Words and Other Wonders: Papers on Lexical and Semantic Topics 9783110219128, 9783110190427

Cognitive Linguistics has given a major impetus to the study of semantics and the lexicon. The present volume brings tog

370 68 19MB

English Pages 503 [504] Year 2006

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Frontmatter
Table of contents
Publication sources
Chapter 1. Prospects and problems of prototype theory
Chapter 2. Where does prototypicality come from?
Chapter 3. The semantic structure of Dutch over
Chapter 4. Salience phenomena in the lexicon. A typology
Chapter 5. Vagueness's puzzles, polysemy's vagaries
Chapter 6. Classical definability and the monosemic bias
Chapter 7. The semantic structure of the indirect object in Dutch
Chapter 8. The interaction of metaphor and metonymy in composite expressions
Chapter 9. Looking back at anger. Cultural traditions and metaphorical patterns
Chapter 10. Beer and semantics
Chapter 11. Cultural models of linguistic standardization
Chapter 12. Caught in a web of irony: Job and his embarassed God
Chapter 13. The lexicographical treatment of prototypical polysemy
Chapter 14. The definitional practice of dictionaries and the cognitive semantic conception of polysemy
Chapter 15. Cognitive grammar and the history of lexical semantics
Chapter 16. The theoretical and descriptive development of lexical semantics
Chapter 17. Idealist and empiricist tendencies in cognitive semantics
Backmatter
Recommend Papers

Words and Other Wonders: Papers on Lexical and Semantic Topics
 9783110219128, 9783110190427

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Words and Other Wonders



Cognitive Linguistics Research 33

Editors Dirk Geeraerts Rene´ Dirven John R. Taylor Honorary editor Ronald W. Langacker

Mouton de Gruyter Berlin · New York

Words and Other Wonders Papers on Lexical and Semantic Topics by Dirk Geeraerts

Mouton de Gruyter Berlin · New York

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

앝 Printed on acid-free paper 앪

which falls within the guidelines of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Geeraerts, Dirk, 1955⫺ Words and other wonders : papers on lexical and semantic topics / by Dirk Geeraerts. p. cm. ⫺ (Cognitive linguistics research ; 33) Includes bibliographical references. ISBN-13: 978-3-11-019042-7 (hardcover : alk. paper) ISBN-10: 3-11-019042-7 (hardcover : alk. paper) 1. Semantics. 2. Lexicology. I. Title. P325.G43 2006 4011.43⫺dc22 2006018773

Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at ⬍http://dnb.ddb.de⬎.

ISBN-13: 978-3-11-019042-7 ISBN-10: 3-11-019042-7 ISSN 1861-4132 쑔 Copyright 2006 by Walter de Gruyter GmbH & Co. KG, 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 mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printed in Germany

Preface

Granunar and lexicon being traditionally separated in mainstream twentiethcentury linguistics, one of the great things about Cognitive Linguistics is the opening it has created towards a renewed study of the lexicon and an integration of lexical semantics into theoretical granunar at large. In fact, the study of the lexicon plays a crucial role in Cognitive Linguistics: questions about conceptual structure and semantic mechanisms that lie at the heart of Cognitive Linguistics, belong to the time-honored expertise of lexical semantics. In the present collection, I bring together a representative sample of my own contributions to lexicology and lexical semantics within the framework of Cognitive Linguistics (including the extension of insights from lexical semantics to other areas of linguistics). In the fifteen years covered by this collection, from 1988 to 2003, the central topics of my research led to a number of monographs: Diachronic Prototype Semantics (1997) synthesizes the research into historical semasiology that dominated much of my work in the 1980s, and two co-authored monographs, The Structure of Lexical Variation (1994) and Convergentie en Divergentie in de Nederlandse Woordenschat (1999) embody the corpus-based studies of onomasiological variation and change that I developed with Dirk Speelman, Stefan Grondelaers and the other members of my Leuven research group in the 1990s, and that continue to dominate our present research efforts. While these two central fields of research are only indirectly represented in the present volume, this collection predominantly brings together a number of papers devoted to lexicological and semantic topics that I dealt with more sporadically or in parallel to these major foci. Thematically speaking, I have consistently tried to choose papers that represent a variety of issues and topics within the broadly defined field of lexicology and lexical semantics. Accordingly, the chapters are organized in thematic groups. The first section deals with prototypicality as a theoretical and practical model of semantic description. The second section discusses polysemy and criteria for distinguishing between meanings. The third section tackles questions of meaning description beyond the level of words, on the level of idioms and constructions. The following section casts the net even wider, dealing with the cultural aspects of meaning. Moving away from the theoretical and descriptive perspective towards applied concerns, the fifth section looks at lexicography from the point of view of Cognitive Linguis-

VI

Preface

tics. The final section has a metatheoretical orientation: it discusses the history and methodology of lexical semantics. Theoretically, the papers are interconnected by a number of accents that mark their specific position within the varied landscape of Cognitive Linguistics: a continued attempt to provide solid empirical foundations for semantics, an emphasis on the multivariate nature of semantic structure, and an insistence on the social and cultural background of language. In the research goals that we are currently pursuing in Leuven, these accents combine in multifactorial corpus-based studies in the field of cognitive sociolinguistics - studies, that is, in which corpus materials provide the firm empirical basis for research, in which linguistic phenomena are statistically analysed with the help of multivariate techniques, and in which social and cultural variation is explicitly included into the multifactorial model. Except for necessary corrections, small additions, and some typographical reorganization (and unless otherwise stated), the chapters are reprinted in their original form. All chapters are reprinted with permission of the original publishers. Each paper is preceded by a brief introduction that situates the text against the period in which it was first written, but that also points to further developments, in my own research or in Cognitive Linguistics at large. In this way, the collection should give the reader an idea of the dynamism of Cognitive Linguistics in the crucial period when it developed from a more or less marginal approach to one that is firmly situated in the mainstream of contemporary linguistics. Knowledge is not produced in isolation, and it befits any researcher to thank those who have provided guidance and motivation in the course of his career. In gratitude, I dedicate this collection to the memory of Brygida Rudzka-Ostyn (1939-1998). Brygida's impact on the international expansion of Cognitive Linguistics can hardly be underestimated. With her infinite enthusiasm and her sense for initiative, she played a crucial rol in the second half of the 1980s in bringing Cognitive Linguistics from its Californian homeground to Europe and beyond. Without the warm stimulus she then provided for a young scholar to place his work in a wider context, none of the papers collected here might ever have been written.

Table of contents

Preface Publication sources

V

IX

Section 1. Prototypicality and salience

1 2 3 4

Prospects and problems of prototype theory Where does prototypicality come from? The semantic structure of Dutch over Salience phenomena in the lexicon. A typology

3 27

48

74

Section 2. Polysemy

5 6

Vagueness's puzzles, polysemy's vagaries Classical definability and the monosemic bias

99 149

Section 3. Constructions and idioms 7 8

The semantic structure of the indirect object in Dutch The interaction of metaphor and metonymy in composite expressions

175 198

Section 4. Meaning and culture

9 10 11 12

Looking back at anger. Cultural traditions and metaphorical patterns Beer and semantics Cultural models of linguistic standardization Caught in a web of irony: Job and his embarassed God

227 252 272 307

Section 5. Lexicography

13 14

The lexicographical treatment of prototypical polysemy The definitional practice of dictionaries and the cognitive semantic conception of polysemy

327 345

Section 6. Theory and method in lexical semantics

15 16 17

Cognitive grammar and the history oflexical semantics The theoretical and descriptive development of lexical semantics Idealist and empiricist tendencies in cognitive semantics

367 398 416

Vlll

Table ofcontents

References Index of names Index of subjects

445 479 485

Publication sources

Chapter 1 - Geeraerts, Dirk. 1989. Prospects and problems of prototype theory. linguistics 27: 587-612. Chapter 2 - Geeraerts, Dirk. 1988. Where does prototypicality come from? In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 207-229. AmsterdamJPhiladelphia: John Benjamins. Chapter 3 - Geeraerts, Dirk. 1992. The semantic structure of Dutch over. Leuvense Bijdragen 81: 205-230. Chapter 4 - Geeraerts, Dirk. 2000. Salience phenomena in the lexicon. A typology. In Meaning and Cognition, Liliana Albertazzi (ed.), 125-136. AmsterdamJPhiladelphia: John Benjamins. Chapter 5 - Geeraerts, Dirk. 1993. Vagueness's puzzles, polysemy's vagaries. Cognitive Linguistics 4: 223-272. Chapter 6 - Geeraerts, Dirk. 1994. Classical definability and the monosernic bias. Rivista di Linguistica 6: 189-207. Chapter 7 - Geeraerts, Dirk. 1998. The semantic structure of the indirect object in Dutch. In The Dative 11. Theoretical Approaches, Willy Van Langendonck and Wiliam Van Belle (eds.), 185-210. AmsterdamJPhiladelphia: John Benjamins. Chapter 8 - Geeraerts, Dirk. 2002. The interaction of metaphor and metonymy in composite expressions. In Metaphor and Metonymy in Comparison and Contrast, Rene Dirven and Ralf Parings (eds.), 435-465. BerlinINew York: Mouton de Gruyter. Chapter 9 - Geeraerts, Dirk, and Stefan Grondelaers. 1995. Looking back at anger. Cultural traditions and metaphorical patterns. In Language and the Construal of the World, John Taylor and Robert E. MacLaury (eds.), 153-180. BerlinINew York: Mouton de Gruyter. Chapter 10 - Geeraerts, Dirk. 1999. Beer and semantics. In Issues in Cognitive Linguistics, Leon De Stadler and Christoph Eyrich (eds.), 35-55. BerlinINew York: Mouton de Gruyter. Chapter 11 - Geeraerts, Dirk. 2003. Cultural models of linguistic standardization. In Cognitive lvfodels in Language and Thought. Ideology, Metaphors and Meanings, Rene Dirven, Roslyn Frank, and Martin Piitz (eds.), 25-68. BerlinINew York: Mouton de Gruyter.

x

Publication sources

Chapter 12 - Geeraerts, Dirk. 2003. Caught in a web of irony: Job and his embarassed God. In Job 28. Cognition in Context, Ellen Van Wolde (ed.), 37-55. Leiden: Brill. Chapter 13 - Geeraerts, Dirk. 1990. The lexicographical treatment of prototypical polysemy. In Meanings and Prototypes. Studies in Linguistic Categorization, Savas L. Tsohatzidis (ed.), 195-210. London: Routledge. Chapter 14 - Geeraerts, Dirk. 2001. The definitional practice of dictionaries and the cognitive semantic conception of polysemy. Lexicographica 17: 6-21. Chapter 15 - Geeraerts, Dirk. 1988. Cognitive grammar and the history oflexical semantics. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 647677. Amsterdam/Philadelphia: John Benjamins. Chapter 16 - Geeraerts, Dirk. 2002. The theoretical and descriptive development of lexical semantics. In The Lexicon in Focus. Competition and Convergence in Current Lexicology, Leila Behrens and Dietmar Zaefferer (eds.), 23-42. Frankfurt: Peter Lang Verlag. Chapter 17 - Geeraerts, Dirk. 1999. Idealist and empiricist tendencies in cognitive linguistics. In Cognitive Linguistics: Foundations, Scope, and Methodology, Theo Janssen and Gisela Redeker (eds.), 163-194. BerlinINew York: Mouton de Gruyter.

Section 1 Prototypicality and salience

Chapter 1 Prospects and problems of prototype theory

Originally published in Linguistics 1989,27: 587-612. Parallel to the research that I did on the application of prototype theory to questions of semantic change (which eventually resulted in my Diachronic Prototype Semantics monograph of 1997), I paid attention on a number of occasions to the theoretical clarification of the notion of prototypicality as such (and of the related notion of polysemy: see the second section of this collection). In the paper reprinted here, I try to bring some systematicity into the many uses of the concept 'prototypicality' by distinguishing between two cross-classifying dimensions. First, I make a distinction between two crucial structural phenomena underlying prototypicality effects: flexibility (i.e. the absence of clear boundaries and demarcations) and salience (i.e. differences of structural weight). Second, I suggest that both prototypicality phenomena may be found on an intensional level (the level of definitions) and on an extensional level (the level of referents). The crossclassification of the two dimensions defines four basic types of prototypicality effects. The chapter originally appeared as the introductory paper of a thematic issue of the journal Linguistics. A section of the original paper describing the various contributions to the thematic issue has been omitted from the present reprint.

1.

Prototype theory within linguistics

The starting-point of the prototypical conception of categorial structure is sununarized in the statement that when describing categories analytically, most traditions of thought have treated category membership as a digital, all-or-none phenomenon. That is, much work in philosophy, psychology, linguistics, and anthropology assumes that categories are logical bounded entities, membership in which is defined by an item's possession of a simple set of criterial features, in which all instances possessing the criterial attributes have a full and equal degree of membership. In contrast, it has recently been argued ... that some natural categories are analog and must be represented logically in a manner which reflects their analog structure (Roscll and Mervis 1975: 573574).

4

Prototypicality and salience

As we shall see in section 2, the exact definition of the concept of prototypicality as used in linguistics is not without problems. The major part of this introduction to the prototypicality-based studies collected here will, in fact, consist of an attempt at clarification of some of the problematic aspects of the way in which the notion of prototype has been used in linguistics. To begin with, however, we shall be concerned with a brief overview of the state of the art in linguistic prototype theory. I The theory originated in the mid 1970s with Eleanor Rosch' s research into the internal structure of categories. (Overviews may be found in Rosch 1978, 1988, and Mervis and Rosch 1981; the basic research is reported on mainly in Heider 1972, Rosch 1973, 1975, 1977, Rosch and Mervis 1975, Rosch, Simpson and Miller 1976, Rosch et al. 1976.) From its psycholinguistic origins, prototype theory has moved mainll in two directions. On the one hand, Rosch's findings and proposals were taken up by formal psycholexicology (and more generally, information-processing psychology), which tries to devise formal models for human conceptual memory and its operation, and which so, obviously, borders on Artificial Intelligence. Excellent overviews of the representational and experimental issues at stake here are Smith and Medin (1981), and Medin and Smith (1984); an interesting sample of current research may be found in Neisser (1987). On the other hand, prototype theory has had a steadily growing success in linguistics since the early 1980s, as witnessed by a number of recent monographs and collective volumes in which prototype theory and its cognitive extensions play a major role (Wierzbicka 1985, Lakoff 1987, Langacker 1987, Craig 1986, Holland and Quinn 1987, Rudzka-Ostyn 1988, Lehmann 1988a, Hullen and Schulze 1988, Tsohatzidis 1989, Taylor 1989). It is with the latter development that we shall be concerned with here. Against the background of the development of linguistic semantics, prototype theory may be defined primarily in contrast with the componential model of semantic analysis that was current in transformational grammar and that is stereotypically associated with Katz and Fodor's analysis of bachelor (Katz and Fodor 1963); in an early defense of a prototypical approach, Fillmore (1975) called this the 'checklist theory' of meaning. The prototypists' reaction against this featural approach had, however, the negative side-effect of creating the impression that prototypical theories rejected any kind of componential analysis. This is a misconception for the simple reason that there can be no semantic description without some sort of decompositional analysis. As a heuristic tool for the description and comparison of lexical meanings, a componential analysis retains its value (a value

Prospects and problems ofproto(vpe theory

5

that, incidentally, it did not acquire with the advent of componential analysis as an explicit semantic theory, but which had been obvious to lexicographers from time immemorial). Rather, the difficulties with the neostructuralist kind of feature analysis that grew out of structuralist field theory lie elsewhere; it is not the use of decomposition as a descriptive instrument that causes concern, but the status attributed to the featural analysis. Two important points have to be mentioned. In the first place, as suggested by the quotation at the beginning of this introduction, featural definitions are classically thought of as criterial, i.e. as listing attributes that are each indispensable for the definition of the concept in question, and that taken together suffice to delimit that concept from all others. In contrast, prototype theory claims that there need not be a single set of defining attributes that conform to the necessity-cum-sufficiency requirement. 3 In the second place, prototype theory is reluctant to accept the idea that there is an autonomous semantic structure in natural languages which can be studied in its own right, in isolation from the other cognitive capacities of man. In particular, meaning phenomena in natural languages cannot be studied in isolation from the encyclopedic knowledge individuals possess; it is precisely the presupposition that there exists a purely linguistic structure of semantic oppositions that enables structuralist and neostructuralist semantics to posit the existence of a distinction between semantic and encyclopedic knowledge. Prototype theory tends to minimize the distinction primarily for methodological reasons: because linguistic categorization is a cognitive phenomenon just like the other cognitive capacities of man, it is important to study it in its relationship to these other capacities. More specific arguments have also been formulated to show that the distinction between an encyclopedic and a semantic level of categorial structure is untenable. 4 For instance, given that the flexible extendibility of prototypical concepts is a synchronic characteristic of linguistic structure, and given the fact that these extensions may be based indiscriminately on allegedly encyclopedic or on allegedly semantic features, the distinction between both kinds of information loses its synchronic relevance. Take the case of metaphor: before lion acquires the meaning 'brave man', the feature 'brave' is not structurally distinctive within the semasiological structure of lion, and hence, it has to be considered encyclopedic according to structuralist theories. But if it can be accepted (and this is of course the crucial point) that the metaphorical extension of lion towards the concept 'brave man' is not just a question of diachronic change, but is merely an effect of the synchronic flexibility of lexical items,

6

Prototypicality and salience

the feature clearly acquires semantic status. If, furthermore, the argument can be repeated in the sense that such synchronic metaphorical extensions may be based on any allegedly encyclopedic attribute, the distinction be5 tween semantic and encyclopedic concepts as a whole falls. The matter need not, to be sure, be settled here. What is important for our introductory purposes is rather to see what exactly prototype theory objects to in componential theories of the Katzian type. First, the suggestion that lexical concepts are criterial in the classical sense, and second, the suggestion that there exists a purely linguistic level of conceptual structuring that is neatly separated from other, 'encyclopedic' forms of conceptual information, and that may thus be studied autonomously, in methodological isolation from other kinds of cognitive research. As against these points of view, prototype theory defends a non-criterial conception of categorial structure, and an interdisciplinary methodological perspective that takes into account relevant research from the other cognitive sciences. (The very transposition of the prototypical approach from experimental psychology to linguistics derives from this attitude.) But this historical positioning of prototype theory with regard to its immediate predecessors within the field of lexical semantics clearly does not explain why it has turned out to be such a successful alternative. Why did (and does) the prototypical approach appeal to a sizeable part of the linguistic community? On the one hand, the historical development of generative grammar had raised a considerable amount of interest in semantic matters. It should not be forgotten, in fact, that it was only after the incorporation of a semantic component into the transformational framework that Chomskyanism became internationally populac the universal appeal of the generative Standard Theory was at least partly due to the promises held by its Katzian semantic component. On the other hand, the promises were not fulfilled. Within the generative paradigm, Generative Semantics (which most strongly embodied the semantic approach) withered in favor of Autonomous Syntax, in which semantics hardly played a role worthy of note. Outside the generative approach, formal semantics of the Montagovian kind was too narrowly restricted to sentential meaning to be able to hold the attention of those who were interested primarily in the internal structure of natural language categories (and not primarily in the way these categories combine into larger unities).6 In short, as far as semantics was concerned, there was a gap in the linguistic market of the early 1980s that was not filled by the major approaches of the day. 7

Prospects and problems ofprototype theory

7

But again, recognizing that there was an interest in the semantics of natural language categories to which prototype theory could appeal does not tell the whole story. Why didn't people simply stick to the componential theory popularized by Katz, or to the rival axiomatic method of representation even if these gradually moved out of the centre of the linguistic attention as Autonomous Syntax and Formal Semantics took over? In general, there are a number of methodological requirements people nowadays expect of linguistic theories: descriptive adequacy (mainly in the form of a broad empirical scope), explanatory depth, productivity, and formalization. Although prototype theory rates much lower on the formalization scale than either the axiomatic or the featural approach, its assets with regard to the other three points are considerable. In the first place, it tackles a number of semantic phenomena that had been swept under the rug by the more structurally minded approaches. The fuzzy boundaries of lexical categories, the existence of typicality scales for the members of a category, the flexible and dynamic nature of word meanings, the importance of metaphor and metonymy as the basis of that flexibility - these are all intuitively obvious elements of the subject matter of semantics that were largely neglected by structural semantics. It is true that they were occasionally pointed at as an indispensable aspect of any fullfledged semantic theory: think, for instance, of Weinreich's remark (1966: 471) that a semantic theory should be able to deal with 'interpretable deviance', or Uhlenbeck's plea (1967) for a dynamic conception of word meaning. 8 These remarks did not, however, have much effect as far as theory formation was concerned. In particular, it is only with the advent of prototype theory that contemporary linguistics developed a valid model for the polysemy of lexical items. This is perhaps the single most appealing characteristic of prototype theory: here at last is a descriptive approach to lexical meaning in which our pretheoretical intuitions about gradedness, fuzziness, flexibility, clustering of senses etc. receive due attention. In the second place, prototype theory appears to be a productive theory not just in the sense that its insights into the structure of lexical categories can be easily applied in various fields of the lexicon, but also in the sense that it may be extended towards other aspects of linguistics. Whereas prototype theory started with being descriptively fruitful in lexical semantics, it soon became theoretically fruitful in the sense that other areas of linguistics were taken into consideration. A few recent examples of such extensions may suffice: phonology (Nathan 1986), morphology (Bybee and Moder 1983, Post 1986), syntax (Van Oosten 1986, Ross 1987), historicallinguis-

8

Prototypicality and salience

tics (Winters 1987, Aijmer 1985), markedness theory (Van Langendonck 1986), theoretical lexicography (Geeraerts 1985b). Through these and similar extensions,9 prototype theory has become one of the cornerstones of Cognitive Linguistics, which tries to account for the interaction between language and cognition on all levels of linguistic structure: one need only have a look at the prominent place attributed to a prototypical conception of categorial structure in Langacker (1987) (one of the basic works of the Cognitive Linguistic approach) to appreciate its importance. IQ In this sense, the development of prototype theory into Cognitive Linguistics contains exciting promises of a unified cognitive theory of linguistic categorization. In the third place, the explanatory depth of prototype theory resides partly in its generalizable character, but also in its interdisciplinary nature. The importance of its genetic link with psycholinguistics can only be fully appreciated against the background of the Chomskyan requirements with regard to theories of grammar. Chomsky' s methodology is, in fact, in the awkward position of declaring linguistics a cognitive science, but refusing to deal directly with the findings of the other sciences of the mind. Roughly stated, Chomskyan linguistics claims to reveal something about the mind, but imperviously prefers a strictly autonomist methodology over the open dialogue with psychology that would seem to be implied by such a claim. Prototype theory's linguistic application of psycholinguistic findings, on the other hand, takes the Chomskyan ideal of cognitive explanatory depth to its natural consequences, viz. of giving up the methodological autonomy of linguistics in favor of an interdisciplinary dialogue with the other cognitive sciences. l1 Prototype theory takes the cognitive claims of Chomskyanism methodologically seriously by its interdisciplinary openness. This is all the more important at a moment when Cognitive Science is emerging as an interdisciplinary cluster of psychology, neuroscience, Artificial Intelligence, and philosophy. It is probably one of the reasons for the appeal of prototype theory that its interdisciplinary connections hold the promise of linking linguistics to the most important development that the human sciences are currently witnessing.

2. Definitional problems, first series: 'Prototype' as a prototypical notion The appeal of prototype theory should not, however, obscure the fact that the exact definition of prototypicality is not without problems. The purpose

Prospects and problems ofprototype theory

9

of this section (and the following) is to analyze the sources of the confusion by making clear that prototypicality is itself, in the words of Posner (1986), a prototypical concept. As a first step, we shall have a look at four characteristics that are frequently mentioned (in various combinations) as typical of prototypicality. In each case, a quotation from early prototype studies is added to illustrate the point. (i) Prototypical categories cannot be defined by means of a single set of criterial (necessary and sufficient) attributes: We have argued that many words ... have as their meanings not a list of necessary and sufficient conditions that a thing or event must satisfy to count as a member of the category denoted by the word, but rather a psychological object or process which we have called a prototype (Coleman and Kay 1981: 43). (ii) Prototypical categories exhibit a family resemblance structure, or more generally, their semantic structure takes the form of a radial set of clustered and overlapping meanings: 12 The purpose of the present research was to explore one of the major structural principles which, we believe, may govern the fonnation of the prototype structure of semantic categories. This principle was first suggested in philosophy; Wittgenstein (1953) argued that the referents of a word need not have common elements to be understood and used in the nonnal functioning of language. He suggested that, rather, a family resemblance might be what linked the various referents of a word. A family resemblance relationship takes the form AB, BC, CD, DE. That is, each item has at least one, and probably several, elements in common with one or more items, but no, or few, elements are common to all items (Rosch and Mervis 1975: 574-575).

(iii) Prototypical categories exhibit degrees of category membership; not every member is equally representative for a category: By prototypes of categories we have generally meant the clearest cases of category membership defined operationally by people's judgments of goodness of membership in the category ... we can judge how clear a case something is and deal with categories on the basis of clear cases in the total absence of information about boundaries (Rosch 1978: 36). (iv) Prototypical categories are blurred at the edges: New trends in categorization research have brought into investigation and debate some of the major issues in conception and learning whose solution

10

Proto(vpicality and salience

had been unquestioned in earlier approaches. Empirical findings have established that ... category boundaries are not necessarily definite (Mervis and Rosch 1981: 109). As a first remark with regard to these characteristics, it should be noted that they are not the only ones that may be used in attempts to define the prototypical conception of categorization. Two classes of such additional features should be mentioned. On the one hand, there are characteristics that do not pertain (as the four mentioned above) to the structure of categories, but that rather pertain to the epistemological features of so-called non-Aristotelian categories. 13 For instance, the view that prototypical categories are not 'objectivist' but 'experiential' in nature (Lakoff 1987) envisages the epistemological relationship between concepts and the world rather than the structural characteristics of those concepts. In particular, it contrasts the allegedly classical view that 'categories of mind ... are simply reflections of categories that supposedly exist objectively in the world, independent of all beings', with the view that 'both categories of mind and human reason depend upon experiential aspects of human psychology' (Lakoff 1982: 99). Such an epistemological rather than structural characterization of natural concepts also has a methodological aspect to it; it entails that prototypical categories should not be studied in isolation from their experiential context. While such an epistemological or methodological conception of prototypical categorization is extremely valuable, we shall take a structural point of view in the following pages; we shall try to determine whether it is possible to give a coherent, structurally intrinsic characterization of prototypical categories. On the other hand, there are structural characteristics of prototypical concepts that can be reduced to the four basic structural features mentioned above. For instance, in my own work on prototypical categorization, I have repeatedly stressed the flexibility of prototypical concepts (1983a, 1985a), together with the fact that a distinction between semantic and encyclopedic components of lexical concepts cannot be maintained in the case of prototypical concepts (l985b). But the flexibility of prototypical categories is linked in a straightforward manner with the fourth characteristic: uncertainties with regard to the denotational boundaries of a category imply that it need not be used in a rigidly fixed manner. Similarly, the absence of a clear dividing line between encyclopedic and purely semantic information follows from this very flexibility together with the first and second characteristic. As illustrated in the previous section, the possibility of incorporating members into the category that do not correspond in every definitional respect with the

Prospects and problems ofprototype theory

11

eXlstmg members entails that features that are encyclopedic (nondefinitional) with regard to a given set of category members may turn into defmitional features with regard to a flexibly incorporated peripheral category member. The resemblance between central and peripheral cases may be based on allegedly encyclopedic just as well as on allegedly 'semantic' features. In short, features of prototypicality that are not included among the ones mentioned in (i)-(iv) may often be reduced to those four, and this in turn justifies a preliminary restriction of the discussion to the latter. A second remark with regard to the four characteristics is concerned with the fact that they are systematically related along two dimensions. On the one hand, the third and the fourth characteristic take into account the referential, extensional structure of a category. In particular, they have a look at the members of a category; they observe, respectively, that not all referents of a category are equal in representativeness for that category, and that the denotational boundaries of a category are not always determinate. On the other hand, these two aspects (centrality and non-rigidity) recur on the intensional level, where the definitional rather than the referential structure of a category is envisaged. For one thing, non-rigidity shows up in the fact that there is no single necessary and sufficient definition for a prototypical concept. For another, family resemblances imply overlapping of the subsets of a category. To take up the formulation used in the quotation under (ii) above, if there is no definition adequately describing A, B, C, D, and E, each of the subsets AB, BC, CD, and DE can be defined separately, but obviously, the 'meanings' that are so distinguished overlap. Consequently, meanings exhibiting a greater degree of overlapping (in the example: the senses corresponding with BC and CD) will have more structural weight than meanings that cover peripheral members of the category only. In short, the clustering of meanings that is typical of family resemblances implies that not every meaning is structurally equally important (and a similar observation can be made with regard to the components into which those meanings may be analyzed). The systematic links between the characteristics mentioned at the beginning are schematically summarized in Table 1. As a third remark, it should be noted that the four characteristics are often thought to be co-extensive, in spite of incidental but clear warnings such as Rosch and Mervis's remark that a family resemblance structure need not be the only source of prototypicality (1975: 599). Admittedly, it is easy to consider them to be equivalent; already in the quotations given above, partial reasons for their mutual interdependence can be found. More systematically,

12

Prototypicality and salience

the following links between the four characteristics might be responsible for the idea that prototypicality necessarily entails the joint presence of all four. Table 1. Characteristics of prototypicality NON-EQUALITY

NON-RIGIDITY

differences in structural weight

flexibility and vagueness

EXTENSIONAL

degrees of representativity

absence of clear boundaries

INTENSIONAL

clusters of overlapping senses

absence of classical definition

First, linking the first to the second characteristic is the argument mentioned above: if there is no single definition adequately describing the extension of an item as a whole, different subsets may be defined, but since the members of a category can usually be grouped together along different dimensions, these subsets are likely to overlap, i.e., to form clusters of related mearungs. Second, linking the second to the third characteristic is the idea that members of a category that are found in an area of overlapping between two senses carry more structural weight than instances that are covered by only one meaning. Representative members of a category (i.e., instances with a high degree of representativity) are to be found in maximally overlapping areas of the extension of a category. (In the example, A and E are less typical members that B, C, and D, which each belong to two different subsets.) Third, linking the third to the fourth characteristic is the idea that differences in degree of membership may diminish to a point where it becomes unclear whether something still belongs to the category or not. Categories have referentially blurred edges because of the dubious categorial status of items with extremely low membership degrees. And fourth, linking the fourth to the first characteristic is the idea that the flexibility that is inherent in the absence of clear boundaries prevents the formulation of an essence that is common to all the members of the category. Because peripheral members may not be identical with central cases but may only share some characteristics with them, it is difficult to define a set of attributes that is common to all members of a category and that is sufficient to distinguish that category from all others. These circular links between the four characteristics are, however, misleading. A closer look at some (familiar and less familiar) examples of prototypicality reveals that they need not co-occur.

Prospects and problems ofprototype theory

13

Bllill

The concept bird (one of Rosch's original examples of prototypicality) shows that natural categories may have clear-cut boundaries. At least with regard to our own, real world, the denotation of bird is determinate; educated speakers of English know very well where birds end and non-birds begin. They know, for instance, that a bat is not a bird but that a penguin is. Of course, the principled indeterminacy described by Waismann (1952) as 'open texture' remains: when confronted with an SF creature (a post-World War III mutant) that looks like a bird but talks like a man, we would not be sure whether it should be called a bird or not. A boundary problem that is typical for a prototypical organization of the lexicon would then arise. As it functions now, however, in present-day English, bird is denotationally clearly bounded, the archaeopteryx notwithstanding. 14 As has been remarked elsewhere (Lakoff 1987), the existence of prototypicality effects in clearly bounded concepts such as bird implies that a strict distinction has to be made between degree of membership and degree of representativity. Membership in the category bird is discrete; something is or is not a bird. But some birds may be birdier than others: the swallow does remain a more typical bird than the ostrich. RED

Color terms such as red constituted the starting-point for prototypical research; drawing on the views developed in Berlin and Kay (1969), Rosch's earliest work is an experimental demonstration of the fact that the borderline between different colors is fuzzy (there is no single line in the spectrum where red stops and orange begins), and of the fact that each color term is psychologically represented by focal colors (some hues are experienced as better reds than others) (Heider 1972, Heider and Olivier 1972). These prototypical characteristics on the extensionallevel are not matched on the definitionallevel. If red can be analytically defined at all (i.e., if it does not simply receive an ostensive definition consisting of an enumeration of hues with their degree of focality), its definition might be 'having a color that is more like that of blood than like that of an unclouded sky, that of grass, that of the sun, that of ... (etc., listing a typical exemplar for each of the other main colors)'. Such a definition (cp. Wierzbicka 1985: 342) does not correspond with either the first or the second characteristic mentioned above. ODD N1J};ffiER

Armstrong, Gleitman and Gleitman (1983) have shown experimentally that even a mathematical concept such as odd number exhibits psychological

14

Prototypicality and salience

representativity effects. This might seem remarkable, since odd number is a classical concept in all other respects: it receives a clear definition, does not exhibit a family resemblance structure or a radial set of clustered meanings, does not have blurred edges. However, Lakoff (1982) has made clear that degrees of representativity among odd numbers are not surprising if the experiential nature of concepts is taken into account. For instance, because the even or uneven character of a large number can be detennined easily by looking at the final number, it is no wonder that uneven numbers below 10 carry more psychological weight: they are procedurally of primary importance. VERS

As I have tried to show elsewhere (1988a), the first characteristic mentioned above is not sufficient to distinguish prototypical from classical categories, since, within the classical approach, the absence of a single definition characterized by necessity-cum-sufficiency might simply be an indication of polysemy. This means that it has to be shown on independent grounds that the allegedly prototypical concepts are not polysemous, or rather, it means that prototypical lexical concepts will be polysemous according to a definitional analysis in tenns of necessary and sufficient conditions (the classical definition of polysemy), but univocal according to certain other criteria. These criteria may be found, for instance, in native speakers' intuitions about the lexical items involved, intuitions that may be revealed by tests such as Quine's (1960) or Zwicky and Sadock's (1975). In this sense, the first characteristic has to be restated: prototypical categories will exhibit intuitive univocality coupled with analytical (definitional) polysemy, and not just the absence of a necessary-and-sufficient definition. Once this revision of the first characteristic is accepted, it can be demonstrated that the first and the second criterion need not co-occur. Lexical items that show clustered overlapping of senses may either confonn or not confonn to the revised first characteristic. An example of the first situation is the literal meaning of bird, an example of the second situation the Dutch adjective vers, which corresponds roughly with Englishfresh (except for the fact that the Dutch word does not carry the meaning 'cool'). Details of the comparison between both categories may be found in the paper mentioned above; by way of summary, Figures 1 and 2 represent the definitional analysis of both items. The distinction in intuitive status between vers and bird can be demonstrated by means of the Quinean test (roughly, a lexical item is ambiguous if it can be simultaneously predicated and negated of something

Prospects and problems ofprototype theory ............................................................................................................························6····· 7

15 ,

kiwi o_s_tr_i_ch_-t_:hiCkeJ

y

3---+-------------+-----+-----'

penguin 5---t--------------' L--------------4------'

1 being able to fly 4 having wings 7 having a beak or bill

2 having feathers 5 not domesticated

3 being S-shaped 6 being born from eggs

Figure J. A definitional analysis of bird

in a particular context). Thus, taking an example based on the corresponding ambiguity in the English counterpart of vers, it would be quite normal to state that the news meant in the sentence there was no fresh news from the jightini 5 is fresh in one sense ('recent, new') but not in another ('in optimal condition'): it makes sense to say that the news is at the same time fresh and not fresh. By contrast, it would be intuitively paradoxical to state that a penguin is at the same time a bird and not a bird (disregarding figurative extensions of the semantic range of bird). Nevertheless, the definitional analyses in Figures 1 and 2 make clear that both concepts exhibit prototypical clustering. In both cases, too, the structural position of the instances just discussed (news, penguin) is not in the central area with maximal overlapping. In short, then, the revised version of the first characteristic need not coincide with the second characteristic. The insight derived from a closer look at the four examples just described may be summarized as in Table 2. It is now easy to see to what extent 'prototypicality' is itself a prototypical notion. There is no single set of attributes that is common to all of the examples discussed here. Rather, they exhibit a family resemblance structure based on partial similarities. In this sense, the set of prototypical concepts characterized by clustering of senses overlaps

16

Prototypicality and salience

e.g. infonnation

1

2

e.g. fruit and other foodstuffs

'---- 1

e.g. air 1 new, novel, recent 2 in an optimal condition, pure, untainted

Figure 2. A definitional analysis of vers with the subset characterized by fuzzy boundaries (because of vers), and so on. At the same time, some concepts are more typically prototypical than others. (Bird and vers are more prototypical than red.) Notice, in particular, that the category fruit makes a good candidate for prototypical prototypicality, in the sense that it seems to combine all four characteristics. It shares the prototypical characteristics of bird, but in addition, things such as coconuts and, perhaps, tomatoes, seem to point out that the denotational boundary of fruit is less clear-cut than that of bird. However, although the examples considered above do not have a set of attributes in common, they do share a single feature, viz. degrees of membership representativity. It is highly dubious, though, whether this feature alone suffices to distinguish prototypical concepts from classical concepts. If the possibility of a single necessary-and-sufficient definition is one of the features par excellence with which the classical conception has been identified, it might justifiably be claimed that degrees of representativity are entirely compatible with the classical conception of categorization. It is, in fact, in that sense that Annstrong, Gleitman and Gleitman (1983) deal with a category such as odd number. The experiments used by Rosch to measure degrees of representativity are not, they claim, indicative of prototypicality since they occur with classical, rigidly definable concepts such as odd number. To say the least, representativity effects are only a peripheral prototypi-

Prospects and problems ofprototype theory

17

cal attribute according to Table 2 (cp. Lakoff 1986). But at the same time, the debate over the status of odd number shows that the concept 'prototypical concept' has no clear boundaries: it is not immediately clear whether a concept such as odd number should be included in the set of prototypical concepts or not. Table 2. The prototypicality of 'prototypicality' P::: ~

Cl P:::

iQ

absence of classical definition clusters of overlapping senses degrees of representativity absence of clear boundaries

+ + +

o:l

P:::

~

+ + +

~

+ +

Cl Cl

0

::2 ~

z

+

Of course, contrary to the situation in everyday speech, such a boundary conflict should not be maintained in scientific speech. A discipline such as linguistics should try to define its concepts as clearly as possible, and the purpose of this section is precisely to show that what has intuitively been classified together as instances of prototypical categories consists of distinct phenomena that have to be kept theoretically apart. In line with prototype theory itself, however, such an attempt at clear definition should not imply an attempt to define the 'true nature' or the 'very essence' of prototypicality. Determining an 'only true kind' of prototypicality is infinitely less important than seeing what the phenomena are and how they are related to each other by contrast or similarity. Still, there might seem to be one way in which decent sense could be made of the question what the true meaning of prototypicality would be. To begin with, let us note that the prototypical character attributed to the concept of prototypicality also shows up in the fact that the notion 'prototype' is an extremely flexible one. This can be illustrated in two ways. First, the lexical item prototypical is spontaneously used to name a number of phenomena that are linked by metonymy, next to the phenomena linked by similarity that are brought together in Table 2. The lexical item does not only characterize structural features of concepts, and the concepts exhibiting those features themselves, but sometimes even particular (viz., highly representative) instances of the categories in question (the robin as a prototypical

18

Prototypicality and salience

bird). Second, context may stress one feature of prototypical organization rather than another (cp. the priming effects in Rosch 1975). The general purpose of one's investigations may lead one to devote more attention to one aspect of the prototypical cluster than to another. To name a few examples: degrees of representativity are important for language development studies (if it is taken into account that most concepts in early language development are acquired via their exemplars), while clustered overlapping of senses will come to the fore in linguistic or lexicographical studies into the structure of polysemy. And a cognitive interest into the epistemological principles underlying natural language will attach more weight to the decoupling of intuitive univocality and analytical, definitional polysemy.16 In this respect, the question with regard to the true nature of prototypicality might be transformed into the question what might be the most interesting (or perhaps even the most important) perspective for studying and defining prototypicality. But here again, the 'ultimate essence fallacy' exposed by prototype theory itself lurks round the corner: there will be different preferences for one perspective rather than another, but there will be no single ultimately and eternally most important conception of prototypicality. In short, the foregoing analysis corroborates Wierzbicka's remark that there are 'many senses' to the notion prototype, and that 'the notion prototype has been used in recent literature as a catch-all notion' (1985: 343). However, a more systematic analysis than Wierzbicka's reveals that this very multiplicity of usage also supports Cognitive Semantics, in the sense that it shows that the same categorization principles may guide common sense and scientific thinking. This is, then, a further indication of the metatheoretical relevance of a cognitive conception of linguistic categorization, which I have explored at length elsewhere (1985b). At the same time, it has become clear that one of the major tasks for the further development of prototype theory is the closer investigation of the prototypically clustered characteristics of prototypicality. A major reference in this respect is Lakoff's attempt (1987: Chapter 4-8) to determine which different kinds of conceptual models may lie at the basis of prototypicality effects.

3. Definitional problems, second Series: 'Prototype theory' as a prototypical notion Whereas the previous section made clear that prototypicality as used in linguistic semantics is a prototypically structured concept, it should now be

Prospects and problems ofprototype theory

19

noted that the prototype-theoretical movement as well is a prototypically structured approach to semantics. There are, in other words, central as well as more peripheral examples of prototypical theories. In particular, there exist a number of theories that combine aspects of the classical approach to semantic structure with aspects of the prototypical conception. In this section, two approaches will be considered that are to some extent semiclassical as well as semi-prototypical; each of both embodies a strategy for reinstating particular aspects of the classical view against the background of an overall cognitive point of view. To begin with, some of the clarity and neatness of the classical approach may be recovered by concentrating on the prototypical centre of a category. If the non-classical indeterminacy of lexical concepts stems primarily from the flexible extendibility of concepts, discreteness may be reinstalled by avoiding the problems of clustered polysemy, i.e., by restricting the definitional analysis to the prototypical centre of the category. This approach is vigorously carried through by Wierzbicka (1985), who explicitly defends the discreteness of semantics by introspectively considering only the clear, salient centre of lexical categories. In a discussion of Labov's experimental investigation into the non-classical characteristics of everyday concepts (1973), she notes: 17 To state the meaning of a word, it is not sufficient to study its applicability to things; what one must do above all is to study the structure of the concept which underlies and explains that applicability. In the case of words describing natural kinds or kinds of human artefacts, to understand the structure of the concept means to describe fully and accurately the idea (not just the visual image) of a typical representative of the kind: the prototype. And to describe it fully and accurately we have to discover the internal logic of the concept. This is best done not through interviews, not through laboratory experiments, and not through reports of casual, superficial impressions or intuitions ... but through methodical introspection and thinking (1985: 19). It should be noted immediately that Wierzbicka's reinstatement of discreteness does not imply that her definitions do in fact always consist of necessary-and-sufficient conditions, and she acknowledges as much (1985: 60). In this respect, Wierzbicka's approach is only partly a departure from the hard core of prototype-theoretical studies: the absence of necessary-andsufficient conditions for the definition of certain core concepts is accepted, but the avoidance of the clustered polysemy problem 'tidies up' the semantic description and reinstates some of the classical neatness. Neither does

20

Prototypicality and salience

Wierzbicka's approach imply that lexical items are always univocal; in her dictionary of English speech act verbs (1987a), several items receive multiple definitions. Each of the definitions does, however, constitute a highly salient meaning, and again, by disregarding peripheral kinds of usage, the clustered or radial structure of the polysemy of lexical items does not enter the picture. The question to be asked, then, is whether Wierzbicka's restriction of the description to the salient meanings of a category is useful and adequate from a cognitive point of view. From a methodological point of view, the periphery of natural, nonuniquely definable categories is as interesting as their salient centre(s), because it is precisely the relationship between both that typically characterizes natural categories. Cognitive Linguistics is not only interested in what constitutes the centre of a category, but also in how this centre can be extended towards peripheral cases, and how far this extension can go. The mechanisms for incorporating marginal cases into a category at the same time restrict the flexibility of that concept; it is only by studying peripheral cases, for instance, that an answer may be found with regard to the question how dissimilar things can be before they are no longer recognized as basically the same. If, in other words, flexible polysemization is indeed one of the major characteristics of natural language categories, a deliberate restriction of the description to the salient meanings of a category is methodologically less propitious, as it may lead to a neglect of this basic feature. 18 A second strategy for salvaging aspects of the classical approach is to invoke sociolinguistic mechanisms such as Putnam's 'division of linguistic labor' (1975). According to Putnam, ordinary language users possess no more than 'stereotypical' knowledge about natural kinds, that is to say, they are aware of a number of salient characteristics, such as the fact that water is a transparent, thirst-quenching, tasteless liquid. The technical definition of water as H 20, on the other hand, is to be located primarily with scientific experts. It is the experts' knowledge that ultimately determines how natural kind terms are to be used. On the one hand, a 'division of linguistic labor' ensures that there are societal experts who know that water is H 2 0, that there is a distinction between elms and beech, how to recognize gold from pyrites, and so on. On the other hand, laymen attune their own linguistic usage to that of the expert scientists, technicians, etc.. The members of the non-specialized group are not required to have expert knowledge, but if they wish to be considered full-fledged members of the linguistic community, they are supposed to know the 'stereotype' connected with a category. A stereotype is, thus, a socially determined minimum set of data with regard to the

Prospects and problems ofproto(vpe theory

21

extension of a category. Given the similarity between Putnam's stereotypes and the prototypes of Cognitive Linguistics (both consist roughly of the most salient information connected with a category), the division of linguistic labor might be used to rescue the classical view of concepts. 19 Expert definitions being classical (they specify an essentialist 'hidden structure' for natural kinds), the stereotypical concepts of everyday language users might now be seen as hardly more than a sloppy derivative of those classically defined expert categories. 'True' (expert) definitions would be classical, and stereotypical/prototypical concepts might be dismissed as sociolinguistically secondary phenomena. It should be remarked immediately that such a reinstatement of the classical view is not as obvious for other words than the natural kind terms for which Putnam' s theory is in fact intended (what is the expert definition of the preposition for?). Moreover, as a sociolinguistic theory about the social factors that determine how lexical items may be used, the 'division of linguistic labor' theory is incomplete to say the least. The primacy of expert definitions would seem to imply that natural language follows the developments and discoveries of science in a strict fashion. In actual fact, however, natural language categorization is not only determined by the state of affairs in the sciences, but also by the communicative and cognitive requirements of the linguistic community in its own right. One of Putnam's own examples may serve as an illustration. Although science has discovered that jade refers to two kinds of materials (one with the 'hidden structure' of a silicate of calcium and magnesium, the other being a silicate of sodium and aluminium), ordinary usage continues to refer to both substances indiscriminately as jade. That is to say, categorization in everyday language is not entirely dependent upon scientific research, but seems to be determined at least in part by independent criteria: if the classificatory exigencies of everyday communicative interaction do not call for a distinction between the two kinds of jade, the scientific splitting of the category is largely ignored. This implies that an investigation into everyday language categorization as an independent cognitive system is justified. More generally, if Putnam's view is seen as a theory about the sociolinguistic structure of semantic norms, his hierarchical model (with experts at one end and laymen at the other) is only one among a number of alternatives, some of which (such as the one described by Bartsch 1985) link up closely with a prototypical conception of categorial structure. At the same time, however, it should be admitted that the relationship between classical scientific categorization and prototypical commonsense categorization may be explored in more depth than is yet the case. 20

22

Prototypicality and salience

To summarize: the confusion associated with the notion of prototypicality is further increased by the fact that more straightforwardly prototypical approaches are surrounded by hybrid theories that contain particular strategies for combining classical discreteness with typically prototypical phenomena. We have discussed two such approaches (one in which the strategy in question is methodological, and another one in which it is sociolinguistic), but 21 this does not mean that these are the only ones that might be mentioned. The two approaches mentioned here are, however, particularly revealing, as they link up with two important currents in the history of Western thought. The first one simplifyingly boils down to the view that the mind is neat (if you look hard enough into it), but that the world is fuzzy: the nondiscreteness that Cognitive Linguistics is concerned with arises from the fact that we have to apply clear-cut mental categories to an external reality that is so to say less well organized. The conception that the world of mental entities is somehow better organized than the outside world is obviously an idealistic one (though it does not constitute the only possible kind of idealism); Wierzbicka herself stresses the Platonist character of her approach. On the other hand, Putnam' s view that science is neat whereas everyday language is fuzzy, links up with the empiricist objectivism of the Ideal Language branch of analytical philosophy: the objective structure of reality is best described by the language of science, and everyday language is at best a weak derivative of scientific categorization, at worst a conceptual muddle teeming with philosophical pseudo-problems. As the previous discussion suggests that hard-core Cognitive Linguistics steers clear of both the idealist and the objectivist option, we have here one more indication 22 for the necessity of a further investigation into the epistemological, philosophical background of the prototypical conception of categorial structure.

Notes 1.

2.

The discussion in section 2 will make clear that the term proto~ype theory should be used with care, since the theoretical uniformity that it suggests tends to obliterate the actual distinctions between the diverse forms of prototypicality discussed in the literature. The term is used here as a convenient reference mark only, to indicate a number of related theoretical conceptions of categorial structure that share an insistence on any or more of the various kinds of prototypicality effects discussed in section 2. Though not exclusively: see Rosch (1988: 386).

Prospects and problems o/prototype theory

3.

4. 5.

6.

7.

8.

23

Notice that this claim applies just as well to the axiomatic, postulate-based form of description that developed as the major representational alternative for Katzian componential analysis. The notion of criteriality is just as much part and parcel of the classical versions of the axiomatic alternative as it is of Katzian feature analysis. See, among others, Haiman (1980a) and Geeraerts (1985b). The distinction between semantic and encyclopedic concepts against which Cognitive Semantics reacts is often misconstrued. In particular, in the statement that there is no principled distinction between semantic and encyclopedic information, the words semantic and encyclopedic are not used (as implied by Lehmann 1988b) in the senses 'as may be found in dictionaries' and 'as may be found in encyclopedias', respectively. Rather, the rejected distinction refers to an alleged distinction within an individual language user's conceptual memory; it involves the presupposition that there is an independent level of semantic information that belongs to the language and that is distinct from the individual's world knowledge. The kind of information that is typically found in encyclopedias involves scientific information ofthe kind 'ovulation triggered by copulation' for the item cat (the example is Lehmann's); but while the distinction between scientific and laymen's knowledge is primarily a social one, this kind of 'encyclopedic' information is only relevant for the psychological perspective of Cognitive Semantics if the individual lexicon to be described is that of someone with a certain amount of scientific knowledge of cats (or if, through sociolinguistic idealization, the average language user's lexicon may be supposed to contain that piece of scientific information). There are, of course, exceptions such as Dowty (1979) to confirm the rule. The historical sketch of the advent of prototype theory given here is treated more thoroughly in Geeraerts (1988a). As the semantic interests of the former audience of Generative Semantics were so to say no longer envisaged by the leading theories of the day, it does not come as a total surprise, from this point of view, to find George Lakoff, one of the leading Generative Semanticists, again as one of the leading cognitivists. These antecedents are not the only ones that might be mentioned. I have elsewhere (1988c) drawn the attention to the similarities between the prestructuralist, historical tradition of semantic research and present-day Cognitive Semantics, but there are other (admittedly non-mainstream) traditions of semantic research with which Cognitive Semantics is methodologically related: think, e.g., of the anthropological research of Malinowski, Firth, and the London School in general. Even a structuralist such as Reichling has held views about the structure of polysemy that come close to the point of view of prototype theory: his influential work on the word as the fundamental unit of

24

9.

10.

11.

12.

13.

Prototypicality and salience

linguistics (1935) contains an analysis of the Dutch word spel that is awkwardly similar to Wittgenstein's remarks about the German equivalent Spiel. The point to be stressed is this: as a theory about the (radial, clustered, dynamically flexible) structure of polysemy, prototype theory is to a considerable extent a rediscovery of views that were paramount during the prestructuralist era of the development of lexical semantics, and that lingered on below the surface in the structuralist and transfonnationalist periods. Because of their large scope, the functionalist approach of Seiler (1986) and the naturalist approach of Dressier (1985) are particularly interesting for the use of prototypicality with regard to various aspects of the fonnal organization of language. A bibliography of work in Cognitive Linguistics is to be found in Dirven (1988). It is worth mentioning that Cognitive Linguistics is currently in a stage of organization: a first international conference of Cognitive linguistics was held in Duisburg in March 1989, and a new journal entitled Cognitive Linguistics, published by Mouton, is scheduled to start appearing in the beginning of 1990. Next to the link with psycholinguistics, there is a connection with Artificial Intelligence research, through the correspondences between the notion of prototypicality and that of frame; see Fillmore (1977a). It needs to be stressed, though, that the link is relatively weak; specifically, the correspondence just mentioned is to a certain extent counterbalanced by Lakoff's criticism (1987) of the objectivist assumptions of mainstream Artificial Intelligence research (but then again, one of Lakoff's current research projects involves a connectionist approach to the fonnal modeling of Cognitive Semantic notions such as metaphorical image schemata). In general, sorting out the relationship between Cognitive Semantics and Artificial Intelligenceoriented Cognitive Science will be one of the major tasks for the further development of Cognitive Semantics. See Lakoff (1987: Chapter 6) for the notion of a radial set and compare Givon (1986) for a comparison between the views of Wittgenstein and those of prototype theory. The stress Givon places on the distinctions between both is slightly exaggerated, as it tends to obscure their mutual rejection of the socalled classical theory. See also the next footnote. The 'so-called' is added to stress, first, that the views of Aristotle also contain features that correspond rather with a cognitive than with a 'classical' approach, and second (more generally), that the philosophical position of prototype theory is in need of further elucidation. The present situation is rather muddled: while the classical Roschian position is to characterize prototype theory as non-Aristotelian and Wittgensteinian, Givon (1986) has argued that prototype theory is non-Wittgensteinian (see the previous note), but whereas Givon also describes prototype theory as non-Platonic, Wierzbicka

Prospects and problems ofprototype theory

14. 15. 16.

17. 18.

19.

25

(to whom we shall come back in section 3) precisely presents an explicitly Platonic version of prototype theory. More generally, the philosophical position of prototype theory has so far been discussed mainly against the background of classical philosophy (Aristotle and Plato), and against the background of contemporary analytical philosophy (see Lakoff 1987). This means that a large part of the history of Western philosophy passes unmentioned; this is to be regretted, as the post-Cartesian period in the history of philosophy is concerned with epistemological questions that are of immediate interest to Cognitive Semantics. In particular, if it can be accepted that one of the major epistemological aspects of a prototypical conception of categorial structure resides in the fact that categories are interpretive schemata that are used flexibly and dynamically in our encounters with reality, a major philosophical reference point for prototype theory will lie with those philosophical theories that recognize the constitutive role of existing knowledge with regard to new experiences. As I have argued elsewhere (1985b), the Husserlian phenomenological movement (as represented, specifically, by Maurice Merleau-Ponty) provides a good starting-point for a further confrontation with philosophy. The archaeopteryx is probably regarded as a species separate from either bird or reptile. The example is taken from the Longman Dictionary of Contemporary English. Considered from this point of view, Lakofrs radial sets as such are not particularly unclassical: structured polysemy as such is entirely compatible with the classical view. Kleiber (1988) offers an insightful discussion of the theoretical consequences of the growing importance of the structure of polysemy in prototype-theoretical research. For a more extended discussion of Wierzbicka's views, see Geeraerts (1988c). Notice that the restriction to the prototypical centre of categories correlates with Wierzbicka's Platonic, introspective methodology: it seems probable that the applications of a category that can be accessed introspectively are only the more salient ones; peripheral cases probably do not always pass the threshold of conscious attention. What is interesting from a cognitive point of view, however, is the way people spontaneously categorize and classify things, not the way in which they introspectively reflect upon their own conceptualizations. Any attempt to describe the peripheral instances of a category together with its prototypical centre can therefore not be restricted to an introspective methodology. This is not say that Putnam actually intended his stereotypical theory as such an attempted rescue: his problems lay with the notion of reference rather than with those of polysemy and categorial structure. My remarks about Putnam

26

Prototypicality and salience

are an investigation into some of the possible consequences of the notion of division of linguistic labor, not an attempt to give an account of Putnam's view in its original setting. Further, it has to be mentioned that some of Putnam's later philosophical views open up entirely different perspectives for a confrontation with Cognitive Semantics; in particular, see Lakoff (1987) on Putnam and anti-objectivism. 20. An interesting contribution to such an exploration is found in Lakoff (1987: Chapter 12), where it is claimed that scientific categories are far from being as classical as is usually assumed. 21. Again, see Lakoff (1987: Chapter 9) for some more examples; they are situated within formal psycholexicology rather than within linguistics. 22. Next, that is, to the remarks made in footnote 13.

Chapter 2 Where does prototypicality come from?

Originally published in Brygida Rudzka-Ostyn (ed.), 1988, Topics in Cognitive Linguistics 207-229. Amsterdam/Philadelphia: John Benjamins. The prototype-theoretical case study reprinted in the present chapter is not just included because it provides a concrete illustration of the theoretical model discussed in Chapter 1. The paper makes two points of a more general nature. First, on the theoretical level, the paper argues for a functional explanation of prototypicality effects, i.e. an explanation in tenns of general requirements on the organization of human cognition: It is cognitively advantageous to maximize the conceptual richness of each category through the incorporation of closely related nuances into a single concept because this makes the conceptual system more economic. Second, on the methodological level, the paper introduces an onomasiological perspective in lexical semantics: it is only by comparing the near-synonyms vernielen and vernietigen that we get a correct insight into the factors that might explain the presence of prototypicality effects. Over the years, I have become more and more convinced that such an onomasiological perspective is essential in Cognitive Linguistics: categorization primarily involve onomasiological choices, i.e. the question 'Why does someone use the category x rather than the category y for talking about phenomenon zT is cognitively more realistic than the question 'Does z belong in the semantic range of application of xT Starting with the monograph The Structure of Lexical Variation (Geeraerts, Grondelaers, and Bakema 1994), the lines that I have tried to set out with our Leuven research group have increasingly been geared towards the analysis of onomasiological variation (see Geeraerts 2002b, Grondelaers and Geeraerts 2003 for a general statement and an overview, and compare Chapter 11).

1.

Hypotheses about the sources of prototypicality

Prototype theory is as it were part of the prototypical core of the cognitive paradigm in semantics, particularly in lexical semantics. I think it is safe to say that it is by now quite obvious that gradience and salience are among the linguistically relevant aspects of semantic structure. One need only recall the early experimental work by Rosch (1973) and Labov (1973) to appreciate

28

Prototypicality and salience

the importance of gradualness and vagueness for the adequate description of word meaning. But what about explanatory adequacy? Can we move beyond the descriptive level and explain why prototypicality exists at all? There are at least four different hypotheses that have been proposed to explain prototypical phenomena. Each of these hypotheses has been formulated (or at least hinted at) by Eleanor Rosch herself; this is an indication that the hypotheses might well be complementary rather than mutually contradictory. I will call these four hypotheses the physiological, the referential, the statistical, and the psychological one. Let us have a look at them. The physiological hypothesis says that prototypicality is the result of the physiological structure of the perceptual apparatus (Rosch 1973). This hypothesis has been formulated with regard to the prototypicality effects in the domain of color terms (the first major field in which prototypicality phenomena have been observed). Particular colors are thought to be focal because the human eye is more sensitive to certain light frequencies than to others. The scope of the physiological explanation is probably fairly limited; it may only be applicable to concepts immediately referring to perceptual phenomena, or at least to bodily experiences that have a distinct physiological basis. Since this is most likely not the majority of cases, additional hypotheses will have to be invoked to explain the prototypical structure of concepts that have no immediate physiological basis. The referential hypothesis states that prototypicality results from the fact that some instances of a category share more attributes with other instances of the category than certain peripheral members of the category (or share attributes with more other instances than these peripheral cases). The peripheral applications of a category share attributes with relatively few other cases, or share only a relatively small number of attributes with other, more central members of the category. This is the family resemblance model of prototypicality (Rosch and Mervis 1975); in psychological terms, it states that the prototypical instances of a category maximize cue validity. I have dubbed this view 'referential' because it considers prototypicality to be an automatic consequence of the structure of the range of application of a concept. Once you know what objects, events etc. a concept can refer to, you can compute differences in salience by comparing the number of shared attributes among those things. One might even say that prototypicality is a secondary phenomenon: it is a side-effect of the mutual attribute relations among the instances in the referential range of application of the concept. Statistical explanations of prototypicality state that the most frequently experienced member of a category is the prototype. At least, this is the sim-

Where does prototypicality come from?

29

pIe fonn of the frequency model. It can also be combined with the family resemblance model; the weight of an attribute within a concept is then not only detennined by its role within the family of applications constituting the category, but also by the relative frequency with which it is experienced (Rosch 1975). The psychological hypothesis is a functional one. It states that it is cognitively advantageous to maximize the conceptual richness of each category through the incorporation of closely related nuances into a single concept because this makes the conceptual system more economic. Because of the maximal conceptual density of each category, the most infonnation can be provided with the least cognitive effort (Rosch 1977). In what follows, I would like to show that the functional explanation of prototypicality is more general than the other ones because it can explain cases of prototypicality that are counter-examples to the other modes. I will elaborate the psychological hypothesis by indicating some more functional sources of prototypicality; I will try to make clear that prototypicality is the outcome of some deep-seated principles of cognitive functioning.

2. A case study in synonymics Dutch has a pair of synonyms vernielen and vernietigen, which both roughly mean 'to destroy'. Though they exhibit some degree of phonetic similarity, their origin is quite diverse. Vernielen is the older fonn. It is already to be found in Middle Dutch, and it is fonned by means of the common verbfonning prefix ver- and the adjective niel, only a few examples of which survive, but which probably meant 'down to the ground'. Etymologically, then, vernielen means 'to throw down to the ground, to tear down'. Vernietigen, on the other hand, makes its first appearance in the sixteenth century; it is fonned by means of the same prefix ver- and the adjective nietig, which is itself a derivation from the negation particle niet (English not) and the suffix -ig (which corresponds with English -y). Vernietigen gradually replaces a third fonn verniefen, which is a straightforward derivation from niet with ver-, and which is extinct by the end of the seventeenth century. Vernietigen literally means 'to annihilate, to bring to naught'. The best way to study both words is to turn to the Woordenboek der Nederlandsche Taal (hence WNT), the major dictionary of Dutch that covers the period from 1500 up to 1920 and that, by the way, is still uncompleted after about a century of editorial work. This dictionary is being com-

30

Prototypicality and salience

piled on the basis of a huge corpus of quotations; there are as yet no equally representative corpora for contemporary Dutch, so that it is rather more difficult to get an adequate picture of twentieth-century usage than it is to study the semantic history of the vocabulary of Dutch. For the purpose of this paper, this is not very important: it suffices to pick out one synchronic period and to see how both words relate to each other in that period. For a number of practical reasons (among others the amount of available material), I will concentrate on the nineteenth century, stretching the temporal borders of that period with approximately one decade at each end. In this way, a period from 120 to 130 years can be considered, ranging from roughly 1790 to 1910. In the light of the history of culture, this seems quite justified; we more or less envisage the cultural period from the French revolution up to the First World War: the nineteenth century in the broadest sense. To get a good picture of the development ofvernietigen and vernielen, it would be necessary to present and discuss the entire articles that I have compiled for the WNT, and the complete set of quotations on which they arc based. For obvious lack of space, I will only give illustrative quotations; bibliographical references are as in the WNT. The set of quotations shows that vernielen and vernietigen can be used indiscriminately with the same range of application. Each numbered pair of quotations gives examples of one particular kind of usage. These examples should be studied from two points of view. On the one hand, the question has to be asked whether vernielen and vernietigen exhibit any syntagmatic differences, i.e. differences in their collocational properties. On the other hand, the question arises whether they are paradigmatically different, i.e. whether they exhibit purely conceptual differences. (More details on the analysis of both verbs can be found in Geeraerts 1985e.) (1)

Dat huis was ... evennmin als de naburige tegen de verwoestende veeten dier tijd bestand. Reeds onder den zoon en opvolger des stichters werd het ... tot den grand toe vernield (Veegens, Hist. Stud. 2, 282, 1869). [Like the neighboring one, this house was not able to stand up against the destructive quarrels of the age. Already under the son of the founder, it was demolished down to the ground.] Alleen zijn de vroegere kruisvensters door vensterramen van nieuweren trant vervangen en hebben de vrijheidsmannen van 1795 ... het wapen des stichters in den voorgevel met ruwe hand vernietigd (Veegens, Hist. Stud. 1, 125, 1864). [Only, the earlier cross-windows have been replaced by windows in a

Where does prototypicality come from? newer style, and in 1795, the freedom fighters demolished the founder's arms in the facade with their rough hands.] (2)

Er gaat dan stroom op den daarvoor gevormden zijweg over, waarbij genoeg warmte ontwikkeld wordt om de draadwindingen in zeer korten tijd te vernielen (Van Cappelle, Electr. 214, 1908). [Electric current is then transferred to the diversion that has been construed to that end, in which case enough warmth is produced to destroy the coils of the wire in a very short time.] Zonder deze voorzorg zou het draadje door de enorme hitte van den gloeidraad vernietigd worden (Van Cappelle, Electr. 295, 1908). [Without this precaution, the wire would be destroyed by the enormous heat of the filament.]

(3)

Hoeveel het wild vernielt wordt door een Engelschman zeer goed uiteengezet bij gelegenheid van een aanval op de beschenning die het wild aldaar geniet (Volksvlijt 1872, 175). [How much is destroyed by game is aptly expressed by a certain Englishman on the occasion of an attack on the protection these animals enjoy in his country.] Bij het vernietigen van de onkruiden door het bewerken dient op hunne voortplanting en ontwikkeling te worden gelet (Reinders, Land. I, 309, 1892). [During the destruction of weeds by cultivating the land, one should bear in mind their reproduction and development.]

(4)

Wel wat hamer! Wordt door zulke sentimentele zotternyen niet al de inwendige kracht vernield? (Wolff en Deken, Blank. 3, 220, 1789). [By golly! Does not such sentiment al foolishness destroy all our inner strength?] Stel mij niet zoo hoog, zei ze onthutst, ik zou daaraan niet beantwoorden; ik zou uw ideaal vernietigen (Vosmaer, Amaz. 175, 1880). [Do not put me on a pedestal, she said disconcertedly, I would not live up to that; I would destroy your ideal.]

(5)

De bergstroom in zijn grammen loop Verscheurt zijn zoom, verdrinkt de dalen: Alzoo vernielt Gij 's menschen hoop! (Ten Kate, Job 53, 1865). [The mountain stream in its angry course rends its banks, drowns the valleys: thus, Thou destroys the hope of Man!] Zy is dan, van kindsbeen af, opgevoed om mynheer Daniel's echtgenote te worden, en nu is die hoop van een geheelleven vernietigd! (Conscience, Kwael d. T. 2, 65, 1859) [She has been raised from childhood to become Master Daniel's wife,

31

32

Prototypicality and salience and now this hope of a lifetime has been annihilated!] Dit toeval vernietigde ons geheele plan (Haafner, Ceilon 103, 1810). [This coincidence annihilated our entire plan.]

(6)

Mij gedenkt ook nog dat Nicolaas Gaal ... mij placht te verhalen ... dat de oude man om deze ontstolen eer zich zoo ontstelde en vergramde, dat het ook scheen ofhij dezen diefwel had willen vernielen (Fruin, Geschr. 1,1974, 1888). [I still remember that Nicolas Gaal was fond oftelling me that the old man used to get so angry and upset about this stolen honor, that it seemed that he would have liked to kill that thief.] Intusschen heeft de Godin de Natuur besloten nu voor altijd de Drijvende Eilanden en al hun inwoners te vernietigen (Quack, Soc. 1,246, 1875). [Meanwhile, the goddess Nature has decided to destroy the inhabitants of the Floating Islands once and for all.]

(7)

De beroerte, die haar zwakke levenskrachten in een half uur tijds vernielde, had reeds in het eerste oogenblik hare spraak verlamd (Beets, C.O. 206, 1840). [The stroke that destroyed her weak life force in half an hour. had from the first moment paralyzed her speech.] Hy moet rusten. Zulke driften vernietigen het sterkste gestel (Wolff en Deken, Leev. 1,290, 1784). [He has to take a rest. Such passions undennine the strongest constitution.]

(8)

De vrouwen, Lus, zijn zonen, al de anderen bleven stom, vernield van ontsteltenis, op hun stoelen genageld (Buysse, NeefPerseyn 4545, 1893). [The women, Lus, his sons, all the others remained silent, destroyed by disconcertedness, nailed to their chairs.] Toen antwoordde zij langzaam met een doffe stem, als vernietigd door haar eigene woorden: 'la, indien het nog mogelijk is' (Buysse, Mea Culpa 68, 1896). [Then she answered slowly, in a dull voice, as if struck down by her own words: 'Yes, if it is still possible'.]

(9)

De uitslag van den stryd was ditmael hem niet gunstig: geheel zyn leger werd vernield of uiteen geslagen (Conscience, Gesch. v. Belgie 110, 1845). [This time, the result of the battle was not favorable to him: his entire army was destroyed or dispersed.]

Where does prototypicality comefrom?

33

Het gansch leger der Turken was vernietigd! (Conscience, Gesch. v. Belgie 352, 1845). [The entire army of the Turks was destroyed! 1 Syntagmatically, we not only see that both words can be used by the same author in the same context without noticeable differences, as in (1), (2), (8) and (9), but also that the range of application of each word can be divided into three identical major groups, which can moreover be subdivided along parallel lines. There is a set of applications in which the words are used with regard to concrete, material objects (1, 2, 3); a set in which they are used with regard to abstract objects (4, 5), and a set in which they are used with regard to persons (6, 7, 8, 9). Within the first set, frequently occurring applications relate to buildings (1), other human artifacts (2), and natural objects, in particular plants and crops (3). With regard to the abstract applications, we can distinguish between the annihilation of the existence of certain abstract objects as such (4), and applications in which the realization or fulfillment of certain abstract notions that contain an aspect of expectation or intention with regard to the future is prevented (5). With regard to persons, (6) expresses their death as such; (7) and (8) indicate how someone's bodily or mental health, respectively, are undermined. (9) expresses how armies are beaten; this application is half-way between the abstract group (the armies cease to exist as functional entities), and the personal group (individual soldiers are killed). The existence of analogous subdivisions within each of the major groups shows that the syntagmatic equivalence of vernielen and vernietigen is not a coincidence, but that it is an essential part of their relationship. Furthermore, the examples also show that there is a paradigmatic, strictly conceptual equivalence between both: they do not only have the same collocational properties, but they also seem to express the same concepts in the same contexts. (The distinction between syntagmatic and paradigmatic meaning is used here for purposes of analysis only; it does not imply any particular view with regard to the theoretical relation between both aspects of lexical meaning and particularly with regard to the question whether selectional restrictions are always an automatic consequence of a concept's paradigmatic characteristics.) As a preliminary step, notice that the concept 'to destroy' does not only appear as the notion 'to annihilate the existence of someone or something, to cause someone or something to disappear out of existence', but that it also exhibits the weaker nuance 'to undermine someone or something with regard

34

Prototypicality and salience

to some aspect of his existence' (without a complete destruction or a complete removal out of existence being implied). The distinction can easily be discovered within the personal group of applications. In (6), a person is killed, taken out of existence, while in (8) (and most likely also in the second quotation from (7), someone's existence is undermined from one point of view or another, but not entirely annihilated. Likewise, we can see that within the abstract group, (4) signifies the suppression of the existence of some abstract things as such, whereas in (5), plans, hopes, and expectations are undermined with regard to their realization and fulfillment: the plan as such is not removed (at least not to begin with), but it is reduced to ineffectiveness and futility. In short, both vernielen and vernietigen express the notions of complete destruction and partial damage, that is to say, the complete removal out of existence of something or someone, and the less drastic undermining in some respect, of the existence of people or objects. According to the syntagmatic context, these notions receive further specifications. For instance, with regard to persons, complete destruction means killing, but with regard to concrete things, destruction signifies material demolition, and so on. (For the sake of completeness it should be added that the equivalence of vernielen and vernietigen is less straightforward in present-day Dutch than it is in nineteenth-century Dutch. Some of the quotations discussed here are now felt to be rather awkward; in particular, it would be difficult to use vernielen with regard to persons. ) On the basis of the foregoing observations, one might be tempted to conclude that the semantic structure of vernielen and vernietigen in nineteenthcentury Dutch is completely identical: both syntagmatically and paradigmatically, they have the same range of application. However, a number of facts testify that both words have different prototypical structures, i.e., that they have different conceptual centers. There are two sets of facts to be considered: corpus-based facts relating to the way in which both words are used in our corpus of quotations, and introspective facts relating to the way in which the words are perceived by the speakers of the language. In general, consideration of these facts will lead to the conclusion that the abstract applications are central within the structure of vernietigen, and that the material applications are central in the cases ofvernielen. As such, each verb has a different semantic structure in spite of the fact that the elements of these structures appear to be the same.

Where does prototypicality comefrom?

35

3. Usage as evidence for prototypicality Five observations support the prototypical hypothesis. In the first place, the abstract group of applications is quantitatively more prominent within the structure of vernietigen than the material set of applications, while the reverse is true of vernielen, in which the material group is the most frequently occurring one. In both cases, the major group is represented by approximately three times the quotations of the less central group. In the second place, the differences in centrality show up in the fact that the prominent applications exhibit specifications and particular nuances that they do not have when they are peripheral within the structure of the lexical item. Thus, the material group of vernielen contains a metonymical extension of the application with regard to plants and crops, towards an application in which the fields and gardens where these plants and crops grow appear as the direct object of the verb. Likewise, the application with regard to buildings receives a figurative extension towards an application with regard to an allegorical 'wall' that separates two people. These extensions are probably not impossible within the concrete set of applications of vernietigen, but the fact that they do not appear there is statistically interesting: it indicates that the concrete application is more productive in the case of vernielen than in the case ofvernietigen. Conversely, the abstract group has nuances and additional specifications in the case ofvernietigen that are lacking in the same group with vernielen, although it is quite easy to imagine that they would in fact occur there. For example, vernietigen has a fairly large set of applications in which social movements, institutions, activities and so on are abolished, one quotation in which it is said that railway transport destroys distances (obviously, distances do not disappear as such, they are only functionally overcome), and one quotation in which a philosopher is said to destroy the soul (again, the soul is not destroyed as such, but the idea that the soul exists is metonymically abolished by the philosopher in question). None of these extensions of the abstract use of the concept 'to destroy' can be found in the case of vernielen, which is indicative of the fact that the abstract use is less prominent in the latter verb than in the semantic structure of vernietigen. In the third place, the salience of the material kind of usage can be derived indirectly from the nominalizations of both verbs. Both vernietiging and vernieling have the verbal sense 'the fact, the act or the process of destroying or being destroyed', but only vernieling exhibits the metonymical extension towards the concept expressing the result of that process or that

36

Prototypicali~yand salience

act, i.e., the concrete damage that issues from it. (In the latter case, the word is typically used in the plural: vernielingen more or less equals the notion 'damage'.) In the fourth place, the internal structure of the set of personal applications reflects the differences in prototypical structure between both verbs. To begin with, notice that the personal group contains concrete as well as abstract applications; to kill someone is clearly more concrete than to undermine someone's psychological well-being or his social position. If we then have a look at the mutual relationship between the abstract and the concrete subgroups of the application with regard to persons, we find that the abstract subgroup is proportionally dominant in the case of vernietigen, whereas the reverse is true in the case of vernielen. Also, we find that extensions of the concrete subgroup of the personal application with regard to other living beings than people or with regard to personifications, are not as strongly present in the case of vernietigen than in the case of vernielen. (It should be added that these observations have to be considered with more care than the previously mentioned points, since there is a general tendency throughout the centuries covered by the WNT-material, to re move the personal application from the structure of vernielen. There are relatively less personal applications in the structure of nineteenth-century vernielen than in the structure of either nineteenth-century vernietigen or sixteenth-century vernielen; as has al ready been mentioned, it is even more difficult to use vernielen with regard to persons in present-day Dutch. In any case, the nineteenth-century material does seem to show that the material subgroup of the personal application of vernielen is more resistant to the tendency in question than the abstract subgroup, as can be predicted from our centrality hypothesis.) Finally, the importance of prototypicality can be derived from the fact that different nuances play a central role within the core of each concept, whereas those nuances are not particularly important within the corresponding group in the other concept. Thus, the destruction of buildings and other human constructions is prominent within the material use ofvernielen, but is only rarely present within the material group of vernietigen. Within the structure vernielen as a whole, demolishing buildings is the single most frequently represented kind of usage, but within the structure of vernietigen, it is merely one among many equally important nuances of the material set of applications. In the same way, the central, abstract group within the structure of vernietigen is itself centered round applications relating to the dissolution, the cancellation, the annulment of agreements, commitments, engagements, obligations, permissions, rights, and so on, and of the laws,

Where does prototypicality comefrom?

37

orders, contracts etc. in which they are contained and through which they come into existence. Whereas vernielen only rarely exhibits this kind of usage, it is the most frequently occurring sense within the abstract group of vernietigen as well as within that word as a whole. In general, these facts of linguistic usage clearly favor the hypothesis that the abstract applications of the concept 'to destroy' are prototypical within the structure of vernietigen, whereas the concrete applications are prominent in the case of vernielen. Taking into account that each central group is itself concentrated round a dominant kind of usage, it seems plausible to say that the latter is the prototypical sense for each of the verbs in question. It should furthermore be noted that these prototypical phenomena seem to be connected with the etymology of the words. On the one hand, the abstract prototype ofvernietigen may well be connected with the abstract character of the words niet 'not', and nietig 'null and void, insignificant', on which it is based. Moreover, the common phrase nietig verklaren 'to declare something to be null and void, dissolve, annul something' corresponds pretty closely with the central notion within the abstract group ofvernietigen. On the other hand, the centrality of the application with regard to buildings in the structure ofvernielen seems to correspond with the etymological meaning 'to tear down, to throw to the ground' that we reconstructed above as the original meaning of the verb.

4. Introspective evidence of prototypicality Before we can deal with the introspective evidence in favor of the prototypicality hypothesis, two preliminary questions have to be answered. In the first place, how trustworthy is the introspective methodology? The paradoxical fact of the matter is that it is exactly the unreliability of introspection that makes it interesting for our purposes. If introspection were able to yield a completely adequate picture of the facts of linguistic usage (which is doubtful), it would simply reduplicate the results reached in the previous paragraphs on the basis of a direct examination of linguistic usage. But given the presupposition that introspection yields only a partial insight into the semantic structure of the words that are investigated, we can also presuppose that it will be exactly the prototypical kinds of usage of those words, that reach the introspective consciousness of the language user. We can use the results of the introspective method as support for the prototypical hypothesis if we presuppose that prototypical kinds of usage (precisely because they are more

38

Prototypicality and salience

salient than other applications) will more easily pass the threshold of conscious attention. Given this presupposition, the introspective judgments of native speakers may shed light on the question which kinds of usage are predominant within a certain concept. In the second place, how can the introspective method be used with regard to historical material? There are no nineteenth-century speakers of Dutch around to be asked what they think is the meaning of particular words, so how are we going to get introspective judgments at all? The fact is that we do have information on how the nineteenth-century speakers of Dutch perceived the near-synonyms that we are investigating, viz. in the form of synonym dictionaries. Synonym dictionaries (at least the older ones) are notoriously unreliable as descriptions of actual patterns of usage; most of the time, the compilers of synonym dictionaries rationalize away the actual identity of words by imposing distinctions that cannot be discovered in the actual facts of usage. However, these rationalizations need not always have proceeded out of the blue: it seems quite plausible that they were guided by the introspective judgments of the compilers. So, if we like to know something of the introspective insights of the nineteenth-century speakers of Dutch, we can have a look at the synonym dictionaries of that time to see whether the distinctions they make between vernielen and vernietigen (however inadequate as a picture of the complete set of possible kinds of usage) do indeed reflect the differences in prototypical structure of both words. And indeed, the nineteenth-century synonym dictionaries of Dutch do distinguish between vernielen and vernietigen along lines that fit into our hypothesis. On the one hand, there are those that draw the line syntagmatically, such as Weiland and Landre (1825), who state that vernielen can only be used with regard to 'lighamelijke dingen' (material things), whereas vernietigen is more widely used, in particular also with regard to 'menschelijke instellingen' (human institutions). De Beer (1897) expresses an analogous point of view. On the other hand, there are those that describe the distinction along paradigmatic lines, so that there would be an actual notional difference between the verbs in question, rather than merely a distinction in selectional restrictions. Whereas vernietigen is defined as 'to bring to naught, to annihilate' , vernielen is defined as 'to damage, to smash to pieces, to tear down'. In this sense, vernietigen implies a complete annihilation whereas there may be some pieces left of the original object in the case ofvernielen. It is easy to see that this paradigmatic point of view, which can be found among others in Pluim (1894), is connected with the previous, syn-

Where does prototypicality come from?

39

tagmatic one: it is precisely because vernielen relates to material things that the notion of remaining debris comes to the fore. Likewise, a complete annihilation (in which the original objects disappear completely) is less likely in the material world of concrete objects, so that the restriction of vernietigen to abstract objects will tend to be related to the notion of complete annihilation. This is in fact done by Weiland and Landre (1825), though not all proponents of the paradigmatic distinction adhere to the syntagmatic distinction. For instance, De Flines (1810) mentions that verniefigen can in fact be used with regard to material objects, but that there is a difference with vernielen in the degree of damage achieved. By and large, these views faithfully reflect the insight into the prototypes of vernielen and vernietigen that we have gained by considering the actual facts of linguistic usage. Syntagmatically, it is recognized that the material context is more important for vernielen, whereas abstract objects are predominant in the case ofvernietigen. Paradigmatically, this is reflected by the fact that vernielen carries overtones of material destruction and damage (think of the relationship between the prototypical usage of vernielen with regard to buildings, and the definitions of that word that bring to the fore the act of smashing and demolishing things), whereas verniefigen calls forth the idea of complete annihilation (as it were, wiping something off the face of the earth). As such, the stubborn efforts of the compilers of synonym dictionaries to find semantic differences among near synonyms seem to be not entirely gratuitous. To the extent that they try to capture the characteristics of the most salient kinds of usage of both lexical concepts, they strengthen our own hypothesis about the differences in prototypical structure among the verbs.

5. The functional explanation of prototypicality There are a number of interesting conclusions to be derived from the above analysis of the near-synonyms vernielen and vernietigen. First, prototypicality is an interesting new point of view in the study of synonyms. It is traditionally well-known in lexical semantics that there are relatively few true synonyms in natural languages, and the ways in which near-synonyms differ can be very diverse. Our discussion of vernielen and vernietigen shows that there is one more factor to be added to the list of differentiating factors: near-synonyms may be distinct with regard to the prototypical structure imposed on an otherwise identical range of application. Once again, the

40

Prototypicality and salience

importance of prototype theory for the traditional concerns of lexical semantics becomes apparent (cp. Geeraerts 1983a, 1983c, 1984a, 1985a); prototype theory opens up new perspectives in the study of synonyms. Secondly, there are some indications that introspective judgments in lexical semantics relate to the prototypically salient instances of concepts rather than to the full range of actual usage possibilities. If this can be confirmed by additional comparisons between introspective perceptions of lexical meanings and actual usage patterns, more will be known about the value of both methodologies (introspective and corpus-based) in lexical semantics. Also, if we maintain the classical view of modem linguistics that it is one of the goals of linguistic theory to account for the introspective judgments of native speakers, and if these judgments appear to be influenced by prototypical phenomena, yet one more reason presents itself for incorporating prototype theory into lexical semantics. Thirdly, the fact that vernielen and vernietigen have the same conceptual and collocational range of application, and yet differ with regard to the core and the periphery of their categorial structure, indicates that there are at least some cases of prototypicality that cannot be explained by means of the referential model. Vernielen and vernietigen refer to the same set of acts and processes; as such, the differences in their prototypical structure cannot be the automatic consequence of their referential range, as is implied by the family resemblance hypothesis. In addition, the physiological and the statistical explanation will not be of much avail either. There is no particular organ or mechanism for the perception of processes of destruction, and even if there were, we would still need two different physiological structures to explain the distinction between both verbs, which is beyond all intuitive plausibility. The statistical explanation is inapplicable for the same reason as the referential hypothesis: since the range of application of both verbs is the same, the frequency of occurrence of the processes referred to is the same for both verbs. That is to say, the frequency with which the demolishing of buildings occurs in reality, relative to the frequency with which, say, agreements are cancelled in reality, has exactly the same effect on both verbs, since these refer to the same objective reality. Because they denote the same things in reality, the structure of reality (either with regard to the frequency of occurrence of its elements, or with regard to the mutual resemblances among those elements) cannot be invoked to explain the distinction in semantic structure between vernielen and vernietigen. In short, we can reject all materialistic explanations of the prototypicality effects observed in the verbs under consideration. Indeed, the physiologi-

Where does prototypicality come from.?

41

cal, the referential, and the statistical hypotheses have this in common: that they try to explain prototypicality on the basis of materialistic data, either the material structure of the human perceptual apparatus, or the material characteristics (statistical or otherwise) of the referential range of the concepts involved. Given that we have to reject these materialistic hypotheses, we can provisionally choose, by elimination, for the psychological, functional explanation of prototypicality. To support this choice, I would like to make clear that the functional hypothesis has some additional advantages, besides the fact that it avoids the problem of the materialistic hypotheses. First, however, three remarks have to be made. To begin with, it might be claimed that a statistical explanation of the prototypicality effects in vernielen and vernietigen can indeed be given, if we take into account, e.g., that the material sense occurs much more frequently with vernielen than with vernietigen, or that the abstract specification of the notion 'to destroy' is statistically much more prominent in the latter verb than in the fornler. However, the frequencies that are mentioned here are linguistic frequencies, not referential frequencies, i.e., they are frequencies of occurrence of words, not of the things those words refer to. Because the frequency at stake here is linguistic rather than referential, it can hardly be invoked to explain prototypicality; as an aspect of linguistic usage, it is one of the things we have to explain, not one of the things that are themselves part of the explanation. We can use linguistic frequencies to determine what instances of a concept are prototypical (that is what we did in section 3), but explaining prototypicality on the basis of linguistic frequency is putting the cart before the horse. Some kinds of usage are not prototypical because they are more frequent; they are more frequent because they are prototypical. The apple is not a prototypical fruit because we talk more about apples than about mangoes, but because we experience apples more often than we encounter mangoes (and this fact, in turn, may be the reason why we talk more about apples). Frequency of linguistic occurrence may be a heuristic tool in the pinpointing of prototypes, but it is not the source of prototypicality as meant in the statistical hypothesis. The second remark has to do with the fact that criticism with regard to the referential, family resemblance model of prototypicality has already been formulated elsewhere. This has been the case in the work of Pulman (1983) and - in more stringent fashion - in the well-known article by Armstrong, Gleitman, and Gleitman (1983). They argue that gradience can be observed in concepts with rigid boundaries (their examples relate to natural numbers), so that family resemblances cannot be invoked to explain the differences in

42

Proto~vpicalityand

salience

salience among numbers. There are two reasons, however, why their argumentation is less relevant than they assume. First of all, they more or less equate prototype theory and the family resemblance model of the sources of prototypicality, whereas it is quite clear that the family resemblance model is merely one of a number of hypotheses concerning the sources of prototypicality: ruling out one hypothesis does not mean that one can ignore the others. And also, I do not think they are successful in presenting a counterexample to the family resemblance model. Even if a concept has rigidly defining characteristics, family resemblances may exist among the nondefining characteristics of the instances of that category. Since Cognitive Semantics is basically encyclopedist in its approach, these non-defining, 'encyclopedic' attributes should be incorporated into the computation of degrees of shared attributes. As Lakoff (1982) has shown, such encyclopedic, experiential factors do indeed occur with regard to numbers, and they can be used to explain the prototypicality ratings found experimentally. My third remark is this: my criticism of the materialistic hypotheses should not be overgeneralized. The fact that they do not work in the case of vernielen and vernietigen clearly does not imply that they do not work in any case, but merely makes clear that next to the physiological, the referential, and the statistical model, there will have to be at least one other source of prototypicality. Let us now come back to the functional model of prototypicality and try to elaborate it. Remember that the psychological hypothesis involves requirements that the cognitive system is to comply with if it is to function efficiently: prototypicality exists because it is cognitively advantageous. As we have seen, Rosch has specified this functional advantage in terms of the economical effect of informational density~ prototypical categories enable one to reach the most information with the least cognitive effort. This functional line of reasoning can be supplemented with some additional (and perhaps even more fundamental) functional reasons for having prototypical categories. We can base the discussion on one of the fundamental insights of cognitive psychology, viz. that cognition should combine structural stability with flexible adaptability. On the one hand, cognition should have a tendency towards structural stability: the categorial system can only work efficiently if it can maintain its overall organization for some time, if it does not change fundamentally any time new information has to be incorporated. At the same time, however, it should be flexible enough to be easily adaptable to changing circumstances. To prevent it from becoming chaotic, it should have a built-in tendency towards structural stability, but this stability should not

Where

doesproto~ypicali~y

come from?

43

become rigidity, lest the system stops being able to adapt itself to new and unforeseen circumstances. This necessity of flexibility is one of the aspects of lexical semantics that was recognized by the prestructuralist tradition of historical semantics, but that has been more or less lost in the meantime, as a result of the structuralist attention for fixed synchronic structures. Be that as it may, it will be clear that prototypically organized categories are particularly well suited to fulfill the double demand for flexible adaptability and structural stability. On the one hand, the fact that slightly deviant nuances can be developed within a particular category indicates that categories have the dynamic ability to cope with changing conditions and changing expressive needs. On the other hand, the same fact (that marginally deviant concepts can be incorporated into existing categories as peripheral instances of the latter) proves that these categories have a tendency to maintain themselves as holistic entities, thus maintaining the overall structure of the categorial system. Prototypical categories maintain themselves by adapting themselves to changing circumstances and new expressive needs; at the same time, they function as expectational patterns with regard to reality: new facts are interpreted in terms of information that is already at the disposal of the individual. The flexibility of the cognitive system does not only show up in the fact that it ran adapt itself to new experiences, but this flexibility is supplemented with the fact that existing categories have a formative influence with regard to experience; new experiences are fitted into the expectational patterns provided by the existing categorial system. Along these lines, prototypicality appears to be the outcome of some fundamental, deep-seated principles of cognitive functioning. The form of the conceptual system appears to be determined by a set of basic functional requirements, and prototypically structured concepts admirably meet these requirements. If this is correct, the same basic principles should also have a role to play in other cognitive disciplines. That is to say, if prototypicality is an emanation of some basic characteristics of all cognition, we should be able to find analogies of the prototypical idea in other fields of cognitive science, next to lexical semantics. I have tried to prove at length elsewhere (1985a) that this is in fact the case: the importance of interpretative schemata mediating between experience and existing knowledge is an idea that can be traced in a number of cognitive disciplines. It is very much apparent in Artificial Intelligence (Minsky's frame notion); it can be found in cognitive psychology, particularly in the work of Bruner, and to some extent in that of Piaget; it can be related to some of the views of the early, Husserlian

44

Prototypicality and salience

phenomenological movement in philosophical epistemology; and it has some important similarities with the paradigmatic conception of scientific enquiry inaugurated by Thomas Kuhn. These are exciting parallels because they suggest that the functional, psychological hypothesis concerning the sources of prototypicality can at the same time be the basis for a truly integrated cognitive science in which the insights of linguistics, Artificial Intelligence, cognitive psychology, philosophical epistemology, and the philosophy of science can be brought together under a common denominator. In this respect, the functional model of prototypicality, even if it does not rule out the possible importance of the physiological, the referential, or the statistical explanation, does seem to be more general than the latter, not just because it is based on fundamental principles of cognition, but also because similar views have been put forward in other branches of cognitive science.

6.

Onomasiological and semasiological aspects of Cognitive Semantics

Unfortunately, the optimistic perspective of the previous paragraph does not solve everything. To round oft the discussion, I would like to show that a complete explanation of all questions to be raised with regard to vernielen and vernietigen is far from available. The picture we have reconstructed so far looks like this: apparently, the linguistic conmmnity at some point in its development finds it convenient to have two distinct categories for the concepts of material and abstract destruction. Thus, a pair of etymologically distinct words becomes available, originally vernielen and vernieten, later on vernielen and vernietigen; their compound character ensures that one of them signifies material destruction, the other abstract annihilation. Gradually, the flexibility that is inherent in all human categorization extends these concepts beyond their etymological usage; as a result, they have the same range of application in the nineteenth century. But now consider the original situation in which these flexible extensions have hardly begun taking place. Is it then not irrational to use vernielen to express abstract annihilation, when you al ready have vernietigen or vernielen to do so? The question can be put in terms of global and local efficiency. As we have argued, the global efficiency of the conceptual system commands its flexible, prototypical organization. But there is, in the case of vernielen and vernietigen, also a local efficiency principle that says that it is uneconomic to have two terms expressing the same things. We are then

Where does proto(vpicality come from?

45

forced to ask: why does not the local efficiency principle stop the application of the global principle? Why is not the prototypical extension of vernielen towards abstract forms of destruction checked or prevented by the consideration that you already have a lexical category expressing abstract destruction? There is yet another way of formulating the problem: prototype-based flexibility is necessary because of the expressive needs of the speaker: he may want to express concepts for which no specific term is available. But why then would he use these flexible mechanisms of semantic extension if such a specific term is in deed available? One kind of answer might simply be that the global principle is stronger than the local principle; the global principle simply supersedes the local principle to the extent that local inefficiencies are created. We are then saying that the global principle is so general that its strength overrules the local principle, and that it applies even where it is not strictly necessary. Still, this does not tell us why the local principle is weaker than the global principle. Also, it is rather awkward to explain a mechanism that is unfunctionally overproductive on the basis of functional considerations. Couldn't we therefore find a more rational explanation of the flexible extensions? The way out, as far as I can see, is to take into account other kinds of expressivity than the purely conceptual one. Using vernielen to express a concept that is commonly expressed by vernietigen may be conceptually superfluous, but that does not mean that doing so may not serve particular expressive purposes. On the level of the linguistic form, for example, it may be quite functional to use another word than the usual one. The varieties of such a formally expressive synonymy are well-known in traditional lexical semantics; near-synonynls may exhibit connotational and emotional differences (as in euphemisms), stylistic differences (as in popular words versus poetic terms), or sociolinguistic differences (as in learned words versus common words). Perhaps we can even say that speakers have an urge for stylistic variation as such, even if the fornlal variants do not carry specific overtones; variation may well be governed by a straightforward desire to avoid monotony, to create new ways of expressing oneself, to experiment with unexpected innovations as a way of stressing one's own individuality. Moreover , it may well be that the importance of metaphor in natural language is determined precisely by its stylistic expressivity; metaphorical expressions would then be created primarily to add expressive weight to the message one wants to convey. (See Rudzka-Ostyn 1988.) It is quite plausible, then, that factors such as these have governed the extension of vernielen and vernietigen beyond their original meanings and into

46

Proto~ypicali~yand salience

each other's etymological range of application. For instance, using vernielen to express a process of abstract cancellation may have been stylistically particularly expressive, because the process of material destruction normally denoted by vernielen carried overtones of physical violence that were less marked in the case of vernietigen. The extended use of vernielen would then have been a case of metaphorical hyperbole. It is, however, very difficult to pinpoint exactly which form of expressivity is the relevant one with regard to the two verbs that we are concerned with here; our historical material for the earliest (Middle Dutch) history of vernielen and vernieten, for instance, is very hard to interpret with regard to such questions. Still, some clear cases may in fact be found. For instance, the first quotation of (5) clearly carries more overtones of violence, force, and intensity than the second quotation in that pair of examples. (This is mainly made apparent by the presence of a simile, marked by alzoo.) This suggests that the verbs highlight slightly different aspects of the situation described, or rather , represent the situation from different points of view (determined by the prototypical core of each verb). (In Langacker' s terminology, the distinction between the two verbs, when used with regard to the same process, might then be characterized as a figure/ground-distinction: vernielen takes the violent process as figure, and vernietigen the destructive result.) This is not an altogether implausible hypothesis, but it is unfortunately hard to confirm for the simple reason that the historical texts used here do not give us enough clues to discern such subtle differences in stylistic or emotional overtones. On the whole, then, what can we conclude from our discussion of this additional problem? On the one hand, it inspires caution with regard to our attempts to explain prototypical phenomena: the linguistic materials at our disposal do not always allow completely satisfactory answers with regard to the questions at stake to be formulated. On the other hand (and this is, I think, the more important conclusion), the discussion suggests that prototype formation may be influenced by other factors than purely conceptual ones. Stretching the meaning of a lexical item may be motivated by the desire to use another form than the one that is usual to express the idea in question; stylistic, sociolinguistic, connotational expressivity rather than purely conceptual needs may determine the flexible use of a category. In such a case, the conceptual coherence of the prototypically structured category (i.e., the fact that the new, peripheral kinds of usage have to be accessible from the prototypical core) constitutes a limit to the desire for formal variation: you can use a particular lexical item to express an idea that is usually signified by another word, but only on the condition that the idea in question is part of

Where does proto~vpicali~v come from?

47

the prototypical potentialities of that lexical item. Basically, you stretch an item's meaning to express something conceptually new, but you can also stretch it to express something conceptually old in a formally new way. This is a very important suggestion, because it implies a warning against a tendency that is a natural characteristic of Cognitive Semantics: the tendency, in fact, to look for purely cognitive or conceptual explanations of the facts one encounters. Taking the cognitive, experiential, encyclopedic nature of linguistic signs seriously should not imply looking only for strictly conceptual explanations. Language is not just content: it is also fornl, and its fonllal side has an expressivity of its own, which does seem to create lexical configurations that can hardly be explained if we only take into account the conceptual expressivity of language. In the traditional terms of lexical semantics, this means that the explanation of prototypicality should not restrict itself to the semasiological perspective (in which each category is considered on its own), but that the onomasiological point of view (in which it is studied how several items may express similar or identical concepts) should be taken into account as well. (conceptual expressivity is basically a factor connected with the semasiological explanation of prototypicality, whereas the onomasiological influences on prototype formation seem to refer to other kinds of expressivity, as was suggested by our study of vernielen and vernietigen. The incorporation of the onomasiological approach does not mean that Cognitive Semantics moves away from the functional perspective advocated in the previous section; non-conceptual expressivity is just as much a functional principle as purely conceptual expressivity and cognitive efficiency. Rather, the incorporation of onomasiology implies that Cognitive Semantics moves much closer to the rich tradition of lexical semantics, in which onomasiological mechanisms and configurations have been thoroughly studied (see Geeraerts 1986a: Chapter 1). Such a link with traditional approaches can only strengthen the linguistic attractiveness of Cognitive Semantics. To summarize: I have tried to argue, on the basis of a case study involving the Dutch near-synonyms vernielen and vernietigen, that the functional point of view is the most encompassing, most promising one for studying prototype formation, though it should not be restricted to purely conceptual expressivity and efficiency, but should also take into account the kinds of functional mechanisms that have traditionally been studied by the onomasiological approach to lexical semantics. The fact, however, that prototypicality may come from a number of diverse sources, also implies that an adequate explanation of conceptual structures will not be easy.

Chapter 3 The semantic structure of Dutch over

Originally published in Leuvense Bijdragen. Leuven Contributions in Linguistics and Philology 1992, 81: 205-230. The preposition over plays a role in Cognitive Semantics that is somewhat comparable to that of bachelor in Katzian semantics: from Brugman (1981, 1988) over Vandeloise (1990), Cuyckens (1991), Deane (1992), and Dewell (1994), to Evans and Tyler (2001), Tyler and Evans (2003), it has been a rallying-point for comparing competing forms of semantic analysis. The present chapter reproduces my own go at the subject. Although 1 would now try to firmly base the analysis on corpus materials rather than on introspectively derived sample sentences, I believe that the basic point of the paper still holds: the semantic structure of lexical items like over has to be seen as a multidimensional structure, in which covariation of semantic shifts along different dimensions constitutes the essential backbone of a prototype-based network. Specifically, the semantic structure of over involves at least two dimensions: on the one hand, the spatially relational dimension in the strict sense (with regard to which three distinct spatial configurations have to be distinguished), and on the other, the motional dimension, with regard to which we have to distinguish between cases of real motion, fictive motion, and zero motion. A third dimension to be added involves the existence of actual contact between the prepositional object and the other entity involved in the spatial relation. This type of analysis contrasts with the more common type of radial set analysis in Cognitive Linguistics, which is much more atomistic: a classical radial set analysis focuses on individual readings and the links that connect pairs of such readings, somewhat neglecting the underlying structure of dimensions that link various readings among each other. Such a focus on the underlying multidimensionality may help to avoid the pitfalls of the polysemy/univocality debate. There are (as argued in Chapter 5) convincing methodological reasons for being careful with the question: the criteria for answering it are neither stable nor consistent. But a practical, descriptive exercise such as the analysis of over may inspire the same conclusion. If you think of semantic structure primarily in terms of individual readings, it is natural to ask the question 'How many meanings does a word have?' (cp. Taylor 1992), i.e. is this a monosemous or a polysemous, or a hyperpolysemous word? If, however, one recognizes that covariation of different dimensions is a crucial characteristic of the structure in question, the individual readings (and the number they have) become less important from a practical and theoretical point of view.

The semantic structure ofDutch over

49

1. Scope and purpose ofthe exercise

The English spatial preposition over occupies an important position within Cognitive Semantics. On the one hand, Brugman's analysis (1981), which also appears with slight modifications in Lakoff (1987), is one of the earliest examples of a radial set analysis of lexical semantic structures. On the other hand, Vandeloise's reaction against the Brugman/Lakoff-analysis (1990) raises a number of fundamental questions concerning the nature of polysemy and the description of polysemy within a Cognitive framework. In the following pages, a similar methodological point concerning lexical-semantic research in Cognitive Semantics will be dealt with on the basis of an analysis of over. However, by presenting an alternative for the treatment in Cuyckens (1991), the paper will focus on the Dutch item over rather than on its English counterpart. Also, the theoretical focus will not be on the possible univocality of over (as discussed by Vandeloise), but on the importance of a detailed definitional analysis oflexical items. Two points will be discussed with regard to which Cuyckens's analysis would seem to be subject to improvement: a moderate lack of comprehensiveness, and an insufficiently specific treatment of the links between the various senses of over. Both of these characteristics seem to follow from a tendency (not altogether surprising in the framework of a prototypetheoretical approach) to concentrate on the most salient instances of use of a category: on the one hand, there will be less attention for the non-typical cases; on the other, the salient cases will tend to be seen as self-contained entities. In contrast with such a synthetic conception of the salient instances of a category, an analytical alternative will be presented that not only tries to determine the conceptual relationship between the various instances of a category as precisely as possible (specifying the various attributes that link the instances to each other), but that also tries to make sense of less obvious cases. There are a number of restrictions on the following treatment of over that should be mentioned explicitly. Only the prepositional uses of over in the strictest sense will be taken into account; predicative and adverbial uses (de pUn is over, de pUn gaat over), or the use of over in compounds (overblijven, overdenken, overgaan etc.) are beyond the scope of the present treatment. The discussion will be limited to the spatial senses of the preposition; figurative and temporal uses (een disGussie over iets, een vergadering over drie dagen) will not be envisaged. Further, a lexical field approach (which would involve comparing over with related prepositions, specifically

50

Prototypicality and salience

with boven) will not be systematically adopted (although there will be incidental remarks about synonymy and near-synonymy). Likewise, there will be no contrastive analysis (with the BrugmanlLakoff-analysis of English over, or with German iiber, which combines the readings of over and boven). Since these are all legitimate topics for investigation that should really be taken up in a more extensive treatment of the preposition, it will be clear that the restrictions are a matter of convenience rather than principle. Like Cuyckens's analysis, the discussion focuses on the core of the semantic structure of over, which would be the basic step in a more detailed description at any rate. There is, however, one restriction on the present treatment of over that does involve matters of principle rather than convenience. In fact, the paper will remain agnostic about the question whether the various kinds of usage that will be described, constitute different meanings or not (in the theoretical sense in which 'meaning' is distinguished from 'vagueness'). How, in fact, to determine whether a particular instance in which a lexical item is used constitutes a different meaning of that item? As argued in Geeraerts (l993a), not only are there various kinds of semantic identity tests that might be invoked (and that have been invoked) to settle the question, but also, these various tests appear to yield mutually contradictory results. To say the least, linguistic semantics has not come up with a universally accepted, unequivocal operational test for distinguishing polysemy from mere referential vagueness. Against the background of this methodological indetenninacy, the present paper will not endeavor to establish the meanings of over (if any such set of clearly delimited semantic entities exists at all), but will rather pursue a more modest goal; it will try to describe the structure of the referential range of application of over in some detail. Which portions of the structure constitute a different meaning in the theoretical sense is a question that is beyond the intended scope of this paper - but conversely, it is a question that can probably only be answered properly on the basis of a referential analysis of the type to be demonstrated here. Finally, there is a restriction on the originality of the following analysis. Although it will try to go one step further than Cuyckens's analysis by presenting an alternative for the overall structure of the item, it also draws heavily on his work when individual applications are at stake (just like Cuyckens himself has drawn on the work of Brugman and Lakoft). The distinctions and correspondences will not be mentioned separately each time they appear, but a global comparison with Cuyckens's proposals will be made in paragraph 6.

The semantic structure ofDutch over

2.

51

The basic referential structure

The absence of a generally applicable operational criterion for meaning discrimination does not imply that the existing approaches could not be used as a starting-point for the kind of structural analysis to be illustrated here. For instance, lexical polysemies could at least in some context show up as a sentential ambiguity. To take over eruse's example (1982), the sentence (1), when uttered at the meeting of a political committee, may mean either that delegate Balder has taken another seat in the conference room, or that he has changed his opinion on the topic at hand. This ambiguity, then, testifies to the polysemy ofposition. (1)

(2)

Bafder has shifted his position! Wafdemar fletst over de fijn

'Waldemar rides his bicycle over the line' Now consider (2).2 Sentence (2) may refer to at least three different situations. First, Waldemar may be riding across the line (as when he moves from one lane to the other). Second, he may be riding on and along the line, staying roughly within its boundaries as he follows it. (He might be doing this as a game, or as a kind of test to enable the police to ascertain whether he is riding under the influence of alcohol). And third, Waldemar may be riding at the other side of the line (seen through the eyes of an implicit observer). For instance, imagine a group of children riding their bikes at a schoolyard, but being forbidden to cross a particular line. Sentence (2) could then be used to point out that Waldemar ignores the prohibition. Notice that the implicit observer turns this third reading into an example of subjectification in the sense of Langacker (1990b): the spatial relation that is expressed by over holds between the bicycle (or at least, between the act of riding the bicycle) and the line, but it is construed with an implicit reference to the subjective presence of an observer. In Figures la, 1b, and lc, the triple ambiguity of (2) is graphically represented. Accordingly, over can be assigned three distinct readings, as defined in [i]-[iii]. With regard to example (2), the definitions specify the spatial characteristics of the activity referred to by the verb fietsen: it is an activity that extends from one point to another (either within the spatial area defined by the prepositional object x, or by crossing that area), or an activity that is situated in a region construed as being to the other side ofx.

52

[i] [ii] [iii]

Prototypicality and salience

Extending from a point or region at one side ofx to a point or region at the other side of x, while crossing x Extending from one point or region within the space defined by x, to another point or region within that space (without going outside of it) Situated at the other side of x, relative to the position of an observer

1b

1a

~, .. _--

1c Figure 1. The basic semantic configurations of over

Now, it is crucial for the semantic structure of over that the three spatial relations as just defined can also occur in other syntagmatic contexts than the one represented by (2). First, the prepositional phrase can indicate the spatial extension of a fictive rather than a real motion. In (3), for instance, there is no real motion of the line; still, the line extends from a point on one side of the prepositional object x to a point to the other side of that object (which motivates the use of over), and further, this extension is construed, through the use of lopen, as a process of fictive motion. 3 Second, if over can be used to express stative spatial relations, it should be possible to combine it with stative verbs that do not express motion. An example is (4), where it is said that the blanket extends from one side of the fence to the other; it thus covers the fence, if one likes. Example (5) presents another syntagmatic context in which the 'zero motion' option is realized; if the prepositional phrase modifies a noun rather than a verb (and if the noun does not refer to a process or an action), the interpretation is automatically stative.

The semantic structure ofDutch over (3)

(4) (5) (6)

53

Er liep een rode streep over het woord verantwoort 'A red line ran across the word verantwoort' (which contains a spelling mistake) De deken hangt over de schutting 'The blanket hangs over the fence' De deken over de schutting is van Godelinde 'The blanket across the fence belongs to Godelinde' Het vliegtuig vliegt over de stad 'The airplane flies over the town'

It will be clear by now that the semantic structure of over involves at least two dimensions: on the one hand, the spatially relational dimension in the strict sense (with regard to which three distinct spatial configurations have to be distinguished), and on the other, the 'motional' dimension, with regard to which we have to distinguish between cases of real motion, fictive motion, and zero motion. A third dimension to be added involves the existence of actual contact between the prepositional object and the other entity involved in the spatial relation. As (6) exemplifies, the reading defined in [i] can also occur when there is no physical contact between trajector and landmark. (In fact, (6) could be shown to be ambiguous along the three readings [i]-[iii]). On the basis of these three dimensions, the referential range of application of over can be charted systematically. As Table I and the examples (7)-(22) show, almost all possible combinations of all values on the three dimensions actually occur. (7) (8) (9)

(10) (11) (12) (13) (14)

De haf roft over de fijn 'The ball rolls over the line' (to the other side of it) Het vliegtuig vliegt over de stad 'The plane flies over the town' (to the next one) Er liep een rode streep over het woord verantwoort 'A red line ran across the word verantwoort' De spoorlijn loopt over de rivier 'The railway line runs across the river' De deken hangt over de schutting 'The blanket hangs over the fence' Er hangen donkere wofken over de stad 'Dark clouds hang over the town' Engefram kuiert over de markt 'Engelram strolls over the market-place' O([ert rijdt over de fijn 'Olfert rides over (on and along) the line'

54

Prototypicality and salience

Table I. Combinations of values in the semantic structure of over FICTIVE MOTION

MOTION

ZERO MOTION

+contact

-contact

+contact

-contact

+contact

-contact

[i 1

(7)

(8)

(9)

(10)

(11)

(12)

[ii 1

(13)(14)

(15)

(16)

(17)

(18)

(19)

[iii 1 (15) (16)

(17) (18)

(19) (20)

(21) (22)

(23)

(20)

(21)

(22)

Het vliegtuig cirkelt over de stad 'The airplane circles above the town' Condensatiestrepen lopen kriskras over de blauwe lucht 'Condensation trails run haphazardly over the blue sky' Ermingards blik dwaalde over het blad 'Ermingard's eye wandered over the sheet of paper' Er figt een armoedig kleedje over de vloer 'A shabby rug lies on the floor' Een dunne streep mist hangt over de rivier 'A small trail offog hangs over (above and along) the river' Over de rivier exerceerde de vijandelijke mifitie 'Across the river, the enemy militia was exercising' Over de rivier loopt een rij bomen fangs de beek 'Across the river, a row of trees runs along the brook' Dordrecht figt over de Moerdijkbrug 'Dordrecht lies across the Moerdijk bridge' Over de brug hangt een ba//on 'Across the bridge, a balloon hangs in the air'

Three refinements are necessary to complete this picture of the basic range of application of over. First, it will be noted that the distinction between and cases is basically irrelevant for the examples of [iii], because the position of the trajector x to the other side of the landmark (the prepositional object) automatically implies that there is no contact between them. It is, of course, possible to interpret the distinction between (22) and (23) as involving a versus situation, but because the contact in question does not involve the bridge (which is the landmark for the relation expressed by over) but rather the ground in general, the distinction does not affect the structure of over. In this sense, then, the gaps in the structure represented in Table 1 are systematic, not incidental ones.

The semantic structure ofDutch over

55

As a second remark, it should be established that the distinction between [i], [ii], and [iii] is not just one between, respectively, a 'directional' sense of over, a 'path' reading, and a 'locational' interpretation. Judging only on the basis of the leftmost column of Table 1, it would seem possible to say that over specifies either the direction of a particular movement, the path along which a movement takes place, or the place where something is located. However, it should be clear from Table 1 as a whole that the readings of [i] and [ii] are just as locational as [iii] as a whole. And third, let us notice that there is a particular complication with resultative readings of the verbs, when a location referred to by over is presented as the result of the movement referred to by the verb. The problem can best be introduced in connection with transitive causative constructions. In such constructions, the moving object is not the subject of the verb, but an entity on which the verbs exerts a direct or indirect force. To begin with, consider cases of the causative construction with what we have called the case of [i], as in (24)-(25). In these cases, the distinction between [i] and [iii] fades away, given that there are various possible paraphrases for the causative construction: while [i] characterizes the directional path of the movement of the affected entity, [iii] simultaneously characterizes the resultative final state of that movement - and both ways of reading the causative construction seem equally appropriate. Thus, (24) could be paraphrased as 'Lidewij causes the ball to move over the fence' (i.e. from one side of the fence to the other, which brings into play the directional reading [i] of over), but also as 'Nimrod causes the ball to reach a position over the fence' (i.e. at the other side of the fence, which activates the locational reading [iii]). If we call the latter interpretation the resultative one, the examples (24)-(25) would seem to suggest that the locational resultative case is epiphenomenal with regard to the directional reading of over. (24) (25) (26)

(27)

Lidewij gooit de bat over de schutting 'Lidewij throws the ball over the fence' Fiobert duwt de fiets over de markt 'Fiobert pushes the bicycle over the market place' Evermoed liet de deken over de schutting vallen 'Evennoed dropped the blanket over the fence' (in the sense that it hangs over the fence) De deken va!t over de schutting 'The blanket falls over the fence' (in the sense that it falls on top of it and subsequently hangs over it)

56

Proto~ypicali~vand salience

In (26), however, the movement of the affected entity cannot be described in terms of any of the cases charted in Table 1, although the resultant state is a simple case of the application of [i]. Consequently, the resultative cases cannot always be reduced to a by-product of a directional reading, and the label in Table 1 should be read so that it includes resultative constructions like in (26). At the same time, we can now also see from (27) that such resultative constructions are not restricted to transitive causatives. The 'resultative' construction undoubtedly deserves to receive a closer look than the one it has been subjected to here; in the context of the present paper, however, it will play only a minor role.

3. Sample syntagmatic restrictions The survey of the range of application of over given in the previous paragraph has to be further elaborated in two different ways. On the one hand, it has to be supplemented with some peripheral cases of semantic relations (which will be the subject matter of paragraph 4). On the other, it has to be refined by an investigation into the syntagmatic restrictions that the various applications of over are subject to. In fact, not any kind of trajector can be associated with any kind of landmark by means of over. Notice, to begin with, that the case of [i] can be used without restrictions. Or at least, objects, surfaces, and lines can move over each other in all possible trajector/landmark combinations, as shown in Table 2 and examples (28)-(45). (Within each cell of the figure, the first example is a one, the second a onel (28)

Berthilde jiefst over de heuvel naar huis

(29)

Fredegonde jietst over het plein naar het begin van de Agnietenstraat

'Berthilde bicycles home over the hill' 'Fredegonde bicycles over the square to the beginning of the Agnietenstraat' (30)

Dietbald jietst over de witte lijn

(31)

De trein stoomde over de heuvel

(32)

De trein stoomde over de vlakte

(33)

De trein stoomde over de grens

'Dietbald crosses the white line with his bike' 'The train steamed over the hill' 'The train crossed the plain' 'The train crossed the border'

The semantic structure ofDutch over

57

Table 2. Trajector-landmark combinations in the structure of over LM OBJECT

LM SURFACE

LM LINE

TR OBJECT

(28)(37)

(29)(38)

(30)(39)

TRSURFACE

(34)(40)

(35)(41)

(36)(42)

TR LINE

(31)(43)

(32)(44)

(33)(45)

(34)

Een Romeins legioen marcheerde over de heuvel

(35) (36) (37)

Een Romeins legioen marcheerde over het plein Een Romeins legioen marcheerde over de grens Het vliegtuig vloog over het monument

(38)

Het vliegtuig vloog over de vlakte naar de stad in het gebergte

(39)

Het vliegtuig vloog over de rivier

(40)

De wolk dreef over het monument

(41) (42) (43)

De wolk dreef over de vlakte De wolk dreef over de rivier Een rij vliegtuigen vloog over het monument

(44) (45) (46)

Een rz/ vliegtuigen vloog over de vlakte Een rij vliegtuigen vloog over de rivier Het vliegtuig cirkelt over de stad

,A Roman legion marched over the hill'

'The plane flew over the monument' 'The plane flew over the plain towards the mountain town' 'The plane flew over the river' 'The cloud drifted over the monument'

'A row of airplanes flew over the monument'

'The airplane circles over the town' In passing, it could be noted that the classification of an entity as either object, line, or surface can be a matter of perspectival construal rather than matter-of-fact objectivity. To give just one example, a town could be seen as a point-like object, but it can also be conceived of as a surface. In a sentence like (46), for instance, assigning [ii] to over automatically entails thinking of the town as a surface: because the movement of the plane has to remain within the (vertically projected) boundaries of the town, thinking of the town as a two-dimensional surface is the natural thing to do. Now, without systematically exploring the restrictions on all the possibilities mentioned in Table 1, let us consider two major examples of limitations on the combination of trajector and landmark. Importantly, the restric-

58

PrototypicaliZv and salience

tions will turn out to be an automatic consequence of the spatial relations defined by [i]-[iii]. (Later on, we shall come across an example of restrictions that are truly ad hoc.) To begin with, let us consider [ii]. Because the spatial relation referred to by over has to remain within the boundaries of the landmark, the prepositional object has to have a line- or surface-like extension: it should be possible to conceive of it as a line or a surface. This shows up quite clearly when considering three-dimensional objects. Whereas (47) and (48) are straightforward cases of lines and surfaces, (49) is an example of a threedimensional object whose third dimension (height) does not prevent it from having a bounded surface that allows for an over-relation. In (50), however, the three-dimensional object's vertical dimension is so important that no surface for aimless wandering can be imagined. As (51) shows, reading [i] of over does not impose this restriction. This does not mean, however, that an object such as schutting could not occur at all with reading [ii] of over. Example (52) shows that the relevant feature is not the objective horizontal or vertical orientation of the surface, but rather the functional characteristic of 'treadability': what is not a surface for free movement for human beings is a natural path for insects. Furthermore, (53) shows that (51) is actually ambiguous between [i] and [ii]5 Examples (51) and (53) are possible with [ii] if a situation is envisaged in which someone crawls along the top of the fence; in that case, the relevant character of the landmark 'fence' is not that of a surface, but that of a line. That is to say, any interpretation of (51) and (53) on the basis of[i] would have to link up with (47) rather than (48). (47)

(48) (49) (50) (51)

(52) (53) (54)

Trudofietste over het pad 'Tmdo rode his bicycle along the path' Sulfrida kuierde over de markt 'Sulfrida strolled over the market place' Kunigonde dwaalde over de berg 'Kunigonde wandered over the mountain' *Dankmar dwaalde over de schutting 'Dankmar wandered over the fence' Gradolfklom over de schutting 'Gradolf climbed over the fence' De spin kroop maar door over de schutting 'The spider kept on crawling over the fence' Odilde kroop maar door over de schuffing 'Odilde kept on crawling over (on and along) the fence' De trein glijdt over de monorail 'The train glides over the monorail'

The semantic structure ofDutch over (55) (56)

(57)

59

Een kolonne soldaten marcheerde over de heuvel 'A line of soldiers marched over the hill' Een rij pukkels over de wang van Adelheid is voor haar een bron van veelzorgen 'A row of pimples on Adelheid' s cheek is a source of much concern to her' Een rij bomen loopt schuin over het veld 'A row of trees mns diagonally over the field'

On the other hand, because the trajector of an over-relationship according to [ii] has to remain within the boundaries of the landmark, it will typically be smaller than the landmark - small enough, at least, to move about on the surface or along the line constituted by the landmark. Specifically in the case of line-like landmarks, however, this does not imply that the trajector has to be smaller than the landmark in all directions. Taking into account that real lines are never merely one-dimensional, but always have a certain width as their second dimension, the trajector does not have to be narrower (but it does have to be shorter). As example (54) shows, the moving object may be wider than the actual line. Two additional remarks have to be made at this point. First, the distinction between line-like and surface-like landmarks in combination with [ii] is not unimportant, because it correlates with a restriction on the kinds of movement that are possible with reading [ii]. Line-like landmarks, in fact, automatically impose an orientation on that movement. Because the movement goes from one point on the line to the other, and because the line restricts the sideward movement of the trajector, the movement follows the line. Automatically, the line defines the path of the movement. Surface-like landmarks do not define such a path: the movement from one place to another on the surface may be random (as in (49), where there is a suggestion of aimless wandering or loss of orientation), but it may also be a straight line. Because, in the latter case, the straight line will typically lead from one side of the surface to the other side, there is a really close relationship with the 'crossing' reading [i]: going over a surface according to [ii] is then the same thing as being in the middle of the process of going over that surface according to [i]. As a second remark, let us note that an example like (55) shows that a line-like trajector may be discontinuous, or, if one wishes, virtual. If we jump to the and applications of [i] for a moment, further examples may be found of such virtual/discontinuous trajectors, as in (56) and (57). Needless to say, examples such as these again

60

Proto(ypicali(y and salience

underline the importance of perspectival construal in natural language: spatially disconnected entities are conceived of as a single entity. Let us now turn to a second major example of syntagmatic restrictions. The restrictions that hold for the reading of [i] are to some extent the reverse of those just described. With regard to [ii], we noted that the landmark was necessarily line-like or surface-like, and that the trajector had to be smaller (in the relevant dimension) than the landmark. In the case that we are concerned with here, the trajector has to be construed as line-like or surface-like, and the landmark has to be smaller (in the relevant dimension) than the trajector. On the one hand, the trajector of the cases of [i] has to have a line-like or surface-like extension of its own. Because the spatial extension that is expressed by the prepositional phrase does not involve the path of a movement (of an entity of any dimensionality), but rather attaches to the entity itself, the trajector cannot be construed as pointlike. On the other hand, because the trajector has to extend from one side of the landmark to the other, it has to be larger than the latter. In the case of line-like trajectors, this means that the landmark is smaller than the trajector in the direction in which the trajector extends. (As the examples (58) and (59) illustrate, the landmark may very well be larger than the trajector in the direction orthogonal to the trajector's.6) (58) (59) (60) (61) (62) (63) (64) (65)

Er ligt een kabel over de weg 'There's a cable lying over the road' Er hangt een deken over de schutting 'There's a blanket lying over the fence' Een deken ligt over het bed 'A blanket is lying across the bed (covers the bed)' Heriswitha legt een vel papier over de vlek 'Heriswitha puts a sheet of paper over the blot' Het rode blad ligt over het gele blad 'The red sheet lies over the yellow one' Ricfridus heefl sproeten over heel z 'n gezicht 'Ricfridus has freckles all over his face' Er ligt een hoop zand over het emmertje 'A heap of sand covers the little bucket' De kaasstolp Iigt over de Camembert 'The cheese cover lies over (covers) the Camembert'

In the case of surface-like trajectors, the trajector may extend from one side of the landmark to the other in various directions, but it should do so in at least one direction. At this point, we come across a phenomenon of the

The semantic structure ofDutch over

61

same kind as the one we encountered when we discussed line-like landmarks in our first example of syntagmatic restrictions. Just like applications of [ii] with line-like trajectors yield the interpretation 'following along x', applications of [i] with surface-like trajectors may result in the interpretation 'covering x', as in (60) and (61).7 Another special case of surface-like trajectors and landmarks is when they are both the same size. When you have a yellow sheet of paper and a red one of precisely the same size, (62) is possible when one sheet lies precisely on top of the other. This is a special case because it is situated at the borderline between [i] and [ii] (both taken in the interpretation, of course): according to [i], the top page should be larger, whereas it should be smaller than the bottom page according to [ii]. Equal size is so to speak the upper limit of [ii] and the lower limit of [i]. However, because it would be natural to say that the top sheet covers the bottom one, and because the 'covering' interpretation is a usual one with [i], there would seem to be at least one indication that special cases of the 'equal size' kind are intuitively conceived of along the lines of [i] rather than [ii]. Note also that the equal size cases also occur with 'virtual' examples that are analogous to (56) and (57). In an example like (63), the virtual surface defined by the freckles is implied to have the same size as the face; characteristically, the freckles cover the face. To round off, let us have a look at (64) and (65). Because the trajectors in these cases are neither clear examples of lines or of surfaces, it might seem that they contradict the earlier statement that the trajectors of the interpretation of [i] are construed as line-like or surface-like. One way out of the problem would be to claim that the relevant 'surface' in these cases is the vertical projection of the maximal expansion of the trajectors; and another would be to point out that surface-like trajectors may involve curved and bended surface. However, rather than trying to salvage the letter of the original statement of the restriction, it would be more in line with the spirit of the original statement to accept that it has to be slightly modified: the basic point is to see that point-like trajectors are excluded (because they have a minimal extension), but this does not rule out object-like trajectors that are construed as masses or spheres rather than points.

4.

Some less systematic cases

Three cases have to be mentioned that fall outside the range of applications charted in Table 1. (Each time, the restrictions on the additional cases will

62

Prototypicali~yand salience

turn out to be ad hoc rather than an automatic consequence of the spatial relations expressed by over, as in the previous paragraph.) The most common case is illustrated by (66)-(69); the meaning expressed by over can be defined as in [iv]. This kind of usage can most easily be seen as a transformation of those that fall under [iii]; specifically, the prepositional object is now the starting-point of a virtual path rather than the path itself. In contrast with [iii], the path is necessarily an open area such as a street or a square (or even a dinner table) rather than just any obstacle that can be crossed: examples such as (66) are ruled out in a situation in which the museum is situated at the foot of a hill and it is your intention to make clear that Gosbert lives over the hill. This is to say that this kind of usage is subject to specific restrictions that do not follow automatically from the spatial relation that is being expressed. The reading of over at stake here is not just a metonymic transformation of [iii], but it is subject to additional restrictions that are absent in the case of [iii]. The restrictions can probably best be understood as involving two people that can see each other in the face (which explains the open space), or entities such as buildings, that are anthropomorphically modeled as looking out towards each other. In fact, the English word facing that can be used to translate over in these cases, expresses the same anthropomorphic metaphor. (66)

Gosbert woont over het museum

(67)

Aan ta/el zat JvJeinol( over JvJombert

(68)

Het museum bevindt zich over de rechtbank

(69)

In de erehaag stond Edmar over Notker

'Gosbert lives across the museum' At the table. Meinolf was sitting facing Mombert 'The museum is situated across the magistrates' court' 'In the double row of honor, Edmar stood facing Notker' [iv1

Situated at the end of a virtual path that extends from x in a straight line across an open space

A second additional case to be mentioned is illustrated by (70)-(72). The meaning of over in these examples can be defined as in [v]. Clearly, this type of usage links up with the cases of [ii]. As in the latter, an entity is situated spatially within the boundaries (through vertical projection) of x, but above x. The distinction between the cases of [ii] and [v] is that the former presupposes that the trajector has a certain extension (as of a line or a surface), whereas the latter specifically allows for point-like trajectors (that are so to speak without extension). This kind of usage, then, is a relaxation (in fact, a generalization) of the

The semantic structure ofDutch over

63

reading described under [iii]. As shown by (73) and (74), the relaxation is subject to the ad hoc restriction that only cases are allowed. It should be remarked that the presence or absence of [v] is particularly sensitive to relatively minor situational differences and differences of perspectival construal. If, for instance, (75) applies to a situation in which the dangling movement of the spider covers a sizeable area, (75) would not be interpreted along the lines of (72), but along those of(15). And whether (76) is a case of [v] or [ii] would seem to depend on the size of the kettle with regard to the fire (or on the extent to which these are taken into account by the observer). Methodologically speaking, the ease with which possible cases of [v] shift towards one of the other readings testifies to the peripheral character of [v]. (70)

De lamp hangt over de tafel

(71)

Het ongeluk gebeurde over de zee

(72)

Het spinnetje hangt sti! over de vensterbank

(73)

*Het spinnetje figt sti! over de vensterbank

(74)

*Het boek rust over de tafel

(75)

Het spinnetje bungelt over de vensterbank

(76)

De ketel hangt over het vuur

'The lamp hangs over the table' 'The accident happened over the sea' 'Motionlessly, the little spider hangs over the window sill' 'Motionlessly, the little spider lies over (on) the window sill' 'The book rests over (on) the table' 'The little spider dangles over the window sill' 'The kettle hangs over the fire' [v 1

Situated above x, within the spatial boundaries of x

A third and final special case is illustrated by examples (77)-(80). In all of these, the trajector has a circular structure, while the landmark is like a pole or stick. While (80), which features a horizontal landmark, falls under the application of [i] in roughly the same way that (11) does, the other cases are problematic because the landmark is in vertical position. A straightforward allocation to the application of [i] is out of the question, because the trajector does not cross or cover the landmark: it does extend from one side of the landmark to the other, but it does so sideways, by going round the landmark rather than by crossing and covering it. According to this line of thought, the examples (77)-(79) illustrate a separate reading [vi], which can be considered a relaxation of the cases of [i]; specifically, the relaxation involves discontinuous trajectors (roughly, with a hole in them) that

64

Prototypicali(v and salience

extend from one side of the landmark to the other by going round it horizontally rather than by going over it vertically. It is important to notice, however, that the syntagmatic restrictions on this type of usage still reflect something of the original meaning [i] from which [vi] appears to be derived. In fact, as the ungrammaticality of (81) illustrates, the landmark cannot be so tall that the trajector could not have reached its position by being pushed downward from above; a trajector that has reached its position of sideways encapsulation by a roundabout motion cannot be said to go over the landmark. Apparently, only the same kinds of covering movement that lead to the configurations illustrated by cases like (60), (61), and (65) can lead to the spatial configuration captured by over in its reading [vi]. And because (60), (61), and (65) are unproblematic examples of [i], the link between [i] and [vi] shows up in the syntagmatic restrictions on over in reading [vi]. At the same time, it will also be clear that these restrictions cannot be predicted from the spatial configuration [vi] itself. They can be derived from the semantic history of the reading [vi] within the structure of over, but from the point of view of the configuration [vi] as such, the restriction is an arbitrary one. (77)

Er hangt een krans over het standbeeld

(78)

De ring ligt over de kegel

'A wreath hangs over the statue'

'The rings lies over the cone' (79)

Wigbert schuift de rol toiletpapier over de pleeboy

'Wigbert slides the toilet roll over the toilet roll stand' (80)

De zwembandjes hangen over een lange haak

(81)

*Een geellint was over de boom gebonden ,A yellow ribbon was tied over the tree'

[vi]

Circularly encapsulating x, extending from one side of x to the other by going round x

'The lifebelts are hanging on a long peg'

To summarize, we have found three special cases that can each be considered a relaxation or generalization of one of the three basic readings discussed in section 2: [iv] is linked to [iii], [v] to [ii], and [vi] to [i]. On the one hand, then, [iv]-[vi] correlate with and testify to the systematicity of the basic range of application of over as discussed in section 2. On the other hand, the less systematic features of [iv]-[vi] show up in the fact that the syntagmatic restrictions that they are subject to CaImot be derived automatically from the semantic content that they express (in contrast with the syntagmatic restrictions on the major readings that were treated in section 3).

The semantic structure ofDutch over

65

5. Prototypicality effects The analysis so far might seem to imply that all applications of over have the same status as far as structural weight is concerned. This is hardly the case, though: there are unmistakable prototypicality effects at work. Intuitively at least, there can be no doubt that (for instance) the cases discussed in the previous section are less preponderant than [i] or [ii]. Given that we do not have psycholinguistic experiments at our disposal to corroborate this intuition, what criteria do we have to establish differences in structural weight among the various instances of over? In the first place, the semantic relations between [i],[ii], and [iii] establish the central role of [i], at least if it can be accepted that that reading is central from which the others can be most economically derived. In particular (restricting, for the sake of the argument, the discussion to cases of real motion), [i] specifies a path across an obstacle while [ii] refers to a part of that path, and while [iii] refers to the endpoint of the path. Both [ii] and [iii] are therefore relatively straightforward metonymical extensions from [i]. Taking into account, however, that metonymy often works both ways, is there any reason why, for instance, [ii] instead of [i] could not be the logical centre along this axis of the structure of over? Notice that the transition from [i] to either [ii] or [iii] can be taken in one step: there is a straightforward type of metonymy (basically of the synecdochal kind) to justify the development. Going from [ii] or [iii] to [i], however, is less straightforward. For instance, while [ii] merely specifies that a certain movement is situated within the boundaries of the landmark, there is no compelling reason why that movement should be intended at crossing the landmark; that is to say, because movement across x always implies movement within the boundaries of x, but not the other way round, deriving [ii] from [i] is more plausible than the alternative. Similarly, while crossing x normally implies that one reaches the other side of x, being located at the other side of x does not imply that one has reached that position by crossing x. In the second place, the syntagmatic restrictions on the trajector/landmark combinations discussed in section 3 point towards the application of [i] as the unmarked case within the structure of over. This is the only case, it will be remembered, that can be used with trajectors and landmarks of all types, and that therefore has the widest applicability. The restrictions on the other applications inevitably mark them off as special cases. At this point, another particular kind of restriction (one we have not discussed before) has to be pointed out in connection with the and cases. Examples like (82)-(85) show that a particular situation that does not involve motion can more easily be expressed as a fictive motion case than as a zero motion case, i.e., using a verb that is originally a motion verb enhances the grammaticality of the sentences. ?

(82)

Er bevindt zich een streep over het bord

'There is a line over the blackboard' (83) (84)

Er loopt een streep over het bord

;A line runs over the blackboard'

Er zijn condensatiestrepen over de lucht

'There are condensation trails over the sky' (85)

Er lopen condensatiestrepen over de lucht

'Condensation trails run over the sky' The phenomenon would have to be investigated more systematically; for instance, the applications of [ii] seem to be more susceptible to the restriction than those of [i]. Even so, the restriction as such seems to support the claim that cases are somehow more marked within the structure of over than cases. In the third place, the existence of alternative expressions for the same referential situation may point to the peripheral status of a particular kind of usage of over, specifically when the alternative expression is intuitively more natural than over in that particular context. The underlying idea is that the basic, central instances of over will also be those that are less like to have strong competitors for the expression of a particular concept. This is, in fact, a particular version of the idea that form and meaning have parallel characteristics: the whole idea boils down to the hypothesis that semasiological centrality (of instances with regard to a category) will be reflected by onomasiological dominance (of a form with regard to a referent). Now, if we take up again some of the examples from Table 1 (and some additional ones discussed in section 4), we can try to deternline whether over has strong competitors in those particular contexts. In examples (86)-(100), the alternative expressions are printed in italics at the end of each sentence. Brackets round the alternative prepositions indicate that they are less likely than over in that context. A plus-sign indicates that they are more natural. In general, the list supports the view arising from the previous criteria, viz. that the upper left corner of Table 1 represents the central set of instances of over. (86)

(87) (88)

(89)

De bal rolt over de lijn: (naar de andere kant van) Het vliegtuig vliegt over de stad: (boven) De deken hangt over de schutting: (van de ene kant naar de andere) Er hangen donkere wolken over de stad: boven

The semantic structure ofDutch over

67

(90)

Engelram kuiert over de markt: (op) Het vliegtuig cirkelt over de stad: boven Er figt een armoedig kleedje over de vloer: op Een dunne streep mist hangt over de rivier: boven (94) Over de rivier exerceerde de vijandelijke mifitie: +aan de andere kant van (95) Over de rivier loopt een rij bomen langs de beek: +aan de andere kant van (96) Dordrecht figt over de Moerdijkbrug: +aan de andere kant van (97) Over de brug hangt een ballon: +aan de andere kant van (98) Gosbert woont over het museum: Hegenover (99) De lamp hangt over de tafel: +boven (lOO) De ring figt over de kegel: rand

(91) (92) (93)

To summarize, we find that there are several indications for a prototypical structure of over pointing in the same direction. It should be mentioned, though, that neither of the criteria used above can be considered a foolproof test of prototypicality; their parallelism, however, does seem to warrant the conclusion we have drawn from them. It is obvious, at any rate, that additional research (of a basically psycholinguistic nature) could be invoked to complement the analysis given here.

6. A comparison with Cuyckens Having charted the range of application of over, let us try to compare our own treatment with the one presented by Cuyckens (1991). In Table 3, the various senses of over distinguished by Cuyckens (1991: 282) are charted within the same framework as the one presented in section 2 (supplemented with the three additional readings described in section 4). Figure 2 is a copy of Cuyckens's own overview of the polysemy of over. In this classification, over} represents the 'above and across' reading, assuming a process of real motion. Over2 refers to the related cases where the movement is conceptualized as being situated within the (vertically projected) spatial boundaries of the landmark. As can be seen in Figure 2, over} and over2 only refer to surface-like landmarks; over3, however, takes care of point-like landmarks in the case of a movement 'above and across'. Next to theses three readings referring to cases, the possibility of contact is recognized in over4, avers and over6. Judging from the graphic representations, 4 and 6 refer to cases where a surface-like or object-like landmark (respectively) is crossed. Overs, on the other hand, considers cases where the movement stays

68

Prototypicali~v

OVER1D

and salience

."

OVER 9

~

."

cb

I--

1



OVER 3 "

~.

.v



. V 7\ V ~. :



t--

"

r~ f--

... :

.

.

0

OVER 2

OVER 1

11

I1

-E)- 8 r--

OVERS

OVER 4

OVER 6 OVER 7

r~

~of'

... . )

OVCR8

x•r~• '.V

....

~

q,",,~ OVCR13

lOVER 12

fI] .

'

: :

'.'

...

.. : '." .

..

'

...

-

Figure 2. The analysis of over in Cuyckens (1991)

@ff]

The semantic structure ofDutch over

69

within the confines of the landmark, but specifically envisages the trajector to pursue a 'fairly rectilinear path' (1991: 280). The addition is not without importance: over]3 refers to those cases where the movement within a surface has no dominant direction, but rather takes a random form. In the analysis presented by Cuyckens, over13 ranges with the 'covering' -readings over11 and over12 rather than with over5. Over11 and over12 themselves involve surfaces (not necessarily of equal size) of which one covers the other. Overl2 specifically takes into account what we have called the 'discontinuous' cases in section 3. Because the graphic representation suggests that the trajector of over11 is at least the same size as the landmark, it should be noted that the examples given by Cuyckens also provide for trajectors that are smaller than the landmark. These have been included as over 11' in Table 3. Other additions are over] and over2', which refer to the counterparts of over] and over2. They are mentioned in the text, but are not counted as separate readings by Cuyckens. The remaining over's represent fairly straightforward cases. Over? is the 'at the other side of, relative to an observer' reading (in all possible combinations); over8 is the 'across from, facing' reading; and overQ and over]O correspond with the 'above' cases that require no particular extension of the trajector. (The distinction between 9 and 10 depends on whether the landmark is a surface or a point.) Table 3. Mapping Cuyckens' analysis onto the multidimensional structure of over MOTION

FICTlVE MOTION

ZERO MOTION

+contact -contact +contact -contact +contact -contact [i]

4.6

1,3

1'

11,12

---->

[vi]

[ii]

5,13

2

2'

11'

---->

[v] 9,10

7

7

---->

[iv] 8

[iii]

7

In comparison with the analysis presented by Cuyckens, there are a number of characteristics that are specific to the one conducted in the previous pages. First and foremost, the present analysis paints a more detailed picture of the conceptual relations between the various applications of over. As can be seen in Table 3, the over's that Cuyckens identifies consist of bundles of co-occurring features (such as a particular type of spatial relation coupled with a particular value on the 'motion/contact' dimension). But while the analysis presented here specifies and isolates those features, Cuyckens treats the bundles that constitute the 'senses' in his description as largely unana-

70

Proto~ypicalityand

salience

lyzable wholes. As can be seen in Figure 2, he does identify a number of larger groupings of related over's, but these groupings do not constitute a systematic exploration of the semantic relations between the various possibilities of use of over. Specifically, the fact that the relevant features determine highly overlapping sets of uses is not prominent in his analysis. (The groupings overlap in the sense that all subsets within the range of application of over that are defined on the basis of the 'motion' dimension cross-classify with the subsets defined on the basis of the 'contact' dimension, and with those defined on the basis of the spatial relations referred to by [i], [ii], and [iii].) Conversely, the analysis presented here precisely tries to describe the semantic structure of over as a cluster of subsets; rather than focusing on the individual readings, it focuses on the structure of the possible groupings of those individual readings. Thus, it not only systematically explores the relations between the various subsets into which the range of application of over can be divided, but it also yields an insight into the question what various kinds of features enter into the conceptual structure of over. Notice, in fact, that there are three major kinds of features involved: spatial relations as defined by [i], [ii], and [iii]; characteristics that pertain to the action, state, or event involving that spatial relation; and characteristics of the trajector and the landmark (and their combination). As a fourth type of a structurally relevant feature, the process of semantic generalization that lies at the basis of [iv], [v], and [vi] could be mentioned. At the same time, it has also become clear that there is much more systematicity in the structure of over than the picture presented by Cuyckens (see Figure 2) suggests: it will be sufficient to compare Table 3 and Figure 2 to appreciate the fact. The difference in focus between the analysis presented by Cuyckens and the one presented here can be captured (and summarized) terminologically by calling his outlook a synthetic one (because the emphasis lies on the individual applications of over as a synthesis of co-occurring features), whereas ours is analytic. Now, the analytic focus leads to two further differences with the synthetic one. First, whereas Cuyckens is rather hesitant to answer the question whether over has a prototype, an exploration of (among other things) the syntagmatic restrictions that the various readings of over are subject to in terms of trajector/landmark combinations unmistakenly supports the suggestion that over has a prototype structure (see section 5). Second, a systematic exploration of the range of application of over leads to greater comprehensiveness than Cuyckens has achieved. Apart from the obvious empty slots in Table 3, it is unclear why some features that are con-

The semantic structure ofDutch over

71

sidered criterial by Cuyckens in one part of the entire range, are not given the same treatment elsewhere. For instance, it is unclear why the distinction between and is considered criterial for distinguishing between over4 and overj, but does not lead to a similar distinction in the case of point-like trajectors (why doesn't Figure 2 contain a counterpart of over] ?).

7.

Methodological concluding remarks

The distinction between the 'synthetic' approach and the 'analytic' approach has to be reinterpreted on a more general methodological level. Prototype theory has (rightly) stressed the important role that the individual members of a category may play in the structure of that category. However, there is a danger that this kind of emphasis may lead to the tendency to restrict the analysis to the salient members of a category, thus underestimating the importance of a definitional analysis that not only tries to indicate the semantic relationships between the salient instances of a category, but that also tries to make sense of less central cases. The salient instances of a category take the form of bundles of co-occurring characteristics, but the foregoing analysis of over suggests that the true semantic structure of a category can only be revealed by systematically unraveling the features occurring in those bundles, their kinds, and their mutual relations. The tendency to concentrate on the salient instances of a category carries with it the danger of neglecting such an analytical approach. At this point, we naturally come back to a question that was raised in section I: how to determine what the readings of over are? Against the background of the analysis summarized in Figures 2 and 4, the same question can be restated: what should we consider to be the 'meanings' of over - higher level groupings such as [i], [ii], and [iii], or lower level applications such as those charted by Cuyckens? In order to make the structure of the problem more apparent, I may refer to a distinction that I introduced in Geeraerts (l987a). On the one hand, the question how to determine what the meanings of a lexical item are may be approached from a psychological angle, taking into account phenomena such as intuitive salience effects. On the other hand, the question may be tackled in a purely definitional, analytical vein; according to the traditional requirements, separate meanings will then be maximally general subsets of the extension of an item that can still be defined by means of necessary and sufficient characteristics 8 Now, the intuitive approach

72

Prototypicali(v and salience

seems to point specifically to the lower level applications of over as distinct meanings: the psychologically salient aspects of over (for instance, the semantic values that can easily be brought to mind by the native speaker) are synthetic entities, bundles of characteristics like the various over's discussed by Cuyckens. Conversely, an analytical, definitional approach automatically shifts the focus towards the more general groupings of the basic applications (as represented by [i], [ii], and [iii], and the like). Is there any way, then, to decide which approach (the intuitive or the definitional one) is most appropriate? I would say there is not. For one thing, as argued in Geeraerts (l993a), the existing criteria for distinguishing true meanings from mere cases of referential vagueness do not yield a clear-cut, unambiguous picture of what 'meanings' are. Rather, the question itself what the meanings of over are might well be suspect: it presupposes that meanings are like entities, stored as clearly separated objects in a word's contents, whereas the semantic flexibility that has been reintroduced into lexical semantics by prototype-theoretical studies could very well (when taken to its extreme) lead to the conclusion that 'meanings' are contextually flexible rather than stable and unaffected by context. Moreover, even if we would reach an agreement about what the concept meaning precisely covers, this would be basically a terminological decision. It would not detract from the fact that (at least from a cognitive perspective) both the psychological phenomena and the definitional analysis are indispensable aspects of any truly adequate study of the semantic structure oflexical items: up to a point, the lexical-semantic structure that Cognitive Semantics would like to describe, resides precisely in the interplay between both types of data. In this sense, my argument in favor of the 'analytic' approach to over does not imply that the 'synthetic' approach can be neglected; rather, the whole point is that both approaches have to be taken into account. To round off, let us note that the analytical type of prepositional description as represented by Figures 2 and 4 can be seen as an application and concretization of the ideas expressed by Leys (1988). Leys, in fact, has suggested that the basic conceptual structure of spatial prepositional relations takes the form [X S/M] R [Y], in which X is the trajector, Y the landmark, R the spatial relation as such, and S/M the stative or motional situation that determines the 'aspect' of the prepositional construction. It will be clear that what I have tried to do is precisely to specify the various values that R can take in the case of over by means of definitions like [i]-[iii], and to identify the various aspectual values of the construction by means of the distinction between , , and cases. Whether the

The semantic structure ofDutch over

73

attempt is as yet entirely successful is another matter, but I hope it has become sufficiently clear that taking the approach advocated by Leys seriously, enables us to make at least some progress in comparison with alternative treatments of over.

Notes 1.

2.

3. 4. 5. 6.

7.

8.

Actually, eruse's example features Charles rather than Balder. However, less commonplace proper names will be used throughout this article to draw the attention to the fact that Odo Leys, to whom this paper and this issue of Leuvense Bijdragen are dedicated, has made some major contributions to the study of proper names apart from his work on the central themes of this collection (and his manifold other research interests). Translations and paraphrases of the Dutch sentences are given in square brackets. Occasionally, round brackets within the square brackets contain explanatory glosses. A classification of the various kinds of fictive motion that could possibly be distinguished is beyond the scope of the present article. Given the system according to which the examples are construed, English translations are not always given. In fact, [iii] is possible as well, but it is irrelevant for the present discussion. The label 'line-like' should not be taken too narrowly in this description. The blanket in (57) could probably also be conceived of as a surface-like trajector; in that case, the requirement that the landmark should be smaller than a linelike trajector in the direction in which the latter extends would have to be modified to provide for oblong surface-like trajectors. It may be noted, however. that a 'covering' interpretation is not necessarily unavoidable in the case of multidimensional over-ness: when er ligt een kleedje over de tarel [a napkin lies over the table] is used to describe a situation in which the napkin is placed like a lozenge, leaving the four corners of the table free, it does not exactly cover the entire surface of the table, although it does lie over it in more than one dimension. (Of course, another way of dealing with this situation would be to accept a relaxed interpretation of 'covering'. so as to include almost complete coverage.) For more details, see the (1987) article, and the (1991) paper. in which the distinction between both types of polysemy criterion is further elaborated.

Chapter 4 Salience phenomena in the lexicon. A typology

Originally published in Liliana Albertazzi (ed.), 2000, Meaning and Cognition 125-136. Amsterdam/Philadelphia: John Benjamins. As argued in Chapter 1, the essence of prototype theory lies in the fact that it highlights the importance of flexibility (absence of clear demarcational boundaries) and salience (differences of structural weight) in the semantic structure of linguistic categories. But whereas prototypicality was initially described for the semasiological structure of lexical items, flexibility and salience may be observed in many other circumstances. In particular, as suggested in Chapter 2, it is crucial for cognitive studies to switch from a semasiological to an onomasiological perspective, i.e. from an analysis conducted on the basis of words to one conducted on the basis of concepts. The present chapter, then, explores the concept of onomasiological salience, and puts it in the context of a systematic analysis of salience phenomena in the lexicon. The article draws on (and partially repeats) a number of case studies included elsewhere in this volume, viz. the over study of Chapter 3 and the beer names study of Chapter 10. Also, like Chapter 6, it includes some materials taken from The Structure of Lexical Variation (Geeraerts, Grondelaers, and Bakema 1994). The detailed empirical investigation of lexical variation (both semasiological and onomasiological) carried through in that monograph constituted the first step in the development of an onomasiological line of research within our Leuven research group (again, see Grondelaers and Geeraerts 2003 for an overview, and compare Chapter 11).

1.

Salience as a lexicological variable

When one considers the question of the impact of Cognitive Semantics (in the sense of Langacker 1990a, Lakoff 1987, or Taylor 1989) on the field of lexicological research, it is probably correct to say that systematic attention to salience phenomena is one of the major innovations of the cognitive paradigm. As an inquiry into language as a system of categories, Cognitive Semantics has received major impetus from the prototype-theoretical insights into the structure of linguistic categories developed in psycholinguistics and anthropological linguistics. And prototype theory is basically concerned with

Salience phenomena in the lexicon

75

salience phenomena: it shows how the various semantic applications that exist within the boundaries of one particular category need not have the same structural weight within that category. The cognitive semantic attention to salience, however, has not yet been as systematic as it might have been. There are forms of salience phenomena that have not yet been investigated as thoroughly and completely as prototype-based salience. The purpose of this paper, therefore, is to provide a tentative overview of the various forms of salience effects that may be usefully distinguished. A typology of salience phenomena will be presented, based on a distinction between the following basic types: perspectival, semasiological, onomasiological, and structural salience. Although terminological hair-splitting and taxonomic overabundance are real dangers for an endeavor such as this one, there is one overriding reason why the present exercise may be useful: salience is the place where structure and use meet. Consider lexical semantics and prototype theory. Traditional semantic analysis is concerned with describing the various meanings of lexical items and the structural relations among those meanings. Katz and Fodor's bachelor example, for instance, describes the readings that 'bachelor' may possibly yield in actual text, and at the same time describes the conceptual relations among those readings, in the form of a hierarchical structure based on shared semantic features (Katz and Fodor 1963). Such analysis is basically a structured list of possibilities: it specifies the space of possible readings that language users may choose from in actual communication. The study of those communication processes, the actual choices made by the language users, is not envisaged. Prototype theory, on the other hand, does incorporate aspects of use in the structural description of the lexical categories: the fact that some readings are more salient than others in the structure of the category reflects the fact that they are more readily chosen when using that category. Admittedly, frequency of occurrence is not the only factor determining prototypicality, but next to conceptual coherence, it is certainly an important one. It should be clear that the combination of structure and use in the study of salience phenomena is essentially also a combination of semantics and pragmatics: semantics as the study of structure, and pragmatics as the study of use, combine when salience is seen as the structural reflection ofpragmatic phenomena. This interlocking combination of the semantic and the pragmatic perspectives implies a slightly different conception of the notion of 'structure' than is customary: whereas linguistic structures would tradi-

76

Prototypicality and salience

tionally be seen merely as ordered sets of possibilities, adding pragmaticsbased salience implies introducing probabilities rather than just possibilities.

2. Perspectival salience

The first step to take involves a definition of the concept of perspectival salience - a type that more or less stands apart from the other cases to be treated, and which will only be mentioned briefly here. Perspectival salience is represented in Cognitive Semantics by such terminological pairs as profile and base, or more generally, jigure and ground (for definitions and applications of these concepts, see Langacker 1990a or Talmy 1978, among many other publications). To mention just the simplest of examples, the lexical item hand evokes a particular body part, but it primarily does so against a particular conceptual background, namely, that of the arm. The arm is presupposed in the concept of the hand, but it is the hand itself that is highlighted against the backdrop of the arm. Crucially, this is a matter of perspectivization: a particular chunk of reality (the hand and the arm) is perspectivized in a particular way. The fundamental distinction between this perspectival type of lexical salience and the ones to be mentioned later, resides in the fact that the other types defined below invariably involve differences of salience among lexicological alternatives (where the alternatives may be meanings, syntagmatic contexts, lexical forms, lexical categories, or distinctive dimensions). In the perspectival type of salience, on the other hand, the salience phenomena involve just one of these lexicological alternatives (and specifically, one particular lexical meaning); within such a single meaning, particular aspects of the extra-linguistic situation referred to receive more weight than others. Perspectival salience, in short, involves differences of perspectival attention within one particular construal of reality, whereas the other type (which we will call variational salience) involves preference relations among different construals of reality. Perspectival salience is defined between the different aspects of the extra-linguistic situation (in the broadest possible sense of 'situation ') invoked by a linguistic expression: 'hand' invokes the entire arm as the background of the conceptualization, but it is only the terminal part of the arm that is saliently conceptualized by 'hand'. Variational salience, on the other hand, is defined between different linguistic expressions or properties of those expressions: the semantic property of 'hand' that it may mean 'the terminal part of a human arm' is more salient than the property that it

Salience phenomena in the lexicon

77

may also mean 'manual worker' (as in 'hired hand') - which is then the less prototypical reading. Confusion should be avoided in particular when the linguistic expressions to be compared from a variational perspective each embody a different perspectival construal of the same chunk of reality. Choosing to describe a commercial transaction by either the verb 'to buy' or the verb 'to sell' involves an onomasiological choice. If one of both terms is the more obvious alternative, there is a difference of variational salience between both verbs: we would then say, for instance, that 'to buy' is a more salient way of talking about a commercial transaction than 'to sell'. On the other hand, each of the two verbs separately attaches perspectival salience to different aspects of the commercial transaction scene. In the case of 'to buy', the person buying is more salient than the person selling, and the reverse is true in the case of 'to sell'. In examples such as these, comparing the differences in perspectival salience of two linguistic expressions is not the same thing as comparing the onomasiological salience of the linguistic expressions themselves, that is, the likelihood that one rather than the other will be used in situations where both are applicable.

3. Semasiological, onomasiological, and structural relations As a further step in the construction of the classification, a distinction must be drawn between semasiological salience and onomasiological salience. Now, the conceptual distinction between semasiology and onomasiology may require some background information, because it involves a terminological pair that has - unfortunately -not made its way into the standard set of linguistic terms to be found in introductory courses. It is a terminological distinction that was introduced at the beginning of this century, at the time of the birth of structural methods of semantic analysis. The tern1S were coined by the Swiss romanist Karl Vossler in 1919. Terminologically speaking, the distinction is still much alive in those branches of (continental) European lexicology that preserve the heritage of structuralist semantics (such as the Coseriu school in Germany). In the Anglo-Saxon world, however, the terminology is hardly known. Basically, the distinction involves two ways of perspectivizing the structuralist link between sign~fie and sign~fiant. This terminological pair itself, to be sure, is not without having problems of its own, but for present purposes it may be sufficient to equate sign~fiant with a word form, and sign~fie with the concepts associated with that word form.

78

Prototypicali(v and salience

A semasiological perspective takes the word as its starting-point and investigates how several concepts (meanings, if one likes) are associated with that word. This is, of course, the regular type of semantic analysis in which the meanings of a word are listed and the semantic relationship between them indicated. Onomasiological research, on the other hand, reverses the perspective, and takes the concepts expressed as its starting-point; it investigates, basically, which various words may express a given concept, and what the structural links between those words are. In actual historical practice, the onomasiological perspective has been realized primarily in the form of lexical field research. Bruno Quadri's bibliographical overview of onomasiological research, published in 1952, for instance, reads as an overview of various types of lexical field analysis (Quadri 1952). Kurt Baldinger (again a Swiss romanist) has succinctly described the difference in the following way: 'Semasiology [... ] considers the isolated word and the way its meanings are manifested, while Onomasiology looks at the designations of a particular concept, that is, at a multiplicity of expressions which form a whole.' (Baldinger 1980: 278). An even more succinct way of putting things would be to say that semasiology involves processes of meaning, whereas onomasiology involves naming. Baldinger's formulation, however, also makes clear that there is a certain ambiguity in the definition of onomasiology. On the one hand, when onomasiology is conceived of in tern1S of 'designation' (or naming), it involves the relationship between semantic entities and the lexical items that name them. If, on the other hand, onomasiology is conceived of in terms of 'a multiplicity of expressions which form a whole', it instead involves the mutual relations between related lexical items. In this conception, onomasiological research will tend to coincide with lexical field research as a purely structural endeavor - that is, as an approach in which the actual choices from among the set of alternatives present within the lexical field are barely envisaged, but in which most if not all of the attention is directed towards an analysis of the relations between those expressions. In actual practice, in fact, so-called onomasiological research has been basically restricted to lexical field research of the kind just described. It may be useful, therefore, to impose a terminological distinction between both kinds of onomasiological research, because it will help us to distinguish between the purely structural and the pragmaticized conception of onomasiology: if the term onomasiology is reserved for the pragmatic approach in which the relationship with actual referents is explicitly envisaged, the other approach may then be called structural.

Salience phenomena in the lexicon

c

79

c

~-ii

R

w w

R

Figure 1. Semiotic relations between the major groups of salience phenomena in

the lexicon The differences between the three perspectives defined in this way, may be presented schematically by starting from two semiotic triangles as in Figure 1. If we maintain a distinction between Word (or lexical item, W in the figure), concept (C in the figure), and referent (R in the figure), then semasiology investigates how a particular lexical item may express various concepts, or alternatively (putting things extensionally rather than intensionally), how it may designate various types of referents. This is expressed by the arrows in the left hand triangle. Onomasiology in the restricted sense defined a moment ago describes how referents or concepts may be designated by various expressions; this is indicated in the right hand triangle. A structural analysis, finally, studies the semantic relationship between expressions. In the figure, this is indicated by the arrow in between both triangles. The arrow actually connects the lines going from C to W rather than just both W's or both C's alone, because the structural analysis involves expressions as combinations of forms and meanings, rather than just word forms or semantic concepts alone. (The double directionality of the arrow between the triangles, therefore, does not indicate a one-to-one correspondence; it merely represents the fact that the structural analysis is a reciprocal one: it looks, so to speak, at one item from the perspective of the other, and vice versa.) Given the distinction between semasiology and onomasiology, and the double interpretation of onomasiology, concepts of semasiological salience, onomasiological salience, and structural salience may now be defined.

4.

Semasiological salience

Semasiological salience is a relationship among the various semantic possibilities of a given lexical item. A lexeme/morpheme may express various semantic values, as when in expresses containment ('the apple in the bowl ') or inclusion/surrounding ('the village in the desert'). Some of the values

80

Proto~ypicali~y and salience

expressed by the lexical element may be more central than others, for instance because they occur more frequently within the range of application of the lexical element, or because they are a semantic centre in a network or family resemblance structure from which the other values are extensions. For example, containment seems to be more central in the structure of in than inclusion (see Vandeloise 1994). Practically speaking, then, semasiological salience coincides with prototypicality. As prototypicality effects constitute the best known type of lexicological salience, it is not necessary to recapitulate the literature here (see Geeraerts 1989c for an introduction, and Taylor 1989 for an extensive overview.) Two things require clarification, though, in preparation for the following discussion of onomasiological salience. First, prototypicality may be defined both as a relationship between a lexical item and its meanings, and as a relationship between a lexical item in one of its meanings and the referents corresponding with that meaning (compare Kleiber 1990). In the former case, it is pointed out for instance that the 'biological species' sense of 'bird' is more common than the slang sense in which 'bird' refers to a girl. In the latter case, it may be pointed out that robins and sparrows are more central members of the 'biological species' sense than, for instance, ostriches and penguins. This distinction between the two possible definitions of prototypicality hinges crucially on the distinction between polysemy and vagueness: of all the multiple applications of a lexical item, some constitute true meaning differences (and hence involve polysemy), while others should merely be considered referential specifications (and hence involve vagueness rather than polysemy). It has recently been suggested (Taylor 1992, Geeraerts 1993a) that the distinction between polysemy and vagueness may well be less stable than is traditionally assumed; if this is correct, the distinction between two different levels of prototypicality (the semantic one and the referential one) marks the endpoints of a continuum, rather than a clear-cut dichotomy. Second, it may be useful to distinguish between the paradigmatic and the syntagmatic aspects of lexical meanings. To begin with, let us note that each of the values expressed by a relational lexical element may occur in a specific syntagmatic frame. For relational spatial concepts, Cognitive Semantics describes such a frame in terms of a 'trajector' and a 'landmark' (see Langacker 1990a). For instance, the landmark of a containment relation has to be a recipient with, say, 'containing potential', whereas the trajector has to be something 'containable' in relation to the landmark. The containment relation itself is a paradigmatic value, whereas the containing potential of the

Salience phenomena in the lexicon

81

landmark is a syntagmatic phenomenon. These restrictions on syntagmatic contexts resemble the selection restrictions of old, except that they now explicitly involve (groups of) referents and not just words. For a more extended example, we may turn to a word that more or less functions as a classic example in Cognitive Semantics, namely the preposition 'over' (see Lakoff 1987). In Geeraerts (l992a) it is argued that there are basically three spatial relations to be distinguished in the semantic structure of the Dutch preposition over. They are defined a, b, and c below and illustrated by the examples (1 )-(3) (lrn stands for 'landmark', tr for 'trajector'). The range of application of over is not restricted to these cases, amongst other things because the spatial relations may show up in situations involving real motion, in situations involving fictive motion, and in examples involving stative relations. (a)

(c)

Extending from a point or region at one side of Im to a point or region at the other side of Im, while crossing hn Extending from one point or region within the space defined by Im, to another point or region within that space (without going outside of it) Situated at the other side ofhn, relative to the position of an observer

(1)

De bal rolt over de lijn

(2)

De agent kuiert over het marktplein

(3)

Over de rivier deed de atleet zijn oefeningen

(b)

'The ball rolls over the line (to the other side of it)' 'The policeman strolls over the market-place' 'Across the river, the athlete was exercising' The relevance of a separate analysis of the syntagmatic restrictions on trajector and landmark becomes clear, for instance, when one compares the selectional restrictions holding for these three basic relations. Without going into all the details, let us briefly note that there are hardly any restrictions on the trajector and landmark of semantic value (a): both trajector and landmark may be one-dimensional (points), two-dimensional (lines), or threedimensional (objects), in any combination. In the case of (b), however, both trajector and landmark are subject to restrictions. On the one hand, because the spatial relation referred to by over must remain within the boundaries of the landmark, the prepositional object must have a line- or surface-like extension: it should be possible to conceive of it as a line or a surface. This shows up quite clearly when considering threedimensional objects. Whereas (4) and (5) are straightforward cases of lines and surfaces, (6) is an example of a three-dimensional object whose third dimension (height) does not prevent it from having a bounded surface that

82

Prototypicality and salience

allows for an over-relation. In (7), however, the three-dimensional object's vertical dimension is so important that no surface for aimless or random wandering can be imagined. As (8) shows, reading (a) of over does not impose this restriction. This does not mean, however, that an object such as schutting 'fence' could not occur at all with reading (b) of over. Example (9) shows that the relevant feature is not the objective horizontal or vertical orientation of the surface, but rather the functional characteristic of 'treadability': something that could hardly be a surface for free movement for human beings may be a natural path for insects. On the other hand, because the trajector of an over-relationship according to (b) must remain within the boundaries of the landmark, it will typically be smaller than the landmark - small enough, at least, to move about on the surface or along the line constituted by the landmark. Specifically in the case of line-like landmarks, however, this does not imply that the trajector has to be smaller than the landmark in all directions. Taking into account that real lines are never merely one-dimensional, but always have a certain width as their second dimension, the trajector does not have to be narrower (but it does have to be shorter). As example (l0) shows, the moving object may be wider than the actual line. (The train is wider than the rail.) (4)

Janjietste over het pad

(5)

Jeanne kuierde over de markt

(6)

Mark dwaalde over de berg

(7)

*Jvfarie dwaalde over de schutting

(8)

Karel klom over de schutting

(9)

De spin kroop maar door over de schutting

(10)

De trein glijdt over de monorail

'John rode his bicycle along the path' 'Jane strolled over the market place' 'Mark wandered over the mountain' 'Mary wandered over the fence' 'Carl climbed over the fence' 'The spider kept on crawling over the fence' 'The train glides over the monorail' Now, the syntagmatic frame of each semantic value may have typical and less typical instantiations. For instance, 'the apple is in the bowl' would probably refer to a typical containment relation, whereas 'the jelly pie is in the suitcase' is an untypical one (even though the containment relation itself is the same as in the typical case). Such typicality relations illustrate the concept of syntagmatic salience. In the over example, those instances of

Salience phenomena in the lexicon

83

reading (b) that more readily invoke a line-like or surface-like landmark will be syntagmatically more salient than the others. In the sample sentences, for instance, (4) and (5) are more typical than (6) or (9). Along the same lines, examples in which the trajector is not entirely contained within the boundaries of the landmark - as in sentence (10) - are syntagmatically less canonical.

5.

Onomasiological salience

Turning now to onomasiological salience, there is a possible misconception to be discarded first. From a very general point of view, salience on the onomasiological level involves a preference for one expression over another as the name for a particular referent or type of referent. The existence of such alternative names would therefore seem to coincide with the kind of formal variation that is customarily studied in sociolinguistics and dialectology. Consider Table 1, which specifies a number oflexical patterns from the clothing terms project described in Geeraerts, Grondelaers, and Bakema (1994). In each of the cases in Table 1, there is a different naming pattern in Belgian Dutch and in Netherlandic Dutch; the clothing types themselves, however, are the same. The choice between legging and caler;on as a designation for leggings, for instance, therefore acquires the same value as the distinction between, for instance, different pronunciations of the same phonological form in different geographical regions or social strata of a linguistic community. This type of onomasiological variation may be called 'formal' or 'synonymy-based', but it is useful to point out that these terms are not ideal in all respects. Terms like legging and caler;on are denotationally synonymous, but they are not absolute synonyms, because they have a different geographical distribution. The two terms have the same meaning in a narrow, denotational sense of the word, but they clearly have a different sociolinguistic value (in a broad sense of 'sociolinguistic '). In a similar way, the term 'formal' variation (to which I shall adhere in the rest of the paper) might suggest that the variation in question is trivial or irrelevant, in the sense that it could be taken to involve mere 'notional variants'. Again, however, the alternation between legging and caler;on is not a question of free variation, but signals a specific stratification of the linguistic community.

84

Proto(vpicali(y and salience

Table J. Differences of naming patterns between Netherlandic and Belgian Dutch ITEMS

BELGIAN SOURCES

NETHERLANDIC SOURCES

calec;on legging leggings

40 (38%) 26 (24.7%) 39 (37.3%)

91 (100%)

blouson jack

13 (46.4%) 15 (53.6%)

85 (100%)

jeans spi}kerbroek

64 (97%) 2 (3%)

38 (70.4%) 16 (29.6%)

The misconception, then, consists of restricting onomasiological variation to this type of 'formal' variation. Next to the synonymy-based variation meant here, there are, obviously, cases where the choice of an alternative name carries a semantic load. Consider the following distinction between Belgian Dutch and Netherlandic Dutch. (11)

Hi} kwam langs de deur naar binnen

(12)

Hi} kwam door de deur naar binnen

'He came in through the door' (Belgian Dutch) 'He came in through the door' (Netherlandic Dutch) In Belgian Dutch, the usual preposition is langs 'via', where the semantic focus is on the path as such. In Netherlandic Dutch, the usual preposition is door 'through', which adds the notion of a passage that is penetrated and crossed. Although the preference for one or the other construction is conventional, it is not conceptually neutral: there is a different perspective involved in the choice of one over the other. This implies that next to the formal salience of synonyms that is at stake in the leggings/cale90n example, we have to distinguish a form of onomasiological variation that is semantic or conceptual in kind: it involves the choice of one semantic category (and not just one lexical item) rather than the other for talking about a particular realworld situation (such as coming through the door). In contrast with the 'formal' kind of onomasiological variation, we will call this type the' categorial' type of onomasiological variation, because it involves the choice of a different category. To be sure, the existence of such a preferred categorial choice is not an entirely novel idea, as it is part of the foundation of the basic level model of taxonomical structure formulated by Berlin and colleagues (see Berlin 1978). There are, however, various points with regard to which the basic

Salience phenomena in the lexicon

85

level model of onomasiological salience may be refined. At least the following two points may be considered. To begin with, the basic level model attaches onomasiological salience to hierarchical levels in the taxonomical tree, rather than to individual words. However, it may be argued (Geeraerts, Grondelaers, and Bakema 1994) that the notion of onomasiological salience should not be restricted to levels, but should be extended - generalized, if one wishes - to individual categories. It may not be necessary to repeat all the arguments here, but Table 2 gives a sample of relevant observations. Table 2. Differences in onomasiological salience among co-hyponyms ITEM

ONOMASIOLOGICAL SALIENCE

broek short-shorts bermuda legging-Ieggings-caler;on jeans-jeansbroek-spijkerbroek

46.47 45.61 50.88 45.50 81.66

t-shirt blouse-bloeze-bloes overhemd topje shirt hemd overhemdblouse

70.61 61.52 31.45 29.62 29.06 22.31 12.74

The data in the figure are based on the following operational definition of onomasiological salience: the onomasiological salience of a lexical category is the frequency of the lexical element naming the category divided by the cumulative token frequency in the database of the semantic values expressed by that lexical element. The rationale behind this definition is as follows. For each semantic value (concept, or type of referents) expressed by a lexical element. there may exist alternative terms (co-hyponyms, hyponyms, hyperonyms). A lexical category is onomasiologically highly salient if it is a likely choice for the semantic values it expresses, that is, if it is stronger than the alternatives. Thus, given a corpus of language use, the onomasiological salience of an item like 'skirt' can be calculated by counting how many times skirts (i.e. the potential referents of 'skirt') are named in the corpus, and then checking how many times these are actually referred to with the lexeme

86

Prototypicality and salience

'skirt', rather than alternative ones. Such alternative terms could be hyponyms like 'miniskirt' or 'wrap-around skirt', or hyperonyms like 'garment' (although the choice of the latter term seems quite unlikely in the case of 'skirt'). The definition of onomasiological salience also takes cases of synonymy into account, but this aspect of the matter need not be spelled out here. The point about Table 2, then, is simply this: the onomasiological salience of different categories on the same taxonomicallevel may differ considerably. For instance, in the upper part of the figure, 'short', 'bermuda', 'legging', and 'jeans' are co-hyponyms, as they all fall under the hyperonymous category broek 'two-legged garment covering the lower part of the body, pants'. However, the onomasiological salience of the different categories differs considerably: that of the jeans-category, for instance, doubles that of the legging-category. This means that a potential member of the category jeans is twice as likely to be designated by an expression that names the category jeans than a member of the legging-category would be likely to be designated by an expression that names the category legging. The second point about the conceptual kind of onomasiological salience involves the recognition that the referential starting-point for the determination of onomasiological salience values is subject to variation. The salience figures given so far represent the likelihood that one specific category, such as legging or t-shirt will be chosen as the name for a particular referent, in those cases where that referent is a potential member of the category. The calculation, to be precise, is made for the category as a whole, that is, for the overall set of referents of the category. However, the calculation may also take its starting-point in a particular subset of the entire range of application of a lexical category. The question asked so far boils down to questions of the type: what is the probability for the set of leggings as a whole that its members will be named with an expression that precisely identifies the category 'legging'? The modified question now reads: given a specific subset of the set of leggings as a whole (for instance, leggings that do not reach down to the ankles but are not longer than the calves), what is the likelihood that it will be named with the item legging or one of its synonyms? Obviously, the answer to the question will be determined by the overall salience of the category legging as calculated a moment ago, but it may also depend on the structural position of the subset under investigation within the semasiological structure of legging. The 'local' onomasiological salience introduced here may be called the 'cue validity' of a particular subset of the range of application of an item: it indicates how good a cue that subset is for

Salience phenomena in the lexicon

87

triggering the name of the category. The suggestion, then, is clearly that this onomasiological cue validity and the semasiological properties of the subset exhibit a positive correlation. Put more simply: an expression will be used more often for naming a particular referent when that referent is a member of the prototypical core of that expression's range of application. When, for instance, a particular referent belongs to the core of item x but to the periphery of y, it is to be expected that x will be a more likely name for that referent than y. (At the same time, of course, the choice for x will be a function of the global salience of the category represented by x, but that is a factor we have already dealt with). A brief and rather impressionistic example may suffice to illustrate the idea (for more detailed, quantitatively elaborated examples, see again Geeraerts, Grondelaers, and Bakema 1994). The kinds of usage that are structurally peripheral in the semasiological structure of Dutch over are more readily replaced by alternative ternlS than the central cases. For instance, in the examples illustrating reading (c) of over, the expression aan de andere kant van 'to the other side of is more natural. In the same way, there is an extension of (c), defined and illustrated below sub (d), for which the alternative expression tegenover 'across, facing' is by far the more natural one. (d)

Situated at the end of a virtual path that extends from lm in a straight line across an open space

(13)

]v/artha woont over het museum

(14)

Aan tarel zat Marianne over Ariane

(15)

Het museum bevindt zich over de rechtbank

(16)

In de erehaag stand Samuel over David

'Martha lives across the museum' .At the table, Marianne sat facing Ariane' 'The museum is situated across the magistrates' court' 'In the double row of honor, Samuel stood facing David' The notion of local onomasiological salience may be subject to two different interpretations, involving the same distinction between polysemy and vagueness that was discussed above in cOlmection with the concept of prototypicality. The referential subset from the range of application of a lexical item that is taken as the starting-point for a calculation of local onomasiological salience mayor may not correspond to a meaning, in the traditional sense, of that item. There is no reason to restrict the investigation of onomasiological salience to subsets that actually correspond with 'meanings'. This is not only because (as mentioned before) the distinction between meanings (underlying polysemy) and mere referential subsets (representing vagueness)

88

PrototypicaliZv and salience

may well be unstable, but also because prototype-theoretical research has established beyond doubt that even referents that do not constitute a 'meaning' in the strict sense (for instance, central members of the category), may play an important role in the semantic structure of a lexical item. The meanings whose local onomasiological salience is calculated need not necessarily constitute denotational meanings: any lexically expressed semantic concept (whether it be of a denotational, emotive, stylistic, or discursive kind) can be subjected to the approach.

6. Structural salience We may now turn to the third major perspective distinguished above. Remember that a purely structural conception of onomasiological lexical relations involves the mutual relationship between lexical categories. Salience effects along this perspective, then, will have something to do with the weight of the distinctive dimensions that distinguish various categories from each other. A semantic feature or a semantic dimension will be structurally salient when it often occurs in the structure of the lexicon; the structure itself, in fact, coincides with the totality of distinctive relations in the lexicon. For an example, let us consider the lexical field of names for beer current in Belgium (see Geeraerts 1999a). The methodology can be summarized as follows: the more a particular feature is expressed in the names of beers, the more it is cognitively salient. When, for instance, the presence of fruit flavors never surfaces in the names of the beers that have such a flavor, it is unlikely that fruit flavor is a preponderant feature of beers. Of course, we have to be more precise about what it means for a distinctive feature or a distinctive dimension 'to be expressed in the names of beers'. In most cases, there is no problem: when you find a reference to kriek in a name, and when you know that kriek is the name for a type of cherry, there is no difficulty in concluding that the addition of cherry juice is mentioned in the name of the beer. In other cases, the reference is more indirect. A striking example is the following: in the French speaking part of Belgium, a peach-flavored beer is produced with the name La pecheresse, which translates as 'the female sinner, the sinning woman'. When you realize, however, that peche is the French for 'peach', it will be clear that the name contains a pun with an indirect reference to the presence of peach flavor. In general, a feature is expressed by a name (or part of a name) when that name (or the relevant part of it) only occurs in connection with that feature.

Salience phenomena in the lexicon

89

Some more examples may further illustrate the point. The fact that a particular beer is low on alcohol may be expressed directly by expressions like alcoholarm 'low on alcohol' or alcoholvrij 'without alcohol'. Indirectly, however, the item talelbier 'table beer' also signals the light character of beer, to the extent that it only occurs within the group of very light beers. Similarly, lambiek and geuze can be taken as expressions of the 'spontaneous fermentation' character of beers, to the extent that they only occur in names for beers with that characteristic. Table 3. Salience characteristics of sample referential features in the field of Belgian beer names REFERENTIAL FEATURE

REFERENTIAL FREQUENCY

CHARACTERISTIC NAMES

wheat (instead of rye) very light low alcohol seasonal beers added herbs raspberry flavor cherry flavor trappist beer

139 110 86 59 40 19 16

91 (65.5%) 25 (22.7%) 86 (100%) 4 (6.8%) 38 (95%) 18 (94.7%) 16 (100%)

Following these guidelines, we can show that not all features of beers are equally important. There are clear differences in the extent to which specific characteristics are expressed in the names. Table 3 presents a sample to demonstrate the kind of differences in salience that occur. The table should be read as follows. The first column of figures indicates the number of beer brands in the data set that has the feature mentioned to the left. The rightmost column indicates how many of those beers actually carry a name that refers directly or indirectly to the feature in question. The relationship among the figures mentioned to the right gives an indication of which features carry distinctive weight. Trappist origin, for instance, seems to be highly valued, while the addition of herbs is not an especially individuating feature. In the spatial realm, the featural salience that is at stake here would involve, for instance, the distinctive weight of dimensions like shape, or movement, or geometry, or the role of the observer across items (or even across languages). (See Taylor 1988 for an overview of potentially relevant dimensions).

Prototypicali~y and salience

90

7. A typology of lexicological salience

In the following overview, each of the seven major cases of lexicological salience as presented before is defined. Also, a shorter name is suggested for each of the various types; of course, these names are just suggestions that should be treated with care as long as there is no uniform lexicological terminology. 1. Perspectival salience: highlighting = the differences of perspectival attention attached to different parts of the overall chunk of extralinguistic reality evoked by a particular concept. 2. Paradigmatic semasiological salience: proto~ypicality = the preponderant structural weight of specific senses or members within the semasiological range of application of a lexical category 3. Syntagmatic semasiological salience: canonicality = the dominant occurrence of specific selection-restrictional frames within the syntagmatic range of a paradigmatic reading of a lexical category 4. Formal onomasiological salience: sociolinguistic prevalence = the preference for a lexical item, in comparison with its synonyms, in a specific language variety or a specific pragmatic context 5. Global categorial onomasiological salience: entrenchment = the preference for a specific lexical category as a designation for its range of application, taken as a whole 6. Local categorial onomasiological salience: cue validity = the preference for a specific lexical category as a designation for a subset of its range of application 7. Structural salience: distinctiveness the preponderant recurrence of a semantic dimension or a semantic feature as a distinctive dimension or feature within the semantic structure of the lexicon

=

By way of summary, the seven types of lexicological salience are brought together in their mutual taxonomical relations in Figure 2. At this point, we may try to chart the seven types against the background of the semiotic triangle used in Section 3. However, the inadequacy of the semiotic triangle for dealing with more complex linguistic relations is wellknown; see Hofinann (1993) for a brief history of the matter, and compare Geeraerts (1983b).

Salience phenomena in the lexicon

91

lexicological salience variational salience

perspectival salience semasiological salience

~

paradigmatic

syntagmatic

structural salience

onomasiological salience

~

formal

categorial

global

~ local

Figure 2. Taxonomical relations among the various types of lexicological salience

An attempt (but certainly no more than attempt) to graphically represent the six variational types of salience may be based on Figure 3. In Figure 3, three levels of analysis are distinguished: the level of words W, the conceptual level of meanings C of those words, and the level of the referents R corresponding to those meanings. The word W2 is polysemous; Wl and W2 are synonymous with regard to meaning Cl. The referential range of words may overlap in yet another case besides synonymy: R4 may be conceptualized either as C3 or as C4 (as when a short wrap-around skirt might be called either a 'miniskirt' or a 'wrap-around skirt'). The referential range of CS is represented in a special way to highlight the syntagmatic contexts S in which the referents of CS occur. The horizontal arrows represent the features that distinguish between the concepts C. Salience effects may now be represented by using bolder lines; the differences between onomasiological and semasiological salience are expressed by having bottom-up and topdown arrows respectively. In Figures 7 to 12, the six types of variational salience are represented graphically against the background of Figure 3. (The parts of the figure that are irrelevant are represented by dotted lines.) Figure 4 indicates that C2 (rather than Cl or C3) is the prototypical meaning of W2, and that R3 (rather than R2 or R4) is the prototypical type of referent of C2. Figure S shows that R-Sl is the canonical syntagmatic context for CS. Figure 6

92

Prototypicality and salience

represents a sociostylistic context in which W2 is the prevalent synonym for Cl. Figure 7 indicates that W4 is a more highly entrenched category than W 1, in the sense that W4 is more likely to be used (in those cases where it can in principle be used) than Wl is likely to be used (in those cases where it is a relevant category). Figure 7 does not contain cases in which Wl and W4 overlap, but obviously a comparison of entrenchment values could just as well be applied to such a situation. Figure 8 expresses that R5 has a higher cue validity with regard to C4 than with regard to C3, that is to say, W3 is more likely to be used to categorize R5 than W2. Figure 9 indicates that the dimensions or features that distinguish C3 from C4 are structurally more important than the others (for instance, because they recur elsewhere in the lexicon).

Wl

W3

W2

Rl

I

I

I Cl

W4

_----_~

R2

C2

R3

.4--_~

R4

C3 _ _ C4 _

C5

V/\ R-Sl

R5

R-S2

Figure 3. Sample network for representing salience phenomena

Wl

W2

Cl

C2

Rl

R2

R3

C3

R4

R5

W3

W4

C4

C5

R-Sl

R-S2

Figure 4. Schematic representation of paradigmatic semasiological salience

Salience phenomena in the lexicon

Wl

W2

Cl

C2

Rl

R2

R3

C3

R4

W3

W4

C4

C5

93

1\

R5

R-S2

R-Sl

Figure 5. Schematic representation of syntagmatic semasiological salience

Wl

W2

W3

W4

C4

C5

1________ C2

Cl

Rl

R2

R3

C3

R4

R5

R-Sl

R-S2

Figure 6. Schematic representation of formal onomasiological salience

Wl

W2

W3

W4

C4

C5

i

I.······· Cl

................

C2

. .... C3

...

1

Rl

R2

R3

R4

R5

1\

R-Sl

R-S2

Figure 7. Schematic representation of global categorial onomasiological salience

94

Prototypicality and salience

Wl

W3

W2

.~ Cl

Rl

C2

R2

R3

,

...........

R4

W4

t

C4

C3

\/ R5

C5

R-S2

R-Sl

Figure 8. Schematic representation of local categorial onomasiological salience

Wl

Cl

Rl

W3

W2

W4

_ - - - -__ C2 . 4 - -__ C3 ++C4 _ _ C5

R2

R3

R4

R5

R-Sl

R-S2

Figure 9. Schematic representation of structural salience

8. Salience and snow The classification presented in the previous pages is undoubtedly only a first step towards an exhaustive and systematic treatment of lexicological salience. This first step was taken for the purpose of clarification, but it may be interesting to round off by adopting a slightly different perspective: could the classification throw new light on other linguistic phenomena? As one example, let us consider the question of conceptual universals. If semantic universals are likely to be statistical universals rather than absolute ones, the notion of semantic salience will be crucial in establishing the statistical universality of a particular phenomenon: semantic phenomena that are cognitively basic are likely to be salient across genetically unrelated languages. Questions of cultural specificity (the logical counterpart of questions of universality) likewise involve salience phenomena. Consider, to address one perennial member of the lexicological stock, the alleged existence of various words for snow in Eskimo. Let us suppose for a moment that there

Salience phenomena in the lexicon

95

are no empirical problems with Boas's statement (1911) that the Eskimo have four words for snow, whereas English and Aztec have only one (but see Martin 1986). Boas mentions the following four: aput 'snow on the ground' gana 'falling snow' piqsirpoq 'scurrying snow' qimuqsuq 'snowstorm, blizzard'.

It has been repeatedly pointed out that this does not imply that people camlOt notice differences in snow merely because they speak English. Clark and Clark (1977: 555), for instance, draw attention to the fact that Englishspeaking expert skiers may have specific words for snow such as 'powder', 'corn', and 'ice'. More generally, there is always the possibility of ad hoc compounds and word groups: 'snow glue', 'drizzle snow', 'icicle flakes', or the translations of the Eskimo words given above ('falling snow' etc.). The possibilities of expression of a language are not restricted to the vocabulary, and neither are the possibilities of conceptualization of the speakers of the language restricted to what is included in the mental lexicon. This attenuated point of view, however, does not imply that the cultural differences between the Eskimo and the Aztec disappear altogether. The cognitive salience of a concept like 'falling snow' may well be higher for the Eskimo than for the Aztec: it is plausible that they will more easily and more readily conceptualize a particular white substance as 'falling snow' rather than just 'snow'. In the terms of the classification presented in this paper, this expectation can be translated into the hypothesis that the entrenchment (the global categorial onomasiological salience) of 'falling snow' will be higher for the Eskimo than for the Aztec. The relationship between salience and universals may be specified from yet another point of view. In her work on semantic universals, Wierzbicka (1992) demonstrates that such abstract concepts as emotions and moral notions may be of a culturally specific nature~ at the same time, she insists on the possibility of expressing such concepts in any given language by means of a language-independent metalanguage: no matter how culture-specific a term might be, it can be translated in the 'alphabet of human thoughts' that she has been researching for decades. The foregoing pages, then, suggest one way of looking for such an 'alphabet of human thought' that might usefully supplement Wierzbicka' s analytic approach: if these concepts really have the universal status attributed to them, they should in one form or another have a high degree of lexicological salience in a large number of natural languages.

96

Proto~ypicali~v and

salience

The universality question also invokes the problem of pre-linguistic forms of salience, namely the objective correlates of the linguistic phenomena. To what extent is linguistic categorization determined by the structure of the world itself, or at least by the structure of our perceptual apparatus? Ifnaming is often grounded in an act of perception, how do the independent properties of such perceptual acts influence the linguistic phenomena? The typology presented here has steered clear of these perceptual forms of salience, but it certainly does not mean to exclude the possibility of a pre-linguistic level of salience. Rather, a cognitive approach to the study of language will inevitably have to consider the possible impact of such a pre-linguistic type of salience. Tackling that question, however, is beyond the scope of this article: before one starts looking for explanations of the linguistic phenomena, one has to be clear about those phenomena themselves.

Section 2 Polysemy

Chapter 5 Vagueness's puzzles, polysemy's vagaries

Originally published in Cognitive Linguistics 1993,4: 223-272. Polysemy is the basic phenomenon in lexical meaning structures, and arguably the basic contribution of Cognitive Linguistics to lexicology is the renewed attention for the structure of polysemy that it generated. But how exactly should polysemy be defined? What does it mean for a reading of a word to be a different meaning? In this paper, I argue that the distinction between vagueness and polysemy (which is crucial to traditional conceptions oflexical-semantic structure) is unstable: what appear to be distinct meanings from one point of view turn out to be instances of vagueness from another. In the present chapter, an argumentation for this thesis is developed in three steps. First, three basic kinds of polysemy criteria are distinguished: a logical, a linguistic, and a definitional test. Second, it is shown that these three kinds of tests may yield mutually contradictory results. Third, it is argued that such inconsistencies also occur within each criterion, in the sense that there are contextual variations in the polysemy results that they yield. As a consequence, Cognitive Linguistics will have to start from a highly flexible conception of meaning. In the period after the publication of the original article, polysemy research has moved in various directions relevant to the issues addressed in this chapter. First, Cognitive Linguists have explored ways for incorporating the flexibility and instability of meaning into the models of semantic structure that they work with. Levels of schematicity in the Langackerian sense (i.e. the idea that the meaning of a linguistic unit can be described as a schematic network in which more specific and more abstract levels of description co-occur) play a crucial role in such attempts: see Tuggy (1993), Dunbar (2001), Taylor (2003), Zlatev (2003). Second, attempts have been made to link polysemy research in Cognitive Linguistics with psycholinguistic studies of polysemy (Sandra and Rice 1995), but this has somewhat surprisingly, given the very cognitive orientation of Cognitive Linguistics - not developed into a dominant trend; see the discussion between Croft (1998), Sandra (1998), and Tuggy (1999). Third, the interest in polysemy issues has spread outside the field of Cognitive Linguistics: see Ravin and Leacock (2000), Peeters (2000) and Nerlich, Todd, Hennan and Clarke (2003) for samples. In particular, a major new trend in polysemy research has arisen in the domain of computational linguistics, in the form of so-called Word Sense Disambiguation algorithms: computational and statistical analyses of large corpora that take the

100

Polysemy

co-occurrence behavior of lexical items as indications of their semantic range (see Stevenson 2001, Agirre and Edmonds 2006 for overviews). Together with detailed corpus-based studies as will be illustrated in the next chapter, psycholinguistic and computational studies of the type just mentioned will, I think, be of crucial importance for the further development of polysemy research in Cognitive Linguistics: the theoretical issues have been greatly clarified, but there is still plenty of room for methodological progress.

1.

Prototypicality and polysemy

In my paper 'On necessary and sufficient conditions' (1988a), I pointed out that there is an intimate connection between the proper evaluation of the novelty of the prototype-oriented movement in lexical semantics, and the long-standing issue how to determine whether a word has one or more senses. In fact, while the revolutionary nature of prototype theory is often said to reside in the fact that it rejects the classical view that lexical categories can be defined by means of a sufficient set of necessary conditions for category membership, the possibility that some lexical items cannot receive a definition in terms of necessary-and-sufficient attributes is not exactly absent from the classical tradition. Rather, it is traditionally taken as an indication of the polysemy of the item in question. Therefore, if prototype theory is to maintain its revolutionary claims, it will have to redefine them, or at least make clear that the absence of necessary and sufficient definitions that it reveals does not simply signal good old-fashioned polysemy. In his recent monograph La semantique du prototype (1990), Georges Kleiber has taken up this suggestion, incorporating it into a discussion of the various trends that exist within linguistic prototype theory. However, as I already briefly noted in my review of Kleiber's thought-provoking book (1992), his treatment of the consequences of polysemy for prototypical theorizing seems to assume that the concept of polysemy is methodologically unproblematic; at least, his book does not deal explicitly with the question how to identifY the polysemous readings of a lexical item. The present paper is devoted precisely to that issue. It will compare various criteria for distinguishing between vagueness and polysemy, and explore the consequences of this comparison both for Kleiber's views and for prototype-oriented lexical semantics at large. In particular, it will point out two sets of problems for the usual polysemy criteria: first, the various types of criterion will appear to be mutually inconsistent, and second, any single criterion yields internally inconsistent results. That is to say, I will try to show that the distinction

Vagueness's puzzles, polysemy's vagaries

101

between vagueness and polysemy is not stable, in the sense that what ap-

pears to be a distinct meaning in one context is reduced to a mere case of vagueness in another. The paper is organized as follows. First, the importance of the polysemy issue for prototype-oriented theories is discussed in more detail. Second, a classification of types of criteria for distinguishing between vagueness and polysemy is presented. As a third step, the problems with the criteria will be presented, first by comparing the criteria among each other, and second by discussing each criterion separately. In contradistinction with existing treatments (such as Zwicky and Sadock 1975 and Cruse 1982), the emphasis will not just lie on the evaluation of each criterion in isolation, but also on their mutual incompatibilities; another point of contrast with existing work is that the distinction between vagueness and polysemy will not simply be assumed, but will itself turn out to be somewhat doubtful. After a complementary discussion of some additional problems, the consequences of these findings will be explored; it will be argued that the instability of the distinction between vagueness and polysemy not only suggests a reorientation of our conception of lexical meaning, but also raises important questions about the objectivity of the methodology of lexical semantics. Although the paper will throughout assume some familiarity with the concepts of linguistic prototype theory (such as they may be found in Taylor 1989), the paper will proceed at a relatively slow pace. The reduced speed is, to be sure, not a sign of leisure, but rather reflects the difficult and often uncharted character of the territory that we will be venturing into.

2.

Intercategorial and intracategorial prototypicality

In the course of the development of prototype studies in linguistics, various definitions of prototypicality have been suggested. Apart from the absence of necessary and sufficient definitions, prototypicality has been said to involve (among others) categorization on the basis of the salient members of a category, vague category boundaries, and a radial set relationship among the various applications of a category. (Compare Lakoff 1987 and Geeraerts 1989c for attempts to chart the different kinds of prototypicality.) It is now generally accepted that these characteristics need not coincide; they rather represent prototypicality effects that a single category may exhibit in various degrees, and that may arise from various sources. Precisely because the various prototypicality effects need not be co-extensive, some of them might

102

Polysemy

be typical of single meanings (the separate senses of a lexical item), whereas others could be typical of the meaning of lexical items as a whole (i.e. of structured sets of senses). Specifically, the difficulty noted in the introductory paragraph may be resolved by making clear that the absence of definitions characterized by necessity-cum-sufficiency is in fact a true novelty when it applies to the single senses of a lexical item (whereas it would only be a traditional indication of polysemy when it applies to the lexical item as whole). For instance, if it turns out to be impossible to define bird by means of a single sufficient set of necessary attributes, this is hardly surprising if the biological sense of bird is taken together with a metaphorical reading as in A silver bird took the happy couple from Chicago to Santa Cruz. However, if the biological sense taken by itself (given that it is a single, separate sense of bird) cannot be defined in the traditional way, this is a result contradicting traditional beliefs about the definability of word senses. Kleiber (1990) has pushed this kind of reasoning one step further by bringing other prototypicality effects into the picture. If we distinguish between the lexical level (consisting of lexical items as forms, i.e. of signifiants), the conceptual level (consisting of cognitive categories that are at the same time the senses oflexical items), and the referential level (consisting of real- or possible-world entities), various types of prototypicality effects can be contrasted. On the one hand (considering the relation between the lexical and the conceptual level), there is prototypicality in the radial sets or family resemblance sense, defined as a relationship among the various senses of a lexical item (each of those senses being a category on the conceptual level). On the other hand (considering the relation between the conceptual and the referential level), there is prototypicality in the original sense of Rosch's psycholinguistic research, involving typicality ratings and degrees of membership (and also, in the sense of an absence of classical definitions). Here, the internal structure of one of the distinct meanings of a lexical item is involved (like the biological sense of bird, in contrast with a metaphorical reading as when an airplane is called a silver bird). Prototypicality is, then, either a model for the polysemy of lexical items (involving the relationship between the lexical and the conceptual level), or a model of categorization (involving the relationship between the conceptual and the referential level). In the latter case, the prototypicality effects are intracategorial; in the former, they are intercategorial. I This model of prototypicality (which is as attractive as it is simple) potentially considerably diminishes the importance of linguistic prototype theory. On the one hand, it could be argued that the 'categorization' relation-

Vagueness's puzzles, polysemy's vagaries

103

ship between the conceptual and the referential level is the proper domain of psychology rather than linguistics; linguistics is primarily concerned (in the structuralist line of thought) with the relationship between signifi£s and signifiants, rather than with the relationship between language and the world. On the other hand, the importance of the 'polysemy' relationship between the lexical and the conceptual level is only a relative novelty in the history of linguistics; both the distinction between central and peripheral meanings, and the idea that those meanings are related by motivated links (like metaphor and metonymy) are far from new in the history of lexical semantics (see Geeraerts 1988a). However, the second part of this interpretation is hardly disconcerting. Even if prototype theory as a model of the polysemy of lexical items is a rediscovery of older concerns rather than a complete novelty, it is important as a rediscovery. In the article that I just referred to, I have argued that prototype theory links up primarily with prestructuralist historical semantics. Granting, then, that the structuralist approach in lexical semantics has (in its mainstream development) downplayed the importance of polysemy, the revivification of the polysemic concerns of prestructuralist semantics is an important step towards a more adequate conception of lexical-semantic structure. Also, there can be no doubt that prototype theory is more than a rediscovery of older ideas; the importance of typical individual members of a category for the structure of that category is, for instance, hardly to be found in the prestructuralist conceptions of lexical structure. The other point is potentially more damaging. If the' categorization' relationship as defined above does not belong to the proper domain of linguistics, the claim of Cognitive Linguistics that it is a full-fledged theory of categorization will be undermined, because it would have to leave that aspect of cognitive structure in which the relationship between the conceptual level and the referential level is studied, to psychology. That is to say, either Cognitive Linguistics would indeed be studying the 'categorization' relationship together with the 'polysemy' relationship (but then it would be trespassing on psychological territory), or it would forsake the study of 'categorization' (but then its cognitive importance would largely evaporate, because the relationship between the referential level and the conceptual level is the basic one in the formation of cognitive categories). This argumentation rests, among other things, on the debatable assumption that existing disciplinary demarcations should be maintained. The fact that linguistics and psychology now exist as separate disciplines should not preclude the formation of an interdisciplinary cognitive science in which

I04

Po~ysemy

intercategorial and intracategorial prototypicality effects are studied together. But then again, independent arguments should be adduced for explaining why they should be studied together. It may well be interdisciplinarily legitimate in principle to study them together, but what is the positive motivation for actually doing so? The most drastic reply possible would be to challenge the distinction between intercategorial and intracategorial polysemy itself. Methodologically speaking, the model that we have derived from Kleiber's views requires a coherent way of establishing polysemy, i.e. of making a distinction between referential differences that do and those that do not correspond with conceptual differences. If, however, what is merely a conceptually neutralized referential distinction at one point turns into a case of polysemy at another, the categorial structure is not stable, and it would be natural (nay, unavoidable) to study intracategorial prototypicality together with intercategorial prototypicality. In fact, the very distinction between both types would have to be rejected. What is at stake, in other words, is not just the classification of prototypicality effects into intercategorial and intracategorial ones, but also the unity of prototype theory, that is to say, the legitimacy and the relevance of an interdisciplinary study of cognitive structures, in which the 'linguistic' study of polysemy and the 'psychological' study of categorization are inescapably intertwined. Proponents of a prototype-oriented perspective in lexical semantics therefore have every reason to have a closer look at the concept of polysemy.

3. Criteria for polysemy - A classification The distinction between vagueness and polysemy involves the question whether a particular piece of semantic information is part of the underlying semantic structure of the item, or is the result of a contextual (and hence pragmatic) specification. For instance, neighbor is not polysemous between the readings 'male dweller next door' and 'female dweller next door', in the sense that the utterance my neighbor is a civil servant will not be recognized as requiring disambiguation in the way that she is a plain girl ('ugly' or 'unsophisticated'?) does. The semantic information that is associated with the item neighbor in the lexicon does not, in other words, contain a specification regarding sex; neighbor is vague (or general, or unspecified) as to the dimension of sex.

Vagueness's puzzles, pozvsemy's vagaries

105

The existence of altemative terms (see Norrick 1981: Ill) for the phenomenon that we are concemed with is related to the fact that the term vagueness has various applications that do not coincide with the kind of vagueness meant here. Specifically, it is important to distinguish the categorical lack of specification illustrated above from the referential indeterminacy (or 'vagueness') that may characterize the individual members of a category or the category as a whole. The former, individual type of referential indeterminacy is illustrated by knee: it is impossible to indicate precisely where the knee ends and the rest of the leg begins. The class type is illustrated by any color term: it is impossible to draw a line within the spectrum between those hues that are a member of the category red and those that are not. In order to bring some structure into this set of interpretations of the notion 'vagueness', it should be noted that they involve indeterminacy on three distinct, hierarchically related levels. The knee example is situated on the level of individual entities; the question where a particular entity (such as a knee) ends is specified as the question whether a potential part of the entity (say, a specific area of the leg) is part of the entity or not. The red example is situated on the level of groups of entities, i.e. conceptual categories; the question conceming the boundaries of a particular category is specified as the question whether a particular entity (such as a particular hue) is part of the category or not. The neighbor example of vagueness, finally, is situated on the level of polysemous lexical items, considered as groups of related conceptual categories; the question where exactly the boundaries of a specific polysemous item lie is specified as the question whether a potential conceptual category (such 'male person living next door') is indeed a member of the set of meanings of a particular item (such as neighbor). While we concentrate here on the third kind of vagueness, it is mainly the second kind that has received a lot of attention in the recent literature on the subject of vagueness. 2,3 The difficulties created by the terminological divergences are enhanced by the existence of various kinds of tests for distinguishing between vagueness and polysemy. A definitive treatment of this subject does not yet exist; despite its importance for theories of semantic structure, it has received relatively little systematic and continuing attention. Without discussing all specific tests that have been suggested, three types of criteria will be distinguished here. (The most comprehensive treatment of tests for lexical polysemy that I am aware of is to be found in eruse 1982, which will furnish a basic reference point for the following discussion 4 .5 Other relevant works will be mentioned in passing in the course of the following pages.)

106

Polysemy

First, from the truth-theoretical point of view taken by Quine (1960: 129), a lexical item is polysemous if it can simultaneously be true and false of the same referent. Considering the readings 'harbor' and 'fortified sweet wine from Portugal' of port, the polysemy of that item is established by sentences such as Sandeman is a port (in a bottle), but not a port (with ships). Or, in the words of Quine: 'An ambiguous term such as light may be at once clearly true of various objects (such as dark feathers) and clearly false of them' (1960: 129). Second, linguistic tests involve semantic restrictions on sentences that contain two related occurrences of the item under consideration (one of which may be implicit or deep-structural); if the granmlatical relationship between both occurrences requires their semantic identity, the readings that can be attributed to the resulting sentence may be an indication for the polysemy of the item. For instance, the identity test described by Zwicky and Sadock (1975) applies to constructions that were assumed in the Chomskyan 'standard theory' to involve transfomlations such as conjunction reduction and so-reduction, which require the semantic identity of the items involved in the reduction. (It may be noted that constructions such as these are nowadays no longer discussed in transformational terms. However, as the current term 'identity-of-sense anaphora' indicates, the idea that there are semantic restrictions on the construction remains intact. Note also that the type of test mentioned here did not originate with Zwicky and Sadock; in particular, see Lakoff 1970.) Thus, at midnight the ship passed the port, and so did the bartender is awkward if the two lexical meanings of port are at stake; disregarding puns, it can only mean that the ship and the bartender alike passed the harbor (or, perhaps, that both moved a particular kind of wine from one place to another). A 'crossed' reading in which the first occurrence of port refers to the harbor, and the second to wine, is normally excluded. Conversely, the fact that the notions 'vintage sweet wine from Portugal' and 'blended sweet wine from Portugal' can be crossed in Vintage Naval is a port, and so is blended Sandeman indicates that port is vague rather than polysemous with regard to the distinction between blended and vintage wmes. Third, the definitional criterion (as infomlally stated by Aristotle in the Posterior Analytics II.xiii) says that an item has more than one lexical meaning if there is no minimally specific definition covering the extension of the item as a whole, and that it has no more lexical meanings than there are maximally general definitions necessary to describe its extension. Definitions of lexical items should be maximally general in the sense that they should

Vagueness's puzzles,

po~vsemy 's

vagaries

107

cover as large a subset of the extension of an item as possible. Thus, separate definitions for 'blended sweet fortified wine from Portugal' and 'vintage sweet fortified wine from Portugal' could not be considered definitions of lexical meanings, because they can be brought together under the definition 'sweet fortified wine from Portugal'. On the other hand, definitions should be minimally specific in the sense that they should be sufficient to distinguish the item from other non-synonymous items. A maximally general definition covering both port 'harbor' and port 'kind of wine' under the definition 'thing, entity' is excluded because it does not capture the specificity of port as distinct from other items.

4.

Criteria for polysemy - Remarks and refinements

There are three important (sets of) remarks to be made with regard to this classification of types of criteria for distinguishing between polysemy and vagueness. First, a number of tests have to be briefly mentioned that have not been included in the classification because their insufficiency is generally accepted. Second, the basic types enumerated above may be supplemented with variants (a possibility that will turn out to be of particular importance for the logical test). And third, there are some remarks about the theoretical background of the tests (and in particular, the definitional test) that are worthwhile mentioning. 4.1. Omissions For the sake of completeness, let it be noted that a number of tests whose insufficiency or impracticability has been sufficiently established elsewhere have not been included in the classification presented above. In particular, these involve the 'indirect' criteria distinguished by Cruse (1982), which refer to the relations that an item may contract with other items in the language (or in other languages, such as when the existence of different translations for a lexical item is taken as proof for the polysemy of that item). (The criteria discussed and rejected by Herringer 1981: 117-118 can be included in the same class.) As an illustration, consider the view according to which the existence of distinct antonyms for specific readings of a lexical item serves as proof of its polysemous nature. In the case of light, the reading 'not having much weight' contrasts with heavy, and the reading 'having a bright color' contrasts with dark. Although, in other words, the antonymy

108

Polysemy

test seems to work for light, there are various counter-examples. On the one hand, there are items with identical antonyms for distinct readings, as in the case of old, which has new as its opposite both in the reading 'former' and in the reading 'advanced in age'. On the other hand, there are items with distinct antonyms for what are merely distinct referential specifications rather than distinct meanings. For instance, when the Dutch word vers 'fresh' is applied to bread, it contracts a relationship of anton;my with oud 'old' but also with oudbakken 'stale'; when vers applies to meat, the antonyms would be oud and adellijk 'high, gamy'. Intuitively and definitionally, however, the two applications of vers constitute the single reading 'recent, and as such, optimal for consumption'. Examples such as these lead eruse to the evaluation that the 'indirect' criteria are less satisfactory than the 'direct' ones, which focus directly on an item's behavior in various contexts 6 4.2. Variants A second major point concerning the classification of polysemy criteria involves the presence, within each type, of other tests than the ones mentioned above. The three types of criteria are meant precisely as types: classes of tests that have something crucial in conunon. If, then, it is relevant to group polysemy criteria into the three classes identified above, we should be able to place specific tests that have not yet been discussed into one of the three classes. Because this point is specifically important for the first criterion, let us now have a look at the distinct varieties of the truth-functional approach, trying to bring some order into the scattered suggestions made by other authors. The starting-point for distinguishing between subtypes of the truthfunctional test is the recognition that Quine' s proposal rests on two features: first, polysemy is signaled by truth-conditional differences between the two alleged meanings, and second, these truth-conditional differences are revealed by applying the two readings to a single situation. Truth-functional criteria for polysemy that do not strictly coincide with the Quinean test as defined above, can now be introduced by taking these two features as a guide-line; basically, they contain variations on those features as involving the questions 'What kind of truth-conditional differences are we looking for?', and 'How do we diagnose them?', respectively. With regard to the first feature, Quine himself already noted that next to the oppositeness of truth value implied by the 'p and not p' criterion, a third truth value 'nonsensical' or 'irrelevant' may have to be taken into account. Questions and chairs can both be hard, but it is difficult to find a situation in

Vagueness's puzzles,

po~vsel1lY 's

vagaries

109

which both readings of hard are relevant (but in which one would be applicable and the other not). Although Quine seems to suggest that such a situation could be treated as a case of vagueness rather than polysemy, it is much more in line with the essence of the truth-functional approach to meaning discrimination, to assume (with Norrick 1981: 113) that a single sentence such as This book is sad may signal the polysemy of sad. The fact that the reading 'evoking sadness' of sad may render the sentence true or false according to the circumstances, whereas the reading 'experiencing sadness' yields a nonsensical interpretation, implies that there is a crucial distinction in the truth-functional properties of This book is sad under both interpretations. Given that the distinction depends on a distinction in the meanings we attribute to sad, this should be as good evidence for the polysemy of sad as a straightforward 'p and not p' case. The truth-functional differences involved in the test should, in other words, be taken broadly so as to include the difference between 'either true or false' and 'nonsensical'. Because the reading 'this book is experiencing sadness' turns This book is sad into a semantically ill-formed construction, it is difficult to have a single well-formed sentence of the 'p and not p' type in which the relevant interpretation of sad and the nonsensical interpretation are combined. This means that (as far as the second feature of the basic Quinean test is concerned) the differences in truth-functional properties that we are looking for will not be signaled by the spontaneous acceptability of a sentence of the 'p and not p' type, but will involve an active consideration of the properties of potentially distinct readings. The 'p and not p' test does not initially require a lot of conscious thinking about different meanings: if it is spontaneously clear that a feather can be light and not light, the polysemy of light is probably firmly established, psychologically speaking; it springs to mind so easily that the non-contradictory nature of the 'p and not p' construction is automatically recognized. In this sense, the 'p and not p' test has the same attractiveness that grammaticality judgments have in syntax: the judgment is unmediated. In the case of This book is sad, however, there is a lot of conscious mediation and active introspection involved, mainly because one has to actively conceptualize the different potential readings of the item and evaluate their truth~conditional consequences. Once this more active and more mediated approach is accepted (and it will have to be accepted if it is correct that the 'p and not p' frame cannot be applied to cases involving a nonsensical reading), varieties of this active search for truth-conditional differences may be explored. I shall mention two, but I do not want to claim that these are the only possibilities. We shall see later (in section 6.2) that

110

Polysemy

the variants of the logical criterion presented here run into trouble, but for the moment, we shall merely present them. To begin with, if a non-contradictory use of the 'p and not p' frame indicates the presence of polysemy, so does a non-redundant use of the 'p and p' frame. If p has two meanings, it should be possible to assert 'p and p' without raising feelings of redundancy or awkwardness. That is to say, while the main 'p and not p' version of the logical test considers cases in which a sentence can be used in the same situation with opposite truth values, the 'p and p' variant considers cases in which p is used in the same situation in two readings with the same truth value. The obvious question then is how we can know whether p indeed has two distinct interpretations. Apart from the possibility of asserting 'p and p' without redundancy or awkwardness (a possibility that is seldom noted in the literature), there is the introspective psychological phenomenon that in the process of interpreting the sentences in question, our attention seems to be switching continuously between one reading and the other. Take the sentence Charles has changed his position, in which Charles is an official delegate at a political summit meeting (the example is Cruse's ; cp. Cruse 1982: 70). The readings involved can be paraphrased as 'Charles has moved to another place in the conference hall' and 'Charles has changed his opinion concerning the topic discussed at the conference'. Now, even in those cases where both readings are true, it is difficult to mentally conceive of them together, i.e. to fuse them into a single reading. Rather, at anyone time, we seem to be able to have only one of the readings at the same time in mind. In the words of Herringer (1981: 95): Wichtig ist also dieses 'entweder/oder', und das, was man das Kippen zwischen den Bedeutungen genannt hat: Es gibt keine fliessenden Dbergange zwischen den Bedeutungen, sondern einen Aspektwechsel, der sich pl6tzhch und oft iiberraschend einsteUt. [What is important, then, is this either/or situation, and what has been called the tilting between meanings: There is no fluent transition between the meanings, but an aspectual change that imposes itself suddenly and by surprise.] Herringer (1981) also discusses a second variant of the basic logical criterion: the truth-conditional differences between sentences can be revealed by taking into account the implications of potentially distinct meanings: if these are indeed distinct, they should have different implications. For instance, the French sentence in (1) contains the polysemous word louer, which may mean either 'to rent' or 'to praise'. In one reading, (1) entails (2), but in

Vagueness's puzzles,

po~vsemy's

vagaries

III

another, it entails (3). However, if we combine these facts into the single implication in (4), and then compare that sentence with (5), where an unsuspected case of vagueness is involved, it becomes clear that the format of the entailment as such does not always distinguish between vagueness and polysemy: the necessity to express the entailments of a sentence in disjunctive form does not sufficiently distinguish between both phenomena. (1) (2) (3) (4)

(5) (6) (7)

11 a loue son appartement. 11 s 'est oblige de payer son appartement. 'He has taken on the obligation to pay the apartment.' 11 a dit du bien de I 'appartement. 'He has spoken well of the apartment.' S'il a loue I 'appartement, il s 'est oblige de payer ou il en a dit du bien 'If he has loue the apartment, either he has taken on the obligation to pay for it, or he has spoken well of it. ' Si c 'est un enfant, c 'est un garr;on ou unefille. 'If it is a child, it is a boy or a girl. ' lfit is not a child, it is not a boy and it is not a girl. If they did not reach the bank in time, either they did not reach the money-bank in time, or they did not reach the river-bank in time.

This problem can be avoided, however, when negation is added to the test (cp. Kempson and Cormack 1981, Cruse 1982). When a vague predicate is negated, the specifications of the predicate are both negated as well. In (6), for instance, the distinction between male children and female children (an example of vagueness with regard to child) lies at the basis of a conjunction of negations. On the other hand, in the case of polysemy, the negation need not extend to both readings at the same time. Thus, in (7), the negated ambiguous readings occur as a disjunction. Notice, in this respect, that the disjunction is necessary if the 'p and not p' frame is to be applicable in the case of polysemy: if PI and P2 are the readings involved, and if the negated form 'not p' would automatically entail both 'not PI' and 'not P2', an interpretation of 'p and not p' as 'PI and not P2' or 'P2 and not PI' would be excluded. 4.3. Specifications To round off the discussion of the classification of polysemy criteria, something more has to be said about the third criterion, if only because it is absent in the works we have been referring to. The criterion is of particular importance because it so to speak looks into the lexical items to determine

112

Polysemy

how the various meanings of the item are constituted. Whereas the other two criteria are at best diagnostic tools that signal the presence of polysemous readings, the third criterion is an analytical one that examines the internal constitution of the meanings of the item. In order to be better able to appreciate the impact of this definitional approach, let us have a closer look at Aristotle's description of it. We should look at what are similar and undifferentiated, and seek, first, what they all have that is the same: next, we should do this again for other things which are of the same genus as the first set and of the same species as one another but of a different species from those. And when we have grasped what all these have that is the same, and similarly for the others, then we must again inquire if what we have grasped have anything that is the same - until we come to a single account; for this will be the definition of the object. And if we come not to one but two or more accounts. it is clear that what we are seeking is not a single thing but several. I mean, e.g., if we were to seek what pride is we should inquire, in the case of some proud men we know, what one thing they all have as such. E.g. if Alcibiades is proud, and Achilles and Ajax, what one thing do they all have? Intolerance of insults; for one made war, one waxed wroth, and the other killed himself. Again in the case of others, e.g. Lysander and Socrates. WelL if here it is being indifferent to good and bad fortune, I take these two things and inquire what both indifference to fortune and not brooking dishonor have that is the same. And if there is nothing, then there will be two sorts of pride (Barnes 1984: 161). [An alternative, more literal reading for the Greek megalopsychia is 'greatness of the soul' rather than 'pride'.]

Granting that Aristotle does not seem to make a distinction between the definition of the word pride and the definition of the thing 'pride' (granting, in other and more traditional words, that his approach to definition is a realist rather than a nominalist one), the quotation helps to see the close connection between the definitional criterion of polysemy and the classical conception of definitions as sufficient sets of necessary criteria for class membership. The whole criterion could, in fact, be rendered in simplified form as the statement that an item is polysemous (or, alternatively, that what was thought to be a single category in fact consists of several categories) if there is no single set of necessary and sufficient attributes to describe the extension of the item or the alleged category. The full criterion does not, however, coincide entirely with this simplified version, because the latter only specifies the lower limit of an item's polysemy: if there is no single set of necessary and sufficient attributes that is minimally specific, the item has at least two meanings. But the definitional criterion of polysemy also sets an

Vagueness's puzzles, polvsel1ly 's vagaries

113

upper limit to the polysemy of an item, by specifying that an item has no more meanings than the number of maximally general definitions that are needed to cover its entire range of application. If, for instance, the two definitions 'indifference to fortune' and 'intolerance of dishonor' are sufficient to cover all examples of proud men that you know of, mega/opsychia does not represent more than two meanings. Notice that this 'upper limit' specification of the simple 'lower limit' rule that an item is polysemous if it cannot be defined by means of a single definition, is an entirely natural one: does not the very fact that we seem to be looking for single definitions first suggest that the crucial thing about definition is generality? In popular parlance, the definitional criterion might then read as follows: 'if you cannot achieve maximal generality, settle for the next degree of generality that you can attain'. After Aristotle, the definitional criterion for polysemy became a standard part of treatises on logic as the requirement that definitions should be neither too broad nor too narrow (see, for instance, Copi 1972: 138). The transition is not a straightforward one, however, to the extent that a requirement on definitions is not necessarily a criterion for polysemy: imposing restrictions on definitions would seem to imply that the definer already knows for certain what the definiendum is, i.e. what the lexical meanings are that are to be defined. However, if the definiendum is taken to be a lexical item rather than an independently discerned lexical meaning, the requirement actually turns into a criterion for the polysemy of the item. In that case, the requirement that a definition should not be too narrow embodies the principle that maximal generality should be aspired at, whereas the requirement that a definition should not be too broad reflects the necessity of minimal specificity as defined above. In actual practice, the fonnulations of the requirement that may be found in logical treatises waver between an interpretation of the target of the definition as a word and an interpretation as a word meaning. For instance, Copi's example ofa definition that is too narrow (1972: 138) is said to involve 'the word shoe', but practically, it seems that only the meaning 'outer foot-covering for a person' is at stake, and not (for instance) the meanings 'horseshoe' or 'thing that is like a shoe in shape or use (such as wheel-drags, sockets, ferrules etc.)'. It would seem, in other words, that the crucial distinction between both interpretations of the standard requirement on definitions is not always made. 7 It should be emphasized that the definitional polysemy test is particularly important in comparison with the other two: it is the only one that embodies a hypothesis about the principles of categorization that human beings em-

114

Polysemy

ploy. It suggests, in fact, that one of the basic characteristics of naturallanguage categorization is a tendency towards generality (or, in other words, towards maximal abstractions). Thus, while the other two classes of polysemy criteria merely contain diagnostic tools for establishing whether interpretations of lexical items constitute distinct meanings, the definitional test gives us an explanatory surplus, because it is based on an assumption about why particular senses are distinguished. It is all the more regrettable, then, that the definitional test has hardly received any explicit attention in the recent linguistic literature on the subject of lexical polysemy.

5. Problems with the polysemy tests - First series Given the existence of three major types of tests for lexical polysemy, we can now try to determine their mutual relationship. We will concentrate primarily on the tests that were earlier presented as the basic representative of each type, rather than bringing all possible variants of each type (as discussed in the previous paragraph) into the comparison. 5.1. Pairwise divergences among the criteria Just a few lexical items may suffice to show that the three tests, taken pairwise, may yield different results. First, consider an autohyponymous word such as dog (i.e. a word one of whose senses is a superordinate category with regard to one of the other senses). In the reading 'Canis familiaris', dog is a hyperonym of dog in the reading 'male Canis familiaris'; in the hyperonymous reading, dog contrasts with (for instance) cat, whereas the proper antonym is bitch in the hyponymous reading. But are these readings really distinct meanings? Applying the tests, it will be easily verified that dog is not polysemous according to the definitional criterion, but does exhibit polysemy according to the logical criterion. Taking Lady (from Disney's Lady and the Tramp) as a prime example of a female dog, it is possible to assert Lady is a dog alright. but she is not a dog. On the other hand, according to the definitional criterion, the hyponymous 'male Canis familiaris' reading is always a proper subset of the superordinate reading, so that no definition of that subset could ever claim the maximal generality that is required. Now, what about to the linguistic test? According to the linguistic test, a 'crossed' reading of sentences containing identity of sense anaphors (or similar constructions) should come out grammatically awkward. That is to say, interpreta-

Vagueness's puzzles, po~vsemy 's vagaries

115

tions of such a sentence in which the full occurrence of the item under consideration receives meaning x, and the anaphorical element receives meaning y, are ungrammatical. In the case of dog, however, it could be doubted whether such an ungrammaticality would ever surface as positive evidence in favor of the polysemy of dog, given that the anaphorical sentence would always automatically be interpreted as a grammatical construction involving just a single one of the two readings of dog. For instance (assuming that Fido is a male dog), Fido is a dog and so is Lady is spontaneously understood as the correct assertion 'Lady and Fido are both members of the species Canis familiaris', or conversely, as the false assertion that both animals are male members of that species. In neither case is the sentence spontaneously felt to be odd in the sense in which At midnight the ship passed the port, and so did the bartender is. Of course, it could be argued that the very fact that Fido is a dog and so is Lady can be either true or false (depending on the reading we attach to dog) in the same situation, testifies to the polysemy of dog. This is correct, but the invocation of truth-functional differences indicates that we are then dealing with a variant of the logical test, rather than with the linguistic grammaticality test that we are concerned with. Let us accept, therefore, that the linguistic test does not testify to the polysemy of dog. We will then have established a distinction between the definitional test and the linguistic test on the one hand, and the logical test on the other. Next, let us consider a case in which the linguistic and the definitional test are at odds. Examples are easy to find (in particular, see Nunberg 1979 and Deane 1988). In the sentence The newspaper has decided to reduce its size, the item newspaper refers first to the management or the board of directors of the paper, and then to the paper as a material object. The perfect grammaticality of the sentence does not allow us to spot any polysemy on linguistic grounds. Definitionally, on the other hand, it does not seem possible to find a description that fits the 'board of directors' usage and the 'actual. material publication' usage into a single well-defined category. This implies that there is a definitional polysemy in the concept newspaper that is obscured in the sentence The newspaper has decided to reduce its size. And this in turn implies that we now have examples establishing for each of the three pairs of tests that can be construed out of the basic set of three criteria, that the members of those pairs may be at odds with each other: while the dog example dealt with the definitional/logical and the linguistic/logical pairs, the newspaper example takes care of the definitionalllinguistic couple.

116

Polysemy

To complete the analysis of the newspaper example, let us notice that applying the logical test also yields polysemy. Speaking of the overpowering presence of a certain editor-in-chief, we could easily utter something like That man is the newspaper, meaning that he is the driving force of the whole business. Because the utterance would be nonsensical if newspaper were interpreted in the 'material publication' sense, the variant of the logical test that we discussed earlier in connection with This book is sad reveals the polysemy of newspaper. It can now also be appreciated that our earlier difficulties with the application of the linguistic test to dog are of relative importance. If we were to reject our earlier views on the matter and accept that, by some kind of mental aerobatics, a crossed reading of Fido is a dog and so is Lady would not only be possible but would also be recognized as odd, 8 the dog example could no longer be invoked to establish that the linguistic test and the logical test are not co-extensive. Even then, however, we could rely on the newspaper example to bring home the point. 5.2. Three-way divergence in a single item For further illustration, let us now consider a category that is a typical example of prototypicality. The category fruit is, in fact, among the categories originally studied by Rosch (see, in particular, Rosch 1975 and Rosch and Mervis 1975 for experimental results). An immediate difficulty withfruit is the vagueness of membership in the category (a problem that is less outspoken, if not absent, in the case of newspaper and dog): is a coconut or an olive a fruit? Notice, first, that we are not concemed with the technical, biological sense offruit, but with our folk model of fruit as a certain category of edible things. Technically, any seed-containing part of a plant is the fruit of that plant; as such, nuts in general are fruit. In ordinary language, on the other hand, nuts and fruit are basically distinct categories (regardless of the possible boundary status of the coconut): nuts are dry and hard, while fruits are soft, sweet, and juicy; also, the situations in which nuts and fruits are eaten are typically different. Second, remember that category membership is not the same thing as typicality: a penguin is undoubtedly an uncharacteristic kind of bird, but it is a bird nonetheless; as to the olive, the question is not just whether it is a typical fruit, but rather whether it is a fruit at all. This indeterminacy has an inm1ediate bearing on the applicability of the various polysemy tests. To begin with, the borderline status of olives shows up in the following exchange: Daddy, is an olive a fruit? - Well, it is and it is not (that is to

Vagueness's puzzles,

po~vsemy 's

vagaries

117

say, it has some of the characteristics of the things that we typically call fruits, but it also differs from the clear cases on a number of important attributes). The 'logical' polysemy that shows up here is similar to the use of hedges: 'up to a point', or perhaps even 'strictly speaking', an olive is a fruit, but seen from another angle, it is not. Applying the linguistic polysemy test, on the other hand, does not seem to yield indications of polysemy. Depending on one's conception of the category fruit, the sentence An orange is frUit and so is an olive would either seem to be simply true or false, but not ungramrnatical (as the linguistic test would have it if there is polysemy). Further, a definitional analysis is initially hampered by the uncertainty surrounding the extensional boundaries of the category. If there is consensus that olives are not fruit, we should not include the olive in an analysis of frUit. Conversely, if an olive is considered to be a fruit (however peripheral and uncharacteristic), it will have to be included. (We might then expect the definitional analysis to reveal the same polysemy that is illustrated by Daddy, is an olive afruit? - Well, it is and it is not.) To circumvent the problem with olives and their likes, let us restrict the definitional analysis to clear cases of fruit, i.e. cases for which doubts about membership do not play a role. Even for these clear cases, it can be shown that ji'uit cannot be defined by means of a single minimally specific set of necessary attributes. Because such absences are often assumed but seldom demonstrated in the literature, it may be useful to be a bit more specific on this point than is usual. A starting-point for a consideration offruit can be found in Wierzbicka's (characteristically lengthy) definition of the category (1985: 299-300). In order to show that this is not a classical, necessary-andsufficient definition, it has to be established, on the one hand, that not all attributes of fruit as mentioned by Wierzbicka are general (even within the set constituted by the examples of fruit that are high on Rosch's typicality ratings), and on the other hand, that the remaining set of general attributes is not minimally specific, i.e. does not suffice to distinguish fruit from, for instance, vegetables. The following characteristics mentioned by Wierzbicka are not general. (I repeat Wierzbicka's formulations, though not in the order in which she presents them.) (a) They have a skin harder than the parts inside. (b) They have some small parts inside, separate from the other parts, not good to eat. These parts put into the ground could grow into new things of the same kind growing out of the ground.

118

Polysemy

(c) They are good to eat without being cooked, without having anything done to them, without any other things, and people can eat them for pleasure. (d) Eating them uncooked makes one feel good. (e) Before they are good to eat they can be sour. (f) They have a lot ofjuice. (g) Their juice is good to drink. (h) They are also good to eat dried. Characteristic (a) is contradicted by the strawberry, which has no skin worthy of that name. Strawberries likewise do not have the seeds mentioned in (b); bananas are another case in point. Attributes (c) and (d) indicate that fruit can be eaten (with pleasant results) without further preparation, but this does not seem to hold for the lemon, whose sour taste generally requires sugaring. (Even if this counter-example were not accepted, adding (c) and (d) to the list of attributes that are general for fruit would not solve the problem that that list does not suffice to distinguish fruits from some vegetables and nuts.) Attributes (e) and (t) are not valid for the banana: first, an unripe banana is bitter rather than sour, and second, there is no juice in a banana. Because the generality of (g) depends on the generality of (t), it may likewise be discarded. Finally, as far as (h) is concerned, it is difficult to imagine a dried melon as being good to eat. Next, there is a set of characteristics whose non-generality seems to be accepted (or at least, implied) by Wierzbicka herself. (i) Wanting to imagine such things, people would imagine them as growing on trees. (j) They can be small enough for a person to be able to put easily more than one thing of this kind into the mouth and eat them all at the same time, or too big for a person to be expected to eat a whole one, bit by bit, at one time, but wanting to imagine such things, people would imagine them as too big for a person to put a whole one easily into the mouth and eat it, and not too big for a person to be expected to eat a whole one, bit by bit, at one time, holding it in one hand. (k) After they have become good to eat they are sweet, or slightly sweet, or sour but good to eat with something sweet. (I) Wanting to imagine such things after they have become good to eat, people would imagine things which are slightly sweet. (m) Things on which such things can grow can also grow in some places where people don't cause them to grow, but wanting to imagine such things, people would imagine them as growing on things growing out of the ground in places where people cause them to grow.

Vagueness's puzzles,

po~ysemy 's

vagaries

119

While (k) is a disjunctively defined attribute (i.e., is a superficial combination of two characteristics that are each not general when taken separately), the other features are introduced by the formula wanting to imagine such things, people would imagine them as; this would seem to indicate that the attribute is merely typically associated with the concept, rather than being general. For instance, the sweetness mentioned in (I) does not hold for lemons, and berries do not grow on trees, in contradistinction with the feature involved in (i)9 The set of general characteristics that is left over after the elimination of the previous sets contains the following features. (n) They grow as parts of certain things growing out of the ground. (0) They don't grow in the ground. (p) They become good to eat after they have grown long enough on the things growing out of the ground. (q) Before they are good to eat they are green or greenish outside. (r) People cause things of this kind to grow in many places because they want to have those things for people to eat. (s) They are good to eat cooked with sugar, or cooked as part of some things which have sugar in them. Is this set minimally specific? Up to characteristic (r), the set applies not only to fruit, but also to nuts, herbs, and large collections of vegetables (though not to the ones that grow in the ground, like carrots), so that the crucially distinctive attribute would be (s). However, if one takes into consideration the use of almonds and other nuts in certain types of pastry, the use of herbs (such as tansy) in pancakes, and the habit of cooking rhubarb with sugar, it soon becomes clear that there are counter-examples with regard to (s) in each of the three categories (nuts, herbs, and vegetables). All in all, most of the attributes mentioned by Wierzbicka are not general, whereas those that are, taken together, apparently do not suffice to exclude nonfruits. Given, then, that we cannot define the uncontroversial core applications of fruit in a classical, necessary-and-sufficient fashion, let us have another look at the other two polysemy tests. The linguistic test still does not yield indications of polysemy: An orange is a fruit and so is a banana is straightforwardly grammatical. Furthermore, the logical test does not seem to signal polysemy either: a sentence like A strawberry is a fruit and not a fruit is simply false, in spite of the fact that the strawberry does not belong to all the maximal subsets that we can distinguish in the structure offruit at the same time. That is to say, if each of those maximal subsets is a distinct meaning

120

Polysemy

according to the definitional approach, it should be possible to reveal that polysemy by applying the logical criterion; but apparently, such an interpretation is impossible for A strawberry is a fruit and not a fruit. However, consider the following exchange: 1 to take the bull by the horns, to grasp the nettle' Aan de weg timmeren 'To practice carpentry at the roadside, to work in public> to attract attention by one's activities, to be in the limelight'

(2)

Met spek schieten 'To shoot with bacon> to tell a tall story, to boast' Een wit voetje bij iemand hebben 'To have a white small foot with

someone> to be in someone's good books, enjoy someone's favours' (3)

JI/fet de handen in het haar zitten 'To sit with one's hands in one's hair

> to be at one's wit's end, to be in trouble' Dat heeft niet veel om het lijTThat does not have much around the

body> there's nothing to it it does not mean very much' (4)

De kat de bel aanbinden 'To tie the bell to the cat> to bell the cat, to

take the lead in a dangerous activity' Als puntje bij paaltje komt 'When point reaches pole> when it comes

to the crunch, when all is said and done, when you get down to the nitty-gritty, Systematically, the idiomatic readings in (1) are both isomorphic and motivated. Those in (2) are isomorphic but not motivated, those in (3) motivated but not isomorphic, and those in (4) neither isomorphic nor motivated. The isomorphic nature of de kae bi} de harens vatten follows from the fact that a consistent one-to-one mapping can be defined between the elements of the global meaning and the meanings of the constituent parts of the expression. If we paraphrase the idiomatic meaning as 'to tackle a problem or a difficulty at the central, most dangerous or difficult point', it becomes clear that the cow maps onto the problem in its entirety, while the horns represent the most tricky part of it; taking hold of the horns further symbolizes tackling the core of the problematic situation. Similarly, met spek schieten is isomorphic because the tall tales that are told can be seen to correspond with spek, while the telling of the tales corresponds with schieten. Conversely, it is difficult to identify those aspects within a situation of being at one's wit's end that could map isomorphically onto the various aspects of the situation described by met de handen in het haar zitten: what would be the hands, and what would be the hair, for instance? As far as motivation is concerned, however, it can be readily appreciated that met de handen in het haar zitten is a metonymic expression for a situation of being in trouble; taking one's

The interaction ofmetaphor and metonymy

203

head between one's hands and pondering the situation is precisely what one does in the circumstances. In the same vein, it is easy to see that the literal situation described by de koe bi) de horens vatten is a metaphorical image for tackling a problem at its most difficult spot. But it is unclear why shooting with bacon should come to indicate boasting, or why a point reaching a pole should have anything to do with things becoming serious. For further clarification of this basic classification, four remarks have to be made. In the first place, the isomorphic relations identified above should not be confused with the question whether the isomorphically mapped readings of the constituent elements of the idiomatic expressions are themselves motivated. For instance, while the lexical item koe in de koe bi) de horens vatten maps onto the 'problem' part of the global figurative reading 'tackle a problem by its most difficult aspect', there is no independent motivation for extending the semantic range of koe towards the meaning 'problem'; a semantic shift from 'cow' to 'problem' is not a conventional aspect of the meaning of koe, nor is there a readily conceivable independent metaphor that leads from 'cow' to 'problem'. In other words, the bottom paradigmatic lines in Figure 1 are not present in the case of de koe bi} de horens vatten. 4 It should now also be clearer why I suggested to take into account isomorphism as a non-directional concept of compositionality. Although de koe bi} de horens vatten exhibits isomorphism, the idiomatic meaning 'tackle the problem by its most difficult aspect' could never be arrived at by means of a bottom-up compositional process, because the building blocks for that process (for instance, an interpretation 'problem' for koe) cannot be reached independently. For lack of an independent paradigmatic motivation at the bottom of the prism, the input for a possible compositional process can only be retrieved when the output of the process (the global figurative meaning of the idiom as a whole) is already available. In the second place, motivation and isomorphism may be partial. Given a paraphrase 'to give the orders' of de lakens uitdelen (mentioned under (5) below), it is isomorphically possible to map lakens onto 'orders' and uitdelen onto 'give'. At the same time, it is possible to imagine a situation in which the person responsible for distributing the sheets is generally in charge; as such, the idiomatic meaning is motivated. But the motivational link is weak: distributing sheets is not the kind of situation that is typically associated with being in charge (or at least, not any more: apparently, the image derives from the dominant position of the lady of the house, whose control over housekeeping is symbolized by her control over the Iinencupboard). In de kogel is door de kerk, an event (such as the making of a

204

Constructions and idioms

decision) whose coming about has been delayed or hindered by some kind of obstacle, has finally occurred. The general picture is fairly clear: a material obstacle (the church) obstructs and slows down the movement of the bullet, in the same way that various difficulties obstruct and slow down the materialization of the long-awaited event. But why a church and a bullet? Again, the transparency of the motivating image is only partial. (5)

De lakens uitdelen 'To hand out the sheets> to run the show, be the boss, play the first fiddle' De kogel is door de kerk 'The bullet has gone through the church> the dice has been cast, a decision has been reached, things have taken a final turn'

(6)

Met spek schieten 'To shoot with bacon> to tell a tall story, to boast' Abraham gezien hebben 'To have seen Abraham > to be over fifty' Als puntje bij paaltje komt 'When point reaches pole> when it comes to the crunch, when all is said and done. when you get down to the nitty-gritty'

(7)

VU de bol gaan 'To go out of one's head> to blow one's top, to go out of one's mind with excitement' Het hoofd verliezen 'To loose one's head' Niet goed bij z'n hoofd zijn 'Not to be well in the head> to be soft in the head' Het hoofd loopt mij om 'My head is going round' Z'n hoofd ergens bijhouden 'To keep one's head to something> to remain attentive, to keep one's mind on something' Buiten zichze!fzijn 'To be beside oneself Vit z 'n vel springen 'To jump out of one's skin> to be beside oneself (with rage)' Vitbarsten 'To burst out, explode' Exploderen/ontplojJen 'To explode' In de wolken zijn 'To be in the clouds> to be overjoyed' In de zevende hemel zijn 'To be in the seventh heaven> to be on cloud nine' In de put zitten 'To sit in the pit> to be downhearted, be in the dumps, feel down' Door een dal gaan 'To go through a valley> to go through an abyss, to suffer a depression' Erbovenop zijn 'To be on top of it> to have overcome one's troubles'

In the third place, loss or weakening of motivation often results from cultural changes. More often than not, the background image that motivates the figurative shift is an aspect of the material or the immaterial culture of a

The interaction ofmetaphor and metonymy

205

language conununity - and when the culture changes, the imagistic motivation may lose its force. A clear example is met spek schieten (repeated under (6». Apparently, enemy ships were shot at with bacon (and similar fat substances) to facilitate setting them afire; the interpretation 'to boast' can then be reached through the intermediary of an interpretation 'to subject someone to verbal aggression, to overpower someone verbally'. In this case, the relevant knowledge belongs to the material culture of old-time warfare at sea. In Abraham gezien hebben, the relevant piece of knowledge belongs to the realm of the inunaterial rather than the material culture. Because a meaning 'to be no longer young or inexperienced' can be derived to the extent that the idiom is interpreted as a hyperbolic expression with the reading 'to have seen someone from a long time ago', the motivation is only partial: why is the borderline set at 50? In fact, it takes a good knowledge of the Bible to recognize the background of the expression, which derives from the gospel of John 8: 57. It should also be clear by now, that the motivated nature of an expression is subject to considerable individual variation (depending, among other things, on individual differences in one's familiarity with the historically motivating context). This is not to say, however, that a vast encyclopedic knowledge will always suffice to recover the motivation behind an expression. The latter may indeed be near to irretrievable, which is typically the case when professional etymologists disagree on the origin of an expression. In als puntje bij paaltje komt, for instance, some think of a transformation of an older expression als putje bij paaltje komt 'when the pit comes to the pole, when it comes to putting the pole into the pit', while others think of a variation on de puntjes op de i zetten 'to dot the i > to be meticulous about the details' . In the fourth place, the motivating image need not be specific for the expression in question; moreover, the motivating image may be complex. At this point, we can link up with the generalized metaphor research in the line of Lakoff and Johnson (1980). Without going too deep into the matter, consider uit de bol gaan in (7). The expression seems to be motivated by a combination of at least three images that are each generalized ones, in the sense that they provide a general motivation for various specific expressions. First, THE HEAD IS THE LOCUS OF ONE'S SELF-CONTROL underlies het hoofd verliezen, niet goed bij z 'n hoofd zijn, het hoofd loopt mij om, z'n hoofd ergens bijhouden. Second, LOSING ONE'S SELF-CONTROL IS LEAVING THE BODY is to be found in buiten zichze(fzijn, uit z'n vel springen, uitbarsten, exploderen, and ontplojfen. And third, UP IS POSITIVE / DOWN IS NEGATIVE

206

Constructions and idioms

is present in in de wolken zijn, in de zevende hemel zijn, in de put zitten, door een dal gaan, erbovenop zijn. 2.3. Local motivation and absence ofliteral isomorphism In the previous section, only part of the original representation in Figure 1 was envisaged. Enlarging our perspective to Figure I as a whole, there are two additional phenomena to be dealt with. On the one hand, the global motivation of an expression can be contrasted with the 'local' motivation of each of the elements in the expression, i.e., motivation along the top paradigmatic line of the prism has to be supplemented with motivation along the bottom line. On the other hand, isomorphism at the figurative back of the prism can be contrasted with isomorphism at the literal front of the prismatic representation. Pare Is voor de zwijnen in (8) below is an example of an idiom in which Figure 1 is fully realized. The global image is motivated (it is easy to appreciate what it means to throw valuable things at the feet of unworthy beings), and the figurative meaning is isomorphic (parels maps onto the valuable things that feature in the idiomatic meaning, and zwijnen maps onto the lowly beings that they are surrendered to). At the same time (and this a major contrast with the de koe bij de horens vatten example that we discussed earlier), both the transition from pare! to 'valuable thing' and from zwijn to 'unworthy person' is motivated on the basis of an evaluative metaphor. Up to a certain point, the metaphorical transition may even be said to be lexicalized (in the sense of being conventional): zwijn is a regular term of invective in the same way that pig is, and een pareltje is a term of praise in the same way that a gem, a jewel, a pearl are. (This is not to say, however, that the metaphorically motivated readings of parel and zwijn in the idiom coincide exactly with the lexicalized metaphorical meanings. For instance, as a term of abuse, zwijn normally implies that the person in question lives an immoral life, whereas no such implication need be present in parels voor de zwijnen. The important point to see is that the lexicalized existence of zwijn in a reading like 'unworthy person, specifically because of his immoral behavior' strengthens the motivated nature of the related reading that the word receives in the context of the idiom.) (8)

Parels voor de zwijnen gooien 'To cast pearls before swine'

(9)

lemand de loe!a(steken 'To take the wind out of someone's sails> to

get the better of someone, to deprive someone of an advantage' Tegen heug en meug 'Against heug and meug > against one's will, re-

The interaction ofmetaphor and metonymy

207

luctantly' T/an hot noch haar weten 'To know neither hot nor haar > to be totally

ignorant' (l0)

lets aan de kaak stellen 'To put something at the jaw> to expose, de-

nounce something' lets op touw zetten 'To put something on the rope> to organize, plan,

start, launch something' Given this analysis of pareIs voor de zwijnen as a fully motivated and fully isomorphic idiom, two additional remarks can be made. To begin with, it will be appreciated that the figurative reading of fully motivated and fully isomorphic idioms can be arrived at along two interpretative routes: either the global literal meaning is derived first and then transferred into the figurative realm, or the shift from the literal meaning of the individual words to their transferred meaning is effectuated first, to be followed by their compositional combination into the idiomatic meaning of the expression as a whole. In terms of Figure 1, the interpretation can go from the bottom to the top first, and then to the back, or it can go to the back along the bottom first, and then move up. Or, in still other words, either the syntagmatic dimension is gone through first, or the paradigmatic one takes precedence (starting, each time, at the lower front of the prism). Which of both interpretative paths is psychologically real (or merely preponderant) is another matter, but it is important to see that both are at least in principle possible. Further, the 'local', lexical motivation may be partial, i.e., it need not involve all the items constituting the expression. When iemand de loef afsteken is interpreted as 'to deprive someone of an advantage (like an initial superior position)', loef maps onto 'the advantage', while afsteken can be associated with the notion of deprivation. In the latter case, the association is motivated: the metaphorical transfer from the literal meaning 'to cut off of afsteken to 'to deprive' is a transparent one. In contrast, no such motivating link is possible in the case of loef, because the latter does not have a literal meaning for the average speaker of Dutch. Notice also that there are gradations in the degree of motivation,S Conventionalized shifts of meaning like the ones we mentioned in the case of pareI and zwijn are stronger than the shift from 'to cut off to 'to deprive' in the case of a.f.~teken: the latter extension is possible and perhaps even plausible, but it is not a conventional one. Together with iemand de loefafsteken, the other expressions under (9) illustrate the second major extension of the basic classification presented in section 3. Because loe/. heug, meug, hot, and haar are cranberry morphs, they illustrate the case in which there is no isomorphism on the literal level: a

208

Constructions and idioms

global literal meaning cannot be computed because one or more of the lexical building blocks lacks a meaning of its own. (Incidentally, the earlier discussion of de loef afsteken has made clear that the absence of isomorphism on the literal level can co-occur with isomorphism on the figurative level.) In addition, it may be useful to mention the existence of 'hidden cranberry morphs' such as kaak and touw in (l0). In the context that motivates the idiom, the words have their older meanings 'pillory' and 'loom' respectively; at present, these meanings are no longer common usage. The homonyms kaak 'jaw' and touw 'rope', on the other hand, are high-frequency words. The expressions aan de kaak stellen and op touw zetten, then, can receive an interpretation incorporating the readings 'jaw' and 'rope' (as suggested in (l0», but this is obviously not the kind of literal reading that could motivate (on the level of the expression as a whole) the figurative interpretation of the idiom. To the extent that kaak 'pillory' and touw 'loom' survive only in the expressions aan de kaak stellen and op touw zetten, they are like ordinary cranberry morphs; to the extent, however, that they formally coincide with the homonyms kaak 'jaw' and touw 'rope', they can be called 'hidden cranberry morphs' . It appears, in short, that the associative links presented in Figure 1 may be present in various combinations. A full-fledged investigation into the semantics of idioms, then, will have to include an overview of the various ways in which the model sketched here may be partially realized. (See Geeraerts and Bakema 1993 for an example of what such an overview may look like.) 2.4. Reinterpretation processes The examples of isomorphism on the figurative level that were mentioned in section 2.2 yield secondary, non-original interpretations for the items involved in the process. For instance, the isomorphic association between lakens and 'orders' in de lakens uitdelen yields a contextually determined interpretation for lakens, but because the transition from laken 'sheet' to laken 'order' is neither conventional nor motivated (in the sense that the latter reading is a plausible and transparent semantic extension of the former), the contextually isomorphic interpretation of laken as 'order' is not likely to acquire much structural weight in the lexicon. The process of contextual reinterpretation within the idiom is not always, however, without structural importance. 6

The interaction ofmetaphor and metonymy (11)

209

Spek'Verkoper 'Person who sells bacon> person who boasts' Van heinde en verre 'From far and wide, from everywhere' Met zijn talenten woekeren 'To make the most of one's talents> to ex-

ploit one's gifts' In fact, the reinterpretation process can be shown to be real when the item in question (in its secondary reading) comes to be used in isolation from its original idiomatic context. This has happened with spek in met spek schieten, which has led to the compound spekverkoper, as in (11) below. The formation of the compound noun can only be explained if it is accepted that spek in its idiomatically contextual reading 'boasting speech, tall tale' has been isolated from the original idiom. S'pekverkoper as such cannot be directly explained as an original metaphor of its O\\Il1: there is no way in which selling bacon can be associated with boasting except through the intem1ediary of met spek schieten. This type of 'semantic back-fommtion' is also quite visible in the case of cranberry morphs. When people are asked for an interpretation of heinde in van heinde en verre, it appears that a majority understands the item as a synonym of near-synonym ofverre (more or less like the relationship between far and wide in the English counterpart of the expression). Etymologically speaking, however, heinde and verre are antonyms rather than synonyms; heinde is related to hand and basically means 'what is near, what is at hand, what can be found in the immediate neighborhood'. When the etymological relationship with hand and the semantic relationship with nearness is lost, however, the overall meaning 'from everywhere' of the idiomatic expression enables heinde to be reinterpreted as a synony111 of verre. Met zijn talenten woekeren is an even clearer case. Whereas talent in its original biblical context referred to a particular coin, the reading 'personal ability, capacity, aptitude' that it received in the figurative interpretation of the expression is now the major one; it occurs freely in separation from the original expression. From a very general perspective, reinterpretation processes such as these? indicate that a search for isomorphism (defined as syl1tagmatic transparency) is an active force in the mind of the language user. Apparently, isomorphism on the figurative level of the idioms is not just real when it is given on the basis of the literal meanings of the constituent elements of the expression, but it is also real in the sense of being sought for when it is not given. If this can be accepted, it also means that interpretation processes are not always bottom-up, but that they can also be top-down: the overall meaning of met zijn talenten woekeren determines the specific meaning of talent that has become the item's major meaning.

210

Constructions and idioms

(12)

Vi/en naar Athene dragen 'Carry owls to Athens/Athena > do something irrelevant, useless. superfluous' De kat de bel aanbinden 'Tie the bell to the cat> bring something out into the open, make something public, ring a bell about something'

Similarly, evidence can be found that the search for paradigmatic motivation too is a real one. Of course, the search for motivation is quite outspoken when etymological researchers disagree on the actual motivation of an expression, as in uilen naar Athene dragen, which is usually interpreted in terms of the owl as the symbol of the goddess Athena and the city Athens, but which some see as being motivated by the simple fact that there were many owls in Athens. But etymologists are professional searchers for motivation, so their interpretative creativity does not tell us too much about the spontaneous occurrence of such interpretative activities when ordinary language users use idioms. More important in this respect are, first, psycholinguistic investigations of the type reported on in Gibbs (1990) and later work, which show that motivating images for idioms are psychologically real. And second, evidence for actual reinterpretations along the paradigmatic axis may be found, i.e. evidence for new meanings coming about through the search for motivation. For instance, the older idiomatic meaning of de kat de bel aanbinden (viz. 'to bell the cat, to take the lead in a dangerous activity') refers to the old fable of the cat and the rats. Nowadays, however, it seems to be shifting towards the interpretation 'to bring something into the open, to make something public, to ring a bell about something': on the one hand, the older association with taking the responsibility in a dangerous action (in favor of other people) disappears into the background; on the other hand, the notion of drawing the public attention to something (in particular, something scandalous or negative) is foregrounded. Given that de kat de bel aanbinden is largely unmotivated for most speakers, the association between the bell referred to in the expression and the notion of making something public (of making it heard, that is) enhances the motivated character of the idiom. The search for greater motivation leads to a shift in the interpretation. 2.5. Summing up What I have tried to indicate in the previous pages can be summarized in three points. First, an adequate description of the various forms of semantic specialization that occur in composite expressions requires that a number of distinctions are taken into account: the distinction between the syntagmatic and the paradigmatic aspects of meaning (which can both be found on the

The interaction o.f metaphor and metonymy

211

level of the literal meaning and on that of the figurative meaning of the idiom), and the distinction between bottom-up and top-down semantic processes. Specifically, the concepts of isomorphism (defined as syntagmatic transparency) and motivation (defined as paradigmatic transparency) are of primary importance for describing the semantics of composite expressions. Second, semantic specialization is a matter of degree. More specifically, the classificatory framework defined on the basis of the conceptual distinctions just mentioned allows for a ranking of the degree of specialization involved. Least specialized are fully isomorphic and fully motivated cases like parels voor de zwijnen gooien. Somewhat less specialized are cases like de koe bij de horens grijpen and met de handen in het haar zitfen, which may not be derivable on a word-per-word basis, but which are entirely transparent along the upper side of the prismatic structure. Still further down the line, we find cases like met spek schieten and de kat de bel aanbinden, that lack the global motivation of the previous examples. Most specialized, finally, are cases where the literal meaning of the expression cannot even be recovered, such as iemand de loeI afsteken and other idioms containing cranberry morphs. In each of these cases, matters are further nuanced by the existence of degrees of motivation. And third, semantic interpretation is not just a question of bottom-up compositionality or literal-to-figurative transfer. The reinterpretation processes that can be observed point to the existence of top-down and figurativeto-literal interpretations. It is not just the case that literal meanings determine figurative ones; figurative meanings also determine literal ones. And it is not just the case that the meaning of the parts determines the meaning of the whole; the meaning of the whole also determines the meaning of the parts.

3, Interactions between metaphor and metonymy in composite expressions

3.1. The metaphor/metonymy continuum If there is a continuum between metonymy and metaphor, this implies that there are in-between cases between expressions that are fully metonymical and expressions that are fully metaphorical. Composite expressions as well can be fully metaphorical or fully metonymical, when the motivational links that are present within the semantic architecture of the expression are only

212

Constructions and idioms

metonymical or only metaphorical. Pare!s voor de zwijnen gooien, as discussed above, is fully metaphorical: the top level shift from 'to throw pearls at swine' to 'to present unworthy people with valuable things' is a metaphorical one, and so are the bottom level shifts from pare! 'pearl' to 'valuable thing',8 and from zwijn 'pig' to 'unworthy person'. Conversely, compounds of the bahuvrihi type, like roodhuid 'redskin', are well-known cases of metonymical compounds: the link between the initial, compositional meaning ('red skin') and the derived reading ('Indian, seen as one with a red skin') is metonymical link of the possessed/possessor type. In the roodhuid case, to be sure, motivational links at the bottom level fail, because the derived reading 'redskin' cannot be considered isomorphic. So, given that composite expressions can be either metaphorical or metonymical, how can we chart the in-between cases? How do metaphor and metonymy occur in mutual combination in compounds and idioms? I will argue that there are three basic cases to be distinguished: cases in which metaphor and metonymy occur consecutively, cases in which they occur in parallel, and cases in which they occur interchangeably. In the following sections, each of these cases is presented separately. (More, and more intricate, examples of the interaction between metaphor and metonymy in expressions may be found in Gevaert 1994 and Feyaerts 1997.) 3.2. Consecutive interaction of metaphor and metonymy A consecutive interaction between metaphor and metonymy occurs when one of the motivational links in the semantics of the composite expression involves a sequence of two semantic extensions. A first example is presented in Figure 2, which contains an analysis of the compound schapenkop. Literally, the word means 'sheep's head' (and the word could actually be used in this sense, in contrast with some of the other compounds that we will analyze presently, in which the literal reading is not conventionalized). The derived reading of schapenkop is 'dumb person', and this reading seems to involve two steps: first, 'sheep's head' is metaphorically extended towards the reading 'a (human) head like that of a sheep, a stupid head', and second, a metonymical step leads to 'a person with a head like that of a sheep, a stupid person'. (The representation 9 in Figure 2 can be completed on the bottom level of the prism, but that is a step that will be taken in section 3.3.)

The interaction ofmetaphor and metonymy

METAPHOR

--8---

213

METONYMY

1 Sheep's head 2 Sheep 3 Head 4 (Human) head like that of a sheep 5 Stupid person

Figure 2. The prismatic structure of schapenkop

A similar consecutive combination occurs in an idiomatic expression such as groen achter de oren zien. Literally, the reading is 'to be green behind the ears', which is then metaphorically interpreted as 'to be young, (as if people are like fruit that have a green color in the first stage of their existence, before they reach maturity). As with so many other expressions indicating young age, the expression next receives a further extension to the reading 'inexperienced, naive'. We may also note that the consecutive sequence need not always involve an alternation of metaphor and metonymy. In an example like hanglip, for instance, two consecutive metonymical steps may be identified. The literal reading is composed of the noun lip 'lip', and the verbal stem hang 'to hang'; the literal reading can therefore be paraphrased as 'hanging lip'. A first metonymical extension (involving the metonymical relationship between a specific feature and the bearer of that feature) leads to 'a person with a hanging or protruding (lower) lip'. A second metonymical extension (involving the metonymical relationship between a typical effect and the usual cause of that effect) leads to 'an unhappy, sulky, pouting person' .

214

Constructions and idioms

3.3. Parallel presence of metaphor and metonymy A parallel presence between metaphor and metonymy occurs when there is a difference in type among the different motivational links that occur in the semantics of a composite expression. If, for instance, the motivational link at the top level of the prismatic structure is metaphorical and one of the links at the lower level is a metonymical one, metaphor and metonymy work in parallel to produce the derived reading of the expression, or at least, both a metaphorical and a metonymical path could be reconstructed to arrive at the derived reading.

1 Sheep's head 4 Sheep-like 7 Stupid

2 Sheep 5 Head 8 Person

3 Head 6 (Human) head like a sheep 9 Stupid person

Figure 3. The prismatic structure of schapenkop completed

As an initial step, let us complete the picture for schapenkop. At the bottom level of the prismatic structure, the intermediate reading 'a (human) head like that of a sheep, a stupid head' can be considered isomorphic: the kop-part does not undergo a semantic change, and the 'sheep-like' reading is related by metaphorical similarity to the original reading of schaap 'sheep'. The ultimate reading 'stupid person' maintains the isomorphism: there is a metaphorical link from 'sheep-like' to 'stupid', and there is a conventional part/whole-metonymy linking 'head' to 'person'. (For instance, de koppen

The interaction a/metaphor and metonymy

215

tellen 'to count the heads' is a conventional expression for 'counting the individuals in a group'.) The example shows that the motivating links at the bottom level of the prismatic structure need not be the same as the ones at the top level: the identity link is absent at the upper level. In some cases, then, we get an alternation between metaphorical and metonymical links. A case in point is the expression in de stront zitten, which may be analyzed as in Figure 4.

1 To sit in the shit 4 To be in great trouble

2 To sit (in) 5 To be situated (in)

3 Shit 6 Trouble, unpleasantness

Figure -I. The prismatic structure of in de stront zitten

The top level shift is a metonymical one of the cause/effect-type: if you are literally surrounded by excrements, you are typically in an unpleasant, troublesome situation. At the bottom level, however, the motivational links are of a metaphorical type. Zitten 'to sit' is a conventional metaphor for 'to be characterized by, to experience': dat zit goed 'that sits well' means as much as 'that is okay', and in moeilijkheden zitten 'to sit in difficulties' equals 'to have, to experience difficulties'. Stront is likewise a conventional expression for anything extremely nasty. As a slightly more complicated example, let us consider droogkloot 'boring person, bore', which can be analyzed as in Figure 5. The compositional literal reading 'dry testicle' is the basis for a roodhuid-type extension, yielding the possessive compound 'person with dry testicles'. This reading, however, is itself the input for a further metaphorical extension, leading to the 'boring person' sense. At the same time, the derived reading is isomorphic:

216

Constructions and idioms

the aspect 'boring' correlates with the 'adjective droog, and the aspect 'man, person' correlates with the noun kloot. Such an isomorphic analysis is further supported by the motivational link between the initial and the derived readings of the constituent parts. Droog is in fact conventionalized in the reading 'boring, dull, dreary' (een droge klaas is 'a tedious fellow', where klaas is originally a proper name), and kloot is a conventional derogatory expression for 'man'.

1 Dry testicle 4 Person with dry testicles 7 Unworthy man

2 Dry 5 Boring person

3 Testicle 6 Lifeless, dull

Figure 5. The prismatic structure of droogkloot

3.4. Interchangeability of metaphor/metonymy analyses The prismatic model 10 implies that the meaning of composite expressions may sometimes be construed in two different ways: from bottom to top and then from front to back at top level (which is undoubtedly the standard pathway) or conversely from front to back at bottom level, and then from bottom to top. The 'non-uniqueness of semantic solutions' (to borrow the words introduced by Nunberg 1979 in a slightly different context) goes even further, if we can show that different prismatic analyses may - with equal or near-equal plausibility - be construed for one and the same expression. To

The interaction ofmetaphor and metonymy

217

the extent that such alternatives involve different configurations of metaphorical and metonymical motivational links, we shall say that we have interchangeable metaphor/metonymy. Compounds pointing in this direction are the following. Badmuts literally means 'swimming cap' but is also used jocularly for a bald person. The shift can be analyzed in two ways. Either 'swimming cap' leads metonymically to 'a person with a swimming cap' and from there by metaphorical similarity to 'a person who looks as ifhe was wearing a swimming cap, a bald person'. Or 'swimming cap' is directly metaphorized as 'a head that looks as if it is covered by a swimming cap, a bald head' and from there metonymically to 'a bald-headed person'. The reconstruction of the semantic process can go either way, and there is no principled way to favor one analysis over the other. An example with an idiom rather than a compound is over de rooie gaan 'to go over the red one', in which 'the red one' is a red mark on a gauge indicating the point of maximal pressure. The derived reading 'to explode with anger' may be arrived at in either of the following ways. To go beyond the point of maximal pressure is the metonymical cause of a material, literal explosion (as when a boiler explodes), and the literal explosion can then be metaphorically used to conceptualize an emotional outburst. Conversely, the notion of crossing the point of maximal pressure may be directly metaphorized into the psychological domain ('go beyond the point of maximal emotional strain'), and this event may then causally (i.e. metonymically) lead to an outburst. It should be clear that the alternatives need not always involve different sequences of metaphors and metonymies. It may also be the case, for instance, that two metonymies occur in alternative orders. Zultkop provides us with an example. Literally, it means 'head filled with or made from brawn'; the derived reading is again 'stupid person'. The consecutive steps could be from the literal reading to 'stupid head' via a metonymicallink (the presence of brawn rather than brain is the cause of the idiocy), and from there to 'stupid person' via another metonymical link of the part/whole-type. Or the sequence might involve an initial part/whole-metonymy producing 'a person with a head full of brawn' and hence to 'stupid person' through the intermediary of the effect/cause-metonymy.

218

4.

Constructions and idioms

Metaphtonymy and prismatic semantics

Within the context of recent metonymy studies, there is an obvious link between the phenomena described above and the notion of metaphtonymy introduced by Louis Goossens in 1990. The two types of metaphtonymy (i.e. interaction between metaphor and metonymy) distinguished by Goossens resemble the two basic interaction types identified above: what Goossens calls metaphor from metonymy refers to a sequential operation of the two mechanisms that can be linked to the consecutive type described in section 3.2, and what Goossens calls metonymy within metaphor/ metaphor within metonymy can be associated with the simultaneous, parallel type of interaction described in section 3.3 of the present article. But how far exactly does the correspondence go? Goossens introduces metaphor from metonymy by referring to the polysemy of giggle. The verb initially means 'to laugh in a nervous way', but this meaning can be used metonymically in a context like 'Oh dear', she giggled, 'I'd quite forgotten', in which giggle comes to mean 'say while giggling'. A further extension towards 'to say as if giggling' then constitutes the 'metaphor from metonymy' reading. Whereas the consecutive operation of a metonymical and a metaphorical shift links up with the cases discussed in section 3.2, it will also be clear that the approach in the present paper has a wider scope than Goossens's. We have identified not just successions of metonymies followed by metaphors, but we have illustrated a larger variety of sequences: metaphors followed by metonymies, metonymies followed by metonymies, etc. From a broader point of view, it is important to realize that neither our 'consecutive interaction of metaphor and metonymy' nor Goossens's 'metaphor from metonymy' can be considered real innovations in the context of lexical semantics. The recognition that mechanisms of semantic extension such as metaphor and metonymy may operate in succession (and in fact, in series with multiple steps) is a natural and time-honored one in diachronic semantics (cp. Geeraerts 1997a). What is being added to that idea in the prismatic model described above, is precisely the importance of a second dimension for an adequate description of composite expressions. As to Goossens's 'metonymy within metaphor', it involves cases like catch someone's ear 'ensure someone's attention'. Such examples (which invariably involve idiomatic expressions rather than single lexemes) receive a straightforward interpretation in the context of the model sketched in the present paper, as can be gathered from the analysis in Figure 6.

The interaction ofmetaphor and metonymy

219

1 Take hold of someone's organ of hearing 2 Take hold of, capture 3 Organ of hearing 4 Force to listen, obtain the attention 5 Attain, obtain 6 Attention Figure 6. The prismatic structure of to catch someone's ear

The literal meaning 'take hold of someone's organ of hearing' is metonymically extended to 'to obtain someone's attention'. Goossens basically sees a metaphorical shift at the level of the expression as a whole, but at the same time allows for a metonymic interpretation. In the context of the prismatic model, the metonymic interpretation would seem to be more plausible: materially taking hold of someone's ear is metonymically conceptualized as a cause (or at least, a contributing factor) for getting someone's attention. At the same time, there is indeed a metaphorical aspect to the expression, but it involves the development of catch at the bottom level of the two-dimensional structure: the verb undergoes a metaphorical shift from a material to an immaterial reading. To be sure, non-uniqueness surfaces again, to the extent that it could also be said that obtaining something is the result of taking hold of it (and of course, the relationship between action and result, or cause and effect, is a metonymical one). Ear, finally, is metonymically linked to the notion of attention: the hearing organ is one of the media for channeling a person's attention. In the light of this analysis, the advantage of the prismatic model can be defined as follows: it draws the attention to the fact that the more specific

220

Constructions and idioms

semantic development is not restricted to one of the constituent parts (as might be suggested by Goossens's examples) but actually has to be determined for all of them. At least for composite expressions, then, the prismatic model appears to have a wider scope than the notions introduced by Goossens, specifically because it allows for other sequences than just metaphor from metonymy. More importantly, the model combines the intuitions behind 'metaphor from metonymy' and 'metonymy within metaphor' / 'metaphor within metonymy' by bringing the two relevant aspects of the development of composite expressions together: on the one hand, the semantic development of the expression as a whole, on the other, the role of the constituent parts of the expression and their independent development. The model thus allows for a uniform and more detailed description of the semantics of composite expressIOns.

Notes 1.

2.

In the Saussurean tradition, syntagmatic relations involve associations between linguistic expressions that exist in presentia, whereas paradigmatic relations involve associations that exist in absentia. In an expression like dames en heren, for instance, the association between dames and heren is realized in the expression dames en heren itself. The semantic association between dames 'ladies' and vrouwen 'women', on the other hand, exists even if it does not show up explicitly in the expression being used as such. Paradigmatic relations may be of various sorts; they do not just include semantic associations of the type just mentioned, but also morphological relations between a lexical base and the derivates or compounds in which it features. Among the semantic paradigmatic associations, metaphor traditionally features prominently (see among others Jakobson 1971: 74); note that in this case, the association exists not beween to different words, but between two readings of the same word. The paradigmatic relations that will be envisaged in this paper are precisely of the kind illustrated by metaphor: semantic associations between different readings of one linguistic expression. Apart from metaphor, the relevant associations involve semantic relations like metonymy, generalization, and specialization. The bottom-up interpretation is present in quotations like the following: 'Whatever linguistic meaning is, there must be some sort of compositional account of the interpretation of complex expressions as composed from the interpretations of their parts and thus ultimately from the interpretations of

The interaction ofmetaphor and metonymy

3.

4.

221

the (finitely many) simple expressions contained in them and of the syntactic structures in which they occur' (Chierchia and McConnell-Ginet 1990: 6). In most work in the tradition of Cognitive Semantics, the concept of motivation is used in a slightly broader way than the way in which it is defined here. In Lakoff (1987) and related work, for instance, 'motivation' involves the principles that explain (or make plausible) why a particular linguistic expression means what it does. The concept is explicitly introduced as an alternative to the more traditional notion of predictability: even if meanings are not entirely predictable, they may be motivated by existing tendencies and schemata. The distinction that is drawn here between 'motivation' and 'isomorphism' tries to be more specific about the general concept of motivation by distinguishing between its syntagmatic and its paradigmatic form. The concept of 'isomorphism', on the other hand, links up with existing work within the Cognitive tradition relating to the iconicity of grammar (see e.g. Haiman 1980b). Isomorphism as used here is a form of iconicity to the extent that features of meaning (in particular, its complex nature) are reflected by features of the linguistic form (viz., its composite nature). The question might be asked how we can put koe as interpreted in the figurative context into correspondence at all with koe as interpreted in the literal plane. On the one hand, we would still be willing to say that koe 'problem' corresponds with koe 'cow'. But on the other hand, there is no associative semantic link from 'cow' to 'problem'. So can we say at all that koe maps onto 'problem'? What is there to stop us from mapping bi} de horens vatten onto the 'problem' part of the idiomatic reading of the expression? In this particular instance, of course, the link between vatten 'to seize (literally)' and the figurative reading 'to tackle' is not unmotivated, and hence, byelimination. koe is easily mapped onto 'problem'. But even if this paradigmatic link between both interpretations of vatten were to be just as non-transparant as that between 'cow' and 'problem', the syntactic structure of the expression (as interpreted literally) would favor a figurative interpretation of koe as a noun, and one of vatten as a verb. This would seem to lead to the conclusion that there is always some paradigmatic link at the bottom of the prism between the literal readings of the constituent items and their figurative interpretation: at the very least, the literal reading would motivate the figurative reading because the latter is consistent with the word class of the former. In principle, such a weak form of motivation can be accounted for by accepting degrees of motivation: it will be made clear further on in the text that this is a useful step to take in any case. Empirically speaking, however, it remains to be seen whether figurative readings are always consistent with the word class of the constituent elements of the literal expression. (Notice that at least in the realm of morphology, reinterpretative processes may violate the initial

222

5.

6.

7.

8.

Constructions and idioms

syntagmatic structure: the isomorphically metanalytic reanalysis of hamburger as ham + burger violates the initial formal structure.) The representation in Figure 1 might be adapted to take into account degrees of motivation by drawing thicker or thinner lines, or by attaching a numeric weight to them. The reinterpretation processes mentioned in this paragraph constitute one kind of proof of the cognitive reality of the semantic structure embodied in the prismatic model of Figure 1. Other kinds of support for the validity of the model will have to be explored in further research. Two main alternatives have to be envisaged. First, psychological investigations (involving on-line processing tasks, or on the basis of questionnaires) may be invoked to establish the psychological reality of a specific analysis. Second, synchronic linguistic phenomena (rather than the diachronic reinterpretation processes mentioned here) may point to the structural importance of the model. Consider, for instance, the possibility of incorporating anaphoric elements in the expression. As a working hypothesis, it would seem that only those idioms that are isomorphic on the figurative level allow for the introduction of anaphoric demonstrative pronouns referring to a previous instantiation of the figuratively interpreted concepts. In this way, it is quite plausible to have a sequence like the following: 'Then came the problem of formatting the text according to the style sheet. To take this bull by the horns appeared to be much more difficult than applying the revisions required by the editors'. Because bull maps onto 'the problem' that is to be tackled, this may be introduced to refer to a previous identification of that problem. In the case of met de handen in het haar zitten 'to sit with one's hands in one's hair> to be at one's wit's end', however, the absence of a clear interpretation for handen in the idiomatic context makes sequences like the following: 'Toen moesten er camera-ready kopieen van de figuren gemaakt worden. Met deze handen in het haar te zitten bleek veel erger dan het schrijven van het oorspronkelijke artikel geweest was (Then came the problem of producing camera-ready figures. To sit with these hands in one's hair appeared to be much more taxing than writing the original paper had been)' rather implausible. This is, to be sure, just an example of the type of phenomena to be studied, but it illustrates how additional evidence for the linguistic reality of the prismatic model may be sought. An example of a reinterpretation of compounds is provided by the element scharrel-, the verbal stem of scharrelen 'rummage about, scratch, scrape'. From the compound scharrelkip 'free-ranging chicken' it is extrapolated to compounds like scharrelei 'an egg ofa free-ranging chicken': scharrel- is reinterpreted as 'produced by biological farming'. There might be some discussion with regard to this specific case: it could perhaps also be considered a generalization. Examples of competing and in-

The interaction ofmetaphor and metonymy

223

terchangeable motivations will be discussed explicitly further on in the article. 9. In Geeraerts and Bakema (1993), the prismatic representation of compounds looks slightly more complicated than in the representation used here. For instance, the compositional reading 'sheep's head' would be analyzed as a specialization of a more general reading 'head having something to do with sheep'. The motivation behind this approach is the underdetennination of compounds. Schapenkop could in fact mean many things: a sheep's head, but also 'a head with a talent for or a specific interest in sheep' (just like a studiekop is 'a bright head, a head with a talent for of a specific interest in study'). The construction of nominal compounds in Dutch does not formally differentiate between the possessive reading and the alternatives; by contrast, the syntactical construction of idiomatic expressions is much more specific as to the semantic role of the constituent parts. This underdetermination of a compound XY can be expressed by merely defining the initial compositional reading as 'a Y that has something or other to do with an X'. The next step is then invariably a semantic specialization yielding specific readings like 'a sheep's head' or 'a head with a talent for or a specific interest in sheep'. In the present article, this complication at the front end of the prismatic diagrams for compounds has not been included, because it does not add very much to the line of thought that is relevant in this text. 10. It may be useful to point out that the 'prismatic model' is a model precisely because it involves a certain degree of abstraction: the semantics of the composite expressions is rendered in a schematic way (in the sense, for instance, that the number of constituent elements is systematically reduced to two).

Section 4 Meaning and culture

Chapter 9 Looking back at anger. Cultural traditions and metaphorical patterns

Originally published as Dirk Geeraerts and Stefan Grondelaers, 1995, in John Taylor and Robert E. MacLaury (eds.), Language and the Construal o[the World 153-180. BerlinJNew York: Mouton de Gruyter. This paper addresses the question of the influence of the Medieval doctrine of the four humors and the four temperaments on our contemporary vocabulary. Specifically, given the psychological part of the humoral theory, is there any way in which the influence of the theory can be felt in the way emotions are being talked about today? Concentrating on the concept of anger, the paper shows that there is. By taking a closer historical look at Kovecses' analysis of expressions for the emotional concept 'anger' (1989), it is argued that his a-historical method obscures the possible role of cultural traditions as a source of emotion concepts. In the context of the development of Cognitive Linguistics, the paper helped to reveal the possible tension between a universalist and a culture-specific interpretation of the 'experientialist' nature of Cognitive Linguistics. Cognitive Linguistics, by its very cognitive nature, has a tendency to look at language from a psychological point of view, i.e., at language as (part of) the organization of knowledge in the individual mind. But if cognition is embodied, and if our bodies are basically the same. then underlying cognitive patterns will be largely the same for all. In Kovecses' analysis of anger metaphors in terms of the ANGER IS HEAT pattern, the universalist assumption is that the pattern is physiologically motivated. However, if the experientialist nature of language and cognition includes the cultural and historical 'situatedness' of human experience, cultural and historical factors are likely to influence our cognitive patterns. That is precisely what the present paper tries to show. In more recent publications, Kovecses (2000. 2002) accepts the point that cultural determination may play a role in shaping cognitive patterns, and investigating the universality (or the cultural specificity) of metaphorical patterns has now become a standard topic in metaphor oriented cognitive linguistic research. Such studies focusing on anger include Taylor and Mbense (1998), Yu (1998), Mikolajczuk (1998), Forceville (1999), Barcelona (2001), Gevaert (2001), Harkins and Wierzbicka (2001), Soriano (2003), and Maalej (2004). More generally, there is a growing interest in Cognitive Linguistics in the social aspects of meaning and cognition. A number of researchers (Palmer 1996, Sinha and Jensen de L6pez 2000, Harder 2003. Itkonen 2003. Tomasello 2003. Geer-

228

Aleaning and culture

aerts 2005 and others: see Kristiansen and DilVen, forthcoming) emphasize that the experientialist nature of Cognitive Linguistics does not only refer to material factors (taking a notion like 'embodiment' in a physical and physiological sense) but that the cultural environment and the socially interactive nature of language should be recognized as primary elements of a cognitive approach. Further, laying the foundations for a true cognitive sociolinguistics, variational phenomena are being studied empirically in work such as Kristiansen on phonetic variation (2003), Berthele (2004) on differences in syntactic construal between dialects, and Grondelaers (Grondelaers 2000) on grammatical phenomena whose distribution is determined by a combination of internal (structural or semantic) and external (contextual or sociolinguistic) factors. (See also the work mentioned in the introduction to Chapter 11.)

1.

The mysteries of masturbation

In the course of 1989, the Belgian Department of Education started a school campaign against truancy. The major slogan of the campaign read Van spijbelen word je doof 'Playing truant makes you deaf - a jocular (but probably inefficient) reference to the old belief that excessive masturbation could cause deafness. It is not likely that this belief itself is still very much alive in our post-sexual revolution, sex education era, but the very fact that knowledge of it was assumed in the campaign seems to indicate that it is still around, and that it was being handed down from educators to pupils not too long ago. But what was the origin of that belief (which, incidentally, came in a number of variants, in the sense that next to deafness, blindness and deterioration of the spinal marrow were cited as the sinful results of promiscuous self-indulgence)? Was it just a conspiratorial invention of priests and parents, intended to keep personal frustration and public morality up? Or was there an actual basis for it? Let us turn to a specialist for an answer. In 1772, the honorable doctor Tissot (member of learned societies in London, Basel, Bern, and Rotterdam) published a lengthy treatise entitled L 'Onanisme. Dissertation sur les Maladies produites par la Masturbation (Grasset, Lausanne), in which all is revealed about this 'crime obscene'. After an extensive treatment of the detrimental influence of masturbation, he asks the question 'Col11l11ent une trop grande emission de semence produitelle tous les maux que je viens de decrire?' [How does an excessive emission of sperm produce all the evils that I have just described?]. On page 69, he begins his answer with a reference to the father of medicine, Hippocrates of

Looking back at anger

229

Kos (approx. 460-377 BC). (The relevant passages are in the Hippocratic treatise known as De Genitura.) Hippocrate a cru qu'elle [la semence] se separoit de tout le corps, mais surtout de la tete. La semence de l'homme vient, dit-il, de toutes les humeurs de son corps, elle en est la partie la plus importante.... Il Ya des veines & des nerfs qui de toutes les parties du corps vont se rendre aux parties genitales; quand celles-ci se trouvent remplies & echauffees, elles eprouvent un prurit, qui se communiquant dant tout le corps, y porte une impression de chaleur & de plaisir; les humeurs entrent dans une espece de fermentation, qui en separe ce qu'il y a de plus precieux & de plus balsamique, & cette partie, ainsi separee du reste, est portee par la moele de l'epine aux organes genitaux. [Hippocrates thought that semen secreted itself from the entire body, but specifically from the head. A man's semen, he says, comes from all the humors of his body, of which it is the most important component. ... There are veins and nerves that go towards the genital organs from all over the body, and when these organs are filled up and wanned up, they experience an urge that communicates itself through the entire body, producing an impression of warmth and pleasure. The humors then enter into some sort of fermentation that separates out the most precious and balsamic substance they contain, and this part, when it is separated from the rest, is carried to the genital organs by the spinal marrow.] If, in other words, semen is produced by a process of fermentation and distillation of the 'humors', it is plausible that an overproduction of sperm weakens the body, given the vital importance of those 'humors'. Specifically, the role of the head and the spinal cord in this process explains why, in particular, the functions of the head (such as seeing and hearing), and the spinal marrow may suffer the detrimental effects of excessive sperm production. In this sense, it all falls it into place - but it only does so against the background of the doctrine of the four humors. In Classical and Medieval physiology, the human body was thought to contain four kinds of fluids, or humors, which regulated the body's functioning and whose disproportionate presence could cause illness. Admittedly, conflicting views were expressed within the humoral framework about the origins of sperm, and the humoral doctrine as a whole was no longer valid in its original Hippocratic form by the end of the eighteenth century. But, although Tissot hardly follows Hippocrates in detail, he quotes him approvingly, and like most of his contemporaries does retain the basic idea that the production of semen (and the harm-

230

Meaning and culture

ful effects of overproduction) involves the extraction of the seminal substance from the vital bodily humors. Against the background of this historical link, extending from classical antiquity well into the modem era, the lingering belief in the negative effects of masturbation appears to be a recently deceased (or at least moribund) renmant of what was once solid science. And to be sure, it is not the only relic. In the Dutch-speaking part of Belgium, for instance, the concept 'cold, inflammation of the mucous membrane of nose and throat' is expressed in a majority of the dialects by the word valling. While valling is morphologically complex (being a nominalization of the verb vallen 'to fall '), it is not semantically transparent to the majority of speakers. Historically speaking, however, its formal complexity makes perfect sense in the framework of the theory of humors: when the nose runs or sputum is expectorated, what 'falls' is nothing else than phlegm, one of the four humors. Given that the head was considered to be the major locus of phlegm (and taking into account that phlegm was described as a cold humor, in contrast with, for instance, blood as one of the warm humors), a valling or a cold is nothing but a precipitation of the brain's fluid. And of course, in non-historical parlance, the English word phlegm now refers precisely to the thick semi-fluid secretion of the mucous membranes of the respiratory passages. Similarly, we speak of catarrh (derived from the Greek katarrheo 'to flow down'), and - in the case of another disease attributed to an excess of phlegm - of rheumatism, in which the Greek verb rheo 'to flow' can be discerned (cp. Siegell968: 323). Faced with examples such as these, we would like to address the question of the influence of the humoral doctrine on our contemporary vocabulary more systematically. What other relics of the old beliefs can we find? Specifically, given the psychological part of the humoral theory, is there any way in which the influence of the theory can still be felt in the way we talk about emotions? Concentrating on the concept of anger, we will try to show that there is. By taking a closer historical look at Kovecses' analysis of emotional expressions in terms of generalized metaphors (1989), we shall argue that his a-historical method obscures the possible role of cultural traditions as a source of emotion concepts. Our purpose, in other words, is factual to the extent that we will try to establish the importance of the old humoral theory for our contemporary emotional vocabulary, critical to the extent that we will try to qualifY Kovecses' analysis, and methodological to the extent that we will stress the methodological importance of a diachronic perspective for linguistic studies with a cultural orientation. Before dealing with the

Looking back at anger

231

specific linguistic part of the investigation, we will first give some more information on the theory of humors and its historical importance.

2. The history of the humors In this section, we will briefly (and simplifyingly) present the humoral doctrine, and sketch its historical development. On various aspects of the history of medicine at large and the humoral theory in particular, more information may be found in among others Lindeboom (1985), Godderis (1988), Beek (1969), Irwin (1947), Siegel (1968), Major (1954), Schafer (1966), and Diepgen (1955). Klibansky, Panofsky and Saxl (1964) deserve to be mentioned separately for their detailed history of the humoral doctrine up to the seventeenth century (with special emphasis on the concept of melancholy). The foundations of the humoral doctrine were laid by Hippocrates of Kos. Three aspects of his approach should be mentioned: the physiological, the psychological, and the medical. Physiologically, the four humoral fluids regulate the vital processes within the human body; the secretion of the humors underlies the dynamical operation of our anatomy. Psychologically, on the other hand, they define four prototypical temperaments, i.e. a person's character is thought to be determined by the preponderance of one of the four vital fluids in his body. Thus, the choleric temperament (given to anger and irascibility) is determined by a preponderance of the yellow bile, while the melancholic, gloomy and fearful, suffers from a constitutional excess of black bile. The phlegmatic personality is typically placid and unmoved, while the sanguine temperament (defined in correlation with blood, the fourth humor) is passionate, optimistic, and brave. The singular combination of physiological and psychological concepts that characterizes the theory of humors also shows up in the fact that a disequilibrium of the fluids does not only characterize constitutional temperaments, but also causes temporary diseases - which are then typically described in bodily, biological terms as well as in psychic terms. For instance, an overproduction of yellow bile may be signaled by the patient's vomiting bile, but also by his dreaming of fire. In the same line, an excess of blood shows up in the redness of the skin and swollen veins, but also in carelessness and a certain degree of recalcitrance. In this sense, the humoral theory is a medical doctrine: it identifies diseases and their symptoms, and defines a therapy. Obviously, the basic therapeutic rule will be to restore the balance of the humors, given that a disturbance of their well-balanced proportion is the basic cause of the pathological situa-

232

Meaning and culture

tion. The long-lasting popularity of blood-letting, for instance (a standard medical practice that continued well into the nineteenth century) has its historical origins in the theory ofhumors. The connection between yellow bile and fire that was mentioned a moment ago is not accidental. It is part of a systematic correlation between the human, anatomical microcosm and the macrocosm, thought to be built up from four basic elements. Thus, yellow bile, black bile, phlegm, and blood corresponded with fire, earth, water, and air respectively. In the Aristotelian elaboration of the Hippocratic doctrine, a 'componential analysis' was added to these correlating sets of microcosmical and macrocosmical basic elements. They were defined, in fact, as combinations of four basic features: cold, warm, wet, and dry. (Needless to say, these four features are themselves related along two dimensions.) Blood was thought to be warm and wet, phlegm cold and wet, yellow bile warm and dry, and black bile cold and dry. The classical humoral doctrine received the form in which it was to dominate the Middle Ages in the work of Galen (129-199). His incorporation of the humoral approach into an encompassing theory of the human digestive system is of particular interest. Galen distinguishes between three successive 'digestions'. In the first digestive process, food is transformed into chyle in the stomach; the residue of this first digestion is faeces. In the second step, the humoral fluids enter the picture. For instance, by the transformation of chyle in the spleen, black bile is produced, while the liver refines chyle into blood; the residue of the second digestion is urine. The third step takes the blood and carries it through the body, sustaining the growth of the body; the residue of this third digestion is perspiration. But while the substance that ensures the growth and maintenance of the body is known as the nutrimental spirit, there are also two other spirits to be taken into account in this third step. In a continuing and cumulative refining process, the heart produces the vital spirit (which regulates the temperature of the body and controls the passions), and the brain produces the animal spirit (Aristotle's pneuma psychicon, which commands the movement of the body, but also feeling and the workings of the mind). Further, Galen's digestive anatomy leads to a dietary pharmacology. All plants (and foodstuffs in general) could be characterized by one of four degrees of warmth, cold, wetness, and dryness. Given that diseases are caused by an excess of one of the four humors, and given that these are themselves characterized by the four features just mentioned, the basic therapeutic rule is to put the patient on a diet that will ensure a decrease of the superfluous

Looking back at anger

233

humor. For instance, because yellow bile is hot and dry, patients suffering from choleric diseases should avoid plants such as garlic and ginger, which are both warm in the fourth degree and dry in the fourth degree. Rather, they should resort to plants such as opium (wet in the fourth degree) and henbane (cold in the fourth degree). Table 1. A system of humoral correspondences

CHARACTERISTIC ELEMENT TEMPERAMENT ORGAN COLOR TASTE SEASON WIND PLANET ANIMAL

PHLEGM

BLACK BILE

YELLOW BILE

BLOOD

cold & moist water phlegmatic brain/bladder white salty winter North moon turtle

cold & dry earth melancholic spleen black sour autumn West Saturn sparrow

warm & dry fire choleric liver/stomach yellow bitter summer South Mars lion

warm & moist air sanguine heart red sweet spring East Jupiter goat

In the course of the Middle Ages, the Galenic framework was further developed into a large-scale system of signs and symbols. In a typically medieval analogical way of thinking, widely divergent phenomena (ranging from the ages of man to astrological notions such as the system of the planets and the signs of the zodiac) were fitted into the fourfold schema presented by the medical theory. In Table I, an overview is given ofa number of those correlations. It should be mentioned, however, that the system was not entirely without unclarities (which is not surprising for a system that was to a large extent devised independently of empirical observation). For instance, while there was general agreement on the core of the system, authors would differ as to the more peripheral elements (such as the question which planet correlates with which humor; in particular, the associated animals are highly unstable across authors). Also, the system so to speak contained its own sources of confusion. There is, for instance, a marked ambiguity in the use of the concept 'blood', which was not only considered to be one of the four basic fluids, but which was also thought to transport the other humors, and which could hence also be used to refer to the mixture of the humors that was carried through the body (cp. Schiifer 1966: 4). And while the basic

234

Meaning and culture

color associated with yellow bile is obviously yellow, it was believed that yellow bile turned red when heated (and black when it was entirely burned up), so that the color 'red' could receive multiple interpretations within the system. The humoral edifice began to be undermined as soon as the Renaissance introduced renewed empirical medical investigations. Harvey's discovery of the circulation of the blood, for instance, was in direct contradiction with the traditional position of the blood in the Galenic 'digestive' system. However, the disappearance of the theory from the medical scene was only very gradual, and it took approximately another three centuries before the last vestiges of the humoral framework were finally removed. The standard view of the historians of medicine is, in fact, that only in the middle of the nineteenth century (and more particularly, with the publication of Rudolf Virchow's Die Cellularpathologie of 1858) did the humoral pathological conception receive its final blow. This 'final' character only holds, of course, for the official medical science: we have already seen in the introductory section about masturbation that traces of the old doctrine continued to exist much longer in popular belief. Along the same lines, it could probably be shown that the contemporary revival of herbalist medicine at the fringe of official medicine has direct links with the Galenic dietary pharmacology.

3. Anger in art As we have seen, the humoral doctrine had developed into a full-fledged semiotic system in the course of the Middle Ages: an ordered set of signs for medical and psychological interpretation. As a first indication of the fact that this semiotic system was not confined to the field of medicine, let us see how it influenced the artistic production of the Renaissance. We shall give two examples, one from the pictorial arts, and one from the dramatic arts. Cesare Ripa's Iconologia of 1593 was undoubtedly one of the major reference works for the seventeenth-century graphic artist. It contained a thematic inventory of the emblematic subject-matter of art, that is to say, of the topics, motifs, and symbols that could be used in paintings, drawings, engravings, and the like. There is a separate section in Ripa devoted to the four temperaments, with a detailed enumeration of the iconography associated with each of the four types. This is how Ripa introduces the choleric temperament.

Looking back at anger

235

Vn giouane magro di color gialliccio, & con sguardo fiero, che essendo quasi nudo tenghi con la destra mano vna spada nuda, stando con prontezza di voler combattere. Da vn lato (cioe per terra) sar. .. vno scudo in mezo del qual sia dipinta vna gran fiamma di fuoco, & dall'altro lato vn feroce leone. [A thin young man in yellow, with a ferocious face, almost naked, holding a drawn weapon in his right hand, ready to fight. From one side, a shield will be placed on the ground. with a flame of fire painted in the middle, and from the other side, a ferocious lion.]

Each of these characteristics is then further explained and elaborated in the course of Ripa' s expose, which is interspersed with references to and quotations from authorities such as Oalen, Ovid, Seneca, and Avicenna. In general, the attributes mentioned in Ripa' s description can be easily related to the characteristics mentioned in Table 1. Basically, the irascibility of the choleric person is symbolized by depicting him as a battle-prone warrior. Note that each of the details subtly contributes to the meaning of the whole; in particular, the fact that the young man is naked, and the fact he is not carrying his shield but that it is merely lying on the ground, indicate the impulsiveness of his hot-tempered nature: in his fits of rage, he does not even think about his own protection. If this impulsiveness is the negative side of his personality, the braveness symbolized by the lion is its positive side. Further elements that can be traced easily are the fire, and the yellow calor (corresponding, of course, with the yellow bile that is the physiological basis of this type). Less clear perhaps is the leanness of the young man's body, but this is an expression of the consuming character of the dry heat that is typical of the choleric physiology. It is worthwhile noticing that Ripa's description contains only the basic iconography of the four temperaments. It suffices to have a look at Klibansky, Panofsky and Saxl's (1964) magisterial monograph on Diirer's well-known wood-cut Melancolia I to get an idea of the intricacies and subtleties that arise when the humoral iconology is used and transformed by a truly creative artist. But the influence of the humoral semiotic system was not confined to the graphic arts. For instance, it has been described by various authors (Campbell 1930, Cruttwell 1951, Draper 1965, Schafer 1966, Pope 1985, Kail 1986) how the psychology of Shakespeare's dramatic characters unmistakenly refers to the theory of humors. Just a few quotations from The Taming of the Shrew suffice to demonstrate this.

236

Meaning and culture

(1) (2) (3)

Were I not a little pot and soon hot [IV: 1:5] Is she so hot a shrew [IV: 1: 17] I tell thee, Kate, 't was burnt and dried away, and I expressly am forbid to touch it, for it engenders choler, planteth anger; and better it were that both of us did fast, since, of ourselves, ourselves are choleric [IV: 1: 156] Gm. What say you to a neat's foot? Kath. 'Tis passing good. I prithee let me have it. Gm. I fear it is too choleric a meat. How say you to a fat tripe finely broil'd? Kath. I like it well. Good Gmmio, fetch it me. Gm. I cannot tell. I fear 'tis choleric. What say you to a piece of beef and mustard? Kath. A dish that I do love to feed upon. Gm. Ay, but the mustard is too hot a little [IV:3:25]

(4)

The conceptualization of anger in these quotations conforms to the model furnished by the theory ofhumors: anger is caused by choler (3), the production of which may be stimulated by certain kinds of food (3,4); while a choleric temperament is a permanent personality trait (3), the main attribute of the choleric personality is hotness (1,2). (This is not to say, by the way, that Shakespeare's use of the humoral doctrine is unoriginal: see Pope 1985 on the vivid and original way in which he handles the humoral concepts.) The fact that passages such as the ones quoted above can be multiplied from the work of Webster, Marlowe, or Jonson, leads Schafer (1966) to the conclusion that the humoral conception of physiology and psychology is something of a true fashion in Elizabethan drama. He attributes this to the fact that it is only in the middle of the sixteenth century that the doctrine became known to a wider audience that that of learned men who could read the medical authorities in their Latin and Greek originals. It is only, in other words, after the invention of printing that works such as Thomas Elyot's Castel of Helthe (1539), Andrew Boorde's A Breuyary of Helth (c. 1542) and A Compendyous Regyment or A Dyetary of Helth (c. 1542), or Thomas Vicary's A Profitable Treatise o.lthe Anatomie ofMans Body (1548) could be widely distributed, and that they could contribute to the spreading of the humoral doctrine to the community at large. (In Chapman 1979: 277, the wide distribution of almanacs is mentioned as a specific factor contributing to its popularity.) But if this dissemination of the doctrine of humors from the realm of learned knowledge to that of popular belief implies that it is technically a piece of gesunkenes Kulturgut, the question arises how far it

Looking back at anger

237

actually sank. In particular. how deep did it become entrenched in the language itself?

4. The lexical legacy We have already seen, in the case ofvalling, catarrh, rheumatism, and even cold, that single lexical items that are current today may be traced back to the humoral doctrine. These items are not isolated cases. In Table 2, we have systematically brought together a number of items and expressions in three European languages (English, French, and Dutch) that can be considered a part of the legacy of the theory ofhumors. Table 2. Lexical relics of the humoral doctrine

PHLEGM

ENGLISH

FRENCH

DurCH

phlegmatic

avoir un jlegme

valling

imperturbable calm, cool, apathetic, undisturbed to be imperturbable BLACK BILE

YELLOW BILE

BLOOD

(dialect) cold, infection of the nose

spleen

melancolie

zwartgallig

organ filtering the blood; sadness

sadness, sorrow, moroseness

sad, depressed (lit. black-bilious)

bilious

colere

Z 'n

angry, irascible, hot-tempered

anger, bad temper, wrath, rage

to vent (lit. to spit out) one's gall

full-blooded

avoir du sang dans les veines

warmbloedig

vigorous, hearty, sensual

to have spirit, luck

gal spuwen

passionate (lit. warm-blooded)

It will be noticed that the items exhibit various kinds of etymological or semantic relationships with regard to the older medical vocabulary. To begin with, there are items like melancolie, colere, andjlegme that refer directly to the original Latin denominations of the four basic fluids or types of personality. Next, there are items such as bilious and zwartgallig, which are based on a synonym (bile) or a translation in the vernacular (gal) of the technical term for the humor in question. Finally, there are items that have a more indirect relationship with the humors, in the sense of being metonymically related with them. Thus, spleen and valling are not formed on the basis of

238

Meaning and culture

the name of one of the basic fluids, but rather refer, respectively, to the organ typically associated with black bile (and hence to the associated temperament), and to a physiological effect thought to involve phlegm. Not included in the figure but equally revealing are items such as French humeur 'temperament; mood', that involve the generic term for the four fluids. Ifwe zoom in on one of the cells of Table 2, still further examples may be found. According to Roget's Thesaurus, the items listed under (5) all refer to anger or related concepts (the glosses are our own). (5)

choler 'anger' gall 'anger' rouse one's choler 'to elicit anger' stir one's bile 'to elicit anger' galling, 'vexing, causing anger' choleric 'irascible' liverish 'irascible' splenetic 'irascible' hot-blooded 'irascible'

.fiery 'irascible' hot-headed 'irascible' Although we will return to methodological problems more systematically further on in the paper, a brief methodological remark may be useful at this point. The basis for quoting a particular item as evidence for the influence of the humoral doctrine is the degree of etymological or semantic motivation that may be attributed to the item in question when it is interpreted in that historical light. Because items such as gall, liverish, choler(ic), and to stir one 's bile would simply remain etymological puzzles if the historical medical background were not taken into account, a humoral interpretation has explanatory value for them. But not all of the cases mentioned under (5) are equally clear. Two kinds of more or less problematic cases can be distinguished. In the first place, there are items whose global motivation in terms of the theory of humors is plausible, but whose local motivation within the theory is not entirely clear. Take the case of hot-blooded: if the typical fluid associated with anger is yellow bile, how come this expression contains a reference to the warming up of blood rather than bile? However, we have already seen that blood had a highly specific position in the whole doctrine: it is not only a humor in itself, but also carries the other humors through the body. If, then, blood can also refer to the mixture of the four humors as it circulates

Looking back at anger

239

through the body, it is not surprising that the warming up that causes anger may be metonymically said to involve the entire mixture. A similar but less easily explainable case of a possible local lack of motivation is splenetic: although this expression supports our general point that the influence of the humoral theory on our contemporary emotional vocabulary can be demonstrated, there seems to be a contradiction in the fact that items referring to the spleen may be either related to the melancholic temperament (see Table 2) or to the choleric temperament, as in (5). Given that the link between the spleen and melancholy is the orthodox one in the framework of the humoral approach, there are at least two ways in which the association between the spleen and anger could be explained. First, the association between the spleen and anger could be an effect of the confusion about aspects of the humoral doctrine that we have already drawn the attention to. Second, the association could be motivated by specific subtleties and refinements of the theory that we have not yet dealt with. In fact, Klibansky, Panofsky and Saxl (1964: 88) draw the attention to a passage in Avicenna where a distinction is made between the natural, primary form of melancholy, caused by an overproduction of black bile, and a secondary form of melancholy caused by a combustion of one of the other humors; thus, there is a specifically 'choleric' form of melancholy, which typically expresses itself as a state of frenzy (compare Starobinsky 1962, Jackson 1986 for the history of melancholy). At stake here is the notion of 'adust melancholy', which was thought to lead to more aggressive behavior and less fearfulness and sorrow than the natural melancholy that was engendered in a straightforward manner by an overabundance of black bile. (On the distinction between natural melancholy, adust melancholy, and choler - and on the confusion it leads to among scholars - see Soufas's 1990 argumentation that Don Quixote is an adust melancholic rather than the choleric type he has been made out to be in earlier humoral interpretations of Cervantes's work.) Our intention here is not to choose between these alternatives, but to make the methodological point that settling the question requires a detailed diachronic analysis of the development of the humoral theory and of its influence on our emotional vocabulary. If it is in general clear that the historical motivation behind the meaning 'irascible' of splenetic has to be sought in the older physiological-psychological conceptions of the theory ofhumors, a closer historical look at the development of that theory would yield valuable information about the specific history of splenetic. If, on the one hand, an explanation is sought in the less central aspects of the humoral theory, the historical analysis should be able to show how, for instance, the dissemina-

240

Meaning and culture

tion of Avicenna's view that was mentioned above led to the lexical association between the spleen and anger. If, on the other hand, that association is the result of a confusion, it is probably a later development, caused by impurities in the dissemination of the theory from its learned origins to the common people (or, perhaps, by the fact that the theory became less transparent when it gradually lost its scientific and medical respectability). If, then, the cases that are characterized by an apparent local lack of motivation merely establish the need for more detailed historical research, the items that raise global motivational questions are potentially more damaging for the humoral hypothesis. Consider an example like fiery: the (metaphorical) reference to heat could be attributed to the lingering influence of the humoral doctrine, but it could also be motivated on entirely different grounds. Suppose, in fact, that increased body heat is a physiological effect of being in a state of anger. and that anger is metonymically conceptualized in terms of its physiological effects. Rather than an historical motivation as a relic of a now abandoned medical theory, an expression such as fiery would then have an a-historical physiological motivation. At this point, we can include another set of expressions for the concept 'anger' into the discussion. In an analysis that has been published in several places (Kovecses 1986, Lakoff and Kovecses 1987, Lakoff 1987, Kovecses 1989), conventionalized phrases such as those in (6) have been subsumed by Kovecses and Lakoff under the general metaphor ANGER IS HEAT, which is further specified into ANGER IS THE HEAT OF A FLUID IN A CONTAINER when the heat applies to fluids, and into ANGER IS FIRE when the heat is applied to solids. (We will base our discussion on Kovecses 1989; there are only minimal differences in any case between the four published versions of the analysis). (6)

I had reached the boiling point. She was seething with rage. He lost his

cool. You make my blood boil. He wasfoaming at the mouth. He'sjust letting ofsteam. Don't get hot under the collar. Bi/~v's a hothead. They were having a heated argument. When I found out, I almost burst a blood vessel. He got red with anger. She was scarlet with rage. I was fuming. When I told him, he just exploded. Smoke was pouring out of his ears. He was breathing fire. Those are inflammatory remarks. That kindled my ire. He was consumed by his anger.

At a still lower level of analysis, these and many similar expressions are grouped together under labels such as when the intensity of anger increases, the fluid rises (his pent-up anger welled up inside him), intense anger produces steam (I was fUming), and when anger becomes too intense, the person explodes (when I told him, he just exploded). Next to the basic general

Looking back at anger

241

metaphor ANGER IS HEAT, less elaborate metaphorical patterns such as ANGER IS lNSANITY, ANGER IS AN OPPONENT, ANGER IS A DANGEROUS ANI-

and CAUSING ANGER IS TRESPASSlNG are identified. It will be obvious that the general metaphor ANGER IS THE HEAT OF A FLUID IN A CONTAlNER neatly fits into the humoral views: the body is the container of the four cardinal fluids, and anger involves the heating up of specific fluids (either yellow bile as the direct source of ire, or blood as the mixture of the four humors). However, given the alternative explanation of the general metaphor in terms of a physiological metonymy, is there any way in which we can say that the humoral hypothesis provides a better explanation of the motivation behind this particular subset of our emotional vocabulary? Apart from the general methodological point that a humoral explanation achieves greater generality by being able to combine an explanation of the cases under (5) with an explanation of those under (6), we have to consider two specific reasons for preferring it over a purely physiological explanation. First, it seems better able to motivate the reference to fluids in the expressions. Kovecses explains these references in the following terms: 'The fluid version [of the basic metaphor] is much more highly elaborated. The reason for this, we surmise, is that in our overall conceptual system we have the general metaphor the body is a container for the emotions' (1989: 53). The latter is illustrated by expressions such as he was filled with anger and she could not contain her joy. However, it is not clear how this metaphor combines with the basic ANGER IS HEAT metaphor to yield the application to fluids (as Kovecses claims it does): the fact that the body is a container for the emotions does not predispose the interpretation towards a conception of the emotions as fluids; after all, the contained emotions could just as well be solids or gases as far as the container metaphor is concerned. We will presently have more to say about the 'solids' version of the basic metaphor, but it can already be remarked here that in Kovecses's view of the matter, one would not expect the fluid version to be more elaborate than the solid version, because the container metaphor that is invoked as an explanation does not seem to favor the one over the other. Second, the humoral interpretation may help us to make sense of cases that are beyond the reach of a physiological explanation. In general, one is tempted to argue that a physiological interpretation entails that like physiological effects lead to like patterns of lexicalization. In this sense, it would be a counterargument for the physiological approach that an emotion such as shame, which is no less characterized by redness in the face (flushing) and a

MAL,

242

Meaning and culture

subjective impression of increased body temperature than anger, is not lexicalized by the same set of expressions as anger. It would make no sense, for instance, to say that one's blood boils with shame, or that someone is fuming with shame. However, Kovecses has rightly pointed out that there need not be a simple correlation between physiological effects and linguistic patterns, and that motivation does not equal prediction (1989: 85). The physiological effects of anger motivate our anger vocabulary, but because they do not predict the linguistic situation, emotions with similar physiological effects may be differently conceptualized. This element of caution does not, however, work as easily in the other direction: similar physiological effects need not have similar lexical reflections, yet similar patterns of lexicalization had better correlate with similar physiological effects if the physiological explanation is to have any generality. To take up an example, there exists a rather hackneyed set of expressions to the effect that love is a fire: you can let the flame of your love die out, you can have a steadily burning devotion for someone, and you can feel warm towards that person. On the one hand, this accords well with the humoral belief that love is one of the 'hot' emotions. On the other hand, it is physiologically unlikely that persons in love have a permanently raised skin temperature (we, at least, are not aware of physiological research to that effect). Granting, in other words, that the methodological key element is motivation rather than prediction, it does seem to be the case that taking into account the historical humoral background may lead to better motivational success. (A related point that will only be mentioned in passing here concerns the subjective experiential prominence of the physiological effects thought to underlie our emotional vocabulary. Kovecses refers to the experimental results of Ekman, Levenson and Friesen (1983) to prove that anger indeed correlates with higher skin temperature, whereas fear correlates with a decrease. However, independent evidence is needed to show that these objective increases and decreases correlate with subjective experiences of warm and cold. This question is relevant because the changes that were measured were rather small: an increase of O. 15 degrees in the case of anger, and a decrease of 0.01 degrees in the case of fear. Are these changes noticed at all by the individuals concerned? Are they sufficient to cause the subjective experiences that could influence our vocabulary?) On the other hand, let us now play the devil's advocate. A possible objection against the hypothesis that the ANGER IS HEAT metaphor is a legacy of the humoral theory could be based on those cases in which the basic metaphor is applied to solids rather than to body fluids: there is no reference to

Looking back at anger

243

solids, after all, in the original medical doctrine. Notice, however, that most of the expressions cited by Kovecses as evidence for the 'solids' interpretation, as listed under (7), refer to fire as the source of combustion rather than to a solid substance that is being warmed up; this is aptly rendered by Kovecses's labeling of this metaphorical subpattern as ANGER IS FIRE. By contrasting heat as applied to solids with heat as applied to fluids, Kovecses's formulations suggest that in the former case, the solids have the same function within the metaphorical image as the fluids in the latter case. But while the fluids are the object of the process of warming up, there is no reference to solids as things that are being warmed up in most of the expressions in (7). On the contrary, we mainly find references to fire as the source of the process of warming up. Therefore, while there would be an incompatibility in the images referring to fluids and to solids as the object of the heating process, there is merely a complementarity between the expressions referring to fluids as the object of the heating up and the expressions referring to fire as the source of the combustion process. And of course, these complementary images dovetail with the hypothesis that the expressions historically have humoral origins. (7)

Those are inflammatory remarks. She was doing a slow burn. He was breathing fire. Your insincere apology just added fuel to the fire. After the argument, Dave was smolderingfor days. That kindled my ire. Boy, am 1 burned up! He was consumed by his anger.

But what about the last two expressions in (7)? In the metaphorical image, there is an unmistakable reference to the person's body as a solid substance being consumed. But either in the presupposition that the fire of anger naturally takes its fuel from the body, or in the presupposition that it may detrimentally spread to the whole body and consume substances that are not its natural source of fuel, there is again no contradiction with the humoral conception of anger. On the contrary, we have already seen in Ripa's description of the choleric that a consumptive burning up of the body, resulting in leanness and thinness, is part and parcel of the original views. We see no reason, in short, to argue that the 'solids' interpretation of the ANGER IS HEAT metaphor endangers a humoral interpretation of the motivation behind that metaphor.

244

Meaning and culture

5. Attenuating the analysis The foregoing does not imply, to be sure, that the humoral interpretation of our emotional vocabulary is without problems. In order to avoid misunderstanding, we would now like to specify a number of views that we explicitly do not intend to propagate. First and foremost, we do not think that our analysis could not be further corroborated. Specifically, because we claim that a sizeable portion of our contemporary anger vocabulary is part of the lexical legacy of the theory of the four humors, we are convinced that a historical analysis of the development of our emotional vocabulary is necessary to supplement the foregoing remarks. It would have to be shown, in this respect, that the conceptual model of anger that we attribute to the humoral doctrine has indeed entered the language under the influence of the popular dissemination of the latter, and further, that there is a continuous tradition from that period to ours. Because the (a-historical) physiological model does not impose such restrictions on the historical development of the language (assuming at least that the physiological correlates of our emotions are historically stable), the ultimate test for the humoral hypothesis consists of a diachronic lexicological analysis. The purpose of this paper, to be sure, is not to carry out this diachronic analysis, but merely to show how it follows in an obvious manner from the humoral hypothesis. Further, we do not want to create the impression that the whole of our emotional vocabulary can be motivated in humoral terms. We are well aware that various patterns of conceptualization can be discerned in our emotional vocabulary~ the fundamental importance of Kovecses's research is precisely that it takes a major step towards the identification of those patterns. As a consequence, we do not claim that the humoral doctrine has had the same amount of impact on every possible emotion concept. Our basic claim so far has merely been that such an influence cannot be disregarded if a proper insight is to be obtained into the motivation behind our contemporary emotional vocabulary. Precisely how far the humoral influence goes is another matter, and one that can only be solved by further research. Specifically, we do not wish to imply that physiological factors are unimportant for the structure of our emotional vocabulary, nor that they could not interact with the historical humoral influences. Such an interaction could take various forms, one of which is that the physiological factors have a marked influence on the reinterpretation process that expressions with a humoral origin undergo in the course of time. The fact that a number of contemporary emotional expressions has its historical origin in the theory of

Looking back at anger

245

humors does not imply, to be sure, that the theory synchronically determines the interpretation of those expressions: though our vocabulary for the concept of anger may still bear the imprint of ancient medical theories, we no longer believe in the theory as such (notwithstanding isolated relics like the masturbation beliefs mentioned above). This implies that the expressions have gone through a process of reinterpretation. Pope (1985: 179) correctly identifies this reinterpretation process as one in which expressions that were once taken literally acquire a figurative interpretation: Though it [the humoral doctrine] now may be dead in our minds it is far from dead on our tongues. We have been taking each other's temperatures for over a hundred years and finding them steady at around 98.4 OF, but we still use and understand the language of humoral psychology. The only difference is that when we describe somebody as having hot blood or a cold heart or a dry wit we realize that we are talking metaphorically, whereas in the past we would have believed ourselves to have been talking about physical qualities. We would suggest, then, that the physiological factors that Kovecses concentrates on could be a crucial factor in this reinterpretation process. As the original literal motivation gradually disappears, the elements of our emotional vocabulary could receive a new interpretation as figurative expressions of the physiological effects of particular emotions. Such a physiological reinterpretation would not be automatic, however: in some cases, the expressions could simply loose all transparency (following the valling model), while in others, the new figurative meaning could be purely metaphorical rather than metonymical along the 'physiological effects' line. For instance, taking for granted that the origins offiery in the sense 'irascible' are humoral, and also taking for granted that the expression has not become totally opaque in the way in which valling has, its contemporary reinterpreted meaning could be based on the physiological metonymy that anger causes body heat (as Kovecses would suggest), but it could also be the case that fiery is synchronically interpreted on the basis of a metaphorical image: the propensity of the irascible person to burst out abruptly could be compared with the fire's tendency to flare up suddenly. Methodologically, what is required here is an investigation of the way in which the expressions in our contemporary emotional vocabulary are actually interpreted: what kind of interpretation (if any) do people associate with them? Again, we wish to emphasize that a closer scrutiny of the reinterpretation process also naturally includes a historical analysis - if only because the reinterpretation process is a historical phenomenon. By following the histori-

246

Nfeaning and culture

cal development of our emotional vocabulary on a step by step basis, information about the reinterpretation process can be obtained. In particular, it can be hypothesized that those lexical items that are not easily reinterpreted will sooner disappear from the language than others. On the basis of this assumption, the relevance of more specific hypotheses can be determined. Is it correct, for instance, that expressions that are easily reinterpreted along the physiological lines set out by K6vecses, are more resistant to a process of lexical loss? The obsolescence of a number of the expressions mentioned under (5) could be an illustration of the same reinterpretation process: if the expressions under (5) are synchronically less lively than those under (6), this may very well signal a historical shift from a humoral to a physiological interpretative framework. It is beyond the scope of this paper to deal with the question in detail, but we hope to have made clear that historical questions such as these follow logically from a consideration of the possible humoral origins of our contemporary emotional vocabulary. Finally, it should be pointed out that we have not tried to answer the question where the humoral theory itself comes from, conceptually speaking. Obviously, it is based on anatomical observations concerning the bodily fluids, but is there any reason why, within the theory, the concept of anger should be specifically linked to the yellow bile, and to fire? Within a physiological conception of our emotion vocabulary, it seems attractive to postulate that the humoral theory itself draws on a pre-theoretical physiological experience of the emotions. At the time of its conception, the humoral theory would then be a literalization of a pre-existing, physiologically motivated metaphorical understanding: the conceptualization of anger as fire, for instance, would then primarily be a physiological metaphor that is later turned into a literal statement in the framework of the medical theory of humors. This is a position that is implicit in K6vecses (1995): accepting the possible influence of culture-specific influences on the emotion vocabulary available in a specific language, he argues that there exist cross-culturally uniform factors of a physiological nature that constrain and stabilize the cultural conceptions. K6vecses argues that such a conception contradicts the suggestions made in the present paper, but this is a conclusion that we explicitly have to oppose: as should have become clear from the foregoing remarks, we do not claim that only cultural factors are important, and that physiological factors could not play a role in the development of our emotion vocabulary. It is important, in this respect, to distinguish between the methodological and the substantive part of the present article. The article purports to do two things:

Looking back at anger

247

to argue for the importance of culture-specific historical research when present-day emotion vocabularies are being considered, and to put forward a particular hypothesis (the humoral one) within such a perspective. Kovecses (1995) accepts the methodological point to the extent that he explicitly recognizes the potential influence of cultural factors on the development of our emotion vocabulary, but he exaggerates the weight we would like to attach to the humoral theory. In particular, we claim that including the humoral doctrine into the picture is important for accounting for our present emotion vocabulary, but we have made no statement about the origins of the doctrine itself. In general, we would therefore like to leave open the possibility that Kovecses is right when he suggests that the humoral doctrine is a culturespecific rationalization of a universal physiologically-based metaphorical understanding of the emotions. However, we would like to stress that there is once again a methodological point to be made: if the physiological conceptualization of anger (as typically embodied in physiological metaphors) precedes the humoral theory, the only way to establish this is by doing historical research. If the suggestion implicit in Kovecses (1995) is correct, the preHippocratic conceptualization of anger in Classical Greek should be based on physiological metaphors. It is beyond the scope of this article to test the hypothesis, but it is methodologically important to see that it is an empirical hypothesis that can be tested through historical research. Here again, our conclusions on the methodological level are more important than those on the substantive level: regardless of whether the origins of the humoral theory are indeed physiologically metaphorical or not, the very question about the origins of the humoral doctrine calls for historical research. To summarize, the present paper is to a large extent hypothesis-forming: we claim that it is necessary to take into account the historical background of our emotion concepts to get a clear picture of the present-day situation, and in particular, that it is necessary to include the humoral doctrine into the investigation. Also, this investigation naturally entails a longitudinal historical analysis of the development of our emotional vocabulary. But if we hope to have established the necessity and the attractiveness of such a research programme, we certainly do not pretend that we have already carried it out.

6. Methodological musings In the previous sections, we have presented an alternative for Kovecses's analysis of the general metaphor ANGER IS THE HEAT OF A FLUID IN A CON-

248

Meaning and culture

TAlNER. Instead of a straightforward physiological interpretation, we suggest that it has undergone the influence of the humoral doctrine, but that the original set of humoral expressions has been subjected to a process of reinterpretation and obsolescence. We are now in a position to make some further methodological remarks. Two related topics will be discussed: the use of conventionalized language, and the relationship between folk models and scientific knowledge. Kovecses explicitly takes the conventionalized way in which a particular culture talks about the emotions as an indication of the way in which that culture conceptualizes the emotions; the conventionalized language under scrutiny includes idioms, cliche's, sayings, proverbs, collocations, and set expressions in general (1989: 43). Now, while Kovecses states with some emphasis that 'each and every expression related to a concept has to be examined if we wish to uncover the minute details of the concept' (1989: 44), the question arises why expressions such as those that are mentioned under (5) are not included in the observational basis of his treatment of the concept 'anger'. Why has Kovecses picked out for consideration the particular set of expressions that he actually concentrates on? It could be hypothesized that Kovecses has explicitly restricted his analysis to those expressions that are the most transparent ones for a contemporary audience, i.e. those expressions whose metaphorical nature is still a live one, or, more generally, those expressions that are most readily considered to be motivated by today's speaker's of English. The question of motivation can be illustrated by comparing an expression such as valling with an item such as to make one's blood boil. Although they have a common historical motivation in terms of the theory of humors, valling is entirely fossilized and opaque for the contemporary language user, whereas to make one's blood boil could possibly receive a motivation along the physiological lines set out in the previous section. The problem that is at stake here is the same as the one mentioned there; also, it has been identified several times in connection with the 'generalized metaphors' approach of Lakoff and Johnson (1980), on which Kovecses' s identification of metaphorical patterns is based: if generalized metaphors are cited as evidence for our contemporary way of conceptualizing the world, it does not suffice to identify the metaphor, but it has to be shown on independent grounds that the metaphors are not just dead ones (see among others Traugott 1985, Geeraerts 1981). Those 'independent grounds' could be the researcher's intuition, but also, for instance, psycholinguistic experiments in the line of Gibbs (1990). So, what would have to be shown before expressions such as those in (6) are cited as evidence for our present-

Looking back at anger

249

day way of conceptualizing emotions is not just that they are not dead metaphors (in the sense in which valling is opaque and fossilized), but also that the motivation that they actually receive is in tenns of the physiological effects of anger. (On related questions, see also Ortony 1988.) Now, it would be unfair to claim that Kovecses ignores the question whether the conventionalized language we use to talk about the emotions actually reflects our current beliefs. He discusses the problem by making a comparison with our astronomical model of the world: expressions such as the sun came up and the sun went down cannot be used as evidence for a geocentric folk model, basically because fonnalized education has profoundly influenced our way of thinking. This is, says Kovecses, in marked contrast with the situation on the emotional field. As a result of certain scientific discoveries, our educational system has spent several centuries on changing our geocentric view of the relationship between the earth and the sun. The consequence is that, despite our language use, anyone with at least some elementary education would refuse the geocentric view as his or her folk model of the earth/sun relationship. Nothing like this has been the case with the emotions. No such large-scale attempts have been made to change our thinking about them (. .. ). As a result, we pretty much believe what we say about them. (... ) It seems then that, as far as the emotions go, we still live by and think in terms of a geocentric emotional universe (1989: 45-46). There are various things to be said about this view. To begin with, it does not invalidate the methodological problem identified above. Because Kovecses reaffinns the necessity to take into account all expressions - and to take them at face value - the fact remains that he does not follow his own methodological dictum. But if he would have followed it, the unmistakable presence of items derived from the theory of humors might have led him to the conclusion that we indeed still (partially) adhere to a 'geocentric' view of emotions (i.e. to a prescientific, medieval theory), but that the theory is a humoral one rather than the physiological one he suggests. So we are faced with a dilemma: either Kovecses is right in affinning that our present-day views about the emotions have not been influenced by scientific discoveries (but then his method of taking expressions at face value would lead to the conclusion that we still have a humoral conception of anger), or he would have to reconsider his statements about the influence of scientific theories on our emotional vocabulary, (together, in fact, with his belief that emotional expressions can be taken at face value, i.e. without considering the possibility of reinterpretations).

250

lvfeaning and culture

The fact, on the other hand, that we clearly no longer take the humoral expressions literally (the fact, that is, that we no longer believe the theory) can only be attributed to the same kind of dissemination of scientific theories that led to the downfall of the geocentric view in astronomy. Because we have learned about the new anatomical and physiological discoveries, we have abandoned our earlier folk models (by and large, i.e., not counting relic beliefs like those about masturbation that we started off with). And even this older folk model itself was not a pure folk model. As we have seen, it was a piece of high, Latinate culture that was gradually incorporated into the common culture through the intermediary of popularizing publications. There is no reason, in short, to believe that our emotional vocabulary is free of scientific influences, and there is no ground for a methodological exploitation of such a conception of the specificity of our emotional vocabulary.

7.

Culture and cognition

To sum up, we have tried to establish the following points. First, the medieval physiological-psychological theory of the four humors and the four temperaments has left its traces on our emotional vocabulary. Second, the ANGER IS THE HEAT OF A FLUID IN A CONTAINER metaphor identified by K6vecses (1989) can be seen as one of those traces. It is then not motivated directly by the physiological effects of anger, as K6vecses suggests, but it is part of the historical (but reinterpreted) legacy of the humoral theory. Third, K6vecses's neglect of the historical background of our emotional vocabulary prevents him from appreciating the possible impact of the humoral theory; once this possible impact is taken into account, inconsistencies in K6vecses's methodology become apparent. Fourth, a further corroboration of the historical-humoral hypothesis requires a longitudinal scrutiny of the historical development of our emotional vocabulary. Because they are the most wide-ranging, we consider the methodological consequences of our investigation to be of primary importance. The basic point as we see it is this: an adequate analysis of the motivation behind cultural phenomena in general and language in particular has to take into account the diachronic dimension. Cultural models, i.e. the more or less coherent sets of concepts that cultures use to structure experience and make sense of the world are not reinvented afresh with every new period in the culture's development. Rather, it is by definition part of their cultural nature that they have a historical dimension. They can only fulfill their role of shaping a

Looking back at anger

251

conununity's life if they have a historical pennanence, that is, if they can be transmitted from generation to generation, assuring continuity over and above an individual's and an individual generation's activities (though not, to be sure, unaffected by them). If cognitive models are cultural models, they are also cultural institutions, and as such, they carry their history along with them: their institutional nature implies their historical continuity. It is only by investigating their historical origins and their gradual transfonnation that their contemporary fonn can be properly understood. Now, while one of the major steps forward taken by Cognitive Semantics has been to put the study of meaning back into its cultural and experiential context, it would seem that the natural consequence of including the diachronic dimension into the investigation has perhaps not yet been fully appreciated. There is an instructive parallel to be drawn here between the cognitive-semantic study of single lexical concepts (as in prototype theory) and the cognitive-semantic research into supra-lexical structures such as the cultural models of emotion that K6vecses concentrates on. In the case of purely lexical research, the emphasis on the mechanisms of semantic flexibility that underlie the structure of polysemy (such as metaphor and metonymy) naturally entails a renewed interest in diachronic semantics (see Geeraerts 1988a): to a large extent, the synchronic polysemy of lexical items is a reflection of their diachronic development. The point that we are trying to bring home here is that an awareness of the synchronic reflection of diachronic patterns is just as natural and just as important in the case of supralexical cognitive structures as in the case of lexical concepts. If cultures are only cultures because they have a tradition, and if, therefore, cognitive models are only cultural models if they have a chronological continuity and a historical pennanence, an awareness of the history of ideas is methodologically indispensable for Cognitive Semantics.

Chapter 10 Beer and semantics

Originally published in Leon De Stadler and Christoph Eyrich (eds.), 1999, Issues in Cognitive Linguistics 35-55. Berlin/New York: Mouton de Gruyter. This paper is a revised version of the opening address presented on July 18, 1993 at the 3rd International Cognitive Linguistics Conference. Both the tone and the topic of the paper bear witness to the original occasion: the topic introduced the opening reception of the conference (which featured a representative sample of choice Belgian beers), and the tone is accordingly somewhat lighter than linguistic papers tend to be. Even so, the present analysis is of interest beyond the original context, for two reasons. First, it illustrates some of the salience phenomena that were described in Chapter 4 of the present volume. Structural salience was there defined in tenns of the weight of the distinctive dimensions that differentiate between lexical items: a semantic feature or dimension is more structurally salient to the extent that it occurs more often in the structure of the lexicon, i.e. to the extent that it is more often expressed or evoked. In the present paper, this is illustrated both with regard to the purely descriptive and with regard to the metaphorically evocative brand names for Belgian beers. The analysis of the non-metaphorical brand names shows how some distinctive characteristics of specific beer types are more readily expressed than others. Beers that are brewed in trappist monasteries, for instance, are invariably identified as such, but the addition of herbs for flavoring is only seldom a motif in naming. More importantly, the analysis of the metaphorically evocative names reveals that beer is associated with a culturally specific cluster of conceptual metaphors in the Lakovian sense: it combines pleasure, prestige, local identity and tradition - and a sense of adventure and transgression. Second, the paper adds a specific perspective to the study of proper names in Cognitive Linguistics. Proper names have not yet been studied intensively by cognitive linguists, but to the extent that they have, stress has fallen on the theoretical question what kind of meaning, if at all, is involved in the use of proper names (Van Langendonck 1999). In addition, studies like Jakel (1998, 1999) have looked at the metonymical mechanisms at work in proper names. The present paper adds the perspective of lexical salience and that of metaphor analysis as ways of uncovering underlying patterns in name-giving. More work on beer branding may be found in Wagner (2002, 2003), who studied the image of beer in German advertisements.

Beer and semantics

253

1. Belgium as beer's own country If Cognitive Linguistics is about the relationship between language and culture, Cognitive Linguistics conferences should be concerned with learning about cultures as much as they are concerned with learning about language and languages. That is why it seemed appropriate to start off the 3rd International Cognitive Linguistics Conference with an investigation into a central aspect of Belgian culture, namely beer. In approaching a culture, in fact, it seems only fair to tackle those areas of behavior first that are experienced as being important by the members of the culture themselves. And it seems that there is indeed sufficient evidence for the claim that beer is part of the pride and joy of the inhabitants of that minuscule patch of land along the coast of the North Sea. First, some factual data may demonstrate the importance of beer in Belgium. To begin with, the consumption of beer per head of the population is relatively high. Data provided by the lnterbrew breweries in 1987 list the following per capita beer consumption in twelve European countries: Sweden 5lliters, Norway 51, Germany 145, Czechoslovakia 130, Belgium 121, Austria 118, England Ill, The Netherlands 85, Ireland 75, Spain 65, Portugal 40, Italy 25. The figures demonstrate the existence of a so-called European beer-belt, situated geographically between the spirit-belt in the north of Europe, and the wine-belt in the south. The figures, of course, do not show that Belgium is beer country number one as far as per capita consumption is concerned: it comes third after Germany and Czechoslovakia. The key concept regarding Belgium's status as a beer country, however, is not quantity, but quality. And the quality shows up specifically in the enormous variety and the abundant diversity of the types of beer produced in Belgium. In the data set that was used for this investigation (and which is taken from Peter Crombecq's annual overview of Belgian beers for the year 1992) no less than 1454 different brands are distinguished. In Crombecq's own classification of these beers according to taste and alcoholic strength, no less than 304 distinct types of taste appear. A comparison with Holland, Belgium's nearest neighbor, strengthens the impression of diversity. Against the 1454 different brands to be found in Belgium, there are only 354 Dutch ones. And while Crombecq distinguishes 304 different Belgian tastes, he only tastes 94 Dutch ones. In short, there is sufficient evidence for the variety of Belgian beers. At least as important as these factual data is the recognition that Belgians identify themselves with the country's production of high quality beers. If

254

A:feaning and culture

you ask people from Belgium what their country is good at, they will probably mention beer and chocolates. The recognition that beer, and the quality of Belgian beer, is an explicit topic in Belgian culture, may be further illustrated in two ways. One is the following slogan from a beer advertisement: Het bier van het land van het bier, i.e. 'the beer from the country of beer', or, more freely, 'the beer from beer's own country'. Belgium as beer's own country - this is apparently a live notion. The second illustration consists of the number of coffee table books that are published in Belgium on the topic of beer (although the name coffee table books is obviously somewhat paradoxical in this respect). Here is a list of some more or less recent ones; if nothing else, it at least shows that there is a live interest in beer, and that beer is considered a quality product worthy of luxurious publications of the type that is usually restricted to wine and art. M. Jackson [1977]. Spectrum bieratlas (translation of The world gUide to

beer). Utrecht!Antwerpen. W. Patroons [1984]. A lies over Belgisch bier. Antwerpen/Weesp. 1. Tulfer [1986]. Belgische biergids. Antwerpen. M. Jackson [1991]. De grote Belgische bieren. Een volledige gids en een hulde aan een unieke cultuur (translation of The great beers ofBelgium).

Anwerpen. G. van Lierde [1992]. Bier in Belgie. Gids voor bieren en brouwerijen.

Roeselare.

2. Methodological considerations Given the importance of beer in Belgian culture, the cognitively interesting question is this: how is beer conceptualized in Belgium? What are the cognitively salient concepts associated with it? And more specifically, starting from a Cognitive Linguistic perspective: what does language reveal about the conceptualization of beer? The two crucial aspects of this question that I would like to focus on are the following. First, a general methodological question: how can salience effects in lexical fields be measured? Second, a specific question about beer: what are the specific values associated with beer? Before I try to be more specific about these questions, I have to clarify a few points. To begin with, there are a number of restrictions on what I will be doing next that have to be mentioned explicitly. The vocabulary that I will be talking about consists basically of brand names, i.e. the names of types of beer.

Beer and semantics

255

This means that various parts of the lexical beer field will pass unnoticed. For instance, I will not have the opportunity to mention the names of pubs, inns and cafes (in spite of such gems as The ship o.ffools, The last judgment, The kingdom of heaven, The ultimate hallucination). The restriction also implies that the vocabulary for talking about beer will not enter the picture. I will have nothing to say, that is, about predicates like earthy, fruity, refreshing, complex, seductive, and others that may be used to describe beers. (Surprisingly perhaps, this vocabulary is far less extended than the one that exists for wine and that has been analyzed so elegantly by Lehrer 1983.) Another consequence of the restriction to brand names is that the everyday names of the beers are not considered systematically in what follows. The brand names, if you like, are full names, constituting a truly individual identification of the beers. The everyday names, on the other hand, are like Christian names: shorter but at the same time more general forms that are communicatively sufficient in most circumstances. In the technical terms of lexical semantics, the brand names are certainly not basic level terms: they are used much less frequently than the everyday names, and they are taxonomically much more specific. In most cases, the generic basic level denominations are based on overall types of beer. Whereas the full names are clearly proper names, the everyday names are common nouns. When you ask for a geuze in a cafe, it will mostly be of no interest to you whether you are served an Eylenbosch gueuze Iambic, a Drie Fonteinen geuze, a Mort Subite geuze or any of the other brands of geuze. But by a typical prototypicality effect, the everyday name is not necessarily originally a common name. When you ask for one of those typical white beers made on the basis of wheat you would probably ask for a Hoegaarden, which is the most typical representative of this class - and you probably would not object if you were then served a Dentergems witbier, which is another example of the same class. It could be argued, no doubt, that the basic level terms are somewhat more interesting for lexical analyses than the brand names, precisely because they are cognitively more salient. Note, however, that systematic research into the actual use of the everyday names would require extensive observation in pubs and cafes, and I am afraid that the investigation would then be diverted very soon onto scientifically unfruitful although extremely pleasant side-roads. Moreover, precisely because the brand names are more specific than the everyday names, they may reveal more of the conceptualization of beer. Again, it could be remarked that the conceptualization they reflect is

256

Meaning and culture

not the consumer's but the brewer's, but even so, the brewers are likely, for obvious commercial reasons, to choose names that have a broad appeal to the consumers. Even though we study names that are not actually given by the consumer, the names can be expected to fit the consumers' image of what beer should be. From a methodological point of view, the field of beer terms is interesting for two reasons - two complications, in fact. The first methodological point to note involves the fact that the names found in the data set subclassify into two fundamentally distinct subsets, which each require their own type of analysis. The majority of the brand names conforms to a general pattern: they consist of an identification of the firm producing the beer, and a specification of the type of beer involved. Here are some examples: Louwaege pils, Liefmans frambozenbier, Belle Vue geuze. There is considerable variation in this pattern, not least because neither the first nor the second element is absolutely necessary. In some cases, a single reference to the origin of the beer suffices. This is the case, for instance, with some of the beers produced in monasteries (I will come back to these further on). A name like Orval identifies the monastery together with the single kind of beer produced there: there is only one Orval kind of beer, and it is a trappist beer. Conversely, only the type of beer may be specified, without reference to the brewer, but then, the generic name will be subclassified by the addition of various specifications as in the following: Dentergems witbier, Aarschotse bruine, where a placename indicates that this is the type of white beer or brown beer or whatever as typically brewed in Dentergem, or Aarschot, or wherever. The subclassification may also occur in combination with an identification of the brewer, as in Liefmans geuze gefilterd. Liefmans geuze ongefilterd, where it is indicated that the geuze beer produced by the firm Lindemans comes in two varieties: a filtered one and an unfiltered one. In cases like the Orval-type, where the element identifying the producer at the same time suggests the kind of beer involved, the subclassifying element occurs without an explicit mention of the generic type of beer. In the following set, the various additions identify the varying degrees of alcoholic strength of the beers. The name Westvleteren is again a name for a monastic brewery; it identifies the beer in question as a trappist beer: Westvleteren dubbel 4, Westvleteren special 6, Westvleteren extra 8, Westvleteren abt 12. All the names mentioned so far stand in dramatic contrast with a type of name illustrated by the following examples: Duvel, Luc!!er, Judas, Verboden Vrucht ('Forbidden fruit'). The basic distinction between this type

Beer and semantics

257

(which fonns a sizeable minority of the total set) and the previous one resides in the fact that the type 1 names are literal names, whereas the type 2 names are figurative, or at least associative names. A name like Lie/mans geuze gefilterd is made up of various infonnational elements that identify aspects of the beer: the fact that it is of the geuze type, the fact that it is produced by the Liefinans brewery, the fact that it has gone through a filtering process. Each of these infonnational elements is identified directly in the name of the beer, i.e. by means of words whose literal meaning it is to express the infonnational elements in question. The literal meaning of the type 2 names, on the other hand, does not as such identify the relevant infonnational element. The name Duvel literally means 'devil', but the infonnation that is relevant for the beer name is only indirectly connected with this literal meaning: it is a concept that has something to do with the devil, and that for one reason or another is also appropriate to talk about this kind of beer. Not surprisingly, the identification of the relevant infonnational element is not always easy with type 2 names. The beer may have something devilish, but what exactly is that? Does it lead you to destruction? Is it a sin to drink it? Does it overpower you in the way the devil might? One could say, perhaps, that the type 2 names attribute somewhat vague qualities to the beers, whereas the type 1 names describe rather clearly defined features. In summary, the type 2 names are associative or, if one wishes, evocative names, because they function indirectly, and because they involve less clearly defined characteristics than the type 1 names. The consequence of the distinction is that the type 2 names have to be analyzed on a different basis than the type 1 names. In the case of the type 1 names, one can base the analysis rather straightforwardly on the literal readings of the tenns. In the case of type 2 names, one has to take into account the relationship between the original literal meaning and the evocatively associated meaning. This type of analysis, of course, is well-known within Cognitive Linguistics: it is analogous to what we all do when we analyze metaphorical patterns in Lakovian style, i.e. to identify source and target domains and pinpoint the precise motivational link that connects them. In the vocabulary subset that interests us here, the associative names are not all metaphorical, but the analytical procedure used for metaphors can be generalized. The second complication that I hinted at before involves the fact that one of the things that we are crucially interested in are salience effects in lexical .fields. Given a set of categories that belong together in a conceptual field (like types of beer), what we would like know is which categories are conceptually more important than others. This is not a question that is ade-

258

Meaning and culture

quately answered in traditional lexical field theory, but I think we should recognize that even Cognitive Linguistics, with its outspoken interest in salience effects, has not yet developed a proper methodology for treating this kind of issue. The salience effects Cognitive Linguistics has been primarily interested in are salience effects within separate categories, an interest that is epitomized by prototype-theoretical research (cp. Geeraerts 1989c). Apart perhaps from Berlin's basic level model (see Berlin 1976, 1978), not much systematic attention has been devoted to salience effects that involve not the various applications of a single category, but rather the relationship between various lexical categories as altemative names for the same kind of referent. There is, to be sure, a nuance to be added to this picture. As far as the associated values as expressed by Type 2 names are concemed, we do have a clear-cut idea of how to measure salience: an associated value is more salient to the extent that it recurs throughout the set of evocative names. This is basically the measure of cognitive salience behind the search for generalized metaphors introduced by Lakoff and Johnson (1980). Note, however, that it involves only Type 2 names, so that it remains important to try to find a measure of salience that specifically applies to Type I names.

3. The origins of the diversity To see what is involved in tackling the methodological question raised a moment ago, we should first have a closer look at the structure of the referential field itself, i.e. at the variety that I hinted at before. In order to get a better grasp of it, it is important to know something about how beer is made. Basically, you put barley into water and let it ferment. However, an elaborate beer culture could not have developed if this process had not been subject to refinement. Schematically, what actually happens can be divided into three steps. First, the barley is let to soak and germinate, and the budding barley is then dried or sometimes roasted. The product of this first step is known as malt, which reveals, by the way, that whisky is the distilled counterpart of beer. In a second step, the malt is crushed and cooked, so that the natural sugars that it contains become free. During the cooking stage, additional substances for flavoring are added, such as hop or extra sugar. The product of this second step is called wort. When yeast is added to the wort, it can then ferment, and at the end of this third stage, you get beer. At each of the three stages, altematives with regard to the standard process exist. Instead of starting off with barley, for instance, you can begin with

Beer and semantics

259

wheat, or rye, or a mixture. In the second stage, various other flavoring substances can be added. Typical for a number of Belgian beers is the addition of fruit flavors like cherry and raspberry, or certain herbs like ginger, licorice, coriander, or mint. And in the third stage, different forms of fermentation exist. This is an important point that needs some elaboration. Basically, there are three forms of fermentation: high fermentation, low fermentation, and spontaneous fermentation. In the first case, the fermentation process is relatively short and intense, producing strong ales. In the case of low fermentation, the process is a slow one, the ripening process takes a long time, and the resulting beer is less heavy than in the case of high fermentation. These are the beers of the regular pilsener or lager type. Spontaneous fermentation, on the other hand, is much less common than high or low fermentation, and is really extremely typical for one portion of the Belgian beers. In the surroundings of Brussels, the art of producing beers by spontaneous fermentation has been refined in ways not known anywhere else in the world, and some of the most typical names associated with Belgian beers have to do with spontaneous fermentation. Specifically, I first have to mention lambiek, which is the product of 100% spontaneous fermentation. As lambiek as such is rather sour, you will not find very often in its pure form. Usually, sweet flavors are added in the form of fruit extracts, which yields names like kriek lambiek 'cherry Iambic'. Another additional process that Iambic may go through is a second fermentation, which yields the well-known geuze, which is sweeter than Iambic, and which has a unique, champaign-like sparkling character.

'____ _

_____'f-----r-----'~~ germination, drying/roasting

;----.-~'___I_______'f---.---~~ addition of hop, addition of sugar etc., cooking

addition of yeast, fermentation

Figure 1. Schematic representation of beer production

The variety does not stop there, however. At least two additional points have to be mentioned. For one thing, some Belgian beers are brewed only on a special occasion. Traditionally, this could be the harvesting season, or Christmas, or the annual festival of the local patron saint. Nowadays, these

260

Meaning and culture

gelegenheidsbieren or 'occasional beers' also include beers made to commemorate specific historical events, like the anniversary of the foundation of one or another city, or the birth of a historical figure. Another additional factor besides the temporal one is a geographical one, or at least, it involves the place where the beer is being brewed. This is particularly important for abbeys and monasteries producing beer. In fact, refining the art of brewing is one salient aspect of the civilizing influence that monasteries and abbeys have exerted in the Low Countries. Traditionally, all monasteries brewed their own beer, like many farmers once did. Today, there are still five monasteries left where the monks brew their own beer. As they all belong to the Trappist order, their beer is commonly known as trappist. The monasteries in question are those in Orval, Chimay, Rochefort, Westvleteren and Westmalle. These beers are all of the strong, high fermentation type, but there are many differences among their individual flavors. It is useful, by the way, to distinguish the trappist beers from the larger category of abdijbieren or 'abbey beers'. These are beers of basically the same type as the trappist beers, but they are being produced by commercial breweries who have taken over the original monasterial breweries, or who have merely paid for the license to use the abbey's name. Examples are Leffe, Grimbergen, Maredsous, Affligem, Corsendonck, and Tongerlo. On the basis of what I have explained so far, we can now define, on a generic level, some of the most common types of Belgian beers. pils: a light blond beer of the pilsener or lager type, produced through low fermentation (common brands: Stella, Jupiler) witbier: 'white beer', a light beer on the basis of wheat rather than barley, with a blond, somewhat cloudy appearance (common brands: Hoegaarden, Dentergems) geuze: a sparkling, spontaneously fennented beer with a reddish color and a sour-sweet taste (common brands: Belle-Vue, Mort Subite) kriek: a moderately sweet beer obtained by adding cherry juice to geuze or lambiek (common brands: Belle-Vue, Mort Subite) trappist: strong, mostly dark brown beer of the high fermentation type, produced in Trappist monasteries (common brands: Orval, Chimay, Rochefort, Westmalle). (Sometimes, the name trappist will also be used for the comparable commercial abbey beers).

Beer and semantics

261

4. Salience effects for type 1 names Although it would seem that the list at the end of the previous section gives us a nice and neat set of categories to start the onomasiological investigation with, things are less clear than they seem. Once we start looking at more specific categories, there appears to be a lot of overlap and crossclassification among the various groupings of beers. Consider the following example. As we have seen, we can distinguish between geuze, lambiek, and kriek in the category of spontaneously fermented beers. Among the beers with high fermentation, there are the whitebeers, based on wheat, but also a number of barley beers, such as the popular, very strong beer called Duvel ('devil'). Geuze and lambiek have in common with the whitebeers that wheat is used in the production process. Duvel and geuze, on the other hand, have in common that their fermentation process continues when they are already bottled. And further, there are kriek beers that are not based on geuze or lambiek. like the Felix kriek oudbruin. The resulting picture is one of multiple overlapping, as shown in Table 1. Table 1. Multiple overlapping in the lexical field of beer names

-. . ~

Q

high fermentation spontaneous fermentation fermentation in bottle wheat based cherry f1avor

+

"-

.~

.s;:::: "-..::,:

~

;;:

~ .~

+

+

Cl"::':

+ +

..::,:

..::,:

>,

:.c;