ADVERBIAL MODIFICATION STUDIES IN LINGUISTICS AND PHILOSOPHY formerly Synthese Language Library Managing Editors: ROBIN COOPER, University of Wisconsin ELlSABET ENGDAHL, University of Wisconsin RICHARD GRANDY, Rice University Editorial Board: EMMON BACH, University of Massachusetts at Amherst JON BARWISE, CSL!, Stanford JOHAN VAN BENTHEM, Rijksuniversiteit Groningen DAVID DOWTY, Ohio State University, Columbus GERALD GAZDAR, University of Sussex, Brighton EWAN KLEIN, University of Edinburgh BILL LADU SA W, University of California 'at 'Santa Cruz SCOTT SOAMES, Princeton University HENRY THOMPSON, University of Edinburgh VOLUME 28 M. J. CRESSWELL Department of Philosophy Victoria University of Wellington ADVERBIAL MODIFICATION Interval Semantics and Its Rivals D. REIDEL PUBLISHING COMPANY A MEMBER OF THE KLUWER ACADEMICPUBUSHERSGROUP DORDRECHT / BOSTON / LANCASTER /TOKYO Library of Congress Cataloging in Publication Data Cresswell, M. J. Adverbial modification. (Studies in linguistics and philosophy; v. 28) Bibliography: p. Includes indexes. 1. Grammer, Comparative and general-Adverb. 2. Semantics (Philosophy) I. Title. II. Series. P284.C74 1985 415 85-24368 ISBN-13: 978-90-277-2060-3 e-ISBN-13: 978-94-009-5414-4 001: 10.1 007/978-94-009-5414-4 Published by D. Reidel Publishing Company, P.O. Box 17,3300 AA Dordrecht, Holland. Sold and distributed in the U.S.A. and Canada by Kluwer Academic Publishers, 190 Old Derby Street, Hingham, MA 02043, U.S.A. In all other countries, sold and distributed by Kluwer Academic Publishers Group, P.O. Box 322, 3300 AH Dordrecht, Holland. All Rights Reserved © 1985 by D. Reidel Publishing Company, Dordrecht, Holland and copyrightholders as specified on appropriate pages within No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical including photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner Softcover reprint of the hardcover 1s t edition 1985. TABLE OF CONTENTS ORIGINAL PUBLICATION DETAILS vi PREFACE vii INTRODUCTION CHAPTER I / Adverbs and Events 13 CHAPTER II / Adverbs of Space and Time 41 CHAPTER III / Interval Semantics and Logical Words 67 APPENDIX TO CHAPTER III (1985) 85 CHAPTER IV / Prepositions and Points of View 97 CHAPTER V / Interval Semantics for Some Event Expressions 143 CHAPTER VI/Adverbs of Causation 173 CHAPTER VII / Adverbial Modification in Situation Semantics 193 BIBLIOGRAPHICAL INDEX 221 GENERAL INDEX 226 ORIGINAL PUBLICATION DETAILS Chapter I: in Synthese 28 (1974) pp. 455-481. Chapter II: in Formal Semantics and Pragmatics for Natural Language (ed. F. Guenthner and S. J. Schmidt), Reidel, Dordrecht, 1978,pp.171-199. Chapter III: in On the Logical Analysis of Tense and Aspect (ed. Ch. Rohrer) Gunter Narr, Tiibingen, 1977, pp. 7-29. (Appendix specially written for this volume.) Chapter IV: in Linguistics and Philosophy 2 (1978), pp. 1-41. Chapter V: in Semantics from Different Points of View (ed. R. Bauerle, U. Egli and A. von Stechow), Springer-Verlag, Berlin, 1979, pp.90-116. Chapter VI: in Words, Worlds and Contexts (ed. H-J. Eikmeyer and H. Rieser), W. de Gruyter, Berlin, 1981, pp. 21-37. Chapter VII: specially written for this volume. PREFACE Adverbial modification is probably one of the least understood areas of linguistics. The essays in this volume all address the problem of how to give an analysis of adverbial modifiers within truth-conditional semantics. Chapters I-VI provide analyses of particular modifiers within a possible worlds framework, and were written between 1974 and 1981. Original publication details of these chapters may be found on p. vi. Of these, all but Chapter I make essential use of the idea that the time reference involved in tensed sentences should be a time interval rather than a single instant. The final chapter (Chapter VII) was written especially for this volume and investigates the question of how the 'situation semantics' recently devised by Jon Barwise and John Perry, as a rival to possible-worlds semantics, might deal with adverbs. In addition I have included an appendix to Chapter III and an introduction which links all the chapters together. I would like to thank the following for permission to reprint the articles which form Chapters III, V and VI: Verlag Gunter Narr, Tilbingen (Chapter III); Springer-Verlag, Berlin and Heidelberg (Chapter V) and W. de Gruyter, Berlin (Chapter VI). I would also like to thank the German Science Founda tion, whose funding supported projects or conferences in which I had the opportunity to participate. Much of what is in this book was the result of the contacts I made on those occasions, particularly at the University of Stuttgart and the University of Konstanz. More specific acknowledgments will be found in the notes to particular chapters. I can only hope that pre senting this material in one place will encourage further work on these difficult topics. Wellington, New Zealand March 1985 vii INTRODUCTION I once described as the 'most certain principle' of semantics the fact that if two sentences a: and {3 are such that a: is true and {3 is false then a: and {3 do not mean the same [10, p. 69] . It is perhaps more a slogan than a prin ciple, and, like all slogans, requires explanation and qualification. Neverthe less, it still does seem to me the most certain thing I know about meaning, and any adequate semantic theory must, in my view, give content to this principle and respect it. One qualification is the fact that most, if not all, sentences of ordinary language do not have truth values in isolation. To know whether (1) Last summer was windy is true or false we must know at least the time and place the sentence refers to, and to know whether (2) France is hexagonal is true we must know what are the acceptable standards of precision required for hexagonality in the particular context. So a theory of meaning must at least provide for a sentence to have a truth value relative to a context of use, whatever that may turn out to be. Another qualification that many authors would introduce is that truth and falsity only apply to indicative sentences (see for instance [15, p. 252f]). They would say that it makes no sense to ask whether (3) Bring me some hokey pokey ice cream or (4) Did you ever meet Dean Goffin? can be true or false. My own view about (3) or (4), as expressed in [6, pp. 228-238] or [8, pp. 21-25], is that such sentences do have truth values, and that the reason that they are often thought not to, is that people think of truth and falsity as a property, not of a sentence relative to a context, but as a property of utterances being used as assertions which purport to be giving information. Those who disagree with me on this will have to see 2 INTRODUCTION the articles in this book as applying only to indicative sentences, with adapta tions to be undertaken later for the other moods. The most certain principle is not the only seman tical principle. Equally important is what in [6, p. 19] I called Frege's Principle, to the effect that the meaning of a complex expression is a function of the meanings of its simpler parts.1 This principle, in conjunction with the most certain principle, imposes quite substantial constraints on a semantical theory. Look first at an illustration from propositional modal logic. The classical propositional calculus may be regarded as the minimal seman tical theory to respect the most certain principle. For in PC the meanings of sentences may simply be identified with truth values, and the operators which make more complex sentences out of simpler ones have as their meanings functions from truth values to truth values - in other words, they are all truth functional. Now if we add a symbol 0 to mean necessarily, we can see that the combination of the most certain principle and Frege's principle shows that meanings cannot be identified with truth values - for the following reason: Let p be a proposition which is true but only contingently so, say that Waikanae is a district community. And let q be a proposition which is true by necessity, say that KM has the finite model property. Then, if the mean ings of p and q were just their truth values, by Frege's principle the meanings of Dp and Dq would be identical. But Dp is false and Dq is true and so, by the most certain principle, the meanings of Dp and oq are not identical, which means that the meanings of p and q cannot be truth values. This argument tells us what the meanings of p and q are not, but it does not tell us what the meanings of p and q are. To answer that question we have to proceed in a piecemeal way. We ask questions like: what should the meaning of p and q be to get the truth conditions of Dp and Dq right? And we then ask this same question of other words as well. In the case of 0 the answer is that we need to know the truth value of p and q not only in the actual world but in every other possible world. We can then say that Dp is false because, although Waikanae is a district community in the actual world, it could easily have been the case either that its popUlation was big enough for it to be declared a borough, or so small that it was not incor porated at all. Notice that in order to know that this is possible we do not need to know what is actually the case. One plausible view of semantic competence is the view that knowing the meaning of a sentence just is knowing the conditions under which it is true. And one way of giving content to this view is to claim that the meaning of a sentence in a context is just a set of possible worlds. It should be clear that one could know that a INTRODUCTION 3 sentence would be true in this or that possible world without knowing whether that world is the actual one. So, knowing meanings does not imply knowing truth values. The operator 0 represents the adverb necessarily. This adverb is a senten tial modifier, which makes a more complex sentence out of a simpler one. The argument just presented can be restated by saying that, in order to get an adequate semantics for necessarily which satisfies both the most certain principle and Frege's principle, the meaning of a sentence must be at least a set of possible worlds. The argument is not completely watertight because of course necessarily might not be a genuine sentential operator, it might be a predicate of sentences, and there might be hidden quotation marks round the p. Arguments of the kind just outlined are always relative to having isolated the correct 'logical form' of the expressions being studied. What I try to do in this book is set out what I consider to be the simplest logical forms for sentences with adverbial modifiers and ask what sorts of entities their meanings must be in order to determine the correct truth conditions of the sentences in which they occur. One of the advantages of possible-worlds semantics is that in making the meaning of a sentence in context a set of possible worlds, an explanation is automatically given of such logical relations between sentences as logical incompatibility and logical entailment. These relations emerge, as I have argued in [8] and [9] that they should, as relations of meaning or content and not of form. Two propositions are incompatible if they have no world in common, and one proposition entails another if it is included in that other. The reader will not find in this book any 'logic of modifiers' in the sense of structural principles willch explain logical relations. This is because I believe that very few, if any, logical relations have a structural explanation, and none are independent of the meanings of words. The con text-dependence is important too, because logical relations do not hold between sentences in isolation, but at best between sentences in context by virtue of the relations between the propositions they express in those contexts. I have explained this in some detail in [7] . Adverbial modification presents a challenge to logically-based semantical frameworks for the reasons stated on p. 41f, and it is therefore important to see how it may be accommodated. There are basically two approaches. One has its origin in a paper by Donald Davidson [12], and construes adver bial phrases as being predicates of events. Davidson formalizes tills in ordinary first-order logic and he and his followers (e.g. [21] and [24]) think of this as one of the attractive features of this way of dealing with adverbs.