that when the FPS is used the speaker makes two predictions; when the PPS is used he makes only one.146

It is not possible to make hard and fast judgments as regards the (un)acceptability of mutually substitutable PPS and FPS forms, be-cause it is largely a matter of emphasis whether one form rather than the other is used. All that can be done is point to the different shades of meaning without making any claims about whether a given form must be rejected as unacceptable:

Prediction vs. fact 229

(4) a. As an experiment the Government propose to make arrangements under which money will be advanced to non-profit-making housing associations which are prepared to build houses to let at economic rents. (LOB)

b. As an experiment the Government propose to make arrangements under which money will be advanced to non-profit-making housing associations which "will be prepared" to build houses to let at economic rents.

When the future tense is used, the speaker predicts that some housing associations will be prepared to cooperate in the government's pro-ject. When the present tense is used, the element of prediction is weaker. In other words, repetition of a FPS form reminds the speaker of the fact that what is said is a speculation, a prediction or an ex-pectation. When the future tense is followed by a PPS form, the speaker first makes a prediction and it is assumed (for the sake of the argument) that it is true. When the domain is expanded, one seems to "forget" that it is "merely" a speculation that is given in the RC. Ex-ample (5) illustrates the same tendency:

(5) a. If only order is known, medians will yield conclu-sions which are correct so far as order is concerned. (LOB)

b. If only order is known, medians will yield conclu-sions which "will be" correct so far as order is concerned.

Paraphrasing the sentences with the present tense and future tense underlines the difference in meaning between the two forms:

(a) present tense: the speaker identifies the kind of conclusion he has in mind: one that is correct as far as order is concerned.

(b) future tense: the speaker predicts: the conclusion will be correct as far as order is concerned.

There is a similar difference in effects between the present and the future tense in (6):


But though I believe that gods and God in any meaningful non-Pickwickian sense are destined to disappear, the stuff of divinity out of which they have grown and developed remains and will provide much of the raw material from which any new religions will be fashioned. (LOB) But though I believe that gods and God in any meaningful non-Pickwickian sense are destined to disappear, the stuff of divinity out of which they have grown and developed remains and will provide much of the raw material from which any new religions "are fashioned".

When the present tense is used, one gets the impression that the speaker takes it for granted that new religions will arise (6b). When the future tense is used, one interprets the sentence as meaning that, in the speaker's opinion, new religions may come into existence.

When the RC refers to a situation which is W-anterior to the SUPC situation, it is possible to exploit the difference in meaning under discussion to an even greater extent: in examples of this type, there is reference to a situation which precedes a future time of orientation and from that point of view might be said to acquire a certain degree of factuality (cf. chapter 5, section 5.3.2.4.1.). If the speaker decides to use the FPS, it can only be because he wants to stress the fact that he is making a prediction:

(7) a. The widow of a man who has paid graduated con-tributions will get a graduate addition to her flat-rate retirement pension equal to one-half the graduated part of the pension which her husband had earned or was drawing when he died. This will be on top of any graduated pension which she herself has earned. (LOB)

b. The widow of a man who has paid graduated con-tributions will get a graduate addition to her flat-rate retirement pension equal to one-half the graduated part of the pension which her husband had earned or was drawing when he died. This will be on top of any graduated pension which she herself will have earned.

(6) a.

b.

Explicit location in time 231

The present perfect tense cannot be misinterpreted as referring to a situation which lies before t0 (chapter 5, section 5.3.2.4.2.) because the RC situation belongs to the intensional post-present domain created in the preceding sentence. In other words, it is not necessary to use the FPS to ensure a correct temporal interpretation. The future perfect conveys a strong element of personal prediction of future factuality. It is somewhat unusual to use this form because it implies that the speaker is certain that the person in question will have earned a graduated pension. It follows from the conditional context of the sentence that this is not necessarily the case.

6.3. Explicit location in time - explicit indication of temporal relation

6.3.1. Past sector

RRCs give restricting (non-)identifying information; NRRCs give additional, relevant information. It is up to the speaker to decide how he will provide the restricting and relevant information. If he believes that the fact that the referent was involved in a certain past sector situation is sufficient to restrict or characterize the referent, he will use the past tense. In this case, the speaker gives characterizing information without bothering to explicitly indicate the temporal relation between the SC situation and the SUPC situation. If, on the other hand, he believes that it is the time when the referent of the NP was involved in the situation which should perform the function of giving restricting/relevant information, he will use the past perfect (when the RC situation is W-anterior to the SUPC situation):

(8) a. Mr. Frederick Manning, the man who piloted Sir Gerald Nabarro, Tory M.P. for Kidderminster, through four successful election campaigns, died at the wheel of his car yesterday. (SEU)

b. Mr. Frederick Manning, the man who ?? "had piloted" Sir Gerald Nabarro, Tory M.P. for Kid-derminster, through four successful election campaigns, died at the wheel of his car yesterday.

Although the election campaigns precede Frederick Manning's death, the past perfect is only marginally possible in this sentence. The speaker has to represent the RC situation as a past fact, as Mr.


Manning is dead. However, there is no need to represent the RC situation as a "past in the past": it is normal when talking about someone who is dead to assume that the things he did preceded his death. The information needed is a characterization of Frederick Manning; an explicit location in time of the characterizing information is not required.147 The same tendency can be observed in (9):

(9) a. The puzzle was solved when Fred Sanger, who won two Nobel awards, (...), applied his technique to it. (SEU)

b. The puzzle was solved when Fred Sanger, who "had won" two Nobel awards, (...), applied his technique to it.

The relevant information about Fred Sanger is simply that he won two Nobel awards; the use of the past perfect does not make the statement more relevant. The past perfect is therefore not used in this context, although it is an acceptable form, i.e. it does not change the temporal relation between the RC situation and the SUPC situation. The following is an example from Bouscaren et al. (1982):

(10) a. In the very beginning there was that accident that occurred about a week after I was born (...) My mother's knees and elbow were badly scraped. I appeared to be un-injured. The blanket I "had been wrapped" in had protected my face and only a very little blood, as if from a small scratch, had come from my nose. (In the beginning, Chaim Potok) (Bouscaren et al. 1982: 84-85)

b. In the very beginning there was that accident that occurred about a week after I was born (...) My mother's knees and elbow were badly scraped. I appeared to be un-injured. The blanket I was wrapped in had protected my face and only a very little blood, as if from a small scratch, had come from my nose.

The authors comment "Avec la forme HAD-EN, on fait appel au processus stabilise en etat resultant. L'aspect "resultant" d'un processus semble l'emporter sur l'etat exprime par le PAST. Le changement qui serait introduit par un past le montre bien"


(Bouscaren et al. 1982: 84). [When the past perfect is used, there is reference to the process as well as to the resulting state. When the past tense is used, the focus is on the state. The change that would be brought about by a past tense clearly demonstrates this point.] The blanket I was wrapped in indicates which blanket they are talking about. Because the information that the speaker was in it (as a result of the wrapping) is relevant in this case (i.e. he did not get badly hurt), the past perfect is used in preference (cf. section 6.4.). Was wrapped refers to the action of wrapping and is at first interpreted by the hearer as referring to the next action in the series, i.e. the situation following I appeared to be un-injured, an interpretation which the hearer has to alter as the speaker continues his utterance. The use of the passive voice may also contribute to the impression that the RC is given to characterize the referent without bothering to locate the situation exactly in time:

He achieved the task for which he had been nominated - getting the budget passed - and therefore he can claim that he has accomplished his mission, (adapted from SEU) He achieved the task for which he "was nominated" - getting the budget passed - and therefore he can claim that he has accomplished his mission. He achieved the task for which they had nominated him - getting the budget passed - and therefore he can claim that he has accomplished his mission. He achieved the task for which they "nominated" him - getting the budget passed - and therefore he can claim that he has accomplished his mission.

The use of the past tense will be even more highly preferred in (1 lb) than in (1 Id): choosing the passive voice means that it is the situation as such that is important, and not the information by whom it was brought about. A corollary of this may be that an explicit location in time is not urgently required either. As appears from the passive RC in (10), in which the resultative implicatures associated with the past perfect contribute to the preference for the use of that tense, this is a tendency rather than a hard and fast rule.

(11) a.

b.

c.

234 Difference in contextual effect between mutually sub statutable forms

6.3.2. Pre-present sector

In the pre-present sector as well, the fact that RRCs give restricting information about the antecedent may result in a relatively free choice between the past, past perfect or present perfect tense. If it is not the time when the referent NP was involved in a particular situation which provides restricting information, but the fact that the referent NP was involved in a certain situation as such, the speaker will not be too worried about the tense used, as long as it does not hamper a correct interpretation:

(12) a. They have commended the Gospel to young people who have been brought up in irreligious homes and regard belief in the supernatural as intellectually disreputable, (adapted from SEU)

b. They have commended the Gospel to young people who "had been brought up" in irreligious homes and regard belief in the supernatural as intellectually disreputable.

The primary function of the RC is to give more information about the antecedent; the exact location in time of that information is only of secondary importance. No matter whether the past perfect or the present perfect is used in (12), there is reference to a particular kind of people, i.e. those brought up in irreligious homes. Example (13) is similar:

(13) a. The Govt, are prone to spring decisions on dele-gations: they have announced the 70 m.p.h. limit to a delegation of chief constables who had come to the Home Office to discuss the breathalyser, (adapted from SEU)

b. The Govt, are prone to spring decisions on dele-gations: they have announced the 70 m.p.h. limit to a delegation of chief constables who "came" to the Home Office to discuss the breathalyser.

c. The Govt, are prone to spring decisions on dele-gations: they have announced the 70 m.p.h. limit to a delegation of chief constables who "have come" to the Home Office to discuss the breathalyser.


In (13), a particular group of people is indicated irrespective of the tense used in the RC, i.e. those who were informed about the speed limit and who discussed the breathalyser. If the chief constables are still at the Home Office, the present perfect is likely to be used; if not, the past perfect will probably be preferred. The past tense leaves it vague whether or not they are still present. The following examples illustrate the same tendency:

(14) a. He has recovered some of the estate duty which has been paid, (adapted from SEU)

b. He has recovered some of the estate duty which "had been paid".

(15) a. They have used a long stem which has borne a perfect flower,

b. They have used a long stem which "had borne" a perfect flower.

In (12) to (15), t0 functions as binding TO for the present perfect tense in the RRC, while the past perfect takes a past situation (the SUPC situation) as TO. This does not hamper the identification of the referent, as it is clear from the context that the RC situation is W-anterior to the SUPC situation.

In pre-present sector NRRCs there is less freedom (cf. chapter 4, section 4.4.3.). Their function is to give relevant information: if the SUPC situation has clear links with the present, the NRRC will probably have to gear the information it gives to this fact. From that point of view, temporal information is more important than in some RRCs. Example (16) is a case in point:

(16) a. They have attacked the trade unions, whose restrictive practices have prevented price reductions, (adapted from SEU)

b. They have attacked the trade unions, whose restrictive practices had prevented price reductions.

c. They have attacked the trade unions, whose restrictive practices prevented price reductions.

The past perfect implicates that the effects of the restrictive practices are no longer felt at t0, the present perfect implicates that they still are. Depending on the context, one of the two forms will be ruled out. The past tense suggests that the restrictive practices held at the moment when the trade unions were attacked (simultaneity with the


SUPC situation) (cf. chapter 4, section 4.2.2.1.); it does not teli us whether the restrictive practices and their effects still hold or not.

6.3.3. Post-present sector

Examples of the type under discussion also occur in the post-present sector. The same piece of information may be represented in two different ways: in the one case it will be explicitly located in the post-present sector (FPS), in the other the characterizing piece of information will also be located in the post-present sector, though not explicitly by means of a future tense (PPS):

(17) a. That will be a generation which grows up in an urban environment and amid many difficulties but does not experience, whilst growing up, the assault of the mass Press as it is known today, of the wireless and television, of the ubiquitous cheap cinemas, and so on. (adapted from SEU)

b. That will be a generation which "will grow up" up in an urban environment and amid many difficulties but "will not experience", whilst growing up, the assault of the mass Press as it is known today, of the wireless and television, of the ubiquitous cheap cinemas, and so on.

The restricting information (i.e. the generation will grow up in an urban environment, it will experience difficulties, ...) is explicitly located in the post-present sector when the future tense is used. The latter also carries the implicature that the speaker is talking about a generation which does not yet exist; this implicature is obviously not as explicit when the present tense is used. Under the PPS, the antecedent is restricted by giving characterizing information not explicitly located in the post-present sector. The present tense RC no doubt inherits the implicature from the SUPC. However, this still means that it is only indirectly implicated that the situation does not yet hold. It will be clear that it is not always possible to make a neat distinction between the examples classified under the heading "prediction vs fact" (section 6.2.) and the examples given in this sec-tion. As regards the example just given, it may also be argued that when using the future tense the speaker explicitly indicates that he is making a prediction whereas the present tense suggests that there is a


greater degree of certainty that the generation in question will grow up in an urban environment. The same tendency can be observed in (18):

(18) a. Some pockets will have a flap to prevent the batten from coming out, but many will have a short length of thin line sewn in which is passed through two holes in the end of the batten, and the ends reef-knotted together, (adapted from SEU)

b. Some pockets will have a flap to prevent the batten from coming out, but many will have a short length of thin line sewn in which "will be passed" through two holes in the end of the batten, and the ends reef-knotted together.

When the present tense is used in the RC, it seems that a certain type of pocket is defined in general terms. The future tense in the SUPC makes it clear that there is reference to a pocket that still has to be sewn; it indicates that the RC has future time reference. When the FPS system is used in the RC, the sentence explicitly tells us two things about the future: (1) the pockets will have a short length of thin line sewn in (SUPC); (2) the line which is sewn in will be passed through two holes in the end of the batten (RC).

It is especially (a) when there is reference to a class of peo-ple/things or a particular type or (b) when the NP is attributive (Donnellan 1966: 285) that the FPS is less likely to be used. Although, again, it is not impossible, it seems more natural to define a certain class of people by giving descriptive information not explicitly relocated in the post-present sector by means of the future tense. The future tense in the SUPC indicates that there is reference to a future class of this particular type of people. The same line of reasoning applies to attributive RRCs. In the following example, a type of person is defined:

(19) a. He will despise anyone who earns money, (adapted from SEU)

b. He will despise anyone who ? will earn money.

Although people who earn money exist at t0, the RC will be interpreted as referring to the post-present sector because the SUPC indicates that future time reference is intended. A prediction is made about this particular class of people, and therefore there is reference


to a future group of people/things of the type defined in the RC. The RC defines a certain class of people and does so by making use of the unmarked present tense. The RC also has conditional connotations, which explains the questionable status of the future tense in the RC. The following examples are similar:

(20) So the arrival of any particular shoal fish at a certain spawning ground may be accidental. Yet there will always be some thai find (will find) it if enough are travelling, just as the robin who gets back exactly may be only one among thousands who do not. (LOB)

(21) "How can you say such a thing? There will be thousands who will thrill (thrill) to the loveliness of Alastor." (BR)

(22) And it is also a fact of life that there will always be youngish half-educated people around who will be dazzled, (are dazzled) by the glitter of what looks like a literary movement. (BR)

(23) Miss Bantin will be there in person ready to give her friendly advice to all who care (will care) to take advantage of her visit to the Capitol. (LOB)

(24) Because of this, it would appear inevitable that an increasing percentage of strategic missiles will seek self-protection in mobility - at least until missile defenses are perfected which have (will have) an exceedingly high kill probability. (BR)

In generic NPs it is sometimes very difficult to use the FPS. Generic NPs are by definition omnitemporal. It is therefore self-evident that a "neutral tense" should be used (cf. chapter 5, section 5.3.2.5.2., section 5.3.2.6.). The following examples, in which the RC situation is W-anterior to the SUPC situation, illustrate this point very clearly:

(25) a.

b.

(26) a.

b.

He will feel like a soldier who has not been to France, (adapted from SEU) He will feel like a soldier who ? will not have been to France. He will create the impression of a man who has been run down by an articulated lorry, (adapted from SEU) He will create the impression of a man who ? will have been run down by an articulated lorry.

Resultativeness 239

6.4. Resultativeness

Because of its semantic structure, the past perfect is very suitable for expressing resultative implicatures. If one looks from a certain situation at a situation that held before, it must be that one considers the two situations as being related.

The semantics of a sentence (e.g. when the sentence is telic) as such may imply resultativeness. However, even in sentences of this type, the resultative implicatures will be more explicit when the past perfect is used than when the past tense is used. From this it follows that in such examples the past perfect is very often felt to refer to a W-anterior event and at the same time indicate a state holding at T02, i.e. the one resulting from the event.148 Hüllen (1989) argues in connection with noun clauses that the past perfect, unlike the past tense, "adds a particular meaning of final result to the subordinate part of the sentence" (Hüllen 1989: 621). He seems to be thinking of the same effect as we have in mind. However, not all his examples are well-chosen:

(27) a. She complained that nobody helped her. (Hüllen 1989:621)

b. She complained that nobody had helped her. (ibid.)

He argues that in (27a) "the clause which is in indirect speech reports something put of the past leaving open a possible change in the future. In [27b], however, the same clause reports something out of the past which has come to a final result and cannot be changed any more" (Hüllen 1989: 621). He claims that the same difference can be observed in the examples in (28):

(28) a. You said he was drunk. (Hüllen 1989: 621)


b. You said he had been drunk once, (ibid.)

The most important difference between the (a) and the (b) alternative in (27) and (28) is that the past perfect is. understood as referring to a situation which is anterior to the SUPC situation, whereas the unmarked interpretation of the past tense is that the RC situation is simultaneous with the SUPC situation. A past time unbounded reversible situation (as in the (a) sentences) is indeed still subject to change, whereas a past time bounded reversible situation (as in the (b) sentences) cannot be changed because it is completely over. The idea of "final result" (Hüllen 1989: 621) Hüllen refers to has more to do with reversibility and boundedness than with the indication of result.

It is clear from the following examples that the past perfect indicates possible results in a more explicit way than the past tense:

(29) a. Mr Wall, who was smuggled into the debating chamber of the students' union, was greeted with shouts of 'Sieg Heil' and standing students giving the Nazi salute. (SEU)

b. Mr Wall, who "had been smuggled" into the debating chamber of the students' union, was greeted with shouts of 'Sieg Heil' and standing students giving the Nazi salute.

The past tense RC seems to provide an answer to the question: "How did Mr Wall get into the debating chamber?"; there is ref-erence to two actions following each other quite quickly, i.e. smuggling and greeting (cf. chapter 3, section 3.2.2.2.). The past perfect seems to be more an answer to the question: "What factors lie at the origin of Mr Wall's presence (= state) in the debating chamber?" The past perfect refers to an event and at the same time stresses the resultant state: Mr. Wall was, against all expectations, present as a result of the students' smuggling him into the room. The past perfect gives more the impression that this was difficult to achieve. The following example is similar:

(30) a. In particular, households with insufficient irrigated land or with no land at all had little option but to work for the large farmers, many of whom were able to reinvest their profits. They bought more land

Resultativeness 241

from poorer people who had fallen into debt, (adapted from SEU)

b. In particular, households with insufficient irrigated land or with no land at all had little option but to work for the large farmers, many of whom were able to reinvest their profits. They bought more land from poorer people who "fell" into debt.

The past perfect clearly has resultative implicatures; it is precisely the result of falling into debt that is important: people are in need of money and consequently sell their land. The idea of result is also present when the past tense is used, but less strongly so. In this case, one gets the impression that very little time elapses between falling into debt and selling land (cf. section 6.5.). There is a similar difference in contextual effects between the past perfect and the past tense in (31):

(31) a. George McLean took a free kick 25 yards out, Reaney headed the ball away, but Bobby Wilson, who had appeared from nowhere, so to say, forced it past Harvey. (SEU)

b. George McLean took a free kick 25 yards out, Reaney headed the ball away, but Bobby Wilson, who "appeared" from nowhere, so to say, forced it past Harvey.

The past tense and past perfect are both acceptable in this example. When the past perfect is used, two situations are referred to: the fact that he appeared (event) and the resulting fact that he is there (state). The past tense also conveys the idea of result (Bobby Wilson is there), but less explicitly so; moreover, it suggests that not much time elapses between his appearance and scoring the goal (cf. section 6.5.).

The following non-RC example given in Bublik (1983) illustrates very well the function of the past perfect under discussion:

(32) Oppressed with heavy anxiety he ran his large, firm workman's hand between his collar and neck. Suddenly, he saw himself going downstairs again. Then he had already knocked and opened the door. (Bublik 1983: 33)


"Karl will Anna sehen, aber er hat Angst und möchte umkehren - da klopft er schnell und öffnet die Tür: Es gibt keine Rückkehr mehr" [Karl would like to see Anna, but he is afraid and wants to leave - he quickly knocks and opens the door: there is no turning back any more.] (Bublik 1983: 33). The past perfect is the tense par excellence to convey the idea of no return; as a result of his knocking on the door and entering, he finds himself in a state he cannot escape from. The use of the past tense in this sentence would have been interpreted as a mere sequence of events. The non-RC example in (33) also illustrates the same principle:

(33) She told me his name after he had left (left). (Bouscaren et al. 1982: 77)

(34) He died after he had been (? was) ill for a long time, (ibid.)

Bouscaren et al. explain the difference in acceptability between the past tense forms in terms of the tighter semantic links between illness and death in the sentence in (34) than between leaving and telling a name in (33). Salkie (1989) subscribes to this view: "Using the past tense instead suggests that the speaker is not asserting any relevance relation between the two events, but in that case, why mention the illness at all, let alone in a subordinate clause straight after mentioning the death" (Salkie 1989: 5). Whereas after he left can be used as a mere temporal anchor for the main clause, "it would be odd to just use the illness as a temporal anchor given the close relation between the illness and dying" (Salkie 1989: 5). As I see it, it is the resultative implicatures conveyed by had left in (33) (i.e. the room was empty, they were safe,...) which lead to a preference for that verb form. The past tense suggests a quick succession of two events. In sentence (34), on the other hand, the past tense cannot be used because for a long time, i.e. from the moment he became ill until his death, requires the use of a tense which establishes a continuative reading.

6.5. Gradual development vs. quick succession

The idea of resultativeness especially associated with the past perfect sometimes goes hand in hand with an impression that the past perfect situation takes up a considerably long period of time. The effect of the past perfect may indeed be to stress the length of time taken up by the RC situation. The use of the past tense in examples of this

Gradual development vs. quick succession 243

type again suggests that there is a quick succession of the RC and the SUPC events:

(35) a. They implemented the reforms on which they had reached general agreement, (adapted from SEU)

b. They implemented the reforms on which they "reached" general agreement.

In example (35) the past perfect suggests that a certain amount of time has elapsed between the agreement on the reforms and their im-plementation. The past tense, on the other hand, suggests that as soon as an agreement is reached on the reforms, they are implemented. Another difference is that the past tense suggests that a series of de-cisions are taken, every one of which is implemented at once (= dis-tributive reading). The past perfect refers to a number of decisions taken at a particular time and then collectively implemented. Already can only be used in the past perfect RC, which again illustrates that the past tense does not allow the collective reading. When the past tense is used, no distinction is made between all the decisions and all the implementations:

(35) c. They implemented the reforms on which they had already reached general agreement,

d. They implemented the reforms on which they ?? already reached general agreement.

The same difference in meaning exists between the past tense (quick succession) and the past perfect in the following examples:

(36) a. In particular, households with insufficient irrigated land or with no land at all had little option but to work for the large farmers, many of whom were able to reinvest their profits. They bought more land from poorer people who had fallen into debt, (adapted from SEU) (cf. section 6.4.)

b. In particular, households with insufficient irrigated land or with no land at all had little option but to work for the large farmers, many of whom were able to reinvest their profits. They bought more land from poorer people who/e// into debt. (SEU)

(37) a. George McLean took a free kick 25 yards out, Reaney headed the ball away, but Bobby Wilson,


who had appeared from nowhere, so to say, forced it past Harvey. (SEU) (cf. section 6.4.)

b. George McLean took a free kick 25 yards out, Reaney headed the ball away, but Bobby Wilson, who appeared from nowhere, so to say, forced it past Harvey.

6.6. Recent vs. distant situation

a. As pointed out before, the difference in effect between the use of the present perfect and the past perfect tense to refer to a W-anterior situation in the pre-present sector can be captured as follows: when the W-anterior situation is referred to by means the present perfect, it seems to be closely related to the binding TO of the SUPC clause (the "now" indicated by t0); if the W-anterior situation is referred to by means of a past perfect, it seems to be more closely related to the "then" indicated by the SUPC situation. It has been explained that the past perfect represents a situation as holding in the past sector, before the SUPC situation, whereas the present perfect focuses on the links the RC situation has with t0. If the context is compatible with both readings, it is a matter of choice on the part of the speaker whether to use one form rather than the other. Depending on how important the link between the RC situation and t0/SUPC situation is considered to be, the present perfect/past perfect respectively will be used:

(38) a. Perhaps this is because most of the working-class people they have known have been of an unusual and self-selected kind, and in special circumstances, young men and women at Summer Schools and the like, exceptional individuals whom the chance of birth has deprived of their intellectual inheritance, and who have made remarkable efforts to gain it. (SEU)

b. Perhaps this is because most of the working-class people they have known have been of an unusual and self-selected kind, and in special circumstances, young men and women at Summer Schools and the like, exceptional individuals whom the chance of birth "had deprived" of their intellectual inheritance, and who had made remarkable efforts to gain it.

Recent VJ. distant situation 245

The sentence with the present perfect seems to provide an answer to the question "What are the people they knew like?", whereas the past perfect sentence answers the question "What were the people they knew like?" The principle illustrated here is the same as that described in section 4.2.1.1. and 4.2.3.1. in chapter 4, the difference being that in the examples given there one option is ruled out altogether. In the present examples, however, both the present perfect and past perfect are acceptable verb forms; no matter whether the RC situation is represented as having links with "now" or with "then", it makes sense to represent the SUPC situation as having links with the present, i.e. to use an indefinite perfect. The following examples also bear testimony to the principle just formulated:

(39) a. She has talked about the arrival of women undergraduates at Trinity College and about women in publishing, of which she has had long personal experience, (adapted from SEU)

b. She has talked about the arrival of women undergraduates at Trinity College and about women in publishing, of which she "had. had" long personal experience.

The past perfect suggests the following: at a past moment in time (SUPC situation) she had easy access to the knowledge about female undergraduates as a result of her previous experience. The use of has talked will probably lead the speaker to use the present perfect in the RC as well, which indicates that her experience affects the present so that she can talk about it now. The difference may be represented schematically in the following way:


had had has talked

b. The fact that the present perfect establishes links with "now" and the past perfect with "then" may lie at the origin of the following im-pression: if one and the same situation can be referred to by means of the present perfect as well as the past perfect, the latter tense repre-sents the situation as considerably removed in time whereas the for-mer represents the situation as fairly recent. Although use of the past perfect as such does not necessarily imply that the situation referred to lies at a great distance from t0, choosing to use the past perfect rather than the past tense or the present perfect to refer to a situation lying before the SUPC situation may have been partially determined by the wish to represent the situation as being considerably removed from t0. This can be seen, for instance, in the following example, in which just, indicating recent past with respect to t0, requires the use of a present perfect tense:

(40) a. We have also enclosed a copy of an appointment of the Receiver which we have just received, (adapted from SEU)

b. We have also enclosed a copy of an appointment of the Receiver which we * had just received.

If the letter indicating who will function as Receiver has arrived re-cently, the present perfect will have to be used.

It should be stressed that a past perfect situation is not necessarily further removed from t0 than a present perfect situation in terms of absolute location in time, i.e. it is not the case that the past perfect has to be used as soon as the situation lies at a certain distance from t0 (cf. Binnick 1991: 55). Even so, when a situation which lies before a pre-present TO is referred to by means of a past perfect rather than the present perfect, it will be experienced as being further removed from the present. The fact that the events denoted by the present per-fect may be relatively recent (cf. e.g. McCoard 1978: 35, Sorensen 1964: 80-81) follows from the general semantics of the perfect. If one decides to look at a past time situation from a present point of view, it must be that one considers there to be a link of some sort

Recent vi. distant situation 247

between the event and now.151 The further removed the past situation is, the more difficult it may be to establish such a link:

There is a general tendency of self-centredness in human speech, whereby, unless otherwise specified, we understand a word or phrase to refer to something close at hand rather than distant. (...) If we recognize that the indefinite past meaning always involves a period leading up to the present, it is easy to see how this period can become reduced, by subjective assumption, from 'always' to 'within the last few days' or even 'within the last few minutes'. (Leech 1971: 38)

That is why the answer to "Have you seen my slippers?" could hardly be "Yes, six months ago" (McCawley 1973: 267).152·153

It may be concluded that although recency is not a necessary con-dition for the use of a present perfect,154 it may be a sufficient condi-tion155 provided there is no adverb referring to a definite past time in the same intonation group. The use of the past perfect implies that the speaker does not consider the RC situation to have (m)any links with the present, which may well result in an impression of the situ-ation being further removed from t0. This effect may have to do with the structure of the tenses: the structure of the past perfect is such that there is always a TO, which lies between t0 and TOsV No matter how close TOsit, T0 2 ana t0 are, they will never allow the same de-gree of closeness between t0 and TOsit as a present perfect situation, because there is no intervening TO, in the latter tense.

The "remoteness" of the past perfect situation may be accompanied by an impression of a higher degree of factuality:

(41) a. They haven't shown the pitfalls into which intelli-gent men have fallen, (adapted from SEU)

b. They haven't shown the pitfalls into which intelli-gent men "had fallen".

The present perfect, unlike the past perfect, suggests temporal close-ness. The present perfect statement is more tentative because have is interpreted as it has happened: it has happened that intelligent men -in spite of what might be expected- have fallen into pitfalls; however, these accidents have not happened frequently. There is no such touch of tentativeness or suggestion of infrequency when the past perfect is used. The use of the past perfect in the RC makes one think more of


the individuals who actually fell into the pitfalls and the numerous occasions on which they did (cf. section 6.1.).

In any case, it is no wonder that Dahl (1985) gives the example of the development of the present perfect and past perfect into tenses indicating a recent and remote past respectively to illustrate his ob-servation that "if some condition happens to be fulfilled frequently when a certain category is used, a stronger association may develop between the condition and the category in such a way that the condi-tion comes to be understood as an integral part of the meaning of the category" (Dahl 1985: 11).

The above list of differences in contextual effects is probably not complete. Even so, it presents the most important tendencies to be observed in the corpus at my disposal.156

Chapter 7 The principle of unmarked temporal interpretation

7.1. Introduction An idea which has been stressed in chapter 3 is that a speaker wishing to conform to the principles of communication will use a shift of domain form only if it does not hamper the correct inter-pretation of the chronological order between the SUPC situation and the RC situation. That is, a shift of domain is possible in the follow-ing cases:

a. Sometimes the chronological order is clear from the use of one or more adverbials:

(1) King Constantine of Greece, who arrived in London on Monday, did not attend a concert in the Royal Albert Hall last night because of the Panagoulis scene, said a friend of the King. (SEU)

b. In other cases, our knowledge of the world tells us whether the RC situation comes before or after the SUPC situation (cf. e.g. Stegu 1983: 322-323):

(2) a. The blood tests which we did for the bruising which mother reported, but which I actually never saw, were quite normal. (SEU)

b. Slocum chose to follow the route that Magellan took. (Diver 1963: 167)

c. A luxury jet yesterday brought Imelda Marcos to New York, where she checked into the Waldorf Astoria Hotel. (Davidsen-Nielsen 1990: 124)

c. Even when neither of these factors helps to establish the correct chronology, the hearer will often derive the order from the contextual knowledge available to him:

(3) The man who went on a holiday to Spain invented a mosquito deterrent.

250 The principle of unmarked temporal interpretation

Out of context, it is not clear whether the invention happens after or before the trip. The use of the definite NP indicates that the speaker assumes that the hearer can identify the referent; the hearer could therefore also be in a position to establish the chronological order be-tween the RC and the SUPC situation. In any case, the speaker will only use a sentence of this type if he is sure that the hearer can interpret it correctly (cf. Grice's cooperation principle and maxims). It is possible that the hearer does not have the contextual information he needs to recover the chronological order between the situations, e.g. if the hearer has not read anything about the invention of the deterrent, it will not be clear to him which situation comes first. One might wonder if any regularities can be observed in the way sequences are temporally interpreted in situations like these.157 Sev-eral linguists have pointed out that some sort of underlying "guidelines" exist.

In what follows, these guidelines, which Declerck calls the prin-ciple of unmarked temporal interpretation (1991a: 119, 138-139) (henceforth PUTI), will be examined in detail. Section 7.2. sum-marizes the PUTI as it is formulated by Declerck. Section 7.3. gives a survey of how this issue has been dealt with by other linguists. The RC examples with which Declerck illustrates the PUTI for embedded clauses are presented and commented on in section 7.4. In section 7.5., a more detailed analysis is given of the PUTI in RCs. The claims made are based on the results of an elicitation test, which is described in the appendix. Special attention is paid to the possible influence of the following factors on the temporal interpretation: (a) RRC vs. NRRC, (b) the order in which clauses are uttered (RC < SUPC or SUPC < RC), (c) the sector in which the situations are located (past sector, pre-present sector or post-present sector).

7.2. The PUTI (Declerck 1991a): theory

When neither adverbials nor pragmatic knowledge/contextual information indicate the temporal relations between the situations referred to in a sequence of main clauses that establish their own domain, the following conversational implicature is said to regulate the temporal interpretation:

The PUTI (Declerck 1991a): theory 251

Principle of unmarked temporal interpretation (for unem-bedded clauses): In a sequence of unembedded clauses where each clause establishes its own domain (within the same sector) and where there is no adverbial, contextual or pragmatic indication of temporal order, the unmarked temporal interpretation is as follows: (a) when the situations are represented as bounded, they are interpreted as following each other in the order in which they are reported; (b) when the situations are represented as unbounded, they are interpreted as simultaneous with each other; (c) when one situation is represented as bounded and the other as unbounded, the bounded situation is interpreted as temporally included in the unbounded one. (Declerck 1991a: 119)

The bounded/unbounded distinction has been dealt with in detail in chapter 1; a situation is bounded if it represents a situation as having reached a temporal boundary. If it does not, it is unbounded. The following examples are given to illustrate (a), (b) and (c) respec-tively:

(4) John went to the door and knocked three times. The door opened slowly. (Declerck 1991a: 124)

(5) John stood by the window. Mary was in the kitchen. (Declerck 1991a: 126)

(6) John went out of the room. He was feeling very tired. (Declerck 1991a: 128)

A slightly different version of the above principle is said to govern the interpretation of embedded clauses:

Principle of unmarked temporal interpretation (for embedded clauses) (a) If both the head clause and the subclause shifting the domain are bounded, the two situations tend to be interpreted as holding in the order in which they are reported only if the subclause is one of the clauses that carry the action forward, (i.e. that have a foregrounding function) (Declerck 1991a: 138)


(b) When the two clauses are unbounded, the unmarked interpretation is that of simultaneity. (Declerck 1991a: 139) (c) When one clause is unbounded while the other is bounded, the unmarked interpretation is that the bounded situation is included in the unbounded one. This is the case irrespective of whether the unbounded clause is the head clause or the subclause and irrespective of the order of the clauses. (Declerck 1991a: 139)

Before discussing the RC examples with which the PUTI for SCs is illustrated in Declerck (1991a), it is still necessary to clarify what is meant by "clauses that carry the action forward" or "clauses that have a foregrounding function". This will give us the opportunity to review the literature on PUTI-like principles, as the fore-ground/background distinction is the label under which this topic is dealt with in most other works.

7.3. The PUTI in other works

a. A first set of observations on a principle similar to the PUTI occurs in the field of discourse analysis (henceforth: discourse-discussions). These discussions examine how we perceive the movement of time in narratives, a narrative being "one method of recapitulating past experience by matching a verbal sequence of clauses to the sequence of events which actually occurred" (Labov -Waletzky 1967: 20). Narrative clauses are "characterized by temporal sequence: their order cannot be changed without changing the inferred sequence of events in their original semantic interpreta-tion" (Labov - Waletzky 1967: 21). In this framework, the PUTI is reflected in the foreground-background distinction. The foreground is "linguistic material which charts the progress of a narrative through time, while the background is durative and descriptive material which serves to embellish and elaborate upon the foreground" (Ehrlich 1987: 363). The following examples, in which the foregrounded clauses are italicized, should suffice as an illustration of the distinction:

(7) a. [Dozing a little, Alleyn sat slumped forward in his seat] A violent jerk woke him. The train had slowed down. He wiped the misty windowpane, shaded his eyes, and tried to look out into this new country. The

The PUTI in other works 253

moon had risen. He saw arching hills, stumps of burnt trees, some misty whiteflowering scrub, and a lonely road. It was very remote and strange ...He turned to see Susan dab at her eyes with a handkerchief. She gave him a deprecatory smile. (Dry 1981: 234)

b. Nick opened the door and went into the room. Ole Anderson was lying on the bed with all his clothes on. He had been a heavyweight prizefighter and he was too long for the bed. He lay with his head on two pillows. He did not look at Nick. 'What was it?' he asked. (Reinhart 1984: 783)

The two basic characteristics determining the foreground/background status of a sentence are said to be:

1. Achievement and accomplishment sentences (bounded telic sentences) belong to the foreground (i.e. refer to W-posterior situations); states and activities (unbounded atelic sentences) belong to the background (i.e. refer to W-simultaneous situations).

2. Main clauses carry the action forward, subclauses do not.

Other characterizing features are:

3. Foregrounded clauses are usually associated with what is "humanly important" (Fleischman 1985: 857). (rejected by Reinhart 1984: 787)

4. Foregrounding events are important for plot development (Fleischman 1985: 858). (rejected by Reinhart 1984: 789)

5. The greater the degree to which an element is "unpredictable or unexpected in a given context", the more likely it is to have a foregrounding function (Fleischman 1985: 859).

6. "The focus structure of the backgrounded clause is different from that of the foregrounded clause. In backgrounded clauses, there is a greater likelihood of topic changes and of new information being introduced in the preverbal position (i.e. indefinite subjects). In foregrounded clauses, on the other hand, it is unusual for completely new information to be introduced in the subject: more often, subjects are highly presuppositional, and the new material in the story is


introduced in the predicate, either in the verb or in the combination of the verb plus complement" (Hopper 1979a: 215).

7. "Strictly speaking, only foregrounded clauses are actually NARRATED. Backgrounded clauses do not themselves narrate, but instead they support, amplify, or COMMENT ON the narration" (Hopper 1979a: 215).

8. "Events in the foreground are expressed via lexical verbs with perfective aspect (...) or in English, in the simple form (...). Imperfective aspect, or in English the progressive form, is reserved for events which are part of the background (...), or which belong to some other section (e.g. orientation, evaluation) of the narrative" (Couper-Kuhlen 1989b: 8).

It is basically characteristic 1 of the foreground which is reflected in Declerck's PUTI for unembedded clauses: achievements and accomplishments, i.e. bounded telic clauses refer to W-posterior situations (provided a bounded clause precedes), whereas states and activities, i.e. unbounded atelic clauses refer to W-simultaneous situations. The question whether it is (un)boundedness or (a)telicity which is important for the PUTI will be considered in section 7.5.1. As foregrounded SCs are exceptional (cf. characteristic 2), the set of SC examples which constitute a sequence (PUTI for embedded clauses (part a)) is likely to be fairly limited.

b. Analyses of how sentences are temporally interpreted also fea-ture within the discussions of reference time (RT) (henceforth: RT-discussions), which is understood as the situation which functions as position of evaluation for the interpretation of an utterance. The following quotation from McCawley (1973) illustrates how RT should be understood:

(8) The Lone Ranger broke the window,, with the barrel of his gun, took , aim, and pulled. the trigger. (McCawley 1973: 269)

t2 contains implicit reference to t. ('shortly after t{') and t3 contains an implicit reference to t2 ('shortly after t2). If we in fact say that t] is the antecedent of t2 and that t2 is the antecedent of tv then Langacker's constraint [a pronoun must be preceded or commanded by its antecedent] explains why [8] is not equivalent to:

The PUTl in other works 255

[9] The Lone Ranger pulled the trigger, took aim, and broke the window with the barrel of his gun. (McCawley 1973:269)

It is basically characteristic 1 of the foreground-background dis-tinction which is reflected in the rule of R-progression in the RT-discussions, although it is formulated in slightly different terms: an event (cover term for accomplishments and achievements) is said to trigger a shift of reference point, whereas a state or an activity does not (cf. e.g. Hinrichs 1986: 68, 70, Partee 1984: 254-256). It will be clear that in Declerckian terms, the second part of the above claim (a state does not trigger R-progression) implies that there is temporal subordination if R is to be equated with what Declerck calls TOsit. From that point of view, this claim does not fit into a discussion of the PUTI, as this principle only holds for sequences of clauses in which the domain is shifted. However, some caution is necessary when making comparisons of this kind: it seems that the term reference time is not synonymous with TOsit in many of the RT-discussions. The claim that a state does not trigger R-progression means: when a state is introduced, we do not move the position at which we had arrived after processing the last sentence, i.e. we "stay where we are", whereas in the case of events, we move forward. Let us take an event,-state-event2 sequence to illustrate the principle of R-progression. The introduction of an event coincides with the introduction of a new reference point (position of evaluation) which lies just after event,. From this position we will evaluate the state which follows and event2 (as states do not trigger R-progression). The introduction of event2 again coincides with the introduction of a new reference point, which lies after the event itself. Partee (1984: 254) visualizes the principle as follows:

Jameson entered the room, shut the door carefully and switched off the light. It was pitch dark around him, because the Venetian blinds were closed, (example from Hinrichs)

r0 e \ r \ e2 r2 ri S\ S2 rs

e 2< r 2 < r s


3 3 s

r} c s2

ex: Jameson enter the room

e2:....

(Partee 1984: 254-255) [r0 is the given past reference time and rs is the present reference time ]

The fact that a state includes the RT of the preceding situation but not necessarily the event that led to the introduction of the RT explains why a state need not overlap with the preceding event. The inchoative state It was pitch dark around him in Partee's example is a case in point: s, overlaps with r3 but not with e3. Dowty's (1986) and Nerbonne's (1986) versions of the rule of R-progression differ slightly from Partee's proposal. They argue that R moves irrespective of the nature of the newly introduced situation; pragmatic knowledge will decide whether it overlaps with the preceding one or not. In this way, examples with inchoative states can also be accounted for. This approach seems to be more closely related to that of Declerck: the claim that there is R-progression probably means that there is a sequence of shifts of domains, in which case pragmatic principles (i.e. the PUTI) play a very important role in determining the temporal interpretation.

c. Most of the above-mentioned characterizing features of the foreground-background distinction/rule of R-progression have been challenged. The following counterexamples are also relevant to the PUTI:

1. A set of event clauses may be interpreted as referring to W-simultaneous situations (cf. e.g. Couper-Kuhlen 1989b: 18, Dowty 1986: 58, Houweling 1986: 188, Nakhimovsky 1988: 41, Partee 1984: 256, Richards 1982: 86):

(10) At the signal, everyone went to work at once. Mary searched the room for any of the items on the list that might be there. John went to the next door to do the same in Bill's apartment. Susan organized the rest of the players to canvass the block. (Dowty 1986: 58)

The PUTl in other works 257

2. An event may overlap partially with a preceding event (cf. e.g. Dowty 1986: 58, Vet - Molendijk 1986: 143-144):

(11) Nadina sat down and read the letter. She was surprised to notice that the letter touched her very much. (Vet -Molendijk 1986: 143)

3. An event may refer to a W-anterior situation (cf. e.g. Cooper 1986: 33, Nerbonne 1986: 92-93, Ter Meulen 1991: 525):

(12) The choir sang "Glory to the God in the highest". Margaret sang the recitative. (Cooper 1986: 33)

4. Non-progressive states that are interpreted inchoatively may belong to the foreground (cf. e.g. Cooper 1986: 34, Couper-Kuhlen 1989a: 17-18, Dry 1978 in Dowty 1986: 51, Nakhimovsky 1988: 41, Schöpf 1991: 247-248, Smith 1982: 180, 1983: 488):161

(13) a. Away, in front, the engine whistled. Trees, hills and road, slid sideways and were gone. (Dry 1978 in Dowty 1986: 51)

b. She looked at him, smiling. Then she was in his arms and he was kissing her. (Couper-Kuhlen 1989a: 17)

c. John went over the day's perplexing event. Sud-denly, he was fast asleep. (Dowty 1986: 50)

5. Activities may be W-anterior to the preceding event (cf. e.g. Vet -Molendijk 1986: 150-151):

(14) They lifted the minister from his bed. He incited the people to revolt. (Vet - Molendijk 1986: 151)

6. Activities may push the action forward (cf. e.g. Dowty 1986: 53, Hatav 1989: 513):

(15) The king was born, grew up, reigned and died. (Hatav 1989:513)

7. SCs may belong to the foreground (cf. e.g. Aristar Dry 1983: 31, Couper-Kuhlen 1987: 23, 1989a: 12, 20):


(16) I didn't move. She gave me another cute glance and went on towards the front door. She had her hand on the knob when we both heard a car coming. She looked at me with questions in her eyes. I shrugged. The car stopped, right in from of the house. (Couper-Kuhlen 1987: 23)

Declerck (1991a: 119), Dowty (1986: 58), Nerbonne (1986: 91, 94) point out that "RT progression"-like rules are conversational implicatures: the pragmatico-semantic meaning of an utterance may override the general rule that events refer to subsequent situations and states and activities overlap with the preceding situations. However, I do not believe that the so-called counterexamples just given provide evidence for the claim that the rule of RT-progression/foregrounding is an implicature. As the temporal order follows from the semantics of the situations, there is no need to appeal to rules of implicated temporal order. There is indeed a discrepancy between the claim that the R-progression rule/PUTI is an implicature and the examples given to substantiate that claim, in which the semantics imply -almost without exception- a certain chronology (cf. Schöpf 1991: 248-249):

(17) a. John went to the door and knocked three times. The door opened slowly. (Declerck 1991a: 124)

There is a logical order between the situations: it is logical to move closer to a door before knocking on it and knocking on a door usually occurs before it is opened. A similar comment applies to the following example:

(17) b. He moved to London and found a job there as a teacher. He courted a girl named Mary and married her. (Declerck 1991a: 124)

The adverbial there in the second clause implies that finding a job happens after moving to London. Similarly, it is a usual rule to court a girl before marrying her. Couper-Kuhlen (1987: 10) points out that in Al went to New York. The others were there once too, the second situation may hold after, before or at the same time as the situation in the first clause. However, she fails to notice that due to the presence of the adverbial once an interpretation in terms of W-anteriority will probably be preferred. Evidence for the implicature-status of PUTI-like rules can only be derived from examples in which the contents

The PUTI in other works 259

or pragmatics do not contribute at all to establishing a chronological order:

(18) a. She met the man of her life and got married.

The unmarked interpretation of this sentence which follows from the rule of R-progression (the rule of foregrounding or the PUTI) is that she marries the man of her life. However, this temporal order is im-plicated, witness the fact that it can be cancelled:

(18) b. She met the man of her life and got married, but not in that order.

It is only examples such as (18) that justify the claim that PUTI-like rules are implicatures. Two more points should be added:

a. It is striking that most of the observations are based on past tense sentences. The almost exclusive consideration of past tense examples in the discourse-discussions may be related to the topic of investiga-tion, i.e. narrative, which typically uses the narrative past.162 How-ever, the exclusive focus on past tense examples in the RT-discus-sions (apart from a few exceptions) cannot be justified by an intrinsic reason, as RT is important for any sentence type. This point provides a first topic for investigation: are the discourse interpretation princi-ples (and Declerck's PUTI) applicable to sequences located in sectors other than the past sector?

b. Most analyses do not make an explicit distinction between tem-porally subordinated clauses and shift of domain clauses, although it is clear that this observation is vital to any discussion of this type. It is useful to point out as a preamble to a treatment of this topic that the question of how temporally subordinated clauses (that are bound directly) are interpreted is not directly relevant to the PUTI/the inter-pretation of narrative, as the tenses as such perform the function of indicating the temporal relations.163 A principle similar to the PUTI is only needed when no other elements suggest a chronological order. Declerck's analysis is more accurate in this respect because he ex-plicitly points out that the PUTI applies to shift of domain sequences.


7.4. The PUTI for embedded clauses (Declerck 1991a): examples

7.4.1. Bounded + bounded

Declerck exemplifies the (a)-part of the PUTI for SCs with the following set of NRRC examples:

(19) a. John hit Bill, who hit him. (Declerck 1991 a: 139) b. John, who hit Bill, was hit by him. (ibid.) c. Bill hit John, who hit him. (ibid.)

The situations in the above sentences tend to be interpreted in the way in which they are reported, "particularly if (...) [they] are couched in a context consisting of clauses that carry the action forward" (Declerck 1991a: 138). As SCs do not often carry the action forward, the applicability of this part of the PUTI is fairly limited; the chronological order between the two bounded situations of a complex sentence will in most cases be left unspecified (Declerck 1991a: 138):

(20) Bill was killed on the same spot where John had his accident. (Declerck 1991a: 138)

In Declerck's opinion, (20) allows the interpretation in which John's accident precedes Bill's death, as well as that in which Bill's death precedes John's accident. However, it may be that same contributes to establishing a preference for the former interpretation. Example (21) also illustrates the (a)-part of the PUTI for SCs:

(21) a. The man who killed the rat bought us a beer. (Declerck 1991a: 138)

b. The man who bought us a beer killed the rat. (ibid.)

Declerck argues that (21a) and (21b) do not suggest a different chronology, "the temporal order of the two situations is simply unspecified" (Declerck 1991a: 138). The same applies to (22):

(22) a. The man came to my office who won the first prize yesterday. (Declerck 1991a: 138)

b. The man who won the first prize yesterday came to my office, (ibid.)

The PUTI for embedded clauses (Declerck 1991a): examples 261

As I see it, the presence of the adverbial yesterday will tip the scales in favour of a W-anterior interpretation of winning the prize.

It is striking that when the RC follows the SUPC, the RRC does not suggest a W-posterior reading (cf. (20) and (22a)) whereas the NRRC does (cf. (19a) and (19c)). Declerck writes: "Not all sentences establishing a domain have a foregrounding function. For example, nonrestrictive relative clauses are mostly background" (1991a: 140-141). The examples in (19) show that NRRCs can belong to the foreground. In what follows, it will become clear that, contrary to what Declerck claims, bounded (especially telic) NRRCs almost always belong to the foreground. It should also be added that principle (a) of the PUTI for embedded clauses includes the danger of circularity: Declerck claims that a sequence of a SUPC and a SC is chronologically ordered provided "the subclause is one of the clauses that carry the action forward (i.e. that have a foregrounding function)" (Declerck 1991a: 138). As the foreground is defined as a sequence of chronologically ordered events, Declerck's statement is tautologous:164 Why are SCs interpreted as a sequence? Because they belong to the foreground. - Why do the SCs belong to the foreground? Because they are chronologically ordered.

Four issues need to be discussed in more detail:

1. Do NRRCs always carry the action forward? 2. Do RRCs ever carry the action forward? Is there indeed no

preferred reading associated with RCs such as those in (20) to (22)?

3. Does the order in which the clauses are reported influence the temporal interpretation?

4. Is it (a)telicity or (un)boundedness which is the decisive factor in the PUTI?

7.4.2. Unbounded + unbounded

The following RC examples illustrate principle (b) of the PUTI for embedded clauses (Declerck 1991a: 139), i.e. when the two situations are unbounded, the unmarked temporal interpretation is one of simultaneity:

(23) I didn't see the girls that wore red hats. (Declerck 1991a: 139)

(24) You won't see the girls that will wear red hats, (ibid.)


(25) Tom, who lived in London, knew the way to the Strand, (ibid.)

(26) Tom, who will then live in London, will know the way to the Strand, (ibid.)

The question brought up by (23) and (25) is whether examples of this type really illustrate the PUTI, i.e. is the verb form in the RC an absolute tense or a relative tense? Declerck himself gives two reasons why the past tense in a similar sequence of unembedded clauses is an absolute one:

(27) a. John stood by the window. Mary was in the kitchen. (Declerck 1991a: 126)

a. The first reason is that in the corresponding post-present sector example, a FPS form is used:

(27) b. John will stand by the window. Mary will be in the kitchen. (Declerck 1991a: 126)

It has been argued in chapter 4 (section 4.3.) and chapter 5 (section 5.2.) that it is difficult to make cross-sectorial generalizations as regards the use of tense. The system of temporal subordination in the post-present sector is indeed subject to a constraint not applicable to the past sector: the use of the past/past perfect (shift of domain vs. temporal subordination) to refer to a W-anterior past sector situation can never result in the location in a different sector (no matter whether the past or the past perfect is used, the situation is located in the past sector) whereas the use of the future perfect/present perfect or future/present tense (shift of domain vs. temporal subordination) to refer to a W-anterior or W-simultaneous post-present situation can: the present perfect/present tense may be misinterpreted as a statement about a pre-present/present sector situation. As has been explained before (cf. chapter 3, section 3.3.), it is not completely safe to rely on a comparison with the post-present sector to determine the status of the past tense. Although a FPS verb form is used in the post-present counterpart of (23) and (25) (cf. (24) and (26)), I hesitate to draw conclusions about the status of the past tense example on the basis of that observation.

b. The second test Declerck mentions is the backshifting test:

The PUTI for embedded clauses (Declerck 1991a): examples 263

(27) c. The witness explained that John had stood by the window and that Mary had been in the kitchen. (Declerck 1991a: 126)

As it is possible to backshift the original past tense without changing the temporal relations between the situations, the original past tense was in (27a) must be an absolute tense, as relative past tenses do not allow the backshift (for more comments on this test cf. chapter 3, section 3.3.)· When we apply the backshifting test to (23) and (25) it appears that the backshift may result in a different temporal location of the RC situation. The unmarked interpretation of the past perfect in the RC is that the situation is anterior to the SUPC situation in both (23b) and (25b):

(23) b. He said he hadn't seen the girls that had worn red hats.

(25) b. He said that Tom, who had lived in London, had known the way to the Strand.

The semantic links between living in London and knowing the way to the Strand are tighter than those between seeing girls and wearing red hats. This may explain why the past perfect RC is more likely to be interpreted as W-simultaneous with the SUPC situation in (25b) than in (23b). In any case, the past perfect allows two different interpretations. This means that the backshifting test does not solve the problem of the status of the past tense. If backshifting is not applied in (23) and (25) wore and knew are interpreted either in terms of a shift of domain or as temporally subordinated verb forms:

(23) c. He said he had seen the girls that wore red hats. (temporal subordination with respect to either the SUPC situation or the matrix clause situation or shift of domain)

(25) c. He said that Tom, who had lived in London, knew the way to the Strand, (temporal subordination with respect to either the SUPC situation or the matrix clause situation or shift of domain)

Declerck points out (personal communication) that if one transposes (23c) to the post-present sector, the future perfect cannot be used in the RRC, which indicates that wore is a relative past:


(23) d. He will say that he didn't see the girls that wore (* will have worn) red hats.

However, the use of the future perfect is not very felicitous in the NRRC either (unless it is interpreted as a comment by the speaker and does not belong to the original utterance of the subject):

(28) He will say that he didn't see Mary and Susan, who wore (?? will have worn) red hats.

Moreover, as has been pointed out before, the intrinsic nature of the tense system in the post-present sector implies that any argument aimed at showing that there is a difference between RRCs and NRRCs will benefit from a comparison with the post-present sector.

As no conclusive proof can be given for the claim that wore in (23), knew in (25) and was in (27a) are absolute tenses, it becomes difficult to maintain the hypothesis that the status of the past tense verb form is determined by the status of the clause in which it is embedded (SC vs SUPC). One might indeed wonder why the status of wore in a RRC would be different from that in an unembedded clause. As the function of the tense is the same in the RC and in the independent clause in the sense that there is a relationship of W-simultaneity between the situations referred to in the two clauses in (23) as well as in (29), wore might be argued to be a relative tense in both cases:

(23) I didn't see the girls that wore red hats. (29) I saw the girls. They wore red hats.

As I see it, the "functional" argument just given is evidence in favour of a unitary treatment of the past tense. If the past tense is a relative tense in (23) and (25), the latter examples do not illustrate the PUTI.

7.4.3. Unbounded + bounded or bounded + unbounded

Principle (c) of the PUTI for embedded clauses runs as follows: "When one clause is unbounded while the other is bounded, the unmarked interpretation is that the bounded situation is included in the unbounded one. This is the case irrespective of whether the unbounded clause is the head clause or the subclause and irrespective of the order of the clauses" (Declerck 1991a: 139). Accordingly,

The PUTI ((a)-part) and RCs 265

there is a relationship of inclusion between the RC situation and the SUPC situation in the following examples:

(30) a. Bill, who seemed to know everything about the stranger, told us who he was. (Declerck 1991a: 139)

b. Bill, who told us who the stranger was, seemed to know everything about him. (ibid.)

Again, the question arises as to whether the second clause is temporally subordinated to the first or not. As will be clear from the previous section, there is no completely watertight evidence against analyzing the RC past tense as temporally subordinated to the other situation.

The following issues for further analysis have arisen in the dis-cussion of the PUTI (part a) and will be examined in section 7.5.:

Question 1. Is it (a)telicity rather than (un)boundedness (or a combination of both factors) which determines the temporal interpretation? (section 7.5.1.) Question 2. Do bounded NRRCs always refer to W-posterior sit-uations? (section 7.5.1.) Question 3. Can RRCs push the action forward (and hence refer to a W-posterior situation)? Is there any preferred reading in a sequence consisting of a bounded-telic SUPC and a bounded-telic RC, or is the chronological order indeed vague, as Declerck claims? (section 7.5.2.) Question 4. Does the order in which the (N)RRCs are uttered influence the temporal interpretation? (section 7.5.3.) Question 5. Do different sectors suggest different temporal in-terpretations? (section 7.5.4.)

7.5. The PUTI ((a)-part) and RCs

When choosing examples, a number of criteria should be taken into account: one has to take care that the sentences do not pragmatically imply a certain order (e.g. The man who fell off the stairs broke his leg). However, there must be some sort of semantic link between the SUPC situation and the RC situation for the sentence to make sense: sentences of the type John married the girl who killed a rat can hardly be used out of context. A further restriction is that it should be


possible for the RC situation to follow as well as precede the SUPC situation. In order to prevent interpretations in terms of partial overlapping, the sentences should be telic bounded (if possible punctual). However, if we are to check the validity of the PUTI, which is formulated in terms of bounded/unbounded situations and not in terms of telic bounded/atelic unbounded situations, some of the examples will necessarily have to be bounded atelic. The points made are to a great extent based on the results of an elicitation test carried out in the English Department of University College London, in which 64 informants participated (cf. appendix 2 for a description of the test). However, due to practical reasons,165 I am not in a position to support all the claims with a large number of native speakers' judgments. It is therefore better to consider those observations as tendencies rather than strict rules.

7.5.1. Question 1 and question 2

In this section, an answer will be formulated to the following questions: 1. Is it (un)boundedness or (a)telicity which determines whether the action is pushed forward?, 2. Do NRRCs always carry the action forward?

Some linguists use the achievement/accomplishment (event) vs. state/activity dichotomy, while others prefer the telic/atelic distinction when discussing PUTI-like principles. The majority speak in terms of state (including activities) vs. event, but it is difficult to see whether this dichotomy corresponds to the bounded/unbounded distinction rather than to the telic/atelic distinction. The examples given to illustrate the rules of R-progression and foregrounding are mostly bounded telic. Similarly, Declerck uses only bounded telic situations to illustrate principle (a) of the PUTI (for unembedded clauses and for embedded clauses), although the rule is formulated in terms of (un)boundedness. Although this may be a mere coincidence, it is interesting to check whether the PUTI for embedded clauses also applies if the situations are bounded atelic: is it not (a)telicity rather than (un)boundedness which is the crucial factor in connection with the temporal interpretation of a set of shift of domain sentences?

Although foregrounding and RT-progression rules are mostly formulated in terms of the Vendlerian Aktionsart classes, the question whether it is (a)telic situations rather than (un)bounded situations which trigger a shift of RT has not gone entirely unnoticed. As was mentioned in section 7.3., several linguists have pointed out


that states that are interpreted inchoatively may trigger R-progression/move narrative time. Declerck correctly points out (1991a: 127) that in examples of this kind the situation is bounded to the left: from that point of view these examples follow the PUTI (cf. Dry (1981) and Eisterhold (1986: 258) for the same observation). Hatav (1989) argues that bounded atelic situations may be "located on the time line" (i.e. may push the action forward). She puts it as follows: "Situations have end points iff they are contained in their reference time, and only such situations can appear on the time line. Events are always contained in their R-time and hence are always candidates for the time line, but states are contained in it only when (a) they are interpreted as inchoatives or (b) their duration is restricted by overt linguistic marking, for example, adverbials such as 'for three hours'" (Hatav 1989: 487). In her opinion, it is the refer-ence to the endpoint of a situation which determines whether a situation appears on the time line or not. The logical outcome of such an approach is that bounded states may also be foregrounded. Hatav illustrates her claim with examples from Biblical Hebrew, Russian and French:

(31) Quand on fut extenue, on fit la paix. (Hatav 1989: 491)

The first clause is bounded to the right. The sentence means that peace can only be made once a certain stage has been reached, i.e. when they are tired. This implies reference to a point of transition: they no longer feel like fighting (= end-point), they are tired (= initial point). In other words, the example in (31) does not really differ from the inchoative state/activity examples (cf. (13)); the situation in the first clause is not bounded on both sides. Even so, the fact that such situations may belong to the foreground (may trigger R-progression) can already be taken as an indication that the important criterion for the PUTI is indeed (un)boundedness rather than (a)telicity. Couper-Kuhlen (1987: 16) also explicitly points out that (a)telicity is too narrow a concept to account for all the examples containing foregrounded clauses; she rejects Nerbonne's idea that (a)telicity is the determining factor.166 Couper-Kuhlen's examples should prove that "certain non-telic [atelic] predicates will also be understood to refer to events in succession in narration" (Couper-Kuhlen 1987: 16):

(32) The balloon popped. The child jumped. (Couper-Kuhlen 1987: 16)


The first clause is bounded telic. If the second clause is interpreted as a single action (bounded telic), there is reference to a sequence. However, the second clause could also refer to an iterative situation, in which case the child starts jumping as a result of the sudden explo-sion of the balloon, i.e. the second situation is bounded to the left. A similar comment applies to the following example:

(33) The guide looked up at the sky. The tourists looked up at the sky. (Couper-Kuhlen 1987: 16)

If the situations in (33) are interpreted as telic and bounded, there is reference to a sequence: the guide looks up at the sky and as a result all the tourists look up at the sky. The difference with the example in (32) is that the situation in (33) is a "punktuelle Veränderung" [punctual change] (Schöpf 1984: 241), i.e. it refers to an action plus a resulting state of affairs, the balloon popped in (32) is a "punktuelles Ereignis" [punctual event] (Schöpf 1984: 243). The course of events in (33) could be represented as follows:

χ

X

On the bounded telic reading, the actions of looking up at the sky follow each other, the resulting states are W-simultaneous. When look up at the sky is interpreted as atelic and unbounded, the sit-uations occur W-simultaneously. In other words, the sentences in Couper-Kuhlen's examples are not unequivocally bounded on both sides either. Aristar Dry (1983) also touches upon the (un)boundedness-(a)telicity issue. When describing the factors which result in a situation being understood as foregrounded, Aristar Dry (1983) rejects the view expressed in an earlier article of hers (Dry 1981) that it is a change of state that triggers the movement of RT: "Although that assertion is true within limits, I have come to see that it should be broadened: references to changes of states necessarily constitute references to points ~ the beginning and ending points of situations — but sentences like (5) indicate that reference to se-quenced points which are not changes of state also trigger perception of time movement:

(5) (a) At twelve the sky was threatening (b) and at one there was no change.


So this paper argues that those constructions that move time refer to sequenced points, not changes of state" (Aristar Dry 1983: 23). Although Aristar Dry does not use the terms (a)telic or (un)bounded, the following quote also indicates that it is (un)boundedness she has in mind: "It is reference to sequenced temporal points which triggers the illusion of time movement; and (...) the points most often referred to are initial and final points of situations" (Aristar Dry 1983: 47). The above observations prove that foregrounding/movement of R should be formulated in terms of (un)boundedness rather than (a)telicity or Vendlerian situation types. However, the sentences with which that claim has so far been substantiated are not bounded "on both sides"; the states or activities in (31) to (33) are bounded to the left or to the right. The following sentence does contain a bounded atelic situation and proves that it may indeed belong to the foreground:

(34) There was a small ivory push button beside the door marked '405'. I pushed it (...) and waited for what seemed a long time. Then the door opened noiselessly about a foot. (Couper-Kuhlen 1989a: 15-16) (italics mine)

Apart from showing that atelic bounded sentences may belong to the foreground, this example again shows that the semantics and pragmatics of the clauses play a crucial role in determining whether narrative time moves forward or not. Even if for what seemed a long time is not added to the verb waited, our knowledge of the world will tell us that opening the door puts an end to the situation of waiting. In other words, any kind of situation in a main clause can move the narrative time if such a shift is dictated by our knowledge of the world. In any case, inchoative state examples, such as that in (35), also justify the conclusion that it is boundedness rather than telicity which is important for the PUTI (cf. Declerck 1991a: 128):

(35) Mr Darby slapped his forehead, then collected himself and opened the door again. The brush man was smiling at him hesitantly. (Hinrichs 1986: 69-70)

The other question to be answered is whether NRRCs always refer to W-posterior situations. When trying to list NRRC examples of the type telic bounded SUPC + atelic bounded RC which refer to a sequence, the most likely possibility seems again to be sentences


whose contents imply a certain order. The sentence in (36) refers to the usual pattern of signing a contract before taking up a job:

(36) He signed a contract with IBM, for whom he worked for 3 years. Results of elicitation test: 1 - 2 :50% 1 - 2 (2 - 1) : 19% 2 - 1 :20% 2 - 1 ( 1 - 2 ) : 11%

The informants could choose from four options: 1-2: the situation mentioned first happened before the situation mentioned secondly, 2-1: the situation mentioned first happened after the situation mentioned secondly, 1-2 (2-1): both orders are possible with a preference for the 1-2 order, 2-1 (1-2): both orders are possible, with a preference for the 2-1 order. The results of the elicitation test indicate a clear preference for the W-posterior interpretation of the RC in (36). 69% of the informants thought that this was the only or preferred reading of the sentence.167 In other words, the NRRC is most likely to be read as a continuative RC (Jespersen 1961, III: 105).

However, if no order is suggested by the pragmatico-semantic meaning of the clauses, and, strictly speaking, the PUTI applies to examples in which the lexical content does not imply a certain order, it becomes more difficult to determine the chronological order between the NRRC situation and the SUPC situation:

(37) Tom met his girlfriend in London, where he lived for three years.168 (W-anteriority, W-simultaneity or W-posteriority) 1 - 2 : 19% 1 - 2 ( 2 - 1 ) : 31% 2 - 1 :30% 2 - 1 ( 1 - 2 ) : 20%

The results of the elicitation test indicate the lack of preference for either the W-anteriority reading (50%) or the W-posteriority reading (50%). However, the figures are somewhat misleading in that the informants could not choose the option of W-simultaneity, which maybe many would have preferred. NRRCs give relevant, "background" information: in (36), the relevant information very clearly refers to a series of situations, which is not the case in (37). In


the latter example, the relevant, characterizing information is not one of two consecutive situations. The difference is also clear from the following versions of (36) and (37):

(36') John signed a contract with IBM. He worked there for 3 years.

(37') John met his girlfriend in London. He lived there for three years.

The following example is similar to that in (36):

(38) John married Mary, who stayed with him for 5 years. 1 - 2 :74% 1 - 2 (2 - 1) : 19% 2 - 1 :4% 2 - 1 ( 1 - 2 ) : 2%

Example (38) is like (36) in that the order "getting married - living with someone" is the unmarked order still deeply rooted in most people's minds. In other words, although the order "living with someone - getting married" is not unlikely, this option may have been chosen less often because the other pattern prevails in the soci-ety to which the informants belong. From that point of view, it is the "unmarked temporal patterns" prevalent in our minds rather than the syntactic independence of the NRRC which explains why the sentence gets a W-posterior reading. In (39), no chronological order is inherent in the meaning of the NRRCs, but unlike in (37), the atelic bounded NRRC precedes the SUPC:

(39) John, who lived in London for three years, met his girl-friend in Paris. 1 - 2 : 56% 1 - 2 ( 2 - 1 ) : 36% 2 - 1 :0 % 2 - 1 (1 -2) : 8%

The figures are revealing because they show that a first-mentioned NRRC will be read as part of a sequence whenever that reading is possible, hence the outspoken preference (92%) for the "1-2" option in (39). In (40), the NRRC follows the SUPC (cf. (37)), but unlike the NRRCs of (37) and (39), it is bounded telic:


(40) John married Mary, who took a course in chemistry. 1 - 2 :24% 1 - 2 (2 - 1) : 53% 2 - 1 :3% 2 - 1 ( 1 - 2 ) : 20%

Again, the results are somewhat misleading in that the simultaneity option could not be chosen. However, a comparison with the results of (37) shows that (a)telicity influences the choice of chronological order. In (40) (77%), in which the NRRC is bounded telic, unlike in (37) (50%), in which the NRRC is bounded atelic, there is a clear preference for a continuative, i.e. W-posterior interpretation. The above examples support the claim that the chronological order will be one of W-posteriority whenever the latter is easily compatible with the semantics and pragmatics of the sentences.

The following conclusions can be drawn: 1. Bounded NRRCs do not always carry the action forward (cf.

possible W-simultaneity interpretation of (37)). 2. If (a) the NRRC is first-mentioned and can be read as part of a

sequence169 (cf. (39)) or if (b) the NRRC allows a continuative reading (especially if it is bounded telic, as in (40) and (41), but also bounded atelic, as in (36) and (38)), it is likely to belong to the foreground and will result in a W-posterior interpretation of the NRRC:

(41) John married Mary, who got pregnant.

3. It also follows from the above discussion that it is (un)boundedness which is the determining factor for the PUTI.

Before we turn to the next issue, the possible effect of zero-bounded/zero-telic situations on the temporal interpretation should be mentioned. As pointed out in chapter 1 (section 1.4.), some sentences allow a bounded and telic as well as an unbounded and atelic interpretation. Depending on the option chosen, the action will (not) be pushed forward. Although no generalizations have been made in this respect, this observation has been made before, i.e. a number of examples seem to allow two interpretations depending on whether a certain clause is interpreted as an "event" or a "state" (cf. e.g. Couper-Kuhlen 1987: 15, 1989a: 16-17, Dowty 1986: 38, 50):

(42) John entered the president's office. The president realized why he had come. (Dowty 1986: 38) (italics mine)

The PUTl ((a)-part) and RCs 273

If realized refers to a state in which the president finds himself, it is W-simultaneous with John's entering. If, however, realized captures the moment in time at which the president suddenly realizes why "he" had come (i.e. realized is interpreted as telic and bounded), there is reference to a sequence. A similar remark applies to the following example:

(43) Dan Rolff was in the living room. A water glass and a brown bottle with no label stood on the table in front of him. He sat straight up in his chair, staring at the bottle. (Couper-Kuhlen 1987: 15)

Accordingly, zero-bounded RCs also allow two possible interpre-tations:

(44) John called the children, who crossed the street.

There may be reference to a sequence of events: John calls the children, upon which they cross the street (telic bounded). The other interpretation is that John calls the children while they are crossing the street (telic unbounded). However, a speaker conforming to the rules of communication would probably use a progressive form (were crossing) if he wanted to indicate W-simultaneity.

7.5.2. Question 3

This section tackles the following question: Do RRCs ever carry the action forward? A comparison will be made between the temporal interpretation of RRCs and that of NRRCs.

7.5.2.1. Other works on foregrounding in RCs

It follows from the general characterization of the foreground that the subordinate status of RCs should prevent them from pushing the action forward (cf. characteristic 2), narrative temporal clauses (cf. e.g. Couper-Kuhlen 1989a, Declerck 1991a: 145-148) being the only exceptions to the general rule that foregrounding is restricted to main clause. Reinhart (1984) and Aristar Dry (1983) are the only two publications which touch upon the question whether (N)RRCs can belong to the foreground. Reinhart argues that RCs, no matter


whether they are restrictive or non-restrictive, never indicate foregrounded information. She illustrates her claim with the following examples:

(45) a. The man who stole Rosa's dog beat him. (Reinhart 1984: 796)

b. Israel said 'Goldman likes ancient music too,' and Eliezra, who was playing with her glasses, said 'but I don't like him.' Israel gazed again at her neck and arms which he knew so well from Goldman's pho-togiaphs. Goldman used to take pictures of her in all possiUe positions, standing, undressing, or lying naked o/> the sofa, and from time to time he would show these pictures to Israel who got upset by Eliezra's last words ... (J. Shabtai, present imperfect, p. 198) (Reinhart 1984: 796-797)

In Reinhart's opinion, "something has gone wrong" (Reinhart 1984: 797) in the last clause in (45b) in the sense that "information, which does belong to the time axis, is reported in a relative clause, while a relative clause usually signals material not belonging to the foreground" (Reinhart 1984: 797). As the information in the NRRC refers to the next step in the series of events that is described, I do not see why the NRRC should be denied the foregrounding status. Aristar Dry (1983) does not share Reinhart's view. However, she points out that only a particular type of RC can trigger time movement, namely RCs that modify nouns in object position:

(46) a. The old man returned with a few lumps of coal, which he placed here and there on the fire. (Aristar Dry 1983: 36)

b. The old man ... returned with two candlesticks which he thrust one after the other into the fire and carried to the table, (ibid.)

c. He extended his hand, which Little Chandler took, (ibid.)

d. He made a circle in the air with his arms and paused. Everyone laughed or smiled at Aunt Kate and Aunt Julie and Mary Jane who all turned crimson with pleasure. Gabriel went on more boldly ... (ibid.)


e. She seemed to be somewhat disappointed at my refusal and went on quietly to the sofa, where she sat down behind her sister, (ibid.)

Aristar Dry argues that this is not surprising: the object position is typically reserved for the communication of new information (cf. characteristic 6) and therefore conforms to the traditional informa-tional pattern associated with foregrounding, i.e. foregrounded clauses provide new information. RRCs embedded in a NP in subject position "communicatefs] presuppositional material that is likely to be anaphoric in the discourse" (Aristar Dry 1983: 51) and therefore can not be foregrounded:

(47) The man who walked up to him offered him a cigar. (Aristar Dry 1983: 51)

However, Aristar Dry adds that this line of reasoning fails to explain why NRRC embedded in a NP in subject position can belong to the foreground:

(48) The man, who walked up to him gracefully, like a dancer walking, offered him a cigar. (Aristar Dry 1983: 51)

I do not agree with Aristar Dry when she claims that the RRC situa-tion in (47) is not foregrounded: the situation in the RC precedes that in the main clause. As there is reference to a sequence of situations, there seems to be no reason why the foregrounded status should not be attributed to the RRC.

7.5.2.2. Corpus evidence

It has already become clear in the course of the discussion that it is not hard to find examples in which the RC situation refers to the next step in a sequence of events. The kind of NRRC traditionally labelled continuative typically establishes a foregrounded reading:170

(49) a. The first time he was shot in the hand as he chased the robbers outside. The second time he identified two robbers, who were arrested and charged. (WSJ)

b. Before that, he was an executive with a manufac-turing concern. At Continental he cut money-losing


operations, which helped produce a modest profit in this year's second quarter. (WSJ)

c. So Brooks gave the dirty work to Coin Wrap Inc., which came up with an unusual solution. (WSJ)

The italicized clauses answer all but one of the characteristics of the foreground: (a) they are bounded situations and (b) the situation they refer to is W-posterior to the SUPC situation. The only deviant feature is that they are SCs. RRC situations may also be posterior to the SUPC situation:

(50) a. That goal was just the tonic Wolves needed, and in the sixty-third minute Murray, capping a slick combined move, made the score 3-1. Three minutes from the end a typical bit of Woosnam Soccer technique laid on a ball from which Sealey scored West Ham's second goal. (LOB)

b. A search was made, but she was not found until she was seen walking towards her home having come off the afternoon train from Carlisle. Interviewed, Dixon made a statement which was put in as evidence and the Constable alleged that he had a clear conscience. (LOB)

c. A writer who had dinner with Dickens says the menu was Whitstable oysters, a brown sole, a baked leg of mutton with oyster & veal stuffing and a gin punch. The same man went to see Carlyle, and, after mentioning that he had dined with Longfellow told the sage a very funny story which made Carlyle absolutely laugh; but all the Chelsea philosopher did in return was to ask if his guest would have a cup of tea! (LOB)

7.5.2.3. Restriction on foregrounding in RRCs

However, there appears to be a constraint on the possibility of RRCs to represent foregrounded information. Some of the RRCs that represent foregrounded information when couched in an indefinite NP no longer do so when they are embedded in definite NPs: in the latter case the situation is interpreted as W-anterior rather than W-posterior to the main clause situation (cf. (51b)); in some cases, this


preferred anteriority interpretation results in an anomalous sentence (cf. (52b)):

(51) a. Rummaging among the remains of our provisions I found a bottle with which I made numerous marks on the ground surrounding our tents and a few yards into the jungle, as far as I dared venture. (LOB)

b. Rummaging among the remains of our provisions I found the bottle with which I made numerous marks on the ground surrounding our tents and a few yards into the jungle, as far as I dared venture.

(52) a. General relief was expressed, but monsieur V (this was actually reported in the next issue) returned home and shot himself, leaving a note which again left his household goods to the superintendent. (LOB)

b. ? General relief was expressed, but monsieur V (this was actually reported in the next issue) returned home and shot himself, leaving the note which again left his household goods to the superintendent.

Most of the examples at my disposal are of the type illustrated in (51a). The unmarked interpretation suggested by the semantics of the sentences is that the RC situation is W-posterior to the main clause situation, although a W-anteriority reading is not ruled out alto-gether. When the RRC is embedded in a definite NP the latter interpretation is preferred. The acceptability of the example in (52b) is questionable because there is a clash between the W-posteriority reading which follows from the semantics and pragmatics of the clauses and the fact that the definite NP tries to impose an interpretation whereby the RRC information is W-anterior to that in the SUPC (cf. section 7.5.2.5.). Still, there appear to be a number of cases in which RRCs in definite NPs can contain foregrounded situations:

a. if the RRC precedes the main clause situation (cf. section 7.5.3.):

(53) a. Investors in stock funds didn't panick the weekend after mid-October's 190-point market plunge. Those who left stock funds simply switched into money marked funds, (adapted from WSJ)


b. A simple enough question. But the passengers who heard it turned to see who asked it. (adapted from LOB)

c. Parents were asked whether they expected the school to guide their child's behaviour as well as teach 'school subjects', and those who answered 'Yes' were asked to state what kinds of behaviour they expected the school to encourage (LOB)

b. if the RRC contains a temporal adverbial which explicitly indi-cates the temporal relations:

(54) a. Police leave has been cancelled and secret plans prepared to deal with the mass sit-down rally planned for Sunday in Parliament-square by the Committee of 100, the anti-nuclear arms group. It was Mr Butler who authorized the action which ended yesterday in 32 members of the Committee of 100 being imprisoned for inciting a breach of the peace, (adapted from LOB)

b. Fair wages were those 'generally accepted as cur-rent'. Trade unionists, however, agitated for many years for the change which was finally accepted in the revision of 1909. (LOB)

c. if the status of the RC is not unequivocally restrictive (i.e. although the antecedent is one which typically combines with a NRRC, this interpretation is not supported by the presence of a comma between the antecedent and the RC):

(55) The table of drink bottles came down. They ended against the record-player which tottered but did not fall, releasing a confetti of gramophone discs on them. (LOB)

d. if the semantics and pragmatics of the situations are not com-patible with a W-anterior reading of the RRC:

(56) a. Three minutes from the end a typical bit of Woosnam Soccer technique laid on the ball from which Sealey scored West Ham's second goal. (adapted from LOB)

The PUTI ((a)-part) and KCs 279

b. A search was made, but she was not found until she was seen walking towards her home having come off the afternoon train from Carlisle. Interviewed, Dixon made the statement which was put in as evidence and the Constable alleged that he had a clear conscience. (LOB)

7.5.2.4. Comparison RRC-NRRC

There is no such restriction on NRRCs, i.e. the (in)definiteness of the NP does not affect the foregrounding possibilities:

(57) a. Frankfurt prices fell across the board amid con-tinued rumors the Bundesbank will call for an emergency meeting Thursday to raise interest rates. Earlier in the day, a central bank spokesman denied the rumor (certain rumors), which unsettled some fi-nancial markets in Europe, (adapted from WSJ)

b. The second time he accused two (the) robbers, who were arrested and charged. (WSJ)

No matter whether the antecedent is definite or indefinite, the NRRC is understood as referring to a W-posterior situation. The following examples also prove that definite RRCs and NRRCs differ in terms of the unmarked temporal order they establish:

(58) a. John fell in love with the girl who won a beauty contest. 1-2 :6% 1-2(2-1) : 11% 2-1 :64% 2-1 (1-2) : 19%

b. John fell in love with Mathilde, who won a beauty contest. 1-2 :25% 1-2(2-1) : 39% 2-1 : 16% 2-1 (1-2) : 20%

The preference for the W-posteriority (64%)/W-anteriority (83%) reading in the NRRC/RRC is statistically significant. Examples of


this type, in which the semantics do not impose a particular reading, show that RRCs and NRRCs differ in terms of their foregrounding capacity. The same difference in unmarked temporal interpretation can be observed in the following examples:

(59) a. John married the girl who got pregnant, b. John married Mary, who got pregnant.

(60) a. John hit the man who hit him. b. John hit Bill, who hit him (= John).

It still needs to be added that RRC embedded in indefinite NPs must answer a particular requirement in order to be understood as W-posterior to the main clause situation: there must be lexical items which establish a W-posteriority reading or it must be clear from the semantics of the clauses in general (cf. (61)) that the RC situation follows the main clause situation. Otherwise, the RC situation is interpreted as being W-anterior to the main clause situation (cf. (62) and (63)):

(61) Sam married a woman who left him for another. (Heny 1982: 116)

(62) John married a girl who got pregnant.

The following example with a RRC in an indefinite NP also indicates a preference for a W-anterior interpretation of the RRC situation:

(63) Yesterday Bill met a man who robbed a bank. (Gabbay -Rohrer 1979: 100) 1-2 : 16% 1-2(2-1) : 16% 2-1 :40% 2-1 (1-2) : 28%

In NRRCs embedded in indefinite NPs, on the other hand, there need not be lexical or semantic clues in order for the SC situation to be interpreted as following the main clause situation. This is particularly clear when we compare (62) with (64):

(64) John married a 25-year old secretary, who got pregnant.

The semantics and pragmatics of the situations referred to in (62) and (64) do not impose a particular order: pregnancy may be the reason


for as well as the result of getting married. However, whereas the reason interpretation is preferred in (62) (i.e. RRC situation is W-anterior to the main clause situation), the unmarked interpretation of (64) is that the NRRC situation is W-posterior to the main clause sit-uation.

7.5.2.5. Restriction on foregrounding in RRCs explained

So far, it has been established that situations in RRCs do not always belong to the foreground: RRCs embedded in definite NPs can be foregrounded provided (a) the RRC precedes the main clause or (b) the RRC contains a temporal adverbial which makes the temporal relations explicit or (c) a W-anterior reading is not compatible with the semantics and pragmatics of the RRC. RRCs in indefinite NPs can be foregrounded provided it follows from the semantics of the clauses or general pragmatic knowledge that the RC situation is W-posterior to the main clause situation. The constraint on foreground-ing in RRCs described so far cannot be explained in terms of the syntactic subordination of the RRC as such, because in that case, it should not be possible for NRRCs, which are also embedded in a NP, to present a foregrounded situation. A more promising track can be pursued by considering the function of the relevant clauses, which is reflected in their syntactic status (i.e. loose (NRRC) vs tight (RRC) bonds with the SUPC): a RRC provides restricting (non-) identifying information in the same information unit as its antecedent. The in-formation will be identifying if the NP is definite, as it helps the hearer find out who the speaker is talking about. If the NP is indefi-nite, the RRC information does not have an identifying function, but all the same narrows down the number of possible referents. No matter whether the NP is definite or indefinite, the fact that the speaker chooses to add the information in the same information unit as the antecedent indicates that he believes that the hearer cannot do without it in order to find out who or what the speaker is talking about. The hearer knows the speaker provides him with more infor-mation so that he might establish reference. It follows from the co-operative principles that the speaker will select the kind of informa-tion that is most easily accessible to the hearer: for the provision of restricting information the speaker is likely to use information that is W-anterior to the main clause situation. This mutual presumption (the speaker uses information which he is assumes is most easily accessible to the hearer and the hearer trusts that the speaker will use


information that is easily accessible to the hearer) explains why a W-anteriority reading arises whenever the semantics and pragmatics of the situation do not impose a sequential reading. RRCs embedded in a definite NP differ from those embedded in an indefinite NP in that the urge to interpret the RC information as W-anterior to the SUPC information is even greater in the former that in the latter. This is clear from examples such as (52): although the semantics and prag-matics of the sentence suggest a W-posterior reading, the influence of the definite article, which prefers W-anterior situations to perform its reference assigning function, and accordingly tries to impose such a reading, is so strong that the resulting sentence is anomalous. The function of a NRRC in a NP differs considerably from that of a RRC. The NRRC adds relevant information in a separate information unit: the speaker assumes that the hearer is familiar with the referent of the NP, therefore, he need not bother about the "reference-restricting function" of the RRC. Bounded situations are interpreted as W-pos-terior to the main clause situation unless this reading is ruled out by the semantics and pragmatics of the situations.

The conclusion to be drawn from this section is that the unmarked chronological order between a bounded telic SUPC situation and a bounded telic RRC situation is one of W-anteriority unless a W-pos-terior reading is imposed by the contents of the sentence. In NRRCs, on the contrary, the W-posterior reading is the preferred interpreta-tion.

7.5.3. Question 4

This section considers the influence of the order in which the clauses are produced. As NRRCs are usually interpreted in the order in which they are given, it is obvious that changing the sequence "SUPC + RC" into "RC + SUPC" will coincide with a change in temporal interpretation:

(65) a. John hit Bill, who hit him. (Declerck 1991a: 138) b. John, who hit Bill, was hit by him. (ibid.)

As regards telic bounded RRCs, we may expect that the RRC situa-tion will be interpreted as happening before the SUPC situation, no matter whether it is mentioned before (cf. (65d)) (cf. Houweling 1986: 174) or after the SUPC (cf. (65c)):


(65) c. John hit the man who hit him. d. The man who hit John was hit by him (= John).

The results of the elicitation test confirm the hypothesis formulated for RRCs and NRRCs:

Sally married the man who won a scholarship. 1 - 2 6% 1 - 2 ( 2 - 1 ) 8% 2 - 1 53% 2 - 1 (1 -2) 33% The man Sally married won a scholarship. 1 - 2 64% 1 - 2 ( 2 - 1 ) 28% 2 - 1 2% 2 - 1 (1 -2) 6% John, whom Sally married, won a scholarship. 1 - 2 44% 1 - 2 (2 - 1) 34% 2 - 1 11% 2 - 1 (1 -2) 11%

The RRC situation is preferably interpreted as W-anterior, no matter whether it precedes the SUPC (66b: 92%) or not (66a: 86%). The first-mentioned NRRC is interpreted as W-anterior to the SUPC situation (78%). The same tendencies are clear from (67):

(67) a. A new mosquito deterrent was invented by the man who went on a holiday to Spain. 1 - 2 :3% 1 - 2 (2 - 1) : 16% 2 - 1 :48% 2 - 1 ( 1 - 2 ) : 33%

b. The man who went on a holiday to Spain invented a new mosquito deterrent. 1 - 2 :55% 1 - 2 ( 2 - 1 ) : 34% 2 - 1 :5% 2 - 1 ( 1 - 2 ) : 6%

c. John, who went on a holiday to Spain, invented a new mosquito deterrent. 1 - 2 :45%


1 - 2 (2 - 1) : 36% 2 - 1 : 11% 2 - 1 ( 1 - 2 ) : 8%

The difference between bounded telic NRRCs and bounded telic RRCs which precede the SUPC is that the former, unlike the latter, give the impression that the hearer is commenting on a sequence of events taking place before his eyes.

Dowty (1982) gives the following example in which the RRC precedes the SUPC:

(68)The woman that stole the book saw John. (Dowty 1982: 43)

(...) [68] not only allows the time of seeing to be the same as the time of stealing, but also allows the time of seeing to be later or even earlier than the time of stealing (...) Of course, we have a strong feeling that if the time of stealing was in fact earlier than the time of seeing, the appropriate sentence to utter is (48), not [68]: (48)The woman that had stolen the book saw John. But if [68] is indeed vague with respect to the ordering of the two events, while (48) unambiguously places the stealing earlier, then Grice's Maxim of Quantity tells us that (48) should always be used in preference to [68], if the speaker knows that the stealing was earlier. (Dowty 1982: 43)

Dowty's example differs from the examples in (66) and (67) because stole may also be interpreted as was stealing and could therefore be argued to express simultaneity with saw. However, a speaker conforming to the rules of communication will indeed prefer was stealing or had stolen to express simultaneity or anteriority. Dowty goes on to say that The woman that stole the book saw the man that robbed the bank "allows the stealing, seeing, and robbing all to be at different times, and one can continue to construct even more complicated examples" (Dowty 1982: 43). Robbed and stole can again be interpreted as was robbing and was stealing, which explains why these situations can be understood as W-simultaneous with saw (cf. Hintikka 1982: 11).

The conclusion to section 7.5.3. is that the order in which the clauses are reported affects the chronology only if the RC is a


before the SUPC situation (unless the semantics of the sentences impose a W-posterior order cf. section 7.5.2.4.), whereas NRRCs that follow the SUPC are interpreted as happening after the SUPC situation.

7.5.4. Question 5

It was pointed out in section 7.3. that most of the discourse-dis-cussions and RT-discussions concentrate on past tense examples. It is therefore interesting to examine whether the choice of temporal sector influences the temporal interpretation (question 5).

7.5.4.1. Future tense

Couper-Kuhlen (1989a) gives future tense examples to illustrate the rule of R-progression as "other tense in sequence (including Past Perfect, Future and Present) can be used equally well as the Past to represent a chain of events iconically:

"(...) You'll be disappointed at first. Then, without being able to say how or when it happened, you'll find you've forgotten your disappointment, and the first thing you know you'll be telling her your life's story, and all your troubles and hopes..." (Couper-Kuhlen 1989a: 11-12)

Declerck (1991a) also illustrates his claims with past sector as well as with post-present sector examples. The following post-present counterparts of examples (20) and (58) to (60) indeed seem to obey the principles observed in connection with past tense examples:

(69) a. John will marry the girl who will get pregnant, b. John will marry Mary, who will get pregnant.

(70) a. John will hit the man who will hit him. b. John will hit Bill, who will hit him.171

(71) a. Bill will be killed on the spot where John will have his accident.

b. Bill will be killed in Baker Street, where John will have his accident.

(72) a. John will fall in love with the girl who will win a beauty contest.172


1 - 2 : 14% 1 - 2 (2 - 1) : 16% 2 - 1 :39% 2 - 1 ( 1 - 2 ) : 31% Compare: John fell in love with the girl who won a beauty contest. (= example (58)) 1 - 2 :6% 1 - 2 ( 2 - 1 ) : 11% 2 - 1 :64% 2 - 1 ( 1 - 2 ) : 19%

b. John will fall in love with Mathilde, who will win a beauty contest.

The RRC situations are likely to be interpreted as W-anterior to the SUPC situation, whereas the NRRC situations will be interpreted as W-posterior to the SUPC situation. The preference for the W-anterior interpretation of (72a) (70%) corroborates this claim, although the preference for the W-anterior interpretation is stronger in the past sector counterpart (83%). Another difference with the past sector examples is that more people believe both chronological orders are possible (47% for the post-present example vs. 30% for the past sector example).

7.5.4.2. Past perfect

Opinions diverge as to whether R-progression/the movement of narrative time can be observed in a sequence of past perfect sen-tences. Adelaar - Lo Cascio (1986) claim that "for relative clauses the direction is not predictable. Cf:

(49) a. Paolo aveva visto il ragazzo che gli aveva dato il libro che lui aveva regalato a sue moglie.

b. Paul had seen the boy who had given him the book which he had given as gift to his wife." (Adelaar -Lo Cascio 1986: 290)

which can be interpreted in 3 ways: Cf. :


El

R —ι E2 ι R — E3 E3 ι—. R

El —

E 2 — R

El - S

E2 I R

R ___ E3

Couper-Kuhlen (1987), on the other hand, points out that the kind of temporal organization usually illustrated with past tense examples also occurs in past perfect sequences (cf. also Binnick 1991: 415, Couper-Kuhlen 1989a: 12, Defroment 1973: 51, Dowty 1986: 57-58):

(73) a. While we were talking about it, a pair of plain-clothes men brought in the red-faced bird who had stopped the slug I had missed Whisper with. It had broken a rib for him, and he had taken a back-door sneak while the rest of us were busy. Noonan's men had picked him up in a doctor's office. (Couper-Kuhlen 1987: 24)

b. After he had left his apartment and had made sure no one was following him, he had hurried to the bank, had withdrawn all his funds, and, before anyone had noticed, had skipped town. (McCoard 1978:185)

The above sequences of unembedded clauses indeed show that the PUTI (for unembedded sentences) also applies to sequences of past perfect sentences. When we convert the tenses of the SUPC - RC sequences in (69) to (72) into past perfect tenses, both the NRRCs and the RRCs are interpreted as bound directly, i.e. there is temporal subordination: all the RC situations are interpreted as anterior to the


main clause situations, irrespective of whether the RC is restrictive or non-restrictive:

(69) c. John had married the girl who had got pregnant, d. John had married Mary, who had got pregnant.

(70) c. John had hit the man who had hit him. d. John had hit Bill, who had hit him.

(71) c. Bill had been killed on the spot where John had had his accident.

d. Bill had been killed in Baker Street, where John had had his accident.

(72) c. John had fallen in love with the girl who had won a beauty contest.

d. John had fallen in love with Mathilde, who had won a beauty contest.

It is only if the lexical content of the NRRC implies that its situation takes place after the SUPC situation that the NRRC past perfect will be interpreted as bound indirectly and as referring to a situation that is W-posterior to the SUPC situation:

(74) a. He had been brought up in an orphanage, from which he had run away.

b. They had taken him to an orphanage, where he had stayed for five years.

c. He had lived with Mary, whom he had left after five years.

7.5.4.3. Present perfect

Haegeman (1989) points out that the present perfect differs from the past tense as regards the temporal interpretation triggered by the relevant forms (sequence vs absence of sequence):173

(75) a. He didn't finish his Ph.D. and he lost his job. (Haegeman 1989: 310)

b. He hasn't finished his Ph.D. and he has lost his job. (ibid.)

"[75a] strongly suggests that the one event causes the other, [75b] could be taken as a mere list" (Haegeman 1989: 310).174 Declerck


argues that this observation follows from the nature of the present perfect: "It is typical of the present perfect that it is interpreted as relating the situation to t0, not to any other TO. (...) In such sequences the different clauses just express that there has been an occasion before t0 when this particular situation held. The precise temporal location of these occasions is treated as irrelevant, and the hearer does not, therefore, assume the situations to be temporally related in a particular way" (Declerck 1991a: 131-132). Declerck probably assumes that this observation also applies to PUTI for embedded clauses, but he does not give any present perfect examples.

A conversion of examples (69) to (72) to the pre-present sector results in the following sequences:

(69) e. John has married the girl who has got pregnant, f. John has married Mary, who has got pregnant.

1 - 2 :20% 1 - 2 (2 - 1) : 53% 2 - 1 :9% 2 - 1 ( 1 - 2 ) : 18%

(70) e. John has hit the man who has hit him. f. John has hit Bill, who has hit him.

1 - 2 :20% 1 - 2 ( 2 - 1 ) : 22% 2 - 1 :36% 2 - 1 ( 1 - 2 ) : 22%

(71) e. Bill has been killed on the spot where John has had his accident.

f. Bill has been killed in Baker Street, where John has had his accident. 1 - 2 :31% 1 - 2 (2 - 1) : 16% 2 - 1 :42% 2 - 1 ( 1 - 2 ) : 11% Compare: Bill was killed in Baker Street, where John had his accident. 1 - 2 :9% 1 - 2 ( 2 - 1 ) : 11% 2 - 1 :46% 2 - 1 ( 1 - 2 ) : 34%

(72) e. John has fallen in love with the girl who has won a beauty contest.


f. John has fallen in love with Mathilde, who has won a beauty contest. 1 - 2 :23% 1 - 2 (2 - 1) : 33% 2 - 1 : 17% 2 - 1 ( 1 - 2 ) : 27% Compare: John fell in love with Mathilde, who won a beauty contest. 1 - 2 :25% 1 - 2 ( 2 - 1 ) : 39% 2 - 1 : 16% 2 - 1 ( 1 - 2 ) : 20%

The present perfect definitely interferes with the usual patterns. Unlike in the related post-present example given below (86%), there is no statistically relevant preference for the W-posterior reading (42%) in (70f):

(76) He will hit John, whom Sue will punch in the stomach. 1 - 2 1 - 2 ( 2 2 - 1 2 - 1 (1 - 2 )

1) 64% 22% 8% 6%

The present perfect NRRCs in (71) and (72) are chronologically vaguer than their past time counterparts: there is no strong preference for either a W-anterior or a W-posterior unmarked interpretation:

Present perf

47% (71 ρ (NRRC) option 1-2 and option 1-2 (2-1) option 2-1 and 53% option 2-1 (1-2) (72f) (NRRC) option 1-2 and 56% option 1-2 (2-1) option 2-1 and 44% option 2-1 (1-2)

Past tense

20%

80%

64%

36%


The notional relation between the NRRC situation and the SUPC situation in example (71)) (i.e. Bill got killed on a dangerous spot, i.e. the one where John had his accident) is less effective in imposing a chronological order and cancelling the "unmarked" W-posterior reading than in its past sector counterpart. The differences between the past sector version and the present perfect tense version of (72) are less outspoken.

The post-present and the past sector do not show major differences as far as the interpretation of shift of domain sequences is concerned. In sequences consisting of a past perfect SUPC and an indirectly bound NRRC the RC situation is interpreted as W-posterior provided this order is implied in the contents of the sentence. The chronological order between present perfect NRRCs and their SUPCs appears to be vaguer than that in their past sector counterparts.


This chapter has revealed the following points: 1. On a theoretical level, some arguments have been listed which

question the validity of the distinction between so-called absolute and relative (unbounded) simple past forms in W-simultaneous RRCs and NRRCs, respectively. All the italicized simple pasts may be argued to be relative tenses expressing simultaneity with the preceding situation:

(77) a. I met John. He was happy. b. I met John, who was happy. c. I met the man who was happy.

In any case, the predictions made by this approach are the same as those following from the (b)- and (c)-principles of the PUTI (i.e. the second situation is W-simultaneous with the first situation). One may therefore wonder whether the processing mind, guided by the princi-ple of attaining maximal effects for minimal effort, will not consider it more economical to interpret the verb form in the second clause in terms of temporal subordination rather than calling to life a new principle (the (b)-part and (c)-part of the PUTI) to establish the temporal relation between the situation of the first clause and that of the second.


2. A comparison has been made between the PUTI, the rule of R-progression and the discourse principle of foreground/background. It has been pointed out how these three principles differ and in which respects they overlap.

3. As far as the applicability of the PUTI to bounded RCs is con-cerned, it has been shown that the order in which clauses are reported, the sector in which they are located and the difference RRC/NRRC affects the interpretation of shift of domain bounded RCs. The views expressed are to a considerable extent based on an elicitation test, in which 64 native speakers of English participated.

Chapter 8 Choice of binding TO in RCs

In chapters 3 to 5, it has been examined whether RRCs and NRRCs differ in the way in which they use tenses to express temporal relations. Chapter 7 dealt with examples in which the tenses do not provide information about the chronological order between SUPCs and (N)RRCs; it formulated an answer to the question whether the two types of RC differ in terms of the unmarked temporal relationship they establish in those cases. In this chapter, the focus is on the binding TO, more specifically on whether or not there are any differences between RRCs and NRRCs as far as the choice of binding TO is concerned.

8.1. TO for (N)RRC situations

In this section, it will be examined whether past perfect RRCs and NRRCs differ as regards the elements which can function as binding TO (T02).

8.1.1. Possible candidates for T02 function

The following is a survey of the possible options for which an example can be found in the corpus of RCs at my disposal.175

Considerable attention will be given to illustrating the notion of indirect binding, which will then be examined in detail in section 8.2.

8.1.1.1. Finite SUPC binds SC situation

Along with Adelaar - Lo Cascio (1986: 255), Hornstein (1981: 140), Lo Cascio - Rohrer (1986: 236) and Smith (1978: 62), Declerck (1991a: 142)176 points out that a SC situation is usually bound by its SUPC situation.177 This claim is borne out by the following examples:

294 Choice of binding TO in RCs

(1) a. Ί must have taken a chill,' Houston said. He didn 't know what had come over him, but he knew that it was not a chill. (SEU) (RRC)

b. The Govt, are prone to spring decisions on dele-gations: they announced the 70 m.p.h. limit to a delegation of chief constables who had come to the Home Office to discuss the breathalyser. (SEU) (RRC)

c. The [Name] authorities do not supply confidential reports on expatriate lecturers, but our [Word Word] has commented that Mr [Name] was a very useful and sensible man, whose wide experience in the field of English language teaching had stood him in a good stead at [Name]. (SEU) (NRRC)

Whenever a SC is bound by its SUPC, it is said to be bound directly (Declerck 1991a: 62).

8.1.1.2. Nonfinite SUPC binds SC situation

Not only finite clauses can function as binding TO for a RC situ-ation. In the following examples, a nonfinite clause performs that function:

(2) a. The second phase, which is the subject of this Report, began with the presentation of the Memorandum on the Reorganization of the Uni-versity and has been partly occupied in imple-menting the reforms on which general agreement had been reached. (SEU) (RRC)

b. On this night of the disintegration of United's defence Astle struck again in the seventy-fifth minute, heading in a pass from Lovett who had come on as a substitute for Rees. (SEU) (NRRC)

8.1.1.3. T0 2 = implicit

T02 may be implicit (cf. Declerck 1991a: 92-97), as in the following examples:

TO for (Ν)RRC situations 295

(3) a. If a local pressure group can achieve results, the interest may gather momentum. Individuals who had never considered taking part in public debate will be attracted to the success of certain ventures. (SEU) (RRC)

b. If a local pressure group can achieve results, the interest may gather momentum. Mr Smith, who had never considered taking part in public debate will be attracted to the success of certain ventures. (NRRC)

8.1.1.4. T02 = noun

Nouns indicating an event can be used as T02:178

(4) a. For the second time in two months Oxford City Police are investigating the disappearance of a young woman who had hired a horse from a local riding stables. (Zandvoort 1964: 79)179 (RRC)

b. The article deals with the exploitation of slaves that had been imported form Africa. (RRC)

c. Indeed, the discoveries of the 'sixties and 'seventies have made it difficult to account for event the simples modifications. The genome has turned out to be so intricate, so much is going on within it which had never been suspected, that if we succeed in grasping the principles at work, genetics will be lifted on to a new plane and all may become clear. (SEU) (RRC)

d. This is often done ironically: Moll Flanders, after her second theft which is of a necklace from a child returning from dancing school, follows her description of the theft with a sudden burst of indignation against the parents who had sent the child out wearing such an expensive piece of jew-ellery. (SEU) (RRC)

e. He will confess the murder of Susan Sharpe, whom he had met in the bar. (NRRC)

The nouns functioning as binding TO are usually nominalizations. The latter term is to be understood as nouns that are morphologically (cf. to disappear, to exploit, to discover, to murder) or semantically


(cf. to steal) related to a dynamic verb. Fraser's terms action nomi-nate (1970: 84) and substantive nominals (1970: 85) cover the former class of nouns.

8.1.1.5. T0 2 = time indicated by adverbial

Adverbials may indicate the time which functions as T02; the TE (i.e. the established time, "a time that is established by an adverbial or by the context" (Declerck 1991a: 251)) the adverbials refer to is used as binding TO:

(5) a. Lady Robins, who had recovered from her disease by the end of February, started making plans to visit her daughter in Paris in June of that year.

The following non-RC example illustrates the same principle:

(5) b. Now England stood in her "finest hour". With an army half of whose arms and equipment had had to be left on the beaches of France, she awaited invasion, and with a handful of fighter planes, Spit-fires and Hurricanes, she challenged - and by the end of September had broken - the power of Germany's Luftwaffe. (Churchill, The man of the century) (Bouscaren et al. 1982: 79)

Example (6) is only slightly different:

(6) Smolensk was not captured until 1667, by which time Russian colonization had reached the Pacific. (SEU) (NRRC)

In (6), it is either the noun referred in the RC pronoun (time) which functions as binding TO for the RC situation or the time referred to by the antecedent.

This does not mean, however, that whenever there is an adverbial in a sentence, it functions as TOz (cf. Declerck 1991a: 268):

(7) a. In recent years, Shell Oil has been increasing the level of its capital expenditure year by year, and in

TO for (N)RRC situations 297

1967, for example, its capital outlays were more than double what they had been in 1963. (SEU)

In (7a), in 1963 establishes the TE which includes the TOsjt of the RRC situation, but it does not function as binding TO for the RC situation; the SUPC situation functions as T02. The following non-RC example from Bertinetto (1986) (taken from Korzen - Vikner 1980: 110) shows that the adverbial may be used as binding TO even if there is a SUPC situation which might also perform that function:

(7) b. Martine etait convaincue que Jean-Michel avait termini le tableau le 14 juillet. (Bertinetto 1986: 65)

If the SC means that Jean-Michel had finished his painting by the 14th of July, this adverbial indicates the time which functions as binding TO. This example proves that it will be very difficult to set up a hierarchy in the elements functioning as T02, i.e. to predict the likelihood with which a certain constituent indicates T02.

8.1.1.6. T02 to be found in extra-linguistic context

In some examples, a situation not verbally expressed in the preceding context provides the binding TO:

(8) a. A deformed shoplifter caught by an Ealing store detective claimed he had received his disability in an Oxford Street bombing which killed his wife. In fact South African-born Indian George Benjamin, 49, who had been born without most of his fingers, had already used his deformity to defraud insurance companies of thousands of pounds. (SEU) (NRRC)

The T02 of had been born is probably the time when Benjamin was caught stealing. As will be pointed out below (cf. section 8.1.1.7.4.), it makes sense to defend the idea that the hearer is likely to look for a suitable T02 in the linguistic context which precedes or in the extra-linguistic context. It is only when these do not provide a satisfactory binding TO that the hearer will look for a T02 in the linguistic context which follows. From that point of view, it is not very likely that had used will act as binding TO for had been born. The RC should be ascribed to the narrator rather than to Benjamin himself. If


it were Benjamin who had said that he had been born without most of his fingers, it would be claimed which functioned as T02. (8b) is a similar example:

(8) b. Only she understood that her blindness was all part of a much greater plan, and that Tarquin's little fingers were the instruments of a fate like that which had overtaken Ruth, the torturer's daughter in Bologna, during the reign of King Hildebrand of Evil Memory. In the crazy jigsaw of her mind the last piece had been fitted with a deftness which only the Gods could have contrived, and she was triumphant in the knowledge of her superior, solitary perception. (SEU) (RRC)

In the crazy jigsaw ... is a comment by the narrator; it is not incor-porated in the domain created by understood. In my opinion, T02 is the extra-linguistic time when she had these thoughts.

Bertinetto (1986) also points out that an extra-linguistic situation may function as T02. He remarks that it is possible to start a conversation with the sentence:

Look! This is the town where I had attended primary school) (The speaker is pointing to a spot on a map). What I want to suggest here is, instead, that in such a case [the sentence] would not be an absolute beginning, but rather a rejoinder to a previous conversation which was already per-formed by the same individuals. Consider, to this effect, Coming back to what I said yesterday ... Look! This is the town where I had attended primary school. In this case the hearer can easily build in his or her mind the temporal map on which he or she will fix the R required by the compound tense. (Bertinetto 1986: 49)

Bertinetto also gives the following non-RC example:

(9) 'Who had built this house?' (The speaker is pointing to the photo of a building which does not exist anymore) (Bertinetto 1986: 59)

"This sentence would sound odd if the speaker and the hearer were not able to construe some relevant R, according to contextual


information at their disposal. The most natural solution would presumably be that in [9] R coincides with the time interval delimited by the coming into existence of the building and its subsequent destruction" (Bertinetto 1986: 60).

8.1.1.7. Indirect binding

In 8.1.1.3. up to 8.1.1.6., a constituent other than a tensed clause takes up the function of T02. In 8.1.1.1. and 8.1.1.2., a (non)finite SUPC situation functions as T02. However, it is important to point out that when a clause functions as T02 for the SC, the SC need not be bound directly, i.e. the SUPC need not always function as T02. If a SC is bound by a clause other than its SUPC, it is said to be bound indirectly (Declerck 1991a: 63). It is not specified in Declerck (1991a) when RCs may rather than must be bound by the matrix clause. This issue will be discussed in section 8.2., as soon as all the possible cases of indirect binding in RCs (that I have come across) have been listed. As pointed out before, the terminology used is that from Quirk et al (1985: 987-991). The SUPC is the clause that immediately dominates the SC. The matrix clause is the ultimate SUPC. As will be gathered from the quotations given, Declerck's terminology is different (i.e. he uses matrix in the sense of SUPC).

8.1.1.7.1. T0 2 = matrix clause situation

Some SCs are not bound by their SUPC but by their matrix clause: "if the situation-TO of the matrix [SUPC] is itself a bound TO, such subclauses may {or, sometimes, must) represent their situation-TO as bound, not by the situation-TO of their matrix [SUPC], but by the TO which also binds the situation-TO of the matrix [SUPC]" (Declerck 1991a: 142) (italics mine) (cf. Adelaar - Lo Cascio 1986: 255). The type of indirect binding in which two clauses are bound by the clause in which they are both embedded is called rebinding (Declerck 1991: 64). Declerck gives the following example:

(10) John knew that Mary had said something which had shockedBül (Declerck 1991a: 142)

In (10), it is very clearly the situation of the matrix clause which functions as T02 for the RC situation. If the RC situation were


temporally subordinated to its SUPC, the past tense would be used; as the RC situation follows the SUPC situation very closely in time, the RC situation is experienced as (sloppily) simultaneous (cf. Declerck 1991a: 41) with the SUPC situation. However, in this sentence, the speaker prefers to express the relationship of anteriority between the matrix clause situation and the RC situation rather than that of (sloppy) simultaneity between the SUPC situation and the RC situation.

In the following set of examples the situations referred to in the RC and its SUPC are bound by a clause dominating both:

(11) a. Then he remembered about Coulter : that he had no family at all, had been brought up in an orphanage, from which he had run away at the age of eleven, to join the circus. (SEU) (NRRC)

Coulter runs away from the orphanage in which he has been brought up. The relationship of W-posteriority (bringing up - running away) is not linguistically expressed. The speaker concentrates on the things which he remembers about Coulter, which are all anterior to the time of remembering. (1 lb) is a similar example:

(11) b. Emma was referring to the fact that she had received two letters from Louis Bates, in which he had confessed his mad passion and sought to drag her some way towards the altar. (SEU) (NRRC)

If the RC situation were bound directly, a form expressing simul-taneity would be used (confessed). Instead, the speaker has chosen to temporally subordinate the RC situation to the matrix clause situation.

Sometimes when the matrix clause situation functions as T02, the RC is conjoined to a SC which uses a relative tense. In that case the RC may be bound by the same T02 as the situation of the clause to which it is conjoined (cf. Adelaar - Lo Cascio 1986: 263-266, De-clerck 1991a: 143):

(11) c. He was a dedicated and highly successful teacher, who had always been willing to meet the challenge of new situations and methods, and who had always shown himself deeply committed to his work, (adapted from SEU)180 (NRRC)


(11) d. Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and hadn't expected to have a maid to run and fetch for her the whole time. (SEU) (NRRC)

Had shown (1 lc) and hadn't expected (1 Id) are likely to be bound by their matrix clauses.

8.1.1.7.2. SC and SUPC situation bound by implicit T0 2

Declerck (1991a: 143) points out that in some examples both the SUPC situation and the SC situation may be bound by an implicit TO:

(12) John had been in bed during the preceding week because he had been ill.

The example is given out of context; it is therefore not possible to find out whether T02 is implicit or given in the preceding context. If it is the beginning of a story, both clauses are indeed bound by an implicit TO.

The first sentence in the opening paragraph of The Adventure of the Sussex Vampire clearly demonstrates that T02 may be implicit:

(13) Holmes had carefully read a note which the last post had brought him. Then, with the dry chuckle which was his nearest approach to a laugh, he tossed it over to me. (The Complete Sherlock Holmes, p. 1033) (Reinhart 1984: 800) (RRC)

Had brought is either bound by the same implicit TO, in which case the RC is bound indirectly, or by the SUPC, in which case there is direct binding.

8.1.1.7.3. Clause mentioned before matrix clause/SUPC binds RC situation

The matrix clause situation and the RC situation may both be bound by the situation-TO of a clause preceding the matrix clause and the


RC (cf. e.g Declerck 1991a: 149, Rohrer 1986: 90, 95, Smith 1978: 65):

(14) a. Something protruded not wood but rope. The coffin had been wrapped around with two inch manilla rope which had become iron hard as the F guano had permeated it. Rhennin said that's the sort of thing Shelborne would do if he wanted to hide a body. (SEU) (RRC)

b. He was quite relaxed. He had left Hawton, where he had been so unhappy, and his future movements would be decided for him by other people in some other place. (SEU) (NRRC)

It is difficult to say whether had been is bound directly or indirectly in (14b). If had left functions as binding TO for had been, the RC is bound directly. If it is was, the RC is bound indirectly.

8.1.1.7.4. RC binds SUPC situation

Declerck (1991a: 148) points out that when a RC precedes its SUPC, the former situation may bind the latter:181

(15) a. A boy who arrived yesterday had been mugged in the underground the day before. (Declerck 1991a: 148)

b. A boy had been mugged who arrived yesterday. (Declerck 1991a: 148)

(16) a. The copy that was destroyed today had only been made a couple of hours earlier. (Declerck 1991a: 148)

b. A copy had been made a couple of hours earlier which was destroyed today. (Declerck 1991a: 148)

"In [15a] the situation-TO of had been mugged is bound by that of arrived. The same thing is not possible in [15b], where the only possible interpretation is that the mugging took place before some (unspecified) time other than the arrival. The past tense arrived then establishes a shift of domain to some other time interval in the past. The same difference in interpretation is also to be observed in [16a-


b]" (Declerck 1991: 148). Smith (1981) and Hüllen (1989) give similar examples:182

(17) a. A woman John met on Thursday was marrying in 3 days. (Smith 1981: 232)

b. A woman John talked to on Wednesday was leaving 3 days later. (Smith 1981: 234)

c. Many of the men who were made redundant last month had worked in that factory for more than 20 years. (Hüllen 1989: 610)

The following corpus examples also show that the RC situation may function as binding TO for the SUPC situation when the RC precedes the SUPC:

(18) a. The Namib desert was timeless but the ancient place to which the springbok came had changed. (SEU) (RRC)

b. Everything she knew or surmised about him, which had made him alluring and romantic afar off, were now the very things that made him formidable. (SEU)183 (NRRC)

c. The information he needed, however, had been assembled in 1964 by Robert W. Holley, who determined the nucleotide sequence in the corre-sponding RNA. (SEU) (RRC)

However, not all RCs preceding their SUPCs automatically bind the latter: in example (19a), the time when Benjamin was caught stealing (i.e. the extra-linguistic context) binds the RC situation and its SUPC situation (cf. section 8.1.1.6.):

(19) a. A deformed shoplifter caught by an Ealing store detective claimed he had received his disability in an Oxford Street bombing which killed his wife. In fact South African-born Indian George Benjamin, 49, who had been born without most of his fingers, had already used his deformity to defraud insurance companies of thousands of pounds. (SEU) (NRRC)


In (19b), the RC situation is bound by the situation referred to in the preceding sentence and there is a shift of domain to the pre-present sector in the SUPC:

(19) b. Under the rules of the game, every single piece of government legislation on its way through the House at the end of the last parliament also had to be dropped. As a result at least three major measures (dealing with agriculture, the land com-mission and building controls) on which a certain amount of useful progress had been made, have all equally been driven back to square one. (SEU)184

(RRC)

The examples just given relate to the importance of the order in which the clauses are reported: is the order of the clauses (SC + SUPC or SUPC + SC) more decisive in choosing a T02 than the syntactic status of the clause (SC vs SUPC)? It is not possible to formulate strict rules as to when it is the preceding RC that binds the SUPC situation rather than some other clause preceding both the RC and the SUPC. Even so, in my opinion, the order in which the clauses are reported is more important than the syntactic status of the clauses. That is, the fact that the RC precedes the SUPC is more important than the fact that the RC is a subordinate clause; a hearer will always look for a binding TO in the stretch of discourse that has already been given. If he does not find a T02 there, two options are possible: either an implicit TO or an extra-linguistic TO will function as T02 or the hearer will wait for the speaker to continue his utterance and will look for a TOL in the stretch of discourse to fol-low.185 Taking into account the linear processing of information, I would again be inclined to say that the speaker will opt for the first alternative. The strategy just described is comparable to the principle of proximal concord: just as the number of the verb is determined by the Ν which is closest to the verb rather than the head of the NP which functions as subject, a constituent which has already been mentioned rather than one which is higher in syntactic status is likely to function as binding TO.

It is very difficult to determine which element functions as T02 in a number of examples with the order SC + SUPC:

(20) a. Oh, I meant to tell you. After you'd got out, the Glamorous Blonde (who had ignored my existence


all through Swanwick) said "Did I hear you'd taken Icelandic at the University ? " - - all very friendly and gushing. (SEU) (NRRC)

The situation when the relevant people were together might be argued to function as binding TO for the RC situation. The general situation has been introduced, the speaker is familiar with the time, place and persons involved. Therefore, it seems reasonable to say that he will use this knowledge to his advantage and use it as T02 for the RC situation rather than wait for the speaker to go on before he chooses a T02, which in that case would be said. There are also two alternative approaches to (21b):

(20) b. When challenged by the store detective, Benjamin, who had often previously claimed to be a doctor, told the staff; "I wouldn't steal; I am disabled." (SEU) (NRRC)

The most adequate (i.e. requiring the least processing effort) option is probably for the hearer to use the adverbial clause as binding TO. However, others might claim that told functions as T02.186

It is difficult to solve the problem alluded to in the examples under discussion. One can never be completely sure as to which option will be considered as the most economical one by the processing mind. Even so, the following conclusion seems warranted: the clause which is first mentioned and the SUPC mostly coincide, i.e. the SUPC usually precedes the SC. In other words, the linear parameter and the syntactic status parameter are often fused into one. When the SC precedes the SUPC, the former will either bind the latter or they will both be bound by a clause that precedes. This boils down to saying that linear order is more important than syntactic status for the choice of binding TO.

In fact, many examples exhibit indeterminacy as far as the choice of binding TO is concerned: in (21a), it is not clear whether it is the SUPC (direct binding) or its matrix clause (indirect binding) that binds the RC situation:

(21) a. He did not care to tell of how he had pored over every drawing, spun out every hour, like a smoker, committed to abstaining, with his last packet of cigarettes. He would not reveal with what reluctance


he had returned to the task which he had promised so earnestly to perform. (SEU)

Had promised may be bound by its SUPC situation (had returned) or by the matrix clause situation (would not reveal). It is equally difficult to determine which element functions as binding TO in (20b):

(21) b. The ideas for discussion by the city council are detailed in a document, part of the Lancaster Local Plan, entitled "The Special Problems of Traffic." A council spokesman said it was a report on the progress of the 10-year plan which the authority had a statutory duty to prepare and which had been impeded in one fundamental area - the problems of traffic and transport. (SEU)

It is not possible to prove whether had and had been impeded are both subordinated to was (direct binding) or whether had been is temporally subordinated to had. As regards the latter possibility, it remains to be found out whether the situation of the first conjoin of a clause can function as binding TO for the situation of the second conjoin, and if it can, whether this is to be considered as a special case of direct binding. In the following example, had put may be temporally subordinated to had been wearing or to realized:

(21) c. Very funny last night - couldn't get to sleep for a bit & was of course thinking bout you - thought Beaulieu & my stone - and suddenly realized that I had just been wearing the jacket in the pocket of which I had put the stone and didn't remember it being there when I took off jacket and removed things from pockets. (SEU)

In (21d), it is either wondering, had been or had turned on which is taken as binding TO for had washed:

(21) d. He had turned on the taps in the bathroom and splashed drinking water again and again over his face, let it slide coolly down his wrists, wondering at it. For so long, there had been only the tins of green water, stinking of chlorine, to drink, and the grey


scum in which someone else had washed before him and the foul water at the bottom of the shell holes, before the sun of June and July had come to dry them out. (SEU)

The above survey has certainly revealed that it is possible for a wide range of elements to serve as binding time of orientation. From this it follows that it is hardly possible to formulate hard and fast rules about the mechanisms underlying the choice of binding TO.187

As soon as one takes a look at a stretch of discourse (cf. Schöpf 1989), one realizes that endeavours to do so (cf. the linguists who work within the Partee/Kamp tradition of automatic shift of reference point) are bound to fail. I therefore subscribe to Schöpf s view:

We do not believe that any kind of automatic progression of the reference time triggered by event sentences represents a realistic picture of the way narrative texts are processed by the listener or reader. Instead, we propose that the processing of narrative texts should be looked upon as a calculation procedure in which every new sentence or clause is evaluated against the background of (all the) information supplied by the language system in question and our knowledge of extra-linguistic reality. One of the basic rules seems to be the instruction to identify the reference time of a newly introduced sentence with the reference time via which the preceding sentence was tensed, unless this is prevented by semantic or pragmatic incompatibility of the two events reported in the two sentences. (Schöpf 1987: 217)

As will be pointed out below, the choice to bind a RC to the matrix clause rather than to the SUPC is, apart from some cases (in which one of the options changes the chronological order between the situations cf. section 8.2.2.1.), a subjective one: it is governed by the meaning one wants to convey, e.g. is the W-simultaneity relation between the SUPC and the RC situation more/less important than the relation of W-anteriority between the RC situation and the matrix clause situation? The principles determining (in)direct binding are in many respects similar to those determining temporal subordina-tion/shift of domain in the past sector. The alternatives of the former set differ from those of the latter set in that the direct/indirect binding options, unlike the temporal subordination/shift of domain


alternatives, both imply there is temporal subordination. However, the choice between direct and indirect binding, like the choice be-tween temporal subordination and shift of domain (e.g. to refer to a past sector W-anterior situation) boils down to a choice between a past tense and a past perfect,188 the difference being that in the case of direct binding the past tense is relative whereas in the case of shift of domain, the past tense is an absolute tense.

8.1.2. Choice of binding TO in RCs: Declerck (1991a)

In his discussion of binding rules for SCs, Declerck makes the following remark relating to RCs: "A relative clause that is embedded into a clause with a relative tense form (and belongs to the same domain as the latter) is normally bound by the TO that also binds the latter" (Declerck 1991a: 149):

(22) a. John, who had been ill, had been in bed during the preceding week. (Declerck 1991a: 149)

b. The boy that had been in hospital for three weeks had had a car accident. (Declerck 1991a: 150)

c. The man had lived in the town where Monroe had been born, (ibid.)

Declerck claims (1991a: 150) that (22b) is the clearest proof that the RC situation is not bound by the SUPC situation as the RC situation is interpreted as "holding later" than the SUPC situation. The order in which the clauses are given (SC before SUPC) no doubt plays a role in this respect. As was pointed out above, a RC situation is less likely to be bound by its SUPC situation if the latter follows the former. The importance of the order is also clear from (22c): the unmarked interpretation of the RC is probably that it is anterior to the SUPC situation and not to some other TO to which the SUPC is subordi-nated. Example (22a) most clearly shows that the SC and the SUPC are bound by the same TO: the SC situation and the SUPC situation are W-simultaneous, which, if there was temporal subordination, would require the use of the past tense in the SUPC. In fact, only (22a) clearly illustrates Declerck's claim that the SUPC situation and the RC situation are both bound by the same TO when the RC is embedded in a SUPC with a relative tense. In (22b) and (22c), the clause first mentioned might be argued to function as binding TO for the W-anterior situation that is mentioned secondly.189 A similar


point cannot be made in connection with (22a): the RC situation cannot bind the SUPC situation because this would result in the use of a different tense, i.e. was.

Declerck goes on to say that in indirect speech examples the RC embedded into a clause with a relative tense form need not be bound by the clause that binds the SUPC: "the relative clause can be interpreted as bound by its matrix [SUPC in my terminology]" (Declerck 1991a: 150):

(23) a. Tom said that he had visited the town that had been destroyed by the earthquake,

b. Tom said that he had spoken to the man who would be sent to Mars.

Depending on whether the RC information belongs to Tom's original utterance or not, it is bound by the SUPC/matrix clause respectively. However, if the information does not belong to Tom's original utterance, has been destroyed and will be sent can also be used in (23a) and (23b). The following intensional domain RC example also illustrates the principle formulated by Declerck:

(24) Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and hadn't expected to have a maid to run and fetch for her the whole time. (SEU)

Had been much cheaper may be bound by the SUPC clause (had got rid) or by the matrix clause (began to regret). Declerck would probably argue that had got rid functions as TO for had been cheaper (provided we can extend Declerck's observation for indirect speech to other intensional domain examples). However, if we fol-low the principle formulated by Declerck on page 149 (cf. supra), began to regret functions as binding TO for had been much cheaper}90 Indeed, the predictions made by the principle referred to on page 150 differ from those following from a principle formulated earlier on: "A relative clause that is embedded into a clause with a relative tense (and belongs to the same domain as the latter) is normally bound by the TO that also binds the latter" (Declerck 1991a: 149). Whereas the latter principle predicts that in a sequence matrix clause + SUPC + SC, the matrix clause will function as binding TO, the principle Declerck (1991a) formulates on page 150 predicts that it is the SUPC situation which functions as binding TO.


8.1.3. General conclusion

In section 8.1., a survey has been given of elements that can function as T02 for a RC situation. That it is possible for nouns to function as binding TO is an observation which has escaped other linguists' attention. The survey shows (a) that a constituent which precedes the RC is more likely to bind the RC situation than one which follows the RC and (b) that in some examples, more than one element can perform the function of T02. In any case, it is clear that any attempt to formalize the choice of T02 must be bound to fail. The above survey has also revealed that in the RC examples at my disposal, no differences can be detected between RRCs and NRRCs as regards the choice of binding TO: the speaker is guided by the principle of minimal effort when looking for a binding TO for a past perfect RRC/NRRC. Although it is not possible to give comprehensive rules determining the choice of binding TO, there are some clear tenden-cies governing the choice direct/indirect binding in a particular subset of examples. These will be dealt with in the next section.

8.2. (In)direct binding in (N)RRCs

This section addresses the final topic which should make it possible to answer the question whether the tense system in RRCs differs from that in NRRCs: are the constraints on (in)direct binding different for RRCs and NRRCs? First, a summary will be given of Declerck's ideas on this issue (section 8.2.1). In section 8.2.2.1., it will become clear that the tenses as such are not the only criterion on which the classification of a sentence as bound indirectly or bound directly should be based. In section 8.2.2.2., the need for a constraint on direct binding in NRRCs in some cases will be explained.

8.2.1. (In)direct binding in (N)RRCs: Declerck (1991a)

When pointing out that conjoined SCs can both be bound by the situation of the matrix clause in which they are embedded, Declerck observes that when the conjoined SCs are NRRCs, "both of them are normally bound by the TO of the head clause. (...) It is only when we add an adverbial representing the situations of the conjuncts as simultaneous with each other that the past tense can be substituted for the past perfect" (Declerck 1991a: 144). NRRCs are listed among

Indirect binding in (N)RRCs 311

the exceptions to the general principle that direct binding is the un-marked binding, possibility in SCs (Declerck 1991a: 142-150):

(25) This piece of information was given by John, who had had a conversation with the manager and had heard a couple of interesting things. (Declerck 1991a: 144)

The past tense cannot be used unless an adverbial representing the situations of the two conjoins (Quirk et al. 1985: 46) as W-simultaneous with each other is added:

(26) This piece of information was given by John, who had had a conversation with the manager and (had) heard a couple of interesting things in the course of that conversation. (Declerck 1991a: 144)

This observation seems to suggest that NRRCs behave differently from RRCs as regards the choice of binding TO. This is in keeping with the examples Declerck gives (1991a: 63) when explaining the concept of direct/indirect binding:

(27) I said I would go to John, who would be back by then. (= indirect binding) (* was back) (Declerck 1991a: 63)

(28) She had been to Brussels for a flying visit to a nephew who was / had been staying at the Hilton. (= direct or indirect binding) (ibid.)

(29) What we hoped was that we would receive aid from John, who would have been informed by then. (= indirect binding) (* had been informed) (ibid.)

"Whereas the time clauses (...) must be bound directly (...), the nonrestrictive relative clauses in [27] and [29] must be bound indirectly, while the restrictive relative clause in [28] allows the two possibilities" (Declerck 1991a: 63).

However, when we convert examples (25) to (29) into their (N)RRC counterparts, the same acceptability judgements apply:

(25) a. This piece of information was given by the man who had had a conversation with the manager and had heard / ?? heard a couple of interesting things.

(26) a. This piece of information was given by the man who had had a conversation with the manager and (had)


heard a couple of interesting things in the course of that conversation.

The above examples remind us of those discussed in the investi-gation of the use of tense in pre-present sector RCs. It was pointed out that in some sentences the past tense is not likely to be used because it needs an anchoring time which is linguistically realized by means of an adverb (cf. chapter 4, section 4.2.2.2.).191 The RCs in (25) and (25a) illustrate the same tendency: the past tense cannot be used unless a time adverbial is added to the sentence (cf. (26) and (26a)). Moreover, the relevant clauses are coordinated. As will be pointed out below (cf. 8.2.2.2.3.), if the first clause uses a past perfect, this leads to a preference for the use of a past perfect tense in the second clause as well. In (27) and (29), the RC situation is W-posterior to, and the SUPC situation is posterior to, the matrix clause situation. In (27), the RC situation is W-simultaneous with, and in (29) W-anterior to, the SUPC situation:

(27a) I said I would go to the man who would be back by then. (* was)

(29a) What we hoped was that we would receive aid from the man who would have been informed by then. (* had been informed)

It makes sense to represent the RC situation as a prediction rather than as a fact (i.e. to use a conditional tense in the RC (cf. chapter 3, section 3.4.3.), because it is not completely certain at the time of speaking that the man will be back (27a)/will have been informed (29a). It is the presence of the adverbial by then which blocks the use of the past tense (27a) or past perfect (29a) (cf. chapter 5, section 5.3.2.4.1.). The fact that the sentences are embedded in a clause introduced by a verb of saying/belief may also contribute to a preference for the conditional tense. Like the RRC in (28), its non-restrictive counterpart also allows the use of the past perfect:

(28a) She had been to Brussels for a flying visit to John, who was staying / had been staying at the Hilton.

The progressive past tense and progressive past perfect give the same temporal information. Was staying can be interpreted in two ways: either as a relative tense bound by the SUPC situation or as an absolute tense establishing a W-simultaneous domain. When the past

Indirect binding in (NjRRCs 313

perfect is used, any misunderstanding of the kind that his stay is anterior to her visit would seem implausible, because it would make no sense for her to go to Brussels if John (28a)/her nephew (28) is no longer there. The use of the progressive past perfect simply stresses the fact that he had already been staying there for a while.

The conclusion to be drawn so far is that the examples given in Declerck (1991a) do not prove that RRCs and NRRCs behave differently as far as (in)direct binding is concerned. Indeed, in the corpus of RCs at my disposal, there is often similarity between RRCs and NRRCs as regards the choice between direct and indirect binding. In the following examples for instance, there is no diffe-rence in acceptability or meaning of verb forms between the two types of clauses:

(30) a. A deformed shoplifter caught by an Ealing store detective claimed he had received his disability in an Oxford Street bombing which killed (had killed) his wife. (SEU) (RRC)

b. A deformed shoplifter caught by an Ealing store detective claimed he had received his disability in an Oxford Street bombing, which killed (had killed) his wife. (NRRC)

(31) a. Then he remembered about Coulter : that he had no family at all, had been brought up in an orphanage where he was (had been) maltreated, (adapted from SEU) (RRC)

b. Then he remembered about Coulter : that he had no family at all, had been brought up in an orphanage, where he was (had been) maltreated. (NRRC)

(32) a. The vetoed bill had been sponsored by Sen. Pat McCarran, the renegade Nevada Democrat with whom Truman had fought (* fought) since his Senate days. (WSJ) (RRC)

b. The vetoed bill had been sponsored by Sen. Pat McCarran, with whom Truman had fought (* fought) since his Senate days. (NRRC)

(33) a. Lady Foxglove began to regret that she had ever got rid of the nanny who had been (was) much cheaper and hadn't expected (didn't expect) to have a maid to run and fetch for her the whole time. (RRC)

b. Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been (was) much


cheaper and hadn't expected {didn't expect) to have a maid to run and fetch for her the whole time. (SEU) (NRRC)

However, occasional differences can be observed:

(34) a. He said he would contact someone who would be (was) willing to help him. (RRC)

b. He said he would contact John, who would be (was) willing to help him. (NRRC)

Was willing in the NRRC (34b), unlike was willing in the RRC (34a), cannot be interpreted as bound by its SUPC. A similar comment applies to (35):

(35) a. He was angry with the colleague who would betray the boy who had played truant,

b. He was angry with the colleague who would betray Jonathan Steel, who had played truant.

Had played in (35b) can only be interpreted as bound by the matrix clause situation. In (35a), had played allows two interpretations: anteriority with respect to the matrix clause situation or the SUPC situation.

In what follows, an analysis will be given of indirect binding past sector examples in which the RC is bound, not by the situation of its SUPC (SUPC1) but by the situation of a clause (C3) which precedes the RC and its SUPC (no matter whether they are embedded in it or not). In some examples, the RC precedes the SUPC situation and both situations are bound by a clause that precedes the RC: clause A - RC - SUPC. In examples of this type, I will concentrate on why the SUPC situation is bound by the situation of the A clause rather than by the RC situation. The examples are all clauses in which the tenses themselves prove that there is indirect binding. This implies that sentences consisting of a matrix clause (C3), a SUPC (SUPC1) and a RC (RC) between whose situations all the following temporal relations hold will not be examined:

RC situation is W-anterior to SUPC1 situation RC situation is W-anterior to C3 situation SUPC1 situation is W-anterior to C3 situation


= RC-situation W-anterior to SUPC'situation W-anterior to C3 situation

Examples of this type will not be discussed for the following reason: if the RC situation is temporally subordinated to the SUPC1 situation, the past perfect will be used (direct binding). If the RC situation is temporally subordinated to the C3 situation, the past perfect will also be used (indirect binding). In other words, the tenses do not provide any indication as to whether it is the SUPC1 situation or the C3

situation that binds the RC situation (cf. chapter 2, section 2.1.8., example (44), cf. Adelaar - Lo Cascio 1986: 261-262). The following example is a case in point:

(36) Paul had seen the boy who had given him the book which he had given as [sic] gift to his wife. (Adelaar - Lo Cascio 1986: 290)

If the RC situation is W-anterior to the SUPC situation and to the matrix clause situation, both clauses may function as binding TO for the RC situation. The types of cases to be examined are the following:

a. RC situation is W-simultaneous with SUPC1 situation RC situation is W-anterior to C3 situation SUPC1 situation is W-anterior to C3 situation

(37) It was well-known that she had lived with a man who had collaborated (collaborated) with the Germans.

ο


Why does the speaker choose to temporally subordinate the RC situation to the C3 situation (i.e. use the past perfect) rather than to the SUPC1 situation (i.e. use the past tense) or vice versa?

b. RC situation is W-simultaneous with SUPC1 situation RC situation is W-posterior to C3 situation SUPC1 situation is W-posterior to C3 situation

(38) On Monday John said that on Wednesday he would try and intercept that letter that would arrive (arrived) that day.


Why does the speaker choose to temporally subordinate the RC situation to the C3 situation (i.e. use the conditional tense) rather than to the SUPC1 situation (i.e. use the past tense) or vice versa?

c. RC situation is W-posterior to SUPC1 situation RC situation is W-anterior to C3 situation SUPC1 situation is W-anterior to C3 situation

(39) It was well-known that John had said something which would make (had made) her leave her husband a few years later.

Why does the speaker choose to temporally subordinate the RC situation to the C3 situation (i.e. use the past perfect) rather than to the SUPC1 situation (i.e. use the conditional tense) or vice versa?

d. RC situation is W-anterior to SUPC1 situation RC situation is W-posterior to C3 situation SUPC1 situation is W-posterior to C3 situation

(40) John announced that he would be helped on Tuesday by the man who would kill (had killed) the inspector on Monday.


Why does the speaker choose to temporally subordinate the RC situation to the C3 situation (i.e. use the conditional) rather than to the SUPC1 situation (i.e. use the past perfect tense) or vice versa?

8.2.2. RRCs vs NRRCs: different systems?

8.2.2.1. (In)direct binding in RRCs

A first point to be made is that in a sentence of the type "matrix -SUPC - RC" in which either the past tense or the past perfect can be used in the RC, one should not automatically classify the past tense as the directly bound (DB) form and the past perfect as the indirectly bound (IB) form:

(41) a. John met the man who had located the source of gas that caused the explosions,

b. John met the man who had located the source of gas that had caused the explosions.

(42) a. She was happy because her husband had invited a colleague she had a crush on.


b. She was happy because her husband had invited a colleague she had had a crush on.

(43) a. She was angry with her husband because he had invited a colleague who was their daughter's boyfriend.

b. She was angry with her husband because he had invited a colleague who had been their daughter's boyfriend.

(44) a. John met a man who had lived with people who collaborated with the Germans,

b. John met a man who had lived with people who had collaborated with the Germans.

In each of these examples, the effect of replacing the past tense by a past perfect is that the RC situation is represented as bounded (past perfect) rather than unbounded (past tense). In some cases, this change affects the temporal information; the unmarked interpretation is that the past perfect situation refers to one that is over at the time of the SUPC clause situation, whereas the past tense clause does not explicitly communicate this. However, this observation does not imply that both the past perfect and the past tense situations are temporally subordinated to the matrix clause situation. The temporal relations in (41) to (44) may be described as follows:

I. with past tense in RC: RC simultaneous with SUPC (DB) or RC simultaneous with matrix (IB) with past perfect in RC: RC anterior to matrix (IB) or RC anterior to SUPC (DB)

The unmarked interpretation of (41a) is that the explosions are represented as completely over at the time of the meeting (RC situation simul SUPC situation); there is pragmatic information indicating that the RC situation is W-simultaneous with the SUPC situation, i.e. the man in question is likely to look for the source of gas at the time when the explosions are still happening. On this reading, there is direct binding. Accordingly, the past perfect RC will be understood as bound by the matrix, i.e. the RC situation referred to by means of the past perfect will be interpreted as expressing anteriority with respect to the matrix clause situation (indirect binding). However, the past tense could also be bound indirectly, in which case the explosions are still going on at the time of the meeting. The past perfect also allows an interpretation in terms of


direct binding; on this reading, the explosions are represented as anterior to the matrix clause situation.

II. with past tense in RC: RC simultaneous with matrix (IB) with past perfect in RC: RC anterior to SUPC (DB)

In (42) and (43), on the other hand, the past tense RC situation will normally be interpreted as W-simultaneous with the matrix clause situation; it is pragmatically more likely that the happiness (anger) is simultaneous with (occurs as a result of) the fact that she has a crush on this colleague (he is their daughter's boyfriend) than that the RC situation is simultaneous with the SUPC situation. In other words, the past tense RC is temporally subordinated to the matrix clause situation for reasons of pragmatic plausibility (indirect binding), whereas the past perfect RC is interpreted as anterior to the SUPC situation (direct binding).

III. with past tense in RC: RC simultaneous with SUPC (DB) with past perfect in RC: RC anterior to SUPC (DB) or RC anterior to matrix (IB)

In (44a), the past tense RC situation (collaborated) will be inter-preted as simultaneous with the SUPC situation. The people concerned collaborated with the Germans at the time when the man lived with them rather than at the time John met the man. The past perfect situation may be interpreted as bound directly, i.e. anterior to the SUPC situation. If this interpretation clashes with the state of the world, the listener will alter his interpretation; he will choose the T02 of the SUPC situation (which is TOsit of the matrix clause situation) as binding TO, to which the RC situation is anterior. In that case, there is indirect binding and his knowledge of the world will have to guide him in revealing that the RC situation is W-simultaneous with the SUPC situation. The unmarked interpretation of the past tense and the past perfect is temporal subordination to the SUPC (direct binding): the past RC is interpreted as simultaneous with the SUPC situation, the past perfect RC as anterior to the SUPC situation.

The following survey summarizes the points made:

(41) (42), (43)

past tense

DB (IB) IB

past perfect

IB (DB) DB


(44) DB DB (IB)

The above examples indicate that pragmatic and contextual factors to a great extent determine how the tenses are interpreted. The tense forms in themselves (past tense vs past perfect tense) do not allow any straightforward classification into a past tense that is always the directly bound form and a past perfect that is always the indirectly bound form in pairs of the type under discussion. Put differently, the examples indicate that verb forms in themselves cannot be used as a basis for classifying sentences in terms of direct and indirect binding. The tenses themselves are neutral on this score. The observations in 8.2.2.1. show that it is sometimes the SUPC situation, sometimes the matrix clause situation that functions as binding TO; in still other cases either alternative is possible. Some alternatives may be ruled out (i.e. the directly and the indirectly bound forms are not always mutually substitutable) because the way in which they locate situations in time does not correspond with the situations' actual location in time.

In the next section, I will demonstrate that the decision to use a directly bound rather than an indirectly bound verb in a RC has in some cases more to do with the effects brought about by the relevant verb forms than with the syntactic difference between relatively independent NRRCs and relatively dependent RRCs (cf. sections 8.2.2.2.1. to 8.2.2.2.5.). There are, however, a number of cases in which the choice to bind the RC situation directly rather than indirectly is not free; as will be shown in sections 8.2.2.2.1. and 8.2.2.2.6., RRCs and NRRCs sometimes differ in the binding possibilities they allow.

8.2.2.2. Factors determining the choice between direct and indi-rect binding

8.2.2.2.1. Indirectly bound past perfect: situation is over vs. directly bound past tense: situation may be over (there are no adverbials in the RCs)

In the following RCs, the indefinite past perfect situation both assigns a characteristic to the antecedent and explicitly indicates that the situation no longer holds at T02. The past tense performs the former function and implicates the latter element of meaning. This difference may lead to a preference for one of the two forms:


(45) a. We met the girl who had lived with a man who had grown marihuana. (RRC)

b. We met the girl who had lived with a man who grew marihuana.

It must be pointed out that in the NRRCs given below, the past tense may be said to be an absolute past. Therefore, the choice may be one not between direct binding (relative past tense) and indirect binding (past perfect) but between shifting the domain (absolute past tense) and indirect binding (past perfect). No matter what the status of the past tense is, there is a difference in effect between the past perfect and the past which is similar to that in (45a) and (45b):

(46) a. There had been a further definition, Μ sb.2.sb, but this had depended on an assumed broad difference in character between the smaller, retail-type, time deposits in clearing banks, which were in practice transferable to current accounts on demand, and larger, money-market, wholesale time deposits in other banks. (SEU) (NRRC)

b. There had been a further definition, Μ sb.2.sb, but this had depended on an assumed broad difference in character between the smaller, retail-type, time deposits in clearing banks, which had been in practice transferable to current accounts on demand, and larger, money-market, wholesale time deposits in other banks

In the original example, the past tense is used. This implies that the speaker prefers to attribute a characteristic to the smaller clearing banks in general terms, i.e. without explicitly indicating that the situation no longer holds. The explicit location in time by means of the past perfect is not important for the characterization. (47) is another NRRC example:

(47) a. Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and hadn't expected to have a maid to run and fetch for her the whole time. (NRRC)


b. Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and didn't expect to have a maid to run and fetch for her the whole time. (SEU)

The SUPC situation (had got rid) implies that the RC situation no longer holds. The NRRC assigns a characteristic to Nanny Bill and explicitly locates it at the time when she worked for Lady Foxglove (hadn't expected), a period of time that is over. The past tense (didn't expect) gives the same piece of information without bothering to indicate the chronological order between the RC situation and the matrix clause situation. The fact that the RC is coordinated to another clause with a past perfect makes it very likely that the past perfect will be used in both conjoins (cf. section 8.2.2.2.3.). A similar difference in meaning between the past perfect and the past tense can be observed in the corresponding RRC, the only difference with the NRRC being that the RC information has a restricting function:

(47) c. Lady Foxglove began to regret that she had ever got rid of the nurse who had been much cheaper and hadn't expected to have a maid to run and fetch for her the whole time, (adapted from SEU) (RRC)

d. Lady Foxglove began to regret that she had ever got rid of the nurse who had been much cheaper and didn't expect to have a maid to run and fetch for her the whole time.

A similar difference in effect can be observed between the past tense and the past perfect (when it is interpreted as bound indirectly) in the following RRC example:

(48) a. I ran into a man who had collaborated with Ger-mans who had lived with him.

b. I ran into a man who had collaborated with Ger-mans who lived with him.

When the past tense is used, the Germans may still be living with the man in question. However, under the influence of the past perfect in the SUPC (had collaborated), which indicates a period that is over, the past tense is more likely to be interpreted as referring to a situation that is also over. Even so, it is only when the past perfect is


used that it is explicitly indicated that the Germans no longer live with the man referred to.

In the examples given so far, the past perfect verbs in the RC are atelic (be in practice, grow marihuana, expect, live). It follows from the Constraint on the Replacement of the Past Perfect (cf. chapter 3, section 3.2.1.) that the replacement of the past perfect by the past tense coincides with a change from interpreting a situation as bounded to interpreting a situation as unbounded. Situations of the latter type usually enter in a relationship of simultaneity with another situation unless there is an adverbial or pragmatic knowledge which imposes a certain order (cf. He said he had loved her vs. He said he loved her). However, this change does not always affect the accept-ability in the examples under discussion in the same way as it did in those discussed in chapter 3 (shift of domain vs. temporal subordina-tion in the past sector); if the past perfect is used, the situation is represented as anterior to the SUPC situation (a), or, if required by the context, to the matrix situation (b). When the past tense is used, the situation is represented as simultaneous with its SUPC situation (c) or matrix clause situation (d). The same W-relations hold be-tween the different situations in options (b) and (c) (i.e. the RC situa-tion is W-anterior to the SUPC situation and W-simultaneous with the matrix situation), as will be clear from the schematic representa-tions given below. This explains why substituting a past perfect for a past tense in examples of the type under discussion does not always coincide with a change in the temporal location of the situations:

(a) X

X /


(c) χ

X past tense

(d) to

X X past tense

In (49) and (50), for instance, the semantic bonds between the RC situation and the SUPC situation safeguard a correct temporal interpretation, i.e. the past tense RC situations in (49b) and (50b) are temporally linked with the SUPC situation (cf. (c)) rather than with the matrix clause situation (cf. (d)), the past tense is therefore interchangeable with the past perfect (cf. (b)):

(49) a. Munyard left his jitters behind. Long conferences had been held between Harry Turton and his Admiral in which various alternations had been debated. (LOB)


b. Munyard left his jitters behind. Long conferences had been held between Harry Turton and his Admiral in which various alternations were debated.

(50) a. For an eternity, it seemed he'd been at strife with the elements and at strife with men. With axes and steam horses he had fought the creeping Black Frosts which had tried to burden his ship with their deadly weight of ice. (LOB)

b. For an eternity, it seemed he'd been at strife with the elements and at strife with men. With axes and steam horses he had fought the creeping Black Frosts which tried to burden his ship with their deadly weight of ice.

In these examples the CRPP does not impose restrictions on the choice of verb form because it is the SUPC situation rather than the matrix clause situation that functions as binding TO for the past tense.

However, if the semantics and pragmatics of the situations do not impose a particular chronological order, the unmarked interpretation of the past tense is that it is bound by the matrix clause situation: in (51), the use of the past tense results in a relationship of simultaneity being established between the RC situation and the matrix clause situation (cf. (d)) rather than between the RC situation and the SUPC situation (cf. (c)). In that case, the past tense will be ruled out:

(51) a. Rodgers knew that Oxenham was merely fishing when he suggested that Tarrant might be the limping man who had been at Brighton. Plenty of people walked with a limp, and no link had been discovered between Tarrant and the white-haired man who had shown an interest in Haines. (LOB)

b. Rodgers knew that Oxenham was merely fishing when he suggested that Tarrant might be the limping man who had been at Brighton. Plenty of people walked with a limp, and no link had been discovered between Tarrant and the white-haired man who 'showed an interest in Haines.

The RC situation referred to in the past tense allows two inter-pretations: it is either simultaneous with the SUPC situation of not finding a link or simultaneous with the time referred to by Rodgers


knew. On the latter reading, the past tense is not substitutable with the past perfect.

The differences in effect between the past perfect and the past tense mentioned so far can be summarized as follows:

a. The use of the past perfect for indirect binding focuses on the temporal relation between the matrix situation and the RC situation, whereas the use of the past tense concentrates on the temporal relation between the SUPC situation and the RC situation.

b. The past perfect indicates that the RC situation is over at T02, while the past tense implicates that it is. However, the presence of a past perfect in the SUPC, which indicates the end of a certain situation, reduces the difference in effect between the verb forms: as the past tense situation is simultaneous with a situation (SUPC situation) that is over, it becomes more likely that the RC situation no longer holds either.

c. One particular verb form may be ruled out because it brings about a change in the chronological relations between situations.

d. A valid conclusion in any case is that it is not possible to state rules of the kind that indirect binding is compulsory in NRRCs or that direct binding is compulsory in RRCs in examples of the type under discussion.

8.2.2.2.2. Resultative implicatures

Perfect tense sentences implicate that the (possible) result brought about by the situation still applies at TU (in the case of the present perfect) or T02 (in the case of the past perfect) (cf. chapter 4, sections 4.2.1.1. and 4.2.3.1.). If the speaker wants to stress the gradual process leading to a result, he will use a perfect tense (in this case, the past perfect), the tense par excellence for expressing resultativeness:

(52) a. But Hahnemann had had extraordinary successes in curing diseases which had quite baffled the conventional remedies. (LOB) (RRC)

b. But Hahnemann had had extraordinary successes in curing diseases which quite baffled the conventional remedies.

The process and its results are emphasized through the use of the past perfect.


Example (53) is a similar NRRC example:

(53) a. We had given too much credence to German propaganda, which had built up in our minds a picture of widespread fanaticism that might well entail prolonged operations of a type that would call for most careful handling. (LOB)

b. We had given too much credence to German propaganda, which built up in our minds a picture of widespread fanaticism that might well entail prolonged operations of a type that would call for most careful handling.

8.2.2.2.3. Conjoined clauses

When one clause is conjoined to another in which a (continuative, as in (54) or indefinite, as in (55)) past perfect is used, it is very likely that the same tense will be used in the two clauses if they are to be interpreted as W-simultaneous (cf. Adelaar - Lo Cascio 1986). De-clerck writes that "a situation expressed by a subclause which is conjoined to a subclause which is also marked by relative tense may be bound by the situation-TO of the latter clause" (1991a: 143). In my opinion, indirect binding is the unmarked option (general rule) and not just a mere possibility:

(54) a. He was a dedicated and highly successful teacher, who had always been willing to meet the challenge of new situations and methods, and who had always shown himself deeply committed to his work, (adapted from SEU) (NRRC)

b. He was a dedicated and highly successful teacher, who had always been willing to meet the challenge of new situations and methods, and who always showed himself deeply committed to his work.

(55) a. Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and hadn't expected to have a maid to run and fetch for her the whole time. (SEU) (NRRC)


b. Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and didn't expect to have a maid to run and fetch for her the whole time

The fact that the last clause is conjoined to the preceding clause con-tributes to the preference for the use of the past perfect. However, the past tense is not unacceptable; it is interpreted either (a) as a shift of domain tense establishing a W-simultaneous domain or (b) as tempo-rally subordinated to the TOsi of the first conjoin. Adelaar - Lo Cascio (1986: 263-266, 293-294) argue that the tense of an anaphoric clause is never bound by the anaphoric tense to which it is coordi-nated. Declerck (1991a: 143) gives the following counterexample to this claim:

(56) An hour ago a witness told the police that the girl who was found killed last night had missed the train home on Tuesday night and would try to get home by hitch-hiking.

It is basically because the situation of the second conjoin (would try) is W-posterior that the clause is bound by the situation-TO of the first conjoin (had missed). When had tried is used, the first conjoin no longer performs the function of TO. In this case, was found killed functions as binding TO.

8.2.2.2.4. Continuative perfect SUPC

In the following example, the RC situation and its SUPC situation start before the matrix clause situation and lead up to it. The W-si-multaneous SUPC situation will be temporally subordinated to the matrix situation rather than to the RC situation, with which it is W-simultaneous, provided that the W-anteriority link (before and up to) between the matrix and the SUPC situation is considered to be more important than the relationship of W-simultaneity between the RC and the SUPC situation:

(57) a. The secret transpired: the girl that he had been living with ever since he graduated was seeing John at the same time.


b. The secret transpired: the girl that he had been living with ever since he graduated had been seeing John at the same time.

Her dating John while living with someone else may be considered to be more important than the fact that the former situation is W-anterior to the matrix situation. In this case the past tense will probably be used. If the speaker wants to stress that her seeing John has been going on for quite some time, he will probably use a continuative past perfect.

8.2.2.2.5. Directly bound conditional tense: prediction vs. indirectly bound past perfect tense: fact

So far, examples of the following type have been dealt with:

RC situation is W-simultaneous with SUPC1 situation RC situation is W-anterior to C3 situation SUPC1 situation is W-anterior to C3 situation

In this section, examples of the following type will be commented on:

RC situation is W-posterior to SUPC1 situation RC situation is W-anterior to C3 situation SUPC1 situation is W-anterior to C3 situation

It is important to point out that in RRCs and NRRCs alike, the use of the past perfect may be ruled out if the semantics of the sentence or pragmatic knowledge does not impose a W-posterior reading. The RC situation will be interpreted as W-anterior to the SUPC situation in (58a) and (58c):

(58) a. He remembered that Coulter had been brought up in the orphanage which had been sued for child abuse. (RRC)

b. He remembered that Coulter had been brought up in the orphanage which would be sued for child abuse. (RRC)


c. He remembered that Coulter had been brought up in the orphanage in Surrey, which had been sued for child abuse. (NRRC)

d. He remembered that Coulter had been brought up in the orphanage in Surrey, which would be sued for child abuse. (NRRC)

Neither in (58a) nor in (58c) is the past perfect likely to be inter-preted as referring to a W-posterior situation. It is not clear, though, whether the past perfect in RRC and the NRRC is bound directly or indirectly. Another problem to be referred to here is that of determining whether would be sued is bound directly or indirectly. If it is bound directly, the RC situation is understood to be posterior to both the situation referred to in SUPC1 and that referred to in C3.

Even when the context makes it clear that the RC situation lies after the matrix clause situation, the past perfect and the conditional tense may differ in acceptability. The change of "representation of the situation as a fact" (past perfect) into "representation of the situation as a prediction" (conditional tense) is clear from the fol-lowing example:

(59) a. In doing so they were not only following their own conscience - since the vote was a free one - they were also providing a reflection of public opinion. The polls had, in fact demonstrated, that of all those who were questioned on the desirability of reform, 63 per cent had come out in favour of change. (SEU)

b. In doing so they were not only following their own conscience - since the vote was a free one - they were also providing a reflection of public opinion. The polls had, in fact demonstrated, that of all those who were questioned on the desirability of reform, 63 per cent ?? would come out in favour of change.

Had demonstrated is temporally subordinated to the TO i( of were following and were providing. The situation in the last cfause (had come out) is W-posterior to the situation of the previous RC situation (were questioned). However, the conditional is not used, because the factuality of the situations is relevant in this context: there is reference to past results of a poll. Therefore, had come out is temporally subordinated to the TOsit of were following and were


providing. In (60) as well, it is necessary to use an indirectly bound past perfect rather than a directly bound conditional tense:

(60) a. Then he remembered about Coulter: that he had no family at all, had been brought up in an orphanage, from which he had run away at the age of eleven, to join the circus. (SEU) (NRRC)

b. Then he remembered about Coulter: that he had no family at all, had been brought up in an orphanage, from which he ? would run away at the age of eleven, to join the circus.

When the RC situation is temporally subordinated (would run away) to the situation of the preceding clause (had been brought), it is represented as a prediction. Would run may be argued to be (marginally) acceptable if it is considered to be a piece of infor-mation given by the narrator and not as something that Coulter remembers. There is no reason why there should be a shift of attention from what Coulter thought to what the author knows, hence the marginal acceptability of the conditional tense. Example (61) is another example of this type:

(61) a. Something protruded, not wood but rope. The coffin had been wrapped around with two inch manilla rope which had become iron hard as the guano permeated it. (SEU)

b. Something protruded, not wood but rope. The coffin had been wrapped around with two inch manilla rope which ?? would become iron hard as the guano permeated it.

The next example shows that in some cases, the meaning inherent in the conditional tense is compatible with the semantics and pragmatics of the sentence:

(62) a. This view was strengthened by the fact that he had written only two years before a paper on National Education, through which he had become one of champions of middle class education. (LOB)

b. This view was strengthened by the fact that he had written only two years before a paper on National


Education, through which he would become one of champions of middle class education.

8.2.2.2.6. Indirect binding the only possibility in the NRRC vs. direct or indirect binding in the RRC

RRCs and NRRCs sometimes differ as far as (in)direct binding is concerned. In examples of the type given below, the following temporal relations hold between the different situations:

RC situation is W-simultaneous with SUPC1 situation RC situation is W-posterior to C3 situation SUPC1 situation is W-posterior to C3 situation

(63) On Monday John said that on Wednesday he would try and intercept that letter that would arrive (arrived) that day.

Examples of this type constitute the past sector counterparts of the W-simultaneous post-present examples discussed in chapter 5. The past sector W-posteriority relation between the NRRC situation and its matrix situation has to be expressed, otherwise the NRRC loses its


W-posterior reference (just as the present tense does not establish future time reference in NRRCs). In RRCs, the use of a past tense may have the same effect (just as PPS forms may be misinterpreted in post-present sector RRCs); however, unlike in NRRCs, this is not always the case. The basic principle is that an unbounded past tense form in a NRRC is interpreted as establishing a relationship of W-simultaneity with the past tense matrix situation, rather than with the SUPC situation. The reason is similar to that given in chapter 6: NRRCs provide relevant information. As facts are considered to be more important than predictions, the past tense will be interpreted as simultaneous with a situation represented as a past fact (i.e. the matrix clause situation) rather than with a predicted one (i.e. the SUPC situation). The same line of reasoning applies to RRCs. How-ever, the tighter syntactic links between the RRC and its SUPC may establish a W-posterior reading:

(64) a. Richard regretted that he would go on holiday to a place which would be full of tourists. (RRC)

b. Richard regretted that he would go on holiday to a place which was full of tourists. (RRC)

c. Richard regretted that he would go on holiday to the Costa del Sol, which would be full of tourists. (NRRC)

d. Richard regretted that he would go on holiday to the Costa del Sol, which 'was' full of tourists. (NRRC)

The past tense in (64d) indicates that the holiday resort is already overcrowded at the time Richard regrets the fact that he will go there. Unambiguous W-posterior location of the situation requires the use of a conditional tense. Apart from allowing this interpretation, the past tense in the RRC (65b) also allows a W-posterior (i.e. si-multaneity with the SUPC situation) interpretation, although this is not the unmarked one. The possible adverbs which can be added to the relevant sentences illustrate the differences:

(65) a. Richard regretted that he would go on holiday (in June) to a place which would be full of tourists (in June). (RRC)

b. Richard regretted that he would go on holiday (in June) to a place which was full of tourists (in June / always). (RRC)


c. Richard regretted that he would go on holiday to the Costa del Sol (in June), which would be full of tourists (in June). (NRRC)

d. Richard regretted that he would go on holiday to the Costa del Sol, which 'was' full of tourists (* in June / always). (NRRC)

In June can be used in (65d) if it gets a habitual interpretation, i.e. every year in June, the Costa del Sol is very crowded. The RRC in (65b) allows a non-habitual interpretation. Example (66) also shows that a RRC may differ from a NRRC in that the former's tight syntactic relation with the SUPC often allows a W-posterior directly bound interpretation of the past tense:

(66) a. The Packwood-Roth proposal would reduce the tax depending on how long an asset was held. It also would create a new individual retirement account that would shield from taxation the appreciation on investments made for a wide variety of purposes. (WSJ) (RRC)

b. The Packwood-Roth proposal would reduce the tax depending on how long an asset was held. It also would create a new individual retirement account that shielded from taxation the appreciation on investments made for a wide variety of purposes. (RRC)

c. The Packwood-Roth proposal would reduce the tax depending on how long an asset was held. It also would create a new individual retirement account, which would shield from taxation the appreciation on investments made for a wide variety of purposes. (NRRC)

d. The Packwood-Roth proposal would reduce the tax depending on how long an asset was held. It also would create a new individual retirement account, which * shielded from taxation the appreciation on investments made for a wide variety of purposes. (NRRC)

Shielded in the RRC will be interpreted as a directly bound form indicating simultaneity with the SUPC situation. The fact that there is reference to a new account contributes to establishing this


interpretation (cf. chapter 5, section 5.3.2.3.)· The use of a past tense in a NRRC is not possible for two reasons: first, the past tense will be understood as bound indirectly, i.e. it expresses simultaneity with TO, rather than with the TOsit of the SUPC situation. In other words, it changes the temporal location of the RC situation. Moreover, even in the latter interpretation, the past tense is not acceptable, because of the semantics of the example. There is reference to something which will happen in the future; hence it does not exist at the time functioning as T02 for the SUPC situation, with which the past tense expresses simultaneity in this interpretation. Therefore, the past tense is ruled out altogether. In the next example, the past tense in the NRRC is interpreted as expressing simultaneity with noted rather than with would prefer:

(67) a. Douglas Stoneman, vice-president of Shell Canada, noted that producers would prefer to avoid hearings into competing proposals that would lengthen the regulatory review process and bogged down development. (WSJ) (RRC)

b. Douglas Stoneman, vice-president of Shell Canada, noted that producers would prefer to avoid hearings into competing proposals that lengthened the regulatory review process and bogged down development. (RRC)

c. Douglas Stoneman, vice-president of Shell Canada, noted that producers would prefer to avoid hearings into competing proposals, which would lengthen the regulatory review process and bogged down development. (NRRC)

d. Douglas Stoneman, vice president of Shell Canada, noted that producers would prefer to avoid hearings into competing proposals, which 'lengthened the regulatory review process and bogged down development. (NRRC)

The NRRC past tense indicates a particular characteristic of hearings into competing proposals, namely they lengthen the review process and bog down development. Apart from allowing this interpretation, the past tense in the RRC can also be interpreted as a verb form that is bound directly: it defines a particular characteristic which the future competing proposals will have.


RRCs and NRRCs also differ as far as (in)direct binding is concerned when the following temporal relations hold between the situations:

RC situation is W-anterior to SUPC1 situation RC situation is W-posterior to C3 situation SUPC1 situation is W-posterior to C3 situation

Examples of this type constitute the past sector counterpart of post-present sector present perfect or past tense RCs. Just as the present perfect or past tense is interpreted as an absolute tense in the post-present sector NRRC examples, the past perfect is interpreted as bound indirectly in the NRRC, i.e. as bound by the "factual" matrix clause situation:

(68) a. John announced that he would be helped on Tuesday by the man who would kill the inspector on Monday. (RRC)

b. John announced that he would be helped on Tuesday by the man who had killed the inspector on Monday. (RRC)


c. John announced that he would be helped on Tuesday by Dr. Swallow, who would kill the inspector on Monday. (NRRC)

d. John announced that he would be helped on Tuesday by Dr Swallow, who had killed the inspector on Monday. (NRRC)

The NRRC can only be bound by the "factual" matrix clause sit-uation, and not by the "predicted" SUPC situation: the only meaning that can be communicated by (68d) is that Dr Swallow killed the inspector before the time of speaking. In the RRC, on the other hand, the past perfect can be used to communicate the information that the murder happens after John's announcement.

8.2.3. General conclusion

In section 8.2., an analysis has been offered of the following situ-ations in which indirect binding occurs:

type 1:


(69) He had lived with people who had collaborated (icollaborated) with the Germans.

type 2:

(70) He would go to someone who would be willing (was willing) to explain the issue to him.


(71) But when they'd been together in Berlin three weeks and no job had appeared, Diana had got hold of a banker, who'd taken (would take) her off with him to Paris. (Isherwood, Goodbye to Berlin)

type 4:

(72) He was disappointed because he knew his student would not be admitted to the university he had applied (would apply) to.

Differences can be observed between RRCs and NRRCs when the RC situation is W-posterior to the matrix clause situation and W-


simultaneous with the SUPC situation (type 2) or when the RC situation is W-anterior to the SUPC situation and W-posterior to the matrix clause situation (type 4). It has been pointed out that these two types of sentences constitute the past sector counterpart of future sector present tense/present perfect and past tense NRRC examples. In other words, the importance of this chapter is that it enables us to complete the areas in which RRCs and NRRCs are likely to differ: if it is possible to interpret a particular situation as a past fact rather than as a past prediction, the former option will be chosen.

As far as the examples of type 1 and type 3 are concerned, the following conclusions can be drawn:

a. The past perfect has the following characteristics which the past tense does not have: (1) The past perfect is very well suited to express resultative implicatures. (2) The indefinite past perfect (of an atelic verb) explicitly indicates that the state no longer holds at T02, whereas the past tense only does so by implicature. Accordingly, depending on which element of meaning the speaker wishes to convey, the past tense or past perfect will be chosen.

b. In the case of type 1, it may also be the relative importance of the semantic link of W-anteriority (between the matrix and the RC situation) rather than W-simultaneity (between the SUPC and the RC situation) which determines the choice of verb form.

c. When two clauses (i.e. the SUPC and the RC) referring to sit-uations that are W-anterior to the matrix clause situation are conjoined, it is very likely that the past perfect will be used in both conjoins.

d. In examples of type 2, the preference to represent the W-pos-terior RC situation as a fact rather than as a prediction may induce the speaker to use a past perfect (indirect binding) rather than a conditional tense (direct binding). The examples considered in this chapter again show clearly that pragmatic factors play a very important role in the choice of binding TO. This explains why it is not possible to formulate strict rules; what is possible in one context may not be possible in another.

General conclusion

It will be clear from the preceding chapters that there are respects in which the expression of temporal relations differs in RRCs and NRRCs. In other words, the tight syntactic bonds between the RRC and its SUPC vs the loose syntactic bonds between the NRRC and its SUPC indeed affect the tense system in some cases. The foregoing analysis enables us to specify the contexts in which differences are likely to occur:

1. Temporal subordination vs shift of domain

1.1. Possibilities

The following scheme shows whether the possibilities of temporal subordination (sub.) and shift of domain are allowed in the different sectors in RRCs and NRRCs:

Past Pre-present Post-present Sub. RRCs + + +

NRRCs + + -

Shift RRCs + + + NRRCs + + +

It should immediately be added that the fact that there may be either temporal subordination or a shift of domain in most sectors does not imply that the choice between these options is unrestricted. However, the most important conclusion to be drawn is that post-present sector NRRCs do not allow temporal subordination. As explained in chapter 5, this is due to the convergence of two factors: (a) NRRCs give relevant information, and (b) factual statements are considered more relevant than predictions. This will result in interpreting as absolute tenses forms which could be interpreted as PPS forms. However, although temporal subordination is possible in RRCs, the temporally subordinated verb forms can also be interpreted as absolute tenses, the difference with NRRCs being that the likelihood of "misinterpretation" is much smaller in RRCs. The tight syntactic bonds between the SUPC and the RRC, which safeguard a post-present interpretation, lie at the origin of this difference.


1.2. Constraints on the use of particular verb forms

As far as constraints on the options in the different sectors are con-cerned, the following distinction must be made:

a. A particular form may be ruled out because it changes the temporal location of the RC situation; e.g. temporal subordination (cf. chapter 4, section 4.2.1.) or shift within the sector (cf. chapter 4, section 4.2.3.) in the pre-present sector, temporal subordination in post-present RRCs and NRRCs (cf. chapter 5, section 5.2 and section 5.3.1.).

b. A particular form may be ruled out because it changes the temporal relation between the RC situation and the SUPC situation ; e.g. the CRPP (chapter 3, section 3.2.1.), the CRPrP (chapter 4, sec-tion 4.2.2.1.), and the CRCT (chapter 3, section 3.4.2.1.) predict that the past tense cannot replace the past perfect, present perfect, conditional tense, respectively, in RRCs and NRRCs when the verb is atelic or zero-telic, because the temporal relation of W-anteriority (CRPP and CRPrP) or W-posteriority (CRCT) will be changed into one of W-simultaneity (provided the semantics or the pragmatics of the situation does not impose a temporal order).

c. A particular form may be ruled out in a NRRC because it is not consistent with the semantics and pragmatics of the situation. It has been pointed out on several occasions that the restricting function of the RRC allows more "temporal liberty" than the function of giving relevant information performed by the NRRCs. In order to be rele-vant, it may be necessary for the NRRC to gear the temporal information it provides to the temporal location of the SUPC situa-tion. In cases like those indicated on chapter 3, section 3.2.2.3. and chapter 4, section 4.4.3. there are differences between RRCs and NRRCs in the acceptability of particular verb forms. However, these are not due to possible differences in (a) the temporal location between RRCs and NRRCs, or (b) the temporal relation established between the RRC situation and SUPC situation, on the one hand, and the NRRC situation and the SUPC on the other hand.

d. A particular form may be unacceptable because its semantics is not compatible with the given situation. Apart from the examples referred to in (c), RRCs and NRRCs are in most cases subject to the same restrictions (cf. the examples with which the principles have been illustrated).

e. Certain forms are ruled out because they are ungrammatical, i.e. incompatible with other lexical material (e.g. the presence of a par-ticular adverbial may block the use of a certain verb form).

344 General conclusion

2. The principle of unmarked temporal interpre-tation

The most important conclusion to be drawn from chapter 7 is that bounded (a)telic shift of domain NRRCs tend to be interpreted in the order in which they are reported, whereas RRCs of this type usually refer to W-anterior situations.192 If the NRRC follows the SUPC, it is likely to be interpreted as continuative. RRCs can belong to the foreground provided the W-posterior relation can be derived from the semantics and pragmatics of the sentences.

3. Choice of TO No differences have been detected between NRRCs and RRCs as far as the items that can function as binding TO are concerned.

4. Indirect binding

The following cases of indirect binding have been analysed:

type 1:


(1) He had spoken to a demonstrator who had looked (looked) suspicious.

type 2:

(2) But when they'd been together in Berlin three weeks and no job had appeared, Diana had got hold of a banker, who'd taken (would take) her off with him to Paris. (Isherwood, Goodbye to Berlin)

type 3:


\

(3) He would complain to the delegate who would be (was) responsible for that issue.

Type 3 constitutes the past sector counterpart of the expression of W-simultaneity in the post-present sector; direct binding is not possible in NRRCs of this type.

type 4:

(4) He was angry at the colleague who would betray the student who would play (had played) truant.


Type 4 constitutes the past sector counterpart of the expression of W-anteriority in the post-present sector; direct binding is not possible in NRRCs of this type either.

Type 1 Type 2 Type 3 Type 4

DB RRC + + + + NRRC + + - -

IB RRC + + + + NRRC + + + +

Again, the possibility of both direct and indirect binding in most cases does not imply that there are no restrictions of any kind. The impossibility of direct binding in type 3 and type 4 NRRCs can be explained in terms of the two factors mentioned above: (a) NRRCs give relevant information (b) past time facts are considered more im-portant than past time predictions. Accordingly, the NRRC situations are understood as bound by the "factual" matrix clause situation rather than the "predicted" SUPC situation. However, the post-present restriction and the indirect binding restriction are not entirely similar; using a particular verb form results, in the case of post-present sector NRRCs, in locating the NRRC situation in a different sector, and, in the case of (in)direct binding, in changing the temporal relation between the NRRC situation and the SUPC situa-tion.

The above survey indicates that the tense system in RRCs and NRRCs differs in post-present sector, W-posterior indirectly bound and W-posterior shift of domain contexts. However, taking into account the foregoing discussion, the claim that there is also a high degree of similarity between RRCs and NRRCs is not difficult to substantiate. A general principle which overrides the restrictions is that pragmatic factors play a very important role as far as the tem-poral interpretation of RCs is concerned; any unmarked interpreta-tion may be overruled by situational or contextual knowledge.

Apart from dealing with the main topic of research and laying bare a number of general principles in connection with the use of tense (chapters 3 to 6), this investigation into the tense system in RCs has also been useful with respect to the following issues:

a. New light has been thrown on the boundedness-unboundedness and telicity-atelicity distinctions, especially by incorporating the ef-fect of tense and adverbials into the discussion (chapter 1).


b. Another conclusion on the general theoretical level is that within the tradition of descriptive linguistics, Declerck's theory of tense is a major step forward in comparison with the many other approaches that I have taken stock of. Twentieth century linguistics has been plagued by an urge to create as many theoretical models as there have been linguists, combined with a lack of zeal in testing out the practical relevance of the theories. In order not to fall into the same trap, I have chosen to select a theory that inspired confidence and have put it to the test. I have found that Declerck's model passes the test with flying colours even when confronted with the most exacting language material: that of a real life corpus. Although critical reflec-tions have been offered and a few refinements have been suggested (cf. chapter 3 - section 3.3., chapter 4 - section 4.3., chapter 7 - sec-tion 7.4., chapter 8 - sections 8.1.2. and 8.2.1.), it has above all been established that the framework is valid, and that it enables the re-searcher to analyse systematically a range of issues whose discussion has hitherto remained fragmentary.

c. Frequent reference to the Gricean Cooperation Principles stresses their importance for the use of tense.

d. A considerable degree of indeterminacy (cf. (in)direct binding, classification of verb forms as absolute or relative) has been brought to light.

e. It has been shown that the choice of a particular verb form is determined by a number of interlinked factors, some of which play a more significant role than others.

Appendix

Original examples chapter 3

(8) For domestic consumption the agenda will demonstrably have been produced by a Reagan Administration and not by a Mondale opposition which has been canvassing the maintenance of contacts with the Kremlin in every state primary. (SEU)

(29) Scientists from the HSE were this week extending their in-(30) vestigations underground and were planning to drain a water

tunnel from the River Lüne to the Abbeystead plan in an at-tempt to locate the source of methane gas which caused the explosion. (SEU)

(42) I first saw the above on 8.11.22 with a story very much (53) resembling polymyalgia rheumetica which she said had come

on fairly suddenly eight weeks previously. (SEU) (63) James Wilkinson became apprenticed to a celebrated London

gunsmith called Henry Nock, who had set up his business in Ludgate Street in 1772. (SEU)

(64) On this night of the disintegration of United's defence Astle struck again in the seventy-fifth minute, heading in a pass from Lovett who had come on as a substitute for Rees. (SEU)

(68) This latest success has most recently been acknowledged by the Design Council who selected the Bonded shaving system for its award for technological innovation and excellence in design. (SEU)

(108) I have put her on Naprosyn 250 mgm., q.d.s., which I hope will relieve at least some of her symptoms until I see her again in one month's time. (SEU)

(121) I gave her a short prescription for Ponstan Forte which may help a bit. (SEU)

(123) At the moment it seems to have settled down but she has some medial ligament tenderness which I have injected and she has also developed a troublesome left tennis elbow which I have also treated with local steroid and have arranged to see her again. (SEU)

(131) If you wish to replace from block you now select the from block option which will insert the text you have saved to buffer using the cut paste option. (SEU)

350 Appendix

Original examples chapter 4 (17) In particular, they appear to accept that Name has come as

close as any scholar to solving the problems of [Word Word],+ <-A/a->nd explaining the rationale which this work would have followed, had it been completed. (SEU)

(21) Rang Ma who got back from Painswick today. (SEU) (22) You may claim for any necessary travelling expenses which

you had to pay in the course of your duties (that is in actually carrying out your work); also any other necessary expenses you have paid and which have arisen wholly and exclusively in carrying out your duties. (SEU)

(24) We are likely to be struck by the extent to which working-class people have improved their lot, acquired more power and more possessions; we are likely to be even more im-pressed by the degree to which they no longer feel themselves members of 'the lower orders', with a sense of other classes, each above them and each superior in the way the world judges. (SEU)

(28) New Jersey's own highest court remains a liberal bulwark against major restrictions on abortion, but the U.S. Supreme Court ruling, Webster vs. Missouri, has engaged voters across the nation who had been insulated from the issue. (WSJ)

(30) Or if public controversies do not intrude themselves to upset (53) the balance there may be private controversies and rivalries,

or professional preoccupations, such as the desire of a scholar to promote a theory of his own devizing or to destroy one that has been promoted by a rival; and, as if these sources of trouble were not enough, scholars have sometimes been known to develop irrational predilections for one dead individual or departed race and an equally irrational dislike for another. (SEU)

(32) Grants will only be issued to students who produce valid College identity cards. (SEU)

(33) What is not a matter of taste but one of necessity is that the technique of historical criticism should be applied to any account of matters that have occurred in the past to which any importance is attached. (SEU)

(34) They have not go the guts to attack the trade unions, whose restrictive practices have prevented price reductions, and are deceiving the public by saying this Bill will reduce the cost of living. (SEU)

Original examples chapter 4 351

(38) In this chapter we shall examine (44) rice cultivation in South India, focusing in particular on the

complex of technological changes that has occurred since the late 1960s, and which has been termed the Green Revolution. (SEU)

(39) The more that can be charged against the estate the better it will be from the estate duty point of view and as there are one or two other adjustments to be made for bills which came in after we had applied for Probate I will prepare a Corrective Affidavit which should enable us to recover some of the estate duty which has been paid. (SEU)

(40) Here are to be found items which have been selected for their value and attractiveness. (SEU)

(41) But alongside Mr. Tosh's article the journal also publishes the experience of authorities who have lost all faith in corporal punishment. (SEU)

(42) The Committee considered that its first step should be to con-(49) tinue the process of consultation which it had already

initiated. (SEU) (43) Mr. [Name] asked me to see this man on 30.1.22 in the hope

that it might be possible to relieve the pain in his right lower limb which he said he has had since [2222]. (SEU)

(45) Men departed for the front and returned from it, and he saw those who came with them to say goodbye, and those who walked in agitation up and down the platforms, waiting, saw partings and greetings, saw the waving and the tears. (SEU)

(46) The Swedish economy has been unexpectedly buoyant, confounding most economic forecasters, who had warned of a first-half downturn. (WSJ)

(48) She seems to have made a surprisingly good response, either to the Indomethacin which I prescribed at her first visit, or to the fact that she has now started at a Special School and has been released from the stresses and strains associated with her previous school-phobia; (I am inclined to think that the latter is the more likely explanation). (SEU)

(50) He had been even less keen when the Minister had hinted that this private enterprise Emperor would inevitably follow the same primrose path that had already been blazed by other British long-range airliners designed since the war. (SEU)

(51) Thus, if devaluation has - as the Prime Minister claimed - had the effect of extricating the U.K. from the economic straitjacket that had previously encased it, the result has not

352 Appendix

been to make Britain's economic future look as promising as that of some of the more growth-conscious of the overseas £ area countries. (SEU)

(52a) Here are to be found items which have been selected for their value and attractiveness. (SEU)

(52b) We went into a fight like it was all a joke, but I've seen Fat sit down beside a guy that'd been bumped off and cry like it was his own brother. (Defroment 1973: 77)

(5 2d) In particular, they appear to accept that Name has come as close as any scholar to solving the problems of [Word Word],+ <-A/a->nd explaining the rationale which this work would have followed, had it been completed. (SEU)

(54) Grants will only be issued to students who produce valid College identity cards. (SEU)

(56) Several thousand have been killed on both sides as Indian troops fought with the sniping tribesmen who used the jungles as cover. (SEU)

(57) But, unhappily, we can see no early prospect of recovering "Agapenor" and "Melampus" which have been trapped in the Suez Canal since June - a harsh penalty for maintaining schedules by allowing our average frequency of almost a daily transit to run unchecked before the Canal was closed. (SEU)


(12) All left by 8.45 except Donald Payne, who stayed a little later. (SEU)

(13) Leeds eventually<sic> changed into top gear, and after a bewildering movement, in which everyone except Harvey appeared to take part, Gray delivered a splendid shot, which brought a save no less worthy from Donaldson. (SEU)

(21) Mr Ranbir Singh, a Transport and General Workers' Union senior shop steward at the factory, said: "We did not refuse to work with Mr Marshall because he was white, but because he jumped a long queue of coloured men who were in line for this job. (SEU)

(25) He suddenly knew envy for the easy happiness with which she had returned to life, while his own return was so lonely and uneasy. (SEU)


(31) Mr. Wilson might, for example, have been defeated - in which case Mr. Heath would have no place at all for a land commission or any other legislation that a Labour government had left on the stocks. (SEU)

(34) All power, therefore, to the elbows of those working Not-tinghamshire men working to encourage groups of miners in all parts of the country, who are fed up with a strike in which they have been denied a say, and who dearly want to create the opportunity for a return to work, especially now it is so clear that the strikers cannot be allowed to win. (SEU)

(35) Mr. [Name] was with us only for the academic year [2222/22], when he had to resign because of family diffi-culties which had affected him throughout his time with us. (SEU)

(40) I was Head of the Department in which she studied for her B.A. degree at [Name] college, so I know her well. (SEU)

(44) Though Arco won't describe the chemical properties of its new unleaded gasoline, it has hinted to some distributors that it will be promoted to customers who now fill up with leaded as well as to owners of newer models. (WSJ)

(47) A Left-wing student seized the microphone but was quickly cut off by an official who cut the cord. (SEU)

(48) Lady Foxglove crossed out "Lilli Marlene" and substituted in her characteristic sloping hand "Bless This House, but not too loud or there might be complaints", and looked up with a busy smile at the enormous figure who was clearing his throat importantly: "Urs summum ze door," he explained and beamed when Lady Foxglove said: "You mean there's someone at the door? (SEU)

(49) There is also a local management committee with repre-sentatives from people in the district who give financial advice. (SEU)

(52) She retained in the vitality of her spirit, in the vigour of her language, in the occasional peasant quality of her humour, a strength which her children had not and towards which they had at times something of a sophisticated and urbanized 'neshness' (soft squeamishness). (SEU)

(56) Courts have ruled that taxpayers must submit to TCMP audits, but the IRS will excuse from the fullscale rigors anyone who was audited without change for either 1986 or 1987. (WSJ)

(60) Although she did not have the requisite research achievement to warrant her applying for this chair, she did give proof of

354 Appendix

considerable powers of organisation and leadership and created a very solid foundation on which I was able to build when I came here. (SEU)

(61) A few scuffles broke out among the 600 students who were allowed into the hall before university ushers locked the door. (SEU)

(68) Its report, which was accepted by the Senate in February 1966, has been generally welcomed for the constructive solutions it provides to the main problems which have hith-erto hindered full co-operation in the joint use of the research facilities of these institutes. (SEU)

(71) Reports from Alexandria said that police surrounded the city's university area as students of the engineering faculty continued a sit-in protest which began yesterday. (SEU)

(73) The building of Brasilia inspired the first long forest road -the 2,123 km from Belem to Brasilia brought into existence 110 towns in the last decade, a jump in population from 100,000 to two million and some five million cattle which did not exist before. (SEU)

(74) The governor was an elderly man, and he had who colonel the clearest possible recollection of the British who colonel who had come with Colonel Younghusband forty-five years before. (SEU)

(78) Meanwhile their rivals who by now had huge populations of fruit-flies which they were subjecting to stress from heat, chemicals and anything else they could think of, were be-ginning to map on the chromosomes the actual positions of some of the mutated genes. (SEU)

(81) In the whole field of discrimination against women, the area which causes the most active, immediate suffering, is the economic insecurity of wives. (SEU)

(82) One aspect which concerns the Falklanders is Lord Chalfont's - and the British Government's - assumption that they suffer great inconvenience through Argentina's cold-shouldering of them. (SEU)

(83) One of the major problems to which constant reference has been made by Schools is the need to alter the composition of the Senate so as to give a greater degree of institutional representation. (SEU)

(85) Cases which are relevant to this consideration are [Name] Ltd. v. [Name Name Word Ltd] [2222] where the new employer was not even aware of the breach of confidence by the


employee and [Word] & [Word] Ltd v. [Name] [2222], where it was held that the remedy was available even where the know how and confidential information was not contained in a document. (SEU)

(86) For instance, if - as in what follows in this chapter - it is necessary to talk at length about the resemblances between two things (or the 'plausibility' of connecting them) then the sheer frequency with which the word 'resemblance' (unqualified) appears on the page creates a disposition to think that this is the most important aspect of the relation between the two things brought together in a metaphor. (SEU)

(93a) The human beings and events which emerge from rigorous historical enquiry and after severe historical criticism are often stranger, more idiosyncratic and in reality more interest-ing than what is normally supplied by the commonplaces of romanticism, or of political partisanship, for they have been quarried from the hard deposit of unpredictable fact that lies behind and gives strength to, the writing of all real history, and are not simply compacted of prejudices of which we already know too much. (SEU)

(93b) 'The brick lining which stores the heat is nice and solid, and opening the doors has virtually no effect on the temperature,1

said Mr Coombs. (SEU) (93c) The lesser persons who appear are also usually vi<-./v->idly

described ; the old hermit to whom the Red Cross Knight goes, with his beautiful white hair like spangles of frost; the wandering old Ignaro; small detailed touches of people in a crowd, like the mother whose child touched the dead dragon; and, particularly vivid, the pageant of sins in Bk I, surely one of the cleverest <-s/p->ortrayals of them ever. (SEU)

(93d) The people who are promoting it are reasonably cultured & 'right' - 1 hope. (SEU)

(93e) The people who are showing signs of enthusiasm are looking at those modern problems which centre on the affluent society and at groups with special needs. (SEU)

(93g) It is salutary to remember that even in the case of Jesus Christ, although He was the incarnate Son of God, and filled with the Holy Spirit, not everyone who saw Him recognized Him as the Son of the Most High, <fnn.2.fnn> not everyone who heard His word acknowledged it to be the Word of God, not everyone who watched His miracles accepted them as acts of God. (SEU)

356 Appendix

(93h) You will wish to know and be able to verify my company's identity and corporate substance, for the most important factors that determine a security purchase decision are ; that the products source is reputable and stable; that the product is effective, reliable and guaranteed; and supported by advice and service throughout it's life. (SEU)

(94) Those Labour members who use this issue as a stick to beat the Government are as much in the wrong, or almost as much so, as those tireless Tories who seek to use it in their selfish battle against nationalization and their self-interested attack on the Civil Service. (SEU)

(100) 3. The objects for which the company is established are: (A) To acquire the freehold property known as 18 South Villas in the London Borough of Camden and to hold the same as an investment for the benefit of the lessees of the flats comprised therein. (SEU)

(101) Newspapers gave it splash treatment, but the arguments that followed sounded exactly like those I had heard when I was living there 10 years earlier. (SEU)

(103) When this happened, the gene which they invaded was in-capacitated and very often the next gene in sequence was also affected. (SEU)

(104) It is recorded that when the invading armies of the Allies reached Belsen, the troops who witnessed the horrors of the camp for the first time were so affected by the sights and the smells and the testimony to human depravity that many of them were taken ill and had to be sent home on leave. (SEU)

(108) The blood tests which we did for the ?? bruising which mother reported, but which I never actually saw, were quite normal so I have not written to offer her any further ap-pointments in our department but will only be too pleased to see her again at your discretion. (SEU)

(109) The manservant who opened the door to Gerald was as old as his master. (SEU)

(110) I dont think they'd even heard of C.Day Lewis, and the poor man who read from Louis MacNeice got a poor reception. (SEU)

(112) Flooding results in iron, aluminium, manganese and calcium becoming more soluble, and as a result these elements chemically release the phosphorus which under aerated, dry-soil conditions is chemically bound up with them. (SEU)



(I) The rivers gain the majority of their supply from snow which -falls- melts in the spring. (SEU)

(3) Applications will be accepted from men and women in any profession or field of study who, at the time of application are British Citizens and who have completed a minimum of four years' secondary schooling </as well as/> the major part of their further education (or equivalent professional experience in lieu of further education) within the United Kingdom. (SEU)

(II) He has achieved the task for which he was nominated - get-ting the budget passed - and therefore he can claim that he has accomplished his mission. (SEU)

(12) Anybody who has had experience of our universities and grammar schools knows how difficult it is to commend the Gospel to young people who have been brought up in irreli-gious homes and regard belief in the supernatural as intel-lectually disreputable. (SEU)

(13) The Govt, are prone to spring decisions on delegations: they announced the 70 m.p.h. limit to a delegation of chief con-stables who had come to the Home Office to discuss the breathalyser. (SEU)

(14) The more that can be charged against the estate the better it will be from the estate duty point of view and as there are one or two other adjustments to be made for bills which came in after we had applied for Probate I will prepare a Corrective Affidavit which should enable us to recover some of the estate duty which has been paid. (SEU)

(16) They have not got the guts to attack the trade unions, whose restrictive practices have prevented price reductions, and are deceiving the public by saying this Bill will reduce the cost of living." (SEU)

(17) This was a generation which grew up in an urban environment and amid many difficulties but did not experience, whilst growing up, the assault of the mass Press as it is known today, of the wireless and television, of the ubiquitous cheap cinemas, and so on. (SEU)

(18) Some pockets have a flap to prevent the batten from coming out, but many have a short length of thin line sewn in which is passed through two holes in the end of the batten, and the ends reef-knotted together. (SEU)

358 Appendix

(19) "He doesn't respect her for it, in fact he despises her, but that's because he despises anyone who earns money, and he treats her badly. (SEU)

(25) Now, he felt like a soldier who had not yet been to France, because of the cane: people looked at him and he wanted to shout at them, Ί have been before, I have been and now I am going back. (SEU)

(26) He created the impression of a man who has been run down by an articulated lorry. (SEU)

(30) In particular, households with insufficient irrigated land or (36) with no land at all had little option but to work for the large

farmers, many of whom were able to reinvest their profits by buying more land from poorer people who had fallen into debt. (SEU)

(35) The second phase, which is the subject of this Report, began with the presentation of the Memorandum on the Reor-ganisation of the University and has been partly occupied in implementing the reforms on which general agreement had been reached. (SEU)

(39) She will talk about the arrival of women undergraduates at Trinity College and about women in publishing, of which she has had long personal experience. (SEU)

(40) P.P.S. We also enclose a copy of an appointment of the Re-ceiver which we have just received. (SEU)

(41) In addition to this it is probably desirable to study the meth-ods of certain historians, sometimes to learn what should be done and sometimes to learn what should not be done and to be shown the pitfalls into which may<sic> intelligent men have fallen. (SEU)


(1 lc) He is a dedicated and highly successful teacher, who has al-(54) ways been willing to meet the challenge of new situations and

methods, and who has always shown himself deeply committed to his work. (SEU)

(31) Then he remembered about Coulter: that he had no family at all, had been brought up in an orphanage, from which he had run away at the age of eleven, to join the circus. (SEU)

Description of elicitation test 359

(47c) Lady Foxglove began to regret that she had ever got rid of Nanny Bill, who had been much cheaper and hadn't expected to have a maid to run and fetch for her the whole time. (SEU)

Description of elicitation test

This section describes the elicitation test and gives a survey of the re-sults on which my observations are based about the relevance of (a) the order in which the clauses are produced, (b) the distinction be-tween a RRC and a NRRC, (c) the sector in which the situations are located.

As will be clear from chapter 7,1 have not been able to support the claims made about the temporal interpretation of a particular sen-tence with a large number of native speakers' judgments. I was in a position to carry out an elicitation test only once; consequently, I have had to make a selection of a number of sentences, the more so as the number of examples submitted to my informants could not ex-ceed 25. The results have already been analysed in detail in chapter 7; I will therefore limit myself to a description of the elicitation test, which was carried out on 3 November 1992 in the English Depart-ment of University College London.193 My informants were 64 first-year English students, whose mother tongue was English. The lectur-ers were asked to read out the sentences to the students, who were given 15 seconds to write down the preferred order and were each time told to turn the page. The following is the set of instructions given to the relevant lecturers and discussed with them before the elicitation test took place.

Instructions to lecturers'.

1. The lecturer explains the purpose of the elicitation test to the stu-dents: it features within research relating to the temporal interpreta-tion of sentences. (The possible influencing factors (e.g. RRC vs. NRRC, tense used, order of the clauses) should not be mentioned to the students.) The booklets are distributed. The students are asked not to read any of the sentences until they are told to do so. 2. The students write down their nationality and mother tongue on the first page.

360 Appendix

3. The lecturer tells the students to read the second page. They should not start the exercise yet. He/She asks if there are any questions and clarifies any issues if necessary. 4. The lecturer tells the students that he/she will read out the sen-tences. After +/- 15 seconds the students will be told to turn the page. 5. 'Turn the page. John will fall in love with the girl who will win a beauty contest.' Lecturer waits for 15 seconds. 'Turn the page. John has married Mary, who has got pregnant.'...

Instructions to informants:

The following are the instructions given to the students on the second page of the booklet containing the sentences.

Each of the sentences below contains two clauses. Put a number in the box under each clause. Write 1 in the box under the clause which indicates the first event. Write 2 in the box under the clause which indicates the second event. If you think both orders are possible, indicate the alternative order beneath the boxes that are given. Put the preferred sequence in the boxes.

Do the exercise as quickly as possible. Do not look back.

Example: She spent the money that she got for the job.

She gave a present to John, who was surprised by this gesture.

She saw the picture which John bought.


Results of elicitation test:

The following are the results of the elicitation test i.e. the preferred chronological readings which the native speakers assigned to the sentences:

1.

3.

4.

7.

John will fall in love with the girl who will win a beauty contest. 1 - 2 1 - 2 , (2 -1 ) 2 - 1 2 - 1 , ( 1 - 2 )

14% 16% 39% 31%

2. John has married Mary, who has got pregnant. 1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

20% 53% 9% 18%

Bill signed a contract with the company for which he worked for 3 years. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

22% 17% 45% 16%

A new mosquito deterrent was invented by the man who went on a holiday to Spain. 1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

3% 16% 48% 33%

5. John fell in love with the girl who won a beauty contest. 1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

6% 11% 64% 19%

6. He will hit John, whom Sue will punch in the stomach. 1 - 2 1 - 2, (2 2 - 1 2 - 1 , ( 1 - 2 )

1) 64% 22% 8% 6%

He signed a contract with IBM, for whom he worked for three years. 1 - 2 : 50%

362 Appendix

11.

1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

19% 20% 11%

8. Bill met a man who robbed a bank. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

16% 16% 40% 28%

9. Sally married the man who won a scholarship.

10.

1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

6% 8% 53% 33%

John, who went on a holiday to Spain, invented a mosquito deterrent. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

45% 36% 11% 8%

Tom met his girlfriend in London, where he lived for three years. 1 - 2 1 - 2, (2 2 - 1 2 - 1 , ( 1 - 2 )

1) 19% 31% 30% 20%

12. The man Sally married won a scholarship. 1 - 2 1 - 2, (2 2 - 1 2 - 1 , ( 1 - 2 )

1) 64% 28% 2% 6%

13. Tom got killed in the place where Bill had his accident. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

8% 20% 48% 24%

14. John has hit Bill, who has hit him. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

20% 22% 36% 22%


15.

19.

21.

John, who lived in London for three years, met his girlfriend in Paris. 1 - 2 1 - 2 , (2 -1 ) 2 - 1 2 - 1 , ( 1 - 2 )

56% 36% 0 % 8%

16. John, whom Sally married, won a scholarship. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

44% 34% 11% 11%

17. Bill was killed in Baker Street, where John had his accident. 1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

9% 11% 46% 34%

18. John fell in love with Mathilda, who won a beauty contest. 1 - 2 1 - 2, (2 2 - 1 2-1,(1

1)

2)

25% 39% 16% 20%

Bill has been killed in Baker Street, where John has had his accident. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

31% 16% 42% 11%

20. John married Mary, who stayed with him for five years. 1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

74% 19% 5% 2%

The man who went on a holiday to Spain invented a new mosquito deterrent. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

55% 34% 5% 6%

22. John married Mary, who took a course in chemistry. 1 - 2 1 - 2, (2 2 - 1

1) 24% 53% 3%

364 Appendix

23. 2 - 1 , ( 1 - 2 ) : 20% John has fallen in love with Mathilda, who has won a beauty contest. 1 - 2 1 - 2, (2 - 1) 2 - 1 2 - 1 , ( 1 - 2 )

23% 33% 17% 27%

24. Bill was killed on the spot where John had his accident. 1 - 2 1 - 2 , ( 2 - 1 ) 2 - 1 2 - 1 , ( 1 - 2 )

3% 9% 72% 16%

Notes on the statistical interpretation of the results:

As was pointed out in chapter 7, the figures have been interpreted statistically in the Mathematics Department of the KULAK194 and it is on the absence/presence of statistical relevance that the claims made are based. The statistical tests applied to the results were aimed at finding out whether (a) a certain percentage (e.g. 69%) is high/low enough to justify the claim that there is a/no preference for a particular option (b) the differences in interpretation brought about by changing the tense, the order of the clauses, the syntactic status (RRC vs. NRRC) or the type of situation (bounded telic vs. unbounded telic) were sta-tistically relevant. I will not go into the actual formulae used to calculate the (non-) rel-evance. Suffice it to say that a normal test for proportions with big, random samples was used; the level of significance being 5.

Notes

1. Declerck, R. 1991a. Tense in English: Its structure and use in discourse. London: Routledge.

2. It is not always possible to reflect the conceptual difference between relations in time (extra-linguistic) and relations in time which are expressed by the use of tense (linguistic) in an adequate terminology. When talking about the order between situations a distinction can be made between chronological order and temporal order. However, chronological does not collocate with just any noun. This is why, for example, the terms temporal relations and temporal information will be used even when talking about extra-linguistic location in time.

3. All the technical terms will be explained in chapter 2. 4. The following are labels which have been used to refer to

situation types: Vendler (1967): state, activity, accomplishment, achievement Heinämäki (1978): durative, non-durative, accomplishment Partee (1984): process, state, achievement, accomplishment Mourelatos (1981), Zydatiss (1976): process, event, state Garey (1957): telic vs. atelic Kenny (1963), Potts (1965): activities, performances, states Schöpf (1984): initial-determinierte Prozesse, initial- und fi-nal-determinierte Prozesse, punktuelle Ereignisse, punktuelle Veränderungen, unquantifizierte gerichtete Prozesse, initial-determinierte gerichtete Prozesse, accomplishment- und achievement Prädikate. Gabbay - Moravcsik (1980): state vs. event Allen (1966), Dahl (1981): bounded vs. nonbounded Cf. Binnick (1991) (chapter 4 and 5) for a history and discus-sion of the concepts aspect, Aktionsart, Aristotelian aspect.

5. Cf. footnotes in Depraetere (1993: 61-101) for references to articles and books in which similar points have been made.

6. Canavan (1983) argues that Vendler's achievements are not always telic because they are not necessarily goal-oriented. However, they do imply a resultant state and should therefore

366 Notes

not be equated with punctual actions: The lightning flashed is punctual though not telic, because it "implies neither a con-tinuing state resulting from the action of flashing nor any prior activity connected with flash" (Canavan 1983: 80). In Canavan's opinion, it cannot be called an achievement either, because it is not associated with a resulting state. In my clas-sification, punctual situations and all achievements are con-sidered to be telic because they imply a natural endpoint be-yond which the situation cannot continue. Smith (1991: 29) also explicitly makes the remark that the endpoint need not be under the control of an agent. My definition of (a)telicity dif-fers from Mommer's (1986) definition of verbs having the CP (culmination point) property ("A CP is a natural endpoint or goal inherent in the structure of a situation type" (Mommer 1986: 59) in that I also classify situations with an intended endpoint, which implies agentivity, as telic situations. Cf. Dahl (1981: 83-87) for notes on the lack of definitional accu-racy of terminal point.

7. Declerck defines boundedness as follows: a sentence is bounded when it "represent[s] a situation as involving a ter-minal point which has to be reached if the situation is to be what it is claimed to be" (Declerck 1991a: 120, 262), a situa-tion is bounded if "it is represented as reaching a terminal point" (Declerck 1991a: 261). I prefer the latter definition to the former because the definition on page 120 is very closely related to that of (a)telicity. It is important to be aware of the distinction between having an inherent or intended terminal point (telicity) and having reached a terminal point (boundedness).

8. Not everyone agrees that punctual situations are bounded. I would argue that non-progressive sentences with a punctual situation are inherently bounded as the beginning and end of the situation coincide. When a punctual proposition is used in a progressive sentence, its punctual character disappears. Consequently, the situation is no longer automatically bounded. Cf. e.g. Allen (1966: 199), Comrie (1976: 42-44), Hirtle (1987: 89), Smith (1991: 29-30), Zydatiss (1976: 107) for observations on whether punctual situations should be considered as having any duration at all.

9. As will be pointed out in chapter 2 (section 2.1.5.2.), a dis-tinction should be made between sentences with a present

Notes 367

perfect which gets an indefinite interpretation (She has dropped the vase with the flowers on the floor.) and sentences with a present perfect which gets a continuative interpretation (She has been complaining about the bad weather since she arrived.) From now on, the terms continuative perfect and in-definite perfect will be used to refer to what strictly speaking should be called the continuative/indefinite use of the present perfect.

10. Bartsch (1988/89: 134-136) also touches upon a distinction similar to that between (un)boundedness and (a)telicity.

11. This comparison exemplifies the terminological confusion which characterizes the Aktionsart discussions: Moens' use of the term bounded could either correspond to what we call bounded or telic. His use of the term telic diverges from ours in that he reserves it for situations that have consequences, i.e. it indicates a subclass of what we call telic situations: on our approach, situations need not necessarily bring about a change of state (e.g. The light flashed) in order to be called telic.

12. Exceptional contexts allow the use of a/or-adverbial:

(i) The Atomic bomb exploded for an hour. (Descles 1989: 174)

(ii) The firecracker exploded for five minutes. (Smith 1991: 62)

13. Allen (1966) makes a distinction between bounded and un-bounded nouns on the basis of this characteristic:

The difference between nouns like cake, egg and apple and nouns like pencil and car seems to be that the for-mer refer to Entities that are "suffusive": a cake is a cake all the way through - that is, any part of a cake, no matter how small, is still cake. The same is true of ap-ple·, an apple is apple throughout. No matter how small a piece one cuts off, one still has apple. But this is not true of a pencil: a very small piece of a pencil may be just wood or lead - the label "pencil" is applied only to a specific kind of unit, not to every part of the unit. Non-suffusive nouns always occur as the nuclei of bounded nominals; suffusive nouns may occur as the

368 Notes

nuclei of either bounded or non-bounded nominals. (Allen 1966: 193-194)

The distinction Allen makes should be applied to mass nouns: an apple or the apple is not homogeneous, i.e. every part of an apple is not an apple. Cf. Smith (1991: 31-33). Mourelatos (1981) draws a further parallel with nouns: a telic sentence can be rephrased as a count NP, an atelic sentence is rephrased as a mass NP:

(i) Vesuvius erupted three times. There were three eruptions of Vesuvius. (Mourelatos 1981:204)

(ii) Mary capsized the boat. There was a capsizing of the boat by Mary, (ibid.)

(iii) John pushed the cart for hours. (Mourelatos 1981: 206) For hours there was pushing of the cart by John.

(iv) Jones was painting the Nativity, (ibid.) There was painting of the Nativity by Jones.

14. This has often resulted in the claim that (a)telicity (i.e. the varying labels under which this characteristic has been dealt with) is a property of predications (Allen 1966: 198) or sen-tences (Declerck 1979: 764, Dowty 1986: 39, 43, Jackendoff 1991: 38-39, Moens 1987: 130, Smith 1991: 7, Vlach 1981: 272, Zydatiss 1976: 43) and not of verbs or verb phrases.

15. Adverbial particles may also influence the classification: cf. e.g. Brinton (1985), (1988: 163-184), Duskova (1974: 71), Kittredge (1969: 75-83), McCoard (1978: 142), Smith (1991: 73). So do verbal suffixes, cf. e.g. Binnick (1991: 141-142).

16. I am concentrating on sentences without temporal adverbials. 17. As will be pointed out below (cf. section 1.3.4.3.), bounded

telic non-progressive past tense sentences imply that the situ-ation is over: * I read the book, and I still do.

18. TS stands for time of the situation, i.e. "that subinterval of the full situation (possibly the whole situation) that is located in time by the sentence, i.e. that is represented as simultaneous with the situation-TO" (Declerck 1991a: 18). In fact, this no-tion should not be defined as referring to a time-interval but to a situation as such. Declerck has opted to use the notion STO as a means to indicate that every situation is simulta-

Notes 369

neous with a time, in his research following Declerck (1991a).

19. Cf. chapter 2, section 2.1.3.: "Roughly speaking, it [sit-TO] is the time with which the situation is represented as simulta-neous" (Declerck 1991a: 18).

20. Cf. chapter 2, section 2.1.3: TO, is the basic TO, i.e. "that TO in the structure of a tense from which the temporal relations expressed by the tense begin to be computed" (Declerck 1991a: 252).

21. Jackendoffs (1991) point in connection with (un)bounded NPs captures fairly well the fact that the point of view from which one assesses something influences its classification:

A speaker uses a -b[ounded] constituent [NP] to refer to an entity whose boundaries are not in view or not of concern; one can think of the boundaries as outside the current field of view. This does not entail that the entity is unbounded in space or time; it is just that we can't see the boundaries from the present vantage point. (Jackendoff 1991: 19)

22. The statements also apply to the past perfect. I concentrate here on the present perfect because this gives me the oppor-tunity to make a comparison with the past tense.

23. The confusion in connection with the different nature of (un)boundedness and (a)telicity is clear from the following statement from Mittwoch (1988): "This raises the question, what sort of Aktionsart [(a)telicity] does the perfect operator [influences (un)boundedness] define?" (Mittwoch 1988: 214).

24. If John lived in London only for a subpart of 1983, Β would include the other places where John lived in his answer to A's question.

25. This is why I do not agree with Woisetschläger (1976), who claims that I read from 10 p.m. to 11 p.m. does not "leave[s] open the possibility that some reading preceded or followed the time interval specified" (Woisetschläger 1976: 29). Tak-ing into account the above discussion, it might be argued that depending on whether the adverbial provides given or new information in I read from 10 p.m. to 11 p.m. the situation of

370 Notes

reading is likely/not likely to continue after the period of time indicated.

26. The fact that the situation continues after TE may be contra-dicted by the linguistic context which follows. Cf. e.g. Dowty (1977: 59), König (1980: 273).

27. Even if Β replied He lived there for example in 1983, the sentence would be classified as bounded. The addition of for example indicates that the hearer knows John lived there at other times as well but he does not know when or he does not want to indicate the other periods. The difference with the re-ply in (62) is that it is more likely that John still lives in Lon-don at t0, especially if the hearer is trying to hide information.

28. The following are exceptional interpretations in which the punctual adverbial does have a bounding effect: if Β noticed John was in the library at 5 o'clock while Β walked past the open library door, one might say that the situations in (65) and (66) are bounded. Another possible (but rather unlikely) bounded interpretation of (65) and (66) is that John was working in the garden/library for such an extremely limited amount of time that it warrants the conclusion that he only did so for a second, i.e. at 5 o'clock.

29. Remember that this does not necessarily mean that the sen-tence is bounded. In order to determine the (un)boundedness of a past tense sentence, one has to consider the situation from the point of view of TOsit (cf. section 1.3.3.1.).

30. "Make your contribution as informative as is required (for the current purposes of the exchange)" (Grice 1975: 45).

31. Quirk et al. (1985) reject the possibility of a past situation still holding at t0, as they claim that one of the two features of meaning of the past tense is that "the event/state must have taken place in the past, with a gap between its completion and the present moment" (Quirk et al. 1985: 183).

32. Cf. also Bache (1985: 268), Dowty (1977), Moens (1987: 91), Mommer (1986: 87-88).

33. Cf. e.g. Bhat (1977: 118), Dowty (1986: 44), Moens (1987: 75), Verkuyl (1972), Zydatiss (1976: 82). Heinämäki (1978) correctly points out that negated telic clauses can take phrases such as in a year, whereas negated atelic clauses cannot:

Notes 371

(i) * Paul did not live in Austin in a year. (Heinämäki 1978: 11)

(ii) Vapply did not solve the problem in an hour, (ibid.) (iii) The answer did not come in a week, (ibid.)

34. Cf. e.g. Bennett - Partee (1978: 14), Binnick (1991: 139), Couper-Kuhlen (1987: 15), (1989b: 16-17), Dowty (1986: 38, 50-51), Jacobson (1980: 54-55), Langacker (1982: 274-275), Mittwoch (1980: 207), Moens (1987: 87), Mommer (1986: 26, 68-73), Mourelatos (1981: 196), Schöpf (1984: 105-106, 242), (1991: 238), Smith (1991: 22, 63, 74). Examples of this type constitute a separate class in the taxonomies of Mommer (1986: 68-73) and Schöpf (1984: 105).

35. This example can also be classified as bounded to the right if it is interpreted as follows: John has agreed to pick up Mary at the station and at five o'clock he arrives at the station.

36. Boogaart (1995) reserves the term (un)boundedness for the cases in which the context imposes temporal boundaries.

37. The term situation is used to refer to the "various 'things' that can be expressed in a clause (viz. actions, processes, states, events ...)" (Declerck 1991a: 14).

38. See Declerck (1991a: 15-16) for exceptions to this general rule.

39. Other linguists have used the label deictic or primary tenses. 40. Relative tenses have been variously called anaphoric, sec-

ondary or relational tenses. 41. I assume that the past tense is a relative tense, although, theo-

retically, it could also be an absolute tense establishing a new domain which is W-simultaneous with the main clause situa-tion. Cf. section 2.1.5.1.

42. In Declerck (1991a), no terminological distinction is made between W-anterior/W-simultaneous/W-posterior and anter-ior/simultaneous/posterior or between temporal and chrono-logical.

43. The label conditional tense, although widely used, is some-what misleading in that it suggests that the use of would al-ways implies conditional meaning. This is not the case, wit-ness the following example:

372 Notes

(i) He left. He would never come back.

In this example, posteriority in the past sector is expressed without any conditional meaning. A comparison with German illustrates the two possible uses of the "conditional" tense more clearly, as würde with conditional meaning is inter-changeable with Konjunkiv II (cf. (ii)) whereas Futur der Vergangenheit würde is not (cf. (iii)):

(ii) Ich würde wiederkommen (käme wieder), wenn ich könnte.

(iii) Er wusste nicht, dass er niemal wiederkommen würde (* wiederkäme).

For lack of another, more adequate label, I will stick to the use of the term conditional tense, bearing in mind that it does not necessarily imply conditional meaning.

44. From now on, I will use the terms indefinite/continuative perfect to refer to what strictly speaking should be called the indefinite/continuative use of the perfect.

45. Other examples illustrating the shift of perspective are:

(i) The train leaves at 5 p.m. (shift from the post-present to the present sector)

(ii) I am leaving tomorrow, (shift from the post-present to the present sector)

(iii) The 10 o'clock news says there is going to be a storm, (shift from the past to the present sector)

(iv) I hear you have been promoted, (shift from the pre-present sector to the present sector)

Strictly speaking, had felt in (42b) may also be interpreted as anterior to had left, in which case there is direct binding. Would be willing in (45a) could also be interpreted as posterior to would go, in which case there is direct binding. The concepts referred to in the text will all be defined in the following chapters. Declerck himself (1991a: 14-156) gives fairly comprehensive references.

46.

47.

48.

49.

Notes 373

50. As pointed out before, the terms W(orld)-anterior, W(orld)-posterior and W(orld)-simultaneous are used to refer to the "real-world" temporal relations between the actual situations. In other words, they do not refer to the grammaticalization of these temporal relations by means of special verb forms. The labels anterior, posterior and simultaneous are used to indicate the latter phenomenon. In order to indicate the sequence in which the clauses are reported (i.e. the RC/SUPC is given first), the paraphrase "the RC/SUPC comes be-fore/after the RC" or "the RC/SUPC is mentioned first/second" will be used. In this case, the syntactic unit clause is under discussion and not the situation referred to in the clause.

51. The order in which the clauses are mentioned is important when the SUPC and the NRRC both refer to bounded (especially telic) situations (cf. section 3.2.2.2. and chapter 7).

52. The form given in brackets is the adapted verb form which I have added. When alternatives are given in the original text, a slash is put between them: e.g. verb/verb.

53. As will be pointed out below (cf. examples in (40)), the past tense, unlike the past perfect, can also be understood as referring to a series of events.

54. It would be more accurate to say "the situation-TO with which the RC situation is simultaneous" (cf. chapter 2, section 2.1.3.). For reasons of simplicity, this more accurate though long-winded paraphrase will not be used.

55. Cf. e.g. Allen (1966: 169-170), Bache (1985: 205-206), Berezovsky (1980: 87), Breithutovä (1968: 133), Couper-Kuhlen (1987: 25), Curme (1931: 361), Davidsen-Nielsen (1990: 134-135), Declerck (1991a: 183, 368), (1991b: 119), Dubos (1985/86: 159), Fenn (1987: 219), Fijn van Draat (1912: 155, 160), Hamann (1989: 101), Hüllen (1989: 621), Jespersen (1961, IV: 74-75, 83), (1962: 247), Kruisinga (1931, II-1: 409), (1932, II-3: 437), Leech (1971: 48), McCoard (1978: 186), McGhie (1978), Poutsma (1926: 277), Schöpf (1985: 90), Smith (1976: 7) Steedman (1982: 145-146), Tregidgo (1974: 100-101), Vasudeva (1971: 104).

Notes

Hamann ( 1 9 8 9 ) is more explicit than most other linguists. She claims that the past perfect is redundant in after-clauses which refer to "punctual events, quantified processes, quanti-fied changes including accomplishments and also achieve-ments. It does not hold for event notions which provide only ^[initial] o r n o P o i n t a t a11" ( 1 9 8 9 : 1 0 1) · Quirk et al. ( 1 9 8 5 ) mistakenly believe that the shift of domain is never possible in noun clauses: "There is one construction in which the past perfective clearly could not be replaced, as a means of refer-ring to past time, by the simple past. This is an indirect speech construction, in which the past perfective indicates backshift to the more remote past.

(i) I told her the parcel had not arrived" (Quirk et al 1985: 197).

Hill (1958: 214) writes: "Curiously, there are no good exam-ples of sentences in which either the had phrase or a simple past is required, and all situations in which one is possible can employ the other." Diver (1963) correctly rejects Hill's claim by giving the following example, in which the past per-fect cannot be replaced by a simple past:

(ii) He approached the bridge with some caution, not en-tirely convinced of its safety, even though engineers had tested it thoroughly only two days before. (Diver 1963: 150)

As will become clear in the course of the discussion that fol-lows, there are basically two reasons why the past perfect has to be used: 1. There is a tendency for a verb to copy the structure of the adverbial {two days before). 2. The use of the past tense turns the past perfect bounded sentence into an unbounded one. This results in changing the temporal order between the situations, i.e. the past tense ex-presses a relationship of simultaneity between the SC situ-ation and the SUPC situation (cf. section 3.2.1.). In the ex-ample, this would bring about a clash between the chrono-logical order suggested by the adverb and that suggested by the verb form.

Notes 375

57. The articles and books which touch upon the past perfect (cf. e.g. Bertinetto 1986, Canavan 1983: 143-151, Chalker 1985: 48-49, Dahl 1985: 144-150, Fenn 1987: 217-222, Hopper 1979b: 40, Hornstein 1977: 531, Huddleston 1969: 785-786, 792-793, Lo Cascio - Rohrer 1986, McCawley 1973: 262, McCoard 1978, Mittwoch 1988: 216, Ota 1963: 46-47, Salkie 1987: 92-94, 1989, Schöpf 1984: 331-332, Smith 1978, Van-neck 1958: 239, Wachtel 1982: 339) centre around the fol-lowing issues: how is the binding TO established?, the inter-action of the past perfect and adverbials, the (dis)similarity between the present perfect and the past perfect, the counter-factual reading of perfect before-clauses and the function of the past perfect in narrative style. McGhie (1978) also de-votes an article to the past perfect but the scope of her discus-sion is rather limited. She claims that the use of the past per-fect is optional in RCs in "just over half of the occurrences" (McGhie 1978: 71). She neither indicates how many sen-tences her corpus consists of nor gives many examples.

58. Cf. e.g. Hopper (1979b: 40), McGhie (1978: 73). The fol-lowing quotation also illustrates the type of vague comment which is mostly given: "We choose the past perfect when we are establishing a significant relationship between two differ-ent past times. (This often occurs in association with after-clauses, and before-clauses, and with adverbial expressions like the day before and the previous year. The past perfect is also common in relative clauses, and in novelistic flashbacks to events preliminary to the main narrative" (Tregidgo 1974: 101). (italics mine)

59. In fact, a third possibility might be added, i.e. the speaker may focus on a past part of a situation which in real life leads up to the SUPC situation:

(i) A deformed shoplifter caught by an Ealing store detective claimed he had received his disability in an Oxford street bombing which killed his wife. In fact, South African-born Indian George Benjamin, 49, who had been born without most of his fingers, had already used his deformity to defraud insurance companies of thousands of pounds. (SEU)

376 Notes

This sentence could be used even if Benjamin was still using his deformity to defraud insurance companies at the time when he was caught. In (i), the speaker focuses on a part of the situation lying before the moment when the shoplifter is caught.

60. The original examples on which the adapted versions are based are listed in the appendix.

61. For the time being, it is assumed that the past tense in the RRC is a relative tense, whereas the past tense in the NRRC is an absolute tense. This issue will be taken up in section 3.3. of this chapter. Huddleston (1969) touches upon this issue when he says that was in He didn't go because he was ill may be interpreted as a deictic past or as a "present in the past": "It would be a mistake to think that these amount to the same thing: in the former analysis the temporal relation between the two processes is made explicit, whereas in the second analysis each process has an independent temporal relation to the axis defined by the speech act. The first would more readily enable us to account for the unacceptability of (105) * He didn't go last month because he was ill three

days later." (Huddleston 1969: 796)

62. a. For the time being, the possibility of sloppy simultaneity is disregarded, i.e. a relative tense expressing simultaneity is used to refer to situations which follow each other:

(i) A Left-wing student seized the microphone but was quickly cut off by an official who cut the cord. (SEU)

Strictly speaking, the cutting of the cord precedes the being cut off in (i). However, the speaker concentrates on the fact that there is a logical relation between the two events, which closely follow each other. The speaker "treats the two situations as belonging to the same 'occasion' and hence as falling within the same time interval" (Declerck 1991a: 42). The following example is similar:

(ii) The picture I gleaned from talks I had with lawyers during my first visit to Ireland was one of complete travesty of justice: people were tried by special courts in camera, without a jury and were not even allowed to

Notes 377

hear the testimony of the prosecution witnesses. (Kupetz 1981: 328)

It is only after a talk with someone that one can form a picture of a situation. In this example, however, the two situations are represented as happening simultaneously, b. Examples with indirect binding will not be considered in this section either. In (iii), a relative tense expressing anteriority is used to refer to a situation which is W-simultaneous with another anterior situation. This topic will be dealt with in chapter 8:

(iii) I met a man who had lived with people who had gradually fallen into debt.

The second RC expresses anteriority with respect to the matrix clause situation and not with respect to the SUPC situation, with which it is W-simultaneous.

63. With a view to retaining the anonymity of the persons involved, the dates have been replaced by 2222 and names have been replaced by Name or Word in all the examples from the SEU. All these "anonymity labels" have been put in square brackets.

64. Costa (1972: 43) gives the following example: Sue suspected that Bill left/had left before the police arrived. Riddle (1975: If in Binnick 1991: 367) observes that the two sentences are not completely synonymous: "Bill left before the police arrived is ambiguous, one reading being that Bill intention-ally left so as to avoid the police (in fact they might never have arrived); but Bill had left before the police arrived lacks this reading" (Binnick 1991: 367). Riddle is right. However, the question with which we are concerned here is whether there is a difference in the chronological order established by the past tense and the past perfect. The answer is no.

65. The difference between zero-telic and purely telic verbs recalls a distinction between English and Dutch pointed out by De Vuyst (1985) in connection with sentences such as

(i) a. Jan liep naar huis toen de bom explodeerde. (De Vuyst 1985: 138)

b. John went home when the bomb exploded, (ibid.)

378 Notes

De Vuyst correctly comments that the English sentence can only refer to a sequence of events (explosion - John goes home) whereas its Dutch equivalent can also get the interpretation that John was going home when the bomb exploded. (Schlyter (1979: 117) makes a similar remark in connection with the German sentence Urn 8 Uhr ging Oskar, which either means Oskar was walking at 8 o'clock or Oskar left at 8 o'clock.) In the light of the distinction between zero-telic sentences and purely telic sentences, de Yuyst's observation must be slightly modified, as it appears that some English VPs also allow the "Dutch double interpretation":

(ii) When he spoke to the chairman, the latter knew he was trying to make fun of the organization.

Apart from indicating to a sequence of events (speaking -knowing), the sentence can also be interpreted as referring to two simultaneous situations.

66. The past perfect has the implicature that the effect of the situation referred to still holds at T0 2 (cf. Declerck 1991a: 343).

67. The use of the past is questionable in the non-habitual reading of theRC.

68. Cf. e.g. Adelaar - Lo Cascio (1986: 275-276), Bertinetto (1986: 53), Comrie (1986a: 299), Declerck (1991a: 289), Tregidgo (1979: 193) for comments on whether or not the structure of adverbs (i.e. whether they are anchored to t0 or to some future or past time as TO) should be matched by a tense with the same structure.

69. Cf. Lo Cascio - Rohrer (1986) summarized in Adelaar - Lo Cascio (1986): "The anaphoric verbal tense of a complement clause, especially if controlled by a verb of saying is always evaluated with respect to the time denoted by its matrix" (Adelaar - Lo Cascio 1986: 256). Cf. also Bertinetto (1986): "Example (29a) [* When I came to town yesterday, I heard that John broke his nose the day before] is quite odd: I heard that works easily as an R with regard to a previous E, and this rules out as ill-formed all sentences containing a simple tense, which does not call for an intrinsic temporal reference" (Bertinetto 1986: 53).

Notes 379

70. The presence of a verb of prepositional attitude as such does not always lead to temporal subordination of the embedded clause. The following linguists have pointed out that a distinction should be made between different types of verbs. These classes behave differently as regards the requirement of temporal subordination: Costa (1972: 46) distinguishes between "Α-verbs" (factives and entailment verbs and a few non-factive verbs of linguistic communication) and "B-verbs" (manner verbs of saying and non-factive verbs of linguistic communication), the latter re-quiring temporal subordination, the former group not necessarily. Cf. Riddle (1978: 47-51) for counterexamples. Declerck (1991a: 185-186) gives an example in which believe, unlike say, requires temporal subordination to refer to a W-simultaneous situation. A similar observation is made in Tregidgo (1979): "Past-subordination is particularly likely following verbs of believing e.g. know and think. (...) In cases like Ί knew it was you.1 on opening the door to a visitor, we cannot change was into is" (Tregidgo 1979: 193). Leech (1971) also points out that backshifting applies "more regularly with verbs such as knew, think, realize, forget than with verbs such as say and tell" (Leech 1971: 107). The examples he gives contain W-simultaneous SC situations. Cf. also Rigter (1982: 114-115). In Declerck's opinion "it is never possible to use an absolute tense to report a past expectation concerning the future" (1991a: 186), no matter whether the introductory verb is say, knew or believe. Cf. also Huddleston (1969: 795) and McCoard (1978: 15).

71. A speaker conforming to the rules of commmunication would probably use a continuous tense to indicate a relationship of simultaneity.

72. I will concentrate on situations that are W-anterior to the situation establishing the intensional domain. The remarks made about domains of this kind often relate to the requirement of temporal subordination to refer to W-posterior situations. Cf. e.g. Comrie (1986a: 281-282), Declerck (1991a: 186), Heny (1982: 126).

73. Cf. Declerck (1991a: 91), Ehrlich (1990), Huddleston (1969: 793), Rohrer (1986: 81).

380 Notes

1A. Ehrlich (1990: 60) correctly points out that the use of the past perfect does not necessarily imply that there is a sequence of RST.

75. The circle indicates the set of situations that belong to the same domain.

76. Salkie (1989) also questions some of the points made by Bouscaren et al.

77. Strictly speaking, atelic bounded situations should also be mentioned in this respect. However, in this section, I will concentrate on examples in which it is the perfect tense which establishes a bounded reading; the bounded character is not supported by the presence of an adverbial with a bounding effect. Accordingly, the corresponding atelic past tense examples will all be unbounded, as there are no adverbials which impose temporal boundaries. Atelic bounded examples will be dealt with in chapter 7 (section 7.5.1.).

78. The students and they are coreferential in (65b). 79. The term is to be understood to mean an argument induced

with a view to retaining a parallelism between the expression of anteriority in the past sector and the expression of simultaneity in the past sector.

80. To express the continuative meaning which the present perfect can have in English, Dutch in most cases uses the simple present in combination with words that convey continuative meaning:

(i) Ik woon hier al sinds 1982. * Ik heb hier al sinds 1982 gewoond.

81. The Dutch present perfect refers to a bounded situation that lies completely before t0 unless it is accompanied by an adverbial that indicates that the situation leads up to t0:

(i) Ik heb daar nooit gewoond. (I have never lived there.)

(ii) Ik heb altijd van je gehouden. (I have always loved you.)

82. De Vuyst (1982) argues that the Dutch past tense is transparent. The Dutch present perfect is opaque, which

Notes 381

means that "binnen een opaak beschreven situatie kan verder niets worden aangeduid" (De Vuyst 1982: 103) [nothing else can be referred to within a situation that is described opaquely], witness ? Hij heeft in de tuin gewerkt toen de telefoon ging (De Vuyst 1982: 102) [? He has worked in the garden when the telephone rang]. "Opgrond van de karakteriseringen die werden gegeven van opaak en transparant valt te verwachten dat de Engelse simple past van een state equivalent zal zijn aan de Nederlands o.v.t. van een state, aangezien beide transparante beschrijvingen zijn. Die verwachting komt ook uit. Het Nederlandse equivalent van He knew English is Hij kende Engels en niet Hij heeft Engels gekend, zoals blijkt uit:

(20) The man I talked to knew English. De man met wie ik sprak kende Engels. * ? De man met wie ik sprak heeft Engels gekend." (De Vuyst 1982: 103)

[On the basis of the way in which opaque and transparent have been characterised it is expected that the English simple past of a state will be equivalent to the Dutch "onvoltooid verleden tijd" of a state, as both forms refer to transparent situations. This expectation is borne out by the facts. The Dutch equivalent of He knew English is He knew English and not He has known English as is clear from:

(20) The man I talked to knew English. The man I talked to knew English. * ? The man I spoke to has known English."]

83. As pointed out in footnote 43, the term conditional is somewhat misleading because it suggests that the context in which the verb form is used has conditional meaning. However, as it is widely used and as there is no better alternative, I will continue to use this label.

84. As Declerck (1991a) points out on page 27-28, there are other forms which can be used to express posteriority in a past domain, e.g. be going to, was to. Like Declerck, I will concentrate on the use of the conditional tense as a means to indicate that a situation is posterior to another past sector situation.

382 Notes

85. "Make your contribution as informative as is required (for the current purposes of the exchange)" (Grice 1975: 45).

86. As will be explained in section 3.4.2., the reasons why the past tense cannot be used may be different for RRCs and NRRCs.

87. The past tense RC situation in (104b) is zero-telic because it is not clear whether the focus is on the activity phase of the situation (atelic) or on the situation as a whole (telic).

88. In Declerck's opinion "it is never possible to use an absolute tense to report a past expectation concerning the future" (1991a: 186), no matter whether the introductory verb is say, knew or believe. McCoard (1978: 15) points out that believe, unlike say, requires temporal subordination when the SC situation is W-posterior.

89. Cf. Vetter (1973: 107). 90. It follows from the observations made in section 3.4.2.1. that,

apart from telic NRRC examples which refer to a sequence, the examples discussed in this section are all sentences in which the order of the RC and the SUPC situation is made clear by the semantics of the clauses and/or pragmatic knowledge.

91. Haegeman observes (1989: 313) that the sentence The Queen would arrive three hours later is false in the event that the Queen did not arrive. Declerck (1991a) shares her opinion: "The absence of a superordinate clause can only mean that the situation is represented as it is viewed by the speaker, i.e. as a factual past situation which is seen as posterior to another past situation" (Declerck 1991a: 383). This explains why would cannot replace be going to in the following example:

(i) Bill was going to leave / * would leave when Mary came in and forced him to stay at home. (Gabbay -Rohrer 1978: 108)

Gabbay - Rohrer (1978: 109) point out that would is not acceptable in the following example either:

Notes 383

(ii) John said that yesterday at noon Bill was going to I * would leave three hours later. (Gabbay - Rohrer 1978: 109)

Although Bill may actually have left at 3 o'clock, there is ref-erence to his intention to do so, which requires the use of was going to.

92. Be nominated is zero-telic because it can refer to the act of nominating (telic) as well as to the state of being nominated (atelic).

93. Although his claims are about RCs in general, the exclusive use of RRC examples probably indicates that he is only con-cerned with RRCs.

94. Like Declerck, Hornstein believes that the tense used to tem-porally subordinate a W-posterior situation to a present per-fect matrix clause is the conditional (Hornstein 1990: 221).

95. Some linguists reject the view that the present perfect is a tense and deal with it under the label aspect (cf. e.g. Brecht 1974: 501, Charleston 1955: 267, Comrie 1976: 52, Dillon 1973: 271, Hirtle 1967, 1975, Huddleston 1984: 164, Leech 1971: 35, Mommer 1986: 246, Quirk et al. 1985: 189, Riviere 1977: 235), while others qualify the present perfect as a tense (cf. e.g. Curme 1931: 358, Declerck 1991a: 319-322, Jespersen 1961, IV, Kaiuza 1984:323, Poutsma 1928, Schöpf 1984: 20, Zydatiss 1978: 339). A third group of linguists do not deny its link with tense but all the same refuse to use this label; they use the label phase (Davidsen-Nielsen 1990: 68-69, Feigenbaum 1981: 393, Huddleston 1969: 777, Joos 1964: 138, Palmer 1974: 34), status (Bauer 1970: 197) or in-clusion (McCoard 1978: 152). Fenn (1987) writes: "the recognition that the perfect is a verbal category separate from tense and aspect (...) [is] an important one, and deserves em-phasis" (1987: 249), but he cannot think of an adequate label. Comrie (1976) does not believe that the present perfect quali-fies as "normal" aspect: "Given the traditional terminology in which the perfect is listed as an aspect, it seems most conve-nient to deal with the perfect in a book on aspect, while bearing in mind continually that it is an aspect in a rather dif-ferent sense from the other aspects considered so far" (Comrie 1976: 52). Cf. also Huddleston (1984): "Clearly,

384 Notes

however, the system of perfect aspect is very much closer to a tense system than is that of progressive aspect" (Huddleston 1984: 164). It should also be pointed out that there is a differ-ence in how the quoted linguists define the notion phase. Palmer (1965: 61) does not explicitly use the label phase. He states: "Tense and perfect/non-perfect are initially best han-dled together, in order to make the point, not usually made, that both are essentially concerned with time relations." Palmer (1974: 34) introduces the notion phase in his discus-sion of the present perfect.

96. Cf. e.g. Comrie (1976: 55), Dowty (1982: 40-41), Huddleston (1969: 795), (1984: 162), Ladusaw (1977: 95-96), Leech (1969: 143-144), Rigter (1982: 117), Thomson and Martinet (1980: 157-158).

97. This principle runs counter to a claim made by Huddleston: "Because the present is included in this way [through the use of a present tense form] the perfect does not trigger back-shifting - cf. * Kim has said the match started tomorrow" (Huddleston 1984: 162). Cf. also Huddleston (1969: 795), Nehls (1974: 20). Rigter also argues that there is no temporal subordination in noun clauses introduced by the present per-fect, although his view is less absolute than Huddleston's: "Embeddings of present tenses under clauses with a present perfect would seem to be more frequent than embeddings of the present tenses under clauses with a past tense. This is why a more detailed investigation into the present perfect and em-beddings under the present perfect is required" (Rigter 1982: 117). Rohrer (1986: 87-88) makes the same point for French noun clauses introduced by a present perfect tense. Dinsmore (1981) shares Declerck's view. He explains the un-acceptability of the second sentence in the example Harry has been in Borneo. * He has enjoyed it a lot. as follows: "At the point at which it [the second sentence] is uttered the oc-currence of the event of Harry's being in Borneo is presup-posed and the event described by the second sentence must be simultaneous to the event of Harry's being in Borneo" (Dinsmore 1981: 488). Cf. also Comrie (1976: 55), Fenn (1987: 180), Hornstein (1990: 220, 221), Inoue (1979: 586), Leech (1969: 143-144), Penhalluriack (1981: 216), Thomson and Martinet (1980: 157-158).

Notes 385

98. Rohrer (1986) makes the same remark in connection with French sentences such as:

(i) Le docteur a constate que Rummenigge s'est foule le pied et ne pourra pas jouer ä Hamburg ce soir. (Rohrer 1986: 88)

(ii) Vous n'avez pas compris (ne compreniez) pas que les decouvertes atomiques ont change la face du monde. (ibid.)

99. As will be pointed out below (cf. section 4.2.1.1.), the native speakers consulted on this point believe that the present per-fect stresses the (large) number of occasions on which the sit-uation held. Cf. Zandvoort (1932: 12) for examples of this type in which a present perfect is used.

100. Huddleston (1969: 784) makes the same point in connection with Palmer's example Every time I've seen them, they've been swimming. Cf. also Bertinetto (1986: 73), Palmer (1965: 102-103), (1974: 78-79).

101. Cf. Rigter (1980): "There is no question of SOT [sequence of tense] in (12) [Mary has known for a week now that Charles stayed here for two days last summer]" (Rigter 1980: 418).

102. The W-anteriority relation I will concentrate on is that of the completely before type and not of the before and up to type (cf. chapter 3, section 3.1.).

103. The present perfect has the implicature that the situation brought about by the present perfect sentence still holds at t0. Cf. Declerck (1991a: 343).

104. Maxim of Quantity (part a): "Make your contribution as in-formative as is required (for the current purposes of the ex-change)" (Grice 1975: 45).

105. The original examples on which the sentences are based are listed in the appendix.

106. That it is the use of the perfect which makes one think of the (large) number of subsituations is also clear from the follow-ing sentence from an article by Harkness (1987): "Any American informants we have had have not agreed with Quirk et al" (1985: 106).

386 Notes

107. Vermant (1983) is -as far as I know- the only linguist who has investigated the collocation of the present perfect with personal pronouns systematically. 41.5% of his present perfect examples have a first person subject. He correctly points out: "There is no real correspondence between verb form and (...) personal pronouns: (...) "you" may very well occur with a present perfect, just as (...) "I" [is] compatible with a preterite. There is however a clear tendency in our corpus in favour of (...) "I" in collocation with the present perfect" (Vermant 1983: 89). The statistical study by Ver-mant also reveals a preference of the present perfect for deictic elements such as this, here, ...: 88.2% of the present perfect sentences with a demonstrative element contain this, these, and here. This does not imply, however, that that, those and there never occur in present perfect sentences or that the former set of words never occur in a past sentence:

(i) That's the first good thing that's happened to us for weeks. (Vermant 1983: 89)

(ii) All this happened long ago. (Vermant 1983: 89)

Cf. also Guenther (1977): "The use of indexicals (/, here, now, etc.) is particularly appropriate in connection with the present perfect, since they "refer" to aspects of the actual point of speech" (Guenther 1977: 94-95). McCoard (1978) makes a similar remark: "We should not expect demonstrative adverbials [this time, that time, this winter, ...] to match neatly with the time-frame of the verb in all cases — (...) An overall correlation remains, however" (McCoard 1978: 138). (Cf. also McCoard 1978: 83).

108. I concentrate on the indefinite interpretation of the present perfect.

109. As will be pointed out below, the chronological order established between the SUPC situation and the RRC situation differs from that established between the SUPC situation and the NRRC situation when the situation is purely telic (cf. section 4.4.2. and chapter 7).

110. It has been pointed out by e.g. Bach (1980: 30), McCawley (1973: 269), Moens (1987: 21), Papp (1985: 107-108), Penhalluriack (1981: 44), Schöpf (1984: 318), (1987: 186), (1991: 242) that the past tense can only be used provided a

Notes 387

definite past time adverb or contextually determined past time is available.

111. Cf. Rohrer (1986: 88) for a similar remark in connection with French.

112. Examples (42) and (43) differ from the others in this chapter in that the anteriority relation is one of the before and up to type.

113. Definite time adverbials are acceptable when the sentence allows a repetitive interpretation or when have is heavily stressed. Cf. e.g. Comrie (1976: 54), (1985: 78-79), Dinsmore (1981: 492), Kittredge (1969: 12), McCoard (1978: 80), Smith (1976: 10-11), N.V. Smith (1981: 261). Davidsen-Nielsen (1990: 214) lists some examples with a non-repetitive present perfect tense used in combination with a definite time adverbial.

114. They are exceptions in the sense that they do not allow the unmarked verb form (i.e. the temporally subordinated form), which can always be used in the past sector.

115. It has been observed by e.g. Brecht (1974: 501), Comrie (1989: 56), Hornstein (1990: 86, 129-130), Ladusaw (1977: 95), Richards (1982: 89), Tregidgo (1979: 194), Quirk et al. (1985: 1010) that the present tense is used to refer to a situation that is simultaneous with a future tense situation.

116. It has been pointed out by e.g. Brecht (1974: 503), Chodorow (1982: 56), Ejerhed Braroe (1974: 58), Hornstein (1990: 125), Partee (1973: 608), Rigter (1982: 129-130), Smith (1977: 154, 159), (1978: 56), (1991: 138) that the past tense may express anteriority with respect to a future point of orien-tation. Lo Cascio (1986) claims that in dirä che oggi e venuto per cortesia. (He will say that today he came to put in an appearance.), came "indicates an event time which is prior to the time of the matrix clause, i.e. prior to the moment of "dire" (say) but also prior to the GPT [Given Primary Time]" (Lo Cascio 1986: 203). It will be clear that, unlike what Lo Cascio claims, he came is not necessarily anterior to t0: if there is a party at 7 o'clock on Friday and if I have a meeting with Peter the next day, and I utter the above sentence at 2 o'clock on Friday, the time of Peter's putting in an appearance does not lie before tn.

388 Notes

117. Cf. e.g. Canavan (1983: 169), Davidsen-Nielsen (1990: 136-137), Kato (1985: 160-161), Ladusaw (1977: 95), Rigter (1980: 436-437).

118. Tregidgo (1974) observes that the future tense can be used in a conditional clause provided the SC situation comes after the SUPC situation. Although Tregidgo (1979: 196) rejects the analysis of future tenses in ij-clauses he defended in Tregidgo (1974), he is one of the few linguists to have made the generalization that the future tense may be used to express posteriority to another future situation. Similarly, Wekker (1980: 98) calls 71 do in Write and tell me what you'll do an after-future. Smith (1978) points out that the future tense in the complement in Stuart will announce tomorrow that he will enter the debate in a week "is not taken to be simultaneous with the matrix" (Smith 1978: 56), but does not make any claims about the binding TO of the SC future tense. Ladusaw (1977: 95) also observes that the future tense is ambiguous in John will find the unicorn that will walk.

119. The fact that the future tense is not used in temporal and conditional clauses is common knowledge (cf. e.g. Binnick 1991: 90, Comrie 1989: 56, Davidsen-Nielsen 1990: 118, 136, Matthiessen 1983: 387, Rigter 1980: 454, Schöpf 1984: 339, Woisetschläger 1976: 103). However, apart from Allen (1966), Declerck (1991a, 1991b), Wekker (1976, 1980) and (to a certain extent) Kruisinga (1931), no one tries to explain the use of tense by means of a concept similar to temporal subordination.

120. The references given in footnotes 115, 116, 117 and 118 are for the greater part comments on a single example; no general principles are formulated.

121. Cf. e.g. Binnick (1991: 115), Comrie (1985: 72-73), (1989: 60), Davidsen-Nielsen (1990: 59), Klein (1992: 527), Leech (1971: 59), Penhalluriack (1981: 219), Vikner (1985: 88). The following linguists claim that a future perfect situation always lies in the post-present sector: Davidsen-Nielsen (1985: 74), Hüllen (1989: 613), Morrissey (1973: 67), Schöpf (1984: 338).

122. The following do not consider will/shall + infinitive as a tense: En? (1987: 634), Feigenbaum (1981: 393), Fenn (1978: 177) (only will + stative verb is a pure future), Nehls (1988:

Notes 389

298), Quirk et al. (1985: 213), Smith (1978: 49). Their opinion is not shared by e.g. Allen (1966: 132), Davidsen-Nielsen (1987: 58), (1990: 68), Declerck (1991a: 9-11), Hornstein (1990: 38), Katuza (1984: 324), Kirsten (1972), Leech (1971: 57), Rauh (1984: 14-15), Schöpf (1984: 182), Tregidgo (1974: 106), Vasudeva (1971: 90), Wekker (1976: 9-13).

123. Cf. also Allen (1966: 177). His classification is discussed in section 5.4.

124. The original examples are listed in the appendix. 125. Notice also that if the RC has conditional connotations, knows

the answer tomorrow becomes acceptable:

(i) I will ask anyone who knows the answer tomorrow. (ii) I will ask those who know the answer tomorrow.

Prince (1982: 455) argues that I know the answer tomorrow is acceptable if the state is interpreted inchoatively.

126. The opposite line of reasoning seems to hold for the expression of anteriority in a past domain: whenever the shift of domain (the past tense) form does not hamper a correct interpretation (one of W-anteriority), it is likely to be used instead of the temporally subordinated verb (the past perfect), as the former tense seems to require less processing effort.

127. Cf. e.g. Binnick (1991: 124), Declerck (1991a: 56), Εης (1987: 641), Ladusaw (1977: 95), Leech (1971: 65), Partee (1973: 608), Rigter (1980: 436-437), Tregidgo (1979: 194-195), Wekker (1976: 91), (1980: 103).

128. Another possibility is the highly conventionalized use of the future perfect to refer to a situation which could also be evaluated from a present or past time of orientation:

(i) Ronald Reagan will have been the first president to defend this policy. (Declerck 1991a: 87)

(ii) It will have cost a good deal, I suppose. (Kruisinga 1931: 497 in Declerck 1991a: 87)

129. Cf. McCawley (1973: 271). Curme (1931) remarks in general that "the future perfect is avoided still more in the subordinate clause. Here it is usually replaced by the present

390 Notes

or the future, the present perfect, the past or the past perfect" (Curme 1931: 372).

130. If the temporal location is mutually manifest to the speaker and hearer, it means that they both know in which absolute sector the relevant situation is located (cf. chapter 3, section 3.1.).

131. Diver (1963) points out that the use of the definite/indefinite article plays a role in the interpretation of w/ien-clauses:

(i) When he has read the book, he will complain about it. (Diver 1963: 179)

(ii) When he has read a book, he will complain about it. (ibid.)

Diver claims that only (i) can have future reference, (ii) having a habitual (generic) interpretation. However, it is clear that when (ii) is couched in a context in which someone has to read a number of books, it can be uttered with future time reference. Diver goes on to say that there is no similar difference in meaning brought about by the definite/indefinite article in the following because-clauses, which both have a single event past sector reading:

(iii) Just because he has read the book, he will complain about it. (Diver 1963: 179)

(iv) Just because he has read a book, he will complain about it. (ibid.)

In fact, it is not really possible to make a comparison between (i) and (ii) on the one hand and (iii) and (iv) on the other, as a temporal w/ien-clause in the present perfect can only have post-present sector reference, whereas there is no such con-straint on the use of the present perfect in because-clauses.

132. Harkness (1987: 91) points out that the sentence The minister will announce at midnight that he burned the documents an hour ago allows two interpretations: one according to which the documents were burned an hour before the moment of speech (the past tense is an absolute tense), the other indicates that the burning took place an hour before the announcement. It is a matter of discussion whether adverbials of the type X ago are necessarily anchored to tQ or can also

Notes 391

have a past/future time anchor. Cf. e.g. Declerck (1991a: 290), Peterson (1980: 115), Smith (1977: 158), (1978: 84).

133. Theoretically speaking, a third option should be added:

(i) absolute tense: the RC situation is located in the past sector.

(ii) PPS form: the RC situation is anterior to the post-present TO and posterior to t0 (i.e. the RC situation is located in the post-present sector).

(iii) PPS form: the RC situation is anterior to the post-present TO and anterior to t0 (i.e. the RC situation is located in the past sector).

The third option is theoretically possible although the first option seems the unmarked one to refer to a situation located in the past sector. Accordingly, in this discussion, option (iii) will not be taken into consideration.

134. Cf. e.g. Arnovick (1990: 92), Charleston (1955: 264), Coates (1983: 61), Haegeman (1983: 54-55), Huddleston (1984: 172-174), King (1983: 111-112), Leech (1971: 57), Lyons (1977: 677, 815); Nehls (1988: 303, 305-306), Penhalluriack (1981: 303-304).

135. However, the conditional is sometimes used because of the aspect of prediction it brings with it (cf. chapter 3, section 3.4.3.):

(i) For example, we made a survey of building societies' attitudes towards granting mortgages where the wife is wage-earner, and recommended improvements which would lead ("led") to many building societies changing their practices. (SEU)

(ii) General Peltz, the commander who would conduct ("conducted") the "Little Blitz" on London in 1944 ... (Brown: F 02071) (Tottie - Övergaard 1984: 155)

136. It is possible to devise a special context in which the extra processing effort is justified: e.g. Bill thinks it is very unsuitable for John to take out a girl just two days after he first met her. In this context Bill could say: Will you believe it! Tomorrow John will take out a girl that he met only two days before. Notice that the girl has been changed into a girl.

392 Notes

When the definite article is retained, the sentence is odd even when used in the context just described.

137. Cf. Harkness (1987), who claims in connection with the sentence The minister will announce at midnight that he burned the documents an hour before that "the last is in itself ambiguous, since burning could have taken place at 11 p.m. or an hour previous to some time referred to here anaphorically ('before that time we are talking about')" (Harkness 1987: 92).

138. One native speaker pointed out that will work can be used if "the character of the method does not show itself until the method is put into effect".

139. Cf. e.g. Comrie (1982: 148-152), Declerck (1991a: 198-222), Haegeman (1983: 146-161), Jacobsson (1984), Leech (1971: 65), Tregidgo (1974: 105) for a discussion of will in if-clauses. If the RC antecedent contains a universal quantifier {every, any, all), the RC often gets a conditional connotation, which results in the use of PPS (cf. Hirtle - Curat 1986: 62).

140. "Be perspicuous" (Grice 1975: 46). 141. I do not want to say that relevance is not restricted in time; it

is clear that what is relevant at one time may be irrelevant later on. What I am pointing out here is that relevant cases as such exist at any time.

142. "A speaker who uses a definite description attributively in an assertion states something about the whoever or whatever is the so-and-so. A speaker who uses a definite description referentially in an assertion, on the other hand, uses the description to enable his audience to pick out whom or what he is talking about and states something about that person or thing" (Donnellan 1966: 285).

143. If the speaker wanted to indicate a process going on at t0, he would use a present progressive tense.

144. "The primary difference is that they [adjuncts and disjunct] differ syntactically in that disjuncts are peripheral to the clause to which they are attached" (Quirk et al. 1985: 1070).

145. The examples on which the adapted versions are based are listed in the appendix.

Notes 393

146. In order to have an indication that two domains are involved I have chosen to draw two circles representing the W-simulta-neous post-present sector domains, which, strictly speaking, coincide completely.

147. At first sight, it may seem that the low acceptability of the past perfect is due to the order in which the clauses are given: the RC precedes the SUPC and therefore, there is no clause preceding the sentence which can serve as TO. However, in other examples with the same order the past perfect can be used:

(i) Mr Wall, who was smuggled (had been smuggled) into the debating chamber of the students' union, was greeted with shouts of 'Sieg Heil' and standing students giving the Nazi salute. (SEU)

(ii) The Governor, who joined (had joined) Lord Chalfont's party yesterday, was flown off the Endurance by heli-copter to head the reception party. (SEU)

(iii) The down-turn in our trade with Malaysia and Singa-pore, which we foreshadowed {had foreshadowed) last year, took place, not least because depression in the price of rubber, on which so much of their economy depends, was not fully off-set by the end of "confrontation" with Indonesia, while the British share of the import market dwindled further as tariff pref-erences disappeared and local manufacture continued to grow. (SEU)

148. According to Berezovsky (1980), "Both had and has are re-lated to a (d)-event [the morpheme marking for past tense is (d) - Berezovsky (1980: 24)] other than the one their verb de-scribes because had is an event of the past with past rele-vance, and has is an event of the past with present relevance" (Berezovsky 1980: 82).

149. The following linguists have pointed out that just indicates temporal proximity in present perfect sentences: Lee (1987: 390), Leech (1971: 38), Palmer (1965: 101), (1974: 77), Thomson and Martinet (1980: 152). (For a somewhat diver-gent view cf. Inoue (1978: 169)). From that point of view it seems strange that in American English just is compatible with the use of past tense: cf. Dietrich (1969:407), Duskova (1976: 55), Leech (1971: 38-39), McCoard (1978: 34),

394 Notes

Neutze (1982: 159), Palmer (1965: 101), (1974: 77), Quirk et al. (1985: 194), Vanneck (1958: 239).

150. King (1983) stresses the same point in connection with the present perfect/past tense opposition: "The relational present takes any expanse of time and makes judgments on the occurrence(s) of a situation up to the present. If that expanse of time is very narrow (i.e. begins rather close to the TOC [time of communication]), the distinction in meaning is not all that revealing in a particular context, i.e. either form is likely to occur:

(i) Did you take the meatloaf out of the oven? Have you taken

(ii) Have you seen that movie? Did you see

Context may reveal that the past situation is near to the TOC regardless of the form of the verb" (King 1983: 121-122). For similar remarks cf. e.g. Serensen (1964: 80).

151. Cf. Comrie (1985): "It is more likely that recent events will have current relevance than more remote events, whence the tendency, out of context, to interpret the perfect as referring to a more recent event than the simple past" (Comrie 1985: 25).

152. For similar examples cf. e.g. Declerck (1991a: 329), Leech (1969: 155-156), (1971: 38), Quirk et al. (1985: 193), Wilson (1990: 17-20).

153 It has been pointed out by several linguists that "hot news" is reported in the present perfect, the reason being that the event is recent and that therefore the hearer is not familiar with it. For interesting remarks relating to "hot news" perfects. Cf. e.g. Comrie (1976: 60-61), Dinsmore (1981: 489-490), Fenn (1987: 129-131), Inoue (1979: 586), King (1983: 125).

154. "If John's leg is currently broken, then the perfect can be used no matter how long ago the break took place, as in John has broken his leg - it happened six weeks ago, and it still hasn't healed' (Comrie 1985: 25). Cf. also Mommer (1986: 255), Young (1980: 196).

Notes 395

155. Cf. Comrie (1976): "While present relevance does not imply recentness, recentness may be a sufficient condition for present relevance" (Comrie 1976: 60). Cf. also Comrie (1985: 25), Huddleston (1984: 160), McCoard (1978: 33-34).

156. Tregidgo (1979) remarks that the past perfect is preferably used when "the information reported earlier is now regarded as superseded, e.g.

He once told us that Shakespeare had died in 1716 - he

was always making mistakes like that.

Compare also:

Why was John late this morning? (a) He told me he fell downstairs on his way to work. (b) He told me he had fallen downstairs on his way to

work." (Tregidgo 1979: 193)

In Tregidgo's opinion, (b) is "particularly likely to be used if the speaker has now discovered John's story to be false ('... but actually he overslept)" (Tregidgo 1979: 193). Interesting though this observation seems, it has not been corroborated by the native speakers consulted. Another variation in effects between the past and the past perfect for which I have not found conclusive evidence is that sometimes "the pluperfect seems to have been used in preference to the preterite as representing the facts described in more lively colours than the latter [the past]" (Poutsma 1926: 283). He gives the following example: "You have such pretty feet, mother!" Instantly, with a woman's instinct she had hidden them (Poutsma 1926: 283). Jespersen (1962: 247) and Sweet (1891, II: 105) make a similar remark.

157. Adelaar - Lo Cascio (1986) do not seem to believe there are any regularities: "The temporal order between states of affairs belonging to the same temporal sector is established with the help of temporal adverbia which have a connective function, or with the help of the knowledge we have about the world. So, where both are lacking we are not able to state a temporal order and the text remains ambiguous or not interpretable" (Adelaar - Lo Cascio 1986: 260).

396 Notes

158. R stands for reference time. R-progression in the RT-discussions corresponds to movement of narrative time or foreground in the discourse-discussions.

159. The examples given suggest that the term Event is synonymous with a bounded telic situation.

160. Dowty speaks in terms of states/events, whereas Nerbonne's discussion makes use of the atelic/telic distinction. However, it seems that the labels cover the same content.

161. Cooper (1986: 34) and Dowty (1986: 54) point out that progressive sentences do not easily allow an inchoative reading. Couper-Kuhlen (1989b: 14, 23), Eisterhold (1986: 12) and Hinrichs (1986: 69-70) do not consider examples of this type "exceptional". In Nerbonne's example, unlike in Couper-Kuhlen's and Hinrichs', the progressive clause pushing the action forward contains an adverbial which indicates a W-posterior time.

162. As will be pointed out below (section 7.5.4.), some people explicitly indicate that the generalizations made are not merely relevant to sequences of past tense sentences but to any tense-type sequences which qualify as narrative. Partee (1984) is also aware of the limitation which an exclusive focus on past tense sentences implies: "As before, we are limiting our attention to sentences in narratives in which only simple past tenses occur, with clauses describing either events or states (thereby bypassing a number of issues that would be important for a more comprehensive treatment)" (Partee 1984: 256).

163. Comrie's (1986b) claim that the use of a past perfect does not always mean that the second situation is anterior to the first does not contradict the point made:

(i) Old Sam's boat approached the other side of the pier. He had docked before I began fishing (Comrie 1986b: 18)

The past perfect is anterior to some binding TO, although not necessarily that of the preceding sentence. I agree with Comrie when he claims that "tenses have relatively abstract meanings, which provide a very underdetermined notion of the chronological structure of a text; in order to come up with

Notes 397

a full interpretation of the chronological structure, the listener must refer to a range of other clues, such as other grammatical categories (in particular verbal aspect), the lexical meaning of verbs, the meanings of the structures into which verbs fit, and even extra-linguistic knowledge" (Comrie 1986b: 12).

164. Eisterhold (1986: 32-33) also points to the circularity in some definitions of the foreground-background distinction.

165. In order to arrive at trustworthy judgments by the informants, not more than 25 sentences should be presented at a time. This, together with the fact that I was only once in a position to present all the sentences to a large number of native speakers, has resulted in a selection of examples.

166. Couper-Kuhlen's observations are based on Nerbonne's 1984 Ohio State University Dissertation, which I have not been able to consult. In his 1986 article, which is available, he explicitly equates atelicity with activities and states (Nerbonne 1986: 91).

167. If I say that there is a clear (no) preference for a particular option, it means that the choice in favour of a certain order is (not) statistically relevant. I am grateful to S. De Roo for helping me assess the relevance of the results in statistical terms. Cf. appendix for a note on the statistical interpretation.

168. The RC situation is interpreted as bounded atelic and not as bounded telic, which, theoretically speaking, is also possible (cf. chapter 1, section 1.3.4.1.).

169. NRRCs which occur in parenthetical position do not yield a W-posterior reading:

(i) We left the clerk, who copied the letter, at his desk. (Daalder 1989: 202)

Daalder writes: "The point of nonrestrictive relatives generally is to offer, within the meaning of the sentence as a whole, something like a motivating circumstance (or also an adverse one) that happens to reside in some property or activity of one of the participants of the sentence. Accord-ingly, as a specification of what we left, the explanation of the elements the and clerk suffices (...). This constructed

398 Notes

meaning explanation is then again specified as to motivation by the explanation of who copied the letter (...) To put it bluntly, we left the clerk alone because he was taking such a long time in copying the letter" (Daalder 1989: 202).

170. NRRCs with bounded situations are not always of the continuative type; the following examples show that the semantics of the NRRC may establish a W-anterior reading:

(i) NEC, one of its largest domestic competitors, said it bid one yen in two separate public auctions since 1987. In both cases, NEC lost the contract to Fujitsu, which made the same bid and won a tie-breaking lottery. (WSJ)

(ii) After breaking into a factory at Soho Hill, Handsworth, Birmingham, he set an alarm clock belonging to one of the staff and went to sleep in the managing director's chair. He stole about +3 from the canteen, which he entered with the help of factory tools, and also helped himself to eggs and milk. (LOB)

171. The following related sentence was presented to my informants:

(i) He will hit John, whom Sue will punch in the stomach. 1 - 2 :64% 1 - 2 (2 -1): 22% 2 - 1 :8% 2 - 1 ( 1 - 2 ) : 6%

172. If the indefinite NP a beauty contest is changed into a definite NP, the relation of W-anteriority appears to be more clearly present.

173. A similar point is made by Duskova (1974):

(i) I've read, I've listened to the radio, I've watched TV -but I haven't enjoyed anything so much as just sitting and doing nothing.

"Here the act described may refer to a different occasion, and, moreover, may denote a repeated act. Thus the acts are expressed not as parts of the same sequence but as isolated events. Sequential interpretation of past acts denoted by the

Notes 399

perfect is conceivable only as a result of the presence of temporal adjuncts indicating repetition" (Duskova 1974: 70). She draws the conclusion that "the preterite and the perfect differ not only in their respective temporal reference to pat acts or events but also in the essentially sequential nature of the former and the non-sequential nature of the latter" (Duskova 1974: 70). Cf. also Kirsten (1986: 61), Ter Meulen (1991: 522).

174. However, Haegeman (1989) adds in a note: "One anonymous referee did not seem to agree with this parallelism. Other native speakers I consulted shared my intuitions. Perhaps this is another instance where for some speakers the distinction between tense forms are being blurred?" (Haegeman 1989: 310). This observation points to the necessity of presenting examples of this type to native speakers.

175. When possible, I have italicized the phrase which functions as binding TO.

176. All the references from Declerck (1991a) relate to SCs in general, unless otherwise indicated.

177. A more accurate formulation is: the TOsit of a SC situation is usually bound by the TOsit of its SUPC situation. However, in what follows, I will refrain from using this more accurate but long-winded paraphrase. The SC (situation) is bound by the SUPC (situation) will be used instead.

178. Vet - Molendijk (1986) argue in connection with the sentence Le grand-vizier mourüt ä l'äge de 88 ans. C'etait un homme tres sage, that "the only antecedent which remains is the individual" (Vet - Molendijk 1986: 150). However, their notion antecedent is not to be equated with T02.

179. Zandvoort uses hired in his example. 180. The original sentences on which the adapted examples are

based are listed in the appendix. 181. Adelaar - Lo Cascio (1986) do not believe that a SC can ever

function as binding TO: "Even if we stated that a tense in a subordinate clause can express a deictic relation it remains the case that it marks an event which in turn cannot provide an evaluation time or reference time for another event in the discourse" (Adelaar - Lo Cascio 1986: 268). "Tenses express

400 Notes

a temporal relation only with respect to temporal entities, functioning as evaluation times, which belong to a higher po-sition in the (syntactic and semantic) hierarchy" (Adelaar - Lo Cascio 1986: 269). It is probably because they only have in mind sequences of the type "SUPC precedes SC" that they do not include the possibility in which the SUPC situation is bound by the SC situation. Adelaar - Lo Cascio (1986) formulate the following rule for main clauses with a relative tense: "If the anaphoric tense marks a main clause, then it will find its antecedent in a clause marked by a deictic tense and functioning as head of the Mu [= Major Unit: "a deictic tense together with those anaphoric tenses controlled by it (Adelaar - Lo Cascio 1986: 253)] instead of finding the antecedent in the immediately preceding clause, marked by an anaphoric tense" (Adelaar -Lo Cascio 1986: 255)

e.g. That evening Paul said that John had arrived. The day before he had seen him in the book store. (Adelaar - Lo Cascio 1986: 293)

The second sentence is a short version of Paul said that the day before he had seen John in the book store This explains why the second sentence is very likely to be interpreted as temporally subordinated to the main clause of the first sen-tence. It is not difficult to find counterexamples to Adelaar -Lo Cascio's rule:

e.g. It was clear that John had already arrived. He had stayed in London for 2 years and had visited all the art galleries while he was there.

Had stayed may be interpreted as temporally subordinated to had arrived, although was clear could also function as bind-ing TO for had stayed. However, if the speaker is guided by the principle of maximal effect for minimal effort, it is likely that had arrived will function as T0 2 because this is the first possible binding TO candidate the speaker encounters when looking for a T0 2 in the preceding context.

182. Couper-Kuhlen (1987: 23), Tregidgo (1979: 192) and Bertinetto (1986: 67) also give examples in which it is "the main clause event with respect to which the temporal clause

Notes 401

event is located" (Couper-Kuhlen 1987: 23) or in which a subclause is the "main time" (Tregidgo 1979: 192). However, we must be cautious when drawing parallels with claims in a Declerckian framework, because it is not entirely clear if the points they are making also apply to binding TO in the sense of Declerck (1991a).

183. Declerck points out (personal communication) that in his opinion, were is the central TO of the domain. It is difficult to determine which element functions as binding TO for had made. I would argue that in this example knew or surmised is more likely to function as T0 2 because those verbs precede had made.

184. Declerck points out (personal communication) that in his opinion, had been made expresses anteriority with respect to have been driven back. As will argued in this chapter, linear sequence is probably more important than syntactic status (main clause - SC) for the choice of binding TO: a hearer will prefer to use a preceding clause as binding TO rather than a clause in which the RC is embedded but which follows.

185. Bertinetto (1986) takes up an example of McCoard's (1978: 185-186):

e.g. After he had left his apartment and made sure no one was following him, he had hurried to the bank, had withdrawn all his funds, and, before anyone had noticed, had skipped town. (Bertinetto 1986: 66)

"We might for instance continue with the following sentence, which provides an explicit R:

(67)... He himself said this to me while we were sitting on the train, a few hours later, last Monday." (1986: 67)

However, McCoard's example is given out of context. It would probably not be used as the beginning of a conversation. This means that, when couched in a context, there is very likely to be an element preceding the sequence of past perfect sentences which will function as binding TO.

186. Declerck (personal communication) believes that binding TOs are in the first place realized by verbs.

402 Notes

187. Although her concept of relative tense is different from ours, Contini-Morava (1983), who discusses Swahili examples, similarly stresses the fact that "any and all of this information, both linguistic and non-linguistic, can contribute to a speaker's identification of time orientation for a relative tense (...), moreover, (...) there is no reason to believe that it is possible to specify in advance what contextual factors or features will determine time orientation" (Contini-Morava 1983: 9). Adelaar - Lo Cascio (1986) endorse Lo Cascio - Rohrer's (1986: 239-241) view that RCs are relatively free in their choice of TO: "The anaphoric verbal tense of a RC can be evaluated with respect to a time denoted by an arbitrary clause and is therefore free of the restrictions imposed by subordination. In some cases the antecedent can be given by the matrix if the verbal tense of the RC and of the matrix clause express the same type of temporal relation" (Adelaar -Lo Cascio 1986: 256).

188. As will be pointed out below (cf. section 8.2.2.1.), the past tense is not always a relative tense in so-called direct/indirect binding sets of examples.

189. Declerck points out (personal communication) that he does not reject these interpretations. However, in his opinion, the other alternative (the SUPC is bound by a TO which acts as binding TO for the RC situation) is "more normal".

190. Declerck points out (personal communication) that he would also allow both possibilities in this example.

191. The following was one of these examples:

(i) We have focussed in particular on the complex of technological changes that has occurred since the late 1960s, which ?? was termed the Green Revolution. (SEU)

Cp. (ii) We have focussed in particular on the complex of technological changes that has occurred since the late 1960s, which at the time was termed the Green Revolution.

192. A situation is either telic, atelic or zero-telic. The CRPP, CRPrP and CRCT have shown that RRCs and NRRCs behave in a similar way when the clauses are atelic or zero-telic. The

Notes 403

PUTI is important for the cases in which the situations are telic and shows that in this respect, there are differences between NRRCs and RRCs.

193. I wish to thank Dr. B. Aarts, V. Adams, A. Rosta, S. Boase and the first-year English students for their cooperation.

194. I wish to express my thanks to Prof. L. Vermeire and S. De Roo for their help.

References

Abraham, Werner - Theo Janssen (eds.) 1989 Tempus - Aspekt - Modus: die lexikalischen und

grammatischen Formen in den Germanischen Sprachen. Tübingen: Niemeyer.

Adelaar, Mascia - Vincenzo Lo Cascio 1986 "Temporal relation, focalization and direction", in:

Vincenzo Lo Cascio - Co Vet (eds.), 251-297. Allen, Robert

1966 The verb system of present-day American English. The Hague: Mouton.

Andersson, Sven-Gunnar 1991 "Tense functions in scientific texts", in: Jadranka

Gvozdanovifc - Theo Janssen (eds.), 1-16. Aristar Dry, Helen

1983 "The movement of narrative time", Journal of Lite-rary Semantics 12: 19-53.

Arnovick, Leslie K. 1990 The development of future constructions in English.

The pragmatics of modal and temporal will and shall in Middle English. New York: Peter Lang.

Bach, Emmon 1980 "Tenses and aspects as functions on verb-phrases", in:

Christian Rohrer (ed.), 19-37. Bache, Carl

1985 Verbal aspect: a general theory and its application to present-day English. Odense: Odense University Press.

Bache, Carl - Leif Kirstgaard Jakobsen 1980 "On the distinction between restrictive and non-re-

strictive relative clauses in modern English", Lingua 52: 243-267.

References 405

Bartsch, Renate 1988/89 "Tenses and aspects in discourse", Theoretical

Linguistics 15: 133-194. Bauer, Gero

1970 "The English 'Perfect' reconsidered", Journal of Lin-guistics 6: 189-198.

Bäuerle, R. - Urs Egli - A. von Stechow (eds.) 1979 Semantics from different points of view. Berlin:

Springer. Beach, Woodford A. - Samuel E. Fox - Shulamith Philosoph (eds.)

1977 Papers from the thirteenth Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Lin-guistic Society.

Bennett, Michael - Barbara Partee 1978 Toward the logic of tense and aspect in English.

Bloomington: Indiana University Linguistics Club. Berezovsky, Helen Carroll

1980 A paraphrastic analysis of the temporal relationships in the English verb system. Ph.D. Dissertation, Uni-versity of Pennsylvania.

Bertinetto, Pier Marco 1986 "Intrinsic and extrinsic temporal references: on re-

stricting the notion of 'reference time'", in: Vincenzo Lo Cascio - Co Vet (eds.), 41-78.

Bertinetto, Pier Marco - V. Bianchi - Mario Squartini (eds.) 1995 Tense, aspect and actionality. Torino: Rosenberg &

Sellier. Bhat, D.N.S.

1977 "A note on ambiguity in negative sentences contain-ing durative adverbials", Glossa 11: 115-121.

Binnick, Robert 1991 Time and the verb - A guide to tense and aspect. Ox-

ford: Oxford University Press.

406 References

Boogaart, Ronny 1995 "Towards a theory of discourse aspect", in: Pier

Marco Bertinetto - V. Bianchi - Mario Squartini (eds.)·

Bouscaren, Janine (ed.) 1982 Cahiers de recherche en grammaire anglaise. Tome I.

Paris: Ophrys. Bouscaren, Janine - Jean Chuquet - O. Herlin

1982 "La forme HAD-en ou le PAST PERFECT", in: Janine Bouscaren (ed.), 73-114.

Brecht, Richard 1974 "Deixis in embedded structures", Foundations of Lan-

guage 11:489-518. Bree, D.S.

1985 "The durative temporal subordinating conjunctions: since and until", Journal of Semantics 4: 1-46.

Breithutovä, Helga 1968 "The function of tenses used in Modern English tem-

poral clauses", Brno studies in English 7: 129-148. Brinton, Laurel

1985 "Verb particles in English: aspect or Aktionsart?", Studia Linguistica 39: 157-168.

1988 The development of English aspectual systems: as-pectualizers and post-verbal particles. Cambridge: Cambridge University Press.

Brisau, Andre 1977 "Conditioned and non-conditioned future in English

and Netherlandic", in: Yvan Putseys (ed.), 51-69. Bublik, Anna I.

1983 "Vergleichende syntagmatische Characteristik der Perfektivformen in der neueren englischen und deutschen Sprache", Zeitschrift für Anglistik und Amerikanistik 31: 32-42.

Buyssens, Eric 1968 Les deux aspectifs de la conjugaison anglaise au XXe

siecle. Brussels: Presses Universitäres de Bruxelles.

References 407

Canavan, John 1983 The English tense system: a study of temporal mean-

ing and reference. Bonn: Bouvier Verlag Herbert Grundmann.

Chalker, Sylvia 1985 "Is the past perfect a before-past?", English Language

Teaching Journal 39: 48-49. Charleston, B.M.

1955 "A reconsideration of the problem of time, tense, and aspect in modern English", English Studies 36: 263-278.

Chodorow, Martin S. 1982 "What time do you presuppose it is?", Papers in Lin-

guistics 15: 51-57. Coates, Jennifer

1983 The semantics of modal auxiliaries. London: Croom Helm.

Cochrane, N. 1977 "An essential difference between momentary and du-

rative events", in: Woodford A. Beach - Samuel E. Fox - Shulamith Philosoph (eds.), 93-103.

Cole, Peter - Jerry L. Morgan (eds.) 1975 Syntax and semantics. Volume 3: Speech acts. New

York: Academic Press. Comrie, Bernard

1976 Aspect. Cambridge: Cambridge University Press. 1982 "Future time reference in the conditional protasis",

Australian Journal of Linguistics 2: 143-152. 1985 Tense. Cambridge: Cambridge University Press. 1986a "Tense in indirect speech", Folia Linguistica 20: 265-

296. 1986b "Tense and time reference: from meaning to interpre-

tation in the chronological structure of a text", Journal of Literary Semantics 15: 12-22.

1989 "On identifying future tenses", in: Werner Abraham -Theo Janssen (eds.), 51-63.

408 References

Contini-Morava, Ellen 1983 "Relative tense in discourse: the inference of time ori-

entation in Swahili", in: Flora Klein-Andreu (ed.), 3-21.

Cooper, Robin 1986 "Tense and discourse location in situation semantics",

Linguistics and Philosophy 9: 17-36. Costa, Rachel

1972 "Sequence of tenses in that-clauses", in: Paul M. Peranteau - Judith N. Levi - Gloria C. Phares (eds.), 41-51.

Couper-Kuhlen, Elizabeth 1987 "Temporal relations and reference time in narrative

discourse", in: Alfred Schöpf (ed.), 7-25. 1989a "On the markedness of "narrative temporal clauses",

in: Olga Miseska Tomic (ed.), 359-372. 1989b "Foregrounding and temporal relations in narrative

discourse", in: Alfred Schöpf (ed.), 7-29. Curme, George

1931 Syntax. Boston: D.C. Heath and Company. Daalder, Saskia

1989 "Continuative relative clauses", Sprechen und Hören, 195-207.

Dahl, Osten 1981 "On the definition of the telic-atelic (bounded-un-

bounded) distinction", in: Philip Tedeschi - Annie Zaenen (eds.), 79-90.

1985 Tense and aspect systems. Oxford: Basil Blackwell. 1992 "Review of Hornstein (1990)", Language 68: 645-

650. Davidsen-Nielsen, Niels

1985 "Tense in modern English and Danish", Papers and Studies in Contrastive Linguistics 20: 73-84.

1987 "Has English a future?", in: Ishrat Lindblad - Magnus Ljung (eds.), 53-59.

References 409

1990 Tense and mood in English - a comparison with Da-nish. Berlin: Mouton, de Gruyter.

Declerck, Renaat 1979 "Aspect and the bounded/unbounded (telic/atelic) dis-

tinction", Linguistics 17: 761-794. 1988 Studies on copular sentences, clefts and pseudo-clefts.

Leuven: Leuven University Press. 1989 "Boundedness and the structure of situations", Leu-

vense Bijdragen 78: 275-304. 1991a Tense in English: its structure and use in discourse.

London: Routledge. 1991b A comprehensive descriptive grammar of English.

Tokyo: Kaitakusha. Declerck, Renaat - Ilse Depraetere.

1995 "The double system of tense forms referring to future time", Journal of Semantics 12.

Defroment, H. 1973 Les constructions perfectives du verbe anglais con-

temporain. The Hague: Mouton. Depraetere, Ilse

1990 Temporal when-clauses in written English: a corpus-based analysis. Unpublished MA dissertation, Univer-sity College London.

1993 The tense system in English relative clauses. Ph.D. dissertation. Katholieke Universiteit Leuven.

Descles, Jean-P. 1989 "State, event, process and typology", General Lin-

guistics 29/3: 159-201. De Vuyst, Jan

1982 "Situatiebeschrijvingen in het Engels en het Nederlands", Tabu 12: 97-107

1983 "Situation-descriptions: temporal and aspectual se-mantics", in: Alice Ter Meulen (ed.), 161-176.

1985 "The present perfect in Dutch and English", Journal of Semantics 4: 137-163.

410 References

Dietrich, Gerhard 1969 "Ein Schmerzenkind der englischen Grammatik - Der

Gebrauch des Präterits und des Perfekts im Engli-schen", Praxis des neusprachlichen Unterrichts 4: 402-412.

Dillon, Georges L. 1973 "Perfect and other aspects in a case grammar of En-

glish", Journal of Linguistics 9: 271-279. Dinsmore, John

1981 "Tense choice and time specification in English", Lin-guistics 19: 475-494.

Dirven, Rene - Wolfgang Zydatiss - Willis J. Edmondson (eds.) 1978 A user's grammar of English: Word, sentence, text,

interaction. (Parts C & D). Frankfurt am Main: Lang. Diver, William

1963 "The chronological system of the English verb", Word 19: 141-181.

Donnellan, Keith S. 1966 "Reference and definite descriptions", The Philo-

sophical Review 75: 281-304. Dowty, David

1977 "Toward a semantic analysis of verb aspect and the English 'imperfective progressive'", Linguistics and Philosophy 1:45-77.

1982 "Tense, time adverbs, and compositional semantic theory", Linguistics and Philosophy 5: 23-55.

1986 "The effects of aspectual class on the temporal struc-ture of discourse: semantics or pragmatics?", Linguis-tics and Philosophy 9: 37-61.

Dry, Helen 1979 "Four temporal effects of narrative now", Texas Lin-

guistic Forum 15: 60-69. 1981 "Sentence aspect and the movement of narrative

time", Text 1: 233-240.

References 411

Dubos, Ulrika 1985/86 "Le plus-que-parfait et la notion d'accompli

d'anteriorite: contribution ä la discussion sur la valeur de had + ed\ Sigma 9: 145-168.

Duskova, Libuse 1974 "The perfect tenses in English vs the perfective aspect

in Czech", Philologica Pragensia 17: 67-91. 1976 "On some differences in the use of the perfect and

preterite between British and American English", Prague Studies in Mathematical Linguistics 5: 53-68.

Ehrlich, Susan 1987 "Aspect, foregrounding and point of view", Text 7:

363-376. 1990 Point of view: A linguistic analysis of literary style.

London: Routledge. Eisterhold, Joan Carson

1986 Verbal aspect and background/foreground distinc-tions in radio commentaries. Ph.D. Dissertation, Northwestern University (Evanston, Illinois).

Ejerhed Braroe, Eva 1974 The syntax and semantics of English tense markers.

Stockholm: University of Stockholm, Institute of Lin-guistics.

Εης, Murvet 1987 "Anchoring conditions for tense", Linguistic Inquiry

18:633-657. Feigenbaum, Irwin

1981 "The uses of the English perfect", Language Learning 31:393-407.

Fenn, Peter 1978 "Dangers and fallacies in the explanation of future

time references", Neusprachliche Mitteilungen 31: 175-181.

1987 A semantic and pragmatic examination of the English perfect. Tübingen: Narr.

412 References

Fijn van Draat, P. 1912 "The preposition since", Anglia 35: 155-164.

Fleischman, Suzanne 1985 "Discourse functions of tense-aspect distinctions in

narrative: towards a theory of grounding", Linguistics 23:851-882.

1989 "Temporal distance: a basic linguistic metaphor", Studies in Language 13: 1-50.

Fraser, Bruce 1970 "Some remarks on the action nominalization in En-

glish", in: Roderick A. Jacobs - Peter S. Rosenbaum (eds.), 83-98.

Gabbay, Dov - Christian Rohrer 1978 "Relative tenses: the interpretation of tense forms

which occur in the scope of temporal adverbs or in embedded sentences", in: Christian Rohrer (ed.), 99-110.

1979 "Do we really need tenses other than future and past?", in: R. Bäuerle - Urs Egli - A. von Stechow (eds.), 15-20.

Gabbay, Dov - Julius Moravcsik 1980 "Verbs, events, and the flow of time", in: Christian

Rohrer (ed.), 59-83. Garey, Howard B.

1957 "Verbal aspect in French", Language 33: 91-110. Gelhaus, Hermann

1974 "Untersuchungen zur Consecutio Temporum im Deutschen", in: Hermann Gelhaus - Sigbert Latzel (eds.), 1-127.

Gelhaus Hermann - Sigbert Latzel (eds.) 1974 Studien zum Tempusgebrauch im Deutschen.

Tübingen: Narr. Givon, Talmy (ed.)

1979 Syntax and semantics. Volume 12: Discourse and syntax. New York: Academic Press.

References 413

Greenbaum, Sidney - Geoffrey Leech - Jan Svartvik, (eds.) 1980 Studies in Linguistics for R. Quirk. London: Long-

man. Grice, Paul H.

1975 "Logic and conversation", in: Peter Cole - Jerry L. Morgan (eds.), 41-58.

Guenthner, F. 1977 "Remarks on the present perfect in English", in:

Christian Rohrer (ed.), 83-98. Gvozdanovii , Jadranka - Theo Janssen (eds.)

1991 The function of tense in texts. Proceedings of the colloquium 'The function of tense in texts', Amster-dam 10-12 april 1990. (Koninklijke Nederlandse Akademie van Wetenschappen) Amsterdam: North-Holland.

Haegeman, Liliane 1983 The semantics of 'will' in present-day British English:

a unified account. Turnhout: Brepols. 1989 "Be going to and will·, a pragmatic account", Journal

of Linguistics 25: 291-317. Haegeman, Liliane - Herman Wekker

1984 "The syntax and interpretation of futurate conditionals in English", Journal of Linguistics 20: 45-55.

Hamann, Cornelia 1989 "English temporal clauses in a reference frame

model", in: Alfred Schöpf (ed.), 31-154. Harkness, Janet

1987 "Time adverbials in English and reference time", in: Alfred Schöpf (ed.), 71-110.

Hatav, G. 1989 "Aspects, Aktionsarten, and the time line", Linguistics

27: 487-516. Heinämäki, Orvokki

1978 Semantics of English temporal connectives. Bloomington: Indiana University Linguistics Club.

414 References

Helm, June (ed.) 1967 Essays on the verbal and visual arts. (Proceedings of

the 1966 Annual Spring Meeting of the American Ethnological Society). Seattle: University of Washington Press.

Heny, Frank 1982 "Tense, aspect and time adverbials: part 2", Linguis-

tics and Philosophy 5: 109-154. Hill, Archibald A.

1958 Introduction to linguistic structures: from sound to sentence in English. New York: Harcourt, Brace and World.

Hinrichs, Erhard 1986 "Temporal anaphora in discourses of English", Lin-

guistics and Philosophy 9: 63-82. Hintikka, Jaakko

1982 "Temporal discourse and semantical games", Linguis-tics and Philosophy 5: 3-22.

Hirtle, Walter 1967 The simple and progressive forms: an analytical ap-

proach. Presses de l'Universite Laval, Quebec. 1975 Time, aspect and the verb. Presses de l'Universite

Laval, Quebec. 1981 "Meaning and form in "when"-clauses", in: Andre

Joly - Walter Hirtle (eds.), 217-228. 1987 "Events, time and the simple form", Revue quebecoise

de linguistiquc 17/1: 85-106. Hirtle, Walter - V.N. Curat

1986 "The simple and progressive "future" use", Transac-tions of the Philological Society: 42-84.

Hopper, Paul J. 1979a "Aspect and foregrounding in discourse", in: Talmy

Givon (ed.), 213-241. 1979b "Some observations on the typology of focus and as-

pect in narrative language", Studies in Language 3: 37-64.

References 415

Hopper, Paul J. (ed.) 1982 Tense-aspect: between semantics and pragmatics.

Amsterdam: John Benjamins. Hornstein, Norbert

1977 "Towards a theory of tense", Linguistic Inquiry 8: 521-557.

1981 "The study of meaning in natural language: three ap-proaches to tense", in: Norbert Hornstein - David Lightfoot (eds.), 116-151.

1990 As time goes by. Tense and universal grammar. Cam-bridge, Mass.: MIT Press.

Hornstein, Norbert - David Lightfoot (eds.) 1981 Explanation in linguistics: the logical problem of lan-

guage acquisition. London: Longman. Houweling, Frans

1986 "Deictic and anaphoric tense morphemes", in: Vin-cenzo Lo Cascio - Co Vet (eds.), 161-190.

Huddleston, Rodney 1969 "Some observations on tense and deixis in English",

Language 45: 777-806. 1971 The sentence in written English. Cambridge: Cam-

bridge University Press. 1984 Introduction to the grammar of English. Cambridge:

Cambridge University Press. Hüllen, Werner

1989 "Text and tense", in: Rene Dirven - Wolfgang Zydatiss - Willis J. Edmondson (eds.), 599-624.

Hüllen, Werner - Rainer Schulze (eds.) 1985 Tempus, Zeit und Text. Heidelberg: Carl Winter Uni-

versitätsverlag. Inoue, Kyoko

1978 "How many senses does the English present tense have?", in: Donka Farkas - Wesley M. Jacobsen -Karol W. Todrys (eds.), 167-178.

1979 "An analysis of the English present perfect", Linguis-tics 17: 561-589.

416 References

Isherwood, Christopher 1969 Goodbye to Berlin. (Seventh edition.) London:

Hogarth. Jackendoff, Ray

1991 "Parts and boundaries", Cognition 41: 9-45. Jacobs, Roderick A. - Peter S. Rosenbaum (eds.)

1970 Readings in English transformational grammar. Waltham, Mass.: Ginn and Company.

Jacobson, Sven 1980 "Some English verbs and the contrast incomple-

tion/completion", in: Sidney Greenbaum - Geoffrey Leech - Jan Svartvik, (eds.), 50-60.

Jacobsson, Bengt 1984 "Notes on tense and modality in conditional if-

clauses", Studia Linguistica 38: 129-147. Jarvella, Robert J. - Wolfgang Klein (eds.)

1982 Speech, place and action: studies in deixis and related topics. Chichester: John Wiley & Sons Ltd.

Jespersen, Otto 1961 A modern English grammar on historical principles.

Volume III. London: Allen and Unwin, Copenhagen: Ejnar Munksgaard.

1961 A modern English grammar on historical principles. Volume IV. London: Allen and Unwin, Copenhagen: Ejnar Munksgaard.

1962 (f.p. 1933) Essentials of English grammar. London: Allen and Unwin.

Joly, Andre - Walter Hirtle (eds.) 1981 Langage et psychomechanique du langage: etudes

dediees ä Roch Valin. Lille: Presses Universitäres de Lille.

Joos, Martin 1964 The English verb: forms and meanings. Madison,

Wis: The University of Wisconsin Press.

References 417

Kaluga , Henryk 1984 "English has two tenses", Kwartalnik Neofilologiczny

Warszawa 31: 323-327. Kato, Kazuo

1985 "The first time and various tense forms", Journal of English Linguistics 18: 153-162.

Kenny, Anthony 1963 Action, emotion and will. London: Routledge and

Kegan Paul Ltd. King, Larry D.

1983 "The semantics of tense, orientation and aspect in En-glish", Lingua 59: 101-154.

Kirsten, Hans 1972 "Gibt es im Englischen ein Futur?", Zeitschrift für

Anglistik und Amerikanistik 20: 229-236. 1986 "In what sense is the present perfect present and not

past?", ELT Journal 40: 61. 1988 "Will + Infinitiv in kommunikativer Sicht", Wis-

senschaftliche Zeitschrift der Wilhelm-Pieck -Univer-sität Rostock Gesellschafts-und Sprachwis-senschaftliche Reihe 37/5: 26-29.

Kittredge, Richard Irwin 1969 Tense, aspects, and conjunction: some inter-relations

for English. Ph.D. Dissertation, University of Penn-sylvania.

Klein, Wolfgang 1992 "The present perfect puzzle", Language 68: 525-552.

Klein-Andreu, Flora (ed.) 1983 Discourse perspective on syntax. New York: Aca-

demic Press. König, Ekkehard

1980 "On the context-dependence of the progressive in En-glish", in Christian Rohrer (ed.), 269-291.

418 References

A handbook of present-day English. Volume II, En-glish Accidence and Syntax 1. (fifth edition.) Groningen: Noordhoff. A handbook of present-day English. Volume II, En-glish Accidence and Syntax 3. (fifth edition.) Groningen: Noordhoff.

"Zur Relevanz temporaler Elemente und Relationen in englischen Texten dargestellt an einem Text der Jour-nalistik", Zeitschrift für Anglistik und Amerikanistik 29: 320-329.

Labov, William - Joshua Waletzky. 1967 "Narrative analysis: oral versions of personal experi-

ence", in: June Helm (ed.), 12-44. Ladusaw, William

1977 "Some problems with tense in PTQ", Texas Linguistic Forum 6: 90-102.

Lakoff, George 1971 "Presupposition and relative well-formedness", in:

Danny D. Steinberg - Leon Jakobovits (eds.), 329-340.

Lakoff, Robin 1970 "Tense and its relation to participants", Language 46:

838-849. Langacker, R.

1982 "Remarks on English aspect", in: Paul J. Hopper (ed.), 265-304.

Lee, David 1987 "The semantics of just", Journal of Pragmatics 11:

377-398. Leech, Geoffrey

1969 Towards a semantic description of English. Bloomington, Indiana: Indiana University Press.

1971 Meaning and the English verb. London: Longman.

Kruisinga, E. 1931

1932

Kupetz, Rita 1981

References 419

LePore, Ernest (ed.) 1987 New directions in semantics. London: Academic

Press. Lindblad, Ishrat - Magnus Ljung (eds.)

1987 Proceedings from the Third Nordic Conference for English Studies, Hässelby, Sept. 25-27, 1986. Vol. I. Stockholm: Almqvist and Wiksell.

Lo Cascio, Vincenzo - Christian Rohrer. 1986 "Interaction between verbal tenses and temporal ad-

verbs in complex sentences", in: Vincenzo Lo Cascio - Co Vet (eds.), 229-249.

Lo Cascio, Vincenzo - Co Vet (eds.) 1986 Temporal structure in sentence and discourse.

Dordrecht: Foris. Lyons, John

1977 Semantics. Cambridge: Cambridge University Press. Malcolm, Lois

1987 "What rules govern tense usage in scientific articles?", English for Specific Purposes 6: 31-43.

Matthiessen, Christian 1983 "Choosing primary tense in English", Studies in Lan-

guage 7: 369-429. McCawley, James

1973 Grammar and meaning. Tokyo: Taishukan. McCoard, Robert

1978 The English perfect: tense-choice and pragmatic in-ferences. Amsterdam: North-Holland.

McGhie, Caroline 1978 "Teaching the past perfect", RELC Journal 9: 69-77.

Meid, Wolfgang - Hans Schmeja (eds.) 1983 Philologie und Sprachwissenschaft. Innsbruck: Insti-

tut für Sprachwissenschaft der Universität Innsbruck. Miseska Tomic , Olga (ed.)

1989 Markedness in synchrony and diachrony. Berlin: Mouton, de Gruyter.

420 References

Mittins, W.H. 1962 A grammar of modern English. London: Methuen and Co

Ltd. Mittwoch, Anita

1980 "The grammar of duration", Studies in Language 4: 201-227.

1988 "Aspects of English aspect: on the interaction of per-fect, progressive and durational phrases", Linguistics and Philosophy 11: 203-254.

Moens, Marc 1987 Tense, aspect and temporal reference. Ph.D. Disser-

tation, University of Edinburgh. Mommer, Kerri Ellen

1986 Theoretical issues concerning inherent aspect and the perfect in English, Cebaari and Swahili. Ph.D. Dis-sertation, Northwestern University (Evanston, Illi-nois).

Morissey, Michael D. 1973 "The English perfective and still/anymore", Journal of

Linguistics 9: 65-69. Mourelatos, Alexander

1981 "Events, processes and states", in: Philip Tedeschi -Annie Zaenen (eds.), 191-212.

Nabokov, Vladimir 1972 Transparent things. New York: Vintage International.

Nakhimovsky, Alexander 1988 "Aspect, aspectual class and the temporal structure of

narrative", Computational linguistics 14/2: 29-44. Needham, Paul

1976 "The speaker's point of view", Synthese 32: 309-327. Nehls, Dietrich

1974 "Bemerkungen zur Zeitenfolge im Englischen", Neusprachliche Mitteilungen 27: 19-24.

1988 "Modality and the expression of future time in En-glish", International Review of Applied Linguistics 26: 295-307.

References 421

Nerbonne, John 1986 "Reference time in narration", Linguistics and Philos-

ophy 9: 83-95. Neutze, Hans-Henning

1982 "Zum Gebrauch der Vergangenheitstempora bei just, recently und in recent years in "Newsweek"", Lebende Sprachen 27: 158-162.

Olofsson, Arne 1981 "Relative junctions in written American English",

Gothenburg Studies in English 50. Ondaatje, Michael

1992 The English patient. London: Picador. Ota, Akira

1963 Tense and aspect of present-day American English. Tokyo: Kenyusha.

Palmer, Frank Robert 1965 A linguistic study of the English verb. London: Long-

man. 1974 The English verb. London: Longman.

Papp, Nändor 1985 "The expression of present and past time in the

English predicate", Papers and Studies in Contrastive Linguistics 20: 95-114.

Partee, Barbara 1973 "Some structural analogies between tenses and pro-

nouns in English", Journal of Philosophy 70: 601-609.

1984 "Nominal and temporal anaphora", Linguistics and Philosophy 7: 243-286.

Penhalluriack, John Murray 1981 A form-content analysis of the modern English sys-

tems of time and occurrence. Ph.D. Dissertation, Columbia University.

422 References

Peranteau, Paul M. - Judith N. Levi - Gloria C. Phares (eds.) 1972 Papers from the eighth Regional Meeting of the

Chicago Linguistic Society. Chicago: Chicago Lin-guistic Society.

Peterson, Thomas 1980 "The law of time conversion: an algebraic property of

time adverbs", Theoretical Linguistics 7: 111-119. Petterson, Thore (ed.)

1979 Aspectology. Stockholm: Almqvist and Wiksell. Potts, T.C.

1965 "States, performances and activities", Aristotelian So-ciety Supplementary Volume 39: 65-84.

Poutsma, H. 1926 A grammar of late modern English. Part II. The parts

of speech. Section II. Groningen: P. Noordhoff. 1928 A grammar of late modern English. Part I. The sen-

tence. 1st half. (2nd edition.) Groningen: P. Noord-hoff.

1929 A grammar of late modern English. Part I. The sen-tence. 2nd half. (2nd edition.) Groningen: P. Noord-hoff.

Prince, Ellen F. 1982 "The simple futurate: not simply progressive futurate

minus progressive", in: Kevin Tuite - Robinson Schneider - Robert Chametzky (eds.), 453-465.

Putseys, Yvan (ed.) 1977 Aspects of English and Netherlandic grammar. Leu-

ven: Acco. Quirk, Randolph - Geoffrey Leech - Sidney Greenbaum - J. Svartvik.

1985 A comprehensive grammar of the English language. London and New York: Longman.

Rauh, Gisa 1984 "Tempora als deiktische Kategorien. Eine Analyse der

Tempora im Englischen und Deutschen. Teil I.", In-dogermanische Forschungen 89: 1-25.

References 423

Reichenbach, Hans 1947 Elements of symbolic logic. New York: the Free Press;

London: Collier-Macmillan. Reinhart, Tanya

1984 "Principles of gestalt perception in the temporal orga-nization of narrative texts", Linguistics 22: 779-809.

Richards, Barry 1982 "Tense, aspect and time adverbials: part 1", Linguis-

tics and Philosophy 5: 59-107. 1987 "Tenses, temporal quantifiers and semantic inno-

cence", in: Ernest LePore (ed.), 337-384. Riddle, Elizabeth

1978 Sequence of tenses in English. Ph.D. Dissertation, University of Illinois at Urbana-Champaign.

Rigter, Bob 1980 "Time diagrams and rules for tense and perfect" in

English, in: Wim Zonneveld - Fred Weerman (eds.), 411-458.

1982 "Intensional domains and the use of tense", Journal of Semantics 1: 95-145.

1986 "Focus matters", in: Vincenzo Lo Cascio - Co Vet (eds.), 99-132.

Riviere, Claude 1977 "Encore le present perfect", Langues Modernes 71/3:

231-247. Rohrer, Christian

1977 "Tempusmetaphern in Relativsätzen", Zeitschrift für französische Sprache und Literatur 87: 43-48.

1977 (ed.) On the logical analysis of tense and aspect. Tübingen: Narr.

1978 (ed.) Papers on tense, aspect and verb classification. Tübingen: Narr.

1980 (ed.) Time, tense and quantifiers. Tübingen: Niemeyer.

1986 "Indirect discourse and 'consecutio temporum'", in: Vincenzo Lo Cascio - Co Vet (eds.), 79-97.

424 References

Rutherford, William E. 1970 "Some observations concerning subordinate clauses in

English", Language 46-1: 97-115. Salkie, Raphael

1987 "Review of Dahl (1985)", Lingua 72: 77-99. 1989 "Perfect and pluperfect: what is the relationship?",

Journal of Linguistics 25: 1-34. Sampson, Geoffrey

1971 "Subordinate future deletion and hyperclauses", Lin-guistic Inquiry 2: 587-589.

Scheurweghs, Gaston 1961. (f.p. 1959) Present-day English syntax. London:

Longman. Schlyter, Suzanne

1979 "Point of observation and time indications with movement verbs", in: Thore Petterson (ed.), 111-126.

Schöpf, Alfred 1984 Das Verzeitungssystem des Englischen und seine

Textfunktion. Tübingen: Niemeyer. 1985 "Das Verzeitungssystem des Englischen und seine

Textfunktion", in: Werner Hüllen - Rainer Schulze (eds.), 83-101.

1987 "The past tense in English", in: Alfred Schöpf (ed.), 177-220.

1987 (ed.) Essays on tensing in English. Vol. I: Reference time, tense and adverbs. Tübingen: Niemeyer.

1989 "The temporal structure of narrative texts", in: Alfred Schöpf (ed.), 247-290.

1989 (ed.) Essays on tensing in English. Vol. II: Time, text and modality. Tübingen: Niemeyer.

1991 "The analysis and reconstruction of the temporal structure of narrative texts", in: Jadranka Gvozdanovit - Theo Janssen (eds.), 237-253.

References 425

Shi, Ziqiang 1990 "On the inherent aspectual properties of NPs, verbs,

sentences and the decomposition of perfectivity and inchoativity", Word AMY. 47-67.

Smith, Carlota 1976 A theory of auxiliary 'have' in English. Bloomington:

Indiana University Linguistics Club. 1977 "The temporal interpretation of complements in En-

glish", Texas Linguistic Forum 6: 150-168. 1978 "The syntax and interpretation of temporal expres-

sions in English", Linguistics and Philosophy 2: 43-99.

1981 "Semantic and syntactic constraints on temporal in-terpretation", in: Paul Tedeschi - Annie Zaenen (eds.), 213-237.

1982 "Aspect and aspectual choice", Texas Linguistic Forum 19: 167-199.

1983 "A theory of aspectual choice", Language 59: 479-501.

1991 The parameter of aspect. Dordrecht: Kluwer Aca-demic Publishers.

Smith, Neil V. 1981 "Grammaticality, time and tense", Philosophical

Transactions of the Royal Society of London. B. 259: 253-265.

S0rensen, Holger Steen 1964 "On the semantic unity of the perfect tense", in En-

glish studies presented to R. W. Zandvoort on his 70th birthday. Amsterdam: Swets and Zeitlinger. 74-83.

Sperber, Dan - Deirdre Wilson. 1986 Relevance. London: Blackwell.

Steedman, Mark 1982 "Reference to past time. Speech, place and action", in:

Robert J. Jarvella - Wolfgang Klein (eds.), 125-157.

426 References

Stegu, Martin 1983 "Zur Interpretation von Verbalhandlungen als

'aufeinanderfolgend' oder 'gleichzeitig'", in: Wolfgang Meid - Hans Schmeja (eds.), 319-329.

Steinberg, Danny D. - Leon Jakobovits (eds.) 1971 Semantics: an interdisciplinary reader, Cambridge:

Cambridge University Press. Sutton, Laurel A. - Christopher Johnson - Ruth Shields (eds.)

1991 Proceedings of the seventeenth annual meeting of the Berkeley Linguistics Society. University of California, Berkeley, California.

Sweet, Henry 1891 New English grammar. Part I. Oxford: Clarendon

Press. Tedeschi, Philip - Annie Zaenen (eds.)

1981 Syntax and semantics. Volume 14: Tense and aspect. New York: Academic Press.

Ter Meulen, Alice 1983 (ed.) Studies in modeltheoretic semantics. Dordrecht:

Foris. 1991 "Shifting of reference time and perspective", in:

Laurel A. Sutton - Christopher Johnson - Ruth Shields (eds.), 520-530.

Thomson, A.J. and A.V. Martinet. 1980 A practical English grammar. Oxford: Oxford

University Press. Tottie, G. - G. Övergaard.

1984 "The author's would - a feature of American English", Studia Linguistica 38: 148-164.

Tregidgo, P.S. 1974 "English tense usage: a bull's eye view", ELT Journal

28: 97-107. 1979 "Tense subordination", ELT Journal 33: 191-197.

References 427

Tuite, Kevin - Robinson Schneider - Robert Chametzky (eds.) 1982 Papers from the eighteenth Regional Meeting

Chicago Linguistic Society. Chicago: Chicago Lin-guistic Society.

Vanneck, Gerard 1958 "The colloquial preterite in Modern American En-

glish", Word 14: 237-242. Van Voorst, Jan

1987 "The structure of aspect", The Canadian Journal of Linguistics 32/3: 261-277.

Vasudeva, Harender Nath 1971 Tense and aspect in English. Ph.D. Dissertation, Uni-

versity of Michigan. Vendler, Zeno

1967 Linguistics in Philosophy. Ithaca, New York: Cornell University Press. 97-121.

Verkuyl, Henk 1972 On the compositional nature of the aspects.

Dordrecht: Reidel. 1989 "Aspectual classes and aspectual composition", Lin-

guistics and Philosophy 12: 39-94. Vermant, Stefan

1983 The English present perfect: a dynamic-synchronic approach. Antwerp: Universitaire Instelling Antwer-pen, Dept. Germaanse, Afdeling Lingui'stiek.

Vet, Co - Arie Molendijk 1986 "The discourse function of the past tenses in French",

in: Vincenzo Lo Cascio - Co Vet (eds.), 133-159. Vetter, David C.

1973 "Someone solves this problem tomorrow", Linguistic Inquiry 4: 104-108.

Vikner, Sten 1985 "Reichenbach revisited: one, two, or three temporal

relations?", Acta Linguistica Hafniensia 19/2: 81-98.

428 References

Vlach, Frank 1981 "The semantics of the progressive", in: Paul Tedeschi

- Annie Zaenen (eds.), 271-292. Wachtel, Tom

1982 "Some problems in tense theory", Linguistic Inquiry 13: 336-341.

Wekker, Herman 1976 The expression of future time in contemporary British

English. Amsterdam: North-Holland. 1980 "Temporal subordination in English", in: Wim Zon-

neveld - Fred Weerman (eds.), 96-103. Wilson, Deidre

1990 "Pragmatics and time", Linguistics Department, Uni-versity College London. [Unpublished MS.]

Woisetschläger, Erich 1976 A semantic theory of the English auxiliary system.

Bloomington: Indiana University Linguistics Club. Wood, Frederick T.

1961 The groundwork of English grammar. London: Macmillan.

Young, David J. 1980 The structure of English clauses. London: Hutchin-

son. Zandvoort, R.W.

1932 "On the perfect of experience", English Studies 14: 11-20, 76-79.

1964 (f.p. 1945) A handbook of English grammar, Groningen: J.B. Wolters.

Zonneveld, Wim - Fred Weerman (eds.) 1980 Linguistics in the Netherlands 1977-1979, Dordrecht:

Foris. Zydatiss, Wolfgang

1976 Tempus und Aspekt im Englischunterricht. Kron-berg/Ts: Scriptor.

Index

Index of subjects

Absolute past tense 81, 112-120, 262-264

Absolute sector, definition 55 Absolute tense, definition 56-

57 Absolute-relative tense 65-66,

171 Anchored time adverbial 195,

201-205 Anteriority

in the past sector 84-112 in the pre-present sector 141-159 in the post-present sector 179-186, 195,201-204

(A)telicity definition 6-10 and NPs 25-27 and directional PPs 27 andthePUTI 266-273 and the use of tense in the post-present sector 186-190 and adverbials 34-37 and progressive 10,47-49

Backgrounding cf. foregrounding

Binding TO, definition 58 Conditional meaning 151,

168-169, 180, 200, 204, 214, 217,218, 227,231,238

Conditional tense 60, 120-131, 133-134, 138, 197-198

Contextual effect, definition 83-84

Corpus, description of 75 CRCT 123-124 CRPP 84-98

definition 84-91 and adverbials 91-96 and verbs of propositional attitude 96-98

CRPrP 146-153 Difference RRC-NRRC 342-

348 Past sector 106-112, 131-133 Pre-present sector 162-167, 235 Post-present sector 171-176, 196, 221-225 andthePUTI 273-282 and binding TO 308-309, 333-338

Domain-internal relations Past sector 60-61,77-82, 120-122 Pre-present sector 61-63, 137-140, 159-162 Present sector 64 Post-present sector 64-66, 168-171, 176-186

Elicitation test 359-364 Established time (TE),

definition 29 Foregrounding 111, 260-261

and (un)boundedness, (a)telicity 266-273 definition 252-254

430 Index

exceptions 256-258 inRCs 273-282

Future perfect 3, 65-66, 180-185, 197-203,230-231,263-264

Future Perspective System 65-66

Gnomic 218-220 Gradual development 242-244 Habitual situation 18, 19, 32,

35, 94, 101, 164, 172, 173, 191, 196, 226-227

Heterogeneous situation 20-22 Homogeneous situation 20-22 (Im)perfectivity 11-12,14 Implicature past tense 42-47,

321-324 (In)direct binding 68-72, 1 Π-

Ι 18, 310-338 Influence of adverbials

on (un)boundedness 38-42, 90 on (a)telicity 34-37

on tense in post-present sector 194-205

Intensional context 98-101, 124-126, 133, 177, 179, 184, 188, 231

Linear order RC-SUPC and use of tense in post-present 205-218 andthePUTI 282-285

Main clause, definition 82 Matrix clause, definition 82 Mutually manifest 83-84 Mutually substitutable 83 NPs

attributive 213-214,218,237

referential 218, 392 and (a)telicity 25-27 and (un)boundedness 25-27 and the use of tense in the post-present 190-194 andthePUTI 276-282

Perfect Continuative 31,62,80-81, 102-103, 140, 150, 152, 184-186, 242, 329-330 Indefinite 61-62, 137-139, 179-184 Repetitive 62-63, 139, 144 Status of perfect 383-384

Posteriority in the past sector 120-134

Prediction 121, 126-131, 173, 174, 181-184, 193, 197-198, 202, 209,215, 228-231,236, 237, 330-333

Present Perspective System 64-65

Processing effort 83-84 Progressive 9, 10, 11-12, 14,

31,39-41,47-49, 87, 186-190,212,217-218

Punctual situation 18-19,22-23, 24 90

PUTI definition 66-68, 250-252, 260-265 and tense 285-291 and linear order RC-SUPC 282-285 and (un)boundedness 266-272 and (a)telicity 266-273

Recency 244-248 Re-establishment of domain

170

Index 431

Relative clause continuative 107, 129, 133, 270, 272, 275 definition 74 Cf. difference RRC-NRRC pushdown 96-97,98-100

Relative tense definition 56-57 past tense 81 112-120

Repetitive situation 24, 34, 35, 37, 50, 62-63, 88-89, 139, 144, 159

Represented Speech and Thought 99-101, 128, 130

Resultativeness 239-242, 327-328

Rule of R-progression 307 Definition 254-256 Exceptions 256-258

Shift of domain, definition 58-59

Shift of perspective 61-62, 64-65, 73, 126, 137-138, 139, 146-147, 159-162, 168-169, 173, 175-176, 179, 202

Simultaneity in the past sector 112-120 in the pre-present sector 138, 140 in the post-present sector 178-179, 186-193,205-220

Situation time of orientation 57-58

Sloppy simultaneity 73, 115, 130, 151, 196, 300, 376-377

Specificational sentences 206-209,215

Subinterval property 20-22

Superordinate clause, definition 82

Temporal vs. chronological 59-60

Temporal domain, definition 56-57

Temporal zero-point 55 Temporal subordination,

definition 56-57 Tests

for (un)boundedness 17-24 for (a)telicity 17-24 for relative and absolute tenses 113-120,262-264

Time of orientation, definition 57-58

Time of the situation (TS) 368-369

Time sphere, definition 55 TO2 57-58, 293-310 Unanchored time adverbial

195-201 (Un)boundedness

definition 6-11 and NPs 25-27 and directional PPs 27 and adverbials 38-42,90 and progressive 9-12, 47-49 and negation 49 left boundedness 7, 24-25, 29, 42, 52-53, 267, 268 right boundedness 7, 24-25, 52-53, 267 and tense 28-33,89-91

Ungrammaticality 83 (World-Relations, definition

59-60 Zero-boundedness 24, 50-52

432 Index

Zero-telicity 18,22,23-24,35, 37, 50-52, 89,91,95, 106-107, 123, 124, 131, 147, 152, 155, 272-273

Index of authors

Adelaar - Lo Cascio 186, 286, 293,299, 300,315,328, 329, 378, 395, 399,400,402

Allen 2,3,31,47, 169-170, 200, 222-223, 365, 366, 367-368, 373, 388, 389

Aristar Dry 257, 268-269, 274-275

Bache - Jakobsen 74 Bauer 30, 383 Berezovsky 79, 373, 393 Bertinetto 297, 298-299, 375,

378, 385,400, 401 Binnick 30,48, 51-52, 180,

287, 365,368, 371,377, 388, 389

Bouscaren et al 79, 99, 102-105, 232-233, 242, 296, 380

Brecht 79, 383, 387 Brinton 368 Buyssens 93,211 Comrie 12,28,47,171,174,

180-181,201-203,204,366, 378, 379, 384, 387, 392, 394, 395, 396-397

Costa 79, 90, 377, 379 Couper-Kuhlen 73, 254, 256,

257, 258, 267-268, 269, 272, 273,285, 287, 371,373, 396, 400-401

Dahl 10, 13-15, 134,248,365, 366, 375

Declerck 6, chapter 2, 80, 98, 105, 112, 113, 117, 119, 137-140, 144, 147, 150, 160-162, 168-171, 172, 174, 186, 204, 206, 209, 218, 220, 250-252, 255, 256, 258, 260-263, 264-265, 266, 267, 269, 273, 282, 285, 288-289, 293, 294, 296, 299, 300, 301, 302, 303, 308-309,310-311,328-329, 366, 368, 369, 371,372, 373,376, 378, 379, 381,382, 383, 384, 385,388, 389, 391,392, 394, 399,401,402

Declerck - Depraetere 177, 179

Defroment 30, 95, 159, 287 Depraetere 117,365 De Vuyst 10, 377-378, 380-

381 Dietrich 393 Dillon 20, 383 Donnellan 213, 237, 392 Dowty 22,49,52,78, 121,

134, 256, 257, 258, 272, 284, 287,368, 370, 371,384, 396

Dry 52, 94, 253, 257, 267, 268 Ehrlich 100, 125, 252, 379,

380 Eisterhold 11-12, 267, 396 397 Feigenbaum 383, 388 Fenn 78, 88, 373, 375, 383,

384,388, 394 Fleischman 174, 253 Grice 45, 83, 121, 142,207,

250 Guenther 386

Index 433

Haegeman 204,282,288,391, 392, 399

Haegeman - Wekker 2, 73, 222-223

Hatav 10,42,257,267 Heinämäki 23, 24, 365, 371 Hinrichs 53,255,269 Hirtle 12, 17, 169, 223, 384 Hopper 254, 375 Hornstein 133-135, 293, 375,

384, 387, 390 Huddleston 73, 74, 375, 376,

379, 383,384, 385,391,395 Inoue 384,393,394 Jespersen 107, 129, 270, 373,

383,395 Joos 383 Klein 388 Kruisinga 169,373,388,391 Labov - Waletzky 252 Ladusaw 86, 119, 171,384,

387, 388, 39 Lo Cascio - Rohrer 73, 170,

171, 177, 180, 293, 375, 378 Leech 32, 169,218,247,373,

379, 383, 384, 388, 389, 391, 392, 393, 394

McCawley 247, 254-255, 375, 386, 389 McCoard 23, 138,287,368,

373, 375, 379, 382, 383, 386, 387, 393, 395, 401

Mittwoch 20, 369, 371, 375 Moens 10,15-16,19,20,32,

367,368, 370, 371,386 Mommer 20, 23, 24, 32, 33-

34, 47, 116, 366, 370, 371, 383,394

Mourelatos 25, 365, 368, 371 Nehls 384,388,391 Nerbonne 256, 257, 258, 267,

396, 397 Ota 2, 375 Palmer 383, 384, 385, 393,

394 Partee 73, 180, 255-256, 365,

387, 389, 396 Penhalluriack 204, 384, 386,

388, 391 Poutsma 72, 139, 373, 383,

395 Quirk et al 2, 11,34,38,68,

82, 87,91,92, 96, 222, 223, 370, 374, 383, 385, 387, 389, 392, 394

Reichenbach 30 Reinhart 253, 273-274, 301 Richards 121,171,186,256,

387 Riddle 28, 43, 47, 194, 377,

379 Rigter 121,379,384,385,387,

388, 389 Rohrer 76, 302, 379, 384, 385,

387 Salkie 78,88, 131,242,375,

380 Schöpf 20, 33, 34, 52, 194-

195, 257, 258, 268, 307, 365, 371,373,375,383, 386, 388, 389

Shi 26 Smith 10, 11, 19,23,24,30,

33, 46, 47, 92, 120, 257, 293, 302, 303, 366, 367, 368, 371, 373,375, 387, 388, 389, 391

434 Index

Sperber - Wilson 83,84 Sweet 395 Tregidgo 2, 223, 224, 373,

375, 378, 379, 387, 388, 389, 392, 395, 400, 401

Vasudeva 32, 373, 389 Vendler 365 Verkuyl 6, 20, 21, 23, 370 Vermant 140,386 Vet - Molendijk 257, 399 Wekker 169, 172, 388, 389 Zandvoort 295, 385, 399 Zydatiss 20,21,23,47-48,

365, 366 , 368, 370, 383

Documents

3110146851_avax.pdf