Words and Other Wonders (Cognitive Linguistic Research)

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

>

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 <+contact> 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 <-contact> versus <+contact> 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 <+contact> one, the second a <-contact> 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 <-contact> 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 <no contact> 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 <+contact> 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 <+contact> 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
66

Prototypicali~y and salience

tion> 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 <-contact> 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 <+contact> and <-contact> 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 <+contact> 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 <motion>, , 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<; a penguin a bird? - Well, it is and it isn't (that is to say, strictly speaking it is a bird, but it is not a typical bird). Taking into account that there is no doubt about the membership status of penguins in the category of birds, the 'p and not p' construction here involves differences in typicality: it sets off the prototypical application from peripheral cases. We now see that we could also have exchanges like the following: Is a mango a fruit? - Well, it is and it isn't (strictly speaking it is, but it does not have the typical characteristics of typical fruits like oranges and apples). If it is correct that the logical test does reveal a polysemy, and if this polysemy precisely involves the distinction between the typical and the non-typical cases of a category, it is also understandable that the 'p and not p' frame does not reveal any polysemy when applied to a highly typical case such as strawberry (even though it is perhaps not the single most typical case). But although the logical criterion does, in this case, yield polysemy, it nevertheless still does not lead to precisely the same kind of polysemy as the definitional test: if at all, the polysemy test can at most distinguish between core and periphery, but it cannot single out any and each of the definitionally maximal subsets that constitute the senses of fruit according to the definitional approach. In other words, the fruit example presents us with a case in which the application of all three criteria for polysemy yields different results. What are distinct meanings from one point of view (that is, according to one test) are reduced to cases of vagueness from another point of view.

6. Problems with the polysemy tests - Second Series At this point, two possibilities for further developments open up. In the first place, the recognition that the various criteria lead to different answers to the question what the meanings of a word are, may lead to the question what exactly those criteria are criteria of. If they result in contradictory statements as to what the meanings of a word are, not all of them can actually be criteria for meaning discrimination; in fact, only one could be the true test of polysemy. Pursuing this line of analysis, then, would mean eliminating the wrong criteria until the only true one is left over. Such an approach would still presuppose that words have a stable set of meanings, though. Hence, the

Vagueness's puzzles, polysemy 's vagaries

121

second road to follow would be to undennine this assumption even more than we have done already. More specifically, the belief in the stability of meanings will turn out to be ultimately spurious if the vagueness of the boundary between vagueness and polysemy can also be established when we restrict the analysis to just one of the three basic types of polysemy test (and not just when we compare the three among each other). Because this is a more radical and more fundamental departure from our traditional conceptions than the view that there is only one true kind of polysemy test, it is worthwhile pursuing this line of research first. The hypothesis, then, will be this: for each of the polysemy criteria, examples can be found in which what is a distinct meaning in one context, is reduced to a case of vagueness in another context (and vice versa). 6. I. The linguistic criterion Contextual influences on the linguistic test have been (implicitly or explicitly) noted by several authors. In fact, the recognition occurs relatively early in the literature on the subject. When Lakoff (1970) introduced the and soconstruction as a criterion for polysemy, he argued that hit is ambiguous between an intentional and an unintentional reading, because John hit the wall and so did Fred would constitute an anomalous utterance in situations in which John hit the wall intentionally but Fred only did so by accident (or the other way round). Catlin and Catlin (1972), however, noted that the sentence could easily be uttered in a context involving imitation. A situation in which John hits his head against the wall after stumbling over his vacuum cleaner and is then comically imitated by Fred, might very well be described by the sentence in question. Nunberg (1979) further drew the attention to sentences such as The newspaper has decided to change its size, which features intuitively distinct senses of newspaper ('management, board of directors' and 'material publication'). Of course, the point would be to show that the distinction between both senses is not just a matter of intuition, but that the linguistic test itself also supports the polysemy of the item under consideration. This can be done by concentrating on the asymmetries pointed out by Deane (1988). On the one hand, there are cases like the well-known The ham sandwich is sitting at table 5 and (with anaphora) The ham sandwich i tasted lousy, so hei demanded the check. These cases show neutralization of the distinction between the literal interpretation of ham sandwich and the customer metonymically connected with it. On the other hand, The ham sandwich i demanded the check because iti tasted lousy is ruled out.

122

Polysemy

While the examples presented so far involve anaphoric references, similar cases can be found involving coordination. For instance, Norrick (1981: 115) contrasted the (decidedly odd) sentence Judy's dissertation is thought provoking and yellowed with age with the (perfectly natural) construction Judy's dissertation is still thought provoking though yellowed with age. If the coordination generally requires that dissertation be used in the same sense with regard to both elements of the coordinated predicate, the sentences show that the distinction between the dissertation as a material product and its contents mayor may not play a role. Cruse (1982), in his turn, noted that none of the following series of sentences containing coordination produces feelings of oddity: (8) (9) (10) (11) (12)

(13)

John John John John John John

likes blondes and racehorses. likes racehorses and fast cars. likes cars and elegant clothes. likes elegant clothes and expensive after-shave. likes expensive after-shave and vintage port. likes vintage port and marshmallows.

Coordinating the first item in the series with the last, however, does produce an awkward sentence. So, while the awkwardness of John likes blondes and marshmallows would normally be taken as evidence for the polysemy of like, the pairings mentioned above suggest that there is a continuum of meaning rather than a dichotomy. Cruse concludes 'that readings which are close together can be coordinated without zeugma, but if they are sufficiently far apart, they are incompatible. If this picture is correct, it does not make sense to ask how many senses of like there are: there is just a seamless fabric of meaning-potential' (1982: 79).10 The previous examples sufficiently establish the context-dependent nature of the linguistic test. These contextual effects are fairly strong, in the sense that some contexts appear to be able to bridge the gap between widely divergent meanings. Deane (1988: 345) points out that explicit comparison is such a context; it can even combine actual homonyms. Examples (14) and (15) reproduce his examples. They can be contrasted with cases such as (16), in which the same meanings cannot be combined. (14) (15)

(16)

The arm ofa giant resembles that ofan ocean. Financial banks resemble those that you find by rivers: they control. respectively, the flow ofmoney and o[water. *The giant put his arm into that o[the ocean.

Vagueness's puzzles, po~vsemy 's vagaries (17)

(18) (19)

(20)

123

When Von Mi1nchhausen said a gigantic arm appeared before him when he came out ofthe jungle in Xanadu, he actually did mean the arm ofa giant, not that ofan ocean. The bank John had trouble finding again was not the one where he had cashed the cheque, but the one where he had moored his boat. Daddy, what exact!.v do you call a bank: the place where we moor the boat or the place where I bring my savings? - Well, son, the place where we moor the boat is a bank, but so is the place where you bring your savings. Daddy, what precisely is a race: a group ofpeople with same ~vpe of body, or a competition in which you try to be the fastest.? - Well, darling, competition in speed is a race, and so is a group ofpeople with the same type ofbody.

Taking Deane's observations one step further, let us note that there are other, comparison-like contexts that have similar effects. First, there are contexts of explicit contrast. For instance, in my earlier description of the difficulty of combining an intentional with a non-intentional reading, I used a sentence that did precisely that: I talked about situations in which John hit the wall intentionally but Fred only did so by accident. The case can be generalized, as illustrated by (17) and (18). A second type of context that exhibits the same strong neutralizing effect involves definition and explanation. Consider the exchanges in (19) and (20). Although it would be correct to remark that this is at least to some extent a case of metalinguistic use, this is hardly consequential for our interpretation of the possibility of identity-ofsense anaphora: the metalinguistic explanation or definitional comment is not realized as a metalinguistic statement about the words bank and race, but rather takes the foml of a categorial statement about category membership.ll At this point, two important remarks suggest themselves. First, we now have a basis for an evaluation of the position taken by Bosch (1979). His views are important, because they are the only ones (to my knowledge, at least) that explicitly take into account the contextual focusing effects that we have been considering here. Bosch's suggestion boils down to replacing the dichotomy between vagueness and polysemy by a threefold distinction, based on the presence or absence of contexts that bridge the gap between two interpretations of a predicate. Concentrating on do so ellipsis, Bosch proposes to talk of polysemy when there is no focusing context that could bring together the two distinct interpretations, of vagueness when there is no context in which they can be kept apart, and of contextually determined meanings when there is at least one context in which the contextual focus allows for a combination of both readings. But as the foregoing suggests that there is

124

Polysemy

always such a context (viz. when we are focusing on comparison, contrast, or explanation), the tripartite distinction is undermined. And even if the neutralizing effects of these contexts were not universal, they seem to be so strong that stable polysemy in the sense outlined by Bosch would turn out to be a relatively marginal case in the lexicon as a whole. Second, the kind of examples that Nunberg and Deane concentrate on raises the question whether the linguistic test (in its identity-of-sense anaphora form) is a test of polysemy at all. Is it correct to talk about such anaphoric constructions as involving 'identity of sense', when they bring together such blatantly different things as newspaper directors and paper copies of the newspaper, or even a river bank and a financial bank? Using them as evidence in a semantic analysis means treating them as categorial statements - but are they? Although Deane (1988) continues to treat these constructions as examples of polysemy, his own explanatory analysis suggests an alternative. On the basis of relevance-theoretic concepts and an explicit theory of attention, he arrives at a theory that predicts which anaphors will be acceptable. What appears to be at stake, however, is less sense identity than what might be called 'referential accessibility'. Crucial for the acceptability of constructions like The newspaper decided to reduce its size is neither the semantic (categorial) identity of the newspaper and it, nor their coreferentiality, but the question whether the referent of it can be accessed on the basis of the interpretation of the newspaper (and given certain restrictions on the flow of attention). This reinterpretation of 'identity of sense' anaphora as 'accessibility of reference' anaphora is an example of the reanalysis of the semantic tests that was suggested in the introductory passage of this paragraph. As already indicated, we will not pursue this line of thinking systematically, but the example given here may help to clarify what is involved in such a reinterpretation process. 6.2. The logical criterion There are two types of internal inconsistencies within the logical approach to polysemy. On the one hand, readings that are indistinguishable in most contexts may be separated in specific circumstances. On the other hand, readings that are mostly kept apart may be combined into a single sense in specific contexts. We have already seen an example of the first type. Whereas the concept 'bird' is usually used univocally when the biological category is meant (in the sense in which a sparrow is a bird and not a bird is unintelligible), it is

Vagueness's puzzles,

po~vsemy's

vagaries

125

possible to come across exchanges such as (21), in which non-prototypical exemplars are implicitly contrasted with typical exemplars. Furthermore, hedges and definitional specifications provide contexts in which 'p and not p' constructions can be fitted easily: see the examples (22), (23) and (24). (21)

(22) (23)

(24)

Is a penguin a bird.? - Well, it's a bird and not a bird. Is a whale ajish? - Irell, technical~y speaking it isn't, but loosely speaking you might say it is. Is a kiwi a bird? - Well, strict~v speaking it is, but it certainly isn't a bird par excellence. 1 bought this really expensive jade necklace last year, and it is certainly worth its price for all its beau~y and srylish appearance. Still, chemically speaking, is this rea/~v jade? - Well, it is jade i{you mean nephrite, but not ifyou mean jadeite.

The final example may require some comment. The traditional precious stone category jade may refer to two chemically different substances: nephrite (silicate of calcium and magnesium), and jadeite (silicate of sodium and aluminium). In outward appearance (hard and green) and function (used as ornament or for implements), both substances are similar; as such, ordinary parlance maintains the traditional unitary, non-polysemous character of the category. However, in specific circumstances such as the one evoked in the example, the chemical differences between both substances may become important the conceptual distinction that is then profiled obviously influences the possibility of the 'p and not p' construction. For the second type of internal instability of the logical criterion, it is natural to turn to cases in which two conceptually distinct readings occur together in one context. First, let us have a closer look at the second major variant of the test that we introduced in section 4.2. It can be shown, in fact, that it runs into difficulties precisely in cases where the two readings whose relationship we are trying to establish, occur together. Following a suggestion made by Kempson and Cormack (1981), we noted in 4.2 that the negation of a vague predicate p leads to conjunctive implications of the 'not PI and not p/ kind, whereas the negation of a polysemous predicate entails a disjunctive implication of the 'not PI or not P2' kind. It is easy to see that the latter disjunction is not exclusive; the negative sentence They did not reach the bank in time would still be true if they neither reached the financial bank nor the river bank. To take this fact into account, the implication frame for polysemous predicates could be rendered as 'not PI or not P2 or not (PI and P2)'. The contextual transformation of polysemy into vagueness then takes place in those situations in which the single alternatives that feature in the

126

Polysemy

disjunction ('not PI but possibly P2' on the one hand, and 'not P2 but possibly PI' on the other) are discarded in favor of the third alternative ('neither PI nor P2')' This will be the case, for instance, when They did not reach the bank in time is uttered in a situation in which the financial bank to be reached is situated on the bank of a river. Because in this particular context, the implication would be of the 'not Pl and not P2' type, bank would have to be considered vague rather than polysemous in the context, given at least that an implicational frame of the 'not PI and not P2' type is taken to be an indication of vagueness. Further, let us consider the other variant of the logical test that we introduced in 4.2. Remember that we assume that there is polysemy if the 'p and p' construction is not considered awkward or redundant. However, there are indications that this criterion needs further refinement. Let us have a closer look at the Dutch word vers, which we already briefly encountered in section 4.1 (for a full analysis of vel'S, see Geeraerts 1989a). Consider the following (quoted) examples. (25) (26)

(27)

Wie eet er de beste kost, wie koopt er verse vis en wild? 'Who can eat the finest food, who can buy fresh fish and garneT Nu meende zij, dat zij aan alles zou kunnen wennen, behalve aan gemis van beweging en verse lucht. 'Now she thought that she would be able to get used to everything, except lack of exercise and fresh air. ' Voor een goede verpakking dienen de verse sigaren tot zolang in vochtige, weke toestand te blijven. 'In order to be packed properly, the newly produced cigars have to remain soft and moist until packing. '

Example (25) illustrates the most frequent reading of vers, the one that is invariably the first to be found in dictionaries and to be remembered when informants are asked what vers means. It is used with regard to foodstuffs, and indicates that the latter are recently produced (caught, bought, harvested etc.), and are therefore in optimal condition for human consumption; one typically speaks of fruit, meat, bread etc. as being vel's. Quotations (26) and (27) show that the two notions that are combined in (25) can occur separately and independently. In (27), the notion of recency occurs without the notion of optimality, and in (26), the reverse is the case. Intuitively speaking, the central applications of vers constitute a distinct meaning: there is no spontaneous recognition of an ambiguity when one mentions fresh food, to the extent that it would be absurd to ask' Vel'S in what sense 7' when there is

Vagueness's puzzles, po~vsemy 's vagaries

127

talk of vers voedsel. Is it possible, then, to apply the logical test in such a way that the non-polysemy of the central applications is recognized? In all the circumstances in which vers could be used in the central meaning, the maximal readings whose overlap actually constitutes the central application apply. So either the 'p and p' construction can be used without awkwardness, and then the singularity of the central applications is not recognized by the logical test, or the 'p and p' frame cannot be used, and then we have a case of contextual neutralization of the distinction between the 'recency' application and the 'optimality' application (a distinction that can be firmly established when the logical test is applied in other contexts). Now, it would seem that the first alternative applies. By uttering, for instance, Dat verse brood dat ik net van de bakker heb meegebracht smaakt lekker vers ('the fresh bread that I just brought home from the baker's tastes deliciously fresh') one can enforce a construction that combines the meaning 'newly produced' in the first occurrence ofvers, with the meaning 'optimal for consumption' in the second occurrence. However, in such a case, we could salvage the logical test by imposing a restriction on it: regardless of the possibility of uttering 'p and p' without awkwardness or redundancy, the simultaneous applicability in a particular context of two interpretations of p does not constitute a case of polysemy if the two interpretations are more closely connected than merely by juxtaposition. What we would like to recognize is the 'fusion' of the two interpretations into a single one, and such a fusion is signaled precisely by the fact that there is more to the combination of those interpretations than merely their simultaneous presence. In the case of vers, for instance, the two separate readings are fused in the central app lications, because there is a causal connection between them: fresh fish is not just recently caught and (by coincidence) optimal for consumption, it is optimal in taste and nutritional value precisely because it is recent. If this is accepted, the logical approach appears to be instable: in its 'p and not p' form, it establishes the distinction between the readings 'optimal' and 'recent' of vers, but when we look at the simultaneous applicability of these interpretations, the restriction just formulated leads to the conclusion that there are contexts in which the distinction between both interpretations is overruled. It may be noted that we have now arrived a point where the logical test leads to difficulties no matter how we look at it. Suppose, on the one hand, that we simply stick to the 'p and not p' form of the test (ignoring the 'p and p' version), or that we accept the test in its 'p and p' format, but without the revision just suggested. In these cases, the intuitive univocality of the central

128

Polysemy

cases ofvers is not recognized. On the other hand, a revision of the 'p and p' test leads to the conclusion that the central application of vers is a separate meaning, and not just an ambiguous combination of the other two. As such, the distinction between the 'optimal' reading and the 'recent' meaning is neutralized in the contexts constituted by the central applications of vers. So either we repeat the observation from paragraph 5 (viz., that the results of the logical test may contradict those of other polysemy criteria), or we have to accept the contextual instability of the test. Now, while such a neutralized reading might seem to be less than damaging if it is permanently available as a third major meaning ofvers, it is unfortunately the case that the neutralization can be entirely contextual, in the sense that it is incidental. As an example of such a contextual neutralization that is like the vers case in all respects except for being purely incidental, let us have a look at the following passage from P.c. Hooft's play Granida. (Hooft is one of the major seventeenth-century Dutch authors.) The shepherd Daifilo talks to his master about his love for the princess Granida. (28)

Mijn heer, den laetsten nacht passeerd'jck bijder straten Onder de venster van mijn vrouw, - (... ): Mijn Vrouwe stondt voor 't glas, en soo sij mij sach gaen, Sondt sij haer voester at, om mij te roepen aen. Mijn Vrouw vernederde'haer, en quam mij selve vraeghen, Wat oorsaack inder nacht mij daer ontrent mochtjaeghen. [My lord, last night I passed through the street Under the window of my lady: My Mistress was at the pane, and when she saw me going there, Sent her servant down to call me. My Mistress humbled herself, and came to ask me herself What reason might have driven me there at night.]

The word we are concerned with is vernederen. Literally, it means 'to bring down' (one may recognize the English nether in the Dutch neder), but already in the seventeenth century, the figurative meaning 'to humiliate, to humble' is the primary one. In the quotation, vernederen is used reflexively; both the literal and the figurative reading are relevant. On the one hand, Granida comes down from the upper storey to meet Daifilo, but on the other hand, she humbles herself precisely by taking this step: as a lady, it is below her status to go out to meet someone of humble standing like Daifilo. Because she abandons her figuratively superior position by leaving her literally superior position, the two readings of (zich) vernederen are fused, and we have a context in which basically separate senses are combined into a single

Vagueness's puzzles,

po~ysemy 's

vagaries

129

reading, in the same way in which the two basic meaning components of vers are fused in the central meaning of that item. And as in the case ofvers, the fusion involves a causal connection between the fused readings: Granida humbles herself precisely because she comes down to meet Daifilo. Although most studies on polysemy would probably ignore cases such as this one as deliberate literary puns, the remarkable character of the simultaneous applicability of both readings of vernederen need not be based on the semantics of the construction, but may simply be a consequence of the novelty of this kind of use. Admitting that there is indeed something punning in the passage quoted above (on this point of literary style, the authors of the Dutch Golden Age are no different from their English counterpart), I do not think that we are obliged to attribute the pun to the polysemy of the construction. If a punning effect may simply be the result of unexpectedness, the remarkable character of vernederen may be a straightforward result of the unusual character of the fusion. In contradistinction with the vers case, the fused combination is not well-entrenched in the mental lexicon - but that need not mean that it is less a distinct meaning, at least not at this point in the discussion. (In section 7.1, we will explore the consequences of restricting the epithet 'meaning' to readings that are pennanently stored in the mental lexicon.) To sununarize, let me repeat the steps in the argumentation. First, the logical test is not only applicable according to a 'p and not p frame', but should also take into account cases in which two readings of a predicate p are simultaneously applicable. Second, in the latter case, polysemy does not occur when the two interpretations of p fuse coherently into a single one, as in the case of vers. Third, the incidental character of the fusion that we find in the case of vernederen underlines the unstable character of the distinction between vagueness and polysemy. More generally, vernederen is an example of contexts that may join readings that are usually separated, in contrast with the jade example and similar ones, \vhich distinguish what is usually considered identical. 6.3. The definitional criterion Let us now turn towards the contextual variability of the definitional test. Though it might superficially seem that this polysemy criterion is entirely objective and decontextualized, this is hardly the case: because the criterion requires that all the definitions that we take into account in order to establish the polysemy of an item are themselves unambiguous, the test may yield

130

Polysemy

different results according to whether we are able to accept a definition as monosemic or not. The requirement in question again refers to a traditional one in the literature on definitions: 'if the definiens is itself ambiguous, the definition obviously fails to perform its function of explaining the definiendum' (Copi 1972: 138). If the definition of what is allegedly a single sense of a lexical item is itself polysemous, the definition may either leave in the dark which of the various polysemous alternatives actually applies to the definiendum, or it may cover up the actual polysemy of the latter. The first difficulty occurs, for instance, when dark is defined as 'not light': because not light is ambiguous between 'dark' and 'heavy', the definition of dark does not prevent the meaning 'heavy' to be attributed to that word. It is the second difficulty, however, that is of interest to us here. Suppose you are describing a language with a word darvy that may be used both with the meaning 'dark' and with the meaning 'heavy'. A definition of darvy as 'not light' might then be pragmatically useful in a dictionary, because it would make economical use of the polysemy of the English expression not light to describe the range of application of darvy. In the context of the definitional test of polysemy, however, defining darvy as 'not light' would be misleading: instead of establishing the monosemy of darvy, the superficially unitary character of the definition would merely hide the actual polysemy (or even homonymy) of darvy. A real example of this kind of hidden polysemy may be found in Wierzbicka (1990). It will be recalled that Fillmore (1982) showed that bachelor (an old favorite of lexical semanticists) can only be defined as 'unmarried male' if the definition is made relative to a normal model of the world in which there is a social institution of marriage as a union of two people of opposite sexes, in which people have to reach a certain age before they can get married, and so on. Prototypical fuzziness in the application of bachelor then arises when this background model is applied to cases where the background conditions fail. Is a member of a homosexual couple (or a male adolescent of 13 or 14) a bachelor? In the sense that he is unmarried, perhaps, but certainly not in the prototypical sense of the word. Wierzbicka then argues that it is not necessary to make a distinction between the actual definition of bachelor and the related set of background assumptions (the Idealized Cognitive Model in the sense of Lakoff 1987) to explain the fuzziness of the word: a simple monosemous definition of bachelor as 'an unmarried man thought of as someone who could marry' (1990: 349) suffices, because the additional phrase 'thought of etc.' takes care of the fuzziness. Notice, however, that this is a prime example of hidden polysemy, if only because it

Vagueness's puzzles, po~ysemy's vagaries

131

includes the verb can. If one utters Yes, John can swim, various things could be meant that clearly need not all be true at the same time. It might mean (among other things) that John is objectively speaking not in a situation that impedes his swimming (he is not lying on a hospital bed with two broken legs), that he knows how to swim (he has acquired the skill), that he has the permanent right to swim (he is a member of the sports club that owns the pool), or that he has the momentaneous permission to swim (the bath attendant has made it clear that the pool is no longer overcrowded and that John may now dive in). Now, if these are distinct meanings of can, the definition of bachelor as 'an unmarried man thought of as someone who could marry' copies the polysemy into bachelor, and hides it at the same time. Also, the distinction between the various ways in which a person could marry is not irrelevant for the internal structure of bachelor, in the sense that judgments about bachelorship are related to it. On the one hand, consider Tarzan, who cannot marry because the objective conditions for marrying (such as the presence of a judge, parson, priest or whatever) are not met. On the other hand, consider a young man who does not get his parents' permission to marry his beloved. Although both cases involve a person who cannot marry in one of the various senses of can, the epithet bachelor can be applied much more easily to the latter case than to the former. If the example can be generalized, this means that the 'permission' sense of can is less important for the structure of bachelor than the 'objective possibility' sense. 12 The crucial point to make is this: if the definitional test is applied unthinkingly, a decision on the polysemy of the lexical item bachelor will contextually depend on our ability to recognize the polysemy of can - and such a recognition is initially an intuitive, individual matter: Wierzbicka, for one, seems not to have noticed it. But couldn't the recognition of such ambiguities be made less intuitive? If we are using a definitional criterion for determining the polysemy of a lexical item but have to determine independently whether the definitions that we consider are not themselves polysemous, two basic possibilities are available. First, as in our example, we may resort to other types of polysemy tests: for instance, when we establish that Yes, John can swim is polysemous because it could be both, in a given situation, true and false, we are relying on the logical test. But this would mean that the reliability of the definitional test of polysemy could be reduced to the reliability of the other criteria. Now, not only do we have independent evidence for the contextual instability of these other tests, but also, at this point in our argumentation we are specifically concerned with evaluating the definitional test in itself and by itself.

132

PoZvsemy

Second, then, an iterative application of the definitional test would not really help either, since this would merely set us off on a series of endless repetitions. Or at least, at some point an intuitive recognition of the univocality of the words we are using in our definitions would have to intervene to stop the endless regression. But precisely because we are relying on intuitive judgments of ambiguity rather than on tests and operational criteria, the possibility of contextual and individual oversight is not excluded. Moreover, if decisions about polysemy ultimately depend on an intuitive recognition of univocality, why could we not simply apply that intuition directly to the lexical item whose polysemy we are trying to establish? The whole point about considering operational tests of polysemy is to get away from the immediacy of simple intuitive judgments about polysemy. So either we accept that the ultimate validity of the definitional test is based on intuition (but then we have not got much of a test left), or we invoke the other tests (but then the singularity of the definitional test disappears, and all the problems connected with the other tests loom as large as ever).13

7.

Additional perspectives - The representational escape hatch

Because the extreme contextuality of meaning that is implied by the internal and the external inconsistencies of the three polysemy tests implies a rather drastic departure from traditional conceptions of semantic structure, it is important to proceed carefully and investigate any remaining possibility of salvaging a stable notion of lexical meaning. Taking for granted, then, that the tests for polysemy that we have discussed so far are sensitive to contextual effects, an entirely different approach suggests itself. If the 'meanings' we have been investigating are contextual surface phenomena, what model of semantic memory do we have to assume to explain the occurrence of those meanings? If the model is to be flexible, it would have to contain rules to derive flexible interpretations, together with elements that the rules operate on. What surfaces in a particular utterance is then nothing else but either one of the permanently stored elements, or an interpretation derived from applying the rules for meaning extension to one of those stored elements. In such a model, the surface-structure interpretations could be distinguished from the deep-structural meanings,14 and the polysemy problem would boil down to the question how to delimit the proper domain of the storage component from that of the rule component. For instance, if literal senses are stored, figurative ones could be derived by rule and need not be stored. In this

Vagueness's puzzles, po~vsel7lY 's vagaries

133

framework, we might try to circumvent the problematic question 'What is a distinct surface interpretation?' by merely asking 'What is a distinct meaning on the deep level of storage?'. The answer to the question is, in fact, easy in principle: an interpretation represents a distinct meaning if it cannot be derived by rule. Before trying to evaluate this approach, let us note, first, that it can be seen as belonging in the class of representational criteria for meaning discrimination: what ultimately decides on the semantic status of an alleged meaning depends on our way of representing lexical-semantic phenomena. Instead of consisting of simple definitions (as in the type of representational criterion that we discussed before), the representation now takes a 'distributed' form, part of the representational burden resting on the storage component of the mental lexicon, the other part being taken care of by the rule component. Second, it may be noted that this representational approach is not uncommon in Cognitive Linguistic debates. For instance, Sweetser (1987) is an attempt to reduce the polysemy of lie as described in Coleman and Kay (1981) to a set of contextually derived specifications of a single Idealized Cognitive Model for lie. And Vandeloise (1990) confronts the analysis of over that is included in Lakoff (1987) (and that is itself based on Brugman 1981); against the multiplication of the polysemy of over that he spots in Lakoff's description, Vandeloise tries to give a monosemic description of prepositions. What we will be concerned with here is not the detail of these debates, but the general strategy that consists of, first, defining the polysemy issue as a question involving penllanently stored meanings rather than surface structure interpretations, and second, assuming that what can be derived procedurally from stored information need not itself be stored. It is important to stress that the following objections against this view do not involve the distinction between stored and derived information as such, but rather the idea that the strategy as just defined somehow solves the polysemy Issue. The two components of the strategy can be evaluated separately. With regard to the first component, one reason for being careful with the representational approach is the fact that it seems to be based on a false analogy with phonology. In the same way that allophones can be contextually derived from underlying phonemes, contextually different interpretations of a lexical item are thought to be derived from underlying, stored meanings. Furthermore, there seems to be a functional similarity between allophony and semantic vagueness: the distinction between allophones (either in the form of free variation or in the form of contextual determination) is not a functional

134

Polysemy

one, and similarly, the contextual distinction between a male and a female interpretation of neighbor is not semantically functional, in the sense, for instance, that no truth-conditional differences are considered to correlate with it. Not surprisingly, then, the contextually determined readings are sometimes called allosemes (among others, see Deane 1988). The analogy breaks down, however, when the functional status of the allo-entities is taken into closer consideration. Take the well-known case of conjunctions. In Horn (1985), it is argued that sentences such as 5J'he 's not happy, she's sad versus She's not happy, she's ecstatic do not feature a distinct meaning of not, because the differences of interpretation can be derived from the context. Prominent among Horn's reasons for merely calling not 'pragmatically ambiguous' rather than polysemous is the fact that its actual interpretation can be derived from pragmatic principles. Now, there is a clear truthconditional distinction between the two sentences that corresponds precisely with the 'pragmatic ambiguity' of not: as the addition of the sentences' tails indicates, different truth conditions hold for she's not happy according to whether not is taken in its ordinary reading or in its metalinguistic reading. We may conclude, in other words, that not all allosemes are functionally neutral; as such, alloseme is a somewhat misleading term. To be sure, this is not an ultimately compelling objection against the representational conception of polysemy. It does not show that it is impossible to proceed in the representational way, but rather that it may be less important to do so than we might think on the basis of the phonological analogy. If the allo-entities are not structurally functional, they should not receive most of the emphasis in linguistic descriptions. But if, in the lexical-semantic domain, we conceive of contextual interpretations on the model of allophones, we run the risk of underestimating their functional importance. A similar risk is inherent in the representational reinterpretation of the tenninological pair vagueness/polysemy: because vagueness is considered the non-functional phenomenon in comparison with meaning, redefining meaning in terms of storage may make us lose sight of the fact that some of the non-stored readings are just as functionally important as the stored ones. In other words, the question 'What is a distinct meaning?' is then no longer more linguistically relevant than the question 'What is a distinct interpretation?'. Or, to put it in yet a different way, if we think of meanings as the semantically functional phenomena, it would be wrong to equate meanings and stored readings. The storage question is not an unimportant one, but it is not necessarily a question about polysemy, if polysemy is defined in functional terms. Even regardless of the terminological matter whether meaning should be defined in

Vagueness's puzzles, po~vsemy 's vagaries

135

such a functional way or not, an investigation into the functions of utterance interpretations cannot be avoided. IS So, if the difficulties we have encountered so far are interpreted as difficulties concerning the question 'What is a different surface interpretation?' (rather than 'What is a different stored meaning?'), they remain unresolved. 16 Now, with regard to the second component of the representational strategy, let us note that the representational approach as defined above is an example of the 'rule/list fallacy' (Langacker 1987: 29). If the impossibility of deriving an interpretation by rule is the criterion for distinguishing polysemy, the crucial assumption is that the set of stored meanings and the set of derived interpretations do not overlap. For what would be the status of a reading that is both stored and derivable? It would contradictorily be both a meaning and not a meaning. So, if the rule/list distinction can be shown to be untenable in this lexical-semantic context, our problems with the question 'What is a distinct meaning?' remain unresolved. Consider, in this respect, the fact that lexical items may change their prototypical centre in the course of their diachronic development. The process typically involves a shift from the periphery to the centre of one of the applications that is initially an extension of the original central sense. (Detailed examples of such processes may be found in, among others, Winters 1989, Melis 1990, Geeraerts 1990b). At the same time, this shift typically involves frequency effects: even though dominant frequency of occurrence is neither a sufficient nor a necessary criterion for the centrality of an item's reading, rising frequencies tend to correlate with growing centrality (cp. Geeraerts 1990b). If a particular peripheral meaning gets to be used more often, and if, at the same time, the originally central meaning gradually lapses into disuse, a conceptual reorganization takes place resulting in a shift of that application towards the centre of the category. But if the frequency with which an application occurs influences its prototypical status, this implies that there is some mechanism in our conceptual memory for keeping a count (however roughly) of the application's frequency of occurrence - and this would be the case regardless of the application's rule-governed nature. If changes in the frequency with which we use an application (or hear it being used) may result in a process by which the application becomes more central in our semantic memory, we are somehow aware of that reading as an individual element in our mental lexicon: the account we apparently keep of the reading's frequency of occurrence presupposes a representation of that reading - even if it is entirely derivable from a given stored meaning, a particular context, and a specific procedure for semantic extension. In short, there are reasons to

136

Polysemy

doubt that a representational conception of univocality, founded on a dichotomous conception of the rule/list distinction, will yield satisfactory results. 17

8.

Consequences

The consequences of the argumentation in the foregoing pages can be traced on two different levels. First, we may have to change our (pre-theoretical or theoretical) conception about what meaning is. More specifically, we will have to determine the impact of the instability of the distinction between vagueness and polysemy on Kleiber's model of prototypicality effects. Second, there may be an influence on our model of what semantics is, i.e. on our conception of the methodology of lexical-semantic research. And here as well, an interesting consequence for prototype theory may be envisaged. 8.1. Consequences on the pre-theoretical and the theoretical level If it can indeed be accepted that the distinction between vagueness and polysemy is not well-defined, the primary conclusion to be drawn relates to the problem that initially engendered our discussion of polysemy. Because the instability of the distinction between vagueness and polysemy precludes a strict dichotomy between intercategorial and intracategorial semantic multiplicity, Kleiber's conception of prototype theory need not worry us too much for the time being. It will be appreciated, however, that this conclusion is considerably hedged in. On the one hand, the fact that we have not found a stable criterion for a distinction between vagueness and polysemy does not mean that such a criterion is in principle impossible to find. If we have not found a stable operational test yet, we have perhaps not looked hard enough. Of course, our pre-theoretical notion of what a distinct meaning is, is not entirely clear, and in that sense, any search for polysemy criteria will be hampered by the fact that we do not know precisely what it is we want a criterion for. But this is not an uncommon situation in linguistics, and so further research into (new and old) polysemy criteria retains its importance. Needless to say, research into polysemy criteria also implies an investigation into the question what the criteria we originally considered to be polysemy criteria are actually criteria of, if not criteria of distinctness of meaning. On the other hand, our rejection of Kleiber's attempt to bring some order in prototype theory does not mean

Vagueness's puzzles, polysemy 's vagaries

137

that all is well in prototype country. Here too, further research into the relationship between the various sorts of prototypicality effects and their sources remains necessary. The main point we can claim on the basis of the foregoing pages, is that Kleiber's attempt to bring order into the prototypical matter seems to meet with fundamental difficulties - not that prototypicality is in no need of further investigation. This is all the more so because our own findings suggest modifications in our conception of lexical structure. After all, the idea that meanings (as distinct from mere referential specifications) do not exist, is rather disconcertingly at odds with what we traditionally believe. So let us try to determine what other consequences this reversal (conditional though it may be) leads to. One major thing to be changed, I would claim, is our metaphorical, pretheoretical image of what meanings are. Come to think of it, the very wording of the previous sentence reveals something about the way we talk about meaning: the plural meanings suggests that meanings are entities or entitylike things, presumably stored somewhere in the big storage room that we conceive the mental lexicon to be, to be carried by word forms to an interpreter who unpacks and understands them. This metaphorical image has, of course, been identified and labeled before: it is both that of the conduit metaphor of verbal communication, first described by Reddy (1979), and of the lexemist conception of the mental lexicon, more recently analyzed by Hoffinann (1989). The common characteristic of both metaphorical complexes is that of reification: meanings are things, prepackaged chunks of information that are contained in and carried about by word bags. But what kind of image could replace this reificational picture? The tremendous flexibility that we observe in lexical semantics suggests a procedural (or perhaps 'processual') rather than a reified conception of meaning; instead of meanings as things, meaning as a process of sense creation would seem to become our primary focus of attention. The image I would like to propose to make this conception more graspable, is that of a floodlight: words are searchlights that highlight, upon each application, a particular subfield of their domain of application. If this domain of application is seen, in model-theoretical fashion, as a set, each time the word is used a particular subset is selected. But while our traditional view of the distinction between vagueness and polysemy entails that the number of subsets that can be lit is fixed and restricted, we now have evidence that the verbal searchlight has much more freedom. The freedom is not absolute, surely, and there will be preferential subsets for each word; even so, the distinction between what can and what cannot be lit up at the same time is not stable. IS

138

Polysemy

Although we should obviously try to move beyond a mere metaphorical description of our new conception of lexical- semantic structure, the actual development of a model of categorial structure that takes into account the shifting nature of meaning lies beyond the reach of this paper. It may nevertheless be noted that, within the broad range of prototype-oriented studies, the 'vantage point theory' presented by Robert MacLaury (1991) is perhaps the one that is most congenial to this enterprise (most explicitly so because it provides for perspectival shifts of the prototypical centre of a category). 8.2. Consequences on the methodological level Accepting that the distinction between vagueness and polysemy is subject to contextual variation highlights the interpretative nature of semantics: meanings are interpretations that arise out of a particular context. But if meaning is radically context-dependent, what is the methodological status of meaning - given that it is us, the interpreters, who choose or provide the contextual perspectives that lead to a particular interpretation? Are meanings phenomena that are objectively out there and that can be discovered, or are they subjectively construed by the interpreter? Are meanings found or are they made? The philosophical problem tumbling head over heels into the discussion here is that of the objectivity of lexical semantics - not just the objectivity of the object of semantic research, but the objectivity of its method. Cognitive Linguistics generally follows Lakoffs (1987) characterization of the enterprise as an anti-objectivist one, but it is seldom realized that this characterization can be taken in two different ways. Usually, the anti-objectivism is taken to be a characterization of the object of the enquiry. But even if linguistic meaning is conceived of as a subjective, experiential, psychological phenomenon, does this also mean that the method of linguistic semantics is non-objective? After all, even if the meanings that we investigate are subjective entities, they might still be there objectively: we could find the meanings as they are, and we would simply have to extract them from the messages in which they are transmitted to us. However, this picture smacks of the reified conduit metaphor and lexemist model that we rejected earlier. What, then, if we do not find objective meanings ready-made, but construe them subjectively? First, this would be naturally in line with the rejection of the reificational lexemist/conduit model: while we primarily rejected the model as a picture of how meanings are put into words, we can now extend this rejection to the way in which meanings are extracted from words. And second,

Vagueness's puzzles, polysemy 's vagaries

139

extending the anti-objectivist, anti-reiticational stance to the receiver's end of the communicative process makes good sense of the difficulties we encounter when trying to fix the meaning of utterances like the ones we have been discussing throughout the paper. What we started looking out for was a unique meaning, i.e. the meaning of the utterance, the authentic and original one that the author put in the message (or perhaps: highlighted) and that is objectively there for us to be unpacked (or alternatively: observed). But, disconcertingly, we have serious difficulties in pinning down the unique meaning of an utterance, not least because the various perspectives that we adopt yield different answers to our semantic question. However, if we abandon the vestiges of objectivism in our methodological self-conception, the presupposition that there is a unique meaning itself can be rejected. Rather than a single unique meaning, there would only be the interpretations that we impose on the material - and our interpretative activities need not yield a unique result. It is not the intention of this paper to settle this question, but just to further illustrate its importance, two points are worth mentioning. First, the interpretative nature of the meanings we attribute to utterances has a bearing on the status of the prototypical structure that we attribute to lexical categories. In those cases where we do not rely on 'hard' techniques such as statistical analysis to determine the core meaning(s) of an item, but rather use traditional linguistic methods of data gathering and description, the question often arises 'How exactly do you determine what the prototypical centre of an item is?'. Now, if the prototypical core meaning of an item is no more objectively traceable than the meaning of a word in an individual utterance, could it not be the case that we choose that meaning as prototypical that gives us the best starting-point for interpreting the various applications in which we encounter that item? Such a meaning should not only be applicable in a straightforward way to the majority of the applications in which we encounter the word, but should also be suited to arrive at an interpretation of novel applications by applying regular mechanisms of semantic extension to the prototype. In that case, the prototype is that meaning that best enables us to make sense (literally, as an interpretative process) of the various ways in which words are used. Or, in other words, the prototype would have to be thought of as an interpretative perspective that helps us to interpret the uses of a word. (Our analysis could then make claims to objectivity to the extent that it mimics the interpretative structure other language users impose on the category.)

140

Polysemy

Second, lurking round the corner here are three major philosophical strands in the development of modem thinking. First, there is the hermeneutical tradition as it stretches from Schleiermacher over Dilthey to Gadamer, Habemlas, and Ricoeur (for an English introduction to the tradition, see a.o. Bleicher 1980 or Ricoeur 1981: 43-62). A crucial aspect of the hermeneutical tradition (specifically as embodied in the Diltheyan distinction between Geisteswissenschaft and Naturwissenschaft) is the methodological characterization of the human sciences (explicitly including linguistics) as being based on processes of interpretation rather than on the objectivist methodology of the natural sciences. In the methodological self-conception of linguistics, this idea has hardly been more than an undercurrent in the major part of the twentieth century; see, however, Itkonen (1978, 1983) and Anttila (1985) for a defense of a hermeneutical conception of linguistics. Second, there is pragmatism, specifically in the form that our cognitive activities are guided by pragmatic interests. For if our interpretations depend on differing perspectives, the choice of a particular perspective may be guided by various interests. Third, the whole set of problems touched upon here extends towards the philosophy of science: the falsificationalistl paradigmatic theories of science (in the line of Kuhn 1970 and Lakatos 1970) precisely emphasize the idea that all scientific thinking starts from perspectivist assumptions. (See Geeraerts 1985a for a discussion of the relationship between prototype theory and this line of thinking in the methodology of science.) However briefly evoked, these philosophical connections highlight the fundamental issue raised by the polysemy problem. If the difficulties we have with the determination of lexical meanings are indeed indicative of the ultimately non-objectivist, perspective-bound, hermeneutically interpretative nature of linguistic semantics, a methodological reexamination of the enterprise imposes itself. If our understanding is perspectival, what perspectives do we actually use, and what perspectives should or could we use? And what pragmatic interests could lexical semantics serve? It is beyond the scope of this article to answer these questions, but their far-reaching character should be sufficient to establish the importance of a systematic exploration into the methodological ramifications of the polysemy issue.

Vagueness's puzzles, polysemy's vagaries

9.

141

An inconclusive overview

The preceding pages have taken us on a tour through the mysterious world of lexical meaning. An examination of three basic classes of criteria for distinguishing between polysemy and vagueness has made clear, first, that those three types of criteria are in mutual conflict (in the sense that they need not lead to the same conclusion in the same circumstances), and second, that each of them taken separately need not lead to a stable distinction between polysemy and vagueness (in the sense that what is a distinct meaning according to one of the tests in one context, may be reduced to a case of vagueness according to the same test in another context). Rather than leading to the assumption that only one of the three tests is the true criterion for distinctness of lexical meaning, these observations give rise to the suspicion that the distinction between vagueness and polysemy is indeed unstable. If so, lexical meanings are not to be thought of as prepackaged chunks of information, but as moving searchlights that may variously highlight subdomains of the range of application of the lexical item in question. Furthermore, the recognition that the distinction between vagueness and polysemy is not unproblematic is not just important for lexical semantics in a very general way, but has specific consequences for prototype-theoretical approaches to the study of word meaning. In particular, the suggestion (as put forward by Kleiber 1990), that various types of prototypicality effects should be kept apart on the basis of their intracategorial or intercategorial status, can now be seen to rely uncritically on the stability of the distinction between vagueness and polysemy.19 In general, in spite of its vital importance for any theory of word meaning, lexical semantics does not seem to have arrived at a consistent operational definition of polysemy yet. Within the non-objectivist framework of Cognitive Semantics, this observation leads to the suggestion that the methodological uncertainty surrounding the concept of polysemy is a side-effect of the interpretative nature of linguistic semantics. However, rather than trying to argue for this thesis as such, the paper has merely tried to evoke the methodological set of problems in the context of which it can arise as a hypothesis. By describing the operationally problematic nature of polysemy, it has tried to make clear that the methodological status of lexical semantics is in urgent need of further analysis.

142

Polysemy

Notes 1.

2.

3.

This formulation presupposes that category as used in the terms intercategorial and intracategorial is defined as 'lexical meaning, sense of a lexical item'. This is not terminologically unquestionable, because in actual practice, category may sometimes be used to refer to lexical items (whether polysemous or not) just as well as to the individual senses of such items; in the latter case, one might speak of 'semantic categories', in the fonner of 'lexical categories'. If such lexical categories are polysemous, they constitute a cohesive cluster of semantic categories. Although the il1ter-/intracategorial terminology equates category with 'semantic category', the distinction between both types of prototypicality that is at stake here, could also be expressed by distinguishing prototypicality within 'semantic categories' from prototypicality within 'lexical categories'. This is not just the case for prototype theory's investigation into degrees of membership. In the context of a logical approach to meaning, the vagueness of predicates like large is discussed in connection with the Sorites paradox: removing one grain from a large heap of sand does not imply that the heap stops being large, but after an indefinite number of steps, the process can no longer be iterated. How, in other words, should the meaning of large be described to accommodate this fact and avoid paradoxes? See Ballmer and Pinkal (1983) for a sample of this line of research. It may be noted that there is yet a fourth type of vagueness to be mentioned, viz. the interpretative indeterminacy that occurs when there is a contextually unresolved polysemy in an utterance. The types of vagueness mentioned in the text involve indeterminate relations between lexical items and their referents (the relation between language and the world, if you wish). But indeterminacy may also include the relationship between the observer (say, the linguist) and the language, in those cases where it is difficult to determine which reading of an item is realized in a particular context. If, in a particular context, it is difficult to make out whether a particular predicate p exhibits reading PI or pz, this could either be because the distinction between PI and P2 is semantically neutralized in that context, or because the context does not provide enough information to choose between both. Whereas the first case represents the kind of vagueness that we will be concerned with in the rest of the paper (the speaker or writer intended to be neutral with regard to the distinction between PI and P2), the second case represents unresolved ambiguity (the addressor of the message intended to communicate either PI or P2, but, for lack of sufficient clues, the addressee cannot decide which is the case). An examination of the relationship between both types (which is interesting primarily for the methodological issues raised in section 8.2) is beyond the scope of the present text. It might be added at this point that the tenninologi-

Vagueness's puzzles, polysemy's vagaries

4.

5.

143

cal difficulties that are mentioned with regard to the term vagueness also occur with regard to the term ambiguity: while it sometimes used as a synonym of polysemy, it may also occur with different readings (for instance, as a specific term for the type of unresolved polysemy discussed in this footnote.) Because of the comprehensiveness of Cruse's (1982) treatment, it may be useful to give a brief indication of the relationship between the tests distinguished by him and the major types distinguished here. While he does not devote attention to the definitional approach, Cruse's tests Band D correspond with our logical and linguistic test respectively. His criterion A will be incorporated here as a specific subtype of the logical test (viz.. involving negation; see section 4.2). As for Cruse's criterion C, I would argue that it as well is a variant of the logical test to the extent that it involves the possibility of spontaneously recognizing (and asserting), in the same situation, two readings with different truth-conditional properties for a given sentence. Specifically, this same situation here takes the form of a sentence conjoined to the one featuring the predicate whose polysemy we are trying to establish. Consider the anomaly of The children wore red shirts, and the boys, blue ones in comparison with The dogs wore red collars, and the bitches, blue ones. The latter sentence is true-or-false in the reading in which dogs means 'male Canis familiaris', but anomalous in the reading in which dog means 'Canis familiaris' in general; the anomaly derives from the fact that a general reading for dog entails that bitches wear red collars, which is in contradiction with the second conjunct, specifying that bitches wear blue collars. (This is, of course, merely an encyclopedic contradiction, not a logical one.) So, in the same situation (the one in which bitches wear blue collars), the sentence The dogs ~j.'ore red collars can be spontaneously recognized as either true or false, or anomalous. As explained in 4.2, this is one of the realizations of the logical test. In the case of The children wore red shirts, on the other hand, no two interpretations with distinct truth-conditional properties can be spontaneously recognized in the situation in which the boys wear blue. The classification of polysemy criteria presented in Cruse (1986: Chapter 3) differs on two fundamental points from the one in the article of 1982. First, the direct criteria are divided into three classes. One of these corresponds with our 'linguistic' class, while a second one includes our class of 'logical' and related approaches. The third criterion is a novel one, to the extent that it invokes a distinction between semantic specifications that are derivable from context, and those that are context-independent. We will come back to this criterion in paragraph 7: in particular, see note 16 below. Second, and more importantly, Cruse' s argumentation in the book of 1986 seems to be partly convergent with the argumentation in the present article. to the extent that his text contains some observations questioning the idea of a clear dichotomy

144

6.

7.

8. 9.

Polysemy

between polysemy and vagueness; more specifically, he describes his own view as a 'compromise position' (1986: 81). Note that the vers case has more drastic consequences than the old case. It could be argued, in fact, that we should not consider relations such as antonymy separately, but that the entire set of relations that an item contracts with other items should be taken into account. Thus, although old may have identical antonyms in the two readings involved, the relationship of synonymy might be invoked to disambiguate between both readings, among other things because the synonym former only applies in one of the two. Even if this global approach would enable us to circumvent the class of problems exemplified by old (which would, to be sure, have to established separately), it would not help against the type of difficulties illustrated by vers: taking into account the entire set of lexical relations that an item contracts might avoid relational underspecification of meanings (i.e. the situation that a relational approach is unable to spot semantic distinctions), but it could not help against relational overspecifications (i.e. seeing more meanings than is feasible). On the contrary, it would probably strengthen the danger of overspecification. In Cruse (1986: 57), thin is used as an example of the vers-type of problem. If a longitudinal analysis confirms that this confusion was indeed a regular feature of the history of logic, part of the explanation might be that the interest in the phenomenon of ambiguity - at least judging from Rosier (1988) has traditionally always gone more to 'amphiboly', i.e. the rhetorical use and abuse of sentential ambiguity, than to the operational definition of lexical polysemy. A good candidate is the following sentence from Cruse (1986: 64): ? Dogs become pregnant at J2 months, but mature later than bitches. But could it not be argued that attributes such as those involved in (1) and (i) are general precisely as typical characteristics? This is (if I read her correctly) the position defended by Wierzbicka (1990), where she argues for the inclusion of 'mental characteristics' in definitions. The reasoning can, I think, be rendered in the following way: although not all fmits grow on trees (and although, therefore, growing on trees is not a general characteristic of fmit) , it is a general characteristic of fruit that people usually think of it as growing on trees. Now clearly, including prototypical characteristics into definitions is not what we would be arguing against from a cognitive point of view. Still, how exactly should we constme Wierzbicka's argument? On the one hand, 'usually growing on trees' as such is not a general characteristic of fruit: if 'usually growing on trees' is a general characteristic of fruit, any fmit should usually grow on trees - but of course, one cannot say that a strawberry usually grows on trees. In other words, if 'usually growing on trees' is constmed as part of our knowledge about fruits, it clearly is not a general attrib-

Vagueness's puzzles, poZvsemy 's vagaries

10.

11. 12.

13.

145

ute. On the other hand. suppose we construe 'usually growing on trees' more radically in the mental sense intended by Wierzbicka: it is then part of our knowledge of fruit that (other) people tend to think of fruit as growing on trees. But again, this attribute (which involves not features of x, but knowledge of features of x) can only be general if, for any thing that belongs to the category fruit, other people tend to think of it as growing on trees - and we probably would not claim that other people tend to think of raspberries as growing on trees. In short. attributes that take into account prototypicality phenomena by incorporating operators like usually do not restore generality. It might perhaps be remarked that Crose's argumentation is less compelling than he suggests. If the dichotomy involves (roughly defined) a 'gustatory' and a 'non-gustatory' sense of like, it could be argued that there is no sentence in the set that actually combines both readings (as would be required to establish the contextual neutralization of the polysemy). The only candidate would be (12), but a crossed interpretation of this sentence is not the first one that comes to mind; rather, it would seem that John non-gustatorily likes both items as status symbols or life style indicators. However, in Cruse (1986: 72), examples of sense-spectra are given for which the specific difficulties with like do not occur. Moreover, examples may be found in which metalinguistic overtones are entirely absent; see David Tuggy's paint example (Tuggy 1993). Furthermore, observe that Wierzbicka' s definition misses the point that apart from ability, willingness to marry also plays a role in judgments about bachelorship: an unmarried male person living together with a woman on a permanent basis is not (if at all) a good example of a bachelor, although he could marry in all relevant respects. This discussion does not imply that there could be no other grounds for establishing the contextual variability of the definitional approach. McCawley (1981: 7), for instance, notes that one person may more easily come up with a general definition than another one: until someone pointed out to him that the two interpretations of queen ('female ruler' and 'wife of male ruler') could be covered under the general definition 'enthroned female', he had been unable to define the word non-enumeratively. In general, the failure to come up with a single definition may be the result of insufficient ingenuity on the part of the definer, and ingenuity is obviously a characteristic that can change from one person to another. Yet another possibility for establishing the contextual variability of the definitional approach is based on the potential co-existence of two distinct definitional analyses of the same item. Suppose, for instance, that Aristotle's cases of megalopsychia are analyzed in such a way that the two groups for which a general definition can be given are constituted by Alcibiades, Achilles, Ajax, and Lysander on the one hand (megalopsychia being 'military ambition in the service of one's country'),

146

Polysemy

and Socrates on the other (megalopsychia being 'the unwillingness to waver from the quest for truth under the pressure of others'). It will be obvious that what is a distinct meaning in the context of this analysis does not have a separate semantic status in the context of Aristotle's definitional analysis. 14. The distinction is not new in lexical semantics; see, for instance, Schmidt's (1963) distinction between lexikalische and aktuelle Bedeutung. 15. There is yet another reason for not forgetting the contextual interpretations. If the stored meanings are by that very fact unobserved entities, and if, conversely, the things that we can actually observe (however indeterminatedly) are surface structure interpretations, getting an idea of what the surface structure meanings are is epistemologically prior to determining the background entities from which they are derived. If representations plus rules are postulated to explain the occurrence of particular surface phenomena. these phenomena have to be determined first before one can start postulating explanatory entities. 16. At this point, we may have a look at the third criterion for polysemy introduced by Cruse (1986); see note 5 above. In essence, the criterion relies on a distinction between interpretations that are totally conditioned by context and ones that are context-independent. For instance, the reading 'female sovereign. queen' of monarch in The Ruritarian monarch is expecting her second baby does not represent a case of polysemy, because the sex of the monarch can be derived from the contextual information that she is having a baby. The argumentation can be reconstructed as follows. 'On the level of surface interpretations, monarch may be interpreted as either 'male sovereign' or 'female sovereign'. These are distinct surface interpretations, because there is at least one context in which they are not compatible, i.e. in which the logical criterion for semantic distinctness applies. (In fact, they are incompatible in any context). These distinct surface interpretations do not represent distinct meanings. because the question which interpretation applies always depends on contextual information (such as the presence of predicates that can only refer to women)'. Now, consider the question what are the constraints on the kind of contextual information that is envisaged in this argumentation. There seems to be no reason to restrict it to purely linguistic contextual information, if only because any relevant piece of extralinguistic. encyclopedic or situational information could be turned into a piece of linguistic context by expressing it in words. But then, the same argumentation that is used to defend the claim that monarch is vague rather than polysemous with regard to the distinction between the readings 'male sovereign' and 'female sovereign' can also be applied to cases of genuine polysemy: 'On the level of surface interpretations, position may be interpreted as either 'occupied place' or 'mental attitude'. These are distinct surface interpretations. because there is at least one context in which they are not compatible, i.e. in which the logical crite-

Vagueness's puzzles, polysemy's vagaries

147

rion for semantic distinctness applies. These distinct surface interpretations do not represent distinct meanings, because the question which interpretation applies always depends on contextual information (such as the referents involved)'. Therefore, if the reconstruction of the argumentation and the generalization of 'context' are justified, there are reasons for doubting whether the criterion is sufficient yet to distinguish vagueness from polysemy. 17. Another failing escape hatch that should be mentioned could be dubbed 'the ploy from infinity'. In fact, could it not be argued that the distinction between vagueness and polysemy is necessary in principle, lest lexical semantics become impossible? The traditional line here is to state that the distinction between polysemy and vagueness is necessary to avoid endless polysemy. If there is no stop to referential distinctions turning up as distinct meanings, a word would have an infinity of meanings. If, for instance, neighbor is polysemous between the readings 'male neighbor' and 'female neighbor' merely because there is a referential difference between the members of both subsets of the extension of neighbor, what is there to stop us from claiming that 'neighbor who owns a Volvo' and 'neighbor who owns a Chevrolet', or 'neighbor who is 1.73 m tall' and 'neighbor who is 1.74 m tall' are distinct meanings? After all, the referential distinctions involved define subsets just like the distinction between 'male' and 'female'. Furthermore, the fact that such distinctions could be multiplied ad infinitum would yield an endless polysemy. So, it might be concluded, even if we do not know yet how to determine the distinction between referential differences that do and those that do not turn up as separate senses, we are at least sure that such a distinction has to exist. However, the argument from infinity assumes that any referential difference would eo ipso turn out to be the basis of a semantic distinction: only in that case could there be infinity. What we have discovered, on the other hand, is the instability of the distinction between polysemy and vagueness, not its complete absence. The distinction remains valid in the sense that referential differences do not automatically surface as distinct meanings: there is always a preliminary process of categorization involved that picks out particular referential attributes as the basis of categorial similarities or contrasts. The attributes that are being picked out may not be entirely stable (in the sense that a distinction that is disregarded in one context is promoted to semantic distinctiveness in another), but there is no automatic mechanism that turns referential differences into semantic oppositions. 18. This is not the first time a 'searchlight' or 'highlighting' metaphor has been used in lexical semantics. The searchlight image has been used by Weisgerber (1954) to specify the way in which various items illuminate portions of lexical fields. The concept of highlighting is used by Crose (1986: 53) to indicate the contextual modulation of an item's meaning, in the way in which the car needs servicing and the car need~ washing focus on different parts of

148

Polysemy

the car. It will be clear in what way our own terminology differs from Crose's and Weisgerber's. While the latter uses the term as an onomasiological concept, ours is a semasiological one. Crose's way of using the term, on the other hand, is referential rather than semantic: what is highlighted is less a part of the total set of semantic possibilities of an item than an aspect of the referent of that item in one of its senses. (This referential conception is not entirely separated from the semantic conception, however. In particular, Crose's approach seems to rely on a distinction between contextual modulation and contextual selection of senses that may not be tenable in the light of our present argumentation. The central tendency of the present paper leads to the suggestion that what are variations of one sense at one point, may turn out to be distinct senses at another. For instance, while Crose's examples refer to a situation in which both aspects of cars ('carriage' and 'engine') are simultaneously applicable but in which one of them is backgrounded, there are contexts in which both aspects are incompatible: an abandoned and demolished car with the engine removed would seem to be a car in the sense of the car needs washing but not in the sense of the car needs servicing. That is to say, if Crose's conception of highlighting presupposes that the backgrounded and the highlighted reading are compatible in the relevant context, our notion of highlighting does not carry that presupposition; moreover, interpretations that stand in a relationship of compatibility-plus-focusing in one context. may turn out to be incompatible in another.) 19. Admittedly, by focusing on the problematic, unstable cases we may have exaggerated the instability. It could still be the case that in actual practice, the problematic cases constitute merely a peripheral phenomenon. Even so, they constitute a constitute a problem in principle for semantics, to the extent that they show that the definition of lexical meaning is not as clear as might be hoped.

Chapter 6 Classical definability and the monosemic bias

Originally published in Rivista di Linguistica 1994.6: 189-207. Lexical semantics cannot at the same time retain a monosemic bias and the ideal of classical definability: there exist cases where sticking to the intuitively plausible idea that a particular lexical item is univocal leads to the conclusion that it cannot be classically defined. In the present chapter (which is based on section 3.2 of Geeraerts, Grondelaers, and Bakema 1994), this claim is supported by means of a corpus-based case-study involving denotational data based on non-elicited actual usage. Within this methodological framework, a procedure for determining classical definability will be demonstrated that operationalizes the requirement that definitions be both general and distinctive. While the previous chapter, then, looked at problems of definability from a methodological point of view. with an eye to the criteria that may be used for establishing either polysemy or univocality, the present paper takes a more practical approach, illustrating definitional problems with detailed examples. Crucially, the paper argues for a rigorous procedure in answering definability questions. If we assume that monosemy can be established by providing a classical definition (a definition in terms of necessary and sufficient features that is both general and distinctive with regard to the item to be defined), there are two requirements that discussions of definability should strictly adhere to. First, the generality of classical definitions implies that no disjunctive features are included in the analytical definitions, and second, the distinctiveness of classical definitions implies that they should make the right predictions about the lexical relations among the items involved. As the examples in the present paper show. complying with these demands requires a careful. step by step procedure that stands in sharp contrast with what semanticians usually do when they posit univocality. The present version of the text is considerably expanded with regard to the originally published article. The discussion of legging, as an example of classical definability. was missing from the version in the journal.

1.

Polysemy and definability

Definability plays an important role in current discussions about lexical categories. if only because it is intimately related to the problem of polysemy. Lexical-semantic research is to a large extent guided by a mono-

150

Polysemy

semic bias: the idea that postulating a single meaning for a lexical item is, on the whole, preferable over postulating multiple meanings. Authors who have recently defended this position with varying degrees of explicitness include Charles Ruhl (1989b), Anna Wierzbicka (1985, 1991), and Claude Vandeloise (1990). As Chris Sinha has pointed out (p.c.), the monosemic bias is related to Grice's 'Modified Occam's Razor': the principle that senses are not to be multiplied beyond necessity (Grice 1989: 47). The monosemic bias is not universal, however. In my own work on lexical semantics, for instance, I have had a tendency to stress the presence of polysemy (for instance, see Geeraerts 1989a, 1990a) and it has been argued (Landheer 1991) that I have so to speak exhibited an undue polysemic bias. Now, the monosemy/polysemy issue is complicated by various factors, two of which stand out. First, there may be various criteria for deciding about the presence of monosemy. In Geeraerts (1993a), an overview of some of the most common criteria is given (and it is shown that they are mutually inconsistent). Second, there is variation as to the level of semantic analysis with regard to which the monosemy question is defined. Most linguists, in fact, would seem to be willing to accept a two-stage model in which a distinction is maintained between, roughly, permanently stored 'deep stmcture meanings', and 'surface stmcture meanings' that can be derived from them on the basis of cognitive and pragmatic principles of semantic extension interacting with situationally and textually contextual factors. If the monosemy issue is defined with regard to the stored readings, different solutions will be arrived at in comparison with a situation in which it involves surface stmcture interpretations. In the former case, for instance, one could say that something is not a different reading when it can be derived by mle from a stored reading (cp. Kempson 1977). In the latter case, on the other hand, one of the possible criteria for semantic difference rests on the possibility of capturing all surface readings of an item under a single definition. Without trying to solve the problem of monosemy as a whole, the present paper will try to shed some light on the latter test of polysemy, viz. the definitional test ofpolysemy that says that two of more instances of a word do not constitute a different reading when they can be defined together as a single meaning. In particular, it will be demonstrated, first, how the criterion can be applied to usage-based data about the referential range of application of lexical items, and second, that applying the criterion involves more comparative computation than is usually assumed. Third, this demonstration will support the claim that lexical semantics cannot at the same time preserve the monosemic bias and the ideal of classical definability: cases will be

Classical definability

151

presented where sticking to the intuitively plausible idea that a particular lexical item is univocal leads to the conclusion that it cannot be classically defined. This claim will be illustrated on the basis of a corpus consisting of 9000 Dutch clothing terms, taken from fashion magazines and general magazines published in 1991. A full description of the project is presented in Geeraerts, Grondelaers, and Bakema (1994); the present paper corresponds largely with section 3.2. of the monograph. In the database, each garment is present in the form of a componential description. The descriptions are based on photographs (or, occasionally, drawings) that appear in the magazines; only words referring to such pictures have been included in the database. The componential system of description differs somewhat according to the major subtypes of clothing. All types of trousers, for instance, are described by means of the following set of features. First dimension: LENGTH 1 The garment does not reach further down than the groins 2 The garment reaches down to the thighs 3 The garment reaches down to the knees 4 The garment reaches down to the calves 5 The garment reaches down to the ankles Second dimension: WIDTH AND CUT 1 The garment is tight-fitting 2 The garment has a straight cut, neither tight-fitting nor wide 3 The garment has a loose, wide cut 4 The garment is tight-fitting round the hips but has widening legs 5 The garment is loose-fitting round the hips but has straight or tightfitting legs 6 The garment has a regular straight cut as far down as the knees, but has widening legs below the knees 7 The garment has a loose cut from the hip to the knee, but has tight-fitting legs below the knees Third dimension: 1 The ends of the 2 The ends of the 3 The ends of the 4 The ends of the 5 The ends of the

END OF LEGS

legs exhibit no special features legs fit tightly by means of an elastic band legs fit tightly by means of tied laces legs fit tightly by means of buttons legs have an elastic band that fits under the feet

Fourth dimension: 1\1ATERIAL 1 The trousers are made of cotton, linen or a similar smooth cloth

152

Po~ysemy

2 The trousers are made of wool or a similar coarse or fluffY material

3 The trousers are 4 The trousers are 5 The trousers are 6 The trousers are 7 The trousers are 8 The trousers are

made of denim made of corduroy made of silk or a similar smooth and shining material made of smooth, shiny leather or an imitation of it made of rough, mat leather or an imitation of it made of stretch

Fifth dimension: DETAILS I Seams and/or pockets are strengthened by metal buttons 2 The waist part of the garment has folds, pleats or creases 3 The garment has a very low crotch 4 The garment is made up of several visible strips of cloth 5 The legs have sharply pressed creases 6 The garment has an elastic band in the waist A specific garment may then be described, for instance, by the featural configuration H51431, where H indicates that the garment belongs to the group of trousers, and where 51431 indicates that it is a pair of trousers with value 5 on the first dimension, value 1 on the second, 4 on the third, and so on. It should be clear from this presentation that the approach followed here is a referential, denotational one: lexical meanings are studied not primarily through introspection, but on the basis of their ranges of application as they appear in actual usage. The referential, usage-based investigation followed here is not accepted unconditionally by all lexical researchers. In actual practice, lexical studies with a Cognitive Semantic orientation exist both in the form of introspective analyses, and in the form of corpus-based research. Schmid (1993: 272) even considers the corpus-based approach in work such as that of Rudzka-Ostyn (1988, 1989), Schulze (1988, 1991), and Dirven (1985, 1990) to be typical of the European branch of the Cognitive Linguistic movement, in contrast with the more introspectively conducted studies of American researchers of a Cognitive Linguistic persuasion. From a more theoretical point of view, explicit attention for the way words are actually used would seem to follow straightforwardly from Langacker's characterization of Cognitive Linguistics as a usage-based model that rejects the Chomskyan neglect of linguistic performance (1987: 46). However, Wierzbicka (1985) has coupled a prototypeoriented form of lexical research with an explicit defense of the introspective method. What advantages, then, are there to a referential, usagebased approach that avoids relying exclusively on introspection? Next to the advantages of a usage-based investigation that are enumerated by Ruhl

Classical definability

153

(1989: 13-16), let us note that the introspective method may succeed rather well in pinning down the prototypical core of the items under investigation, but is hardly able to capture the peripheral uses to which the core meanings appear to give rise in actual usage. This is a point to which we will return in section 5.

2. Criteria for classical definability The definitional criterion only works properly when restrictions on definitions exist: there has to be a way of assessing what is a good definition before definability can be invoked as a measure of polysemy. Classical definability implies that a definition can be found that characterizes all the members of the category to be defined, and only those. The definition has to be general, in the sense that it applies to all the members of the category, and it has to be distinctive, in the sense that it adequately distinguishes the category from all others. For instance, let us assume that we are trying to define the category 'bird' (as a biological species). We will then have to list the attributes that all birds have in common, if there are any; further, we will have to make out whether this list of attributes (or any subset of it) suffices to distinguish birds from mammals, reptiles, and fishes, to say the least. As illustrated in Geeraerts (l987a), the attributes that one would be inclined to mention as general characteristics of birds, often do not have the required commonality. On the other hand, the attributes that do seem to be general among birds do not suffice to distinguish birds from other species; even when the features in question are taken together, the duck-billed platypus is a counter-example to the alleged definition. It may be useful to point out here that there are various other ways of terminologically indicating the classical nature of definitions. One is to say that classical definitions define all and only the members of the category, while another is to say that they uniquely define the category. More importantly, however, it has to be noted that applying the definition meets with particular problems in the case of our material. Before turning to actual examples, let us consider each of the two requirements in more methodological detail, with specific reference to the kind of material incorporated in the database described above. The first part of the joint requirement of generality and distinctiveness would seem to be easy to check: the componential description of the referents of each lexical item allows us to check whether there are any attributes

154

PoZVsemy

that these members have in common. There is an important reason, however, for rejecting such a straightforward and mechanical procedure. The descriptive features that define the various configurations in the referential range of a lexical item cannot be taken at face value, but have to be interpreted (as an automatic consequence of which, the issue of classical definability cannot be settled mechanically). There are basically two forms of interpretation to be taken into account. For ease of reference, they will be called the quantitative and the qualitative one. The quantitative interpretation involves numerical dimensions, i.e. dimensions whose values constitute a graded continuum. The crucial point here is to see that it is not the individual value of a specific referent with regard to that dimension that is definitionally important, but rather the range of values with which the dimension occurs. If, for instance, a dimension like WIDTH receives the values [2], [3], and [4] in the semasiological range of application of an item, we should not say that the item has no common feature on the dimension WIDTH, but we should rather say that the width of the referents of the item in question ranges from value [2] to [4]. Although the presence of the values [2], [3], and [4] would superficially suggest that the referents of the item do not have common characteristics as far as their width is concerned, they do upon closer inspection: all of them fall within the range defined by the interval [2]-[4]. On the other hand, a qualitative reinterpretation of the superficially given values involves hidden variables. In particular, whereas all the dimensions in the database are visual ones, there may be covert dimensions of a fimctional nature. For instance, if the MATERlAL dimension of an item features the values [silk] and [cotton], there is again, superficially speaking, no common characteristic. If, however, both silk and cotton are used as light materials serving the purpose of keeping the person cool in warm weather, the common functional feature [light and cool] reduces the original variation on the MATERlAL dimension to epiphenomenal status. The distinctiveness criterion for classical definability should be handled with equal care. To begin with, notice that the distinctiveness requirement crucially involves negative evidence. If a definition is to hold for all and only the members of a particular category, the definition should not apply to any specific thing that does not belong to the category. The distinctiveness of the definition is contradicted, in other words, if we can find a referent that falls within the scope of the definition but that falls outside the scope of the category. This does not mean, to be sure, that the items falling within the scope of the definition could never occur as members of other categories than the one to be defined. For instance, let us define the attributes plusquint and

Classical definability

155

deciminus of natural numbers. A natural number is plusquint if it is larger than five; it is deciminus if it is smaller than ten. Both definitions are classical: they are as mathematically precise as you can get. At the same time, both categories naturally overlap: the natural numbers 6, 7, 8, and 9 fall within both categories. This means that the number 7 may sometimes be called a plusquint and sometimes a deciminus. Suppose further that we have actually encountered both ways of speaking in our corpus of mathematical language, and that we are trying to define the word deciminus. We have noted that all deciminuses share the property of being smaller than ten, and therefore propose to define deciminus accordingly; we also notice, however, that 7 is sometimes called a plusquint. When confronted with such a plusquint instance of 7, could we then say (repeating the sentence introduced above) that 7 falls within the scope of the definition of deciminus, but that it falls outside the scope of the category (because it is not then called a plusquint)? Of course not: falling outside the scope of the category means never occurring within it. The number 7 is not a counter-example to the proposed definition of deciminus because it is occasionally called a plusquint, but it would be a counter-example if it were never called a deciminus. In the same way, the duck-billed platypus is only a counter-example to the classical definability of bird because it is never categorized as a bird. At the same time, the fact that not all deciminuses can be called plusquint implies that the definition of the latter category should not be so broad as to include the entire range of application of deciminus; it must include those deciminuses that are sometimes called plusquint, but it must exclude those that are never so called. As a practical consequence of this observation, we will have to check any alleged classical definition of a lexical item against the words with which the item referentially overlaps. In particular, the definition should not be overly general, in the sense that the entire overlapping category (rather than just the intersecting part) is drawn into the category to be defined. The relationship between two items that share referents may, however, take other forms than the kind of overlapping (partial co-referentiality, semi-synonymy) that is illustrated by the plusquintldeciminus case. Systematically, there are three other relations to be envisaged. When the items are synonymous, no problem arises when the definition of the definiendum covers the entire range of application of the second item. Similarly, when the definiendum is a hyperonym of the second item, the definition may (in fact, must) cover all the referents of the second item. But when the definiendum is a hyponym of the other word, a definition that exceeds the referential boundaries of the hyponymous

156

Polysemy

item will have to be rejected. To summarize, the distinctiveness criterion does not apply to the synonyms and the hyponyms of the definiendum; in the case of overlapping and hyperonymous categories, it should only be applied to the overlapping and hyperonymous categories as a whole, not to those subsets of the latter that they share with the definiendum.

3. Legging: a case of classical definability Now that we have a better idea of how the classical definability of lexical items can be established, actual examples can be considered. In the following pages, the lexical item legging will be considered in detail as an illustration of the methodology outlined above. The other lexical items that will be analyzed further on in this section will be treated more succinctly; rather than concentrating on the analytical procedure itself, we will then concentrate on the results of the analysis. Table 1. The semasiological range of legging H3118.v H4118.v H41186v H4211.v

[3] [36] [3] [ 1]

H5118.v H51186v H5128.v H5154.v

[58] [2] [1] [1]

H5211.v H52115v H52116v

[3] [ 1] [1]

The semasiological information that can be extracted from the database takes the form of a list of referential descriptions of the form illustrated in Table 1. The table lists all the referential configurations with which the item legging occurs, together with their respective frequencies, which are mentioned between square brackets. The numbers refer to the five dimensions illustrated in the previous section. The feature v specifies that the item is worn by a woman. And a dot instead of a bumber indicates that the relevant feature is not relevant; in this case, it means that the dimension Detail that was introduced above is not applicable to the item in question. Finding out whether legging can be classically defined would now seem to follow a straightforward procedure: first, it will have to be established whether there are any characteristics that are common to all referents of legging, and second, it will have to be investigated whether the resulting definition is sufficient to distinguish legging from all other categories that are neither hyponyms nor synonyms of legging. On the basis of this proce-

Classical definability

157

dure, legging would definitively turn out not to be classically definable. A glance at the table suffices to appreciate that the only truly general characteristics of all the listed instances of legging are the fact that they all involve trouser-like gannents, as represented by the feature [H], and the fact that they are worn by women, as represented by the feature [v]. At the same time, the database contains various trousers worn by women that are never called legging; specifically, wider and shorter types of trousers for women fall outside the category. But obviously, we have not yet subjected legging to the quantitative and qualitative reinterpretation process that we described above. Let us now try to establish whether we can salvage the classical definability of legging by using a more refined approach. A quantitative reinterpretation is important for the dimensions WIDTH and LENGTH. The width of the referents of legging varies between the values [1] and [2], which is to say that leggings are either tight or narrow. The dimension LENGTH has a range between values [3] and [5]; the referents of legging reach down at least to the knee, but they may also cover the entire leg down to the ankle. The impact of a qualitative reinterpretation can be appreciated when we have a look at the dimension MATERlAL. The fact that the predominant value on this dimension, viz. [8], refers to stretchy fabrics suggests that there is a causal connection with the dimension WIDTH: leggings are mostly tight or narrow precisely because they are made of elastic material. Could it be the case, then, that the feature [elastic] allows for a reduction of the variation on the dimension MATERlAL? A renewed consultation of the original pictures on which the database records were based, reveals that this is indeed the case. On the one hand, the records that contain the value [1] on the dimension MATERlAL appear to be made of a finely woven tricot that is at least moderately elastic. On the other hand, the single record that features the value [4] for MATERIAL involves a stretchy, very tight-fitting corduroy. In other words, although not all leggings have the same degree of elasticity, they do share a certain amount of stretchiness. The resulting picture of the common characteristics of all instances of legging may be summarized as follows: a legging is a two-legged outer garment for women covering the lower part of the body from the waist down, ranging in length from the knee to the ankle, and made from elastic materials such that the width of the legs ranges from tight to narrow. The next step involves checking whether this set of common features is sufficiently distinctive to act as a definition of legging. In order to get an idea of the lexical items that have to be included in an analysis of the distinctiveness

158

Polysemy

of the definition of legging that was given above, Table 2 lists the onomasiological alternatives with which legging-configurations occur in the database. For each of the various configurations that are situated within the semasiological range of legging, Table 2 specifies the other lexical items referring to that configuration, together with their frequency. Table 2. The onomasiological alternatives for legging H3118.v

broekje J, kniebroekje J, legging 3, piratenbroek J, wielrennersbroek J

H4118.v

broek 12, broekje I, calec;on J, denimbroek J, jeans J, kuitbroek J, legging 36, leggings 9, piratenbroek 3, stretchbroek J, tricotbroek J broek 2, calec;on J, legging 3, leggings 2

H41186v H4211.v

broek J 7, calec;on J, kuitbroek J, legging J, pantalon J, streepbroek J

H5118.v

broek J2, broekje J, legging 58, leggings 23, skibroek 3, stretchbroek J, stretchleggings J, tricotbroek 5

H51186v

broek 6, calec;on 2, legging 2, leggings 7, pantalon J

H5128.v

broek J, joggingbroek J, legging J, skibroek 2

H5154.v

legging J

H5211.v

bloemenbroek J, broek 74, calec;on J, jeans JJ, legging 3, pantalon 2

H52115v

broek 24, legging J, pantalon Uricotbroek J

H52116v

legging J, broek 3

The status of the onomasiological alternatives in terms of their lexical relationship with regard to legging is specified in Table 3. In accordance with the methodology set out above, we need not worry about a possible lack of distinctiveness of the definition of legging with regard to hyponyms like kabellegging ('cabled !egging', legging decorated with a cable pattern) and stretchlegging ('!egging in stretchy fabric'), or synonymous expressions like leggings and cale90n. (In passing, it may be noted that the latter term is geographically restricted to the Belgian sources, in contrast with the magazines published in The Netherlands.) The hyperonym broek does not present a real danger either, because the definition of legging specifies an actual subset of the range of application of broek (which may, in particular, refer

Classical definability

159

to two-legged gannents that are much shorter and much wider than the referents of legging). Table 3. The relations between legging and its onomasiological alternatives

Synonyms Hyperonym Hyponyms

leggings, caler;on broek kabelleggings, stretchleggings

Semi-synonyms

kuitbroek, kniebroekje, broekje jeans, denimbroek, stretchbroek, tricotbroek streepbroek, bloemenbroek piratenbroek, wielrennersbroek, skibroek, joggingbroek, pantalon

As suggested in Table 3, various subsets may be distinguished within the set of semi-synonyms. A first set comprises kuitbroek 'calf-long pair of trousers', kniebroekje 'knee-long pair of trousers', and broekje 'short pair of trousers, shorts', which each delimit a specific area within the range of application of broek on the basis of the length of the gannent. The second set consists of jeans, denimbroek, stretchbroek, and tricotbroek, which each refer to trousers made of a particular fabric. In the third set we find items that refer to a particular type of motif or decoration: streepbroek indicates the presence of stripes, and bloemenbroek signals the presence of flowers. The fourth subset contains piratenbroek 'pirate's trousers', wielrennersbroek 'cyclist's trousers', skibroek 'ski pants', joggingbroek 'jogging pants', and pantalon 'pair of trousers'. There is an interesting distinction between the fourth subset and the previous three, in the sense that the items in the latter may be adequately defined as a hyponym of broek by referring to a single dimension (length, material, and motif respectively). These dimensions detennine the morphological structure of the words, in the sense that the first member of the compound refers to the specific value on the relevant dimension that is crucial for the item. For the items in the fourth subset, however, various dimensions have to be mentioned at the same time in order to specify their proper position. In the cases involved in the first three sets, establishing the referential overlap with the semasiological range of legging is a relatively straightforward matter. The items are actual semi-synonyms to the extent that they may refer to pieces of clothing that are not legging-like on any dimension that is irrelevant for the item in question. For instance, referents of kuitbroek that

160

Polysemy

are not tight-fitting enough to fall within the definition of legging are not so called either. Kuitbroek is characterized on the basis of the dimension LENGTH, and its specific value on this dimension happens to fall within the range of lengths that is definitional for legging. Because of its 'onedimensional' nature, however, it is not definitionally specified with regard to other dimensions that are subject to restrictions in the case of legging (such as, in the example, WIDTH), and it may therefore refer to pieces of clothing that are definitely too wide for leggings. We have to make methodological allowances, though, for the fact that the number of records we have for the various items does not always suffice to establish their overlapping status beyond all doubt. That is to say, we may not have enough examples of an item like bloemenbroek to establish whether it does indeed occur with referents that are not leggings. In these cases, we have nevertheless listed the item as a semi-synonym on the basis of the assumption that the morphological structure of the item is a good indication of its semantics - on the basis of the assumption, for instance, that trousers with a flower motif may be called bloemenbroek regardless of their other characteristics. For the items of the fourth subset identified above, such an abductive underpinning of the classification on the joint basis of intuition and morphological structure is less obvious. Although most of the referents of skibroek are relatively tight-fitting, the widest ski pants are definitely wider than the widest referent of legging; skibroek, in other words, has a wider range on the dimension WIDTH than legging. The same criterion also distinguishes joggingbroek from legging. For wielrennersbroek, on the other hand, the most important distinctive dimension is LENGTH: whereas legging ranges from dimensional value [3] to [5], the referents of wielrennersbroek are never longer than the knee. Similarly, piratenbroek refers to garments that may be as long as the knees or the calves, but never as long as the ankles.

4.

Vest etc.: absence of classical definability

The discussion of legging shows that a careful analysis of semasiological ranges and lexical relations is necessary to determine the classical definability of lexical items. The legging example produces positive results, in the sense that the item in question appears to be classically definable. The next step will be to discuss a number of cases where the definitional analysis yields negative results, in the sense that the items in question cannot be defined on a classical basis. The discussion will be based on a subset of the

Classical definability

161

field of clothing terminology. In particular, we will consider pieces of clothing that cover the upper part of the body, that can be entirely opened at the front, and that are never worn as the first layer of clothing above the underwear. The items with the highest frequencies in this subset are jack, colbert, blazer, jasje, and vestNL . The distinction between vestNL and vests is necessary because there is a marked difference between the ways in which the item vest is used in the Belgian and the Netherlandic sources; we will come back to this point below. Table 4 gives an overview of the ranges of application of the items. The table is based on individual analyses of the items along the principles demonstrated in connection with legging. That is to say, the dimensions used in the table do not necessarily reproduce the information structure of the database in a straightforward maJmer, but may be the result of a reinterpretation of the stored data or even a reconsideration of the original pictures. Dimensions that are not distinctive within the subset have been left out. For instance, all the types of clothing included in the subset are worn by men and women alike; accordingly, the dimension SEX has not been retained in the table. A plus sign means that a particular dimensional value occurs within the range of application of the item; a minus sign indicates that it never occurs. Thus, plus signs on all values of a particular dimension mean that both values may occur. For instance, the referents ofjasje may either occur with a type of fastening that can be fastened up to the neck, or with a type of fastening that stops on the chest somewhat lower than the neck; by contrast, jack is never used as a name for garments that cannot be fastened entirely. The only dimensions in the overview for which the relevant values might have to be restated in terms of ranges are LENGTH and FASTENING, since both involve measures of length; for instance, the referents of jasje have a fastening whose length ranges from up to the chest to up to the neck. The other dimensions consist of discontinuous values. A first thing to note is that the referential ranges included in Table 4 suggest the existence of certain hyponymy relations. It appears, for instance, that all dimensional values that occur in the range of jack also occur in the range of jasje; at the same time, the latter item exhibits a number of dimensional values that are absent in the case of jack. In this particular case, the suggestion that jack is a hyponym of jasje (because the referential range of the latter word includes that of the former) is supported by the intuition that jasje is a cover-term for the entire set of items included in 4. However, the overview in the table is not really a good way of settling the hyponymy relations among the items, because the referential ranges are being considered in

162

Polysemy

tenus of separate dimensions rather than dimensions in combination. Consider a fictitious case in which an item A is represented by the referential types [ac] and [bd], and an item B by the types [ad] and [bc]. In both cases, the first dimension ranges over the values [a] and [b], and the second dimension over the values [cl and [d]. Judging on the basis of an overview of dimensional ranges, then, A and B would be synonymous, since they have the same dimensional ranges. Judging on the basis of the dimensional values as they occur in combination, however, it becomes clear that there is neither a relationship of synonymy nor hyponymy between both items. It is therefore necessary to establish hyponymy relations on another basis than Table 4 as such, Table .;, The semasiological ranges of jack, blazer, colbert, vestNv jasje, jack

colbert

blazer

vestNL

jasje

+

+ + +

+ + +

+ + +

+ + +

+ + +

+ + +

+

+ +

+ +

+ + +

+

+

+ +

+ +

LENGTH

shorter than the waist as long as the waist lower than the waist

+ +

CUT

blousing wide and straight narrow and straight waisted

+ +

+ +

MATERIAL

woven fabrics knitted leather

+ +

FASTENING

up to the neck lower than neck

+

A corpus-based approach for the recognition of hyponymous relations may be established as follows. If A is a hyponym of B, B may occur as an alternative name for all referents of A. Of course, B need not be as frequent as A for the referential set in question, because A may be more entrenched than B. Also, it may be expected for statistical reasons that the less common referential types of A may not occur in the corpus with B as an alternative

Classical definability

163

denomination; in actual practice, it may be sufficient to establish that B occurs as an alternative for the most common referents of A. Table 5. VestNL andjasje as onomasiological alternatives for blazer. blazer: configurations

B1l22v B1222v B2111m B2111v B2112v B2121m B2121v B2122v B2131v B2132v B2212v B2221m B2222v B2232v C2212v C1312v C2311v C2312v C2322v C2332v C2412v C3212v C3312v C3332v C3412v C3432v

frequency of configuration

vest as alternative

10 10

2 1 17 5

14 48 3 8 12 3 22 5

1 1

+

I 7

+ + + +

1 1 1 1 14 1 4 1

+ +

+

jasje as alternative

+ + + + + + + + + + + + + + + + + + + + + + + + + +

In Tables 5 and 6, such overviews of onomasiological alternatives are given for blazer and colbert. Given the statistical margin that was just mentioned, it can be deduced from the table that colbert is a hyponym of blazer and jasje, and that blazer is a hyponym of ja5je. At the same time, of course, it should be established that there are cases of B that are not named by means of A (lest a situation of synonymy rather than hyponymy obtains). This type of information, however, can be safely derived from overviews like

164

Polysemy

the one in Table 4: for instance, the plus sign on the 'knitted' value of the dimension MATERIAL for blazer as opposed to the minus sign for colbert indicates that knitted referents of blazer never occur with the name colbert, for the simple reason that the referential range of colbert does not include knitted garments of any kind. Table 6. Blazer, vestl\JL andjasje as alternatives for colbert. colbert: configurations B1121m B1122v B2111m B2121m B2121 v B2122v B2131m B2211m B2212m B2221m B2222m B2222v C3212m C3311 v C3312v

frequency of configuration 1 7 2 52 4

17 5 1 1 13 1

blazer as alternative

vest as alternative

jasje as alternative

+ + + + +

+ + + + + +

+

+ + +

4

+

1 3 1

+

+

+ +

Repeating the procedure illustrated in 5 and 6 for the items jack and vest leads to the hyponymy relations that are charted in Figure 1. (The label '+lh' indicates that there is a relationship of hyponymy between the items in question, given that a lexical test of hyponymy as illustrated in 5 and 6 is used. The label '-lh' signals the absence of hyponymy according to the lexical criterion.) The definitional question regarding the five items can now be made more precise: can the items be classically defined on the basis of the overview in 4 without obscuring the lexical relations summarized in 7? This involves reviewing all possible classical definitions of the items and checking whether they respect the relations in 7. Note, however, that only two of the four dimensions included in Table 4 can be used to establish classical definability. Whereas the values on the LENGTH and the FASTENlNG dimension can be expressed in terms of ranges, any reference to the other two dimensions automatically involves disjunctive

Classical definability

165

values; therefore, including these dimensions in the definition inevitably turns the definition into a non-classical one. Thus, it will be sufficient to investigate only the potential definitions that refer to the dimensions LENGTH and FASTENING. A further restriction follows from the specific position of jasje. Both from the point of view of the featural ranges given in 4 and from the point of view of the lexical relations specified in 7, jasje appears to act as a coverterm with regard to the other items: it is a hyperonym of the other four, and the row of plus-signs accompanying jasje in 4 indicates that it indeed includes all definitional possibilities that are relevant for the other items. In this sense, the search for classical, distinctive definitions may be confined to the other four items. In all of the potential definitions mentioned below, then, jasje 'jacket' will be used as a cover-term; roughly, it may be defined as referring to garments that cover the upper part of the body, that can be entirely opened at the front, and that are never worn as the first layer of clothing above the underwear. hyponym jack

co/bert

blazer

vestNL

hyperonym colbert

-Ih

blazer

-Ih

vestNL

-Ih

jasje

+Ih

Figure 1. Hyponymy relations between jack, blazer, colbert, vestNL,jasje.

The definitions that are to be taken into account may be systematically grouped together in three sets: definitions that refer only to the dimension LENGTH, definitions that refer only to the dimension FASTENING, and definitions that involve both dimensions at the same time. For ease of reference, we will call jack] the definition that involves LENGTH, jack2 the definition that involves FASTENING, etc. All in all, the following twelve definitions have to be considered.

166

Polysemy

Jack] Jasje whose length ranges from the region of the hip to the region of the upper part of the legs Jack2 Jasje that can always be fastened up to the neck Jack] Jasje whose length ranges from the region of the hip to the region of the upper part of the legs, and that can always be fastened up to the neck Colbert] Jasje that reaches down to the region of the upper part of the legs Colbert2 Jasje that can only be fastened as far as the chest (but not up to the neck) Colbert] Jasje that reaches down to the region of the upper part of the legs and that can only be fastened as far as the chest (but not up to the neck) Blazer] Ja.sje whose length ranges from the waist to the region of the upper part of the legs Blazer2 Jasje that can only be fastened as far as the chest (but not up to the neck) Blazer] Jasje whose length ranges from the waist to the region of the upper part of the legs and that can only be fastened as far as the chest (but not up to the neck) Vest] Jasje whose length ranges from the waist to the region of the upper part of the legs Vest2 Jasje with a fastening whose length ranges from the up to the chest to up to the neck Vest] Jasje whose length ranges from the waist to the region of the upper part of the legs and that has a fastening whose length ranges from up to the chest to up to the neck

These twelve definitions may occur in 81 (=3 4) combinations. That is to say, the general definability question boils down to 81 questions of the type: if jack is defined asjackj, calbert as calbert], blazer as blazer2, and vest as vest], do the lexical relations that follow from these definitions then conform to the actual relations that are summarized in figure I? Or, in other words, is there at least one combination of the twelve classical definitions that makes

Classical definability

167

the right predictions about the attested lexical relations? Rather than considering all 81 possibilities separately, it can be shown in the following way that the question has to be answered in the negative. First, consider all combinations of two elements from among the set of definitions that refer only to LENGTH. The co-occurrence of jack] and colbert] is to be excluded, because this would counterfactually imply that colbert is a hyponym of jack (as the range of LENGTH for jack as defined includes the range as defined for colbert). The co-occurrence of jack] and blazer] is to be excluded because it would imply that jack is a hyponym of blazer. The co-occurrence of jack] and vest] is to be excluded because it would imply that jack is a hyponym of vest. The co-occurrence of colbert] and vest] is to be excluded because it would imply that colbert is a hyponym of vest. And the co-occurrence of blazer] and vest] is to be excluded because it would imply that blazer and vest are synonymous. Second, consider all combinations of two elements from among the set of definitions that refer only to FASTENING. The co-occurrence ofjack2 and vest2 has to be excluded because it implies that jack is a hyponym of vest. The co-occurrence of colbert2 and blazer2 has to be excluded because it implies that colbert is a synonym of blazer. The co-occurrence of colbert2 and vest2 has to be excluded because it implies that colbert is a hyponym of vest. And the co-occurrence of blazer2 and vest2 has to be excluded because it implies that blazer is a hyponym of vest. Third, consider all combinations of two elements from among the set of definitions that refer to both LENGTH and FASTENING. The co-occurrence of jack] and vest3 has to be excluded because it implies that jack is a hyponym of vest. The co-occurrence of colbert] and vest] has to be excluded because it implies that colbert is a hyponym of vest. And the co-occurrence of blazer] and vest] has to be excluded because it implies that blazer is a hyponym of vest. The alternatives that remain at this point are summarized in Table 7. It is now immediately obvious that there can be no combination of four classical definitions that respects the existing restrictions, if only because all possible definitions of vestNL have already been ruled out. We may conclude, then, that there is no set of classical definitions for jack, colbert, blazer, and vest that sufficiently distinguishes the items among each other and that respects the lexical relations that appear to exist among them. Whether this is the dominant situation in the lexicon is difficult to say on the basis of our material; after all, we have only been able to examine a few lexical categories. One general conclusion, at least, is that indeed not all lexical categories can

168

Polysemy

be classically defined. In addition, it is worthwhile to point out that the definability issue seems to be strongly influenced by the specific subfield of the field of clothing terminology that is being considered. The 'skirts' -subfields, for instance, contains classically definable categories, whereas the subfield consisting of shirts, blouses, t-shirts, and their likes is as unclassical as the subfield analyzed in the previous pages. Table 7. Combinations of classical definitions for jack, colbert, blazer, vest. blazer

vest

2

3 3

5.

Methodological discussion

What we have tried to illustrate in the foregoing pages (apart from the fundamental fact that non-classical definability is a real phenomenon) is the importance of a rigorous procedure in answering definability questions. Apart from the importance of distinguishing between polysemy and univocality, there are two requirements that discussions of definability should strictly adhere to. First, the generality of classical definitions implies that no disjunctive features are included in the analytical definitions, and second, the distinctiveness of classical definitions implies that they should make the right predictions about the lexical relations among the items involved. As both the legging example and the jasje-subfield show, complying with these demands requires a careful, step by step procedure that stands in sharp contrast with two other approaches to the relationship between polysemy and definitionality. In fact, while the approach demonstrated here applies the classical criterion of dejinability to denotational data based on non-elicited actual usage, the two most outspoken recent defenders of monosemy appear to drop either the one or the other of these two features from their methodology On the one hand, Ruhl (l989b) accepts the importance of usage-based data, but gives up on the necessity of actually defining the presumed single meaning of an item. If all the instances of use ofa lexical item are claimed to

Classical definability

169

exhibit the same meaning, the least one should do to make this statement falsifiable is to define the semantic value in question. Ruhl, however, explicitly denies the possibility of doing this. At the end of his analysis of the English verb bear, for instance, he writes: So what does bear mean? It should be clear by now that this question cannot be answered in words: there is no single word or phrase that can comprehensively capture exactly what bear contributes. I hope, by trying to show the unity of bear's contexts, to have revealed a unified meaning; but such a conclusion is inferential (1989: 63). Clearly, such an approach gives up on precision in favor of impressionism. For general methodological reasons of comparability and falsifiability, such a strategy has to be rejected in favor of the attempt to achieve descriptively adequate definitions. If it is not even possible to definitionally identify the allegedly unitary meaning of a lexeme, how can its unitary status be tested at all? This is not, to be sure, a mere rhetorical question: see Geeraerts (1993a) for an overview of other approaches to polysemy apart from the definitional one. Ruhl, however, dismisses the definitional criterion without being explicit about an alternative that is methodologically sound in the sense of being precise enough to allow for falsifiablity. On the other hand, Wierzbicka' s approach retains the importance of precise definition, but rejects the denotational, usage-based approach. Such an explicitly introspective strategy runs the risk of being imprecise with regard to the actual range of application of an item. As an illustration, consider Wierzbicka's definition of dress (1985: 382): A KIND OF THING MADE BY PEOPLE FOR WOMEN AND GIRLS TO WEAR. IMAGINING THINGS OF THIS KIND PEOPLE COULD SAY THESE THINGS ABOUT THEM:

they are made to be worn on the body, below the head, to cover most of the body so that all the parts of a woman's body which people think should not normally be seen are covered with that one thing and to protect most of the body with undesirable contact with the environment and to cause the woman wearing it to look good they are made in such a way that when they are on the body the lower half surrounds the lower half of the woman's body from all sides so that the legs are not separated from one another and so that the genital area of the woman's body seems to be hidden and so that women wearing things of this kind look different from men

170

Polysemy

things of this kind are thought of as something suitable for women to wear in most kinds of places and in most kinds of circumstances. When we have a look at the actual garments that occur in our database as instances ofjurk (the Dutch equivalent of dress), we find cases in our material that do not conform to the description. If, for instance, 'covering most of the body' is interpreted as 'covering more than 50% of the body', then a number of very short summer dresses with open backs and low necklines do not display the feature in question. And if 'the parts of a woman's body which people think should not normally be seen' include the upper part of the thighs, then dresses with long side slits contradict the image. Furthermore, some dresses have such wide armholes and such a plunging decolletage that they could not normally be worn without exposure of the breasts (unless they are worn with an additional t-shirt or blouse underneath). The comparison shows, in other words, that the description proposed by Wierzbicka may well be adequate for the majority of cases in the range of dress, but does not really cover all possible instances. Admittedly, such a comparison is risky for at least two reasons. First, we start from the assumption that English dress and Dutch jurk are equivalent as far as their referential range of application is concerned. As long as we do not have a similar corpus-based analysis of dress as the one we have made for jurk, the comparison will have to remain a conditional one. Second (and more importantly), it is not even certain that Wierzbicka actually intends the definition to apply to all the cases in the extension of dress. By introducing the phrase 'imagining things of this kind people could say these things about them', the perspective is shifted from the objective features of the things that are being called dresses to the subjective image that people say they have about dresses when they are asked for it. In a sense, Wierzbicka defines dress by referring to what people think dresses are. And if what people think dresses are only involves the central cases of the category'dress', then, of course, it makes no sense to complain that the description of this mental image does not apply to non-prototypical dresses: it never intended to do so anyway. On this reading of Wierzbicka's view, its reference to subjective images could be construed as implying a conscious restriction of the description to the prototypical core of the category. And because introspection probably does work efficiently for retrieving such prototypical images, the introspective method may be salvaged. It is not quite clear, however, whether this interpretation ofWierzbicka's position is a valid one. On the one hand, she argues that 'a valid definition must be empirically adequate, that is, it must be phrased in such a way that

Classical definability

171

it covers the entire range of use of a given word, expression, or construction' (1989: 738). On the other hand, if a definition such as that of dress is to be applicable to all things that may be called by that name, people should be able to assert all the characteristics mentioned in the definition any time they see a dress. But surely, when imagining a less prototypical kind of dress than the kind whose features are included in the definition, people will not imagine it as a prototypical case. What people could say about dresses changes when peripheral members of the category are at stake: default dresses, for instance, may well cover most of the body, but that feature may be suppressed when a fancy type of sununer dress is involved. Even if, however, we accept that Wierzbickian definitions of the kind illustrated above are explicitly restricted to the prototypical core of the categories, a counterargument may be advanced in favor of a method based on observing actual usage. Note that it remains a matter to be settled empirically whether the lexical knowledge that people have in their in minds is indeed restricted to a mental image of the core of the category in question. It is not a priori given that the idea of a category that people may introspectively retrieve from memory is an adequate reflection of the extent of that person's actual knowledge of the category. On the contrary, if it is part of his knowledge to produce or accept an application of dress to nonprototypical cases, then he 'knows' more about the category than would be included in his introspectively retrieved idea of the category. That knowledge, to be sure, is not necessarily conscious knowledge; it is less 'knowledge that (lexical item x may refer to entities with such and such characteristics)' but rather 'knowledge how (lexical item x may be successfully used)'. In order, then, to get a better grasp on the lexical 'knowledge how', usagebased investigations of the type illustrated in this article are vital, precisely if it is suspected that conscious knowledge may only partially cover the full extent of a person's 'knowledge how' . In short, the approach demonstrated above is inspired by the desire to steer clear both of the danger of theoretical imprecision that is implicit in Ruhl's strategy, and of the danger of empirical incompleteness that is implicit in Wierzbicka's strategy. But if the approach followed here is indeed to be preferred for general methodological reasons, what are the findings that it leads to?

172

Po~ysemy

6. Conclusions If the methodological framework defended in section 5 is accepted, the investigation carried out in section 4 shows that not all lexical items can be classically defined in terms of necessary, jointly sufficient features. If classical definability is considered a criterion for monosemy, this conclusion further implies that not all lexical items are monosemous. At the same time, however, the investigation suggests that classical definability is not necessarily a psychologically acceptable criterion for monosemy. The items whose definitionally polysemous nature was determined in section 4 are not ones that are intuitively recognized as polysemous. In this sense, the investigation supports and further illustrates the point made in Geeraerts (l987a) and (l993a): the different criteria for polysemy that are quite plausible taken by themselves may be mutually contradictory; specifically, vest etc. are definitionally polysemous, but not intuitively. This implies, in other words, that lexical semanticians cannot at the same time maintain the monosemic bias and the ideal of classical definability. If, on the basis of the monosemic bias, it is accepted that vest's intuitive univocality is a theoretically viable reflection of an actual monosemy, the ideal that an individual meaning can always be classically defined has to be relaxed (because there is no classical definition for vest in the usage under investigation). Conversely, if one sticks to the idea that classical definability is an integral part of what it is to be a distinct meaning, the monosemic bias will have to be relaxed to the point where even an intuitively non-suspect case of univocality like vest is accepted to be polysemous.

Section 3 Constructions and idioms

Chapter 7 The semantic structure of the indirect object in Dutch

Originally published in Willy Van Langendonck and William Van Belle (eds.), 1998, The Dative 11. Theoretical Approaches 185-210. Amsterdam/Philadelphia: John Benjamins. To a considerable extent, Cognitive Linguistics is characterized by a strategy that extrapolates models developed in the study of the lexicon to other areas of linguistics. (Structuralism, by contrast. is characterized by a strategy that extrapolates models of phonological analysis, like the use of distinctive features and distributional analysis, to the rest of linguistic structure.) Specifically in the domain of semantics, productive models of meaning description like prototype theory matured in the field of lexical semantics and were then applied to grammatical phenomena. This may be clearly seen in the area of Construction Grammar as represented by Goldberg (1995), Croft (2001), Fried and Ostman (2004), Ostman and Fried (2005). Constructions are so to speak grammatical patterns treated as lexical items: more or less schematic entities that should be treated as symbolic units in their own right, with properties that cannot just be compositionally derived. But if that is indeed the case, competing models for semantic description such as may be found in lexical semantics are likely to play a role in Construction Grammar as well. In the present paper, I argue that the multidimensional form of prototype-based analysis that is illustrated in Chapter 3 is relevant for the description of constructions just as well as for the description of lexical items. Such a multidimensional model, in which co-variation of semantic shifts along different dimensions constitutes the grid of a prototype-based network, contrasts with a simpler type of radial network analysis (as represented by Goldberg's analysis of the dative construction) in which the crucial elements of the network are points (meanings, readings, senses) rather than dimensions. Methodologically speaking, the paper is based on introspectively derived examples rather than corpus-based materials. and that is clearly not the way in which I would proceed at the moment. Current work by Colleman (2005) shows however that the semantic structure for the Dutch dative construction suggested here may be fruitfully applied to corpus data.

176

Constructions and idioms

1. Aims and background ofthe exercise The purpose of the present article may be defined from two points of view: a more narrow descriptive one, and a broader theoretical one. Narrowly defmed, the paper intends to show that the semantic structure of the indirect object construction in Dutch can be fruitfully described, a) by starting from a spatial prototype (the indirect object as the recipient of a spatial transfer) b) by deriving further uses from this basis by taking into account four specific conceptual mechanisms of semantic change, viz. generalization, metaphor, metonymy, and subjectification c) by systematically exploring the various relations of these mechanisms, i.e. by describing the field of application of the indirect object as a multidimensional space. Broadly defined, the paper will try to bring home three general theoretical points about the semantic structure of syntactic constructions. The first point implies that such constructions can be semantically described by means of basically the same descriptive mechanisms that apply to lexical categories. The second point involves the view that the semantic structure of categories is best described as a multidimensional space. The third point consists of the recognition that the semantic description of syntactic constructions has be intimately linked to their syntactic structure. All three points require some additional discussion. The first point is a straightforward consequence of one of the major methodological strategies employed by Cognitive Linguistics, viz. to model grammatical research on the basis of lexical research: the study of lexical categories is taken as a methodological point of departure for the study of categorization in the grammar at large, and grammatical construction types are considered categories just like lexical items. Given that linguistic categorization is the major focus of attention for Cognitive Linguistics, studying the lexicon first is, in fact, a plausible step to take: the categorizing function of the lexicon has received more attention in the linguistic tradition (and is also, perhaps, easier to investigate) than that of grammatical constructs. The intention to study the grammar of the language along the same lines as its lexicon obviously leads to a specific kind of grammatical theory, viz. one in which the grammar, like the lexicon, is conceived of as an inventory of symbolic elements (i.e. meaningful units). At this point in time, this idea has received its clearest and most elaborate form of expression in Langacker's

The indirect object in Dutch

177

Cognitive Grammar (1987, 1990a, 1991), and in Fillmore's Construction Grammar (Fillmore, Kay and O'Connor 1988, Goldberg 1994). A major difficulty for such a research strategy involves the formal identification of the grammatical constructions. In the domain of lexical research, it is usually quite clear what separate entities have to be distinguished: in spite of (stubborn but relatively marginal) problems surrounding the exact demarcation of homonymy, the fact that lexical items have a constant phonological form by and large suffices to identify what the basic units of the analysis should be. In the domain of grammatical research, however, the formal side of the categories/constructions cannot receive such a straightforward definition. For one thing, the grammatical categories do not necessarily have a constant phonological form: word classes, for instance, may well exhibit a type of formal constancy in terms of their distributional and morphological properties, but these characteristics are already abstract properties that do not correspond in a one-to-one fashion with the presence of an identifiable and constant phonological form. For an other, we tend to think of the formal side of grammatical constructions as being subject to variations and transformations (in a non-technical sense), but it is not immediately obvious which changes and alternations are allowed, and which ones rather undermine the formal identity of the construction. The purpose of the present article is definitely not to resolve this question, but it should be pointed out from the start that the problem has immediate consequences for the task at hand. Traditionally, various construction types have been subsumed under the label 'indirect object': ditransitive constructions featuring both a direct and an indirect object, some intransitive constructions, and some prepositional constructions. Against the background of the methodological question that we just raised, it is not clear whether it is entirely justified to lump together these various constructions: does the indirect object exist at all in its traditional definition, incorporating , some cases of , and constructions of the type or ? The present article will not try to answer the question, but rather circumvent the problem by focusing on the ditransitive constructions (which will belong to the hard core of the construction in any event). There will be some remarks about intransitive indirect objects (that is, indirect objects in a V NP frame rather than a V NP NP frame), but none about the prepositional indirect object. As for the demarcation of the single NP indirect object constructions in contrast with single NP direct object constructions, the paper relies on the groundwork laid by Van Belle and Van Langendonck (1992).

178

Constructions and idioms

The second theoretical point mentioned above involves the question what kind of semantic structure best describes the semasiological range of grammatical constructions (or, as the case may be, lexical items). Although the representational formats used in Cognitive Semantics basically have the same representational power (see Geeraerts 1995a), it is nevertheless the case that some types of representation inadvertently favor a particular type of analysis. Radial set representations of the type promoted by Lakoff (1987), for instance, would seem to suggest a two-dimensional analysis of semantic structures: starting from a prototypical centre, new readings are derived directly or indirectly from the prototype in what ultimately constitutes a two-dimensional plane of interconnected senses. As against this impression of two-dimensionality, it has been argued that a more adequate picture of the semantic structure of lexical items has to take into account the multidimensional nature of semantic extensions: the structure of polysemous categories is characterized by covariation on multiple dimensions. A case in point is the reanalysis suggested in Geeraerts (1992a) of the radial set representation of the Dutch preposition over proposed by Cuyckens (1991); a related argument for a multidimensional approach is to be found in Deane (1993). The point to make, to be precise, is not one about representational formats: it is not impossible, for instance, to adapt a radial set representation in such a way that the multidimensional character of the semantic structure is revealed (see, for instance, the way in which Brugman 1989 enriches a basic radial set representation with a type of componential analysis that reveals the multiple links between the readings in the radial set). Rather, the point is an empirical one: the description of semasiological structures (of lexical items and grammatical constructions alike) requires specific attention for the multidimensional nature of the semantic links among the readings of the item or the construction. In the case that interests us here, a multidimensional analysis of the indirect object construction will be contrasted with two related analyses that do not pay as much attention to this multidimensional character, viz. Goldberg's radial set analysis of the English ditransitive construction (Goldberg 1992), and Rudzka-Ostyn' s hierarchical network analysis of the Polish dative (Rudzka-Ostyn 1996). The third theoretical point mentioned above might seem to be uncontroversial: it is a rather obvious idea that the semantic description of grammatical constructions had better be systematically linked to their formal structure. More concretely, a construction-based definition will have to take a relational form: if the indirect object constituent is explicitly defined as part

The indirect object in Dutch

179

of the constructional frame , then it should receive a definition that includes a reference (however indirectly) to the other constituents in the construction. Given the overall plausibility of this idea, the present paper will highlight (by way of example rather than through elaborate theoretical discussion) the connection between this point and the previous one: it will be shown that the relational character of the definition of the indirect object constituent plays a crucial role in the multidimensional nature of its semantic structure.

2. The overall structure of the analysis The basic goal of the analysis is to show that the indirect object construction is semantically cohesive: the various instantiations of the category are linked to each other by means of semantically plausible links, just like the senses of ordinary lexical items are linked to each other on through semantic associations. The links that will be envisaged here largely belong to the traditional repertoire of semasiology as established by the prestructuralist tradition of diachronic semantics: we will make use of processes of generalization, of metonymy, and of metaphor. In addition, we will invoke the notion of subjectification, which is a more recent addition to the inventory of mechanisms of semantic change. The starting-point of the analysis consists of those cases in which the indirect object refers to the recipient of a benefactive material transfer, as in hi} gal haar een roos 'he gave her a rose'. The prototypical indirect object, in other words, is defined in terms of a frame featuring a process of transfer, something being transferred, and a specific functional character of the transfer. The indirect object itself, of course, is the person to whom the transfer is directed and who benefits from the transfer. The choice of this particular construction as the prototype of the category is not an arbitrary one: it links up with traditional localist conceptions of case relations, in which spatial aspects of meaning are considered to be basic. Although this lies beyond the scope of the present paper, it should be noted that independent evidence for the identification of the prototype might in fact be looked for: a frequency analysis of corpus data, or data of a psychological type involving the ease with which informants produce particular examples of the indirect object construction, might very well confirm the central position of the 'spatial' kind of use selected here. In the context of the present paper, however, the choice for this particular prototypical center is a hypothesis that can be con-

180

Constructions and idioms

finned only indirectly. It will turn out to be a suitable hypothesis, in fact, if it indeed provides a good starting-point for deriving a plausibly coherent analysis of the indirect object construction. It is important for the overall structure of the analysis in the following pages, that the prototype as defined here contains three major components: if the prototypical indirect object constituent is the recipient of a benefactive material transfer, semantic changes may take their starting-point in the 'recipient' part of the definition, in the 'transfer' part of the definition, and in the 'material' or 'benefactive' nature of the transfer. The observation that the prototype may be analyzed into a number of conceptual components has interesting repercussions in the light of the third theoretical point mentioned in the previous section, but discussion of that point will be saved for the final section of the paper. At this point, it suffices to see that mechanisms of extension such as metaphor and generalization may attach to different components of the prototype. And at the same time, of course, each component may be subject to different types of semantic shifts.

3.

Extension 1: Generalization

Starting from the indirect object as recipient of a material transfer as prototype of the construction, the first two dimensions leading to extensions from the prototype involve processes of metaphorization and generalization. Table 1 charts a number of these extensions, in such a way that their mutual combinations become apparent. Generalization features along the vertical dimension in the table, and metaphorization is charted along the horizontal dimension in the table. (1)

Ik geefhem een boek

(2)

Haar werd een hart ingeplant

(3)

De wind blies hem de regen in het gezicht

(4)

Ik geefhet huis een nieuw laagje verf

(5)

Ik schenk u mijn koninkrijk

(6)

Ze hebben je verkeerde ideeen ingeprent

'I give him a book' 'Her was a heart implanted, she received a new heart' 'The wind blew the rain in his face' 'I give the house a new coat of paint' 'I donate my kingdom to you' 'They have impressed- wrong ideas upon you'

The indirect object in Dutch (7)

181

20 'n geluk overkomt je nooit meer

'Such a happiness will never befall you again' (8)

De jury heeft 'De ontdekking van de hemel' de eerste prijs toegekend

(9)

Ik zeg u dat hij komt

'The jury has awarded the first prize to De ontdekking van de hemel' 'I tell you that he comes'

Table 1. Generalization and metaphorization of the Beneficiary role RECIPIENT OF

RECIPIENT OF

RECIPIENT OF

MATERIAL

ABSTRACT

COMMUNICATIVE

TRANSFER

TRANSFER

TRANSFER

EXPERIENCE

(1) (2) (3)

(5) (6) (7)

(9)

AFFECTEDNESS

(4)

(8)

transfer into the sphere of CONTROL USE

Generalization as a traditional mechanism of semantic extension in lexical semantics involves all fonns of semantic shift in which the original reading refers to a proper subset of the extension of the new reading. Classical examples are the development from French arriver 'to arrive at the river"s edge, to land' to the less specific meaning 'to arrive in general', or the development of English moon 'earth"s satellite' to the more encompassing reading 'any planetary satellite. Alternative tenns featuring in the lexical semantic literature for the concept of generalization include 'extension', 'weakening', 'broadening', and 'schematization' (in the Langackerian sense). The process of generalization identified in Table 1 basically involves a weakening of the Beneficiary-role of the recipient of the prototypical transfer. In Table 2, tentative definitions are given of the four positions that are distinguished along the generalization dimension in Table 1: for each of the four labels distinguished along the vertical dimension of Table 1, a componential definition is presented that reveals their mutual relations. In accordance with the definition of generalization, each secondary generalized reading includes the original one as a special case. All cases of 'transfer into the sphere of control', for instance, imply 'transfer into the sphere of experience". The overall structure, then, of the meanings affected by the process of generalization is one of nested subsets. The broadest category distinguished along the generalization dimension merely implies that the indirect object constituent is affected by the transfer

182

Constructions and idioms

referred to by the construction as a whole: the transfer involves a change (in the broadest possible sense) in the indirect object. At the most generalized end of the scale, the indirect object merely passively undergoes the effect of the transfer, without in any way being able to control or manipulate the transferred entity. At the other end of the scale, the indirect object constituent is a beneficiary in the strong sense: a human being who is free to use the received entity according to his own intentions. In between these two extreme positions, a continuum may be discerned going from 'control' over 'use' and 'experience' to 'affectedness'. Precisely because the generalization dimension is conceived of as a continuum, borderline cases are not to be excluded. For instance, because the 'use'-category distinguished in the figure is probably not a very prominent one as far as frequency of occurrence is concerned, it might already be considered a more or less marginal borderline case between the more prominent poles involving active control and manipulation on the one hand, and mere sentient experience on the other hand. In general, it is not necessarily the case that the intermediary positions distinguished in Table 2 are the only ones to be retained on a closer analysis. Table 2. Distinguishing features for control, use, experience, affectedness

CONTROL USE EXPERIENCE AFFECTEDNESS

the transfer involves a change of attributes and characteristics in the indirect object constituent

the transfer involves a sentient, mostly human being who is conscious of the change

the transferred entity is used but not manipulated by the beneficiary

the beneficiary has manipulative freedom in his use of the transferred entity

+ + + +

+ + +

+ +

+

4. Extension 2: Metaphorization The metaphorization dimension charted in Table I features two relatively straightforward patterns of metaphorical change. On the one hand, there is a shift from cases involving material transferred entities to cases involving abstract transferred entities. The link with the non-metaphorical readings

The indirect object in Dutch

183

shows up in the fact that the different positions on the generalization dimension that we distinguished earlier, also appear to be relevant for the metaphorized readings. The various abstract entities that are being transferred, have effects on the recipient that fall into the same categories as when material objects are being transferred. On the other hand, there are metaphorical extensions into the domain of communicative actions; broadly speaking, these transfers seem to rely on the conduit metaphor (Reddy 1979): messages are conceived of as entities that are being transferred from one person to the other. Logically, the recipient of a material transfer may thus be metaphorized into the recipient of a communicative transfer. One question that immediately imposes itself against the background of these metaphorical extensions is the following: why is the distribution of the instances of 'communicative transfer' over the 'control' etc. series so defective? An immediate answer to the question involves the recognition that a concept such as 'control' (as defined along the generalization dimension) simply does not apply to the objects of communicative transfers: the recipient of the message is more passive with regard to the message received than, for instance, the recipient of a present with regard to the present. Similarly, communicative transfers presuppose human (or at least animate) recipients, which excludes the 'affectedness' category as defined along the generalization dimension (because it specifically involved non-human recipients). Going one step further, however, it may be suggested that the asymmetrical distribution of the examples in Table I derives from a shift of perspective with regard to the prototype. Each of both metaphorical dimensions would then capitalize on different aspects of the prototype, and the different distributional patterns of both metaphors would be a reflection of that shift of perspective. Pursuing this line of thought, let us first identify the two perspectives involved. If the prototype case of the indirect object construction involves the indirect object as the recipient of a material transfer, the focus may lie either on the functional aspects of the transfer (the transfer as a process with specific effects for the recipient), or on the spatial aspects of the transfer (the transfer as a material change of place of an object). The simultaneous presence of both perspectives is reflected by the fact that the indirect object constituent in the prototypical case may be either identified as a Beneficiary (according to the functional perspective) or as a Recipient (when the focus is shifted somewhat to the spatial point of view). According to the functional perspective, the generalization continuum described in Table 2 is of prime importance: the various positions on the scale identify dif-

184

Constructions and idioms

ferent forms of functional effects that the indirect object constituent may be subject to. Conversely, distinctions between 'sphere of control', 'sphere of experience', 'sphere of affectedness' etc. recede to the background when the more purely orientational, spatial perspective is taken. Given, then, that the conduit metaphor takes its starting-point in the orientational conception of the literal prototype case rather than in the functional conception, it is not surprising to find that the distinction between the different positions on the generalization dimension turns out to be largely irrelevant for the metaphorical shifts into the field of communicative transfers. It may be noted that comparative evidence for the multiple perspectivizability of the prototype case might be found in languages where the dative case (the formally marked indirect object) expresses not just the notion 'recipient of transfer', but also the concept 'goal, direction of movement', in a clear-cut spatial, orientational sense. Such an extension of the category's range is a generalization of the spatial aspect of the transfer prototype, rather than of its functional aspects: the orientation of the movement is entirely detached from the notion of transfer. Incidentally, if we consider the dative in languages like Latin a morphologically marked indirect object, and if we further take into account that this morphological dative in Indo-European seems to be a development of an original locative case, it becomes clear that the diachronic relationship between the goal-reading and the recipient-reading may well be the converse of the synchronic relationship sketched above: historically speaking, the spatial aspects may have been primary with regard to the functional aspects. Such prototype shifts are not a theoretical anomaly, to be sure (compare Geeraerts 1997a).

5.

Extension 3: Metonymy

Two kinds of metonymy may play a role in the structure of the indirect object. First, we find indirect identifications of the indirect object entity: in a pattern like hij schrijft {het pa/eis, het stadhuis, het Instituut, de Wetstraat} een brief 'he writes {the palace, the town hall, the Institute, Downing Street 10} a letter', the human recipient of the communicative transfer is indicated by means of a reference to something (a building or dwelling-place) that is characteristically associated with the recipient in question. This is a form of metonymy that is quite wide-spread, and that is in no way specific for the

The indirect object in Dutch

185

indirect object construction. It will not be envisaged any further in what follows. Second, and more importantly, we find metonymical ramifications of the transfer aspect of the core indirect object meaning ('recipient of functional transfer'): instead of naming the recipient in an actual process of transfer, the indirect object constituent identifies entities or persons involved in actions or states of affairs that are metonymically associated with processes of transfer. Two important pattems may be discemed. On the one hand, there are states of affairs following acts of transfer: resultant states, effects, effectuated changes. If one says, for instance, dit cadeau is mU een kostbaar bezit 'this present is a precious possession to me', one describes oneself as the beneficiary of an act of transfer, but focusing on the situation after the actual transfer: the offering of the present itself is implied, but it is not explicitly mentioned. In such cases, the indirect object may be broadly defined as 'the entity affected by an (explicitly mentioned) state resulting from an (unmentioned) functional transfer (such as would be named in a core example of the indirect object construction)'. On the other hand, we find references to acts and states of affairs preceding acts of transfer: conditions, preparations, preliminary actions, intentions. If, for instance, one promises something to someone (ik beloof hem het koninkrijk, 'I promise him the kingdom'), the transfer has not yet taken place, but one describes someone as the beneficiary of a preliminary action (the promise) that will normally lead to the actual transfer. In these cases, the indirect object may be roughly defined as 'the entity affected by an (explicitly mentioned) action that is preparatory with regard to an (unmentioned) functional transfer of which the entity is the recipient'. Table 3 features examples of these two metonymical pattems ('resultant state' metonymies and 'preceding conditions' metonymies) for most of the types of use that we distinguished previously. In order not to clutter the figure too much, however, some simplifications have been applied: along the generalization dimension, the 'use' category has been left out; likewise, no instances of communicative transfers are included. (10)

Hij schenkt haar ko/fie in

'He serves her coffee' Ik gun hem het boek '1 grant him the book' (11)

Ik be/aa/hem het koninkrijk '1 promise him the kingdom'

186

Constructions and idioms

(12)

Er stond hem een regenbui te wachten

(13)

Zij gunnen hem dat genoegen niet

(14)

De burgemeester wenste het nieuwe museum veel bezoekers toe

(15)

Ik wens deze nieuwe theorie niets dan mislukte experimenten toe

(16)

Dit cadeau is mij een kostbaar bezit

(17)

Het rijk behoort haar toe

(18)

De hagel was hem onaangenaam

(19) (20)

Het valt mij moeilijk dat toe te geven 'Tt is for me difficult to admit that' De nieuwe kleur staat het huis goed

(21)

De aanpassingen waren de theorie tot voordeel

'A shower of rain was awaiting him' 'They do not grant him that pleasure' 'The mayor wished the new museum many visitors' '1 wish this new theory nothing but failing experiments' 'This present is a precious possession to me' 'The empire belongs to her' 'The hail was unpleasant to him'

'The new colour becomes the house' 'The changes were to the advantage of the theory' Table 3. Metonymical extensions of the indirect object construction

CONTROL EXPERIENCE AFFECTEDNESS

material abstract material abstract material abstract

PRECEDENT ACTIONS

RESULTANT STATE

METONYMY

METONYMY

(10) (11) (12) (13) (14) (15)

(16) (17) (18) (19) (20) (21)

It may be noted that the 'resultant state' cases typically involve the intransitive indirect object. This is probably an automatic, semantically driven consequence of the shift of perspective involved in the 'resultant state' metonymies: when the attention is shifted to the outcome of the transfer, the initiator of the initial transfer recedes to the background of the attention. A second point to note involves the fact that on top of the cases charted in Table 3, an additional type of metonymical extensions has to be distinguished. Let us notice, to begin with, that metonymical extensions need not imply that the original meaning is always relevant in those contexts where

The indirect object in Dutch

187

the new metonymical reading is applied. On the one hand, we find cases like he drank a glass: what he drinks is a 'quantity of water in a glass', but there is always at the same time a glass 'recipient of specific form and substance' around. On the other hand, we get cases like barrel: a barrel is a 'specific quantity of oil contained in a barrel', even if it is, in the circumstances, contained in a tanker or a pipeline. In the same way, to fall signifies 'to be killed by a weapon' even if the person does not fall down ('the parachutist who got caught in the tree fell under enemy fire'). Although this is not traditionally done, developments such as those affecting barrel and to fall could be seen as a generalization based upon an initial metonymy. For instance, barrel would undergo a metonymic shift from the initial 'recipient' reading to the reading 'amount of oil in such a recipient', and then further by generalization to 'same amount of oil, even if not actually in the recipient'. Now, regardless of the exact kind of analysis that one may wish to apply to such cases, it is crucial not note that the kind of extension exemplified by barrel and to fall also obtains in the indirect object construction. When, in fact, the metonymical 'resultant state' cases contained in Table 3 occur without reference to processes of transfer, 'pure' cases of 'possessor', 'experiencer', and 'affected entity' readings arise. For instance, the prototypically active recipient with powers of control and manipulation that features in the central cases of benefactive transfer, turns metonymically into an active, controlling entity in a stative relationship resulting from such a transfer; by a further process of generalization, it becomes an active, controlling entity in a stative relationship (regardless of whether an initial transfer took place or not). Table 4 contains a number of examples of the shift intended here. The examples are restricted to cases that link up with the 'resultant state' metonymies, but it should be noted that similar examples may be found with 'precedent conditions' metonymies. For instance, Ik gun hem zijn overwinning 'I grant him his victory, I do not begrudge his victory' may be said by someone who has no power at all to bring about (or to hamper) the victory. (22) (23) (24) (25)

Van al zijn lichaamsdelen was zijn linkerhand hem het dierbaarst 'Of all the parts of his body, his left hand was most dearest to him' Pannekoeken zijnmij het lekkerst 'Pancakes are the most tasteful to me' Jungles zijn tropische landen eigen 'Jungles are typical for tropical countries' Je vermogen om zeer snel te lezen zal je goed van pas komen 'Your capacity for reading very fast will be an advantage to you'

188

Constructions and idioms

(26)

AlIe vriendschap leek hem nutteloos

(27)

Zwaarmoedigheid is alle romantische muziek eigen

'All friendship seemed useless to him' 'Melancholy is common to all romantic music' Table 4. Secondary metonylllical extensions

CONTROL EXPERIENCE AFFECTEDNESS

6.

MATERIAL

ABSTRACT

(22) (23) (24)

(25) (26) (27)

Extension 4: Subjectification

On the basis of the 'experiencer' and 'affected entity' readings in the stative relations described at the end of the previous paragraph, a further extension leads to the so-called ethical dative, as in the idiomatic expression het is me wat 'that's something [to me, as far as I am concerned]!': the infonnation contained in the utterance is part of the sphere of experience of the indirect object (which is identical with the speaking subject), who is concerned, involved, interested with regard to the objective situation or event described by the utterance. Characteristically, the verb phrase need not refer to a process of transfer, which is an indication of the fact that the ethical dative is indeed a further development from the 'stative' cases described at the end of the previous paragraph. Against the background of the framework sketched so far, the ethical dative may be explained as a shift from the referential domain of the utterance to the discursive context. The interested party is not part of the referential domain in which the action takes place, but of the pragmatic domain in which the utterance is made. This shift is an example of the subjectification processes as described by Langacker (1990b) or Traugott (1989). (There are some differences between Langacker's and Traugott's conception of subjectification, but the differences need not detain our attention here). In general, subjectification involves a change in the conceptual construal of a particular relationship: from an objective construal (in which the relationship is construed without reference to the speech situation), a change is made to a subjective construal, in which the speaking subject (or another element from the speech situation) features as one of the relata in the relationship in question.

The indirect object in Dutch

189

In our case, the relationship could be, for instance, 'affectedness'. In the more common cases of the indirect object construction, the affected entity is part of an objective referential situation that is conceived of as being independent of the speech situation. In the case of the ethical dative, on the other hand, the affected entity is the speaking subject, as a basic element of the speech situation.

7. Extension 5: Perspectival switching A final extension to be noted involves cases in which the transfers that feature in the prototypical cases change polarity, so to speak, and are transformed into cases of blocked transfers or deprivations. Rather than receiving something, the indirect object entity is deprived of something: something is taken away from him that he already possessed, or he is denied something that he wishes for or that he expects to receive. The nuances that this reading occurs with are largely similar to the ones that were identified before. Without trying to be exhaustive about the possibilities, Table 5 lists examples involving actual transfers (where 'negative transfer' equals deprivation), together with a few examples involving stative relations; in each case, both material and abstract instantiations are included. (28)

Hij nam haar het speeltje a(

(29)

(30)

Een wolk ontnam ons het laatste straaltje zon ,A cloud took the last ray of sunshine away from us' Dit bericht ontnam haar de laatste kleur die haar gelaat nog sierde

(31)

Het ontbreekt hem aan het nodige geld om dat huis te kopen

(32)

Ze ontnamen hem aUe rechten

(33)

Het bericht ontnam Hemline haar gevoel van rust

(34)

De nieuwe snelweg ontneemt het landschap zijn laatste aantrekkelijkheid

'He took the toy from her'

'This message took the last shade of colour away that adorned her face' 'He lacks the necessary money to buy that house' 'All his rights were taken away from him' 'That message took Hermione's quiet feelings away'

'The new highway deprives this landscape from its last point of attractiveness' (35)

Het ontgaat haar dat het hoogtepunt van het generativisme achter de rug is

'She fails to see that the heyday of generativism is over'

190

Constructions and idioms

Table 5. Perspectival switches in the indirect object construction

negative transfer from the sphere of control negative transfer from the sphere of experience negative transfer from the sphere of affectedness stative relation of affectedness

MATERIAL

ABSTRACT

(28) (29) (30) (31)

(32) (33) (34) (35)

The explanation of the shift of perspective (or, if one wishes, the shift of polarity) that is illustrated by these examples is less straightforward than in the previous cases. Probably the most plausible analysis is to see the shift as the result of a generalization process affecting the 'transfer' component of the prototype case. Earlier on, we saw that the transfer component was subject to a metonymical shift in cases like Ik gun hem het back 'I grant him the book'. In the cases that concern us here, the transfer component is subject to a different type of change: it is generalized by dropping the 'benefactive' aspect that was associated with the prototype. The transfers envisaged here, in fact, do not add something to the sphere of control (etc.) of the indirect object constituent, but rather take something away: instead of being transferred in the direction of the indirect object constituent, the transferred entity moves away from the indirect object. Obviously, the indirect object can then no longer be characterized as a 'recipient', but had better be characterized as 'affected entity' or 'person concerned'. In the cases intended here, in other words, the generalization of the transfer component correlates with a shift in the 'recipient' component. Incidentally, this discussion shows that the label 'benefactive' that was used initially to characterize the prototype case may be slightly misleading, in the sense that it combines (in typically prototypical fashion) two different features that co-occur in the prototype case, but that are nevertheless separable. A benefactive transfer is typically one in which something good is given to someone: a positive for the indirect object constituent combines with a directionality oriented towards the indirect object. At the other extreme, the cases of deprivation illustrated in Table 5 involve transfers in which something good is taken away from someone. Notice, however, that giving something bad to someone (hi) gaf hem een klap 'he hit him') combines negative functionality with an orientation towards the indirect object. In the same way, taking something bad away from someone (haar antwoord ontnam hem zi)n somberheid 'her answer took away his melancholy mood') combines a positive, beneficial effect with an orientation of the transfer away

The indirect object in Dutch

191

from the indirect object constituent. The occurrence of all four combinations of directionality and effect shows, in short, that the generalizing extension of the 'benefactive transfer' aspect of the prototype actually covers two different processes. On the one hand, there is a perspectival change involving the directionality of the transfer. On the other, the functional aspect of the prototypical transfer may similarly (but independently) undergo a perspectival switch. Needless to say, this discussion as a whole once more establishes the importance of distinguishing between the functional ('benefactive') and the spatial (directional, orientational) aspects of the prototype case - a point that was already stressed in our discussion of the metaphorical extensions based on the conduit metaphor.

8. The resulting picture

If the prototype case of the indirect object construction is defined as 'the active recipient (with controlling power) of a benefactive transfer of material entities', various semantic extensions appear to start from each of the aspects of the prototype. An overview of the various extensions is presented in Figure 1. (For reasons of graphical economy, the extension towards the ethical dative is not included in the figure. Neither have the nuances of the perspectival switch discussed at the end of the previous section been explicitly represented. ) Crucially, the points in the network sketched in Figure 1 are not separate 'senses' of the indirect object construction, but conceptual components that co-occur in various combinations: Figure 1 only indicates the dimensions along which the indirect object construction may vary, but it does not chart all the individual readings that arise from the combination of the different positions that occur along the variational dimensions. In this sense, the representational fonnat of Figure 1 is fundamentally different from the 'radial sets' representations that are common in cognitive linguistic research: while the nodes in radial networks a la Lakoff (1987) constitute individual readings of the category in question, the representation of Figure 1 gives an overview of the conceptual components that may enter into such individual readings. For instance, a sentence like Zo 'n ge/uk overkomt je nooit meer (see Table 1) combines a shift from active recipient to passive experiencer along the generalization dimension (represented by the line going from the prototype to the lower right hand corner of Figure 1) with a shift from mate-

192

Constructions and idioms

rial to abstract transfer along the metaphorization dimension (the axis drawn vertically downward in Figure I).

affected entity instead of acti ve recipient

deprivation and blocked transfer instead of beneficial transfer

expenencer instead of active recipient

ill

communicative transfer instead of benefactive functional transfer

ACTIVE RECIPlENT OF BENEFICIAL TRANSFER OF MATERlAL ENTITY

preconditions and resultant states instead of transfers

stative relations instead of transfers

abstract entities instead of material entities

ill

GENERALIZATION

~

METAPHOR

Cl)

METONYMY

Figure I. The multidimensional structure of the indirect object construction

The indirect object in Dutch

193

Conversely, it is not necessarily the case that any point in the network that results from the combination of the dimensions in Figure 1, should be considered a different 'sense' (in the technical reading of that word). Already in the field of lexical semantics, distinguishing polysemy from vagueness is not always an easy task (compare Geeraerts 1993a). In syntax, then, the difficulties are even greater, because the criteria to apply are even less clear than in the field of lexical studies. Broadly labeled categories like Beneficiary or Affected Entity (Patiens), that occupy specific regions of the overall network, could certainly be considered candidates, but then again, according to what methodology should these categories be distinguished and demarcated? The road taken in the present paper has not been to try to answer the polysemy question, but rather to sketch the network of different uses with regard to which such a polysemy would have to be defined.

9.

Further steps to be taken

The analysis of the Dutch indirect object construction presented here does not pretend to be the final word on the subject, but merely tries to make a number of points that may prove stimulating for further research. In principle, there are two possible areas for a further elaboration of the model presented here. In the first place, the analysis as such will have to be refined. Specifically, by confronting the analyses given here with corpus data illustrating actual language use, it may be checked whether any additional semantic nuances have to be added to the picture drawn in the previous pages. Also, corpus data will yield frequency data that may confirm the prototypetheoretical structure assumed here. Is it the case, for instance, that the shades of meaning that are considered peripheral according to the present analysis, do indeed occur less frequently? In the second place, the model may be elaborated beyond the boundaries of the analysis given here. For one thing, the question arises whether the model has any typological and comparative validity: can the same structure be used to describe developments in other languages? Note that this question has two aspects to it: what languages share a similar construction with the same prototype as the one identified for the Dutch indirect object (see, for instance, Rudzka-Ostyn's identification of the prototype of the Polish dative: Rudzka-Ostyn 1996)? And - given such a prototypical similarity - can the extensions from the prototype that occur in the different languages be modeled along lines that are similar to the ones set out for Dutch? Another area

194

Constructions and idioms

for elaboration concerns the whole field of onomasiological research. The present analysis is a semasiological one: it describes the semantic structure of a single construction. At least for a number of the senses within the semasiological range of application of the indirect object construction, alternative modes of expression exist. First among these are the 'prepositional indirect objects', i.e. constructions of the type or . Assuming that these may indeed be considered different constructions (see the discussion in the first section of the article), the question arises whether the exact relationship among these onomasiological alternatives might be deternlined. In particular, would it be possible to analyze the choice between the ditransitive construction and the prepositional indirect object along the same quantitative lines set out in Geeraerts, Grondelaers, and Bakema (1994)? In that monograph, it is shown that the choice among lexical alternatives is determined simultaneously by semasiological and onomasiological salience effects (which may be calculated, given a sufficiently large corpus). Now, if the basic research strategy of Cognitive Linguistics (viz. of modeling the semantic analysis of grammatical constructions on the example provided by lexicological analyses) is carried through systematically, an extension of the analysis towards the onomasiological domain will have to be envisaged sooner or later.

10. Conclusion and discussion The major conclusion to be drawn from the present analysis involves the recognition that it is indeed possible to systematically and coherently describe the semantic structure of the (ditransitive) indirect object construction in Dutch on a 'lexical' basis, that is to say, by using models for semantic description developed in the lexicological domain. In the present analysis, this takes the form of starting from a prototypical meaning and deriving the other readings of the construction on the basis of well-known mechanisms of semantic extension such as metaphor and metonymy. As it stands, this general conclusion is not entirely original: the same idea lies at the basis of other recent analyses of datives and ditransitives, such as Goldberg (1992) and Rudzka-Ostyn (1996). On specific points of the analysis as well, there are many parallels with existing studies in a functional or cognitive linguistic framework. (These parallels have not been systematically indicated here, but see Rudzka-Ostyn 1996 for relevant references.) In what sense, then, does our analysis differ from these other proposals - in particular, from Goldberg

The indirect object in Dutch

195

(1992) and Rudzka-Ostyn (1996)? Restricting the comparison to these two articles, and focusing on fundamental matters rather than questions of detail, three points should be mentioned. First, instead of restricting the analysis to (or at least focusing it on) metaphorical links, as in Goldberg's (1992) radial set analysis of English ditransitives, other types of semantic extensions have to be considered. All the classical and less classical mechanisms of semantic extension (such metonymy, generalization, and subjectification next to metaphor) have to be envisaged in the analysis. The 'lexical' structure of the semantics of the indirect object construction cannot be adequately described by starting from an impoverished conception of semantic structure. This is an approach that the present analysis shares with Rudzka-Ostyn's (1996). Second, in contrast with Rudzka-Ostyn's proposal, the present analysis attaches no special structural position to extensions based on processes of generalization (which appear in Rudzka-Ostyn's proposal in the form of 'schematizations'). Just like Goldberg's use of a radial set representation seems to favor focusing on metaphorical links (given that the radial set model is predominantly used in the Lakovian tradition of metaphor-based analyses), Rudzka-Ostyn's adoption of a Langackerian schematic network representation seems to have predisposed her analysis towards overstressing the role of generalizations ('schematic' readings). Generalization, however, is but one of the many mechanisms in which the semantic coherence of a category may be grounded, and there does not seem to be any specific reason for making it more important than others. In particular, a category need not receive a single schematic reading to remain coherent. The obligatory search for such schematic readings easily leads to forced conclusions, and RudzkaOstyn's analysis is a case in point. Her identification of the topmost schematic reading of the Polish dative as a 'landmark' is, in fact, too schematic: it insufficiently distinguishes the dative from other constructions that also function as landmarks. The whole point, of course, mirrors the discussion in lexical semantics about the possibility of finding single definitions for lexical categories. Such definitions have to be maximally general and minimally specific: maximally general in the sense of applying to all the cases falling within the range of the category, and minimally specific in the sense of distinguishing the category from other, non-synonymous categories. A definition of the dative as a 'landmark', then, does not conform to the requirement of minimal specificity. On a more fundamental level, the point may be formulated as follows: given that lexical semantic research has established beyond doubt that monosemic, 'schematic' definitions of lexical categories

196

Constructions and idioms

need not always be possible (see Geeraerts, Grondelaers, and Bakema 1994), one has to beware of a forced search for such definitions in the case of grammatical constructions as well. Third (and foremost), instead of linking extended individual meanings directly to the prototype or to each other, a multidimensional structure has to indicate how non-prototypical readings are linked among each other by the co-occurrence of semantic shifts along various dimensions. Both Goldberg's radial set analysis and Rudzka-Ostyn's schematic network analysis interconnect individual readings of the construction in question, but insufficiently succeed in indicating the multidimensional structure underlying the interconnections between the individual readings. To appreciate the importance of these underlying dimensions, we may have another look at Figure 1. The representation used there visualizes an observation that was already brought to the fore in the second section of the article: the dimensions along which the semantic extensions take place, relate to specific components of the prototypical reading. For instance, assuming that the prototype case may indeed be defined as the recipient of a benefactive material transfer, one dimension of change pertains to the 'recipient' component, another to the 'transfer' component, and so on. We should now go one step further, however, and recognize that the major components in the definition link up with specific components of the syntactic frame in which the indirect object construction occurs. Focusing once more on the ditransitive -frame, it will be readily appreciated that the 'transfer' component of the definition links up with the verbal part of the formal construction, and the 'material' aspect of the definition with the direct object noun phrase (there is a material transfer when the direct object noun phrase refers to a material object). The 'recipient' itself, of course, is the second noun phrase, the indirect object as such. The definition that we have used, in other words, is a relational one that explicitly takes into account the syntagmatic dimension of the indirect object constituent. One could also say, alternatively, that the description is aframe-based one, in which the indirect object is defined as part of a larger conceptual structure. (Incidentally, while it might be suggested that this type of analysis characteristically distinguishes the description of verbal lexical items and grammatical constructions from that of nominal ones, this is not entirely adequate. Any relational concept, in fact, including relational nouns, will be most adequately described by incorporating the relational aspects into the definition.) What does not explicitly appear in Goldberg's and Rudzka-Ostyn's approaches, then, is certainly not the idea that the indirect object or the dative have to be de-

The indirect object in Dutch

197

scribed as (part of) a grammatical construction, but rather the insight that it is precisely this relational description that underlies the multidimensional nature of the semantic structure of the indirect object or dative. To summarize, what we have done, and what largely constitutes the specificity of the present approach, is a) to define the indirect object constituent as part of a construction: the components of the definition systematically evoke the other constituents that feature in the grammatical construction; b) to reveal the multidimensional semantic structure that correlates with this relational definition: the polysemy of the indirect object is determined by covariation of semantic extensions affecting the various components of the relational, construction-based definition; c) to stress the lexical nature of the semantic organization of the indirect object construction: an adequate picture of the changes affecting the components of the definition can only be acquired by envisaging all the mechanisms of semantic extension that have traditionally been identified in lexical semantics.

Chapter 8 The interaction of metaphor and metonymy in composite expressions

Originally published in Rene Dirven and Ralf Parings (eds.), 2002, Metaphor and Metonymy in Comparison and Contrast 435-465. BerlinlNew York: Mouton de Gruyter. Idioms and compounds have similar semantic characteristics: both types of expressions are semantically composite, in the sense that their meaning is (at least in principle) composed of elementary building blocks, i.e. the constituent parts of the expressions. Idioms have received considerable attention from Cognitive Linguistics (see e.g. Gibbs 1986, 1994, 1995, Fillmore, Kay and O'Connor 1988, Dobrovol'skij 1995, Feyaerts 1999), compounds much less so (see e.g. Ryder 1994, Rudzka-Ostyn 2003). The mechanisms of semantic compositionality as well have received only limited attention, with the exception of work in the line of mental space and blending theory, as in Sweetser (1999) or (Coulson 2001). The present paper takes a closer look at the semantics of composite expressions, without however attempting to provide a systematic comparison with the blending analysis: drawing on Geeraerts and Bakema (1993) and Geeraerts (l995b), it describes the interaction between the syntagmatic and the paradigmatic axes in the meaning of idioms and compounds, and then charts the various ways in which metaphor and metonymy can interact along these axes. The crucial idea is this: when we study compositional processes in the framework of a semantically oriented grammar like Cognitive Linguistics, we should not only describe the way in which the constituent elements work together to produce a composite meaning, but we should also take into account the semantic shifts that each of the elements (and their combination) may undergo in its own right.

1.

Metaphor and metonymy in compounds and idioms

The semantic architecture of idioms and compounds is identical: in both types of expressions, meaning is compositional to the extent that it is built up out of the constituent parts of the expression; at the at same time, this compositional meaning is usually but a first semantic step, to the extent that processes of meaning extension produce a figurative reading from the compositional one. This paper will explore the various ways in which metaphor

The interaction ofmetaphor and metonymy

199

and metonymy may interact in expressions with this type of semantic architecture. The first part of the paper describes the architectural characteristics, i.e. the paradigmatic and the syntagmatic axes of composite expressions and the way in which they interact. This part of the paper reproduces the major part of Geeraerts (1995 b). The phenomena in question will be introduced with regard to idioms in particular, but the extrapolation to the semantics of compounds is a straightforward one. The second part of the paper charts the various ways in which metaphor and metonymy combine and interact within the structure defined by the syntagmatic and paradigmatic axes. In particular, it will be shown that metaphor and metonymy may occur either in a consecutive or in a parallel sequence. This part of the paper is loosely based on Geeraerts and Bakema (1993). All the examples illustrating the argument are taken from Dutch.

2. The prismatic architecture of composite expressions

2.1. Isomorphism and motivation The paradigmatic and the syntagmatic dimension 1 of idioms are both twofold, in the sense that both can be considered with regard to the original, literal meaning of the idiom, and with regard to the derived, figurative meaning. The paradigmatic dimension of idioms primarily involves the relationship between the original meaning of the idiomatic expression as a whole and its derived meaning. Secondarily, it involves the relationship between the original, literal meaning of the constituent parts of the idiomatic expression, and the interpretation that those parts receive within the derived reading of the expression as a whole. The syntagmatic dimension of idiomatic expressions involves the relationship between the interpretation of the constituent parts of the expression on the one hand, and the interpretation of the expression as a whole on the other, but clearly, this syntagmatic dimension can be envisaged both with regard to the original meaning and with regard to the derived meaning. Systematically, then, the semantic relations in idioms might be charted by means of a prismatic structure as in Figure 1. (The figure is intended to serve purposes of clarification and reference only; it is obviously not a formal representation as envisaged in formal grammar. To keep matters simple, the expression is assumed to contain only two lexical items.)

200

Constructions and idioms

1 Expression as a whole in its literal reading 2 First constituent item in its literal reading 3 Second constituent item in its literal reading 4 Expression as a whole in its derived, idiomatic reading 5 First constituent item in its derived reading 6 Second constituent item in its derived reading Figure 1. The prismatic structure of idioms

In addition, the notion of compositionality may either receive a dynamic or a static interpretation. Within the dynamic interpretation, compositionality is thought of as a syntagmatic derivational process in the course of which the meaning of a compound expression is computed on the basis of the meanings of the constituent parts of the expression'" In contrast with this dynamic, bottom-up conception, a static interpretation can be envisaged that merely notes that a one-to-one correspondence between the parts of the semantic value of the expression as a whole and the meanings of the constituent parts of the expression can be detected, regardless of the question whether this correspondence has come about through a process of bottom-up derivation or through a top-dO\vn interpretative process. Examples of such interpretative processes will be given later on; at this point, it suffices to see that a 'non-directional' conception of compositionality is not excluded. Now, because the term compositionality, through its processual connotation, strongly calls up the idea of bottom-up derivation, a different term might be helpful for the neutral, non-directional interpretation. I propose to use the

The interaction ofmetaphor and metonymy

201

term isomorphism here: what is ultimately at stake, is a one-to-one correspondence between the formal structure of the expression and the structure of its semantic interpretation, in the sense that there exists a systematic correlation between the parts of the semantic value of the expression as a whole and the constituent parts of that expression. A third preparatory step can be made by refining the notion of derivation. Although statements to the effect that idiomatic meanings are not compositional, that they are specialized, and that they cannot be derived straightforwardly would seem to be interchangeable, it should now be clear that we have to distinguish in principle between the syntagmatic underivability of the meaning of an expression (which is, of course, its non-compositionality), and its paradigmatic non-derivability. The latter involves the transparency of the semantic extension that leads from the original meaning of an expression to its transferred reading. It is illustrated by the interpretative difficulty that we noted with regard to (3): along the top line of the prism in Figure 1, the transition from the literal meaning of iemand iets op de mouw spelden to its idiomatic meaning is opaque. The latter cannot be derived on the basis of the former, because the motivating image is lost. I propose to call this type of derivability motivation (in contrast with dynamic compositionality as a syntagmatic kind of derivability). To round off the preparations, let us note that isomorphism and motivation as defined here 3 share a common characteristic: both involve the transparency of some of the links indicated in Figure 1. More specifically, isomorphism coincides with syntagmatic transparency, whereas motivation can be defined as paradigmatic transparency. 2.2. Combinations of isomorphism and motivation We can now arrive at a basic classification of the specialized nature of idioms when we consider the various combinations that isomorphism and motivation can occur in. Isomorphism and motivation can, of course, each be considered at two points in the relational structure sketched in Figure 1. On the one hand, motivation may refer both to the paradigmatic top line of the prism and to the two paradigmatic bottom lines. On the other hand, syntagmatic isomorphism may involve the front triangle or the back triangle of the figure. Because the entity that we are primarily interested in is the idiomatic meaning of the expression as a whole (the top backside corner of the figure), we will for now only consider motivation and isomorphism as they directly relate to that part of the structure, i.e., we shall consider isomorphism within

202

Constructions and idioms

the figurative plane, and motivation on the global level. Extensions towards other parts of the structure will follow in the next sections. Consider, then, the following expressions. (1)

De koe bij de horens vatten 'To take the cow by the horns> 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

-. <\l

;>. ~

Q

high fermentation spontaneous fermentation fermentation in bottle wheat based cherry f1avor

+

"-

.~

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

~

;;:

~ .~

+

+

Cl"::':

+ +

..::,:

..::,:

>,

:.c;

<\l

..::,::.c;

<\l

lE::

~

0;:: ::::

I:lI)

+ +

+

<\l

~

<\l

<\l

..::,:~

+ +

Each beer type, in other words, can be considered a bundle of features, each of which can be the basis for the name of the beer (or part of the name of the beer). But how then can we go about determining the conceptual salience effects that we are interested in? If each beer type were to belong uniquely to a neatly separated, taxonomically well-behaved category, we would just have to investigate those categories. But if the categories themselves are extremely varied and exhibit multiple overlapping, what should one do? I earlier referred to Berlin's basic level model as the only systematically elaborated model for salience within lexical fields. But note that Ber-

262

Meaning and culture

lin's model deals with the salience of levels in a taxonomy, while we are interested here in salience phenomena among categories on fundamentally the same level of a taxonomy. So it seems that it may be worthwhile to suggest ways of dealing with intercategorial (rather than intracategorial) salience effects in a way that goes beyond the basic level model. (The ideas presented here are developed in more theoretical and empirical detail in Geeraerts, Grondelaers, and Bakema 1994. The notion 'intercategorial salience' is briefly introduced in Geeraerts 1993b; it is basically a quantitatively operationalized version of the concept of entrenchment defined by Langacker 1987: 59-60. A useful presentation of the distinction between intracategorial and intercategorial entrenchment may also be found in Kleiber 1991.) The obvious thing to do, I would suggest, is to start from the features themselves that constitute the bundles representing each individual type of beer. A practical methodology can then be formulated 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. On a general methodological level, I would like to suggest the following operational definition of conceptual salience in lexical fields: a category in a lexical field is more salient to the extent that its members are more often identified by names that are typical for the category. Of course, we have to be more precise about what it means for a feature '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 ofa name) when that name (or the relevant part of it) only occurs in connection with that feature. 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 tafelbier 'table beer' also signals the light character of beer, to the extent that it only occurs within the group of very light beers.

Beer and semantics

263

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. Kriek, however, does not express the feature of spontaneous fermentation, because (as we have seen in Figure 3) it also occurs in the name of beers of high fermentation. 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 2 presents a sample to demonstrate the kind of differences in salience that occur. I should add immediately that it is practically impossible to systematically explore all possible features, mainly because their presence is often uncertain: even persistent tasting (to the extent that the researcher can muster it) does not always suffice to establish the objective characteristics of a particular brand. 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. Table 2. Salience characteristics of sample referential features REFERENTIAL FEATURE

REFERENTIAL FREQUENCY

NAME FREQUENCY

based on wheat very light, low alcohol occasional 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%)

The relationship among the figures mentioned to the right gives an indication of which features are considered special enough to merit a name of their own, or at least, special enough to be referred to in the beer names. Trappist origin, for instance, seems to be highly valued, while the addition of herbs is not a terribly individuating feature. It is tempting, of course, to speculate about the origin of these preferences. It is not unlikely that they reflect historical traditions within the culture, but this is, to be sure, something that cannot be established on the basis of this scanty set of examples alone - one would need a longitudinal diachronic analysis to get any certainty on this point.

264

Meaning and culture

On the other hand, the examples at least allow for the falsification of one hypothesis. It might be suggested that the salience of a particular feature is inversely correlated with its referential frequency: an infrequent feature like the presence of cherry flavor is lexically expressed to a high degree, precisely because it is a marked, unexpected feature. This phenomenon probably does play a role in the lexical field as a whole: no one would expect a reference to water in the name of beers, precisely because it is obvious to anyone with the least encyclopedic knowledge of the domain that all beers contain water. However, the figures in the table clearly establish that such an inverse correlation with referential frequency could at most explain part of the data. There seem to be at least some real preferential phenomena at play.

5. Salience effects for type 2 names I would now like to tum to the analysis of what I have earlier called the associative values of the beer denominations. In Figure 5, five salient associative motifs are identified. They are recurrent ones not just because they show up in various individual names, but also because they generalize over more than one source domain. It would lead us too far to discuss all the names separately. but here is a sample analysis of some illustrative cases. A first recurrent pattem represents beer as something prestigious. In its most direct form, the high status of beer is evoked by borrowing names from the lexical field of wine (the prestigious drink par excellence), as in Vlaamse Bourgogne 'Flemish burgundy'. The same value may be evoked more indirectly by referring to important people, like kings (Grand Monarque 'Great monarch') or members of the aristocracy (Ridder 'Knight'). A link with the religious motif that occurs a number of times in the set is constituted by names referring to hierarchically important people in religious organizations (like Monseigneur 'Monsignor' and Kapittel abt 'Chapter abbof). Another set of prestigious figures involves historically important people, who may be political leaders like emperor Charles V (Charles Quint) or culturally prominent people like the painters Breugel, Rubens, and Rembrandt. In most cases, these figures have a direct historical link with the Low Countries, but names like Napoleon and Da Vinei show that this is not an absolute precondition. Next prestigious people, expensive materials and other valuables may be evoked, as in Salir; a cross-references to one of the other subcategories already mentioned appears in a name like Gouden Carolus. which is originally the name of the gold coin minted by emperor Charles V.

Beer and semantics

265

A second major motif represents beer as an adventure at the edge of the accepted norms. Up to a certain point, the notion of adventure (danger, physically extreme conditions, violence) and transgressive behavior (flouting existing norms) may be separated, but because they so often occur together, they are here treated as one category. The norms that are violated may be basically religious or social. The first case yields names referring to sin (Verboden Vrucht, 'Forbidden fruit'), witchcraft (Sorcieres d 'Ellezelles 'Witches of Ellezelles'), and the devil (Lucifer, Satan, Duvel 'Devil'). The second case yields names referring to pirates (Piraat, Boucanier), smugglers (Smokkelaar), highwaymen (Bokkereyer) and related unorthodox professions. Also, names like Schavuit, Deugniet, and Stouterik 'rascal' indicate moderated mischievousness; they are typically the kind of name to be used hypocoristically to a child. Brigand and Kerelsbier are particularly interesting cases. For instance, in the Belgian context, Brigand refers to the peasants and farmers who rebelled against the French occupation at the end of the eighteenth century (the Boerenkrijg 'peasants' war'). Looking back at the historical figures mentioned above, it now becomes clear that they too often (as in the case of Ambiorix, Artevelde, Breydel and De Coninck) represent a struggle for independence against a foreign oppressor. (It will be suggested below that an irreverent or even rebellious attitude with regard to established authority may be a general characteristic of Belgian culture.) Names like Stoeren Bonk 'sturdy fellow, son-of-a-gun' evoke virility and force. Against the background of what was just said, it seems plausible that these names do not however evoke machismo pure and simple: rather, the image is at least in part one of strong and independent men who stand up for their rights. A final major source domain in this group points to the physical effects of overabundant consumption of beer, as in Delirium Tremens. The idea probably is that you have to be a strong person (of the type evoked in the previous group) to take such strong drinks. Note, in addition, that multiple associations may again be active: Houten Kop 'wooden head' may not only mean 'hangover', but also stubbornness, which is again related with the spirit of independence. Names like Flierefluiter 'Iayabout, idler', Lekkerbek 'gourmet', and Pallieter (the name of a famous literary figure who has become symbolic for an epicurean attitude) are typical for a third major motif, in which beer is heralded as a source of pleasure and as part of an easy-going way of life. Characteristically, the religious motif appears once again when beer is so to speak presented as a foretaste of the pleasures that may await us in heaven. This is the case in among others Engeltjesbier 'little angels' beer'.

266

Meaning and culture

The religious source domain also contributes to the large group of names in which beer is represented as a link with the traditional culture of the past. Religion, to be sure, is such an aspect of the tradition, in the same way in which the guilds of old (as in Gildenbier 'beer of the guilds' are. References to old methods of producing beer (Traditions des Moines 'The monks' traditions') serve a similar function. In A la ferme 'On the farm', Verte campagne 'Green countryside', and Bosbier 'Beer of the woods' rural if not pastoral associations are activated: to the extent that beer is part of the legacy of the past, it is clearly the pre-industrial era that is at stake. Historical references, to be sure, are a recurrent phenomenon in the set presented here. Apart from the historical figures mentioned before and the references to local history mentioned below, there are straightforward references to the past in names like Vieux Temps 'Old times'. In Druide 'Druid', l'iguanodon 'The iguanodon', and Blonde du Menhir 'The blonde one of the menhir', the historical reference takes a remarkable prehistoric, pre-Christian shape: the historical reference is so to speak pushed to its extreme, but at the same time, the notions of primitivity and force that seem to accompany these references echo some of the vigorous, vitalistic values connected with the second motif discussed above. The final major motif represents beer as an aspect of local identity, by referring to people and things associated with a particular region. The relevant region can change in surface, to be sure. It may be Belgium as a whole (as with historical figures like the painters Rubens and Breugel, or the cartographer Mercator), or it may be one of the main regions of Belgium (as when the historical figures Artevelde, Breydel, and De Coninck evoke the past of Flanders in particular, and when Elckerlyc and Pallieter refer to figures from Flemish literature); very often, it is a specific town or village (as in the many names that mention a particular placename, or the place's patron saint). The latter case is also represented by those names that make reference to aspects of local folklore, like famous local figures (Ketje) , mysterious beings that appear in legends and traditional stories (Antigoon, Cnudde), or salient landmarks (Manneken Pis). Linguistically interesting, to round off, are those names where a local identity is evoked by dialect forms (which are often at the same time archaic). The relevant linguistic features include inflection (Stoeren Bonk would be Stoere Bonk in standard Dutch), pronunciation (Duvel instead of dUivel), or spelling (as the typically archaic sch in Oude Tongerschen Bruinen). These features may, of course, be combined with each other and with values discussed before. The result may be quite surprising: in particular, notice how the orthographical form Lowie Kators

Beer and semantics

267

for the name of the French king Louis Quatorze (cp. the source domain of royalty and aristocracy) evokes a Dutch, dialect-like pronunciation, and as such ironically combines the lofty position of the foreign king with the downto-earth status of the local dialect. Table 3. Recurrent associative values in beer names Beer is a prestigious entity Wine tenninology

Vlaamse Bourgogne, Cuvee Chateau des Flandres, Caves, Rose de Cambrinus

Royalty and aristocracy; VIPs

Ada Royal, Grand Monarque, King's Ale, Regal, Ridder, Royal pils, Marquise Monseigneur, Moeder Overste, Kapittel abt

Historical figures

Ambiorix, Artevelde, Breydel en de Coninckbier, Hertog Jan pils, Charles Quint Breugelbier, Rubensbier, Rembrandt's bier, Vondel, Mercator Lowie Kators, Napoleon, Da Vinci

Wealth and riches

Fortuyn, Gold Pils, Gouden Carolus, Parel Pils, Silver Pils, Safir, Toison d'or

Beer is an adventure at the edge ofthe accepted norms Transgression of religious norms

Duivels bier, Duvel, HopduveL Judas, Lucifer, La Pecheresse, Satan, Verboden Vrucht, Sorcieres d'Ellezelles

Transgression of social norms

Boucanier, Piraat, Smokkelaar, Schavuit, Stouterik, Averechtsen, Bokkereyer, Deugniet, Bakelandt, Brigand, Kerelsbier, Voyoll

Virility and force

Het geheim van Jan de Sterke, Kastaar, Bink, Brugse Straffe Hendrik, Stoeren Bonk, De Soldaat

Strong bodily effects

Delirium Tremens, Houten Kop, Katerke, Killer Pils, Mort Subite

Beer is a source ofpleasure Carpe diem

Bacchus, Cupido, Drongens plezierke, Flierefluiter, Joyeuse blonde, Lekkerbek, Pallieter, Plaisir, Relax, Cote d'azur speciale

268

Meaning and culture

Heavenly pleasure

Engeltjesbier, Bons voeux, Sanctus, Onze vader z'n bier + various names of saints

Beer constitutes a link with the traditional culture ofthe pre-industrial era Traditional methods of production, old trades

Brouwmeester, Gildenbier, Keurbier, Traditions des Moines, Oerbier, Retro oude methode, Natuurbier, Saison 1900

Rural origins

A la fenne, Bosbier, Boskeun, Verte campagne, Fenne de Grand Pre, Pays vert, Peerdevisscher, Oelens boerke

History and prehistory

Antiek, Vieux Temps, Druide, Blonde du Menhir, l'Iguanodon; and see below ('Local history') and before ('Historical figures')

Religion

as before: Monseigneur, Moeder Overste, Kapittel abt Engeldesbier, Bons voeux, Sanctus, Onze vader z'n bier + various names of saints

Beer is an aspect oflocal identity Dialect forms and archaic forms

Duvel, Lowie Kators. Stoeren Bonk, Peerdevisscher, Drongens plezierke, Oude Tongerschen Bruinen

Local history

Ambiorix, Artevelde, Bakelandt. Breydel- en de Coninckbier, Kerelsbier, Hertog Jan pils Breugelbier, Rubensbier. Rembrandt's bier, Vondel, Mercator Gouden Carolus, Toison d'or Elckerlyc, Pallieter, Prutske bier, Hercule Poirot

Local folkore

Antigoon, Cnudde kriek, Het geheim van Jan de Sterke, Ketje, Rosse Lei, Rooie Rietje Stekselbier, Tineke van Heule, Manneken Pis, Geitenbier

There are three additional remarks to be made with regard to this overview of associated values. First, it is not always easy to get the etymological origin of the names right, or rather, the actual etymological origin is not necessarily the one that is relevant for the value that seems to be evoked now by the name. For instance, Mort Subite as 'sudden death'. To the extent that it is interpretable at all, it could only be in tenns of the physiological impact of strong beers. Originally, however, the name was that of a particular cafe plus brewery in Brussels; the name of the cafe itself apparently referred to a

Beer and semantics

269

particular type of backganunon popular in the pub. Some methodological caution, in other words, is necessary when dealing with this kind of material. As a second remark, it will be noted that the list contains both French and Dutch names. In principle, it might have been possible to analyze both sets of names separately, on the assumption that we might then get an insight in the differences between the culture of the Dutch speaking part of Belgium (Flanders) and its French speaking part (the Walloon country). I will, however, proceed on the assumption that, as far as beer goes, Belgian culture is fairly uniform, This is supported by the recognition that there is no unambiguous link between the language of the brad name and the geographical position of the brewery. Some French names, for instance, derive from breweries in Flemish territory. Moreover, it appears from sociological surveys (most recently, Kerkhofs, Dobbelaere, Voye, and Bawin-Legros 1992) that the opinions and behavior that distinguish Flemings from Walloons are by far outweighed by what they have in conunon in contrast with the Dutch, Germans, or French (to name only the inunediate neighbors). A third remark is simply that a specific name may be motivated by various evocative values at the same time, as has become apparent at various points in the preceding discussion. A name like Ambiorix, for instance, refers to the leader of one of the Celtic tribes that rebelled against Julius Caesar's invasion of this part of Gaul. In this sense, the name Ambiorix combines the motifs of local identity, of prestigious leadership, of the far historical past, and of resistance to an imposed authority. The fourth remark is the crucial one. If beer is as central to the Belgian self-conception as I claimed earlier, do these aspects of Belgium's conception of beer reveal something of Belgium's self-conception? Methodologically, answering such a question requires going beyond linguistics and cooperating with sociologists, historians, and anthropologists. That is more than could be done in the space of this paper, but let us note that linguistics is not entirely unequipped to deal with the question. At least one thing we can do as linguists is to compare the Belgian situation with that in other countries. Taking The Netherlands as a point of comparison, some differences do appear that support the idea that the pattern that emerges for Belgium is a culturally specific one. For instance, while the prestige motif is as outspoken in The Netherlands as it in Belgium (which may be due to a universal conunercial strategy of appraisal), the link with the pleasure motif is much weaker, and references to local identity (at a level below that of the country as a whole) are rather scarce. So are references to religion. Most tellingly, per-

270

Meaning and culture

haps, references to anti-authoritarian transgressive behavior are almost entirely absent. These differences can be meaningfully interpreted against the background of the different cultural and political history of both countries. (These differences can only be summarily evoked here. For more background information, see Verdoodt 1978, where a sketch of the cultural and political history of the Low Countries provides the backdrop for a sketch of the divergent linguistic evolution of Belgian Dutch and Netherlandic Dutch.) First, the protestant, calvinistic religion that was dominant in Holland from the sixteenth century on did not favor the enjoyment of earthly goods, while the Catholic religious atmosphere in Belgium was much less austere. This may explain why the link with the religious motif and with the pleasure motif is almost absent in the brand names found in The Netherlands. Second, Holland became a nation, a unified national state in the seventeenth century, while the Belgian provinces remained under foreign (Spanish, Austrian, French) influence until the nineteenth century. This correlates with the fact that Dutch people identify more readily with the national state, whereas the inhabitants of Belgium traditionally have a genuinely irreverent attitude with regard to the political authorities at large. Apparently, then, this difference in mentality also shows up in the different patterns of associated values for beer names in both countries.

6. Concluding remarks Possible conclusions to be drawn from this field trip through lexical beer country include the following. First, a systematic investigation into salience phenomena in lexical fields should be high on the list of methodological priorities of Cognitive Linguistics, if it is to reach its aim of doing cultural analysis by linguistic means. Second, in analyzing such salience phenomena, a distinction has to be maintained between the salience of 'literal' referential features and the salience of associated values. In the foregoing pages, this distinction showed up in the distinction between 'Type l' and 'Type 2' names. Third, proper names (and in particular, commercial brand names) are amenable to a cognitive analysis in terms of recurrent associative and metaphorical values. Fourth, the wealth and diversity of Belgian beers is reflected in the variety of beer names; moreover, there is at least some evidence that the semantic patterns that emerge from the linguistic analysis correspond with real culturally specific aspects of behavior and mentality. And fifth, the

Beer and semantics

271

lexicology of beer names can be fun - but whether it is more fun than the tasting of beer itself is for the reader to explore.

Chapter 11 Cultural models of linguistic standardization

Originally published in Rell(~ DilVen, Roslyn Frank and Martin Piitz (eds.), 2003, Cognitive Models in Language and Thought. Ideology, Metaphors and Aleanings 25-68. Berlin/New York: Mouton de Gruyter. What are the models that shape our thinking about language as a social phenomenon? What are the paradigms that we use to think about language, not primarily in terms of linguistic structure, but in terms of linguistic variation: models about the way in which language varieties are distributed over a language community and about the way in which such distribution should be evaluated? In this paper, I argue that two basic models may be identified: a rationalist and a romantic one. I chart the ways in which they interact, describe how they are transformed in the course of time into a nationalist and a postmodern model, and explore how the models can be used in the analysis of actual linguistic variation. The paper constitutes a link between the type of semantic studies illustrated in the previous chapters, and the variationist, sociolinguistically oriented studies that grew out of the shift towards an onomasiological form of lexicology as described in Chapter 4 and elsewhere in this volume. There are various substrands to this type of research, which is the main focus of our Leuven research group Quantitative Lexicology and Variational Linguistics: sociolectometrical research into the relationship between language varieties (as illustrated in the final part of the present paper, and in studies such as Geeraerts, Grondelaers and Speelman 1999; Geeraerts 200la, 200lb; Grondelaers, Geeraerts, Speelman and Tummers 2001; Geeraerts 2002a; Speelman, Grondelaers and Geeraerts 2003; Van Gijsel, Geeraerts and Speelman 2004), statistically advanced, corpus based analyses of multifactorial sociolinguistically sensitive phenomena (as in Grondelaers, Brysbaert, Speelman and Geeraerts 2002; Grondelaers, Speelman and Geeraerts 2002; De Sutter, Speelman, Geeraerts and Grondelaers 2003; Tummers, Speelman and Geeraerts 2004), and theoretical arguments for incorporating social variation into the scope of Cognitive Linguistics (Geeraerts 2002b, 2005).

1.

Cultural models and language variation

In line with well-known trends in cultural theory (see Burke, Crowley and Girvin 2000), Cognitive Linguistics has stressed the idea that we think about social reality in terms of models - 'cultural models' or 'folk theories': from

Cultural models ofstandardization

273

Holland and Quinn (1987) over Lakoff (1996) and Palmer (1996) to Dirven, Hawkins and Sandikcioglu (2001) and Dirven, Frank and Hie (2001), cognitive linguists have demonstrated how the technical apparatus of Cognitive Linguistics can be used to analyze how our conception of social reality is shaped by underlying patterns of thought. But if language is a social and cultural reality, what are the models that shape our conception of language? Specifically, what are the models that shape our thinking about language as a social phenomenon? What are the paradigms that we use to think about language, not primarily in ternlS of linguistic structure (as in Reddy 1979), but in terms of linguistic variation: models about the way in which language varieties are distributed over a language community and about the way in which such distribution should be evaluated? In this paper, I will argue that two basic models may be identified: a rationalist and a romantic one. I will chart the ways in which they interact, describe how they are transformed in the course of time, and explore how the models can be used in the analysis of actual linguistic variation. There are two preliminary remarks that I should make in order to situate the present paper against a wider background. First, the analysis is a marginal offshoot of a more central interest in empirical methods for studying linguistic variation and change. The work that I have been doing over the last ten years or so with my research group has specifically focused on various aspects of lexical variation and change: diachronic semantics (Geeraerts 1997a), the relationship between semantic and lexical variation (Geeraerts, Grondelaers, and Bakema 1994), and lexical variation within pluricentric languages such as Dutch (Geeraerts, Grondelaers and Speelman 1999). Within the latter line of research, we have been particularly concerned with the development of quantitative techniques for measuring lexical variation and processes of lexical standardization (see section 4 below). There are two ways, then, in which the present more or less essayistic paper links up with the more rigorous descriptive and methodological work that is my basic field of interest. For one thing, an investigation into linguistic usage needs to be complemented by an investigation into the way in which the users of the language perceive the actual situation. The cultural models that I will be talking about define, in a sense, basic language attitudes - and an adequate interpretation of language variation should obviously take into account language attitudes along with language behavior. At the same time, both perspectives (the behavioral and the attitudinal) have links with Cognitive Linguistics. Whereas the attitudinal approach draws inspiration from the Cognitive Linguistic analysis of cultural models

274

Meaning and culture

and folk theories, the descriptive approach is a further development of the Cognitive Linguistic interest in lexical-semantic variation as represented by prototype theory. Underlying the publications mentioned above is a logical line of development from semasiological prototype theory (Geeraerts 1997a) to a model of lexical variation encompassing onomasiological variation (Geeraerts, Grondelaers, and Bakema 1994), which then further broadens to the investigation of 'external', sociolectal and dialectal factors of variation (Geeraerts, Grondelaers and Speelman 1999). As a second preliminary remark, is there a difference between a 'cultural model' and an ideology? It is a common idea in Cognitive Linguistics that the cultural models underlying reasoning and argumentation are to some extent idealized entities (see, for instance, the notion of ICM's or 'Idealized Cognitive Models' as introduced in Lakoff 1987). Actually occurring phenomena and situations usually differ to a smaller or a greater extent from the models that act as cognitive reference points: the models themselves, then, are to some extent abstract, general, perhaps even simplistic, precisely because we use them to make sense of phenomena that are intrinsically more complicated. With regard to social phenomena, this means that cultural models may turn out to be not just idealized entities, but also ideological ones. Cultural models may be ideologies in two different respects: either when their idealized character is forgotten (when the difference between the abstract model and the actual circumstances is neglected), or when they are used in a prescriptive and normative rather than a descriptive way (when they are used as models of how things should be rather than of how things are). In the latter case, an ideology is basically a guiding line for social action, a shared system of ideas for the interpretation of social reality, regardless of the researcher's evaluation of that perspective. In the former case, an ideology is always to some extent a cover-up, a semblance, a deliberate misrepresentation of the actual situation, and a description of such ideologies will of necessity have to be critical. The distinction is of course well-known in ideology research, and there is an extensive linguistic literature probing the relationship between language and ideology. There are two basic (and to some extent overlapping) approaches here: on the one hand, all forms of critical discourse analysis, as represented by Van Dijk (1998), Wodak and Meyer (2001), or Blommaert and Bulcaen (1997); and on the other, the 'ideologies oflanguage' approach, as represented by Joseph and Taylor (1990), Woolard, Schieffelin and Kroskrity (1998), and Schiffman (1996). The former approach critically

Cultural models ofstandardization

275

analyzes any text with regard to its position in the social power play - with regard to the way, that is, in which it reproduces or counteracts existing social relations. The latter approach concentrates on how beliefs about language variation and specific linguistic varieties manifest themselves explicitly (as in language policies) or implicitly (as in educational practices), and how they interact with group identity, economic development, social mobility, political organization. In the following pages, I will not take a critical approach, but rather start from a neutral and descriptive conception of linguistic cultural models. Rather than critically analyzing specific practices and policies as ideological, I will try to explore the underlying structure and the historical development of the competing cultural models that lie at the basis of such practices and policies as well as their critical analysis.

2.

The rationalist and the romantic model

In this section, I will present the two basic cultural models that I think need to be distinguished if we want to get a grip on the logic of standardization debates: the rationalist one and the romantic one. I will present them in mutual contrast, showing how they are to a large extent each other's counterpart, and how they are dialectically related. The present section will not however exhaust the comparison between both models. In the next section, the comparison will be further expanded, leading to the identification of two historical transformations of the basic models, in the form of a nationalist and a postmodem model. 2.1. The rationalist model So what are the characteristics that are ideally (and perhaps ideologically) attributed to standard languages? The most conspicuous feature is probably the generality of standard languages. Standard languages, in contrast with dialects and other restricted languages, are general in three different ways. They are geographically general, in the sense that they overarch the more restricted areas of application of dialects. Further, they are socially general because they constitute a common language that is not the property of a single social group but that is available to all. Finally, they are thematically universal in the sense that they are equipped to deal with any semantic domain or any linguistic function. More advanced domains of experience in

276

Meaning and culture

particular (like science or high culture) fall outside the range of local dialects. Because of their generality, standard languages have two additional features. First, they are supposed to be a neutral medium, with a mediating function, in an almost philosophical sense of 'mediation'. Standard languages, in fact, transcend social differences: they ensure that men and women from all walks of life and from all corners of the nation can communicate freely. In that sense, they are a medium of participation and emancipation. Because of their neutrality and because of their functional generality, standard languages are a key to the world of learning and higher culture: functional domains par excellence for standard language use (or, reversing the perspective, functional domains that cannot be accessed on the basis of dialect knowledge alone). Perhaps even more importantly, standard languages are supposed to contribute to political participation. The possibility of free communication is a feature of a democratic political organization, in the sense of the ideal 'herrschaftsfreie Kommunikation' as described by Jurgen Habermas. If then linguistic standardization contributes to mutual understanding and free communication, it is a factor of political emancipation just as it is a factor of social emancipation when it contributes to the spreading of culture and education. By contrast, if you believe in the beneficial effects of standardization, dialects are mere relics of an obscurantist social and political system that opposes democracy and emancipation. In a context of postmodern ideological debunking, such a positive conception of standardization is definitely suspect, but it is crucial for my line of argumentation that at least in the context in which it originated (that of the eighteenth-century Enlightenment), there was a genuine positive appraisal of standardization. To illustrate, let us have a look at some excerpts from reports presented to the revolutionary Convention in France. Barere (1794 [1975]) puts matters as follows. (1)

Citoyens, la langue d'un peuple libre doit etre une et la meme pour tous. (1794 [1975]: 297) [Citizens, the language of a free people has to be one and the same for all.]

(2)

Les lumieres portees a grands frais aux extremites de la France s'eteignent en y arrivant, puisque les lois n'y sont pas entendues. (1794 [1975]: 295)

Cultural models ofstandardization

277

[The lumieres, when they are brought with great difficulty to the remote corners of France, die out when they arrive there, because the laws are not understood.] (3)

Laisser les citoyens dans I'ignorance de la langue nationale, c'est trahir la patrie; c' est laisser le torrent des lumieres empoisonne ou obstrue dans son cours; c'est meconnaitre les bienfaits de l'imprimerie, car chaque imprimeur est un instituteur public de langue et de legislation. (1794 [1975]: 296-297) [To maintain the citizens in their ignorance of the national language is to betray the country. It permits the torrent of the lumieres to be poisoned or obstructed in its course. It means disavowing the blessings of the printing press, because all publishers are public teachers of the language and the legislation.]

(4)

Citoyens, les tyrans coalises on dit: l'ignorance fut toujours notre auxiliaire le plus puissant; maintenons l'ignorance; elle fait les fanatiques, elle multiplie les contre-revolutionnaires; faisons retrograder les Fran<;:ais vers la barbarie: servons-nous des peuples mal instruits ou de ceux qui parlent un idiome different de celui de l'instruction publique. (1794 [1975 ]: 291) [Citizens, the allied tyrants have said: ignorance has always been our most powerful helper. It creates fanatics, it breeds counterrevolutionaries. Let's make sure the French degrade into barbarity: let's take advantage of the badly educated peoples or of those that speak a language that is different from that of public education.]

(5)

Les habitants des campagnes n'entendent que le bas-breton; c'est avec cet instrument barbare de leurs pensees superstitieuses que les pretres et les intrigants les tiennent sous leur empire, dirigent leurs consciences et empechent les citoyens de connaitre les lois et d'aimer la Republique. Vos travaux leur sont inconnus, vos efforts pour leur affranchissement sont ignores. (1794 [1975]: 292-293) [The inhabitants of the countryside speak only the Breton dialect. It is with that instrument of their superstitious way of thinking that the priests and the plotters keep them under their thumb, control their minds, and prevent the citizens from knowing the laws of the Republic. Your works are unknown to them, your efforts to bring them liberty are ignored.]

The characteristics that we have attributed to standard languages (generality and communicative neutrality, emancipatory and participatory effects, opposition to obscurantism) can be easily identified in these fragments.

278

Meaning and culture

Fragment (1) expresses the generality and uniformity of the standard language. Fragments (2) and (3) stress the emancipatory function of knowledge of the standard: citizens who only know their dialect will not understand the laws of the Republic (the assumption being, of course, that these have a liberating effect), nor will they, more generally speaking, be able to profit from the benefits brought by the printed press. Fragments (4) and (5) associate dialects more directly with counter-revolutionary obscurantism: it is suggested that priests and 'tyrants' deliberately maintain ignorance by preventing the common people from acquiring the standard language. A similar pattern can be found in the following quotes from Gregoire (1794 [1975]), who actually presents an entire educational project to the Convention to 'abolish the dialects and generalize the use of the French language'. (His notion of 'dialect' actually includes not just the dialects of French, but also the different languages spoken in the territory of France, like German in the Alsace region, Flemish in the northern area, or Breton in Brittany.) (6)

Mais au mains on peut unifonner le langage d'une grande nation, de maniere que tous les citoyens qui la composent puissent sans obstacle se communiquer leurs pensees. Cette entreprise, qui ne fut pleinement executee chez aucun peuple, est digne du peuple fran~ais, qui centralise toutes les branches de I'organisation sociale et qui doit etre jaloux de consacrer au plut6t, dans une Republique une et indivisible, 1'usage unique et invariable de la langue et de la liberte. (1794 [1975]: 302) [But at least one can standardize the language of a great nation, to the extent that all its citizens can mutually communicate their thoughts unhindered. Such an enterprise, which no people has fully achieved as yet, is worthy of the French nation, which centralizes all aspects of the social organization and which must endeavor to endorse as soon as possible, in a Republic that is one and indivisible, the sole and invariable use of language and freedom.]

(7)

'Il y a dans notre langue, disait un royaliste, une hierarchie de style, parce que les mats sont classes comme les sujets dans une monarchie'. Cet aveu est un trait de lumiere pour quiconque reflechit. En appliquant I'inegalite des styles cl celle des conditions, on peut tirer des consequences qui prouvent I' importance de man projet dans une dernocratie. (1794 [1975]: 316)

['There is in our language, a certain royalist said, a hierarchy of styles, because the words are classified just like the citizens in a monarchy'. This confession constitutes a ray of insight for any thinking person. If we apply the inegality of the styles to the inegality of the conditions un-

Cultural models ofstandardization

279

der which people live, we may come to conclusions that prove the importance of my project (of linguistic standardization through an educationallanguage policy) in a democracy.] (8)

Tous les membres du souverain sont admissibles a toutes les places; il est a desirer que tous P4issent successivement les remplir, et retourner a leurs professions agricoles ou mecaniques. eet etat de choses nous presente l'alternative suiv~nte: si ces places sont occupees par des hommes incapables de s'enoncer, d'ecrire dans la langue nationale, les droits des citoyens seront-ils bien garantis par des actes dont la redaction presentera l'impropriete des termes, l'imprecision des idees, en un mot tous les symptomes de l'ignorance? Si au contraire cette ignorance exclut des places, bientot renaitra cette aristocratie qui jadis employait le patois pour montrer son affabilite protectrice a ceux qu'on appelait insolemment les petites gens. [... ] Ainsi l'ignorance de la langue compromettrait le bonheur social ou detruirait l'egalite. (1794 [1975]: 303) [All members of the sovereign people are eligible for all positions. It is desirable that all may successively fill these positions, and afterwards return to their agricultural or industrial professions. This state of affairs yields the following alternative. If the positions are taken up by men incapable of expressing themselves or of writing in the national language, will the rights of the citizens be safeguarded by laws that are characterized by improper choice of words, by imprecise ideas, in short by all symptoms of ignorance? If on the contrary this ignorance prevents people from taking up office, then soon enough we will witness the rebirth of that aristocracy that once used the dialects to demonstrate its affability with regard to those that it insolently named 'the small people'. [... ] Thus, ignorance of the language either compromises social happiness or destroys egality.]

Fragment (6) points to the communicative generality of the standard language: having a unitary language not only symbolizes the unity of the nation, but it also ensures that the citizens can freely communicate their thoughts. Fragment (7) symbolically links the absence of standardization to the prerevolutionary situation: the existence of hierarchically ordered varieties within the language mirrors the hierarchical organization of society. Fragment (8) aptly describes the politically emancipatory function of standardization. The egalitarian ideal implies that any citizen can take part in the government of the nation; in fact, the ideal would be that all citizens successively fulfill political functions and then return to their professional environment. However, in order to be able to fulfill these functions, a thorough knowledge of the common language is necessary. People should not be pre-

280

Meaning and culture

vented from taking up office by their ignorance of the language. Hence, an educational effort to ensure standardization is necessary: Gregoire is an ardent defender of the 'Ecole publique' as a standardizing force. In section 3, I will describe the transformations that the rationalist, Enlightenment ideal of standardization goes through in the course of the last two centuries. Even in its transformed shape, however, the positive evaluation of standardization refers to one or another of the features mentioned here: a neutrally mediating conmmnicative function, and an emancipatory and participatory effect, both of these supported by an educational system geared towards the spreading of the standard language. 2.2. The romantic model The romantic conception of standardization may be easily defined in contrast with the two dominating features of the rationalist model. First, as against the emancipatory and participatory goals of the enlightened view, a romantic view will tend to point out that standard languages are themselves instruments of oppression and exclusion. At this point, of course, the analysis of standardization takes the form of an ideological criticism: it will argue that the enlightened ideals are not often realized, and that, in fact, processes of standardization typically achieve the reverse of what they pretend to aim at. Although the term is not often used, this type of critical discourse boils down to a demonstration that linguistic standardization exemplifies what Horkheimer and Adomo (1947) called the 'Dialektik der Aufklarung' - the (negative) dialectic of Enlightenment. Horkheimer and Adomo argue that rationalist positions have a tendency to lead to their own dialectical counterpart (in the sense, for instance, in which a growing technical mastery of man over nature may lead to the destruction of the natural world). Now, if we look back at the three types of generality that standard languages are supposed to characterize, it is easy to see that the actual realization of the ideal may tend to contradict the ideal - which is then a case in point of the 'Dialektik der Aufklarung'. First, standard languages are supposed to be geographically neutral, but in actual practice, processes of standardization often have their startingpoint in a specific region that is economically, culturally, and/or politically dominant. For people in the other, outer provinces, then, the standard language is not an impartial medium, but it rather affirms the dominance of the leading province. Standard French, for instance, is not just an unbiased language coming out of the blue; it is the language of the upper and the middle

Cultural models ofstandardization

281

classes of Paris and the Ile-de-France, and it is associated with the role that the central province has played since the medieval era. Second, standard languages are supposed to be functionally general, but in actual practice, they are typically used in cultural, educational, scientific, administrative, and political contexts - at least in those circumstances in which a language community is not entirely standardized. Non-standard varieties may then naturally acquire additional, contrastive overtones. For one thing, if the standard language is the language of public life, the nonstandard varieties will be appreciated as the language associated with intimacy, familiarity, the personal rather than the public sphere. For another, if the standard language functions in typically intellectual contexts (education and science), non-standard varieties will be invested with emotional values. For speakers of a dialect, the dialect is often the language of the emotions, of spontaneity, of naturalness, in contrast with the official and educational language. Ironically, the functional generality of standard languages engenders a functional specialization, separating the public sphere from the personal, and the emotional sphere from the intellectual. Third, standard languages are supposed to be socially neutral, but in actual practice, they are typically the language of an elite. The link between an economical, cultural, or political elite and the standard language is in fact an inevitable side-effect of the functional generality of standard languages. If standard languages are typically used in cultural, educational, scientific, administrative, and political contexts, then those speakers of the language that act in these contexts will more easily learn the standard language or adopt it as their first language than speakers who remain foreign to these functions. The outsiders may then perceive the greater linguistic proficiency of the elite as a factor contributing to social exclusion. In Gregoire's view, knowledge of the standard language contributes to social mobility, but conversely, the real social distribution of standard language functions may turn the standard language into an instrument of discrimination. We can see, in other words, how the alleged generality of standard languages actually takes the form of a series of specializations. The process of standardization takes its starting-point in the language of specific regions, specific groups of speakers, specific domains and functions, and this largely inevitable fact may subvert the very ideal that standardization was supposed to serve. When that happens, the original ideal may be critically unmasked as an ideological pretence. Needless to say, this dialectical reversal may also affect the educational system. If the standard language is recognized as an instrument of oppres-

282

Meaning and culture

sion, discrimination, social exclusion, the educational system will likewise be rejected as contributing to such processes of social exclusion. Rather than seeing the school as an institution that spreads knowledge of the common language (and knowledge in general), creating possibilities for social mobility, it will then be pointed out that the educational system, relying on perhaps more than contributing to the knowledge of the language, favors those language users whose background makes them more familiar with the standard language, and thus reproduces rather than neutralizes social inequality. But why call this critical reversal of the appreciation of the standard language a 'romantic' model? Why not simply call it a 'realistic' or a 'critical' or an 'anti-ideological' one? The reason is that this critical stance is often (though not necessarily always) accompanied by a second feature, that may be contrasted with the second characteristic of the rationalist model. That is to say, we have just seen how a critical approach questions the emancipatory, participatory conception of the Enlightenment model. But what about the second feature? What about the communicative aspects of the rationalist model? We get a truly 'romantic' model of language variation when the critical attitude towards official standards is coupled with a view of language as expression rather than communication. According to the Enlightenment perspective, languages are means of communication, and a standard language is a superior communicative tool because it is functionally general and socially neutral. According to a romantic perspective, languages are primarily expressive rather than communicative. They express an identity, and they do so because they embody a particular conception of the world, a world view or 'Weltanschauung' in the sense of Herder. The link between this well-known romantic conception of the relationship between language and thought and the standardization debate will be clear. If languages or language varieties embody a specific identity, then a preference for one language or language variety rather than another implies that the specific identity of a specific group of people is neglected or denied. Not recognizing the language is not recognizing the language users. If some language varieties are relegated to second rate status through the existence of a standard variety, then the speakers of those language varieties are denied a fundamental right: the right to express themselves in their own language - the only language, in fact, that could do justice to their individual identity, according to the romantic conception of the relationship between language and identity.

Cultural models oJstandardization

283

A correlate of this position is the positive evaluation of variety. Whereas the rationalist approach cherished linguistic unifonnity as the symbolic expression of a free and open community in which all citizens have equal rights to speech, the romantic approach values diversity as a recognition of a fundamental respect for different identities. In short, a fully romantic view of language variation and linguistic standardization opposes the Enlightenment view of language as communication with a view of language as the expression of an individual identity. It opposes the emancipatory and participatory rationalist ideal with a critical view of standardization as a tool of discrimination and exclusion, and it opposes the positive appreciation of education as an instrument for the dissemination of linguistic knowledge with a fundamental distrust of schools as part of a system reproducing social inequality. In order to illustrate this model, I will not (as I did in the case of the rationalist model) use an historical example, but I would like to have a brief look at the current debate about linguistic genocide and the international position of English. In that interlinguistic fonn of variation, English replaces the standard language of intralinguistic variation, and minority languages threatened with disappearance replace the non-standard varieties. All the objections that a romantic approach would level against a dominating standard variety, could then be applied against the international domination of English. Consider, as an example, the following excerpts from an abstract of Skutnabb-Kangas (2000) (The abstract, by the author herself, may be found on the author's homepage). (9)

Indigenous peoples and minorities are the main bearers of linguistic and cultural diversity in the world - over 80% of the world's languages exist in one country only and the median language has no more than 5,000 speakers. Some of the direct main agents of linguistic (and cultural) genocide today are parts of what we call the consciousness industry: formal educational systems and the mass media. [... ] The book shows that the education of most minorities and indigenous peoples in the world is organized in ways which both counteract sound scientific principles and lead to the disappearance of linguistic and cultural diversity. [... ] Schools are every day committing linguistic genocide. [... ] They also do it by forcibly moving children from one group (indigenous or minority) to another group (the dominant group) through linguistic and cultural forced assimilation in schools. [... ] This inevitably includes a consideration of power relations. The book shows how the formal educational systems participate in maintaining and reproducing unequal power relations, here especially between linguistic minorities

284

Aifeaning and culture

and others, but also more generally, and how the ways of doing this have changed and are constantly changing, and how control and domination are resisted and alternatives are constantly created and negotiated, managed and controlled, and recreated. The deficiency-based models that are used in most minority education invalidate the linguistic and cultural capital of minority children and their parents and communities. They make the resources of dominated groups seem handicaps or deficiencies, instead of valued and validated non-material resources, or they render them invisible and therefore not possible to convert into material resources and positions of structural power. This happens just as much in global international relations and the Mcdonaldization of the world as it happens in ESL classrooms. Regardless of whether Skutnabb-Kangas is right or not (see Joseph forthcoming for a critical reaction), the components of the romantic approach are conspicuously present in her statement. First, the disappearance of languages and the disappearance of cultures are equated. The very notion of 'linguistic genocide' in fact invokes the extermination of an entire people (or at least culture) together with its language. As opposed to this process of forced assimilation and disappearance, different cultures have to be accepted as fundamentally equal, and diversity should be treasured as an end in itself. Second, the international dissemination of English does not lead to emancipation and participation, but rather serves purposes of international oppression, notably by multinational companies. And third, the text has explicit misgivings about the role schools play in this linguistic and cultural power play. 2.3. The rationalist and the romantic model and the origins of language

In the previous pages, I have not only sketched the rationalist and the romantic model of standardization, but I have also indicated that they exhibit a specific and narrow relationship. Not only is one the counterpart of the other, but there is a dialectical relationship between the two, in the sense that actual processes of standardization seem to be caught in a negative 'dialectic of Enlightenment' in the course of which the positive rationalist ideals tend to be subverted, thus giving way to a romantic, critical appreciation of the standardization process. This dialectical relationship, which is summarized in Table 1, does not however exhaust the links that exist between the two basic conceptions.

Cultural models ofstandardization

285

Table I. The rationalist and the romantic models of standardization THE RATIONALIST MODEL

THE ROMANTIC MODEL

language as a medium of communication

language as a medium of expression

conception of standardi- a democratic ideal: zation: standard language as a neutral medium of social participation

anti-ideological criticism: standard language as a medium of social exclusion

language variation as an impediment to emancipation

language variation as expressing different identities

linguistic-philosophical basis:

conception of language variation:

In this section, I will have a brief look at the theoretical linguistic background of the basic models: is there a specific conception of language that goes hand in hand with either of the perspectives? I will argue that eighteenth-century theories about the origins of language complement the picture painted so far. It was suggested above that the romantic and the rationalist approach differ in their theoretical conception of language: the former starts from an expressive conception of language and the latter takes its startingpoint in a communicative conception. The distinction between both perspectives can get more relief if we have a brief look at some of the theories on the origins of language that were formulated in the eighteenth century - in the period, that is, in which we situate the intellectual sources of the rationalist and the romantic conception of language variation. The expressive conception of the origins of language may be illustrated on the basis of Herder's views, while Condillac may serve as an illustration of the conununicative VIew. Herder's conception of language as the expression of an inner view ofthe world shapes the way in which he describes the birth of language: language arises when man expresses his understanding of the world. In the following passage, Herder explains how man, in giving names to the animals (in this case, a sheep) identifies and expresses a specific feature (in this case, the sheep's bleating). (10)

Weiss, sanft wollicht - seine besonnen sich iibende Seele sucht ein Merkmal, - das Schaaf blocket! sie hat Merkmal gefunden. Der innere Sinn wiirket. Dies Bl6cken, das ihr am starksten Eindruck macht, das sich von alien andern Eigenschaften des Beschauens ul1d Betastens los-

286

Meaning and culture riss, hervorsprang, am tiefsten eindrang, bleibt ihr. Das Schaaf kommt wieder. Weiss, sanft wollicht - sie sieht, tastet, besinnet sich, sucht Merkmal - es b16ckt, und nun erkennet sies wieder! 'Ha! du bist das Blockende!' fuhlt sie innerlich, sie hat es Menschlich erkannt, da sies deutlich, das ist, mit einem Merkmal, erkennet und nennet (1772 [1978]: 33)

[White, soft, woolly - its cautiously exercizing soul looks for a characteristic feature. The sheep bleats! It has found a feature. The inner sense is working. This bleating. which impresses it to the utmost, which detaches itself from all other visual or tactile features, which springs forth, which makes an impression, stays with it. The sheep comes back. White, soft, woolly - the soul sees, feels, reflects - the sheep bleats, and now it recognizes it! 'Ha! You are the bleating one', it feels within itself, it has recognized it in human fashion, because it has recognized and named the sheep clearly, that is, by means of a characteristic feature.] Herder's story may be contrasted with Condillac's view about the origins of language. Whereas Herder looks for the origins of language in a cognitive, epistemological urge to identify things (and more specifically, to identify them in a specific way that reflects man's understanding of the things in question), Condillac points to a communicative situation. In his story about two children, he describes how a cry for help by one is recognized by the other, and is gradually conventionalized as a specific sign. (11)

Celui qui souffroit parce qu'il etoit prive d'un objet que ses passions lui rendoient necessaire, ne s'en tenoit pas a pousser des cris: il faisoit des efforts pout l'obtenir, il agitoit sa tete. ses bras. et toutes les parties de son corps. L'autre. emu a ce spectacle. fixoit les yeux sur le meme objet; et [... ] il souffroit de voir souffrir ce miserable. Des ce moment il se sent interesse a le soulager, et il obeit a cette impression, autant qu'il est en son pouvoir. Ainsi, par le seul instinct, ces hommes se demandoient et se pretoient de secours [... ]. Cependant les memes circonstances ne purent se repeter souvent, qu'ils n'accoutumassent enfin a attacher aux cris des passions et aux differentes actions du corps, des perceptions qui y etoient exprimees d'une maniere si sensibles (1746 [1973]: 194-195).

[A person who suffered from being deprived of an object that his passions made necessary for him, did not restrict himself to shouting: he made an effort to obtain it, shook his head and arms and body parts. The other person. moved by this spectacle, fixed his gaze on the same object, and suffered from seeing the miserable other person suffer. From

Cultural models ofstandardization

287

this moment on, he grows interested in coming to his relief, and he obeys this sentiment to the best of his powers. In this way, by instinct alone, men mutually ask and give help. [... ] But such events could not repeat themselves often before people got used to attaching to the passionate cries and bodily actions the views that they expressed so tangibly.] The difference between Herder and Condillac correlates with a difference in their view of the relationship between language and culture. Unlike Condillac, Herder does not think it possible to explain the origins of human language from instinctive emotional cries. Understanding needs to be present before human, conscious language can arise. In Condillac's account understanding or 'reflection' takes shape gradually, through the use of signs in a communicative situation. For Herder, the birth of language presupposes the existence of a human culture, a culture that is being expressed in the language. From Condillac's point of view, human culture develops from instinctive beginnings through the use of signs and the development of language. (Compare Sapir 1907 for a more elaborated view.) If it is indeed correct that the intellectual origins of our cultural models of linguistic standardization have to be sought in the eighteenth century, quotations (l 0) and (ll) suggest that there was a link with the linguistic theorizing of that period. This recognition opens the way towards the broader historical study of the relationship between theories in linguistics and the philosophy of language on the one hand, and language policies on the other. To what extent have theories about the origin, nature, structure of language influenced the way in which people (and governments) have thought about matters of language variation and linguistic standardization? The existing literature offers many points of departure but mostly concentrates on standardization processes and language policies within a single language or nation. One interesting way to use the cultural models identified above, could be to see whether they can be used as the basis for a more synthetic view of these historical links between linguistic theory and standardization.

3.

The historical transformation of the models

The present section will have a look at various distinctive moments in the development of the competing models, in particular charting the transformations that they go through in the nineteenth and the twentieth centuries. First, I will argue that the nationalist model of standardization that rose to promi-

288

Meaning and culture

nence in the nineteenth century constitutes a specific blend of the rationalist and the romantic model. Further, I will have a look at the way in which our contemporary postmodern awareness influences the competition between the rationalist and the romantic model. It is an interesting question, by the way, whether the models ever occur in their purest form. Even in the examples from the French revolutionary period, a link with patriotic nationalism is present. If this is indeed the case, the models presented in the previous section are to be seen as analytic reference points, as 'idealized cognitive models' in the sense of Cognitive Linguistics. 3. 1. The nineteenth century and the nationalist model Both the rationalist and the romantic model have a problem with the level at which they should be situated. If the rationalist model is carried to its extreme, it implies the necessity of a universal, international language. If the driving force behind standardization is maximizing mutual communication, then a universal language that transcends all existing language variation is to be recommended: the neutralization of interlinguistic variation complements the neutralization of intralinguistic standardization. And of course, the ideal of a universal, ideal language (Esperanto, Volapilk and the like) is precisely the historical realization of this consequent interpretation of the rationalist approach. In actual practice, however, Esperantist movements and the like remained marginal. The real level at which standardization processes took place, lay at a lower level - that of the nation. Starting from the Enlightenment model, there is a simple logic to this (which can, in fact, be identified in the quotations from Barere and Gregoire that we discussed): if standardization aims at democratic, political participation, then obviously the nation, as the ideal form of political organization, becomes the locus of standardization processes and the educational efforts supporting them. In itself, then, a link between nationalism and the rationalist view of standardization cannot come as a surprise. Linguistic standardization is primarily standardization within a nation, because it is within the nation that the processes of political decision making take place that linguistic standardization is supposed to contribute to. A terminological clarification may be useful at this point. 'Nationalism' is the political ideology in which a state, as a political organization, derives its political legitimacy from its people, rather than from tradition, divine right, or the like. A state that lives up to this requirement is a nation. Nationalism, in other words, claims that any state should be a nation. The national-

Cultural models ofstandardization

289

ist relationship between the people and the state may be conceived of in two different ways: according to a distinction that is customary in the literature on nationalism, we may make a distinction between civic nationalism and identity nationalism. On the one hand, civic nationalism is the conception of nationalism in which the nation derives its legitimacy from the active participation of its citizens, through a system of political representation. This is the liberal, rationalist conception of nationalism. On the other hand, identity nationalism is the conception of nationalism in which the nation derives its political legitimacy from the cultural identity of the people. This is the romantic conception of nationalism. 'Nationalism' also refers to the claim and the efforts of a particular group to become a nation. Existing states are not necessarily nations according to the nationalist view: either because they do not achieve democratic legitimacy (the liberal point of view), or because they do not recognize the cultural identity of certain groups (the romantic point of view). Historically speaking, then, 'nationalist movements' may be either movements trying to establish a liberal democracy, or movements claiming independence for a specific group or region. (In contemporary usage, though, the focus tends to lie more on the latter type.) The link between nationalism and language that we described above clearly involves the liberal, rationalist version of nationalism: if the nation derives its legitimacy from the active participation of its citizens, then maximizing mutual communication through standardization is an instrument of participation. But if we turn to identity nationalism, nationalism has a similar, and maybe even stronger link with a romantic conception of language. Whereas the rationalist perspective contains a tendency towards universality, the romantic perspective has a tendency towards individuality. If carried to its extreme, the romantic conception of language variation implies that each person may have his or her own language. Just like the rationalist perspective tends to maximize communicability, the romantic perspective tends to maximize individual variation. Again, in actual practice, this is an extreme position that can hardly be realized as such. Except perhaps in the romantic admiration for the 'individual voice of the poet' and the like, the romantic conception deals with the language of groups rather than with the language of individuals. The identity that is expressed by the language is the identity of a community, and the community is a nation when it acquires political autonomy. Hence the well-known romantic link between nationalism and language: see, among many others, Deprez and Vos (1998). On the one hand, language correlates with identity according to the romantic model,

290

Aleaning and culture

and on the other, nations may derive their legitimacy from the cultural identity of the people (which is not to say that all nationalism is linguistic nationalism: as is well known, the sense of identity may come from many other sources, like religion or ethnicity). From two different angles, then, nationalism links up with language, and this recognition may be linked to the distinction between two basic types of nationalism that is often made in political theory. On the one hand, civic nationalism is the conception of nationalism in which the nation derives its legitimacy from the active participation of its citizens, through a system of political representation. In such a liberal, rationalist conception, the common language is the medium of participation. On the other hand, identity nationalism is the conception of nationalism in which the nation derives its political legitimacy from the cultural identity of the people, and language is one of the factors establishing such identity. The actual alliance between both forms of reasoning may be briefly illustrated by the following quotes from Verlooy (1788). A Dutch-speaking lawyer in Brussels, which was then under Austrian rule, Verlooy argues against the growing use of French in public life and in favor of the use of the native Dutch tongue. In 1789, Verlooy played a role in the 'Brabantse Omwenteling', an (ineffective) insurrection against the Austrians. His pamphlet of 1788 may be read as the intellectual basis of his nationalist stance of the next year. But what is the role attributed to language in Verlooy's nationalism? (12)

Het is zonder twyffel een goed voor eenigelyk wel ter tael en ter spraek te zyn, en zyne redens vaerdig en onbelemmert te voeren. Doch hier toe is een' zekere frankheyd noodig. Maer, gelyk by ons gezien en geplogen is, wanneer wHen wy frank zyn in die vremde tael? (1788 [1979]: 58)

[Without any doubt it is good for any person to be able to speak fluently, and to engage in conversation freely. But to achieve this a certain candor is necessary. However, as can be observed in our case, when will we obtain such candor in this foreign language?] (13)

Door ons frans schynen wy van die middelbare geleertheyd en borgerIyke wysheyd af geheeI het gemeyn, onze bestgemoedde en weetgirige borgers, ambachtslieden, akkennans, en onze vrouwen: die 't frans teenemael niet, of ten minsten zoo verre niet en weten, dat-ze 't met vennaek of zonder moyelykheid konnen lezen: die daer door als als gedoemt schynen tot een' gezogte onwetendheyd. (1788 [1979]: 49)

Cultural models ofstandardization

291

[By speaking French, we separate from this common knowledge and this civic wisdom all the common people, our well-humored and inquisitive townsmen, craftsmen, farmers, and our women: who do not know French, or at least not well enough to read it easily and efficiently, and who therefore seem to be condemned to ignorance.] (14)

Voor het vaderlanderschap eener natie is zeer dienstig zoo veel eygen en bezonder te hebben als mogelyk is (... ) en zelfs hoe meer een' zaek uytwendigs heeft, gelyk de taeL dragten, toneelen, godsdienst, zeker plechten: hoe meer zy de gemoederen van 't volk zal aentrekken. [... ] Waerom werken wy dan om zoo bekwamen band van vaderlanderschap, de moederlyke tael, te bannen? (1788 [1979]: 59-60) [For a feeling of national identity within a nation, it is useful to have as many common and specific features as possible, and these features will more readily attract the hearts of the people to the extent that they can be externally observed, like the language, the attire, the theater and the public entertainments, the religion. Why then do we endeavor to discard our mother tongue, which constitutes such a strong tie of patriotism?]

Quotation (12) emphasizes the individual and emancipatory perspective: it is important for people to be able to express themselves freely, and this can only be guaranteed in their mother tongue. In the same vein, quotation (13) stresses the importance ofa common language for an open communication within a given society and for the dissemination of knowledge: the further use of French would engender an undesirable rupture between the middle classes and the lower classes. By contrast, quotation (14) stresses the importance of a common identity for nation-building. Both rationalist and romantic themes, in other words, may appear in the discourse of proponents of nationalist movements. With the development of the national movements in the nineteenth century, though, the nationalist emphasis tended to fall more and more on the romantic notion of national identity. Minorities aspiring towards independence naively assume or explicitly construct an identity, and nation states may blatantly enforce a common identity, linguistic or otherwise. These processes are well known from the nationalism literature (in the line of Hobsbawm, Anderson, Gellner, Smith). For our present linguistic purposes, the crucial point is to see that this romantic nationalism reveals the paradoxes of the romantic cultural model that we identified above. The transition from the romantic model as described earlier to the nationalist model constitutes so to speak a 'Dialektik der Romantik' that parallels the 'Dialektik der Aufk-

292

Meaning and culture

Hirung', i.e. an almost natural process through which the original romantic model becomes subverted and contradicts and least some of its own startingpoints. The paradox of the romantically inspired nationalist model, in fact, is this. On the one hand, it claims recognition of diversity, equal rights, political independence for one (linguistic) group with regard to other groups. On the other, it has to assume an internal homogeneity within that group, for the simple reason that within the romantic logic, it is the identity of the group that legitimatizes the claim for recognition. And so, the identity may have to be imposed or constructed, and dissident voices within the group may have to be stifled. The romantic model, then, is no less prone to contradictory developments than the rationalist one. In the linguistic debate, the specific form of the romantically nationalist position is a concern for the purity of the language. Defending the common language against foreign influences (loan words, basically) is at the same time a defense of the cultural identity of the people. In the nationalist subversion of the initial romantic model identities are not only expressed, but they are also made permanent. Again, the link between purism and nationalism is well-known, and there is an extended literature on purism. What I would like to stress, in this respect, is less the phenomenon as such, but rather how it fits into the overall pattern that defines the paradoxicallogic of the rationalist and the romantic model of language variation. This paradoxical logic, to sum up, resides in the following points. First, although the basic models are opposites, they find a common ground in the notion of nationalism. Because the rationalist model cannot easily realize its extreme universalist claims, and because the romantic model cannot easily realize its radical individualist claims, both models meet on a middle ground where groups of people claim political identity and independence. Second, this coalescence of the models does not annihilate the tensions that exist between them: the history of the past two centuries brims with examples of conflicts between a more rationalist Staatsnationalismus ( civic nationalism at the level of the nation-state) and a more romantic Volksnationalismus (ethnic or cultural identity nationalism). Third, in addition to the tensions between the models, we have to take into account tensions within each model: the rationalist model is subject to the danger of a discriminatory 'Dialektik der AufkHirung', and the romantic assumption of internal homogeneity may likewise have oppressive side-effects.

Cultural models ofstandardization

293

3.2. The twentieth century and the postmodern awareness Living as we do in the aftermath of the nationalist era, we should complete our overview of the historical transformations of the cultural models of language variation by charting what changes are brought to the debate by our post-nationalist environment. The current situation can best be characterized by two overlapping developments: globalization and postmodernism. Globalization is economic and political (to the extent that the growing importance of international organizations diminishes the older importance of the nation state). But it is also linguistic: the international spread of English almost realizes the old rationalist's dream of a universal language. The postmodern awareness, on the other hand, resides in two features. First, the so-called 'disappearance of the Great Narratives' signals a weakening of the older patterns of interpretation. There is a great deal of suspicion with regard to the rationalist model of a smooth, emancipatory progress as well as (and perhaps more dominantly so) with regard to the nationalist model. Postmodern thinking is the self-consciousness of the late twentieth century: progress is not automatic, and nationalism is dangerous. This critical attitude entails a second feature: if the old models are no longer self-evident, a dehierarchicahzation and informalization occurs. If, for instance, the original hierarchical ordering of high culture and low culture is rejected as part of the old models of interpretation, then popular culture may claim equal rights with regard to high culture. These two features imply that postmodernism is to a large extent a renewed form of the original romantic attitude: it renews the critical, 'countercultural' attitude with regard to the official stories, and it revives the claims for diversity. The interesting question from our point of view is whether these changes lead to a fundamental transformation of the cultural models that are used to discuss language variation. Basically, there is a growing emphasis on the international relationship between languages rather than the national relationship between language varieties. The initial models of standardization are essentially models of standard languages in comparison with dialects or other varieties of the same language. In the nationalist era, the debate sometimes involves national languages as opposed to minority languages, but it is only in our days that the debate concentrates on the international relationship between different languages, viz. the relationship between English as a world language in comparison with local, possibly endangered languages.

294

Meaning and culture

Now, to the extent that the position of global English is at stake, the old opposition between rationalist and romantic attitudes receives a new impetus. I have shown above how Skutnabb-Kangas's argumentation about the treatment of minority languages is largely situated within what I would call a 'romantic' frame. At the same time, it is not difficult to see what form the basic pattern of a rationalist reply with regard to the position taken by Skutnabb-Kangas would probably take. First, against the identification of language and culture, the rationalist could point to cases where the same language is unproblematically shared by different cultures, or conversely, where the same culture unites people with different languages. Second, against the allegation that the international dissemination of English is discriminatory, the rationalist might want to stress the actual emancipatory effects of a knowledge of English. If English is indeed the key to international communication (and if, indeed, acquiring English is possible for all), then it can only be welcomed that more and more people are able to participate in that kind of communication. At the same time, though, the contemporary discussions seem to lead to the development of a model based on a functional differentiation between the varieties involved - an 'and/and' -model rather than an 'either/or' -modeL so to speak. In the discussion about the international situation in particular, there is a growing recognition that multilingualism is a natural situation. Interestingly, the shift towards multilingualism as a (so to speak) dialectic synthesis of the opposite forces may be derived from the rationalist as well as from the romantic model. In the previous section, we saw that the shift towards nationalism fitted into the logic of both basic models, if account was taken of the 'problem of levels'. At this point, we can see in a similar way that a new focus on multilingualism fits in with both models. On the one hand, a multilingual solution seems to presuppose some form of functional distribution: one language is used for a specific set of circumstances, and the other for another set of circumstances. Such a diglossic or polyglossic situational specialization is not incompatible with the original rationalist model. After all, the rationalist model is motivated by a desire to assure maximal democratic participation in what are sometimes called 'secondary domains' of social life: specific, public domains of experience, to begin with higher education and political life. For the 'primary domains', beginning with the more private aspects of life, the existence of less uniform, more local language varieties does not fundamentally contradict the ideological basis of the model.

Cultural models o.fstandardization

295

On the other hand, the postmodern twist of the romantic model entails a new attitude towards the question of personal identity. It is often said, in fact, that one of the hallmarks of the postmodern mentality is the fragmentation of identity. People no longer experience a single personal identity, but they exhibit a number of different, possibly shifting identities, of a professional, social, ethnic, cultural nature. Different languages may then, following the original 'expressive' logic of the romantic attitude, express this fragmentation (or perhaps rather multiplication) of identities. Although multilingualism would thus appear to provide a possible synthesis of the initial models, the multilingual solution does not, however, completely remove the tensions. Just like the nationalist convergence of the models in the nineteenth century engendered a tension between nation states and minorities, the multilingual convergence entails tensions about the exact functional and situational distribution of the language varieties. A simple case in point is the current reform of higher education in Europe. The imposition of a uniform Bachelor/Master system is intended, among other things, to stimulate student mobility, and this in turn increases the pressure to introduce English as a language of instruction at least at the Master level. But many, of course, are reluctant to accept such a functional restriction on the original national language. Likewise, internal tensions within each model continue to crop up. For instance, those welcoming international English as an opportunity for all to participate equally in a global culture may easily be blind to the fact that access to English is not equal for all. And the champions of linguistic diversity may readily overlook the fact that the people they purport to defend often prefer the educational and professional opportunities provided by the non-native language (as appears to be the case, for instance, in most African countries that are ex-colonies of Britain). All in all, then, we may sum up the present situation in a way that largely parallels the summary at the end of section 3.2. On the one hand, just like nationalism allowed for a coalescence between the rationalist and the romantic model, multilingualism may constitute a point of convergence for the post-nationalist manifestations of the models. On the other, just like nationalism did not abolish the basic tension between the models nor the internal tensions within the models, a multilingual model does not cancel out the tension between rationalistically seeing global English as a communicative and educational opportunity and romantically seeing it as a threat to diversity and local identity. At the same time, though, we should keep in mind that the multilingual model is only beginning to emerge, and that the posi-

296

Meaning and culture

tions in the current debate have not yet crystallized as much as they have in the older nationalism debate. 3.3. Summarizing the models and their transformations We can now identify the pattern that emerges from the discussion in the previous pages. Referring to the philosophical and cultural climate of the eighteenth century, we have distinguished between a rationalist and a romantic basic model of linguistic standardization. Starting from a communicative conception of language, the fonner stresses the emancipatory function of a common language as an instrument of political and educational participation. Starting, on the other hand, from an expressive conception of language, the romantic model stresses how the imposition of a standard language may discriminate specific cultural identities. There is, then, a tension between the models to the extent that they are each other's counterpart. That tension is enhanced by the 'Dialektik der AufkHirung', the mechanism, through which the implementation of the rationalist ideals may generate its own opposite. Table 2. Cultural models of standardization and their historical transformations 18TH CENTURY:

19TH CENTURY:

LATE 20TH CENTURY:

the basic models

nationalism

postmodernism

the rationalist the common lanthe nation as the basis of a liberal position: guage as an instrument of political and democracy educational participation

global English as an opportunity / multilingualism as functional specialization

the romantic position:

the standard language as a discrimination of specific identities

global English as a the nation as a focus of cultural or threat / multilingualism as the expresethnic identity sion of fragmented, postmodern identities

the tension between both positions:

opposition between the models. enhanced by the 'Dialektik der AufkHirung'

conflict between nation states and ethnic/cultural groups

what is the exact shape of the functional specialization?

Cultural models ofstandardization

297

In the successive transformations that the models undergo in the nineteenth and the twentieth centuries, we have not only identified variants of the two models as such, but we have also indicated how the tension that exists between them reappears in different forms. Table 2 charts the various positions. The nineteenth century witnesses a partial convergence of the models round the concept of nationalism. Both basic models do, in fact, have a problem of levels: how general should the communicative community of the rationalist model be? And how specific can the identity get that is expressed according to the romantic model? Rather than going for (respectively) universalist or individualist extremes, the nationalist interpretation of the models focuses on an intermediate level. From the rationalist point of view, the nation is, by definition, the level at which democratic participation should be guaranteed. From the romantic point of view, the identity to be expressed is a common identity, the identity of the group that constitutes a nation. These two forms of nationalism are well known, of course: liberal nationalism (civic nationalism, Staatsnationalismus) sees the nation as the basis of a liberal democracy, and romantic nationalism (identity nationalism, Volksnationalismus) sees the nation as a focus of cultural or ethnic identity. The tensions between both are equally well-known: the liberal nation state need not coincide with a single cultural or ethnic group, and these groups may then claim a status as an independent nation. Needless to say, linguistic differences may play a crucial role in the definition of the groups involved. Note, in addition, that a federal political model constitutes a rather frequent compromise for the tension between claims of nation states and those of nationalist groups. The late twentieth century is characterized by a process of political and economic globalization that has its attitudinal counterpart in a postmodem view of the world, and that has its linguistic counterpart in the global spread of English. The debate accordingly shifts towards the position of English vis a vis local, possibly endangered languages. Although the process has not perhaps reached its culmination yet, the debate seems to find a new focus in the concept of multilingualism as a way of reconciling the different positions. In the same way in which the nationalist focus of the nineteenth century followed logically from the initial models (through the problem of levels), the focus on multilingualism can be equally motivated on the basis of both models. For the rationalist model, multilingualism involves an acceptable functional specialization of different languages: if language is an instrument of communication, different communicative situations may require

298

Meaning and culture

different languages. For the romantic model, multilingualism correlates with the fragmented identity of the postmodern individual: if people may so to speak have different identities, they may use different languages to express those identities. However, the shift towards multilingualism does not eliminate the tension: the exact functional specialization of the languages involved remains a cause for conflict.

4. Applying the models

The rationalist, romantic, nationalist, postmodern models that we have identified may basically be used in two different ways in linguistics. On the one hand, they may be used as the basis for an analysis of any discourse concerning problems of standardization. The previous pages have already given some examples of how the cultural models can be used in the analysis of the logic of standardization debates. In general, such an analysis identifies the components of a given discourse as expressing, amending, opposing aspects of the basic models. We could go one step further by identifying standard patterns of statements and replies that may be expected in this kind of discussion, but it is not my intention to do this systematically here. On the other hand, language attitudes based on one or the other of the models may influence the actual development of standardization processes. In the following sections, I will illustrate this second domain of application. I will do so by having a closer look at the work that I have been doing with my research team on the internal variation of contemporary Dutch. In section 4.1, I will summarize the main results. (This section could be skipped by readers less interested in the technicalities of the project.) In section 4.2, I will discuss the way in which the cultural models of standardization may be brought to bear on the findings of the project. 4.1. Contemporary Dutch as a pluricentric language Dutch basically comes in two varieties: Dutch as used in The Netherlands, and Dutch as used in the Flanders region of Belgium (sometimes referred to as 'Flemish'). The situation of the standard language in both countries is somewhat different. In Flanders, the standardization process that started off (as in most European countries) in the Early Modem Period was slowed down as a result of Flanders' political separation from The Netherlands during the Eighty Years' War. Standard Dutch developed in The Nether-

Cultural models ojstandardization

299

lands in the course of the seventeenth century, but as Flanders was politically separated from The Netherlands, remaining under foreign (Spanish or Austrian) rule, it did not link up with this process of standardization. Rather, French was used more and more as the language of government and high culture, a practice that received an important impulse after the birth of the Belgian state in 1830. Dutch then survived basically in the form ofa variety of Flemish dialects. However, as a result of a social and political struggle for the emancipation of Flanders and the Flemish-speaking part of the Belgian population, Dutch again gained ground as a standard language (the language of learning, government, and high culture) in Flanders. This process started somewhat hesitantly in the late nineteenth century as a typically romantic movement, gained momentum during the first half of the twentieth century, and finally made a major leap after World War 11 and during the booming 1960s. Still, most linguists agree that the standardization process has not yet reached its final point, or at least, that the level of standardization has not reached the same height as in The Netherlands. The latter observation is the starting-point for our research: can we quantify the relationship between Belgian Dutch and Netherlandic Dutch (and the internal stratification of both varieties)? Can we calculate how close or how distant both varieties of Dutch are with regard to each other? More specifically (given that our research team is primarily interested in lexical variation), how does one quantify lexical convergence or divergence between two language varieties? In Geeraerts, Grondelaers and Speelman (1999), a measure of lexical overlap was developed, based on the notions onomasiological prOfile and unUormity. Table 3. Onomasiological profiles for shirt

hemd overhemd shirt

Belgian Dutch 1990

Netherlandic Dutch 1990

31 %

17 % 46% 37%

69 %

0%

The onomasiological profile of a concept in a particular source is the set of synonymous names for that concept in that particular source, differentiated by relative frequency. Table 3 contains the onomasiological profiles for OVERHEMD 'shirt' in the Belgian and the Netherlandic 1990-database.

300

Meaning and culture

Uniformity is a measure for the correspondence between two onomasiological profiles. Our computation of uniformity has its starting-point in the idea that a common language norm triggers unifornl linguistic behavior. In its most extreme form, lexical uniformity in the naming of a concept obtains when two language varieties have an identical name for that concept, or several names with identical frequencies in the two varieties. Much more frequent than these examples of 'ideal' uniformity, however, are such partial correspondences as illustrated in Table 3. Let us, for the sake of illustration, assume that the relative frequencies in Table 3 represent 100 actual naming instances in each of both profiles, rather than percentages. The partial overlap between the profiles in Table 3 is quantified by counting the naming instances for which there is a counterpart in the other profile. In the ideal scenario outlined above, each of the 100 naming events in each of both profiles has its counterpart in the other profile, yielding a maximal uniformity of 100 %. In Table 3, however, 14 instances of hemd in B90 have no counterpart in N90, 23 Belgian overhemd's have no Netherlandic counterpart, and there are no Belgian counterparts for the 37 Netherlandic shirts. On the grand total of 200 naming events in the two profiles, only 200-(14+23+37) = 126 instances have counterparts in the other profile, which yields a uniformity of 126/2=63%. For the sake of quantitative convenience, it should be noticed that this percentage equates the sum of the smallest relative frequency for each alternative term, i.e. 17+46+0=63%. If more than one concept is investigated, a uniformity index U is defined as the average of the uniformity indexes of the separate concepts, whereas uniformity index U' is defined as a weighted average, in which the relative frequency of each concept in the investigated samples is taken into account. In the present context, we will focus exclusively on the weighted uniformity U', in which high frequency concepts have a more outspoken impact on the overall uniformity. The empirical foundation of the research project consisted of 40.000 observations of language use. We collected the different names (and their frequencies) used to denote thirty concepts, fifteen from the field of clothing terminology, and fifteen from the field of football (i.e. soccer) terminology. The resulting database allows us, for instance, to calculate the proportion in Belgian and Netherlandic sources of the term buitenspel 'offside' and the loanword offside for the concept OFFSIDE; in the case of the concept JURK 'dress', we can determine whether the lexical choices involve a preference for either jurk, japon, or kleed. The core of the observed material consisted of magazine and newspaper material recorded in 1990.

Cultural models ofstandardization

301

This core was extended in two ways. In the first place, similar material was collected for 1950 an 1970, which enabled us to carry out a 'real time'investigation of lexical convergence or divergence processes. In addition, the stratification of language use was taken into account. Between standard and dialect, there are a number of 'strata' on which register differences may cooccur with an increasing geographical specialization. For an investigation of the relationship between Belgian and Netherlandic Dutch, these strata - viz. the regionally colored informal variants of the standard language - are extremely relevant: it can be expected that the linguistic differences between Belgium and The Netherlands will increase on this regiolectic level. This intermediate level between dialect and written standard language was represented by the clothing terms we collected from labels and price tags in shop windows in two Belgian (Leuven and Kortrijk) and two Netherlandic towns (Leiden and Maastricht). The intended audience of this form of communication is more restricted than the national or binational audience which is the target of the magazines from which the core material was selected. The fact that we are dealing with written language in a semi-formal situation, on the other hand, ensures that we steer clear of the purely dialectal pole of the stratificational continuum. Table 4. U' values comparing Belgium and The Netherlands (1950-1970-1990)

and comparing written data from magazines and newspapers with local shop window data (1990) B50 / N50: B70 / N70: B90 / N90: B90 / Bsw 90: N90 / Nsw 90:

69,84 74,59 81,70 45,90 67,75

Given this database, what can we expect to find with regard to the relationship between the various language varieties? With respect to the status and the development of Belgian Dutch, two uncontroversial hypotheses can be found in the linguistic literature. First, there is an expectation of diachronic convergence between Belgian and Netherlandic Dutch. The standardization process in Flanders is characterized by an explicit normative orientation towards Netherlandic Dutch: the standardization of Belgian Dutch took the form of an adoption of the Dutch standard language that existed already in The Netherlands. In addition, the unfinished character of the standardization of Belgian Dutch is believed to manifest itself in a larger

302

Meaning and culture

synchronic distance between local and national language in Belgium than in The Netherlands; even to the untrained observer, it is obvious that the differences between regional and supraregional registers are much larger in Belgium than in The Netherlands. The diachronic and the synchronic hypothesis may now be made operational in terms of uniformity values as defined above. Diachronically, convergence and divergence can be quantified as increasing or decreasing uniformity. Synchronically, the larger distance between national and local language we expect in Belgian Dutch, will manifest itself in a smaller uniformity between magazine and shop window material in Belgian Dutch than in Netherlandic Dutch. Table 4 contains the relevant results. B50 stands for 'Belgian data from 1950', N50 stands for 'data from The Netherlands from 1950'. Bsw 90 refers to the shop window materials in Belgium, in contrast with B90, which stands for the data taken from magazines and newspapers. The data in Table 4 unambiguously confirn1 the diachronic as well as the synchronic hypothesis. Diachronically, the increase in uniformity between Belgian and Netherlandic Dutch suggests an evident lexical convergence between both varieties: U'(B50,N50) 69.84

< <

U'(B70,N70) 74.59

<

<

U'(B90,N90) 81.70

Synchronically, the delayed or unfinished standardization of Belgian Dutch manifests itself in a distinctly lower unifonnity between the Belgian magazine and shop window data than between the Netherlandic magazine and shop window material: U'(B90, B sw 90) 45.90

< <

U'(N90, Nsw 90) 67.75

4.2. Cultural models and the evolution of Belgian Dutch The overall situation of the different varieties of Dutch may be summarized as in Figure 1. The left-hand side of the figure represents Netherlandic Dutch, whereas the right-hand side represents Belgian Dutch ('Flemish'). The vertical dimension represents the internal stratification of both national varieties, distinguishing between the register of standard speech, the colloquial register, and the local dialects. Distances in the figure symbolize linguistic distances. In the upper register, both national varieties are close to each other, but the distances increase in the lower registers. Crucially, the

Cultural models ofstandardization

303

colloquial Belgian Dutch (colloquial Flemish) is much further apart from the high register in Belgium than the distance that exists between colloquial speech and polite speech in The Netherlands. So how would Belgian Dutch evolve? Basically, there are three possibilities. First, the present situation is merely a transitory one in a gradual process of convergence and standardization. We have seen that there is some evidence for such a converging development, at least in the upper stratum of the language, and it might be assumed that this evolution will continue in the lower strata. In that case, colloquial Flemish is likely to move upwards in the figure, perhaps until it reaches a position that is similar to that of Netherlandic Dutch. Second, the process of convergence might be reversed. Typical features of colloquial Flemish might be incorporated into the standard register used in Belgium, and this informalization of the standard language may in turn lead to a growing distance in the highest stratum between Netherlandic Dutch and Belgian Dutch. And third, the situation might simply be stable. Netherlandic 5tandard D.1tch

Belgian standard

D.1tch

/

Netherlandic colloCPJial D.1tch

colloquial Flemish

dialects

\

dialects

Figure 1. The stratificational structure of contemporary Dutch

Now, what I would like to suggest is that the actual choice from among these three logical possibilities may be determined at least in part by the extent to which the three models of standardization are prevalent within the language community.

304

Meaning and culture

The first possibility (in which colloquial Flemish moves upwards in the direction of the upper stratum) would receive the strongest impetus from a rationalist attitude towards processes of standardization. A growing tendency to participate in those activities that are typically associated with standard language use would go hand in hand with a further dissemination of the standard language in its present form. Whether this process would imply a further convergence of the upper register with Netherlandic Dutch is less clear: apart from the overall universalist aspect of the rationalist stance, i.e. the tendency to broaden the conmmnicative scope of the language, the functional motivation for such a convergence would seem to be relatively weak. We should note that a process of continuing convergence could also be triggered by a nationalist attitude, if the Flemish and the Dutch feel as one supra-national people, united by a common language across the borderline between Belgium and The Netherlands. However, although there exists a minor tradition of people advocating such an 'ethnic' union of Holland and Flanders, it plays no significant role in popular opinion. The feelings that exist between the Dutch and the Flemish are feelings of competition and gentle animosity rather than ethnic, language-based solidarity. The second possibility (an increasing internal uniformity within Belgian Dutch, achieved through an informalization of the highest register) links up with the nationalist model. Flanders has achieved a political emancipation within Belgium, and this might be reflected in a growing linguistic independence - in the development, in other words, of an independent standard language nornl that is increasingly different from the imported Netherlandic Dutch that was the initial reference point for the standardization process in Flanders. If the nationalist model prevails (and if it is oriented towards Flanders alone and not towards an 'ethnic' union of Holland and Flanders), the speakers of Flemish might develop a growing preference for the indigenous language forms that they encounter at the colloquial level rather than for the imported forms in the higher registers. The third possibility (a status quo) has a romantic and postmodern ring to it, given that the romantic attitude in its purest form involves antiestablishment feelings. The romantic appreciation of standardization (if it is not lifted to the nationalist level) is basically a suspicious one: standardization is seen as a threat to diversity in general and to one's own local or individual identity in particular. Distancing himself from the official standard, the romantic has several options: he can plead for diversity (the acceptance of all varieties as equal) or he can try to impose his own language as the standard (the nationalist option), but he could also simply maintain the dis-

Cultural models o.fstandardization

305

tance, accepting that there is a language for official occasions with which he does not identify, and a colloquial language in which he feels at home and that is sufficiently distant from the official language to be perceived as an expression of his own 'tme' identity. Are there any indications at all that Flanders might indeed have a mentality that favors such a stable form of diglossia? Its own popular self-image would certainly seem to support such a claim. Flanders perceives itself as being mildly anarchistic, in the sense that it is highly suspicious of official authorities, and in the sense that it does not take laws and regulations all too seriously. Tax evasion, for instance, is frequently described as a national sport. Also, historical reasons are often cited to explain this mentality. As mentioned above, Flanders has a history of foreign mle, and this is supposed to have engendered a mentality of critical distancing with regard to all forms of official authority. Postmodem tendencies towards informalization and dehierarchicalization would of course strengthen such an attitude. So, could it be that this mentality supports a romantically diglossic attitude in linguistic matters? The suggestion is certainly fascinating enough to warrant further empirically sound research about the mentality of the Flemings and their cultural history. To be sure, I am not able to make any predictions about the probable course of events in Flanders. The evolution will certainly not only be determined by the attitudinal models mentioned here, and I do not have any attitudinal data yet that determine the strength of one or the other attitude. What I have tried to show, though, is that it makes sense to think about actual processes of standardization in terms of the cultural models identified in the first part of this paper. Cultural models of standardization are real not only to the extent that they shape public debate, but also to the extent that they lie at the basis of the language attitudes of the language users.

5.

Conclusions

I have tried to do a number of things in this paper. First, I have argued that our thinking about questions of language variation and linguistic standardization can most often be reduced to two underlying, opposing cultural models: a rationalist one and a romantic one. Second, I have suggested that the models originated in the eighteenth century, and then went through a number of transformations in the course of the following two hundred years. Specifically, the nineteenth century witnessed

306

Meaning and culture

the development of a very influential nationalist model that combines elements of both basic models, and the late twentieth century witnessed a shift towards questions of globalization and the international position of English. At each of these stages, the models exhibit a certain tendency towards a coalescence, but at the same time, tensions remain - tensions between the models, and tensions within the models. Finally, I have tried to demonstrate that the models in question cannot only be used in discourse-analytic fashion to scrutinize argumentations and discussions, but that they may also be seen as shaping the language attitudes of linguistic communities. If they have a real presence, they do not only show up in the way in which people talk about standardization, but also in the way in which they act. Each of these points may be further developed. The identification of the models may be expanded towards a systematic map of standardization discussions, charting recurrent patterns of statements and replies. The historical sketch might be developed into a synthetic historical overview of standardization processes, standardization debates, and their relationship with linguistic theorizing. And the suggested link between models and language attitudes could lead to empirical attitudinal research. All of these possible developments, though, should contribute to a common goal: a better understanding of the underlying logic of standardization debates.

Chapter 12 Caught in a web of irony: Job and his embarassed God

Originally published in Ellen Van Wolde (ed.), 2003, Job 28. Cognition in Context 37-55. Leiden: Brill. One of the best indications for the success of Cognitive Linguistics is the way in which its models of analysis spread to disciplines that lie outside linguistics in the traditional sense, like literary studies (as represented by Steen 1994, Freeman 1995, 2000, Turner 1996, Brandt 1998, Hamilton 2002, Danaher 2003, Popova 2003, Hiraga 2003) or the study of religious language (see among others Biebuyck, Dirven, and Ries 1998, Boeve and Feyaerts 1999, Feyaerts 2003, Tabakowska 2000, Jakel 2002). The present paper belongs to the latter group; it derives from a conference held in Amsterdam in 2002, in which biblical scholars and cognitive linguists convened to discuss the Book of Job. Admittedly, the principles of Cognitive Linguistics put at work in the present study are only high-level ones, not very specific modes and models of analysis. Still, the paper illustrates nicely how a contextualized reading (both in terms of the internal context of the text and the cultural context in which it was written) may enhance understanding.

1.

Dealing with discrepancies

In an ironical but at the same time esthetically pleasing way, the position of the reader with regard to the Book of Job is analogous to the position of Job vis a vis God's behavior: for neither of them, it is an easy matter to arrive at a satisfactory interpretation. But if both God and the text are difficult to make sense of, it should come as no surprise that the Book of Job has been subject to postmodern analyses. Postmodern criticism is very much concerned with two related characteristics of texts that feature prominently in the Book of Job: on the one hand, the absence of a definitive interpretation (what is, ultimately, the significance of human suffering?), and on the other, the presence of incongruities and discrepancies (why, for instance, does God answer Job by basically repeating a number of the things that Job himself has already said?) In Job scholarship, these two characteristics seem to lie at the basis of two different strategies of interpretation, one somewhat less

308

lvleaning and culture

extreme than the other. Whereas a more moderate approach shows how the tensions in the text contribute to meaning rather than subverting it, the more radical form of postmodern interpretation takes its starting-point in the absence of a definitive interpretation, and dialectically turns this absence into the very message of the text: the impossibility of arriving at an ultimate meaning is the meaning of the text. A postmodern approach, however, is merely one way of making sense of the alleged inconsistency and inconclusiveness of the text. More traditional is the earnest philological habit of attributing difficulties to textual corruptions and interpolations, and there have of course been many suggestions as to the philologically heterogeneous character of the Book of Job. I am not in a position to judge the philological value of such arguments, but I will assume that it is good hermeneutic practice to apply an interpretative Occam's razor: if we can find a coherent meaning, that option is to be preferred over postulating different textual sources. The same methodological guideline applies, to be sure, to radical postmodernism: renouncing a coherent textual interpretation in favor of a meta-textual reading is something to be done only as a last recourse, when all other possibilities have been rejected. So have they? What I propose to do in this paper, then, is to steer clear of both the Scylla of postmodem extremism and the Charybdis of philological facility. Incoherence, a lack of harmony, clashes between different aspects of the text may in fact be interpreted in yet a third way: as symptoms of humor. As a literary genre, humor is the form par excellence to harbor incongruities. So would a humorous reading be an adequate way of dealing with the interpretative difficulties of the Book of Job? To what extent can it be read as an ironic text? The question is not new, and what I will be able to add to the debate is merely a specific reading of God's speech in chapters 38-41. I will develop my argument in three steps. First, I will review some of the evidence for the presence of irony in the Book of Job. Second, I will chart the ironists' positions with regard to the theophany of chapters 38-41: for those writers who have taken the ironical aspects of the Book of Job seriously, the nature of God's speech is as much a matter of debate as it for those writers who have simply taken the Book of Job seriously. And third, I will have a closer look at chapters 38-41, arguing that they can indeed be interpreted ironically. The outcome is a reappraisal of the dramatic figure of God: I will argue that the God-character is best interpreted as an embarrassed God.

Caught in a web ofirony

309

2. Methodological preliminaries

The methodology that I will use in the following pages is not specifically linguistic in any technical sense. (It is not even specifically lexical, in spite of my personal specialization in lexical semantics and lexicology.) There is, however, a consistent inspiration from Cognitive Linguistics, which it may be useful to make explicit at this point. (For an introduction into Cognitive Linguistics, see Taylor 1995, Ungerer and Schmid 1993. Foundational works are Langacker 1987, 1991 and Lakoff and Johnson 1981, Lakoff 1987.) As a first step, let us note that my analysis involves three specific thematic discussions: an investigation into the concept of wisdom as it appears in the Book of Job, a pragmatically oriented analysis of Job's speech, and an attempt to answer the question whether God's reply to Job can be considered ironical. These topics are situated at different levels of linguistic structure: the analysis of the notion of wisdom looks at a single category, whereas the other two involve utterances as a whole. As I will try to make clear presently, the three topical questions illustrate a number of views on meaning that are cherished by Cognitive Linguistics. This does not mean, to be sure, that a similar analysis could not be arrived at by other means, i.e. starting from a different framework. I have since long claimed, for instance, that Cognitive Semantics links up in fundamental respects with a traditional philological approach, and (notwithstanding the recognition that Cognitive Linguistics has developed a battery of analytical notions and techniques that go far beyond anything to be found in the philological tradition), I could very well conceive of the following analysis as coming out of a more traditionally oriented philological approach. The relevant views on meaning, then, are the following. Each of them relates specifically to one of the three topical discussions. I Meaning involves categorization. Linguistic meaning is not a straightforward reflection of the world, but it is a way of shaping reality, of making sense of the world. In the case of a concept like 'wisdom', what is to be understood by that term is not a priori given. It is, as we shall see, a category that derives its experiential content from its relationship with other concepts. In the Book of Job, wisdom is not just 'the quality of being wise' or some such lexicographical definition, but it is closely connected with age and experience: I will argue that it is crucial for an understanding of the Book of Job that wisdom comes with age. If we merely think of lexical concepts in the way in which they tend to be economically defined in dictionaries, such

310

Meaning and culture

aspects of a concept might easily be considered of secondary importance. If one thinks of meanings as categories of experience, however, the entire canvas of associations that the concept entertains becomes part of the meaning. 2 Meanings may invoke culturally specific background knowledge and assumptions. As categories making sense of the world, specific categories are situated against a wider context of lived experience, which may be either physiological/biological or cultural. In the pragmatic analysis of Job's speech, I will suggest that Job seems to be flaunting a culturally specific pragmatic politeness hierarchy that has its roots in a patriarchal context. 3 Meanings are contextually flexible. The meanings that are realized in actual speech contexts need not be completely identical with the meanings that are more or less permanently stored in our linguistic memory. The interplay between specific circumstances and the linguistic message may create new or different readings. On the level of individual categories, this flexibility is well-known in the form of prototype effects (Geeraerts 1997a). In the Book of Job, the importance of context becomes specifically clear if, at the level of utterances rather than individual categories, we look at the way in which irony comes about. Irony is typically a case in which an expression receives a different interpretation in one context rather than another - even extremely so: one reading is the opposite of the other, and moreover, the contrast between the two readings is part of the effect that the speaker wants to achieve. When, for instance, Job replies to Sofar and the other friends: 'wisdom will die with you '(12:2), the utterance receives an ironic interpretation because we know that Job is far from happy with the comfort that his friends purport to give. In the context, Job's reply acquires a sarcastic reading. The impact of context in the constitution of irony is complicated by the fact that the knowledge that triggers the ironic reading is not necessarily shared by all participants (and the reader, to be sure, is one of those participants).

3. Humor in the Book of Job Before focusing on the possible irony of God's speeches in chapters 38-41, let us first establish that a reading of the Book of Job as a humorous text is not an altogether implausible option. Humor is far from absent in the Old Testament (Good 1965, Knox 1969, Friedman 2000), and a number ofauthors (Whedbee 1977, Robertson 1973, 1977, Hoffmann 1983) have specifi-

Caught in a web ofirony

311

cally analyzed the Book of Job with regard to its humorous content. The most radical position in this respect is that of Robertson and Whedbee, who argue that the text as a whole belongs to the genre of comedy. I will discuss the views of Whedbee, Robertson, and Hoffinann more thoroughly in the next section. At this point, I will merely enumerate some of the humorous aspects of the text. Many of the points enumerated here may be found in the works of the authors just mentioned, but I have tried to bring some order into the matter by grouping the observations in four categories: the conversational style of the book, the characters, the plot as a whole, and specific details of the plot. First, throughout the conversations and speeches that constitute the main body of the text, there are sarcastic remarks and ironical interchanges among the main characters. For one thing, Job regularly scorns his friends for their failure to provide an adequate answer to his misery. (Quotes are from the Revised Standard Version of 1947.) 12:2 No doubt you are the people, and wisdom will die with you. 13:5 Oh that you would keep silent, and it would be your wisdom! 26:2 How you have helped him who has no power! How you have saved the ann that has no strength! 3 How you have counseled him who has no wisdom, and plentifully declared sound knowledge! 4 With whose help have you uttered words, and whose spirit has come forth from you? For an other, God mocks Job's claim to wisdom (the claim, that is, that he could explain something to God that he does not know yet): 38:4 Where were you when I laid the foundation of the earth? Tell me, if you have understanding. 5 Who determined its measurements-surely you know! Or who stretched the line upon it? and Job's challenging attitude: Deck yourself with majesty and dignity: clothe yourself with glory and splendor. 11 Pour forth the overflowings of your anger, and look on every one that is proud, and abase him. 12 Look on every one that is proud, and bring him low; and tread down the wicked where they stand. 13 Hide them all in the dust together; bind their faces in the world below. 40: 10

312

Meaning and culture

14 Then will I also acknowledge to you, that your own right hand can give you victory. Second, the characters have caricatural features. To begin with, Job's friends are a caricature of the wise counselor: they are supposed to comfort Job and provide him with good counsel, but blatantly fail in their task. Next, Elihu is even more ridiculous: there is a comical contrast between the tension and the expectation that is created when Job so to speak calls God's bluff, and the unexpected appearance of Elihu, who is portrayed as a bombastic, long-winded and self-important young upstart. In contrast with the awesome God whose appearance is expected, Elihu is a joker who relieves and at the same time prolongs the tension. Finally, even Job himself, working himself stubbornly from a state of utter desolation into a frenzy of rebellion in which he challenges God to a showdown, displays an obsessive single-mindedness that would suit any of Moliere's comical archetypes. Third, the overall narrative structure of the text corresponds with that of a comedy (not in the sense of an amusing story with funny characters and farcical incidents, but in the sense of a literary work with a happy ending). The overarching structure moves from a blissful, fairytale-like beginning to a situation in which the main character lapses into wretchedness and misery. When the main character has gone through the test (and has changed his ways accordingly), a happy end follows. And fourth, at a more detailed level of the plot, the progress of events is characterized by comical repetitions and comical reversals. The dominant example of repetition is obviously the conversation between Job and his friends. Time and again, the attempts to comfort and convince Job fail, but time and again, the pattern of futile argument and emotional reply is repeated. A specifically ironical touch is the way in which the pattern of repetitions gradually comes to a halt, with ever shorter interventions of the friends, and finally the complete absence of the expected third speech by Sofar. The ineffectiveness of the interventions, in other words, is reflected by the gradual fading away of the friends' speeches: they literally have no arguments left. An ironic reversal occurs in Job's attitude with regard to God: after God has only once raised his voice, Job's rebellious challenge abruptly changes into subservience. Similarly, there is an ironical reversal in Job's relationship with his friends: while they were supposed to act as intercessors for Job but failed to do so properly, Job, after his restoration, functions as an intercessor for them.

Caught in a web a/irony

4.

313

The irony of God's speech I

Both microstructurally and macrostructurally, then, humor and irony seem to play an important role in the construction of the Book of Job. But all interpretations of the book stumble over God's speech - the final chapters of the poetical part of the text that are so hard to interpret that some critics have even suggested that they do not belong to the original text. Writers like Hoffmann, Whedbee, and Robertson, who have devoted specific attention to the humorous and ironic aspects of the book, are likewise divided as to the meaning to be attributed to the intervention of God. In this paragraph, I will try to show that the positions they take correspond to a specific pattern. At the same time, it will appear that there is a hole in the pattern, and the next paragraph will suggest how the hole can be filled. The position of Robertson and Hoffmann may be contrasted with that of Whedbee: whereas the former argue that God's speech is ironic and evasive, and therefore does not constitute a proper answer to Job's question, the latter underlines the non-ironic, relevant aspects of God's answer. Let us take a closer look at the various positions. Robertson (1973) interprets God's speech against the larger context of the book, and specifically in the light of chapter 9, where Job predicts what God would do in a face-to-face encounter: God would overawe Job with his intellectual superiority and his overwhelming powers. And when God does indeed appear, he acts precisely as predicted by Job: Job simply does not stand a chance, and he acknowledges as much. God then appears as an inconsiderate tyrant rather than a just and merciful ruler: 'So God's rhetoric, because Job has armed us against it, convinces us that he is a charlatan God, one who has the power and skill of a god but is a fake at the truly divine task of governing with justice and love' (1973:464). Job's repentance, in turn, can only be a non-authentic stance, faking rather than feeling submissiveness. The overall meaning of the Book of Job, then, is that of curing man's fear of fate, his destiny, the unknown by 'ridicule of the object feared' (1973: 468): 'So we know of him what we know of all tyrants, that while they may torture us and finally kill us, they cannot destroy our personal integrity' (1973: 469). Whedbee (1977), as we have seen, accepts Robertson' s overall premise that the Book of Job belongs to the genre of comedy, but challenges Robertson's reading of God's speech. Robertson, he claims, underplays the significance of God's intervention by merely concentrating on the ironic interplay between Job's prior speeches and God's speeches. God does in fact go fur-

314

lvfeaning and culture

ther than just echoing Job's predictions: he redirects them towards a positive vision of the universe. Yes, God is all-powerful (as Job predicted), but whereas Job paints a bleak picture of God's overwhelming strength, God himself brings, as Whedbee states it, a 'playful, festive note in the portrayal of creation', stressing the magnificence and the order of the universe rather than the chaos and oppression that Job highlighted. Job's confession, then, is a genuine recognition that he saw creation in the wrong light, and the restoration of Job, rather than being ludicrous in all in its implausibility (as Robertson would have it) is a true happy ending. Robertson (1977), incidentally, has answered Whedbee, but in a very half-hearted way, admitting that 'though neither theory is free from defects, both are about as coherent and comprehensive interpretations of the Book of Job as we are likely to get'. Hoffmann's position (1983, 1996), to round off the overview, is an interesting variant of Robertson's. Hoffmann too recognizes God's speech as ironical, but it is not irony of God with regard to Job, but rather irony of the author with regard to himself. It is worthwhile to quote Hoffmann (1983: 19) in full: The author has ironically presented himself as someone who knows the answer to the questions he has raised. However, in the manner of irony, that which is written (i.e.: here is the well-ordered answer to the questions raised heretofore) implies the exact opposite. What is really being said is: I cannot answer the difficult questions I have presented, and they are still as problematic as before. What we have here is selfirony, in that throughout the book the author creates the impression that he knows the answer and will tell us at the end. ( ... ) Hence we arrive at the self-irony of man: All I. as a human being, can do is to delude myself that there is a solution to the problem of divine recompense and mask my delusions one way or another. The ironic truth is that I must accept my fate, come what may. The inaptness of God's words with regard to Job's question turns the text as a whole into an ironic statement. All along, the text seems to work its laborious way towards an answer, but when the answer finally comes, it is way off the mark. The pattern that emerges from the positions of these three authors is a simple one: either God's speech is an appropriate albeit indirect response to Job (Whedbee), or it is ironic and evasive (Robertson and Hoffmann). But are these alternatives mutually exclusive? At least in one way they are not: the view that God's speech is a corruption of the original text in the form of a later insertion can be equated with the defeatist view that it is neither ap-

Caught in a web a/irony

315

propriate nor ironic, that it is, in other words, uninterpretable. The true hole in the pattern, then, involves the possibility that God's speech is both appropriate and ironic. What would happen if we pursued this line of thought? Can such an interpretation be maintained? I will explore the possibilities in the next paragraph, but I first have to indicate how the approach that I will propose differs from the positions discussed so far. On the one hand, I agree with Whedbee that Robertson underestimates the impact of God's speeches: they do not just mimic Job's earlier speeches, but they add something to them. I disagree with Whedbee, however, as to what it is that is being added. Even though there is a element of playfulness in God's speech, the speech as a whole is not really a straightforward hymn to creation. If it convinces Job, it has to be on other grounds than just the festive nature of creation. A closer scrutiny of God's speech is in order, then. On the other hand, I entirely agree with Hoffmann that we have to take into account different levels of irony in the text. Whereas Robertson only envisages the irony of God with regard to Job, we have to allow for forms of irony that are perceived only by the author and the audience, and not by the characters. However, I think we can be more precise about the conspiracy between the author and the audience than Hoffmann is, by taking a closer look at the narrative framework of the text. If the audience knows more than the characters, it is through the narrative setting introduced in the prologue. So how does God's speech relate to the prologue?

5. The appropriateness of God's speech I

There are two points of a pragmatic nature that need to be elucidated first. To begin with, God's speech is definitely an indirect speech act, and indirect speech acts can constitute extremely appropriate answers. If you ask me: 'Will you attend Ron Langacker's lecture?', and I reply: 'Have men walked on the moon?', then my rhetorical question is not a direct answer, but it does constitute an appropriate response of the' obviously' -type. Further, appropriate responses to speech acts may take the form of questioning the felicity conditions of those speech acts. Felicity conditions are criteria that have to be fulfilled if a speech act is to be properly made. They may involve the sincerity of the speaker, the authority invested in the speaker, the ability of the speaker to live up to the consequences of his commitments, and so on. In all cases, satisfaction of the felicity conditions is a

316

Meaning and culture

necessary prerequisite for a certain speech act to be pragmatically valid. This implies that certain acts may be appropriately countered by pointing out that the felicity conditions of the act do not hold. Imagine a 15-year old at the box office of a theater showing only X-rated movies, asking whether the film is shown with or without a break. Ifhe gets the answer: 'Sorry, but you are under age', then it is clearly not a direct answer to his question, but it is an appropriate answer, because it makes clear that he was not in a position to ask the initial question to begin with. The presuppositions that are necessary for the question to be a relevant one, are not met, and pointing that out is a conversationally adequate answer. My suggestion now is that God's speeches may be read as indirect speech acts questioning the felicity conditions of Job's question. I will argue for this position in two steps. First, I will analyze the concept of wisdom in the Book of Job, because Job's speech act is fundamentally franled as an investigation into knowledge and truth. If Job is asking for insight but is asking for it in the wrong way, then we will first have to be clear about the conception of knowledge and insight that permeates the Book of Job (and that is insufficiently heeded by Job). Second, I will have to deal with the bipartite structure of God's speech, because each of those parts addresses a different aspect of Job's pragmatically inappropriate behavior. The concept of wisdom (and insight, as both terms are often mentioned together: see 11:6, 15:8-9, 26:3, 28: 12, 32:7-8, 34:34-35, 38:36, 39: 17) is basically presented with three features (apart from the fact, of course, that wisdom is difficult attain). First, wisdom is a condition for appropriate speech. See e.g. 11 :47,20:3, 26:3-4, 32:7-8, 34:35, 36:4 and other passages, of which God's comment in 38:2-4 is perhaps the most poignant one. In itself, this association between wisdom and the pragmatic validity of a given discourse is not surprising. In the dramatic setting of Job's attempt to justify himself, speaking the truth, expressing wisdom, formulating the right insight is probably the motor concept par excellence of the whole text. We will see presently, however, that this unsurprising link between possessing wisdom and speaking appropriately is the basis for a specific pragmatic logic. Second, wisdom correlates with age. Take the second speech by Elifaz, lines 15:7-10. Elifaz questions Job's insight by pointing out that Job is not of infinite age. 7 Are you the first man that was born? Or were you brought forth before the hills? 8 Have you listened in the council of God? And do you limit wisdom to

Caught in a web ofirony

317

yourself? 9 What do you know that we do not know? What do you understand that is not clear to us? 10 Both the gray-haired and the aged are among us, older than your father. Superior knowledge, in other words, presupposes superior age. Similar lines are to be found in Sofar's speech in 20:4, in Elihu's speech in 32:7-9, perhaps also in Job's words in 28: 13, and most certainly in God's remark in 38:4. Job cannot have infinite wisdom because he does not have infinite age. If we accept that the experience needed for ultimate wisdom exceeds the bounds of human life, then, of course, wisdom cannot be found in the land of the living. The patriarchal association of wisdom with experience and age is, in other words, extrapolated towards God. In the ordinary context of the patriarchal society, the elders know best. God however is a superhuman hyper-patriarch who has to know even better, if only because of his superior age. Third, it follows from the combination of the first and the second feature that age is a prerequisite for appropriate speech, or at least, that there is likely to exist a pragmatic hierarchy of speaking rights based on age and assumed wisdom: the young do not speak before the old. As a form of a politeness hierarchy, such a hierarchical restriction on the structure of conversation would be a conventional part of a patriarchal society, but in the context of the Book of Job, it acquires a specific depth through the way in which the text focuses on the concept of wisdom. This is particularly apparent in the speech of Elihu. On the one hand, Elihu presupposes the patriarchal model, expressing his disappointment in the older people who have spoken before him, but who have not exhibited the wisdom that might be expected of them: see 32:7-9. On the other hand, Elihu breaks the hierarchy. According to the patriarchal politeness hierarchy, Elihu speaks before his time. He is a young person challenging and the elders, and moreover challenging them on their own domain: that of wisdom. That is why Elihu can be depicted as a fool, making a lot of wind, inflating himself, making himself more important, and continually re-affinning his own importance. Elihu, in short, challenges the patriarchs without being in a position to do so. But ironically, this is precisely what Job does with regard to God. The position of Elihu with regard to Job and his patriarch friends is structurally analogous to the position of Job with regard to the divine super-patriarch. The epistemological hierarchy of wisdom and age, coupled with the pragmatic hierarchy of politeness and conversational turn-taking, leads to a lin-

318

Meaning and culture

ear ordering where Elihu is lower on the ladder than Job and where Job is lower on the ladder than God. But then, Job calling God to account is not observing the linear ordering just as much as Elihu is in reprimanding the elders. In challenging God, Job is making a non-felicitous speech act, because he is not observing the hierarchy of politeness and conversational rights, and in claiming that he can teach God something about his innocence, he is ignoring the hierarchy of wisdom that is itself (as we have seen) intimately connected with the pragmatic hierarchy. God, to be sure, makes all this apparent, putting his mighty finger on Job's breach of felicity conditions. God makes clear (in a roundabout way, by pointing out rather than by stating explicitly) that Job is going beyond the pale. Note that we have just identified two relevant conditions that Job is not observing: one having to do with a hierarchy of politeness of conversational rights, and the other having to do with a hierarchy of wisdom and knowledge. That is why we have two speeches of God: in each of the speeches, God addresses one of Job's pragmatic trespasses. The first speech addresses the hierarchy of wisdom (i.e. Job's flaunting of the presupposition that superior wisdom correlates with superior age, and that ultimate wisdom relies on infinite age). The second speech addresses the hierarchy of power (i.e. Job's flaunting of principles of conversational politeness).

6. The appropriateness of God's speech 11 But the binary nature of God's speech is one of those aspects of the text that have troubled the interpreters. It has been claimed that the double speech of God is a flaw in the text, an interpolation perhaps, and in any case a superfluous repetition. Why are there two speeches of God? Why is there a second speech if Job has already submitted after the first one? Is God just sadistically rubbing it in, delivering a final blow to an already subdued enemy? I would argue that there are subtle differences between the two speeches that indicate that they have a different focus. There are three basic observations that I would like to advance in support of this idea. First, God's opening lines, setting the topic of the speeches, unmistakenly mark a difference. The opening of the first speech takes Job at issue for wanting to teach God something, as if God were saying: 'So you want to instruct me? Well okay, be a man, and let's see if you can teach me anything'. The second speech puts it no less explicitly in temlS of power rela-

Caught in a web ofirony

319

tions: 'You want to challenge me? Well okay, let's see how far you can get.' Whereas 38:2 invokes knowledge (and 'words without knowledge') as the theme of the speech, 40:8 invokes condemnation and justification, i.e. the notion of authority. Needless to say, the two aspects that we have identified above involve precisely the way in which Job counteracts or at least ignores the prevailing models of knowledge and authority. Second, Job's reaction is fitting with regard to the topic of the speeches. After the first speech, he basically admits that he has no knowledge to speak of, and then he shuts up - which is an entirely appropriate reaction in a cognitive context, i.e. in a context focusing on knowledge and wisdom. After the second speech, by contrast, Job does something rather than say something: he submits and repents. The reaction is not a cognitive one, but a symbolic one; it involves making symbolical amends. Whereas Job's first reaction is a cognitive recognition of ignorance, his second reaction is a symbolical recognition of impudence, followed by a ritual act of making amends. And third, the text of the speeches, the argument that each of them contains, is clearly different. In the second speech, we find images of power the two monsters - and of power over power. God conjures up the terrible image of Leviathan and Behemoth, and further shows (in a positively hilarious way) that he controls the monsters: 'Look, I can draw a straw through the crocodile's nose and lift it up. Can you?' In the first speech, on the other hand, we get images of the birth of the cosmos and the principles that organize it. Given the model of wisdom that we have been able to identify, this is entirely appropriate: God displays his superior knowledge which is grounded in his superior age, i.e. the fact that he (and not Job; compare 15:7) witnessed and engineered the creation of the world. Once again, God is overwhelming Job: 'Look, I know all of all this. Do you? And if you do not, how can you question me?' In short, then, God seems to be saying something like this: 'Well, Job, you are asking me for a chance to defend yourself, but actually, you are not in a position to ask. You challenge me to explain what is going on, but come to think of it, you do not have enough wisdom to understand what is going on, and moreover, you are not powerful enough to challenge me. ' There is, however, another step to take. What is the function of the second part of the first speech? Chapter 39 does not contain cosmic images but rather describes a number of animals (though not powerful monsters as in chapters 40 and 41). One way of reading 39 is just as an addition to 38, as further proof of God's powers of creation (and hence his unsurpassed knowledge). I would argue, however, that God is adding a nuance to the first

320

Meaning and culture

part of his first speech. To be sure, we cannot really say that God is noticeably subtle in his speeches. He uses extremely powerful images of cosmic and animal forces, and Job is duly overpowered; predominantly, God is just putting Job down. But in 39, a touch of subtlety does creep in. Overall, the chapter presents multiple images of carefree and proud animals. What is the function of this? I would suggest that the key lies right in the center of the chapter, where God describes the ostrich. 13 The wings of the ostrich wave proudly; but are they the pinions and plumage of love? 14 For she leaves her eggs to the earth, and lets them be warmed on the ground, 15 forgetting that a foot may crush them, and that the wild beast may trample them. 16 She deals cruelly with her young, as if they were not hers; though her labor be in vain, yet she has no fear; 17 because God has made her forget wisdom, and given her no share in understanding. 18 When she rouses herself to flee, she laughs at the horse and his rider. The passage (and particularly 39: 17, which seems to mirror 28:20) adds a crucial nuance to the conclusion that Job arrives at in 28:28. The ostrich is presented as carefree (39: 13) but at the same time as careless: it does not mind leaving its children in the desert (39: 14-15) nor treating them harsWy (39: 16). Moreover (39: 17), it does not mind if its efforts are useless, and it thinks higWy of its own strength, lightly attacking humans, i.e. beings that are higher on the scale of creation (39: 18). The ostrich, then, is the ironical counter-image of Job himself: Job does worry, conspicuously so, about his children coming to harm and about the fruit of his life's work being destroyed. And even if Job stands up against a higher, divine power, he does not do so laughingly. This ironical context gives specific depth to line 39: 17, which reads as an explanation of the ostrich's behavior against the background of the hierarchy of knowledge: the ostrich is carefree precisely because it has no insight. This could be called the 'tree of knowledge' argument: because man has eaten from the tree of knowledge, he is no longer happy and carefree like the animals. By implication, the very fact that Job does worry establishes that he is not entirely devoid of insight. God, in other words, refers to the hierarchy of wisdom to point out to Job that he got his presuppositions wrong (chapter 38), but at the same time, he subtly suggests (chapter 39) that the hierarchy of wisdom has to be extended

Caught in a web ofirony

321

in another sense as well: starting from Job, the Great Chain of Being has to be extended upwards from man to God, but it also has to be extended in a downward direction from man to animal. So, we have to modify our summary of God's first speech slightly: 'Job, you are challenging me on the point of wisdom. Well, you're in no position to do so, because my knowledge is infinite and yours is limited. But if it can be of any comfort to you, remember that your limited knowledge is precisely what distinguishes you from the animals.' Job is asking questions that he cannot answer, but the very fact that he can ask them demonstrates that he is closer to God than he may be inclined to think.

7. The irony of God's speech 11 In an indirect, roundabout speech, replete with images and innuendo's, God

suggests that the felicity conditions for Job's challenging attitude are not met, with regard to the question of wisdom just as well as with regard to the question of power and authority. This is a relevant answer, given the hierarchical model of knowledge and authority that permeates the text. And because Job clearly agrees, both in thought and in behavior, God's reply can only be considered an effective one. But even if God's discourse is an appropriate and effective one, yielding an answer that Job can genuinely (though perhaps uncomfortably) live with, God is still being evasive and dismissive. He is sending Job off none the wiser, or at least, just a tiny bit wiser. Shouldn't we expect a different answer from God? It is important to make a distinction at this point between the expectations we may have as a reader and the expectations we may have as a believer. As believers, identifying with Job, we may either find God's answer unsatisfactory, or we may accept the idea that God's logic escapes us. As readers, however, we need not even ask the question, because we know what the answer is. This is a recognition that is remarkably absent from most discussions. If Job's question is: 'Why do you torment me so while I am ilIDocent?', we as readers know the answer, and we have known it from the very start of the book: God is tormenting Job merely because God has been persuaded by the satan to test Job. God has been bragging about Job, and being challenged by the satan to prove himself right, God lets the satan have his way with Job. As readers, then, we know very well that God's answer to Job, though pragmatically relevant and conversationally effective, is not the true answer.

322

Meaning and culture

So why doesn't God give the true answer? At least one plausible solution, it would seem to me, is to recognize that God may well be embarrassed. Even if the satan in the story were not the devil of later Judaism and Christianity but rather a member of God's courtly council (see Pagels 1995), God is not exactly well-behaved by common human standards. He has been playing a bad joke on Job, and he is not really in a position to confess to it. Job's insistence creates a painful moment for God: he could not really admit that he has been agonizing Job because he is his most faithful servant, could he? The suggestion of an embarrassed God, outlandish though it may seem, does not only follow from the narrative context. There are two other features of the very final chapter of the book that fit in well with such an interpretation. First, how can God declare that Job has spoken right of him when Job himself admitted that he had no understanding and that he had been talking nonsense? God's declaration in 42:7 can hardly apply to Job's earlier statements, which were dismissed in 38:2 as lacking understanding. At the same time, it can hardly apply to Job's newly acquired submissive attitude (and silence), because it is not different enough from the deferential attitudes advocated earlier by Job's friends, who are now being punished. Logically speaking, then, God's statement that Job has spoken right of him would seem to be paradoxical on any account. However, if we take it as a gesture of atonement, as a comforting signal that everything is alright now, as a way of saying that in spite of everything, Job is still God's best friend, then of course it matches the assumption that God somehow feels indebted to Job. Second, the restoration of Job carries a hint of overcompensation. The reversal is so abrupt and the reparation so exaggerated that it can only be a sign of God's bad conscience. God is trying to make amends, and his feeling of unease induces him to do so lavishly. On the whole, then, coming back to our earlier discussion of the views of Whedbee, Robertson, and Hoffmann, we can conclude that the opposition between Whedbee on the one hand and Robertson and Hoffmann on the other is not an exclusive one. God's speech is appropriate, but at the same time, his behavior appears to be profoundly ironic, in a sense that may be perceived by the audience and God himself, but hardly by Job. There is an ironic discrepancy between God's show of power with regard to Job, and his actual state of embarrassment. God's authority, his wisdom, and his strength overpower Job, but the reader can recognize them as a disguise of God's vulnerability. Likewise, there is an ironic discrepancy between the message

Caught in a web o{irony

323

of God's indirect speech act ('you're in no position to ask') and the message that we assume God is trying to avoid ('I am in no position to answer').

8. The theology of God's embarrassment

Within the dramatic structure of the text, then, God's behavior exhibits an ironical embarrassment. But what does this mean for our interpretation of the text as a whole, if we think of it as a religious text? How should we interpret God's embarrassment? I would like to suggest that we have two basic alternatives. If we contrast the conunon expectation of an onmipotent God with the image drawn in the Book of Job - the image, that is, of an all too human, fumbling, embarrassed God - then the text has an overall ironic meaning. We could even conjecture that we are dealing here with a form of what we would now identify as Jewish humor, with the author mocking his own initial, pre-existing image of God, and his own subordination to a God that he does not entirely fathom. But as we know next to nothing about the way in which the original audience may have understood the text (how old, in fact, is Jewish humor?), it may be cautious to contrast the radically ironic interpretation with a more charitable one: what if the Book of Job is trying to make a specifically subtle point about the relationship between God and man? At this point, I am again drawn to the ostrich passage as a key for an alternative interpretation. On the ladder of Creation, human intelligence has a specific place. It alone can recognize the wonders of creation and the power of God, but it can only do so at a tremendous price - the loss of innocence and carefreeness. God's embarrassment at the enormity of the price, in return, would seem to ask for much more drastic measures than the mere restoration of Job. Perhaps the Job episode, this practical joke that almost got out of hand, convinced God that he needed to set things straight with mankind on a much grander scale. And so Job would be vindicated, as a prefiguration ... It should be clear that the context that could disambiguate between these alternatives can hardly be found within the text itself. I have so far tried to formulate an interpretation that is as consistent as possible with what we know about the internal features of the text and the external context (such as the patriarchal social structure) in which it originated. But there are natural limits to this approach, and the choice between the alternative interpretations

324

Meaning and culture

suggested here far exceeds the contextualized methodology that I have tried to follow. But that, I would say, is merely one more irony of this profoundly ironical text.

Section 5 Lexicography

Chapter 13 The lexicographical treatment of prototypical polysemy

Originally published in Savas L. Tsohatzidis (ed.), 1990, Meanings and Prototypes. Studies in Linguistic Categorization 195-210. London: Routledge. Conceptions of semantic structure should not only be theoretically interesting, they should also fit the applied perspective of practical lexicography: large-scale descriptive work of the lexicographical type provides a crucial testing ground for theoretical conceptions. In this paper, I argue that a prototype-theoretical conception of semantic structure sheds an interesting light on specific problems of dictionary making (like the linearization problem, involving the difficulty of mapping a multidimensional semantic structure onto a two-dimensional ordering of definitions). The paper is one of a number of publications in which I deal either with specific questions about the lexicography of Dutch (1986b, 2000b, Geeraerts and Janssens 1982) or with matters pertaining to theoretical lexicography at large (l985b, 1987b, 1989b, 1997b, 2000a, 2003). There's a biographical background to this interest in theoretical lexicography: I started my professional career working as a historical lexicographer on the Woordenboek der Nederland~che Taal, the Dutch counterpart (eventually completed in 1998) of the historical Oxford English Dictionary. and later, I acted as editor-in-chief of the 14th edition of the Van Dale Groot woordenboek van de Nederlandse taal, which is the main contemporary dictionary of Dutch (Den Boon and Geeraerts 2005). My views on the importance of prototype theory for semantic change (as laid down in my 1997 book on diachronic prototype semantics) would never have come about had it not been for the broad experience with historical meaning phenomena that I acquired when working on the Woordenboek der Nederlandsche Taal.

1. Lexicography and prototype semantics There have hitherto been two main directions in the application of the prototypical conception of semantic structure to the problems of lexicography. On the one hand, Wierzbicka (1985, 1987a, 1987b) has presented a method of semantic analysis which replaces traditional dictionary definitions by exhaustive, in-depth analyses of the prototypes underlying natural language

328

Lexicography

concepts. On the other hand, I myself have argued that the prototypical model of lexical polysemy is an indispensable aspect of any theory trying to classify the diverse forms of semantic information that may be incorporated in lexicographical reference works (Geeraerts 1985b, 1987b). Both transpositions of prototype theory to the field of applied linguistics differ basically with regard to the type of lexicographical problem they address. In fact, while Wierzbicka is concerned with the microstructural presentation problem, I have tried to (re)formulate the theoretical background of the microstructural selection problem. The terminology used here is based on the assumption that the problems of lexicography are concentrated on two times two domains. On the one hand, dictionaries contain a macrostructure (a set of lexical explananda) next to a microstructure (explanatory data for each of the elements of the macrostructure); on the other hand, lexicographers are faced with problems of selection and representation with regard to each of those two sets of elements (cp. Geeraerts 1984b). That is to say, while the macrostructural selection problem basically involves the question 'What lexical items should be incorporated into the dictionary?', the question 'What information should be presented with regard to the selected items?' is raised on the level of microstructural selection. Similarly, while the macrostructural presentation problem involves, among other things, the choice between an alphabetical and a thematic organization of the selected items, questions with regard to the arrangement of the distinct meanings of a particular word (should they, for instance, be listed chronologically, logically, or according to frequency?) belong to the field of microstructural presentation. Questions concerning the definitional techniques to be used with regard to individual meanings likewise belong to the microstructural presentation problem, and that is why Wierzbicka's approach is to be primarily situated there as well. By contrast, my own attempts to identify and classify the different types of semantic information (prototypical, stereotypical, technical) that can be incorporated into dictionaries are situated on the microstructural selectional domain. In this respect, both approaches are basically complementary with regard to each other. Whereas Wierzbicka's research is concentrated round the question 'Given that one decides to incorporate the prototypical meaning of natural language categories into one's dictionary, how should one define them?', mine addresses the question what other kinds of meanings, next to the prototypical one, might be selected for lexicographical treatment. This basic methodological complementarity between the two approaches does not imply, however, that the points of view developed in each case are necessar-

Lexicographical treatment o/prototypical polysemy

329

ily in accordance with each other. In particular, there are two remarks that I believe should be made with regard to the practical relevance of Wierzbicka's point of view. In each case, the pragmatic purposes of dictionaries play an important role in the argumentation. A first and minor remark involves Wierzbicka's answer with regard to her own basic question. Should prototypical meanings indeed receive the full definitions of 'gargantuan length' (1985: 37) defended and admirably illustrated by her? From a theoretical point of view, she is undoubtedly right: prototypical concepts are 'encyclopedic' entities that should be described in all relevant aspects, and that cannot be reduced in the structuralist fashion to minimal sets of contrastive features. This is a position that I have elsewhere defended at length myself (1985b), and which I see no reason to take theoretical issue with. From the practical, applied-linguistic point of view of lexicography, however, the question arises whether definitions should always be as lengthy as Wierzbicka proposes. The point to be made is this: the kind of information to be incorporated into a dictionary is ultimately determined by the purposes the dictionary is likely to fulfill for its intended audience. For certain types of audiences, it may not be necessary to have definitions specifying the full conceptual detail of prototypical categories. This is particularly the case if the dictionary user can be expected to be familiar with the concept in question; the definition should then do no more than enable the dictionary user to identify the concept in question, that is, to retrieve it from his own conceptual memory. Adult native speakers of English may, for instance, be assumed to know what houses are; by definition, they are supposed to know everything that is contained in a Wierzbickian analysis of the concept, because this analysis is intended precisely to describe the native speaker's conceptual knowledge. A definition of house in a dictionary for adult native speakers of English may therefore be restricted to a brief gloss identifying the concept in question Ca building for people to live in'). While such a minimal rather than a maximal Wierzbickian definition may help the dictionary user to distinguish between the entries devoted to house as a verb and house as a noun, the general question nevertheless arises why such familiar words should then be incorporated into the dictionary at all. One should not forget, however, that the entry in question will usually also contain information that is more likely to lead to a consultation of the dictionary to the extent that it is less familiar. (People may, for instance, wonder what eat someone out of house and home, or bring the house down mean.) This brings us straight to the second remark to be made with regard to Wierzbicka's application of prototype theory to lexicography, but before going

330

Lexicography

into that, it may be useful to fonnulate on a more general level the conclusion that may be drawn from the present remark. Most important is the recognition that there is a distinction between theoretical semantics and practical lexicography; a definitional technique that is optimally justified from a theoretical point of view need not be so from the pragmatic point of view of practical dictionary making. In this respect, it is not my intention to claim that Wierzbickian definitions could never be used when prototypical concepts are to be defined for adult native speakers; the only point I am trying to make is that the theoretical preference for such definitions cannot be transferred automatically and without further (pragmatic) justification to practical lexicography. Now, the impression that Wierzbicka tends to underestimate the distinction between practical lexicography and theoretical semantics is further corroborated by the fact that she explicitly restricts her analyses to the prototypical concepts at the centre of natural language categories. Her solution for the microstructural presentational problem thus seems to be coupled with a very specific answer with regard to the question of microstructural selection, viz. that it is the salient senses of a lexical category that should be defmed. It is not entirely clear whether she indeed believes that such an approach should be generalized to everyday dictionaries, but if so, this point of view should receive some nuances. As suggested a moment ago, the kind of things people may want to consult a dictionary for will very often involve precisely the peripheral, less familiar applications of a category. From a practical point of view (that is to say, with an eye to the pragmatic purposes a dictionary has to meet for a particular audience), restricting the microstructural selection to prototypes may be useful for certain types of dictionaries, but is certainly not generally applicable. Dictionaries have to provide infonnation about the polysemy of words just as well as about their basic meanmgs.

2.

The linearization problem and its background

Once it is accepted (on the microstructural selectionallevel) that dictionaries may wish to describe cases of prototypical polysemy, the microstructural presentational problem presents itself: how can such a prototypically polysemous cluster of meanings be adequately treated in the framework of a traditional dictionary? In general, the problem is the following: whereas lexicographical microstructures basically consist of a list of neatly separated,

Lexicographical treatment ofproto~vpical polysemy

331

consecutively numbered lexical meanings, the structure of prototypically organized concepts is characterized by clustering and maximal overlapping. How then can the multidimensional structure of prototypical concepts be mapped onto the linear order of the dictionary? What I would like to show here, is that there are a number of tricks and techniques that are intuitively used by lexicographers to solve the linearization problem. This is entirely as may be expected, or at least, if prototype theory is right, one may expect lexicographers to be troubled by the linear order of the dictionary. If the prototypical conception of categorization is correct, lexicographers should be looking for ways to make the description more faithful with regard to the perceived semantic structure of the category. That this is indeed the case may be demonstrated by having a look at a dictionary entry for an indubitably prototypical concept. Before, however, there are some preliminary remarks to be made. In the first place, all dictionaries are not equally deviant with regard to the canonical linear order. Once again, the pragmatic purposes of lexicography appear to be very important. If it is sufficient for a particular dictionary merely to enumerate the basic meanings of an item, without bothering too much about their semantic interrelations, no specific attempts to render the prototypical structure of the concept as adequately as possible should be expected. The only intrusion of prototypicality will then probably be the fact that the preponderant meaning will be enumerated first, less frequent ones being mentioned near the end of the entry. But if a dictionary intends to give a highly detailed, linguistically adequate description of the vocabulary of a language, matters will be different. Not only will the prototypical structure of the category become more apparent to the extent that more semantically distinct examples of usage are incorporated into the description, but also, the lexicographer will be more inclined to give an adequate description of prototypical polysemy to the extent that the scholarly level of his work rises. That is to say, first, if a dictionary concentrates on describing only the more salient senses of a word, the conceptual distance between those senses will be greater than if it also incorporates less frequent kinds of usage; the borderline cases in which the clustered structure of the prototypically organized category shows up most clearly will be precisely the less frequent applications. In this sense, one may stick roughly to the rule that larger dictionaries are more likely to show the effects of the linearization problem. Second, the pragmatic intention to render the semantic structure of a category as adequately as possible is stronger in those dictionaries that are compiled at least partly with a scholarly purpose. Whereas the ordinary,

332

Lexicography

everyday dictionary serves a primarily pedagogical purpose (transmitting practical, communicatively useful information about a specific language or a language variety), a number of dictionaries are explicitly conceived with the primarily scientific rather than pedagogical purpose of giving a linguistically adequate description of a language or a language variety. Clear cases are historical dictionaries such as the Dictionary of the Older Scottish Tongue and Bosworth and ToIler's An Anglo-Saxon Dictionary, or etymological dictionaries: these works are first and foremost independent pieces of linguistic research, in which pedagogical considerations only play a minor role. Obviously, then, the burden of linearization will be more felt in dictionaries for which linguistic descriptive adequacy is more important than pedagogical efficiency. This means that our example should preferably come from a large, 'scholarly' dictionary. Specifically, we shall use the entry vers from the Woordenboek der Nederlandsche Taal, which is the major historical dictionary of Dutch, ranging from 1500 up to the beginning of the twentieth century. (Actually, the entry is labeled versch in accordance with the older spelling of Dutch, but the contemporary spelling shall be used here.) In its scope, its methods, and its nineteenth century origins, the Woordenboek der Nederlandsche Taal is the Dutch equivalent of Murray's Oxford English Dictionary and Grimm's Deutsches Worterbuch. Whereas the latter have been completed for some time, the Woordenboek der Nederlandsche Taal, however, has not yet reached completion. This is at least partly due to the fact that the level of detail at which the dictionary works, is even higher than in the other two; having progressed up to the letter W, the Woordenboek now totals 33 volumes with an average of 1500 pages per volume. (For more information about the Woordenboek, see Heestermans 1979.) As a second preliminary remark, it should be noted that the linearization problem has something to do with the traditional form of the dictionary as a written text, but does not derive from it in any absolute fashion. On the one hand, the necessity to list meanings in a linear way is imposed by the basic linear structure of written text; on the other hand, graphic representations such as those used in this article (see Figure 1) might circumvent the greater part of the difficulties arising from the linearization problem. The fact, then, that such graphic representations are never used in dictionaries should probably be attributed partly to the force of lexicographical traditions, partly to economical reasons. It should not be forgotten, in fact, that most dictionaries are written under the pressure of financial constraints; with regard to his possible audience just like with regard to his financial sources, it is in-

Lexicographical treatment o/prototypical polysemy

333

dispensable for the lexicographer to keep the final costs down. As the incorporation of figures showing the conceptual relations among the senses of a word greatly enhances the required printing space, and hence, the cost of the dictionary, this may indeed be a reason for not using them. In this respect, it should be mentioned that computerized dictionaries open up new perspectives for dealing with prototypically clustered polysemy. Not only is storage space hardly a problem any more, but also, the access to the different meanings of an item becomes much more flexible. It is not difficult to imagine, for instance, how a graphic representation such as Figure I (together with explanatory glosses) would form the basic semantic information in a particular entry; further information at varying levels of specificity would then be retrievable by highlighting specific areas and points in the figure. But developing blueprints for changes in lexicographical practice is not the primary intention of this paper. Rather, as we are dealing with an analysis of existing practices, the question now arises whether the linearization problem is only created by the practical constraints just described, or whether the canonical linear order also reflects a particular conception of semantic structure. Is the linear dictionary model of semantic structure a direct consequence of the classical theory of categorization against which prototype theory reacts? Lakoff's propagandistic denunciation of the classical conception, which 'in the course of two thousand years ... has become so entrenched that we tend not to even notice its presence' (1982: 14) seems to suggest a positive answer to this question: if the classical theory has indeed been as universally dominant as Lakoff claims, it is natural to suppose that lexicographers have been influenced by it. However, as I have tried to show on a number of occasions, traditional conceptions of categorization are far from being as monolithically 'classical' as Lakoff believes. First, the prestructuralist historical-philological tradition of semantic research in linguistics is characterized by a fundamental methodological similarity with the contemporary cognitive approach (Geeraerts 1988a). Second, Kantian and early phenomenological conceptions of philosophical epistemology likewise correspond in basic respects with the view of human knowledge implied by the prototypical theory of categorization (Geeraerts 1985a). And third, clustered overlapping of lexical meanings is not in principle ruled out by the classical theory. A closer look at the Aristotelian sources of the classical conception reveals that what is new about prototype theory is not the discovery of clustered overlapping as such, nor the recognition that lexical categories as a whole may not be definable by means of a single set of necessary and sufficient characteristics, but rather

334

Lexicography

the fact that the overlapping is as a rule maximal and structural instead of minimal and incidental, and more importantly, the fact that applications that are intuitively considered to represent a single meaning, may not be analytically definable by means of a single set of necessary and sufficient attributes (Geeraerts 1987a). Given these nuances with regard to Lakoff's simplified view of the tradition, it is not entirely sure whether the canonical linear form of the dictionary is indeed a straightforward consequence of the classical theory of categorization. On the one hand, the classical expectation of minimal overlapping among lexical meanings will probably have exerted a certain degree of influence, but on the other hand, it is equally probable that lexicographers, through their extensive empirical familiarity with actual semantic structures, have been long aware of the difficulties surrounding the classical expectations. This view of things is strengthened by the fact that lexicographical manuals explicitly recognize these difficulties. It may be sufficient, for instance, to see how Zgusta's influential Manual oflexicography discusses the existence of peripheral and divergent uses of a word (1971: 47-49) to appreciate the fact that the classical view is not necessarily the lexicographically canonical one. And obviously, if the influence of theoretical conceptions on the linear order of dictionaries is indeed restricted, the material constraints mentioned earlier acquire more weight.

3. The structure of the problem In order to see how the representational constraints imposed by the linear order of the dictionary may be circumvented, we shall have to compare the lexicographical description of our example vers with an independent semantic analysis of that word. The latter may be found in Figure I, which gives a prototypical analysis of the nineteenth century material for vers that is available in the corpus of the Woordenboek der Nederlandsche Taal. The restriction to the nineteenth century material is motivated by the traditional attempt to avoid diachronic distortions in what is basically a synchronic analysis. Because prototype theory undermines the dichotomous distinction between diachrony and synchrony (cp. Geeraerts 1985d), the selection and description of a single synchronic period as a static unity is clearly an idealization, but it is a necessary one for the analysis to start at all. There is a further idealization to be mentioned: for the sake of simplicity, a small number of highly peripheral applications of the category has not been included in

Lexicographical treatment o/prototypical polysemy

335

the figure. As it stands, the figure sufficiently illustrates the prototypical character of the category. For reasons of succinctness, a justification of the analysis based on a detailed discussion and mutual comparison of the available quotations (which takes about thirty pages) is left out; by the same token, no illustrative quotations are given. Whereas the Arabic numerals refer to specific, 'low level' meanings of vers, the Roman numerals indicate larger, 'higher level' groupings of those basic applications.

,..... i

r-iii r--

ii

.....-y1:3

12

11

xxi14 17

5

I

~I~

~M

15

16

4 6 3

xii 7

a

10

18

19

~

20

21

9

Figure J. A prototypical analysis of the nineteenth-century material for vers avail-

able in the corpus of the Woordenboek der Nederlandsche Taal

336

ii iii iv v VI

vii (I)

(2) (3)

(4) (5)

(6)

(7)

(8)

(9) (10) (11)

Lexicography In an optimal condition: not characterized by decay consisting in the loss of certain attributes that are considered optimal Optimal in a functional sense; optimal for use; optimal for the function it has to fulfill in a particular context Optimal in a material sense; in an optimal condition with regard to its physical, material characteristics Optimal for consumption; in the full possession of the useful and/or pleasant attributes that motivate its consumption Recent, new; not having been present previously or for a long time in a particular context New as such; not having existed for a very long time New by contrast; distinct from members of the same category that are contextually implied or explicitly mentioned With regard to foodstuffs: recently produced and hence optimal for consumption as far as taste and nutritive value are concerned; recently harvested, caught, prepared etc., and hence not yet subject to decay With regard to foodstuffs: not preserved: not smoked, dried, salted, sugared, frozen etc. to prevent decay With regard to organic materials in general: recently harvested, picked etc., and hence not yet contaminated by a natural process of organic decay With regard to artifacts: only recently produced, purchased, taken into use etc., and hence not yet subject to wear or damage With regard to things that are inherently characterized by decay and contamination, viz. with regard to manure: recently produced and hence most powerful in its effects With regard to things that are inherently characterized by decay and contamination, viz. with regard to wounds: recently made and hence most painful With regard to organic materials: not yet contaminated by a natural process of organic decay, in contrast with similar things that are mentioned or implied in the context With regard to people or animals: physically in an optimal condition, in good shape, in contrast with other people or other animals that are mentioned or implied in the context With regard to artifacts: not subject to wear or damage, in contrast with other things that are mentioned or implied in the context With regard to foodstuffs: optimal for consumption, in contrast with other foodstuffs that are mentioned or implied in the context With regard to foodstuffs: optimal for consumption, regardless of their inherently or contextually recent character

Lexicographical treatment o/prototypical polysemy (12) (13) (14) (15)

(16) (17) (18) (19) (20) (21)

337

With regard to people: in an optimal physical condition, well rested, not tired, regardless of any contrast with other people With regard to abstract things (such as memory impressions): vivid, intense, powerful (regardless of recency) With regard to abstract things (such as memory impressions): having a recent origin, and hence vivid, intense, powerful With regard to abstract things (such as courage): intense, powerful, in contrast with other examples of the same category that are mentioned or implied in the context With regard to people: without moral stains, uncorrupted, in contrast with other people that are mentioned or implied in the context With regard to abstract things (such as one's conscience): recently come into existence, and hence still free of moral stains Recently present or available in a particular context, regardless of any contrast with other things, and regardless of inherent novelty Recent as such, inherently new, regardless of any contrast with other things Recent as such, inherently new, and hence distinct with regard to other things that are mentioned or implied in the context Recently present or available in a particular context, in contrast with other things that are mentioned or implied in the context, though regardless of inherent newness

In general, the method followed in drawing the figure is based on the two criteria mentioned at the end of the previous paragraph: on the one hand, there is the analytical attempt to define subsets of the observed applications of vers, on the other, there are intuitive observations as to the distinctions and similarities between those applications. For more details about the relationship between these two criteria for meaning discrimination, I refer once again to the previously mentioned article (l987c). Briefly, it may be argued that the basic discovery of prototype theory is the fact that what is intuitively recognized as a single meaning need not be so from an analytic, definitional point of view; for instance, the 'biological species' -reading of bird is intuitively recognized as a single lexical meaning of the word, in spite of the fact that there is no single definition in terms of necessary and sufficient properties to cover precisely that reading. That is to say, the reading in question is not ambiguous from an intuitive point of view (which may e.g. be demonstrated by applying the identity test described by Zwicky and Sadock 1975), but it is ambiguous from a definitional point of view (given the absence of a single necessary-and-sufficient definition). It should be noted that the converse relationship between the intuitive and the definitional criterion also

338

Lexicography

exists: autohyponymous terms such as dog are intuitively ambiguous according to the identity test, but definitionally univocal according to the necessity-cum-sufficiency criterion. Now, once it is accepted that semantic analyses are based on the interplay between both criteria, the representational dilemma facing the lexicographer in the case of vers may be analyzed in more detail. A linear representation of the semantic structure of vers might, in fact, enumerate either the specific meanings or the higher level groupings. Both alternatives, however, raise problems of adequacy. In its simplest form, the problem shows up in the following way. On the one hand, merely enumerating the higher level groupings obscures intuitively valid distinctions within those groups. On the other hand, listing only the lower level specifications ignores the fact that these may often be definitionally subsumed under a more general reading, and would thus have to be considered cases of vagueness rather than ambiguity according to the analytical criterion. In actual fact, the structure of the problem may be a bit more intricate, for instance in the sense that intuitive distinctness need not be restricted to the lower level applications, but may also apply to larger groupings. For expository purposes, however, we shall restrict the discussion to the problem in its simplest form. An illustration of the problem may then be found at the very centre of the category. To begin with, let us note that vers has roughly the same meaning as the English word fresh, except for the fact that the meaning 'moderately cool' is absent from the semantic structure of verso As in English, vers prototypically applies to foodstuffs; fresh fruit and vegetables are in an optimal condition for consumption because they have not been produced, harvested, picked etc. for a very long time. As can be seen in Figure 1, this central meaning is constituted by the overlapping of two concepts that may also occur independently: on the one hand, the notion 'in an optimal condition', on the other, the notion 'new, recent'. The intricacy of the semantic structure of the word derives largely from the fact that each of these two notions may receive further specifications and nuances which may then, of course, be combined in the overlapping area. Moreover, additional distinctions may occur within each of the areas determined in the manner just described. For instance, the most central meaning 'with regard to food: recently produced (etc.), and therefore optimal for consumption', which is represented by [1] in the figure, gives rise to the particular nuance represented by [2]. Here, vers means something like 'recently produced, and as such, not (yet) smoked, salted, dried, frozen etc.; that is to say, not having been subjected to a preserving treatment' .

Lexicographical treatment o/prototypical polysemy

339

From an intuitive point of view, [l] and [2] are distinct; something may be vers in the sense of [I] though not in the sense of [2]. The expression vers spek, for instance, is ambiguous. According to [2], it is a fixed expression for the notion 'pork, unpreserved meat from pigs', contrasting with gerookt spek and gezouten spek, 'bacon'. According to [I], however, vers spek might for instance simply mean 'recently purchased pork or bacon' (regardless of preservation). Similarly, it would not be contradictory to say het verse spek is niet vers meer (literally 'the fresh pork is no longer fresh') meaning that the pork has gone off. From an analytical, definitional point of view, however, the distinction between [1] and [2] is a case of referential vagueness rather than semantic ambiguity, as both concepts may be subsumed under the notion 'recently produced'. In fact, as I have tried to show in the 1987c article, the distinction between vagueness and ambiguity is analytically determined by the rule of thumb that the senses of a polysemous item should be defined as extensionally broad as possible. In this way, different shades of red do not constitute different meanings of red, as they can be subsumed under a general definition of the color in question; the distinction between the political reading of red and its interpretation as a color term does, on the other hand, represent a case of ambiguity rather than vagueness, since both readings cannot be subsumed under a general definition sufficient to distinguish red from other adjectives. Applying this criterion (which again may be traced to Aristotle) to vers does not only yield the conclusion that the distinction between [1] and [2] is not a case of ambiguity, but actually reduces the semantic analysis of the word as a whole to an enumeration of the largest groupings in Figure 1. Disregarding some of the peripheral applications that are not included in the figure (see however Figure 2), vers would then basically only have the meanings 'in an optimal condition' and 'recent, new', as these are the only ones that cannot be reduced to each other. It should by now be clear how the Iinearization problem arises. Linearly enumerating only low level specifications such as [1] and [2] is analytically inadequate, as they analytically represent cases of vagueness rather than truly semantic ambiguity (and it is, of course, ambiguity with which linguistics should be concerned, not vagueness). Linearly enumerating only high level groupings such as [i] and [v] is intuitively inadequate, as the intuitively valid distinction between cases such as [1] and [2] is obscured. (In fact, the whole prototypical area formed by the overlapping of [i] and [v] could never be listed separately, as it can always be subsumed under either of both.) Given, then, that a semantic analysis should indeed take into account ana-

340

Lexicography

lytical just as well as intuitive considerations, a simple linear representation is out of the question. What semanticians and lexicographers should do, is to describe the basic specifications together with the different higher level groupings to which they give rise. And this automatically implies a description of the clustered nature of those groupings.

4.

Circumventions of the Iinearization problem

The primary step to be taken by the lexicographer will obviously be the imposition of a hierarchical structure on the semantic material. This is a way of describing basic specifications together with higher level groupings that is still rather close to the 'classical' conception of semantic structures, at least in the sense that the ideal of rigid boundaries between different applications of a concept may be maintained. As shown in Figure 2 (which is a translation of part of the structural description of vers in the Woordenboek der Nederlandsche Taal) , the hierarchical description may reach considerable depth, containing up to 6 distinct levels. (It should be mentioned that the partial structure given in Figure 2 contains some of the minor nuances that were not incorporated into Figure 1.) The point to be made is, of course, that the neatness and the rigidity of the hierarchy do not adequately represent the actual semantic structure of prototypical concepts, since the larger groupings of basic specifications do not take the form of a taxonomy, but are rather characterized by, first, multiple overlapping, and second, uncertain boundaries. That is to say, the taxonomic model of semantic structure meets the prototypical model half-way, in that it recognizes the necessity to describe higher level groupings alongside basic specifications. On the other hand, it still implies that each of the hierarchical subsets is well-defined, and that there is no overlapping among them. How then can lexicographers remedy these defects of a linearly hierarchical model? In particular, how does the Woordenboek der Nederlandsche Taal cope with vers? In the first place, the overlapping between the different groups of meanings is explicitly indicated in the dictionary entry for vers by using crossreferences to other parts of the hierarchical structure. A prime example can be found at the highest level of the structure. As mentioned in the previous paragraph, the prototypical application of vers with regard to foodstuffs is constituted by the overlapping of the concepts 'in an optimal condition' and 'new, recent'. This fact appears to be mirrored in the structure of the dic-

Lexicographical treatment ofprototypical polysemy

341

tionary entry by the fact that I is explicitly described as the conjunction of the concepts represented by II and m. In the second place, the vers-entry in the Woordenboek der Nederlandsche Taal uses different means to indicate that each of the groupings in the hierarchical structure may have unclear boundaries in that it may harbor slightly deviant nuances. To begin with, it may be explicitly pointed out that a particular group of senses have a 'core meaning' in common (see, e.g., the definitions of II and m in Figure 2). Such a definitional technique is a warning for the dictionary user that the applications that are grouped together need not be strictly identical with regard to the attributes mentioned in the definition, but should rather be considered nuances of a central meaning (which is then given by the superordinate definition). The subordinate senses in the hierarchical structure are grouped together on the basis of similarity rather than strict identity with regard to the concept identified by their superordinate definition; the superordinate definition indicates the basis of the similarity that holds among the subordinate senses, not the basis of their identity. Within a taxonomical conception of a hierarchical structure, subordinate senses merely add attributes to the superordinate definition; within a prototypical conception, however, the subordinate specifications need not possess all the attributes identified by the higher level definition, in the sense that they may be related by similarity rather than taxonomic identity. This fact, then, may be lexicographically signaled by using specific definitional techniques, such as the explicit mentioning of the existence of a core meaning rather than an essentialist definition in the classical sense. Further, the non-taxonomical grouping of senses is achieved by using definitions that are themselves combinations of similar but distinct concepts. A very clear example is to be found at I-A-l in Figure 2. The use of the and/or-specification, the presence of an optional element in the definition (the one between brackets), and the incorporation of an open-ended enumeration clearly mark the definition as a disjunctive rather than an essentialist one. Again, what lies at the basis of the grouping is a similarity that allows for variation and deviation rather than a taxonomical relationship that implies essentialist identity. And again, non-classical definitional means are used to convey this fact to the dictionary user. Finally, there are structural next to definitional means for indicating that hierarchically grouped senses may be related rather than taxonomically identical. In fact, their relatedness is not always based on similarity (as may have been suggested by the foregoing discussion). For instance, as witnessed by the lowest level in the partial hierarchical structure as presented in Figure 2,

342

Lexicography VERS

I) In applications in which the notions 'new' (cp.lI) and 'in optimal condition' (cp. Ill) occur in combination

11) In application with the core meaning 'in an optimal condition, not characterized by decay consisting ofthe loss of certain optimal characteristics that the thing in question possesses'

Ill) In application with the core meaning 'new, recent', either in the sense 'not having come into existence or having appeared for a very long time yet', or in the sense' other, distinct from what was mentioned or implied before'

A) In the central application, with regard to foodstuffs

B) In applications that link up with the specification 'not having come into existence of having appeared for a long time yet' of the basic notion 'new': because of its recent character not yet subject to the loss of its original, optimal characteristics

C) In applications that link up with the specification 'other, distinct' ofthe basic notion 'new': distinct from what was mentioned or implied before, viz. because of its better condition

I) Recently produced and hence, through its taste and/or nutritive value, (pre-eminently suited for consumption); recently harvested, caught, killed, perpared etc., and hence not subject to decay

2) Through specialization of I): unpreserved, not having been subject to a preserving treatment; not smoked, dried, salted, sugared, frozen etc.in order to prevent contamination and decay

a) In general

b) In specialized applications

10 Often with regard to fish (viz. herring), and then more particularlyas a technical term in the fishing industry: not salted

2° Verse worst: unsalted and unsmoked sausage

3° Verse waar: garbage from slaughtered animals, meat from animal organs, specifically when sold as food for dogs or cats

I

- Metonymically with regard to industrial fishing: concerned with fish that is not meant to be salted or otherwise preserved, but that is meant to be brought directly to the market

Figure 2. Translation of part of the structural description of vers in the Woordenboek der Nederlandsche Taal

Lexicographical treatment ofproto~ypical polysemy

343

conceptual nuances that are structurally grouped together may be related through metonymy rather than similarity. Within the group of applications 1A-I-b-I o, the metonymical nuance is structurally singled out: on the one hand, its close relationship to the application defined by I-A-I-b-Io is indicated by its subordinate status with regard to the latter, but on the other hand, the fact that it is a metonymical extension that is not exactly identical with I-A-I-b-Iois made apparent both by its distinct structural position within that group, and by its being explicitly labeled as metonymical. In this sense, the subordinate position is not an indication of a taxonomical relationship, but rather of a close relationship that is yet to be distinguished from true identity. There is ample evidence, in short, that lexicographers use hierarchical structures in a non-taxonomical way; various provisions are made to represent the indeterminacy and overlapping of the hierarchical groupings. To round off the discussion, notice that Figure 2 contains an example where both structural characteristics are interwoven. The definition under III is one in which a core meaning is explicitly identified, while at the same time, the two major nuances of that core application are mentioned; as such, it is made clear that the grouping of meanings under III (the structure of which is not represented in detail in the figure) is based on similarity rather than strict identity. But as I is constituted by the overlapping of 11 and Ill, the main nuances of III should also be identified within I. This is indeed the case: I-B and I-C contain an explicit cross-reference to the main nuances that are grouped together under Ill.

5. Conclusions

There are a number of points that may be retained from the foregoing discussion. Basically, there is no reason to identify the classical conception of categorization with an allegedly lexicographical model of definitional structures. An examination of an actual dictionary entry reveals that neither a classically linear or a semi-classical taxonomic model is followed. Rather, a hierarchical structure is used to accommodate the tension between analytical and intuitive distinctness of meaning, i.e., to incorporate both basic specifications and higher level groupings into the entry. At the same time, the hierarchical structure is modified in several ways in order to bring out most clearly the indeterminate and clustered nature of those groupings. It follows from this observation that prototype semantics is well suited as a theoretical

344

Lexicography

basis for a lexicographical metatheory, since it accurately models the kind of semantic phenomena that lexicographers have to face up to in their pragmatically restricted descriptive activities. In this respect (and given the results mentioned at the beginning of the article), a further rapprochement between the recent, cognitive developments in linguistic semantics on the one hand and lexicography on the other may be envisaged. For one thing, lexicographical metatheories had better abandon all hope of building an adequate theory of lexicography on a linguistic basis that does not conform with their own empirical experience; in particular, extant attempts to impose a structuralist conception of semantic structure on the methodology of lexicography should be rejected as misguided: there are now simply much more suitable semantic theories available. But the rapprochement could not be unidirectional: it would seem that prototype semantics might profit a lot from a closer acquaintance with the detailed descriptions of prototypical structures contained in large-scale traditional dictionaries such as the Woordenboek der Nederlandsche Taal.

Chapter 14 The definitional practice of dictionaries and the cognitive semantic conception of polysemy

Originally published in Lexicographica 2001,17: 6-21. This chapter puts the previous one in a wider perspective. First, while the previous paper focuses on a specific type of dictionary (the large-scale diachronic dictionary of the Oxford English Dictionary type), the present takes its examples from common desk dictionaries. Second, while the previous chapter deals with the relationship between prototypicality and lexicographical definitions only, the present refers to other lexicographically relevant aspects of Cognitive Semantics as well. And third, while the previous chapter basically addresses an audience of linguists, trying to persuade them of the interest of lexicographical data for semantic theorizing, the present paper takes the reverse perspective, presenting relevant aspects of Cognitive Semantics to an audience of lexicographers who are not necessarily acquainted with recent developments in theoretical semantics. (Section 2 of the paper, then, will be quite familiar for readers who have progressed up to this point in the present volume.) In the years between the initial publication of the article and the present publication, the situation has changed drastically with regard to at least one of the desiderata mentioned at the end of the article: in general, the use of corpus data in Cognitive Linguistics has increased significantly (see Tummers, Heylen, and Geeraerts 2005 for a critical overview), and more specifically, the use of existing collocational techniques has been popularised through the 'collostructional' approach of Gries and Stefanowitsch (among many other publications, see Stefanowitsch and Gries 2003, Stefanowitsch and Gries 2003).

1.

Gist of the argument

Cognitive Semantics has had a major impact on lexical studies: more than any other recent theory, it has in the last fifteen years led to a renewed interest in lexical research. But what about lexicography? What is the relationship between Cognitive Semantics and lexicography? And specifically, in the context of the present issue of this journal, what is the impact of Cognitive Semantics on the lexicographical treatment of polysemy? In what follows, I will try to answer that question - succinctly and perhaps a bit sketchily, but

346

Lexicography

also as fair as possible to both the theoretical, lexicological and the practical, lexicographical part of the comparison. I emphasize my attempt at fairness, because I deliberately want to avoid any form of theoretical imperialism. It is definitely not an assumption of this paper that theoreticians in principle have the answers, and that lexicographers simply have to follow. I have pointed out earlier (Geeraerts 1997a: 5) that there should be a relationship of mutual inspiration between both disciplines, and even though the present paper looks from lexicological theory to lexicographical practice rather than the other way round, I still hold the reverse perspective to be equally important. In general, I will argue that a number of existing definitional and descriptive practices in the dictionary that are somewhat suspect from an older theoretical point of view receive a natural interpretation and legitimacy in the theoretical framework offered by Cognitive Semantics. More specifically, there are three aspects of the Cognitive conception of lexical semantic structure that have to be discussed: the importance of prototypicality effects for lexical structure, the intractability of polysemy, and the structured nature of polysemy. I will argue that each of these points inspires a specific conclusion for lexicographical practice, or at least, that it vindicates existing aspects of lexicographical practice. First, the importance ofprototypicality effects for lexical structure blurs the distinction between semantic information and encyclopedic information. This does not entail that there is no distinction between dictionaries and encyclopedias as types of reference works, but rather that references to typical examples and characteristic features are a natural thing to expect in dictionanes. Second, the intractability ofpolysemy involves the absence of a coherent set of criteria for establishing polysemy; a more charitable way of wording things would be to say that distinctiveness between senses of a lexical item is to some extent a flexible and context-based phenomenon. Dictionaries, then, will use various definitional techniques to accommodate the flexibility of meamng. Third, the structured nature of polysemy involves, basically, the radial set structure of polysemy. While lexicography has certainly never denied the existence of links between the various readings of a lexical item, Cognitive Semantics has added a number of new insights: the clustered nature of polysemic structures is now being analyzed in more detail than ever. For lexicography, this implies a recognition of the linearization problem that traditional dictionaries face.

Dejinitional practice ofdictionaries

347

Given these three focal points, the paper has a more or less deductive structure. I will first present the facets of Cognitive Semantics that I want to focus on. (I will not, however, try to present an overall introduction to the Cognitive approach: see Taylor 1995, Ungerer and Schmid 1996, Palmer 1996, Violi 1997, Dirven and Verspoor 1998 for introductory volumes. The discussion in section 2 will be largely superfluous for readers familiar with Cognitive Semantics.) Next, I will identify the specific expectations with regard to lexicographical practice that may be deduced from this theoretical analysis, and then proceed to show that these predicted features are indeed part and parcel of actual lexicographical practice - in spite of what might be expected on the basis of other theoretical approaches to semantics. The paper closes with an attempt to place these observations in a wider context the relationship between lexicography and Cognitive Semantics is not exhausted by the discussion of polysemy.

2.

A primer of Cognitive Semantics

What are the structural characteristics of semasiological structures highlighted by Cognitive Semantics? A good starting-point to define some of the topics central to Cognitive Semantics is provided by the well-known distinction between the level of senses and the level of referents (in logic-semantic parlance, between intension and extension). Consider the word fruit. This is a polysemous word: next to its basic, everyday reading ('sweet and soft edible part of a plant, containing seeds'), there are various other readings conventionally associated with the word. In a technical sense, for instance ('the seed-bearing part of a plant or tree'), the word also refers to things that lie outside the range of application of the basic reading, such as acorns and pea pods. In an expression like the fruits of nature, the meaning is even more general, as the word refers to everything that grows and that can be eaten by people (including, for instance, grains and vegetables). Further, there is a range of figurative readings, including the abstract sense "the result or outcome of an action' (as in the fruits ofhis labor or his work bore fi'uit), or the somewhat archaic reading 'offspring, progeny' (as in the biblical expressions thefruit ofthe womb, thefruit o(his loins) Each of these readings constitutes a separate sense offruit, but in turn, each sense may be thought of as a set of things in the outside world. The basic sense of fruit, for instance, corresponds with a set including apples,

348

Lexicography

oranges, and bananas (and many other types of fruit). If you think offruit in this central sense as a category, the set consists of the members of the category. These members are 'things' only in a broad sense. In the fruit example, they happen to be material objects, but in the case of verbs, they could be actions, or situations, or events; in the case of adjectives, they could be properties; and so on. Also, the 'things' featuring in the set need not exist in the real world. The set contains all real and imaginary apples and oranges (etc.) that fruit could possibly name, in the same way in which goblin will have a set of members associated with it, regardless of whether goblins are real or not. Given the distinction between the intensional and the extensional level of semasiological analysis, we can now describe three structural characteristics that receive specific attention within a Cognitive Semantic framework. 2.1. Differences of structural weight Differences in salience involve the fact that not all the elements at one level of analysis have the same structural weight. On the semantic level, for instance, the everyday reading of fruit occupies a more central position than the archaic reading 'offspring' or the technical reading. Various indications may be adduced for this central position. For one thing, the central reading more readily springs to mind when people think of the category: on being asked what fruit means, you are more likely to mention the edible parts of plants than a person's offspring. For another, the 'edible part' reading is more frequent in actual language use. In addition, the 'edible part' reading is a good starting-point for describing the other readings. It would probably be more easy to understand the expression frUit of the womb (if it is new to you) when you understand the 'edible part' reading than the other way round. The basic reading, in other words, is the center of semantic cohesion in the category; it holds the category together by making the other readings accessible. Three features, in short (psychological salience, relative frequency of use, interpretative advantageousness), may be mentioned as indications for the central position of a particular reading. Centrality effects are not restricted to the level of senses, however, but may also be invoked at the referential level. When prompted, Europeans will more readily name apples and oranges as types of fruit than avocados or pomegranates, and references to apples and oranges are likely to be more

Dejinitional practice ofdictionaries

349

frequent in a European context than references to mangos. (This does not exclude, to be sure, cultural differences among distinct parts of Europe.) The terminology used to describe these differences of structural weight is quite diverse, and the description in the foregoing paragraphs has featured such (intuitively transparent) terms as salience, typicality, and centrality. The most technical term however is prototypicality: the central reading of an item or the central subset within the extensional range of a specific reading is the prototype. The linguistic literature on prototypes is by now vast. Apart from the introductory works mentioned above, see Mangasser-Wahl (2000) for an interesting overview of the development of the approach. 2.2. Demarcation problems The elements at one particular level of the semasiological analysis need not necessarily be clearly distinguishable with regard to each other. As an illustration, let us consider the question whether the central sense offruit can be delimited in a straightforward fashion. Such a delimitation will take the form of a definition that is general and distinctive: it is general in the sense of naming characteristics that are common to all fruits, and it is distinctive in the sense of being sufficient to distinguish the category fruit (in the relevant sense) from any other category. (If a definition is not distinctive, it is too general: it will cover cases that do not belong in the category to be defined.) Now, many of the characteristics that one might be inclined to include in a definition of the central reading offruit do not have the required generality: they are not necessarily sweet (lemons), they do not necessarily contain parts that are immediately recognizable as seeds (bananas), they are not necessarily soft (avocados). There are, to be sure, a number of features that do have the required generality: all fruits grow above the ground on plants or trees (rather than in the ground); they have to ripen before you can eat them, and if you want to prepare them (rather than eat them raw), you would primarily use sugar, or at least use in them in dishes that have a predominantly sweet taste. Taken together, however, these features do not suffice to prevent almonds (and other nuts), or a vegetable like rhubarb (which is usually cooked with sugar), from being wrongly included into the category that is to be defined. We have to conclude, then, that the central sense ofjruit cannot receive a definition that is both general and distinctive. If we shift the attention to the referential level, similar effects may be observed: the borderline of categories

350

Lexicography

is not always clearly delineated. For instance, is a coconut or an olive a fruit? Definitional difficulties like the one just illustrated are enhanced by the existence of various kinds of tests for distinguishing between vagueness and polysemy - and specifically, by the fact that the existing tests may yield results that are to some extent divergent with regard to each other. To briefly illustrate the main point, and without discussing all specific tests that have been suggested, three types of criterion can be distinguished. First, from the truth-theoretical point of view taken by Quine (1960,129), a lexical item is polysemous if it can simultaneously be clearly true and clearly 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). Second, lingUistic tests involve acceptability judgments about sentences that contain two related occurrences of the item under consideration (one of which may be implicit or deep-structural); if the grammatical relationship between both occurrences requires their semantic identity, 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 transformations 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.) 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 dejinitional criterion (as informally stated by Aristotle in the Posterior Analytics II.xiii) says that an item has more than one lexical

Dejinitional practice ofdictionaries

351

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 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 things. Now, the existence of various polysemy tests is non-trivial to the extent that they need not always (in contrast with the port example) yield the same results. In the case of autohyponymous words, for instance, the definitional approach does not reveal an ambiguity, whereas the Quinean criterion does. In fact, given that dog is autohyponymous between the readings 'Canis familiaris' and 'male Canis familiaris', the latter definition is not maximal, because it defines a proper subset of the Canis familiaris' reading; the sentence Lady is a dog, but not a dog, on the other hand, is not ruled out. Such divergences between polysemy tests occur on a larger scale: see Geeraerts (1993a), and the further discussion in Tuggy (1993), and specifically Cruse (2000a). Following up on the lead provided by Geeraerts (1993a), Cruse (2000a) systematically explores various configurations of divergence and convergence of the polysemy tests. From the point of view of the present article, this line of discussion is important because it lends further support to the recognition that there is not necessarily a unique and optimal solution to drawing dividing lines around and between the meanings of a lexical item. 2.3. Multidimensional structural relations The relationship that exists between the various elements at each level of the analysis is not restricted to the quantifiable phenomena described in section 2.1: the links between those elements may also be described in a more qualitative way. On the level of senses, in particular, it appears that the relationship between the meanings of a word may be described in terms of a more or less limited set of basic conceptual links. The senses ofjruit, for instance, do

352

Lexicography

not exist in isolation, but they are related in various ways to the central sense and to each other. The technical reading (,seed-containing part') and the sense illustrated by the fruits ofnature are both related to the central meaning by a process of generalization. The technical reading generalizes over the biological function of the things covered by the central meaning, whereas the meaning 'everything that grows and that can be eaten by people' focuses on the function that those things have for human beings. The figurative uses, on the other hand, are linked to the other meanings by a metaphorical link, but notice also that the meaning 'offspring' is still closer to the central sense, because it remains within the biological domain. The overall picture, in short, takes the form of a cluster of mutually interrelated readings. This observation is, of course, a familiar and time-honored one in lexical semantics: the terminology used to describe the links among senses originated with diachronic semantics in the late nineteenth century, with the very birth of lexical semantics as a separate subdiscipline of linguistics. What is new in Cognitive Semantics, though, is the emphasis on the overall structure of the related meanings rather than on the individual links: an emphasis on the multidimensional nature of the overall structure, and an emphasis on the cohesive role of prototypical centers within such structures. In fact, multidimensional analyses of the semantic structure of lexical items are a common feature of Cognitive Semantics: they are a crucial feature of the radial set model of semantic structure that has become popular through the work of Brugman (198 I) and Lakoff (1987) and many others. Apart from prototype theory as described above and the conceptual metaphors introduced by Lakoff and Johnson (1980), the radial set model of semantic description may indeed be the most widely known feature of Cognitive Semantics. 2.4. Summarizing the position of Cognitive Semantics The semasiological model arising from the previous pages may be summarized in the graphical format of Figure 1. Without in any way being exhaustive, the picture shows how a word like fruit can, on a first level of analysis, be associated with various senses. On a second level of analysis, each of those senses is itself associated with a set of referents. These sets are represented in a form that resembles the representation with Venn-diagrams that is usual in mathematics. Examples of entities at the referential level are included only in the set associated with the sense 'edible part'. This is, of course, a matter of graphical economy rather than principle. At each level, specific structural characteristics have to be taken into account. Of the three

Dejinitional practice ofdictionaries

353

basic characteristics mentioned above, two have received a graphical expression in the picture. The differences of centrality and structural weight among the elements at each level are indicated by drawing them in different sizes and by topologically ordering them in a way that reflects the dine from center to periphery. The structured nature of the relations between the elements is indicated by an explicit identification of the relevant links. (Again for reasons of graphical economy, this is restricted to the level of senses.) fruit

seed-bearing part of plant ..

GENERALIZATION

edible part of plant

METAPHOR

•

result of an action

apple, orange, pear, banana

pomegranate Figure I. The semantic structure offruit

We have now reached the point where we can summarize the specific approach taken by Cognitive Semantics in just a few points. First, it was suggested above that there is an outspoken homology between the structure of the referential level and the structure of the semantic level: the same structural characteristics shape both levels. This suggests that both levels are less far apart than has traditionally been thought. The structuralist view of lexicology in particular tends to suggest that only the

354

Lexicography

semantic level (the level of senses) is worthy of linguistic analysis. By contrast, the so-called Cognitive Semantic conception that has meanwhile arisen in opposition to this earlier structuralist view, stresses the fact that the referentiallevel has to be included in the analysis. Second, while structuralist approaches to semantics tend to be reluctant to take into account differences of structural weight and demarcational fuzziness, Cognitive Semantics readily accepts these phenomena as relevant aspects of semantic structure. And third, linking up with prestructuralist semantics, Cognitive Semantics puts a new emphasis on the multidimensional, clustered nature of semasiological structures.

3. From theory to practice Now, what would be the consequences for lexicographical practice? Or rather, if the Cognitive conception of semantic structure is by and large correct, what could we expect to find in actual dictionaries? The three characteristics highlighted in the previous paragraph lead to the following hypotheses. To begin with, if it is correct that the referential level of semantic structure is part and parcel of a proper semantic description, we may expect dictionaries to include references to that level - in spite of the traditional, strict distinction between the semantic and the encyclopedic level of description. In particular, we may expect dictionaries to refer to prototype instances of categories or to typical (rather than general) features of the members of those categories. Further, if it is correct that the description of meaning has to come to terms with fuzziness, demarcation problems, and non-uniqueness, we expect dictionary definitions to use definitional methods that take into account these characteristics. Instead of definitions that rigidly take the form of separately general and mutually distinctive features, we expect the intrusion of unorthodox definitional methods such as enumerations, disjunctions, and the accumulation of near-synonyms. And finally, if it is correct that semantic structures predominantly take the form of a multidimensional radial set structure, we may expect dictionaries to face a linearization problem: how can the multidimensional nature of the semantic structures be mapped onto the linear order of the dictionary?

Dejinitional practice ofdictionaries

355

In the following subsections, these expectations will be confronted with actual examples. It will be shown that the expectations are basically correct. 3.1. Pratotypicality effects in lexical structure Consider the following definitions (of separate meanings or idiomatic expressions) fram the New Shorter Oxford English Dictionary (cd-ram version, 1997). abiogenesis The production of organic matter or compounds, other than by the agency of living organisms; esp. the supposed spontaneous generation of living organisms. baritone A I The male voice between tenor and bass, ranging typically from lower A in the bass clef to lower F in the treble clef; a singer having such a voice; a part written for such a voice. cup b An ornamental vessel, typically of silver and comprising a bowl with a stem and base, that is offered as a prize in a competitive event. defoliate Remove the leaves from; cause the defoliation of, esp. as a military tactic. dwarf A I b Any of a mythical race of diminutive beings, typically skilled in mining and metalworking and often possessing magical powers, figuring esp. in Scandinavian folklore. hear! hear! An exclam. calling attention to a speaker's words, e.g. in the House of Commons, and now usu. expressing enthusiastic assent, occas. ironical derision. heart 5 A central part of distinct confonnation or character, e.g. the white tender centre of a cabbage, lettuce, etc. honours of war Privileges granted to a capitulating force, e.g. that of marching out with colours flying. model 2 a (fig.) A person or thing resembling another. esp. on a smaller scale. tea 5 A meal or social gathering at which tea is served. Now esp. (a) a light afternoon meaL usu. consisting of tea, cakes, sandwiches, etc. (also more fully afternoon tea, five o'clock tea); (b) (in parts of the UK, and in Australia and NZ) a main meal in the evening that usually includes a cooked dish, bread and butter, and tea (also more fully high tea)

356

Lexicography

tee A conical metallic structure, usually hung with bells, surmounting the pagodas of Myanmar (Burma) and adjacent countries. thimblerig A sleight-of-hand game or trick usually played with three inverted thimbles and a pea, the thimbles being moved about and bystanders encouraged to place bets or to guess as to which thimble the pea is under. In each of these definitions, words such as especially, e.g., typically, usually and often introduce descriptive features that are not general but that rather identify typical (prototypical, if one likes) characteristics or instances of the category. Within a structuralist conception of semantics, this would be inadmissible, because these elements belong to the 'encyclopedic' level rather than the semantic level. In actual practice, however, this prototypeoriented definitional technique can hardly be called exceptional in the context of the dictionary as a whole. The expression esp., for instance, is used no less than 28335 times in 18274 entries in the dictionary as a whole. Does this mean, by the way, that the difference between dictionaries and encyclopedias is a spurious one? The question asks for a brief excursion. An early discussion of the question between Haiman (1980a) and Frawley (1981), with a further reply by Haiman (1982), provides a good startingpoint for delimiting the Cognitive point of view (for a more recent discussion of the theoretical question, see the contributions in Peeters 2000). On the one hand, the theoretical basis for a distinction between dictionaries and encyclopedias cannot be provided by the structuralist approach (as in Lara 1989): it is a crucial aspect of Cognitive Semantics that the distinction between the two levels of description is not as strict as presupposed by the structuralist doctrine. On the other hand, there is a practical difference between dictionaries and encyclopedias that need not be abolished: there is a difference in scope and content between, say, the Encarta or the New Shorter Oxford English Dictionary, or between the Encyclopaedia Britannica and the Oxford English Dictionary, and no cognitive linguist would argue against the distinction. This distinction basically resides in two features. Macrostructurally, the encyclopedia focuses on proper names, nouns, and maybe a number of other elements from open word classes, whereas the dictionary includes all word classes (typically excluding all or most proper names). Microstructurally, the encyclopedia focuses on expert information as provided by scientific, technical, or professional experts, whereas that information is only one of the types of semantic description that the dictionary may include, together with the more everyday uses of the words.

Definitional practice ofdictionaries

357

But if Cognitive Semantics accepts this distinction, how can it justify it? As a theoretical background for the distinction between the type of information typically included in encyclopedias and that included in dictionaries, we need a 'sociosemantic' theory: a theory about the distribution of semantic knowledge within a linguistic community. Scientific, technical, professional information is, in fact, primarily information that is produced and certified by a specific group of people - the experts, who are recognized by the community as such and on whom the community relies when expert knowledge is at stake. Although no such 'sociosemantic' theory is as yet available with any reasonable degree of comprehensiveness, a starting-point is provided by Putnam's theory of the 'division of linguistic labor' (1975), which explicitly distinguishes between extensional concepts (the expert's knowledge) and stereotypes (the basic semantic knowledge that language users are supposed to possess if they are to count as full-grown members of the linguistic community). A combination of Putnam' s approach with prototype theory is not impossible (see Geeraerts 1985b, 1987b): if a prototypically organized concept combines all the various nuances with which a lexical item may be used within a linguistic community, then extensional and stereotypical concepts are particular members of the full prototypical set of applications of an item. Extensional concepts are characterized by their expert nature, whereas stereotypes represent the minimal amount of semantic knowledge that the language user is supposed to possess if he is to count as mastering the language. Roughly speaking, stereotypes are likely to coincide with the most common, most central senses within a prototypical cluster: what people are primarily supposed to know are the central readings of the cluster. This recognition of a possible theoretical combination of prototype theory and a theory of the division of linguistic labor yields a theoretical framework for reference works that naturally provides a place for both the encyclopedia and the dictionary (see Geeraerts 1985b, 1987b). In fact, three basic types may be distinguished. First, technical, professional, scientific expert knowledge is treated in encyclopedias and terminological dictionaries. Second, the full prototypically organized set of senses of a lexical item, including nuances and less frequent or more specialized readings, is treated by large-scale dictionaries, of the size represented by (to name just a few) the New Oxford Dictionary of English or Merriam Webster's Collegiate Dictionary, and any dictionary beyond that size.

358

Lexicography

And third, standard desk dictionaries can be related to the notion of stereotype: they make a selection from the full prototypical set by presenting only the most central, most frequent senses. Closing the excursion, we may conclude that a Cognitive Semantic conception of the relationship between semantic and encyclopedic knowledge does not preclude a theoretical justification for the distinction between dictionaries and encyclopedias as different types of reference works. 3.2. The intractability of polysemy Definitional demarcation problems show up in the fact that dictionaries appear to use definitional techniques that are 'unorthodox' from the point of view of a traditional conception of meaning. Consider the following set of entries, again from the New Shorter Oxford English Dictionary. (The entries are rendered in a reduced form: etymologies, quotations, dates, and a number of labels have been left out.) primer I 1 A prayer-book or devotional manual for the laity. 2 An elementary textbook (orig. a small prayer-book) used in teaching children to read. b A small introductory book on any subject; fig. something introducing or providing initial instruction in a particular subject, practice, etc.. c (A child in) an elementary class in a primary school. 3 A size of type. Chiefly & now only in great primer, long primer. primer 11 I a = priming-wire. b A cap, cylinder, etc., containing a compound which responds to friction, electrical impulse, etc., and ignites the charge in a cartridge etc.2 A substance used as a preparatory coat on previously unpainted wood, metal, canvas, etc., esp. to prevent the absorption of subsequent layers of paint or the development of rust.3 A person who primes something. 4 Aeronaut. A small pump in an aircraft for pumping fuel to prime the engine. 5 a Biochem. A molecule that serves as a starting material for a polymerization. b Zoo/. & Physio/. A pheromone that acts initially on the endocrine system, and is thus more general in effect than a releaser. primer III I First in order of time or occurrence; early; primitive. 2 First in rank or importance; principal, chief.

Dejinitional practice ofdictionaries

359

In almost half of the fourteen senses or subsenses presented here, we fmd definitional techniques that would seem to be inadmissible if one assumes that meanings have to be defined in terms of necessary and sufficient, general and distinctive characteristics. To begin with, we find disjunctions in I 1 ('A prayer-book or devotional manual for the laity'), in I 2b ('something introducing or providing initial instruction in a particular subject, practice, etc. '), in 11 2 (' A substance used as a preparatory coat on previously unpainted wood, metal, canvas, etc., esp. to prevent the absorption of subsequent layers of paint or the development of rust'), in III I ('First in order of time or occurrence'), and in III 2 ('First in rank or importance'). From a traditional point of view, disjunctions are barred from definitions, because they fail to capture the common aspects of the category to be defined. In a similar way, open-ended enumerations should be avoided: they may illustrate or partially demarcate a category, but they do not define it, if you assume a rigid conception of definitions. In the examples, however, quite a number of open-ended enumerations appear: in I 2b ('something introducing or providing initial instruction in a particular subject, practice, etc. '), in 11 1 a CA cap, cylinder, etc., containing a compound which responds to friction, electrical impulse, etc., and ignites the charge in a cartridge etc.'), in 11 2 CA substance used as a preparatory coat on previously unpainted wood, metal, canvas, etc. '). Finally, we may note that the juxtaposition of near-synonyms is yet another way of loosening up the definitions. In the example III 1, the nearsynonyms early and primitive do not have exactly the same meaning (what is early is not necessarily primitive, and vice versa). At the same time, they add something to the analytical definition; in particular, the near-synonym primitive adds a nuance of lack of sophistication that is not explicit in the definition 'First in order of time or occurrence'. Lexicographical practice, in short, appears to be in accordance with the lexicological observation that the distinction between meanings need not be clear-cut. This fact has not escaped the lexicographers themselves, to be sure: among others, see Ayto (1983), Stock (1983), Hanks (1994). In the neighboring field of computational lexicography, similar voices may be heard: Kilgarriff (1997). 3.3. The structured nature of polysemy Let us consider the first seven senses of the adjective fresh in the Oxford English Dictionary, 2nd edition. (In the overview below, the definitions are

360

Lexicography

sometimes rendered only partially. Some meaning nuances have been left out.) fresh I New, recent 1. a. New, novel; not previously known, used, met with, introduced, etc. b. In weaker sense: Additional, another, other, different, further. 2. Recent; newly made, recently arrived, received, or taken in. 3. Making one's first acquaintance with a position, society, etc.; raw, inexperienced; unsophisticated, 'green'. n. Having the signs of newness. 4. Of perishable articles of food, etc.: New, in contradistinction to being artificially preserved; (of meat) not salted, pickled, or smoked; (of butter) without salt; (of fruits, etc.) not dried or preserved in sugar or the like 5. Of water: Not salt or bitter; fit for drinking. 6. Untainted, pure; hence, possessed of active properties; invigorating, refreshing. Said esp. of air 7. Retaining its original qualities; not deteriorated or changed by lapse of time; not stale, musty, or vapid. The article exhibits a linear ordering of the meanings, with a higherorder, taxonomical structure of three levels. Even a cursory inspection of the defInitions reveals that the hierarchical ordering does not make explicit all the relations that exist among the different senses. In the fIrst place, the senses 1-3 within group I are related by similarity, with sense 1 probably as the prototypical center of the group. Roughly, sense I can be paraphrased as 'new according to the perspective of a beholder'. Sense 2 is 'new as such, newly produced'. Sense 3 may receive the paraphrase 'new in a specifIc context, new in a given position or function'. The senses within group II are likewise related by similarity, but 7 seems to be a more encompassing one than the others: if 7 is paraphrased as 'retaining its originally optimal character', then both the 'pure and strong' reading of 6 and the 'optimal for consumption, still in possession of all its nutritional value' reading of 4 are specializations of7. Sense 5 'fIt for drinking', on the other hand, belongs together with 5. In short, the linear order within group I and within group II does not have an identical value, or at least, the semantic relations within each group are more specifIc than can be expressed by a mere linear ordering. In the second place, the relationship between group I and group II is a metonyrnical one: having the features of newness is a causal result of being new, in whatever sense. However, such a metonyrnical relationship also

Dejinitional practice ofdictionaries

361

appears within group I. The nuance 'raw, inexperienced, unsophisticated' that appears after the colon in definition 3 is as much a 'sign of newness' in sense 3 as the meanings 4-7 are signs of newness in the sense defined by 2. We see, in other words, that the same type of relationship is not always treated in the same way. This also holds for the relationship of semantic specialization that links 7 to 4, 5, and 6. Notice, in fact, that reading lb is a semantic specialization of Ia. The things that are fresh in Ib are not just novel from the point of view of the beholder, they are novel in comparison with a set or series of similar things. All in all then, the semantic structure of the item is a multidimensional one. A further, more detailed analysis would undoubtedly reveal more dimensions, but at this point, it may be sufficient to take into account the three dimensions that came to the fore in our cursory analysis: the relationship of similarity between 1, 2, and 3; the relationship of specialization that exists between 7 and 4,5,6 on the one hand and between la and lb on the other; and the metonymical relationship between 2 and 7, and between 3 and 3' (where 3' refers to the reading 'raw, inexperienced, unsophisticated'). The point, to be sure, is not that the linear order in the Oxford English Dictionary should be condemned as an inadequate rendering of the underlying semantic structure. The point is rather that any traditional form of linear ordering cannot do full justice to the multidimensional nature of semantic structures. In an earlier article (Geeraerts 1990a), I called this the lexicographical linearization problem: the fact that lexicographers compiling traditional dictionaries have to project a multidimensional, clustered semantic structure onto the linear order of the dictionary. In that article, I presented a detailed analysis of the word vers (the Dutch counterpart of English fresh) and its treatment in the Woordenboek der Nederlandsche Taal (the Dutch counterpart of the Oxford English Dictionary), and analyzed the various mechanisms (like hierarchical groupings, labels, and cross-references) that lexicographers may employ to circumvent the problem. The main point then, as now, was not a practical but a theoretical one: if the linearization problem is indeed a recurrent problem for practical lexicography, then a lexicographical metatheory had better start from a linguistic theory that explicitly recognizes the underlying semantic multidimensionality.

362

Lexicography

4. The wider picture The discussion in the previous pages suggests that the conception that Cognitive Semantics has of polysemy and semantic structure is consonant with the actual practice of dictionaries. I deliberately use the word suggests, because the few examples taken into consideration here could hardly tell the whole story. Even so, what Cognitive Semantics seems to offer to lexicography is a conception of semantic structure that is perhaps in a number of respects more realistic than what many other semantic theories (in particular, theories of a structuralist persuasion) can provide. This recognition does not, however, exhaust the interaction between Cognitive Semantics and lexicography. There are at least three further points that should be mentioned to put the present contribution in a wider context. First, the previous discussion was restricted to the way in which Cognitive Semantics encompasses a theoretical perspective that so to speak vindicates an existing definitional practice. However, Cognitive Semantics may also suggest ways of dealing with the links between the senses of lexical items that go beyond common practice. Swanepoel (1992, 1998) and Van der Meer (2000), for instance, argue for devoting more explicit attention to the motivational link between core senses and figurative subsenses. Such motivational links could specifically involve conceptual metaphors in the Lakovian sense (Van der Meer, Swanepoel), or even image schemata (Swanepoel). Interestingly, Van der Meer's suggestion is part of a critical appraisal of the New Oxford Dictionary of English, which is perhaps the first dictionary to refer explicitly to prototype theory as the basis of its organizing principles (cp. Hanks 1994). Up to a point, then, Van der Meer's comments can be read as the suggestion that an even greater influence of the Cognitive Linnguistics approach could be lexicographically useful. Second, there are aspects of lexicography that Cognitive Semantics has touched upon, but that fall outside the scope of the present paper, focusing as it does on problems of polysemy. In particular, frame theory has proved a highly stimulating framework for the description of verbal meaning, both theoretically and lexicographically: see Fillmore and Atkins (1992, 2000) for the more theoretical side of the approach, and compare the website description of the Berkeley FrameNet project for the lexicographical applications. And third, to complete the picture, it should also be mentioned that there are important aspects of current lexicographical practice that Cognitive Semantics has only marginally touched upon, in spite of the fact that the lexicographical approach has proved extremely rewarding for lexical analysis at

Dejinitional practice ofdictionaries

363

large and for the study of polysemy in particular. Specifically, a collocational approach to polysemy, identifying different meanings through differences in collocational patterns, is a methodological focus for many current lexicographical projects (see among others Moon 1998). However, although the use of corpus materials is ardently advocated by a number of linguists working in the tradition of Cognitive Semantics (Geeraerts, Grondelaers, and Bakema 1994, Barlow and Kemmer 2000), collocational methods to get a grip on polysemy are not (yet) among the standard equipment of cognitive linguists. In short, although Cognitive Semantics appears to offer an exciting perspective for the further development of lexicography and lexicographical theory, the real interaction has clearly only started to emerge.

Section 6 Theory and method in lexical semantics

Chapter 15 Cognitive grammar and the history of lexical semantics

Originally published in Brygida Rudzka-Ostyn (ed.), 1988, Topics in Cognitive Linguistics 647-677. Amsterdam/Philadelphia: John Benjamins. Innovative approaches may tend to overstate the novelty of their approach, and in the early years of Cognitive Linguistics, sweeping statements could be found about the revolutionary impact of the new framework (as they may perhaps still be in the rhetoric of converts and novices). An awareness of the history of linguistics is a welcome antidote against such exaggerations. In the article reprinted here, I argue that Cognitive Semantics. rather than being a radical break with some two thousands years of thinking about language, is in a number of important respects a return to the prestructuralist, diachronically oriented fonn of semantics that reigned supreme before the advent of structuralist semantics in the first half of the twentieth century. In the meantime, many other studies have discussed the position of Cognitive Linguistics in the context of the history of linguistics: apart from the extensive work done by Nerlich (1990, 1992), examples to be cited include Swiggers (1988), Lutzeier (1992), lakel (1997), Taylor (1999), Elffers (1999), and Haser (2005). As a companion piece to the present paper, I published an article entitled 'Cognitive Semantics and the history of philosophical epistemology' (Geeraerts 1993c) in which I argued that from a philosophical point of view as well, the exact implications of the cognitive framework can only be properly understood against the background of the history of the discipline. The present article is most profitably read in conjunction with the next chapter, which adds a number of important points to the historical picture painted here.

L

Introduction De semasiologie of semantiek is dus dM deel der taalwetenschap, dat zich ten doel stelt de levensgeschiedenissen van alle woorden stelselmatig te verzamelen, te schiften en te klassificeeren, om daarna te trachten al de diepere psychologische en sociale oorzaken van die levensgeschiedenissen te achterhalen niet alleen, maar weer onderling te vergelijken, om zoo tot de kennis van eenige weinige algemeen geldige semasiologische wetten te

368

Theory and method

komen. Daar nu de woorden alleen leven van ons menschelijk psychisch en sociaal leven, zullen dit dus psychische en sociaal-psychische wetten moeten zijn (Van Ginneken 1912: 6). [Semasiology, or semantics. is that part of linguistics that tries to systematically collect the histories of all words, to select and classifY them, in order not only to try to discover the deeper psychological and social causes of those histories, but also to compare these among each other, so that an insight into some few generally valid semasiological laws can be attained. But because words can only exist and develop through our human psychological and social life, these will be psychological and socio-psychological laws.] The view of semantics expressed in this quotation has an immediate appeal to proponents of the new school of Cognitive Grammar that is exemplified by the studies in this volume. Indeed, studying the semantic history of words to learn something about the 'laws' of the human mind can easily be appreciated as a particular realization of the more general attempt to study linguistic phenomena at large as an emanation and concretization of human cognition and its working principles. Is Van Ginneken merely an incidental precursor of Cognitive Grammar, or are his views typical of a broader tradition of semantic research that can be shown to have some remarkable affinities with Cognitive Grammar? In this paper, I will try to demonstrate the latter. I will try to point out some revealing similarities between Cognitive Grammar and the historicalphilological tradition of semantic research to which Van Ginneken belongs (albeit with the restriction that I will be concerned exclusively with lexical semantics). Also, I will try to make clear that Cognitive Grammar may profit from a greater awareness of the historical-philological tradition of semantics, and I will place the observed similarities against the background of a brief survey of the history oflexical semantics as a branch of linguistics.

2. Historical-philological semantics The historical-philological approach can be exemplified by major works such as Paul (1880), Darmesteter (1887), Breal (1897), Wundt (1900), Nyrop (1913), Wellander (1928), Sperber (1923), Carnoy (1927), and Stem (1931). It can be seen as the first truly linguistic fornl of semantics, i.e. as a form of semantic research that is embedded in linguistics, as an autonomous empirical discipline, rather than being a part of speculative philosophy or of

History oflexical semantics

369

rhetoric (as is the case with most of the semantic studies carried out before the nineteenth century). The historical-philological approach is characterized by three prominent features. I In the following, these will be illustrated with quotations from Breal (1897), not because Breal is the first or the most important exponent of historical semantics, but because his work clearly expresses the major methodological features of the historical-philological approach 2 It should be mentioned that the three characteristics listed below need not be simultaneously present in all of the works belonging to the historical-philological era. Individual differences of emphasis notwithstanding, they do, however, adequately characterize the basic methodological outlook that is shared by most of the semantic studies in this period.

1. Already on the first page of Breal's Essai de semantique. the historical orientation of semantics is indicated as an intuitively obvious matter of fact. Talking about linguistics, he notes: Si 1'on se borne aux changements des voyelles et des consonnes, on reduit cette etude aux proportions d'une branche secondaire de I'acoustique et de la physiologie; si I'on se contente d'enumerer les pertes subies par le mecanisme grammatical, on donne I'illusion d'un edifice qui tombe en ruines; si 1'on se retranche dans de vagues theories sur I'origine du langage, on ajoute, sans grand profit, un chapitre a l'histoire des systemes. 11 y a la, iI me semble. autre chose a faire ... La linguistique parle a I'homme de lui-meme: elle lui montre comment il a construit, comment il a perfectionne, a travers des obstacles de toute nature et malgre d'inevitables lenteurs, malgre meme des reculs momentanes, le plus necessaire instnullent de civilisation (1897: 1-3).

[If one restricts oneself to the study of the changes of vowels and consonants, this discipline is reduced to a secondary branch of acoustics and physiology; if one merely enumerates the losses suffered by the grammatical mechanism, one creates the illusion of a building tumbling into ruins; if one hides behind vague theories about the origin of languages, one adds, without much profit, a chapter to the history of systems. There is, it seems to me, something else to be done... Linguistics talks to man about himself: it shows how he has constructed, how he has perfected, through difficulties of all sorts and in spite of an inevitable inertia, in spite even of temporary retreats, the most indispensable tool of civilization.]

It is even the case that an adequate understanding of words in their contemporary meaning requires a thorough knowledge of their semantic history: 'L'histoire peut seule nous donner aux mots le degre de precision dont nous avons besoin pour les bien comprendre' (Only history can give to the words

370

Theory and method

the degree of precision that we reqUIre to understand them adequately) (1897: 124). This historical orientation is, of course, a general aspect of nineteenthcentury linguistics. As an autonomous empirical discipline, linguistics comes into being as a form of historical research, and the birth of historical linguistic semantics in the nineteenth century is merely one more aspect of the overall diachronic outlook of the first phase in the development of modem linguistics. It should be noted, however, that the birth of semantics within that young linguistic science was not just a question of completeness, but rather of necessity. The study of meaning was not simply taken up out of a desire to study linguistic change in all of its aspects, but a thorough knowledge of the mechanisms of semantic change appeared to be a prerequisite for adequate historical investigations into the formal aspects of languages. Indeed, one of the most important branches of historical linguistics was the etymological reconstruction of the history of words and phrases. The methodology of comparative reconstruction requires that the word forms from different languages that are to be compared with each other, be semantically related. But such a relationship is not always obvious. For instance, throughout the older Germanic languages, there is a fairly systematic formal resemblance between words for the concept 'beech', and words for notions such as 'book' and 'letter'. (Compare, e.g., Old High German buohha 'beech' and buoh 'book', or Old Saxon bOka 'beech' and bOk 'book, writing tablet'.) Now, in order to justify a reconstruction of these forms as being related to the same Proto-Germanic root, their semantic relationship has to be clarified. (In this particular case, an awareness of the frequent metonymical relationship between names for substances and the name of objects made of those substances will have to be combined with archaeological evidence showing that wooden tablets were used for writing purposes.) Considering a number of lexical forms as cognate requires that their semantic relationship can be plausibly established, and this in turn requires an overview of the regular mechanisms of semantic change. As such, diachronic semantics was not merely taken up as an end in itself, but also as an auxiliary discipline with regard to etymological studies. 3 It should also be noted that the enormous success of historical semantics in the nineteenth century and the first decades of the twentieth, does not only show up in the large number of articles and monographs devoted to it, but also in the major lexicographical projects undertaken during that period. Voluminous historical and etymological dictionaries such as the Deutsches Worterbuch, the Oxford English Dictionary, and the Woordenboek der

History oflexical semantics

371

Nederlandsche Taal testify that the early interest in the semantic histories of words has led to an hitherto unsurpassed amount of empirical semantic research. 2. The psychological orientation of historical-philological semantics has two aspects to it. In the first place, lexical meanings are considered to be psychological entities, i.e., (a kind ot) thoughts or ideas: '(Le langage) objective la pensee' (Language makes thought objective.) (Breal 1897: 273). The mental status of lexical meanings links up directly with the overall function of thinking, i.e., with the function of cognition as a reflection and reconstruction of experience. As such, language expresses thought as a storeroom of encyclopedic knowledge: 'Le langage est une traduction de la realite, une transposition ou les objets figment deja generalises et classifies par le travail de la pensee' (Language is a translation of reality, a transposition in which particular objects only appear through the intermediary of the generalizing and classificatory efforts of thought.) (1897: 275). Language, then, is not autonomous; it is embedded in and linked with the total set of cognitive capacities that enable men to understand the world with ever more refined conceptual tools. In the second place, historical-philological semantics tries to explain meaning changes as the result of psychological processes. The general mechanisms of semantic change that can be derived from the classificatory study of the history of words constitute patterns of thought of the human mind. Breal caBs these mechanisms 'les lois intellectueBes du langage', but he hastens to add that 'law' means something different here than in the natural sciences: a law of semantic change is not a strict rule without exceptions, but it represents a tendency of the human cognitive apparatus to function in a particular way. In a passage that opposes the restriction of linguistics to the study of the formal aspects of language, he remarks: Nous ne doutons pas que la linguistique, revenant de ses paradoxes et de ses partis pris, deviendra plus juste pour le premier moteur des langues, c'est-a-dire pour nous-memes, pour I'intelligence humaine. Cette mysterieuse transfonnation qui fait sortir le franc;ais du latin, comme le persan du zend et I'anglais de I'anglo-saxon, et qui presente partout sur les faits essentiels un ensemble frappant de rencontres et d'identites, n'est pas le simple produit de la decadence des sons et de I'usure des flexions; sous ces phenomenes ou tout nous parle de ruine, nous sentons I'action d'une pensee qui se degage de la fonne a laquelle elle est enchainee, qui travaille a la modifier, et qui tire souvent avantage de ce qui semble d'abord perte et destruction. Mens agitat molem (1897: 338-339).

372

Theory and method

[We do not doubt that linguistics, giving up its paradoxical prejudices, will treat more fairly the primary forces in languages, i.e. ourselves, human intelligence. The mysterious transformation that makes French grow out of Latin (just as Persian out of Zend, and English out of Anglo-Saxon), and that everywhere shows a remarkable set of similarities and parallelisms with regard to its essentials, is not simply the product of the decay of sounds and the wearing off of endings. Behind these phenomena in which everything seems to speak of destmction, we feel the active efforts of human thought liberating itself from the form in which it is constrained, trying to modifY it, and very of ten turning to its advantage what at first sight appears to be mere loss and decay. Mind moves matter.] The moving force of the human mind also shows up in the fact that the fundamental factor that brings the psychological mechanisms of semantic change into action, consists of the communicative needs of the language user. Languages change because people try to express their thoughts as accurately and satisfactorily as possible. Le but, en matiere de langage, c'est d'etre compris. L'enfant, pendant des mois, exerce sa langue a proferer des voyelles, a articuler des consonnes: combien d'avortements, avant de parvenir a prononcer clairement une syllabel Les innovations grammaticales sont de la meme sorte, avec cette difference que tout un peuple y collabore. Que de constmctions maladroites, incorrectes, obscures, avant de trouver celle qui sera non pas l'expression adequate (il n'en est point), mais du moins suffisante de la pensee (1897: 8).

[The goal, as far as language is concerned. is to be understood. During months, the child exercises his tongue to produce vowels, to articulate consonants: how many failures, before he can clearly pronounce a syllable. On the grammatical level, innovations are of the same sort, with this difference that an entire people is involved. How many clumsy, incorrect, obscure constmctions, before the one is found that will be, not the perfect expression of thought (there is none), but at least a sufficient expression of it.] Note, however, that the expressive, communicative intent that lies at the basis of linguistic change need not always have the conscious, voluntary characteristics attributed to it by Breal. Sperber (1923), e.g., stresses the importance of subconscious emotional factors. Also, the overall psychological orientation of historical-philological semantics leaves room for authors such as Meillet (1906), who stresses the social rather than the psychological background of semantic change. Likewise, the earliest classifications of semantic changes, relying heavily on traditional

History oflexical semantics

373

rhetorical classifications (as is still the case in, e.g., Darmesteter 1887), were 4 much less psychologically orientated than later ones. 3. Methodologically, historical-philological semantics is a hermeneutical discipline. In the following quotation, Breal does not simply repeat the point that semantics is a historical science, but he also has something to say about the way in which that scientific project is put into practice: Si I'on admet une difference entre les sciences historiques et les sciences naturelles, si I'on consictere I'homme comme fournissant la matiere d'un chapitre a part dans notre etude de I'univers, le langage, qui est l'oeuvre de I'honune, ne pourra pas rester sur I'autre bord, et la iinguistique, par une consequence necessaire, fera partie des sciences historiques (1897: 278). [If one admits that there is a distinction between the historical and the natural sciences, that is, if one considers man as being the subject matter of a separate chapter of our study of the universe, language (which is the product of man), cannot stay on the other side, and linguistics wil1 inevitably be a branch of the historical sciences.]

Clearly, the natural sciences also study historical processes (as in geology or the study of biological evolution), and that is why the difference between the natural and the human sciences that is mentioned in the quotation has to be sought on the methodological level rather than on the level of the object matter of both approaches. The distinction made by Breal probably refers to the theories of the German philosopher Wilhelm Dilthey, whose views on the relationship between the natural and the human sciences (Naturwissenschaft versus Geisteswissenschaft) were widely popular near the end of the nineteenth century. 5 The methodological autonomy of the human sciences with regard to the natural sciences resides in the fact that they try to understand, by means of an empathic process of interpretation (Einfuhlung), the cultural forms of expressions in which men have, throughout history, laid down their experience of the world. The natural sciences, on the other hand, try to explain the characteristics of the material world by means of rigid laws. Next to being historically and culturally orientated, the human sciences in the Diltheyan sense are hermeneutical par excellence: they try to reconstruct the original experience that has lain at the basis of particular forms of human expression that have been transmitted from earlier times to the present day; they look for the expressive intention behind historical forms of expression. The connection between the Diltheyan conception of the human sciences and the kind of linguistic semantics sketched above will be clear: through its historical approach, through its cultural, experiential orientation (language is

374

Theory and method

an expression of a particular experience of the world - see the quotation given above from Breal 1897: 275), and through the importance it attaches to the expressive intentions of language users as the source of linguistic change, historical-philological semantics fits nicely into the Diltheyan view of the human sciences. This is reflected on the methodological level. Because linguistic semantics is a historical discipline, its primary material consists of texts from dead languages or from previous stages in the development of a living language. Its basic methodological act is therefore the interpretation of those texts, only afterwards can changes between periods (and the mechanisms guiding them) be recognized, classified, and explained. The primary methodological step of the historical semantician is that of the historical lexicographer and the philological scholar: to interpret historical texts against the background of their original context by trying to recover the original communicative intention of the author.

3. Cognitive Semantics With Rosch's prototypical conception of categorial structure as one of its leading notions (cp. Rosch 1977, Mervis and Rosch 1981), Cognitive Semantics is exemplified by studies such as Fillmore (1977a, 1977b), Lakoff (1982a, 1982b), Lakoff and Johnson (1980), Langacker (1983), Talmy (1983, 1985b), Coleman and Kay (1981), Lindner (1981), Brugman (1981, 1984), Hawkins (1984), Sweetser (1984), Vandeloise (1985a), Dirven (1985), Rudzka-Ostyn (1985), Geeraerts (1983a, 1985a). The characteristics of the cognitive trend in lexical semantics can be summarized by contrasting it with a view of categorization that has, implicitly or explicitly, been fashionable for some decades. This view is sometimes referred to as the classical or 'Aristotelian' view (cp. Rosch 1975). As far as the history of lexical semantics is concerned, this is a somewhat misleading terminology. The purpose of this paper is precisely to show that the 'classical' approach to lexical structure has not been dominant throughout the whole history of lexical semantics as a branch of modem linguistics, but is rather typical of the structuralist period between the older historical-philological era and the birth of Cognitive Semantics. As far as Aristotle is concerned, the predicate 'Aristotelian' is justified by the Aristotelian (and scholastic) metaphysical distinction between accidental and essential attributes, with the latter constituting the true definition of a concept. One should not forget, however,

History oflexical semantics

375

that in his poetical and rhetorical treatises, Aristotle also has a lot of attention for topics such as metaphor, analogy, and the use of concrete examples or 'paradigms' in rhetoric, that come close to the interests of cognitive semanticists. The following list of characteristics gives only a rough overview of Cognitive Semantics. I trust that the other papers in this volume will give the reader an exemplary view of the cognitive approach. I will not try to prove or defend the theses listed below: I merely present them as a summary of what seem to be the major tendencies in cognitive semantic research on the lexical-semantic level, a summary that may serve as an aide-memoire to the comparison of Cognitive Semantics with the historical-philological tradition that will be the subject matter of section 4. It should be mentioned that there is a lot of variation within Cognitive Semantics: the views listed below need not all be held by the specific studies that can be situated within the cognitive framework. In this respect, Cognitive Semantics as a theoretical paradigm corresponds with its own views on categorization. One of its basic tenets is that lexical concepts are not rigidly defined by a set of essential attributes shared by all the instances of those concepts. Likewise, a rigid definition of Cognitive Semantics, consisting of an enumeration of attributes shared by all cognitive semantic studies is hard to give. Notice that this is in keeping with the paradigmatic conception of science defended by Kuhn, Lakatos and others, which considers scientific research programmes to be clusters of interconnected theories. (The methodological consequences of this analogy between the theoretical views of Cognitive Semantics and the metatheoretical views of the Kuhnian, paradigmatic theory of science are explored in Geeraerts 1985a.) What, then, are the central theses of Cognitive Semantics? 1. Lexical concepts have vague boundaries, in the sense that they contain peripheral zones round clear conceptual centers (versus: lexical concepts are discrete, well-defined entities). - The prototypical view of categorization implies that categories may have marginal instantiations that do not conform rigidly to the central cases. This extensional fact is paralleled by the intensional phenomenon that it may not be possible to define a concept by means of a set of attributes applicable to all the in stances in the extension of that concept. Intensionally as well as extensionally, then, concepts may be characterized by the existence of a chiaroscuro-zone, either with regard to the category membership of instantiations of a category, or with regard to the definitional status of attributes.

376

Theory and method

2. Lexical concepts are polysemous clusters of overlapping semantic nuances (versus: the various senses of a lexical item can always be strictly separated from each other). - The vagueness that characterizes conceptual categories as a whole is mirrored by their internal structure. The blurred edges of lexical concepts as a whole are paralleled by the possible absence of clear boundaries among the senses that constitute the category as a whole. The most common expression of this view is the family resemblance model of categorial structure (Rosch and Mervis 1975). 3. The distinction between analyticity and syntheticity, and that between essential and accidental attributes cannot be rigidly maintained (versus: there is a strict distinction between analytical and synthetic statements, and between essential or accidental attributes). - This is a definitional consequence of the previous points. There need not be a set of attributes that is applicable to all the members of a category (and only those), and that defines the 'essence' of that category (all other attributes being accidental). Since the Aristotelian distinction between essential and accidental attributes is paralleled by the Kantian distinction between analytic and synthetic statements (at least on the lexical level), the dichotomous distinction between the latter cannot be maintained either, although there may be a gradient pattern of more or less inherent attributes. 4. Lexical concepts may be disjunctively defined (versus: lexical concepts are necessarily defined conjunctively). - This is again a consequence on the level of definitions of points 1 and 2. In the classical view, a concept is defined by a set of essential attributes all of which can be predicated of the members of the category in question. Within a prototypically organized category, however, categorial membership may be based on sufficient similarity rather than identity; some of the allegedly necessary ('essential') attributes of the central exemplars of the category may appear to be optional at the periphery. As Labov (1973) has remarked, traditional lexicography very often uses disjunctive rather than conjunctive definitions, a fact that clearly indicates the practical importance of disjunctive definitions for semantic description. 5. Attributes within (or exemplars of) a category may have different degrees of salience (versus: all attributes within the definition, or exemplars in the extension of a concept have an equal degree of salience). - Attributes not shared by all the members of a category (or a majority thereof) are less important definitionally than attributes that do appear in all or most of the exemplars of a category. These frequent attributes carry more weight within

History oflexical semantics

377

the semantic structure of the category. Analogously, some members of a category carry more weight as better exemplars of the category, in contrast with the peripheral, less salient exemplars. 6. Lexical concepts function in a flexible and analogical manner (versus: lexical concepts function in a rigid, algorithmical fashion). - The fact that category membership can be defined by similarity rather than identity entails that conceptual categories can be used in an extremely flexible way; making the criteria for the use of a category less strict enhances the usage potential of that category. One aspect of this extremely important characteristic of lexical concepts is discussed in my other contribution to this volume (Geeraerts 1988d); there, I try to make clear that the flexibility inherent in prototypical categories as interpretative schemata with regard to experience, may well be one of the major functional, psychological reasons for having such a type of categorial structure. Another point to be mentioned is this: the flexibility of prototypical concepts bridges the gap between synchrony and diachrony (a moot point in structuralist conceptions of language) by incorporating the dynamism that is usually considered to be characteristic for a diachronic conception of language among the synchronic features of lexical concepts. This is, of course, a point that is of particular interest with regard to the comparison of historical-philological semantics and Cognitive Semantics. 7. Lexical concepts have to be studied as a proper part of human cognition at large (versus: lexical concepts have to be studied as a part of an autonomous linguistic structure). - What is at stake here is the fundamentally structuralist claim that the lexicon of a natural language contains a proper semantic structure that is independent of the broader cognitive organization of the human mind, and which it is the task of linguistic semantics to reveal; this implies that semantics is an autonomous discipline. 6 By contrast, Cognitive Semantics claims that lexical concepts can only be studied adequately against the background of the human cognitive capacities at large, and in particular , that there is no specifically linguistic, semantic organization of knowledge, separate from conceptual memory in the broadest sense. If there is no autonomous linguistic-semantic memory, lexical semantic research cannot be done autonomously, but it has to be conducted in close cooperation with other sciences studying the human mind. Of these, psychology is obviously the most important one, but there are others such as Artificial Intelligence, neurophysiology, and cultural anthropology.

378

Theory and method

8. There is no distinction between semantic and encyclopedic knowledge (versus: the semantic definition of a lexical concept is to be distinguished from the encyclopedic data that can be connected with the latter). - This point is a specification of the former. The distinction between semantic and encyclopedic knowledge (which parallels that between essential/analytic and accidental/synthetic information) is based on the distinction between an autonomous linguistic conceptual structure on the level where lexical items are defined, and conceptual memory in the broadest sense. If the latter distinction is discarded, the former also falls. 9. Semantic studies cannot ignore the experiential and cultural background of the language user (versus: semantic phenomena should be studied apart from user- or culture-specific background data). - This is a further consequence of the non-autonomous approach mentioned in point 7. If language is one of the basic cognitive tools of man, it should not be studied autonomously, but it should be considered in the light of this cognitive function, i.e. of interpreting, ordering, retaining, and expressing human experience. This experience may, of course, be individually or culturally specific. This implies, e.g., that attention will be given to cultural differences in the metaphorical patteming of experience (see Lakoff and Johnson 1980), and that semantic universals will be seen as general strategies for coping with experience rather than as substantive universals.

4.

Comparing the cognitive and the historical-philological tradition

In the light of the foregoing sections, it should not be too difficult to recognize the similarities between the cognitive and the historical-philological approaches, particularly when they are set off against the 'classical' approach. I. A first point of primary importance is the psychological orientation of both approaches. Both consider meanings to be primarily psychological entities, that is to say, they identify 'meanings' and 'mental concepts', as against an approach that strictly separates linguistic meanings and (encyclopedic) psychological concepts. At the beginning of his Griechische Bedeutungslehre, Max Hecht sums up the disciplinary position of historicalphilological semantics: Insofem sie zugunsten der Lexikographie die Bedeutungen in zeitlicher Folge ordnet und im Interesse der Etymologie die Gesetze der Bedeutungs-

History oflexical semantics

379

anderung aufstellt, hat sie sprachwissenschaftlichen Weft. Soweit sie aber diese Gesetze aus der Natur des Geistes herleitet und eine Geschichte der Vorstellungen gibt - Bedeutungen sind Vorstellungen -, rallt sie auf das Gebiet der empirischen Psychologie 15 (1888: 5). [(Semantics) is linguistically valuable to the extent that it chronologically classifies meanings in the interest of lexicography, and writes down the laws of semantic change in the interest of etymology. To the extent, however, that it derives these laws from the nature of the mind and that it writes a history of ideas - meanings are ideas -, it falls within the realm of empirical psychology.] In a vigorously polemical article that is properly to be regarded as the first theoretical statement of structural semantics (with Trier's 1931 monograph on semantic fields being the first descriptive achievement), Leo Weisgerber (1927) blames Hecht for relegating semantics to the field of psychology. Treating meanings as psychological 'representations' (Vorstellungen) , says Weisgerber, entails that meanings cannot be studied autonomously, as a proper part of linguistic structure.? This structuralist desire for an autonomous approach to linguistic semantics is then again discarded by Cognitive Semantics in favor of a whole-heartedly psychological approach in which the study of linguistic structure is required to be in accordance with what is known about cognitive processing in general (see Langacker 1983: 1/6-7). An immediate consequence of the rejection of an autonomous approach is the fact that neither the historical-philological nor the cognitive tradition imposes a strict distinction between encyclopedic and purely semantic data in lexical analyses. One may say that, at the time of historical-philological semantics, the distinction had not yet gained wide acceptance, whereas it is no longer generally accepted in contemporary lexical semantics (see Haiman 1980a, Langacker 1983: 11/63). Another consequence of the psychological approach is the fact that experimental psychological research will have an influence on linguistic conceptions of categorial structure. Taking the cognitive nature of language seriously entails taking psychological studies of meaning seriously. It is, in fact, possible to identifY major psychological influences on the philological and the cognitive research programmes. The latter's conception of categorial structure has to a large extent been shaped by Eleanor Rosch' s experimental work on prototype semantics (cp. supra), whereas the former has been profoundly influenced by Wilhelm Wundt's apperceptionalist psychology (Wundt 1900). It is true, though, that Wundt's views have not been generally accepted by all semanticists of the historical-philological persuasion, but

380

Theory and method

even when they are systematically criticized (e.g., Marty 1908), the tendency is rather to substitute another psychological conception for Wundt's, than to abandon the psychological outlook as a whole. A final consequence of the psychological orientation of the research programmes that we are trying to compare is their common attention for the experiential background of semantic phenomena. In historical-philological semantics, this is embodied in what I have called its hermeneutic character: it considers lexical meanings as an expression of a particular (individual or cultural) view of the world. As for Cognitive Semantics, its experiential nature has probably been brought to the fore most vigorously by George Lakoff in his attack on the classical, 'objectivist' conception of categorization (but see also, e.g., Langacker 1983: 1/6-7). In the following quotation, the link between the experiential nature of semantics and its psychological, non-autonomous orientation is clearly made apparent: After a generation of research in which it was implicitly assumed that language could be described on its own terms, it has become more interesting to ask how much of the structure of language is determined by the fact that people have bodies with perceptual mechanisms and memory and processing capabilities and limitations, by the fact that people have to try to make sense of the world using limited resources, and by the fact that people live in social groups and have to try to communicate with each other (Lakoff 1982b: 155).

Comparing the importance of experiential factors in historicalphilological and in Cognitive Semantics, there may be a shift in emphasis from meanings as the expression of subjective experience to conceptual categories as interpretative schemata shaping experience, but the basic point is the same: language is an integral part of the intellectual, emotional, and social life of human beings, and this is reflected by the nature of linguistic categories. As such, these cannot be considered apart from the human endeavors to make sense (literally) of the world. 2. Next to the fundamental correspondence in the methodological orientation of historical-philological and Cognitive Semantics, it should also be mentioned that there is a lot of similarity between both approaches with regard to their conception of the nature of semantic categories, although this theoretical similarity is not as complete as the basic methodological one. One point that is quite clear, is the importance of polysemy for categorial structure. Both approaches tend to conceive of lexical concepts as polysemous clusters of senses, and to study them accordingly. In particular, if the main descrip-

History oJlexical semantics

381

tive efforts of historical-philological semanticists are directed at the classification of semantic changes, this is because the mechanisms of semantic change are responsible for the polysemy of lexical items. (One might say: mechanisms of change such as metaphor and metonymy are the structural links between the various senses of an item.) It is no surprise, then, that the cognitivists' interest in polysemy leads to a revival of the research into some of the earlier semanticists' favorite topics. This is, of course, most clearly true for the study of metaphor, but the phenomenon is not restricted to it. (There is, e.g., the rediscovery of sound symbolism by Lakoff 1983; see Kronasser 1952: 160-168 for an introduction to the older tradition's treatment of it.) Furthermore, the notions of vagueness and centrality that figure so prominently in prototypical semantics are far from absent in historicalphilological theories. They may not receive the same emphasis as in presentday theories, but they are clearly present in a number of works in the historical-philological vein. They can be found, e.g., in Paul's influential classification of semantic changes (1880), and in Erdmann (1910), whose particularly lucid formulation of the characteristics of lexical concepts justifies an extensive quotation. Denn welche Theorien iiber Wesen, Bedeutung und Entstehung der Begriffe man auch vertreten mag: vom Standpunkt der Logik wird man immer fordern miissen, dass sie eine unzweideutige, klare Grenze aufweisen, dass sie einen bestimmten Inhalt und Umfang haben. Und Begriffe dieser Art werden durch Worte nicht ohne weiteres bezeichnet. Worte sind vielmehr im allgemeinen Zeichen fur ziemlich unbestimmte Komplexe von Vorstellungen, die in mehr oder minder loser Weise Zllsammenhangen.... Die Grenzen der Wortbedeutung sind verwaschen, verschwommen, zerfliessend. Treffender aber noch wird meines Erachtens der Sachverhalt gekennzeichnet, wenn man iiberhaupt nicht von Grenzlinien des Umfangs redet, sondern ... von einem Grenzgebiet, das einen Kern einschliesst. Veranschaulicht man sich gewohnlich den Umfang eines logisch vollkommenen Begriffs durch eine scharfe Kreislinie, wie sie annahrend ein gutgespitzter Bleistift erzeugt, so kann man sich die Abgrenzung eines Wortumfanges durch einen mehr oder minder breiten, in sich zumcklaufenden Streifen versinnlichen, wie ihn ein in Farbe getauchter Pinsel auf einer Flache hinterlasst (1910: 4-5). [But whatever theories about the essential nature, the meaning and the origin of concepts one may adhere to, from the point of view of logic one will always have to require that they exhibit unambiguous, clear boundaries, that their extent and content are clearly specified. But words do not simply

382

Theory and method

indicate concepts of that kind. Words in general are rather signs for fairly unspecific complexes of psychological representations that belong together more or less loosely. '" The boundaries of word meanings are vague, unclear, indeterminate. The situation is, I think, even more adequately described if one simply does not talk about the borderline of the range (of a word), but ... if one talks about a border area that includes a central area. If one usually represents the range of a logically perfect concept by means of a sharp borderline, such as is approximately drawn by a well-sharpened pencil, one can represent the boundary of the content of a word by means of a more or less wide, gradually narrowing band, such as is produced on a flat surface by a brush that has been dipped into paint.] The correspondence with the contemporary prototypical view is too obvious to need much comment. The basic correspondence established in the previous points does not imply that there are no differences between the two research traditions compared here. The basic difference is the shift from a diachronic to a synchronic approach to semantics, mirrored by the fact that the emphasis shifts from the mechanisms of semantic change to the internal prototypical structure of polysemous lexical concepts. This shift is, of course, part of the legacy of structuralism, but one that is not accepted without fundamental alterations. In fact, the extent to which Cognitive Semantics differs from structuralism with regard to the diachrony/synchrony-distinction is more important than its primary adherence to that distinction. A strictly synchronic approach as favored by structuralism tends to overlook the fact that languages are in constant flux. 8 By incorporating the dynamic principles that structuralism can only conceive of diachronically into the synchronically flexible nature of lexical concepts, cognitive semantics can avoid the structuralist paradox that the true nature of linguistic structure is considered to be a steady state, whereas such a steady state can hardly bc found in linguistic reality. The prototypical view of natural categories is so to say the synchronic side of the diachronic coin; what appears as diachronic change from one point of view, is merely the realization of synchronically available possibilities from another. In short, flexibility is the intersection of diachrony and synchrony at the level of lexical structure, and that is a point of view that may not have been completely absent in historical-philological semantics, but that still deserved to be emphasized as it is now done in Cognitive Semantics. In this respect, then, Cognitive Semantics is a fundamental improvement over the older tradition. But does this relationship also work the other way around, i.e., can Cognitive Semantics profit from the historical-philological

History oflexical semantics

383

tradition? The latter constitutes such a huge body of observations, classifications, and explanations (indeed, its gigantic proportions may well be the very cause of its present-day neglect), that it is hard to imagine that it contains nothing of interest for Cognitive Semantics. That a confrontation with historical-philological semantics may be very fruitful is a point I would like to illustrate on two different levels, first with regard to a very specific topic of research, and second with regard to a rather more general, and more fundamental aspect of semantics. a. Both Van Ginneken (1912) and Sweetser (1984) study the figurative transfers of (among many others) expressions connected with auditory experience. Both do so from an essentially cross-linguistic, etymological point of view, but their perspective is slightly different. Sweetser concentrates on words for the act of hearing itself, and notices among other things that verbs for the concept 'to hear' shift towards the notion 'to listen, to heed', and further to 'to obey'. She also mentions that derived words coming from 'hear' -roots regularly mean 'tale, report, fame, glory, news'. Van Ginneken considers attributes for the things we hear , and points out that auditory characteristics may be transferred, among others, to visual sensations (loud colours, aloud dress), to tactile sensations (Dutch dovenetel 'dead nettle' from door, 'deaf), or to emotions (to quieten someone's fear). A confrontation of Van Ginneken's and Sweetser's results yields a particular question for further research: do the patterns of figurative transfers in both sets of expressions correspond with each other, and if they do not, what cognitive reason can there be for conceptualizing one set differently from the other? For instance, if 'loud' is an attribute for things that can be heard well, and if 'to be heard' shifts towards 'to be famous', is it reasonable to expect words for the notion 'loud' to be applicable to fame or famous things and persons? Conversely, if attributes for noises are applicable to non-auditory perceptual sensations, is it also possible to express the notion 'to experience a non-auditory perceptual stimulus' with words that etymologically mean 'to hear'? These questions need not, of course, be answered here; the important thing is to see how they arise as specific research hypotheses from a confrontation of parallel studies in Cognitive Semantics and in the prestructuralist tradition of historical semantics. b. One of the major questions in prototypical semantics concerns the internal coherence of natural categories. What keeps a prototypically organized category together, and more particularly, how far can the senses in a polysemous cluster be apart from each other or from the prototypical centre before they

384

Theory and method

'drop out' of the category? The question can be refonnulated in tenns of the relations that hold among those senses: what kind of relations are acceptable? Are they gradable, and if so, to what extent can they hold? $0 far, cognitive semantics has answered these questions almost exclusively in tenns of similarity (obviously a gradable notion); think of a notion such as 'family resemblance', of the statement that prototypically organized concepts are clusters of overlapping senses (the overlapping senses are similar to the ex tent that they have a number of features in common), and of the cognitivists' interest in metaphor. Now, it would not be implausible to equate the relations that can hold among the diverse senses within a natural category, with the mechanisms of semantic change identified by diachronic semantics. In fact, if the flexibility of prototypical categories is indeed a phenomenon of the same type as the semantic changes of old (as I have argued above), the same type of mechanisms should govern both. The mechanisms that have been studied by historical-philological semantics as types of diachronic change are then incorporated into synchronic categories as the mechanisms that underlie their flexibility and that specify the relations between the original senses and the new ones that arise from the flexible use of the category. These mechanisms detennine and restrict the range of possible extensions of a concept, given a prototypical centre or an existing cluster of applications. As such, they are interpretative principles, strategies that help one get from existing knowledge to new, contextually functional applications of a concept. But historical-philological semantics has revealed the existence of much more mechanisms of change than mere similarity (though the latter is undoubtedly one of the major types). The associations principles leading from existing senses to new ones also comprise relations of contiguity, which fonn the basis of what is traditionally known as metonymy (including synecdoche), and relations that might be called mimetic because they presuppose the existence of another expression that already has the new sense of the word we are investigating, and with which the investigated word is associated for one .reason or another. A typical example of this mimetic type of change is semantic borrowing: the Gothic word daupjan is associated with the Greek word baptizein through their common base meaning 'to cause to go down, in particular, to submerge' , and next copies the secondary sense 'to baptize' of the Greek word. Other examples of mimetic change are ellipsis (in which case the association is syntactic), the semantic merging of homonyms (where phonetic associations are involved), and parallel devel-

History oflexical semantics

385

opments among antonyms (again a case of an initial semantic association). (See Geeraerts 1986a: 38-41 for a more thorough treatment.) I believe it will be one of the major tasks for lexical semantics in the next few years to study the relevance of these traditional associationist mechanisms of diachronic change for the cognitive theory of categorization. Do they have the same cognitive status as associations through (literal or metaphorical) similarity, in providing principles that guide the interpretative efforts of the hearer, enable the speaker to use lexical concepts flexibly, and define the structural coherence of natural categories? Or do prototypical categories only allow certain types of relations, so that semantic extensions based on principles of contiguity or lexical mimesis immediately lead to the creation of another category, distinct from the original one? Again, the question need not (in fact, could not) be answered here. It suffices to see that the confrontation with the historical-philological tradition opens up a major area of research for Cognitive Semantics. In short, then, the affinity between Cognitive Semantics and historicalphilological semantics also seems to entail that the former may profit from a closer acquaintance with the latter. To deepen our insight into the affinities between both approaches, we shall now have a closer look at the history of lexical semantics as a whole.

5. A survey of the history of lexical semantics

Near the end of his book A History o.fSemantics, W.T. Gordon remarks that semanticists have tended to take a tabula rasa approach to their discipline, either out of a conscious rejection of previous theories, or because they were simply unaware of the work of their predecessors (1982: 244). However, Gordon's remark is more indicative of his own failure to identify major schools and traditions in the history of semantics, than of the actual absence of such major trends. At least in lexical semantics (and let us not forget that up to the incorporation of semantic problems in transformational grammar, semantics had been primarily lexical), it is possible to point out the existence of a number of global methodological perspectives that shape the semantic endeavors of a majority of researchers within a particular period. Table 1 gives an approximation of what such a methodological history of lexical semantics might look like. The justification for the figure can only be dealt with summarily here; a more detailed treatment of the major historical trends in lexical semantics can be found in Geeraerts (1986a).

386

Theory and method

Next to the historical-philological and the cognitive approach, the figure identifies two other basic methodological approaches to lexical semantics. The logical approach to lexical semantics is clearly a minor tradition. When it is not primarily a formalization of descriptive ideas developed within structuralist semantics (cp. Dowty 1979), its descriptive contribution to lexical semantics resides mainly in the analysis of lexical items corresponding with logical operators (such as quantifiers and connectives), and in the analysis of the syntactic meaning of word classes (the 'types' of Montague semantics). The proponents of Montagovian semantics (which is the most sophisticated form of logical semantics) have themselves expressed serious doubts with regard to the applicability of intensional logic to word meaning beyond the level of the syntactic meaning of word classes (cp. Thomason 1974: 48). Regardless of the results achieved by logical semantics in other domains, its contributions to lexical semantics are as yet too restricted to consider it a major trend in the history of lexical semantics. Structuralist semantics is undoubtedly the most diverse of the four approaches mentioned in the figure. Disregarding the distributionalism of, among others, Dubois (1964) and Apresjan (1966), which was never very successful, there are three main approaches to be mentioned within structuralist semantics. Two of the alternatives are clearly present in the debate between the proponents of a decompositional and an axiomatic approach to the formalization of lexical meaning, such as was held within the framework of transformational granunar in the course of the 1970s (see, e.g., Katz and Nagel 1974 versus Fodor, Fodor and Garrett 1975). For reasons to be explained in the following section, I consider transformational semantics to be the methodological culmination of structuralist semantics. The axiomatic and decompositional alternatives within transformational semantics, then, are related to earlier forms of structural semantics. Componential analysis is an outcome of the semantic field approach made popular by Trier (1931), in the sense that the conceptual dimensions from which components are derived, are primarily the distinctive oppositions within a semantic field. The field origin of componential analysis is less clearly visible in Katz and Fodor's introduction of the method into transformational granunar (1963), but it is quite obvious in Lounsbury's (1956) and Goodenough's (1956) anthropological studies, in which the componential method for meaning analysis was applied for the first time, and from which Katz and Fodor seem to have borrowed the idea. The field-theoretical origin of componential analysis is equally apparent in the introduction of the method by European linguists such as Pottier (1963) and Coseriu (1964). It is not quite clear. incidentally,

History oflexical semantics

387

whether this European branch of componential analysis came into being independently of Lounsbury, Goodenough, and/or Katz and Fodor. Table 1. Dominant theories in the history oflexical semantics MAIN

THEORETICAL

OBSERVATIONAL EMPIRICAL

EXPLANATORY

PERIOD

FRAMEWORK

PRINCIPLE

PRINCIPLE

18701930

HISTORICAL-

language as the statements about self-expression meaning change of an individual or a community

PHILOLOGICAL SEMANTICS

19301975

1970-

STRUCTURAL SEMANTICS

LOGICAL SEMANTICS

1975-

COGNITIVE SEMANTICS

language as an autonomous structure

DOMAIN

principles of efficiency and expressivity

statements about meaning decompoparadigmatic and sition, or meaning postulates syntagmatic semantic relations

language as statements about reference to the truth condipossible worlds tions of propositions

algorithmical compositionality

language as a cognitive tool

natural categorization (prototypicality) and cognitive strategies

statements about the flexibility, internal structure, experiential nature, encyclopedic character of lexical concepts

On the other hand, the axiomatic (or 'meaning postulate') approach is methodologically related to earlier purely structuralist work by Lyons (1963, 1968), in which relations such as hyponymy, synonymy, and antonymy among unanalyzed lexical items are taken to be the core structure of the lexicon (rather than analytic relations among the items in semantic fields). It is extremely important to realize that the decompositional and the axiomatic approach are compared here, not as formal methods of description, but as methodological perspectives on word meaning. In particular. they methodologically define different observational bases for lexical semantics: the componential approach focuses on the structural relations among

388

Theory and method

the items in semantic fields, whereas the axiomatic approach focuses on structural relations of the type represented by hyponymy and synonymy. As such, the point to be made here is simply that structural semantics may focus on different types of semantic structure. Notice in this respect that componential analysis and meaning postulates as formal methods of description are not exclusively tied up with the different observational approaches with which they are primarily connected. In fact, componential analysis can be used to explain the synonymy, antonymy etc. of lexical items (this is the Katzian position), whereas meaning postulates may serve as a descriptive method for indicating distinctive oppositions within semantic fields. Also, as methods of formalization, the origins of the decompositional and the axiomatic approach do not lie with structural semantics, but in phonology and logic, respectively. In short, the methodologically important point is not the formal method of description, but the difference in observational basis with which it is usually connected: structuralists may focus on semantic field structures, or on structural relations such as hyponymy, antonymy, and the like. Next to these two types of paradigmatic structures, they may direct their attention at a third type of structure, viz. the syntagmatic semantic relations that have become known in transformational grammar as selectional restrictions, but that had actually been discussed already by Porzig (1934).9 There will be much more to say about structural semantics in the following section, but first, our brief discussion of Table 1 has to be concluded with a note on the chronological periods distinguished in it. These indicate the heyday of each of the traditions mentioned, i.e., the period in which each of the methodological perspectives was most intensively applied and in which it constituted (one ot) the major approach(es) to lexical semantics of that era. This implies that studies within a particular methodological paradigm may well be found outside the periods mentioned in the figure, at a time when that paradigm is no longer (one ot) the major one(s) in lexical semantics. Thus, componential approaches to meaning description can still be found today (see, e.g., Lipka 1985), but it would be misleading to say that structuralist componential analysis is the leading, theoretically most advanced conception of lexical semantics. Likewise, the fact that the classification takes its starting-point around 1870 does not mean that semantic studies are completely absent in the previous period, but merely that it is around that date that semantics be gins to receive frequent and continuous attention , and that lively discussions on semantic matters begin to take place. Still, I would like to stress that the classification is intended to cover

History oflexical semantics

389

only the history of linguistic lexical semantics, i.e., semantic studies that can be situated within linguistics as a discipline that combines empincal description with theoretical analysis. In this sense, the older (roughly, pre-nineteenth century) traditions of semantics (speculative philosophy, practicallexicography without theoretical interests, and the rhetorical classification of stylistic tropes) are not envisaged here.

6.

Main lines in the history of lexical semantics

The survey of Table 1 is based on a methodological analysis of the various approaches. The three headings used to distinguish among them (empirical domain I observational principle I explanatory principle) are well-known notions in the philosophy of science. They correspond roughly with the questions 'What kind of phenomena does a theory deal with?', 'From what point of view are those phenomena described?', and 'What are the basic principles used in explaining the empirical observations?'. As with any classificatory perspective, this methodological approach entails certain choices that might be reversed from another point of view. For instance, Generative Semantics and Interpretive Semantics are not mentioned separately because they are distinguished mainly by the position they want a semantic component to take within a transformational grammar, but otherwise share the basic methodological assumptions of transformationalist semantics. Even the latter is not singled out for separate treatment, in spite of its mentalist views, which seem to contradict the initial anti-psychological position of structuralism. Initially, however, the mentalist self-definition of transformational semantics hardly played a major methodological role, among other things because introspection had always been tacitly accepted as one of the methodological techniques of semantics, even by structuralists. (Distributionalists are the possible exception, but then, their failure merely shows the difficulty of avoiding the introspective method.) Methodologically speaking, transformationalist semantics is a culmination of, rather than a departure from, the structuralist approach. The Katzian definition of the methodology of lexical semantics (1972) is typically structuralist in that it assumes the existence of an autonomous level or semantic structure that is distinct from cognition at large and that underlies the linguistic judgments of native speakers on such matters as implication, homonymy, synonymy and the like. The basic methodological tenet of structuralist semantics, the fundamental point by which it defines itself in contrast

390

Theory and method

with the older historical-philological tradition is the following (see Weisgerber 1927): meaning is not primarily a psychological, but a linguistic phenomenon; hence, it is by studying linguistic phenomena that semantic structures can be discovered. The whole point is easily illustrated by the notion 'distinctive opposition'. Given a number of semantically related words within a lexical field, a structural analysis will have to reveal the distinctive oppositions among the items in question, and stop there, discarding other attributes as linguistically irrelevant. In a restricted set of items such as ram: ewe :: stallion : mare, the opposition ± MALE is distinctive. The fact that ram and ewe are covered with relatively long hair from which clothes can be made, whereas stallion and mare are not, is then considered to be linguistically irrelevant: it is part of our encyclopedic knowledge of the world, but it is not a part of the structure of language. Structural semantics originates by making precisely that distinction, viz. between semantic structure (a linguistic phenomenon involving necessary and sufficient conditions for explaining lexical and semantic relations) and cognition (a psychological phenomenon involving encyclopedic knowledge). By adhering to the idea that there is such an autonomously linguistic level of semantic structure, transformational semantics is basically structuralist. Next, what makes it a culmination of the structuralist approach is the range of linguistic facts it indicates as the observational basis of lexical semantics. In fact, it brings together the three major kinds of facts to which structuralist semanticists had addressed themselves: the semantic field relations with which structuralist semantics originated and which dominated structuralist semantics up to the beginning of the 1960s, the lexical relations highlighted by Lyons (1963), and the syntagmatic aspects of semantic structure which had once been indicated by Porzig (1934), but which had more or 10 less been forgotten until they were taken up simultaneously in the componential theories of Katz and Fodor (1963) and Pottier (1963). It is of course to these syntagmatic aspects of semantic structure, embodied in the notion of selection restrictions, that most of the transformationalist attention was directed, at least in the sense that it was precisely the role of semantics with regard to syntax that engendered the debate between Generative and Interpretive Semantics. However, the methodologically basic point is that the Katzian definition of the methodology of lexical semantics (1972) incorporates into the observational basis of semantics all three kinds of structural relations that had been studied separately by pretransformationalist structuralists. It is the linguistic judgments of native speakers with regard to these relations that reveal the autonomous semantic structure of language; con-

History oflexical semantics

391

versely, a fonnal description of that autonomous level of semantic structure explains the behavior of the native speakers. It is true, though, that Generative Semantics gradually moved towards a position in which the autonomy of linguistics appeared to be threatened. The growing importance given to semantic factors within the transfonnational framework runs parallel with a move away from the original ideal of linguistic autonomy (an ideal that is no less paramount in the Chomskyan conception of grammar than in the structuralist conception of lexical semantics). One need only recall, e.g., Lakoff's incorporation of pragmatic considerations in the fonn of transderivational constraints (1970) to appreciate the extent to which the Generative Semantics movement slid away from the autonomist ideal. To the extent that Generative Semantics threatened to violate that basic ideal, it is right to say, following Newmeyer (1980), that Generative Semantics destroyed itself, but only with the proviso that it destroyed itself as a branch of transformational grammar. It is remarkable, in fact, that some of the most prominent figures of Cognitive Grammar (in particular, Lakoff and Langacker) were fonnerly much closer to the Generativist than to the Interpretivist approach. As an approach within the transfonnational paradigm, Generative Semantics disappeared because it became incompatible with some of the basic assumptions of transfonnational grammar, but the primacy it gave to meaning now reappears within a new school of thought that has abandoned the autonomist creed. We can see, then, that the transition from transfonnational to Cognitive Semantics does not occur at random, but is detennined by substantial points of debate. Let us therefore have a closer look at the developmental lines that interconnect the traditions identified in Table 1. The transition from the first to the second stage has already been described: in trying to establish an autonomously linguistic approach to the study of meaning, structuralist semantics rejected the historical-philological 'confusion' of encyclopedic cognition and semantic structure. And of course, the structuralist primacy of synchronic analysis also favored abandoning the older, primarily diachronic approach. To see how structural semantics gave way to the logical paradigm, it is once more the incorporation of lexical semantics into transfonnational grammar that has to be taken into account. In line with the general tendency of generative grammar, this incorporation led to an increased interest in the fonnalization of semantic analyses. Justified criticism with regard to Katz's fonnalizations (see e.g. Bierwisch 1969) induced people like McCawley (1971) and Lakoff (1972) to take over the more sophisticated fonnalisms of

392

Theory and method

predicate logic. When it was recognized that logical and semilogical formalizations alike have to be referentially interpreted (see e.g. Lewis' 1972 wellknown sarcasm with regard to the uninterpreted 'Markerese' of earlier transformationalist semanticists), the gradually spreading interest in logic eventually led to the adoption of the intensional-logical format of Montague grammar. The transition from the structuralist to the cognitive paradigm can be attributed to two factors, one of which has already been mentioned: the Generative Semanticists' tendency to broaden the empirical scope of generative grammar contained a movement away from the autonomy aspired at by Chomskyan linguistics. Next to this, Cognitive Grammar can be considered as taking seriously the methodological consequences of the mentalism proclaimed so vigorously by transformational semantics. Although (as we have seen) the basic methodological position of transformational semantics is in line with the autonomist views of structuralism, its mentalist pretensions did become methodologically relevant at a particular point in its development. In fact, the mentalist strand in transformational semantics became important in the debate between the decompositional and the axiomatic approach to meaning representation when the choice between both approaches came to be seen as relying on the psychological reality of a set of primitive, nondecomposable semantic features (the basic vocabulary of componential representations). As might be expected, the discussion about the psychological reality of the set in question moved away from the transformational framework to a more purely psychological approach. That is to say, through the appeal to experimental psychological evidence, the discussion on meaning representation was taken over by psychologists pure and simple. (See, e.g., the debate between Glass and Holyoak 1975 and Smith, Shoben and Rips 1974.) But psychologists obviously do not bother about the ideal of linguistic autonomy: in their survey of psycholexicology, Smith and Medin (1981) explicitly suggest that a psychologically adequate representation of linguistically encoded concepts may have to rely on a complete representation of all the cognitive capacities of man. In this sense, taking the psychological implications of the transformationalist mentalism methodologically seriously carries one away from the autonomist approach of transformational grammar. This is, of course, precisely the position taken by Cognitive Semantics. If semantics (as grammar in general, according to Chomsky) is indeed a psychological discipline, linguistic meaning can no longer be studied in isolation from the other cognitive capacities of man, but it will have to be con-

History oflexical semantics

393

sidered in the light of its basic psychological function, e.g. as one of the basic cognitive tools of the human mind. In this sense, then, transformational semantics represents the grandeur et decadence of structural semantics, On the one hand, it is the methodological culmination of the latter, but on the other, its predilection for formalization and its mentalist pretensions embody the springs of its own dissolution. In particular, by giving up the earlier anti-psychological views of structuralism in favor of a mentalist self-conception, transformational semantics undermines its own autonomistically structuralist methodology. The developmental lines set out here are summarized in Figure 1; in bringing out the pivotal position of transformational semantics, the figure clearly refines the classification of Table 1, in which transformational semantics is incorporated into the structural approach without further nuances, On a more general level, we can now see that the history of lexical semantics exhibits an oscillation between an autonomous and an experiential, psychological approach. Structuralist semantics reacted against the psychological perspective of historical-philological semantics, but nowadays, the pendulum is swinging back (partly out of an attempt to take the mentalist claims of neostructuralist transformational semantics methodologically seriously). It should furthermore be noted that, with regard to this basic dichotomy, logical semantics ranges with structural semantics for two interconnected reasons. First, logical semantics is definitely not a psychological approach. Its subject matter (the truth-functional relationship between linguistic expressions and the world) is defined regardless of the cognitive subject: the truth conditions of a proposition exist regardless of the person(s) entertaining that proposition, and can be described accordingly. Second, by abstracting away from subjective, psychological factors, structuralism and logical semantics alike tend to be objectivist approaches. In both cases, meaning is primarily defined as an objective entity that is independent of the individual mind: the supra-individual structure of language, or possibleworld truth conditions. As opposed to this. historical-philological semantics and Cognitive Semantics take into accoullt the subjective, individual Sitz im Leben of language. The opposition between 'objectivist' and 'subjectivist' approaches leads us back to the Diltheyan distinction between, respectively, 'natural' and 'human' science (see section 2). Against the background of the history of lexical semantics, the fundamental methodological affinity between cognitive and historical-philological semantics seems to reside primarily in the fact that, through their experiential, psychological orientation, both

394

Theory and method

are subjective, 'human' sciences in the Diltheyan sense, in contrast with the objectivist tendencies of structural and logical semantics, which seem to take

HISTORICALPHILOLOGICAL SEMANTICS

i

meaning should be studied from a synchronic, nonpsychological, autonomous point of view

~ STRUCTURAL SEMANTICS

paradigmatic and syntagmatic relations reveal the autonomous semantic structure of the language

TRANSFORMATIONAL SEMANTICS

-

mentalism

a psychologically adequate study of meaning requires a non-autonomistic approach

fonnalization

-

a fonnally adequate representation of meaning requires a model-theoretic interpretation

1

1

COGNITIVE

LOGICAL

SEMANTICS

SEMANTICS

Figure I. Stages and transitions in the history of lexical semantics

History oflexical semantics

395

their methodological mode Is primarily from the 'natural' sciences. This paper is obviously not the place to decide the issue between both basic approaches; the important thing is merely to see how such a major methodological dichotomy underlies the similarities between prestructuralist and Cognitive Semantics that we have described in section 4.

7.

Conclusion

In the foregoing pages, the following theses have been developed. Prima, Lakoff's statement that 'The classical theory of concepts and categories has been accepted in the West for two thousand years' (l982a: 99) overstates the novelty of the cognitive approach. Within the history of lexical semantics as a branch of linguistics, it is rather a partial return to the methodological position of the prestructuralist, historical-philological tradition of semantic research. Secunda, the major methodological alternative in the history of lexical semantics is a Diltheyan dichotomy between semantics as an objectivist, autonomist, and as an experiential, psychological approach. Tertia, the historical-philological tradition contains a huge amount of empirical observations, classificatory mechanisms, and explanatory hypotheses that may be extremely valuable for the further development of Cognitive Semantics. It is precisely because of the large quantity of studies in the historicalphilological paradigm, that this article has had to be restricted to some main lines. Nevertheless, I hope that it will have become clear that a closer look at prestructuralist semantics will be worthwhile. This conclusion should not, however, obscure the fact that Cognitive Semantics also takes some fundamental steps forward in comparison with the older tradition. For instance, Gestalt psychologists have long been aware of the figure/ground alignment in perception, but its importance for lexical-semantic analysis has come to be appreciated only recently through the work of Langacker and Talmy. Similarly, the prototypical conception of lexical structure opens up hitherto scarcely explored possibilities for bridging the gap between the synchronic and the diachronic approach to linguistic analysis. Although Cognitive Semantics shares its fundamental methodological position with historicalphilological semantics, it is much more than a mere return to prestructuralist semantics.

396

Theory and method

Notes 1.

2.

3.

4.

5.

It may be useful to point out that I use the tenn 'philological' mainly in a

methodological sense; I intend it to be the indication of a methodological approach to the study of semantic phenomena, and not just of a topic or field of research. In this sense, the studies grouped under the label 'philological' need not be concerned with the study of a particular text or a particular literary tradition; rather, they are by and large supposed to share some important methodological characteristics with philological studies in the narrow (literary) sense. In particular, the most typical representatives of this line of semantic research are concerned with the historical interpretation oflinguistic material as the expression of thought, philology in the narrow sense being concerned mainly with the historical interpretation of literary works as the expression of an individual author's views. (Needless to say, this fonnulation refers to the characteristics of literary philology in its prestructuralist stage; in our own century, literary hermeneutics has changed just as much as linguistic semantics. The changes are at least partly similar: by moving away from a roman tic insistence on the individuality of the author to a modernistic interest in the literary work as an autonomous 'textual object', the development of literary scholarship parallels the linguistic transition from the historical towards the structural study of meaning.) Neither does my treatment of BnSal imply that his ideas are completely novel within the history of linguistics at large. As was pointed out to me by Pierre Swiggers (to whom I am gratefiJl for a number of interesting remarks with regard to an earlier draft of this paper), it may be revealing to link Breal's views with eighteenth-century thought. In view of the chronological limits of this paper, this is a line which will not be pursued here. Different approaches within the etymological study of words might be used to identify particular schools, trends, and developments within the historical tradition of lexical research that have not been mentioned separately in this article. Thus, the W6rter und Sachen-movement, and the dialectgeographical method introduced by Gillieron deserve explicit treatment in a complete history of lexical semantics. Still, readers who may have doubts as to the pervasiveness of the psychological outlook in diachronic semantics will find Weisgerber's attack on historical semantics in general and psychologism in particular (1927) extremely instructive. At the same time, Breal's insistence on the fact that linguistics is a human instead of a natural science, is a reaction against the biological naturalism of Schleicher, according to which language is an organism evolving independently of man.

History oflexical semantics 6.

397

This formulation explicitly uses the phrase fundamentally structuralist claim instead of fundamental structuralist claim to account for the fact that the Bloomfieldian, behaviorist branch of structuralism does not make the claim, but instead discards the semantic description of linguistic forms from the domain of linguistics. The point is this: to the extent that structuralists accept the feasibility of a semantic description of language, they typically posit the existence of an autonomously linguistic, non-encyclopedic lexical-semantic structure, but this feasibility is not accepted by all forms of structuralism. References to structuralist semantics in this paper obviously exclude those structuralists that do not accept the possibility of a linguistic semantics. (Moreover, it should be kept in mind that not all semantic studies by linguists that take an overall structuralist approach need be typically structuralist on the methodological level: see note la.) 7. There are two additional remarks to be made with regard to Weisgerber's comments on Hecht. In the first place, nineteenth-century linguists sometimes make a distinction between BegrUf(concept in the most neutral sense) and Vorstellung (concept as a psychological entity, mental representation). As such, Hechfs use of the term Vorstellung is itself an indication of his psychological outlook. In the second place. Weisgerber's major objection against the older tradition of semantics is the following: identifying meaning and cognition entails that meanings exist and develop independently of language; meanings (thoughts) have an independent psychological life, and words are simply attached to them as labels. It may be good to mention that this objection is ill-founded. Identifying semantic and encyclopedic knowledge need not imply that language may not help to shape and support thought, nor does it entail that cognition develops autonomously, without the help of the symbolical tools provided by language. On the contrary, it is precisely the acceptance of an intimate connection between language and cognition that makes it possible to envisage a relationship of reciprocal influence between them. (For an elaboration of this point, see Geeraerts 1985a, sections 1.4. and 1.6.) 8. To avoid misunderstandings: the 'synchronic' character of the structural method does not exclude diachronic studies, but conceives of change as a succession of distinct synchronic stages. As such, the investigation into synchronic structures is taken to be methodologically basic with regard to the study of the transition between synchronic stages. However, such stages are practically very hard to identify, since linguistic systems change neither en bloc nor abruptly. 9. The structuralist views of Reichling (1935), which come close to the prototypical view (see Geeraerts 1983d), will not be discussed as they do not belong to 'mainstream' structuralism. 10. 'More or less': the Firthians might be mentioned (with some provisos) as an exception.

Chapter 16 The theoretical and descriptive development of lexical semantics

Originally published in Leila Behrens and Dietmar Zaefferer (eds.). 2002, The Lexicon in Focus. Competition and Convergence in Current Lexicology 23-42. Frankfurt: Peter Lang Verlag. The picture of the history of lexical semantics painted in Chapter 15 is restricted in two ways. First, it could only take into account the very beginnings of Cognitive Semantics, focusing in particular on prototype theory. Second, it concentrated on the relationship between Cognitive Semantics and prestmcturalist semantics. Although it did give stmcturalist semantics its rightful place in the chronology, it paid relatively little attention to the offshoots of the stmcturalist and generativist tradition that exist alongside Cognitive Semantics. Almost fifteen years after the initial article, I returned to the topic in a brief attempt to remedy these two points. In the present chapter, which was originally published in a volume featuring examples of the major traditions of lexicological research, a specific position is reserved for the 'neostmcturalist' and the 'neogenerativist' approaches that continue the lines of the tradition that dominated before Cognitive Semantics entered the scene. Further, the history of lexical semantics is not just described as a succession of theoretical approaches, but also as a gradual widening of the descriptive domain of semantics in which all major traditions play an important role. It is shown that there is an underlying pattern to this development of empirical broadening, going from semasiology to onomasiology, and from purely stmcture-oriented to usage-oriented phenomena. The paper rounds off with a consideration of the possibilities for future convergence between the different theoretical approaches: an awareness of history is not just important for recovering past achievements that might otherwise be lost, but also for seeing more clearly what the future might bring.

1.

Introduction

Existing treatments of the history of linguistic semantics, such as Nerlich (1992) and Gordon (1982), concentrate on the analysis ofthe views of individual authors or separate periods rather than on the general lines of development that characterize the development of semantics. Counterbalancing

Theoretical and descriptive development

399

such more or less atomistic (but clearly indispensable) analyses of individual views and separate stages in the development of the discipline, I earlier published a number of articles (specifically, 1988a and 1988b) that tried to present a rational reconstruction of the overall progression of linguistic lexical semantics. In these articles, I concentrated particularly on the way in which the development of lexical semantics was pushed forward by competing theoretical conceptions of word meaning. In the present paper, I will have a look at the history of lexical semantics from a slightly different angle: I will reconsider the classification into different theoretical approaches, taking into account the developments that occurred in the course of the 1990s, and I will then consider the way in which these various theoretical traditions concentrate on different empirical domains. At the same time, I would like to show that these shifts in descriptive emphasis fit together in an encompassing conception of the empirical domain of lexical semantics: it can be shown, I suggest, that the various theoretical approaches that may be distinguished in the history of the discipline gradually occupy and fill out the potential range of investigation that could ideally be associated with lexical semantics. A similar view of the empirical development of lexical semantics was presented in Geeraerts (l999b). In comparison, the picture in the present article is a more comprehensive and more synthetic one. As I will inevitably have to deal with quite a number of different approaches, I will, for the sake of succinctness, assume a basic familiarity with most of them. The various approaches will be briefly presented, but it is obviously beyond the scope of the present contribution to do so in much conceptual or bibliographical detail.

2. Theoretical currents in the history of lexical semantics For starters, I would like to suggest that four major theoretical currents shape the development of lexical semantics: prestructuralist semantics, structuralist and neostructuralist semantics, generativist and neogenerativist semantics, and Cognitive Semantics. (These currents, by the way, involve linguistic semantics only: psycholinguistic or philosophical approaches to semantics are not included in the overview.)

400

Theory and method

2.1. Prestructuralist semantics Prestructuralist diachronic semantics dominated the scene from roughly 1870 to 1930. This does not mean, to be sure, that there is no form oflexical semantics before 1870. It is only by the middle of the nineteenth century, however, that research into word meaning establishes itself as a distinct subdiscipline of the new science of linguistics. At the level of individual researchers, prestructuralist diachronic semantics is represented by Breal and many others, like Paul, Darmesteter, Nyrop, and (at the end of the relevant period) Carnoy and Stem. A systematic treatment of the main figures and currents may be found in Nerlich (1992). At the methodological level, the type of semantic research meant here may be characterized by the following three features. First, in line with the overall nature of nineteenth-century linguistics, the orientation is a diachronic one: what semantics is interested in, is change of meaning. Second, change of meaning is narrowed down to change of word meaning: the orientation is predominantly semasiological rather than onomasiological (or grammatical, for that matter). Third, the conception of meaning is predominantly psychological, in a double sense. Lexical meanings are considered to be psychological entities, that is to say, (a kind of) thoughts or ideas. Further, meaning changes (the main domain of research for prestructuralist semantics) are explained as resulting from psychological processes. The general mechanisms that are supposed to underlie semantic changes and whose presence can be established through the classificatory study of the history of words, correspond with patterns of thought of the human mind. A concept like metonymy is not just a linguistic concept, it is also a cognitive capacity of the human mind. 2.2. Structuralist and neostructuralist semantics The origins of structuralist semantics are customarily attributed to Trier (1931), but while Trier's monograph may indeed be the first major descriptive work in structuralist semantics, the first theoretical and methodological expose of the new approach (which, needless to say, took its inspiration from De Saussure) is to be found in Weisgerber (1927). Weisgerber criticizes prestructuralist historical semantics precisely on the three characteristic points mentioned above. First, the study of meaning should not be atomistic but should be concerned with semantic structures. Second, it should be synchronic instead of diachronic, and third, the study of linguistic meaning

Theoretical and descriptive development

401

should proceed in an autonomously linguistic way. Because the meaning of a linguistic sign is detennined by its position in the linguistic structures of which it is a part, linguistic semantics should deal with those structures directly, regardless of the way in which they may be present in the individual's mind. Because the subject matter of semantics consists of autonomous linguistic phenomena, the methodology of linguistic semantics should be autonomous, too. The realization of this attempt to develop a synchronic, non-psychological, structural theory of semantics depended on the way in which the notion of semantic structure was conceived. In actual practice, there are mainly three distinct definitions of semantic structure that have been employed by structuralist semanticians. More particularly, three distinct kinds of structural relations among lexical items have been singled out as the proper methodological basis of lexical semantics. First, there is the relationship of semantic similarity that lies at the basis of semantic field analysis, inaugurated by Trier and ultimately leading to componential analysis in the work of anthropological linguists such as Goodenough (1956) and Lounsbury (1956), and, apparently independently, in the work of European structuralists such as Pottier (1964). Second, there are unanalyzed lexical relations such as synonymy, antonymy, and hyponymy; these were for the first time systematically selected as the methodological basis of structural semantics by Lyons (1963). Third, syntagmatic lexical relations were identified by Porzig (1934) under the name of 'wesenhafte Bedeutungsbeziehungen'; they later reappeared as selectional restrictions in the neostructuralist kind of semantics that was incorporated into generative grammar by Katz and Fodor (1963). The contemporary descendants of structuralist semantics (which may be tenned 'neostructuralist' to distinguish them from the original approaches) link up with these three distinct interpretations of the notion of lexical structure. First, lexical field theory in its original fonn is most explicitly further developed by Coseriu and his followers (Coseriu and Geckeler 1981). The approach does not, however, seem to generate a lot of new work anymore, and its own advocates admit to a certain stagnation (Geckeler 1993). Second, the relational approach has been further developed by Cruse (1986); in his most recent work (2000b), however, the structuralist origins of his approach are attenuated by the impact of Cognitive Semantics. A consequent elaboration of the relational perspective is to be found in the work of Mel9uk and his collaborators (1988, 1996), whose purpose is to identify a universally valid set of relations that allows for the description of

402

Theory and method

all semantic associations between lexical items in the vocabulary of any given language. The approach is not a purely theoretical one: it involves the preparation of a number of formalized dictionaries (Melyuk 1984). In addition to Melyuk's work, a major descriptive achievement based on a relational conception of lexical structure is the WordNet project (Fellbaum 1998). WordNet is an online electronic lexical database organized on relational principles, which now comprises nearly 100,000 concepts. Third, following the lead provided by Firth (1957), the syntagmatic conception of lexical relations developed into a highly productive tradition of collocational analysis (where 'collocations' are, roughly defined, sets of regularly co-occurring words). Many different specific methods of collocational analysis exist, and there is no single name that can be mentioned as the leading figure in this form of semantics. In the present context, it may be sufficient to mention the incorporation of collocational analysis in the framework developed by Halliday (Halliday and Hasan 1976), together with the highly successful incorporation of the approach into lexicography (among many others, see Sinclair 1987, Moon 1998). Because a collocational analysis naturally involves a description of the facts of lexical cooccurrence in actual language use, this approach links up directly with corpus linguistics (which, as we shall see, is an important point for the overall line of the present article). 2.3. Generativist and neogenerativist semantics All through the second half of the 1960s and the major part of the 1970s, the neostructuralist, generativist model of lexical-semantic description introduced by Katz and Fodor (1963, later developed by Katz, specifically in Katz 1972) formed a reference point for studies in lexical semantics. The appeal of Katzian semantics was at least partly due to its incorporation into generative grammar; it profited from the superior position that the generative paradigm occupied in linguistic theorizing in the period in question. Intrinsically (i.e. as a theory of lexical semantics), Katzian semantics combined a culmination of the structuralist approach with two characteristics that were intimately connected with its incorporation into generative grammar. First, Katz took over the Chomskyan requirement that linguistic analyses be rigidly formalized. In particular, componential analysis in the Katzian model was at the same time a method of descriptive analysis and a formal apparatus that seemed indispensable to comply with the requirement of algorithmic formalization imposed by Chomskyanism.

Theoretical and descriptive development

403

Second, Katzian semantics took over the mentalist self-conception of Chomskyanism. By defining the subject matter of semantics as the competential 'ability to interpret sentences' of the language user, semantics came to share the promises of explanatory adequacy that constituted so much of the appeal of generative grammar. Third, Katzian semantics brought together the three types of semantic relations that could lie at the basis of structuralist semantic theories. In the first place, paradigmatic similarity relations along the lines of lexical field theory showed up in Katz and Fodor's adoption of componential analysis. In the second place, syntagmatic restrictions on the combination of words were captured in selectional restrictions (such as that the direct object of eat has to refer to something edible). And in the third place, the paradigmatic lexical relations highlighted by Lyons (1963) were not discussed in the 1963 article, but in 1972 (apparently as a result of the publication of Lyons's book), Katz pointed out explicitly that semantic theory should be concerned with lexical relations such as synonymy, antonymy, and hyponymy. In short, Katzian semantics was a singular combination, within the framework of generative grammar, of a basic structural semantic methodology, a mentalist philosophy of language, and a formalized descriptive apparatus. The further development of lexical semantics was characterized by two tendencies. In each case, semantics moved away from the structuralist pole of the Katzian synthesis towards one of the other two poles. On the one hand, the demands of formalization diminished the structuralist influence in semantics in favor of logical approaches to meaning analysis. On the other hand, attempts to take the mentalist position of Katzian semantics seriously led to a straightforward psychological, cognitive orientation in semantic studies. For reasons to be explained, the former approach can be called 'neogenerativist'; it will be described presently. The latter, cognitive approach is introduced in the next section. In the context of an historical overview of lexical semantics, the evolution towards logical semantics basically means a shift of emphasis from lexical semantics to sentential semantics. Truth conditions are the primary point of interest of logical semantics, and truth conditions are properties of propositions, not of individual words. Thus, Thomason's statement that a semantic theory need not specify the way in which items such as walk and run differ in meaning (1974: 48) is typical for the shift in interest from lexical relational structures to sentential structures. As a result of this shift in interest, the logical approach to lexical semantics initially remained somewhat of a minor tradition - within the field of lexical semantics, that is.

404

Theory and method

In the 1990s, however, Pustejovsky (1995) gave a new impetus to the Katzian ideal of a formalized semantic representation by basing it on a logical rather than a featural formalism. At the same time, he goes way beyond the Katzian approach by emphasizing the necessity of building a lexicon that is truly generative - in the sense that it does not just consist of an enumeration of word senses, but incorporates the possibility of formally deriving new readings from stored ones. In this sense, Pustejovsky's approach (which links up closely with lackendoff's) is a 'neogenerativist' one: it develops the Katzian ideal of a formal semantic representation by introducing semantic flexibility and a logical formalism. 2.4. Cognitive Semantics The Katzian approach is characterized by a specific tension between theoretical starting-point and actual methodology. Using psycholinguistic data could be the ultimate consequence of the mentalist position of Katzian semantics; if semantic descriptions do indeed have psychological reality, it should be methodologically acceptable to use all kinds of psychological evidence. On the other hand, there are two other aspects of Katz's position that opposed this methodological extrapolation of the mentalist stance. Both the generative notion of competence and the structuralist attempt to develop an autonomous method for linguistic semantics led to reluctance with regard to the use of psychological data. In the light of Weisgerber's charges against psychological approaches in semantics, it can established that Katz's mentalism is not a straightforward methodological position: it characterizes the object of the investigation as something that is psychological real, but it apparently does not influence the method used to study that object. That method remains structural, in the sense that it is based on static relations between linguistic elements, rather than on actual psychological processes. However, other linguists than Katz himself actually pursued the nonautonomous methodological approach that could be derived from the Katzian position. Linking up with existing work on natural language categorization and meaning representation as carried out in psycholinguistics and artificial intelligence, this kind of research opens up the field of lexical semantics to the more general question 'What does an adequate model of man's use and knowledge of meaning look like?'. To the extent that researchers in psycholinguistics and artificial intelligence generally do not estimate that the linguistic capacities of man can be studied in isolation from his other cognitive capacities, the autonomist methodological ideal of struc-

Theoretical and descriptive development

405

turalism is abandoned, not just by using psycholinguistic data, but more generally by incorporating the study of natural language semantics into cognitive science at large. In linguistic semantics, this cognitive reorientation has led to the school of Cognitive Linguistics of which Lakoff (1987) and Langacker (1987, 1991) are the leading figures. Cognitive Semantics has proved itself to be a highly productive paradigm for linguistic research. As far as lexical semantics is concerned, it is perhaps the most productive of the current approaches, as will become clear in section 3.2. (see Violi 1997 for an introduction and further refinements, and compare Ungerer and Schmid 1996). Among the major conceptual contributions of Cognitive Semantics to the study of lexical meaning, the following should be mentioned. The prototypical theory of categorial structure developed in psycholinguistics by Eleanor Rosch is taken as the basis for models of the internal structure of natural language categories (see Taylor 1995, Geeraerts 1997a, Mangasser-Wahl 2000). (Although it is impossible to elaborate the point in the present context, I should add that I see the Wierzbickian approach to semantics - see Wierzbicka 1996, Goddard 1998 - as one of the variants of the prototype-theoretical conception.) Along the lines of Lakoff and Jolmson (1980), a renewed interest in metaphor has led to a new wave of research into the epistemological role of cognitive models and their experiential background. Conceptual metaphors, as sets of related metaphorical expressions with the same source and target domain, are defined as metaphorical conceptions that transcend the boundaries of individual lexical items. This type of research has further developed into new investigations of semantic structuring at large and metonymy in particular (Panther and Radden 1999). Fillmore's scenes-and-frames approach to semantics takes its startingpoint in the assumption that the human conceptual apparatus does not consist of isolated concepts, but is organized into larger, internally structured wholes. These larger chunks of knowledge, comprising coherent sets of human beliefs, actions, experiences, or imaginations, are called 'scenes'. 'Frames', on the other hand, are the linguistic means available to refer to (aspects of) the scene. Frame theory has proved a stimulating framework for the description of verbal meaning, both theoretically and lexicographically: see Fillmore (1977a), Fillmore and Atkins (1992, 2000).

406

Theory and method

The background of the basic level hypothesis (see Berlin 1978) is the ethnolinguistic recognition that folk classifications of biological domains usually conform to a general organizational principle, in the sense that they consist of five or six taxonomical levels, one of which is the 'generic' level. To the extent that the generic level is the core of any folk biological category, it is the basic level. The generic level, in other words, is the most salient one within the taxonomy: within the lexical field defined by the taxonomy, the generic level specifies a set of preponderant items. The basic level thus embodies a set of naming preferences: given a particular referent, the most likely name for that referent from among the alternatives provided by the taxonomy will be the name situated at the basic level. 2.5. An overview of the developments Given this bird's eye overview of the traditions that have dominated the course of development of lexical semantics, Figure 1 suggests a schematic representation of the relevant lines of development. Each of the boxes in the figure represents one of the approaches identified above. While the vertical axis is basically a historical one, the bottom line of the figure charts the approaches that are currently around. The lines connecting the boxes indicate the relationships between the approaches. The double arrow between prestructural historical semantics and structural semantics indicates that the latter is a reaction to the former. The single arrow between structuralist semantics and generativist semantics emphasizes the methodological continuity between both; as described above, though, generativist semantics also adds some crucial new features to the study of the lexicon. A similar relationship links generativist semantics to neogenerativist semantics: there is a fundamental continuity, but there is also much that is new in Pustejovsky's approach. The relationship between structuralist and neostructuralist semantics is a somewhat more straightforward line of development, while the relationship between generativist and Cognitive Semantics is rather one of contrast (hence the double arrow). Two additional features of the development sketched in the figure should be highlighted. First, the approach inaugurated by Katz appears to have had a pivotal function in the history of lexical semantics: it sununarizes the structural approach, but at the same time, the two innovations that it added to a more traditional structuralist methodology eventually led to forms of semantics

Theoretical and descriptive development

407

that went far beyond the initial structuralist position. The shift towards neogenerativist semantics on the one hand and toward Cognitive Semantics on the other is a consequence of the attempt to take, respectively, the formalized and the mentalist character of Katzian semantics seriously.

Prestructuralist Semantics

Structuralist Semantics

Generativist Semantics

Ir

Neostructuralist Semantics

Cognitive Semantics

Neogenerativist Semantics

Figure 1. Main lines in the history oflexical semantics

Second, as I have explained at greater length elsewhere (1988a), the cognitive approach is to a large extent a return to the questions and interests of the prestructuralist tradition of diachronic semantics. For one thing, the cognitive revival of diachronic semantics links up with the gigantic amount of historical work done in the prestructuralist stage of the development of lexical semantics. Further, Cognitive Semantics and traditional historical semantics share, by and large, a psychological conception of meaning. And finally,

408

Theory and method

both approaches start from an encyclopedist conception of meaning, in the sense that lexical meaning is not considered to be an autonomous phenomenon, but is rather inextricably bound up with the individual, cultural, social, historical experience of the language user. On these three points, then, there is a basic methodological contrast between Prestructural and Cognitive Semantics on the one hand, and structuralist semantics (and its immediate neostructuralist descendants) on the other. Against the background of this fundamental methodological distinction, the neogenerativist approach shares with Cognitive Semantics an interest in the flexibility and dynamism of meaning. Pustejovsky's work, as we mentioned above, is very much concerned with the mechanisms of polysemy, just like a lot of Cognitive Semantic work. However, the Pustejovskian approach is reluctant to adhere to what is, in contrast, one of the basic assumptions of Cognitive Semantics, viz. the encyclopedic nature of meaning - the idea that it is hardly useful for the analysis of natural language to adhere to a strict distinction between the level of linguistic meaning stricto sensu and the level of world knowledge. By maintaining the distinction, formalized semantics of the lackendoff/Pustejovsky school remains close to the tradition of structural semantics. (For a good illustration of the consequences of accepting or rejecting the distinction, see the discussion between Taylor 1996 and Jackendoff 1996.) In short, the history of lexical semantics is characterized by a succession of different theoretical approaches that are related by lines of similarity and extrapolation as well as mutual opposition. As a basic underlying distinction, we may single out the contrast between an encyclopedist approach on the one hand, and a more restrictive approach on the other.

3. The empirical scope of lexical semantics Given the main traditions of lexical semantics as identified in the previous paragraph, what has been their contribution to lexical semantics in terms of the empirical field of research covered? Answering that question requires a conceptual map of the field of lexical semantics, and such a map, I believe, should be based on two fundamental distinctions: the distinction between semasiology and onomasiology, and the distinction between the qualitative and the quantitative aspects of lexical structures. Once the conceptual map has been set up, we can chart the contribution that was made by the various theoretical currents as described above.

Theoretical and descriptive development

409

3.1. A conceptual map of lexical semantics Before we can go on to present the distinction between semasiology and onomasiology, and that between the qualitative and quantitative aspects of lexical structure, it should first be pointed out that these dimensions do not in themselves exhaust the possibilities for distinguishing between different forms of doing lexical semantics. At least two additional distinctive dimensions would have to be included to get a truly comprehensive picture: the distinction between diachronic semantics and synchronic semantics, and the distinction between referential and non-referential types of meaning. These two additional dimensions cross-classify with the three basic ones and with each other: one can study onomasiological changes, for instance, or changes of non-referential meaning. The reason for not including them, then, is one of practice rather than principle: in order to get a first picture of the field, I think it may be acceptable to start off with those dimensions that relate to the core area of lexical semantics, viz. the synchronic (rather than diachronic) structure of referential (rather than non-referential) meanings. Also, the dimensions that are not explicitly included in the classification are quantitatively less important in actual lexical semantic research: diachronic semantics was a dominant field of investigation only in the first stage of development of lexical semantics, and non-referential types of meaning (like the emotive or stylistic value of words) have never occupied a central position within the actual research activities of lexical semanticists. Let us now have a closer look at each of the core dimensions. The distinction between onomasiology and semasiology is a traditional one in Continental structural semantics and the Eastern European tradition of lexicological research, but it has hardly found its way to the canonical English terminology of linguistics. It does not appear, for instance, in most standard textbooks of linguistics. Essentially, it equals the distinction between meaning and naming: semasiology takes its starting-point in the word as a form, and charts the meanings that the word can occur with; onomasiology takes its starting-point in a concept, and investigates by which different expressions the concept can be designated, or named. Traditionally, the onomasiological perspective leads to an investigation into lexical structures, i.e. to the study of semantically related expressions (as in lexical field theory, or the study of the lexicon as a relational network of words interconnected by links of a hyponymical, antonymical, synonymous nature etc.). However, studying the designations of a particular concept also opens the way for a contextualized, pragmatic conception of ono-

410

Theory and method

masiology, involving the actual choices made for a particular name as a designation of a particular concept or a particular referent. This shift of perspective involves a shift from an investigation of structure to an investigation of use. The structural conception deals with sets of related expressions, and basically asks the question: what are the relations among the alternative expressions? The pragmatic conception deals with the actual choices made from among a set of related expressions, and basically asks the question: what factors determine the choice for one or the other alternative? Such a perspective automatically leads to a 'quantitative' approach: some onomasiological choices will be more frequent than others. Turning the shift of perspective the other way round, these differences in the probability of occurrence of certain expressions will be reflected on the structural level: we can then say, for instance, that certain expressions occupy a dominant or central role within the set of available expressions. More concretely, what are the different types of investigation that relate to the qualitative and quantitative aspects of onomasiology? The qualitative onomasiological question takes the following form: what kinds of (semantic) relations hold between the lexical items in a lexicon (or a subset of the lexicon)? The outcome, clearly, is an investigation into various kind of lexical structuring: field relationships, taxonomies, lexical relations like antonymy and so on. The quantitative onomasiological question takes the following form: are some categories cognitively more salient than others, that is, are there any differences in the probability that one category rather than another will be chosen for designating things out in the world? Are certain lexical categories more obvious names than others? This type of quantitative research is fairly new. The best-known model to date is Berlin and Kay's basic level model, which involves the claim that a particular taxonomical level constitutes a preferred, default level for categorization (see above). In Geeraerts, Grondelaers, and Bakema (1994), the quantitative approach to onomasiology is generalized: it is shown that onomasiological preferences do not attach to taxonomical levels (as suggested by the basic level hypothesis) but to individual categories. At the same time, corpus-based methods are introduced for determining the factors that influence actual lexical and categorial choices in language use. The distinction may be further illustrated by extrapolating it to semasiological structures. Qualitative aspects of semasiological structure involve the following questions: which meanings does a word have, and how are they semantically related? The outcome is an investigation into polysemy, and the

Theoretical and descriptive development

411

relationships of metonymy, metaphor etc. that hold between the various readings of an item. Quantitative aspects of lexical structure, on the other hand, involve the question whether all the readings of an item carry the same structural weight. The outcome, obviously, is an investigation into prototypicality effects of various kinds. Needless to say, the qualitative perspective is a much more traditional one in semasiological lexicology than the quantitative one, which was taken systematically only recently, with the birth and development of prototype theory. The overall picture of the field of lexical semantics that emerges from the foregoing is charted in Table 1.

Table 1. A conceptual map of lexical semantics SEMASIOLOGY 'QUALITY':

investigating structure (elements and relations) 'QUANTITY':

investigating use and differential weights within structures

senses and semantic links among senses (metaphor, metonymy etc.) prototypicality as involving salience phenomena, core versus periphery

ONOMASIOLOGY

semantic relations among lexical items (fields, taxonomies, networks etc.) differences in cognitive salience between categories, basic level phenomena and entrenchment

3.2. The contribution of the traditions The next step to take involves the recognition that each of the major areas in the overview presented in the previous paragraph is typically covered by a specific theoretical tradition from the history of lexical semantics. Each of the three major traditions mentioned in the first paragraph focuses specifically on a particular area of the conceptual map. The prestructuralist tradition of diachronic semantics, to begin with, deals predominantly with the 'qualitative' aspects of semasiology - with processes like metaphor and metonymy, that do not just function as mechanisms of semantic extension, but that constitute, at the same time, the synchronic links between the various readings of a lexical item. Structuralist semantics, on the other hand, focuses on 'qualitative' phenomena of an onomasiological (a structural onomasiological) kind, such as field relations, taxonomical hierarchies, and lexical relations like antonymy. Cognitive Semantics, finally, shifts the attention towards the 'quantitative' aspects of lexical structure: all forms of

412

Theory and method

prototypicality effects in the semasiological realm, and basic levels on the onomasiological side. This overview of the different descriptive focus of the various traditions of lexical semantics also shows how the conceptual map presented in Table 1 got gradually filled out by different theoretical approaches with different descriptive priorities. There is a historical progression, in fact, from 'qualitative' semasiology to 'qualitative' onomasiology, and hence to an interest in 'quantitative' phenomena on the semasiological as well as on the onomasiological level. The actual historical development from prestructuralist semantics over structuralism and generativism to Cognitive Semantics thus constitutes a gradual extension of the descriptive scope of lexical semantics. The overview of the descriptive preferences of the major traditions of lexical semantics is summarized in Table 2. Table 2. Dominance of the theoretical traditions within lexical semantics SEMASIOLOGY

ONOMASIOLOGY

'QUALITY':

investigating structure (elements and relations)

prestructuralist semantics

structuralist semantics

Cognitive Semantics

Cognitive Semantics

'QUANTITY':

investigating use and differential weights within structures

In order to avoid misunderstanding, it is important to strike a note of warning: the figure charts dominant domains of activity rather than strictly confined fields. The various traditions, that is, are certainly not restricted in their activities to the empirical areas mentioned in the overview. Let us consider some important nuances with regard to each of the major traditions. First, prestructuralist semantics does not have an exclusively semasiological orientation. Onomasiological considerations may enter the classification of types of semantic change wherever necessary. The notion of analogical changes of meaning, in particular, presupposes an onomasiological perspective. 'Analogical change' is the general name for semantic extensions that copy the meaning of another, related word. Analogical changes can be subclassified on the basis of the associative onomasiological link that exists between the expression that is subject to the semasiological extension, and the expression that furnishes the model for the extension. In cases of seman-

Theoretical and descriptive development

413

tic loans, for instance, the associative link between the source expression and the target has a semantic character. In cases of ellipsis (when a word takes over meanings from a syntactically complex expression of which it is a part), the relationship between the two expressions involved in the analogical change is syntagmatic rather than semantic. Second, structuralist semantics does not have an exclusively onomasiological orientation. By focusing on distinctiveness as a basic phenomenon to be considered in lexical semantics, it generally proclaims the primacy of onomasiological phenomena over semasiological phenomena. The research strategy of componential analysis, for instance, is basically to derive the semasiological description of the meaning of individual lexical items from an analysis of the onomasiological relations that the item in question entertains with other items in the same lexical field. Semasiological analysis, in short, is far from absent from the structuralist approach, but depends on an initial onomasiological analysis. Third, in the generativist and neogenerativist approaches that grew out of structuralism, the semasiological orientation is even more outspoken. Katz's componential analysis of bachelor, for instance, focuses on the distinctiveness of the various readings of a lexical item among each other, rather than on the distinctiveness of the items in a lexical field. And Pustejovsky's interest in regular polysemy and the dynamism of meaning has an obvious semasiological orientation. Fourth, Cognitive Semantics is not restricted to prototype theory and basic level research. Specifically, there are at least two important contributions that Cognitive Semantics has made to the structural onomasiological perspective (both of which have been mentioned above): on the one hand, the development of the Fillmorean frame model of semantic analysis, and on the other hand, the introduction of generalized metaphor research in the line of Lakoff and Johnson. Both approaches are in fact extensions of the set of lexical relations studied within the realm of onomasiology. On the one hand, conceptual metaphors involve onomasiological sets of metaphorically related expressions. On the other, the typical perspective of Fillmorean frame semantics is of a metonymical kind (cp. Koch 1999). To consider an example, studying verbs like buy and sell by bringing buyers and sellers and goods and prices into the picture, equals studying the 'referential syntagmatics' of a lexical item: the way in which the referent of the item (in this case, a particular commercial transaction) occurs in reality in temporal and spatial and functional conjunction with other entities, locations, processes, activities, or whatever.

414

Theory and method

Given these nuances with regard to the picture drawn in Table 2, we now have a fairly accurate idea of the contribution made by the various traditions of research to the overall domain of lexical semantics. In Table 3, the major nuances are incorporated into the picture presented by Table 2. Table 3. The contribution of the traditions to the development of the field SEMASIOLOGY 'QUALITY':

investigating structure (elements and relations)

prestructuralist semantics: mechanisms

of semantic change neogenerativist semantics: regular

polysemy

ONOMASIOLOGY

structuralist semantics:

taxonomies and lexical relations; lexical fields; syntagmatic relations and collocations Cognitive Semantics:

conceptual metaphors and metonymies; frames 'QUANTITY' :

Cognitive Semantics:

Cognitive Semantics:

investigating use and differential weights within structures

prototype theory

basic levels and entrenchment

4.

Competition and convergence

Essential to the topic of the present paper is the recognition that there is empirical progress in the development of the discipline. The actual historical development from prestructuralist semantics over structuralism and generativist semantics to Cognitive Semantics constitutes a gradual extension of the descriptive scope of lexical semantics. This form of empirical progress constitutes a nuance with regard to the historical picture that was painted before: while the theoretical development of lexical semantics is characterized by shifts and oppositions, the empirical development is characterized by at least some degree of complementarity and accumulation. In the context of the present volume, it seems legitimate to ask whether this empirical complementarity may lead to a convergence of the various theoretical approaches. I would venture the suggestion that such a process of convergence - if it will take place at all - could find a focal point in a pragmatic, usage-based perspective to lexical semantics. Three observations lie at the basis of this suggestion.

Theoretical and descriptive development

415

First, the currently most productive offshoot of structuralist semantics (the corpus-based analysis of collocational patterns) involves a shift from an analysis of structure as such (the initial approach of structuralism) to an analysis of patterns of actual language use. In structuralist parlance, this constitutes a shift from the level of langue to the level of parole - or, if one likes, a pragmatic shift towards a usage-based perspective. Second, a somewhat similar shift characterizes the interest in regular polysemy of Pustejovskian neogenerativist semantics. Regular patterns of polysemy distinguish between stored meanings - meanings that have to be included in the mental lexicon - and derived meanings - meanings that can be determined through the operation of rules for regular polysemy and contextual interpretation. This constitutes a shift from an interest in semantic representations as such to an interest in semantic representations as the basis of semantic interpretations - and the process of interpretation is obviously to be situated at the level of usage. And third, although the overview presented above makes clear that the major conceptual contributions to the 'quantitative' perspective have so far come from Cognitive Semantics, it is equally clear that this area of research is far from exhausted. Investigations into the actual onomasiological choices made by language users and the factors influencing such choices have only started to emerge, so that a continuing interest from the side of Cognitive Semantics for this field of investigation is far from unlikely. In various guises, then, a pragmatic turn characterizes many of the contemporary approaches to lexical semantics. The usage-based aspects of lexical semantics would therefore seem to be a primary starting-point for exploring the possibility of theoretical convergences - or, as the case may be, for establishing fundamental incompatibilities.

Chapter 17 Idealist and empiricist tendencies in cognitive semantics

Originally published in Theo Janssen and Gisela Redeker (eds.), 1999, Cognitive Linguistics: Foundations, Scope, and Methodologv 163-194. BerlinfNew York: Mouton de Gruyter. Two metatheoretical threads run through the papers collected in this volume. First, as an antidote against the danger of reinventing the wheel, publications like the previous two chapters stress the importance of situating Cognitive Linguistics against a wider background: the history of linguistics, and the interdisciplinary field of philosophy, psychology, and cognitive science at large. Second, as an antidote against idle speculation, it is imperative for Cognitive Linguistics to do research on a firm empirical basis. In the current line of research of the Quantitative Lexicology and Variational Linguistics research group that I lead in Leuven, this methodological conviction takes the form of statistically advanced forms of corpus linguistics. In the final chapter of the present volume, the question is tackled from a theoretical rather than a practical point of view: how can one defend the value of an empirical approach against a more idealistically, introspectively oriented method? The discussion takes the uncharacteristic but somewhat ironic form of a platonic dialogue, with a prologue and an epilogue.

Prologue The methodological situation in present-day Cognitive Linguistics is characterized by the existence of two methodological extremes. On the one hand, there is the idealistic approach most conspicuously advocated by Anna Wierzbicka in her numerous publications (among which 1985, 1992, 1996). On the other hand, there exist various tendencies to objectivize the methods used in Cognitive Linguistics. Roughly, there are three main tendencies at this end of the methodological opposition: psycholinguistic research (as in Sandra and Rice 1995 or Gibbs 1994), neurophysiologial modeling (as in Regier 1995), and quantitative corpus-based analysis (as in Geeraerts, Grondelaers, and Bakema 1994). Most actual work in Cognitive Linguistics is probably situated somewhere in-between both extremes, or rather, the

Idealist and empiricist tendencies

417

practical methodology used is more often introspective rather than datadriven, but without the outspoken idealistic commitments that Anna Wierzbicka explicitly describes. Whence the title of the following dialogue (taken from Machado de Assis' QUincas Borba): even those cognitive linguists who have never reflected on the epistemological underpinnings of their practical work, are implicitly caught up in the debate between the idealistic and the more empiristically minded approaches. My own position in the debate is clearly not on the idealist side, but the following dialogue (which suitably takes a platonic form) tries to establish why that choice is less self-evident than it may appear once one has made it. THE CAT, THOUGH IT HAS NOT READ KANT, MAY WELL BE A METAPHYSICAL ANIMAL

A Philosophical Dialogue Dramatis Personae: DUODECIMUS EMPIRIC US, a cognitive linguist specializing in corpus-based techniques of analysis and experimental methods; author of the book Quantitative Methods in Cognitive Semantics.

a cognitive linguist of a platonic persuasion; author of the book Universal Fundaments o(Meaning.

ANTIPODA PROLIPHICA,

teacher and friend to Duodecimus and Antipoda, a perpetual source of knowledge, inspiration, and academic tolerance to both of them.

HISTRIO POLYMORPHUS,

The scene is set in the garden of Histrio' s villa, where the three of them meet regularly to discuss on a theme set by Histrio. D. Well, master, the lavishness of last week's feast has more than soothed the hunger of our stomachs, but our minds are as thirsty for knowledge as ever. What then will be the topic for today's meeting? To what surprising views will your argumentation lead us, as unexpected as the vistas opening from your garden's paths?

H. Gardening, my young friend, is a matter of patience. Before the bushes and lawns, flowerbeds and trees acquire their ultimate architectural form, years of preparatory work are necessary. The wilderness has to be conquered, the grounds have to be laid out, and the plants have to be guided and protected year after year before the global pattern finally emerges.

418

Theory and method

Allow me, therefore, to introduce you indirectly, with some sort of a detour, to the theme that I have in mind. A. Most certainly so, master, but do allow us to drink to your health first. Such a long journey as you promise us should not be undertaken without invigorating the body first. D. Right you are, Antipoda: may Histrio's thoughts continue to flower like a cherry tree in spring! H. Your are too kind to a man in the autumn of his life. But let us take the first step - a step that brings us to physics rather than metaphysics. Do you know the story of the ether? A. We know ether is a colorless liquid used as a solvent, or as an anesthetic. Are you referring to the way it was discovered? H. The word ether actually has yet another meaning, not well known nowadays, but extremely important in the physical theories of the nineteenth century. By the end of the nineteenth century, the majority of physicists was convinced of the existence of an ether: an odorless, colorless and tasteless substance that filled the universe and that functioned as the material bearer of waves of light. If light is a wave, it indeed requires a medium through which its waves are transported, just like the waves of the sea move through the material medium that is water.

D. So the ether was supposed to be a medium that filled the gaps between particles of matter? This sounds like a physical kind of horror vacui! H. Exactly. But the analogy with a better known medium like water led to further refinements of the concept. A body like the earth, then, that moved through the ether would inevitably cause an ether wind: the ether particles in front of the body would be pressed together, while behind it an area with lowered pressure, so to speak, would be formed.

D. I can see that, but if this ether is odorless, colorless and tasteless, how could its presence be detected? As you know, such methodological questions particularly interest me. H. Scientists tried to experimentally measure the ether wind: a light wave moving in the same direction as the earth would travel more slowly than a light wave that went along with the ether wind, because the former would have to overcome the resistance of the tightly packed ether particles in

Idealist and empiricist tendencies

419

front of the moving earth. In 1881, this hypothesis was put to an experimental test by Michelson and Morley. Unfortunately, it appeared that the beam of light that was sent out in the same direction as the earth did not go faster than the one that shone in the opposite direction. D. Perhaps the experimental equipment used had not been sophisticated enough? H. I can see you have some experience with experimental research, dear Duodecimus: your suggestion is precisely what Michelson and Morley thought. So the experiment was repeated in 1887 with more accurate instruments and with a truly ingenious design (which I will not describe, not to tax your patience). This new design did not bring the expected results either, however. A. I can very well imagine the reason why not, master: don't we all know that the speed of light is invariable? H. Certainly. But at the time of the experiments, science was not sufficiently advanced yet to have realized this: we all know now that the speed of light is constant, but that assumption as such is neither self-evident nor empirically obvious. Only in 1905, in fact, did Einstein proclaim the inexistence of the ether wind together with the invariability of the speed of light. D. I am rather surprised that you say that Einstein 'proclaimed' the inexistence of the ether wind. Should we not rather say that he discovered its inexistence? H. From our point of view, we can certainly say so - because we no longer believe in the ether either. But for Michelson and Morley things were not that obvious: they actually continued to believe in the existence of the ether a long time after Einstein described his alternative view, and they tried out ever more refined experiments - without much success, as you can Imagme. A. So were they not just being stubborn? H. Not necessarily. The interesting thing about the story, in fact, is its methodological structure. When an experiment fails, two courses can be followed: either you assume that there is something wrong with the observational apparatus with which the experiment was carried out, or you assume that the theoretical assumptions of the experiment should be sub-

420

Theory and method

jected to severe criticism. In the former case there is uncertainty concerning the actual result of the experiment. The experiment is considered undecided, and it is assumed that the uncertainty can be ironed out by using more refined experimental techniques. In the latter case the experiment is recognized as an actual rejection of an hypothesis. Michelson and Morley, by their repeated endeavors to make the experimental environment better fit to measure the existence of an ether and the presence of an ether wind, followed the first tack. Einstein, on the other, took the opposite course, by taking Michelson and Morley's experimental failure as a straightforward rejection of the hypothesis that there existed an ether. D. I must confess that the first surprise in your story comes earlier than expected. Are you trying to tell us that falsification of scientific hypotheses is not a straightforward matter? That there are cases where it is difficult to disprove a hypothesis directly? H. Well, try to look at it in this way. The existence of alternative ways of dealing with negative experimental results implies that, at least to some extent, theoretical presuppositions define what is a theoretically relevant fact: if one sticks to the existence of an ether wind, then the experiment simply does not yield an acceptable observation; due to the presumed problems with the experimental design, no theoretically interpretable fact at all is observed. If, on the other hand, one starts from the presupposition that there is no ether wind at all, the experiment produces a clear result. D. But surely, one could not go on indefinitely ignoring the failure of the experiments? H. I would like to leave open the question how long the first position can be maintained after repeated failures: the crucial point to see is that observational results do not exist in isolation, but that 'the facts' depend at least partly on the intensity with which one cherishes certain theoretical concepts - or conversely, on the degree to which one may be ready to give up such concepts. Of the utmost importance here is the notion of a 'theory of interpretation' as the embodiment of the view that there are no bare facts: facts only become facts through the intervention of background assumptions and interpretative principles.

A. When I let my initial surprise subside, I think I see a relationship with what we, cognitive linguists, know as 'internal realism'. Following ideas

Idealist and empiricist tendencies

421

from Hilary Putnam, George Lakoff (1987) has characterized the epistemological position of Cognitive Linguistics as a rejection of both hardcore objectivism and radical idealism: there is no doubt about the existence of the outside world, but at the same time, the world as we know it is shaped by the concepts we impose upon it. Well, I need not repeat this to you, of course. As I see it now, the ideas that you just introduced define a kind of internal realism on the level of scientific methodology. Facts do not exist as such, but only within a particular framework? H. I knew you would be clever enough to see that yourself, Antipoda. A. But that does not mean I believe in it, yet.

H. Knowing your independence of mind, I had not expected otherwise. Before we start discussing alternatives, however, I have to say something about the status of such a position of internal realism within the philosophy of science. There is, in fact, a particular conception of the philosophy of science that embodies such an internal-realistic view, viz. the paradigmatic theory of science as represented by Thomas Kuhn, Imre Lakatos, and Paul Feyerabend. Internal realism is present in Kuhn's conception of scientific paradigms in the following way. Apparent counterevidence with regard to a paradigmatic theory is not simply accepted as such, but may merely be considered an indication of a problem that will be solved when the theory is developed further. Rather than occasioning an immediate rejection of the existing theory, counterevidence (if it is not simply disregarded as based on irrelevant experiments) leads to the attempt to explain away or solve the problems in the framework of the existing paradigm. Paradigmatic theories are given credit, they are given time to solve puzzles that arise.

D. That suggests to me that the structure and the epistemological role of scientific paradigms (seen as a set of theories clustered round a number of central assumptions) are similar to those of natural language categories (seen as clusters of conceptual applications concentrated round a prototypical center). Since both have a formative influence with regard to, respectively, new theories and new concepts, prototypical categories may be seen as a reflection, on the microstructural level of individual natural language concepts, of the same 'paradigmatic' principles that govern the development of science on the macrostructural level of scientific theories.

422

Theory and method

H. I would say so, yes. But notice that the similarity between both positions should not be thought of in a simplistic way. Cognitive Linguistics characterizes the relationship between natural language (and natural thinking, if you like) and the world in terms of internal realism. But Cognitive Linguistics itself is a scientific theory that may be characterized by internal realism. D. So there are two epistemological levels involved: the level of thinking about the world (the level of folk knowledge, so to speak), and the level of scientifically thinking about thinking. And you suggest that there is a homology between both levels: the relationship between thinking and its object is the same on both levels, even if the object on the second level is thinking itself. Where is the problem with that? H. As you will remember from the times when we discussed the concept of self-reference, interactions between different epistemological levels may yield remarkable results. In our case, questions about the objectivity of semantics arise. If not only our thinking as such, but also our thinking about thinking is influenced by pre-existing conceptualizations that ftmction as interpretational schemata, what kind of a scientific enterprise is semantics? If it does not only deal with the way in which the speakers of a language construct interpretations for the expressions they encounter, but if semantics itself imposes interpretations on its subject matter, what claims to objectivity can it make? It's like the Liar's paradox, is it not? If you are Cretan and claim that all Cretans are liars, you undermine your own statement. In the same way, if you claim that all human knowledge is a non-objectivist construction, you inevitably undermine your own position, to the extent that you suggest that your own general statement is something of a construction. A. I get the impression we are reaching the central topic for today. Makes me thirsty, actually. Shall I fill your glasses? H. Please do.

A. Here you are. To be honest, I am less convinced there is a problem than you are. After all, you should not forget that there is a crucial difference between the two levels. On the first level, there is an ontological distinction between the object and the subject of the epistemological relationship: the known object is the outside world, and the knowing subject is a human person (a mind, if you wish). But on the second level, subject and

Idealist and empiricist tendencies

423

object coincide. The object is the way we humans conceptualize the world, that is to say, the way in which the world is known to us in our knowing minds. At the same time, it is that very same mind that investigates its own knowledge of the outside world. On the first level, we deal with consciousness, but on the second level, the discussion is about selfconsciousness. And because subject and object coincide in the case of self-consciousness, there is a transparent relationship between them. D. So you believe in the unmitigated powers of introspection, the full epistemological presence of the mind to itself, the ultimate transparency of self-knowledge? !

A. There is no reason to be ironical. You know my methodological startingpoint perfectly well. I am convinced that the best guide to and the ultimate ground in semantic analysis is an examination of one's own consciousness - what one knows of the concepts to be analyzed. 'Chasing the phantom of 'objectivity' through supposedly scientific methods one loses the only firm ground there is in semantics: the terra firma of one's own deep intuitions' (Wierzbicka 1985: 43). D. That makes me an objectivist phantom-chaser, then. You know perfectly well that I practice semantics in the opposite way: not by denying the importance of intuition, to be sure, but by supplementing and supporting any reliance on introspection with corpus analysis or experimentation. When I make a claim about a word's meaning, for instance, I want to check that claim against the way in which people actually use the word. Or rather the other way round: if! try to define a word's meaning, I need evidence about the way the word is being used first. For instance, a definition should fit the referential range of application of a word: if you do not know what objects a word can refer to, how can you define the word? A. But if you know the word, you know how to use it, do you not? So you will not have to look any further for evidence about how it is to be used: just consult your own knowledge. D. No, that will not do, because there is sufficient evidence that an introspective analysis of how people actually use words does not necessarily conform to what people actually do. I could give you several examples, but let us maybe just restrict the discussion to a single case that both of us have already done research on. At one point (Wierzbicka 1985: 382), you defined dress in the following way:

424

Theory and method

A KIND OF THING MADE BY PEOPLE FOR WOMEN AND GIRLS TO WEAR. IMAGINING THINGS OF THIS KIND PEOPLE COULD SAY THESE THINGS ABOUT THEM:

they are made to be worn on the body, below the head, to cover most of the body so that all the parts of a woman's body which people think should not normally be seen are covered with that one thing and to protect most of the body from undesirable contact with the environment and to cause the woman wearing it to look good they are made in such a way that when they are on the body the lower half surrounds the lower half of the woman's body from all sides so that the legs are not separated from one another and so that the genital area of the woman's body seems to be hidden and so that women wearing things of this kind look different from men things of this kind are thought of as something suitable for women to wear in most kinds of places and in most kinds of circumstances. Now, as you know, I did a corpus-based study with my collaborators exploring the actual referential use of clothing terms in Dutch. When we have a look at the actual garments that occur in our database as instances ofjurk (the Dutch equivalent of dress), we find cases in our material that do not conform to the description. If, for instance, 'covering most of the body' is interpreted as 'covering more than 50% of the body', then a number of very short summer dresses with open backs and low necklines do not display the feature in question. And if 'the parts of a woman's body which people think should not normally be seen' include the upper part of the thighs, then dresses with long side slits contradict the image. Furthermore, some dresses have such wide armholes and such a plunging decolletage that they could not normally be worn without exposure of the breasts (unless they are worn with an additional t-shirt or blouse underneath). The comparison shows, in other words, that the description you propose may well be adequate for the majority of cases in the range of dress, but does far from cover all possible instances.

A. This is irrelevant. What my definition describes is not the objective features of the things that might be called dresses, but the conceptual knowledge that people have of dresses. Not an extension, if you like, but rather an intension. In fact, by introducing the phrase 'imagining things of this kind people could say these things about them-in the definition, the perspective is shifted from the objective features of the things that are being

Idealist and empiricist tendencies

425

called dresses to the subjective knowledge that people know they have about dresses when they examine their own consciousness. D. So you define dress by referring to what people think dresses are?

A. Yes. Is that not a natural thing to do for a so-called 'cognitive' analysis? What I try to define is the concept 'dress' that people have, not actual dresses. Let us maybe look at another example. You once made similar remarks about my analysis of the category fruit. In my definition of the category, I included the feature that people, 'wanting to imagine such things [fruits], would imagine them as growing on trees'. You then commented that this feature does not apply to all fruit: raspberries, for example, are fruit, and yet they do not grow on trees. 'In my view, however, this indisputable fact does not disprove the existence of a conceptual link between 'fruit' and 'trees' (just as the fact that ostriches don't fly does not disprove the existence of a conceptual link between 'birds' and 'flying') .... From the fact that people think of ostriches as birds, and of birds as flying, it doesn't follow that they think of ostriches as flying' (Wierzbicka 1996: 164). And from the fact that people think of fruit as growing on trees, and of raspberries as fruit, it does not follow that they think of raspberries as growing on trees. D. That amounts to saying that a definition should not necessarily cover the entire range of the category to be defined. But such a position contradicts what you once said, namely that 'a valid definition must be empirically adequate, that is, it must be phrased in such a way that it covers the entire range of use of a given word, expression, or construction. It is not good enough if it only fits the more common or more typical cases' (Wierzbicka 1989: 738).

A. But again, do not forget that the features in my definition refer to mental states, so to speak, not to objective features of dresses or fruit or birds. What I claim is that people think of trees any time they think of fruit. D. I am afraid I cannot follow you there. Even if we think of the features in your definition as truly conceptual features, they still do not have the required generality. For instance, if a definition such as that of dress is to be applicable to all the things that may be called by that name, people should be able to assert all the characteristics mentioned in the definition any time they see a dress. But surely, when imagining a less typical kind of dress than the kind whose features are included in the definition, peo-

426

Theory and method

pie will not imagine it as a typical case. Or would they? What people could say about dresses changes when peripheral members of the category are at stake: default dresses, for instance, may well cover most of the body, but I would say that that feature is suppressed when a fancy type of summer dress is involved. Or, in the case of fruit, what would be the semantics of a sentence like A raspberry is a fruit? If you paraphrase it as 'a raspberry is one of those things we think about (among other things) as growing on trees', you do imply that raspberries are thought of as growing on trees, and that is obviously not the case. And similarly for ostriches and birds: if An ostrich is a bird reads 'an ostrich is one of those living beings thinking about which we would imagine them as being able to fly', then the sentence would be false. Which amounts to saying that the type of definition you use is incapable of accounting for categorization judgments. The definition offruit that you suggest is unable to explain why a raspberry belongs to the category in question.

A. Do not jump to conclusions, dear Duodecimus. I would say there are two aspects to your remark. First, how is it we can say that raspberries are fruit, if our concept offruit does not allow us to decide? Second, is a sentence like Raspberries are fruit contradictory? To tackle the first point first, the referential range of a category like fruit might be settled in a conventional way, more or less in the same way in which Hilary Putnam (1973) describes the semantics of natural kind terms. Whether a particular thing is to be called tiger is one thing, and the knowledge that people have of tigers is quite something else. On the one hand, there is a social convention about the names of things, on the other, there is a social convention about the knowledge (called 'stereotypical' by Putnam) that the speakers of a language are supposed to have of the category. Crucially, it is not what you know about jade that allows you to determine whether something is jade or not. 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 H2 0, 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 beeches, how to recognize gold from pyrites, and so on. On the

Idealist and empiricist tendencies

427

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 extension of a category. D. Well, I am fairly familiar with Putnam's theory, and I do not think ,that his particular model of the division of linguistic labor is empirically adequate. Also, his model is restricted to the semantics of natural kind terms, so we should be careful with generalizing it. But I admit that that is less relevant for what you are trying to say: if you assume that naming is a conventional act that need not be backed up by conceptual knowledge, then yes, you need not adapt your definition offruit to allow for raspberries. But what about the other objection? Is a sentence like Raspberries are fruit contradictory? A. Let me ask you a question in return. Do you think that the semantic knowledge that we possess at one particular moment is always sufficient to categorize the entities, situations, events, processes or whatever that we encounter in reality?

D. I do not see yet where this is leading to, but no, I do not think so. One of the crucial aspects of human cognition is the way in which it can adapt itself to changing circumstances in the outside world. The knowledge that we have has to be applied in flexible ways to new experiences. Incidentally, that is precisely the reason why prototype-theoretical models of semantic structures exert such an appeal on cognitive linguists. At least to some extent, the peripheral, slightly deviant cases of a category are precisely those that result from the application of an existing category to cases that are similar enough to be included, but that may differ in subtle ways from the existing concept. A. So you accept that there is a distinction between stored knowledge and actualized knowledge? Between the semantic knowledge that is mentally stored and the meanings that appear in actual utterances?

D. Absolutely. This is, to be sure, intimately connected with the 'cognitive' nature of Cognitive Linguistics itself. Cognitive Linguistics is cognitive in the same way that cognitive psychology is: by assuming that our inter-

428

Theory and method

action with the world is mediated through infonnational structures in the mind.

A. Now if you accept that, Raspberries are fruit is not contradictory once you realize that my definition of fruit describes the knowledge that we have of the concept 'fruit' on the deep-seated level of stored meanings. It tells you what we know about fruit out of context, so to speak. Conversely, Raspberries are fruit is an actualization of that knowledge on the surface level, where the conceptual interaction between what we know about raspberries and what we know about fruit may lead to a contextual attunement of the underlying concept. D. So you have a semantic deep structure, and a semantic surface structure, and semantic transfonnations that change deep structure into surface structure?

A. Your tenninology has unfortunate historical connotations, but okay, let us accept this way of putting it as a convenient figure of speech. An alternative tenninology could be to distinguish a level of 'deep semantics' from a level of 'contextual' semantics. D. Ab yes, but then I do not quite understand why you concentrate your powers of analysis and description (which are considerable, let me be quite clear about that) on the deep-structural, deep-semantic level. The surface level, after all, is the level where most of the semantic action is to be found; I mean, most of the phenomena that a semanticist could be interested in are surface level phenomena. It follows from your own analysis of Raspberries are fruit, for instance, that truth conditions are to be investigated on the surface level. And the same holds for polysemy. Standard tests for polysemy, like the zeugma test, or Quine's p and not p-test, can only be properly applied to the utterance level.

A. Really? But if you define polysemy in tenns of definability (two meanings are different if you cannot define them as being identical), and if definition refers to the deep-structural, stored readings, then clearly polysemy questions can be answered on the basis of an introspective methodology. D. Still, it remains a matter to be settled empirically whether the lexical knowledge that people have in their minds confonns to the definition that you retrieve introspectively. It is not a priori given that the idea of a category that people may introspectively retrieve from memory is an adequate

Idealist and empiricist tendencies

429

reflection of the extent of that person's actual knowledge of the category. On the contrary, if it is part of his knowledge to produce or accept an application of dress or fruit to cases that do not conform to the stored concept as introspectively defined, then he 'knows' more about the category than would be included in his introspectively retrieved idea of the category. That knowledge, to be sure, is not necessarily conscious knowledge; it is less 'knowledge that (lexical item x may refer to entities with such and such characteristics)' but rather 'knowledge how (lexical item x may be successfully used)'. In order, then, to get a better grasp on the lexical 'knowledge how', usage-based investigations of the type illustrated by my own studies are vital, precisely if it is suspected that conscious knowledge may only partially cover the full extent of a person's 'knowledge how'. So I really believe that your position will only remain defensible if you provide us with an explicit model that describes how stored meanings are transformed into contextual readings at the utterance level.

A. I am sorry to disappoint you, but it may take some time before I come up with such a model. After all, why should I? The crucial phenomenon is the underlying concept, and the only proper way to investigate it is by introspection. So why should I spend my energy on what is (from my point of view) a secondary phenomenon? D. One reason could be that your approach implies that your position becomes practically immune to criticism if you do not provide a full model of what happens when stored conceptual knowledge is contextualized. If an introspective examination of your own consciousness is the ultimate basis of semantics, how can your analyses be falsified at all? Is not your introspective methodology a gigantic immunization strategy? To put it slightly differently, how can you ensure that what you come up with in the analysis has anything to do at all with the knowledge of the language that we share, as speakers of the same language?

A. 'My own experience in lexicographical analysis carried out over many years in seminars and group discussions has led me to the conclusion that the problem of objectively identifying the 'shared knowledge' and the shared notions may in fact be more apparent than real [...] Certainly if you ask a group of people what a cup or a skirt is, they will at first come up with overlapping rather than identical responses. If, however, each of them is induced to investigate his or her own concepts in depth, investing

430

Theory and method

in the task a large amount of time and effort, then the differences in their responses will gradually diminish. If they are given a chance to discuss their differences the agreement will grow even further and the residual differences will often disappear. In the process, it will become clear that the underlying concept is the same for all the participants and that the initial differences in their responses were due to the fact that they had not yet explored sufficiently their own consciousness' (Wierzbicka 1985: 4243). D. I doubt whether the results of such a discussion will always be as uniform as you pretend, but I can see that, within your framework, intersubjectivity would indeed be a way of avoiding excessive .... Master?! Are you feeling alright? Quickly, pour him something to drink! H. Thank you, thank you children, but I am perfectly okay. I was actually just closing my eyes to get a better view of your argumentation. Quite a nice pattern is emerging from your lively conversation, I must say.

A. Surely not as well-balanced as the pattern of your pleasant garden, I dare say. H. Oh yes, there is symmetry in what you have said. If you have not been carried off entirely by your ardent interchanges, you will remember that your discussion was triggered by my little story about the ether wind. Recall that the essential point about the story was this: ontological assumptions, i.e. assumptions about what is, need not be subject to straightforward falsification. Now, in what both of you have said so far, I discern two different 'paradigms' in the Kuhnian sense; two different ways of considering cognitive phenomena that start from more or less opposing ontological assumptions. Your position, dear Antipoda, is idealistic in a double sense: an ontological sense and a methodological one. Ontologically, you believe in the existence of well-defined, deep-seated meanings that are the true essence of our conceptual knowledge - a platonic realm of ideas, really, but with a psychological rather than a metaphysical status. Methodologically, your position is an idealist one to the extent that you believe in the transparency of the mind to itself - in the epistemological primacy of introspection, in other words.

A. That is a fair picture of my position. What about Duodecimus? H. I would say that his position is the opposite of yours. He believes in the epistemological primacy of utterance meaning. Ontologically speaking, I

Idealist and empiricist tendencies

431

think he would say that meaning is to be situated primarily at the level of performance. Meanings exist primarily as realized meanings, that is to say, as meanings inherent in utterances, in actual events of using language. Methodologically, this view correlates with the idea that utterances are the place where meanings are closest to us, where they are so to speak most palpable. D. Precisely. Corpus-based studies and experimental psycholinguistic research are important to see what actually happens when people use the language. It has sometimes been said that Cognitive Linguistics is a usage-based model, but I feel cognitive linguists sometimes neglect to draw the methodological consequences of that recognition - viz. that actual language use is the methodological basis of Cognitive Linguistics. A. But since you do believe in the importance of stored meanings (as you explained earlier), what is the status of such stored meanings? D. That is an interesting point, and I must confess that I have not yet arrived at an entirely satisfactory picture of what is going on. Methodologically speaking, the matter is quite clear: stored semantic knowledge is inferred; it can hardly be observed directly, but it functions as a hidden entity that helps one to explain the actual behavior that one finds reflected in the usage data. There is nothing extraordinary about that, surely: making reference to hidden entities as hypothetical constructs is common practice to quite a number of sciences. Ontologically speaking, however, things are less straightforward. Traditionally, cognitive scientists have conceived of such stored meanings in a symbolic fashion, as information encoded in some form of mental language or 'Mentalese'. But that is a far-reaching assumption that should be handled with great care. Connectionism in particular has made it clear that we had better be careful with thinking about stored knowledge as something encoded in a symbolic fashion. So, to be on the safe side, I would not want to go any further at this point than simply say that the stored knowledge is the mental apparatus that enables us to understand and produce meanings (where 'meanings' are to be found on the level of usage). As far as I am concerned, it is not even certain that it is useful to speak of 'meanings' as a specific kind of entity on this level at all. At least as far as I can see, we have only a very rudimentary view of how to think about mental meaning, which gives me one more motivation to start from the other end - that is, the level of utterances, performance, usage.

432

Theory and method

A. I can see clearly now how our views mirror each other. But I am not yet convinced that the ether wind story is applicable to our situation. Master, how exactly did you see the link? H. Let us look back at the conversation the two of you had a moment ago, and reconstruct it along the lines of my earlier story. Antipoda started from the initial assumption that, for instance, the word dress had a specific unitary meaning. You, Duodecimus, formulated an experiment that could prove Antipoda wrong. The 'experiment' (an empirical investigation into real language use, actually) did not establish the correctness of Antipoda's view, just like Michelson and Morley's experiment did not establish the existence of the ether. But instead of accepting your experiment as a straightforward rejection, Antipoda used different possibilities to salvage her original position: either to claim that your methodology was not the most appropriate one (meaning can only be grasped by an introspective methodology), or to make slight theoretical adjustments (the meaning to be defined is a 'deep-structural', 'deep-semantic' phenomenon). Michelson and Morley, as I mentioned, took the road of methodological adjustment, but in the case of the ether just as in the case of Antipoda's meanings, some people took the road of theoretical adjustments. (One such approach involved so-called 'Lorentz Fitzgerald contraction' the idea that the effect of the ether wind could not be measured because the moving objects underwent a contraction.) And the ironical thing, Duodecimus, is that you cannot even blame Antipoda for doing so, because her strategy is a legitimate one, if you are willing to extrapolate your position of internal realism from the theoretical to the metatheoretical level.

D. Good. You have established that deciding between the two models is not self-evident. But a decision should be possible, should it not? After all, the debate about the ether was decided eventually. I mean, ontological assumptions cannot be selected at random: if I start from the ontological assumption that the earth has two moons, the world itself will prove me wrong. You have made the point that the extrapolation of the notion of internal realism to the methodological level creates a broad margin for theoretical and methodological differences; quite different assumptions may make it more difficult to assess competing theories in a straightforward way. But there clearly are limits to that flexibility, are there not? If you have a theory that produces a model that systematically describes how surface readings are related to stored semantic knowledge, then

Idealist and empiricist tendencies

433

surely the pure scope of such a theory would be a reason for preferring it over a model that restricts itself to stored meanings only! A. The point is rather too hypothetical for my taste: you will have to admit that at this point, neither of the two approaches has developed such an encompassing model yet. So it is rather irrelevant to compare them on this point. There is another thing, however, that the two of you seem to forget. The relationship between both approaches, in fact, is not as symmetrical as Histrio's picture suggests. I am convinced, dear Duodecimus, that you are somewhat of a closet introspectivist. D. Again a surprise! What do you mean? A. Let us consider the clothing terms project that you referred to earlier. The methodological starting-point of the project was to investigate to which actual referents specific terms referred. Right? D. Most certainly so. By choosing clothing terms as the field of investigation, we could get independent access to the referents of the terms in question, for instance by examining the photographs that the clothing terms refer to in fashion magazines and the like. Then we described the various features of the referents of any specific clothing term, and on the basis of this featural description, described the structure of the semasiological range of application of the different garment names. A. Now notice that your approach is not completely free of intuitive aspects, in the sense that your own understanding of the instances of language use under investigation is entirely ignored or suppressed. More precisely, the referential approach does not entail that the investigation proceeds in a purely objective fashion, without any recourse to interpretative activities on the part of the investigator. The point may be illustrated by considering the initial selection of the descriptive features included in the componential system. The choice of those features cannot be dictated automatically by the referents of the words themselves. In principle, an infinite number of characteristics could possibly be included in the descriptive framework. In the case of trousers, for instance, it would be possible to refer to the presence of lining in the legs, to whether the hip pockets have a flap or not, or to the number of nooses in the waist intended to hold a belt. The fact that, in actual practice, you decided not to include these features in the componential system is undoubtedly determined by assumptions about their relevance for the description. Ultimately, I would

434

Theory and method

say that it is your own pre-existent knowledge of the field of Dutch clothing tenns, and your own intuitions about what would be pertinent features for describing that field, that have played their role in the choice of a particular componential system. D. Oh, but I have no wish to deny that any semantic analysis has to rely at least to some extent on an intuitive understanding of linguistic expressions. But we opt for a referentially enriched corpus (as we prefer to call it) precisely to guide our intuitions, and to restrict the possibilities of an uncritical or unmitigated reliance on intuitive judgments. A. But if you accept that meanings are only realized mentally, and that you need intuition anyway to grasp them at all, why not just accept introspection as the basic method of research? After all ...

H. Excuse me for interrupting you, but I feel we are reaching the heart of the matter here, and I would like you both to see quite clearly what is at stake here. D. Master, we are eminently interruptible when you summon us. Pray explain the stakes to us. H. It will be a pleasure. But shall we perhaps leave our seats and stroll around the garden for a while? I feel we could use a little exercise. A. We certainly can. Could we perhaps walk to the sundial down yonder? I have never had a chance to take a good look at it.

H. Certainly. But I suggest not to follow the path that leads directly to it; we could take the path through the orchard instead. In fact, just as I introduced the topic of our conversation by making a detour, I would now like to follow another side-path, to see whether another excursion can help us to bring the discussion to a close.

D. Fine. Let us go peripatetic then. H. What Antipoda's last remarks make clear, is the fact that semantics is basically a hermeneutic enterprise. Lexical description, for instance, does not simply consist of recording referents, but of trying to detennine what features of the referents motivate or license the use of a particular item, in short, of understanding expressions. I would now like to show to you that the opposition between both of you is not uncommon in the henneneutic sciences.

Idealist and empiricist tendencies

435

D. 'The henneneutic sciences' - what do you understand by that? H. From a very general point of view, henneneutics is the 'theory of interpretation', a methodological reflection on the principles underlying interpretative activities concerned with literary products, works of art, juridical texts, and so on. It is, in fact, from the sources of biblical exegesis and philological scholarship that henneneutics as a theoretical enterprise came into being in the course of the nineteenth century. More specifically, henneneutics is a philosophical tradition, inaugurated by Wilhelm Dilthey at the end of the previous century, that takes interpretation to be the basic methodological concept of the human sciences (Geisteswissenschaften). Dilthey's explicit aim was to provide the human sciences with a philosophical understanding of their methodological basis that could be set off against the positivist self-understanding of the natural sciences. D. So the henneneutic sciences coincide with the human sciences, because they use an interpretative methodology? H. More accurately, Dilthey's views consist of a characterization of the subject-matter of the human sciences, and of a characterization of their method. The products of the human mind that constitute the subjectmatter of the human sciences are the expression of an inner experience, an understanding of life in tenns of the intellect, the will, and the emotions that all human beings possess. The method of the human sciences, in return, consists of an interpretative attempt to recover the original experience behind the expressions. In this sense, the logic of the human sciences is fundamentally different from that of the natural sciences: whereas the latter are based on 'Erklarung' (positivistic explanation in tenns of the immutable laws of nature), the fonner are based on 'Verstehen' (understanding of lived experience through its expressions).

A. This emphasis on experience must imply a natural affinity between Cognitive Linguistics and Diltheyan henneneutics, then. H. All the more so because Dilthey thinks of that experience in tenns that will sound familiar to cognitive linguists. I think you will also be happy to note, for instance, that Dilthey mentions the study of language as the epitome of the henneneutic enterprise. And because experience is a mental phenomenon, there is a link between henneneutics and psychology. This is not to say that the experiences that are being expressed are always and only highly personal phenomena. Rather, human experience is also a

436

Theory and method

cultural and historical phenomenon, and products of human life such as systems of laws, rituals, and everyday conventions clearly embody the historically transmitted experience of cultures rather than just individuals - but even then, the cultural message would cease to exist if it were not incorporated into the personal experience of individuals. At the same time, the henneneutic methodology is just as encyclopedic as human experience itself. While the experience that is being expressed is an encompassing phenomenon in which feelings, thoughts, memories, expectations etc. come together, the interpretative methodology of the human sciences will ideally be an interdisciplinary fusion of historical, psychological, anthropological, and linguistic research. D. I understand there is an affinity with Cognitive Linguistics, but we have had many conversations before in which you revealed that Cognitive Linguistics has many predecessors, both in philosophy and linguistics. What makes Dilthey special? H. You are absolutely right. If we restrict the perspective to philosophy, Dilthey is but one representative of a larger number of theoreticians who worked round the turn of the century and who discussed epistemological problems that are quite relevant to us today. Think of all those, like Brentano and Husserl, who explored the concept of intentionality. But Dilthey is special in the context of our present discussion, because he has explicitly drawn the attention to the methodological importance of interpretation for the human sciences. Moreover, and this is the reason that I mentioned his name at all, both his own work and the further development of philosophical henneneutics foreshadow the debate between the two of you.

A. Neither of us ever had the intention of being absolutely original, I am sure. H. Please do not feel diminished - I would be the last to overestimate historical originality. But I think I had better explain to you where the tension in Dilthey's work may be situated. If the method of the human sciences is based on interpretation (that is, on recovering the experience behind the expressions), can this process of recovery be concluded in an objective fashion? Does it lead to certainty? Can we be sure that our interpretations are correct? Can we truthfully detennine the original experience and the authentic intention behind the expressions? At this point, a particular tension within Dilthey's work becomes apparent. On the one

Idealist and empiricist tendencies

437

hand, Dilthey's attempt to place the human sciences on as secure and as respectable a footing as the natural sciences induces him to stress the objectivity of the human sciences. In his text 'Die Entstehung der Hermeneutik' (1961, originally 1913), for instance, he reacts against the danger of 'romantic arbitrariness' in the conception of the hermeneutical enterprise. If the hermeneutic act is basically an empathic subjective reenactment of an original creative act, objective standards for assessing the success of an interpretation may be hard to come by. Conversely, if hermeneutics is modeled on the example provided by biblical (and generally, philological) textual exegesis, interpretative rules might indeed be found. At this point, Dilthey is quite outspoken that the latter course is the one to take. He emphasizes the importance of subjecting the hermeneutical, interpretative enterprise to critical scrutiny, and stresses that the task of hermeneutics should be to develop a general theory of interpretation, a universally applicable interpretative methodology that specifies the criteria for correct interpretations. A. You spoke of a tension in his work, so there must be other tendencies at work as well. H. In fact, Dilthey's emphasis on the historicity of human experience also introduces a note of relativism into his views. While nature is governed by the necessity of natural laws, human life is governed by purposive action and conscious effort; the realm par excellence of human life is the unfolding of history. As such, can there be a general interpretative method that universally spans the entire range of the history of mankind, disregarding the historical differences between the objects of its interpretative activities? Can all historical products be interpreted on the same basis? More importantly still, if man in general is caught in history, can interpreters free themselves from their own historical background when confronting products from the past? A. I can see the tension quite clearly now, but did he find a solution for it in any way? H. In Dilthey's own work, the tension between the claim to universality implied by the search for objectivity, and the relativist tendency suggested by the emphasis on historicity is far from resolved. In the later development of hermeneutics, however, the non-universalist, non-objectivist tendency becomes most prominent, most clearly so in the developmental line

438

Theory and method

leading from Martin Heidegger to Hans-Georg Gadamer on the one hand, and to Jacques Derrida on the other. D. If I understand you correctly, you are making the point that the differences of opinion that Antipoda and I have do not involve a choice between an idealist and an objectivist conception of linguistic semantics, but rather between two ways of thinking about the interpretative, or if you wish 'hermeneutic' aspects of linguistics? H. Indeed. In your discussion of a few moments ago, Duodecimus tried to find hard and fast evidence that Antipoda's view is straightforwardly wrong, in the way in which it would be indubitably wrong to claim that the earth has two natural satellites. But at least as long as we do not have two full-fledged models to compare, such a straightforward falsification may be difficult to attain at this stage in the development of Cognitive Linguistics. Also, the methodological extrapolation of the concept of internal realism teaches us that theoretical comparison need not be selfevident. So, perhaps, it may be useful to think of your conflicting views from a different angle, viz. as different ways of dealing with the interpretative nature of semantics. Specifically, while Antipoda basically trusts intuitive interpretation, the position taken by Duodecimus is a prolongation of the critical approach that I mentioned in connection with Dilthey. D. I have no objections to thinking about my point of view in this way. When I insist on the importance of corpus-linguistic (or for that matter, psycho-experimental or neurophysiological) data as the backdrop for semantic research, I basically try to achieve a semantic analysis that is consonant with as many factors as possible: the way words are actually applied to things in the world, the way in which language is actually processed, the material embodiment of our linguistic skills, but also all contextual information we have about the historical and cultural background of the speakers of the language - everything, in a sense, in which language may be said to be grounded. Indeed, grounding may well be the essential concept in a Cognitive Linguistic theory of interpretation. A. That would seem to put me in the corner of 'romantic arbitrariness', or even worse, postmodem license a la Derrida. I really have to protest against that impression! There are at least two aspects of my approach that counteract any tendencies towards subjective arbitrariness. First, there is the importance of intersubjectivity which I drew your attention to a few minutes ago already.

Idealist and empiricist tendencies

439

D. And second?

A. Second, I firmly believe that semantic description should be couched in a vocabulary of universal concepts - a set of innate indefinables that is common to all languages of the world and that constitutes the core of their vocabulary. If definitions are required to be formulated in terms of such a natural set of primitive semantic elements, that will sufficiently restrict the process of interpretation, will it not? D. Again a very platonic idea, this universal set of innate ideas. But is it indeed introspectively clear what those innate ideas are? Is it part of our intuitive knowledge that we can identify the universal set of primitive semantic elements? A. I do not think so, actually. The set of universal concepts rather emerges from the definitional analyses themselves. Carefully defining yields a set of undefinables. D. How is that? I would say such an idea contradicts the idea that this set of indefinables has restrictive power. On the one hand, the set of primitives is supposed to restrict the set of possible definitions. On the other hand, the definitional analysis itself is supposed to yield the set of innate concepts. Is that not circular: the definitional analysis yields the set of restrictions to which it is supposed to conform?!

A. I may have expressed myself inaccurately. The actual criterion for the definitions is their cross-linguistic transparency, their understandability, if you like. A superior definition, you will agree, is one that can be understood without prior knowledge of the language to be defined. In actual practice, this is achieved by couching the definition in terms that are cross-linguistically unambiguous, such as SEE and HEAR, THIS, I, YOU, WHERE, GOOD and BAD, SAME and OTHER, or THINK. SO the ultimate restriction is not the innateness of the defining terms, but the crosslinguistic transparency of the definitions. D. I sometimes doubt whether transparency is exactly what you achieve with your type of definitions, but anyway, what strikes me as particularly odd in this conversation as a whole, is the affinity that now becomes apparent between Cognitive Linguistics and formal semantics.

440

Theory and method

A. Now it is my turn to be surprised - flabbergasted, rather. A likeness between Cognitive Linguistics and formal semantics? What in love's name are we to understand by that? D. You may remember that Histrio once tried to convince us that formal semantics was (originally at least) a normative enterprise, a method of stipulating requirements for how to describe meanings. But perhaps, master, you had better tell us about that yourself. Am I overlooking something? H. You are absolutely right from a historical point of view, Duodecimus. Formal semantics, as it was developed by applying formal logic to natural language, was explicitly intended by at least some of its founding fathers (like the members of the Vienna Circle) as a tool for separating metaphysical speculation from empirically viable statements. Logic is concerned with the validity of arguments, and the merits of formalization in logic consist in restricting possible propositions to ones whose validity may be tested, in contrast with mere speculation. And so formal semantics, like Antipoda's approach, tries to solve Dilthey's problem (if we may call it that - I mean the problem of controlling the process of interpretation lest it slide off towards 'romantic arbitrariness') by imposing restrictions on how possible interpretations might be expressed. The clarity of the logical language is a way of ensuring the epistemological viability of statements. What can be legitimately expressed as an interpretation of utterances in a natural language, is restricted by how it can be expressed in the language of formal logic. D. In the same way in which Antipoda wishes to restrict the boundless freedom of interpretation by imposing requirements on how the interpretations are expressed. My approach to finding restrictions, by contrast, is not representational, but rather empirical: I want suggested interpretations to be empirically grounded in what we know of the cultural background, the actual behavior of the language users, the physiology of the human conceptual apparatus.

A. All that is very fine, but I do not exactly feel like a formal semanticist. I think you have an obligation to explain the differences between my position and the formalists. But master, what are you scribbling there in the sand? I hope we are not boring you.

Idealist and empiricist tendencies

441

H. On the contrary. I was thinking about a way to answer your question. Perhaps you would care to have a look at my little sketch here. If we make a distinction between the restrictive principles applied by the various approaches on the one hand, and their major aim on the other, the difference between Antipoda's approach and the formalists' is to a large extent that the latter are hardly interested in cognitive relevance, in the psychological sense in which Cognitive Semantics tends to conceive of that term. Cognitive Semantics is interested in the full range of semantic phenomena that seem to play a role in actual language use - including those that cannot be formally described within the framework of logical semantics. Formalists, by contrast, seem willing to accept the restriction of the description to the truth-functional properties of language alone, given that those may well be (they feel) the only ones that can be described according to the high demands of formalizability that they wish to impose on linguistics. RESTRICTIVE

METHODOLOGICAL POSI-

MAJOR

PRINCIPLE

TION

AIM

empirical restrictions

empiricist Cognitive Linguistics idealist Cognitive Linguistics

representational restrictions I

I

formal semantics

cognitive relevance logical validity

Figure I. Methodological relations between the different approaches to semantics

A. You take the words right out of my mouth, master. 'Despite its name,

'formal semantics' [...] doesn't seek to reveal and describe the meanings encoded in natural language, or to compare meanings across languages and cultures. Rather, it sees its goal as that of translating certain carefully selected types of sentences into a logical calculus. It is interested not in meaning (in the sense of conceptual structures encoded in language) but in the logical properties of sentences such as entailment, contradiction, or logical equivalence' (Wierzbicka 1996: 8). D. Very illuminating indeed - but I notice the sundial is just on the other side of this hedge. What a beautiful object! How should one read it?

442

Theory and method

H. Let me show you. See that shadow here? The number that it falls on along this circular band indicates the hour of the day: what you see is not time itself, but a sign of time.

D. And just so, linguistic behavior is not the conceptualization as such, but only a sign to be interpreted.

A. Hm, if you interpret this shadowy sign correctly, you will notice that the tables are now likely to be being laid out; the hour of dinner is approaching. Shall we perhaps walk back to the house? H. Yes, we should: a happy hour is awaiting us.

Epilogue

In the foregoing philosophical dialogue, I have compared an empiricist and an idealist conception of the methodology of Cognitive Semantics. Against the background of my previous research, this comparison continues the confrontation of Cognitive Semantics with philosophical epistemology that I began in Geeraerts (1985a), where the possible 'backfiring' of Cognitive Semantics' adherence to internal realism was first described. Other work along this line of thought includes Geeraerts (1993c), and the final chapter of Geeraerts (1997a), which includes a discussion of Diltheyan hermeneutics. At the same time, the dialogue is a methodological reflection on the type of work presented in Geeraerts, Grondelaers, and Bakema (1994), which takes a heavily empiricist attitude. The dialogue proceeded in eight steps. First, the notion of internal realism was introduced on the basis of an example taken from the natural sciences, involving the interpretation of Michelson and Morley's experiments. Second, it was recalled that internal realism characterizes the epistemological conception that Cognitive Semantics has of the relationship between language and the world. Third, the adherence to internal realism was shown to raise a methodological problem: can it be extrapolated to the relationship between language and the linguistic description of language? And if so, what safeguards are there against a lapse into radical subjectivism? Empiricist and idealist approaches in Cognitive Semantics were presented as different answers with regard to these questions. Fourth, the initial discussion between the proponents of both approaches remained an open-ended one. Although it was shown that an introspective, idealist methodology does not always yield

Idealist and empiricist tendencies

443

an adequate description of the way in which words are actually being used, there were at least two possibilities for the idealist to escape from the consequences of this discrepancy and to keep the debate open: to dissociate judgments about category membership from conceptual knowledge (in a way linking up with Putnam's theory of natural kind terms), or by imposing a radical distinction between semantic structure and surface structure. Fifth, the open-ended discussion was placed in a wider framework, by interpreting it against the philosophical background of Wilhelm Dilthey' s notion of hermeneutics. It was shown that the Diltheyan conception of language fits Cognitive Semantics very well. Sixth, starting from this recognition, it was suggested that the debate involves a choice between two ways of thinking about the interpretative aspects of linguistic semantics, rather than a straightforward choice between empiricist objectivism and idealist subjectivism in any crude form. More specifically, the two competing approaches were seen as representing alternative ways of avoiding the danger of 'romantic arbitrariness' as formulated by Dilthey. Whereas the 'empiricist' approach tries to achieve maximal empirical grounding for posited interpretations, the 'idealist' approach tries to restrict the set of possible interpretations by couching it in a vocabulary of allegedly universal concepts. Seventh, without enforcing a [mal decision, I suggested that the grounding approach represents the more radical safeguard against arbitrariness. Finally, the two approaches within Cognitive Semantics were compared with formal semantics as yet another answer to Dilthey's problem. Relying on a vocabulary of universal primitives, the idealist approach within Cognitive Semantics shares with formal semantics a 'representational' strategy for imposing restrictions on linguistic descriptions. At the same time, it shares with the empiricist approach in Cognitive Semantics a desire to describe the full range of semantic phenomena that play a role in natural language use, rather than restricting the domain of enquiry to the 'logical', truth-functional properties of language. What emerges from the discussion, then, is not exactly a straightforward answer with regard to the initial question - although I do feel that an empiricist approach imposes stronger demands on the semantic enterprise than the idealist one, and as such, provides a more outspoken answer to Dilthey's problem. The picture emerging from the discussion does, on the other hand, provide an illuminating view of the present-day situation in linguistic semantics. Perhaps surprisingly, the major conceptions of linguistic semantics appear to pattern together as different answers with regard to Dilthey's problem, that is, the problem of what constitutes a legitimate interpretation of natural language expressions. I would like to suggest that this is a good

444

Theory and method

reason for putting the interpretative nature of linguistic semantics in the forefront of theoretical attention.

References

Agirre, Eneko, and Philip Edmonds (eds.) 2006 Word Sense Disambiguation. Algorithms and Applications. Berlin: Springer. Aijmer, Karen 1985 The semantic development of will. In Historical S'emantics Historical Word-Formation, Jacek Fisiak (ed.), 11-21. Berlin,lNew York: Mouton de Gruyter. Anttila, Raimo 1985 Language, cognition, and linguistics. In Linguistics and Philosophy. Essays in Honor of Rulon S. Wells, Adam Makkai and Alan K. Melby (eds.), 11-22. AmsterdamlPhiladelphia: John Benjamins. Apresjan, Juri 1966 Analyse distributionnelle des significations et champs semantiques structures. Langages 1: 44-74. Armstrong, Sharon L., Lila R. Gleitman, and Henry Gleitman 1983 What some concepts might not be. Cognition 13: 263-308. Ayto, John 1983 On specifying meaning. In Lexicography. Principles and Practice, Reinhard R.K. Hartmann (ed.), 89-98. LondonlNew York: Academic Press. Baldinger, Kurt 1980 Semantic Theory. Towards a Nfodern Semantics. Oxford: Basil Blackwell. Ballmer, Thomas, and Manfred Pinkal (eds.) 1983 Approaching Vagueness. Amsterdam: North Holland Publishing Company. Barcelona, Antonio 2001 On the systematic contrastive analysis of conceptual metaphors: Case studies and proposed methodology. In Applied Cognitive Linguistics lJ: Language Pedagogy, Martin Piitz, Susanne Niemeier, and Rene Dirven (eds.), 117-146. BerlinINew York: Mouton de Gruyter. Barere, Bertrand 1975 Rapport du Comite de Salut Public sur les idiomes. In Une politique de la Langue: La Revolution franr;aise et les Patois, Michel de Certeau, Dominique Julia, and Jacques Revel (eds.), 291-299. Paris: Editions Gallimard. Original edition 1794.

446

References

Barlow, Michael, and Suzanne Kemmer (eds.) 2000 Usage-based Models ofLanguage. Stanford: CSLI Publications. Barnes, Jonathan (ed.) 1984 The Complete works ofAristotle. The Revised Oxford Translation. Princeton: Princeton University Press. Bartsch, Renate 1985 Sprachnormen: Theorie und Praxis. Ttibingen: Niemeyer Verlag. Beek, Henri H. 1969 Waanzin in de Middeleeuwen. Haarlem: De Toorts. Berlin, Brent 1976 The concept of rank in ethnobiological classification: some evidence from Aguarana folk botany. American Ethnologist 3: 381400. 1978 Ethnobiological classification. In Cognition and Categorization, Eleanor E. Rosch and Barbara B. L10yd (eds.), 9-26. Hillsdale, NJ: Lawrence Erlbaum. Berlin, Brent, and Paul Kay 1969 Basic Color Terms. Berkeley: University of California Press. Berlin, Brent. Donald E. Breedlove, and Paul H. Raven 1973 General principles of classification and nomenclature in folk biology. American Anthropologist 75: 214-242. Berthele, Raphael 2004 The typology of motion and posture verbs: a variationist account. In Dialectology meets Typology. Dialect Grammar from a Cross-Linguistic Perspective, Bernd Kortmann (ed.), 93-126. BerlinINew York: Mouton de Gruyter. Biebuyck, Benjamin, Rene Dirven, and John Ries (eds.) 1998 Faith and Fiction. Interdisciplinary Studies on the Interplay between Metaphor and Religion. Frankfurt am Main: Peter Lang. Bierwisch, Manfred 1969 On certain problems of semantic representations. Foundations of Language 5:153-184. Bleicher, Josef 1980 Contemporary Hermeneutics. London: Routledge and Kegan Paul. Blommaert, Jan, and Chris Bulcaen (eds.) 1997 Political Linguistics. AmsterdamJPhiladelphia: John Benjamins.

References

447

Boas, Franz Introduction. In Handbook of American Indian Languages, 1911 Franz Boas, 1-83. Washington DC: Government Printing Office. Boeve, Lieven, and Kurt Feyaerts (eds.) 1999 Metaphor and Godtalk. Bern: Peter Lang. Bosch, Peter Vagueness, ambiguity. and all the rest. In Sprachstruktur, Indi1979 viduum, und Gesellschaft. Marc van de Velde and Willy Vandeweghe (eds.), 9-19. Ttibingen: Max Niemeyer Verlag. Brandt, Per A. 1998 Cats in space. Baudelaire's 'Les chats' read by Jakobson and Levi-Strauss. In The Roman lakobson Centennial Symposium 1996, Per A. Brandt (ed.), 339-347. Copenhagen: C. A. Reitzel. Breal, Michel 1897 Essai de Semantique. Paris: Hachette. Brugman, Claudia 1981 The story of over. MA thesis, University of California at Berkeley. 1984 The very idea. A case study in polysemy and cross-lexical generalizations. In Papers from the Parasession on Lexical Semantics, David Testen, Veena Mishra, and Joseph Drogo (eds.), 2138. Chicago: Chicago Linguistic Society. 1988 The Story o.fOver: Po(vsemy, S'emantics and the Structure ofthe Lexicon. New York: Garland. Burke. Lucy, Tony Crowley, and Allan Girvin (eds.) 2000 The Routledge Language and Cultural Theory Reader. LondonINew York: Routledge. Bybee, Joan, and Carol L. Moder 1983 Morphological classes as natural categories. Language 59: 251270. Campbell, Lily B. 1930 Shakespeare's Tragic Heroes: Slaves of Passion. Cambridge: Cambridge University Press. Carnoy, Albert 1927 La Science du Mot: Traite de Semantique. Leuven: Universitas. Catlin, Jane-Carol, and Jack Catlin 1972 Intentionality: a source of ambiguity in English? Linguistic Inquiry 3: 504-508.

448

References

Chapman, Allan 1979 Astrological Medicine. In Health, Medicine, and Mortality in the Sixteenth Century, Charles Webster (ed.), 275-300. Cambridge: Cambridge University Press. Chierchia, Gennaro, and Sally McConnell-Ginet 1990 Meaning and Grammar. An Introduction to Semantics. Cambridge, Mass.: MIT Press. Clark, Herbert R, and Eve V. Clark 1977 Psychology and Language. An Introduction to Psycholinguistics. New York: Harcourt Brace Jovanovitch. Coleman, Linda, and Paul Kay 1981 Prototype semantics: the English word lie. Language 57: 26-44. Colleman, Timothy 2005 Central and less--central subsenses of the Dutch ditransitive: Evidence from a collexeme analysis. Paper presented at the International Cognitive Linguistics Conference 9, Seoul, Korea. Condillac, Etienne Bonnot de 1973 Essai sur l'Origines des Sciences humaines. Charles Porset (ed.). Auvers-sur-Oise: GaliIee. Original edition 1746. Copi, Irving M. 1972 Introduction to Logic. 4th ed. New York: MacMillan. Coseriu, Eugenio 1964 Pour une semantique diachronique structurale. Travaux de LingUistique et de Litterature 2 : 139-186. Coseriu, Eugenio, and Horst Geckeler 1981 Trends in Structural Semantics. Tiibingen: Gunter Narr Verlag. Coulson, Seana 2001 Semantic Leaps: Frame-shifting and Conceptual Blending in Meaning Construction. Cambridge: Cambridge University Press. Craig, Colette (ed.) 1986 Noun Classes and Categorization. AmsterdamJPhiladelphia: John Benjamins. Croft, William 1998 Linguistic evidence and mental representations. Cognitive Linguistics 9: 151-173. 2001 Radical Construction Grammar: S:vntactic Theory from a Typological Perspective. Oxford: Oxford University Press. Cruse, D. Alan 1982 On lexical ambiguity. Nottingham Linguistic Circular 11/2: 6580. 1986 Lexical S'emantics. Cambridge: Cambridge University Press.

References

2000a

449

Aspects of the micro-structure of word meanings. In Polysemy, Yael Ravin and Claudia Leacock (eds.), 30-51. Oxford: Oxford University Press. 2000b Meaning in Language. Oxford: Oxford University Press. Cruttwe11, Patrick 1951 Physiology and psychology in Shakespeare's age. Journal of the History ofIdeas 12: 75-89. Cuyckens, Hubert 1991 The Semantics of Spatial Prepositions in Dutch. PhD thesis, Universitaire Inste11ing Antwerpen. Danaher, David S. 2003 A cognitive approach to metaphor in prose: Truth and falsehood in Leo Tolstoy's 'The Death of Ivan Illich'. Poetics Today 24: 439-470. Darmesteter, Arsene 1887 La Vie des Mots etudifie dans leurs Signijications. Paris: Delagrave. De Beer, Taco 1897 Tijdscluiften. Noord en Zuid 20: 460. De Flines, Quirijn 1810 Proeve van Hollandsche of Nederduitsche Synonyma. Amsterdam: Bolt. De Sutter, Gert, Dirk Speelman, Dirk Geeraerts, and Stefan Grondelaers 2003 Woordvolgordevariatie in tweeledige werkwoordelijke eindgroepen. Naar een statistische evaluatie van zes factoren. In Artikelen van de Vierde Sociolinguistische Conferentie, Tom Koole, Jacomine Nortier, and Bert Tahitu (eds.), 111-119. Delft: Eburon. Deane, Paul D. 1988 Polysemy and cognition. Lingua 75: 325-361. 1992 Polysemy as the consequence of internal conceptual complexity: The case of over. Proceedings of the Eastern States Conference on Linguistics (ESCOL), 9. Ohio State University, CLC Publications: 32-43. 1993 At, by, to and past: a study in multimodal image theory. In Proceedings of the Nineteenth Annual Meeting of the Berkeley Linguistics Society, Joshua Guenter, Barbara Kaiser, and Cheryl Zol1 (eds.). 112-124. Berkeley: : Berkeley Linguistics Society. Den Boon, Ton, and Dirk Geeraerts 2005 Van Dale Groot Woordenboek van de Nederlandse Taa/. 14th ed. Utrecht, Antwerpen: Van Dale Lexicografie.

450

References

Deprez, Kas, and Louis Vos (eds.) 1998 Nationalism in Belgium: Shifting Identities. Basingstoke : Macmillan. Dewell, Robert B. 1994 Over again: On the role of image-schemas in semantic analysis. Cognitive Linguistics 5: 351-380. Diepgen, Paul 1955 Geschichte der Medizin. Berlin, New York: De Gruyter. Dilthey, Wilhelm 1961 Die Entstehung der Hermeneutik. In Gesammelte Schriften v: 317-331. Leipzig: Teubner. Original edition 1913. Dirven. Rene 1985 Metaphor and polysemy. Cahiers de l'lnstitut de Linguistique de Louvain 11: 9-27. 1988 Bibliography of Cognitive Linguistics. Duisburg: Linguistic Agency of the University of Duisburg. 1990 Metaphorical divergences in the evolution of Dutch and Afrikaans. In Diachronic Semantics (Belgian Journal of Linguistics 5), Dirk Geeraerts (ed.), 9-34. Brussels: Editions de l'Universite de Bruxelles. 2001 The metaphoric in recent cognitive approaches to English phrasal verbs. Metaphorik.de 1: 39-54. Dirven, Rene, Bruce Hawkins, and Esra Sandikcioglu (eds.) 2001 Language and Ideology 1. Theoretical Cognitive Approaches. AmsterdamJPhiladelphia: John Benjamins. Dirven, Rene, Roslyn Frank, and Cornelia Hie (eds.) 2001 Language and Ideology JJ. Descriptive Cognitive Approaches. AmsterdamJPhiladelphia: John Benjamins. Dirven, Rene, and Marjolijn Verspoor (eds.) 1998 Cognitive Exploration of Language and Linguistics. AmsterdamJPhiladelphia: John Benjamins. Dobrovol'skij, Dmitri 1995 Schiss und Espenlaub: Idiome der Angst. Folia Linguistica 29: 317-346. Dowty, David 1979 IVord kfeaning and IvJontague Grammar. Dordrecht: Reidel. Draper, John W. 1965 The HUl1lors and Shakespeare's Characters. New York: AMS Press. Dressier, Wolfgang 1985 Morphophonology. Ann Arbor: Karoma.

References

451

Dubois, Jean 1964 Distribution, ensemble et marque dans le lexique. Cahiers de Lexicologie 4 : 5-16. Dunbar, George 2001 Towards a cognitive analysis of polysemy, ambiguity and vagueness. Cognitive Linguistics 12: 1-14. Ekman, Paul, Robert W. Levenson, and Wallace V. Friesen 1983 Autonomic Nervous System Activity Distinguishes among Emotions. Science 221: 1208-1210. Elffers, EIs 1999 Predecessors of cognitive semantics and speech act theory? In History of Linguistics 1996. Selected Papers from the Seventh International Conference on the History of the Language Sciences 11. From Classical to Contemporary LingUistics, David Cram, Andrew R. Linn and Elke Nowak (eds.), 307-320. AmsterdamlPhiladelphia: John Benjamins. Erdmann, Karl Otto 1910 Die Bedeutung des Wortes. Leipzig: Haessel. Evans, Vyvian, and Andrea Tyler 2001 Reconsidering prepositional polysemy networks: The case of over. Language 77: 724-765. Fellbaum, Christiane (ed.) 1998 WordNet. An Electronic Lexical Database. Cambridge, Mass.: MIT Press. Feyaerts, Kurt 1997 Die Bedeutung der Metonymie als konzeptuellen Strukturprinzips. Eine kognitiv-semantische Analyse deutscher Dummheitsausdrticke. PhD thesis, University ofLeuven. 1999 Metonyl11ic hierarchies: The conceptualization of stupidity in German idiomatic expressions. In Metonymy in Language and Thought, Klaus-Uwe Panther and Giinter Radden (eds.), 309332.. AmsterdamlPhiladelphia: John Benjal11ins. Feyaerts, Kurt (ed.) 2003 The Bible through Metaphor and Translation: A Cognitive Semantic Perspective. Bern, Peter Lang. Fillmore, Charles J. 1975 An alternative to checklist theories of meaning. In Proceedings of the First Annual Meeting of the Berkeley Linguistics Socie~y, Cathy Cogen, Henry Thol11pson, and James Wright (eds.), 123131. Berkeley: Berkeley Linguistics Society.

452

References

1977a

Scenes-and-frames semantics. In Linguistic Structures Processing, Antonio Zampolli (ed.), 55-81. Amsterdam: North Holland Publishing Company. 1977b Topics in lexical semantics. In Issues in Linguistics, Richard Cole (ed.), 76-138. Bloomington: Indiana University Press. 1982 Towards a descriptive framework for spatial deixis. In Speech, Place, and Action: Studies in Deixis and Related Topics, Robert Jarvella and Wolfgang Klein (eds.), 31-59. Chichester: Wiley. Fillmore, Charles 1., Paul Kay, and Kathryn O'Connor 1988 Regularity and idiomaticity in grammatical construction: The case of let alone. Language 64: 501-538. Fillmore, Charles 1., and Beryl T.S. Atkins 1992 Towards a frame-based lexicon: the semantics of risk and its neighbors. In Frames, Fields and Contrasts, Adrienne Lehrer and Eve Kittay (eds.), 75-102. Hillsdale, NI: Lawrence ErIbaum. 2000 Describing polysemy: the case of crawl. In Polysemy. Theoretical and Computational Approaches, Yael Ravin and Claudia Leacock (eds.), 91-110. Oxford: Oxford University Press. Firth, John R. 1957 Papers in Linguistics 193-1-1951. London: Oxford University Press. Fodor, Janet, Jerry A. Fodor, and Merrill Garrett 1975 The psychological unreality of semantic representations. Linguistic InqUiry 6: 515-531. Forceville, Charles 1999 The metaphor COLIN IS A CHILD in Ian McEwan's, Harold Pinter's, and Paul Schrader's 'The Comfort of Strangers'. Metaphor and S'ymbo/14: 179-198. Frawley, William 1981 In defense of the dictionary. A response to Haiman. Lingua 55: 53--61. Freeman, Margaret H. 1995 Metaphor making meaning: Dickinson's conceptual universe. Journal ofPragmatics 24: 643-666. 2000 Poetry and the scope of metaphor: Toward a cognitive theory of literature. Afetaphor and Metonymy at the Crossroads: A Cognitive Per.spective, Antonio Barcelona (ed.), 253-281. BerlinlNew York: Mouton de Gruyter. Fried, Mirjam, and Jan-Ola Ostman (eds.) 2004 Construction Grammar in a Cross-Language Perspective. AmsterdamlPhiladelphia: John Benjamins.

References

453

Friedman, Hershey 2000 Humor in the Hebrew Bible. Humor. International Journal of Humor Research 13: 258-285. Geckeler, Horst 1993 StruktureUe Wortfeldforschung heute. In Studien zur Wortfeldtheorie. Studies in Lexical Field Theory, Peter R. Lutzeier (ed.), 11-21. Ttibingen: Max Niemeyer Veriag. Geeraerts, Dirk 1981 Review of George Lakoff and Mark Johnson, Metaphors we Live by. Quaderni di Sel1lantica 2: 389-396. 1983a Prototype theory and diachronic semantics: a case study. Indogermanische Forschungen 88: 1-32. 1983b Review of Kurt Baldinger, Semantic Theory. Lingua 60: 231238. 1983c Reclassifying Semantic Change. Quaderni di Semantica 4: 217240. 1983d Lexicografie en linguistiek: Reichling gerehabiliteerd. Tijdschrijl voor Nederlandse Taal- en Letterkunde 99: 186-206. 1984a Type en prototype. Tijdschrijl voor Tekst- en Taalwetenschap 4: 69-8 1984b Dictionary classification and the foundations of lexicography. ITL. Review ofApplied LingUistics 63: 37--63. 1985a Paradigm and Paradox. Explorations into a Paradigmatic Theory of Meaning and its Epistemological Background. Leuven: Universitaire Pers. 1985b Les donnees stereotypiques, prototypiques et encyclopediques dans le dictionnaire. Cahiers de Lexicologie 46: 27-43. 1985c Cognitive restrictions on the structure of semantic change. In Historical Semantics - Historical Word Formation, Jacek Fisiak (ed.), 127-153. Beriin/New York: Mouton de Gruyter. 1985d Diachronic extensions of prototype theory. In Meaning and the lexicon, Geer Hoppenbrouwers, Pieter Seuren and Ton Weijters (eds.), 354-362. Dordrecht: Foris. 1985e Preponderantievreschillen bij bijna-synoniemen. De Nieuwe Taalgids 78: 18-27. 1986a Woordbetekenis. Een Overzicht van de Lexicale Semantiek. Leuven: Acco. 1986b Lexicografische vernieuwing in de vroege achttiende eeuw. Documentatieblad Achttiende Eeuw 1-14. 1987a On necessary and sufficient conditions. Journal of Semantics 5: 275-291.

454

References

1987b

1988a

1988b

1988c 1988d

1989a 1989b

1989c 1990a

1990b 1992a 1992b 1993a 1993b

1993c

Types of semantic information in dictionaries. In A Spectrum of Lexicography, Robert Ilson (ed.), 1-10. AmsterdamlPhiladelphia: John Benjamins. Cognitive Grammar and the History of Lexical Semantics. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 647-677. AmsterdamlPhiladelphia: John Benjamins. [This volume, Chapter 16.) Katz revisited: aspects of the history of lexical semantics. In Understanding the Lexicon, Werner Hiillen and Rainer Schulze (eds.), 23-35. Tiibingen: Max Niemeyer Verlag. Review of Anna Wierzbicka, Lexicography and Conceptual Analysis. Language in Society 17: 449-455. Where does prototypicality come from? In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 207-229. AmsterdamlPhiladelphia: John Benjamins. [This volume, Chapter 2.] Wat er in een Woord Zit. Facetten van de Lexicale 5,'emantiek. Leuven: Peeters. Principles of monolingual lexicography. In Worterbiicher. Ein internationales Handbuch zur Lexikographie, Franz Josef Hausmann, Oskar Reichmann, Herbert Ernst Wiegand and Ladislav Zgusta (eds.), 1: 287-296. BerlinlNew York: De Gruyter. Prospects and problems of prototype theory. Linguistics 27: 587612. [This volume, Chapter I.] The lexicographical treatment of prototypical polysemy. In Meanings and Prototypes. Studies in Linguistic Categorization, Savas L. Tsohatzidis (ed.), 195-210. LondonlNew York: Routledge. [This volume, Chapter 13.] Homonymy, iconicity, and prototypicality. Belgian Journal of Linguistics 5: 49-74. The semantic structure of Dutch over". Leuvense Bijdragen 81: 205-230. [This volume, Chapter 3.) Prototypicality and polysemy. On Georges Kleiber, La semantique du prototype. Cognitive Linguistics 3: 219-231. Vagueness's puzzles, polysemy's vagaries. Cognitive Linguistics 4: 223-72. [This volume, Chapter 5.] Generalised onomasiological salience. In Perspectives on Language and Conceptualization (Belgian Jounal of Linguistics 8), Jan Nuyts & Eric Pederson (eds.), 43-56. Brussel: Editions de l'Universite de Bruxelles. Cognitive semantics and the history of philosophical epistemology. In Conceptualizations and Mental Processing in Language,

References

1995a 1995b

1997a 1997b

1999a

1999b

2000a

2000b

455

Richard Geiger and Brygida Rudzka-Ostyn (eds.), 53-80. BerlinINew York: Mouton de Gmyter. Representational formats in cognitive semantics. Folia Linguistica 39: 21-41. Specialisation and reinterpretation in idioms. In Idioms. Structural and Psychological Perspectives, Martin Everaert, Erik-Jan van der Linden, Andre Schenk, and Rob Schreuder (eds.), 5774. Hillsdale, NJ: Lawrence Erlbaum. Diachronic Prototype Semantics. Oxford: Clarendon Press. Entfernung, Fortschritt und UnvoIIstiindigkeit. Das historische VerhaItnis von Lexikologie und Lexikographie. In Linguistische Theorie und lexikographische Praxis, Klaus-Peter Konerding and Andrea Lehr (eds.), 7-16. Tiibingen: Max Niemeyer Verlag. Beer and semantics. In Issues in Cognitive Linguistics, Leon De Stadler and Chris Eyrich (eds.), 35-55. BerlinINew York: Mouton de Gmyter. [This volume, Chapter 10.] Hundred years of lexical semantics. In Aetas do jO Encontro Internacional de Linguistica Cognitiva, Mario Vilela and Fatima Silva (eds.), 123-154. Porto: Faculdade de Letras. Adding electronic value. The electronic version of the Grote Van Dale. In The Ninth Euralex International Congress Proceedings, Ulrich Heid, Stefan Evert, Egbert Lehmann and Christian Rohrer (eds.), I: 75-84. Stuttgart: Institut fur MaschineIIe Sprachverarbeitung. Honderd jaar Belgisch in Van Dale. Nederlands van Nu 48: 1923.

2001a

2001a

2002a

2002b

Everyday language in the media. The case of Belgian Dutch soap series. In Sprache im Alltag. Beitrage zu neuen Perspektiven in der Linguistik Herbert Ernst Wiegand zum 65. Geburtstag gewidmet. Matthias Kammerer, Klaus-Peter Konerding, Andrea Lehr, Angelika Storrer, Caja Thimm, and Werner Wolski (eds.), 281-291. BerlinlNew York: WaIter de Gmyter. On measuring lexical convergence. In Linguagem e Cogni~ao. A Perspectiva da Linguistica Cognitiva, Augusto Soares da Silva (ed.), 51-61. Braga: Associac;ao Portuguesa de Linguistica. De 19de eeuw als lexicale breuklijn. Verslagen en Mededelingen van de Koninklijke Academie voor Nederlandse Taal- en Letterkunde 112: 489-507. The scope of diachronic onomasiology. In Das Wort. Seine strukturelle und kulturelle Dimension. Festschrift fur Oskar Reichmann zum 65. Geburtstag, Vilmos Agel, Andreas Gardt,

456

References

Ulrike Hass-Zumkehr and Thorsten Roelcke (eds.), 29-44. Tiibingen: Max Niemeyer Verlag. 2003 Meaning and definition. In A Practical Guide to Lexicography, Piet Van Sterkenburg (ed.), 83-93. All1sterdamIPhiladelphia: John Benjamins. 2005 Lectal data and empirical variation in Cognitive Linguistics. In Cognitive Linguistics. Internal Dynamics and Interdisciplinary Interactions, Francesco Ruiz de Mendoza Ibaiiez and Sandra Peiia Cervel (eds.), 163-189. BerlinINew York: Mouton de Gruyter. Geeraerts. Dirk, and Peter Bakell1a 1993 De prismatische semantiek van idioll1en en composita. Leuvense Bijdragen 82: 185-226. Geeraerts, Dirk, Stefan Grondelaers, and Peter Bakema 1994 The Structure ofLexical Variation. Meaning, Naming, and Context. BerlinINew York: Mouton de Gruyter. Geeraerts, Dirk, Stefan Grondelaers, and Dirk Speelman 1999 Convergentie en Divergentie in de Nederlandse Woordenschat. Een Onderzoek naar Kleding- en Voetbaltermen. Amsterdam: Meertens Instituut. Geeraerts. Dirk. and Guy Janssens 1982 Wegwijs in Woordenboekel1. Een Kritisch Overzicht van de Lexicograjie van het Nederlands. Assen: Van Gorcum. Gevaert, Caroline 1994 Prismatische breking bij idiomen. Het prismatische betekenismodel in de diachrone studie van rei'nterpretatieprocessen bij idiomen. MA thesis, University of Leuven. 2001 Anger in Old and Middle English: A 'hot' topic? Belgian Essays on Language and Literature, 2001: 89-101. Gibbs, Raymond W. 1986 Skating on thin ice: Literal meaning and understanding idioms in conversation. Discourse Processes 9: 17-30. 1990 Psycholinguistic studies on the conceptual basis of idiomaticity. Cognitive Linguistics 1: 417-451. 1994 The Poetics of Mind. Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. 1995 Idiomaticity and human cognition. In Idioms: S'tructural and Psychological Perspectives, Martin Everaert, Erik-Jan Van Der Linden, Andre Schenk, and Rob Schreuder (eds.), 97-116. Hillsdale, NJ: Lawrence Erlbaum.

References

457

Gibbs, Raymond W., and Jennifer E. O'Brien 1990 Idioms and mental imagery: The metaphorical motivation for idiomatic meaning. Cognition 36: 35-68. Gibbs, Raymond W., Josephine M. Bogdanovich, Jeffrey R. Sykes, and Dale J. Barr 1997 Metaphor in idiom comprehension. Journal of Memory and Language 37: 141-154. Givon, Talmy 1986 Prototypes: between Plato and Wittgenstein. In Noun Classes and Categorization, Colette Craig (ed.), 78-102. AmsterdamJPhiladelphia: John Benjamins. Glass, Amold and Keith Holyoak 1975 Altemative conceptions of semantic theory. Cognition 3: 313339. Goddard, Cliff 1998 Semantic Ana(vsis. Oxford: Oxford University Press. Godderis, Jan 1988 Galenos van Pergamon over Psychische Stoornissen. Leuven: Acco. Goldberg, Adele 1992 The inherent semantics of argument structure: the case of the English ditransitive construction. Cognitive Linguistics 3: 3774. 1995 Constructions: A Construction Grammar Approach to Argument .)'tructure. Chicago, University of Chicago Press. Good, Edwin M. 1965 Irony in the Old Testament. Philadelphia: The Westminster Press. Goodenough, Ward H. 1956 Componential analysis and the study of meaning. Language 32: 195-216. Goossens, Louis 1990 Metaphtonymy: The interaction of metaphor and metonymy in expressions for linguistic action. Cognitive Linguistics 1: 323340. Gordon, W. Terrence 1982 A History of Semantics. Amsterdam/Philadelphia: John Benjamins. Gregoire, Henri-Baptiste 1794 [1975] Rapport sur la necessite et les moyens d'aneantir les patois et d'universaliser l'usage de la langue franc;;aise. In : Michel de Certeau, Dominique Julia, and Jacques Revel (eds.), Une

458

References Politique de la Langue: La Revolution franr;aise et les Patois, 300-317. Paris: Editions Gallimard.

Grice, Paul Studies in the Way of Words, Cambridge, Mass., Harvard University Press. Gries, Stefan Th. 2003 Multifactorial Analysis in Corpus Linguistics: A Study ofParticle Placement. London: Continuum Press Grondelaers, Stefan 2000 De distributie van niet-anaforisch er buiten de eerste zinsplaats. Sociolectische, functionele en psycholinguistische aspecten van er's status als presentatief signaal. PhD thesis, University of Leuven. Grondelaers, Stefan, and Dirk Geeraerts 2003 Towards a pragmatic model of cognitive onomasiology. In Cognitive Approaches to Lexical Semantics, Hubert Cuyckens, Rene Dirven and John Taylor (eds.), 67-92. BerlinJNew York: Mouton de Gmyter. Grondelaers, Stefan, Dirk Geeraerts, Dirk Speelman, and Jose Tummers 2001 Lexical standardisation in internet conversations. Comparing Belgium and The Netherlands. In Proceedings of the First International Conference on Language Variation in Europe, Josep M. Fontana, Louise McNally, M. Teresa Turell, and Enric Vallduvi (eds.), 90-100. Barcelona: Universitat Pompeu Fabra. Grondelaers, Stefan, Dirk Speelman, and Dirk Geeraerts 2002 Regressing on er. Statistical analysis of texts and language variation. In 6iemes Journees Internationales d'Analyse Statistique des Donnees Textuelles - 6th International Conference on Textual Data Statistical Analysis, Anne Morin and Pascale Sebillot (eds.), 335-346. Rennes: Institut National de Recherche en Infonnatique et en Automatique. Grondelaers, Stefan, Marc Brysbaert, Dirk Speelman, and Dirk Geeraerts 2002 Er als accessibility marker: on- en offline evidentie voor een procedurele interpretatie van presentatieve zinnen. Gramma/TTT 9: 1-22. John Haiman 1980a Dictionaries and encylopedias. Lingua 50: 329-357,1980. 1980b The iconicity of grammar. Language 56:515-54. 1982 Dictionaries and encylopedias again. Discussion. Lingua 56: 353-355. Halliday, Michael A.K., and Ruqaiya Hasan 1976 Cohesion in English. London: Longman. 1989

References

459

Hamilton, Craig A. 2002 Conceptual integration in Christine de Pizan's 'City of Ladies'. In Cognitive S'tylistics. Language and Cognition in Text Analysis, Elena Semino and Jonathan Culpeper (eds.), 1-22. AmsterdamlPhiladelphia: John Benjamins. Hanks, Patrick 1994 Linguistic norms and pragmatic exploitations, or why lexicographers need prototype theory, and vice versa. In Papers in Computational Lexicography, Ferenc Kiefer, Gabor Kiss, Julia Pajzs (eds.), 89-113. Budapest: Hungarian Academy of Sciences. Harder, Peter 2003 The status of linguistic facts. Rethinking the relation between cognition, social institution and utterance from a functional point of view. Mind and Language 18: 52-76. Harkins, Jean, and Anna Wierzbicka (eds.) 2001 Emotions in Cross-linguistic Perspective. Cognitive Linguistics Research 17. BerlinINew York: Mouton de Gruyter. Harris, Catherine L. 1989 A connectionist approach to the story of over. In Proceedings of the Fifteenth Annual Meeting of the Berkeley Linguistics S'ociety, Kira Hall, Michael Meacham, and Richard Shapiro (eds.), 126-137. Berkeley: Berkeley Linguistics Society. Haser, Verena 2005 Metaphor, Metonymy, and Experientialist Philosophy: Challenging Cognitive Semantics. BerlinINew York: Mouton de Gruyter. Hawkins, Bruce W. 1984 The semantics of English spatial prepositions. PhD thesis, University of California, San Diego. Hecht, Max 1888 Die griechische Bedeutungslehre, eine Aufgabe der klassischen Philologie. Leipzig: Teubner. Heestermans, Hans 1979 Das Woordenboek cler Nederlandsche Taa!. In Praxis der Lexikographie, Helmut Henne (ed.), 68-82. Tiibingen: Max Niemeyer Verlag. Heider, Eleanor R. 1972 Universals in color naming and memory. Journal ofExperimental Psychology 93: 10-20. Heider, Eleanor R., and Olivier, D.e. 1972 The structure of color space in naming and memory for two languages. Cognitive Psychology 3: 337-354.

460

References

Herder, Johann Gottfried 1978 Abhandlung iiber den Ursprung der Sprache. Wolfgang Pross (ed.) . Munchen: Hanser. Original edition 1772. Herringer, Hans Jurgen 1981 Die Unentscheidbarkeit der Ambiguitiit. In Logos S'emantikos, Wolf Dietrich and Horst Geckeler (eds.), Ill: 93-126. Madrid: Editorial Gredos, BerlinINew York: De Gruyter. Hiraga, Masako K. 2003 How metaphor and iconicity are entwined in poetry: A case in Haiku. In Jconicity in Language and Literature IlI. From Sign to Signing, Wolfgang W. G. Muller and alga Fischer 3: 317-335. AmsterdamIPhiladelphia: John Benjamins. Hullen, Wemer, and Rainer Schulze (eds.) 1988 Understanding the Lexicon. Meaning, Sense, and World Knowledge in Lexical Semantics. Tubingen: Max Niemeyer Verlag. Hoffmann, Yael 1983 Irony in the Book of Job. Immanuel 17: 7-21. 1996 A Blemished Perfection: The Book ofJob in Context. Sheffield: Sheffield Academic Press. Hoffmann. Robert R. 1989 Some ambiguities in the study of ambiguity. In Resolving semantic ambiguity, David S. Gorfein (ed.), 204-222. New York: Springer. Hofmann, Thomas R. 1993 Realms of Meaning. An Introduction to Semantics. London: Longman. Holland, Dorothy, and Naomi Quinn (eds.) 1987 Cultural Models in Language and Thought. Cambridge: Cambridge University Press. Horkheimer, Max. and Theodor W. Adorno 1947 Dialektik der Aujklarung. Amsterdam: Querido. Horn, Laurence R. 1985 Metalinguistic negation and pragmatic ambiguity. Language 61: 121-174. Irwin, James R. 1947 Galen on the Temperaments. The Journal of General Psychology 36: 45-64. Itkonen, Esa 1978 Grammatical Theory and Metascience. AmsterdamIPhiladelphia: John Benjamins. Causality in LingUistic Theory. London: Croom Helm. Bloom1983 ington: Indiana University Press.

References 2003

461

What is Language? A Study in the Philosophy of Linguistics. Turku: Abo Akademis tryckeri.

Iwata, Seizi 2004 Over-prefixation: A lexical constructional approach. English Language and Linguistics 8: 239-292. Jackendoff, Ray 1996 Conceptual semantics and cognitive linguistics. Cognitive Linguistics 7: 93-129. Jackson, Stanley W. 1986 Melancholia and Depression. From Hippocratic Times to Modern Times. New Haven: Yale University Press. Jakel,Olaf 1997 European predecessors of a cognitive theory of metaphor. In Human Contact through Language and Linguistics, Birgit Smieja and Meike Tasch (eds.), 69-86. Frankfurt, Peter Lang. Metonymy in onomastics. In Metonymy in Language and 1999 Thought, Klaus-Uwe Panther and Giinter Radden (eds.), 211289. Amsterdam/Philadelphia: John Benjamins. Hypotheses revisited: The cognitive theory of metaphor applied 2002 to religious texts. Me taphorik.de 2: 20-42. Jakobson, Roman 1971 Selected Writings 2: Word and Language. The Hague: Mouton. Joseph, John, and Talbot Taylor (eds.) 1990 Ideologies ofLanguage. London: Routiedge. Joseph, John forthc. Linguistic identities and language shift in modern Europe. In Linguistic Identities, Language Shift and Language Policy in Europe, Bert Cornillie, Jose Lambert and Pierre Swiggers (eds.). Leuven/Paris: Peeters. Kail, Aubrey C. 1986 The Medical Mind of Shakespeare. Balgowla: Williams and Wilkins. Katz, Jerrold J. 1972 Semantic Theory. New York: Harper and Row. Katz, Jerrold J., and Jerry A. Fodor 1963 The structure ofa semantic theory. Language 39, 170-210. Katz, Jerrold J., and Richard Nagel 1974 Meaning postulates and semantic theory. Foundations of Language 11: 311-340. Kempson, Ruth, and Annabel Connack 1981 Ambiguity and quantification. Linguistics and Philosophy 4: 259-309.

462

References

Kempson, Ruth 1977 Semantic Theory, Cambridge, Cambridge University Press. Kerkhofs, Jan, Karel Dobbelaere, Liliane Yoye, and Bernadette Bawin-Legros 1992 De Versnelde Ommekeer: de Waarden van r'1amingen, Walen en Brusselaars in de Jaren '90. Tielt: Lannoo. Kilgarriff, Adam 1997 I don't believe in word senses. Computers and the Humanities 31: 91-113. Kleiber, Georges 1988 Prototype, stereotype: un air de famille? DRLA V Revue de Linguistique 38: 1--61. 1990 La Semantique du Prototype. Categories et Sens Lexical. Paris: Presses Universitaires de France. 1991 Hierarchie lexicale: categorisation verticale et tennes de base. Semiotiques 1: 35-57. Klibansky, Raymond, Erwin Panofsky, and Fritz Saxl 1964 Saturn and Melancholy: Studies in the History of Natural Philosophy, Religion, and Art. London: Nelson. Knox, Israel The traditional roots of Jewish humor. In Holy Laughter, Con1969 rad Hyers (ed.), 150-165. New York: The Seabury Press. Koch, Peter 1999 Frame and contiguity. In Metonymy in Language and Thought, Klaus-Uwe Panther and Giinther Radden (eds.), 139-167. BerIin/New York: Mouton de Gruyter. Kbvecses, Zoltan 1986 Metaphors of Anger, Pride, and Love. Amsterdam/Philadelphia: John Benjamins. 1989 Emotion Concepts. New York: Springer. 1995 Anger: its language, conceptualization, and physiology in the light of cross-cultural evidence. In Language and the Cognitive Construal of the World, John Taylor and Robert E. MacLaury (eds.), 181-196. Berlin/New York: Mouton de Gruyter. A1etaphor and Emotion. Language, Culture and Body in Human 2000 Feeling. Cambridge: Cambridge University Press. Aletaphor: A Practical Introduction. Oxford: Oxford University 2002 Press. Kristiansen, Gitte 2003 How to do things with allophones: linguistic stereotypes as cognitive reference points in social cognition. In Cognitive Models in Language and Thought: Ideologies, Metaphors, and Mean-

References

463

ings, Rene Dirven. Roslyn Frank, and Martin Piitz (eds.), 69120. Berlin/New York: Mouton de Gruyter. Kristiansen, Gitte, and Rene Dirven (eds.) forthc. Cognitive S'ociolinguistics. Berlin/New York: Mouton de Gruyter. Kronasser, Heinz 1952 Handbuch der Semasiologie. Heidelberg: Winter. Kuhn, Thomas 1970 The Structure ofScientific Revolutions. Chicago: The University of Chicago Press. Labov, William 1973 The boundaries of words and their meanings. In New Ways of Analysing Variation in English, Charles 1. Bailey and Roger Shuy (eds.), 340-373. Washington: Georgetown University Press. Lakatos, Imre 1970 Falsification and the methodology of scientific research programmes. In Criticism and the Growth of Knowledge, Imre Lakatos and Alan Musgrave (eds.), 91-196. Cambridge: Cambridge University Press. Lakoff, George 1970a A note on vagueness and ambiguity. Linguistic Inquiry 1: 357359. 1970b Global rules. Language 46: 627-639. 1972 Linguistics and natural logic. Semantics of Natural Language, Donald Davidson and Gilbert Hannan (eds.), 545-665. Dordrecht: Reidel. 1982a Categories and cognitive models. Trier: Linguistic Agency of the University of Trier. Also as: Berkeley Cognitive Science Report no. 2 (Berkeley: Institute for Human Learning). 1982b Experiential factors in linguistics. Language, Mind, and Brain, Thomas W. Simon and Robert 1. Scholes (eds.), 145-156. Hillsdale, NJ: Lawrence Erlbaum. 1983 Elements of the sound symbolism system. Lecture at the Codoc symposium on Cognitive Linguistics. Brussels, 26 March 1983. 1986 Classifiers as a reflection of mind. In Noun Classes and Categorization, Colette Craig (ed.), 13-51. Amsterdam/Philadelphia: John Benjamins. 1987 Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. Chicago: The University of Chicago Press. 1996 Moral Politics. What Conservatives Know that Liberals Don't. Chicago: The University of Chicago Press.

464

References

Lakoff, George, and Mark Johnson 1980 Metaphors we Live by. Chicago: The University of Chicago Press. Lakoff, George, and Zoltan Kovecses 1987 The cognitive model of anger inherent in American English. In Cultural Models in Language and Thought, Dorothy Holland and Naomi Quinn (eds.), 195-221. Cambridge: Cambridge University Press. Landheer. Ronald 1991 Tekstsemantiek en lexicale semantiek. Jaarboek van de Stichting Instituut voor Nederlandse Lexicologie 1990: 53-66. Langacker Ronald W. 1983 Foundations of Cognitive Grammar. I: Orientation, 11: Semantic Structure. Trier: Linguistic Agency of the University of Trier. 1987 Foundations of Cognitive Grammar f. Stanford: Stanford University Press. Concept, Image, and Symbol. The Cognitive Basis of Grammar. 1990a BerlinINew York: Mouton de Gruyter. 1990b Subjectification. Cognitive linguistics 1: 5-38. 1991 Foundations of Cognitive Grammar If. Stanford: Stanford University Press. Lara, Luis Fernando 1989 Dictionnaire de langue, encyclopedie et dictionnaire encyclopedique: le sens de leur distinction. In Worterbucher. Dictionaries. Dictionnaires, Franz Josef Hausmann, Oskar Reichmann, Herbert Ernst Wiegand, Ladislav Zgusta. (eds.), I: 280-287. Berlin, New York: Waiter de Gruyter. Lelunann, Winfred P. (ed.) 1988a Prototypes in Language and Cognition. Ann Arbor: Karoma. Lehmann Winfred P. 1988b Review of R. Langacker, Foundations of Cognitive Grammar. General linguistics 28: 122-30. Lehrer, Adrienne 1983 Wine and Conversation. Bloomington: Indiana University Press. Leys, ado 1988 Some remarks on spatial prepositional structure. In Worlds behind Words. Essays in Honour of F.G. Droste on the Occasion ofhis Sixtieth Birthday, Frans Heyvaert and Frieda Steurs (eds.), 77-84. Leuven: Leuven University Press.

References

465

Lewis, David 1972 General semantics. In Semantics of Natural Language, Donald Davidson and Gilbert Harman (eds.), 169-218. Dordrecht: Reidel. Lindeboom, Gerrit A. 1985 Inleiding tot de Geschiedenis der Geneeskunde. Amsterdam: Rodopi. Lindner, Susan 1. 1981 A lexico-semantic analysis of English verb-particle constructions with up and out. PhD thesis, University of California, San Diego. Lipka, Leonhard 1985 Inferential Features in Historical Semantics. Historical .s'emantics - Historical Word-Formalion, Jacek Fisiak (ed.), 339-354. BerlinlNew York: Mouton de Gruyter. Lounsbury, Floyd 1956 A semantic analysis of Pawnee kinship usage. Language 32: 158-194. Lutzeier, Peter R. 1992 Wortfeldtheorie und kognitive Linguistik. Deutsche .s'prache 20: 62-81. Lyons, John 1963 Structural Semantics. Oxford: Blackwell. 1968 Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. Maalej, Zouhair 2004 Figurative language in anger expressions in Tunisian Arabic: An extended view of embodiment. Metaphor and Symbol 19: 51-75. MacLaury, Robert E. 1991 Social and cognitive motivations of change: measuring variability in color semantics. Language 67: 34-62. Major, Ralph H. 1954 A History ofMedicine. Springfield, Ill.: Thomas. Mangasser-Wahl, Martina 2000 Von der Proto~vpentheorie zur empirischen ,)'emantik. Frankfurt: Peter Lang Verlag. Martin, Laura 1986 Eskimo words for snow. American Anthropologist 88: 418-23. Marty, Anton 1908 Untersuchungen zur Grundlegung der a/lgemeinen Grammatik und .s'prachphilosophie. Halle: Niemeyer.

466

References

McCawley, James D. 1971 Where do NP's come from? In Semantics, Danny D. Steinberg and Leon A. Jakobovits (eds.), 217-231. Cambridge: Cambridge University Press 1981 Everything that Linguists Have Always Wanted to Know about Logic - but IVere Ashamed to Ask. Oxford: Blackwell. Medin, Doug L., and Edward E. Smith 1984 Concepts and concept formation. Annual Review of Psychology 35, 113-138. Me19uk, Igor 1984 Dictionaire Explicat~r et Combinatoire du Franryais Contemporain. Montreal: Les Presses de L'Universite. 1988 Semantic description of lexical units in an explanatory combinatorial dictionary: basic principles and heuristic criteria. International Journal ofLexicography 1, 165-188. 1996 Lexical functions: a tool for the description of lexical relations in a lexicon. In Lexical Functions in Lexicography and Natural Language Processing, Leo Wanner (ed.), 37-102. Amsterdam/Philadelphia: John Benjamins. Melis, Ludo 1990 Pronominal verbs in Old and Modern French, or How prototypes can be restructured on the basis of permanent meaning effects. Belgian Journal ofLinguistics 5: 87-108. Mervis, Carolyn B., and Eleanor Rosch 1981 Categorization of natural objects. Annual Review of Psychology 32: 89-115. Mikolajczuk, Agnieszka 1998 The metonymic and metaphorical conceptualisation of anger in Polish. Speaking of Emotions: Conceptualisation and Expression, Angeliki Athanasiadou and Elzbieta Tabakowska, 153190. BerlinlNew York: Mouton de Gruyter Moon, Rosamund 1998 Fixed Expressions and Idioms in English. A Corpus-based Approach. Oxford: Oxford University Press. Nathan, Geoffrey S. 1986 Phonemes as mental categories. In Proceedings of the Twelfth Annual Meeting of the Berkeley Linguistics Society, Vassiliki Nikiforidou, Mary VanClay, Mary Niepokuj, Deborah Feder (eds.), 212-224. Berkeley: Berkeley Linguistics Society.

References

467

Neisser, Ulrich 1987 Concepts and Conceptual Development. Ecological and Intellectual Factors in Categorization. Cambridge: Cambridge University Press. Nerlich, Brigitte 1990 Change in Language: Whitney, Breal and Wegener. London: Routledge. 1992 Semantic Theories in Europe 1830-1930. From Etymology to Contextuality. AmsterdamIPhiladelphia: John Benjamins. Newmeyer, Frederick J. 1980 Linguistic Theory in America. New York: Academic Press. Norrick, Neal R. 1981 Semiotic Principles in Semantic Theory. AmsterdamIPhiladelphia: John Benjamins. Nunberg, Geoffrey 1979 The non-uniqueness of semantic solutions: polysemy. Linguistics and Philosophy 2: 143-184. Nyrop, Kristoffer 1913 Grammaire Historique de la Langue Franr;aise IV. Semantique. Kopenhagen: Gyldendalske Boghandel. Ortony, Andrew 1988 Are emotion metaphors conceptual or lexical? Cognition and Emotion 2: 95-103. bstman, Jan-Ola, and Mirjam Fried (eds.) 2005 Construction Grammars: Cognitive Grounding and Theoretical Extensions. AmsterdamIPhiladelphia: John Benjamins. Pagels, Elaine 1996 The Origin ofSatan. Harmondsworth: Penguin. Palmer, Gary 1996 Toward a Theory of Cultural Linguistics. Austin: University of Texas Press. Panther, K1aus-Uwe, and Giinther Radden 1999 Metonymy in Language and Thought. Berlin/New York: Mouton de Gruyter. Paul, Hermann 1880 Prinzipien der Sprachgeschichte. Halle: Niemeyer. Peeters, Bert (ed.) 2000 The Lexicon-Encyclopedia Interface. Oxford: Elsevier Science. Pluim, Teunis 1894 Keur van Nederlandsche synoniemen. Purmerend: Muusses.

468

References

Pope, Maurice 1985 Shakespeare's Medical Imagination. In :;;hakespeare Survey, Stanley Wells (ed.), 38: 175-186. Cambridge: Cambridge University Press. Popova, Yana 2003 'The fool sees with his nose': Metaphoric mappings in the sense of smell in Patrick Siiskind's Perfume. Language and Literature 12: 136-151. Porzig, Waiter 1934 Wesenhafte Bedeutungsbeziehungen. Beitrage zur Geschichte der Deutschen Sprache und Literatur 58: 70-97. Posner, Michael 1986 Empirical studies of prototypes. Noun classes and categorization, Colette Craig (ed.), 53-61. Amsterdam/Philadelphia: John Benjamins. Post, Michael 1986 A prototype approach to denominal adjectives. In Linguistics across Historical and Cleographical Boundaries, Dieter Kastovsky and Aleksander Szwedek (eds.), 11: 1003-1013. BerlinlNew York: Mouton De Gruyter. Pottier, Bernard 1963 Recherches sur I'analyse semantique en linguistique et en traduction mecanique. Nancy: Faculte de Lettres et Sciences Humaines. 1964 Vers une semantique moderne. Travaux de Linguistique et de Litterature 2: 107-137. Pulman, Steve 1983 Word A1eaning and Belief London: Croom Helm. Pustejovsky, James 1995 The Generative Lexicon: A Theory of Computational Lexical semantics. Cambridge, Mass.: MIT Press. Putnam, Hilary 1975 The meaning of Meaning. In Mind, Language, and Reality. Philosophical Papers n, Hilary Putnam, 215-271. Cambridge: Cambridge University Press. Quadri, Bruno 1952 Aufgaben und Methoden der Onomasiologischen Forschung. Bern: Francke. Quine, Willard v.o. 1960 Word and Object. Cambridge, Mass.: MIT Press.

References

469

Ravin, Yael, and Claudia Leacock (eds.) 2000 Polysemy. Theoretical and Computational Approaches. Oxford: Oxford University Press. Reddy, Michael 1979 The conduit metaphor - a case of frame conflict in our language about language'. In Metaphor and thought, Andrew Ortony (ed.), 284-384. Cambridge: Cambridge University Press. Regier, Terry 1995 A model of the human capacity for categorizing spatial relations. Cognitive Linguistics 6: 63-88. Reichling. Anton 1935 Het Woord. Zwolle: Tjeenk Willink. Ricoeur, Paul 1981 Hermeneutics and the Human Sciences. Edited, translated and introduced by John B. Thompson. Cambridge: Cambridge University Press, Paris: Editions de la Maison des Sciences de I'Homme. Robertson, David 1973 The book of Job: A literary study. Soundings 56: 446-469. 1977 The comedy of Job: A response. Semeia 7: 41-44. Rosch, Eleanor 1973 On the internal structure of perceptual and semantic categories. In Cognitive Development and the Acquisition of Language, Timothy E. Moore (ed.), 111-144. New York: Academic Press. 1975 Cognitive representations of semantic categories. Journal of Experimental Psychology: General 104: 192-233. 1977 Human categorization. In Studies in Cross-cultural Psychology J, Warren, Neil (ed.), 1-49. New York: Academic Press. 1978 Principles of categorization. In Cognition and Categorization, Eleanor Rosch and Barbara B. Lloyd (eds.), 27-48. Hillsdale, NJ: Lawrence Erlbaum. 1988 Coherences and categorization: a historical view. In The Development of Language and Language Researchers. Essays in Honor of Roger Brown, Frank Kessel (ed.). 373-392. Hillsdale, NI: Lawrence Erlbaum. Rosch, Eleanor, and Carolyn B. Mervis 1975 Family resemblances: Studies in the internal structure of categories. Cognitive Psychology 7: 573--605. Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray. David Johnson, and Penny Boyes-Braem 1976 Basic objects in natural categories. Cognitive P.~vchology 8: 382-439.

470

References

Rosch, Eleanor, Carol Simpson, and Scott R. Miller 1976 Structural bases of typicality effects. Journal of Experimental Psychology: Human Perception and Peformance 2: 491-502. Rosier, Irene (ed.) 1988 L'Ambiguite. Cinq Etudes Historiques. Lille: Presses Universitaires de Lille. Ross, John R. 1987 Islands and syntactic prototypes. In Papers from the 23rd Annual Jvfeeting of the Chicago Linguistic Society, Barbara Need, Eric Schiller, and Anna Bosch (eds.), 309-320. Chicago: Chicago Linguistic Society. Rudzka-Ostyn, Brygida 1985 Metaphoric processes in word formation. The case of prefixed verbs. The Ubiquity of Metaphor, Wolf Paprotte and Rene Dirven (eds.), 209-241. AmsterdamJPhiladelphia: John Benjamins. 1988 Semantic extensions into the domain of verbal communication. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 507-553. AmsterdamJPhiladelphia: John Benjamins. 1989 Prototypes, schemas, and cross-category correspondences, the case of ask. Linguistics 27: 613-662. 1996 The Polish dative. In The Dative 1. Descriptive Approaches, Willy Van Langendonck and William Van Belle (eds.), 341-394. AmsterdamJPhiladelphia: John Benjamins. 2003 Word Power: Phrasal Verbs and Compounds. A Cognitive Approach. BerlinlNew York: Mouton de Gruyter. Ruhl, Charles 1989 On Monosemy. A Study in LingUistic Semantics. Albany: State University of New York Press. Ryder, Mary E. 1994 Ordered Chaos: The Interpretation ofEnglish Noun-Noun Compounds. Berkeley: University of California Press. Sandra, Dominiek 1998 What linguists can and can't tell you about the human mind: A reply to Croft. Cognitive Linguistics 9: 361-378. Sandra, Dominiek, and Sally Rice 1995 Network analyses of prepositional meaning: Mirroring whose mind - the linguist's or the language user's? Cognitive Linguistics 6: 89-130.

References

471

Sapir, Edward 1907 Herder's 'Urspmng der Sprache'. Modern Philology 5: 109142. Reprinted in Historiographia Linguistica 11: 355-388 [1984]. Schafer, Jiirgen 1966 Wort und Begriff Humour in der Elisabethanischen KomOdie. Miinster: Aschendorff Verlag. Schiffman, Harold 1996 Linguistic Culture and Language Policy. London: Routledge. Schmid, Hans-Jorg 1993 Cottage und Co., idea, start vs. begin. Die Kategorisierung als Grundprinzip einer diJferenzierten Bedeutungsbeschreibung, Tiibingen: Max Niemeyer Verlag. Schmidt, Wilhelm 1963 Lexikalische und Aktuelle Bedeutung. Berlin: Akademie-Verlag. Schulze, Rainer 1988 A short story of down. In Understanding the Lexicon. Meaning, Sense, and World Knowledge in Lexical Semantics, Werner Hiillen and Rainer Schulze (eds.), 395-414, Tiibingen: Max Niemeyer Verlag. 1991 Getting round to (a)round, towards the description and analysis of a 'spatial' predicate. In Approaches to Prepositions, Gisa Rauh (ed.), 253-274. Tiibingen: Gunter Nan Verlag. Seiler, Hansjakob 1986 Apprehension: Language, Object, and Order. Tiibingen: Gunter Nan Verlag. Shibatani, Masayoshi, and Sandra A. Thompson (eds.) 1996 Grammatical Constructions: Their Form and Meaning. Oxford: Oxford University Press. Siegel, Rudolph E. 1968 Galen's System ofPhysiology and Medicine. Basel: Karger. Sinclair, John 1987 Looking up. London and Glasgow: Collins. Sinha, Chris, and Kristina Jensen de Lopez 2000 Language, culture and the embodiment of spatial cognition. Cognitive Linguistics 11: 17-41. Skutnabb--Kangas, Tove 2000 Linguistic Genocide in Education ~ or lVorldwide Diversity and Human Rights? Mahwah, NI: Lawrence Erlbaum. Smith, Edward E., and Doug Medin 1981 Categories and Concepts. Cambridge, Mass.: Harvard University Press.

472

References

Smith, Edward E., Edward J. Shoben, and Lance 1. Rips 1974 Structure and process in semantic memory: a featural model for semantic decisions. Psychological Review 81. 214-241. Soriano, Cristina 2003 Some anger metaphors in Spanish and English. A contrastive review. International Journal ofEnglish Studies 3: 107-122. Soufas, Teresa Scott 1990 Melancholy and the Secular Mind in Spanish Golden Age Literature. Columbia: University of Missouri Press. Speelman, Dirk, Stefan Grondelaers, and Dirk Geeraerts 2003 Profile-based linguistic uniformity as a generic method for comparing language varieties. Computers and the Humanities 37: 317-337. Sperber, Hans 1923 Einfuhrung in die Bedeutungslehre. Bonn: Schroeder. Starobinsky, Jean 1962 History ofthe Treatment ofMelancholy from the Earliest Times to 1900. Basel: Geigy. Steen, Gerard 1. 1994 Understanding Metaphor in Literature: An Empirical study. London, Longman. Stefanowitsch, Anatol, and Stefan Th. Gries 2003 Collostructions: Investigating the interaction between words and constructions. International Journal of Corpus linguistics 8: 209-243. Stern, Gustav 1931 Meaning and Change ofMeaning. G6teborg: Elanders. Stevenson, Mark 2001 Word Sense Disambiguation. Combining Knowledge Sources for 5,'ense Resolution. Stanford: CSLI Publications. Stock, Penelope F. 1983 Polysemy. In LEXeter '83. Proceedings of the Exeter lexicography conference, Reinhard RK. Hartmann (ed.), 131-140. Tilbingen: Max Niemeyer Verlag. Swanepoel, Piet 1992 Linguistic motivation and its lexicographical application. South African Journal ofLinguistics 10: 49--60. 1998 Back to basics: prepositions, schema theory, and the explanatory function of the dictionary. In Euralex '98 Proceedings, Thierry Fontenelle, Philippe Hiligsmann, Archibald Michiels, Andre Moulin, Siegfried Theissen (eds.), 655--666. Liege: Universite de Liege.

References

473

Sweetser, Eve E. 1984 Semantic structure and semantic change: English perception verbs in an Indo-European Context. Trier: Linguistic Agency of the University of Trier. 1987 The definition of lie: an examination of folk models underlying a semantic prototype'. In Cultural models in language and thought, Dorothy Holland and Naomi Quinn (eds.), 43-66. Cambridge: Cambridge University Press 1999 Compositionality and blending: Semantic composition in a cognitively realistic framework. In Cognitive Linguistics: Foundations, Scope, and Alethodology. Theo Janssen, Gisela Redeker (eds.), 129-162. BerlinlNew York: Mouton de Gruyter. Swiggers, Pierre 1988 Grammatical categories and human conceptualization: Aristotle and the modistae. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 621-646. Amsterdam/ Philadelphia: John Benjamins. Tabakowska, Elzbieta 2000 Grammar and the cult of the virgin. A case study of Polish religious discourse. In Evidence for Linguistic Relativity, Susanne Niemeier and Rene Dirven (eds.), 223-233. AmsterdamJPhiladelphia: John Benjamins. Talmy, Leonard 1978 Figure and ground in complex sentences. In Universals of Human Language 4. Syntax, Joseph H. Greenberg (ed.), 625-49. Stanford: Stanford University Press. 1983 How language structures space. In Spatial Orientation: Theory, Research, and Application, Herbert Pick and Linda Acredolo (eds.), 225-282. New York: Plenum Press. 1985 Force dynamics in language and thought. In Papers .from the Parasession on Causatives and Agentivity at the Twenty-First Regionallv/eeting of the Chicago Linguistic Socie(v, William H. Eilfort, Paul D. Kroeber, and Karen L. Peterson (eds.), 293-337. Chicago: Chicago Linguistic Society. Taylor, John 1988 Contrasting prepositional categories: English and Italian. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 299-326. Amsterdam/Philadelphia: John Benjamins Publishing Company. 1989 Linguistic Categorization. Prototypes in linguistic Theory. Oxford: Clarendon Press.

474

References

How many meanings does a word have? S'tellenbosch Papers in Linguistics 25: 133-168. 1995 Linguistic Categorization. Proto~ypes in Linguistic Theory. 2nd ed. Oxford: Oxford University Press. 1996 On running and jogging. Cognitive Linguistics 7: 21-34. 1999 Cognitive semantics and structural semantics. In Historical Semantics and Cognition, Andreas Blank and Peter Koch (eds.), 17-48. Berlin/New York: Mouton de Gruyter. 2003 Polysemy's paradoxes. Language Sciences 25: 637-655. Taylor, John, and Thandi G. Mbense 1998 Red dogs and rotten mealies: How Zulus talk about anger. Speaking of Emotions: Conceptualisation and Expression. Angeliki Athanasiadou and Elzbieta Tabakowska (eds.), 191 -226. Berlin/New York: Mouton de Gruyter. Thomason, Richmond (ed.) 1974 Formal Philosophy. Selected Papers ofRichard Montague. New Haven: Yale University Press. Tomasello, Michael 2003 Constructing a Language. A Usage-based Theory of Language Acquisition. Harvard: Harvard University Press. Traugott, Elisabeth C. 1989 On the rise of epistemic meanings in English: an example of subjectification in semantic change. Language 65: 31-55. 1985 'Conventional' and 'dead' metaphors revisited. In The UbiqUity of Metaphor, Wolf Paprotte and Rene Dirven (eds.), 17-56. Amsterdam/Philadelphia: John Benjamins. Trier, Jost 1931 Der Deutsche Wortschatz im Sinnbezirk des Verstandes. Die Geschichte eines Sprachlichen Fe/des. Heidelberg: Winter. Tsohatzidis, Savas L. (ed.) 1989 Meanings and Prototypes. S'tudies on Linguistic Categorization. London: Routledge and Kegan Paul. Tuggy, David 1993 Ambiguity, polysemy, and vagueness. Cognitive Linguistics 4: 273-290. 1999 Linguistic evidence for polysemy in the mind: A response to William Croft and Dominiek Sandra. Cognitive Linguistics 10: 343-368. Tummers, Jose, Kris Heylen, and Dirk Geeraerts 2005 Usage-based approaches in Cognitive Linguistics: A technical state of the art. Corpus Linguistics and Linguistic Theory 1: 225261. 1992

References

475

Tummers, Jose, Dirk Speelman, and Dirk Geeraerts QuantifYing semantic effects. The impact of lexical collocations 2004 on the inflectional variation of Dutch attributive adjectives. In Le poids des Mots. Actes des tes Jourm'!es Internationales d'Analyse Statistique des Donm'!es Textuelles, Gerald Purnelle, Cedrick Fairon, and Anne Dister (eds.), 1079-1088. Louvain-laNeuve: Presses Universitaires de Louvain. Turner, Mark 1996 The Literary Mind: The Origins of Thought and Language. Oxford: Oxford University Press. Tyler, Andrea, and Vyvian Evans 2003 Reconsidering prepositional polysemy networks: The case of 'over'. In Polysen~v: Flexible Patterns of Meaning in Alind, Brigitte Nerlich, Zazie Todd, Vimala Hennan, and David D. Clarke (eds.), 96-160. BerlinINew York: Mouton de Gruyter. UWenbeck, Eugenius M. 1967 The dynamic nature of word meaning. Actes du Xe Congres International des Linguistes It 679-685. Bucarest: Editions de l'Academie. Ungerer. Friedrich, and Hans-Jorg Schmid 1996 An Introduction to Cognitive Linguistics. London: Longman. Van Belle. William, and Willy Van Langendonck 1992 The indirect object in Dutch. Leuvense Bijdragen 81: 17-43. Vandeloise, Claude 1985a La preposition dans et la relation contenantlcontenu. Leuvense Bijdragen 74: 193-208. Van der Meer. Geart 2000 Core. subsense, and the New Oxford Dictionary of English. In Euralex 2000 Proceedings, Ulrich Heid, Stefan Evert. Egbert Lelllnann, Christian Rohrer (eds.), 419-431. Stuttgart: Universitat Stuttgart. Van Dijk, Teun A. 1998 Ideology: A Multidisciplinary Approach. London: Sage. Van Gijsel, Sofie. Dirk Geeraerts, and Dirk Speelman 2004 A functional analysis of the linguistic variation in Flemish spoken commercials. In Le Poids des Mots. Actes des tes Jourm'!es Internationales d'Analyse Statistique des Donnees Textuelles, Gerald Purnelle, Cedrick Fairon, and Anne Dister (eds.), 11361143. Louvain-la-Neuve: Presses Universitaires de Louvain. Van Ginneken, Jac 1912 Het gevoel in taal- en woordkunst. Leuvense Bijdragen 9: 265356,10: 1-156, 173-273.

476

References

Van Hecke, Pierre 2001 Polysemy or homonymy in the root(s) r'h in Biblical Hebrew. A cognitive-linguistic approach. ZeitschriJt fur Althebraistik 14: 50-66. Van Langendonck, Willy 1986 Markedness. prototypes, and language acquisition. Cahiers de l'lnstitut de Linguistique de Louvain 12 : 39-76. Van Oosten, Jeanne 1986 The nature of subjects. topics, and agents: a cognitive explanation. Bloomington: Indiana University Linguistics Club. Vandeloise. Claude 1990 Representation, prototypes, and centrality. In Meanings and Prototypes. Studies in Linguistic Categorization, Savas L. Tsohatzidis (eds.), 403-437. London, New York: Routledge. 1994 Methodology and analyses of the preposition in. Cognitive Linguistics 5: 157-84. Verdoodt, Albert (ed.) 1978 Belgium. (International Journal of the SOCiology of Language 15). The Hague: Mouton. Verlooy Jan B. C. 1979 Verhandeling op d'Onacht der Moederlyke Tael in de Nederlanden. Jozef Smeyers and Jan Van den Broeck (eds.). Den Haag: Martinus Nijhoff. Original edition 1788. Violi, Patrizia 1997 Significato ed Esperienza. Milano: Bompiani. Wagner, Doris 2002 Papier-Bier. Adjektive zur Beschreibung des Bieres in der deutschen Bierplakat-Werbung. SKY Journal ofLinguistics 15: 251282. 2003 Bier gut, Abend gut! Neuphilologische Mitteilungen 104: 133147. Waismann, Friedrich 1952 Verifiability. In Logic and Language (First ,c.,'eries), Anthony Flew (ed.), 117-144. Oxford: Basil Blackwell. Weiland, Petrus. and George Landre 1825 Woordenboek der Nederduitsche synoniemen. 's-Gravenhage: Allart. Weinreich. Uriel 1966 Explorations in semantic theory. Current trends in lingUistics Ill: 395-479.

References

477

Weisgerber, Leo 1927 Die Bedeutungslehre: ein Irrweg der Sprachwissenschaft? Germanisch-Romanische Monatsschrift 15: 161-183. 1954 Die Sprachfelder in der geistigen Erschliessung der Welt. In Festschrift lost Trier zu seinem 60. Geburtstag, Benno von Wiese and Karl Heinz Borck (eds.), 34-49. Meisenheim: Hain. Wellander, Erich 1928 Studien zum Bedeutungswandel im Deutschen. Uppsala: Berlings Boktryckeri. Whedbee, William 1977 The Comedy of Job. S'emeia 7: 1-39. Wierzbicka, Anna 1985 Lexicography and Conceptual Analysis. Ann Arbor: Karoma. English speech act verbs: a semantic dictionary. Sydney: Aca1987a demic Press. 1987b Predict - Prophesy - Forecast. Semantics and lexicography. In Aspects of Language. Studies in Honour of Mario Alinei 11, Roberto Crespo, Bill Dotson Smith, and Henk Schultink (eds.),. Amsterdam: Rodopi. 1989 Prototypes in semantics and pragmatics, explicating attitudinal meanings in terms of prototypes. Linguistics 27: 731-769. 1990 Prototypes save: on the uses and abuses of the notion 'prototype' in linguistics and related fields. In Meanings and Prototypes, Savas L. Tsohatzidis (ed.), 347-367. London: Routledge. 1991 Semantic rules know no exceptions. Studies in Language 15: 371-398. 1992 Semantics, Culture, and Cognition. Universal Human Concepts in Culture-Specific Configurations. Oxford: Oxford University Press. 1996 Semantics: Primes and Universals. Oxford: Oxford University Press. Winters, Margaret 1987 Syntactic and semantic space: the development of the French subjunctive. In Papers from the Seventh International Conference on Historical Linguistics, Anna G. Ramat, Onofrio Carruba, and Giulano Bernini (eds.), 607-618. AmsterdamJPhiladelphia: John Benjamins. 1989 Diachronic prototype theory: on the evolution of the French subjunctive. Linguistics 27: 703-730. Wittgenstein, Ludwig 1953 Philosophische Untersuchungen. Philosophical Investigations. Oxford: Basil Blackwell.

478

References

Wodak, Ruth, and Michael Meyer (eds.) 2001 Methods of Critical Discourse Analysis. London: Sage. Woolard, Kathryn, Bambi Schieffelin, and Paul Kroskrity 1998 Language Ideologies: Practice and Theory. Oxford: Oxford University Press. Wundt, Wilhelm 1900 Volkerpsychologie J: Die Sprache. Leipzig: Engelmann. Yu, Ning 1998 The Contemporary Theory of Metaphor: A Perspective from Chinese. AmsterdamlPhiladelphia: John Benjamins. Zgusta, Ladislav 1971 Manual ofLexicography. The Hague: Mouton. Zlatev. Jordan 2003 Polysemy or generality? Mu. In Cognitive Approaches to Lexical Semantics, Hubert Cuyckens, Rem: Dirven, and John Taylor 447-494. Berlin/New York: Mouton de Gruyter. Zwicky, Arnold, and Jerry Sadock 1975. Ambiguity tests and how to fail them. In Syntax and Semantics .t. John Kimball (ed.), 1-36. New York: Academic Press.

Index of names

Adorno, Theodor W., 280 Agirre, Eneko, 100 Aijmer, Karen, 8 Anttila, Raimo, 140 Apresjan, Juri, 386 Aristotle, 24, 106, 112, 113, 145, 232, 339, 350, 374 Armstrong, Sharon, 13, 16,41 Atkins, Beryl T.S., 362, 405 Ayto, John, 359 Bakema, Peter, 27, 74, 83, 85, 87, 149,151,194,196,198,199, 208,223,262,273,274,363, 410,416,442 Baldinger, Kurt, 78 Ballmer, Thomas, 142 Barcelona, Antonio, 227 Barere, Bertrand, 276, 288 Barlow, Michael, 363 Barnes, Jonathan, 112 Bartsch, Renate, 21 Bawin-Legros, Bernadette, 269 Beek, Henri, 231 Berlin, Brent, 13,84,258,261,406, 410 Biebuyck, Benjamin, 307 Bierwisch, Manfred, 391 Bleicher, Joser. 140 Blommaert, Jan, 274 Boas, Franz, 95 Boeve, Lieven, 307 Bosch, Peter, 123 Brandt, Per A., 307 Breal, Michel, 368, 369, 371. 372, 373,374,396,400 Bmgman, Claudia, 48, 49, 50, 133, 178,352,374 Bmner, Jerome, 43

Brysbaert, Marc, 272 Bulcaen, Chris, 274 Burke, Lucy, 272 Bybee, Joan, 7 Campbell, Lily B., 235 Carnoy, Albert, 368,400 Catlin, Jack, 121 Catlin, Jane-Carol, 121 Cervantes, Miguel de, 239 Chapman, Allan, 236 Chierchia, Gennaro, 221 Chomsky, Noam, 8, 392 Clark, Eve V., 95 Clark, Herbert H., 95 Coleman, Linda, 9, 133, 374 Colleman, Timothy, 175 Condillac, Etienne Bonnot de, 285, 286,287 Copi, Irving M., 113, 130 Cormack, Annabel, Ill, 125 Coseriu, Eugenio, 77, 386, 401 Coulson, Seana, 198 Craig, Colette, 4 Croft William, 99, 175 Crowley, Tony, 272 Cmse, D. Allan, 51,73,101, 105, 107,110,111,122,143,144, 145,146, 147,351,401 Cmttwell, Patrick, 235 Cuyckens, Hubert, 48, 49, 50, 67, 68,69,70,71,178 Danaher, David S., 307 Dannesteter, Arsene, 368, 373, 400 De Beer, Taco, 38 De Flines, Quirijn, 39 De Sutter, Gert, 272 Deane, Paul D., 48, 115, 121, 122, 123, 124, 134, 178

480

Index a/names

Den Boon, Ton, 327 Deprez, Kas, 289 Dewell, Robert B., 48 Diepgen, Paul, 231 Dilthey, Wilhelm, 140, 373, 435, 436,437,438,440,443 Dirven, Rene, 24, 152, 198, 228, 272,273,307,347,374 Dobbelaere, Karel, 269 Dowty, David, 23,386 Draper, John W., 235 DressIer, Wolfgang, 24 Dubois, Jean, 386 Dunbar, George, 99 Edmonds, Philip, 100 Einstein, Albert, 419, 420 Ekman, Paul, 242 Elffers, EIs, 367 Erdmann, Karl Otto, 381 Evans, Vyvian, 48 Feyaerts, Kurt, 198, 212, 307 Feyerabend, Paul, 421 Fillmore, Charles J., 4, 24, 130, 177, 198,362,374,405 Firth, John R., 23, 402 Fodor, Janer, 386 Fodor, Jerry A., 4, 75, 386, 390, 401,402,403 Forceville, Charles, 227 Frank, Roslyn, 272, 273 Frawley, William, 356 Freeman, Margaret H., 307 Fried, MiIjam, 175 Friedman, Hershey, 310 Friesen, Wallace v., 242 Gadamer, Hans-Georg, 140,438 Galen, 232, 235 Garrett, Merrill, 386 Geckeler, Horst, 401 Geeraerts, Dirk, 8, 23, 25, 27, 30, 40,47,50, 71, 72, 74, 80, 81, 83, 85,87,88,90,101, 103, 126,

135, 140, 149, 150, 151, 153, 169,172,178,184,193,194, 196, 198, 199,208,218,223, 227,228,248,251,258,262, 272,273,274,299,310,327, 328,333,334,345,346,351, 357, 361, 363, 367, 374, 375, 377,385,397,399,405,410, 416,442 Gevaert, Caroline, 212, 227 Gibbs, Raymond W., 198,210,248, 416 Girvin, Allan, 272 Givon, Tahny, 24 Glass, Arnold, 392 Gleitman, Henry, 13, 16,41 Gleitman, Lila, 13, 16,41 Goddard, Cliff, 405 Godderis, Jan, 231 Goldberg, Adele, 175, 177, 178, 194, 195, 196 Good, Edwin M., 310 Goodenough, Ward H., 386, 401 Goossens, Louis, 218, 219, 220 Gordon, Terrence W., 385, 398 Gregoire, Henri-Baptiste, 278, 280, 281, 288 Grice, Paul, 150 Gries, Stephan Th., 345 Grondelaers, Stefan, 27, 74, 83, 85, 87, 149, 151, 194, 196,227,228, 262,272,273,274,299,363, 410,416,442 Habennas,Jurgen, 140,276 Hailnan,John, 23, 221, 356, 379 Halliday, Michael A.K., 402 Hamilton, Craig, 307 Hanks, Patrick, 359, 362 Harder, Peter, 227 Harkins, Jean, 227 Harvey, William, 234 Hasan, Ruqaiya, 402

Index ofnames Haser, Verena, 367 Hawkins, Bruce, 273, 374 Hecht, Max, 378, 379, 397 Heesterrnans, Hans, 332 Heider, Eleanor R, 4, 13 Herder, Johann Gottfried, 282, 285, 286,287 Herringer, Hans-Jiirgen, 107, 110 Heylen, Kris, 345 Hippocrates, 228, 229, 231 Hiraga, Masako K., 307 Hoffmann, Robert R, 137 Hoffmann, Yael, 310, 313, 314, 315, 322 Hofmann, Thomas R, 90 Holland, Dorothy, 4, 273 Holyoak, Keith, 392 Horkheimer, Max, 280 Horn, Laurence R., 134 Hiillen, Werner, 4 Husserl, Edmund, 436 Hie, Cornelia, 273 Irwin, James R, 231 Itkonen, Esa, 140,227 Jackendoff, Ray, 404, 408 Jackson, Stanley W., 239, 254 Jake!. Olaf, 252, 307, 367 Jakobson, Roman, 220 Janssens, Guy, 327 Jensen de L6pez. Kristina, 227 Johnson,Mark, 205.248, 258, 309, 352,374,378,405,413 Joseph,John, 274,284 Kail, Aubrey, 235 Katz, Jerrold 1.,4, 7, 75, 386, 390. 391,401,402,403,404,406,413 Kay, Paul, 9,13.133,177,198,374, 410 Kemmer, Suzanne, 363 Kempson, Ruth, Ill, 125, 150 Kerkhofs, Jan, 269 Kilgarriff, Adam, 359

481

Kleiber, Georges, 25, 80, 100, 102, 104, 136, 141,262 Klibansky, Raymond, 231, 235, 239 Knox, Israel, 3 10 Koch, Peter, 413 Kovecses, Zoltan, 227. 230, 240, 241,242,243,244.245,246, 247,248,249,250,251 Kristiansen, Gitte, 228 Kronasser, Heinz, 381 Kroskrity, Paul, 274 Kuhn, Thomas, 44, 140, 375, 421 Labov, Williall1, 19,27,376 Lakatos, Ill1re, 140, 375, 421 Lakoff, George, 4, 10, 13, 14, 17, 18,23,24,25,26,42,49,50,74, 81, 101, 106, 121, 130, 133, 138, 178,191,205,221,240,248, 258,273,274,309,333,334, 352,374,378,380,381,391, 395,405,413,421 Landheer, Ronald, 150 Landre, George, 38 Langacker, Ronald W., 4,8,46,51, 74.76.80, 135, 152, 176, 188, 262,309,315,374,379,380, 391.395,405 Lara, Luis Fernando, 356 Leacock, Claudia, 99 Lehmann, Winfred P., 4, 23 Lehrer, Adrienne, 255 Levenson, Robert W., 242 Lewis, David, 392 Leys, Odo, 72, 73 Lindeboom, Gerrit A., 231 Lindner, Susan 1., 374 Lipka, Leonhard, 388 Lounsbury, Floyd, 386,401 Lutzeier, Peter R., 367 Lyons, John, 387, 390,401.403 Maalej, Zouhair, 227 Machado de Assis, Joaquim, 417

482

Index ofnames

MacLaury, Robert E., 138,227 Major, Ralph H., 231 Mangasser-Wahl, Martina, 349, 405 Martin, Laura, 95, 272, 438 Marty, Anton, 380 Mbense, Thandi G., 227 McCawley, James D., 145,391 McConnell-Ginet, Sally, 221 Medin, Doug L., 4, 392 Melc;;uk, Igor, 401 Melis, Ludo, 135 Mervis, Carolyn B., 3, 4,9, 10, 11, 28, 116, 374, 376 Meyer, Michael, 274 Michelson, Albert, 419, 420, 432, 442 Mikolajczuk, Agnieszka, 227 Miller, Scott R, 4 Minsky, Marvin, 43 Moder, Carol L 7 Montague, David, 386, 392 Moon, Rosamund, 363, 402 Morley, Edward, 419, 420, 432, 442 Nagel, Richard, 386 Nathan, Geoffrey S., 7 Neisser. Ulrich, 4 Nerlich, Brigitte, 99, 367, 398, 400 Newmeyer, Frederick 1.,391 Norrick, Neal R, 105, 109, 122 Nunberg, Geoffrey, 115, 121, 124, 216 Nyrop, Kristoffer, 368, 400 Olivier, D.e., 13 Ortony, Andrew, 249 bstman, Jan-Ola, 175 Pagels, Elaine, 322 Palmer, Gary, 227, 273, 347 Panofsky, Erwin, 231, 235. 239 Panther, Klaus-Uwe, 405 Paul, Hermann, 368, 381, 400 Peeters, Bert, 99, 356 Piaget, Jean, 43

Pinkal, Manfred, 142 Plato, 22,25,416, 417,430, 439 Pluim, Teunis, 38 Pope, Maurice, 235, 236, 245 Popova, Yana, 307 Porzig, Waiter, 388, 390. 401 Posner, Michael, 9 Pottier, Bernard, 386. 390, 401 Pulman, Steve, 41 Pustejovsky, James, 404, 406, 408, 413 Putnam, Hilary, 20, 21. 22, 25, 357, 421,426,427,443 Quadri, Bruno, 78 Quine, Willard V.O., 14, 106, 108, 350,428 Quinn, Naomi, 4, 273 Radden, Giinther, 405 Ravin, Yael, 99 Reddy, Michael, 137, 183, 273 Regier, Terry, 416 Reichling, Anton, 23, 397 Rice, Sally, 99, 416 Ricoeur, Paul, 140 Ries, John, 307 Ripa, Cesare, 234, 235, 243 Rips, Lance 1., 392 Robertson, David, 310, 313, 314, 315, 322 Rosch, Eleanor, 3, 4, 9, 10, 11, 13, 16, 18,22,27,28,29,42, 102, 116, 117,374,376,379,405 Rosier, Irene, 144 Ross, John R, 7 Rudzka-Ostyn, Brygida, 4, 27, 45, 152, 178, 193, 194, 195, 196, 198, 367, 374 RuhL Charles, 150, 152, 168, 169, 171 Ryder, Mary E., 198 Sadock, Jerry, 14, 101, 106,337, 350

Index a/names Sandikcioglu, Esra, 273 Sandra, Dominiek, 99,416 Sapir, Edward, 287 Saxl, Fritz, 231, 235, 239 Schafer,Jtirgen, 231, 233, 235, 236 Schieffelin, Bambi, 274 Schiffman, Harold, 274 Schleiennacher, Friedrich, 140 Schmid, Hans-Jarg, 152,309,347. 405 Schmidt, Wilhelm, 146 Schulze. Rainer, 4, 152 Seiler, Hansjakob, 24 Shakespeare, William, 235, 236 Shoben, Edward 1.,392 Siegel, Rudolp Eo, 230, 231 Simpson, Carol, 4 Sinclair, John, 402 Sinha, Chris, 150, 227 Skutnabb-Kangas. Tove, 283, 284, 294 Smith, Edward Eo, 4, 291, 392 Soriano, Cristina, 227 Soufas, Teresa Scott, 239 Speelman, Dirk, 272, 273, 274, 299 Sperber. Hans, 368, 372 Starobinsky, Jean, 239 Stefanowitsch, Anatol. 345 Stem, Gustav, 368. 400 Stevenson, Mark. 100 Stock, Penelope, 359 Swanepoel, Piet, 362 Sweetser, Eve Eo, 133, 198,374,383 Swiggers, Pierre, 367, 396 Tabakowska, Elzbieta, 307 Talmy, Leonard, 76, 374, 395 Taylor, John, 4, 48, 74, 80, 89, 99, 101,227,309,347,367,405,408 Taylor, Talbot, 274 Thomason, Richmond, 386, 403 Tissot, Samuel, 228, 229 Tomasello, Michael, 227

483

Traugott, Elisabeth c., 188,248 Trier, Jost, 379, 386, 400, 401 Tuggy, David, 99, 145,351 Tummers, Jose, 272, 345 Turner, Mark, 307 Tyler, Andrea, 48 Uhlenbeck, Eugenius Mo, 7 Ungerer, Friedrich, 309, 347, 405 Van Belle, William, 175, 177 Van der Meer, Geart, 362 Van Dijk, Teun A., 274 Van Gijsel, Sofie, 272 Van Ginneken, Jac, 368, 383 Van Langendonck, Willy, 8, 175, 177.252 Van Oosten, Jeanne, 7 Vandeloise, Claude, 48, 49, 80, 133, 150, 374 Verdoodt, Albert, 270 Verlooy, Jan B.c.. 290 Verspoor, Mmjolijn, 347 Violi, Patrizia, 347,405 Virchow, Rudolph, 234 Vos, Louis, 277, 289 Vossler, Karl, 77 Voye. Liliane, 269 Wagner, Doris, 252 Waismann, Friedrich, 13 Weiland, Petrus, 38 Weinreich, Uriel, 7 Weisgerber, Leo, 147.379, 390, 396,397,400.404 Wellander, Erich. 368 Whedbee, William, 310, 313, 314, 315,322 Wierzbicka. Anna, 4, 13, 18, 19,22, 24,25.95,117,118.119,130, 131. 144, 145, 150, 152, 169, 170, 171,227,327,328,329, 330.405,416,423,425.430,441 Winters, Margaret, 8. 135 Wittgenstein, Ludwig, 9, 24

484

Index ofnames

Wodak, Ruth, 274 Woolard, Kathryn, 274 Wundt, Wilhelm, 368, 379 Yu, Ning, 227

Zgusta, Ladislav, 334 Zlatev, Jordan, 99 Zwicky, Arnold, 14, 101, 106,337, 350

Index of subjects

alloseme, 134 ambiguity, 14,51,53,58,78,106, 111,121,126,128,130,132, 134, 142, 144,233, 337, 338, 339,351 analogy, 33, 38, 61, 133, 134, 257, 307, 317, 375,418 anaphora, 106, 121, 122, 123, 124, 222, 350 anthropology, 3, 23, 74, 377, 386, 401,436 antonymy, 107, 114, 144,209,385, 387,388,401,403,410,411 applied linguistics, 328 artificial intelligence, 404 basic level hypothesis, 84, 85, 255, 258,261,406,410,411,412, 413,414 categorization, 5, 8, 9, 10, 16, 18, 21,22,27,44,96,101,102,103, 104,113,147,176,309,327, 331,333,334,343,374,375, 380,385,387,404,410,426 classical definability, 12, 14, 17, 102, 149, 150, 153, 154, 155, 156,157,160,164, 166, 167, 168, 171,341 cluster, 8, 18, 70, 142, 252, 330, 352, 357, 383, 384 cognitive linguistics, 8, 20, 21, 22, 24,27,48,99,100,103,138, 152, 175, 176, 191, 194, 198, 227,252,253,257,258,270, 272, 273, 274, 288, 307, 309, 345,367,405,416,421,422, 427,431,435,436,438,439, 440,441

cognitive science, 6, 8, 24, 43, 44, 103,405,416 cognitive semantics, 18, 23, 24, 25, 26,42,44,47,48,49,72,74,75, 76,80,81,141, 178,221,251, 309, 345, 346, 347, 352, 353, 354, 356, 357, 362, 363, 367, 374, 375, 377, 379, 380, 382, 383,384,385,387,391,392, 393,395,398,399,401,404, 405,406,407,408,411,412, 413,414,415,416,417,441,442 collocation, 30, 33, 40, 248, 345, 363,402,414,415 colloquial language, 305 componential analysis, 4, 6, 7, 23, 151,153, 178, 181,232,386, 387,388,390,392,401,402, 403,413,433 composite expression, 198, 199, 210,211,212,214,216,218, 220,223 compositionality, 198,200,201, 203,207,211,212,215,220, 223, 387 compound, 44, 49, 95,159,198, 200,209,212,215,217,220, 222,223,355,358,359 conceptual metaphor, 252, 352, 362, 413,414 connotational meaning, 45, 46, 200 construction, 36, 55, 56, 71, 77, 84, 106,109,114,120,121,122, 124,125,126,127,129,171, 173, 175, 176, 177, 178, 179, 180. 182, 183, 185, 186, 187, 189,190, 191, 192, 193, 194,

486

Index ofsubjects

195,196,197,223,313,350, 372,422,425 construction grammar, 175, 177 convergence, 295, 297, 299, 301, 302,303,304,351,398,414 core meaning, 139, 153,341, 342, 343 corpus linguistics, 30, 34, 40, 48, 85,100,149,151, 152, 155, 162, 170,175, 179, 193, 194,272, 334,335,345,363,402,410, 415,416,417,423,424,431, 434,438 critical discourse analysis, 274 cue validity, 28, 86, 90, 92 culture, 30, 94, 95,204, 225, 227, 228,230,246,248,250,251, 253,254,258,263,265,266, 268,269,270,272,273,274, 275,276,281,283,284,287, 289,290,291,292,293,294, 295,296,297,298,299,302, 305,307,310,349,373,377, 378,380,408,436,438,440 dative, 175, 178, 184, 188, 191, 193, 195, 196 definiendum, 113, 130, 155 definiens, 130 definition, 3, 4, 5, 8, 10, 11, 12, 13, 14,16,17,18,19,20,21,39,42, 49,51, 71, 72, 76, 78, 80, 85, 99, 100,101,102,106,107,112, 113, 114, 115, 117, 120, 123, 125, 129, 130, 131, 132, 133, 141, 143, 144, 145, 148, 149, 150,153, 154, 155, 156, 157, 158, 160, 164, 165, 166, 167, 168,169,170,171,172,177, 178,180,181,195,196,197, 223,250,262,297,309,327, 328, 329, 330, 337, 339, 341, 343,345,346,349,350,351,

354, 355, 356, 358, 359, 360, 361, 362, 374, 375, 376, 378, 389,390,401,403,423,424, 425,426,427,428,439 denotationa1 meaning, 10, 11, 13, 16,83, 88, 149, 152, 168, 169 diachronic semantics, 43, 103, 179, 218,251,273,352,369,370, 383,384,396,400,406,407, 409,411 diachrony, 3, 5, 135, 179, 184,218, 222,230,239,244,250,251, 263,273,301,302,327,334, 345, 352, 370, 377, 382, 384, 385,391,395,396,397,400, 407,409,411 dialect, 237, 266, 268, 274, 276, 277,278,281,301,396 Dialektik der Aufklarung, 280, 292, 296 dictionary, 20, 23, 25, 29, 38, 39, 126, 130, 309, 327, 328, 329, 330,331,332,333,334,340, 341,343,345,346,354,355, 356, 357, 358, 359, 361, 362, 370,402 discreteness, 13, 19,22,375 distinctiveness, 5, 76, 88, 89, 90, 119, 147, 149, 153, 154, 156, 157, 158, 160, 161, 165, 168, 175,252,287,346,349,354, 359,386,388,390,409,413 ditransitive construction, 177, 178, 194 divergence, 116,299,301,302,351 division of linguistic labor, 20, 21, 26,320,347,357,426,427 emancipation, 276, 284, 285, 299, 304 empirical method, 7, 10, 74, 95, 171,178,233,234,247,262, 273,300,306,334,344,368,

Index ofsubjects 370,371,379,387,389,392, 395,398,399,408,412,414, 416,432,440,441,443 empiricism, 22, 416, 441, 442, 443 encyclopedic meaning, 5,6, 10, 23, 42,47, 143, 146,205,264, 329, 346,354,356,357,358,371, 378,379,387,390,391,397, 408,436 Enlightenment, 276, 280, 282, 283, 284,288,292,296 entrenchment, 90, 92, 95,262,411, 414

epistemology, 10, 18, 22, 25, 44, 286,317,333,367,405,417, 421,422,423,430,436,440,442 etymology, 37, 44, 46, 209, 210, 237,238,268,332,370,379, 383, 396 evidence, 35, 37, 109, 115, 122, 124,131,137,154,179,184, 210,222,238,242,243,248, 249,253,270,303,308,343, 370,392,404,423,438 expert knowledge, 20, 21,95,356, 357,427 expressivity, 43, 45, 46, 47, 282, 285,295,296,372,373,374,387 extension, 3, 5, 1L 12, 13,20,21, 35,45,52,58,60,61,62,69,71, 81,87,106,112,117,132,135, 139, 147, 150, 170, 180,181, 182, 184, 187, 188, 189, 191, 194, 195, 197, 198,201,207, 208,213,215,218,343,347, 348,349,351,357,375,376, 411,412,414,424,427 falsificationalism, 140 family resemblance, 9, 11, 14, 15, 28,29,40,41,80, 102,376,384 felicity conditions, 315, 316, 318, 321

487

figurative meaning, 15, 35, 49, 128, 132,198, 199,202,203,204, 206,207,208,209,211,221, 222,245,257,347,352,362,383 figure / ground, 29, 30, 37. 46, 54, 76,95,117,118.119,235,250, 292,299,320,327,349,395,423 flexibility, 3, 5, 7, 10, 12, 17,19,20, 24,42,43,44,45,46,72,74,99, 132,137,251,310,333,346, 377,382,384,387,404,408, 427,432 formal semantics, 6, 439, 440, 441, 443 frame semantics, 6, 24, 43, 80, 82, 109, 110, 111, 120, 125, 127, 129,177, 179, 196,294,362, 405,413 frequency, 29,40,41, 75, 85, 89, 135,158,163,164. 179, 182, 193,208,263,264,299,300. 328, 348 function, 27, 29, 33, 39, 41, 42, 43, 44,45.47, 58, 82, 87, 108, 109, 115. 125, 130, 133, 134, 154, 176,179, 183, 184, 185, 191, 194, 195,243,257,266,275, 276,278,279,280,281,294, 295,296,297,304,319,320, 336,352,360,371,377,378, 384,393,406,411,413,422, 441,443 fuzziness, 7,13.16,22, 130,354 Geisteswissenschaft, 140, 373 generality, 113, 114, 118, 145, 149, 153, 168,241,242,275,276, 277,279,280,281,349,425 generative grammar, 6, 23, 389, 390,391,392,401,402,403,404 globalization, 293, 297, 306 gradience, 27, 41

488

Index ofsubjects

henneneutics, 140, 308, 373, 380, 396,434,435,436,437,438, 442,443 historical linguistics, 8, 370 historical semantics, 43, 103, 179, 218,251,273,352,369,370, 383,384,396,400,406,407, 409,411

historical-philological semantics. 333,368,371,372,373,374, 375,377.378,379,380,381, 382, 384, 385, 386, 387, 390, 391. 393, 395 homonymy, 122, 130, 177,208, 384, 389 humoral theory, 227, 229, 230, 231, 232,233,234,235,236,237, 238,239,240,241,242,243, 244,245,246,247,248,249,250 hyperonymy, 85, 86, 114, 155, 158, 159, 165 hyponymy, 161, 163, 164, 165,387, 388,401. 403 idealism, 22, 416, 417, 421, 430, 438,441,442,443 identity and language, 38, 50, 106, 114, 123, 124,177,215.252, 266,268,269.275,282,283, 289,290,291.292,295,296, 297, 298, 304, 337, 341, 343, 350,376,377 ideology, 272, 274, 275, 276, 280, 281,282,285,288,294 idiom, 173, 198, 199,200,201,203, 205,206,207,208,209,210, 211,212,217,222,248 indeterminacy, 13, 19,50.105. 116, 142, 343, 382 indirect object, 175, 176, 177, 178, 179,180,181, 182, 183, 184, 185,186, 187, 188, 189, 190, 191,192, 193, 194, 195, 196, 197

informal language, 301, 302, 303, 304, 305 intension, 3, 11, 12, 347, 348, 375, 386.392,424 interdisciplinarity. 6, 8, 103, 104, 416,436 interpretation, 43, 52, 55, 58, 61, 73, 79, 87, 103, 105, 109, 110, Ill, 113, 114, 115, 120, 121, 123, 124, 127, 129, 132, 133, 134, 135, 138, 139. 140, 141, 142, 143, 145, 146, 148, 150, 154,170, 199,200,201,203, 205,207,208,209,210,211. 218,219,220,221,222,227, 234,238,239,241,243,244, 245,248,273,274,288,293, 297,307.308,310,313,314, 315,322,323,339,346.348, 373, 374, 377, 380, 384, 385, 396,401,415,420,422,433, 435,436,437,438,439,440, 442,443 intracategorial polysemy, 10 1, 102, 104,136, 141, 142,262 introspection, 19,25,34,37,38,40, 109, 110, 152, 169, 170, 389, 417,423,428,429,430,432. 434,442 irony, 307, 308, 309, 310, 311, 312, 313,314,315,320,321,323, 324.355,423,432 isomorphism. 199,201,202,203, 206,207,208,209,211,212, 214,215,221,222 landmark, 53, 54, 56, 57, 58, 59, 60. 63,65,67,69,70,72,73,80,81, 82,83, 195 lexical field, 49, 78, 88, 147, 254, 257,261,262,264,270,390, 401,403,406,409,413,414

Index ofsubjects

lexicography, 8, 18, 309, 325, 327, 328,329,330,331,332,333, 334, 343, 345, 346, 347. 354, 359, 361, 362, 363, 370, 376, 379,389,402,429 lexicology, 74, 76, 77, 80, 90, 91, 94,95,99, 194,244,271, 272, 309,346,353,359,398,409. 411,416 lexicon, 7, 13,23, 74, 79, 88, 90, 92, 95. 104, 124, 129, 133, 135, 137,167,175, 176,208,252, 377,387,398,404,406,409, 410,415 literal meaning, 14, 112, 121, 128, 132, 184, 199,200,201, 203, 206,207,209,211,212,213, 215,217,219,221,245,246, 257,270,385 logic, 3,19,65.94,99,107,110, 113, 114, 115, 116, 117, 119, 120, 124, 125, 126, 127, 129, 131,142,143,144,146,274, 275,288,292,294,295,298, 303,306.316,321,347,381. 386.387.388,391,393,403, 404,435,440,441,443 logical semantics. 386, 387, 393, 403,441 macrostructure, lexicographical-, 328,421 markedness, 8, 46, 66, 161, 184, 233,244,249,264 membership, 3, 9, 12, 13, 16, 100, 102, 112, 116, 117, 120, 123, 142,375,376,377,443 metaphor, 5, 7, 24, 45, 46. 62. 102, 103,137,138,147,176,179, 180,182,183,191, 194, 195, 198,203,205,206,207,209, 211,212,213,214,215,216, 217,218,219,220,227.240,

489

241,242.243,245,246,247, 248,250,251,252,257,270, 352,375,378,381,384,385, 405,411,413 metaphtonymy, 218 methodology, 5, 6, 7, 8, 10, 20, 22, 25,27,37,48,49,50,71,88, 100,101, 136, 138, 139, 140, 141, 142, 149, 153, 156, 158, 160,168,169,171, 172, 176, 177,193,230,238,239,241, 242,246,247,248,249,250, 254,256,258,262,269,270, 273, 308, 309, 324, 328, 333. 344,363,369,370,373,374, 380, 385, 386, 387, 388, 389, 390,392,393,395,396,397, 400,401,403,404,406,408, 416,418,419,421,423,428, 429,430,431,432,433,435, 436,437,438,441,442 metonymy, 7, 17,65, 103, 176, 179, 184,186, 187, 194, 195, 198, 199,211.212,213,214,216, 217.218,220,241,245,251, 343.381,384,400,405,411 microstructure, lexicographical-, 328,330,421 monosemy, 14, 18, 20, 48, 49, 127, 130, 132, 136, 149, 150, 168, 172,338 morphology, 7,159,160, 177, 184, 220,221 motivation, 104, 199,201,202,203, 204,205,206,207,210,211, 221,222,223,238,239,240, 241,242,243,244,245,248, 250,304.431 multidimensional structure, 48, 69, 73, 175, 176, 178, 179, 192, 196, 197,272,327,331,351,352, 354, 361

490

Index ofsubjects

multilingualism, 294, 295, 296, 297 naming, 78, 83, 84, 85, 87, 96, 185, 252,300,349,406,409,427 nationalism, 272, 275, 287, 288, 289,290,291,292,293,294, 295,296,297,298,304,306 native speaker, 14, 38, 40, 72, 329, 330, 389, 390 Naturwissenschaft, 140, 373 near-synonymy, 50 necessary and sufficient criteria, 9, 11,14,16,19,71,100,101,112, 117,119,149,333,337,359,390 neogenerativist semantics, 398, 399, 402,403,404,406,407,408, 413,414,415 neostructuralist semantics, 5, 393, 398,399,400,401,402,406,408 network, 48, 80, 92, 99, 175, 178, 191,193,195,196,409 neuroscience, 8 non-equality, 12 non-referential meaning, 409 non-rigidity, 11, 12 norms oflanguage, 274, 300, 301, 304,440 objectivism, 10, 22, 24, 26, 138, 140, 141,380,393,395,421, 422,423,437,438,443 onomasiology, 27, 44, 47, 66, 74, 75, 77, 78, 79, 80, 83, 84, 85, 86, 87,88,90,91,93,94,95, 148, 158, 159, 163, 194,261,272, 274,299,300,398,400,408, 409,410,411,412,413,414,415 paradigmatic relations, 33, 38, 44, 80,90,91,92, 140, 198, 199, 201,203,206,207,210,220, 221,375,387,388,403,421 perception, 40, 96, 395

perspectival salience, 57, 60, 63, 75, 76,77,90,91, 138, 140, 189, 190,191 philology, 48, 308, 309, 333, 368, 369,371,372,373,374,375, 377, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 390, 391,393,395,396,435,437 philosophy, 3, 8, 9,22, 24,26,44, 138, 140,276,285, 287, 296, 333,367,368,389,399,403, 416,417,421,435,436,442,443 phonology, 7, 29, 83, 133, 134, 175, 177,228,384,388 physiology, 28, 40, 41, 42, 44, 228, 229,231,235,236,238,239, 240,241,242,244,245,246, 248,249,250,268,310,369,440 politeness, 310, 317, 318 polysemy, 3, 7,14, 18, 19,23,25, 48,49,50,51,67,73,80,87,91, 97,99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, Ill, 112, 113, 114, 115, 116, 117, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 140, 141, 142, 143, 144, 145, 146, 147, 149, 150, 153,168, 169, 172, 178, 193, 197,218,251, 327, 328, 330,331,333,339,345,346, 347,350,351,358,359,362, 363,376,380,382,383,408, 410,413,414,415,428 postmodernism, 272,275,276, 288, 293,295,296,297,298,304, 305,307,308,438 pragmatics, 75, 78, 90, 104, 134, 140,150,188,310,315,316, 317,318,329,330,331,391, 409,414,415

Index ofsubjects

preposition, 21, 48, 49, 81, 84. 178 prestructuralist semantics, 23, 43, 103, 179,333,354,367.383, 395,396,398,399,400,407, 411,412,414 prismatic semantics, 199, 200, 206, 211,213,214,215,216,218, 219,220,222,223 proper name, 73, 216, 252, 255, 270, 356 prototype theory, 1, 3, 4, 5, 6, 7, 8. 9. 10, 11, 12, 13, 14, 15, 16, 17, 18, 1~ 21,22,23,24,25,27,28, 29.34,35,36,37,38,39,40.41, 42,43,44,46,47,48,49,65,67, 70,71,72,74,75,77,80,87,90, 91, 100, 101, 102, 103, 104, 116, 120, 125, 130, 135. 136, 138, 139, 140, 141, 142, 144, 152, 170, 171, 175. 176, 178, 179, 180, 181, 183, 184, 189, 190, 191,193.194,196,231,251, 255,258,274,310,327,328, 329,330,331,333,334,335, 337, 339, 340, 341, 343, 345. 346, 349, 352, 354, 355, 356, 357,358,360,362,374.375. 377, 379, 381. 382. 383, 384, 385,387,395,397.398,405, 411,412,413,414,421,427 psycholinguistics, 4, 8, 24, 65, 67, 74,99, 100, 102.210,248,399, 404,405,416,431 psychology, 3, 4, 6, 8, 9, 10, 13,23, 28,29,36,41,42,43,44,71,72, 103,104,110,138,179,217, 222,227,230,231,234,235, 236,239,245,250,348,368, 371,372,377,378,379,380, 382, 389, 390, 392, 393, 395, 396,397,400,401,403,404, 407,416,427,430,435,441

491

purism, 292 quantitative method, 300, 302 radial network, 9, 14,20,24, 25, 48, 49,101,102,175,178,191,195, 196,346,352,354 rationalism, 272, 273, 275, 280, 282,283,284,285.288,289, 290,291.292,293,294,295, 296,297,298,304,305 referential meaning, 11,28,40,41, 42,44,50,51.53,66,72,80,86, 87,89,91, 102, 103, 104, 105, 108, 124, 137, 147, 148. 150, 152, 154, 155, 156, 159, 161, 162. 164, 170, 188, 258. 263, 264, 270, 339, 348, 349, 352, 353,354,409,413,423,424, 426,433 register, 301, 302, 303, 304 reinterpretation, 124, 134, 154, 157, 161,208,209,211,222,244, 245, 248 relational semantics, 48, 53, 80, 144,178. 196, 197,201,401, 403,409 representativity, 9, 12, 13, 14, 16, 17, 19, 30, 114, 252, 255, 436 romanticism, 188, 272, 273, 275. 280,282,283,284,285,288, 289,291,292,293,294,295, 296,297,298,299.304,305, 437,438,440,443 rule/list fallacy, 135 salience, 1,3, 19,20,25,27,28,35, 38,39,40,42,49,65,71,74,75, 76, 77, 79, 82, 83, 84, 85, 86, 87, 88,89,90,91,92,93,94,95,96, 101,194,207,231,252,254, 255,257,258,260,261,262, 263,264,266,270,330.331, 348,349,376,406,410,411,426 scenes-and-frames.405

492

Index ofsubjects

semantic feature, 5, 75, 88. 90. 252, 392 semasiology, 5,44,47,66, 74, 75,

77,78,79,80,86,87,90,91,92, 93, 148, 154, 156, 158, 159, 160, 162,178, 179, 194,274,347, 348, 349, 352, 354, 368. 398,

400,408,409,410.411,412, 413,414,433 semiotics. 79, 90, 234, 235 signifiant, 77 signifie, 77, 103 snow, 94, 95 social aspects of meaning, 21, 23, 30, 35, 36, 83, 94, 95, 130, 204.

227,230,246,250,251,265, 267,270,272,273,274,275, 276,278,279,281,282,283,

284,285,287,289,290,291, 292,293.294,295,296,297,

298,299.302,305,307,310. 323, 349. 355, 368, 372, 373,

377,378.380,408,426,436, 438,440 sociolinguistics, 20, 21, 22, 23, 45, 46,83.90,228 speech act. 20, 315, 316, 318, 323 Staatsnationalismus, 292. 297 standard language. 275, 276, 277,

279,280,281,282,283,285, 293,296,298,299,301,303,304 standardization, 272, 273, 275, 276,

279,280,281,282,283,284, 285,287,288,289,293,296,

298,299,301,302,303,304, 305, 306 statistics, 28, 40, 41, 42. 44, 94, 99. 139, 162, 163 stereotype, 20, 25, 328, 357, 358, 426 stratification of language, 83, 299,

301,302

structural salience, 75, 79, 88, 90, 91,94,252 structuralist semantics, 5, 23, 43, 77, 103, 329, 344, 353, 354, 356, 362,367,374,377.379,382, 386, 387, 388, 389. 390, 391,

392,393,397,398,399,400, 401,402,403,404,406,408, 411,412,413,414,415 subjectification, 51, 176. 179, 188, 195 subordination. 323 synecdoche,65,384 synonymy, 27, 29, 38, 39, 45, 47, 50, 83, 84, 86, 90, 91, 92. 107, 143, 144, 155, 156, 158, 159, 162, 163, 167, 195,209,237, 299,351,354,359,387,388, 389,401,403,409 syntagmatic relations, 30. 33, 34,

39,52,56,60,61.64,65,70.76, 80,81,82,90,91,93,196, 198. 199,200,201,207,209,210, 220,221,222,387,388,390,

401,402,403,413,414 syntax, 6, 7, 109, 193,390 trajector, 53, 54, 56, 57. 59, 60, 62. 63,65,69,70,72,73,80,81.82, 83 transfonnational grammar, 4, 6, 106, 350, 385, 386. 388, 389,

390,391,392,393 transparency. 20, 201, 204, 207, 208,209,211,230,240,245,

248,349,423,426,430,439 truth conditions, 108, 109, 110, 134,

143,387,393,403,428 turn-taking, 317 universality, 94, 96, 227, 289,437 univocality. 14, 18,20,48.49, 127, 130, 132, 136, 149. 150. 168, 172,338

Index ofsubjects usage, 18,20,21, 30, 35, 36. 37. 38, 39,40,41,44,46,50,62.64,66, 87, 115, 149, 150, 152, 168, 169, 171, 172,208,273,278,289, 331,377,398,414,415,427, 429,431 vagueness, 12,28, 50, 72, 80, 87, 99, 100, 101, 104, 105, 106, 107, 109, Ill, 116. 120, 121, 123, 125,129,133, 134, 136, 137, 138,141,142, 144, 146, 147, 193,257,338,339,350,369, 375, 376, 381, 382 vantage theory, 138

493

variation, 27, 45, 46, 74, 83, 84, 86, 133,138, 150, 154, 157, 175, 205,228,256,272,273,274, 275,282,283,285,287,288, 289,292,293,298,299,305, 341, 375 vocabulary, 30, 95, 227, 230, 237, 239,241,242,244,245,246, 247,249,250,254,257,331, 392,402,439,443 Volksnationalismus, 292, 297 Woordenboek der Nederlandsche Taa1, 29, 327, 332, 334, 335, 340,341,342,344,361,371