Metonymy in language and thought

Metonymy in Language and Thought HUMAN COGNITIVE PROCESSING is a forum for interdisciplinary research on the nature a...

Author: Klaus-Uwe Panther and Günter Radden (editors)

380 downloads 7120 Views 12MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Metonymy in Language and Thought

HUMAN COGNITIVE PROCESSING is a forum for interdisciplinary research on the nature and organization ofthecognitive systems and processes involved in speaking and understanding natural language (including sign language), and their relationship to other domains of human cognition, including general conceptual or knowledge systems and processes (the language and thought issue), and other perceptual or behavioral systems such as vision and non-verbal behavior (e.g. gesture). 'Cognition' should be taken broadly, not only including the domain of rationality, but also dimensions such as emotion and the unconscious. The series is open to any type of approach to the above questions (methodologically and theoretically) and to research from any discipline, including (but not restricted to) different branches of psychology, artificial intelligence and computer science, cognitive anthropology, linguistics, philosophy and neuroscience. It takes a special interest in research crossing the boundaries ofthese disciplines.

EDITORS

Marcelo Dascal (Tel Avi11 Uniw:rsity) Raymond Gibbs (University of California Q1 Santo Cruz) jan Nuyts (University of Antwerp) Editorial address: Jan Nuyts, University of Antwerp, Dept. of Linguistics (GER), Universiteitsplein 1, B 2610 Wilrijk, Belgium, e-mail: [email protected]

EDITORIAL ADVISORY BOARD

Melissa Bowerman (Nijmegen); Wallace Chafe (Santo Barbara, CA) Philip R. Cohen (Portland, OR); Antonio Damasio (lowo City, lA) Morton Ann Cernsbacher (Madison, WI); David McNeill (Chicago, IL) Eric Pederson (Eugene, OR); Fran~ois Recanati (Paris) Sally Rice (Edmonton, Alberto); Benny Shanon Oerusa/em} Lokendra Shastri (Berkeley, CA); Dan Slobin (Berkeley, CA) Paul Thagard (Wo~rloo, Ontario)

Volume4 Klaus-Uwe Panther and GUnter Radden (eds)



Edited by KLAUS-UWE PANTHER GUNTER RADDEN University of Hamburg

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAMJPHILADELPH lA

TM

00

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences- Permanence of Paper for Printed Library Materials, ANSI Z39·48·1984.

Ubrary of Congress Cataloging-in-Publication Data Metonymy in language and thought f edited by Klaus-Uwe Panter, GUnter Radden. p. em. -- (Human cognitive processing, ISSN 1387-6724; v. 4)

Papers presented at a workshop held June 23-24,1996, Hamburg University. Includes bibliographical references and index. 1. Metonyms Congresses. 2. Cognitive grammar Congresses. I. Panther, Klaus-Uwe, 194211. Radden, GUnter. Ill. Series. 1999 PJ01.5.M49M48

401'.41··dC21 ISBN go 272 1356 4 (Eur.)

99-23468

/1

55619 204 5 (US) (alk. paper)

CIP

© 1999- John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. • P.O.BoK 75577 • 1070 AN Amsterdam • The Netherlands John Benjamins North America • P.O.BoK 27519 • Philadelphia PA 19118·0519 • USA

Acknowledgments

The present volume evolved from a workshop on metonymy held at Hamburg

University on June 23 and 24. 1996. This conference wa." probably the first international meeting of scholars from Europe. North America and Asia which

was exclusively devoted to the study of metonymy in language and thought. We would like to express our thanks to all of the participants of the conference. both presenters and audience. for making the workshop an intellectually stimulating and revealing event. The workshop was generously funded by the Deut.Khe Forsclrungsgemeinschaft and the Hansi.Khe Unh·ersitiit.utiftung, Hamburg. whose financial contributions are gratefully acknowledged. For their help in organizing the workshop and preparing the manuscript we are indebted to our secretary Elisabeth Himmler and our student assistants Elizabeth Matthis. Anatol Stefanowitsch and Sebastian Ross-Hagcbaum who devoted many long hours to the conscientious preparation of the volume. Without their dedication and suppon this volume would not have been possible. We also benefited greatly from the insightful comments given to us by two anonymous reviewers. Finally. we would like to thank lhe editors of the series Hwmm Cognitive Proces.\·ing for accepting this volume, in panicular Jan Nuyts. who kindly assisted us in all editorial matters.

Contents Introduction

Klaus-Uwe Panther and Giinter Radden

Part 1: Theoretical Aspects of Metonymy Towards a Theory of Metonymy Giinter Rudden and Zo/uin KOvecses

17

Speaking and Thinking with Metonymy Raynwml W. Gibbs, Jr.

61

Metonymy and Conceptual Integration

77

Gilles Fauconnier and Mark Turner Distinguishing Metonymy from Synecdoche

91

Ken-khi Seto Aspects of Referential Metonymy Bemrice Warren

121

Part II: Historical Aspects of Metonymy Frame and Contiguity: On the Cognitive Bases of Metonymy and Certain Types of Word Formation

139

Peter Koch

Co-presence and Succession: A Cognitive Typology of Metonymy Andrew~

Blatlk

169

viii

Contents

Metonymic Bridges in Modal Shifts Louis GoosJens

193

Metonymy in Onomastics 0/afJdke/

21 I

Part III: Case Studies of Metonymy Grammatical Constraints on Metonymy: On the Role of the Direct Object Richard Waltereit

233

Putting Metonymy in its Place Paul Pauwels

255

Conversion as a Conceptual Metonymy of Event Schemata Rem! Dirven

275

Opposition as a Metonymic Principle Christian VoHhagen

289

Metonymic Hierarchies: The Conceptualization of Stupidity in German Idiomatic Expressions Kurt Feyaerts The Potentiality for Actuality Metonymy in English and Hungarian Klaus-Uwe Panther am/ Linda Thornburg

309

333

Part IV: Applications of Metonymy "Mummy.llike being a sandwich": Metonymy in Language Acquisition Brigitte Nerlic:h, David D. Clarke. Za::}e Todd

361

Rccontextualization of Metonymy in Narrative and the Case of Morrison· s Song of Solomon Anne Pankhllnt

385

Col! tell I.\'

ix

List of Contributors

401

Subjccl index.

405

Author index.

417

Metonymy and metaphor index.

423

Introduction

Klaus-Uwe Panther and Gunter Radden University of Hamburg

1.

Background

Eighteen years after Lakoffand Johnson's (1980) seminal work on the role of metaphor in conceptualilation. which sparked a vast amount of research in cognitive linguistics. it has become increasingly apparent that metonymy is a cognitive phenomenon that may be even more fundamental than metaphor. We believe that the contributions give a fair view of the state of the art in metonymic research. although we are also aware of the fact that a great many questions about metonymy still remain unanswered, some of which will be addressed below. The cognitive understanding of metaphor and metonymy is certainly at variance with both naive and traditional scholarly views. which have strongly been influenced by centuries of rhetorical and literary studies. The cleavage between literal and figurative language. which was taken for granted by traditional rhetoric and linguistics, has recently been challenged by Gibbs (1994: 24-79: and this volume). Still. we owe the first basic insights into the nature of tropes toGreek.Roman and medieval scholars. modem literary critics and linguists. Many different classifications of tropes have been proposed. starting with Aristotle. who subsumed metonymy and synecdoche under metaphor, and more recently by the Gro11pe cle Lii!ge or Groupe Ji. which subsumed metaphor and metonymy under synecdoche (see Schafer and Rice 1977). Some of these ideas on metonymy definitely have a modem. cognitive tinge. Various contributors to this volume (Koch: Blank: and Nerlich. Todd and Clarke) link their cognitive approach to metonymy to this rhetorical tradition.

Klau.'i-Uwe Pa"ther cmd Giimer Radcle"

The authors of the contributions to this volume have different theoretical backgrounds and are affiliated with different disciplines: linguistics. psycholinguistics, psychology and literary studies. Many of them share the assumption that metonymy is a cognitive phenomenon underlying much of our ordinary thinking and that the use of metonymy in language is a reflection of its conceptual status. The conceptual framework within which metonymy is understood in most of these contributions is that of scenes. frames. scenarios. domains or idealized cognitive models (ICMs). Within these models, a metonymic link may be established between two conceptual entities in the broadest sense. This view supersedes the traditional assumption of metonymy as having primarily a referential function. a view which was still held by Lakoff and Johnson ( 1980). The papers read at the conference and collected in this volume address a wide range of topics related to metonymy. The papers have been grouped into four parts. Part I deals with theoretical aspects of metonymy as a cognitive process. Part 2 investigates historical aspects of metonymy within a cognitive framework. Part 3 contains a number of case studies on selected metonymies or aspects of metonymy. Part 4 explores the notion of metonymy in its application to language acquisition and literary criticism.

2.

Contributions to the volume

2.1. Theoretical aspects of metonymy Three papers address the role of metonymy in language and thought from a broader theoretical perspective. The issue of the conceptual nature of metonymy is investigated by GUnter Radden and Zoltan Ki:iv~ses in their paper "Towards a theory of metonymy." Their approach is based on the notion of idealized cognitive model (ICM) as proposed by Lakoff(l987). Metonymy is understood as a conceptual process in which one conceptual entity. the 'target,' is made mentally accessible by means of another conceptual entity. the 'vehicle.' within the same ICM.In principle. either of the two conceptual entities related may stand for the other. i.e .. metonymy is ba'iically a reversible process. There are. however, a number of cognitive principles which govern the selection of a preferred vehicle. These principles lead to natural. or 'default' cases of metonymy and oflen escape our awareness. These principles

Introduction may. however, be overridden by factors such as style. taboo or politeness and may lead to the creation of expressive. or 'non-default' cases of metonymies. In his contribution on "Speaking and thinking with metonymy:· Raymond Gibbs lays the foundations for the study of metonymy in ordinary language as well a literary discourse. He situates metonymy in a larger cognitive context and adduces evidence for the conceptual basis of metonymy. Thus. the Gricean notion of conversational implicature can be seen as being metonymically motivated; metonymic reasoning may also contribute to the establishing of coherence by means of 'conceptual anaphors.' Finally, Gibbs also shows how metonymy is operative in discourse and leads to a better understanding of contextually determined reference. indirect speech acts. and colloquial tautologies. Gilles Fauconnier and Mark Turner explore another conceptual aspect of metonymy. In their paper "Metonymy and conceptual integration,'' the authors investigate the interaction of conceptual blending and metonymy. They demonstrate that Lakoffs and KOvecses' unidirectional model of conceptual metaphor does not account for expressions such as smoke is comint: out of his ears. which can only be understood as resulting from 'blending' the source and the target domains. since.literally. there are no ears in the source domain and there is no smoke in the target domain. Fauconnicr and Turner's approach has far-reaching consequences for the theory of metaphor in that it may very well tum out that most metaphors involve conceptual integration. Also, conceptual entities may be metonymically linked in a blended space. In the well-known symbolic representation of death as The Grim Reaper. the input clements 'scythe.' 'cowl,' and 'skeleton' arc conceptually integrated. Thus the blend 'shortens' the metonymic distance between originally noncontiguous conceptual entities. Ken-ichi Seto's paper "On distinguishing synecdoche from metonymy" argues for a clear conceptual distinction between metonymy and synecdoche. Despite the current interest in these tropes, they have not yet been defined in a satisfactory fashion. According to Seto. the reason for this lack of precision resides in the confusion between taxonomies and partonomies. Taxonomies involve 'kind or relations in a hyponymically-structured conceptual domain (e.g., ct ham sandwkh is a kim/ offood). whereas partonomies involve 'part of relations in the physical world (as in an arm i.{ a part of the body). Seto calls the former rehnions. which are defined by semantic inclusion, C-relations (category relations); the latter relations, which are constituted by spatia-

Klaus-Uwe Pa,ther cmd Gii11ter Radcle11 temporal contiguity between entities in the physical world. are called Erelations (entity relations). Seto proposes to reserve the term 'synecdoche' for C-relations and 'metonymy' forE-relations. On the basis of this distinction. he develops classifications of metonymic and synecdochic relations. In herpaper"Aspectsofreferential metonymy," Beatrice Warren detects structural parallels between referential metonyms on the one hand and nounnoun compounds and denominal verbs on the other hand. These three constructions have in common that they involve two referents and an implicit link connecting them. In noun-noun compounds. the referring item is explicit. whereas it is implied in metonyms.Accordingto Warren. metonymy is basically an abbreviation device and, apart from finding the implicit referring item. il interpretation involves retrieving a relation. Metonymic relations are restricted to a small number of recurrent links. This makes the interpretation of metonymy less demanding. or even effortless. when compared to the open-ended interpretation of metaphor. which involves working out several matching links (of similar attributes) between the conventional and intended referents.

2.2. Historical aspects of metonymy The outstanding role of metonymy in triggering linguistic change has long been recognized and led to various classifications of types of metonymy. some of which are described in the papers below. The cognitive paradigm in linguistics offers a new analytic tool for analyzing historical data. Four papers are devoted to the operation of metonymy in historical processes of language. In his paper "Frame and contiguity: On the cognitive bases of metonymy and certain types of word formation," Peter Koch investigates metonymicallyinduced changes of meaning. He makes use of the conceptual networks provided by frame theory. Frames may be seen as conceptual gestalts and. in metonymic changes of meaning. a new sense is highlighted as the ·figure' while the old sense serves as its 'ground' within the frame. Figure/ground effects also account for the origin of metonymy in discourse. Koch distinguishes between three types of ad hoc metonymic innovation inducing a metonymic change: hearer-based inferential innovations such as the interpretation of 'fireplace' as 'fire.' speaker-based 'imprecise' innovations such as the sense of 'hip' for 'thigh; and expressive innovations such as 'skull' for 'head: Andreas Blank presents further argumenL" for a frame-theoretical approach to metonymy. His paper "Co-presence and succession: A cognitive

Introduction typology of metonymy" critically examines the classifications of metonymy that have been proposed. He claims that most typologies of metonymy are defective in that a number of metonymies do not fit into any of the categories proposed. He argues that all types of metonymic changes can be subsumed under two major types: relations between entities that are co-present within a frame. and those that are successive within one frame or two related frames. His typological model of metonymy comprises three levels of abstraction: the two domains of co-present and successive contiguity at the highest level. schematic types of contiguity at the intermediate level and concrete linguistic metonymies at the lowest level. A specific problem of semantic change is studied by Louis Goossens in his papcr"Metonymic bridges in modal shifts." He investigates the conceptual shift of the English modal must from a deontic to an epistemic meaning. This general shift, which also applies to other modal verbs, has been accounted for in two different ways: as a metaphorical mapping from the sociophysical world onto the epistemic world (Sweetser 1990), or. especially in grammaticalization studies, as a shift triggered by context-induced inference. Both synchronic and diachronic data on the usages of must suggest that the shift from deontic to epistemic is a gradual process. which can be accounted for by metonymic bridges. The metonymic approach to historical linguistics is also relevant to the discipline of onomastics. In his paper "Metonymy in onoma.,;tics," Olaf Jiikel investigates the cognitive motivation underlying naming patterns a.-. evidenced in the etymologies of German surnames. Apart from a number of surnames which are not motivated or whose motivation is obscure. surnames are coined by means of three principal patterns of naming: genealogy. profession and metonymy. Metonymic naming strategies make usc of three types of metonymy: 'utensil metonymy' (IMPORTANT UTENSIL FOR PERSON) as in Bohn.mck 'beanbag: 'quality metonymy' (SALIENT QUALITY FOR PERSON) as in Wwulerlich 'strange: and 'location metonymy' (PLACE OF ORIGIN OR RESIDENCE FOR PERSON) as in La11gacker 'long field.'

2.3. Case studies of metonymy A variety of case studies investigate the operation of metonymy on various linguistic and conceptual levels. One paper is concerned with metonymy in grammar (Wahereit); three papers look at selected lexical and semantic me-

Klaus-Uwe Panther (llld Giimer Rad(/en

tonymies (Dirven; Pauwels: Vo8hagen): one paper studies the metonymic slructure of a particular concept in its cultural context (Feyaerts). and another paper investigates the degree of exploitation of a particular metonymic principle in two genetically unrelated languages (Panther and Thornburg). Richard Waltereit's contribution "Grammatical constraints on metonymic reference: On the primacy of the direct object" demonstrates that metonymy also plays an important role at the level of grammatical relations. First. metonymic transfer can involve the insertion of a participant into a given thematic role such as Le 53 e.\'1 rentri 'No. 53 is back," where the number stands for a hotel guest. Second. it can also involve a transfer of semantically contiguous thematic roles as in Papa va balayer ta c:hambre 'Daddy will sweep your room' vs. Papa n'a pa,\· encore balayi /es dibrisde wrre 'Daddy hasn't swept up the broken glass yet,' where the container (the room) is semantically contiguous with the object contained (the broken gla'>s). Waltereit argues that. with regard to metonymic transfers, the direct object has primacy over the subject and other grammatical relations. The metonymically privileged status of the direct object is caused by three factors: it is semantically opaque and. hence. allows for a number of thematic roles to fit into the direct object slot: it is the argument that is semantically closest to the verb, which entails that its referential autonomy is somewhat weakened: and it is syntagmatically closer to the subject than other (oblique) arguments. In his lexical-semantic study "Putting metonymy in its place," Paul Pauwels investigates the metonymic structure of four related verbs: put ..~el. lay. and place. Pauwels' corpus-based investigation shows that the majority of examples were not of the traditional nominal or referential kind. In his corpus. metonymy often seems to function a'> a 'euphemistic avoidance strategy.' But it can also serve a.. a focusing strategy. which. in extreme cases, may result in dysphemism. The most frequent metonymic type Pauwels encounters in his corpus is based on a relation of inclusion. where a more general concept stands for a more specific concept. or vice versa. In his paper "Conversion as a conceptual metonymy of basic event schemata." Rene Dirven investigates the phenomenon which is generally known as conversion or zero-derivation, in particular. the conversion from nouns to verbs {e.g .. author vs. to author). He shows that the process of conversion is typically found in three event schemata: the action schema, the location and motion schema, and the essive schema. Conversion is regarded as a process in which one participant in the event schema is metonymically focused

l11trodllctio11

upon. but the whole event is conceptually involved. For example. in the action schema the participants patient (fish), instrument (hook), and manner (pearl jishi11g) are most frequently converted into new verbs (to jiJh, to f10ok. to ji.tlr pear/.{, respectively). Five participant types are regularly exploited to yield new verbs. In conclusion. Dirven raises the question if this selectivity is the result of sociocultural saliency or rather a matter of linguistic preference. Another semantic study. Christian VoRhagen's paper "Opposition as a metonymic principle,'' focuses on antonymy as a metonymic relation. This metonymy shows up in irony. where usually a positive concept metonymically stands for a negative concept. and in some conventionalized lexical items such as terribly in It was terribly wmui11g. As a rule. the metonymy applies to evaluative concepts, which are semantically scalar but are reinterpreted as complementary. In his paper "Metonymic hierarchies: The conceptualization of stupidity in German idiomatic expressions:· Kurt Feyaerts studies the metonymic structure of everyday expressions of stupidity. For example. an expression such as Dll hist wohl 11icht wm hier? 'You arc not from here. are you?' exemplifies the metonymic folk model OUTGROUP ORIGIN FOR STUPIDITY. Feyaerts shows that metaphorically and metonymically organized hierarchies have major structural characteristics in common. Higher-level metonymies tend to be cross-culturally valid. while lower-level metonymies are more culture-specific. Klaus·Uwe Panther and Linda Thornburg emphasize the importance of a cross-linguistic comparison of conceptual metonymies. In their paper"The POTENTIALITY FOR ACTUALITY metonymy in English and Hungarian,'' they analyze the extent to which this metonymy is exploited across two genetically unrelated languages. English and Hungarian. They explore its operation in seven conceptual domains: sense perceptions. mental states and processes. hedged performatives. indirect speech acts. (extralinguistic) actions. character dispositions. and acquired skills. In some of these domains. the POTENTIALITY FOR ACTUALITY metonymy is much more productive in English than in Hungarian. The most striking contrast between the two languages emerges in the domain of sense perceptions: whereas English systematically exploit.~ the metonymy in sentences such as I can ta.~te lhe mnilla (for lla.we the mnilla). Hungarian systematically excludes the metonymy and resorts to a non-modal construction in the indicative mood. The authors also discuss the relationship between Griccan maxims. conversational implicaturcs and metonymy.

Klau.'i-Uwe Panther (llld Giinter Radden

2.4. Applications of metonymy Two contributions arc devoted to the significance of metonymy in language acquisition and literary criticism. The role of metonymy in language acquisition is investigated by Brigitte Nerlich. David D. Clarke and Zazie Todd. Their paper '"Mummy, I like being a sandwich': Metonymy in language acquisition" may in fact be the first study on the production and understanding of metonymy in this field. whereas studies on the production and understanding of metaphor in language acquisition proliferate. In child language the usc of metonymy serves two different functions: it is a means of extending the known stock of words to cope with increasing communicative needs and of exploiting ·natural pathways of meanings' creatively. The former use of metonymy is a pragmatic strategy which leads to 'compelled' overextension. Compelled overextcnsions arc typically found up to the age of 2;5, while by the age of four children start producing metonymy for creative purposes. This use of metonymy is referred to by the authors as 'creative metonymical shrinking.' Children's comprehension of metonymy is empirically studied using a group of 2-3 year-olds and a group of 4-5 year-olds. In her paper "Recontextualization of metonymy in narrative and the case of Morrison· s Song of Solomon:· Anne Pankhurst explores the function of metonymy in narrative fiction.ln the novel analyzed. the reader has to activate metonymic strategies in order to understand the impact of an apparently simple object. an earring. The earring serves several metonymic functions: it is. in particular. a means of identifying its wearer and. at a macro-structural level. holds together different episodes of the novel. Pankhurst argues that the complex use of metonymy in this narrative cannot be accounted for by a single theory. The most promising approach to understanding the complex processes of reference and recontextualization in this world of fiction arc provided by Riffaterre·s functional view of metonymy and Gibbs' metonymic models of thought.

3. Perspectives for future research The papers collected in this volume certainly contribute to a deeper understanding of the conceptual nature and function of metonymy. At the same time. the contributors are aware of the fact that many a a cognitive process in which one conceptual entity. the \"Chide. provides mental acce~s to another conceptual entity. the 1Mget. within the same idealized cognitive model.

This working definition is useful in that it allows us to raise further important empirical and theoretical issues. We believe that. amongst others, the following questions need to be addressed in developing a theoretical framework. of metonymy. A first question we need to ask is where do we find metonymy? According to the above definition. metonymy may occur wherever we have idealized cognitive models. We have ICMs of everything that is conceptualized, which includes the conceptualization of things and events. word forms and their meanings, and things and events in the real world. We will refer to these types of conceptualization as 'ontological realms.' ICMs are not restricted to a single ontological realm. but may also interrelate ontological realms. For example, people tend to sec a close relationship between the two entities which establish a sign: the concept of a thing and its name. This ICM crosscuts two ontological realms and. as will be shown below .leads to metonymy. The notion of metonymy which follows from the conceptual definition given above thus has much wider application than that of traditional approaches. A second question which needs to be addressed relates to the 'mental bridge' which allows the conceptualizer to access the desired target. This

22

Giimer Raclden and Zoltdn Kbvecse.\'

question concerns the nature of the relationship between the vehicle and one or more targets. Metonymy tends to make use of stereotypical, or idealized. relationships within an ICM. Thus, certain places tend to be associated with events which typically occur at the place. For example, the expression to go to bed may. depending on the situation. evoke the metonymic targets 'to go to sleep: 'to have sex· or 'to be sick.' All these events arc stcreotypically associated with beds. irrespective of the particular context that triggers the situationally relevant target. More generally, we may describe the conceptual relationship between space and event as one that is entrenched and may be exploited by metonymy. The question that needs to be answered here is what types of conceptual relationships in an ICM may give rise to metonymy. A third question pertains to the choice of vehicle and target. Thus far, we have only considered conceptual relationships between two entities either of which may become the vehicle or the target as shown in the examples of She 'J a pretty face and She's a pretty penon. Unlike metaphorical mappings. which tend to be unidirectional. metonymic mappings arc in principle reversible. This was already implicitly noticed in traditional approaches by listing both directions of a metonymic relationship such as CAUSE FOR EfFECT and EFFECT FOR CAUSE. GENL:S FOR SPECIES and SPECIES FOR GENUS. etc. Such theoretically possible alternatives have to be distinguished from the speaker's choice of a particular vehicle as the 'entry point' into the ICM. We therefore need to ask if there are any preferred metonymic construals and. if this is the case. what 'cognitive principles' govern the selection of one type of vehicle entity over another. Such precedence principles were already hinted at by Langacker ( 1993: 30). To the extent that there are such preferred routes. these will define the unmarked. or 'default.' cases of metonymy. Given that there are such default routes. we need to ask. as a fourth question, if there arc any principles that determine the choice of a vehicle other than by default construal. If this is the case, these metonymic construals yield marked, or 'non-default,' instances of metonymy. The following four sections of this paper will be devoted to finding answers to these central questions which, for convenience. are summarized below: (i) What are the ontological realms in which ICMs and metonymic relationships may occur? (Section 2): (ii) What arc the types of conceptual relationships that may give rise to metonymy? (Section 3): (iii) Arc there any conceptual entities that can better direct attention to an intended target than others? If this is the case, are there any cognitive

23

Towards a Theory of Metoflymy

principles that govern the selection of such 'default' cases of metonymy? (Section 4): (iv) Are there any principles that override the preferred default routes and yield 'non-default' cases of metonymy? (Section 5).

2.

Ontological realms in which metonymy occurs

The following three ontological realms are distinguished for the present purpose: the world of 'concepts.' the world of 'fonns.' in particular. fonns of language, and the world of 'things' and 'events.' These realms roughly correspond to the three entities that comprise the well-known semiotic triangle as developed by Ogden and Richards (1923: II): thought. symbol and referent. The interrelations between entities of the same or from different ontological realms lead to various ICMs and possibilities for metonymy. An important distinction has to be made between ICMs which interrelate entities of different ontological realms within the same semiotic unit and lCMs which interrelate entities of different semiotic units within the same ontological realm or realms. The fanner situation of interrelated ontological realms gives rise to two lCMs: the pairing of a concept and a fonn establishes a sign and may be described a.-. 'Sign lCM': the pairing of a thing or event and a sign. fonn or concept establishes a referential situation and may be described as 'Reference ICM.' In as far as these ICMs lead to metonymy. the metonymies will be described a'\ 'sign metonymy' and 'reference metonymy." respectively. The latter situation of interrelated semiotic units involves concepts, typically in conjunction with fonns. These ICMs will be referred to as 'Concept ICMs.' and a metonymy ba.'\ed on a Concept ICM will be described a'\ 'concept metonymy.· Figure I illustrates the semiotic relationships which lead to the sign metonymy (I) and three types of reference metonymies (2H4) on the one hand and one type of concept metonymy (5) on the other hand. The arrows indicate the direction of the metonymic mapping which will be discussed below.

{3) / /

)'"~~%' {2)----~·--.....: ())

Ji.·------

Thing/Event

·---------------

"

ronn

(4)

FiRUrl' /. Siwr. referellct' mul com·ept me1mr.w11ie.'i

(5)

GUnter Raclden and Zoltdn Kiivecse.\·

24 2.1 . Sign JCMs and sign metonymies

The Sign ICM unites a form and one or more concepts. Thus. the word form dollar or the dollar sign$ arc intimately linked with the 'currency denomina-

tion of dollar.' 'currency.' or 'money· in general. As a rule. the form mctonymically stands for the concept it denotes. (I)

FORM FOR CONCEPT:

dollar for 'money'

The very nature of language is based on this metonymic principle. which Lakoff and Turner (1989: 108) describe as WORDS STANL> FOR THE CONCEPTS THEY EXPRESS. Since we have no other means of expressing and communicating our concepts than by using forms.language ar; well as other communication systems are of necessity metonymic. It is also for that reason that we fail to notice the metonymic character of language. This general FORM FOR CONCEPT metonymy has to be distinguished from a FORM FOR CONCEPT metonymy involving specific signs (sec (47)). The Sign ICM only seems to lead to the FORM FOR CONCEPT metonymy. but not to the reverse metonymy of CONCEPT FOR FORM. A metonymic situation of this kind is difficult to imagine. It may be said to occur in the 'tip of the tongue' experience. in which we have a certain concept but cannot think of the corresponding word-form. or in the context of foreign-language learning. where learners must find a form in the foreign language for a concept with which they are familiar.

2.2. Reference ICMs and reference metonymies Real-world entities may be related to signs. concepts or forms. We thus have three types of Reference ICMs and possible metonymies as shown in Figure Ia. In all three types of reference metonymies. the metonymic target is the real-world thing or event. The standard situation of reference involves signs. i.e., form-concept units. which stand for a thing or event referred to. We thus have the metonymy: (2)

fORM-cONCEPT FOR THING/EVENT:

word cow for a rea)

COW

Strictly speaking. the sign does not refer to the world of reality but to our mental model of reality. Frawley ( 1992: 20fl) illustrates the mismatch be-

Toward.~

a Theory of Metonymy

25

tween these two worlds by comparing the extensional and intensional meanings of the verb to pmu:ll. In the extensional world of reality. an event of punching involves a series ofsubcvcnts: folding one's fist, moving one's arm. bringing it into contact with an object, and recoiling it. A punching event thus has duration. Intensionally, however. to punch is a punctual verb and. as such. cannot be used to describe a durational event a'\ in ""It tookfi\'e minutes to punch him. The projected world we refer to by means of a sign thus only partly matches reality.H We do, however. firmly believe that words refer to the extensional world so that metonymy (2} has psychological validity. Signs can only be said to stand for things or events they refer to - the reverse metonymic situation, THING/EVE."'T FOR FORM-CONCEPT, is hard to imagine. A thing or event may evoke the word denoting it. but it certainly cannot be argued that when we sec a cow and the word cow comes to our mind. the thing 'cow' metonymically stands for the sign cow. The above folk theory of reference has two variants in which either the concept or the form of a sign is focused upon. Lakoff ( 1987: 168f) describes the former situation as "reference via meaning." and the latter as "doctrine of direct reference." According to the Reference-via-Meaning ICM. "words have inherent meanings (called intensiom) and designate objects by virtue of those meanings" (Lakoff 1987: 168f). In this view. the meaning associated with the word cow is assumed to stand for any cow in the world of reality - in contrast to the set-theoretic account. in which 'cow' denotes the set/class of cows. (3)

CONCEPT FOR TiliNG/EVENT: concept 'cow' for a real

COW

The reversal of this metonymic relationship. TI-liNG/EVENT FOR CONCEPT. may occur in special situ non-human: whole> part; concrete> abstract; visible> non-visible;

Towards a Themy of Metonymy

45

etc." Some of the semantic constraints which Cooper and Ross (1975) found to determine the fixed ordering of bipartite expressions are also relevant for default metonymies. 17 The cognitive principles which have been identified here relate to three general determinants of conceptual organization: human experience. perceptual selectivity. and cultural preference. which. however. tend to interact and overlap in several ways. 4.1.1. Human Experience Our bao;ic human experiences are derived from our anthropocentric view of the world and our interaction in the world. In this world. humans take precedence over non-humans, things are looked at from a subjective rather than objective point of view, concrete objects are more salient than abstract entities. things we interact with are selected over things we do not interact with. and functional things are more important to us than things which are nonfunctionaL (i) HUMAN OVER NON-HUMAN. This principle accounts for the default cases of the production. control and possession metonymies. namely (38) PRODUCER FOR PRODUCT (I've got a Ford). (41a) CONTROLlER FOR CONTROLLED (Schwartzkopf defeated Iraq) and (42a) POSSESSOR FOR POSSESSED(/ ha~·e a flat tire). (ii) SUBJJ:X.TJ\'f; OVER OBJECTIVE. Each person's experience is necessarily subjective and leads to a subjective view of the world. The principle SUBJECTIVE OVER OBJECTIVE determines the use of metonymy (30b) PERCEPTION FOR THING PERCEIVED ao; in What a beautiful .~ight for 'thing seen.· This principle may also account for the unidirectionality of the reference metonymies (2) FORM-CONCEPT fOR THING/EVENT and (3) CONCEPT FOR llHNG/EVENT. Here. we access the world of 'objective reality' by means of our subjective world of concepts. (iii) coNCRtTE OVER ABSTRACT. Our basic human experience relates to concrete physical objects, which have more salience for us than abstract objects. The cognitive principle CONCRt.TE OVER ABSTRACT accounts for why we speak of a book wrilten in a careful hand for 'carefully written' or having one's hands 011 sometlri11g for 'controlling something.' Body pans make panicularly 'good' objects. and we routinely access various abstract human domains by reference to our body. Special subcases of the CONCRETE OVER ABSTRACT principle may be described as BODILY OVER EMOTIONAL (heart for 'kindness'), BODILY OVER ACTIONAL (hold )'tJUr tongue for 'stop speaking').

Gilmer Radde11 mul Zt)/tdll Kiivecses

46

BODILY OVER MENTAL (braill for 'inteJJecf) and BODILY OVER PERCEPTUAL (good ear for 'good hearing'). Since concrete objects are visible and abstract

things invisible. the principle also entails VISIBLE OVER INVISIBLE, which is reflected in metonymies such as to .w~·e tme ·.~ .~ki" for 'to save one's life.' Visibility. in its tum. accounts for the default metonymy (43a) coi\ITAJNER fOR CONTENTS,

since things in a container arc typically hidden inside the container

and. thus, invisible. At a more general level. the CONCRETE OVER

ABSTRACT

principle also accounts for metonymies (I) FORM FOR CONCEPT. and (47) WORDS FOR THE CONCEPTS THEY EXPRESS, in which the concrete visual or acoustic shape of a sign stands for the abstract concept denoted by the (word~)form.

(iv) INTERACTIONAL OVER NON~JNFERACT/ON;\L. Our experience of the world mainly derives from our interacting in it. Entities we interact with form better reference points than entities with which we do not interact. We often interact with parts of a whole so that this principle provides a default motivation for PART FOR WHOLE metonymies. For example. the part we interact with in driving is the steering wheel so that we speak of sitting behind the wheel. We mainly use our hands in interacting with the world and hence speak of hamlon demotwration. we use our fingers in typing on the keyboard and thus speak of having the world at our fingertips when we log into the Internet. Our interaction with things is closely related to their function. (V) f"UNCTIONAL OVER NON·f"UN(T/ONAL. As shown by Tversky and Hemenway ( 1984). we attach particular salience to functional parts such as the engine of a car. There is a vested human interest in seeing things. in particular anifacts.from a functional point of view. Thus. the part of a TV-set that appears to be the most functional is its tllbe so that we speak of spending the whole day in front of the tube; the parts of a car that are functional to its driving are the wheels. the motor and the steering wheel so that we speak of a 24-wheeler, a motorway and sitting behind the wheel. Pans which are not important in driving such as the doors. the windshield wipers. or the fenders are. of course. highly unlikely to be selected as metonymic reference points for the car.

4.1.2. Perceptual selectivity A number of cognitive principles are relatable to perceptual salience. which, of course. also determines the way we experience the world as human beings. Our perceptual apparatus is geared towards things in our immediacy and presence. things which are large and big. and things which form good gestalts.

Towards a Themy of Metonymy

47

have clear boundaries and are specific instances. These foci of perceptual selectivity can be stated in the following principles of cognitive preference. (i) JMMEDIATf.' OVER NON-JMMEDJATE. This cognitive principle accounts for selecting stimuli in our spatial. temporal and causal immediacy. The metonymy in /'1/ answer the phone for 'I'll answer the person speaking at the other end of the line' is motivated by spatial immediacy. Metonymies (13) PRESENT FOR HABITUAL as in I alway.~ take the 9 o'clock train and ( 14) PRESENT FOR AITURE as in I am off for 'I will be off are motivated by temporal immediacy. Metonymy (33) EMOTION FOR CAUSE OF EMOTION. as in She i.~ my joy for 'she makes me be happy' is motivated by causal immediacy. Here. the emotion as an effect of some stimulus serves as the metonymic vehicle. In general. effects are more perceptible and affect us more than causes. The immediacy principle also accounts for many emotion metonymies in which physiological and behavioral responses produced by emotions are used to stand for the emotions themselves as in He got cold feet for 'he became frightened' (see KOvecses I 990). (ii) OCCURRf.'NT OVf;R NON-OCCURRENT. This principle reflects our preferential concern with real. factual. and occurrent experiences. It accounlformconcept> concept > reality. Forms rank over concepts due to the principle CONCRETE OVER ABSTRACT. form·concepts occupy a natural intermediate posi· tion between the two, and concepts rank over reality due to the principle SUBJECTIVE OVER OBJECTIVE.

As a rule, several cognitive and communicative principles are involved in a given instance of metonymy. They may be. and often are. in conflict with each other and in this way decrease the naturalness of the metonymy. The following section will briefly investigate the nature of such overriding factors.

5.

Overriding factors

The use of metonymy may be motivated by a speaker's expressive needs or a given social situation. In the former situation. an individual speaker may want to achieve a rhetorical effect by deliberately violating one or more of the cognitive principles in his or her use of metonymy. The latter situation is governed by social norms and usually involves violating the communicative principle CLEAR OVER OBSCURE. Since the cognitive and/or communicative principles arc overridden deliberately. the resulting metonymy is usually also felt to be figurative. or, non·default.

5.1. Rhetorical effects Along with other figurative modes of thought. metonymy is commonly used to produce rhetorical effects as in humor.jargon,literature. persuasion. slang, poetry and the like. The rhetorical effects tend to derive from violations of

Toward.,· a Theory of Metonymy

53

default cognitive and communicative principles. For example, Chaucer's depiction of 'fright' as Myn ltene i.~ .wore afright and the slang expression get out ofmyface for 'leave me alone· violate the principle of GOOD GESTALT OVER POOR GESTALT in that a body part is used to stand for a person. The aesthetic effect of the metonymies in The pen is mightier than the sword derives from the deliberate reversal of the cognitive principle HUMAN OVER NON-HUMAN. Shakespeare's wording Let pride marry her and the journalist"s description Many American li~·e-'" were loJt for 'many Americans died' both violate the principle of CONCRETE OVER ABSTRACT.

5.2. Social-communicative effects Social considerations in a communicative situation may require speakers to override some cognitive or communicative principles. This is especially obvious in the use of metonymy-based euphemisms. in which considerations involving the hearer's 'face' prevent the speaker from using a clear expression. 1M This is the case in the British English expression redundancies for 'dismissals.' The euphemistic word redundancy refers to a precondition that may lead to a worker's dismissal. The metonymy deliberately reverses the cognitive principle CENTRAL OVER PERIPHERAL and, since the target is not clearly identifiable. also violates the communicative principle CLEAR OVER OBSCURE. The expression They clid it for 'they had sex· also overrides the principle CLEAR OVER OBSCURE. but here it is due to its violation of the principle SPECIAC OVER GENERIC. The euphemistic expressions to go to the bathroom and to wa.~h one's hand.~ (for 'to urinate/defecate') focus on the initial and final phases of a complex event and are, thus. motivated by the conceptual principle INITIAL OR ANAL OVER MIDDLE. Since going to the bathroom and washing one's hands are. however. only tangentially concerned with the central and relevant activity. the principles CENTRAL OVER I'ERIPHERAL. RELEVAI'IT OVER IRRELEVANT as well as CLEAR OVER OBSCURE are violated. The two expressions are no longer experienced as instances of metonymic vehicles. Thus, to go to the bathroom is no longer associated with its spatial meaning 'to transport oneself to the bathroom.' but evokes the target sense directly in expressions such as The dog we Ill to the bathroom 011 the living room rug. 1 '~ Metonymic expressions which are no longer felt to mystify a taboo topic tend to be replaced by new nondefault metonymies. This happened to the originally euphemistic word toilet.

GUnter Rudden and Zoltdn Kiivecse.\·

54

which wa'i replaced by bathroom and restroom, which in their tum were supplanted by expressions such a.;;, facilities and comfort station. Violations of the clarity principle also abound in officialese as in eq11al opportllllity employer and other areas that arc prone to using jargon. These types of 'vivid' metonymy have traditionally been studied in rhetoric and literary criticism. In the cognitivist view of metonymy presented here they now appear as non-default ca.~s. in which cognitive principles and/ or communicative principles are deliberately overridden. Since the primary goal of this paper is to isolate the principles which determine default cases. the issue of non-default metonymies shall not be explored any further.

6.

Conclusion

We have attempted to offer a relatively comprehensive and integrated outline of metonymy from a cognitivist point of view. As stated in the beginning, we believe that such a view involves doing at least the following: (i) identifying the ontological levels at which metonymy occurs; (ii) specifying the types of relationships that hold between elements in a

metonymic relationship: (iii) establishing those cognitive principles that explain the most conventionalized or most ·natural' vehicle-to-target routes; (iv) discovering the conditions under which non-default routes can be selected. The paper argues that metonymy is not restricted to language but is a cognitive process which operates within the same idealized cognitive model. Since ICMs may cross-cut ontological realms, we may also expect to find metonymy-producing relationships in and cross-cutting the three ontological realms of concepts, fonns and things/events. We have been able to identify eight ICMs which give rise to 'ontological metonymies.· The metonymy-producing relationships were subsumed under two general conceptual configurations: whole ICM and its part(s) and parts of an ICM. The fonner configuration typically gives rise to metonymies involving things, the Iauer primarily applies to metonymies involving predications. A small number of conceptual relationships only admit metonymization in one direction: the majority of conceptual relationships, however, lead to reversible metonymies.

Toward.~

a Theory of Metonymy

55

Generally. one of these metonymic construals is conceptually preferred. A number of cognitive and communicative principles govern the default selection of the preferred metonymic vehicle. The cognitive principles pertain to the areas of human experience, perceptual selectivity and cultural preferences. The communicative principles include those of clarity and relevance. These cognitive and communicative principles may be overridden for expressive or social reasons. Non-default metonymies. which arise through such overriding factors. violate one or more of the default cognitive and communicative principles, in particular the principle CLEAR OVER OBSCURE. We do not claim that we have carried out this project fully. On the contrary. what we have done here is just the beginning and we are certain that there are others who do not necessarily agree with us concerning either the details of our proposals or even our general claims. However. we believe that the four issues raised are important for understanding the nature of metonymy. If we can come to some consensus regarding these issues. we can address further ones that have not been discussed in the cognitive linguistic literature in the detail or depth they deserve. Among these are the following: I. 2. 3. 4.

5.

What is the precise nature of the relationship between metonymy and other figurative modes of thought. in particular metaphor? What is the function of metonymy. especially in comparison to that of metaphor? Why do people utilize metonymy? What are the restrictions on the use of metonymy? To use Taylor"s (1995: 123) example. why docs the PRODUCER FOR PRODUCT metonymy not apply to Mary was de/idem.\·, meaning by Mary the cheesecake which Mary made? What is the relationship of metonymic speech to literal speech?

To be sure, there are answers to each of these questions in cognitive linguistics. and some of them have been touched upon in the previous pages. However. in light of what we have found. it seems to us that much more can be done. And. indeed. the main goal of this paper was to point out some of the deeper issues concerning metonymy and bring these issues to the attention of the cognitive linguistic community.

Giinler Radden and Zoltdn Kiivec.fes

56

Notes This paper elaborates upon an earlier ver.;ion which appeared in C11KIIitil't' LiiiKIIisticv 91 ( 19911). We are grateful to our anonymous reviewers for their insightful and helpful comments. We also wish to express our thanb to Elizabeth Mathis and Gary Palmer for their valuable linguistic contributions to this paper.

2.

This view corresponds to Dirvcn's (1993: 14) charactcriznlion of metonymy ru; opposed to metaphor:-[ ... ] in metonymy the two domains both remain intoct. but they are seen to be in line. whereas in metaphor only one domain. viz. the target domain is kept. and the other domain. viz. the soun:e domain disappear.;. so to speak."

3.

The attraction of Langackcr's cognitive e,;planation of metonymy lies in the perva~ivc· ness of reference-point phenomena in language structure, in particular posscMive ex pres· sions. The view of mc10nymy a.~ a reference-point phenomenon i~. however. not unproblematic. The pmccs..~ of first making mental oontoct to a reference point before acce~~ing the target ~hould take longer than that of acces.~ing a conceptual entity directly. which. however. has not been confirmed experimentally in terms of proces.sing time (Gibbs 1993).

4.

Sec the discus.~ion of the notion of 'contiguity' in Koch (this volume). The notion of contiguity i~ also present in cognitive definitions a.~ in Croft'~ ( 1993: 347) definition of metonymy n.~ ··a shift of a word meaning from the entity it ~lands for to a ·contiguou~· entity."

5.

For similar views sec Lakoff and Johnson (1980), Lakoff (1987), Croft ( 1993) and Blank (this volume).

6.

The word htuut derives from Latin lhJirpiam 'large rake used a.~ a harrow' and Samnite (hJirpu.f 'wolf,' i.e .. the pins of the harrow were metaphorically related to the wolf~ teeth.

7.

In distingui~hing metaphor from metonymy. Lak.off and Johnson (1980: 36) and Lakoff and Turner ( 19!!9: 103) view metonymy a.~ primarily having a referential function. a.~ opposed to the function of undc~tanding as.~ociatcd with metaphor. But this view of metonymy is at odds with the many other areas in which metonymy is found.

8.

Frawley (1992: 19-20) illustrotcs the difference between the extensional world and thc mentally projected world by the well-known case of tautologies a.~ in George Bush is ThL~ sentence i~ meaningfully interpretable in an attitudinal sense. which may be made explicit by attitude predicates such as wondtr. do11b1. be 1martuin. etc. a.~ in I wondtr if Geor1w Bush is Gtorge B11sh. Gibbs ( 1994: 345-351) ha.~ shown that such tautologies are metonymic in that a category L~ used to refer to specific salient parts or attributes. This analysis is also adopted here (cf. 3.1.(vi)).

Gtorge Bush.

9.

Cf. Langackcr (1993: 31): "it [the octive-zone/profale discrepancy) ha.~ to be regarded a.~ natural and t.f~(·ud rather than pathological." Similarly, Hcrskovit~ (1985: 363) point~ out in her discussion of the salience principle that the metonymic shift to a salient pan "ha.~ come to be so naturulthat it is hard to think that it involves any special proces.~ such as metonymy." Langacker (1991: 191) even argue~ that "II is in fact quite difftcult 10 find convincing example~ [ ... ] where all aspects of the designated entity participate equally in a relationship."

Towards a Theory of Metonymy 10.

57

Cf. the careful distinction between metonymy and synecdoche made by Seto (this volume).

II.

On metonymies relating to the hand, see KOv«ses and SzabO (1996).

12.

S« Langacker (1991: Ch. 3). Prototypical sub.~taoccs are charucterized by intemaJ homogeneity and divisibility. prOI.otypical things are chan~.cterized by internal heteroge· neity and integration of its par1s into a whole.

13.

For a more general distinction between metonymies ba.'ied on co-presence and succes· sian, see Blank (this volume).

14.

The metonymy ACTUAL FOR POTE.,..TIAL aJso accounts for the development of the possibility sense of the modals am and may. The epistemic sense of ,.a,. may have developed from the original sense of 'koow' via 'knowledge how to do and 'being able to act' to

x·

'increased ~sibility of acting.' In a similar chain of sense development, the originaJ sense of may 'might. power' may have led to the modal sense of possibility. Studies on grammaticaJization have convincingly explained the steps involved in these semantic changes by implicature and pragmatic strengthening. discus.~ion

15.

Cf. Lakoff in the

16.

Objects and animals may. of course. also be as~iated with a place. A nice example of metonymic association is the proper name CtJnur;.·l.sftJnds. which goes back to the name Ca11aria given to it by the Romans on account of the many dogs SC(!n there and which later on provided the name for the bird ('anary, which the Spanish found on the islands.

on "Semantic Accommodation" in Cogling. May 7, 1994.

17.

The following semantic constraints

18.

Violation of the communicative principle CLEAR OVER a.~ nouting Grice's maxim of manner.

19.

Cf. Morgan (1978: 263). who anaJyzes this example. which goes back to Robin Lakoff. a~ convcntionaJized conversational implicature.

idcntif~ed by Cooper and Ros.~ (1975) correspond to the cognitive principles a.~ used here: Herr and Now correspond to IMMEDIATE ovat NON· IMMEDIATE, SiiiKllfar correspoDds to SPECIRC OVER GENDIC. AnimtJit' and Agelllil't corre· spond to HUMJ\.'J OVER NON·HUMAN. and C.mnt corresponds to BOUNDED OVER UNBOUNDED. Pos.~ibly aJso the remaining semantic constraints are relevant for metonymy.

OBSCURE

amounts to the same thing

References Cooper, William E. John Robert Ro.~s 1975 World order. Papers from the Para.fes.~ion on F1mctionalism. University of Chicago: Chicago Linguistics Society. 63-111. Croft. William 1993 The role of domains in the interpretation of metaphors and metonymies. Cogniti~·e UliRlli.wics 4: 335-370. Cruse. D. Alan 1986 Lexical Semanric.t. Cambridge: Cambridge University Press.

58 Dirven. Rene 1993

GUnler Radden and Zoltdn Kiivec.fes

Metonymy and metaphor: differcnl mental str.uegies of conccpiUalization. Leuven.fe Bijdrage11 82: 1-28. Frawley. William 1992 lingui.ftic SeiiUI/Ific.t. Hillsdale. NJ: Lawrence Erlbaum. Gibbs, Raymond W .. Jr. 1993 Processes and products in making sense of tropes. In A. Ortony (ed.), Metaphor and Thought. Cambridge: Cambridge University Press. Second edition. 252-276. 1994 The Ponies of Mind: Figurative Tlwughr.LLltlglwf that guide it, metonymy plays an important role in constructing conceptual integration networks. Therefore. the various basic cognitive phenomena that are varieties of conceptual integration - e.g .. framing. provisional category extension. analogy. metaphor. the construction of counterfactual spaces, aspects of grammar - will interact systematically with metonymy according to unifonn principles.

Note We are p:ra1cful1o Dan Gruen for poinling outlhis eKamplc.

References Coulson, Seana 1996 The Menendez Brolhers virus: analogical mapping in blended spaces. In A. Goldberg (ed.). Conceptual Stmcturt:, Di.fcmtrse, and Wnguage. S1anford: Cenler for the Study of Language and lnfonnation. 67-81. 1997 Semanric Uup.f: The Role of Framt>·Jhifting and Conct>plual Blt>nding in Mt>aning Construction. Ph.D. dissertation, UC San Diego. Fauconnier. Gilles 1997 Mapping.f in Thought and Languagt'. Cambridge: Cambridge University Press. Fauconnier, Gilles. Mark Turner 1994 Conceptual projection and middle spaces. UCSD Cogniti,·eScit:llce Technical ReptJrt 940/. San Diego. [Available from http:l/cogsci.ucsd.edu and from http://www.wam.umd.etlu!-mturn) 1996 Blending as a central process of grammar. In A. Goldberg (ed.), Discmtrse and cognition: Bridging the Gt~p. Stanford: Center for the Study of Language and lnfonnalion. 113-130. 1998a Conceplual inlegralion networks. Cogniti1·eSciena 2212: 133-187. 1998b Principles of conceptual integration. In J.-P. Koenig (ed.), Di.Kottr.fe and Cognitio11: Bridf(itlg the Gap. Vol. II. Stanford: Cenler for lhe Study of Language and Infonnation, 269-283 KOvecses, Zoltlin 1986 Metaphor$ tJ/ Anger, Pride 011d Lo1•t>: A Lexical Approoch to the StrUl'tllre of Concepts. Amsterdam and Philadelphia: Benjamins. Lakoff. George 1987 Wome11. Fire. and Dullf(eroll.f Thillf(S.' Wltut Cut~:gorie.~ Re1•eaf ahof/1 the Milul. Chicago: The Universily of Chicago Press.

90

Gilles Fauconnier and Mark Turner

Turner. Mark 1996 The Literary' Mind, New York: Oxford University Press, Turner, Mark, Gilles Fauconnier 1995 Conceptual integration and formal expression. Ml'taphor and Symhofit· Acti\·itr 10/3: 183-203. 1998 ConcePtual integration in counterfactuals. In J.-P. Koenig (ed.). Di.~course and Cognitio11: Bridging the Gap, Vol.//. Stanford: Center for the Study of language and Information, 285-296. in print A mechanism of ucativity. Poetic.~ Today.

Distinguishing Metonymy from Synecdoche

Ken-ichi Seta Osaka City University

1.

Introduction

The purpose of this paper is to address the essential nature of metonymy in relation to some other closely related concepts. particularly ·synccdochc.' 1 Despite the amount of interest shown in metonymy during the past two decades. no precise definition of metonymy ha.mnity On/_1·. A few examples arc from dictionaries and other sources.

12.

In diachronk semantics synecd<Xbe in our sense ha.~ been known a.~ •generalization· and "specialization.' Thus Gccracrts (1994) refel'll to specialization. generalization. me· tonymy. and metaphor a.~ "the cla.~sical quartet.'" which I would like to call 'the cla.~sical trio' in term..~ of synecdoche, metonymy. and metaphor. Nerlich (forthcoming) is clear in appreciating the cognitive difference between metonymy and synecdoche along the line of the present paper. "Metonymy is based on our world·knowlcdge about space and time. cause and effect, part and whole. wherca.~ synecdoche is ba Notes By "not explicitly mentioned" I mean that there i~ no morpheme which rould be taken to represent it. Primer, for instance, would not be considered a mctonym. since the suffix. er can be taken to fonnally - although opaquely - indicate the nominal statu~ of the expres~ion, reflected in the interpretation 'that which prints.'

2.

The tenns 'trigger' and 'target' are adopted from Fauconnicr (19115).

3.

The~

4.

are the Jinks involved in turning count nouns into non-count nouns and vice ver~a. as in tllere is egg on tlrt knife and two ,·offen. pletue. The fonner process is sometimes referred to ;1s 'grinding' and the Iauer as 'ehunking.' Cf. Copestakc and Briscoe (19%).

Nerlich et al. {this volume) sec an unconventional metonymic relation in the compound Tretll'lt people j11ice. fomJCd by her son Mall hew. The connection between Treadt andj11ic·c is that Mallhew wa.~ served a particular juice after having watched a TV play called ''lhc Treacle People.'' In my view,lhe connection fits quite well with the OBJECT·nME pattern: 'that which (practically) co-occurred with The Treacle People.' The expression is ~tmnge. however, in that the play will hardly affect the character of the drink, i.e .. be a salient )ubpartofit.

pet~ple

5.

The ba.~ic structure of both compounds and metonym~ can be expanded. Consider compound within compounds {cowboy IIUI) and metaphors within metonyms or metonyms within rnctaphor.; as in ydlow belly in the sense 'a native of the fens': ·~b who is like ~b who ha.~ a yellow belly {i.e .. a frog).' Cf. Warren {1992: 94-97).

6.

This b a tran~lation from Swcdbh, i.e .. from Ar del ti11Jom /igger i skrirare11?

7.

There are, of cour.;e, other bugs on me.'

H.

An increasing number of linguists do, however, approach metonymy in this spirit. cr. in particular Jackendoff (1997), Nunbcrg {1996), Fa.~s (1997) and Copestakc and Briscoe (1996).

9.

A terminological di~tinction of this kind is not infrequently made. Consider. for instance. Stem (1965). who refer.; to sense shifts developed from the fonncr type of relation as perm111ttlio11s.

pos~ible interpretation~

of {7). viz. 'I am irritated' or 'I have

Bealrice Warren

134 References Bergsten, Nils

1911 A Study of Compmmd Substalllil·es in EngliJh. Uppsala: Almquist and Wikscll. Clark, Eve V., Herbert H. Clark 1979 When nouns surface as verbs.lAnguage 55: 767-811. Copcstake. Ann. Ted Briscoe

1996

Semi·pnxlu~:tive

polysemy and sense extension. In J. Pustejo•;sky. B.

BogurdcV (eds.), Le.tical Semantic.f: Tile Problem of PoJ.v.fem.\'. Oxford:

Clarendon Press, 15-67. Fass.Dan 1997

Processing Metonymy and Metaphor. Greenwich. Connecticut. etc.: Ablex Publishing Corporation.

Fauconnier, Gilles 1985 Mental Space.f. Cambridge. Ma.">sachusens: The MIT Press. Gibbs, Raymond W. 1994 The Poetics of Mind: FiguratiW? TIJOJJgl!t. Langua!{e a11d Under.Utmdin~:. Cambridge: Cambridge University Press. Goossens. Louis 1990 Metaphtonymy: the: interaction of metaphor and metonymy in expre~sions for linguistic action. Cognitive Linguistk.f I: 323-340. Grice, H. Paul 1975 Logic and conve111ation.ln P. Cole, J. Morgan (eds.),Symaxand Semantics. Volumt' 1: Spet'ch Acu. New York: Academic Press, 41-58. Halliday. Michael A.K. 1994 An Jmroductio11 to Funclional Grammar. London, Melbourne, Auckland: Edward Arnold. Jackendoff. Ray 1997 Tile Arcllitecwre of the UJnguage Fund~·. Cambridge. Massachusens: The MIT Press. langacker, Ronald W. 1991 Concept. Image and Symbol. Berlin, New York: Mouton de Gruyter. Leisi,Emst 1985 Praxi.f der Engliscllen St'malltik. Heide1bc:rg: Winter. Leitner, Gerhard 1974 Denomim1fe Verbt1fisienmg im Engfis£·11m. Tiibingen: Niemeyer. Levi. Judith 1978 The S)·nlcu and Semantic.f of Complex Nominal.f. New York. San Francisco, London: Academic Press. Lipka, Leonhard 1988 A rose is a mse is a rose: on simple and dual categorization in natur.d languages. In W. HUIIen, R. Schulze (eds.). Understanding tile Uxicon. TUbingen: Niemeyer. 355-366.

AspectJ of Referential Metonymy

135

Mellenius, lngmarie 1997 11u: Acquisition of Nominal Compounding;,, Swedish. Lund: Lund University Press Nunberg. Geoffrey 1996 Transferofmeaning.ln J. Pustejovsky. B. Boguraev (eds.),Lnical Semantics: The Pmblrm1 of Polysemy. Oxford: Clarendon Press. 107-132. OED 1989 The Oxford Engli.fh Dicti0/101)'. Second edition. Odord: Clarendon Press Stem. Gustaf 1965 (1931) Meaning and Change of Me011ing. Bloomington: Indiana University Press. Ungerer, Friedrich. Hans-lOrg Schmid 1996 An lmrodm·tion to Cogniliw• Linguistit·s. London. New York: Longman. Wam:n. Beatrice 1978 Semantic Pattern.f af Noun-Noun Compmmd.f. Gothenburg Studies in English41. 1992 Sen.fe Dew:lopments. Stockholm: Almquist and Wikselllntemational. 1995 Distinguishing between metaphor and metonymy. In G. Melchers. B. Warren (eds.). Studies in A11glistics. Stockholm: Almquist and Wiks.ell International. 137-151. 1998 What i.f metonymy? In M. Hogg. L. van Bergen (eds.). Historical Linguistics 1995. Amsterdam and Philadelphia: Benjamins. 301-310.

Part II

Historical Aspects of Metonymy

Frame and Contiguity On the Cognitive Bases of Metonymy and Certain Types of Word Formation

Peter Koch University of TUbingen

1.

lntroduction1

II goes without saying that in a rhetorical perspective metaphor and metonymy are usually considered to be close relatives of each other. And in fact. they often get confused. not only by students. but - as to the application of these two theoretical notions - even by scholars. In general. it is metonymies that arc reduced to metaphors. All in all. metonymy appears to be a kind of parente pamu of metaphor (cf. Bredin 1984: 45): metonymy seems less interesting, Jess abstract and is supposed to demand a relatively minor intellectual cffon. Nevertheless, I am convinced that metonymy occurs much more frequently than metaphor and tells us a great deal about our cognitive equipment. So it is worthwhile studying its cognitive bases in some detaiL

2.

Ad hoc metonymy and metonymic polysemy

First of all. let me clarify a terminological problem that constantly arises in connection with metonymy. On the one hand. the term 'metonymy' denotes a rhetorical trope which is applied ad hoc to certain lexical material: compare for instance:

Peter Koch

140

(I)

Lat.

iam proximus ardet Uca/egon 'nearby,(the Jumse oJl Uca/ego11 is already burning' (my translationl (Verg. Aen. 2. 3llf. cit. Lausberg 1973: § 568, lc: my italics)

On the other hand, the term 'metonymy· is used to denote a specific semantic relation between two senses of a given polysemous lexeme. as for instance: (2)

Eng.

bar

<sense 1> <sense~>

<sense,> <sensej>

'counter' 'public house' (metonymic sense in relation to )

To keep things clear. it would be better to call cases like Eng. bar(with respect to its sense\} 'metonymic polysemy.· Nevertheless. there is a close relationship between ad hoc metonymy as illustrated in (I) and metonymic polysemy as illustrated in (2): just as polysemy in general develops through 'lexicalization' of a semantic change ultimately triggered by a contextual and/or expressive ad hoc use of a word in discourse. 2 metonymic polysemy also develops through lexicalization out of ad hoc metonymic usage in discourse - or put the other way around. metonymy constitutes an ad hoc innovation that can potentially induce a 'metonymic change' in the meaning of the lexeme concerned. which thereby becomes (metonymically) polysemous. I will come back to this issue in Section 9.

3.

Metonymy and contiguity

We have to look for the semantic basis of ad hoc metonymy in order to account for metonymic polysemy as well. Traditionally, metonymy (and especially ad hoc metonymy) belongs to the realm of rhetoric. To my knowledge. the earliest definition of metonymy (Lat. denominatio) as a rhetorical trope in its own right is to be found in the Rhetorica ad Heren11ium: (3)

Denominatio est, quae ab rebus propinq11i.f et finitimis trahit orationem. qua possit intellegi res. quae non suo vocabulo sit appcllata.

Frame and Comiguity

141

'Denominatio (i.e., ·metonymy') is a trope that takes its expression from near and clo.~e things and by which we can comprehend a thing that is not denominated by its proper word.' )my translation) (Her. IV: 32.43 =Anonymous 1894: 337; my italics). By re.~ propinquae et finitimae. the author of the Rhetorica ad Herenniwn virtually introduces an associative or cognitive element into his definition, namely the relation of closeness that might remind us of the atlcyyt'~· constituting one of the three relations Aristotle sets up to account for the process of remembering:

(4)

b\0 xui tO fq>E~11c;: Of}QEUo~EV vo~oavtEc;: (mO toU vUv ~ CiUou

nvOc;:. xai 'Of.lolou ~ ivavtlou ylyvnm 1i Uvc:'.t~tVf}Otc;:.

~

toU

atlvcyyv~.

bul toUto

'So we track down the sequence (of our ideas) by starting from the present moment or from something else and from something similar or opposite or close (to it). That is the way remembering comes about.' (my translation] (Aristotle. De memorict et reminiscentia, 45Jb: 18-22 =Aristotle 1975:300: my italics). In view of the fact that the theory of rhetorical tropes and the theory of remembering are, prinlll facie. two very different fields of investigation. the author of the Rhetorica ad Herennium was probably not in any way innuenced by Aristotle. And that is the way the story continued for a long time. On the one hand. we have the classical rhetorical conception of metonymy. more or less clearly involving a relation of closeness (or merely a constant relation of some kind). In the form of the notion of xu.rU.xrpJOt~.the rhetorical theory offers a device for coping even with lexicalized metonymic (or synecdochical or metaphorical) change of meaning (cf. Lausberg 1973: §§562 and 577: for metaphorical xu.TciX(!IJOI~ sec also Koch 1994: 203f. 207). Not very surprisingly. several 19th century historical linguists (Reisig. BrCal. Paul. Darmesteter. Nyrop. etc.) took up the rhetorical basis for describing metonymic (as well as other types of) change of meaning (cf. Nerlich 1992: Blank 1997a: 10-18). On the other hand. we have the modem thread of English sensualist philosophy and asociationist psychology (Hobbes. Locke. Hume. James and John St. Mill. Bain. etc.) relying heavily on the relation of 'contiguity' and, to a lesser degree. on that of 'similarity.' This tradition does not go directly back to Aristotle·s views presented in (4) - despite partly analogous distinctions. 3

142

Peter Koch

A first resolute attempt to bridge the by now traditional associationist relations and the linguistic classification of semantic change (Wundt 1912: 459--627) unfortunately did not dearly outline just metonymy ... Meanwhile. Kruszewski had applied the two associationist relations of similarity and contiguity directly to the functioning of linguistic entities and. thereby, prepared Saussurc's (synchronic) two-axes theory. which opposed ·associative· relations (i.e .. thosc fanned by pure similarity. in absentia) and ·syntagmatic' relations (i.e., those fanned by contiguity. in prae.wmtiu) between linguistic entities (cf. Kruszewski 1884-90: Saussure 1916: 170--180; Holenstein 1975: 143-145: Happ 1985: 37-52). It was only LConcc Roudet ( 1921: 686--692). who. in a most straightforward account, took up as.'lociationist relations and Saussure's twu axes and thereby established the systematics of semantic change as illustrated in Table I. T11blc- I.

S.\·.~temaric.f

11[ .w.'malllif· dlllllf?t' f /...n)lln' Ramlt'rJ 1dea~

wonh

contiguity similarity

To my knowledge, it is here for the fin>ttime that the cognitive comptment of metonymy is explicitly stated in tcnns of associationist psychology (and in accordance with the classical rhetorical tradition): (5)

Chungemems rt!.wlrant d"wre as.wcimion par contigui~t! entre /es idies. Taus Jes changements que ron a appclCs changements par connexite ou par mCtonymie (devcnue inconsciente) appartiennent 3 cctte categoric. (Roudct 1921: 690)

Take our example (2)_ In a public house we nonnally find a counter across which liquor and food are served. Thus. there is a spatial contiguity between the ideas COUNTER and PUBUC HOUSE. Their contiguity must have triggered the metonymic change of meaning (from 'counter· to 'public house') that underlies the metonymil: plllysemy of Eng. bar ·counter: publil: house.' Now let us consider another example {cf. DHLF, s.v_pri.wn).

143

Frame ami Comiguity

(6)

OFr.

Ia priso11 'act of seizing," hence: ·captivity: hence: 'prison.' hence: (Mfr.) 'penalty of imprisonment'

BetWL.'Cn ACT 01' SEIZING and CAPTIVITY there is a temporal and a causal contiguity: and between CAPTIVITY and PRISON there is a spatial contiguity. a-. there is also between PRISON and PENALTY OF IMPRISONMEtfl. These contiguity relations are the cognitive basis of the corresponding metonymic changes {and of the resulting synchronic metonymic polysemies we observe in OFr. Ia pri.w11 'act of seizing: captivity: prison' and MFr. let pri.wm 'prison: penalty of imprisonment'). Roudet's diachronic considerations were not paid much attention to in the discussions of the following decades. which concentrated on synchronic and theoretical issues. A far more influential statement. including thoughts on the functioning of metonymy in general (but not on metonymic change!), is due tu Roman Jakobson (1956). who. inspired by Kruszewski and Saussure. put forward his two-axes theory with far-reaching overall correlations, paralleling contiguity- l')'lllllf:matic axis- {'omhillaticm- Broca ctphaJia- metonymy on the nne hand, and similarity - ptmuligmutic axi::i - selectioll - Wernickr aphasia - metaplwr on the other hand.s The oversimplifications inherent in these correlations have been criticized on several grounds (cf. the criticism in Holenstein 1974: Hpllcll UlfnwrmJ

conllguity -~imilanty

As ·sense· probably corresponds to the Saussurian .~ignifii. we would have to explain metonymic change by contiguity between linguistic meanings. which is not easy to imagine in abstracto. On a more technicaJ level. contiguity between ·sememes· of wurds has been described in terms of operations on sets of semantic features. when treating. for example. rhetorical metonymies or metonymic polysemy (cf. Dubois et al. 1970: 106ff: Martin 1992: 75-86).

Frame and Contiguity

145

According to lhis approach. in our polysemous example (2), we would have to posit a semantic feature (counter} as figuring in the set of features that fonn the sememe 'public house' of the word Eng. bar. Things get even more complicated when we have to describe cases of metonymic change in which the original meaning has not survived - a situation that can occur, sooner or later. with any semantic change (cf. REW. s.v.focus): (7)

Clat. focus 'fireplace·> Vlat.foms 'fire' (d. Fr.feu: lt.fuoco: Sp.fuego, etc.)

Would we have to posit, in this case,a feature (fireplace] as still functioning in the sememic description of Fr. feu. etc.? Obviously, this cannot be a realistic approach to metonymy. Following this line of inquiry. we would be compelled to integrate into our linguistic description of a given lexical item all the infonnation necessary to explain whatever metonymy may occur in this lexical item in the course of - even future language history.ln most cases ofmetonymy.lthink. this method would yield a far too powerful and at the same time an 'overbrcd' linguistic description. Moreover. an intralinguistic solution for contiguity seems inappropriate from the outset. The metonymy Eng. bar 'counter: public house' is possible thanks to our knowledge of public houses and counters and not thanks to our knowledge of the word bar. Similarly, the metonymy Lat.focu.\· 'fireplace: fire' was possible thanks to lhe Romans' knowledge of fireplaces and fire. It is not our knowledge of words (and their semantic features). but our knowledge of the world that detennines contiguities. So metonymy is not a problem of linguistic structure, but a problem that concerns the relation between language and the extralinguistic world. Contiguity has to be considered as constituting a conceptual. extraJinguistic and not an intralinguistic relationship (see Le Guem 1973: 14, 25: Bredin 1984: 52f: Happ 1985: 129: Koch 1991: 284. 1993: 281; Blank 1993: 37, and lhis volume: also lhe notion cotopie simiotique developed by Bonhomme 1987: 46).

5. Contiguity and frames I now come to the psychological problem involved in contiguity relationships. According to Umberto Eco ( 1984: 147), 'contiguity" is a wishy-washy notion:

146

Peter Koch

e concetto abbastanza sfumato I ..)"' (cf. also Weinrich 1987: 107). Brcdin (1984: 47) calls it "a convenient but unrevealing metaphor." Originally. the term 'contiguity' (like Greek mivEyyv~ in quotation (4)) belongs to the conceptual domain of space. Applying it to other conceptual relations (like the temporal or causal one between ACT OF SEIZING and CAMWITY in (6)). seems to involve a metaphor on the metaconceptuallevel: by choosing exactly this term. we conceptualize. on the mctaconceptuallevel. different types of concepiUal contiguity in terms of spatial contiguity. But is this metaphor legitimate on psychological grounds? Or do we run the risk of unduly spatializing cognitive relations of any kind? 7 In my opinion. we can retain the term 'contiguity' because it actually covers a coherent range of cognitive relations. But we are not bound to retain the underlying spatial metaphor. An alternative model accounting for contiguity as a unitary kind of relationship is the 'frame· model. elaborated in cognitive psychology and linguistics in the last two decades (cf.. for example, Minsky 1975: Fillmore 1975, 1985: Tannen 1979: Barsalou 1992: Cordier 1993: 143-147; cf. also the notion 'schema' in Bartlett 1932: 197-214.300304, 311-314). By recurring to frames. we can easily understand metonymic phenomena because frames - and this is a point I would like to stress - are non-linguistic. conceptual wholes. When acknowledging the latter fact. we do not have to overproliferate linguistic-semantic descriptions only for the sake of metonymies. Contiguity is the relation that exists between elements of a frame or between the frame as a whole and its clements. Thus. PUBLIC HOUSE constitutes a frame (Figure I). one of whose clements is the COU/IITER (cf. (2)). '"1-.-1 Ia contiguita

Fig11re /. PUBUC HOU!'Wframe

Frame and Comiguity

With regard to example (7). there is a whose elements is obviously FIRE.

Fig11re 2.

147

FIREPLACE

frame (Figure 2). one of

FIREPUCEjramt'

Similarly, there exists a frame, say. DEPRIVATION OF UBERTY (Figure 3) comprising different elements such as ACT OF SEIZING, CAPTIVITY, PRISON, PRISONER, PENALTY OF IMPRISONMENT, and SO on, that easily accounts for the metonymic changes of OFr. priso11 from 'act of seizing' to 'captivity· and thereafter to 'prison' (cf. (6)).

Figi4Te

3. DEPRIVATION Of" UBERTY fralllt'

Peter Koch

148

Consider also the following, more complex example. In certain cultures at certain times. we lind a MARRIAGE frame (figure 4),comprising elements such as BETIWTHAL,

AANCE(E), TRUST/ENGAGEMENT. SOMEONE who

('unites in matrimony') the

BRIDE

MARRIES 1

and the BRIDEGROOM. who MARRY 2 ('take

each other as WIFE/HUSBAND'), MARRIAGE CONTRACf. VOW/OATH, WEDDING. PRAYER, UNION OF WIFE AND HUSBAND. SET UP HOUSE, MOTIIERHOOD, and SO on.

On the basis of this frame. we can unitarily explain a series of quite different metonymies in different languagcs:M (8)

a. Lat. spon.sus,-a 'fiancC(e).' hence: 'bride(groom)" > VLat. 'husband/wife' (cf. Sp. espo.w.-a: Fr. epoux. -se) b. Du. trouwen 'to entrust s.o. to s.o.: hence 'to betroth.' hence 'to marry 1' (cf. MHG triiwen. NHG tra11en) c. OE weddian 'to engage.' hence 'to marry,· d. Lat. mta 'vows'> Sp. boda(s) 'wedding'Pol. .oilub ·vow,' hence: 'marriage' Goth./iugo 'marriage' (cf. Olr./u(i)ge, Cym.llw, Br./i 'oath') e. (OHG iwa 'law; contract'>) MHG i(we) 'marriage contract,' hence: ·marriage' (cf. NHG Ehe) f. OFr. marier 'to marry 1' =Eng. to marry 'to unite in matrimony.' hence: 'to take s.o. as husband/wife' g. Fr. mariage 'marriage', hence: 'union of wife and husband' and 'wedding' h. Lat. matrimmziw11 (originally) 'motherhood.' hence: 'marriage' (i.e., 'union of wife and husband') >It. matrimo11io 'marriage,' hence also 'wedding' > Sp. matrimonio 'marriage,' hence also 'husband and wife' VLat. casare 'to set up house'>! Sp. ca.wr 'to marry 1' j. Lat. oralio 'prayer'~ Br. eured 'wedding'

Frame and Conliguity

149

Figure 4. MARRIAGE frame

6.

Frames and prototypes

But there still remain some problems with contiguity and even with frames. The original paradigm of associationist philosophy and psychology (see Section 3), which leans heavily on contiguity, takes a rather mechanistic approach.9 Whenever there is contiguity, there has to be an associative link. But is not everything contiguous to everything else'! So why do we not associate everything with something or something with anything or everything'! Frame models seem to be more subtle because they represent non-accidental networks of contiguities. But even a frame model conceived as categorical (something is or is not in a frame: this is inevitable. I think, in Artificial Intelligence) would not serve our purpose. Quite on the contrary: We have to acknowledge (see (2)) that there are perhaps public houses without a counter, which we would nevertheless call bars. We have to acknowledge (see (7)) that there are fireplaces without a (burning) fire and that fire is not confined to fireplaces: and, nevertheless. speakers of Vulgar Latin would have used the word focus to denote FIRE.

Peter Koch

150

We have to acknowledge (sec (8)) that there arc marriages without a wedding. husbands and wives that were never fiance(e)s before. mar· riages without motherhood and vice versa. etc. Nevertheless. all the metonymic changes exemplified in (8) have taken place. So then we have to admit that contiguity relations only hold for ·salient' members of the conceptual categories involved. Frames and the contiguity relations constituting them have 'prototypical' charactcr: 10 A prototypical public house has a counter, and so we call it bar. Prototypically. a fireplace has a fire burning in it. and for Roman people. the prototypical fire may have been that of the fireplace; so they called it joCil.~.

A prototypical marriage begins with a wedding; in certain cultures at certain times. prototypical husbands/wives arc engaged before they get married. etc. It is on the basis of such prototypical frames and contiguities that metonymy works. This should not be misunderstood. Many prototype theorists speak of metonymic 'extension: reducing metonymic change to a prototypical effect within one and the same category. I do not adhere to this kind of purely semasiologically based. 'extended' prototype notion (cf. Fillmore 1982: 32f; Jongcn 1985: 126f: Lakoff 1987: 91-117. 416-461 and passim; Gceracrts 1988, 1997: 21f; Taylor 1989: 99-141; see the astute criticism of Kleiber 1990: 147-183; see also below. Section 7 .I., and cf. Warren 1992: 123; Koch 1995: 37-40.1996a: 129-131. 1996b: 231-234: Blank 1997a: 79-85. 1997b: 89-93). On the contrary .I would like to draw attention to the fact that, from an onomasiological point of view, PUBLIC HOUSE, for instance, although belonging to the same frame, is not a peripheral instance of the category COU!'ITER (nor vice versa). 11 So the two tenns of a contiguity relation. for instance COUI"'TER and PUBLIC HOUSE. are two distinct conceptual categories. both prototypically structured in themselves (figure 5). Strictly speaking. metonymic change involves prototypicality only inasmuch a-. the underlying contiguity relation would not hold but for the salient, prototypical members of one or both of the two categories concerned, and it is then generalized to these categories as wholes. Or, as Simon C. Dik (1977) puts it. these changes of meaning presuppose a sort of 'inductive generalization' (cf. also Geeraerts 1997: 68ft).

Fnmu· and Conriguit.r

151

ftS:III"ci.fiOII.' THIGH

The third type of ad hoc metonymy corresponds to metonymy a.o; a rhetorical trope. •s A Latin speaker. in a very emotional setting. wants to refer to the head of a person he or she dislikes or is angry with. So the speaker does not use lhc neutral word cap11t. but he chooses a more expressive lexical means. the word tesltl 'skull.· It is the figure/ground effect between sKULL (as a salient part) and HEAD (as the whole) that produces expressiveness here (Table 7).

Frame a11d Comiguiry

157

Note the pragmatic differences between these three types of metonymic innovation (irrespective of the fundamental figure/ground constant). In the examples in Tables 4 and 5. it is the hearer (qua virtual speaker) who brings about the innovation, whereas in the examples in Table 6 and in Table 7 it is the speaker. In the cases described in Tablc::s 4 and 5 and in Table 6. the "oddness· of the innovation can be "repaired' on the basis of Grice· s ( 1975) conversatiunal implicatures. whereas in the case described in Table 7 the metonymic trope is meant to be striking and not intended to be "repaired.' Later on, by conventionalization through lexicalization, it loses its striking effect and. hence, its expressiveness.

10. Contiguity, gender change, and word formation So far I have been looking at the role of contiguity in relation to metonymic change of meaning. which. by producing metonymic polysemy. constitutes an important contribution to lexical diversification. Now it is interesting to note that contiguities within frames also enable us to explain other types of lexical diversification. Recall the frame DEPRIVATION OF LIBERTY (Figure 3) that Comprised concepts such as All" Of SEIZING. CAPTIVITY. PRISON. PRISONER. PENALTY OF IMPRISONMENT. etc. and allowed for an explanation of our example (6): as we have seen, several clements of this same frame were designated one after the other by the polysemous item Fr. lc1 priso11 (with feminine gender). Now I would like to draw your attention to lhe item /i pri.wns (with masculine gender). which in Old French enabled speakers lo designate yet another concept. namely PRISONER:

158

Peter Koch

(6') Ofr.

Ia prison 'captivity"-+ li pri,\'OIIs 'prisoner'

This new possibility of designation is ba'iCd once again on a contiguity relation within the same frame. However. it is not simply a case of change of meaning. but the creation of a new lexical item identical with the original one (Ia prison). except for the gender. We have to conclude that frames and contiguities not only account for metonymic change and polysemy. but also for other lexical processes such as gender change. In this area. besides gender change. which is of minor importance, we especially have to think of certain types of word fonnation. Ofr.li prisom (6') has not survived in Modem French. but in competition with it. there existed the item U prisonier.\· 'prisoner." which has survived in Modem French: (6'') OFr.

Ia prison 'captivity' -+ li pri!wniers 'prisoner·> MFr. le prismmier

Semantically. the item pri.wmnier belongs to a type of derivation called 'Ausgriff by Gauger ( 1971: 66--74).1n my analysis. this type of derivation is also founded upon contiguity within frames: a new concept is 'grasped' (and a new word is formed) by 'reaching out' from a contiguous concept designated by a word which already exists. 16 This is a most important insight. Frames and contiguities represent a very fundamental cognitive principle underlying several lexical processes which considerably differ from each other. The processes are: metonymic change of meaning (without any morphological change). gender change. derivation of the · Ausgriff type. and. as we will see in a moment. even a certain type of compositionP Let us consider. as a final example, the f'RUIT-TRI:.E frame (cf. Koch. in print). Ba'icd on the contiguity between FRUIT and TREE. in Italian. we find within this frame the following metonymic change of meaning and the resulting metonymic polysemy: (II) a. It.

limcme 'lemon·. hence: 'lemon tree'

In Spanish and English. on the other hand, we have word formations that arc based on exactly the same contiguity between FRUIT and TREE: (II) b. Sp. limc11l 'lemon'-+ limonero 'lemon tree' (derivation) c. Eng. lemon-+ lemon tree (composition) As (lie) shows. there exists in English a - by the way, very importanttype of nominal composition N 1 + N 2 • in which the concept designated by N 1

Frame and Conliguiry

159

(LEMON FRUIT) and the concept designated by N 1 + N 2 (LEMON TREE) are contiguous (whereas the relation between the concept designated by N, (TREE) and the concept designated by N 1 + N 2 (LEMON TREE) is that of tax~nomic subordination). A purely semasiological approach (see Section 7 .2.) may be able to reconstruct examples like (!Ia), but it would not account for the fundamental cognitive principle underlying examples (!Ia), as well as (lib) and (lie). whose essential conceptual homology can be elucidated only by an onomasiological frame approach.

11. Conclusion In conclusion. when trying to grasp the mechanisms underlying metonymy. it may be useful to integrate four different traditional and non-traditional paradigms of ·cognitive' research: the associationist paradigm, gestalt theory. frame theory, and prototype theory. In certain cases (see Section 9). it is stimulating to take KOnig and Traugotfs (1988) pragmatic strengthening theory into account as well. It is only the integration of these different approaches that enables us to appreciate the nature and the relatedness of frames and contiguities. The salient links between elements of a given frame - as constituting a prototypical conceptual gestalt - are what we call contiguity relations. Along these contiguities, people can produce the figure/ground effects underlying metonymies. But frames and contiguities extend considerably beyond the realm of metonymy: they also help us to understand certain types of gender change and word formation. So. metonymy is a very important - though not the only - rhetorical and lexical device that gives us a clue to the fundamental role of frames and contiguities.

Notes I express my gr.llitude 10 Mary Copple (Berlin) for !he stylistic revision of !his paper. 2.

Concerning !he in1ima1e link belween M!manlie change and poly:-cmy. cf. Br~al (1921: 143ff. 2H4-2117). Wenh (1974: 3710. Koch (1991: 2113). Blank (1993: 31). for me-

Peter Koch

160

tonymy. sec especially Schifko {1979: 148-251). The lexicalization step~ for metaphor are described in Koch ( 1994: 203-209): a general fmmewol'k for lexicalization (and delexicalization) processes within sc:mantic change ha.~ been established in Blank (1997a: 116-130). Note the imponant difCerence between 'innovation' and 'change'

proposed by Coseriu (19511: 44--46). 3.

Cf. Amin (1973: 19-81.espedally 38). The first to explicitly lake up Aristode seem~ to be the Scot~man Thoma.~ Brown, who occupies a somewhat particular position with respect to as.'IOCiationism (cf. ibid.: 12).

4.

Cf. the critical outlines ofWundt's approach in Roudet (1921: 681-686); Nerlich (1992: 77-80): Blank ( 1997a: ISO.

5.

From

Hjelm.~lcv onward.~.

Saussure's rather open concept of 'associative' relation

i.~

replaced by the technical, more specifiC concept of 'pamdigmatic' relation (EiniERI OR-relation of substilution in a given syntagmatic environmenl): cf. Hjelmslev (1963: 33-40); Happ (1985: 52-59). 6.

Cf. Jakob...on (1956: 90f): Holenstein (1974: 81-91); Happ (19115: 75-79. 130--139). According to Raible (1911l),the distinction between semantic(-parodigmatic) contiguity (hilt -thatch) and (semantic-Jsyntagmatic contiguily (/lilt- b11m1 out) is es. ·angel'~ Arab al wa:::ir 'minister: 'mayor' > Sp algullcil 'lower law-court official'

The main concern of the following rencclions is how metonymies can best be classified. In Section 4 I will introduce a couple of older and more recent

Andreas Blank

170

typologies. some of which also include synecdoche or parts of this tropc.3 First, however,( will investigate the psychological foundation of metonymy (Section 2) and sketch its cognitive background (Section 3). These rcncctions arc intended to contribute to a typology of metonymy in light of cognitive linguistics (Section 5).4 As we will see. a cognitive basis is necessary for a better understanding of the metonymic process.

2.

The psychological background of metonymy

In ancient rhetoric. metonymy was already seen as the transfer of a word to a "closely related or neighboring thing" (Anonymous 1954. Rhetorica ad Herrenium IV: 32. 43).5 A similar interpretation is found in modem rhetoric manuals: Fontanier ( 1968 (1830]: 79) calls the different types of metonymy "tropes by correspondence.·· which designate an object by the name of another object, Lausberg (1960: § 565) sees a "real relation" between the meanings of the two words. and Mortara Garavelli (1988: 149) speaks of ··a connection of reciprocal dependency ."6 What kind of relation is this? What gives rise to metonymy? According to Stephen Ullmann (1962; 218ft) the underlying relation is "contiguity of senses." i.e .. an association between (intralinguistic) semantic features of two words. In contrast with metaphor. says Ullmann (ibid.). "metonymy ( ... J arises between words already related to each other.·· This definition. however, raises two serious questions: I. How can semantic features or scmcmes ever be contiguous to each other? The only possibility is to put them next to each other in a representation of word fields. but if we call this contiguity we are confusing the way we conceptualize things with the way they really arc, or - as Coseriu might put it - we arc making a "transitus ab intellccto ad rem" (cf. Blank 1993: 34; for the quotation cf. Coscriu 1958: 9). 2. The second problem is that often there arc not two words being related, but only one lexicalized word which is metonymically used to give a name to an object or an idea that has not been verbalized before. as in (2) or in: (4}

Lpecunia 'cattle (a" a kind of currency}'> 'money'

But even when a concept ha'l already been verbalized. a new metonymy shows

Co-pre.~ence

and Successitm

171

another facet of the old meaning. as in (I} OE sadix 'blessed' > ModE silly 'stupid,' which of course. initially. was intended to be a euphemistic synonym of OE dumb or contemporary synonyms. In all cases. the metonymy relies on extralinguistic world knowledge (cf. Blank 1993.1997a; Warren 1995; Koch, this volume): cattle served as money in former times, angels are God's messengers, and the most positive thing to say about a stupid person is that he or she is blessed by Jesus. 7 Thus. the associative relation is rather between the two concepts BLESSED -STUPID. MESSENGER- ANGEL or CATTLE- MONEY. It is important tO point out that this conceptual contiguity exists before the metonymic transfer is done (cf. Section 3.1.). As Hugh Bredin (1984: 57) says,"( ... ) metaphor creates the relation between its objects. while metonymy presupptJJe,\' that relation." Ullmann's interpretation goes back to a scarcely known article by Leoncc Roudet ( 1921 ). who appears to be the first to have brought the term.'> 'contiguity' and 'similarity' (Roudet says 'resemblance') into historical linguistics. Roudct (1921: 688f) distinguishes between semantic change due to an association "par contigune entre les idees," "par ressemblance entre les idees" as well as "changements resultant des rapports syntagmatiques entre les mots."~ The last type of change is generally called 'ellipsis,' the second 'metaphor· and the first 'metonymy' (cf. Blank 1997b: 157-190, 230-269, 281-302). Roudet's view of an association between 'ideas' is more plausible than Ullmann's 'contiguity of senses' according to our conception of the psychological grounding of metonymy:~ More popular than Roudet' s lucid classification became Jakobson · s ( 1971) application of the terms 'contiguity' and 'similarity' to linguistics (cf. Koch.this volume). The major point of Jakobson"s paper is the construction of a parallel between communicative strategies of aphasic speakers and both metaphor and metonymy. He begins by stating that any linguistic sign can either be 'combined' with other signs or be 'substituted' by others. The first ''mode of arrangement" is called "combination" and is based on the in prae.\·entia contiguity of signs in an utterance. the second is named "selection.. and is based on an in absemia similarity of concepts in a paradigm (1971: 74f). He further states that one type of aphasia (Wernicke and amnesic aphasia) is characterized by a loss of semantic knowledge. the mode of selection being disturbed. These speakers often have recourse to syntagmatic strategies, e.g .• they give a paraphrase of the word they cannot find - or create a metonymy. The other major type of aphasia(Broca) is characterized by a loss of syntactic knowledge.

172

Andrea.f Blank

Thus. syntagmatic combinations become difficult. while selection still works, enabling speakers to give synonyms for the words they want to express and even

create new metaphors (cf. Jakobson 1971: 77-89). According to Jakobson. a disorder in combination leads to strategies based on selection/similarity, which he calls the "metaphoric way.'' whereas a disorder in selection leads to strategies based on combination/contiguity. i.e .. the "metonymic way .. (1971: 90). Thus, by analyzing the strategies employed by aphasic speakers, Jakobson creates a strong interdependence between paradigmatic a-.sociation.

similarity and metaphor on one side. and syntagmatic association, contiguity and metonymy on the other. However. in light of more recent studies on aphasia. Jakobson's view is clearly too simplistic (cf. Happ 1985: 72-89).• 0 Jakobson himself seems to have noticed that his twofold linear interpretation wa-. insufficient. as. in the last chapter of his article. this structure undergoes a fundamental correction from one page to the next here he not only distinguishes between different types of semantic similarity and contrast (such as synonyms. antonyms. metaphors). but he also interprets metonymies such as tlratclr .litter or pm•erty in the sense of 'hut' as a combination of "the positional similarity (i.e .• ·selection': A.B.) with semantic contiguity" (1971: 91). In other words. Jakobson ends up by interpreting metonymy as relying on a ·paradigmatic' association by contiguity! This correction means in every respect a return to Roudet's threefold typology. 11 Both paradigmatic and syntagmatic relations in language can theoretically be combined with the three psychological a"sociations of similarity. contrast and contiguity (cf. Blank 1993: 32ff: 1997b: 131-156).1 2 Contiguity can therefore be realized 'syntagmatically' between words (in a complex word, a phrase or a sentence) and 'paradigmatically' as an association between concepts in the mind. The first combination is the basis for ellipsis (e.g .. notebook comp11ter > notebook), the second can give rise to metonymy. 11

3.

The cognitive and communicative backgrounds of metonymy

3.1. Concepts, frames, scenarios and contiguity In the preceding section. I claimed that the relation upon which metonymies are based is a conceptual one. and that these concepts are contiguous to each

Co-prese11a a11d Sun·essioll

173

other. The nature of this conceptual contiguity has yet to be explained. In metonymy. target and donator (or vehicle) concepts. rather than being isolated, arc quite often parts of greater conceptual networks that have been described as 'frames,' 'scenes,' 'scenarios," 'domains,' etc. (cf. Minsky 1975: Fillmore 1975. 1977. 1985: Croft 1993: for the terminology d. Andor 1985). Frames. scenes or scenarios arc static or dynamic mental representations of typical situations in life and their typical elements. 14 Concepts within frames are related by 'conceptual contiguity." 15 As will be shown later (Section 5.1.). the difference between dynamic and static conceptualizations is of fundamental relevance for a cognitive typology of metonymy. The content and the shape of a frame depends on our everyday experience. on our world knowledge. Beings. things. processes. and actions that generally or ideally occur together are represented in the mind as a frame. A frame is formed by an 'inductive generalization' of extra-linguistic knowledge (cf. Dik 1977: 283: Koch. forthcoming: Section 6.2, this volume). This does not imply that all the concepts of a frame arc to be realized every time it is activated. it is sufficient that they usually are. Figure I shows two typically related English frames, which allow several observations important for the understanding of frames:

FiRure I. Tile EIIRiisll breakfaJI and ltmdt frameJ

Concepts within frames can build a complex network. of contiguity (e.g .. TOAST- BlJITER - HAM - EGG - MILK). but they arc also related to the frame itself (BREAKFAST) and to other contiguous frames (LUNCH). There are even 'intersections' (DRINK. SALT. KETCHUP). When a specific frame is opened or accessed. all concepts that by convention belong to this frame are simuhaneously activated. 1 ~

Andreas Blank

174

Furthermore, this example shows that frames and their contents arc entirely culture-dependent: they not only differ from one linguistic community to another but they can also vary even within the same community. An English breakfast, for example, typically includes bacon and eggs, buttered toast. etc .. whereas in Scotland porridge made of oats is an important part of brcakfast. 17 Finally. these two frames show that even strong contiguity between concepts in a frame does not automatically lead to metonymy .111 Framerelations and frame-networks only build the cognitive foundation necessary for a metonymic transfer. In this case, a salient conceptual relation is highlighted (Croft 1993: 348) and the target concept is verbalized by the word usually related to the concept that serves a'\ donator. As Langacker ( 1993: 30f) puts it, the donator concept serves as a 'reference point' for accessing the target. The metonymic innovation may remain a hapax legomenon or lead to 'lexicalized metonymy· by semantic change when being adopted by a group of speakers or the whole speech community. 19 Thus, we can say that metonymy is a linguistic device ba'\ed on salient conceptual relations within a frame-network. In this frame-network, three fonns of conceptual contiguity are relevant for metonymy: 20 I. Relations between concepts within a frame. as in (5): (5) 2.

(6) 3.

Lpraeco 'messenger'> OSp pregdn. OPt pregiio 'message'

Relations between concepts and the superordinate frame. as in (6): OF travail 'pain'> MF 'work'; ME tramil 'pain' >ModE travel 'journey'

Or even relations between related frames. as in (7): (7)

OF cli.~ner 'to have the first meal of the day'> ModF diner 'to have lunch'> 'to have dinner.'

3.2. The communicative efficiency of metonymies and their lexicalization

Metonymies show a clear-cut cognitive background, which enables us to produce and understand them easily. because the only thing speakers have to share are the same conceptual relations, common world knowledge about how

Co-pre.fena and Succession

175

life is typically organized and how the 'things of life' arc interrelated. This makes metonymies very efficient tools for resolving different ta'iks in communication.21 As we have already seen. metonymies can express new concepts which have not been verbalized before but which are closely related to other concepts in a frame; this was the ca.'ie in examples (2) and (4). We have also learned that they can be used for euphemistic. attenuating speech. This caused the metonymic transfers in (I). (6) and also in (3). where a more prestigious expression was used in addressing an inferior person in order to please him. In time. it became impossible for Spanish speakers to address the 'lower lawcourt official' in any other way than by calling him alguacil. A third possibility offered by metonymy is to renect changes in reality and. consequently. in the conceptual representation of the real world. This was the case in (7). When. in the 16th century. the upper cla'ises of Paris became used to getting up late. they shifted the first (and most important) meal of the day from the middle of the morning to noon. The tcnn diner. however. was kept. The same thing happened a century ago. a'\ people increa.. ingly had to stay away from home all day in order to work and. consequently. moved the most important meal to the evening. Metonymy is further used to tum deverbal nouns into more prototypical nouns, as in: (8)

L prehetnio ·act of seizing someone' > F prison 'captivity' > 'prison'

(9)

F bdtiment 'uct of building'> 'building'

( 10) L civita.f 'citizenship'> 'community of citizens'> OF citt!.I ciuii, Sp ci~tdad, Pt cidade 'town,' 'city' As in the above examples. we often find that names for localities are created by metonymic processes (cf. Koch 1997). Finally. we find that quite often two concepts are strongly related in the speech context so that a single word can accidentally refer to either of them. The shared speech context enables us to understand utterances like The ham sa11dwich i.~ waiting for l!iJ check (Lakoff and Johnson 1980: 35). where a waiter refers to a customer by the dish the customer has ordered. When this conceptual constellation satisfies the communicative needs of a defined group of speakers or the whole speech community. the metonymy or the type of

Andrea.'i Bltmk

176

metonymy (cf. Section 4) becomes conventionalized, as in (5) or in the following examples: (II) Lparare 'to prepare,· 'to arm'> Jtparare. Ptparar 'to ward off. to

repulse' (12) a. Lpficare 'to fold'> Rum a pleca 'to leave' b. Lplicare 'to fold'> Sp 1/egar 'to arrive· (13) F de.ifier 'to break the vassal's oath'> 'to defy'

In (I I) PREPARATION is considered a precondition for REPUL'>ING an enemy; in ( 12a) fULDING THE TENTS in the Balkan shepherd society was connected with LEAVING, while in the marine culture of Spain FOLDING THE SAILS was linked to ARRIVAL.In (13) we can easily imagine that an uucrancc like OF je \'OS desfie, said by the vassal to his lord, pragmatically implied the illocution of DEFYING this lord. Here a performativc meaning is generated by metonymy.22 Linguistic (and even non-linguistic) communication can be seen as a process whereby people try to maximize their communicative success by minimizing their linguistic effort (cf. Sperber and Wilson 1986: 46: Keller 1994: 142f). Metonymy is a response to both dcmands.lt can support communicative success when we use euphemism and it minimizes linguistic effort when we use the same word for concepts that are closely relatedP The selection of the convenient donator concept for expressing the target in question is governed by a number of cognitive principles. GeneraJly. concepts that more closely correspond to our knowledge. that are more salient and/or more concrete. are preferred to less bounded, less important or more abstract concepts (cf. Langacker 1993: 30f: Radden and KOvecses,this volume).

4. Typologies of metonymy Having discussed some of the psychological and cognitive bases of metonymy. we can come to a classification of the different types of metonymy. Let us first consider some traditional and more recent typologies. One approach to classification that can be found throughout ancient and modem rhetoric is to give more or less complex lists of types of metonymy. like INVENTOR FOR THE THING INVENTED, CONTAINER FOR CONTENTS, etc. (cf., e.g.,

Co-prese11a and Sun·essioll

177

Quintilian. JnstitUiionis oratoriae; Dumarsais 1818: Fontanier 1968 [1830); Lausberg 1960: Lakoff and Johnson 1980; Bredin 1984). Another traditional way of classifying has already been referred to in Section I: several types of metonymy are subsumed under a more general type of contiguity, i.e., SPATIAL, TEMPORAL and CAUSAL CONTIGUIIT. Typologies of this kind have been proposed, for example. by Ullmann ( 1962), DuchiiCek (1967) and Schifl:.o (1979). As said above. the problem with this more general typology is that there remain types of metonymy which do not rely on temporal. spatial or causal contiguity. Ullmann ( 1962: 218ft) and Duch3.Cck ( 1967: 123-129) resolve this problem by simply listing those remaining cases. such as KINSHIP REU TIONS. REU TIONS BETWEEN A PERSON AND AN OBJECf, AN INDIVIIJUAL A!'ID A GROUP. A WHOLE AND ITS PARTS and AN ACTIVITY AND RELATED PHENOMENA. Certainly. it has to be questioned whether such a particular list could ever be complete.2"' but a more important problem seems to be the following: temporal. spatial and causal relations are very general"types of contiguity:· which comprise several more specific types: the rest of the list contains. in fact. such specific types! Duch3.Cek and Ullmann thus combine different levels of abstraction. Schifko ( 1979: 245f) groups cases such as these in an extended "causal dimension," which comprises relations like CAUSE INSTRUMENT, AGENT- ACTIVITY, ACTIVIIT -

AF-l·'ECTED OBJECT, ACTIVIIT -

PRODUCT.

It is interesting to note that the types of metonymy in Schifko's "causal dimension" rely on certain case roles (cf. Fillmore 1968) substituting for each other by means of the metonymic process. The idea that metonymy is a caserole substitution is elaborated on by Dan Fass (1991: 43fand Appendix 8) and can also be found implicitly in Warren (1992: 66). Fass analyzes the types of metonymy given by Stem ( 1931) and Lakoff and Johnson ( 1980) and states that substitutions can be bidirectional but that they lead to different resuhsP and that one role (PATIEh"T) can even be substituted for an instance of the same role (e.g .. in PART-WHOLE relations). From the point of view of case role or valence semantics. this is a valid approach. The problem with this theory is that. in order to see the ca-.e-role effect, we need concrete sentences that must contain verbs. Lexicalized metonymies. however. show this metonymic sense without being syntactically actualized. Thus. we can only state that a case-role substitution might have occurred during the past.26 Fass also docs not give a complete list of all metonymic types resulting from case-role substitution. Warren (1992: 65ff. 1995: Section 3, this volume) distinguishes five

178

Andreas Blank

types of contiguity. each of which comprises several subtypes: COMPOSITION > 'sheets'), CAUSATION (e.g., ecstasy 'strong emotion. exaltation'> 'drug'). POSSESSION (e.g .. sceptre 'emblem of regal or imperial power' > 'sovereignty'), LOCATION {e.g., wwn 'densely populated area' > 'town dwellers'), REPRESfNfATION (e.g .. univer,{ity 'building· > 'staff). These five types seem indeed to cover the field of metonymy except for the ASPECT - AITRIBUTE relation. which in Warren (1992: 51--63) is separated from metonymy and called 'implication.' The necessity of this distinction, however. remains opaque, given the fact that the psychological foundation of both is conceptual contiguity, and so it seems more logical to me to subsume 'implication' under metonymy. 27 In summary. we can say that there is a certain tendency to reduce the different types of metonymy to a sma11er number of more general types of contiguity. What is generally lacking in these typologies is an integrated cognitive framework that covers all possible contiguities. Without such a framework as a basis. all lists of types of contiguity and of the corresponding metonymies remain purely empirical. (e.g .. linen 'type of fabric'

5. A cognitive typology of metonymy In this section. I shall try to integrate the theoretical issues on contiguity and frames. as introduced in Section 3, into a typologica1 approach to metonymy different from the taxonomies presented above. Instead of listing types of conceptual relation. the focus will be set on two rather abstract conceptual frameworks that derive directly from the two fundamental ways of conceptualizing real-life situations. viz. as .{tatic frames or as dynamic scenarios.

5.1. Co-presence and succession A typology of metonymy somewhat different from those presented above has been suggested by Marc Bonhomme ( 1987: 58-65). who distinguishes between 'mCtonymies situatives' and 'mCtonymies actancielles,' the Iauer also working on the ground of case-role substitution. What is interesting in the distinction he makes. is that the concepts related in the first group arc spatially and/or temporally 'co-present': they rely on the synchronism of their element'\. The concepts in 'metonymies actancielles' stand in a causal, instru-

Co-prese11a a11d Sun·essioll

179

mental. final or consecutive relation, that is to say, they arc 'successive in time,' although Bonhommc does not explicitly characterize them in this way. With this distinction Bonhommc does not describe actual metonymies. but offers a framework to classify them, or to be more precise: to classify types of contiguity according to the temporal status and the perspectivization of their conceptual relation. This fits in very well with our understanding of the cognitive background of metonymy. We have co-presence in a frame. temporal succession in a dynamic scenario, or even succession between two related frames. 'Co-presence' and 'succession' are two very general issues that hang closely together with two fundamental models of human conceptualization: I. the 'synchronic' model. in which all aspects of a given situation or a ·system' are equally present and where time is excluded; 2. the 'diachronic' model. where the processual. consecutive character of things and events is highlighted. These two 'domains of contiguity' each contain a considerable number of more specific types of contiguity. The latter involve conceptual relations which give rise to concrete metonymies. At first glance. the choice of copresence and succession as superordinated domains of contiguity seems to be arbitrary. I feel. however. that these two meta-conceptual models arc philosophically (cf. Aristotle, CategorieJ 12 and 13) and cognitivcly well rooted (cf. the distinction made in Section 3.1. between frames and scenarios). Their main advantage consists in integrating all possible types of contiguity without any remainder. Co-present relations exist between the ACTORS (people, animals. institutions) interacting in a frame. their ACTIVITY. 1!'-ISTIWMENTS. TOOLS, AFI-"ECil:D OBJECTS or PRODUCTS, the PLACE where an activity takes place. and the TIME at which this activity usually occurs. Another group of co-present relations are typical (essential or implicated) ATIRIBliTES and ASPECTS of persons, objects and activities as well as distinguishable PARTS of persons. bodies. objects and activities. The latter are, in rhetoric, usually subsumed under synecdoche as PART-WHOLE and WHOLEPART relations. But together with the implicated ASPECTS they are psychologically grounded in conceptual contiguity. The difference between this type and other types of contiguity is the fact that here distinct concepts are not just related. but that they 'build' the whole and that the whole 'consists' of them. furthermore, INDIVIDUAL REPRESENTATIVES of a COLLECTIVE BODY and finally

Andreas Blank

180

the

as a whole arc co·presenttypcs of contiguity. Figure 2 shows the co· present conceptual relations detennined so far.

FRAME

FRAME. ACTIVITY and an important STATE OF THINGS appear in the center because ACTIVITY or STATE often build the core of a FRAME and sometimes coincide with the latter. The ensuing list gives examples of metonymic semantic change based on co-present conceptua.J relations.

Examples: OBJECT: L praeco 'messenger' > OSp preg611. OPt pregiio 'message'; F courrier 'postman'> 'mail' (b) INVENTOR - PRODUCT: G Zeppelin 'proper name' > 'long, cylindrical,

(a)

ACTOR -

dirigible balloon filled with gas' 211 (C) ACTOR- TYPICAL ASPECT: QE .~a'Jig 'blessed'> ModE silly 'stupid': OGr aggelos 'messenger'> 'angel' (d) OBJECT" - TYPICAL ASPECT: L pecunia 'cattle (as a kind of currency)' > 'money' (e) OBJECT- FUNcnoN: E target 'linle round shield'> 'object to be aimed at in shooting· (f) ACTIVIIT- OBJECT: L \'estis 'act of dressing'> 'clothes' (g) ACTIVITY -TIME: It \'endemmia 'gathering of grapes'> 'vintage (season)' (h) ACTIVITY- PLACE: F comidie 'comedy'> 'theater (building)" (i) STATE- PLACE: OF prison 'captivity'> 'prison' (j} TYPICAL ASPECT- ACTIVITY: f baiser 'to kiss'> 'to make love' (k) TYPICAL ASPECT- FRAME: ME travail 'pain'> ModE trave/'joumcy'

Co-prnence and Successitm

181

Lfocus 'fireplace'> F feu 'fire' 'tongue'> 'language' INSTRUMENT- ACTOR: f plume 'pen'> 'author' PART- WHOLE: Ptpalavra 'word'> E pala~·er 'talk,' 'conference'; F dme 'soul'> 'human being' (p) PART- PART: L bucca 'check'> VL boc:c:a 'mouth' (q) OBJECT - COLLECTIVE BODY: L civitas 'citizenship' > 'community of citizens' (r) COLLECTIVE BODY- STAll:: L ci\>itas 'community of citizens' >Of cittf 'city'

(I)

PLACE- OBJECT:

(m) (n) (o)

TOOL- OBJECT: L li11gua

Successive relations exist between a STATE and its PREVIOUS and CONSECUTIVE STATE, between an ACTIVITY or a PROCESS and its PURPOSE or AIM, its CAUSE or PRECONDITIONS, or its PRODUCTS or RESULTS. Other succes.o;ive relations exist between PERIODs, different PLACES and,last but not least, related FRAMES. The conceptual types of these successive relations are shown in figure 3. As above. the figure is followed by a list of examples.

PURPOSE AIM

INSTRUMENT

TOOL

PERIOD 1 - - PERIOD 2 - - PERI003 PLACE 1 - - PLACE 2 --PLACE 3 FRAME 1 - - FRAME 2 - - FRAME 3

Figure 3. Succt'J.fion

Andrea.f Blank

182 Examples: (a)

(b)

(c)

(d) (c) (f} (g)

(h) (i)

(j) (k)

ACTIVITY- CONSECUTIVE ACTIVITY: l p/icare 'to fo)d' >Rum a pfeca 'to leave,' Sp 1/egar ·to arrive· ACTIVITY- RESULT: F desfier 'to break. the vassal's oath' > 'to defy'; L prehemio 'act of seizing someone' > F pri.wm 'captivity' > 'prison'; F bcltiment 'act of building'> 'building' ACTIVITY- AIM: F pour 'for (exchanger> 'in order to' CAUSE- AIM: It perchi. Sp porque 'because' > 'in order that" RESULT- CAUSE: L reus 'accused'> Rum rtiu 'bad' PRECONDITION -ACTIVITY: L parare 'to prepare.' 'to arm' > It parare, Pt parar 'to ward off, to repulse'; E to reali:.e 'to make real'> 'to recognize' MATERIAL- PRODUCT: E horn 'hom of an animal' > 'wind-instrument' INSTRUME/'IT- PRODUCT: f plume 'pen'> 'style (of an author)' PERIOD - PERIOD: Sp. mllflana, ME mon,,e, morrow 'morning' > 'the following day· PLACE- PLACE: OL *loukos 'clearing. open space·> L Iucus 'grove.' 'woOO' FRAME - FRAME: OF disner 'to have the first meal of the day' > ModF diner 'to have lunch'> 'to have dinner'

5.2. Three levels of abstraction The two lists of contiguity types arc certainly incomplete. as one might always find metonymies grounded on conceptual contiguities other than those mentioned here. Up to now. however. every type of metonymy I have found belongs to one of the two superordinate domains of conceptual contiguity. as the underlying concepts are either co-present or successive to each other. 29 The typology of metonymy proposed here involves three levels of abstraction (cf. Figure 4): two superordinate 'domains of contiguity' (co-presence and succession). a principally open list of 'types of contiguity' or ·contiguity schemas' (or even 'conceptual metonymies·: cf. Raddcn and Kovecses, this volume).and finally- on the level of the language- concrete metonymies. In our everyday life we observe. experience and Jearn concrete contiguities and we arc subsequently learning and using metonymies that result from this knowledge. For speakers it is important to know which types of contiguity are salient and which are usuaJJy realized by metonymy. This knowledge is

Co-prese11ce a11d Succes.~ilm

183

not only necessary to understand existing metonymies but it is very useful for producing new ones. The rather abstract 'domains of contiguity' are less relevant for speakers than for the purpose of linguistic description. They are kinds of 'meta-frames' which show the same level of abstraction as the conceptual meta-frames behind some types of word formation. 30 Meta-frames generally highlight a more general aspect underlying and structuring different sorts of frames, as shown in Figures 2 and 3.Jl

Fi1.;11re 4. A typofo!{y

ofmetouymie.~

The 'types of contiguity' can best be compared to the 'image schemas' underlying metaphors (cf. Lakoff and Johnson 1980: Johnson 1987), which are highly recurrent and conventionalized. but nevertheless build on an open list. These 'contiguity schemas' are extremely important for the production and understanding of concrete metonymies. The more embodied a certain type of contiguity is in our conceptual knowledge. the more it can be used in concrete metonymies and the higher the chances are that such a metonymy is communicatively successful and perhaps becomes lexicalized within a specific group of speakers or the whole speech community (cf. Section 3.2.). In certain cases. it is only the type of contiguity in question that becomes conventionalized for a group of speakers. e.g., waitresses or waiters. as in the case of ORDERED DISH - cusTOMER. On the other hand. no special dish (not even the famous 'ham sandwich') seems to have become lcxicalized yet in the sense of 'customer.' Some contiguity schemas have restrictions concerning the group of speakers: others have restrictions on the objects that can be metonymically expressed. As Taylor ( 1989: 123) points out. it is. for example. rather uncom-

Andreas Blank

184

mon to use the contiguity schema AliTHOR -

PRODUCT

for referring to a cake by

the name of the baker (in a sentence like Mary wa.~ delidous).n The three lraditional types of metonymy a" mentioned in Section 4 (spatial, temporal and causal relations) arc so to speak "transverse' to the two superordinate types of conceptual contiguity. In both domains we can find spatial relations (e.g .• PART- WHOLE VS. CO!\ITIGUOUS PLACES).

tions (e.g ..

temporal rela-

ACTIVITI -TIME WHEN THE ALnVITY OCCURS VS. PREVIOUS STATE-

CONSECLmVE STATE) and causal relations (e.g .• ACTIVITY - AA-"ECTEU OBJECT VS. PROCESS OF PRODUCTION- PRODUCT).

6.

Conclusion

In this paper I have tried to show that a theory and typology of metonymy first of all has to define the cognitive background within which it works. Furthermore it has to be made clear what is understood by 'contiguity.' Considering this. I come to the following conclusions: I. Metonymy as a linguistic device is the transfer of a word to another concept on the basis of conceptual contiguity between a donator and a target concept. Any spontaneous metonymy can be adopted by the speech community and thus become lcxicalized. 2. Conceptual contiguity results from a relation between these concepts within cognitive frames. between a concept and the frame itself or between two related frames. 3. The conceptual relations in these frames and between frames arc represented as types of contiguity. Any concrete metonymy is rooted in such a conventionalized contiguity schema. 4. All conceptual relations relevant to metonymy are either co-present or successive in time. These two very fundamental aspects in human conceptualization constitute meta-frames which contain typical conventionalized contiguity schemas. The co-presence and the succession domains arc mentally stored abstractions from the schcmas behind actual metonymies. Frame theory and the three-level conception of metonymy make metonymy a very powerful and efficient tool in communication and semantic change. To highlight a given concept we only have to apply the appropriate contiguity schema to the frame in question. In contra'\! to metaphor. no new relation

185

needs to be established (by the speaker) and interpreted (by the hearer), as both frames and contiguity schema. 'pub' (NHG Km~) under the influence of MLG krikh ·pub'). Formal and syntagmatic cnnrrast are the ba~is of aU semiotic system~. Cf. Raible (191U: 6-26): Blank (1997b:l31-156).

13.

Contiguity a~ an as.sociative principle also works in several processes of lexical change other than semantic change. a~ for example in word formation and in the creation of idioms. Cf. Blank (1996. 1997a: 96f. 1998. fonhcoming): Koch (fonhcoming: Section 6.4 .. this volume).

14.

In this sense - but only in this sense - a frumc i~ a prototypical phenomenon (cf. Fillmore 1977: 56; Koch 1995: Section 7), but. in contra~t to Lakoffs theory of ICMs ( 1987: 77-90). I would not ~y that one more or less prominent concept in a frame could be considered a~ a prototypical instance of this frame. In other words: a metonymic transfer like OGr u~gdos ·messenger' > 'angel' docs not occur because an angel i~ the prototype of a messenger. but because 'being a messenger (of God)' is a salient aspect of angels.

IS.

It is obvious that in many ca---C>-C) Fi~:11re

I. Mewplwr jmm metonymy

c®>---C>-C) Fi~:11re

2. From denmic

w epi.wemic

Metonymic

2.

Bridge.~

;, Modal Shifts

195

Research question

With respect to English nuut. the question arises whether there arc sufficient grounds to consider the development of iL'i epistemic sense as a metaphor from metonymy. In the following quotation. Traugott and KOnig clearly argue that the change was induced from contexts which are open to both a deontic and an cpistcmic interpretation and which would therefore give it a metonymic basis: ( .•• ) llllfll in the epistemic sense of 'I conclude that' derived from the obligative sense of 'ought to' by strengthening of conversational inferences and subjectification.lf I say Slle mfl.f/ f,(.' married in the obligation sense, I invite the inference that she will indeed get married. This inference is of course epistcmic. pertaining to a state of affairs that is anticipated to be true at some later time. (1991: 209)

Bybee et al. ( 1994: 196) generally favor the thesis that the overall change from agent-oriented to epistemic "involves the conventionalization of an implicature, by which the inferences that can be made from the meaning of a particular modal become part of the meaning of that modal.'' Also here. in the terminology adopted. we find acceptance of the metonymic origin. In the specific case of English must, however, they are opposed to such an origin: Since the cpistemic usc of 111/W arises in contexb with aspectual interpretations dbtinct from the obligation uses. it appears that metaphor may be at work in this change. Metaphorical change involves a ~hift to a different domain - in this t:ase from the domain of ~ocial obligation~ and physit:al net:essities applied to an agent, to the epi~temic domain that speaks of the necessary conditions under which a proposition can be true. (Bybee et al. 1994: 201)

The argument focuses on the essential difference between a state of affairs that combines with deontic must. which is typically dynamic or at least controlled, whereas for epistemic must we need a 'State.' that is. a state of affairs conceived of as non-dynamic and non-controlled.' As I have pointed out in Goossens ( 1987). the qualification of a controlled/dynamic nuclear predication by such aspcctual notions as Imperfective/Progressive or Perfect gives it a State-character. Traugott and KOnig's example would seem to indicate that there are contexts which are compatible with both the deontic and the epistemic interpretation. But then their example is clearly a constructed one. It remains to be demonstrated whether language use actually provides contexts in which the

196

Louis Goo.Uelfs

two interpretations are simultaneously relevant. and more generally. what the precise basis for the development may have been. In what follows. I will study contexts which arc relevant to a bcucr understanding of this modal shift in must. first in contemporary English (Section 3). then in older language stages (Section 4).

3.

Metonymic bridges for contemporary English must

3.1. The data I have analyzed samples of mu.rt in context from three contemporary data bases: Brown. Lancaster-Oslo-Bergen (LOB), and London-Lund (LOLU). 2

The first two are written corpora (American English and British English. respectively). the third is a spoken British English corpus. Table I arranges these data into five categories:

Decmtic!Nece.uity: This comprises instances which express an obligation (deontic uses) a'i well as a more general necessity. The deontic uses include instances in which the obligation is imposed/backed by the speaker (which is the rule) and instances in which this is not the case, but we have not separated them in our analysis.

Inferable nece.~.\·ity: Another label could be 'objective epistemic' or 'objective inference' (mll.~t expresses an inference which is not defeasible: the speaker draws a 'logical" conclusion: the context contains the necessary information for the hearer to arrive at the inference presented by the speaker). Tramitional: These are indeterminate uses (especially with respect to the difference between deontic and epistemic: the context permits the interpretation to go either way). Epi.~temic: 'Subjective epistemic' or 'subjective inferential' uses: the inference is the speaker's: the hearer cannot be assumed to have the (contextual) information to arrive at this inference 'objectively.'

Undecidable: The context does not permit any conclusive interpretation: these are incomplete utterances from the spoken corpus: instances of this kind arc listed under'?' in the last column of Table I.

Metonymic Bridges in Modal Shifts

197

Tahll' I. Mu.u i11 three comemporary data hauJ

Dcontic/Nct-essity Inferable

Transitional

Epistemic

BROWN (1111/.f/.69)

54

"

48

15

LOB (tmm,75)

2

(11111.ft110t.2)

LOLU (lllll.ft,l21)

60

(111/l.WIIOt.29)

23

52

Before we tackle the instances that are directly relevant to our purposes. allow me to make a number of preliminary observations. For obvious reasons the undecidable ("?')cases do not receive any further discussion. As far as the deontic/necessity examples are concerned. they outnumber the epistemic ones in all the samples (but more clearly in the written ones). Among them there were 23 instances of must not in the spoken data. but only two in the written samples (more specifically in LOB). Of special interest are the four cases of must not which I have assigned to 'subjective epistemic,' because we do not expect epistemic readings of must to occur with twt (cf. Coates 1983: Section 4.1.3). They are all in the interrogative-negative (three out of four come in tags) and have the effect of trying to elicit agreement of the addressee with an inference of the speaker: (2) is a case in point. Neither this instance.however. nor the other examples of this type. are relevant to the issue that concerns us here. (2)

I like to think of those days# and how tough it wa.'i #for the average Englishman#. what a hard life they must have had#- and mustn't there be #. endless stories about this mansion # - 3

It is the 'inferable necessity' and the 'transitional' uses that are our focus. They will be discussed in the next subsection.

3.2. Bridges As can be observed in Table I. there were 19 instances in our sample (out of a total of 296) which somehow provide us with bridges between the dcontic and

198

Louis Goo.Uelfs

the (subjective) epistemic uses. Of these. 9 are examples of 'inferable necessity' while 10 are 'transitional' uses. Let us lirst consider the instances in which both a deontic and an epistemic (inferential) reading are relevant (the 'transitionar uses). As it happens. they all occur in the British English samples. though I do not think that any further significance should be attached to this. (3) and (4) exemplify such transitional uses. (3)

(LOLU)

and conversation. went like this. this sort of conversation (m- have you noticed president II. that. (m- the boiled eggs # at Sunday *breakfast• -always *(laughs)* hard- and president said- ah well- the simple truth is that. if you're going to boil eggs. communally # they must be

hard (4)

(LOB)

''Well, I cannot say that I approve. You must remember that I am

taking your aunl"s hospitality. and. if your plans go right. on entirely false pretences.[ .. .)" In (3) they

m11st

be hard is compatible with an interpretation 'one has to/must

make them hard' (which is dcontic) and with an epistemic reading 'they will of necessity be hard.· The epistemic interpretation is objective rather than subjective (the inference follows from the fact that the eggs are boiled communally. which is contextually also available to the hearer: it is in that sense not a subjective/speaker inference). In (4). you nuut remember can be paraphrased as 'I want you to remember' (deontic). but also as Tm sure you will remember' (subjective cpistcmic: in this case the inference is to be assigned to the speaker). On the other hand. there are bridging instances of a different kind. as illustrated in (5) and (6). (5)

(LOLU)

what Fan didn ·t realize not being a lawyer#. or a Jawyer·s wife #that apparently the wife was put on probation II so that Fan didn't realize that# that she must also have been up before the court# you can·t be put on probation II not unless you're guilty

Metonymic (6)

Bridge.~

ill Modal Shifts

199

(BROWN)

Rossi and Ferguson have been across the street. talking to the kid. They've found some sort of new evidence. a bundle of clothes or something. and it must link the kid even stronger to the crime. Why won't you accept facts? The two kids were together a lot. they ( ... ] Both (5) and (6) present an inference for which the evidence is available contextually. so that the hearer can arrive at the same conclusion as the speaker - although it remains the case that it is the speaker who draws the conclusion (since he/she does the talking). Some five percent of the data have to be accounted for in this way. Although these instances are not metonymic bctw~.--en dcontic and epistemic a-. the 'transitional' cases are. they may nevertheless be considered as stepping stones to the full-scale subjective epistemic use which is typical of present-day English. The diachronic data to be discussed in Section 4 will confirm the transitional character of this type of usage toward a full-scale (subjective) epistemic use of English m11M. The following conclusions can be drawn from this brief discussion: (i) Although the normal understanding of 'epistemic' implies that it is a subjective inference of the speaker (this is. for instance. what is indicated by 'subjectification' in the above quotation from Traugott and KOnig). we also have to consider that inferences may be 'objective.' i.e .. the inferential ground is understood to be available to both speaker and hearer. a'i in instances (5) and (6). Without claiming that these 'objective epistemic' instances provide us with a metonymic bridge between the deontic and the epistemic usage. I would like to hypothesize that somehow they must have played a part in the shift to the full epistemic sense typical of contemporary epistemic Jmlst. (ii) Occasionally an utterance/sentence with must is compatible with both a deontic and an objective, or subjective. inferential reading. A condition for such a 'dual' interpretation is that the state of affairs represented can be taken to be both non-controlled or controlled (or at least controllable). This is the ca'ie for remember in instance (4) (something may spontaneously come to mind. or may be the result of a conscious effort to recall it); the normal interpretation of be hard in (3) is that it is a State. but in the deontic reading assigned to it above this is viewed as controllable (the subject will (have to) treat them in such a way that they become/are hard). Uses like these could be regarded as the metonymic bridge we have been looking for. Armed with these synchronic observations let us now move on to some diachronic data to see whether and how the rise of epistemic 11111.{1 confirms the

Umis Goossen.\·

200

picture that both 'inferable necessity' and 'transitional cpistemic/dcontic' instances play a part in the development of an epistemic sense of English mu.,·t.

4.

'Metonymic bridges': older language stages

4.1. The data Using the categories that proved to be relevant in our investigation of the present-day English data. we have looked at the instances of ""m (mote. mooste, ...) in successive samples of the Helsinki Corpus of English Texts. Obviously the Helsinki samples are too restricted to provide us with the very first instantiations of the new development in focus here (which. moreover. may be expected to have occurred in unrecorded spoken language anyway). but it may reasonably be assumed that such an analysis will help us to understand how obligational/necessity mwil came to develop an epistemic sense.

4.2. Beginnings of epistemic must As it happened. the first clear uses of subjective cpistemic muM were found in sample Early Modern English 3 (1640-1710). The four instances found arc listed here as (7HIO). (7)

(CEDIAR3B'. 457-461) (EMOE3) That was opposed & spoke against with such vchemency by my L. Clarendon (her owne Unkle) as putt himm by all preferments, which must doubtlesse, [{have{] been grcate. as could have ben given him 5

(8)

(CEOFRC3. 180-184) (EMOE3) The manner of doing it. so unsought for and unexpected, must take. with any good mind, more then the thing itselr. and I am sure the sense must always last with mee.

(9)

(CEPRIV3. 1200-1204) (EMOE3) [ ... ] I did thinke never to imploy him, but he being all your tailors I have altered my resolution; beleev I must not have a gown again this scaven yeare this cost soc much: four pound four shillings the outside cost me ready mony [ ... ]

Metonymic

Bridge.~

ill Modal Shifts

201

(10) (CETRAV3B. 342-346) (EMOE3) [ ... )were put to such stress. that had not their Annadocome to their relief. they must have desisted their Enterprize: Upon arrival whereof the new recruits gained so much on the Arab Governors. that[ ... ] In (7). (8) and (9) the epistemic reading is underscored by other epistemic expressions: doubtles.~e in (7),/ am sure in (8}. and be/eel'[= 'I believe') in (9). ( 10), and (7) in its emended version. are counterfactual. For ( 10) the paraphrase may be either would have desi.~ted or would have had to desi.ft: in the second paraphrase the epistemic reading still contains a necessity ingredient, but that does not invalidate the fact that the reading is also the expression of a subjective inference. In all cases the combining state of affairs is non-controlled: in instance ( 10) a~ a result of the qualification by the Perfect aspect. In our search for stepping-stones towards this infrequent, but already finnly established, usage of an epistemic sense in EMOE3. let us now first provide a general survey of the uses of must in EMOE3. as well as in the four samples before EMOE3. viz. EMOE2 (1570-1640). EMOEI (1500-1570). EME4 (1420-1500) and EME3 (1350-1420).

4.3. Must in five successive Helsinki samples Table 2gives the distribution of the different uses of must.classifying them into the categories used for the analysis of the present-day items (deontic/necessity. inferable necessity. transitional. and (subjective) epistemic). For the Middle Tabfe 2. M1ut in fil·e .fucce.ui1·e periods (Helsinki CorpusJ

Wish ME3(1350-1420J (Total: 63) ME4(1420-1500) (Total: 78) EMOEI(I500-1570) (Total: 88) EMOE2( 1570-1640) (Total: 191) EMOE3(1640-1710J (Total: 173)

Deonticl Nel:essity

Inferable Necessity

43

t8

Transitional Epistcmic

70

71

15

174

15

131

22

16

Lo11i.~

202

Goos.wms

English data we need an additional category 'wish.' This usc builds on the older

permission sense: it is clearly recessive in Middle English and disappears completely in the EMOE samples. As such, it will not concern us any further. A first observation is the absolute predominance of the category Dcontic/

Necessity. which is much more pronounced than in present-day English. As pointed out above, it also includes instances which as such do not express an obligation. (II) and (12) illustrate this more general necessity use. (II) (CMGAYTRY. 360-364) (ME4) (... ] for-thynkyngc J>at we hafc of oure syn, with-owtten will of thoghte to tume agayne to it. And l>is sacrament must hauc thre thynges:- Ane es. sorowc in ourc hcrte )lat we hafe synncdc.f ... ] (12) (CEBOETH2. 304-308) (EMOE2) [... )All that is so. long must last & holde togither, as it is one. but must needes perish & decay, whan so it leaves to be;( ... ]

In addition. note that in ME3. ME4 and EMOEI instances, in which the speaker can be assumed to be in authority, or at least strongly associates himself/herself with some other authority/necessity source. are markedly lower in frequency than those where this is not the case. Among those that do

not involve speaker authority. m11M is still regularly used in instances where it is a past tense (is used in past time main clause contexts). a usage which decreases as we come closer to present-day English. The most important observations to be made. however. relate to the other

three categories (inferable necessity. lransitional and cpistemic). As was pointed out in the preceding section, the fully (subjective) epistemic usage does not come in until EMOE3 (four instances). The 'transitional' usage is found from EMOEI onwards, but is lacking in the ME samples. Inferable necessity, however. is well established from our earliest sample (ME3) onwards. and remains so in all the subsequent ones. It is there that inferential must finds its starting-point. as I will try to demonstrate in the next section. 4.4./nferab/e necessity: a 'bridge' to inferential must

Let us first offer a few instances of 'inferable necessity.· one from each period. (13) (CMBOETH. 375-379) (ME3) But I have wei concluded that blisfulnesse and God ben the sovereyn good: for which it mote nede be that sovereyne blisfulnes is sovereyn devynite

Metonymic

Bridge.~

ill Modal Shifts

203

(14) (CMCAXPRO, 348-352) (ME4) The fruytes of vertue ben lmmortall/ Specyally whanne they ben wrapped in the benefycc of hystoryes/ Thenne it muste followe/ That it is mooste fayre to men Mortalle to suffre labours and payne/ for glorye and fame Jnmortalle (15) (CESCIEIB. 729-733) (EMOEI) Also the third square which is N.H.L.P. must ofnecessitee be equal to the square of C.D. and F.B .. bicause those lines be so coupe led that cuery couple are equall in the scucrall figures (16) (CEEDUC2A. 503-507) (EMOE2) I do fully see the evidence of all which you have said. and therefore I must needs be pcrswaded of it. I do heartily thanke God for it,[ ... ) (17) (CEHAND3B. 312-316)(EMOE3) and consequently the bud inoculated in the morning. must be more likely to grow. having the whole days plenty sap to invite it to unite with All these instances of must present a conclusion of some sort. an inference drawn objectively, rather than a subjective epistemic assessment of the combining state of affairs. The state of affairs is a State in (13). ( 15). and (17), a Process6 in ( 14) and (16). Note that supporting adverbials underscore the inferential character: nede in ( 13). ofneces:oitee in ( 15). therefore and 11eeds in (l6).comequemly in (17). where in addition the participial construction adds a reao:;on for drawing the conclusion. Given my claim that 'inferable necessity' provides us with the interface between the categories 'deontic/necessity' and '(subjective) epistcmic,' two questions must be asked next: (i) How do we go from 'deontic/necessity' to 'inferable necessity'? In other words, what is the contextual - the metonymic. if you wish -bridge between these two categories? (ii) What is the transition like from 'inferable necessity' to '(subjective) epistemic'? We tackle these questions in Sections 4.5. and 4.6 .. respectively. 4.5. From 'deontidnecessity' to 'inferable necessity'

Within the category 'deontic/necessity' it is not instances of mu.~t expressing obligation, but rather those in which must expresses a more general type of necessity not rooted in the field of social pressure that must be our startingpoint. The following examples, again one from each sub-period, illustrate this

Umi.~

204

Goossens

type of general necessity. (18) (CMCTPROS, 567-571) (ME3) Wherto and why burieth a man his goodes by his grete avarice.and

knoweth wei that nedcs moste hy dye? For deeth is the cnde of every man ( ... ]

(19) (CMSIEGE, 447-451) (ME4) ( ... (for ~r is no wattur ner i'e cyty t>en flem Jurdane. And when J>ey fynd no freche watur, )>ey most nedys fle, and we wyl folio and sley )'erne all (20) (CEFICTIB. 434-438) (EMOEI) ( ... ] rebuking a wyld roge because he went idclly about. he shewed me that he was a beggar by inheritance - his Grandfather was a beggar. his father one, and he must nedes be one by good reason. (21) (=(12)) (CEBOETH2, 304-308) (EMOE2) ( ... (All that is so. long must last & holde togither. as it is one. but must needes perish & decay, whan so it leaves to be; ( ... ] (22) (CESERM3A, 565-569) (EMOE3)

But now a very acute and scholastical man that would argue that God must needs have done whatever he fancies convenient for the World should be done, might[ ... ] None of these expresses an obligation, at least in the sense that no clear authority source is involved. The states of affairs range from State (in (20); but 'being a beggar' can be taken to be controllable as well as non·controllable) to Process ((18) and (21)) and Action ((19) and (22). but note that in (22) this Action is qualified by Perfect and that as a result the overall state of affairs qualifies as a State). Although none of these is a good instance of an (objective) inference either. I would nevertheless like to argue that an inferential ingredient. be it to varying degrees. can be read into them. In that sense they provide us with stepping stones from '(dcontic )/necessity· to 'inferable necessity.' In all of the instances we find the adverb needs (nedes, IJedy.\·, ueecle.\·). which we also found in some of the 'inferable necessity' cases. In (18) the general necessity expressed is a subclause argument of 'he knoweth'; it is as if this general necessity is presented as inferable from our, as well as the matrix clause

Metonymic Bric/ges ill Modal Shifts

205

subject's, knowledge of the world. In (19) the necessity for them to nee can be inferred from the absence of fresh water (mentioned in the preceding clause). In (20) the necessity for the subject to be a beggar follows 'by good rea-.on' from the fact that he belongs to successive generations of beggars, as is made explicit in what precedes. Also (2 I) and (22) come in a context in which a reasoning is presented. General necessity instances like the ones discussed here can therefore be taken to offer a natural transition between necessity and objective inference/ inferable necessity. 4.6. 'Transitional' uses: from (inferable) necessity to epistemic usage

As wa-. pointed out in Section 4.3., instances which are transitional towards (subjective) epistemic usage do not occur until the EMOE samples, and only sparingly in EMOEl and EMOE2. The following instances are representative of the kind of transition we came across: the first (EMOEI) instance is from an educational treatise. the second (EMOE2) from The Trial ofSir Walrer Raleigh. and the other two (EMOE3) from Pepy.';'s Diary and Tire Earl of £.,·sex's Corre.~pondence, respectively. I add a brief discussion to each instance.

(23) (CEEDUCIB. 71-75) (EMOEI) A childe shall Jeameofthc better of them, that, which an otherdaie. if he be wise. and cum to iudgement. he must be faine to vnleame againe Contrary to what the reader may have been led to expect on the basis of Section 4.4., (23) docs not build on an inferable necessity, but rather on a general necessity with a deontic tinge. since the speaker can be assumed to give his backing to this necessity. Note in this respect that be faine to vnleame againe is a process that may come about in an uncontrolled a-. well as in a controlled way. At the same time. the example can be interpreted epistemically. but the best paraphrase in that interpretation is will rather than mu.~t. In contemporary English. the instance with will would be predictive rather than inferential.

(24) (CETRI2B. 73-77) (EMOE2) Nay, I will prove all: thou art a Monster; thou hast an ("English") Face. but a ("Spanish") Heart. Now you must have Money: ("Aremberg") was no sooner in ("England") (I charge thee ("Raleigh")) but thou incitedst ("Cobham") to go unto him.f ... F

206

Lmtis Goo.uetu

In (24).11ow you must haw Mo11ey is pan of an argumentation in a trial: 11ow indicates the next step in this argumentation. M1ut is used to express the speaker"s inference (as such it is subjective). but the speaker later goes out of his way to give the grounds for his inference. which gives the inference an 'objective' backing. In other words. 'inferable necessity' supports the speaker's inference. (25) (CEDIAR3A. 312-315)(EMOE3) We were full in discourse of the sad state of our times. And the horrid shame brought on the King's service by the just clamours of the poor seamen. And that we must be undone in a little time. This is a nice mixture of inferable necessity and epistemic use (again predic· tivc: here the paraphrase is would rather than lt'il/ because this is reported speech or. more specifically. erlebte Rede 'semi-indirect speech'). A deontic/ general necessity reading is out of the question here. (26) (CEOFFIC3. 735-739) (EMOE3) ( ... ) although in a matter so nice as this yr Excellency must needs bee ye best Judge. I believe ye Parliament is like to sin longer{ ... ] On the one hand. this may express the speaker's (subjective) judgment. but can also be interpreted as expressing a more objective necessity (in line with the presence of the adverb needs). On the other hand we can take it as expressing the speaker's advice. which gives it a (speaker-backed) deontic overtone. Summing up. these transitional uses confirm that the epistcmic usage of must typically builds on inferable necessity 1111m: this is illustrated by (24). (25) and. be it less obviously. in (26). However. there arc transitional cases which exhibit other features. Besides the subjective epistemic ingredient. a deontic interpretation proved possible in (23) and perhaps also in (26). Note also that in three out of four instances. m11st is rather prospective and paraphrasable in contemporary English by will/would. The conclusion at this point seems to be that the rise of epistemic must is rooted in the 'inferable necessity' uses. but that complexes of dcontic and epistemic uses may have played a supportive role. In the next section we add another factor that appears to have been operative to complete this picture.

Metonymic Bridge.\· in Modal Shifts

207

4.7. Subjectification

A factor signaled by Traugott and KOnig (cf. the quotation in Section 2.) is subjectification. For Langackcr this is a crucial ingredient in the semantic make-up of the present-day modal auxiliaries of English: it is this subjectification that turns them into 'grounding predications' (Langacker 1990 and 1991: 6.3).8 Although a detailed account of how must develops into a subjectified item is outside the scope of this paper. I would like to confirm the basic correctness of the claim that subjectification played a part in the rise of epistemic must. More particularly. it can be argued that it is subjectification in the deontic area which paved the way forepistemic must, which, as we have defined the category. is necessarily subjectified (it reflects the speaker's subjective inference). An initial analysis of the deontic/necessity uses in our data revealed that. although speaker-backed deontic uses are already firmly established in the ME samples, it is not until we reach EMOEl that those subjectified deontic instances begin to outnumber the non-subjectified ones in this category. and that in EMOE3. where the first clear instances of epistemic must occur. the ratio is roughly 75% (for the speaker-backed deontic uses) a" opposed to 25% (for the 'not-speaker-backed' deontic/necessity instances). There is no doubt, it seems to me. that the increase and, finally. the predominance of subjectified uses in the deontic area, made the rise of an essentially subjectified epistemic/ inferential must possible.

5. Conclusion (i) The rise of epistemic mu.~t took place against the background of a gradual meaning development of must from a general necessity sense to the expression of inferable necessity. which amounts to an objective. non-defeasible inference. Early instances which are transitional to epistemic muM still show clements of an objective inference. On the other hand. another type of transition is found in instances where a deontic and an inferential interpretation are possible simultaneously. In those ca'\es. however. mu.{t is prospective and parapharasable by present-day English will, rather than tmut. Although they must have contributed to the rise of epistemic mu.{f and indicate that we cannot explain everything in terms of a

Loui.~

208

Goosu11s

single pattern of contextual shift.lhey arc less essential for our understanding of how present-day inferential must developed. What is crucial. however, is

the marked increase in subjectification in the deontic senses. This was an essential factor in the development of subjectified infcrential/epistemic must. (ii) With respect to the research question fonnulated in Section 2. it is clear that our empirical findings speak strongly against a purely metaphorical shift hypothesis. The shifts that we have observed are minimal nnd very gradual. There is no evidence that there was at any given point in the develop-

ment of English mu.~t a single conceptual shift whereby an element from the sociophysical domain was mapped onto the epistemic domain. (iii) More importantly. at least from the point of view of a collective volume whose focus is on the workings of metonymy. the view that must extended its meaning by way of contextualizations in which the deontic and the epistemic sense are intertwined (in other words. the view that the change from deontic to epistemic must is a case of conceptual metonymy) cannot be upheld either. The concatenation that we have witnessed is one that involves shifts that do not reach a cognitive salience which can be interpreted as the conceptualization of one 'clement" in terms of another within the same (complex) domain. Moreover. more than one type of minimal shift appears to have contributed simultaneously. (iv) Finally. we have emphasized the important contribution of a more global shift. namely that of subjectification in the deontic area. This is a conceptualization and grammaticalization operation which eventually affected the inferential domain that had become part of the semantic make-up of muM. This process of subjectification is in need of further investigation. but beyond the scope of what I set out to clarify in this paper.9

Notes My terminology for the differentiation into Furu:tional Grammar (d. Dik l9K9J. ~:oll~ted

state~

of affain; is in line with that of

2.

The samples were and 1996).

randomly by Ludo Lejeune. For details. cf. Lejeune ( 1995

3.

The quOiations from LOLU arc presented in a somewhat simplified form. The diacriti~: sign~ retained arc II. whi~:h marks intonation units. • •, which signals overlap. and those that indicate pauses of varying lcn~ths.

Metonymic Bridges in Modal Shifts

209

4.

The abbreviations for the different texts are taken over from the Hel~inki Corpu~. 1he sources. in the order of their fint occurrence in the example~ quoted a.~ {7H26), are a~ follows: CEDIAR3B =Evelyn, The Diar:r of John El't/yn: CEOFFIC3 = Lctten. Nonprivate: CEPRIV = Lctten, Private: CETRAV38 =Fryer, A Nrw AtTOUfll of &m l"dia: CMGAYTRY = Gaytridge, Dan Jon GaytridKr·.~ Srromn: CEBOETH2 = Elizabeth. Bonhi11.~: CMBOETH "' Oumcer. BO('thi11s: CMCAXPRO = Caxton, The Prologurs and Epilo~:•ws: CESCIE I 8 "'Record, thr Path-way ... of GtmMtrit: CEEDUClA = Brinslcy. Ludw; Literarill:s or thr Grammar Sd1oolr: CEHAND3B = Langford. Plain untl Ftdl lfutrut"lion.~ to Rt~i.ft All Sm1.~ tJf Fruit-trres: CMCTPROS =Chaucer. Tlrt> Par.fOII 's Talr: CMSJEGE"' The SitKt of Jmuulem in Pro.fe: CEACTIB =Hannan. A Cm·eat ... for Conu11..n Cunetors: CESERMJA "' TilloL~on. Srrmmu: CEEDUCIB "' Ascham, tM St·holemustrr: CETRI2B = Tire TrialtJf tht Earl of Em?:c; CEDIAR3A ::: The Diary of Samurl Prpys. For funher details. see KytO (1993).

5.

The notation 1{ .....} J indicates an italicized emendation in the source tell!. Note that the tellt makes sense with and without the emendation; without the emendation we have to read /wen as the equivalent of Modem English IN.

6.

Again in the terminology of Functional Grammar. i.e .. dynamic. but not controlled. Note that in (16) the Process character is the result of pa.~sivization.

7.

The notation (A ..... A) in the Helsinki font in the source text.

Corpu.~

indicates a typographical shift from the basic

8.

For a critical discussion with respect to contempornry m1m, cf.

9.

This

proce~s

Goossen.~

(1996: 2.33)

of subjectification is in focus in Goossens, fonhcoming.

References Bybee. Joan, Revere Perkins. William Pagliuca 1994 Tht> Evolution of Grammar: T~nu. Afprct, and Modality i11 thr Umguaxr.f ofthf' World. Chicago: The Univer..ity of Chicago Press. Coates. Jennifer 1983 The Semamit·s of the Modal Auxiliaries. London: Croom Helm. Dik. Simon C. 1989 Th~ Theory of Functimwl Grammar. Part 1: Tllr Strut·ture of thr Clause. Dordnx:ht. Providence: Foris. Goossens, Louis 1987 Modal shifts and predication types. In J. Van der Auwerd. L. Goossens (eds.), ln.f and Ouu of the Predication. Dordrccht, Providence: Foris. 2137. 1990 Metaphtonymy: the interaction of metaphor and metonymy in expressions for linguistic action. Cogniti\'e Linguistics I: 323-340. (Also in Goossens et al. (1995). 159-174.) 1996 English Modals and Fum·tio110l Models: A Confrolllation. University of Antwerp. UIA. Antwerp Papers in Linguistics 86.

210

Lo11is Goossens

fonhcoming Panems of meaning extension, ··parallel chaining;· subjectification, and modal shifts. In A. Barcelona (cd.). Metaphor and Metonymy at the Cro.uroods. Berlin, New York: Mouton de Gruytcr. Goossens. Louis.Paul Pauwels, Brygida Rudzka-Ostyn, Anne-Marie Simon-Vandcnbcrgcn, Johan Vanparys 1995 By Word of Momh: MetatJhor. Mnrmymy and Ungui.~tic Action in a Cogniti1•e Per.fJ~c·til'e. Amsterdam and Philadelphia: Benjamins. Heine, Bernd 1995 Agent-oriented vs. epistcmic modality: some observations on German modals. In J. Bybee. S. Reischman (eds.). Modality and Grammar in Discouru. Amsterdam and Philadelphia: Bcnjamins, 17-53. KytO. Merja {cd.) 1993 The Hebil•ki Corpu.f of English Te:rts. University uf Helsinki. Dcpanment of English. Langacker. Ronald W. 1990 Subjectification. Cogniti\'e Linguistic.~ 1: 5-38. 1991 Foundation.~ ofCogllitil·e Grammar. Volume II: Descripti\·e Application. Stanfonl. California: Stanfonl University Press. Lejeune. Ludo (ed.) 1995 Databases for the Study of EtiRiisl! A1uiliarie.f. Volume 1: The Modal.~ i11 Outline. Wilrijk, Dcpartemcnt Gennaanse UIA. 1996 Database.fJor tile Smd.\· ofEngli.dr Au.tiliarif'J. Vol11me II: The Modals in Spoke~~ British En~:li.th, Wilrijk. Departemcnt Gennaanse UIA. Sweetser, Eve 1990 From Etymo!OJO' to Pragmatic's. Cambridge: Cambridge University Press. Traugon, Elizabeth C. 1989 On the rise of epistemic meaning: An example of subjectification. LJ11· gua~:e 65: 31-55. Traugott. Elizabeth C., Ekkehard KOnig 1991 The semantics-pragmatics of grammaticalization revisited. In E.C. Traugott, B. Heine (eds.), Apprnache.f to Grammatkali:otion. Volume I. Amsterdam and Philadelphia: Benjamins, 189-218.

Metonymy in Onomastics

Olaf Jakel University of Halle

1.

Introduction

How are surnames generated in German and in other languages? Are there

principles or patterns after all in the realm of anthroponymy? Or can we at best reconstruct the etymology of each single name. finally to establish overall arbitrariness? Like all proper names. surnames belong to a special class of linguistic expressions that is in marked contrast to appellatives. Unlike these. their semantic function lies not in Lhe categorization of objects or in descriptive conceptualization. but in the naming and identifying of individuals. With several people bearing the same first name in one community. unambiguous reference to a ccnain person became a problem. The solution was found in giving by-names, e.g., 'James .fOil of Zebedee.' 'John the Bapti.\·t: 'Richard Uonheart.' 'Erik the Red." or 'Thomas Aquinas.' 1 In bynames like these lies the origin of surnames as we know them, as Adolf Bach confirms: "Our surnames arc by-names inherited from ancestors .. (Bach 1952: 231: cf. Reaney 1967: 190.2 The custom of hereditary surnames did not spread over Germany until the twelfth century. taking almost five hundred years until complete proliferation: ''The custom of bearing a surname was not generally established in Germany until the year 1600 .. (Gottschald 1982: 47). In the field of onomastics. the notion of semantic motivation has always been more than a mere working hypothesis.lt is simply regarded as a ··fact that proper names, at least at the time of origin. are more or less motivated" (Naumann 1987: 16). Any original motivation. though. may later vanish from the conscious knowledge of the linguistic community.' This is why. from

0/afliikel

212

today's point of view. most onomasticians sec proper names as having etymological meaning rather than lexical meaning:' It is this etymological meaning of surnames that we will be dealing with in the following account of various naming principles. All surnames used as examples in this paper have been taken from the Hamburg Telephone Directory of 1993/94 (Deutsche Bundcspost Telckom

1992). which served as corpus material in this investigation. Aflcr a short glance at surnames motivated by 'genealogy· or 'profession· (Section 2) we will tum to metonymic strategies of surnaming after certain 'utensils' or 'qualities· (Section 3). Section 4 focuses on surnaming motivated by 'location

metonymy.' which shows a rich structure of subtypes that can be analyzed profitably from a cognitive linguistic perspective. Section 5 presents a sum· mary of the suggested taxonomy of motivated surnames. As a final point, the evidence of this onomastic investigation is brought to bear on the controver· sial discussion of the role of metonymy currently taking place among cogni· tive linguists.

2.

Motivated surnames

At the outset of an investigation of motivated names it must be noted that in German there is quite a number of common surnames which have to be regarded as semantically unmotivated: (I)

Grw1lf. Techert. Voel;:, Guhr. etc.

Even with the help of etymological research. the history of these and many other surnames remains pretty much in the dark. Though working on the general assumption that all surnames were once motivated. onomasticians find it impossible to reconstruct the original motivation of a number of names:~ The reasons must be seen in multiple naws of diachronic transmis· sion involving articulation or pronunciation. auditory understanding, and spelling. This supplies us with a residual category for those names whose semantic motivation is not provided by any of the naming strategies discussed in what follows. As this investigation is mainly concerned with the 'principal patterns' of naming. we will deal with surnames which do not necessitate longwinded reconstructions of etymological meaning. The naming patterns demonstrated

Metonymy in Onomastics

213

by these examples surely account as well for surnames whose etymology is less transparent. Detailed analyses of such names, though. can be found in the specialized onomastic literature (cf. References). Probably the oldest strategy of motivated surnaming makes use of the descent of the name~bearer. This 'genealogical' naming simply gives the first name of the father as a surname to his son: (2)

Werner. Thomas, Stephan. Christopher. Paul, Hein:.. Heinrich. Karl. Fran:, Jcma.\·, Lore11:, Martin, etc.

This group also comprises some supposedly unmotivated names such as Arendt ( .mbject >other

This hierarchy makes two claims: firstly. verbal polysemies of the kind exemplified in (3) concern the direct object (if there is one). otherwise the subject. In very few cases they may affect other arguments. Secondly. syntactic constructions that inherently involve contiguity allow metonymic reference more easily for the direct object (if there is one) than for the subject. Note that the hieran::hy does not make the general claim that the object sustains metonymic reference better than the subject. I do not intend to affirm that. say, a subject cannot sustain metonymic reference if it appears in a transitive sentence. Already the example (2a) makes obvious that this would not be the case. The following sections will explain the hierarchy (4) in more detail.

3.

Role-level contiguities: verbal polysemy

Let us begin with role-level phenomena. They cruciaJly involve thematic alternations insofar as the roles selected by the verb are modified. As will be seen. any kind of polysemy-based thematic alternation observes the following order: if there is a direct object. the polysemous alternation applies to the direct object. If there is no direct object. the alternation nearly always applies to the subject. In the entire French valence dictionary of Busse and Oubost (1983), I have found only one counterexample to this rule. 5

Grammalical Consrrainls on Me/onymic Refere"ce

237

3.1. Frame-based alternations

Let me exemplify my hypothesis with the following data: (5)

a.

Marc va balayer ta chambre. 'Marc will sweep your room.' b. Marc n·a pas encore balaye les debris de verrc. 'Marc hasn't swept up the broken glass yet.'

Here we have a container-contents polysemy in the direct object: the things that are swept up and the room that becomes thereby cleaned. (6)

a.

Lc public a bisse cette chanson. 'The public called for an encore of that song.' b. Les spectateurs ont bisse sans succes le fameux groupe de rock. 'Without success, the spectators called the famous rock group back for an encore.'

With this verb, too, the direct object allows semantic variation: the artist who is supposed to give an encore and the song slhe is supposed to sing. These roles are semantically contiguous, but neither of them activates a metonymy in the sense that they have to be understood non-literally. Accordingly, the verb hisser can be treated as polysemous.6 The same kind of alternation of roles in the direct object position shows up in the following sentence pairs: (7)

Contiguity of game and player: a. Elle a gagne Ia partie. 'She has won the game.' b. Elle m'a gagnC. ·She has beaten me at the game.'

(8)

Contiguity of observation and looked-out object: a. Elle a epiC tcs all&!s et venues. 'She has spied upon you.' b. Je pao;se mon temps a Cpier une occasion favorable. 'I spend my time on the lookout for a good opportunity.'

(9)

Contiguity of dangerous action and endangered thing: a. Elle allait risquer une demarche fatale. 'She was going to risk taking a fatal step.'

Richard Waltereit

238

b. Vous devriez nc pas risquer votrc reputation. 'You should not risk your reputation.' ( 10) Contiguity of sound and resounding room: a. Ta voix rCsonne dans lc couloir. 'Your voice echoes in the corridor.' b. Cette salle rCsonne trop. 'This hall resounds too much.·

An extreme example comes from English to teach: to teach (mathematic.~/ SumlaJ· school/little children/third grade) (Langacker 1987: 269f), where we are confronted with the fourfold contiguity of the 'school subject,' the 'institution: the 'type of Ieamer." and the 'school grade' in the direct object. If there is no direct object but nevertheless two semantically contiguous thematic roles in the verb's argument structure, then this contiguity nearly always applies to the subject. as is the case of the intransitive predicate in ( 10). The roles of the sound and the resounding room arc contiguous. This results in a verbal polysemy with respect to the subject. The contiguity in all these cases relates to the respective verb's frame or domain (cf. Croft 1993; Koch 1997 and this volume). The sound and the resounding room are entities that belong to the encyclopaedic knowledge activated by the verb rhmmer 'to resound: Why do verbs have these polysemies? A scmantic·pragmatic motivation for the polysemy in. e.g .. (9) could run like this: the 'endangered thing." the 'dangerous thing' and an 'actor' are likely to co.occur whenever people speak of some dangerous action. We may call this the 'frame of the dangerous action: Now, people may wish to verbalize only part of this frame. It is therefore natural that languages should provide them with tools that enable this partial construal. One of these tools is apparently verbal polysemy. The verb allows two different construals. each selecting different entities from the respective frame. The contiguity of the 'dangerous thing' and the 'endangered thing' ensures successful communication and provides the motivation for the polysemy. This is the overall picture emerging from examples (5) to (IO).In these examples, we arc dealing with a contiguity relation between two kinds of thematic roles. The contiguity relation in question is a very specific one, as defined by the verb. 'Specific' means that the contiguity crucially relates to particular properties of the frame activated by the verb. It does not. or not primarily. relate to general properties of action verbs. process verbs. etc. As the semantic nature of these alternations hinges to a large extent on the frame

Grammatical Constrain/.\" on Me/Onymic Reference

239

embodied by the respective verb, we do best to calllhem frame-based ahemations.

3.2. Locative alternations Regularities of a slightly different kind are found with the locative and 'swann"-altemations that have received a lot of attention especially in the generative literature (cf. Fillmore 1968; Anderson 1971: Baker 1988; Olsen 1994). These can always be reduced to some kind of 'local' relation. Locative alternations are of the type illustrated in (II) and ( 12): ( II) a. b. ( 12)

lis ont charge du charbon sur le bateau. 'They have loaded coal onto the ship.' lis ont charge le bateau de charbon. 'They have loaded the ship with coal.'

II faut substituer ce chiffre par un autre. "You have to replace this number with a different one.' b. II faut substituer ce chiffre a un autre. ·You have to put this number into another number" s place.'

What is aJtemating in lhese alternations? The object position sustains two different semantic roles. These two roles arc clearly contiguous to each other. they establish a (be it metaphorical) CONTAINER-CONTENTS relation. This is obvious in (II). where the ship and the coal are actually regarded as a container and its contents. It is less obvious but nevertheless plausible in ( 12). The verb suh.f/it11er 'to substitute' evokes the idea of a base from which the substituted thing is detached as a profile (in the sense of Langacker 1987: 183189). The highlighted profile is part of the base: one can only substitute parts in wholes. The PART-WHOLE relation is in tum a relation of 'topological inclusion.' a CONTAINER-CONTENTS relation (Winston et al. 1987: 431-435). Note that in these cases it is only possible to speak of polysemy with respect to the direct object position. Only in the direct object position are there alternating contiguous roles without any surface indication of the alternation. The oblique objects (.mr le ba1ea11 "onto the ship.' de charbon 'with coal.' etc.) present alternating roles that are in a relation of contiguity, too, but formally indicate the semantic variation by a varying preposition. The formal marking of the oblique object prevents these from being semantically ambigu-

Ric/rani Waltereit

240

ous: and ambiguity would be a prerequisite for polysemy. The ambiguity condition is indeed crucial for my argumentation. Only if an argument of a given verb is ambiguous with respect to its thematic interpretation can it be analyzed as the diachronic resuh of an insertional metonymy. Recall from Section 2 that role-level contiguities are considered as lexicalizations of a non-

literal (metonymic) reading of a detenninatc thematic role. The contiguity of the oblique objects (sur le bateau. de charbon. etc.) cannot be the direct diachronic result of a metonymic interpretation. because their semantic relation to the predicate is formally marked and is therefore unlikely to result from

a non-literal reading.

3.3. The 'swarm'-alternation The so-called 'swarm'-altemation owes its name to the famous example Bees are swarming in the garden/The garden is swarming with bees. The point of this alternation is that two thematic roles can be coded either in the subject position or in a prepositional phrase. Interestingly. the contiguity relation between the two roles is usually a CONTAINER-CONTENTS relation: (13) a. L'eau dCborde de Ia baignoire. 'The water is flowing over the bathtub.' b. La baignoire debordc d'eau. 'The bathtub is overflowing with water.· In (13) there is clearly a contiguity between the contents, the water, and the containcr.thc bathtub. (14} a. La pluie ruisselait sur lcs murs. 'The rain was streaming down the walls.' b. II ruisselait de cold-cream. de sueur ct de vin. (Flaubert) 'He was streaming with cold cream. sweat and wine.' The CONTAINER-CONTENTS relation is less obvious here. but at any rate we arc dealing with a relation of local contact with some surface. The 'container' is, so to speak. 'flat.' With both verbs. the subject can express both the contents and the container. This flexibility has to be explained by polysemy. For a monosemous interpretation to be possible. we would have to lind one label that would appropriately characterize the selection restrictions of the subject. But this

Grammatical Constraints on Mellmymic Refere"ce

241

turns out to be impossible. We cannot identify a reasonable criterion that would integrate both, say. the qurlque dU)sddl' quelqu'm• 'inherit slhJ10 be s.o.'s heir.' In this case. the contiguity of the inheritance and the heir applies to a prepositional object.

6.

An anonymous reviewer objected that these ahemations might simply boil down to active-zone phenomena (Langackcr 1984). Standard examples of active zones are We all trlullfJtt or Don't 1'\"f'r belie\'f' Gl'fald. The semantic content of their direct objects does not precisely coincide with lhe entity designated by these object~. which participates most directly in the process. A more 'literal' version would be WI' all lll'ard lht' sound of tire lrrunpel or Dmr'l e\'t•r ht'lien• what Gt'rald says. Consequently. activezone phenomena overlap with metonymies (cf. Langacker 1993: 31). But there is a crucial difference between active-zone phenomena and the alternations (S) to (10): wherea~ active-zone examples are felt a~ non-literal speech and can be paraphra~ed by a synonymous 'literal' sentence {a~ in the cited examples), frame-based alternations cannot. Take examples (Sa) and (b): the variant~ of hula.'''-'' present slight but important differences in meaning. &layer 'to sweep' in (Sa) ha~ an incremental theme {the cleaned room. place. etc. gets clean 'step by step') and this necessarily implies duration. 1be variant {Sb). however, ha~ a holistic theme (thing.~ may be swept up 'all in one go') and it does not necessarily imply dumtion. Accordingly. it is not always possible to replace the variants one for each other {which again indicates that they actually differ in semantic content). On the one hand, someone who has 'swept up something' ha~ not always thereby 'swept a place: On the other hand, someone who has cleaned an already clean place (i.e., without ~weeping up anything). can nevertheless claim to 'have swept it.' hf'ard the

7.

This (ungrammatical) construction nJUst, of course, not be confused with the impeccable one u prltrn·uper dt: qudq11t: d~tm! 'to worry about something.·

II.

The I'AMT·WHOLE relation is even iconically reflected in the reflexive clitic's slight morphological and phonetic weight. as was suggested to me by Peter Koch. Even on the level of linguistic form, the reflexive's antecedent is bigger than the refle:ooive ibelf.just a~ the whole is bigger than one of ib part.~. Interestingly ,the a~ymmetry between the refle:ooive and its antecedent fades away a.s the reflexive marker grows in phonetic shape (which again confmns the iconic nature of the part-whole relation in question). Prion·uper-type constnr.ctions arc sometimes bencr with a strong reflexive than with a reflexive ditic:fr mt' rlmq11t: moi-mime is bcner thanjt' me t·hoqur 'I shock my !>elf {cf. Bellcni and Rizzi 191!1!: 297f for analogous examples in Italian).

9.

Languages may differ considerably with re~pect to the semantic range permitted by the ~ubjcct and the direct object. Clearly. argumcnb such a~ ~ubjcct and direct object are language-specific categories. The semantic spectrum of a given argument is therefore highly dependent on particularities of the language's grammar which could not be studied hcre.e.g .. lhc case system, won! order rules. elc. The flexibility for metonymic reference is also part of 1hese language-specific propenies. Cf. Hawkins ( 1986: S3-73J for differences of this kind between English and Gennan. and MUIIer-Gotama 1994 for a cro.~s linguistie study on semantic transparency in argument linking.

10.

Koch 19111 offers an articulated theory on abstractness vs. concreteness of descriplion of themalic roles.

Gramnullkal Constraints on Metonymic" Reference

251

References Anderson, Stephen 1971 On the role of deep structure in semantic interpretation. Foundation.f of UJnguage 1: 387-396. Baker. Mark 1988 Incorporation: A Tlreory of Grammatical Function Changing. Chicago: The University of Chicago Press. Barsalou, Lawrence W. 1992 Frames. concepts, and conceptua1 fields. In A. Lehrer, E.F. Kittay (eds.). Fromt's, Fields. and Cm1cepts: New Es.m_\"S in Semantic and l..Lximl Orga· ni;;;ation. HillsdaJe: Erlbaum, 21-74. Belletli. Adriana, Luigi Rizzi 1988 Psych-verbs and -theory. Natural Lnngrwge and Um:ui.wic Tht'ory 6: 291352. Blake, Barry J. 1984 Problems for possessor aM:ension: some Austr.:alian ex.amples. Ungui.~tic.f 22: 437-453. Blank. Andreas 1997 Prin:Jpien des le.tikalischen Bedeutung.nmndeb am Beispiel der roma· ni.n·l!en Spra,·hen. Tilbingen: Niemeyer. Busse, Winfried, Jean-Pierre Dubost 1983 Franzfisischt'.f Verhk1·ikon: Die Kon.~truktion der Verhen im Franzfisischen. Second edition. Stuttgart: Klett. Brtal. Michel 1897 Essai dt' simantiq11e. Paris. Croft, William 1993 lbe role of domains in the interpretation of metaphors and metonymies. Cognitive Ungui.rtit's 4: 335-370. Fa.~s. Dan 1991 Metonymy, case role substitution and sense ambiguity. In D. Fa.~s et al. (ed~.). Proct'edings of the /JCA/-91 Worbhop on Complllationol Appr()(lc/res to Non-literal Lnnguage. Boulder, Colorado {Technical Report CU-CS-SS0-91). 42-51. Fillmore, Charles 1968 The case for ca.~.ln E. Bach, R.T. Harms (eds.). Unil·ersals in Ungui.Jtic Throry. New York: Holt. Rinehart and Winston, 1-88. 1975 An alternative to checklist theories of meaning. Proc~t'dings of the First Confert'nce of the Berkeley Ungui.Jtic.r Sociery·. Berkeley, California: BLS, 123-131. 1976 Frame semantics and the nature of language. Annab of the New York Academy of Science.f 280: 20-32. 1977 Scenes-and-frames semantics. In A. Zampolli (ed.), Unguistic Struclllre.J Proce.uing. Amsterdam: Nonh-Holland, 55-81.

252

Richard Waltereit

1985 Frames and the semantics of understanding. Quaderni di Semantica 6: 222254. Frci.Henri 1972(1939) Sylvie est jolie des yeux. In A. Sechehaye (ed.), Mllange.f de lilrguistique offeru il Clrarle.'i Ball)·. Geneva: Slatkine. 185-194. Giv6n, Talmy 1984 Syntax: A Fum·timrai-T.\'pologicallmroductimr. Vol. /. Amsterdam and Philadelphia: Benjamins. Grimshaw, Jane 1990 Argumem St11cture. Cambridge, Massa~:hussets: The MIT Press. Hawkins. John A. 1986 A Comparatil'e T_1·pology of English and German: Unifi.o'ing the Contra.m. London: Croom Helm. Jackendoff. Ray 1992 Mme Tussaud meets the binding theory. Na111ral Language and Lingui.uic Theory 8, 1-31. Koch. Peter 1981 Verb-Valen::-Verfiigung: Zltr Sat::semmrtil.:. mrd Valen::frmr:.Osischer Verben am Beispiel der Verfogllnf{s-Verbnr. Heidelberg: Carl Winter Universitiitsvcrlag. 1997 La diacronia quale campo empirico della semantica cognitiva. In F. LoPiparo (ed.), Unguaggio e t·og~riz.ione. Alii dl.'l XXVIII Congre.uo. Rome: Bulzoni, 225-246. KOnig, Ekkehard, Martin Haspelmath 1998 Les constructions a possesseur exteme dans les langues de J'Europe. In J. Feuillct (ed.). Actancl.' et \'alenct.' dans les langue.f de /'Europe. Berlin: Mouton de Gruyter. 525-606. Lakoff. George 1987 Women, Fire, and Dangerous Tiring.'i: What Cmegorie.t Re}·eal about tire Mind. Chicago: The University of Chicago Press. Langacker, Ronald W. 1984 Active zones. Proceeding.f of the Tenth Annual Meeting of the Berkeley Unguistics Society. Berkeley. California: BLS, 172-188. 1987 Foundations ofCognitil·e Grammar. Volume 1: Theoretical Prerequisites. Stanford, California: Stanford University Press. 1993 Reference-point constructions. Cog~riti1•e Unguistics 4: 1-38. Le Guem. Michel 1973 Sbnantique de Ia mlwphore et de Ia mitmrymie. Paris: Larous...e. L.e Robert Historique 1992 Dictionnaire historique de Ia langue franrai.re. Paris: Dictionnaires Le Robert. Miiller-Gotama, Franz 1994 Grammatical Relatio11s: A Cross-linguistic Per.'ipectil·e on Tlrt.'ir S.\'nlax ami Semantics. Berlin: Mouton de Gruyler.

Grammalkal Conslrainls on MeiOnymic Refere"ce

253

Nunberg, Geoffrey D. 1995 Transfers of meaning. Journal of Semantics 12: 109-132. Olsen, Susan 1994 Lokativaltemationen im Deutschen und Englischen. Zeit.idrrift fijr Sprat·flwisuriSclwft 13: 201-235. Schank, Roger C .. Robert P. Abelson 1977 Script.f, Plall.f, Goals, and Undf'r.uanding: An Inquiry into Human Knowlf'd!{e Structures. Hillsdale, NJ: Erlbaum. Spanoghe. Anne-Marie 1995 La syma:re de l'appartenarlcf' illalii11ab/e enfrallrais, en espaxnof et en portu~;ais. Frankfurt a.M.: Lang. Tannen. Debor.!h 1979 What's in a fr.amc? Surface evidence for underlying expectations. In R.O. Freedle (ed.). New Diret"tions in Di.fcaurse ProCI'.uing. Norwood, NJ: Ablcx, 137-181. Winston, Morton E .• Roger Chaffin, Douglas Herrmann 1987 A taxonomy of pan-whole relations. Co~;11itire Science II: 417-444. Zribi-Henz. Anne 1978 Economisons-nous: A propos d'une cla.sse de formes Jtflexives mCtonymiques en franc;:ais. Langue Franraise 39: 104-128.

Putting Metonymy in its Place

Paul Pauwels KVH Antwerpen

1.

Characterizing Metonymy1

When Lakoff and Johnson (1980) put figurative language at the top of the agenda. claiming that it involved basic cognitive processes rather than deviant usage, their work gave an enonnous boost to metaphor research, but not to the study of metonymy. After all. they did call their work Metaphors We Live By. and in devoting only a single chapter to metonymy. they seemed to imply it was a minor process in comparison. Moreover. they mainly defined me-

tonymy in comparison with metaphor. Metaphor is principally a way of conceiving one thing in terms of another. and its primary function is understanding. Metonymy. on the other hand. has primarily a referential function. that is. it allows us to usc one entity to .uand for another. But metonymy is not merely a referential device.lt also serves the function of enhancing understanding. ) ... ) determine~ which a:.pe~1 we are focusing on. (Lakoff and John~on 1980: 36)

It is this focusing behavior which gives metonymy its special cognitive status. Examples (I) and (2) illustrate what is meant by 'focusing.· (I)

The Times hasn't arrived at the press conference yet.

(2)

We need some good heads on the project.

In (I), which follows the WHOU:. t-'OR PART pattern (institution for member of the institution). the metonym is said to focus on the status of the reporter thus referred to. In (2). the (body)part which is selected to refer to the whole person. focuses the intelligence of that person as a particularly relevant

256

Paul Pauwels

characteristic. The grounding of figurative expressions is, for Lakoff and Johnson. more obvious in the case of metonymy than in the case of metaphor since it "usually involves direct physical or causal associations"' (1980: 39). Metonymy is clearly the poor member of the family here: it only 'focuses an aspect.' and it is also more obvious and less creative - hence presumably less worthy of the researcher's attention. Lakoff ( 1987) approaches the difference between metaphor and me-

tonymy against the background of an already more elaborate cognitive framework. using the notions of(cognitive) 'domain' (any sort of conceptualization relative to which semantic structures are characterized) and 'idealized cognitive model' (a kind of gestalt which gives structure to reality). For Lakoff. "a metonymic mapping occurs within a single conceptual domain which is struciUred by an ICM:· whereas "metaphoric mapping involves a source domain and a target domain" (1987: 288). The ''direct associations" (Lakoff and Johnson 1980: 39) which characterize metonymy arc here (re)defined in terms of membership within a single cognitive domain. The central clement differentiating metaphor and metonymy seems to be the greater cognitive distance (to put it in spatial terms) between the concepts involved in metaphor. Taylor (1989) remarks that metonymy has received little discussion in comparison to metaphor. He takes a broad view of metonymy which not only encompasses conventionalized metonymy as in (I) and (2) above, but which also accommodates situationally sanctioned examples like Lakoff and Johnson's famous 'ham sandwich' example and common contextual modulations of meaning as in (3): (3)

a. He opened the door. b. He walked through the door.

For him "the essence of metonymy resides in the possibility of establishing connections between entities which co-occur within a given conceptual structure" (1989: 123f). In this view. it is not surprising that "metonymy turns out to be one of the most fundamental processes of meaning extension, more basic. perhaps, even than metaphor" (1989: 124). In reaching this conclusion, Taylor is supported by Goossens (1990. 1995). Looking at diachronic evidence. Goossens comes to the conclusion that metonymy is much more central in our conceptual apparatus than is usually thought.and. on top of that. it is often at the basis of metaphor. The following examples arc taken from Goossens (1990):

Putting Metu11ymy in it.f Place (4)

257

''Oh dear;• she giggled. Td quite forgotten."

(5)

have a word in someone's ear

(6)

I could bite my tongue off.

In cases like (4), for example. it is not clear whether the domain of giggling and that of (light-hearted) linguistic action are conceived separately. We are still aware of the metonymic basis in the metaphorical interpretation. a fact that Goossens (1990: 328) calls 'metaphor from metonymy.' In these cases. the difference between metonymy and metaphor lies in the distinction between a 'while'-interpretation and an ·as if-reading, respectively. In most such cases it seems the 'as if reading is most likely. although the metonymic basis cannot be discounted completely. This is illustrated by (5): speaking privately with someone may involve whispering into his/her ear. but the expression can be used to conceptualize situations where this is not actually happening. Another 'blend' Goossens distinguishes is 'metonymy within metaphor,' where a shared clement bridges the gap between the metaphor's source and target domain. functioning metonymically in the target domain. as in (6). Tongue is used literally in the source domain. but in the framework of the metaphorical usage (depriving oneself of one's speech as some kind of punishment) it functions as a metonym for the speech faculty. It should be clear from the foregoing that, in actual usage. the line between metaphor and metonymy is not always easy to draw. Against this background, Croft (1993) can be read as an attempt to come to a more narrowly defined concept of metonymy. Like Taylor. he discusses the difference between metaphor and metonymy against the background of other processes of meaning extension. He relies on Langacker's notion of 'domain matrix,' i.e .. the situation which occurs when the base against which an expression is profiled refers to the intersection of several domains (which is the case for most expressions). Lakoff's (1987) distinction is rephrasing, anger, or shame. Again we note the tension between specific and general in most examples. In (23) there is a difference in 'scale' between the primary LM (fiJot) and the secondary LM describing its target-location (ki11gdom). It could be argued that this case is an exception: in other examples like i11 thi.,· hou.fe, in this room the tension is less obvious. Still. the fact remains that literal descriptions usually have as secondary LM either a physical object. a substance. or a surface. Similarly. in (24).there is a tension between the primary LM (body part) and the secondary LM (person a.'i a whole), while in literal descriptions the focus is more narrowly on the contact between body parts as in lay a hand on .wmeone·.,. arm. It is this tension which is resolved in the metonymic interpretation. In (22) the tension is between the specificness of the description of the contact between the lips and a glass. and the frequency suggested by the plural glaue.~ in the antecedenl. 4.2. Manipulating human beings A second subset with approximately 20 examples describes human beings as manipulated entities. It should be clear from the start that babies. children. unconscious persons. bodies and willing adults can indeed be lifted. transported and fait/, set, put and placed in a literal sense. In the examples. however. the manipulation stands for a more elaborate kind of interaction. Motivations for these uses can be found in euphemism or. conversely. dysphemism.

Paul Pauwe!J

266

(25) I can lay a woman any time I want.

(26) Lucy's laid him out in his dress uniform. (27) a. She laid him out flat. (cold) b. She was out cold but still breathing.

(28) I ought to put you across my knee. Try it. You take one step towards me and I'll... (29) "But I always put my babies on pots straight away." said old Mrs. Harford. reprovingly. In (25) the sexual act is referred to by means of the prototypical position of the participants (i.c .. lying down). On the one hand the expression is quite graphic - which accounts for the selection of lay to function metonymically. On the other hand. the use of the verb as a two-place predicate without profiling a secondary LM results in a certain vagueness. At the same time the causative aspect in the verb is stressed. The overall effect seems to be one of dysphcmism. which is probably a consequence of the focus on the physical and causative aspects of the interaction. Vagueness and euphemism are apparent in (26), where the preparation of a body for burial (not only a disagreeable topic, but also one which is 'shrouded' in mystery for most people. one can presume). is referred to by invoking the resultant state of a body stretched out. Again the action of bringing the body in this state is a necessary part of the complete process. Example (27a).though similar at lirst sight. is the result of a completely different kind of meaning extension. Here. the usage is based on a metaphorical usage of mtt (cold/flat). as exemplified by (27b). This usage is at the basis of expressions like knock someone Ollt. which is here being metaphorized by means of lay someone out flat/wid. which focuses the result rather than the means by which this result is arrived at. The causative aspect of lay is again foregrounded by this usage. In how far the scene in (26) is invoked in the sanctioning of (27a) is a moot point - that there are possible bridges is obvious. The final two examples (28) and (29) are at the same time vague the really disagreeable bit is not mentioned - and explicit - there is no doubt as to what is being referred to. They probably originated as euphemism, but they can also be used as dysphcmism. In all of the above. the metonyms focus on one stage of a sequence which is characterized by unity of place. time, and participants in the interaction not unlike Fillmore's scenes. The events referred to (having sex. preparing a

P11tting Metunymy in it.f Place

267

body. spanking. toilet-training) are (or used to be) taboo topics. The subdomain of manipulation is not only selected because it avoids the taboo; there is also the concreteness of manipulation which provides the necessary salience. and the focus on the control exerted by one participant over another. which is a central element in the event.

4.3. Manipulating instruments In the following, a selection from 35 examples. more prototypical scenes of manipulation are involved in metonymic patterns. A first set of expressions. with lay as a central element. has 'stopping' as a target domain. (30) It is not difficult to lay down The Conque.~t of Pem or Volume 2 of The Cambridge Mellieval Hi.ftory. once it has been taken up; but it is not so easy to feel altogether happy about never taking it up again. (31) Carpentry is an occupation that lends itself to being laid down at will, either temporarily or permanently. (32) Ron. dcmobbed from home guarding. gladly laying his rine aside. built me a fruit cage. In (30), to stop reading is referred to by means of laying down a book - a physical activity which indeed entails the more general complex of activities. and which seems intrinsically bound up with it. This is in contrast with (3 I). which should probably be considered as metaphor pure and simple: the very general activity of stopping is here being conceived in terms of stopping to manipulate. There still is a metonymic link in the background. since carpentry is prototypically a manual job which requires manipulation of tools and objects. In (32) as well. the link between source and target is less strong: soldiering and rine-carrying are of course related. in so far as one of the main aspects of being a soldier (in the public mind) is the obligation to 'take up· arms, but it is not inconceivable for a soldier never to carry anns. Moreover, there is no 'entailment' relationship between source and target as is the case in (30). (33) Plant plastic gnomes where you walk and even. perhaps. put pen to paper and relate your experiences in the force. (34) I'll cook something, you can wash your hair, you can relax. we'll put on some records.

268

Paul Pauwe/J (35) a. If that'll be all sir, I'll let her in and put a kettle on. b. I'll put on an egg.

A second set of examples. where put is the central verb. all involve the starting up of a process. In (33) the act of writing is referred to by means of the manipulation of the prototypical instrument involved. or rather the first movement of that instrument. In so far as this expression is used to refer to 'manual' writing, it can be considered as metonymy. If it were used. consciously, to refer to writing by means of other instruments. this would be a case of metaphor from metonymy (cf. also Pauwels 1995: 150). Similarly. in (34). a fairly simple act of manipulation which occurs at the beginning of a sequence

of activating a recordplayer and playing a record. is used to refer to the complete sequence. In this case. the metaphorical potential is possibly greater. for two rea..ons. For one. the classic turntable is on its way out and being replaced by the CO-player. where the record is not put on in the literal sense.~ On top of that. this might activate another path via which the usage can be sanctioned, i.e., the metaphorical put wr as in put 011 tire radio. Example (35a) falls into the same pattern as the above. with both the potential of a metonymic reading ('on the stove') and a metaphorical reading ('electric kettle'). An example like (35b) clearly demonstrates a further shift away from the literal, descriptive usage. since the instrument involved in the boiling of the egg (indeed. boiling. rather than any other way of cooking an egg. is referred to thus) is not mentioned. In cases like the above. the metonyms appear to be mainly referring devices. Referring to a complex activity by means of its fairly concrete first or last stage is basically a time-saving strategy. Rather than focus on a specific aspect, or euphemistically avoid another (the main strategies encountered so far). this type of metonymy tends toward imprecise language usc, exploiting Grice's maxims. one of which states one need not be more specific/precise than necessary. The fact that quite a few of these examples involve closely scripted events undoubtedly plays a role here. The selection of lay to conceptualize 'stopping' and put to conceptualize 'starting' relates to the prototypical meanings of these verbs (sec further Section 6) - a full discussion is beyond the scope of this paper. Not all the examples of manipulating instruments are of the same type. however. Example (36) is clearly much more closely linked with the examples in the previous sections as far as the motivating factor behind the metonym is concerned.

Putting Metunymy in

it.~

Place

269

(36) ''Put a pillow on his face and get him out of it,"' said Paul to himself. Although the actual act of putting a pillow on someone's face falls into the pattern of referring to the start of a scripted sequence of actions, there is, in this ca'\e, an additional a.'>pecl of euphemism which provides further motivation for the partial selection. This is also true of (37). which is however even more complex. since there is a complete interpenetration of metaphor and metonymy on different levels. (37) Wipe that smile off your face and get up. before I put you on the governor's report. At the highest level. this is again a case of the first stage in a sequence standing for the complete sequence: writing someone's name down in an official register is used to refer to the consequent punishment. However. the writing down itself is metaphorized in terms of manipulation. where that which is written down is referred to metonymically. The result of the complete sequence is a 'demetonymization' (cf. also Goossens 1990: 335) since in the end the person is literally the affected entity.

5.

Target domain linguistic action

In this section I will take a closer look at the nature of the link between source and target domain from the perspective of the target domain. Of the examples found in the corpus. quite a few. with put a.'\ the central element. have already been discussed in some detail in Pauwels ( 1995). Also. examples from previous sections fit in here. (18) a.

I do put my foot down about discussing them.

(30) It is not difficult to lay down The Conquest of Pem or Volume 2 of The Cambridge Medieval History, once it has been taken up: but it is not so easy to feel altogether happy about never taking it up again. (33) Plant plastic gnomes where you walk and even, perhaps. put pen to paper and relate your experiences in the force. (37) Wipe that smile off your face and get up. before I put you on the governor's report.

Paul Pauwe/.\'

270

(38) The draft was laid before parliament. (39) Let me put your mind at case and say that Gregory has certainly no intention of wanting to marry me. The source domain of {18a). i.e .. manipulating body parts. is not only related to the larget domain of linguistic action (LA) by virtue of the coalescing in time of the two actions. The body is used frequently in non-verbal communication. and it is this background - the communication domain - which supports this metaphor from metonymy. In (30) and (33). the specific subdomains of reading and writing centrally involve the manipulation of instruments. The source domain is clearly in the

matrix of the target. This is also true of special cases of making official statements, where the introduction of a topic to an audience is usually accompanied by the introduction of a written document at the same time. as in (38). Since the addressee is orten not only the listener. but also the undergocrof an LA. it is not surprising that metonyms involving the person addressed/ affected are often embedded in LA metaphors. Especially metaphorizations involving the source domain of manipulation are prime candidates. as is clear from (37). (39) is slightly different. since the metonym does not highlight the 'whole' person, but rather the mind-'part'. In summary. there appear to be three types of metonymy from the target domain point of view. There are two types of event-metonymy. one in which source and target domain are both subsumed by the more general domain of non-verbal communication. as in (18), and a second one where the source domain is subsumed by the target domain of LA. as in (30), (33) and (38). The third type are nominal metonymies embedded in LA-metaphors which activate a more general domain. as in (37), or a more specific domain. as in (39).

6.

Metonymy, generalization and the extension of verbal meaning

In quite a few of the examples of metonymy discussed above where the metonym was invoked to resolve tension at the sentence level between the dependent predication and the autonomous one. the origin of the tension could be traced to a discrepancy in the scale or level of abstraction of the concepts involved. Indeed, the most pervasive patterns of metonymy are those which

Pu"ing Meto11ymy ;, it.f Place

271

have traditionally been described as PART FOR WHOLE or WHOLE FOR PART. Looking at. for example. Lakoff and Johnson's list ( 1980: 38f) it is clear that most of their cases can be construed as more specific kinds of these two basic types. PRODUCER FOR PRODUCT, OBJECT USED FOR USER, CONTROLLER FOR CO!>mtOLLED. INSTITUTION FOR PEOPLE RESPONSIBLE, PLACE FOR INSTITUTION. PlACE FOR EVENT can all be characterized by invoking a 'container· relationship between two domains: a superordinate one and a subsumed one. In analyzing these examples as metonymy. the stand-for relationship is taken to resolve the semantic tension. However, the tension can be, and. judging from the evidence. often is. resolved in another way. Take example (40). which can be described as a straightforward case of metonymy, where table stands for the objects which are being laid on it.

(40) I'll just lay the table. (LET/Ayckbourn) The autonomous predication (table) is the locus of the effect induced by the dependent one (lay). This is however not the full story: the preferred status of goal-objects appears to have resulted in an accommodation of this fact in the semantics of the verb. which has resulted in a generalization. Rather than 'manipulate,'/ay can be taken to mean 'arrange. fix. prepare or make ready.· In other words, the meaning of the dependent predication appears to have shifted as well. Also. the syntax of the verb has been affected by this: rather than profile the primary LM of the lay-prototype (the goal-object which is being moved. a in example (12)). this two-place predicate only profiles the secondary LM of that prototype (the location involved in the movement). That this pattern has proved productive. should be clear from the list of examples under (41). (41) set the table, set the alarm, set a fuse, set the thermostat. set a watch. set the oven to the required temperature ... Frequencies in the corpus seem to indicate that set has evolved furthest in this direction. There are some examples with lay in this pattern. but no examples at all for put and place. For put, this could be explained by the fact that the original prototype concerns bodily movement (pushing. shoving) rather than manipulation. In this prototype. the secondary LM with respect to which the movement can be characterized is the agent her/himself. The use of p11t to profile manipulation of an object with respect to a location is a later. extended use, and in this literal meaning put needs the presence of at least a hidden

272

Paul Pauwels

secondary LM, which can then not move up in the hierarchy, so to speak, to take the place of the primary one. Place, finally, is a much more recent addition to the language: set and lay are attested in OE, put from late OE onwards. while piClce is a 15th century loan from French.

7. Conclusion The previous sections have tried to present a survey of metonymy as it occurred in a specific corpus. From the survey. it soon became clear that the majority of the examples were not of the 'traditionar nominal kind. If anything, it seems like most nominal metonyms occur inside metaphors. Metonymy often seems to function as a kind of 'avoidance strategy.' for reasons of euphemism perhaps. Conversely. it also serves as a 'focusing slratcgy,' which in extreme cases results in dysphemism. The relationship between the domains involved in metonymy seems to be one of inclusion, with either a more general concept standing for a more specific one. or vice versa. There will always be a difference in scaJe or level of abstraction. which is not the case for metaphor. Finally, it should be clear that metonymy cannot be investigated without looking at other processes of meaning extension to which it relates in actual usage. Goossens' (1990. 1995) work demonstrates how metonymy interacts with. gives rise to. and is embedded in metaphor. Furthennore. it appears that generalization and specification are also often related to metonymy - a conclusion which should not surprise us, given the interaction between general and specific inherent in metonymy.

Notes This paper benefited from 2.

~:ommcnts

on an earlier version by L. GoosM:ns.

In thb actual example. taken from a corpus of English from the 196(Js. the 'CD· iniCrprctation · b of course not open; present-day English however eontmues to refer to 'pulling on· a CD; further illustrations of the generali1..ation which has occurred arc provided by 'put on a ca~scue· and 'put on a song."

P11ning Me/Onymy in its Place

273

References Crofl, William 1993 The role of domains in the interpretation of metaphors and metonymies. Cogniti1·~ Lingui.~tic.~ 4: 3.15-370. Geens, D., L.K. Engels, W. Martin 1975 Leuven Dmma Corpus and Frequency List. PAL. KU Leuven. Goossens, Louis 1990 Mctaphtonymy: the interaction of metaphor and metonymy in figurative ex.pressions for linguistic action. Coxnith·~ LinRuistin I: 323-340. 1995 From three respectable horses' mouths: metonymy and conventionalization in a diachronically differentiated data ba-.c.ln L Goossens ct aJ. (eds.). By Word ofMowh: Mnaplwr. M~tonym)' and Linxuistic Action in a Cogniti1·~ P~rsp~t·til·~. Amsterdam and Philadelphia: Benjamins, 175-204. Johansson. Stig. Geoffrey N. Leech. Hellen Good.luck 1978 Lancll.liter-OslofBergcn Corpus of British English, for use with digital computers. Ukoff. George 1987 Women. Fire ami Dangerous Tlrings: What Cat~gorie.f Re1•eal abolll the Mind. Chicago: 'The University of Chicago Press. Lakoff. George. Mark Johnson 1980 M~tapl10r.f W~ Lil·~ By. Chicago: The University of Chicago Press. Langacker. Ronald W. 1987 Foundation.f of Cognitil·~ Grammar. Vol11m~ 1: Theoretical Preuquisite.f. Stanford. California: Stanford University Press. 1994 The limits of continuity: discreteness in cognitive semantics. In C. Fuchs, B. Victorri {eds.), Contimtity ill Li11guistic Semalllic.f. Am~terdam and Philadelphia: Bcnjamins. 9-20. Pauwels, Paul 1995 Levels of metaphorization: the t:ase of pill. In L Goos.-.ens ct al. (eds.), By Word of Mouth: Meraphor, Metonym_\' and Lingui.uic Action in a Cogniti1·e PeTJpectil·e. Amsterdam and Philadelphia: Benjamins. 125-158. Pauwels. Paul. Anne-Marie Simon-Vandenbcrgen 1995 Body parts in linguistic action: underlying schemata and value judgements. In L Goossens eta!. (eds.), B)· Word of Mouth: Metaphor, Meton.mt)" and Li"guistic Action ;, a Cogniti•·e Perspecti•·e. Amsterdam and Philadelphia: Benjamins. 35-70. Taylor, John R. 1989 Ling11istic Caregori::arim1: Prototype.f ill lJngui.ftic Theory•. Oxford: Clarendon Press.

Conversion as a Conceptual Metonymy of Event Schemata

Rene Dirven Gerhard Mercator University of Duisburg

1.

Definition

As a conceptual process of locating a reference point, metonymy belongs to the wider set of strategies of finding a point in the common reference space between a speaker and a listener that can serve as a bridge or link to the intended referent (cf. Langacker 1991: 170f0. One possible strategy used for this purpose in English is the genitive. as in I'll take Dad'.\· mr. whereby Dad is the link known to the son and his girlfriend for identifying the car. In a restaurant scene where the waiter speaks about one of the customers in mct(mymicalterms as the hamburger. we find the same communicative need of identification as with Dad's car. Only now the context docs not provide an identity marker. since the customer's name is unknown and his place in the restaurant may not be a salient one. An alternative would be to give tables and customers numbers. But even then the waiter would not necessarily say ( Ia). but might equally well say (I b). (I)

a. The customer at table 13 is complaining about his hamburger. b. Table 13 is complaining.

If no numbers arc available. the waiter can only take the most relevant feature in the given link between the institutional context of the restaurant and the rea.'ion why people go there. i.e .. in order to get food. Since people tend to take different dishes. these become a salient means of identifying customers. In fact. in the case of the referencing use of metonymy as in the restaurant scene.

Reni Dirven

276

the hamburger ordered or eaten by the customer serves as the identifying description. The only identifying link that the waiter ha" with the customer is by means of the food he has ordered. Notice that the very same process of referencing occurs in other institutional contexts. too. Hospital nurses frequently refer to their patients in terms of the organ affected or the illness they suffer from. Thus. one may hear things like (2): (2)

a. The liver is still waiting outside. b. I'll now take the by-pass back to his room.

Although this metonymical strategy for the sake of referencing is ubiquitous - or perhaps because it is so ubiquitous - we are no longer aware that we apply it constantly. But it is always used in the context of picking out referents in a somewhat amorphous or anonymous setting. For example. teachers usc it with their students (the striped sweater). spectators wilh referees (tire black one) or football players (the red one.the bald one), musicians (the ''iolins, the tuba). pigeons (various shades of gray. or. in some languages. colors), salespersons usc it with customers in a nower shop. not at the counter but in the back room (the fuchsia, the geranium woman). But even if we already have a name for identifying an institution. the tendency to refer to it by way of metonymy remains very strong especially with employees referring to their workplace, which is orten indicated by the street or quarter. This is perhaps a sign of distancing and neutralizing emotional ties and commitment signals which might be contained in using the real name. Thus. Dutch teachers tend to refer in Dutch to their schools as the High Street (Hoogstr'/rr rrrr iaw lrodluc~J

Hrlraslsrlris

I

/Hrnr<JI'rdw rlrrmlrrr11·mfJ; Hrp are professionally concerned with the explication of the notions of truth. knowledge. belief and existence. The fundamental problem for the linguist. a.~ far as reference is concerned. is to elucidate and to describe the way in which we use language to dmw allention to what we are talking about.

In light of these observations, it docs not seem appropriate to make a distinc· tion between two kinds of metonymy. This distinction would implicate that a phenomenon like hyperbole interferes with the semantic (metonymic) struc· ture of an expression. I have tried to show that this is not the case. Therefore I prefer to speak of 'hyperbolically u,,·ed metonymies.·

5.

Metonymic hierarchies

On the basis of this characterization of metonymy .I can now tum to some largescale metonymic structures that generalize over several linguistic instantiations. I will show that metonymic hierarchical systems display similar structural characteristics to the ones I described for metaphoric hierarchies (sec above). First, I will discuss two specific examples and then I will present a schematic picture of the conceptual structure of STIJPID which will demonstrate the structural role metonymy can play as a pervasive conceptual mechanism. (8)

Er steht da wic der Ochse vor dem Scheunentor. 'He stands there like the ox in front of the bam-door.'

This example concerns the conceptual background of a non-conventionalized expression which profiles stupidity in tenns of a particular behavior: a person who is doing nothing else than just standing there like an immobilized object (NOT MOVING). More specifically. an expression such as (8) incorporates an overlap between the two concepts DEVIANT BEHAVIOR and DEVIANT APPEARANCE. A deviant behavior in its tum elaborates both concepts DEVIANT SOCIAL PROPERTY and DEVIANT PHYSICAL PROPERTI. On a higher schematic level, these latter concepts are schematized as SALIENT DEVIANT HUMAN PROPERTY. This

K11r1 F eyaerts

324

conceptual hierarchy is represented in Figure 2. which mainly focuses on the internal structure of the source concept. Accordingly. the metonymic extension to the target is mentioned only once, namely on the level of the linguistic expression (STIJPID). The metonymic relationship involved here can be paraphrased as 'stupidity shows up in deviant behavior: (Broken arrows are used to indicate metonymic extensions.) With respect to the three structural characteristics I distinguished for metaphoric hierarchies. it is obvious that conceptual metonymy can be described on different levels of abstraction as well. Profiling stupidity as a particular way of standing can be subsumed under the highly schematic pattern in which a salient property stands for (expresses) a less salient one, since physical and social properties can be perceived more easily than an abstract mental property. As already mentioned above. conceptual/linguistic metonymies also display patterns of overlap between higher schematic structures. Not surprisingly. the third characteristic applies to metonymically organized systems as well: metonymies of a higher. more schematic level tend to be of a more universal. purely cognitive nature. Lower metonymies on the other hand. appear to be more culturally restricted .

•

I STUPID I Fi~:ure

2. Hierarchical .ftruc·ture of the DEl//ANT BFJIA.l'IOR FOR STIJP/DfTY metonymy

Metonymic Hierarchies

325

I want to illuslfate this cultural dependence with a second example which refers to a typical German social stereotype in which stupidity plays a major role: (9)

Mantafahrer! 'Manta-driver!'

This metonymic model gets its name from a particular kind of car (Opel Manta) which can be characterized as having a sporty look. a low and aerodynamic profile. a powerful engine. etc. Moreover. it is a relatively cheap car which made it a big commercial success. At the same time, however. this popularity ha.'i 'degraded' it to an ordinary car everyone can afford. To (re)gain some exclusiveness many people equip their Opel Manta with fancy, aggressive-looking accessories like chrome spoilers. muscle tires. rally scat belts, high-beam headlamps, racing colors. etc. All these elements contribute to a typical image of this car as an 'ordinary car dressed up with kitschy gadgets.' What is particularly interesting in this context. is the cultural (German) model in which this car is associated with a typical kind of people owning and driving it. This social stereotype characterizes a Manta-driver as macho, having a blond girlfriend or wife. belonging to a lower social class. driving fast and aggressively, and also as not very intelligent but rather stupid. Calling somebody a Mamafahrer thus can be interpreted in a particular context a-. 'he is stupid.' This conceptual extension is clearly metonymic in that salient properties like deviant behavior. appearance and possession (i.e .. the car itself) are conslfued as manifestations of stupidity. Accordingly. these concepts all take part in the source domain structure which serves as the conceptual base for profiling the target STUPID. To put it another way. these source concepts overlap in any stupidity expression that instantiates this Mamafalrrer-modcl. This model is represented schematically in Figure 3. This hierarchy shows how. in its metonymic extension (broken arrow) onto the target concept STUPID. the culturally specific Malllajahrer-concept elaborates (full lines) several schematic concepts. Figure (4) presents a global representation of the schematic conceptual structure of the target domain. The difference between this and previous figures is that in this case I have brought together the most important source concepts in a single system. The many broken arrows indicate that STUPID is essentially metonymically structurcd.u Presented in this figure are only the most schematic levels of the conceptual structure. Not included are several

Kurt Feyaert.\'

326

~

"''.""~'""''

"".""'~"""" MANTAFAIIRER

I STjPID I FiRure 3. Hierard1ict1l strlt£'fllre of till! MtiiiiO/Clllrer·nwdel

lower-level

schema.~

which further elaborate parts of this structure. Notice that

I consider the target structure STUPID as an elaOOration of DEFICIENT MENTAL ABILmES and that. normally. one would expect the abstract and undifferenti-

ated concept STUPID to appear right below this schema. In order to keep this

figure as clear as possible, however. I moved it to the bottom. The numbers between brackets refer to instantiating expressions which correspond to examples mentioned below or elsewhere in the text. An interesting observation concerns the complex source structure which can be observed in (5). On the one hand. not being able to count to three elaborates both mental as well as practical abilities. On the other hand. however. this deficiency also relates metonymically to little children who are not able to count properly so that this notion - albeit not very prominently also takes part in the conceptual structure or the source domain. It thus appears that - in applying the theory or complex conceptual networks - a complex source domain structure may consist or schematically·related. as well as (metonymically) extended concepts. A similar conceptual pattern underlies the expres..'\ion in (10). which instantiates a deviant appearance (the most prominent source structure) and at the same time metonymically refers to a small (newborn) child (DEVIANT AGE) which typically has this kind or appearance. Both conceptual relationships combine into the complex source structure which metonymically stands for the target STUPID.

Metonymic Hierarchies --------------------~

Fi~11re 4. General nmn?J'Iflal JIT/1('/IIre of STUPI{)

(10) Er ist nicht trocken hinter den Ohren. 'He is not dry behind his ears.' (II) Sic hat nicht aile Sinnc bcisammen. 'She does not have all senses together.' {12) Er hat Wasser im Kopf/Er hat ein weiches Gchim. 'He ha" water in his head.' I' He has a weak brain.' (13) Sie ist nichl so dumm wie sie aussieht. 'She is not as dumb as she looks.·

327

328

Kurt Feyaert.\' (14) Da.\· i.~l 11och mrterdem Naclllwiidrler JEr isl ein geisliger

Untergefreiler.

'That is even below the night watchman."/'Hc is an intellectual "'under-corporal".' (I 5} Er ist zu dumm, urn eincn Eimer Wasser umzustoBcn. 'He is even too stupid to knock over a bucket of water.· Most of these examples speak for themselves. except perhaps for ( 13). which is not an expression that profiles stupidity in a direct way. On the contrary, it profiles a positive mental property. However, the image which is used confirms the social stereotype according to which mental ability can be inferred from somebody· s appearance ('appearance as a manifestation of mental abilities·). from a more profound analysis of the data. it follows that the prototypical schematic source structure for STUPID is DERCIENT HEAD/BRAIN (compare ( 12)}. More than the other source structures this schematic concept hides a complex network of multiple interacting conceptual relationships. Correspondingly. this conceptual source domain displays the highest instantiation rate. With regard to a structural comparison between metaphorically and metonymicDIIy organized conceptual hierarchies. it seems that both systems have major structural characteristics in common. Although I am aware of the different conceptual strategies which are involved in both mechanisms (cf. Dirven 1993). figure 4 clearly shows that generalized metonymy,just like its metaphoric counterpart. operates on different levels of schematicity ('first similarity feature'). At the same time this broader focus allows one to identify more easily the patterns of interplay and overlap between different metonymies in the hierarchy ('second similarity feature'). In this respect I have briefly discussed some interesting examples (compare (5). (8). (9). (10)) that profile stupidity on the basis of a complex source domain. Such patterns of overlap can be described as a complex structure situated somewhere in conceptual space. defined by different relationships with other conceptual structures and activated as (part of} the source domain matrix of the profiled concept. The examples cited above display two possible patterns of overlap between concepts/metonymies. The first one resides in the elaboration of two schematic structures. whereas the second one consists of a metonymic association with another concept. With this description. I apply Langacker's ( 1987} view on schematic networks as the organizing principle of our knowledge structures. Accordingly, the conceptual background (domain matrix) of a profile requires "specifications in numerous domains" (ibid.: 163). All these

Metonymic Hierarchie.\·

329

specifications relate in one way or another to the conceptual entity which is profiled by the expression. The entity designated by a symbolic unit (its profile: K.F.) can therefore be thought of a.~ a point of accc:.~ toil network. The ~mantic value of a ~ymbolic unit is given by the open-ended set of relations - :.implc and complex, direct and indirect - in which this access node participates. (ibid.)

Not all domains/concepts which constitute the conceptual background of a profile are activated with the same degree of intensity (cf. also Croft 1993: 344f). Although in this context I did not deal with that issue explicitly, I mentioned that some concepts which are metonymically related to the source structure may be less directly involved in the definition of the profiled concept STUPID. Compare. for example. the association of a concept like MARKED AGE (CHILD) with the conceptual structure which is directly instantiated by the expression in (5). A similar observation holds for the association of a low social class with the central aspects of the Mantafahrer-stereotype. As to the 'third similarity feature.' finally. I have demonstrated that specific, low-level metonymies like the German Mamafahrer-model tend to be culturally more specific with a correspondingly restricted use. Higher-level metonymic mappings on the other hand are more widespread in their use because of their general cognitive. culturally unrestricted nature.

6.

Conclusion

Conceived in a cognitive semantic framework. this article raises two major points concerning the nature and structure of metonymy. First. it is argued that on both a theoretical and methodological level. the notion 'domain matrix· is not precise enough to be used as the primary criterion for the distinction between metaphor and metonymy. The essence of metonymy resides in the nature of the relationship that connects two concepts with each other. Accordingly, I have defined metonymy in terms of a conceptualized contiguity relationship between two entities. Hence. metonymy is categorized primarily as a conceptual rather than a strictly referential phenomenon. Consequently. the main interest of this contribution is to be situated on the level of conceptual structure. In this perspective. the analysis of the data demonstrates the crucial role metonymy plays in structuring our knowledge system. From the examples provided here. it follows that metonymy represents a sophisticated and robust

Kurt Feyaerts

330

conceptual mechanism which must be treated on a par with metaphor. II was also shown, second, that metonymically motivated idioms may cluster into higher-level conceptual metonymies in a similar way that metaphoric expressions do. This comparison has led to the conclusion that both

types of conceptual organization have three major structural characteristics in common. The extrapolation of the notion of multi-level hierarchies of conceptual patterns from the study of metaphors to the sludy of metonymies represents the specific theoretical contribution of this paper to the cognitive semantic study of metonymy.

Notes I wi.~h 10 thank Dirk Gecracrts for valuable help with the prepara1ion of 1he 1alk on which !his anicle is based, GUnler Radden for helpful comments on an earlier version of this paper and Jessica Dobratz for proofreading major pans of the text. This aniclc pre~up poses a cenain acquaimancc with basic theses and tenninology of cognitive semantics and of the Lakovian theory of metaphor in panicular.

2.

Compare KOvecses (19M9: 44): "each and every expres~ion related to a concept ha.~ to be examined if we wish to uncover the minute details of the coocept." but sec also Geeracrts and Gmndelaers (1995: 174ff).

3.

In this respect I refer to Langacker"s (1993: 30) char.~eteri7.ation of metonymy in tenns of reference poinl~; "( ... [ a well-chosen metonymic expression leL~ us mention one entity that is salient and ea.~ily coded. and thereby evoke - essemially automatically - a target that is either of lesser interest or harder to name.[ ... ) Other things being equal. various principles of relative salience generally hold: human > non-human: whole > pan: concrete> abstract: visible> non-visible: etc."

4.

The cognitive basis of person perception makes it a di!!Cipline of great interest for cognitive semanticisl~: ··1n dealing with person perception we arc concerned with it a.~ it occurs in the layman. as a process affecting ordinary human interaction. and not with the person a.~ an object to be diagnosed by the expert" (Tagiuri 19511: xii).

5.

Cf. Heider (191111: 341): ··our main concern is n01 the exact mea.~ure of onc·s own ability. but right away the relation to others. If somebody would tell us exactly what our ability measures. we still would a~k: but how do others do"!'" studie~

6.

Cf. Lakoff(l993) and Kiivec:.cs (1995a) for recent

7.

Several examples illustrate that UFE ts ,, JOURNEY also elaborates the object dual of both schematic metaphors which are involved here. Compare. for instance. expressions like Ht- /n~l hi.~ /ifr- (l'MANGI'. OF STATE IS MOVEMf.NT m· POSSES.~W OBJELI) or Hi.f youth h1.1s

in this respect.

II.

Set- also Ki:i\'CCses (1995b: 193). who distinguishes between a generic. mainly physiologiully motivated coJK.-eptuallevel and a specific. mainly ~:ulturally detcnnined conceptual level.

abundoned him (TIME I'ASStNG IS MOVING E..'ITITY).

Metonymic Hierarchies

331

9.

Argumentation theory providl!s a good basis for an adequate de~cripcion of the conceptual relationships underlying these expres~ions. Crucially, the kind of argumentation (or conceptual linking) involved here is situated on a rfletoriul rather than a strictly logical level.

10.

Obviously. a.~ this theoretical pcr.;pective is inherent in any linguistic research, no linguistic analysis can ever claim to reflect the mental richness of a conceptual structure in full detail.

II.

I refer to Sandra and Rice ( 1995), where the is.~uc of the psychological reality of theoretical notions such a.~ 'domain' is discus.wd in greater detail.

12.

According to Dirven (1993: 18), figurative meaning ··only arises- or atlea.~t can arise - if lhe conceptual distance between the two (sub )domains or things referred to is large enough.'"

13.

In order not to complicate the picture too much, not every relationship between rourcc and target or between one rource concepc and another can be explicated in one single figure. for example.the contiguity relationship and potential metonymy (CALTSE/EFFECT) among the source concepts DEFICIENT SENSORY Al'f'ARATIJS, DEACIEm' HEAD/BRAIN and DI:ACIENT (PRACTICAL) ABIIJnE.~ is not represented in this schema.

References Bruner. J.S .. D. Shapiro. R. Tagiuri 1958 The meaning of tmits in isolation and combination. In R. Tagiuri, L. Petrullo (eds.). Per.wn Perceptimt and Jmerper.wnal 8f'ha1·ior. Stanford: Stanford University Press. 277-288. Croft. William 1993 The role of domains in the interpretation of metaphor.; and metonymies. Cognitive li11gui.~tics 4: 335-370. Davies. C. 1987 Language, identity and ethnic jokes about stupidity./lllernatiollallmmtaf of the Sodo/Of(\' of Umguage 65: 39-52. 1990 Ethnic Humor around thf' World: A Comparati1•e Analysi.L Bloomington: Indiana University Press. Dirven, RenC 1993 Metonymy and metaphor: different mental strategies of conceptualization. l.etll"eme Bijdragen 82, 1-28. Eemeren. F.H. van, R. Gmotendorst. F. Snoeck-Henkemans 1996 Fundamenta!J of Argumentation Theory: A Handbook of Hi.~torical Ba,·k· grounds and Comemporary Developmem.~. Mahwah. NJ: Lawrence Erlbaum. Geeraerts. Dirk 1994 Nondenotational semamic change. In R.E. Asher. J.M.Y. Simpson (eds.). Tile Enc·yclopedia of Language and linguistics. Vol. 5. Oxford: Pergamon Press. 2823-2824.

332

Kurt Feyaerts

Geeraens, Dirk, S. Grondelaers 1995 Looking back at anger: cultural uaditions and metaphorical patterns. In J. Taylor, R.E. MacLaury (eds.), Lang11age and th~ Cognitiw: Con.flrllol of the World. Berlin: Mouton de Gruyter, 153-179. Geiger.RA. 1993 Metacognitive a'ipecls of n:ference: assessing referential coi'Teetness and success. In R.A. Geiger, B. Rudzka·Ostyn (eds.), Conn•pmalizmio11s ami Mental Pmce.uing ini.Ang11age. Berlin: MoUion de Gruyter. 267-289. Goossen.~. Louis 1995 Metaphtonymy: the interaction of metaphor and metonymy in Cltprcssions for linguistic action. In L. Goossens et a!. (eds.), B)· Word of Momh: Metaphor, Mt>tonymy, and lingt1istic Al·tion in a Cognitiw: Per.fpec·ri~·e. Amsterdam and Philadelphia: Benjamins. 159-174. Heider. F .. M. Benesh-Weiner(ed.) 1988 Thl' Notl'books. Vol. 2: Percl'ption. Miinchcn: Psychologic: Verlags Union. Ktivecses. Zolt3n 1989 Emotion Cmlcl'pls. New York: Springer. 1995a American friendship and the scope of metaphor. Coxlliti\"t' l..ingui.uics 6: 315-346. 1995b Anger: its language. conceptualization. and physiology in the light of crosscultural evidence. In J. Taylor. R.E. Maclaury (eds.). Languugl' and thl' Cog11itil'l' C.mstmal of till' World. Berlin: Mouton de Gruyter. 181-196. Lakoff. George 1987 Womm. Firl', and Dangt'rous Thin~:.~: What Catt'f:OFil'.f Rl'\'1'01 abo11t till' Mind. Chicago: 'The University of Chicago Press. 1993 The contemporary theory of metaphor. In A. Onony (ed.). Metaphor and TlwuRht. Second edition. Cambridge: Cambridge University Press. 202-251. Langacker. Ronald W. 1987 Fomulations of Coxnitil"l' Grammar. Vol. /: Thl'oretical Prerequi.~iii'S. Stanford. California: Stanford University Press. 1993 Reference-point constructions. Cognitive l..ingui.uics 4: 1-38. Lyons. John 1977 St'mamics. Cambridge: Cambridge University Press. Sandra. Dominick, Sally Rice 1995 Network analyses of prepositional meaning: mirroring whose mind - the linguist's or the language user's? Cot-:nili\"1' Lingui.fli("S 6: 89-130. Schmid, Hans-JOrg 199.' CotJagt' 1111d Co .. idt'a. starrl'.f. begin: Dil' Katt'gorisierungals Gnmdprin:ip l'iner differm:il'rtl'll Bt'dt'lltung.fht'schrl'ibrmg. Ti.ibingen: Niemeyer. Tagiuri. Rcnato 1958 Introduction. In R. Tagiuri, L. Petrullo {eds.). Person PI'I"CI'ption a11d lnterpt'rsonal Beharior. Stanford: Stun ford University Press. Taylor. John 1989 l..ingui.uk Categori:tttio11: Protot)pl's in Lingui.~tit' Theory. Oxford: Clarendon Press.

The Potentiality for Actuality Metonymy in English and Hungarian

Klaus-Uwe Panther University of Hamburg

Linda Thornburg E6tv6s Lorand University, Budapest

1.

Introduction

In a recent paper (Thornburg and Panther 1997) we postulated and exemplified some general metonymies used by speakers of English. e.g .. the metonymies FORM FOR COt\TENT. EFFECT FOR CAUSE, POTENTIALITY FOR MOTIVATION, NECI::SSITY fOR MOTIVATIOI" and I'OTENTIALITY FOR ACTUALITY. the latter being one of the most pervasive. A natural extension of this research is to investigate the function and exploitation of metonymies across languages. In this paper. we single out the POTENTIALITY FOR Am:AUIT metonymy and explore how it functions in two genetically and typologically unrelated languages. English and Hungarian.

1. 1. Metonymy Metonymy is traditionally regarded as a figure of speech that involves a process of substituting one linguistic expression for another. i.e .. me10nymy is viewed as a relation in which one linguistic expression 'stands for' another. The bcst·known cases of metonymy in this lraditional sense arc expressions that are used for the purpose of indirect referring. For example, there is a convention that the referential noun phrase

Klaus-Uwe Panther am/ Unda Thornburg

334

( 1)

the White House

can be used to refer to the executive branch of the government of the United States, a spokesperson for that branch. or even the President himself but is not synonymous with any of these. More recently. with the advent of cognitive linguistics. it has been recognized that the traditional view of metonymy is too narrow and that metonymy, like metaphor. is a conceptual tool that operates within 'idealized

cognitive models' or 'domain matrices' (e.g .. Lakoff 1987: Croft 1993: Kovecses and Raddcn 1998; Raddcn and Kovecses, this volume). In this view metonymy is "[ ... 1 ll cognitive process in which one nmcepllwl entity, the vehicfe.provides memal acceu to another conceptual emity,the target. within the same idealized cognitive modef' (Radden and KOvecses, this volume). The cognitive approach to metonymy is not necessarily incompatible with the traditional view. At some level the vehicle and the target are equivalent, although they are not semantically synonymous. On the one hand. the metonymic substitution of one expression for another creates a relation of pragmatic equivalence between the substituting and the substituted expression: on the other hand. in accordance with Radden and Kovecses (this volume), we a'sume that both the vehicle and the target are conceptually present when a metonymy is used. We thus consider metonymy to be both a property of conceptual structure and of language usc, with varying degrees of conventionalization in the linguistic system. In this chapter. we will focus on metonymic phenomena that have a nonreferential function. By way of example. consider the propositions expressed in the assertions in (2a) and (2b): (2)

a. She was able to finish her dissertation. b. She finished her dissertation.

Just as the noun phrase vehicle in (I) does not have the same meaning as any of its metonymic targets. sentences (2a) and (2b) arc not semantically synonymous. It is possible to assert (2a) and to deny (2b) without contradiction. Yet on many occasions of usc the speaker of (2a) pragmatically conveys the same propositional content as that expressed in (2b). In this sense the statement (2a) can be used to 'stand for' the statement {2b). the only difference being that in the first case the speaker predicates the ability to finish the dissertation of the subject she. whereas in the second case the speaker predicates the actuality of finishing it. In pragmatic terms. (2b) is a (generalized) conversational

The Polenlialilyfor At·lllalily Melonymy in Engli.~h and Hungarian

335

implicature induced by (2a) (see Grice 1975). In contra•it to the referential metonymy in (I). statement (2a) is an instance of what we call 'prcdicational metonymy' and exemplifies the POTENTIALITY FOR ACTL:ALITY metonymy. A second type of non-referential metonymy is found in the domain of communicative actions. We call this type of metonymy 'speech act metonymy· wherein one illocutionary act stands for another illocutionary act, as, e.g .. in the statement or assertion (3)

a.

I don't know where the bath soap is.l

which may metonymically stand for the question or inquiry (3)

b.

Where is the bath soap?

Utterance (3a) has the direct illocutionary force of an assertion about what the speaker does not know, but in many contexts it is used with the indirect illocutionary force of a question (for the identification of speech acts in conversation. cf. Panther and Thornburg 1998). Searle (1975) hypothesizes that indirect speech acts involve the performance of two illocutionary acts: the secondary illocutionary act relies on the literal force of the utterance whereas the primary illocutionary act corresponds to its intended force. in the case (3a) that of an inquiry. Searle's claim that two illocutionary acts arc performed in an indirect speech act is just another way of saying that both the metonymic vehicle (the secondary illocutionary act) and the target (the primary illocutionary act) are conceptually present in the speaker's mind. In conclusion then, it is obvious that the phenomenon of metonymy is not restricted to its well-known referring function but is much more pervasive in ordinary language usc. In addition to referential metonymies. at least two more types of metonymic relations exist and can be shown to be linguistically relevant. To summarize, we propose the following pragmatic typology of metonymic functions: (4)

a.

propositional metonymies: (i) referential. (ii)predicational: b. illocutionary metonymies.

All of these metonymic functions may co-occur in a single utterance. as in (5): (5)

I don't know whether the first violin was able to pass her driver's test.


336

which contains the metonymic types presented in (6): (6)

a.

propo!~itional:

the noun phrase the first violin may stand for the person in an orchestra playing the first violin: (ii)predicational: the propositional content that The speaker doe.\· not know whether the first violin was abk 10 pass her driver'.\· tes1 may stand for the propositional content that The speaker does not know whether the first \•iolin passed her driver's test; b. illocuticmary: the assenion I dmr 't know whether the first violin was able 10 pw:.\· her driver's te.~t may stand for the question Was the first \'iolin able to pass her driver's te.ft? (i) rejerelllial:

In this paper, we concentrate on the operation of the POTENTIALITY FOR AcruAUTY metonymy in the predicational and the illocutionary domains across two languages. 2

1.2. The scenario approach to metonymy In accordance with recent cognitive approaches to semantics, we assume that meaning. including propositional and illocutionary meaning. is fruitfully analyzed in terms of scenarios consisting of parts which can bear metonymic relations to each other and the whole of the scenario (cf. Blank. this volume).3 Applying this part-whole analysis to speech acts in our prior paper (Thornburg and Panther 1997: cf. also Searle 1969. 1975) we proposed. among other things, the scenario for directive speech acts given in (7): (7)

Scenario for Directive Speech Acts (i) the BEFORE: H can do A. S wants H to do A. (ii) the CORE: S puts H under a (more or less strong) obligation to do A. the RESULT: His under an obligation to do A (H must/should/ ought to do A). (iii)the AF-TER: H will do A.

The scenario approach allowed us to show two things: ( 1) that indirect speech

The Pote"'ialityfor Actuality Meton_vmy in

Engli.~h

and Hungarian

337

acts (and related phenomena) yield to an explanation in terms of general metonymies and (2) that their linguistic forms are highly motivated. To illustrate. an indirect request such as that given in example (8): (8)

John, you will take out the garbage.

instantiates a 'part' of the Directive Scenario. namely the AFfER condition, which metonymically stands for the CORE of the requcstive speech act scenario or the scenario as a ·whole.' Moreover. the form of (8) is motivated since the prospective future action of the hearer is an integral component of that scenario.

1.3. The state-of-affairs scenario From the scenario given in (7) for the specific illocutionary type 'directives' we can abstract a more general State-of-Affairs Scenario. Speech acts are just special instances of what we call 'states of affairs.' e.g., states, events. processes. and actions, which, of course, include linguistic actions. We assume that the distinction between BEFORE conditions. the coRE. the immediate RESULT or Ef-VECT of the core. and the AFfER. which we have already posited for the Directive Speech Act Scenario in (7), also holds for states of affairs. Thus. we propose the scenario given in (9) for propositional contents that describe existing (actual) states of affairs. (9)

The State-of-Affairs (SoA) Scenario (i) the BEFORE: necessary preconditiom: motivations, potentialities. capabilities. abilities. dispositions. etc .. which can bring about the SoA (ii) the CORE: the exiMingltrue SoA the EFFECTs: necessary conseq11ences immediately following from the SoA (iii) the AfTER: mm-necenary col/.{eqllellces of the SoA

As an example of a state of affairs. consider a description of the act of opening a window. such as in ( I0): (10) Mary opened the window. A necessary pre-condition (a BEFORE condition) for the existence of the state of affairs described in sentence (10) is Mary's ability to bring it about. An even

338


more basic BEFORE condition would be that it is at all possible (for anyone) to open the window. Other BEFORE conditions which might be regarded as specific subcases of the ability condition would be. e.g., that Mary has the strength, the time. the patience, etc. to open the window (sec Panther 1981). If Mary is not able to open the window or is not strong enough or docs not have the time to open the window, then the window-opening simply docs not come about. Certain mental states also count as BEFORE conditions in our account. e.g .. the desire or intention to open the window. which can be regarded as motivations for the action. The cORE of the state of affairs is the actual performance of the action. The immediate Err""Ecr of a successful act of opening the window is another state of affairs. that is, that the window is open. An intended but not necessarily following consequence of opening the window (the AfTER) might be that fresh air flows through the window. An unintended consequence might be that flies come through the open window. In general, the Al-TER is only loosely associated with the CORE, whereas the EFF-r:cr is a necessary part of it:~ The SoA Scenario can be metonymically exploited just like the more specific Scenario for Directive Speech Acts. For example. in answering the question (II a) with (lib) the answerer B may exploit the metonymy EA-""ECT FOR CORE:

(II) a. A: Did you open the window? b. B: The window is open. Thus, in some contexts, interlocutor B might wish to convey that she actually has opened the window although. literally, she only states the result of the window-opening.

2. The POTENTlAUTY FOR ACTUAUTY metonymy in English and Hungarians In Section I. we exemplified a metonymic principle that is based on the conceptual contiguity of the modality of possibility and its actualization. It is an empirical question whether and to what extent the POTENTIALITY FOR ACTIJALITY metonymy is operative cross-linguistically, possibly as a universal cognitive principle. As a first step towards answering this question we undertook a comparative analysis of two genetically and typologically unrelated

The Potemialityfor Actuality Metonymy in EngliJh and Hungarian

339

languages, English and Hungarian. We investigated the linguistic relevance and distribution of the POTENTIALITY FOR ACTUALITY metonymy in seven conceptual and communicative domains: perceptual events, mental states and processes, hedged performatives, indirect speech acts, (extra-linguistic) actions. character dispositions, and acquired skins. The analysis revealed interesting contra•ils between the two languages in the exploitation of the POTENTIALITY FOR ACTIJALITV metonymy. As will be demonstrated below. this metonymy is used much more systematically in English than in Hungarian.

2.1. The sense perception metonymy (ABILiTY ro PERCEM FOR AauAL PERCEPnON)

The first conceptual domain that we analyze concerns perceptual events like seeing. hearing. feeling, tasting, and smelling. All the English sentences from (12)-(20). apart from examples (15) and (16), are about actually occurring sense perceptions although. literally. they either question or assert what we call the BEFORE condition of the Sense Perception Scenario. i.e .. they question or assert the ability for the particular sense perception. For example. in (12) and ( 13) the speaker wants to know whether the hearer actually sees someone or something. In ( 17) the speaker actually heard the sneering laughter .In (I 8) the gambler actually has a sense of being lucky. In ( 19) and (20) the taste and smell arc actually experienced. In each of these English examples the ability for the sense perception 'stands for' the actual perceiving. Thus the POTENTIALITY FOR Aid D. Clarke and 7A:.ie Todll

III. Part - whole I. Part - whole 2. Act- complex. act 3. Central factor- institution IV. Container- content I. Container - content 2. Loca1ity - occupant 3. CosiUme - wearer V. Experience - convention I. Experience - convention 2. Manifestation -definition VI. Possessor - possession I. Possessor- possession 2. Office holder -office

To analyze and compare these various typologies would require a separate paper. In what follows. we shall see how one can use these classifications for the study of metonymy in language acquisition.

3.

Compelled metonymical overextensions

Studies of how children produce and, especially. how they understand meta-

phor. proliferate in the literature on language acquisition in general and semantic development in panicular. However. no work has been done up to now on metonymy in language acquisition, and the possibility or having metonymic overextensions in child language has hardly ever been noted berore (cf .• however. Melissa Bowerman's work. e.g .• Bowerman 1978: Gibbs 1994). Our first hypothesis was that metonymical relations might be exploited in overextensions produced by children up to age 2:5. We called these 'compelled metonymical overextensions· because they are based on the ract that at this age a child's vocabulary, category and conceptual systems are still relatively small and unstructured. This scarcity compels them to extend already known words to cope with increasing communicative needs. to comment on what they sec and to request what they want. A good overview or the research done on over- and underextensions is provided by Eve Clark in her book The Lexicon in Acqui.fition. She writes:

'Mummy,/ Like Being a Sandwich'

365

Children do not start out already knowing the meanings of the words they are auempting to pronounce and use. They must first a.~sign some meaning to them. and that takes time. In fact. children's earliest mappings of meanings onto forms diverge from adult usage in a variety of ways. (Clark 1993: 32)

One of the most important ways of mapping the meaning a child want" to express onto a fonn which he or she has already acquired is overextension, that is, by applying a word to members of the adult category and to members of other categories that are perceptually similar, conceptually contiguous or spatio-temporally related. Consider these examples: ( I)

a. ball for balls of all kinds. and also for round hanging lampshades. doorknobs. and the moon (based on perceptual similarity) b. door for corks, jar-lids. box-lids. and gates when wishing to have the relevant object opened or closed (based on functional similarity!

Eve Clark distinguishes between two types of overextensions: (i) ·overinclusions.' where children extend a tenn to other entities from the same taxonomy, e.g.: (2)

a. dada used for both father and mother b. baby used for self-reference and all children c. apple for apples and for oranges

and (ii) 'analogical extensions,' where children use a tenn for objects from other taxonomies on the basis of perceptual similarity, e.g.: (3)

cotty-bar.~ for abacus on the wall and picture of a building with a columned fac;adc b. wmb for centipede

a.

Whereas analogical overextensions are based on recognizing and construing similarities (and can therefore be regarded as the training-ground for use of metaphors). overinclusions are based on both perceptual similarity and conceptual contiguity (we shall later call them synecdochical overextensions, and they are the basis for the construction of taxonomies). That is.the words being elltended arc applied to instcmces of other categories within the same or an adja~;ent conceptual domain. For example. a child may over-extend a term within the domain of animals. as when hone is applied to goats. cows, and sheep or within the domain of vehicles. as when rruck is applied to buses.tractors, and vans: or within the domain of clothing, as when lrar is applied to a crown. (Clark 1993: 34)

Brigitte Nerlich, Dm,id D. Clarke and 7A:.ie Todll

366

However. it seems to us that Eve Clark should have discussed in more depth a third type of overextension ba'\ed not so much on conceptual contiguity. but on spatia-temporal or functional contiguity. As we shall see. such contiguities are based on visual perception and are therefore the precursors to actual metonymies. This type of overextension has been studied from Piaget and Vygotsky onwards under the heading of 'complexive associations: Those studying complcxive associations have claimed that object names initially encode complexive groupings that incorporate actions or locations associated with objects (cf. Huuenlocher and Smiley 1993: 240). Even the words themselves are said to be initially just auditory features of situations where they habitually occur ( ... ].In short, it has been suggested that young children group their experiences in a fundamentally different way than older children or adults - in tenns of co-occurring aspects of situations and that object names, rather than standing for particular types of objects. are just another type of associate. (ibid.: 222)

Huttenlocher and Smiley contest this claim and argue that the extension of children's early object words match those of adults to a large extent. However. after distinguishing more clearly between a word's extension or meaning and its usc, they found that in certain cases other than naming contexts, such as comments or requests. such ·overextensions' did occur: as when a child said toy when seeing a certain bag which habitually contained toys. or when a child said slroe when she saw a foot (ibid.: 2260. This means that children usc certain 'metonymically' based pragmatic strategies. just as adults do. so as to achieve certain communicative effects. When a child comprehends a word correctly but overextends it in production. Clark (1978) concludes the overextension is not the result of an immature or nonadult word concept but is rather a deliberate attempt to communicate concepts for which the appropriate word cannot be remembered. (Pease et al. 1989: 113)

By contrast, the children's semantic system as well as their systems of concepts and categories. although not altogether different from the adult ones. are as yet incomplete and unstructured and will have to be adjusted and modified in the years to come. Overextcnsions are therefore pragmatic strategies which are not completely different from aduh communicative strategies. They are based on three types of relationships which also give rise to polysemy historically. structure

'Mummy./ Like Being a Sandwich'

367

meanings synchronically and allow adull speakers to vary word meanings contextually. These relationships arc metaphorical ones. metonymical ones and synecdochical ones, all three based on Seto 's cognitive triangle. It seems that even very young children begin to see similarities. connections and cla'isinclusions. and that with age they gradually adjust to the way the adults categorize and form associations and restructure their linguistic and conceptual systems accordingly. After this initial training in seeing and understanding similarities, connections and class-inclusions. they can later exploit these connections more creatively. We can therefore distinguish between three types of overextensions: synecdochical: overinclusions exploiting relations such as conceptual contiguity inside a taxonomy (e.g .. GENUS-SPECIES, SPECIESSPECIES and SO on). 1 (ii) metonymical: overextensions based on perceptual, spatia-temporal and functional contiguity. (iii) metaphorical: analogical extensions based on perceptual similarity. (i)

It should be stressed that in the case of compelled metonymical overextcnsions words are not overextended to label an object for which the child ha'i already acquired a more appropriate name. For example, as soon as the right word for 'horse' has been learned. the use of woof or do~-: for 'horse' stops. In the process of language acquisition,the extensions of words are adjusted until the lexical field (and the categorical domain) to which they belong has subdivided in just the same way ao; that of the adult caretakers (cf. Barrett 1982). All this is not the case in creative metonymical shrinking where the metonymically used words may replace an already known word and cut across subdivisions within lexical fields and even across lexical fields. In short. compelled metonymies are used by children to cover up gaps in their tiny lexicons. whereas creative metonymies arc used to express something new by not using the aJready available words in their lexicons. In order to determine which types of metonymy are used in compelled metonymical overcxtensions, we looked at the examples accumulated in Barrett (1982) and the examples we could find in the corpus assembled by Braine (1976). Barrett ( 1982: 322ff) reports on some of the ovcrextensions produced by Hildegard as reported by Leopold (1939, 1949) at the one-word stage. These are of three types: synecdochical. metonymical and metaphorical overexten-

368

Brigitte Nerlich, Dm,id D. Clarke and 7A:.ie Todll

sions. We shall provide a selection of examples for each type: (i)

Synecdochical ovcrextensions:

(4)

a.

Papa for father,grandfather, mothcr(l;O); any man ( 1;2); more appropriate lexical items learned; Mama (I :3). Mwm (I ;.5)

b. Ball for balls (1:0): balloon/ball of yam (1;4); observatory dome (I :8): balls of tin foil and paper (called paper-ball). marbles/ovoid ball (called egg-balf)la spherical bead (called ball-beads) (I; II); more appropriate lexical items learned: balloon ( 1;10). beads ( 1;9) (ii)

Metonymical ovcrextensions:

(5)

a.

Wheel for a wheelbarrow wheel (1:8): a wheelbarrow (1:10): toy wagon/a ring (I ;II); more appropriate lexical item learned: wheelbarrow (I; II) [ba.;ed on part-whole relationship} b. Choo-choo for trains (I ;7); bradyscope (I :9); airplane/wheelbarrow (1;10); streetcar/a trunk (;I 1): more appropriate lexical item learned: airplane (I; II). streetcar {I: 11 ). wheelbarrow (I; II) (based on the relationship between the noise associated with an object and the object itself - one could call this 'onomatopoeic metonymy'}

(iii)

Metaphorical overextensions:

(6)

a. Boatfortoysailboat(I:IO):airship(l:ll) b. Milkbollle for milkbottles (I; 10): bottle containing white toothpowder (I; 10)

We also looked at a corpus of two-word utterances collected by Braine (1976). At this stage in acquisition. overinclusions seem to diminish. whereas compelled metonymies seem to increa.~. In the examples in (7) we indicate the metonymical relation that is exploited. as well as the illocutionary force of the utterance. (7) a. b. c.

David (I ;9}: want pocket. (container-contained} {request} David (1:10): here fix it. (action-object} {comment} David (1;10): here hello. (words-object} {comment} Situation: indicating or identifying toy telephone (cf. also example 81)

'Mummy, I Like Bei11g a Sandwich' d. e. f.

g. h.

j. k.

369

David (1;10): here more book Iinstrument-action 1{request} David (1:10): that hello. (see above) Situation: indicating the toy telephone David (1;10): want more spoon. (instrument-actionl{request} David (1;10); gimme that blow. [instrument-action){requcst} Situation: wants to blow the match out David (I :10): u [=?you) hello. [sec above) Situation: David is instructing me to talk on his toy telephone, there is imperative intonation David (1:10): more put in. [action for place) {request} Situation: has been putting tinker toys in their box, apparently wants to put somewhere the pieces the adults are using Jonathan (I :11.15): more book. (instrument-action) {request} Jonathan ( 1:1 I .15): all gone blow.[instrument-action) {comment} Situation: the match box is empty. i.e .. there arc no more matches to blow out. Jonathan (2;0); daddy eat. (action for object) {comment} Situation: occurred twice in the apparent sense of daddy's food. once it referred to a piece of bread he had taken from my plate

As one can sec. children with limited lexicons focus on one object or attribute of an object to achieve certain speech acts in a metonymical way. They focus on one salient feature in a set framework or frame of repeated interactions with the caregiver or parent. They say book if they want to read, blow when referring to a match. they say hello for telephone, and so on. Gradually. through interaction with the caregiver or parent who grants their request~. the children elaborate on their comments. Finally. the metonymical ovcrextensions, which function as place-markers in the interactional and conceptual framework. will be replaced by the 'proper' words allocated them by the adults, such as read, match and telephone.

4.

Creative metonymical shrinking

Compelled metonymical overextensions normally peter out by the age of 2:5. although they can still occur as the vocabulary continues to develop. Between age 4:0 and 5:0 one can begin to notice a more creative exploitation of the natural pathways of meaning which are metaphor and metonymy. These new

Brigitte Nerlich, David D. Clarke and ZA:.ie Todll

370

metonymies are based on what we call ·creative metonymical shrinking,· as they are usually produced in order to communicate new ideas with the least verbal effort. In contradistinction to compelled metonymical overextension. these metonymies arc produced despite the fact that the message could also have been conveyed using conventional linguistic signs and syntactic constructions. Metonymical shrinking occurs in a wide variety of linguistic forms (nouns. derivations. compounds), as described in Section 4.1 .. and conversion of nouns into verbs as described in 4.2. The following examples are based on Mauhew (Brigitte's son). For the purpose of the collection of examples we have used the concept 'metonymy' in a broad sense, including not just strictly lexical metonymies, but also ·morphological metonymies,' that is. derivations and compounds based on metonymic associations.

4.1.

Creative metonymies (8)

a.

Mauhew (4:11) (The 'Title story") Manhew started S(:hool in January. At first we thought he might eat the Sion

No

due~

CluQ.

Figw·e 2. CompreiU'n.fimr by tt•xt ()1Jt' for both agt• groupJ

This experiment shows that the four and fivc-ycar-olds have a significantly bencr understanding of metonymy than the two and threc-yeur-olds. which is what might be expected from what has been said elsewhere in this paper and from experimental studies of metaphor comprehension (e.g .. Vosniadou 1987). In addition this experiment shows that children arc able to use clues in the text of a story to help them to understand metonymies. More evidence would be needed to establish whether clues such as this enable children to develop some understanding of metonymy in the first place. or if they are just used in some instances once children already have some basic knowledge of how metonymy works. One observation that could be worth following up is that for the metonymy Twemy sailor hat.~ were marching down the gangphmk. most of the children who correctly selected the picture of

'Mumm.v, I Like Being a Sandwich'

379

sailors pointed to the sailors hats rather than another part of the picture. The children may have realized that hats cannot march on their own and used that as the basis for selecting a picture (although clearly there are many children's cartoons and books in which it is possible for inanimate objects to march. walk and even talk). ll may be that children's general cognitive development or understanding of the real-pretend distinction is as important as any understanding of metonymy itself. Further experiments could also be carried out to investigate other aspects of metonymy comprehension and in particular to see if there are links between development in comprehension of metaphor and metonymy.

6.

Conclusion

Understanding the use and comprehension of metaphor and metonymy in language acquisition has important implications for language teaching, especially for reading and comprehension and later on for writing and composition. Without the knowledge of how to understand and use words with multiple meanings. how to exploit subtle shades in meaning to their advantage, and how to multiply the meanings of words themselves, children's chances of acquiring literary skills would be severely reduced. It is necessary to build on what Gibbs has called the 'poetic minds' of children to enhance and develop their semantic competence.

Note An example of a (rather rare) SI'H'IE.~ R>R Gf.NL'S synecdoche can be found on a toy recemly bought at a "Scalife Centre."" It consists of a plastic dolphin. whale. shark or alligator which you put into water where it expands 200 times in size. On the package it ~ays: "PUI scalife imo a pan, sink or pool.. " ··Scalife can grow to 22 inches. wei'' (on ~ynccdoche in general. cf. Nerlich in print and Ncrlich and Clarke in prinl).

References Anonymous [Cicero) 1954 Ad C. Herem1i11m. de ratione tlic·encli (Rhetorim ad Herenllium}. london: Heinemann.

380

Brigitte Nerlich. David D. Clarke and Zazie Todtl

Barrett. Martyn 0. 1982 Distinguishing between prototypes: the early acquisition of the meaning of object names. In S.A. Kuczaj (cd.).IAIIRUOflt' Ot'\'elopment. Vol. 1: Sylllax and Semantics. Hillsdale. NJ: Lawrence Erlbaum, 313-334. Bowerman, Melissa 1978 Systematizing scmanlic knowledge: changes over time in the c:hild's organization of meaning. Child De\·dopmelll 49: 977-987. Draine, Martin D.S. 1976 Cllildren'.t Fir.tt Word Combinations. Monographs of the Society of Resean:h in Child Devclopment41 (Serial No. 164).

BrCal. Michel 1991 (1884) How words are organized in the mind. In G. Wolf (ed.), The Beginnings of Senumtic.f: Es.my.t. Ucmre.t and ReviewJ. London: Duckworth. 145-151. Clark. Eve V. 1978 Strategies for communication. Child Development 49: 953-959. 1982 Th~ young word maker: a case study of innovation in the child's lexicon.ln E. Wanner. L.R. Gleitman (eds.). Language Acqui.filion: Till.' Stale of the Arr. Cambridge. Massachusetts: The MIT Press, 390-428. 1993 The Uxictm in Acqui.fition. Cambridge: Cambridge University Press. E.o;nault, Gaston 1925 L 'imaginalion populaire: Mitaphores occidenlales. Paris: Presses Univer· sitaires de France. Fass.Dan 1991 Metonymy, ca.-;e role substitution and sense ambiguity. In D. Fass et al. (cds.). Proceedinxs of the /JCAJ Wori.:.Jhop ml Computational Approaches to Non-lJteral Language: Metaphor, Metonymy, Idiom, Speech Act.f and /mplicawre. held at the Twelfth International Joint Conference on Anificial Intelligence (24 August 1991. Sydney, Australia). 42-51. Gibbs, Raymond W. 1994 The Poetics of Mind: Figurati1•e Tlroughl, Lnnguage, and Under.ftondinx. Cambridge: Cltmbridge University Press. Hous~holder. Fred W. 1995 Plato and his predecessors. In E.F.K. Koerner. R.E. Asher (eds.), Conci.fe History of the Wnguage Scienas. Oxford: Pergamon, 9-93. Huttenlochcr. Janellen. Patricia Smiley 1993 Early word meanings: the case of object names. In P. Bloom (ed.). l.an· guage Acquisition. New York: Harvester and Wheatsheaf. 222-247. (First published 1987) Lakoff. George, Mark Johnson 1980 Metaplror.f We Li1·e By. Chicago: The University of Chicago Press. Leopold. Werner F. 1939 Speech De1•elopment of a Bilingual Child: A Lingui.f(.f Remrd. Vol. 1: Vm:abulary Growt/1 in the First Two Yean. Evanston, IL: Nonhwestcm University Press.

'Mummy,/ Like Being a Sandwich'

381

Spuch Devdopmem of a Bilingual Child: A Lingui.u".f Ret·ord. Vol. 3: Grammar and General Problems i11 tile First Two Years. Evanston. IL: Nonhwestem University Press. Nerlich. Brigitte in print Synecdoche: a trope. a whole trope. and nothing but a trope? InN. Norrick. A. Burkhardt (eds.). Tmpil-af Trutlt. Amsterdam and Philadelphia: Benja· mins. Nerlich. Brigitte, David D. Clarke 1988 A Dynamic model of semantic change. Joumal of literary Semtmlin 17: 73-90. in print Synecdoche as a cognitive and communicative strategy. In A. Blank. P. Koch {eds.), Hi.florical Semami1·s and Cognition. Berlin: Walter de Gruyter. Norrick, Neal R. 1981 Semiotic Principles in Semamic Theory. Amsterdam and Philadelphia: Benjamins. Nyrop, Kristoffer 1913 Gran111wire Hi.florique de fa Langue Franrai.fe. Vo/4: Sbnatttique. Copenhagen: Gyldendalske Boghandel Nordisk Forlag. Pease, Deborah Myers. Jean Berko Glea.'iOn. Barbara Alexander Pan 1989 Gaining meaning: semantic development. In J.D. Gleason (ed.). The De••ef. opmem of Longua~:e. Second edition. Columbus. etc.: Merrill, 101-134. Sappan. Raphael 1987 The Rhetoricai-Lngica/ Cla.uification of Semantic Changes. Braunton: Merlin Books. Seto, Ken-ichi in print Metonymy and the cognitive triangle. Stem. Gustaf 1968 (1931] MeaninR and ChanRe of Meaning. Bloomington: Indiana Univer· sity Press. Stimson. Joan 1992 Jake and the Jumble Panic.ln J. Stimson, Bedtime Stories for Under Fives. Paignton: Ladybird. Ullmann. Stephen 1957 ( 1951} The Principle.f of Semantic.f. Oxford: Basil Blackwell. Vosniadou, Stella 1987 Children and metaphors. Child Development 58: 870-885. Yamanashi, Masa·Aki 1987 Metonymic interpretation and a.o;sociative processes in naturo~.llanguage.ln M. Naga (ed.), LonRIIOge and Artificiol/ntelligena,/nternationaf S·ympo· .fillm on Language and Artificial Intelligence ( 16--21 March 1986, Kyoto. Japan). Nonh Holland: Elsevier Science Publishers, 77-86. 1949

382

Brigitte Nerlich. David D. Clarke and Za:ie Todd

Appendix The Pirate Story used in the comprehension experiment: 'The metonymic treasure hunt.' Tom and Mary had been promised a special adventure for their binhday. For one day they would be able to live and work on a pirate ship. When they came to the harbor. drcs~ed up

as pirates, there she wa~. the Black Princess swaying gently on the waves. "What a beauty." said Tom. ''But where arc the pirates?" asked Mary. And there they came twenty sailor hats were marching down the gang plank, stood in a row before them and saluted them. Mary and Tom wanted to greet them. but then they reali1.cd that they didn't know their name!>. "What are your names?" asked Tom. "Har-har, my dcars,this is the first mystery you have to solve." said the captain of the ship. who had just stepped in front of the pirates as if from nowhere. "My name is Captain Halty, by the way." announced the Captain. Look at the sailors closely and try to find out what their names are. Peggy (-leg) (Eye) Patchy Stripy Dotty Tatty and Fatty Now the children knew the names of the pirates and everybody went aboard. "All hands on deck:' cried the captain. "set sails!" And off they went. The Black Princess was heading for adventure. When it was lunch time. the ship's cook shouted "Twelve bells. Twelve bells," come and eat your grub. Everybody sat down in the tiny galley. The cook. Fatty. brought the meal. and looking mund. he said jokingly: "Oh dear. there are two more mouths to feed. Let's hope we have enough fish fingers!" To celebrate the arrival of Tom and Mary. the captain got out a bottle of lemonade and everybody got a glass. But just as they were about to speak a toast. the ship wobbled a bit and the lemonade tipped over. but what was worse. the treasure-map. that the captain was just about to show to the children. :.lid on the floor and got a bit dirty. Quickly. Tom jumped up and saved it before it got all dusty. The captain took it and nailed it down on the table. to be on the safe side. After a few days they arrived on a de~n island. Everybody got off the ship and explored the beach. Tom had brought a camera and said to Mary. "Stand there between the rocks and do a Captain Hook for me!" Then they all went into the deep dark jungle. The sun shone down and everybody got thirsty. Mary said: ··1 wish I could have a Ribena." "No Ribena here my deary." said the captain. "you'll have to do with this nice fresh spring water." They all had a drink and went on. Finally they arrived at the spot where there was a cross on the treasure map and started digging. They dug and they dug. But they couldn't find anything. Then Tom had a look at the map that Dotty had pinned to a tree and said: "But it's upside down!"lbc captain took Dotty by his collar and shook him.saying "Don't you have any brain! We have wasted half a day now. let's have a rest. before we start the search again:· They all put their heads down for an hour and then started looking for the treasure again - this time in the right spot. And after a lillie digging. the whole crew shouted: "There it is!" They could all sec a

'M11mmy, I Like Bein~: a Sandwich'

383

huge treasure chest. They opened it and found gold coins, jewels and, what a surprise,toy~ for Tom and Mary. Everybody was very pleased. They went back to the ship and sailed home.

Recontextualization of Metonymy in Narrative and the Case of Morrison's Song of Solomon

Anne Pankhurst University of Edinburgh

1.

Introduction

Metonymic referentiality in narrative fiction is associated with a number of powerful effects. achieved by recurring references to a concept. experience or object. The scope of a referent is extended from a single episode to the entire text, assisting in the development of a communicative linguistic interface between writer and reader. Poetic thought processes like metonymy arc fundamental to our way of thinking. so the daim that we understand poetry by means of our everyday experiences (Lakoff and Turner 1989: Gibbs 1994) is also relevant to the consideration of poetic clements in narrative prose fiction. The reader then activates cognitive strategies which facilitate access to the new context or world of experience proposed in a text. In the case of Toni Morrison·s novel Sottg of So/omo11 (1977 (1989)). scene. plot and characters are opaque to readers unfamiliar with the social and ethnic background to the fictional narrative. Among various means used by Morrison to make this world credible are recurring references to an earring made and worn by the central character. Pilate. The apparently simple object described in the text is a means of identifying its wearer. enabling us to answer the question: who is Pilate? The answer is available within the fiction and at

Amre PwrkhurJt

386

the same time evokes the real world of the reader's experience. It is more than a description. The earring has numerous features which refer the reader to more than one concept. This metonymy also has a number of meta-narrative functions serving to structure the overall development of thematic material in the text. Its recurrence in new contexts is an important element of the narrative. It reactivates the reader's memory and highlights important episodes.

2.

Theories of metonymy in narrative

2 .1 Jakobson on metonymy Theories of the functions of metonymy in narrative arc of interest to the analysis both of literature and of oral narrative, e.g .. Propp's (1928) analysis of Russian folk-tales, Labov's (1972) account of natural narrative. Here. I shall limit the discussion to literary narrative in prose fiction. and to the wellknown theories of Jakobsen and Riffatcrrc. Starting with the linguistic analysis of text, both theories select metonymy as a regular semantic principle of substitution which extends beyond a given linguistic domain and the syntagmatic organization of a sentence. While Jakobsen is concerned with extending the semantic principle of contiguity to the syntactic structure of discourse. Riffaterrc emphasizes the conceptual contiguity created by repetition of a metonymic association once it has been recognized. Jakobson (1956) proposed a theory in which the nature of discourse is defined a

Metonymy in language and thought

Mappings in Thought and Language

Metonymy in Language and Thought (Discourse Approaches to Politics, Society and Culture)

Logic, Thought and Language

Language, Thought, and Reality

Thought and Language

Language and Thought

Paradigms, Thought and Language

Language, Thought and Reference

Logic, Thought and Language

Figurative Language and Thought

Language and Thought

Language and Thought

Thought and Language

Color Categories in Thought and Language

Color Categories in Thought and Language

Cultural Models in Language and Thought

Cultural Models in Language and Thought

Cultural Models in Language and Thought

Language and Thought: Interdisciplinary Themes

Language, Thought, and Information Processing

Aune on Thought and Language

Language, Thought, and the Brain

Metonymy and Pragmatic Inferencing

Thought and Language - Revised Edition

The language of thought

Poetry, Language, Thought

The Language of Thought

Poetry, Language, Thought

The language of thought (The Language & thought series)

Figurative Language and Thought (Counterpoints: Cognition, Memory, and Language)

Metonymy in language and thought

Mappings in Thought and Language