Defining Metonymy in Cognitive Linguistics: Towards a Consensus View 9027223823, 9789027223821

Table of contents :
Defining Metonymy in Cognitive Linguistics
Editorial page
Title page
LCC data
Table of contents
List of contributors
Introduction
1. Background
2. Aim and structure of the volume
References
Reviewing the properties and prototype structure of metonymy
1. Introduction
2. Problematic properties affecting the standard cognitive-linguistic notion of metonymy
2.1 “Entities” or “domains”?
2.2 Is metonymy necessarily connected to an act of reference?
2.3 What is meant by saying that metonymy is a “stand-for” relationship?
2.4 Mapping, “domain highlighting”, activation, and directionality in metonymy
2.5 whole for part metonymies, degrees in metonymicity and prototype-based vs. unitary definitions of metonymy
2.6 Metonymic and non-metonymic “reference point phenomena”
2.7 Is there any difference between metonymy and such phenomena as “contextual modulation”, “facets”, and active-zone / profile discrepancy?
2.8 The conventionalization of metonymy
3. Problematic properties affecting the standard criteria in the distinction between metaphor and metonymy
3.1 Problems derived from the notion of “domain”
3.2 A linguistic expression may sometimes be interpreted, on the basis of context, background knowledge or the interpreter’s choice, as metaphorical, metonymic, or as a combination of both
3.3 Interaction of metaphor and metonymy
4. Summary and conclusions. Revised general definitions of metonymy and metaphor
4.1 Revised definitions of schematic metonymy and of metaphor
References
Part I. Metonymy and related cognitive, semantic, and rhetorical phenomena
Metonymization: A key mechanism in semantic change
1. Introduction
2. Lexical Meaning as Ontologies and Construals (LOC)
3. Metonymization and zone activation
4. Metonymization and change
5. From contentful to contentful meanings
6. From contentful to configurational meanings
7. Conclusion
References
Zones, facets, and prototype-based metonymy
1. Introduction
2. Dogs and cigarettes
3. Kettles and telephones
4. Books and trumpets
4.1 Pragmatics
4.2 Syntax
4.3 Semantics
5. Conclusions
References
Metonymy and cognitive operations
1. Introduction
2. Types of cognitive operation
2.1 Formal operations
2.2 Content operations
3. Conclusion
References
Metonymy, category broadening and narrowing, and vertical polysemy
1. Introduction
2. Vertical polysemy
3. Metonymy
4. Metonymy and vertical polysemy in encyclopedic semantics
4.1 Metonymy in encyclopedic semantics
4.2 Vertical relations in encyclopedic semantics
5. Vertical polysemy and metonymy: Discussion
6. Conclusion
References
Metonymy at the crossroads: A case of euphemisms and dysphemisms
1. Introduction
2. Metonymy in euphemisms and dysphemisms
3. Case studies
3.1 The euphemism body count
3.2 The dysphemism chickenhawk
4. A glance at conceptual and linguistic borrowing
5. Conclusion
References
The role of metonymy in complex tropes: Cognitive operations and pragmatic implications
1. Introduction
2. icms and tropes
3. The role of metonymy in complex tropes
3.1 Irony
3.1 Irony
3.2 Oxymoron
3.3 Overstatement
3.4 Understatement
3.5 Euphemism and dysphemism
4. Conclusion
References
Part II. Metonymy and metonymic chains as mappings or processes within domain matrices/networks
Putting the notion of “domain” back into metonymy: Evidence from compounds
1. The single domain-based approach to metonymy
What do metonymic chains reveal about the nature of metonymy?
2. Why domains are indeed necessary:Evidence from noun–noun compounds
3. Metonymy as a domain network of conceptual relationships
4. Contiguity or association?
5. Domains: Tentative evidence from psycholinguistics
6. Conclusion
References
Metonymic matrix domains and multiple formations in indirect speech acts
1. Introduction
2. Ruiz de Mendoza’s double metonymy
3. Matrix domains and multiple formations
3.1 Matrix domains with one source and various possible target domains
3.2 Matrix domain with various sources and one target
4. Conclusion
References
Appendix 1
Appendix 2
Authors’ biodata
Metaphor and metonymy index
Name index
Subject index

Citation preview

Defining Metonymy in Cognitive Linguistics

Human Cognitive Processing (HCP)

Cognitive Foundations of Language Structure and Use This book series is a forum for interdisciplinary research on the grammatical structure, semantic organization, and communicative function of language(s), and their anchoring in human cognitive faculties. For an overview of all books published in this series, please see http://benjamins.com/catalog/hcp

Editors Klaus-Uwe Panther University of Hamburg

Linda L. Thornburg

Editorial Board Bogusław Bierwiaczonek

Higher School of Economics and Humanities, Poland

Mario Brdar

Josip Juraj Strossmayer University, Croatia

Barbara Dancygier

University of British Columbia

N.J. Enfield

Max Planck Institute for Psycholinguistics, Nijmegen & Radboud University Nijmegen

Elisabeth Engberg-Pedersen University of Copenhagen

Ad Foolen

Radboud University Nijmegen

Raymond W. Gibbs, Jr.

University of California at Santa Cruz

Rachel Giora

Tel Aviv University

Elżbieta Górska

University of Warsaw

Martin Hilpert

Freiburg Institute for Advanced Studies

Zoltán Kövecses

Eötvös Loránd University, Hungary

Teenie Matlock

University of California at Merced

Carita Paradis

Lund University

Günter Radden

University of Hamburg

Francisco José Ruiz de Mendoza Ibáñez University of La Rioja

Doris Schönefeld

University of Leipzig

Debra Ziegeler

Paul Valéry University, France

Volume 28 Defining Metonymy in Cognitive Linguistics. Towards a consensus view Edited by Réka Benczes, Antonio Barcelona and Francisco José Ruiz de Mendoza Ibáñez

Defining Metonymy in Cognitive Linguistics Towards a consensus view Edited by

Réka Benczes Eötvös Loránd University

Antonio Barcelona University of Córdoba

Francisco José Ruiz de Mendoza Ibáñez University of La Rioja

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

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.

Library of Congress Cataloging-in-Publication Data Defining metonymy in cognitive linguistics : Towards a consensus view / edited by Réka Benczes, Antonio Barcelona, Francisco José Ruiz de Mendoza Ibáñez. p. cm. (Human Cognitive Processing, issn 1387-6724 ; v. 28) Includes bibliographical references and index. 1. Metonyms. 2. Cognitive grammar. I. Benczes, Réka, 1974-. II. Barcelona, Antonio, 1952-. III. Ruiz de Mendoza Ibáñez, Francisco José, 1961-. P301.5.M49D44   2011 401’.43--dc22 2011011880 isbn 978 90 272 2382 1 (Hb ; alk. paper) isbn 978 90 272 8676 5 (Eb)

© 2011 – 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. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

Table of contents

List of contributors Introduction Antonio Barcelona, Réka Benczes, and Francisco José Ruiz de Mendoza Ibáñez Reviewing the properties and prototype structure of metonymy Antonio Barcelona

vii 1

7

Part I.  Metonymy and related cognitive, semantic, and rhetorical phenomena Metonymization: A key mechanism in semantic change Carita Paradis

61

Zones, facets, and prototype-based metonymy Dirk Geeraerts and Yves Peirsman

89

Metonymy and cognitive operations Francisco José Ruiz de Mendoza Ibáñez

103

Metonymy, category broadening and narrowing, and vertical polysemy Anu Koskela

125

Metonymy at the crossroads: A case of euphemisms and dysphemisms Tanja Gradečak-Erdeljić and Goran Milić

147

The role of metonymy in complex tropes: Cognitive operations and pragmatic implications Javier Herrero Ruiz

167

Part II.  Metonymy and metonymic chains as mappings or processes within domain matrices/networks Putting the notion of “domain” back into metonymy: Evidence from compounds Réka Benczes

197

vi

Defining Metonymy in Cognitive Linguistics

What do metonymic chains reveal about the nature of metonymy? Rita Brdar-Szabó and Mario Brdar

217

Metonymic matrix domains and multiple formations in indirect speech acts Xianglan Chen

249

Authors’ biodata Metaphor and metonymy index Name index Subject index

269 275 277 281

List of contributors

Antonio Barcelona Catedrático de Universidad Departamento de Filología Inglesa y Alemana Facultad de Filosofía y Letras Universidad de Córdoba 14071 Córdoba Spain [email protected]

Xianglan Chen School of International Studies University of International Business and Economics Huixin Dongjie No. 10 Chaoyang District 100029 Beijing China [email protected]

Réka Benczes Eötvös Loránd University Department of American Studies H-1088 Budapest Rákóczi út 5. Hungary [email protected]

Dirk Geeraerts QLVL, University of Leuven Blijde-Inkomststraat 21 PO Box 3308 3000 Leuven Belgium [email protected]

Mario Brdar University of Osijek Faculty of Philosophy L. Jägera 9 31000 Osijek Croatia [email protected] Rita Brdar-Szabó Eötvös Loránd University Faculty of Arts School of Germanic Studies H-1088 Budapest Rákóczi út 5. Hungary [email protected]

Tanja Gradečak-Erdeljić Department of English Faculty of Philosophy Josip Juraj Strossmayer University of Osijek L. Jägera 9 31000 Osijek Croatia [email protected] Javier Herrero Ruiz (aff. University of La Rioja, Spain) c/ Berones 10, 7ºC 26580 Arnedo-la Rioja Spain [email protected]

viii Defining Metonymy in Cognitive Linguistics

Anu Koskela School of English University of Sussex Falmer Brighton BN1 9QN United Kingdom [email protected]

Yves Peirsman QLVL, University of Leuven & Research Foundation – Flanders Blijde-Inkomststraat 21 PO Box 3308 3000 Leuven Belgium [email protected]

Goran Milić Department of English Faculty of Philosophy Josip Juraj Strossmayer University of Osijek L. Jägera 9 31000 Osijek Croatia [email protected]

Francisco José Ruiz de Mendoza Ibáñez University of La Rioja Department of Modern Philologies c/ San José de Calasanz s/n 26004, Logroño, La Rioja Spain [email protected]

Carita Paradis Lund University Centre for Languages and Literature Box 201 SE-221 00 Lund Sweden [email protected]

Introduction Antonio Barcelona, Réka Benczes, and Francisco José Ruiz de Mendoza Ibáñez Universidad de Córdoba / Eötvös Loránd University / University of La Rioja

1.

Background

When George Lakoff and Mark Johnson published their seminal book, Metaphors We Live By in 1980, their discussion of conceptual metonymy received only a single chapter. Although the authors made note of the fact that “[m]etonymic concepts… are part of the ordinary, everyday way we think and act as well as talk” (p. 37), the title of the book implicated that the main cognitive mechanism guiding human conceptualization was conceptual metaphor. Nevertheless, this partiality towards metaphor has gradually waned over the past thirty years. Perhaps the first major step in this process was the publication of the volume Metonymy in Language and Thought (1999), in which the editors, Panther and Radden, pointed out that “metonymy is a cognitive phenomenon that may be even more fundamental than metaphor” (p. 1). Further collections of papers in especially ­ Barcelona (2003/2000), Dirven and Pörings (2002), Panther and Thornburg (2003) and ­Panther et al. (2009) have confirmed that Panther and Radden’s (1999) speculation on the nature of metonymy can now be stated in the affirmative. Nevertheless, while the cognitive linguistics community is essentially in agreement with both the conceptual nature and the fundamental importance of conceptual metonymy, there remain a number of important disagreements on a number of specific issues. For instance, one of the main problems arises from the . A similar view had also been put forth by Kövecses and Radden (1998). . Empirical research from psycholinguistics has not yet been able to demonstrate the ubiquity/ fundamentality of metonymy in human conceptualization. However, Gibbs (2007) points out that this lack of experimental evidence does not necessarily undermine the claim that people constantly employ metonymy in understanding; it is simply an indication of the technical difficulty associated with creating viable experimental tests for metonymy.



Antonio Barcelona, Réka Benczes, and Francisco José Ruiz de Mendoza Ibáñez

claim that metonymy operates within the same conceptual domain, whereas metaphor connects two different conceptual domains. In response to the existence of problematic areas within the Cognitive Linguistics approach to metonymy, ­Antonio Barcelona and Francisco José Ruiz de Mendoza Ibáñez convened a theme session entitled “What do we want to call a metonymy? An attempt at building a consensus view on the limits of metonymy in Cognitive Linguistics” at the 10th International Cognitive Linguistics Conference (Cracow, Poland, 2007). The aim of the session was to spark scholarly debate – and to come up with a consensus view, if possible – on the following issues related to metonymy, among others: – Is metonymy a relationship between “entities” or a relationship between “domains”? – Is metonymy necessarily connected to an act of reference? – What is meant by saying that metonymy is a stand-for relationship? – Is metonymy a mapping? – Is metonymy unidirectional or bidirectional? – What exactly is meant by claiming that metonymy is a type of “domain highlighting”? – What exactly is meant by claiming that metonymy is a type of activation? – What qualifies as a target in whole for part metonymies? – Are there any degrees in metonymicity? Furthermore, are some metonymies more prototypical than others? – Is there any difference between metonymy and “reference point phenomena”? – Is there any difference between metonymy and such phenomena as active-zone/profile discrepancy, modulation, and facets? – What factors determine the linguistic conventionalization of a metonymy?

2.

Aim and structure of the volume

By virtue of the large amount of research on metonymy within Cognitive Linguistics, it is now accepted that beyond its traditionally acknowledged role in lexical semantics, metonymy plays a crucial part in the motivation of numerous conceptual metaphors, in prototype categorization, in certain types of symbolism and iconicity, in pragmatic inferencing, in blending and conceptual integration, in gestures, etc. However, though most of this research reflects a basic agreement on the notion of metonymy, it is necessary to delimit this notion more precisely so that these research results and the results of future research on the role of metonymy in language and cognition can be fruitfully compared and evaluated. In



Introduction

sum, we now need to arrive at a consensus view on metonymy, or at least to clarify the main points of disagreement and the different ways in which metonymy is defined and treated within the field. The chapters in the present volume are the revised versions of papers presented by most of the participants at the metonymy theme session in Cracow; however, the volume also includes a number of invited contributions from leading researchers on metonymy. The chapters address the core issues of the notion of metonymy on the basis of the latest research results. In this respect, the present volume takes an innovative and unprecedented approach by being devoted exclusively to (i) delimiting the notion of metonymy in the current state of development of the different strands of Cognitive Linguistics (cf. Dirven, 2005; Dirven and Ruiz de Mendoza, 2010), (ii) clarifying the points of divergence between the various contributors with respect to this notion, and (iii) suggesting a consensus view which will hopefully have reverberations in both the Cognitive Linguistics community and the linguistics community at large. Since all of the chapters have theoretical import, as they attempt to delimit the notion of metonymy from one or more perspectives, in addition to exploring its application in certain areas, a neat classification into mutually exclusive thematic sections would seem to be rather difficult, given that many chapters discuss several definitional issues in metonymy theory and are, therefore, closely interconnected with one another. On closer introspection, however, one can see that the chapters can be grouped into two broad sections, depending on what particular topic within the vast field of metonymy they touch upon (and which set of questions listed above they discuss): “Metonymy and related cognitive, semantic and rhetorical phenomena” (comprising Part I) and “Metonymy and metonymic chains as mappings or processes within domain matrices/networks” (comprising Part II). Preceding the two thematic sections, the discussion on metonymy is launched by Antonio Barcelona’s introductory chapter, which “sets the scene” for all the other chapters by tackling the problem areas raised above. Barcelona takes a clear stand on every one of these issues, arguing – for instance – that metonymy is indeed a relationship between domains rather than entities, that some metonymies are indeed more prototypical than others, and that metonymy, as a mapping causing the activation of the target, can only occur if source and target are conceptually linked by a “pragmatic function” within a “functional domain” (i.e., a frame or icm); furthermore, the “mapping” in metonymy is claimed by him to . We would like to thank the three workshop discussants, Günter Radden, Zoltán Kövecses, and Klaus-Uwe Panther, for their highly constructive and debate-conducive participation. The chapters in the present volume have greatly profited from their input.





Antonio Barcelona, Réka Benczes, and Francisco José Ruiz de Mendoza Ibáñez

be “asymmetric” (unlike metaphor, which is a complex mapping resulting in the “symmetric” matching of source–target counterparts) and to impose a conceptual perspective on the target. Barcelona’s chapter not only “plots the field”, as the title of the section indicates, but also reflects upon many of the other chapters in the volume, thereby creating a snapshot of the current debate on metonymy. Then, Part I, “Metonymy and related cognitive, semantic and rhetorical phenomena”, is mainly devoted to how words can acquire new meaning via metonymy. Needless to say, such a topic inevitably cuts across the interrelated issues of active zones, facets, polysemy and the nature of mappings, as well as the question of how metonymy effects linguistic conventionalization and underlies or interacts with certain other tropes. Accordingly, Carita Paradis maintains that metonymy has a fundamental role in the development of new meanings in language change, as it involves the use of a lexical item to evoke the sense of something that is not conventionally linked to that particular lexical item. She postulates a continuum from metonymy to zone activation in the process of meaning change, as it presupposes polysemy, occurs by semantic leaps and is gradual. Dirk Geeraerts and Yves Peirsman compare and contrast profile/zone discrepancy and facetization, and claim that since the former is a broader phenomenon than the latter, its instances should not in all cases be referred to as “metonymy”, as they do not involve a shift in reference, but are based upon various selectional restrictions of the participating entities; on the other hand, they argue that the criteria usually given to systematically distinguish between facets and metonymy do not always work. Francisco José Ruiz de Mendoza Ibáñez argues that it is not sufficient to state that mappings are “sets of correspondences”, because this definition does not reveal how the elements are perceived to correspond. Therefore, he refines this description by taking into consideration a number of possible cognitive operations that can help determine the nature of a correspondence. As a consequence, metonymy is revealed as a cover term for the cognitive processes of domain expansion and domain reduction (the latter of which is the result of either meaning facetization or domain highlighting). Moreover, Ruiz de Mendoza gives an account of the various cognitive operations that reinforce metonymy, such as cueing, selection and integration, and how they individually affect the metonymic process. Anu Koskela concentrates on the relationship between metonymy on the one hand and category broadening and narrowing on the other, and the resulting state of multiplicity of meaning, i.e., vertical polysemy. Broadening and narrowing have been viewed by some authors as being motivated by a metonymic process where a category member stands for the whole category or vice versa, but, according to Koskela, vertical polysemy differs from metonymy both in terms of the contiguity relations and the underlying conceptual structures involved. Tanja Gradečak-Erdeljić and Goran



Introduction

Milić focus on the status and function of metonymies as ­inference schemas in euphemistic and dysphemistic political discourse. According to their analysis, these metonymies are understood as a chain of pragmatically triggered links between the source and target concepts, thereby supporting the notion of the pragmatic function criterion as a guidepost for determining the ensuing pragmatic effect of the metonymic mappings. Concluding Part I, Javier Herrero Ruiz demonstrates that the basic meanings of various tropes (such as irony, oxymoron, overstatement, understatement, euphemism and dysphemism) can be distinguished with the help of metonymy. Part II, “Metonymy and metonymic chains as mappings or processes within domain matrices/networks”, investigates the influence of online meaning construction on our understanding of the notions of “domain” and “mapping”, and through them, on our understanding of the notion of metonymy. Réka Benczes explores the role that domains play in conceptual metonymy, by examining the semantics of metonymical (and metaphorical) neologisms. According to her analysis, domains are best understood as networks of semantic associations, with links to further semantic domains or even other grammatical constructions. Therefore, the chapter proposes that metonymy operates within a domain network, where the domains are interconnected by an intricate, web-like set of semantic associations. According to Rita Brdar-Szabó and Mario Brdar, whether metonymy can be called a mapping should be decided by examining the function(s) of metonymy in discourse, that is, by investigating its role in the online construction of meaning. Based on their rich source of data, the authors come to the conclusion that metonymy should be best approached as inference-based elaboration (either expansion or reduction) of the metonymic source. Chen Xianglan proposes that a metonymic mapping can be viewed as a kind of pragmatic inference. It is claimed in his chapter that cognitive background knowledge, the immediate context and individual pragmatic factors, all of which can be claimed to constitute “metonymic triggers”, influence metonymic thinking. By analyzing various examples of language use, Chen maintains that metonymic inference can be best explained with reference to a matrix domain, where the “metonymic triggers” all guide and simultaneously constrain the metonymic thinking process.

References Barcelona, Antonio (ed.). 2003/2000. Metaphor and Metonymy at the Crossroads: A Cognitive Perspective. Berlin & New York: Mouton de Gruyter. Dirven, René and Ralph Pörings (eds.). 2002. Metaphor and Metonymy in Comparison and Contrast. Berlin & New York: Mouton de Gruyter.





Antonio Barcelona, Réka Benczes, and Francisco José Ruiz de Mendoza Ibáñez

Dirven, René. 2005. Major strands in cognitive linguistics. In Ruiz de Mendoza, Francisco José and Sandra Peña (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (69–100). Berlin & New York: Mouton de Gruyter. Dirven, René and Francisco Ruiz de Mendoza. 2010. Looking back at 30 years of Cognitive Linguistics. In Tabakowska, E., M. Choiński, and Ł. Wiraszka (eds.), Cognitive Linguistics in Action: From Theory to Application and Back. Berlin & New York: Mouton de Gruyter. Gibbs, Raymond W., Jr. 2007. Experimental tests of figurative meaning construction. In ­Radden, Günter, Klaus-Michael Köpcke, Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (19–32). Amsterdam & Philadelphia: John Benjamins. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9(1): 37–77. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago: University of Chicago Press. Panther, Klaus-Uwe and Günter Radden (eds.). 1999. Metonymy in Language and Thought. Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg (eds.). 2003. Metonymy and Pragmatic Inferencing. Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe, Linda L. Thornburg, and Antonio Barcelona (eds.). 2009. Metonymy and Metaphor in Grammar. Amsterdam & Philadelphia: John Benjamins.

Reviewing the properties and prototype structure of metonymy* Antonio Barcelona Universidad de Córdoba

Building on previous research (Barcelona, 2002a, 2003a), the author carefully discusses and takes a stand on the issues he finds problematic in the standard cognitive linguistic notion of metonymy. These include, among others, the status of metonymy as a mapping, as a stand-for relationship, as a type of activation, and as a type of “domain highlighting”; prototype-based vs. unitary definitions of metonymy; and the distinction of metonymy from metaphor, “modulation”, “facets”, and active zones. On the basis of his discussion of those issues, he puts forward a technical unitary (“schematic”) notion of metonymy, which he claims to coexist with three tightly interrelated prototype-based notions of metonymy that he proposes: “purely schematic”, “typical”, and “prototypical” metonymies, all parts of a continuum. Keywords: activation, mapping, problematic issues in metonymy theory, prototype notion of metonymy, schematic notion of metonymy

1.

Introduction

This chapter has been written with the purpose of discussing the problematic properties of what in this chapter will be called the “standard cognitive-linguistic notion of conceptual metonymy”. This notion has been developed over the last * I am grateful to a number of people for their insightful comments on an earlier version of this paper: My two co-editors, especially Réka Benczes, my colleague Carita Paradis, an anonymous external reviewer, and the series editors, Klaus-Uwe Panther and Linda Thornburg. They have helped me to improve the chapter in many ways, but I am solely responsible for any errors or deficiencies. The research reported in this paper was supported in part by a grant awarded by the Spanish Ministry of Science and Innovation for the project FFI2008-04585/FILO, whose head researcher is Antonio Barcelona.



Antonio Barcelona

three decades, starting with Lakoff and Johnson’s seminal insights (Lakoff and Johnson, 1980), which have been refined by such linguists as Lakoff (1987), ­Lakoff and Turner (1989), William Croft (2002/1993), Kövecses and Radden (1998), and Panther and Thornburg (2007), among many others. All of them have made major contributions to the understanding of metonymy. This “standard” notion is, not surprisingly, the most widely accepted account of metonymy in cognitive linguistics, to which I also adhere. It is by no means a completely uniform notion, as there is some disagreement among these authors over a number of issues (some of which will be pointed out in this chapter) but there is full agreement among them as regards the ‘core elements’ of a cognitive linguistic view of metonymy. Those core elements are a number of properties of metonymy proposed by those linguists, which are uncontroversial in cognitive linguistics: (1) the fundamentally conceptual nature of metonymy; (2) the fact that it is experientially grounded; (3) the fact that it can be the root of certain cognitive models; (4) and the fact that it involves experientially and conceptually connected, i.e., “contiguous”, elements. The other properties proposed by those linguists are problematic, in my view. The reasons are that not all of them are universally accepted in cognitive linguistics and that some of them have not even been clearly formulated. This chapter is an attempt at refining the standard notion a bit further. In the rest of this chapter, a long list of the controversial properties will be examined, pointing out the problems that affect them and suggesting some possible answers. Section 2 is concerned with the problematic properties affecting exclusively the standard notion of metonymy. Section 3 is concerned with the problematic properties affecting the usual cognitive-linguistic criteria followed in the distinction between metaphor and metonymy. Section 4 recapitulates the main conclusions reached in the chapter. The ideas presented in this chapter constitute a re-elaboration and an extension of those presented in Barcelona (1999, 2000b, 2002a, and 2003a).

2. Problematic properties affecting the standard cognitive-linguistic notion of metonymy One of the main problematic properties is created by the claim that metonymy operates within the same conceptual domain, whereas metaphor connects two different conceptual domains. As stated above, this property will be discussed in Section 3. There are a number of problematic properties in the standard cognitive view of metonymy that are in principle unrelated to the distinction between metonymy and metaphor. These problematic properties are enumerated below under the guise of a list of questions involving each of them:



Properties and prototype structure of metonymy

– Is metonymy a relationship between “entities” or a relationship between “domains”? – Is metonymy necessarily connected to an act of reference? – What is meant by saying that metonymy is a “stand-for” relationship? – Is metonymy a mapping? – Is metonymy unidirectional or bidirectional? – What exactly is meant by claiming that metonymy is a type of “domain high­ lighting”? – What exactly is meant by claiming that metonymy is a type of activation? – What qualifies as a target in whole for part metonymies? – Are there any degrees in metonymicity? And, related to the previous question: Are some metonymies more prototypical than others? Can we have also a general, unitary definition of metonymy? – Is there any difference between metonymy and “reference point phenomena”? – Is there any difference between metonymy and such phenomena as activezone / profile discrepancy, modulation, and facets? – What factors determine the linguistic conventionalization of a metonymy? In the following, I will address these questions one by one.

2.1 “Entities” or “domains”? This section is concerned with the first question (Is metonymy a relationship between “entities” or a relationship between “domains”?). Metonymy is sometimes presented as a relationship between two “entities” (see Kövecses and Radden, 1998). But what exactly is meant by “entity”? The term seems to allude to an individual of some kind, not necessarily a physical entity (a physical object, an animate being, even a well-defined abstract concept, like “agent”, “action”, “institution”, “emotion”, “category”, “cause”, “effect”, etc.). Abstract entities normally have an intricate conceptual structure. That is, they constitute conceptually complex “domains”. Even the concept of an individual physical entity like a finger constitutes a “domain”, in that it presupposes a certain aspect of knowledge (part of a hand) as the base (domain) against which it is profiled (Langacker, 1987: 147–152; Croft, 2002/1993). Croft’s (2002/1993) most interesting contribution to the study of metonymy is his claim that metonymy highlights a domain within a complex domain matrix. And in Lakoff ’s theory of metonymic cognitive models (Lakoff, 1987: Chapter 5), a subcategory stands for the whole category that includes it. All of this constitutes evidence that the entities linked by metonymy are two domains, rather than two individuals conceptualized in isolation from our detailed knowledge about them (this conceptualization is virtually impossible, on the other hand). If the term



10

Antonio Barcelona

“(conceptual) entity” is used in the definition of metonymy, it should be clear that these entities constitute “domains”.

2.2 Is metonymy necessarily connected to an act of reference? Even though Croft (2002/1993: 179) sets reference as a necessary requirement for metonymies, Lakoff and Johnson (1980: Chapter 8), and later Lakoff and Turner (1989: 100ff) had claimed that referentiality is only the primary purpose of metonymy (without making this purpose a necessary requirement for metonymicity), a position also adopted by Taylor (1995/1989: 124). It is undeniable that metonymy is not necessarily referential. Apart from predicational metonymies, as in (1) (1) She’s just a pretty face. (body part for person; Lakoff and Johnson, 1980: 37)

we should consider Lakoff ’s (1987: Chapter 5) “metonymic models” (i.e., prototype categories whose central category has emerged through metonymy). Some of the most cognitively powerful metonymies he explores are not necessarily used for reference, like the housewife-mother stereotype (ibid.), but principally for making inferences. Additionally, recent research by Klaus-Uwe Panther, Linda Thornburg, Günter Radden, Antonio Barcelona, Francisco Ruiz de Mendoza, Mario and Rita Brdar, and others has revealed the enormous frequency of purely inferential metonymies (see, e.g., Panther and Thornburg, 2003a for a representative collection of papers; see also Barcelona, 2007). In fact, I have suggested (Barcelona, 2005) that reference assignment is actually a less frequent function of metonymy than the guidance of pragmatic inferences or the motivation of constructional (including lexical) meaning and form (though the three functions often interact).

2.3 What is meant by saying that metonymy is a “stand-for” relationship? It is not clear what exactly is meant by Lakoff and Johnson (1980) when they claim that metonymy is a “stand-for” relationship. If they refer to the unambiguous substitution of the target by the source, then there are many undisputed instances of metonymy that do not imply a stand-for relationship in that sense. For instance, in Washington is afraid of possible terrorist attacks, the exact metonymic target of the source Washington (inhabitants, political institutions) depends on the broader context. Compare this example with The kettle is boiling, where the metonymic target of kettle is unambiguous, i.e., the (liquid) content (of the kettle) (Kövecses and Radden, 1998: 40). Probably, what Lakoff and Johnson and others mean by “stand-for” is much the same as what Lakoff and Turner (1989) mean by “(metonymic) mapping” (see below).



Properties and prototype structure of metonymy

2.4 Mapping, “domain highlighting”, activation, and directionality in metonymy This sub-section answers these four questions: Is metonymy a mapping? What exactly is meant by claiming that metonymy is a type of “domain highlighting”? What exactly is meant by claiming that metonymy is a type of activation? Is metonymy unidirectional or bidirectional? The properties of metonymy addressed in all these questions are closely connected. This is why they are treated together in the same sub-section. If metonymy is a mapping at all, what type of mapping is it? Leaving aside the thorny issue of the sameness or distinctness of the domains involved (see below), are metonymic mappings of the same kind as metaphorical mappings? Croft (2002/1993) claims that metonymy is an instance of what he calls domain highlighting, whereas metaphor is an instance of what he calls domain mapping. Domain highlighting consists in highlighting a secondary domain within the domain matrix constituted by a speaker’s encyclopedic knowledge of the meaning of a linguistic expression. (The technical notions “secondary domain” and “primary domain” are borrowed by Croft from Langacker, 1987: 165, 222.) In Croft’s example (2002/1993: 178), here presented as (2), (2) Proust is tough to read.

the subject NP highlights Proust’s work. In our knowledge of Marcel Proust, his literary work is an important domain, but it is not primary in comparison to the person domain, namely our knowledge of the fact that Proust was a person. In this metonymy, then, the secondary subdomain of Proust’s literary work is highlighted within the “domain matrix” for Proust. This characterization of metaphor as a type of mapping and of metonymy as a type of highlighting might lead to Croft’s position being misconstrued as implying that metonymy is not a mapping. Actually, Croft states (2002/1993: 178) that “in metonymy […] the mapping occurs only within a domain matrix”. In my view, Croft’s notion of domain highlighting amounts to the mental activation of a certain (sub)domain, the target, by another (sub)domain, the source, both located within the same domain matrix. But one does not necessarily have to view metonymy in terms of Langacker’s notion of “domain matrix” (this is Croft’s strategy) to take advantage of the useful notions of “highlighting” or “activation” (I prefer the latter term). For instance, if one views metonymy as operating within an Idealized Cognitive Model, an icm (as Kövecses and ­Radden, 1998, do), it is still possible to claim that in every instance of metonymy a (sub)domain is mentally activated via another (sub)domain in the same icm

11

12

Antonio Barcelona

or domain. In my view, this intra-domain activation of one (sub)domain by another is one of the essential properties of metonymy. A mapping, “in its most general mathematical sense, is a correspondence between two sets that assigns to each element in the first a counterpart in the second” (Fauconnier, 1997: 1). The term “mapping” can also be understood, in a ­narrower sense, as the projection of one structure onto another, so that the projected structure imposes (some of) its internal elements, traits, and properties onto their counterparts in the other. The mapping, in this sense, is facilitated if both structures have a basic degree of structural similarity. We find this situation in metaphor, whose source and target share at least part of their abstract imageschematic structure, so that sources are mapped onto sources, goals onto goals, etc. (Lakoff, 1993). In life is a journey, the beginning of the journey maps onto that of life, the obstacles in the journey onto life’s difficulties, etc. But a conceptual structure can be mapped onto another to which it bears no structural similarity, as in metonymy, where the projection proceeds from a whole onto a part, a part onto a whole, or a part onto a part of a domain. Metonymically linked (sub)domains do not normally exhibit any degree of structural similarity or equivalence. Wholes do not exhibit a similar abstract structure to that of their parts, and when the mapping relates two parts of the same icm (e.g., the producer and the product in the “Production icm”, as in I bought a Stradivarius), those parts are seldom, if ever, similar, and never equivalent functionally or relationally (the role of the producer in the icm is not equivalent to that of the product). Given this asymmetry, is it accurate to regard metonymy as a mapping? Some authors (e.g., Feyaerts, 2000: 63–64; Kövecses and Radden, 1998), actually avoid calling metonymy a mapping at all; they just treat it as a “relationship” based on conceptual contiguity; others, like Ruiz de Mendoza (this volume) claim that what are called metonymic mappings are actually a manifestation of two more basic processes, i.e., domain expansion and reduction; see also Brdar and Brdar-Szabó (this volume). My claim, however, is that metonymy constitutes an asymmetrical mapping, whereas metaphor constitutes a symmetrical mapping.

. The notion of “domain matrix”, as put forward by Langacker, should be distinguished from that of icm and also from that of “frame”, although it overlaps with them. Frames and icms are models of areas of experiential knowledge (I call them “functional domains” below) and they are not exclusively attached to lexical or constructional meaning (they can operate in reasoning and inferencing). Domain matrices, as formulated by Langacker (1987: 147) are simply the array of domains relative to which a “predicate” (i.e., the semantic pole of a morpheme; see Langacker, 1991: 1–13) is characterized. . A preliminary discussion of this issue is in Barcelona (2003a) and in Barcelona (n.d.). See also Warren (2002). The postulation of “one-correspondence” metaphors (metaphors ­consisting



Properties and prototype structure of metonymy

The metonymic source projects its conceptual structure onto that of the target, not by means of a systematic matching of counterparts (as in metaphor), but by imposing a conceptual (and linguistic) perspective from which the target is activated; this activation entails a shift in (conscious or unconscious) attention from source to target. In Example (2), the subdomain of Proust’s literary work is activated from the domain of its author (in which personhood is a primary subdomain). A metonymic mapping affects the conceptualization of the target, which is now understood from the perspective imposed by the source. In metonymy, thus, the target is accessed mentally (activated) via the source, which is its “reference point” (Langacker, 1993). The choice of the source (or “vehicle” in the older terminology followed by Kövecses and Radden, 1998) is important, as the perspective imposed by it constrains the way in which the target is viewed. In the Proust example, the literary works are activated mainly as produced by a given author: The metonymy invites viewing Proust’s literary work primarily as an extension of its author’s personality. Metonymy-driven re-conceptualization can be relatively transient and limited, as in this example, or it can leave an extensive permanent trace in cognitive domains and in inference patterns, as in what Lakoff (1987) calls metonymic models. In this example, thus, the source author is mapped onto the target (author’s)­ work, not because of the functional or relational similarity between source and target (as in metaphor), but because the source imposes the conceptual perspective from which the target is being considered. This is the sense (perspective ­imposition) in which metonymy can be regarded not only as a phenomenon of highlighting/activation, but also as a type of mapping. The perspectivization occurring in metonymy is a dimension of construal (in the Cognitive Grammar sense of this term; see Langacker, 1999: 5–7, 207–212), and construal is a manifestation of “viewing” (in the neutral sense, including conception and perception, given to this term in Cognitive Grammar; see Langacker, 1999: 206). One of the main types of perspective is what Langacker calls the selection of a “reference point” (Langacker, 1999: 207–208). On the other hand, to

of just one submapping, hence not of a systematic set of submappings) by Ruiz de Mendoza (2000) would run against this difference. However, these “one-correspondence” metaphors can be easily extended, and in fact they are often extended, to become “multiple-correspondence metaphors”, whereas it is not easy or even possible to extend a metonymy so that it will become a “multiple-correspondence metonymy” (and when that happens, it has become a metaphor, as in metonymy-based metaphors). . On the difference between metonymic source–target connection and the one between reference point-target connection, see Section 2.6.

13

14

Antonio Barcelona

Langacker, metonymy is a common type of reference point selection (more on metonymy as a reference point in Section 2.6 below). In sum, it has been argued so far that metonymy is an asymmetric mapping whereby the target is construed from the perspective of a reference point, namely the source. But there is a further qualification that has to be made to the claim that metonymy is a mapping. We have to inquire on what makes it possible for a given source (Proust as author in (2)) to serve as a reference point (a type of perspective) from which to activate and construe (conceptualize) the target. The answer is a further fundamental property of metonymy, namely the strong experiential (hence pragmatic) link between the roles performed by the source and the target in the “domain” where they operate. (I will be claiming below, in 3.1.2, that metonymy operates within a “functional domain”, i.e., a frame or icm.) Fauconnier (1997: 11) regards metonymy as a “pragmatic function mapping”. A “pragmatic function” (Fauconnier, 1994, 1997) is a strong built-in connection between roles in a domain (cause–effect, author–work, agent–action, etc.). This type of connection is necessary for a metonymic source to activate its target. Kövecses and Radden (1998) point out that even if the nose and the mouth are elements in the same domain (which might be called the human face frame/icm), they do not seem to perform roles connected by what is here called a pragmatic function, so that neither can act as a metonymic source for the other. However, the pragmatic function ­salient body part – person allows the face to act as a metonymic source for the whole person, as in (1), repeated here as (3): (3) She’s just a pretty face. 

(Lakoff and Johnson, 1980: 37)

This shows that being in the same functional domain (see 3.1.2) is a necessary, but not a sufficient condition for metonymicity. A pragmatic function linking source and target is additionally necessary. Kövecses and Radden’s (1998) claim that metonymy arises on the basis of a specific “relationship type” (cause–effect, etc.) between two entities within an icm seems to be a different way of formulating the need for the privileged connection between roles that has been called “pragmatic function” here. . An important issue that is mentioned by some of the chapters in this volume (e.g., Brdar and Brdar-Szabó) is the connection between metonymy and compression (Fauconnier and Turner, 2002). Compression often underlies (or is a result of?) many referential metonymies, but not of all of them. Fauconnier (n.d.) shows how metonymy is connected with compression in examples like The blue car decided to park beside me, or I squeezed in between the two trucks, as shown by the possibility of reflexives anaphorically connected to either source or target (The blue car decided to park itself beside me / I squeezed myself in between the two trucks), which show source and target are compressed to uniqueness. This is not possible with other referen-



Properties and prototype structure of metonymy

In any discussion of mapping, the issue of the directionality of the mapping has to be examined. Metaphor is claimed by Lakoff and Turner (1989: 132–133) to be unidirectional, but they say nothing about metonymy in this respect. In my view, metonymy is also unidirectional (for a contrary view, see Brdar and BrdarSzabó, this volume). Its difference from metaphor is that metonymies tend to be reversible, as Kövecses and Radden (1998: 46ff) have noticed. This does not happen so often in metaphor. Take these two examples from these linguists (Kövecses and Radden, 1998: 50): (a) America (for “United States”, i.e., whole thing for a part of the thing); (b) England (for “Great Britain”, i.e., part of a thing for the whole thing). However, reversibility is not the same thing as bidirectionality, which implies that source and target simultaneously map onto each other. Firstly, the mapping in (a) is very different from the mapping in (b), therefore they constitute different metonymies, not two different manifestations of the same metonymy. Secondly, and more importantly, in neither metonymy does the source and the target simultaneously map onto each other: In (a) the domain America activates the domain United States (as in America will prevail, said by a U.S. President talking about the future victory of his country over its enemies) but the domain United States does not activate the domain America at the same time and in the same sentence; and in (b) the domain England activates the domain Great Britain (as in England finally joined the European Economic Community, where England stands for Great Britain or the United Kingdom), but the domain Great Britain does not activate the domain England at the same time and in the same sentence.

2.5 whole for part metonymies, degrees in metonymicity and prototype-based vs. unitary definitions of metonymy This long section is an attempt at answering the four questions formulated about these issues at the beginning of Section 2. As in the preceding section, they are all closely connected, and it is convenient to treat them in the same section, though tial metonymies: Next to The gastric ulcer wants some coffee [illness→patient] we can have The gastric ulcer cured himself but not *The gastric ulcer cured itself [intended meaning: “The patient with the ulcer cured himself ”]. The connection of non-referential metonymies (including predicational, illocutionary and purely inferential ones) with compression still has to be explored in detail. . Incidentally, the terms “symmetric” and “asymmetric” were not used above in the sense these terms have in formal semantics (as in the technical expression “symmetric predicates” applied to predicates like resemble); therefore I do not claim that metonymies are unidirectional and metaphors bi-directional. Both are unidirectional mappings.

15

16

Antonio Barcelona

in different sub-sections, since the answer to the first question leads to the answer to the second, third, and fourth questions. I have attempted to answer the first question in Section 2.5.1. The reasons for a positive answer to the second, third, and fourth questions are provided in Section 2.5.2. In Section 2.5.3, I briefly examine Peirsman and Geeraerts’ (2006a, 2006b) recent alternative, though complementary, proposal for a prototype account of metonymy.

2.5.1 The target in whole for part metonymies whole for part metonymies often create a definitional problem. The problem is expressed by the first question, which can also be formulated like this: Which requirements must be satisfied by a part to qualify as a metonymic target of a whole? The above Proust example (Example (2)) would be treated by most authors as metonymic. Disagreement occurs in instances like (4): (4) This book is a history of Iraq.

Some linguists would regard the NP This book as metonymic in this case, its target being semantic content, while others would deny it any metonymic status. The answers to this issue have invoked the notion of “secondary domain”. We will first briefly discuss this notion before examining the answers to the first question. As I pointed out above, Croft (2002/1993) characterizes metonymy in terms of domain highlighting, which consists in mentally activating a “secondary” (sub)domain, the target, through another (sub)domain, the source, both within the same “domain matrix”. Ruiz de Mendoza and Barcelona also make use of this notion. The notion is based on Langacker’s (1987: 165) notion of “primary domain” – a “secondary (sub)domain” being a (sub)domain, which is not “primary”. Thus a crucial point is to know which are the factors determining the ranking of subdomains as primary or secondary in a domain matrix. 

. A “primary domain” (Langacker, 1987: 165) is a “domain that is highly ranked” in terms of its “prominence and likelihood of activation”. Langacker cites the difference in meaning between roe and caviar: In the former, the domain of “the reproductive cycle of fish” (p. 164) is obligatorily accessed, hence primary, whereas the “domains pertaining to the preparation and (conspicuous) consumption of foods are peripheral and activated only on a contingent basis. This ranking is reversed in the case of caviar: the domains that construe the designated mass as an item of consumption are salient and obligatorily activated, but the relation of this mass to fish reproduction is secondary (and often suppressed)” (Langacker, 1987: 165). Even though Langacker does not use the term “secondary domain” explicitly, the term and the corresponding notion are implicit in his words: A secondary domain is one which is not obligatorily activated when understanding the meaning of a symbolic unit (i.e., a morpheme, a lexeme or any other grammatical construction).



Properties and prototype structure of metonymy

According to Croft (2002/1993: 179–182), there is only one determining factor: extrinsicness. Secondary domains, he claims, are more extrinsic than primary domains within the relevant domain matrix. Extrinsic domains make reference to entities that are external to the domain matrix for a concept (cf. Langacker, 1987: 160–161); in contrast, intrinsic domains make no reference to such external entities. Croft claims that in Proust is tough to read, Proust’s literary work is relatively extrinsic in the domain matrix for Proust, whereas the person domain is quite intrinsic in that domain matrix. Accordingly, he claims that the literary work domain qualifies as a secondary domain that can be highlighted and thus become the metonymic target of Proust. But domain extrinsicness is not the only factor making a (sub)domain secondary. As Ruiz de Mendoza (2000) reminds us, the really important factor for Langacker (1987: 158–165, especially 164–165), is the fact that primary domains tend to be relatively central (and not simply intrinsic) in the conceptualization of a domain matrix, whereas secondary domains tend to be relatively non-central (or less central); intrinsicness is just one of the factors in centrality. The other factors making a particular (sub)domain more or less central are (i) conventionality, i.e., (not) being conventionally regarded as part of the source (e.g., literary work is conventionally a part of the overall Proust domain matrix, but less so than the person domain); (ii) characteristicness, i.e., (not) being exclusive to the source (e.g., literary work is not exclusively part of Proust); (iii) genericness, i.e., the fact that it is (not) generic knowledge that the target belongs to the source (e.g., it is not generic knowledge that literary work is a part of Proust, whereas it is generic knowledge that person is a part of Proust). . “Shape, for example, is a highly intrinsic property of physical objects, as it reduces to relations between the parts of an object and does not require interaction or comparison with other entities. Size, on the other hand, implies comparison either with other objects or with some scale of measurement; hence it is not quite so intrinsic as shape.” (Langacker, 1987: 160–161) . Here we depend on our general typology of metonymy. For those that only recognize whole for part or part for whole metonymies (i.e., Ruiz de Mendoza, 2000), literary work is a (relatively non-central) part of the Proust matrix. For those that also accept part for part metonymies, literary work may be regarded as a part of the domain matrix constituted by Proust, or it may be regarded as just a part of a broader domain matrix or ICM (the literary production ICM), author being another part in the same matrix with which it has a strong pragmatic connection. . Of course, characterizing Proust’s literary activity as a secondary domain in the overall Proust domain does not mean that Proust’s literary output is not a salient part of the overall domain. It simply means that literary work is not intrinsically, conventionally, characteristically, and generically regarded as central a domain in Proust as the fact that Proust was a person, which is a primary domain in our conceptualization of any human being.

17

18

Antonio Barcelona

Langacker (1987: 165) mentions another major factor, which often joins centrality in determining the ranking of domains as primary or secondary: Convention (to be distinguished from the “conventionality” determining domain centrality). As his example of roe vs. caviar shows, “the ranking of constituent domains is not fully predictable from the centrality factors […], but is to some degree subject to conventional determination”. In other words the primacy of fish reproduction in the basic meaning of roe and of food preparation and consumption in the basic meaning of caviar depends on convention, since both terms “designate a mass of fish eggs, and the same mass of eggs can qualify as either roe or caviar (at least at a certain stage of preparation)” (Langacker, 1987: 164). Returning to Example (4): Can the NP This book be regarded as metonymic, so that the book as a whole maps onto its semantic content domain? Croft (2002/1993) and Ruiz de Mendoza (2000) disagree in their answer to this question. To Croft, (4) is not metonymic, because semantic content is a highly intrinsic domain in the book domain matrix, and therefore constitutes a primary domain in book. To Ruiz de Mendoza, (4) is fully metonymic, because (as was stated above) intrinsicness is to Langacker (in his characterization of secondary domains) just one of the factors involved in the centrality of a domain, which is in turn one of the two main factors determining the primary or secondary status of a domain in a domain matrix; centrality, in turn, depends on the three additional factors mentioned above (conventionality, characteristicness, and genericness). Semantic content may thus be shown to be a relatively secondary domain of book, which is a notion profiling primarily physical objects. I agree with Ruiz de Mendoza’s notion of secondariness. According to him, the target must be a secondary, non-central domain in the source in what he calls “target-in-source” metonymies, i.e., whole for part metonymies (see also Barcelona, 2003a). This requirement ultimately means that the target part must be neatly distinguishable from the source whole. Therefore, Croft and Ruiz de Mendoza would claim that an example like (5) (5) This book is very large.

would not be metonymic, since both linguists regard the physical object domain (which in turn implies size, shape, weight, color, etc.) as a primary, rather than secondary, domain in the book domain. I would also regard physical object as a primary domain in book. But I disagree with Croft and Ruiz de Mendoza with respect to the metonymic status of this example. I would still regard it as at least peripherally metonymic. The reason is that I do not consider secondariness a requirement for metonymic status, but only for one of the degrees of metonymicity that I will be postulating below.



Properties and prototype structure of metonymy

2.5.2 Degrees in metonymicity. An initial unitary, schematic definition of metonymy. The prototype structure of metonymy Some linguists like Jakobson (1971/1956), Dirven (2002), or Radden (2002) have advocated the existence of a continuum between metaphor and metonymy. It may also be reasonable to suggest the existence of a continuum from the most prototypical instances of metonymy to other more marginal instances of it, like (5), and to non-metonymic use. Croft (2002/1993: 180–181) also suggests this continuum. I proposed in Barcelona (2002a, 2003b) to distinguish at least three basic degrees of metonymicity on this continuum: purely schematic, typical, and prototypical metonymies. However, we may wonder whether a unitary definition of metonymy is possible as well, a definition which would be applicable not only to the most prototypical but also to the more marginal instances of metonymy. As I also claimed in Barcelona (2002a, 2003b), such a definition is possible. Metonymy can be provisionally defined in general, “schematic” terms as an asymmetric mapping of a conceptual domain, the source, onto another domain, the target. Source and target are in the same overall domain and are linked by a pragmatic function, so that the target is mentally activated.

Since all metonymies constitute an asymmetric mapping, this property has been included in the unitary, schematic definition of metonymy (“schematic metonymy” for short). The above definition brings together, in my view, all the necessary and sufficient conditions for metonymicity. Therefore, to the extent that the use of This book in (5) satisfies them, it should be considered as a metonymy. It may seem contradictory to claim that a category (in this case, a technical category like “conceptual metonymy”) can be defined in terms of necessary and sufficient conditions and that it is a prototypical category. But in some cases both classifications can co-exist and conceptualizers use them for different purposes, as Rosch made it clear long ago (Rosch, 1983). The reason is that even the few categories which exhibit necessary and sufficient conditions (or are supposed to exhibit them, typically “technical” or “scientific” categories) are at the same time heterogeneous and have good and bad examples of the category, i.e. they also exhibit prototype effects. Langacker (e.g., 1987: 371ff) also claims that, at least as regards the lexicon and the grammar, a “schematic” conception of a category can co-exist with a prototype conception of the same category. In fact he claims that the “two modes of categorization are nonetheless inherently related and describable as aspects of a unified phenomenon” (Langacker,­ 1987: 371). An important additional property of schematic metonymy is that, when it is expressed linguistically, pictorially or otherwise, i.e., when it does not simply

19

20 Antonio Barcelona

remain at the conceptual or processing level,10 the target remains implicit. But since this property only pertains to metonymic expressions and not to conceptual metonymies as such, it has been left out of the definition and will not be discussed at length in this chapter. However, this is one of the properties that will be used below to distinguish metonymic expressions from other reference-point constructions. In the conclusions, a final unitary, schematic definition of metonymy will be provided, which will specify the “overall domain” as a “functional domain” (see 3.1.2). Any conceptual process that satisfies the properties included in the schematic definition is a metonymy. But the conceptual processes that are normally recognized by linguists and rhetoricians as “undoubted metonymies” have some additional properties, some of which, like “secondariness”, are scalar. These undoubted metonymies have been called “typical” metonymies in Barcelona (2003a). A typical metonymy is a schematic metonymy whose target is clearly distinct from the source, either because it is a relatively secondary subdomain of the source, as in certain whole for part metonymies such as (4) or, even more markedly, (6) (6) The use of the pill has reduced the birth rate in many countries. (category (pill) for member (birth control pill))

or because it is not included in it, as in part for whole metonymies like (1) and (3), repeated once more as (7), or in part for part metonymies like (8):11 (7) She’s just a pretty face. (salient body part for person; Lakoff and Johnson, 1980: 37) (8) The coke felt as stimulating a drink as a cup of tea. (part [container] for part [content]; according to Kövecses and Radden, 1998: 57, container and contents are two parts of the “Containment” icm)

10. In discourse (see Barcelona, 2003b, 2005, 2007, in preparation), inference-guiding conceptual metonymies are sometimes not expressed linguistically, like some of the metonymies guiding inferences in humor comprehension discussed in Barcelona (2003b); see also below on “purely inferential metonymies”. Or like the metonymies motivating the stereotypes underlying the use of certain names as paragons: Shakespeare stands as a paragon for the class of immensely talented writers (licensing such expressions as Mike is a real Shakespeare) on the basis of a linguistically unexpressed metonymic model of Shakespeare himself as almost exclusively a talented writer (thus ignoring other facets of the person William Shakespeare); see Barcelona (2004). 11. The general classification of metonymies into whole for part, part for whole, and part for part follows established tradition.



Properties and prototype structure of metonymy

These two degrees of metonymicity (i.e., “purely schematic”, marginal metonymies, like that in (5) and typical metonymies as in (4), (6), (7) and (8)) help us account for the difference between (4), in which the semantic content subdomain of book is activated, and (5), in which the physical domain subdomain of book is activated. Sentence (4) manifests a typical metonymy because semantic content is a clearly distinguishable subdomain (a relatively secondary subdomain) within the book domain, as compared with physical object; as a result (4) is more easily noticed as a mapping, and as a semantic shift, than (5). In (5), the physical object domain is less easily distinguishable from the book domain as a whole (it is felt to be an “essential” domain in it, at least in the traditional, prototypical book, which is first of all a physical object), and the semantic shift is less noticeable; (5) would be a purely schematic rather than a typical metonymy. Typical metonymies in turn admit of many different degrees of typicality, as the distinctness of target vis-à-vis source in whole for part metonymies is often a matter of degree (see Barcelona, 2003a). The third and highest basic degree of metonymicity is constituted by prototypical metonymies. A prototypical metonymy is a highly typical referential metonymy (i.e., a referential metonymy with a high degree of typicality), whose target and referent is an individual entity, or a collection of individual entities. By “high degree of typicality” I mean that the target of these metonymies scores relatively high on all of the factors of centrality and that its secondary status is highly conventional (as discussed in 2.5.1); this notion is implicit in Barcelona 2002a and 2003a). On the other hand, the fact that these metonymies are referential entails that they are expressed linguistically or in other communicative modes (perhaps pictorially or gesturally). Examples (9) and (10) include prototypical metonymies: (9) Paris agreed to a truce. (part [location] for part [institution located in it]; according to Kövecses and Radden, 1998: 58, place for inhabitants, which is similar to location for institution, is a part for part metonymy) (10) We have seen a couple of new faces around lately. (part [salient body part] for whole [person])

What justifies regarding this type of metonymies as prototypical? First, reference is the primary pragmatic purpose of metonymies for individuals (see Lakoff and Johnson, 1980: Chapter 8, where they state that metonymies for individuals are the classical instances of referential metonymy); in fact, it is on the basis of referential metonymies for individuals that the very notion of metonymy arose in rhetoric and linguistics. Second, the referential function of these metonymies is readily perceived by language users. The usual question to be expected from a hearer,

21

22

Antonio Barcelona

if a referential metonymy has not been used quite felicitously by the speaker, is “What/Who do you refer to?” (or “What/Who do you mean?”). In sum, these metonymies seem to function as basic-level cognitive reference points, both for the analyst and for the language user, on the continuum of metonymicity. Note that the metonymy in (4) above, though referring to an individual entity (in this case, to the semantic content of a book) is not prototypical because its degree of typicality is low: Its target is not clearly distinct from the source, since it scores low – though higher than (5) – in secondariness, i.e., on the various factors of non-centrality and in the conventionality of its secondary status. Note also that (6), (7), and (8) above, though with a relatively high degree of typicality, are not at the same time prototypical, because the NPs over which they operate are not referential: These NPs just mention (via metonymy) a category of pills in (6), a category of people in (7), and a category of drinks in (8), but do not refer to them or to particular instances of them. Therefore, the metonymies in these three examples are simply typical. By contrast, the metonymy in (9) is prototypical, since besides being highly typical, it is used to refer to an individual entity (the French government or the French president). The metonymy in (10) is prototypical for the same reason: Its target and referent is constituted by two individual members of the people category. Now compare the metonymies in (9) and (10) with the metonymy in (11) below: (11) A good student should read extensively. (virtual random representative member of a category for the category)

The generic use of a singular count indefinite NP consists of the random selection of a virtual representative member of a category to stand for the whole category (see Radden and Dirven, 2007: Chapter 5; Langacker, 1999: Chapter 8). This member is said to be “virtual” because it has no existence or status outside the mental space called by Langacker (1999: 251) the “structural plane” (the mental space reserved for the speaker’s representation of the general “structure of the world”, where generic statements belong). This type of generic NP is conceptually motivated by a blend (Radden and Dirven, ibid.) of singularity and genericity, which is based on the metonymy member for category (see also Radden, 2005: 21; Radden, 2009).12 The metonymy underlying the genericity of (11) is not

12. In her interesting, insightful paper in this volume, Koskela argues that “vertical polysemy”, consisting in the broadening or narrowing of a lexical item (such as the broadening of ­Aspirin to the sense “any pain killer” or the narrowing of verb drink from “consume liquid” to “consume alcoholic beverages”) is due to vertical categorial shift, which she claims to be due mainly to abstraction and to be essentially different from metonymy, characterized by highlighting (in Croft’s sense). Metonymy, according to her, reduces to connections between the concepts of



Properties and prototype structure of metonymy

prototypical, because, though referential, the NP A good student is not used to refer to an individual but to a class or category. For the same reason, metonymies occurring in non-referential predicative NPs, like the one in (7), and other predicational metonymies (Panther and Thornburg­, 1999: 335–336, 2003b: 5–7, 2007: 246) like that in (12) below, illocutionary metonymies (Panther and Thornburg, 1997, 2003b: 5–7, 2007: 246) like that in (13) below, and “purely inferential” metonymies (i.e., those guiding implicatures; see Barcelona 2002b, 2003b, 2005, 2008, in preparation) like that in (14) below are not prototypical metonymies: None of them is used to refer to individuals. In addition, purely inferential metonymies are often not expressed linguistically, pictorially or otherwise; they simply guide inferential processes of meaning construction that constantly occur in communication, but they are not used to perform an act of reference. The examples of predicational (12), illocutionary (13), and purely inferential metonymies (14) are presented and briefly commented on below. (12) The saxophone had to leave early.

This example has been borrowed from Panther and Thornburg (2007: 246). According to these linguists, this sentence induces in many contexts the interpretation “The saxophone player left early”. They claim that the interpretation, as an actually occurring past event, of the past obligation to carry out an action (which is literally expressed by the syntactic predicate) is motivated by the high-level metonymy potentiality for actuality (which, according to these linguists, also motivates a large number of other phenomena). In this example, the metonymy ranges over the syntactic predicate, as does the metonymy in (7), which motivates the interpretation of the nominal predicate as cueing a category, not as referring objects and other contiguous objects and to part–whole connections and does not motivate vertical categorial shifts. But, on the one hand, she herself admits that even category-member connections can be made the object of “metonymy-type” highlighting (i.e., activation) processes. On the other hand, although (as Koskela correctly points out), it is true that the taxonomic member-category relation is a different phenomenon from metonymy, this does not mean that category for member or member for category metonymies do not exist. The member-category relation is a static conceptual–semantic relationship (based on encyclopedic knowledge) on which a dynamic process like metonymy operates by activating, i.e., bringing to attention, one of the two poles of the static relation. Speakers know that alcoholic beverages are a type of liquid; the metonymy that motivates the semantic narrowing of drink does not create this knowledge, but it exploits it with the result of extending the polysemy of this lexeme. The same applies to static meronymic (i.e., part–whole) semantic connections (which again reflect encyclopedic knowledge): These, too are dynamically exploited by part–whole metonymies (which Koskela regards as undisputed metonymies), so that in some cases the part is used to activate the whole and in other cases we find the opposite metonymy.

23

24

Antonio Barcelona

to an extralinguistic entity; both are predicational. The predicational metonymy in (12) is typical, since it is a part for part metonymy. (We also have another metonymy in this example: The subject NP the saxophone is used to refer to the musician rather than to the instrument thanks to the metonymy instrument for agent, which in this case constitutes a prototypical referential metonymy.) (13) Can you open the window? (said by someone who wants the window to be opened to someone else who is sitting or standing near the window)

The utterance of a sentence like (13) in the appropriate context would be interpreted as an indirect request. According to Gibbs (1994: 351–357), conceptual metonymy is involved in the selection of a salient part of the scenario (i.e., a frame or icm) for directive speech acts (which includes requests). This salient part, in this case a precondition for the actual performance of the requested action (the precondition being the hearer’s ability to perform the action), activates the whole scenario of requesting, with the result that this question is understood pragmatically as a request. The part for whole metonymy guiding the inference to the speech act of requesting in this example is an illocutionary metonymy and is at the same time a typical metonymy. It is not a prototypical metonymy since it is not referential. Thornburg and Panther (1997), Panther and Thornburg (1998), and Ruiz de Mendoza and Otal Campo (2002), among others, have greatly developed the study of illocutionary metonymies. (14) Opposition M.P. (referring to the Prime Minister): “But what can we expect, after all, of a man who wears silk underpants?” Prime Minister (rising calmly): “Oh, I would have never thought the Right Honorable’s wife could be so indiscreet!”

This brief exchange is reported to have taken place at a Parliament in the 1930s. It was analyzed in great detail by Barcelona (2003b). Both the M.P.’s utterance and the Prime Minister’s clever repartee (especially the latter) invite a large number of implicatures. Among the implicatures invited by the Prime Minister’s reply utterance, we may single out the following, all of them guided in part by metonymy: a. “The M.P.’s wife shares a secret with the Prime Minister (the secret being the kind of underwear used by the Prime Minister, a type of information that should not be disclosed to strangers).” Guided by result (being discreet/ indiscreet) for condition (knowing the secret). b. “She has seen the Prime Minister undress.” Guided by fact (knowing the underwear used by someone; this knowledge being the content of the secret) for one of its conventional explanations (having seeing that person undress).



Properties and prototype structure of metonymy

c. “She has had a sexual affair with the Prime Minister.” (Punchline and main inference intended by the Prime Minister). Guided by fact (having seeing someone undress) for one of its conventional explanations (having had a sexual encounter with that person).

None of these metonymies have the purpose of performing an act of reference: They simply connect one proposition to another – the implicature. And only the metonymy guiding implicature a takes as its source domain a proposition linguistically expressed in the conversation (“the M.P.’s wife is indiscreet”), its target being implicature a. The propositional source of the metonymies guiding implicatures b and c is not explicitly manifested in the conversation; this source is rather the target (or part of the target) of the metonymies guiding implicatures a and b in the inferential chain a → b → c.13 The metonymies guiding these implicatures are “purely inferential” metonymies and are, in this case, typical but not prototypical metonymies.14 Like the phrasal metonymies (i.e., metonymies operating over a phrase, normally an NP) in (6), (7), (8), and (11), the predicational (12), illocutionary (13), and purely inferential (14) metonymies examined above are simply typical, as they are not prototypical as well. To these, we could also add those non-referential metonymies that motivate in part the meaning or the form (especially the elliptical forms) of certain syntactic constructions (see, e.g., Barcelona, 2009; Langacker, 2009; Radden and Kövecses, 1999: 28, 33–34, 43). Prototypical metonymies display the highest degree of metonymic prototypicality, with typical and purely schematic metonymies displaying decreasing degrees of it. Typical metonymies that are not prototypical are simply typical (with varying degrees of typicality; compare (4) with (11)). As has been repeatedly implied above, every metonymy is a schematic metonymy, but not every

13. Actually, the inferential chain meant and obviously conveyed by the Prime Minister’s utterance is slightly more complex. For the sake of brevity I present here a simplified version which nonetheless retains the essential elements of the analysis. There is an additional inferential chain which was probably also meant and conveyed by the Prime Minister, and which has not been considered here. For details, see Barcelona (2003b). 14. Every conceptual metonymy is fundamentally inferential in nature or was so at some stage of its application in the history of a language. However, referential, predicational, and illocutionary metonymies serve additional purposes (establishing reference in the case of referential metonymies, or motivating relatively conventional meanings of certain grammatical structures in the case of predicational and illocutionary metonymies). By contrast, “purely inferential” metonymies have the sole purpose of guiding implicatures which are normally not conventionally connected to a certain grammatical structure.

25

26 Antonio Barcelona

schematic metonymy is a typical metonymy, as we saw was the case with (5). For these cases we reserve the label purely schematic metonymies. Despite their peripheral status in the category, some of these metonymies, including the one in (5), become perfectly conventional, and can even motivate established polysemy (see Section 2.8). The treatment of examples like (5) as metonymic has the advantage of underlining the fundamental similarity connecting this example to normally undisputed examples of conceptual metonymies like (6)–(14), and to controversial instances like (4): All of them exhibit the properties of schematic metonymies, namely intra-domain asymmetric mapping and activation of target by source due to their pragmatic function link. Obviously, granting metonymic status to (5) entails that most uses of a linguistic expression (especially of a word) in context will be metonymic in the purely schematic sense. But this is hardly surprising, given that both non-metonymic and metonymic active zone phenomena are the rule rather than the exception (Langacker, 1999: 33–35, 67). We discuss the connection between Langacker’s active-zone / profile discrepancy and similar notions proposed in the literature such as “modulation” (and “facets”) on the one hand, and metonymy on the other hand, in Section 2.7 below. The additional properties exhibited by typical metonymies such as (4) and (6)–(14), namely the greater distinctness of the target with respect to the source, and by prototypical metonymies like (9), (10), and the one operating over the subject NP in (12) (referentiality and individual targets), make them more likely to be noticed as semantic shifts or mappings than purely schematic metonymies. The continuum of metonymicity is a scale of metonymic prototypicality because what are here called prototypical metonymies have traditionally acted as the model of the technical category “metonymy” in rhetoric and linguistics. But, since typical and purely schematic metonymies are equally metonymic in schematic terms, the continuum might also be regarded as a scale of “noteworthiness” in metonymy, measuring the relative ease with which these phenomena are perceived as conceptual and semantic shifts. The continuum of metonymicity is represented graphically in Figure 1.

2.5.3 On a different yet complementary prototype account of metonymy (Peirsman and Geeraerts, 2006a, 2006b) These authors seem to have uncovered a number of striking subtle parallelisms between what they regard as prototypical, core instances of metonymy and other more peripheral types that are linked to the core in terms of similarity. This similarity is both metaphorical similarity “in the form of a shift from the spatial and material domain to temporally characterized entities and to functional



Properties and prototype structure of metonymy

Prototypical

Simply typical

Purely schematic

Typical

Schematic metonymies

Figure 1.  The continuum of metonymicity (adapted from Barcelona, 2003a)

and abstract wholes”, and “similarity in the form of a gradual weakening of the contiguous part–whole relations to looser forms of contact and contiguity”; with respect to the latter type of similarity, they point out important parallelisms between the gradual weakening of the spatial dimensions “strength of contact” and “boundedness” observable in metonymies operating in the spatial and material domains, and the gradual weakening of the counterparts of these dimensions in the temporal domain and in what they call “taxonomies”, “functional assemblies” and “collections”. These two types of similarity are observable in the following examples, adapted from Peirsman and Geeraerts (P&G henceforth), most of which they borrow from other authors on the basis of their extensive survey of the literature on metonymy: (15) Prototypical: Spatial part for whole (highest degree of “strength of contact” between part and whole and of boundedness of part and whole): We need some good heads on the project. (16) Non-prototypical: Extended (non-spatial, non-material), part for whole: subevent for complex event (highest degree of metaphorical “strength of contact” between part and whole and of metaphorical boundedness of part and whole): “putting a king on the throne” where a sub-event symbolizes the more complex event of establishing and recognizing someone as the king. (17) Less prototypical: material for object (highest degree of “strength of contact”, but lower degree of boundedness: Only the target object is bounded): carton ‘cardboard’ for ‘cardboard box’. The material is an unbounded substance, normally coded as a noncount noun.

27

28

Antonio Barcelona

(18) Non-prototypical: action/event/process for state (highest degree of metaphorical “strength of contact” but lower degree of metaphorical boundedness: Only the source is bounded): Harry drinks (used to indicate that Harry is an alcoholic); here an activity (drinking), which is temporally bounded by definition activates a state (being an alcoholic), which is by definition temporally unbounded.

P&G’s prototype account of metonymy is based on their prototype account of the notion of contiguity. All in all, their highly suggestive article presents a detailed, highly structured, prototype-based classificatory scheme for a great many metonymies discussed in the literature, with an indication of the multiple motivational links between core and non-core metonymic types.15 In particular, the strength of contact dimension may be a major factor underlying the relative strength of the metonymic link (as noted by Panther and Thornburg, 1999, 2007), i.e., the relative strength of the pragmatic function link. This might be added as a further dimension in what I have called above metonymic typicality and prototypicality. The dimension of boundedness (both spatial and non-spatial) may be a major additional factor making up the distinctness of the target vis-à-vis the source, which has been claimed above to be a complex dimension along which typical metonymies are distinguished from purely schematic. P&G’s prototype approach can definitely interact with and supplement the prototype approach to metonymy presented in Section 2.5.2 above: It might well be the case that prototypical metonymies are “typical” referential metonymies operating in the spatial and material domains where source and target are two bounded individual entities linked by strict contact. I have, though, a number of reservations and critical comments to make on some aspects of P&G’s general approach to metonymy. P&G mention Barcelona (2002a) as one of the authors explicitly suggesting a prototype approach to the notion of metonymy. This proposal is not simply a “suggestion”, but a carefully articulated proposal, presented in the 2002 paper and with greater detail in Barcelona 2003a (actually written before the 2002a paper). Section 2.5.2 of the present chapter is a summary (and at the same time a refinement) of that proposal. The authors seem to suggest that a non-unitary, prototype-based definition of metonymy is preferable to a unitary, domain-based definition (by the way, P&G’s paper is somewhat self-contradictory, since it does not really manage to get rid of a notion of metonymy as contiguity operating within one single domain, be

15. Croft (2006) reacted to their paper and P&G added a reply, both in the same issue of the journal in which P&G’s paper was published.



Properties and prototype structure of metonymy

it the spatial or material domain, the temporal domain, the event and state domains, or the domain of assemblies and collections). I agree with P&G that a prototype approach to the study of metonymy is highly convenient to uncover the multiple connections holding across different types of metonymies. But that does not mean that an abstract, general definition of metonymy is not possible as well. As in other areas of technical linguistic categorization, a prototypical core may perfectly well co-exist with a general, highly abstract characterization of the category (on this point, see e.g., Langacker, 1987: 373ff). This definition is, in fact, essential to know why all of the cases discussed by P&G are metonymies (prototypical or not) in the first place. An attempt at such a definition is my notion of “schematic metonymy”. It is not clear whether or not metonymy is a mapping to P&G. They actually state that they are interested in the asymmetrical nature of the mapping (the description of metonymy as an asymmetrical mapping was proposed in ­Barcelona 2002a, 2003a, but P&G do not mention this fact). However, they do not examine this issue in their article. Nor do they discuss the related issue of the pragmatic function link between source and target; in fact, Fauconnier, the proponent of the description of metonymy as “pragmatic function mapping”, is not even cited by P&G. P&G do not discuss one of the main properties of metonymy, namely the activation of the source by the target, nor the related notion of metonymy as a reference point (they mention the term “reference point” in passing in p. 282, but it is not clear that they use it in the technical Langackerian sense; see Section 2.6). However, metonymy should be regarded as a process of activation of concepts that are often closely connected in experience, and which tend, therefore, to be mentally activated together or one after the other in quick succession. This is a better way of formulating the intuition expressed by means of the “contiguity” metaphor. P&G indirectly admit that contiguity is not the only factor in metonymicity on p. 310, when they state that the directionality of metonymies is also motivated by pragmatic factors, which they do not discuss, and that “not all contiguous entities can metonymically refer to one another”. The reason, in my view, is that the “conceptual contiguity” giving rise to metonymy must have the additional property of being a privileged pragmatic connection between source and target. P&G’s approach has immediate appeal to anyone familiar with the assumptions of cognitive linguistics. Claiming that temporal metonymies and those ­operating over assemblies and collections are derived by similarity from prototypical cases of strict spatial and material contiguity cannot fail to attract a cognitive linguistic readership. But it is not clear whether the extension really happened historically and if so, when it happened. P&G say (footnote 5) that the chapter “addresses the developments in the use of the term metonymy in the linguistic

29

30

Antonio Barcelona

literature rather than the historical developments of the metonymical patterns themselves”. The parallelism of the links connecting spatial metonymies to each other and, say, those connecting temporal metonymies to each other, seems to be P&G’s best piece of evidence that some extension of spatial metonymies to the other domains may have occurred in the use of the term metonymy by linguists. P&G seem to imply that since there exists a generalized metaphorical mapping of space onto time, it is only natural that the metonymic connections operating in the spatial domain should also occur in the temporal domain (with the limitations imposed by the inherent structure of the temporal domain). But the authors do not provide empirical evidence that this extension of the term metonymy by linguists actually took place. Maybe the term metonymy was used in the spatial and material domains and in the other domains from the beginning. They do not give clear evidence that metonymy was regarded as fundamentally a spatial or material relationship. However, there is overwhelming evidence that the traditional rhetorical theory of metonymy sees it as a device for referring to “entities”, which supports my above claim that the prototype of the technical category of metonymy is represented by referential metonymies for individuals. Whether or not rhetoricians and scholars viewed these prototypical metonymies as primarily involving a spatial part and whole seems reasonable but has to be proved.

2.6 Metonymic and non-metonymic “reference point phenomena” This section is an attempt at answering these questions: Is there any difference between metonymy and “reference point phenomena”? In particular, which systematic criteria can be used to distinguish non-metonymic from metonymic reference points? We stated in Section 2.4 that reference point phenomena are instances of perspective, which is in turn a dimension of construal. Langacker (1993: 5) defines the notion of “conceptual reference point” as the image-schematic “ability to invoke the conception of one entity for purposes of establishing mental contact with another, i.e., to single it out for individual conscious awareness”. He regards metonymy as a reference point phenomenon (Langacker, 1993: 29–35). Other linguistic manifestations of the reference point ability are, according to Langacker, possessive and quasi-possessive constructions, and topic and presentational constructions. The issue is complex, as these other reference point phenomena are fairly different from each other, and a detailed elucidation of the differences between schematic metonymy and these other phenomena requires further research. In fact, Langacker does not always distinguish clearly between metonymy and reference point (see e.g., Langacker, 2009).



Properties and prototype structure of metonymy

In my view, at least one of the differences between these other reference point phenomena and metonymy seems to lie in the fact, already hinted at by Langacker (1999: 198) that at least linguistic metonymy operates fundamentally on the paradigmatic plane (i.e., linguistically, it consists on the selection of a linguistic structure that will activate a linguistically unexpressed conceptual structure, that is, the implicit target). By contrast, the other reference point phenomena operate on the syntagmatic plane, so that a linguistic structure, say a possessive noun phrase such as The boy’s, provides mental access to a linguistically expressed conceptual target, the “possessed” (watch in The boy’s watch). Compare with the metonymy in (19) The Times hasn’t arrived at the press conference yet  (Lakoff and Johnson, 1980: 35)

in which the newspaper company is a reference point mapping onto the implicit reporter(s) by perspectivizing the latter primarily as an employee of that company.

2.7 Is there any difference between metonymy and such phenomena as “contextual modulation”, “facets”, and active-zone / profile discrepancy? Dirven (2002), following Cruse (1986, 2000) and unlike Taylor (1995/1989) would see Examples (4) and (5) as cases of “modulation” (a term borrowed from Cruse 1986), rather than as instances of metonymy, because the focusing on some of the constituent elements of an (abstract or physical) entity does not bring about the creation of new lexical senses. But metonymy need not result in established lexical polysemy. We have metonymy in Washington was not insensitive to the needs of the people without the lexeme Washington thereby acquiring a new established sense such as “the federal government of the United States”. And in any case, as I show later (Section 2.8, end paragraph), book does have several separate senses resulting from the metonymic activation of its primary or its relatively secondary subdomains. What Cruse (2000), Dirven (2002), and Croft and Cruse (2004) call “contextual modulation” results in the activation of what Cruse (2000) calls “facets” of meaning, i.e., aspects of a given lexical sense. We discuss “facets” and their connection to “active zones” and metonymy next. Example (5), This book is very large has been claimed above to be an instance of a “purely schematic” metonymy. But at the same time it is an instance of active-zone / profile discrepancy (“zone activation” for short), where the zone of the complex domain book that is activated is its physical object subdomain. The metonymy in (5) should be labeled whole thing (book) for active zone part (the physical object subdomain). A similar active zone metonymy ­operates

31

32

Antonio Barcelona

in (4), This book is a history of Iraq; the metonymy is whole thing (book) for active zone part (the semantic content subdomain). But an obvious question is: Should all instances of active-zone / profile discrepancy be regarded as metonymic? It is difficult to answer this question in categorical terms. I agree with Paradis (this volume) that there is a continuum from metonymy to zone activation. Probably some of the instances of zone activation discussed by ­Langacker are not typically or even schematically metonymic. Instances like Your dog bit my cat (Langacker, 1999) seem to satisfy some of the criteria for schematic metonymies: same (functional) domain and activation of the target by the source reference point. But it is not clear that the other basic criterion, namely the asymmetric pragmatic-function mapping (perspectivization) of source onto target really occurs. Do we really perspectivize the dog’s set of teeth from the dog in the same way as we perspectivize the United States from America in America will prevail over terrorism? Also the issues of frequency and acceptability are involved: mentioning the dog rather than the dog’s teeth (?Your dog’s teeth bit my cat) is more frequent and acceptable, whereas mentioning America instead of the United States is simply an option which is not necessarily more frequent or acceptable than the opposite option. In any case, instances of active-zone / profile discrepancy such as Your dog bit my cat, if really metonymic at all, would be on the borderline of the metonymy category. Geeraerts and Peirsman (this volume) provide some interesting arguments of a different sort to exclude this and similar instances of zone activation from metonymy. Another important difference is that zone activation does not preclude the actual explicit mention, for communicative purposes, of the active zone in addition to the reference point. Langacker (1999: 64–65) offers such examples as She blinked her big blue eyes, where the active zone of the personal subject she (her big blue eyes) is made explicit. This explicit mention of the target is in principle not possible in metonymy. Paradis (2004, this volume) discusses similar cases, and also some of the above book examples, and suggests that they cannot be metonymic. She regards the book examples not as instances of active-zone / profile discrepancy, but as instances of what she calls “facetization”. “Facets” are, according to her, aspects of the meaning of one lexeme (this shows her tendency to link metonymy with lexical semantics) that can be separately focalized, while not being antagonistic. She claims that real metonymies (where source and target, according to her, are antagonistic, hence incompatible with each other when in coordination or in anaphorical connection) block zeugma, whereas mere facetization (a different phenomenon from metonymy, in her view) allows zeugma. She claims that an example like The book is red and interesting has no metonymy but simply highlights two facets of the meaning of the lexeme book, whereas a “real” metonymy, in her view, would not allow zeugma as in *The piano was in a bad



Properties and prototype structure of metonymy

mood (activation­ of the pianist intended) and in black mahogany (activation of the instrument intended). However, we find unacceptable zeugmatic examples involving some of her favorite examples of facetization, as the examples in Geeraerts and Peirsman (this volume) show. One of them is *The whole department was made redundant and then torn down. As they claim, “the predicate was made redundant highlights the people facet of department, while [was] torn down focuses on the building. Even though these readings should not be antagonistic, zeugma is blocked”. On the other hand, one can occasionally find instances of undoubted metonymies that do not block zeugma, like their example That author is pretty young, but impossible to read, an instance of artist for work. And the observation by Taylor (1995/1989: 124–125) of the acceptability of zeugmatic examples of metonymy like We took the door off its hinges and then walked through it and I painted the window while she was standing in it, and his discussion of the metonymic senses of close and climb (Taylor, 1995/1989: 105–109) are consonant with a “schematic” notion of metonymy, and with Langacker’s notion of “active zone”. Zeugma is, therefore, not a good criterion for metonymicity. The cause for the acceptability or unacceptability of zeugma is probably to be sought elsewhere, perhaps in Ruiz de Mendoza’s (2000) “Domain Availability Principle”, or in Peirsman and Geeraerts’ “Identity of Reference Principle” (see Geeraerts and Peirsman, this volume). In any case, the examples discussed by Paradis (2004, this volume) in connection with zeugma-blocking as a criterion for metonymicity are instances of lexical or phrasal meaning, i.e., instances where either conceptual metonymy or what she calls “facetization” affects the contextual meaning of a particular lexeme or of a phrase (normally a noun phrase). We have seen zeugma-blocking is not a very useful criterion even in these cases. But what it simply cannot be applied is to other manifestations of metonymy, such as illocutionary and purely inferential metonymies (see Barcelona, 2002b, 2003b, 2005; Panther and Thornburg, 2003a). So this criterion is completely irrelevant as a criterion to decide when we have a conceptual metonymy at work or not. It might be useful, at best, to decide when we have a prototypical lexical or phrasal manifestation of a conceptual metonymy. Paradis (2004) regards, on the other hand, zone activation as the “the reverse of metonymy in that the conventional naming is constant both for the whole and the highlighted part”. This is quite wrong, in my view. Following Paradis’ line of thought, the lexeme ‘bracelet’ in (20) (20) A bracelet would be a good idea for Susan’s birthday

(an example which Paradis would regard as a non-metonymic instance of zone activation, and which I regard as an instance of metonymic zone activation) would be equally conventional as a name for the object and as a name for the

33

34

Antonio Barcelona

c­ ontextually relevant active zone, in this case an unmentioned relationship like “buying a bracelet” or “making a bracelet”; and in fact, it would be more conventional as a name for those relationships than expressions like buying a bracelet or making a bracelet. This conclusion would be absurd. According to Paradis, on the other hand, in an example like The piano is in a bad mood, which she would regard as an indisputably metonymic expression,16 the piano is less “constant” as a name for the pianist (the metonymic target meaning of the piano) than the noun pianist, which is the conventional name for this concept. But bracelet in the above example is not a conventional name for the active zone highlighted either. In neither case is the lexeme in question a “constant name” i.e., there occurs a meaning shift in both cases. So “constancy of naming” does not seem to be a valid criterion to distinguish metonymy from zone activation, at least not in all cases.

2.8 The conventionalization of metonymy This issue does not directly affect the notion of “conceptual metonymy” as such: A conceptual metonymy will still be one even if it is not (yet) completely conventional. However, I have included in the present discussion a brief comment on this issue17 because only fully conventional metonymies can be completely effective in communication across a whole culture or society, and not just within a subculture or a small social group (an example of a relatively unconventional metonymy is offered below). The investigation of the conventionalization of metonymy amounts to an investigation of the factors that seem to favor the conventionality of conceptual metonymies, which is in turn responsible in part for the conventionality of their expression (linguistic, gestural, pictorial or otherwise) or the conventionalization of their use in guiding established patterns of inferencing when the metonymy is not expressed (as is the case with purely inferential metonymies). Those factors are still poorly understood and more research on them is required. They can be classed into two main groups (see Taylor, 1995/1989: 122–123): a. The conformity of the metonymy in question to one or more of the default high-level patterns or types of metonymic mappings (types like part for whole, whole for part, producer for product, path for goal, etc.; see Kövecses and Radden, 1998: 48–61 and Radden and Kövecses, 1999: 44–54) for a systematic yet incomplete list of these default patterns).

16. Indisputably metonymic (i.e., not just an instance of zone activation) if one excludes a (less likely) metaphorical interpretation conferring animacy on the piano itself. 17. For further details, see Barcelona (2002a, 2003a).



Properties and prototype structure of metonymy



Simply typical and prototypical metonymies respond to these patterns, but so do purely schematic metonymies. For instance, in (5) we have the purely schematic metonymy whole thing (book) for active zone (its physical object subdomain): the book complex domain, in its role a whole thing, is mapped onto one of its parts, namely its physical object subdomain, in its role as active zone.

b. The social sanction of the metonymy. This is, obviously, the most important factor in conventionalization. Social sanction depends, in turn, on:

i. The number of general cognitive and communicative principles favoring default types of metonymy that apply in the case in question; see Kövecses­ and Radden (1998: 62–71) for these principles (cognitive principles like human over non human, concrete over abstract, and good ­gestalt over poor gestalt, and communicative principles such as clear over less clear or relevant over irrelevant).



ii. The presence or absence of a specific cultural (Taylor, 1995/1989: 123), ­social-interactional, or aesthetic principle (Kövecses and Radden, 1998: 71–74; Langacker, 1999: 199–200) favoring the conventionalization of that metonymy. These specific principles sometimes override the general cognitive and communicative principles.

As an example, take the sentence I have just bought a Picasso, which manifests the low-level metonymy Picasso for his work, in turn a manifestation of the high-level, default pattern author for work. This metonymy is motivated by the cognitive principles human over non-human, concrete over abstract, and good gestalt over poor gestalt, and probably also salient over nonsalient. It also gets conventionalized because there exists a cultural principle whereby works of art are regarded as unique products of the creative genius of artists, as an extension of their personality (ii). However, if my sister Jane paints what I take to be wonderful landscapes, which are, however, only bought by our family and a few friends, the specific realization of author for work intended by our aunt Mary in a sentence like I have just bought a Jane, certainly responds to the general pattern – factor (a) – and is motivated by the same cognitive-communicative principles as in I have just bought a Picasso – factor (b) (i) – but it is not socially conventionalized according to factor (b) (ii), except perhaps within my small family circle (see Taylor, 1995/1989: 123). An important point is that any of the three types or degrees of metonymy discussed in 2.5.2 above may become conventionalized and even give rise to established polysemy, when the metonymy affects lexical items. This normally happens with typical and prototypical metonymies, but it can also happen with

35

36

Antonio Barcelona

purely schematic metonymies. The targets of book in (4) and (5) have become essential parts of several established separate senses of the lexeme book. A standard dictionary like the Webster’s Dictionary (McKechnie, 1978) registers these senses, among others: “1. any literary or scientific composition or treatise which is printed…”; “2. any number of written or printed sheets when bound or sewed together along one edge, usually between protective covers”; “3. a volume of blank paper, or of printed blank forms, intended for any kind of writing, as for memorandums, accounts or receipts”; “13. a booklike package, as of matches, tickets, gold leaf, etc.”.

3. Problematic properties affecting the standard criteria in the distinction between metaphor and metonymy These problems are the following: 1. Cognitive domains often have fuzzy boundaries so that it is not always easy to know whether or not the source and the target domains are in the same superordinate domain. 2. A linguistic expression may often be interpreted, on the sole basis of context, background knowledge, or the purpose of the interpreter, as metaphorical or as metonymic. 3. Metaphor and metonymy very often interact in intricate patterns, a fact which complicates their distinction.

3.1

Problems derived from the notion of “domain”

The most important problem affecting the distinction between metaphor and metonymy in cognitive linguistics is the one that derives from the notion of conceptual domain or cognitive domain. In the encyclopedic view of linguistic meaning that prevails in cognitive linguistics, conceptual domains are normally open-ended. Conceptual domains, which can be defined as structured blocks of knowledge based on experience, are very often presupposed in other apparently separate domains, in an extremely intricate conceptual web (see Langacker, 1987: 147–182, especially 154–166). Langacker claims that only the domains which he calls basic domains do not presuppose other domains for their conceptualization: space, time, vision, pitch, taste and smell, temperature, pressure, pain, the emotions, and perhaps others. This conception of domains creates several problems for the standard domainbased definitions of metaphor and metonymy. There are two types of these:



Properties and prototype structure of metonymy

1. In certain metaphors, the source domain is apparently included in the target domain. 2. In certain metaphors, both the source and the target are included in the same superordinate domain. I discuss them separately below. The importance of these two problems is greatly diminished if one primarily distinguishes metaphor from metonymy in terms of these two other criteria: the nature of the mapping (multiple and symmetric in metaphor and simple and asymmetric in metonymy; see e.g., Feyaerts, 1999: 317) and (in the case of metonymy-based metaphors) the actual occurrence of the source domain in metonymy vs. its non-occurrence in metaphor (though this criterion is not completely predictive either; see Section 3.2). The same/different domain criterion would then simply be an additional criterion that works in most but not in all cases. However, it is necessary to refine this latter criterion, by specifying more precisely what type of “same” or “different” conceptual domain is meant. This will help us to improve our schematic definition of metonymy and also the standard cognitive-linguistic notion of metaphor.

3.1.1 The source domain is included in the target domain This problem typically applies to metaphors arising through the decontextualization or generalization of certain metonymies (Barcelona, 2000c; Radden, 2002). Metonymy-based metaphors of this sort often occur in emotional domains but also in many others. Let us illustrate the problem with a few examples. Probably no native speaker of English would deny that the cause–effect link between an emotion and certain physiological or behavioral responses is included in the overall domain of emotions. Most English speakers would agree that sadness is often manifested, among other responses, by a drooping bodily posture (drooping head, shoulders or facial muscles), and that anger is often manifested (or is believed to be manifested), among other responses, by an increase in body heat (on anger metaphors see Kövecses, 2000 and Soriano, n.d.). As for non-emotional domains, most speakers would agree that the height reached by a fluid in a container or by a pile of objects depends on their respective amounts, so that this cause-effect relationship (more fluid in a container or of objects in a pile results in a higher level of the fluid or of the pile). The behavioral reactions (the bodily postures) or the physiological reactions (the increase in body heat) can be used metonymically to activate the corresponding causal emotions (Kövecses, 1990, 2000); and a level of verticality can be used to activate metonymically a causal level of quantity. In my own study of the concept of depression in American English (Barcelona, 1986), I identified a conventional metonymy whereby, in a given pragmatic context, a drooping bodily posture (as effect) stands for sadness (as cause):

37

38

Antonio Barcelona

(21) a. Mary has a long face.18 b. John drooped his head (sadly). c. She walked with drooping shoulders / downcast eyes after the news of her child’s death.

As for anger and quantity: In (22a) an increase in body heat leading to redness in the face and neck area is used to activate metonymically its emotional cause, i.e., a certain degree of anger (effect for cause), so that the expression could be imperfectly paraphrased as ‘She got too angry’. And in (22b), a high level of verticality (full) reached by the milk in the milkjug (a container) is used by the second speaker to invite the inference that there is a large amount of milk (more than enough to mix with tea for two people): (22) a. She was scarlet with rage.  (Kövecses, 1990: 52) b. “Have we got enough milk for our tea, dear?” (wife to husband) “Yes, the large milkjug is full.”

In (21) the emotional effect downward bodily posture presupposes in turn the domain of verticality, which in turn presupposes that of space. If we stretch our logic but a little, we might end up concluding that verticality and space are also a part of the cognitive domain of sadness. Therefore, the mappings in examples like those in (23) (23)

a. b. c. d. e.

She is in the pits. Mike is in low spirits. I am prostrate. Mary is down in the dumps. Her spirits drooped.

would be metonymic and not metaphorical, because a part (verticality, specifically its lower pole down ) of the overall domain of sadness would be mapped, as a part of the effect of sadness consisting in a drooping bodily posture, onto the overall domain of sadness; this would be then a part for whole metonymy. The same problem would be posed by examples (22a–b): If, as a typical physiological effect of anger, body heat (which presupposes heat) is a part of the anger domain, then heat is included in anger. And if, as a typical effect of an increase in quantity, the level of verticality reached by a fluid in a container or by a substance

18. That is, vertically ‘long’, with drooping facial muscles. To put/have a long face means in English to look sad, dismal or serious. The equivalent expression has a similar meaning in Spanish, but in Dutch it means to look discontent or angry. (I am grateful to René Dirven for the remark on Dutch.) The behavioral effects of sadness do not seem to be represented in the same way even in related European cultures.



Properties and prototype structure of metonymy

or a group of objects in a pile is included in the quantity domain, then verticality as a whole is included in quantity. Again, examples like those in (24) (24) a. You make my blood boil. b. The prices have soared

would be metonymic and not metaphorical. However, most cognitive semanticists would intuitively be inclined to regard the sentences in (23) as manifestations of the metaphor sadness is down and not of the metonymy down (a part of verticality) for sadness, to regard the sentence in (24a) as a manifestation of the metaphor anger is (the) heat (of a fluid in a container) rather than of the metonymy body heat for anger; and to regard the sentence in (24b) as a manifestation of the metaphor more is up rather than of the metonymy verticality scale for quantity scale. Note that the sentences in (23) and (24a), unlike those in (21) and (22a) are used to describe the emotional state of someone who may not actually be displaying a drooping posture behavior or an actual increase in body heat; in these cases, the behavioral or physiological effect would not be used as a conceptual link to activate the emotional cause in the hearer’s mind. And, whereas the reply in (22b) activates the notion of a certain degree of quantity from the notion of a certain degree in height, both domains actually and literally co-occurring in the situation described by the example, only the quantity domain seems to be literally involved in the situation described by (24b).19 There certainly exists, thus, a conventional metaphorical mapping of down onto sadness and of heat onto anger which does not have to co-occur with actual bodily behavior or with an actual physiological effect, and a conventional metaphorical mapping of verticality onto quantity (more is up / less is down), which does not have to co-occur with an actual change in height. This does not mean that in these, as in a great many other cases, the metaphor is not ultimately based upon a metonymy.20 The problem is that the source (verticality, heat) is apparently included in the target (sadness, anger, quantity). 19. The fact that the source domain is not really “present” in the situation described is a strong symptom of metaphoricity. But, as stated above, this criterion is not completely predictive. Therefore, at least some of the sentences in (23) and (24) might be interpreted metonymically by some speakers. See Section 3.2 for the fundamental role of the interpreter in these cases. 20. I have argued elsewhere that the metonymic motivation of metaphor is in fact the rule, rather than the exception (see Barcelona, 2000c, and below). But the fact that metaphors are often based on metonymies does not mean that they can be totally reduced to (post-)metonymies. In this respect, I cannot agree with Riemer (2002), if I interpret his position correctly, since metonymy-based metaphors (see also Goossens, 2002/1990; Kövecses and Radden, 1998; Radden, 2002) are more complex mappings than the metonymies originating them.

39

40 Antonio Barcelona

A possible way out of this theoretical puzzle may be, in my view, to notice that there exists in every culture a conscious icm (Lakoff, 1987) of the taxonomy of domains (a sort of folk classification of the world), which specifies, to a greater or a lesser extent, the hierarchies of domains, and which excludes certain (sub)domains from others. The spatial domain (verticality) may be indirectly included in the domain of sadness (or happiness) via people’s experiential knowledge of the behavioral effects of this emotion (or its opposite). But this does not mean that it is actually included in the overall domain of sadness or happiness by our conventional folk classification of domains: no native speaker of English (or, for that matter, Spanish, Italian, and other European languages) is likely to categorize verticality (and space with it) consciously as a subdomain of sadness or of happiness. If the native speakers of any of these languages are asked, prior to showing them the sentences in (23), whether verticality (let alone space in general) is a part of (the domain of) emotions, they will give a negative answer. These remarks should not be interpreted as claiming that every time the sadness is down metaphor is activated, people are conscious of the fact that verticality belongs to an altogether separate domain. As is well-known, both metaphor and metonymy are typically unconscious mental processes. What I mean to say is that speakers would normally reject the proposition that the notion of sadness includes that of verticality or space in general, if it is explicitly submitted to them. This rejection would be evidence that their conscious folk taxonomy of domains separates sadness (or happiness) from verticality. Therefore, when metaphor is said to be a mapping across two different domains, this should be interpreted as meaning that these domains are not included by that taxonomy in the same overall domain. The same applies to (24). Speakers do not seem to consciously include heat in general as a subdomain of anger, or verticality as a subdomain of quantity.

3.1.2 In certain metaphors, both the source and the target are actually included in the same superordinate domain The refinement of the standard definition that has just been suggested in Section 3.1.1 may help us to distinguish metonymy from metaphor in many cases. However, it would not be of great help in certain others. Few cognitive linguists would regard an example like (25) (25) John is a lion.

as metonymic. And yet no average native speaker of English would deny that both the domain of animals (the source) and the domain of people (the target) are centrally included in the overall taxonomic domain of living beings. I suggest below a further refinement of the criteria used to distinguish metaphor from metonymy. This refinement consists:



Properties and prototype structure of metonymy

1. in distinguishing, to the extent that this can be done, between the experiencebased taxonomic classification of domains and the co-occurrence of domains in an experience-based functional superordinate domain. A domain X may be distinguished from a domain Y in terms of the general taxonomy of domains, and yet be grouped with it in an overall functional experiential domain, i.e., a frame (Fillmore, 1985) or an icm (Lakoff, 1987: 68–77). 2. in taking into account the presence or absence of a pragmatic function (Fauconnier­, 1994, 1997: 11) linking the two domains.21 Now if source domain X and target domain Y are linked by a pragmatic function, then one of them may eventually be mapped onto the other in a metonymic relationship. Irrespective of the fact that they belong or not to the same overall taxonomic domain, this pragmatic link shows that they are grouped in the same functional domain. The pragmatic link in these cases connects a domain and its role in the functional domain to another subdomain and its role in the same functional domain, so that one of the domains activates the other, as a reference point for it. It was briefly stated above (Section 2.4) that the term “functional domain” designates the same type of coherent blocks of knowledge that is designated by the terms “frame” and icm (an “idealized cognitive model”, in Lakoff ’s (1987) terminology). The reason why the term is used here is that this term allows us to specify that the type of “domain” claimed by Lakoff and Turner to be involved in metonymy22 is not a generic, unspecified “taxonomic” domain, but a more specific type of domain, i.e., a frame or icm. Frames and icms are “functional domains” because they are functional in constantly guiding our behavior and communication thanks to the detailed organized knowledge of specific areas of experience that they include, whereas “taxonomic” domains have fundamentally the function of classifying those areas into large groups. For example, governments and buildings can be said to be in two different superordinate taxonomic domains, respectively the domains of abstract entities and of inert physical entities. However, they can be connected by a number of functional domains. The U.S. federal political institutions icm is a functional domain under which the U.S. government is linked to the White House. Furthermore, there exists a pragmatic function linking the U.S. government, in its locatum role, to the White House, in its location role (location ⇒ 21. Gradečak-Erdeljić and Milić (this volume) also favor the relevance of the functional domain and pragmatic function in the definition of metonymy. Kövecses and Radden (1998) also make it clear that metonymy operates within an icm, i.e., a functional domain. 22. “A metonymic mapping occurs within a single domain, not across domains” (Lakoff and Turner, 1989: 103).

41

42

Antonio Barcelona

locatum). The result is that there can be, and in fact there is, a conventional metonymy that could be called White House for U.S. government, as in (26): (26) The White House did not intervene.

A different situation holds in examples like (23) above. Here, although down and sadness are in two different taxonomic domains (space and emotions) down, at the same time, can be said to be functionally included in the sadness domain, no matter how indirectly or unconsciously. However, there is no pragmatic function directly linking down to sadness. The direct pragmatic link is that between downward bodily orientation (in its effect role) and sadness (in its cause role). Thus, the link between down and sadness is not strictly metonymic, but metaphorical (even though this metaphor, being based on a generalization of the metonymy linking the bodily posture and the emotion, is on the very borderline of metaphor and metonymy, and perhaps down for sadness can be regarded as the final metonymic step in the generalization from downward bodily orientation for sadness to sadness is down; see Barcelona (2000c: 43–44); see also Radden (2002: 409) for the “literalness-metonymy-metaphor” continuum). Thus, if source and target are not linked by a pragmatic function, even if they are in the same functional domain, then they cannot stand in a metonymic relationship (remember the example, commented on in Section 2.4, of the nose and the mouth being in the same face icm, yet neither being linked by a pragmatic function to each other and thus not being connected by metonymy). On the other hand, two domains may be consciously included in the same general taxonomic domain, like animals and people, both included in living beings. But if they are not additionally included in the same functional experiential domain, then they may stand in a metaphorical relation, not in a metonymic one. This is what happens in (25), in which the lion domain, or at least its stereotype, is metaphorically mapped onto the human domain. However, the same two domains may be grouped under an overall functional experiential domain and stand in a metonymic relation. Take the horse racing icm, which groups a certain subdomain of the animal domain with a certain subdomain of the people domain. Then, if a pragmatic function connects these two domains, they can stand in a metonymic relationship. The pragmatic function is controlled ⇒ controller, in (27): (27) The white horse acted very cleverly (referring to the rider of the white horse, who managed to win the race with a clever final maneuver).

People and animals are both in the same taxonomic domain of living beings, and also in the same functional domain (the horse racing icm), and they are connected by a pragmatic function.



Properties and prototype structure of metonymy

Like the taxonomic grouping of domains, functional groupings (frames, icms) can be subjected to explicit conscious recognition. That is, speakers are normally aware of the functional connection between certain people and certain animals in the horse racing icm, or of the functional association between the White House building and the executive branch of the U.S. government. The requirement of the existence of a pragmatic function link affects any type of metonymy, including what I have called above purely schematic metonymies (normally certain types of whole for part metonymies), which are not likely to be noticed as a semantic shift. The pragmatic function link between source and target is sometimes quite generic (sometimes just part ⇒ whole, or whole thing ⇒ active zone part), as in (5), in which the whole book domain is mapped onto its physical object subdomain. A similar proposal is made by Dirven (2002), who suggests that the distinction between domains is a matter of construal. In sum, the notion of domain must be flexibly understood in either its taxonomic or its functional sense, with the latter (together with the pragmatic function link) having a decisive role in distinguishing metaphor from metonymy.23

3.2 A linguistic expression may sometimes be interpreted, on the basis of context, background knowledge or the interpreter’s choice, as metaphorical, metonymic, or as a combination of both This situation is very frequent, and has been noted by several authors (e.g., Bartsch, 2002; Goossens, 2002/1990). These are cases like (28) (adapted from Bultinck, 1998): (28) He fell in the First World War. He was buried in Paris.

Depending on the context of interpretation, this sentence could be interpreted metaphorically or metonymically. If it describes the conventional image of the soldier that is seriously wounded, falls and dies, then the meaning could be claimed to be metonymic: a salient event (falling) for a salient successive [or co-occurring] event (dying). If, however, the speaker simply refers 23. Benczes (this volume) claims that metonymy operates in the context of what she calls “domain network”, i.e., not one single domain but a rather complex web of domains that are involved in the interpretation of linguistic constructions (her example are noun–noun compounds). This notion is, on the one hand, close to that of “functional domain” (since a frame or icm does involve multiple “taxonomic” domains) and, on the other, it is close to Langacker’s notion of base, as this constitutes the domain network presupposed by a “predication” (a word or a higher construction).

43

44 Antonio Barcelona

to the fact that the soldier died when he was in the war, irrespective of the fact that he died when he was lying asleep in his bed (perhaps because the barracks was bombed at night), the domain of falling is used as a generalized metaphoric source for the domain of dying at war. Doubtless the generalized metaphorical mapping is grounded in the prototypical metonymic connection between actual falling and dying in the war. This would then constitute an instance of a metaphor motivated by the decontextualization and generalization of a metonymy (see Section 3.3 below); in these cases the same linguistic expression can manifest the metonymy or the metaphor, which results in ambiguity. The interpretation of (28) as metaphorical or metonymic depends of the listener’s awareness of the context (including the co-text) and/or the speaker’s communicative intent. Knowledge of these factors is, then, decisive. This is a general problem with linguistic interpretation in any area. These relatively frequent instances of ambiguity do not undermine the validity of the cognitive-linguistic distinction between metaphor and metonymy, as the interpreter can always freely choose to apply either type of conceptual mapping in the absence of contrary indication from the context. The decisive role of the interpreter’s awareness of context and of the speaker’s purpose is the main reason why metonymy-based metaphor and metonymy cannot always be distinguished in ambiguous cases on the sole basis of the (knowledge of) the presence or the absence of the source. In (28), knowledge of the absence of the source in the situation described by this sentence (i.e., of the fact that the soldier did not literally (i.e. spatially) fall down before or when dying, given that he died in his bed) leads to a metaphorical interpretation, because then falling would be used as a metaphorical source for dying. The fact that the source domain is not really “present” in the situation described is often a strong symptom of metaphoricity (when you say Prices are soaring, you know that the prices are not literally going up in space, but simply increasing). But this symptom is not completely predictive. The source may not be literally present in the situation and the hearer/ reader may still make a metonymic interpretation. At least some of the putatively metaphorical sentences in (23) and (24) might be interpreted metonymically by some speakers. The sentence I am prostrate (Example (23c)) would normally be interpreted metaphorically if the speaker, when uttering this sentence, is not literally adopting the corresponding bodily posture (i.e., “lying at full length or with the body extended flat” – OED; see Benbow, 2002). But it could also interpreted ­metonymically as an activation of sadness (or other mental states such as submissiveness, resignation, etc.) via the activation of the notion of a possible downwardoriented body posture, even if this posture is not actually exhibited. By contrast, (23a–b) and (23d–e) are unambiguous cases of the metaphor sadness is down, and very unlikely to be interpreted as instances of the metonymy downward



Properties and prototype structure of metonymy

bodily posture for sadness. As for (24), (24a) could be read by some interpreters as metonymic, because one of the effects believed to be caused by anger is body heat: Since blood temperature is a type of body heat, the metonymy body heat for anger could apply, even if no actual increase in blood temperature really occurs (in fact, it could not really boil except under lethal circumstances – the expression is hyperbolic at the same time). On the role of knowledge of the context and of speaker’s intention in the recognition of a combination of metaphor and metonymy, see Section 3.3.

3.3 Interaction of metaphor and metonymy The main problems affecting the distinction between metaphor and metonymy are those discussed in Sections 3.1 and 3.2 and the criteria suggested there should help us to distinguish the two types of mapping in most cases. Therefore the topic of this section is less relevant for their distinction, since it reduces to identifying the various patterns of interaction between metaphor and metonymy in linguistic expressions. However, it is necessary to recognize these patterns to avoid classifying their linguistic manifestation (or their manifestation in any other mode) as exclusively metaphorical or exclusively metonymic. The patterns have been discussed in some detail from a cognitive-linguistic perspective by Lakoff and ­Turner (1989) and more systematically by Goossens (see, e.g., Goossens, 2002/1990; or Goossens et al., 1995), though much more research in this area is necessary. A very clear discussion is offered in Kövecses (2002: 156–161). In my view these patterns can be classified into these two main types:24

A. Conceptual motivation of metaphor by metonymy and of metonymy by metaphor In this type of interaction, the process motivating or underlying the other is not linguistically expressed. I use the verb “motivate” (in such syntagms as “motivate a conceptual metaphor” or “motivate a conceptual metonymy”) in a special sense, namely “be a conceptual prerequisite for” (Barcelona, 2000c). That is, a certain conceptual metonymy may motivate a certain conceptual metaphor if the latter is possible, among other factors, thanks to that conceptual metonymy; in other words, if it cannot occur independently from it. Two main patterns of conceptual motivation have been discussed:

24. What follows is a substantial revision of the views expressed on this issue in Barcelona (2000b and 2002a).

45

46 Antonio Barcelona

A1. Motivation of a conceptual metonymy by a conceptual metaphor. In the sense of motivation used here, this pattern is extremely rare, if it exists at all. In Barcelona (2000c: 11–12), guided by sense 6 of ear in the OED (Benbow, 2002) entry for this lexeme,25 I simply assumed that the metonymy ear for attention, which underlies the use of ear in examples such as (29) She caught the Minister’s ear and persuaded him to accept her plan26

could only occur in linguistic expressions of the metaphor attention is a (typically moving) physical entity (that one has to get hold, attract, “call”, win, gain, have, etc.). Unfortunately, one can find at least one instance of the metonymy occurring independently from that metaphor: To be all ears ‘to be eagerly attentive’. On the other hand, what Goossens (2002/1990: 349, 367–368) calls “metonymy from metaphor” (i.e., a metonymy conceptually motivated by a metaphor) is not attested in his corpus. I have not found any instances of this pattern in my studies on metaphor in certain types of discourse (some of them are referred to in Barcelona, 2000a, 2000b, and 2002a) either. A2. Motivation of a conceptual metaphor by a conceptual metonymy. As pointed out by Goossens (2002), Goossens et al. (1995), Barcelona (2000b, 2000c), ­Radden (2002), this is by far the more frequent pattern. In Barcelona (2000c), I distinguished two main subtypes: 1. In one of them, the source and the target are both understood metonymically from a shared subdomain or aspect. For example, in (30) I listened to a sweet ballad



both “sweet” music and sweet edibles are perspectivized metonymically from the pleasurable effect caused by both physical experiences. This metonymic “reduction” and “perspectivization” helps us create an abstract similarity between source and target, which makes the metaphorical mapping possible. This type of metonymic motivation is typical of, among others, synesthetic metaphors like the one in (30) or the one in That’s a loud color.

2. In the other, a metonymy involving the metaphoric target and the metaphoric source becomes decontextualized (generalized). This type of metonymic motivation has been illustrated in Section 3.1.1, with examples of sadness is 25. “Voluntary hearing; listening; attention. Chiefly in phrases like to give ear ‘to listen attentively’; to have (win, gain) a person’s ear: ‘to have (obtain) his favourable attention.” 26. This example was drawn from Goossens’ corpus. He treated it as an instance of “metonymy within metaphor” (Goossens, 2002/1990: 364–365).



Properties and prototype structure of metonymy

down, more is up, and anger is heat (as regards anger is heat, Kövecses­ and Radden, 1998: 61, had already pointed out that this metaphor “arises from a generalization of body heat to heat”). The two main types of metonymic motivation are not mutually exclusive and can often co-occur.

B. Conceptual combination of a conceptual metaphor and a conceptual metonymy resulting in the same expression (linguistic or otherwise) In Lakoff and Johnson’s famous example The ham sandwich is waiting for his check (Lakoff and Johnson, 1980: 35), the food activates the customer who ordered or consumed it. This is a part for part metonymy occurring within the restaurant frame: The sentence might be uttered, for example, by a waiter talking to the person in charge of the cash register. A conceptual metonymy can combine with a conceptual metaphor provided both are compatible (Barcelona, 2002a: 245). In an example like (31) The ham sandwich started snarling

the metonymy food for customer combines conceptually with the metaphor angry human behavior is aggressive animal behavior (which is a submetaphor of people are animals), because both the metonymy and the metaphor have a class of people, or an aspect thereof, as target. The target is a type of people, namely restaurant customers, in the metonymy and a type of human behavior, namely angry human behavior, in the metaphor. It is conceptually possible to predicate of a type of people a type of behavior which that type of people could engage in. This conceptual combinability results in linguistic combinability, with the metonymy’s target being cued by the subject of (31) and the metaphor’s target being cued by the verb started snarling.27 Conceptual compatibility also underlies Goossens’ (2002: 364–365) instances of “metonymy within metaphor”, including the “catch ear” example, which should be regarded as instances of conceptual combination of a conceptual metaphor and a conceptual metonymy.

27. Note that in this example of interaction the metaphor and the metonymy are independent from each other, as each can occur without the other. In The ham sandwich is waiting for his check the metonymy food for (restaurant) customer occurs without the metaphor angry human behavior is aggressive animal behavior. Conversely, in John started snarling, the same metaphor occurs without the metonymy food for customer. If that metaphor is conceptually motivated by a metonymy, it is certainly not motivated by food for customer: It simply combines with it.

47

48 Antonio Barcelona

In (31), the recognition of the metaphor and the metonymy and of their conceptual interaction prevents the interpreter, together with pragmatic principles such as relevance or the cooperative maxims, from dismissing the sentence as nonsensical or incomprehensible. This recognition depends, as stated in Section 3.2, on context and/or background knowledge: The interpreter must know the type of person that uttered the sentence and the context of utterance. A further example of the fundamental role of these discourse factors is (32): (32) The Three Lions Roared.28

The three lions are the three lions depicted on the emblem of England’s soccer team, which were taken from the Coat of Arms of England. Knowledge of the text where (32) occurs (an ad for a DVD with that title on the history of England’s soccer team) and background knowledge about the team’s emblem is required to interpret the noun phrase The Three Lions as manifesting a metonymy mapping on a variety of targets (which are themselves determined with each occurrence of the phrase in the text), ranging from the team’s emblem, the team’s shirt, the team as a whole or its players, or even England as a whole; in the DVD title, it seems to refer to the team as a whole. Once the metonymic target is established, the metaphorical reading is facilitated (the metaphor would map the aggressive behavior of lions (roaring) onto the high quality performance of the team. Both the metonymy and the metaphor are compatible since they share the same target.29 These discourse factors are essential in this case to dismiss the literal reading (“three actual lions roared”), which would be strongly favored by the collocational affinity between lion and roared.

4. Summary and conclusions. Revised general definitions of metonymy and metaphor A number of problematic properties in the standard cognitive-linguistic notion of metonymy have been discussed in this chapter: It has been claimed that metonymic sources and targets constitute “domains” (or subdomains), rather than just “entities” conceived in isolation. In other words, 28. See http://www.lovefilm.com/film/The-Three-Lions-Roared-The-History-Of-England/55748. 29. As usual, blending (Fauconnier and Turner, 2002) often results from metaphor, metonymy or their combination. The blend between a great team (implying among other things the ability to establish their position and to threaten the opponent’s goal) and a lion (which marks its territory and threatens its opponents by roaring, among other actions) is particularly strong here thanks to the frequent collocation of lion with roar in the metaphor’s source domain.



Properties and prototype structure of metonymy

the source and the target include at least the more relevant facets of speakers’ encyclopedic knowledge about them. As for referentiality and “stand-for”, the former notion has been shown not to be a necessary condition for metonymicity, and it has been suggested that the latter notion should be dispensed with and replaced by “mapping”. The properties, often attributed to metonymy, of being a “mapping” and a type of “highlighting” and “activation” have been treated together, since they are closely connected. I have claimed that Croft’s highlighting amounts to activation or mental access; that the mapping in metonymy is an asymmetric mapping and that it consists of the activation and the (transient or permanent) conceptualization (construal) of the target from the perspective set by the source, which is the reference point setting this perspective. It has also been claimed that the source can act as a reference point for the target in virtue of the strong experiential link between the source and the target in the “domain” where they operate. This strong link is the “pragmatic function” connecting the roles of source and target in a given domain (author–work, agent–action, etc.). As for the directionality of metonymic mappings, my position is that metonymy is a unidirectional mapping (i.e., that there are no simultaneously bidirectional metonymic mappings whereby a container at the same time activates a content and vice-versa), though metonymies are frequently reversible. The next cluster of problematic properties discussed in the chapter has required a fairly extensive discussion. One of them is the controversial issue of the requirements that must be satisfied by a part to qualify as a metonymic target of the corresponding whole. The debate concerns examples like (4), This book is a history of Iraq or (5), This book is very large. Can book be considered as metonymic in these cases? The answer suggested in this chapter is positive. In Example (4), the whole (the book) maps onto its semantic content subdomain, which is a relatively secondary domain (in Langacker’s sense of secondariness) within the book domain matrix. Domain secondariness is a multifactorial, scalar property and it does not depend solely or mainly on extrinsicness, as Croft seems to contend. In (5), the whole (the book) maps onto its physical object subdomain, which is a primary subdomain within the book domain matrix. However, both their semantic content and their physical make-up are neatly distinguishable “parts” of books, which can be separately made the focus of attention. In any case, (5) exhibits at least all the basic properties of metonymy (asymmetric pragmatic function mapping within the same domain bringing about the activation of the target). Whether or not the target is a secondary or primary subdomain within the source is less decisive. This broader view of metonymy allows us to realize that the same basic cognitive process is at work in many apparently disparate phenomena, i.e., that there exists a fundamental similarity between (5) and uncontroversial examples of conceptual metonymies like (6)–(14).

49

50

Antonio Barcelona

However, the differences among these phenomena must also be recognized and accounted for. In particular, we must account for the fact that certain metonymies are more readily recognized as such (i.e., as semantic shifts) than others both by linguists and by language users in general. This led me to the proposal some time ago of a prototype approach to metonymy resulting in a continuum of metonymicity. Three main types of metonymy in terms of prototypicality are presented; the more prototypical types presuppose the less prototypical types so that all of them are connected: 1. Purely schematic metonymies: Those satisfying only the necessary and sufficient conditions for metonymicity, i.e., those that are only schematic metonymies. They are extremely common, though they are fairly removed from the prototype of the metonymy category. 2. Simply typical metonymies: Schematic metonymies whose target is clearly distinct from the source, because it is a (relatively) secondary subdomain of the source (in whole for part metonymies), or because it is not included in it (in part for whole or part for part metonymies). 3. Prototypical metonymies. These are highly typical referential metonymies (i.e., scoring high on all the factors of distinctness between source and target) whose target is a particular individual entity or a collection of specific individuals (rather than a class). Non-referential metonymies such as predicative, predicational, illocutionary and purely inferential metonymies are simply typical, not prototypical metonymies.30 With respect to Peirsman and Geeraerts’ (2006a, 2006b) prototype account of metonymy, I have voiced some of my reservations, mainly the need to supplement a prototype approach with a schematic, unitary notion of metonymy, and the possible confusion between the metaphorical extension of the category by linguists and the actual use of metonymy by speakers, which may not be metaphor-based and be simply based on experiential connections. But on the whole I view their approach as compatible with my own. The following problematic property is the status of conceptual metonymy as a “reference point” phenomenon, which raises the issue of how to distinguish conceptual and linguistic metonymy from other linguistic structures that ­Langacker has also treated as reference point phenomena (such as possessive and quasipossessive constructions and topic and presentational constructions). This is a 30. Barnden (2010) claims that an absolute distinction between metaphor and metonymy is not possible because it is difficult, if not impossible, to come up with unitary, general definitions of metaphor or metonymy. But he acknowledges the possibility of prototype-based definitions of both phenomena.



Properties and prototype structure of metonymy

­ ifficult issue. The (provisional) answer suggested here is that linguistic metonyd my operates on the paradigmatic plane and activates a normally implicit target, whereas the other reference point constructions operate on the syntagmatic plane with a normally explicit target. The distinction of “modulation”, “facets”, and zone activation from metonymy is another difficult issue. My position is that “contextual modulation” (which results in the activation of “facets”) is indistinguishable from purely schematic metonymies. One of the reasons is that metonymy does not have to result in established polysemy. Another reason is that, pace Paradis (2004, this volume), zeugma-blocking is not a valid criterion to distinguish between lexical metonymy and facets (even less so between non-lexical or non-phrasal metonymies and facets). As for zone activation, some instances like Your dog bit my cat are doubtfully metonymic, since it is difficult to claim that the dog’s teeth are perspectivized from the dog. Also, the active zone is sometimes explicit, which is not normally possible in metonymy. On the other hand, the criterion, suggested by Paradis (2004), of the “constancy of naming” in zone activation and the lack thereof in metonymy does not seem to work either, since such constancy is missing in certain instances of zone activation. The (un)conventionality of a conceptual metonymy does not impinge directly on its status as metonymy, but it determines its communicative effectiveness across a whole culture or society. This is why this issue is addressed in this chapter. The most important factor is social sanction. Importantly, even though simply typical and prototypical metonymies are more likely to become conventional, some purely schematic metonymies can also become conventional and even give rise to established polysemy. The problematic properties derived from the distinction between metaphor and metonymy in the standard cognitive-linguistic account are the last issue discussed in the chapter. One of the most controversial properties of metonymy proposed by that standard account is its distinction from metaphor in terms of the same/different domain criterion (for a summary see, among others, ­Barnden, 2010 or Peirsman and Geeraerts, 2006a and 2006b). It is suggested in this chapter that this criterion should simply be used to supplement another two criteria, namely the nature of the mapping (“asymmetric” and allied to a “pragmatic function” in metonymy vs. “symmetric” in metaphor) and the actual “absence” of the source in metonymy-based metaphors (this second criterion is not always completely predictive). The main refinement suggested in the chapter of the definition of metonymy in terms of the same / different domain is to view metonymy as a mapping within a “functional domain” (an icm or a frame). Although the criteria suggested in the chapter should help us to distinguish metaphor from metonymy in most cases, the actual interpretation of a linguistic expression as

51

52

Antonio Barcelona

metaphorical, metonymic or “metaphtonymic” sometimes depends exclusively on context, shared knowledge and the interpreter’s choice (and this is why the actual absence/presence of the source does not always work as a criterion). But this fact does not undermine the theoretical distinction between metaphor and metonymy. Nor does the frequent interaction between metaphor and metonymy prevent us in principle from distinguishing them. This issue is included in this chapter because it is necessary to recognize at least the main patterns of interaction to avoid categorizing their manifestation in language or in any other mode as exclusively metaphorical or exclusively metonymic and to avoid unfounded claims on the impossibility of distinguishing metaphor from metonymy. Two main patterns of interaction are recognized in this chapter. One of them is motivational interaction, i.e., either motivation of metonymy by metaphor (very rare if it exists at all) or motivation of metaphor by metonymy (extremely common, with two often overlapping subtypes). The other pattern is conceptual combination of metaphor and metonymy resulting in their combination in expressions, linguistic or otherwise; this pattern is very common and requires the conceptual compatibility of the metaphor(s) and the metonymy/ies.

4.1 Revised definitions of schematic metonymy and of metaphor The preceding discussion has allowed us to elucidate the main properties (i.e., the basic, “schematic” properties) of conceptual metonymy, but also some of the main properties of conceptual metaphor, through our efforts to specify clear criteria to distinguish conceptual metonymy from it. As regards metonymy, its main properties can now be put together in this revised definition of schematic metonymy, which is at the same time a general, unitary definition of metonymy: Metonymy is an asymmetric mapping of a conceptual domain, the source, onto another domain, the target. Source and target are in the same functional domain and are linked by a pragmatic function, so that the target is mentally activated.

This definition modifies the one given in Section 2.5.2 by adding the specification that the common domain is a “functional” domain (icm, frame), a further property proposed in the course of the discussion of the difference between metaphor and metonymy. Panther and Thornburg (2007: 240–241) have also suggested, as a further definitional property of metonymy, its contingency (i.e., its defeasibility). I will not discuss their suggestion in any detail here for lack of space and also because it does not seem to be followed by other cognitive linguists, and thus does not seem to be part of the standard cognitive-linguistic notion of conceptual metonymy. I can only note that, though a great many metonymies do seem to be



Properties and prototype structure of metonymy

contingent (e.g., in Example (27), the speaker might later indicate that she was referring to the horse, not to the rider), many others do not seem to be so (in The kettle is boiling it is difficult to cancel the inference that what is boiling is not the kettle but some liquid inside the kettle or at least on its surface). As regards metaphor, its main properties can be put together in the following revised definition: Metaphor is a symmetric mapping of a conceptual domain, the source, onto another domain, the target. Source and target are either in different taxonomic domains and not linked by a pragmatic function, or they are in different functional domains.

As explained in earlier sections, by “symmetric mapping” I mean a partial mapping of the structure of the source onto that of the target, so that each source element mapped has a counterpart in the target. This definition would exclude cases like The piano is a real genius (obviously intended to apply to the pianist) as metaphorical. Although piano and piano player are in two different taxonomic domains (inert objects and people), they are here in the same functional domain (the concert frame, or the orchestra frame, depending on the discourse context), and linked by a pragmatic function (instrument–agent), hence there is only metonymy, not metaphor, in the interpretation of the piano as the piano player. The definition would include instances like The piano started playing nasty tricks on me in the middle of the concert (the pianist talking about the malfunctioning of her instrument). The domains connected (pianos and ill-willing human beings) are in two different superordinate taxonomic domains (inert objects and people), and in this case they are not linked by a pragmatic function nor are they included in the same functional domain (ill-willing human beings are not obligatory subdomains/elements of concerts or orchestras).

References Barcelona, Antonio. 1986. On the concept of depression in American English: A cognitive approach. Revista Canaria de Estudios Ingleses 12: 7–33. Barcelona, Antonio. 1999. Notas sobre la teoría cognitiva de la metonimia y su poder explicativo. In Ruiz de Mendoza, Francisco (ed.), Perspectivas aplicadas de la organización del conocimiento, procesamiento y uso del lenguaje (913–938). Logroño: Universidad de La Rioja, Spain. Barcelona, Antonio (ed.). 2000a. Metaphor and Metonymy at the Crossroads: A Cognitive Perspective. Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2000b. Introduction. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (1–28). Berlin & New York: Mouton de Gruyter.

53

54

Antonio Barcelona

Barcelona, Antonio. 2000c. On the plausibility of claiming a metonymic motivation for conceptual metaphor. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (31–58). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2002a. Clarifying and applying the notions of metaphor and metonymy within Cognitive Linguistics: An update. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (207–277). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2002b. On the ubiquity and multiple-level operation of metonymy. In Lewandowska-Tomaszczyk, Barbara and Kamila Turewicz (eds.), Cognitive Linguistics Today [Lódz Studies in Language] (207–224). Frankfurt/Main: Peter Lang. Barcelona, Antonio. 2003a. Metonymy in cognitive linguistics. An analysis and a few modest proposals. In Cuyckens, Hubert, Klaus-Uwe Panther, and Thomas Berg (eds.), Motivation in Language: Studies in Honor of Günter Radden (223–255). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2003b. The case for a metonymic basis of pragmatic inferencing: Evidence from jokes and funny anecdotes. In Panther, Klaus-Uwe and Linda Thornburg (eds.), Metonymy and Pragmatic Inferencing [Pragmatics and Beyond New Series] (81–102). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2004. Metonymy behind grammar: The motivation of the seemingly “irregular” grammatical behavior of English paragon names. In Radden, Günter and Klaus-Uwe Panther (eds.), Studies in Linguistic Motivation (357–374). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2005. The multilevel operation of metonymy in grammar and discourse, with particular attention to metonymic chains. In Ruiz de Mendoza, Francisco and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (313–352). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2007. The role of metonymy in meaning construction at discourse level: A case study. In Radden, Günter, Klaus-Michael Köpcke, Thomas Berg, and Peter ­Siemund (eds.), Aspects of Meaning Construction (51–75). Amsterdam & Philadelphia: John ­Benjamins. Barcelona, Antonio. 2009. Motivation of construction meaning and form. The roles of metonymy and inference. In Panther, Klaus-Uwe, Linda Thornburg, and Antonio Barcelona (eds.), Metonymy and Metaphor in Grammar (363–401). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. In preparation. On the Pervasive Role of Metonymy in Constructional Meaning and Form and in Discourse Comprehension: A Corpus-Based Study from a Cognitive-Linguistic Perspective. (Provisional title.) Barcelona, Antonio. n.d. The difference between metaphor and metonymy: A question of asymmetry? (Unpublished manuscript.) Paper presented at the Fourth Conference on Researching and Applying Metaphor. Tunis (Tunisia), 5–7 April 2001. Barnden, John. 2010. Metaphor and metonymy: Making their connections more slippery. Cognitive Linguistics 21(1): 1–34. Bartsch, Renate. 2002. Generating polysemy: Metaphor and Metonymy. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (49–74). Berlin & New York: Mouton de Gruyter. Benbow, T. J. (ed.). 2002. Oxford English Dictionary. Oxford: Oxford University Press. CDROM edition of the Second Edition (1989) and of the Additions Series (1993–1997).



Properties and prototype structure of metonymy

Bultinck, Bert. 1998. Metaphors we die by: Conceptualizations of death in English and their implications for the theory of metaphor [APIL Paper no. 94]. Antwerp: University of Antwerp, Department of Linguistics. Croft, William. 2002/1993. The role of domains in the interpretation of metaphors and metonymies. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (161–205). Berlin & New York: Mouton de Gruyter. (Reproduced with slight changes from the paper with the same title in Cognitive Linguistics 4(4): 335–371.) Croft, William. 2006. On explaining metonymy: Comment on Peirsman and Geeraerts, “Metonymy as a prototypical category”. Cognitive Linguistics 17(3): 317–326. Croft, William and David Alan Cruse. 2004. Cognitive Linguistics. New York: Cambridge University Press. Cruse, David Alan. 1986. Lexical Semantics. Cambridge: Cambridge University Press. Cruse, David Alan. 2000. Meaning in Language. Oxford: Oxford University Press. Dirven, René. 2002. Metonymy and metaphor: Different strategies of conceptualisation. In ­Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (75–112). Berlin & New York: Mouton de Gruyter. Fauconnier, Gilles. 1994. Mental Spaces: Aspects of Meaning Construction in Natural Language. Cambridge: Cambridge University Press. Fauconnier, Gilles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Fauconnier, Gilles. n.d. How compression gives rise to metaphor and metonymy. A paper presented at the 9th Conference on Conceptual Structure, Discourse, and Language (CSDL9) held at Case Western Reserve University, Cleveland, Ohio, October 18–20, 2008. Fauconnier, Gilles and Mark Turner. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books. Feyaerts, Kurt. 1999. Metonymic hierarchies. The conceptualization of stupidity in German idiomatic expressions. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (309–332). Amsterdam & Philadelphia: John Benjamins. Feyaerts, Kurt. 2000. Refining the inheritance hypothesis: interaction between metaphoric and metonymic hierarchies. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (59–78). Berlin & New York: Mouton de Gruyter. Fillmore, Charles. 1985. Frames and the semantics of understanding. Quaderni di Semantica 6: 222–254. Gibbs, Raymond W., Jr. 1994. The Poetics of Mind. Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Goossens, Louis. 2002/1990. Metaphtonymy: The Interaction of metaphor and metonymy in expressions for linguistic action. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (349–377). Berlin & New York: Mouton de ­Gruyter. (Reproduced with slight changes from the paper with the same title in Cognitive Linguistics 1(3): 323–340). Goossens, Louis, Paul Pauwels, Brygida Rudzka-Ostyn, Anne-Marie Simon-Vanderbergen, and Johan Vanparys. 1995. By Word of Mouth: Metaphor, Metonymy and Linguistic Action in a Cognitive Perspective. Amsterdam & Philadelphia: John Benjamins. Jakobson, Roman. 1971/1956. The metaphoric and metonymic poles. In Jakobson, Roman and Morris Halle (eds.), Fundamentals of Language, Vol. 1 (90–96). The Hague & Paris: Mouton. Kövecses, Zoltán. 1990. Emotion Concepts. New York & Berlin: Springer Verlag.

55

56

Antonio Barcelona

Kövecses, Zoltán. 2000. Metaphor and Emotion: Language, Culture and Body in Human Feeling. Cambridge: Cambridge University Press. Kövecses, Zoltán. 2002. Metaphor: A Practical Introduction. New York: Oxford University Press. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9(1): 37–77. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: University of Chicago Press. Lakoff, George. 1993. The contemporary theory of metaphor. In Ortony, Andrew (ed.), Metaphor and Thought (2nd edition) (202–251). Cambridge: Cambridge University Press. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago: University of Chicago Press. Lakoff, George and Mark Turner. 1989. More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: The University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Vol. 2: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 1993. Reference-point constructions. Cognitive Linguistics 4: 1–38. Langacker, Ronald W. 1999. Grammar and conceptualization. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2009. Metonymic grammar. In Panther, Klaus-Uwe, Linda Thornburg, and Antonio Barcelona (eds.), Metonymy and Metaphor in Grammar (45–71). Berlin & New York: Mouton de Gruyter. McKechnie, J. L. (ed.). 1978. Webster’s New Twentieth Century Dictionary of the English Language. Second edition. n.p.: Collins and World. Panther, Klaus-Uwe and Linda L. Thornburg. 1998. A cognitive approach to inferencing in conversation. Journal of Pragmatics 30: 755–769. Panther, Klaus-Uwe and Linda L. Thornburg. 1999. The potentiality for actuality metonymy in English and Hungarian. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (333–357). Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg (eds.). 2003a. Metonymy and Pragmatic Inferencing [Pragmatics and Beyond New Series]. Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg. 2003b. Introduction: On the nature of conceptual metonymy. In Panther, Klaus-Uwe and Linda L. Thornburg (eds.), Metonymy and Pragmatic Inferencing [Pragmatics and Beyond New Series] (1–20). Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg. 2007. Metonymy. In Geeraerts, Dirk, René Dirven, and ­Hubert Cuyckens (eds.), Handbook of Cognitive Linguistics (236–263). Oxford: Oxford University Press. Paradis, Carita. 2004. Where does metonymy stop? Senses, facets and active zones. Metaphor and Symbol 19(4): 245–264. Peirsman, Yves and Dirk Geeraerts. 2006a. Metonymy as a prototypical category. Cognitive Linguistics 17(3): 269–316. Peirsman, Yves and Dirk Geeraerts. 2006b. Don’t let metonymy be misunderstood: An answer to Croft. Cognitive Linguistics 17(3): 327–336.



Properties and prototype structure of metonymy

Radden, Günter. 2002. How metonymic are metaphors? In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (407–434). Berlin & New York: Mouton de Gruyter. Radden, Günter. 2005. The ubiquity of metonymy. In Otal Campo, José Luis, Ignasi Navarro i Ferrando, and Begońa Bellés Fortuño (eds.), Cognitive and Discourse Approaches to Metaphor and Metonymy (11–28). Castelló (Spain): Universitat Jaume I. Radden, Günter. 2009. Generic reference in English: A metonymic and conceptual blending analysis. In Panther, Klaus-Uwe, Linda Thornburg, and Antonio Barcelona (eds.), Metonymy and Metaphor in Grammar (199–228). Berlin & New York: Mouton de Gruyter. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought (17–59). Amsterdam & Philadelphia: John Benjamins Radden, Günter and René Dirven. 2007. Cognitive English Grammar. Amsterdam & Philadelphia: John Benjamins. Riemer, Nick. 2002. When is a metonymy no longer a metonymy? In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (379–406). Berlin & New York: Mouton de Gruyter. Rosch, Eleanor. 1983. Prototype classification and logical classification: The two systems. In Scholnik, Ellin K. (ed.), New Trends in Cognitive Representation: Challenges to Piaget’s ­Theory (73–86). Hillsdale: Lawrence Erlbaum. Ruiz de Mendoza, Francisco. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (109–132). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José and José Luis Otal Campo. 2002. Metonymy, Grammar and Communication [Colección Estudios de Lengua Inglesa 7]. Albolote (Granada, Spain): Comares. Soriano, Cristina. n.d. The conceptualization of anger in English and Spanish: A cognitive approach. PhD thesis. University of Murcia, Spain. Taylor, John. 1995/1989. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press. Thornburg, Linda and Klaus Panther. 1997. Speech act metonymies. In Liebert, Wolf-Andrea, Gisela Redeker, and Linda Waugh (eds.), Discourse and perspectives in cognitive linguistics [Current Issues in Linguistic Theory 151] (201–219). Amsterdam and Philadelphia: John Benjamins. Warren, Beatrice. 2002. An alternative account of the interpretation of referential metonymy and metaphor. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (113–130). Berlin & New York: Mouton de Gruyter.

57

part i

Metonymy and related cognitive, semantic, and rhetorical phenomena

Metonymization A key mechanism in semantic change* Carita Paradis Lund University

Despite a sizable functional literature on variation and change, very little attention has been given to the diachronic mechanisms of meaning shifts and change covering the whole continuum from contentful to configurational meanings. This chapter is an attempt to begin filling this gap. It proposes a unitary approach to the mechanisms of variation and change within the Cognitive Linguistics framework. Two types of construals are considered to be essential for the differences between change and shifts, namely, metonymization and zone activation, both of which are part–whole configurations that profile aspects that are contextually motivated in a given situation in human communication. Metonymization is a construal that operates between senses, while activation of zones operates within senses. Keywords: configuration, construal, part–whole, profile, sense, zone activation

1.

Introduction

Language is most essentially the opposite of mathematics, which exclusively deals with the relations of concepts to each other without consideration of their relation to experience. Language deals with concepts that are strongly related to experience. The “use potential” of a lexical item is a conceptual structure that has been built up and is being built up by its different uses in different situations. In all usage events, only a portion of the total use potential of a lexical item is evoked.

* My thanks to Lena Ekberg, Dylan Glynn, Elizabeth Traugott and the editors of this book for generous comments and discussions on an earlier draft of this chapter. I would also like to extend my thanks to Magnus Levin and Hans Lindquist for reading the last version of the manuscript.

62

Carita Paradis

It is on the occurrence of use in human communication that the more specific focus of attention and the profiling of the meaning of a lexical item in context are being fixed. This means that meanings of words in context are pragmatically motivated and formed by construals operating on their use potential. Two types of construals, which are frequently referred to as metonymy in the literature, are distinguished as metonymization and zone activation (Paradis, 2004/2010). These construals are both based on part–whole / whole–part configurations and select the most salient aspects of meaning of a conceptual structure on the occurrence of use. However, they differ with respect to conventionalization of the profiled meaning. Metonymization holds between senses and activation of zones within senses. This chapter argues that metonymization is instrumental in semantic change, both (1) in sense developments from one contentful meaning to another contentful meaning, e.g., launder (clothes and money), mouse (animal and computer device), or head (body part and leader); and in (2) grammaticalization and pragmatization (subjectification and intersubjectification), e.g., be going to in its spatial movement sense and as a marker of future time in be going to and gonna, in (inter)subjectification, e.g., you know, like and quite! as pragmatic markers in contemporary English. The particular focus of this chapter is on the mechanisms of change rather than on the motivations for change, that is, how, rather than why, new meanings develop, and the claim is that metonymy is key. The treatment of semantic variation and change is couched within Lexical Meaning as Ontologies and Construals (LOC for short, Paradis, 2005) under the Cognitive Semantics umbrella (Talmy, 2000; Croft and Cruse, 2004; Geeraerts and Cuyckens, 2007). While there is no dearth of research on change and directions of change motivated by different kinds of communicative and pragmatic factors within the functional school of thought (e.g., Hopper and Traugott, 1993; Traugott and Dasher, 2005; Fisher, 2008) and on change causing reanalysis of the paradigm settings in the syntactic derivation in the generative school of thought (e.g., Roberts and Roussou, 2003; Lightfoot, 2006), very little attention has been given to mechanisms of meaning-making and change covering the whole continuum from contentful through grammatical to pragmatic meanings. Mechanisms in the present treatment relate to the actual processes that underlie meaning-making in language use in general – processes that may or may not result in change. In her very final words on future directions in diachronic linguistics, Bybee (2007: 981) calls our attention to this imbalance. “Clearly, reference to cognitive factors brings us closer to explanation in both the diachronic and synchronic realms. In diachrony, it is of utmost importance to emphasize not just the motivation for change, but also the mechanisms; that is, in order to establish why changes occur in a certain direction, we also have to understand how changes occur.” (Bybee, 2007: 981).



Metonymization: A key mechanism in semantic change

Metonymization, as it is defined above, involves the use of a lexical item to evoke the sense of something that is not conventionally linked to that particular lexical item (Paradis, 2004/2010). It is affected “on-line” and is an implied, contingent relation that precedes change. In novel uses of form–meaning pairings, couplings between lexical items and their meanings are not conventionalized. Change involves intersubjective entrenchment and conventionalization of metonymical readings and has taken place when form–meaning pairings have been established for certain uses, and focus of attention is again selected through zone activation. In the process of meaning change, there is a continuum from metonymy to zone activation, that is, from non-conventionalized couplings between form–meaning pairs (polysemy) to conventionalized form–meaning pairings and zone activation within senses (Paradis, 2008). Conventionalization is a socio-cognitive phenomenon that requires successful hearer recognition and subsequent acceptance of the speech community. This study centers on metonymization as a mechanism. Needless to say, there is also a motivational side to metonymization, which involves semantic reanalysis and has to do with communicative economy, flexibility in speaker-hearer negotiation and the desire to express oneself at the adequate level on the scale of clarity and specificity. The chapter is structured as follows. Section 2 gives a short description of the foundational Cognitive Linguistics assumptions and of LOC, the model of analysis. Section 3 deals with the notions of metonymization and zone activation in previous treatments and in the present model. In Section 4, I proceed to discuss the relevance of metonymization and zone activation in variation and change. Section 5 gives a more detailed description of the mechanism of metonymization in the development from contentful to new contentful meanings, and in Section 6 from primarily contentful meanings to new primarily configurational meanings. . The notion of polysemy in my model of meaning does not depend on the notion of conventionalization as it does in most treatments. Polysemies may be conventionalized uses or ad hoc uses. This has important theoretical consequences for the aspect of change in that change presupposes conventionalization of polysemies, while non-conventionalized polysemies represent meaning-making without consequences for the speech community. This will become clear as the arguments are being expounded on in the subsequent sections. This view is akin to IINs (Invited Inferences) and GIINs (Generalized Invited Inferences) in Traugott (1999). Intersubjective entrenchment is a term suggested to me by Dylan Glynn. . To avoid confusion, it deserves to be pointed out that reanalysis and analogy are often treated as mechanisms in the literature (e.g., Hopper and Traugott, 1993: 88; Eckardt, 2006). In contrast to those treatments, I see reanalysis and analogy as motivations for mechanism such as metonymization and metaphorization. Reanalysis and metonymization are considered primary and crucial in change. Another mechanism that has been suggested more recently is blending as a mechanism in a diachronic study of compounding (Benczes, 2011).

63

64 Carita Paradis

Finally, Section 7 offers a conclusion, and thereby my contribution to the study of metonymization within Cognitive Linguistics and the enterprise of this volume to reach a consensus on the delimitation of the notion.

2.

Lexical Meaning as Ontologies and Construals (LOC)

The common denominator in cognitive approaches to meanings of linguistic expressions is that they are perspectival and profiled according to a “base” or a “domain” (Langacker, 1987; Benczes, this volume), a “frame” (Fillmore, 1982), or an “idealized cognitive model” of a situation (Lakoff, 1987). All these constructs represent presupposed knowledge that is activated, implicated and inferred in communicative situations. “[W]hat holds together the diverse forms of Cognitive Linguistics is the belief that linguistic knowledge involves not just knowledge of the language, but knowledge of our experience of the world as mediated by the language” (Geeraerts and Cuyckens, 2007: 7). Being a model within the cognitive framework, LOC adheres to the view of the importance of frames/domains as fundamental constructs in linguistic communication. One major theoretical implication of LOC is that it forces a radical perspective on some of the current claims about the nature of meaning in language. Meanings of linguistic expressions are emergent and get their final interpretation in context. There is no principled difference between creative form–meaning pairings and conventionalized pairings from the point of view of how they are construed. Following Cruse (2002), LOC states that lexical meanings are constrained by encyclopedic knowledge, conventionalized mappings between lexical items and concepts, and conventional modes of thought in different contexts and situational frames. Contextual variation is natural and expected in a dynamic usage-based model such as LOC (Paradis, 2005). Cruse (2002) describes the total meaning of a linguistic element as a pattern of readings in conceptual space. He introduces a spatial metaphor that describes readings as bounded regions in conceptual space. According to him, readings tend to cluster in groups and as such they show different degrees of salience and cohesiveness. Between the groups, there are regions that are relatively sparsely inhabited. They are sense boundaries. This take on “sense” and polysemy as a function of distance and boundaries in conceptual space is consistent with the approach in LOC. These boundaries are important for the modeling of the development of new polysemies through meaning change in being the conceptual spaces separating different senses. Such modeling also means that the notion of a sense boundary and boundaries between readings within a sense are closely related to the degree of autonomy of the clusters that the boundaries delimit. Senses exhibit strong signs of autonomy and they are kept



Metonymization: A key mechanism in semantic change

apart by substantial boundaries, while readings within a sense are only weakly autonomous or not autonomous at all. It is the symptoms of autonomy that may be highlighted through various definitional tests that provide the evidence for boundaries (Croft and Cruse, 2004: 109–140, and for the continuum between metonymization through facetization to zone activation in Paradis, 2004/2010). In LOC, conceptual structures are modeled as two types: content ontologies and configurational ontologies, as shown in Table 1. These two types of structure are on a continuum from primarily contentful to primarily configurational; for that reason the line between the two columns is not solid. Ontologies are not set lexical meanings but pre-meaning structures (or purport, to use Cruse’s term, Croft and Cruse, 2004: 100–101) that contribute to the final interpretations of lexical items in context. They are not themselves fully-fledged discourse meanings, but are created at lower levels of conceptual organization on the way to discourse meaning construals. Table 1.  Ontologies and cognitive processes in meaning construction, adapted from Paradis (2005) Ontologies (conceptual structures)

Construals

Contentful

Configurations

– concrete phenomena – events, processes, states – abstract phenomena

part/whole, thing, relation, boundedness, scale, degree, point, frequency, focus, path, order, modality, …

Gestalt: e.g., structural schematization, profiling Salience: e.g., metonymization, generalization, zone activation Comparison: e.g., metaphorization, categorization Perspective: e.g., foregrounding / backgrounding, subjectification

Table 1 divides into Ontologies and Construals. The leftmost column of Table 1 gives the three most general content structures (1) concrete spatial phenomena; (2) spatio–temporal phenomena (events, processes and states); and completely (3) abstract phenomena. These top ontologies, in turn, subcategorize into more fine-grained ontologies, such as concrete phenomena, e.g., tree, woman, stone; event structures, such as walk, die, sad; and abstract phenomena, such as idea, problem, economics (for more details on the model, see Paradis, 2005). The second column under Ontologies shows various examples of configurational constructs or schemas. The configurations are not listed in any particular order, that is, I make no general claims about whether the configurational ontologies are hierarchically organized or not. Configurations are free ontologies in the sense that they are applicable to many different content structures, not in a one-to-one fashion but in a one-to-many fashion. For instance,

65

66 Carita Paradis

configurations such as degree may be associated with all of the three most general types of content ontologies: in much petrol the configuration degree (scale) is in co-operation with the contentful concrete meaning structure substance. In he is running around a lot degree (scale) shapes an activity, that is, a ­spatiotemporal meaning structure. In totally closed, degree (bounded) combines with state. Finally, in the ultimate idea, degree (bounded) is in co-operation with an abstract phenomenon. What is traditionally referred to as encyclopedic knowledge is taken to be represented by contentful meaning structures and linguistic (grammatical) knowledge by configurations. The matching of configurational structures to content structures is motivated by how we perceive of the world in a given situation. Configurational (structural) meanings take precedence over the interpretation of the contentful meanings they modify and determine the final conceptual structure of the whole – not the other way round. For instance, the most common interpretation of car is bounded. However, in combination with a lot of, our understanding of car will shift to unbounded, that is, degree and scale. It is not the case that car will force a non-gradable reading onto a lot of. This does not mean that car is polysemous between the two readings, but rather that its meaning is construed in two different ways. The Construals, given in the third column in Table 1, represent a totally different substance from Ontologies. They are the cognitive processes that operate on the ontological representations in conceptual space and are responsible for the final interpretation of linguistic meaning in context. A system of construals was set up by Croft and Wood (2000: 55–56) to match the construals presented in the cognitive linguistics literature with the cognitive processes from psychology and phenomenology. The Construal operations are special cases of four general cognitive processes, namely (1) Gestalt (constitution), (2) salience (focus of

. Talmy (2000: 24–40), too, distinguishes between two types of conceptual structure, the contentful subtype and the schematic subtype. Open-class meanings represent the former and closed-class meanings the latter. Closed-class meanings are constrained by various neutralities, e.g., bulk neutrality (abstracted away from the bulk of bodies in space and reduced to points, lines and the like), magnitude neutrality, shape neutrality, token neutrality and substance neutrality. . Construals have been described in the cognitive literature by Talmy (2000) in terms of schematic systems, which embrace configurational structure, deployment of perspectives, distribution of attention and force dynamics. Langacker (1987: 99–146, 2000: 3–5) deals with construals under the rubrics of comparison, attention and focal adjustments. The focal adjustments are further subdivided into selection of the facets of a particular scene, the perspective from which a scene is viewed and the level of abstraction or level of specificity. Lakoff and Johnson (1980) treat construals under metaphor.



Metonymization: A key mechanism in semantic change

a­ ttention), (3) comparison (judgment), and (4) perspective (situatedness). The system of Construals in Table 1 is most essentially the same as the system in Croft and Wood (2000), but not identical. The Construals form the dynamic component of LOC, important for our interpretation of different readings of all kinds of linguistic expressions in use. The four classes represent four distinct processes in different realms of experience, which in turn subsume different construal operations as exemplified in Table 1. The four processes are not mutually exclusive, but co-occurring and co-active. The construal, central to the argument of this chapter, is salience, that is, metonymization and zone activation that operate on the ontological pre-meanings. The conceptual similarities and the linguistic differences between metonymization and zone activation are dealt with in Section 3.

3.

Metonymization and zone activation

On the whole, metonymy has received relatively little attention in the synchronic literature, but has recently experienced an upsurge of interest as this volume is evidence of. It has been discussed both in its own right and in relation to metaphor, e.g., Warren (1992, 2003, 2006), Gibbs (1994, 1999), Papafragou (1996), Kövecses and Radden (1998), Panther and Radden (1999), Radden and Kövecses (1999), Dirven and Pörings (2002), Goosens (2002), Barcelona (2003), Panther and Thornburg (2003, 2007, 2009), Ruiz de Mendoza (2003), Paradis (2004/2010, 2008), Haser (2005) and Peirsman and Geeraerts (2006), and in relation to vertical polysemy, i.e., categorical shifts of broadening and narrowing (Koskela, this volume). Metonymization is not restricted to linguistic communication but is also a construal in other communicative modes such as pointing and gesture. For instance, a 3-year old child once visited our house. Pointing to the sofa in the lounge he exclaimed: “Dad?!” Since no man was sitting on the sofa, he concluded that there was no adult male family member. Pointing gestures are precursors of linguistic communication and metonymization is common (Goldin-Medow, 2007; Tomasello, Carpenter and Liszkowski, 2007; Gärdenfors and Warglien, forthcoming). However, the clue to how we as language users cope with contextual flexibility still eludes linguists to a large extent, and contextual readings such as metonymies create problems in computational linguistics where a great deal of effort is currently directed to finding out how such uses can be predicted, e.g., Pustejovsky (1995), Lapata and Lascarides (2003). Langacker (2000: 199) offers a unitary treatment of metonymy as a referencepoint and an activation phenomenon in that “the entity that is normally designated by a metonymic expression serves as a reference point affording mental access to the desired target, i.e., the entity actually being referred to”. The scope of such

67

68 Carita Paradis

a definition is too unconstrained for a more detailed lexico-semantic analysis. In contrast to Langacker, Peirsman and Geeraerts (2006) suggest a prototypically structured non-unitary definition of metonymy as conceptual contiguity. They take spatial part–whole relations as the core of the category of metonymy along three interacting indicators: (1) strength of contact, that is, going from part–whole over physical contact to adjacency; (2) boundedness, that is, from bounded to unbounded wholes and parts; and (3) from spatial and material domains through spatio-temporal domains such as events, actions and processes and into the domain of assemblies and collections. The definition of metonymy in terms of what they refer to as a “logical” prototype structure is an attractive idea that lends itself to hypothesis formulation for empirical studies of psychological prototypicality. It offers tools for further empirical investigation of the meaning structures but no constraints on the linguistic side of the matter. Like Peirsman and Geeraerts (2006), Paradis (2004/2010) proposes an approach to metonymization as a continuum in the conceptual realm, but her ­approach to metonymization on the lexical side of the matter differs from theirs. Paradis (2004/2010) presents three examples: (1) The red shirts won the match, (2) The court had to assume that the statement of claim was true and (3) I have a really slow car. All three are reference-point and salience phenomena at the conceptual level. Salience, as she uses the term, refers to the prominence and degree of activation of certain conceptual structures relative to other possible parts in the cognitive network. However, they differ with respect to the degree of conventionalization of the profiling of the lexico-semantic couplings. (1) is an example of metonymization proper, (2) of facetization and (3) of zone activation. Facetization is a squish category that concerns form–meaning pairing occupying the middle range of the continuum between metonymization and zone activation (Paradis, 2004/2010). As stated in the introduction, this study is restricted to metonymization and zone activation. The status and justification for facetization in between those two poles is also discussed in Barcelona (this volume) and Geeraerts and Peirsman (this volume).

. Geeraerts and Peirsman (this volume) have developed their approach to metonymization and zone activation further by appealing to “reference”, which they seem to use by and large in the same way as I use conventionalization of profiled meanings that are intersubjectively entrenched. I deliberately avoid the term because of the theoretical burden it carries from objectivist and structural approaches to meaning. I prefer “conventionalization” to reference in particular when constructions impose strong restrictions on readings such as in the case of middle constructions. For instance, in This wine should drink well for 10-20 years there are no selection restriction violations because the construction itself demands the set up of participants (Yoshimura and Taylor, 2004; Paradis, 2009a, 2009b).



Metonymization: A key mechanism in semantic change

The argument behind the distinction between metonymization and zone activation is operationalized through conventionalization and runs as follows. The metonymical expression red shirts in (1) denotes red shirts, which refers to a concrete entity that is linked to the intended referents in their capacity of being people (‘football players’). The functional role of ‘people as players’ is highlighted through their red shirts. In metonymization in general, as in (1), one of the concepts, red shirt here, is lexically encoded and thereby foregrounded and focalized. Red shirts has the role of providing access to the inferred concept player, which is being profiled. It is player, not shirt, that is activated in anaphora resolution, e.g., *One of the red shirts came in from the left. It ran towards the goal, but One of the red shirts came in from the left. He/She ran towards the goal (cf. Ruíz de Mendoza, 2003 and this volume). Out of context, “player” and “shirt” represent two different concepts/senses, but in (1) they are used to refer to the same entity by means of a conventional mode of thought triggered by a search for contextual relevance. Metonymization is licensed by pragmatic inferencing (Panther and Thornburg, 2003). Additional examples of metonymizations are the referee pulled out the yellow card (Radden, this volume), How did you get to the airport? I waved down a taxi (Gibbs, 1994: 327), She married money (Warren, 2006: 18), a topless district (Gärdenfors, 2000: 117), or the final whistle for ‘the game is over’ (Levin, 2008: 145), in which pull out the yellow card, wave down a taxi, marry money, topless district and final whistle are all part/whole focalizing shortcuts. The example of zone activation in (3) is similar to metonymization in being a salience phenomenon and a part/whole construal, but it differs in that the focalized reference point is conventional and for that reason does not require much pragmatic inferencing on the part of the addressee (Paradis, 2004/2010; Koskela, this volume). Zone activation is a much more pervasive construal than metonymization proper and concerns all readings on all occurrence of use. It is always the case that only a certain portion of the use potential of lexical items is in focus in linguistic communication. Zone activation happens within senses (in monosemy) and does not involve different senses. For instance, in slow car, the function role of “performance” is the relevant aspect, but that does not give rise to a situation in which we are dealing with two different senses of car. Car is the conventional lexical item for both ‘car as artefact’ and ‘car as performance’. In a similar way, the difference between tall man, lazy man and real man is that the adjectives conventionally activate different zones of the meaning structure of man, that is, physical (tall), functional (lazy) and personality characteristics (real) respectively. It is thus crucial for the argument of this chapter that metonymization proper is a polysemy phenomenon and concerns different senses, where one of the senses is conventionally associated with the lexical item used, whereas the other sense is inferred.

69

70 Carita Paradis

4.

Metonymization and change

The notion of metonymy is by all means not a new notion in the context of variation and change in the historical linguistics literature. On the contrary, it is a commonplace among other notions in both prestructuralist and structuralist work. Nerlich and Clarke (2001: 245) note that “Metonymy has been studied for at least two thousand years by rhetoricians, for two hundred years by historical semanticists, and for about ten years by cognitive linguists.” Various classification schemes for types of change have been suggested in the literature. For instance, Paul (1920), Stern (1931), Bloomfield (1933) and Ullmann (1942) have suggested a number of different types of change such as narrowing, widening, metaphor, metonymy, synecdoche, hyperbole, litotes, degeneration, elevation. What most of these scholars aimed to do was to provide descriptions of chains of successive meanings of lexemes and for that purpose they collected them in classes. However, the problem with many of these classification schemes is that they are exactly that, without offering any strong theoretical explanations for change. The classes are like islands, and most classes can be included in one another. Philology and structuralism dominated the study of diachronic change up to the beginning of the 90s, and it was not until the emergence of Grammaticalization Theory (Traugott, 1982; Hopper and Traugott, 1993) and Cognitive Linguistics (Sweetser, 1990; Blank and Koch, 1999; Traugott and Dasher, 2005) that we have a usage-based theoretical basis and better tools for the formulation of hypotheses with self-consistently explanatory potential of dealing with meaning flexibility and paths of change. Following Traugott (1999), I see innovation as driven by and explainable in terms of a speaker-initiated process of strategic choices using general principles of abductive reasoning (i.e., observations in combination with known laws give rise to hypotheses). This type of reasoning hinges on semantic reanalysis and gives rise to implications of innovative uses that may or may not result in change. Conventionalization and change in turn presuppose acceptance of the speech community. I argue that change proceeds from non-conventionalized mapping between lexical items and their readings, construed through a particular focus of attention that is contextually motivated, i.e., change proceeds through metonymization. Change has taken place when a form–meaning pairing has been established for a certain use, and focus of attention is again selected through zone activation. Change is gradual and involves robust conceptual boundaries between the established sense and the new sense. Variation within senses is sensitive to . For a critique of different types of grammaticalization research in different linguistic traditions and frameworks during the 20th century, see Rosenkvist (2004).



Metonymization: A key mechanism in semantic change

constructional and contextual forces, but differences between different readings are gradient and involve no substantial boundaries in conceptual space. The idea of contiguities of meaning profiling (from metonymization through zone activation) relies on the interaction of domains and discourse frames as established in human communication (Benczes, this volume). The structuring component and the component that makes the definition of metonymy operational is linguistic conventionalization, i.e., intersubjective entrenchment. Change presupposes conventionalization of polysemy. The frames associated with particular foci of attention of meanings evoked by words and expressions in language always involve a complex conglomeration of related notions of contentful representations and configurations. There is very little in the literature on what types of meaning structures function as pivots for meaning change in developments in particular within contentful structures. Most treatments concern developments that develop into principally configurational meanings, that is, become grammaticalized and pragmaticized. In the next two sections I describe the mechanism of metonymization for meaning change from contentful meanings to contentful meaning (Section 5), and meaning change from contentful to configurational meaning (Section 6). Recall again the distinction made between contentful meaning structures and more schematic structures in Table 1). I show that metonymization proceeds across subnotions of thing and subevents and properties of relation, as well as shifts and change from thing to relation and from relation to thing, and change from such mainly contentful meanings into configurations through the foregrounding of, for instance, boundedness and scale, which bring us to changes referred to as grammaticalization and pragmatization.

. Traugott and Dasher (2005: 35) also see metonymization and salience as important factors in language change. It should however be noted that salience in their terminology is different from the way it is used here to refer to a cognitive process. In their treatment salience is associated with social value. Initially, people unconsciously adopt a certain meaning and exploit it rhetorically and only later does the change become subject to conscious evaluation. This view is not incompatible with the present view, but it puts the main focus on the interactional side of the process rather than the semiological side. . The question of whether semantic change is primarily a result of metaphorization or of metonymization has been the focus of much research interest. Clearly, metaphorization is an important factor in language change. The scope of this paper does not allow me to discuss the construals of metaphorization at all. Suffice it to say, that in its capacity of being a construal of salience, metonymization is a prerequisite for metaphorization. Reanalysis is the motivating factor for metonymization, while metaphorization is driven by comparison and analogy. Analogy presupposes reanalysis while the reverse is not the case.

71

72

Carita Paradis

5.

From contentful to contentful meanings

The semasiological development from contentful into contentful meanings with a mechanistic, metonymical focus is an area where a lot work remains to be done. There is very little in the literature on what types of meaning structures function as turning points for metonymizations and change in developments within contentful structures. This section discusses the semasiological development of the meanings of some lexical items in the directions from relation to thing and from thing to relation, using the lexical items crawl, crawler, key, draft and the development of “fear” expressions in Greek (fovame) and English (fear, be afraid). The lexical items are selected to illustrate various focal points of salience in their diachronic development beyond the fairly well described drift from meanings related to physical domains into meanings in the mental world. By doing this, I hope to open up new avenues of research on semantic change. It is likely that there are patternings that have gone missing in previous analyses due to the rather strong research focus on predictions of the directions of change from lexical to grammatical and pragmatical meanings and from physical to mental meanings. It may well be that there is more to the story of variation and change if more consistent corpus investigations across different ontological types were carried out, in particular with respect to change from contentful meanings into new contentful meanings (recall the distinction made between contentful meaning structures and more schematic structures in Table 1). Firstly, the first instance of the verb crawl in the OED is from 1300 with the definition “to move slowly in a prone position, by dragging the body along close to the ground, as a child upon its hands and knees, any short-limbed quadruped or reptile, an insect, serpent, worm or a slug”, as in (1). It also says that crawl probably originated from Norse (cf. Swedish kravla) and that it was a rare word, apparently only northern. Consider Examples (1) through (6):

(1) (2) (3) (4) (5) (6)

The aged Symeon cralls to kyrk.  A green mantling vine That crawls along the side of yon small hill.  Sicknesse posteth to us, but crawleth from us.  Cranmer Hath crawl’d into the fauour of the King.  You make me crawl all over, talkin’ so much about dyin’.  Mr Krushchev was crawling to Washington. 

(1460) (1634) (1654) (1613) (1891) (1966)

In its earliest use, as in (1), six important meaning structures are crucial for our understanding of crawl, namely actual motion in physical space, presupposing an agentive discourse participant, such as the aged Symeon, slow speed, low location and horizontal orientation. These substructures, which represent parts of the whole of the contentful meaning structures crawl construed as a



Metonymization: A key mechanism in semantic change

verbal relation as in (1), are also involved in the interpretations of the uses in (2) through (6), but at different degrees of salience, fostered in different constructions and, as we shall see, in some cases in terms of their opposites. In (2), crawl denotes concrete movement by a concrete entity, a plant, and mental scanning by the conceptualizing speaker. This means that there is no willful human agent involved in the growth of the plant (motion). Due to this difference, a zeugma is created when the two constructions are coordinated: ?Aged Symeon crawled along the path and so did the vine. This antagonism is evidence of sense difference between (1) and (2). The nature of the discourse participants is an important factor in meaning construal and in this case a sufficiently strong factor to create a zeugmatic effect in coordination. Furthermore, the expression in (3) provides a description of sickness as an entity that is capable of crawling, in which case the substructure of crawl has taken on another shape. We are in a semi-physical spatial region, where we are supposed to understand the disappearance of the sickness both as an experience in mental space and in terms of a metaphorical ‘crawling’ feeling in which case the sickness is portrayed as the causing agent. In (5) next, the expression you make me crawl all over is similar to (3) in being primarily experiential (undergoer) rather than agentive (actor). The situation described is quasi-emotion rather than motion, i.e., a sensory perception of a crawling feeling caused by an external agent, who may or may not be aware of the effect he or she has on the experiencer (undergoer). It deserves to be pointed out that the constructions as such in all cases contribute to shape the specific interpretations of the senses and the readings within the senses. The imagined feeling of a crawling creature takes us from motion to quasi-emotion in a very specific causative construction. The aspects of slow (speed), low (location) and horizontal (orientation) are all there but out of focus of attention. The interpretations in both (3) and (5) are strongly tied to the rather particular set-up of the constructions in terms of the personification of a state as the agentive discourse participant in (3) and the causative construction in (5). These constructional differences indicate that identity tests are not enough, but that other methods have to be used to shed light on formal and semantic differences of this sort (Gries, 2006; Glynn, 2009). In (4) and (6), the set of pre-meaning structures is the same as in (1) but they are definitely transferred from the physical world to the mental world in the creation of the new senses. Both expressions presuppose a human agent for the mental movement (motion), in contrast to emotion in (3) and (5). The step from physical space to mental space is a metaphorically motivated construal of comparison. With the agentive mental reading, as in (4) and (6), a clear goal emerges and becomes a crucial part of the conceptual set up, namely false subservience. The combination of slow (speed), low (location) and horizontal (orientation) in

73

74

Carita Paradis

mental (space) evoke negativity that comes with up as emblematic of superiority and down of inferiority and the slow manner of movement on the ground as implicating stealthy, treacherous and insidious motion pertaining to terror-inducing cultural stereotypes such as snakes. The uses of crawl in (1) through (6) above leave us with five reasonably robust sense boundaries created through substructure shifts, with no sufficient boundary between (4) and (6) At the one end we have the meanings with the human agent as in (1) developing into a similar motion pattern for non-human entities such as plants in (2), sickness in (3) and a self-propelled feeling in (5) to the mental movements carried out by human agents in (4) and (6). It deserves to be emphasized again that the constructions as such in all cases contribute to shape the specific interpretations of the senses. Next, crawl came into use as a thing construal in the 19th century for slow creeping motion as in (7), (8) and (9). By way of the thing construal configuration of those form–meaning pairings, they are sufficiently different in terms of conceptual distance from the relation ‘crawl’ in (1) through (6) to be different senses. Yet, the more contentful pre-meaning structures of these nominalizations are the same as in examples (1), i.e., (speed), low (location), horizontal (orientation), movement (motion) in physical space (space) by living creatures (participant): (7) I rather dislike the crawl of the centipede or slime of snail.  (1853) (8) R. Cavill, who revolutionised all ideas about speed swimming for short distances by introducing a further modification of his style, which was at once termed the ‘crawl’ stroke.  (1903) (9) We did a “pub-crawl” in Commercial Road and East India Dock Road. (1915)

With the coining of crawl for a new swim style (8), the same substructures of meaning as in (7) and (9) are represented, but ‘slow’ has been substituted with the opposite side of the of the scale of speed, namely ‘fast’, and the activity is necessarily taking place in water. More recently, the leisurely pace of the walk in gin-crawls, beer-crawls and pub-crawls (9) have become faster through motorized pub-crawls. The horizontal orientation of the participants in the crawl has been used in advertising with pictures of people on their hands and knees on their way to another glass. Also, jokes have been created on the basis of orientation, which is evidence in favor of its relative prominence. A travel agency made use of . speed in most of these uses may be modified by “slow” or “fast”, relating to its relative reference points. In other words, the reference frame for a crawling baby is slow, but within this setting the crawl may be slow or fast, and inversely, fast is the frame of reference for crawling in water, but again within that frame it may be performed slowly or fast.



Metonymization: A key mechanism in semantic change

an advertisement saying: Horizontal holidays: Pub crawls Glasgow.10 The uses of the noun crawl in (7), (8) and (9) represent three different senses that cannot be substituted for one another in truth-tests. We cannot felicitously say that Bill did the crawl and so did Alan, meaning both doing the swim, crawling on the ground or doing a pub-crawl. Also, in the development of the nomen agentis crawler from crawl as relation, the semasiological pattern repeats itself. Crawler is distinct from both crawl as relation and thing. The first instance of the use of crawler in the OED is from 1649 and pertains to crawling creatures such as people, reptiles and insects. Later on, specialized uses referring to humans and artefacts appear, such as cabs moving slowly along the streets in search for fares (1865), and overall garments for children (1891), swimmers doing the crawl-stroke (1912) and later on during the 20th century, vehicles such as crawler tractors. All of those developments make salient the particular functions of slow (speed), low (location), horizontal (orientation­) and movement (motion) in physical space (space) by more or less directly agentive discourse participants.11 In the mid 1800s, crawler, denoting somebody who acts in a servile and despicable way, developed in parallel with the physical meaning for the verb crawl, e.g., that scheming crawler (1856). In addition to these developments, there is a more recent use of crawler (1994) that developed in computing, i.e., a web crawler or a spider. A web crawler is a computer program that visits websites and collects information on the Internet. The focus of attention is on the aspects of speed and the goal of finding the information searched for. However, being the spider in the web, the web crawler is capable of embracing the web and in contrast to most of the other senses it moves fast rather than slowly all over. Thus, in the semasiological development of crawl and crawler, six substructures are crucial both for readings within one sense and for the development of new senses. In the case of crawl and crawler, they are motion, space, participant, speed, location and orientation. In addition to these structures, it has been shown that configuration construals such as either thing or relation also create robust sense boundaries. Also, note that participant becomes part and parcel of the thing construal. This is a fact that is taken for granted in most frameworks, but within a cognitive semantics framework such as LOC where meanings are contextual construals upon use, this becomes an issue and has to be taken seriously. There are clearly generalizations to be made from developments 10. http://www.cartoonstock.com/directory/g/glasgow.asp 11. The reader is referred to Panther and Thornburg (2002) for an account of various different uses of English -er nominals and the thematic roles highlighted by them, in the case of the agentive “crawlers” (e.g., reptiles) or the instrumental “crawlers” (e.g., clothes).

75

76

Carita Paradis

of meanings construed as thing or as relation and the importance of the functioning of the meaning in the world in a very broad sense. This also highlights the importance of the other side of the of the coin, the onomasiological side, that is, what kind of entities and events are named by these words, which is closely connected to the type of configuration construal in terms of thing or relation. The scope of this chapter does not allow me to say anything about that here. For work on the onomasiological side of the matter with reference to contentful meanings, see Geeraerts (1997). The functioning in the world is of crucial importance to a more recent development of draft and key as thing in physical space to their uses as property relations in the mental world, see (10) and (11).12 (10) I think my key point is going to be this: girls are not wired to do that kind of stuff… And an even keyer point is the definition of stuff. (11) This is really quite draft at the moment.

In (10) the telic qualia role of “key” is made salient (Paradis, 2005), and, by implication, crucial importance of the function of “key” emerges and receives the focus of attention. This shift from thing to relation also takes us to the mental world. Draft in (11) has gone through a similar development resulting in a meaning change. In a similar fashion to the change of crawl used as relation (verb) or thing (nominal), these two uses definitely create a robust enough boundary between the uses as different senses. Thirdly, in a diachronic study of “fear” predicates in Greek and English, ­Kitis (2009) outlines the parallel developmental paths through metonymization of ­ fovame (“fear”) in Greek and fear and be afraid in English from motion in physical space to emotion in mental space. Kitis shows that that the evolution of fovame in Modern Greek originates from a strongly agentive transitive verb meaning ‘put to flight’ in Homer’s Iliad in Ancient Greek. Homer also used the same verb in the passive ‘to be put to flight’ and in the middle ‘flee in terror’. In all three constructions there is a contingent association with a feeling of fear and terror that becomes central to and focalized through metonymization in its later emotion sense of being frightened only. Kitis shows how the fleeing sense developed into the fear sense through metonymization, i.e., conventionalization of the focus of attention on fear. The actual separation of ‘fleeing’ and ‘fright’ is nicely captured in an example where the motion and the emotion are rendered by two different lexical items (“having been frightened, the strangers fled”, Example (19) in Kitis, 2009). The definite change from motion in physical space to emotion originated in the middle construction where the affected object, i.e., the 12. The examples are borrowed from Denison (2007); italics are added by CP.



Metonymization: A key mechanism in semantic change

experiencer is the primary discourse participant and the consequent degree of transitivity ranges in the middle.13 As evidenced by both Kitis (2009) and Tissari (2007), the meaning of the “fear” predicate changed through metonymization from emotion into the interpersonal socio-cognitive epistemic sphere as a regulator of communication both in Greek and in English in expressions such as I fear it is too late now. The role of I am afraid is not to express fright, but rather to preface an unwelcome view. Morphosyntactically, this development is aided by the fact that the complementizer clause becomes upgraded to “main clause status” and the more flexible position of I fear, as in e.g., It is too late now I fear.14 The development of the emotion meaning of I fear (that) and I am afraid (that) into epistemic markers in intersubjective uses of I fear and I am afraid are accounted for in detail by Tissari (2007). Both Kitis and Tissari describe the pragmatization of emotional fear as a contingent implicated cause-and-effect relationship between fear and avoidance. In the final stage of the semasiological development of the “fear” meanings, we have crossed the border between primarily contentful meanings and primarily configurational ones. It is to the topic of the crossing of this line that I now turn.15

6.

From contentful to configurational meanings

There is an extensive literature dealing with the semasiological development of meanings of lexical items from contentful to configurational meanings (e.g.,

13. The flow of transfer of activity is described for English middles in Kemmer (1993) and Paradis (2009a). The ontological prerequisites for property ascription through zone activation in the middle construction are described by Yoshimura and Taylor (2004) and Paradis (2009b). The middle is accounted for in terms of an underlying high-level metonymic shift involving process for action for result by Ruiz de Mendoza and Mairal (2007: 46). 14. This pattern is found in the context of other verbs as well, I think (e.g., Aijmer, 2009; Boye and Harder, 2007, 2009) or the role of complement clauses for intersubjectivity in discourse (Verhagen, 2006). 15. In a similar fashion to the tracing of the development of ‘fear’ in English, Tissari (2008) investigates the development of respect (as a noun and as a verb) from the physical domain of visual attention through the mental attention to textual attention and again the regulation of communication. Also in the context of respect she points up the chain of metonymization as the vehicle that takes us from physical vision and cognition to emotion and socio-cognition and interpersonal relations in expressions such as with respect to and in respect of in text and discourse. See also Sweetser (1990) for development from vision and cognition.

77

78

Carita Paradis

Hopper and Traugott, 1993; Traugott and Dasher, 2005).16 A particularly popular area of research has been the development of various types of scope-taking adverbials from contentful meaning structures (e.g., Ungerer, 1988; Nevalainen and Rissanen, 2002; Traugott, 2006, 2007) and the development of expressions foregrounding degree configurations from various contentful sources (Peters, 1993; Paradis, 1997, 2000a, 2000b, 2008; Traugott and Dasher, 2005; Méndez Naya, 2008a17). In order to provide support for my claims concerning change from contentful to configurational meanings, I will make use of a couple of diachronic studies related to the domain of degree. In contemporary English there is a number of reinforcing adjectives that can be shown to have developed degree meanings through metonymization of boundedness and scale configurations at the expense of content proper, e.g., “absolute bliss”, “a complete bitch”, “a perfect idiot”, “total crap”. In combination with nominals such as bliss, bitch, idiot, crap, nonsense, pleasure, mess, coward, muddle and bore, the above adjectives are all interpreted as modifiers of degree (Paradis, 2000b). The development of the reinforcing senses of the adjectives is relatively recent in the history of the English language. Most of them developed this reading during the Early Modern period. Historically, complete, perfect, total and absolute all share a content dimension of ‘completeness’. Complete comes from ‘having all its parts’, ‘entire’, ‘full’, perfect from ‘completed’, ‘accomplished’, total from ‘relating to the whole of something’ and absolute was originally a participle meaning ‘disengaged from’ and ‘free from imperfection’, which later came to be used in the sense of ‘complete degree’ through implication and metonymization of its configuration. As degree modifiers complete, perfect, total and absolute all foreground the configuration of boundary.18 In contrast to the adjectivals expressing totality and boundedness, there is another set of adjectives, awful, dreadful, horrible and terrible, which exhibits the same kind of development from contentful to configurational. In contrast to the

16. Much work has also been carried out on meaning shifts and change from configurational to configurational, (e.g., Jucker, 1995; Brinton, 1996; Buchstaller and Traugott, 2006; Ekberg, 2007, forthcoming). While assuming that such developments also come about through metonymization, I will not discuss such changes at all due to space limitation. 17. Méndez Naya (2008a) is a Special Issue on English Intensifiers with contributions by González Diaz (2008), Lenker (2008), Méndez Naya (2008b), Nevalainen (2008), Paradis (2008), Rissanen (2008) and Tagliamonte (2008). 18. Note that although from slightly different contentful sources, the adjectives form a paradigm of degree modifiers with, by and large, the same function. This is also true of the adverbial degree modifiers and of other operators such as focusing adverbs (Nevalainen, 1991).



Metonymization: A key mechanism in semantic change

first set in which the member are associated with a boundary, the members of this set are all associated with a scale structure. Awful, dreadful, horrible and terrible are all expressive of content structures that are based at the extreme end of a scalar structure. They were originally associated with ‘awe/dread/horror/ terror-causing’, all of which highlight an extreme point on a scale of ‘content X’. For instance a terrible dragon is a dragon that causes terror. Like absolute, complete, perfect and total, they were also recruited to degree modification by implication in combination with meanings that were potentially scalable single property content structures with a relatively salient scale reading. That is, in combination with nominals such as mess, coward, muddle, and bore as in an awful mess, a dreadful coward, a horrible muddle and a terrible bore. In LOC, the process of grammaticalization translates into a development from content to configurational structures, not the other way round. None of the above adjectives in these combinations elaborate a property associated with the content structure of their nominal heads as such in combination with nominal heads that construe properties as thing. For instance, it is not the case that the bore causes terror. The function of terrible is to specify a range of the scale with which the meaning of “bore” is associated. Again, this is made possible through metonymization whereby the superlative content component of “terrible” is generalized into ‘high degree’. As a modifier of degree, terrible are neither felicitous in comparison nor in predicative use: *He is the more/most terrible bore or *The bore is terrible (referring to a person). The role of terrible is to highlight and reinforce the degree of ‘boredom’ in bore. The contentful side of terror has changed and the scalar template has become predominantly salient. In other words, ‘terror-causing’ is metonymically related to ‘high degree’ through the backgrounded scale configuration, which, in turn, gives rise to the degree reading and the foregrounding of the scale configuration of terrible in contexts such as a terrible bore or a terrible nonsense. Change took place when the ‘high degree’ reading of terrible became conventionalized in combinations such as terrible bore, and the construal became one of zone activation and not metonymization, i.e., the relationship had become conventionalized rather than construed on the fly to fit more ad hoc contextual needs.19

19. In addition, what is clear from all the examples of reinforcing adjectives is that they prefer indefinite constructions. This suggests that change is fostered in constructional straitjackets – an observation which is true of all developments from contentful to contentful meanings as well as from contentful to configurational meanings and from configurational to configurational meanings (Paradis, 2000b: 249; Heine, 2002; Traugott, 2003; Eckardt, 2006; Ekberg, forthcoming).

79

80 Carita Paradis

What all the reinforcing adjectives have in common is a conceptual substructure that lends itself to developing a degree sense. Their various contentful structures, in terms of ‘totality’ and ‘extremeness’, are such that they can be used for grading. Also, their meanings are all either bounded or scale configurations, albeit initially as backgrounded structures. The configurational schema underlying absolute, complete, perfect and total is a definitive boundary and the schema that structures awful, dreadful, horrible and terrible is scale. When the above adjectives were recruited as reinforcers, degree was made prominent and their main role was to modify the gradable property expressed by the noun with respect to degree. This reading gained speaker acceptance and change took place through metonymization involving two different senses in all cases. Their incompatibility in identity tests is a symptom of polysemy *The dragon was terrible and so was the bore. Once change has taken place and the use of these adjectivals has been conventionalized, the reading of an expression such as a terrible bore is a case of zone activation as is the case in the terrible dragon where the focus of attention is on the ‘terror’ content as such. The relationship between reinforcing adjectives and their nominal heads is comparable to the relationship between reinforcing adverbs and their adjectival head, e.g., absolutely pure, awfully messy, totally free, completely disastrous, really competent and terribly boring. The difference between awful mess and awfully messy is one of configuration construal of mess as thing, while the configuration of messy is one of relation (Paradis, 2005, 2008). Similar to adjectival degree modifiers, adverbial degree modifiers also developed their grading functions from content–foregrounding form–meaning couplings to couplings that foreground configuration. For instance, the boosters awfully, frightfully, terribly, horribly, really, dead, jolly and very have developed into degree operators from content foregrounding meanings and the same is true of the moderators and downtoners quite, rather, pretty, fairly and a bit, and the maximizers absolutely, quite, totally, perfectly, right and completely (e.g., Stoffel, 1901; Borst, 1902; Brugman, 1984; ­Peters, 1993; ­Paradis, 1997: 71–76, 2000a, 2000b, 2003, 2008; Nevalainen and Rissanen, 2002; Méndez Naya, 2003, 2007, 2008b; Traugott 2006, 2008a). It has been shown again that focus of attention is operative in various meaning construals, and also that constructions play an important role in fostering change and probably also in creating the basis for analogies such as for the various degree modifier paradigms such as awfully, horribly, frightfully, terribly, and for modifiers such as a bit of, a piece of, a shred of (Traugott, 2008b). What is characteristic of the path of development of degree meanings is the conceptualization of the message as increasingly strongly grounded in the situation and thereby more (inter)subjective (Traugott, 1995; Traugott and Dasher, 2005; ­Ekberg, forthcoming).



7.

Metonymization: A key mechanism in semantic change

Conclusion

This chapter offers a unitary approach to the mechanism of variation and change in language within Lexical Meaning as Ontologies and Construals (LOC). The basic assumption that the model rests on is that meanings of words are emergent and formed by construals operating on their use potential on all occasions of use. There are no stable inherent word meanings. This chapter has centered on two types of construals that are crucial for explanations of the difference between change and shifts, namely metonymization and zone activation. These construals are similar in that they are both based on part–whole configurations that profile salient aspects in a given context. However, they differ with respect to the conventionalization of the profiled meanings. Metonymization is a construal that operates between senses, while activation of zones operates within senses. I have argued that metonymization is instrumental in the development of all new meanings in language change irrespective of whether they involve developments from primarily contentful into primarily configurational meaning types, or whether they concern developments of form–meaning pairings within each of these two types of structure. The framework of LOC predicts semantic change from primarily contentful to primarily contentful meanings, from primarily contentful meanings into primarily configurational meanings and from primarily configurational to primarily configurational meanings. In this framework, grammaticalization is the latter developments. Metonymization involves the use of a lexical item to evoke the sense of something that is not conventionally linked to that particular lexical item. It is an implied relation that precedes change and involves novel uses of form–meaning pairings, in which case the connection ­between lexical items and their meanings has not yet been conventionalized. Conventionalization and change require socio-cognitive­ entrenchment of the metonymical readings and acceptance within the speech ­community, and change has taken place when conventional form–meaning pairings have been established for certain uses and focus of attention is again selected through zone activation. In the process of meaning change, there is a continuum from metonymy to zone activation, i.e., from non-conventionalized couplings between form–meaning­ pairs (polysemy) to conventionalized form–meaning pairings and zone activation within senses. While variation in monosemy is fluid and gradient, change presupposes polysemy, occurs by semantic leaps and is gradual. The full understanding of the nature of the ontogeny and phylogeny of lexico-semantic knowledge is dependent on a better understanding of the use of that knowledge with reference to various encyclopedic frames or domains, construals, constructions and collocations. Such insights are likely to be obtained through ­investigations of

81

82

Carita Paradis

the mechanism of shifts and change in terms of the relative salience of meaning structure accommodated under the negatively formulated assumption that there are no absolute and stable meanings.

References Aijmer, Karin. 2009. Seem and evidentiality. In Ekberg, Lena and Carita Paradis (eds.), Special issue on Evidentiality in language and cognition. Functions of Language 16(1): 63–88. Barcelona, Antonio (ed.). 2003. Metaphor and Metonymy at the Crossroads: A Cognitive Perspective. Berlin and New York: Mouton de Gruyter. Benczes, Réka. 2011. Blending and creativity in metaphorical compounds: A diachronic investigation. In Schmid, Hans-Jörg and Susanne Handl (eds.), Windows to the Mind: Metaphor, Metonymy and Conceptual Blending (247–268). Berlin & New York: Mouton de Gruyter. Blank, Andreas and Peter Koch (eds.). 1999. Historical Semantics. Berlin: Mouton de Gruyter. Bloomfield, Leonard. 1933. Language. London: George Allen & Unwin Ltd. Boye, Kasper and Peter Harder. 2007. Complement-taking predicates: Usage and linguistic structure. Studies in Language 31(3): 569–606. Boye, Kasper and Peter Harder. 2009. Evidentiality: Linguistic categories and grammaticalization. In Ekberg, Lena and Carita Paradis (eds.), Special issue on Evidentiality in language and cognition. Functions of Language 16(1): 9–43. Borst, Eugen. 1902. Die Gradadverben im Englischen [Anglistische Forschung 10]. Heidelberg. Brinton, Laurel. 1996. Pragmatic Markers in English: Grammaticalizaiton and Discourse Functions. Berlin & New York: Mouton de Gruyter. Brugman, Claudia. 1984. The very idea: A case study in polysemy and cross-lexical generalizations. CLS XX, Papers from the Parasession on Lexical Semantics (21–38). Chicago: Chicago Linguistics Society. Buchstaller, Isabelle and Elizabeth Traugott. 2006. The lady was al demonyak: Historical aspects of the adverb all. English Language and Linguistics 10(2): 345–370. Bybee, Joan. 2007. Diachronic linguistics. In Geeraerts, Dirk and Hubert Cuyckens (eds.), The Oxford Handbook of Cognitive Linguistics (945–987). Oxford: Oxford University Press. Croft, William and Alan Cruse. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press. Croft, William and Esther Wood. 2000. Construal operations in linguistics and artificial intelligence. In Albertazzi, Liliana (ed.), Meaning and Cognition: A Multidisciplinary Approach (51–78). Amsterdam & Philadelphia: John Benjamins. Cruse, Alan. 2002. The construal of sense boundaries. Revue de Sémantique et Pragmatique 12: 101–119. Denison, David. 2007. Playing tag with category boundaries. In Meurman-Solin, Anneli and Arja Nurmi (eds.), Annotating Variation and Change http://www.helsinki.fi/varieng/ journal/volumes/01/denison/. Helsinki: Research Unit for Variation, Contacts and Change in English (VARIENG). Dirven, René and Ralf Pörings (eds.). 2002. Metaphor and Metonymy in Comparison and Contrast. Berlin & New York: Mouton de Gruyter. Eckardt, Regine. 2006. Meaning Change in Grammaticalization and Inquiry into Semantic Re­ analysis Change. Oxford: Oxford University Press.



Metonymization: A key mechanism in semantic change

Ekberg, Lena. 2007. …sån svensk å blond å sånt du vet. Lexiko-grammatiska drag i Malmöungdomars talspråk. In Ekberg, Lena (ed.), Språket hos ungdomar i en flerspråklig miljö i Malmö (48–77). Nordlund 27. Småskrifter från. Nordiska språk vid Lunds universitet. Ekberg, Lena. forthcoming. Joint attention and cooperation in the Swedish of adolescents in multilingual settings: The use of sån “such” and såhär “like”. In Auer, P., F. Hinskens, and P. Kerswill (eds.), Studies in Language Variation. Amsterdam & Philadelphia: John ­Benjamins. Fillmore, Charles. 1982. Frame semantics. In Linguistic Society of Korea (ed.), Linguistics in the Morning Calm (111–138). Seoul: Hanshin. Fisher, Olga. 2008. On analogy as the motivation for grammaticalization. Studies in Language 32(2): 336–382. Gärdenfors, Peter. 2000. Conceptual spaces: The geometry of thought. Cambridge, MA: MIT Press. Gärdenfors, Peter and Massimo Warglien. forthcoming. The development of semantic space for pointing and verbal communication. In Paradis, Carita, Jean Hudson, and Ulf Magnusson­ (eds.), Conceptual Spaces and the Construal of Spatial Meaning: Empirical Evidence from Human Communication. Oxford: Oxford University Press. Geeraerts, Dirk. 1997. Diachronic Prototype Semantics. Oxford: Clarendon Press. Geeraerts Dirk and Hubert Cuyckens (eds.). 2007. The Handbook of Cognitive Linguistics. Oxford: Oxford University Press. Gibbs, Raymond. 1994. The Poetics of Mind: Figurative Thought, Language and Understanding. Cambridge: Cambridge University Press. Gibbs, Raymond. 1999. Speaking and thinking with metonymy. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (61–76). Amsterdam & Philadelphia: John Benjamins. Glynn, Dylan. 2009. Polysemy, syntax and variation. A usage-based method for Cognitive Semantics. In Evans, Vyvyan and Sandra Pourcel (eds.), New Directions in Cognitive Linguistics. Amsterdam & Philadelphia: John Benjamins. Goldin-Medow, Susan. 2007. Pointing sets the stage for learning language – and creating language. Child Development 78(3): 741–745. González Díaz, Victorina. 2008. Recent developments in English intensifiers: The case of very much. English Language and Linguistics 12(2): 221–243. Goosens, Louis. 2002. Metaphtonymy: The interaction of metaphor and metonymy in expressions for linguistic action. In Dirven, René and Ralph Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (349–377). Berlin: Mouton de Gruyter. Gries, Stefan Th. 2006. Corpus-based methods and cognitive semantics: The many meanings of to run. In Gries, Stefan Th. and Anatol Stefanowitsch (eds.), Corpora in Cognitive Linguistics: Corpus-based Approaches to Syntax and Lexis (57–99). Berlin & New York: Mouton de Gruyter. Haser, Verena. 2005. Metaphor, Metonymy and Experientialist Philosophy: Challenging Cognitive Semantics. Berlin & New York: Mouton de Gruyter. Heine, Bernd. 2002. On the role of context in grammaticalization. In Wischer, Ilse and Gabriele Diewald (eds.), New Reflections on Grammaticalization (83–101). Amsterdam & Philadelphia: John Benjamins. Hopper, Paul and Elizabeth Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. Jucker, Andreas (ed.). 1995. Historical Pragmatics. Amsterdam & Philadelphia: John Benjamins.

83

84

Carita Paradis

Kemmer, Suzanne. 1993. The Middle Voice. Amsterdam: John Benjamins. Kitis, Eliza. 2009. From motion to emotion to interpersonal function. The case of fear predicates. In Pishwa, Hanna (ed.), Language and Social Cognition (433–454). Berlin & New York: Mouton de Gruyter. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9(1): 37–77. Lakoff, George. 1987. Women, Fire, and Dangerous Things. Chicago & London: University of Chicago Press. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago: University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 2000. Grammar and Conceptualization. Berlin & New York: Mouton de Gruyter. Lapata, Maria and Alex Lascarides. 2003. A probabilistic account of logical metonymy. Computational Linguistics 29(2): 263–317. Lenker, Ursula. 2008. Booster prefixes in Old English – An alternative view of the roots of ME forsooth. English Language and Linguistics 12(2): 245–265. Levin, Magnus. 2008. Hitting the back of the net just before the final whistle. High-frequency phrases in football reporting. In Lavric, Eva, Gerhard Pisek, Andrew Skinner and Wolfgang Stadler (eds.), The Linguistics of Football (143–155). Tübingen: Günter Narr. Lightfoot, David. 2006. How Languages Emerge. Cambridge: Cambridge University Press. Méndez Naya, Belén. 2003. On intensifiers and grammaticalization: The case of swiþe. English Studies 84(4): 372–391. Méndez Naya, Belén. 2007. He nas nat right fat: On the origin and development of the intensifier right. In Mazzon, Gabriella (ed.), Studies in Middle English Forms and Meanings (191–207). Frankfurt am Main: Peter Lang. Méndez Naya, Belén (ed.). 2008a. Special issue on English intensifiers. English Language and Linguistics 12(2): 213–219. Méndez Naya, Belén. 2008b. On the history of downright. English Language and Linguistics 12(2): 267–287. Nerlich, Brigitte and David D. Clarke. 2001. Serial metonymy: A study of reference-based polysemysation. Journal of Historical Pragmatics 2(2): 245–272. Nevalainen, Terttu. 1991. But, Only, Just: Focusing Adverbial Change in Modern English 1500– 1900 [Mémoires de la société néophilologique de Helsinki 51]. Helsinki: Société Néophilologique. Nevalainen, Terttu. 2008. Social variation in intensifier use: constraints on -ly adverbialization in the past. English Language and Linguistics 12(2): 289–315. Nevalainen, Terttu and Matti Rissanen. 2002. Fairly, pretty or pretty fair? On the development and grammaticalization of English downtoners. Language Sciences 24: 359–380. Oxford English Dictionary, 2nd ed. 1989. Oxford: Oxford University Press. Online version with revisions: http://www.oed.com. Panther, Klaus-Uwe and Günter Radden (eds.). 1999. Metonymy in Language and Thought. Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg. 2002. The roles of metaphor and metonymy in English -er nominals. In Dirven, René and Ralph Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (279–318). Berlin: Mouton de Gruyter.



Metonymization: A key mechanism in semantic change

Panther, Klaus-Uwe and Linda L. Thornburg (eds.). 2003. Pragmatic Inferencing in Metonymy. Amsterdam & Philadelphia: John Benjamins Publishing Company. Panther, Klaus-Uwe and Linda L. Thornburg. 2007. Metonymy. In Geeraerts, Dirk and Hubert Cuyckens (eds.), The Oxford Handbook of Cognitive Linguistics (236–263). Oxford: Oxford University Press. Panther, Klaus-Uwe and Linda L. Thornburg (eds.). 2009. Metonymy and Metaphor in Grammar. Amsterdam & Philadelphia: John Benjamins Publishing Company. Papafragou, Anna. 1996. On metonymy. Lingua 99: 169–195. Paradis, Carita. 1997. Degree Modifiers of Adjectives in Spoken British English. Lund: Lund University Press. Paradis, Carita. 2000a. It’s well weird. Degree modifiers of adjectives revisited: The nineties. In Kirk, John (ed.), Corpora Galore: Analyses and Techniques in Describing English (147–160). Amsterdam & Atlanta: Rodopi. Paradis, Carita. 2000b. Reinforcing adjectives: A cognitive semantic perspective on grammaticalization. In Bermúdez Otero, Ricardo, David Denison, Richard Hogg, and C. B. McCully (eds.), Generative Theory and Corpus Studies (233–58). Berlin: Mouton de Gruyter. Paradis, Carita. 2003. Between epistemic modality and degree: The case of really. In Facchinetti, Roberta, Manfred Krug, and Frank R. Palmer (eds.), Modality in Contemporary English (197–220). Berlin: Mouton de Gruyter. Paradis, Carita. 2004/2010. Where does metonymy stop? Senses, facets and active zones. Metaphor and Symbol 19(4): 245–264. Reprinted in Hanks, Patrick and Rachel Giora (eds.), Metaphor and Figurative Language: Critical Concepts in Linguistics (in preparation). Amsterdam: Routledge. Paradis, Carita. 2005. Ontologies and construals in lexical semantics. Axiomathes 15: 541–573. Paradis, Carita. 2008. Configurations, construals and change: Expressions of degree. English Language and Linguistics 12(2): 317–343. Paradis, Carita. 2009a. “This beauty should drink well for 10–12 years”: A note on recommendations as semantic middles. Text & Talk 29(1): 53–73. Paradis, Carita. 2009b. Prime time: The middle construction in wine-drinking recommendations. In Bowen, R., M. Moberg, and S. Ohlander (eds.), Corpus and Discourse – and Stuff: Papers in Honour of Karin Aijmer (243–225) [Gothenburg Studies in English 96]. Göteborg: Acta Universitatis Gothoburgensis. Paul, Hermann. 1920. Prinzipien der Sprachgeschichte. Halle: Niemeyer. Peters, Hans. 1993. Die englischen Gradadverbien der Kategorie Booster. Tübingen: Günter Narr Verlag. Peirsman, Yves and Dirk Geeraerts. 2006. Metonymy as a prototypical category. Cognitive Linguistics 17(3): 269–316. Pustejovsky, James. 1995. The Generative Lexicon. Cambridge, MA & London: MIT Press. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought (17–59). Amsterdam: John Benjamins. Rissanen, Matti. 2008. From “quickly” to “fairly”: On the history of rather. English Language and Linguistics 12(2): 345–359. Roberts, Ian and Anna Roussou. 2003. Syntactic Change: A Minimalist Approach to Grammaticalization. Cambridge: Cambridge University Press. Rosenkvist, Henrik. 2004. The Emergence of Conditional Subordinators in Swedish – A Study in Grammaticalization. Lund: Scandinavian Languages.

85

86 Carita Paradis

Ruiz de Mendoza Ibáñez, Francisco José. 2003. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads (109–132). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José and Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Radden, Günter, Klaus-Michael Köpcke, ­Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (33–49). ­Amsterdam & Philadelphia: John Benjamins. Stern, Gustaf. 1931. Meaning and Change of Meaning, with Special Reference to the English Language. Göteborg: Göteborgs Högskolas Årskrift 32.1. Stoffel, Cornelis. 1901. Intensives and downtoners [Anglistische Forchung 1]. Heidelberg: n.p. Sweetser, Eve. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Tagliamonte, Sali. 2008. So different and pretty cool! Recycling intensifiers in Toronto, Canada. English Language and Linguistics 12(2): 361–394. Talmy, Leonard. 2000. Towards a Cognitive Semantics I. Cambridge, MA & London: MIT Press. Tissari, Heli. 2007. Compressing emotion to politeness: On I fear and I am afraid. In Rissanen, M., M. Hintikka, L. Kahlas-Tarkka, and R. McConchie (eds.), Change in Meaning and the Meaning of Change: Studies in Semantics and Grammar from Old to PresentDay English (57–90). Helsinki: Société Néophilologique. Tissari, Heli. 2008. A look at respect: Investigating metonymies in Early Modern English. In Dury, R., M. Gotti, and M. Dossena (eds.), Lexical and Semantical Change: Selected Papers from the Fourteenth International Conference on English Historical Linguistics (139–157). Amsterdam: John Benjamins. Tomasello, Michael, Malinda Carpenter and Ulf Liszkowski. 2007. A new look at infant pointing. Child Development 78(3): 705–722. Traugott, Elizabeth. 1982. From propositional to textual and expressive meanings: Some ­semantic-pragmatic aspects of grammaticalization. In Lehmann, Winfred and Yakov ­Malkiel (eds.), Perspectives on Historical Linguistics (245–271). Amsterdam: Benjamins. Traugott, Elizabeth. 1995. Subjectification in grammaticalisation. In Stein, D. and L. Wright (eds.), Subjectivity and Subjectification. Cambridge: Cambridge University Press. Traugott, Elizabeth. 1999. The role of pragmatics in semantic change. In Verscheuren, J. (ed.), Pragmatics in 1998: Selected Papers from the 6th International Pragmatics Conference (93– 102). Antwerp: International Pragmatics Association. Traugott, Elizabeth. 2003. Constructions in grammaticalization. In Joseph, B. and R. Janda (eds.), Handbook of Historical Linguistics (624–647). Oxford: Blackwell. Traugott, Elizabeth. 2006. The semantic development of scalar focus modifiers. In van Kemenade­, Ans and Bettelou Los (eds.), The Handbook of the History of English (335–359). Oxford & Malden, MA: Blackwell. Traugott, Elizabeth. 2007. The concepts of constructional mismatch and type-shifting from the perspective of view of grammaticalization. Cognitive Linguistics 18(4): 523–557. Traugott, Elizabeth. 2008a. Grammaticalization, constructions and the incremental development of language: Suggestions from the development of degree modifiers in English. In Eckardt, Regine, Gerhard Jäger, and Tonjes Veenstra (eds.), Variation, Selection, Development: Probing the Evolutionary Model of Language Change (219–250). Berlin & New York: Mouton de Gruyter.



Metonymization: A key mechanism in semantic change

Traugott, Elizabeth. 2008b. The grammaticalization of NP of NP patterns. In Bergs, Alexander and Gabriele Diewald (eds.), Constructions and Language Change (21–43). Berlin & New York: Mouton de Gruyter. Traugott, Elizabeth and Richard Dasher. 2005. Regularity in Semantic Change. Cambridge: Cambridge University Press. Ullmann, Stephen. 1942. The range and mechanism of changes of meaning. The Journal of English and German Philology 41: 46–52. Ungerer, Friedrich. 1988. Syntax der englischen Adverbialen. Tübingen: Niemeyer. Verhagen, Arie. 2006. Constructions of Intersubjectivity: Discourse, Syntax and Cognition. Oxford: Oxford University Press. Warren, Beatrice. 1992. Sense Developments. Stockholm: Almquist & Wiksell International. Warren, Beatrice. 2003. The role of links and/or qualia in head-modifier construction. In ­Nerlich, Brigitte, Zazie Todd, Vimala Herman, and David Clarke (eds.), Polysemy: Patterns of Meaning in Mind and Language (233–252). Amsterdam & Philadelphia: John Benjamins. Warren, Beatrice. 2006. Referential Metonymy. Lund: Scripta Minora. Yoshimura, Kimihiro and John R. Taylor. 2004. What makes a good middle? The role of qualia in the interpretation and acceptability of middle expressions in English. English Language and Linguistics 8(2): 293–321.

87

Zones, facets, and prototype-based metonymy Dirk Geeraerts and Yves Peirsman University of Leuven / University of Leuven & Research Foundation – Flanders

This chapter investigates the relationship between metonymy and two related semantic phenomena: profile/zone discrepancy and facetization. Both of these phenomena aim to explain linguistic examples that activate only a specific zone or facet of the entity referred to. However, their precise semantic status, both with respect to each other and to metonymy, remains unclear. In order to resolve the resulting terminological confusion, we claim that profile/zone discrepancy functions as an overarching category that spans a continuum ranging from the classic concept of metonymy, via facetization, to non-metonymical language. We put forward several pragmatic, syntactic, and semantic arguments in support of this position. Keywords: facet, facetization, profile/zone discrepancy

1.

Introduction

Despite the massive popularity of metonymy in Cognitive Linguistics, there still exists considerable disagreement about its precise definition. In addition to the ongoing discussion between supporters of the contiguity-based and the domainbased perspective (see Peirsman and Geeraerts, 2006 for an overview), many questions remain about the relationship between metonymy and a number of similar semantic phenomena. Two of these phenomena, zone activation (Langacker­, 1984) and facetization (Paradis, 2004), are particularly problematic. Their examples are variously classified as metonymical or non-metonymical, giving rise to a terminological muddle around metonymy and its neighbouring phenomena. This chapter aims to address this issue, by comparing the semantic processes involved in zone activation, facetization and uncontroversial examples of metonymy. In particular, we will argue that facetization is a subtype of metonymy, while zone activation appears to be a totally different phenomenon.

90 Dirk Geeraerts and Yves Peirsman

Of the two phenomena under scrutiny, zone activation is the better-established one. Active zones were introduced by Langacker (1984), who defined them as “[t]hose portions of a trajector or landmark that participate directly in a given relation” (Langacker, 1984: 177). Here are a number of examples – some more typical, others less:

(1) (2) (3) (4) (5)

Your dog bit my cat.  Roger heard a noise.  Susan has a cigarette in her mouth.  This book is highly instructive.  The kettle is boiling. 

(Langacker, 1984) (Langacker, 1984) (Langacker, 1984) (Barcelona, 2005) (Langacker, 1999)

In all these sentences, it is claimed that the entities referred to by the italicized phrases do not take part in the relations described as a whole. Instead, one of their salient parts or aspects is activated: the dog’s teeth and the bitten part of the cat in (1), Roger’s ears and brain in (2), the filter end of the cigarette in (3), the semantic content of the book in (4), and the content of the kettle in (5). This activation of a specific zone is by no means exceptional in language: according to Langacker (1984: 178), it even represents the “normal situation”. The general agreement about the frequency of this phenomenon, however, is in stark contrast to the incongruity of its theoretical treatment. Langacker (1999: 67) for instance claims that “[p]rofile/active-zone discrepancy is a special case of metonymy”, while Paradis (2004: 262) describes it as “the reverse of metonymy”, because “the conventional naming is constant both for the whole and the highlighted part”. Moreover, Paradis argues that some typical examples of zone activation, like (4) above, in fact represent a different phenomenon: facetization. Facetization has to be situated somewhere in between metonymy and zone activation. It occurs with concepts that have so-called meaning “facets” (a term borrowed from Cruse, 1995). All these facets are “on a par with one another” (Paradis, 2004: 246), but in a specific context one of them is usually selected. ­Paradis (2004) distinguishes between two groups of concepts that trigger facetization: the book group and the department group. Concepts in the former group have two facets (a concrete, psychical one and an abstract one), those in the latter have three (people, building, and institution). Even though these facets can be focused upon separately, they are not antagonistic, so that zeugma is possible. This is in contrast to metonymy, where the source and target meanings (“two different concepts in a conceptual complex”) are incompatible and therefore block zeugma. Paradis thus arrives at a three-way classification: metonymy, facetization, and zone activation. Ruiz de Mendoza (this volume), however, challenges her approach, by describing facetization as “another label for what Croft (2002/1993)



Zones, facets, and prototype-based metonymy

called domain highlighting” (which is also present in target-in-source metonymies, see below), along with zone activation. Zone activation, as defined by Langacker (1984), may thus be one of the least homogeneous concepts of Cognitive Linguistics. On the basis of the examples above, it appears to take on the form of a continuum. At the one end, we have cases like Your dog bit my cat and Susan has a cigarette in her mouth. These represent the most typical instances of zone activation. They will be discussed in the next section. At the other end, we meet clear instances of metonymy like The kettle is boiling or Washington is not insensitive to the needs of the people. These are the topic of Section 3. In between, there are cases like This book is highly instructive, which are treated by Paradis (2004) as facetization. The discussion of these examples, in Section 4, will take up the major part of this chapter. We will argue that the last two types should be treated as metonymical, while the first type should not.

2.

Dogs and cigarettes

Let us start with the most prototypical type of zone activation, instantiated by Examples (1) to (3). In relation to these examples, it is often claimed that the referents of your dog, my cat, Roger and a cigarette do not participate in the relationships expressed by bite, hear, and have (in somebody’s mouth) as a whole. Instead, it is only the teeth of the dog, the bitten part of the cat, Roger’s ears and a small part of the cigarette that are activated. These examples are therefore sometimes classified as a type of whole-for-part metonymy. According to Langacker, “the entity that is normally designated by a metonymic expression serves as a reference point affording mental access to the desired target (i.e., the entity actually being referred to)” (Langacker, 1993: 30). In other words, there is a discrepancy between “the entity that is normally designated” and “the entity actually being referred to”. We believe that this discrepancy, and the shift of reference it entails, is not present in the examples above, despite the fact that some parts of the referents participate more clearly in the action than others. Examples like these should thus not be seen as metonymical. The shift of reference that takes place in referential metonymies is contextually unavoidable. It is mostly triggered by the selectional restrictions of the verb, but potentially also by more encyclopedic knowledge. Let us give a few instances of such uncontroversial referential metonymies. In Example (6), the verb criticize requires a human, volitional subject. The Times, however, generally refers to a newspaper – an inanimate entity. In order to resolve this problem, we therefore impose a shift of reference on the noun phrase, and take it to refer to one or

91

92

Dirk Geeraerts and Yves Peirsman

more of the journalists writing for this newspaper. This process is called “coercion” in Pustejovsky’s (1995) theory of the Generative Lexicon. In Example (7), an equally uncontroversial metonymy, there is no such clash between the subject of the verb and that verb’s selectional restrictions. Here it is encyclopedic knowledge that tells us it was not Nixon himself who bombed Hanoi. Even though a literal interpretation is allowed theoretically, we know that presidents in times of war leave the bombing of cities to their subordinates, and do not undertake the action themselves. Therefore we again impose a shift of reference on Nixon, from the president to a soldier or group of soldiers under his command. In contrast to purely grammatical selectional restrictions, these “encyclopedic” restrictions are not covered by Pustejovsky’s theory. Nevertheless, both types of reference shift are indispensable – not always in a logical, but certainly in a pragmatic sense, as a prerequisite for arriving at a contextually appropriate reading: the literal interpretations of the nouns lead to an impossible or incorrect sentence meaning. (6) The Times criticized the minister. (7) Nixon bombed Hanoi.

Examples like your dog bit my cat, however, do not require such a shift of reference. Take the grammatical selectional preferences of bite: it requires an animate volitional subject – a condition fulfilled by your dog. Encyclopedic knowledge does not trigger a shift of reference either, as it is indeed a dog that participates in the action: the animal felt compelled to bite the cat, ran to the cat, and finally bit it. What is more, if we do make the shift of reference from the dog to its teeth, we arrive at an impossible situation. Teeth on their own (think of a set of false teeth) are not able to bite. They need an animate, volitional entity of which they are a part and which controls them. As a result, a noun phrase like your dog’s teeth does not correspond to the selectional restrictions of bite. This is noted, too, by Langacker (1999: 62), who gives the infelicitous example *Your dog’s teeth (and jaws, etc.) bit my cat’s tail. The same goes for Examples (2) and (3). It is not just Roger’s ear that participates in the action of hearing – an ear that has been chopped off does not hear much. Similarly, Susan in Example (3) is smoking an entire cigarette, even though only part of it is located inside her mouth. If we shifted the sense of cigarette to “a part of a cigarette”, we could even interpret sentence (3) as referring to a situation in which Susan has broken off a piece of cigarette and started chewing on it. This interpretation is in fact impossible, or at least highly marked. Now compare this situation to Examples (6) and (7), where we can easily make the shift of reference explicit. The sentences The journalists of the Times criticized the minister and Nixon’s soldiers bombed Hanoi are perfectly acceptable. This underpins the observation that a different conceptual process is taking place here.



Zones, facets, and prototype-based metonymy

To sum up, typical instances of zone activation, like Examples (1) to (3), should not be seen as metonymical. The lack of necessity of the reference shift, together with the incompatibility between the verb and the shifted sense of the predicate, lead us to the conclusion that these typical examples of profile/zone discrepancy clearly differ from genuine metonymies like Examples (6) and (7).

3.

Kettles and telephones

While these previous typical examples of zone activation are most easily recognized as non-metonymical – in fact, they probably do not even involve a semantic shift – instances at the other end of the continuum are most easily identifiable as real metonymies. Let us give a few examples: (8) (9) (10) (11) (12)

The kettle is boiling.  Washington is insensitive to the needs of the people.  The author finished a new book.  He picked up the telephone. He filled up the car.

(Langacker, 1999) (Barcelona, 2002) (Langacker, 1984)

These examples differ from the ones in the previous section in that their shift of reference is necessary. Intransitive boiling requires as its subject a fluid, and thus necessarily triggers a reference shift in the kettle. Insensitive is a characteristic that in a literal sense can only refer to people, causing a shift of reference from the city Washington to the members of the American parliament. Finish is a verb that requires an action as its object, but a new book normally refers to an object. Therefore its meaning is shifted to an action in which the new book participates – typically, the action of writing or reading. In pick up the phone and fill up the car, finally, the shift of reference is caused by encyclopedic knowledge. Although it is possible to pick up the whole device with which you make a phone call, and to fill your entire car with petrol, our knowledge about the real-world situations causes us to impose a sense shift on the noun phrases involved, from the whole to one of its parts. These shifts of references set apart these examples from those of the type the dog bit the cat. This classification of Examples (8) to (12) is supported by the current theory of metonymy. Whichever view of metonymy one may take – the contiguity-based or the domain-based (Peirsman and Geeraerts, 2006) – it is the conceptual link between the source and target senses that distinguishes metonymy from other semantic phenomena like metaphor. Examples (8) to (12) represent conceptual links that are clearly metonymical. They instantiate typical metonymical patterns

93

94 Dirk Geeraerts and Yves Peirsman

like container for contents in (8), place for people in (9), participant for action in (10), and whole for part in (11) and (12). These patterns all obey the contiguity or same-(matrix-)domain criterion and have frequently been treated as metonymical elsewhere (Kövecses and Radden, 1998). One mistake we should not make is to see these examples as non-metonymical because of their conventionality. What connects sentences (8) to (12) is that they are so frequent, so well-established, that many people forget to think of them as metonymical. Conventionality per se, however, does not change the type of conceptual link that motivates the sense shift. For instance, even though (13) and (14) are much less conventional than (8) to (12), the semantic relationship that links Washington to the American government, Edinburgh to the Scottish Parliament and Houston to NASA is exactly the same: the city stands for the people working in an institution located there. The link may be a bit stronger with more conventional examples, but its type remains the same. (13) Edinburgh will decide on the future of Scotland. (14) Houston decided to keep the media out.

A similar case, albeit in a different context, has already been made by Panther and Thornburg (2003a). They claim that metonymy occurs both in semantics and pragmatics, since a pattern like salient body part for animate being underlies the conventional redbreast as well as novel, unconventional names like yellowbeak or purplehead. Similarly, the conventional metonymical process behind the polysemous potbelly can lead to new examples like balloonnose or fatface (Panther and Thornburg, 2003a: 7). Conventionality is a result of many factors, like the frequency of use of a word; it does not change the conceptual process behind it. For these three reasons – the necessity of the sense shift, the instantiation of typical metonymical patterns and the irrelevance of their conventionality in our case – we see no reasons to treat the examples above as non-metonymical.

4.

Books and trumpets

It is the intermediate type of examples – the ones called facetization by Paradis (2004) – that deserves the most elaborate discussion. Let us contrast an example like (16) with a metonymy like the one in (15): (15) The red shirts won the match. (16) This book is utterly boring.



Zones, facets, and prototype-based metonymy

In Example (15), a piece of clothing is used to refer to the football players that it is worn by. This is an uncontroversial instance of metonymy. In Example (16), the adjective boring highlights the semantic content of the book, and not the book as a physical object. Since the content can be regarded as an inherent property of books (whereas football players are not inherent properties of the shirts they wear), the status of this example is extremely controversial. Paradis (2004) calls it facetization, Ruiz de Mendoza (2000) treats it as a pure metonymy, and Croft (2002/1993) argues it is non-metonymical. This is because Croft’s theory holds that metonymy involves the highlighting of a secondary subdomain. As he considers semantic content to be a primary subdomain of book, he classifies Example (16) as non-metonymical. Ruiz de Mendoza (2000), by contrast, argues semantic content to be a relatively secondary property of books, and therefore views (16) as metonymical after all. Guided by these two examples, we will now investigate whether there is indeed a difference in the conceptual links underlying the phenomena they instantiate. We will identify the differences and similarities at the pragmatic, syntactic and semantic levels and then show why the dissimilarities – if any – are only superficial, and do not necessarily point to conceptual differences. In Section 4.1, we will look at the pragmatic differences: is there a distinction between the functional, communicative processes that the phenomena trigger? In Section 4.2, we will home in on the syntactic behavior of the examples. We will determine in particular if they behave differently with respect to zeugma. Finally, in Section 4.3, we will scrutinize the semantic aspects. Could it be argued, for instance, that the source and target senses of (16) are inseparably connected, while this is not the case for (15)?

4.1 Pragmatics Examples (15) and (16) serve the same type of referential identification. In both sentences, the shift from a source sense to a target sense has a clear pragmatic function: it serves to correctly identify the referent that the context asks for. If shirt and book in these examples are seen as metonymies, they would thus be of the referential kind. We should add here that metonymy is not the only way of realizing such referential identification. In the case of, let’s say, the Belgian national football team, we have a variety of semantic phenomena at our disposal: – literal noun phrases: de nationale ploeg (“the national team”) – anaphora: die van België (“those from Belgium”), de onzen (“our boys”, lit. “the ours”) – metaphor: de Rode Duivels (“the Red Devils”)

95

96 Dirk Geeraerts and Yves Peirsman

Metonymy is thus merely one of the possible instruments by which referential identification can be achieved. In particular, we can summarize the role of metonymy in this process by a pragmatic rule like the following: In order to identify a certain referent, make use of a feature that economically allows for the unique identification of that referent in the given context.

Thus, if there is a conceptually unique link between X and Y, and if X happens to be more accessible or more noticeable than Y, X can be used for the identification of Y (cf. also Nunberg, 1978). This rule applies to our shirt-type as well as to our book-type examples. Football shirts are such a useful way of referring to football players simply because they allow swift identification of either of the two teams – that is why they were invented to begin with. Similarly, we use the word book to refer to a book’s semantic content because physical entities are more salient in our world than abstract ones. The reasons why certain concepts are more likely to function as metonymical sources have been investigated extensively (see for instance Kövecses and ­Radden, 1998). They hold for all phenomena that we have introduced above. Although the grounds for choosing one concept as the source rather than the other are the same, they do produce slightly different results. With the shirt-type examples, the source is a subdomain of the target: the gaudy-colored shirts are a distinct characteristic in the domain of football players. With the book-type examples, by contrast, the target is a subdomain of the source: semantic content is a salient property in the domain of books. This distinction between target-in-source and source-in-target phenomena was introduced by Ruiz de Mendoza (2000, see also Ruiz de Mendoza, this volume), who treated both as subtypes of metonymy. After all, the reversibility of metonymical patterns like part for whole or container for contained has been a well-known fact since the earliest works on the topic. The relative position of the source vis-à-vis the target should thus not be a reason to treat the two subtypes as entirely different phenomena. What it does entail, however, is a difference in syntactic behavior.

4.2 Syntax One difference between shirt-like and book-like examples that crops up time and again in the literature is their behavior with respect to zeugma. This difference is illustrated by Examples (17) and (18): (17) *The red shirts won the match and had to be cleaned thoroughly. (18) The book is thick as well as boring.



Zones, facets, and prototype-based metonymy

Sentence (17) is ungrammatical: in a conjunction of two predicates, it is not possible for one to refer to the subject’s target sense (won the match) and for the other to refer to its source sense (had to be cleaned thoroughly). Sentence (18), by contrast, is possible, even though again each of the predicates refers to a different sense (or facet) of book. Thick refers to the physical entity, while boring highlights its content. Clearly, there must be a difference between the shirt-type and the book-type phenomena that explains this pattern. For Paradis (2004), this difference is essential enough to classify the two examples as instances of separate phenomena: metonymy and facetization. Examples of facetization, she argues, distinguish themselves, because “[o]n the one hand, their various readings can be separately focalized, but, on the other, the readings are not antagonistic” (Paradis, 2004: 258). It is precisely this last aspect that allows for zeugma. This explanation, however, does not seem to suffice. Let us take an example like department, which instantiates the second group of Paradis’ (2004) facetization concepts. The facets of this concept (the institution, the building, the people) are “not antagonistic”. Therefore we should be able to combine two predicates that select different facets. While Example (19) illustrates this is indeed possible, (20) and (21) show this is not necessarily the case: (19) Linguistics is the biggest and most sympathetic department of the faculty. (20) *The whole department was made redundant and then torn down. (21) *The department was down in the dumps after it burnt down.

In Example (19), biggest refers to the literal sense of department, while most sympathetic selects the people working in that department. As Paradis (2004) correctly predicts, the two adjectives can be combined without any problem. In (20), however, this is not the case. The predicate was made redundant highlights the people facet of department, while [was] torn down focuses on the building. Even though these readings should not be antagonistic, zeugma is blocked. The same observation applies to Example (21), albeit in relation to anaphoric reference instead of zeugma. Down in the dumps again selects the people working in the department, burnt down applies to the building. As a result, anaphoric it clashes with the sense shift imposed by was down in the dumps. This shows that instances of facetization do not behave homogeneously with respect to syntactic phenomena like zeugma. So what explains these puzzling facts? On closer inspection, it becomes clear that these instances of department, book, and the like, actually behave the same with respect to zeugma as typical target-in-source metonymies, like author for work or place for people. The noun phrase the red shirts in Example (17), by contrast, represents a source-in-target metonymy. Consider the following metonymical examples:

97

98 Dirk Geeraerts and Yves Peirsman

(22) That author is still young, but already impossible to read. (23) Belgium is a small, but very friendly country. (24) *Madrid decided to cut taxes in the morning and lost an important football match in the evening.

Examples (22) and (23) are equivalent to (18) and (19). They all have one predicate focusing on the literal aspect of the subject (be thick and be young, biggest and small), and one imposing a sense shift on that subject (be boring and be impossible to read, sympathetic and friendly). Yet, these two predicates can simply be combined, in the metonymical examples as well as in the so-called facetization examples. Similarly, Example (24) is equivalent to (20) and (21). Here both predicates impose a sense shift on the subject. In (20), the sense shifts go from the abstract department to its people and its building. In (24), the sense shifts lead to the interpretation of Madrid as “the Spanish government” first, and to “the Spanish or Real Madrid football team” second. In both cases, these predicates cannot be combined, because they refer to two different sense-shifted interpretations of the subject. As a result, the alleged status of the example (as either a metonymy or an instance of facetization) does not make a difference for its behavior with respect to zeugma. The relevant difference between sentences (17) and (18) lies in the distinction between target-in-source metonymies (like book in Example (18)) on the one hand and source-in-target metonymies (like the red shirts in Example (17)) on the other, not between metonymy and facetization.

4.3 Semantics It has become clear that the pragmatic function and syntactic behavior of examples like (15) and (16) do not point towards any conceptual difference between them. So maybe the difference lies in their semantics. It could be claimed, specifically, that there lies a distinction in the conceptual link we find in the book-type and the shirt-type examples. For instance, the physical entity book and its content seem pretty inseparable: at first sight, it appears impossible to have one without the other. This is in contrast to a shirt and the person that wears it, which can easily be separated from each other. This inseparability, however, should not be taken at face value. Think, for instance, of the recent phenomenon of e-books, which contain the same type of content as their traditional counterparts, but do away with the physical entity. Apparently, the two are not as inseparable as we first thought. Or consider a sentence like Everyone has a novel in them, where again the physical entity is absent, while the content exists, albeit in someone’s fantasy or lived experience. It is not difficult to come up with other examples: there are many theatre companies without



Zones, facets, and prototype-based metonymy

their own theatre building, books without content, dogs without teeth. In short, a criterion like inseparability cannot lie at the basis of a demarcation of facetization or zone activation. Of course, there often exists a difference in “conceptual distance” between the source and target senses. This can be observed in a series like the following: (25) I pick up the trumpet. (= the object) (26) I hear the trumpet. (= the sound produced by the object) (27) I hear the trumpet. (= an imitation of trumpet sound, produced by something else, a synthesizer, for instance)

In this series, (25) is the literal example: the trumpet refers to the physical entity that has dropped to the floor. In (26) and (27), we have distinct target senses: in (26) it is the sound of a genuine trumpet, in (27) it is a trumpet-like sound produced by a synthesizer or a computer. Obviously, the distance between the source and target sense is much bigger in (27) than in (26). In (26), an actual trumpet participates in the extra-linguistic situation; in (27) there is no real trumpet to be seen. On these grounds it could be claimed that only (27) is metonymical, while (26) is an example of a related process like facetization or zone activation. Why? The difference in contextual distance could motivate a distinction between semantic and pragmatic phenomena: the contextual proximity between the source and target senses in (26) ensures that its target meaning can be pragmatically derived, while the referential shift in (27) requires more explanation and is thus of a semantic nature. Following this logic, metonymy could be seen as a semantic phenomenon, facetization, and zone activation as a type of pragmatic inference. Again, there are good reasons to question such a train of thought. First, the conceptual distance between two senses does not say anything about the type of link between them. Second, metonymy and pragmatic inferences often occur together and even interact: not only do prototypical metonymies also make use of pragmatic inferences; even pragmatic inferences may rely on metonymy (Panther and Thornburg, 2003b; Chen, this volume). Let us first show that pragmatic inferences are also present in “uncontroversial” metonymies. Whenever someone uses an example like (27), or any other metonymy for that matter, the hearers have only the context to go by in order to identify the correct referent. This is true even for metonymies that are much more conventional than (27). At a given point, someone must have used the metonymy for the first time. At that time it was new, and the hearer thus had to pragmatically infer the intended sense of the word. This is more or less comparable with the e-book example we gave above: language users make use of their encyclopedic knowledge of electronic media to infer that an e-book is an electronic book that does not take on a physical form.

99

100 Dirk Geeraerts and Yves Peirsman

Conversely, pragmatic inferences often rely on metonymy. In his treatment of pragmatic inferences from subsets of a category, Geeraerts (1997: 83) argues that “mechanisms of change like metaphor and metonymy and change from subsets are not mutually exclusive, but rather regularly interact”. He illustrates this interaction with the example to be in deep waters, “to be in a difficult situation”. Obviously, not everyone is in danger as soon as they swim in water that is literally deep. So why do we automatically interpret the deep waters in the expression as providing someone with difficulties? This interpretation is only possible when we first shift pragmatically from the entire category “situations in deep water” to that subset of situations that involve danger. Only after this process has taken place can we move to the figurative target meaning, by a process of metaphor, metonymy or a similar semantic phenomenon. Thus, pragmatic inferences like the former regularly interact with semantic processes like the latter. Even metaphors require the reader to make pragmatic inferences, after all. As a result, the pragmatic nature of an example like (26) does not allow us to reject metonymy as a possible conceptual process. Even though this case does not involve “change from subsets”, the pragmatic aspect of the inferential process should be kept apart from its semantic nature. This should be no surprise: while pragmatic inferencing takes place as a part of the interpretative activity of the hearer, we are focusing here on the speaker who picks out a suitable name for a referent. Even though we have now established that both (26) and (27) involve pragmatic processes, there is one final way of construing the difference between them that needs to be considered. Suppose we argue as follows: whereas (27) has only one reading, examples like (26) display a semantic ambiguity between a. the sound produced by the object; and b. the object as producing a sound.

Of these, (b) could be seen as the “active zone” or “facet” reading, where we focus on the instrument during the activity for which it is typically used. (a) would then be the genuinely metonymical reading, which occurs independently in an example like (27). In this interpretation, the opposition between metonymy and facetization is reduced to two different readings of the same example – two readings, moreover, that in practice are inseparable from each other. After all, if there is a difference in attentional profiling involved, where would it show up – linguistically or otherwise? While it may be a last resort of the defendants of a facetization or zone activation reading, it would be up to them to show that the distinction between (a) and (b) actually makes a difference. In summary, the distinctions between book­-type and shirt-type examples provide no sufficient reasons for their treatment as two entirely different conceptual phenomena. It rather looks like the relevant differences can be explained with reference to the two types of metonymy introduced by Ruiz de Mendoza (2000).



Zones, facets, and prototype-based metonymy 101

Examples of the book type are instances of target-in-source metonymies, while the shirt-type cases are source-in-target metonymies. This analysis does not only explain their pragmatic characteristics and their syntactic behavior with respect to zeugma; it also takes away all reasons to claim any deeper conceptual differences than just the direction of the sense shift.

5.

Conclusions

In this chapter, we have investigated the relationship between metonymy and two related semantic phenomena, viz. profile/zone discrepancy and facetization. We identified three distinct groups, exemplified by the cases your dog bit my cat, the kettle is boiling, and this book is boring. We argued that the last two of these instantiate metonymy, while the first case does not even involve a shift of reference and should thus be treated as non-metonymical. In your dog bit my cat, there is no reason to impose a shift of reference on the noun phrases. Both the dog and the cat participate in the action as animate, volitional entities (well, the cat may be less volitional than the dog here…), and it is not just the teeth of the dog, or the bitten part of the cat that are relevant to the action. In fact, shifting the dog to the dog’s teeth violates the selectional restrictions of bite, which requires a volitional subject. While it is true that some parts of the participants are highlighted in the conceptualization of the action, the lack of a shift of reference leads us to the conclusion that this conceptual phenomenon is different from metonymy. In the kettle is boiling, the situation is totally different. The kettle violates the selectional restrictions of boil, which requires a fluid as its subject. Therefore the shift of reference from the container to its contents is necessary. In other cases, the shift of reference may be triggered by encyclopedic knowledge rather than selectional restrictions, but it remains indispensable. This crucial difference with examples of the former type shows that we are dealing here with genuine metonymies. This book is boring probably instantiates the most controversial case. However, we argued that again, there is no reason to see in this type a phenomenon different from metonymy, from a pragmatic, syntactic or a semantic perspective. First, it shares with metonymy the pragmatic function of referential identification. Second, we showed that its behavior with respect to zeugma is the same as that of uncontroversial target-in-source metonymies. Third, both metonymy and this book is boring involve semantic as well as pragmatic processes. As a result, we see no reason to set examples like these apart from metonymy. With this investigation, we have wanted to resolve the terminological problems that surround the concept of metonymy and related phenomena. We have proposed a simple classification of a number of phenomena, based on the ­necessity

102 Dirk Geeraerts and Yves Peirsman

of referential shifts, and supported Ruiz de Mendoza’s (2000) analysis of sourcein-target and target-in-source metonymies. In this way, we hope to contribute to a clearer delimitation of the notion of metonymy in Cognitive Linguistics, based on conceptual grounds.

References Barcelona, Antonio. 2002. Clarifying and applying the notions of metaphor and metonymy within Cognitive Linguistics: An update. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast [Cognitive Linguistics Research 20] (207–277). Berlin: Mouton de Gruyter. Barcelona, Antonio. 2005. The multilevel operation of metonymy in grammar and discourse, with particular attention to metonymic chains. In Ruiz de Mendoza, Francisco and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction [Cognitive Linguistics Research 32] (313–352). Berlin & New York: Mouton de Gruyter. Croft, William. 2002/1993. The role of domains in the interpretation of metaphors and metonymies. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (161–205). Berlin & New York: Mouton de Gruyter. (Reproduced with slight changes from the paper with the same title in Cognitive Linguistics 4(4): 335–371.) Cruse, D. Alan. 1995. Polysemy and related phenomena from a cognitive linguistic viewpoint. In Disier, Patrick St. and Evelyne Viegas (eds.), Computational Lexical Semantics (33–49). Cambridge: Cambridge University Press. Geeraerts, Dirk. 1997. Diachronic Prototype Semantics. Oxford: The Clarendon Press. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9(1): 37–77. Langacker, Ronald W. 1984. Active Zones. In Proceedings of the Annual Meeting of the Berkeley Linguistics Society 12: 455–471. Langacker, Ronald W. 1993. Reference-point constructions. Cognitive Linguistics 4: 1–38. Langacker, Ronald W. 1999. Grammar and Conceptualization. Berlin: Mouton de Gruyter. Nunberg, Geoffrey D. 1978. The Pragmatics of Reference. Bloomington: The Indiana University Linguistics Club. Panther, Klaus-Uwe and Linda L. Thornburg. 2003a. Introduction: On the nature of conceptual metonymy. In Panther, Klaus-Uwe and Linda L. Thornburg (eds.), Metonymy and Pragmatic Inferencing [Pragmatics and Beyond New Series 113] (1–20). Amsterdam: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg (eds.). 2003b. Metonymy and Pragmatic Inferencing [Pragmatics and Beyond New Series 113]. Amsterdam: John Benjamins. Paradis, Carita. 2004. Where does metonymy stop? Senses, facets and active zones. Metaphor and Symbol 19(4): 245–264. Peirsman, Yves and Dirk Geeraerts. 2006. Metonymy as a prototypical category. Cognitive Linguistics 17(3): 269–316. Pustejovsky, James. 1995. The Generative Lexicon. London: MIT Press. Ruiz de Mendoza, Francisco. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.) Metaphor and Metonymy at the Crossroads: Cognitive Approaches [Topics in English Linguistics 30] (109–132). Berlin & New York: Mouton de Gruyter.

Metonymy and cognitive operations* Francisco José Ruiz de Mendoza Ibáñez University of La Rioja, Spain

This chapter begins with a distinction made in Ruiz de Mendoza and Peña (2005) between two broad kinds of cognitive operations: formal and content operations. The former are higher-level processes whose activity is necessary for lower-level processes to take place. The latter are used to make inferences on the basis of cues provided by the linguistic expression and its context. The chapter proposes that metonymy can be broken down into two content operations, domain expansion and domain reduction, the latter of which can be further broken down into other two processes: facetization and zone activation. It then explores how these operations relate to or contrast with other content operations and the way in which they are supported by formal operations. Keywords: cognitive operation, facetization, metaphor, zone activation

1.

Introduction

In psychology a cognitive operation is usually understood as the performance of some cognitive activity that has an identifiable effect in terms of how the brain responds to the human need to interact with the world. Examples of cognitive operations range from concept-construction and identification processes (e.g., memory encoding and retrieval; tactile, visual and auditory encoding) through mental imagery manipulation to motor programs and their execution

* Center for Research in the Applications of Language (http://www.cilap.es). Financial support for this research has been provided by the Spanish Ministry of Science and Innovation, grant no. HUM2007-65755. This research is also part of the activities of the international research group Lexicom (http://www.lexicom.es). I am grateful to Antonio Barcelona Sánchez, Réka Benczes, the Human Cognitive Processing series editors and the anonymous referees for their comments on the preliminary versions of this chapter. Of course, any remaining flaw is my own responsibility.

104 Francisco José Ruiz de Mendoza Ibáñez

(cf. ­ Anderson, 1983). In Cognitive Linguistics terms like “cognitive operation” or “cognitive process” have been used in a more restricted way either to refer to construal phenomena – partly inspired by previous work in cognitive psychology – such as figure–ground alignment (Langacker, 1987) and windowing of attention (Talmy, 2000) or to designate conceptual structure creation processes such as metaphoric and metonymic mappings (Lakoff, 1987, 1993) or blending (Fauconnier and Turner, 2002). In the present chapter, we will make use of the notion of cognitive operation in a way which is coherent with the more restricted view and which sets up a link between mental processes that become manifest in language use and their potential to make meaning through inferential activity. Thus, by “cognitive operation” we will refer to any mental mechanism whose purpose is to contribute to the inferential processes that are necessary to derive a full semantic representation out of a linguistic expression or any other symbolic device (e.g., a drawing) in order to make it fully meaningful in the context in which it is to be interpreted. Note that this approach to cognitive operations does not deny their status as such to other cognitive phenomena like the ones listed above. It only attempts to refine the nature and scope of the notion when applied to language-based meaning construction processes. As pointed out above, metaphoric and metonymic mappings seem to be clear examples of cognitive operations. However, saying that a mapping is a cognitive operation, although essentially correct, is an oversimplification. Mappings are sets of correspondences, but saying that there is a correspondence between two elements does not reveal how the elements are perceived to correspond. It is necessary to refine our description by considering all possible cognitive operations that may determine the nature of a correspondence. In previous work (Ruiz de Mendoza and Peña, 2005), a distinction has been made between two broad kinds of cognitive operations: content and formal operations. Formal operations are higher-level processes whose activity is necessary for lower-level processes to take place. Content operations are used to make inferences on the basis of cues provided by the linguistic expression or the context in which it is produced. Among them, we find two basic metonymic operations: domain expansion and domain reduction, already identified in Ruiz de Mendoza (2000). This chapter explores the boundaries of these two operations in connection to other low-level processes, such as correlation, contrast, and resemblance, among others. It also argues that understanding the full gamut of cognitive operations supplies the analyst with solid criteria to delimit the nature of metonymy.



Metonymy and cognitive operations 105

2.

Types of cognitive operation

Following previous work in Ruiz de Mendoza and Santibáñez (2003), we will argue that there is linguistic evidence for a basic distinction between content and formal cognitive operations. As mentioned in the introduction section, the former are lower-level processes that people use to make inferences and thus construct meaning on the basis of linguistic and/or contextual cues. However, by themselves these operations are insufficient to explain how the inferential part of meaning construction takes place. There are other higher-level operations, of a formal nature, which act as prerequisites for content operations. The evidence for the formulation of different cognitive operations derives from the observation of the linguistic behavior and communicative effects of utterances traditionally regarded as containing examples of so-called figures of speech such as metaphor, metonymy, hyperbole, meiosis, and oxymoron. We make use of some of the explanatory tools developed by cognitive linguists over the last three decades to identify the cognitive processes that underlie and constrain the inferential activity arising from the communicative exploitation of these figures.

2.1 Formal operations Formal operations play no direct role in inference making. Inferences are the direct result of lower-level or content operations. However, without formal operations the conceptual material would not be ready for lower-level operations to be possible and inferential activity would be hindered. We will distinguish four basic formal operations: (1) cueing, (2) selection, (3) integration, and (4) abstraction. We examine each operation in turn. Cueing, which is the most basic operation of the four, invariably applies in all meaning construction tasks in order to allow the addressee to select the relevant conceptual structure and to organize it adequately for the interpretive task that is taking place. The concept of cueing has been explored at all levels of meaning construction in Ruiz de Mendoza and Mairal (2008) and Mairal and Ruiz de Mendoza (2009). But for the purposes of the present chapter, it will be enough to discuss it briefly in its connection to metonymy and some related notions such as meaning facetization and active zones. Let us start with the following uses of the word rabbit (cf. Nunberg and Zaenen, 1992): (1)

a. b. c. d.

She loves her little rabbit (‘the animal’). We eat rabbit every now and then (‘rabbit meat’). There was rabbit all over the road (‘the dead flesh’). He wears rabbit regularly (‘the fur’).

106 Francisco José Ruiz de Mendoza Ibáñez

The interpretation of the word rabbit in Examples (1b) to (1d) is regarded by some scholars as a matter of metonymy (Ruiz de Mendoza, 2000; Geeraerts and ­Peirsman, this volume). The same word in (1a), on the other hand, is a clear case of so-called central (or sometimes “literal”) meaning. Other scholars (e.g., ­Paradis, 2004, this volume) would argue that all of the uses of rabbit above are a question of meaning facetization, that is, the highlighting of different aspects or properties of a conceptual characterization that are inherent to it. According to Paradis (2004), in facetization the different aspects of a concept can stand together in the same sentence without producing a semantic clash as they are called upon; that is, zeugma is possible: She eats and wears rabbit. However, zeugma is not possible in the case of metonymy: *The ham sandwich is waiting for his check and getting stale. As Geeraerts and Peirsman (this volume) have shown, however, there are cases of facetization that are not sensitive to zeugma. Compare the possibility of saying The department is big and also friendly (where department refers both to the institution and to its members), but not *The whole department was made redundant and then torn down (where department first refers to the people and then to the building). The solution that Geeraerts and Peirsman offer to this puzzle is to dispense with the notion of facetization, postulate metonymic meaning shifts, and explain sensitiveness to zeugma on the basis of a careful consideration of source-target inclusion relationships, since zeugma is only possible when we combine several target-in-source metonymies, as in She eats and wears rabbit, which can be paraphrased as ‘She eats rabbit meat’ and ‘She wears rabbit fur’. In contrast, the sentence *The whole department was made redundant and then torn down incorrectly combines a target-in-source metonymy (the institution for its members) with a source-in-target metonymy (the institution for the building where its activities take place). However, it is true that facetization exists. In our view, it is another label for what Croft (1993) called domain highlighting, which consists in giving primary status to what is otherwise a secondary conceptual domain. In this connection, Ruiz de Mendoza (2000) has observed that only target-in-source metonymies make use of highlighting, which involves domain reduction, that is, cutting down the amount of conceptual material used to construct the meaning interpretation, while source-in-target metonymies are based on domain expansion, which consists in increasing the amount of relevant conceptual material on the basis of the point of access provided by the metonymic source. A clear case of source-intarget­ metonymy is the use of hand in hand for worker in the sentence We need another hand to complete the job, which we can compare with the non-metonymic­ use in My hand was covered in plaster to immobilize it while the thumb healed. Evidently, the different senses of the word hand in these two sentences are a matter of the textual environment provided for this word. Thus, the predicate “to



Metonymy and cognitive operations 107

complete­ the job” calls for an interpretation of “hand” in the domain of labor, which is possible because of our common experience of using our hands to work, and supports the meaning shift from hand to worker; in turn, the predicate “was covered in plaster” requires an interpretation of “hand” in the domain of its physical properties, especially its size and shape. Note that whether we have a central characterization of a lexical item, as in (1a), or a meaning shift, as in (1b) to (1d), or in the hand for worker metonymy, the interpretation crucially depends on the way the rest of the linguistic expression interacts with the meaning of the lexical item. Thus, the predicates “love”, “eat”, “there was”, and “wear” in the different examples in (1), or the whole predication (“complete the job”) in We need another hand to complete the job, guide us in calling upon the most relevant aspects of the concepts rabbit and hand respectively. This process, which we call cueing, is thus a pre-requisite for the two cognitive operations that we have described: domain reduction based on two forms of highlighting (facetization, zone activation) and domain expansion. The result of cueing is explainable in terms of what Langacker (2000, 2009) has called profile / active zone discrepancies or interpretive situations where the profile of a concept is not the same as its active zone. In Langacker’s view, except for some special cases like time, concepts are profiled against other concepts, which are their base domains. The profile of a concept is the way we construe its referent. For example, the word rabbit may be profiled in the (base) domains of size, shape, color, and weight, among other possibilities. (cf. In the pet shop there was a giant brown rabbit that looked like a huge cat and was too heavy for me to carry.) The active zone of a concept is the portion of the concept that directly participates in a given relation (Langacker, 1984, 1987). In We ate rabbit, the meat is the active zone of the concept rabbit, whose referent is a rabbit construed as an animal that can be used as food (i.e., rabbit is profiled in the domain of edible animals). The active zone (rabbit meat) and profile (the rabbit) do not coincide. However, in She loves her little rabbit, the active zone and profile are the same (the rabbit). An active zone is generally different from a meaning facet. For example, in She held my hand against her cheek, the active zone of hand is the part of the speaker’s hand that touched the protagonist’s cheek; in terms of meaning facets, hand in the example above refers to the whole concrete physical entity (the extremity of one of a person’s two superior limbs), to be contrasted with a non-physical facet, as in He decided to try his hand at shooting. But it is possible to find examples where active zones and meaning facets seem to coincide, which is the case of We ate rabbit, where the meat qualifies as both the active zone and the meaning facet of rabbit. The reason for this degree of overlap between active zones and meaning facets is to be found in the fact that facets are highlighted subdomains

108 Francisco José Ruiz de Mendoza Ibáñez

of concepts, the same as active zones; however, facets are intrinsically secondary subdomains, while active zones can be either central or secondary. In the case of rabbit interpreted as ‘rabbit meat’, the meat is a highlighted secondary conceptual element (i.e., their edibility is a non-central property of rabbits in the sense that we can conceive a non-edible rabbit, but not a rabbit that has no size, shape, color, etc.); the meat is also the part of a rabbit that directly participates in the action of eating or its active zone. As discussed above, in the case of metonymy, cueing results in the reduction or the expansion of a conceptual domain. Reduction is achieved either by highlighting a secondary domain (thus giving rise to meaning facets) or a certain aspect of a primary domain (creating an active zone). In general, linguistic and textual cues serve as guides for the conceptual activation of relevant pieces of world knowledge, and are thus inextricably linked to another formal operation called selection. This formal operation lies at the base of what Fauconnier and Turner (2002) have labeled mental spaces. A mental space is an organized knowledge packet that we construct when we are involved in cognitive tasks, such as language production and comprehension. Thus, in She eats rabbit the verb eat cues for an edible object and in She wears rabbit the verb wears cues for an object that can be worn. The language user is thus led to select in ‘rabbit’ whatever conceptual structure meets these requirements. However, it must be noted that the selection task, even though cued, is not fully determined by the linguistic expression. There are other contextual factors that play a role, such as speaker’s beliefs and previous discourse tasks (cf. Chen, this volume). All in all, cueing is ultimately a matter of conceptual consistency: for a piece of conceptual structure to be cued and subsequently selected as relevant for a given cognitive task, it needs to be conceptually compatible with the cueing item. Because of this peculiarity, cueing also underlies the third type of formal operation we have distinguished: conceptual integration. As of the mid 1990s, conceptual integration has been studied in depth by Gilles Fauconnier and Mark Turner (see for example, Turner and Fauconnier, 1995, 2000; Fauconnier and Turner, 2002 inter alia; see Ruiz de Mendoza, 1998; Ruiz de Mendoza and Santibáñez, 2003; and Ruiz de Mendoza and Peña, 2005, for discussion and revision proposals). In our view, this formal operation, which consists in the combination and/or merging of conceptual structure from any number of cued items, is sometimes necessary for some low-level cognitive operations (e.g., conceptual correlation or resemblance; see 2.2 below) to take place. We will also postulate two forms of conceptual integration: (1) integration by enrichment and (2) integration by combination. The former takes place when a conceptual configuration incorporates into its internal make-up another configuration which is either inherently subsidiary to it or becomes subsidiary by virtue of the enrichment process. In He has more and



Metonymy and cognitive operations 109

more love inside him; in fact, he’s full of love, the explicitly invoked full–empty schema is enriched by the implicit verticality schema, which is necessary for the figurative quantity–height correlation to take place. The verticality schema is inherently subsidiary to full–empty (cf. Peña, 2003, 2008). In She was led into a depression, the conceptual structure of the path schema underlies the figurative motion of the protagonist into a certain state conceptualized as a location. The path schema is enriched by the container schema filling in the end-of-path structural slot, but the path and container schemas do not hold an inherent basic–subsidiary relationship. In the case of integration by combination the concepts to be initially combined are never subsidiary to one another. Integration is then made possible by calling upon selected structure from additional conceptual domains that will provide the necessary structural slots for the integration operation to be possible. A clear example is found in the metaphoric expression Her head is teeming with ideas, where the metaphoric source combines the image of a place crowded with animals with the container image-schema (contrast Her head was filled with ideas, where the animal element is absent). Ideas are often seen as objects contained in the head. In the example, the head figuratively contains animals or people moving around a lot. The concept of animals moving around in a limited area is external to the container–contents relation, but the two conceptual constructs have analogous structure (the animals are to the location as the contents are to the container). The existence of this analogous structure facilitates the combination process, where neither concept is subsidiary to the other. In grammar, a special case of integration by combination is usually called coercion (Michaelis, 2003), as in the case of the adaptation of the meaning and syntax (i.e., the constructional behavior) of talk, smile, and love in He talked me into business, She smiled me into her room, She loved him back into life. Ruiz de ­Mendoza and Mairal (2007) argue that the specific form of subcategorial manipulation that takes place in these uses is a matter of a high-level metaphoric resemblance operation whereby goal-directed activities or actions without a direct physical impact on the object (e.g., communicating, sensing) are seen as having that kind of impact. In coercion we force the insertion of one form of conceptual structure (e.g., a non-causal verbal predicate) into another form of conceptual structure (e.g., the caused-motion construction which in principle is only compatible with causal verbal predicates). As discussed in Ruiz de Mendoza and Díez (2002), metonymy often combines with metaphor in cognitively and communicatively productive ways. In contrast to previous analyses on metaphor–metonymy interaction (cf. Goossens, 1990), Ruiz de Mendoza and Díez (2002) distinguish four basic patterns: metonymic expansion or reduction of the metaphoric source and metonymic expansion or

110 Francisco José Ruiz de Mendoza Ibáñez

r­ eduction of the metaphoric target. That is, metonymic activity occurs either inside the source or the target domain of the metaphor. No use is made of conceptual structure external to either domain, which clearly points to the combination of metaphor and metonymy as a special case of integration by enrichment. To illustrate this idea, take an example of a paragon (Brdar, 2007: 111): (2) Humboldt is the Shakespeare of travelers – as much superior in genius to other travelers as Shakespeare to other poets.

In this view, a paragon involves a metonymic reduction of the source domain of a metaphor. In a paragon we look for an ideal example of a class of items and put it in analogy with another ideal example of a different class of items. Thus, in (2) Shakespeare is seen as the perfect embodiment of excellence in writing, and Humboldt as his analogue in the field of traveling. Since we are focusing on specific features of Shakespeare to the detriment of others, the metaphoric source undergoes domain reduction through highlighting (in this case, facetization). There is enrichment, since (2) does not require us to combine our knowledge about Shakespeare with any other item external to this concept (see Figure 1). Source

Target

Shakespeare as ideal poetry writer

Humboldt as ideal traveler

Superior ingenuity/skills in writing poetry

Superior ingenuity/skills in traveling

Writing poetry

Traveling

Goals as a poet

Goals as a traveler

Figure 1.  Humboldt is the Shakespeare of travelers

Finally, the fourth formal operation is abstraction. This operation has been dealt with by Fauconnier and Turner (2002), in the context of blending theory, in connection to metaphor. These authors have shown that metaphoric correspondences are possible because we are able to derive generic structure common to the source and target domains. Consider two examples: (3) a. There are several small farming hamlets strung out along the valley. b. Prices are soaring.



Metonymy and cognitive operations

In (3a) we have a resemblance metaphor: hamlets are spread out in a line as if they were on a string. The resemblance operation maps the image of stretched strings onto the overall shape of the hamlets as seen from a distance. The operation is possible because of the shared image-schematic structure between the imaginary line imposed by our perceptual systems on the spatial arrangement of the hamlets in the valley and that of a real string. However, contrast (3a) with (3b), where the metaphorical operation is one of correlation between two experiential domains: it maps height onto quantity on the basis of our everyday observations about the rise and fall of levels as more is added or taken away. Fauconnier and Turner (2002) have not discussed abstraction operations in the case of experiential correlation. In fact, it could be argued that, since there is no shared structure between source and target, abstraction plays no role in metaphors of this kind. However, the experiential correlation between quantity and height is itself the result of abstracting common structure from multiple low-level observations. Abstraction operations have been explained by Ruiz de Mendoza and Baicchi (2007) as a matter of the application of the high-level metonymy specific for generic to low-level cognitive models. Once abstract structure has been created it may be applied back to other models through the converse metonymy generic for specific, which readily explains the cognitive grounding of examples like (3b): low-level experience (e.g., prices increase) stands for high-level correlation (e.g., quantity–height) which in turn stands for other cases of low-level experience where the high-level correlation applies (e.g., interest grows/diminishes, income goes up/down). Thus, formal abstraction operations have a metonymic grounding and are preconditions for content (or low-level) metaphoric operations of correlation or resemblance.

2.2 Content operations In previous studies (Ruiz de Mendoza and Pérez, 2003; Ruiz de Mendoza and ­Santibáñez, 2003; Ruiz de Mendoza and Peña, 2005), content operations have been described as basic cognitive mechanisms that regulate inferential activity and can thus be distinguished from one another through their specific communicative impact. Content operations result in different kinds of metaphor, ­metonymy, hyperbole, among other so-called rhetorical figures, and their combinations. The preliminary list includes domain expansion/reduction for the case of metonymy, correlation and resemblance for metaphor, and mitigation for hyperbole. In this chapter, we will add a few more content operations (viz. echoing, strengthening, contrasting, parametrization, and saturation) to this list and we will examine them from the additional perspective of how they put conceptual domains in correspondence.

111

112 Francisco José Ruiz de Mendoza Ibáñez

We distinguish two basic forms of correspondence: a is b, and a for b. To the first kind belong the following operations: correlation, resemblance, strengthening, mitigation, echoing, and contrasting. The second kind features domain expansion, domain reduction, parametrization, and saturation. Let us describe each set of operations. a is b operations are those in which one conceptual domain is used to understand certain aspects of a different conceptual domain. These operations have been described in Cognitive Semantics in connection to metaphor as sets of correspondences between source and target domains where the source (b) allows us to reason, understand and talk about the target (a) (cf. Lakoff, 1993). For example, in affection is warmth, as in She gave me a warm embrace or He’s cold to his friends, which is a correlation metaphor, we talk about emotions in terms of body temperature (probably because body temperature is felt when people come close to us to show affection). Metaphorical correlation is grounded in experiential conflation (as discussed in Lakoff and Johnson, 1999), which consists in the mind envisaging two separate domains as if they were the same on the basis of continued co-occurrence in nature. Thus, we see change in terms of motion (e.g., She’s going from bad to worse) because we tend to correlate certain states with certain locations; e.g., being cool in the shade, warm in bed, safe at home. Or we see quantity in terms of height (e.g., Prices are soaring; World stocks have plummeted overnight) because levels rise and fall as quantity increases or decreases. Other metaphors are based on resemblance operations, as is the case of ­Achilles is a lion, where the behavioral attributes of the fierceness and aggressiveness of a warrior (the metaphoric target domain) are understood in terms of corresponding attributes in an animal (the metaphoric source domain). Or we can talk about rock-hard muscles on the basis of the similarity between the toughness of a rock (the source item) and that of someone’s muscles (the target). Resemblance operations naturally underlie simile, which differs from metaphor in how the resemblance operation is constrained constructionally. This point is evidenced if we compare the resemblance metaphor He is a shark with the two related similes He’s like a shark and He’s as fast as a shark in water. The metaphor is conventionally associated with the nature of sharks as aggressive, predatory creatures, that is, with behavioral features, as is often the case with people are animals in English (cf. She’s a cow for ‘nasty’; He’s a fox for ‘cunning’; He bulled through his demands for ‘pushed by force’). In contrast, the interpretation of the “like” simile is open to all kinds of features, whether behavioral or not (He’s like a shark can be interpreted as ‘He is fierce, aggressive, dangerous, strong, powerful, etc.’) and the “as X as” simile is restricted to just the feature or features that are made explicit linguistically. So, while “like” similes are unconstrained and “as X as” similes are heavily constrained, resemblance metaphors are somewhere in



Metonymy and cognitive operations 113

the middle, ­sometimes ­constrained by general cultural convention (the case of behavioral features in people are animals) or sometimes, if fully unconventional, as open as “like” similes. There are other a is b operations that are not metaphorical. Consider the following examples: (4) a. Nice day today! b. Yeah, right, Mary is an angel! (5) a. This suitcase weighs a ton! b. The lacerations inflicted on my client [said of minor wounds]. (6) a. They live some/a little distance from here. b. It’s just a scratch [said about a bad wound]. (7) a. I must be cruel to be kind. b. He’s a wise fool.

In (4a) and (4b) we have two different cases of what is commonly labeled irony. The ironical interpretation of (4a) requires contextual cueing that triggers off an echoing operation. Irony has been described as an echoic use of language in Relevance Theory. Here, we give echoing the status of a cognitive operation. Think of a context in which the speaker thought it would be a nice shiny day, but to his disappointment he finds it is pouring with rain. In this example, a is a state of affairs (the bad weather) that contradicts the speaker’s previous thoughts or, in other variants of this context, somebody else’s thoughts (the good weather) about the state of affairs; b, which echoes (i.e., imitates) such thoughts, is used to suggest that the situation is the exact opposite of the one that it describes, with the additional implication that the situation is annoying and the speaker feels he was derisively wrong. We could label the resulting mapping bad is good (i.e., we talk about a bad situation in terms of a good situation). The case of (4b) is slightly different since it combines irony and a resemblance metaphor. Here, the notion of goodness (in an extreme form) is represented by the notion of angel. Angels are culturally seen as celestial beings that possess miraculous abilities that they sometimes use to help people. Of course, the ironical component is not only cued linguistically (as marked by “Yeah, right”) but also contextually. In any case, there is irony only if it is mutually manifest to speaker and hearer that Mary is hideous. The examples in (5) and (6) illustrate the related operations of strengthening and mitigation, which act in converse ways depending on whether we take the speaker’s or the hearer’s perspective. In all these cases contextual cueing is largely necessary for interpretation. Let us first discuss (5a), which is a clear example of hyperbole. In the correspondence, a is a real-world suitcase that weighs too much for the speaker, while b is an imaginary suitcase that weighs a ton. In essence, the

114 Francisco José Ruiz de Mendoza Ibáñez

mapping leads us to understand the real suitcase and its impact on the speaker in terms of the impact the imaginary suitcase would have on the speaker. The speaker thus presents the hearer with an extreme case formulation of a scalar feature of a state of affairs; the hearer, on the other hand, needs to bring the formulation down the scale to a point that is compatible with his perception of the state of affairs. The resulting mapping can be labeled lower is upper. In (5b) we have a special case of hyperbole called auxesis. The mapping here is lowest is uppermost. In it, the seriousness of the state of affairs is maximized in a disproportionate way, which has the consequence of putting the credibility of the speaker at risk. In (6a), in turn, we have a clear example of understatement, with (6b) representing a special case called meiosis. Whether we have understatement or meiosis in the correspondence between domains, a is a point in scale and b is another point at a level lower than a. For understatement the mapping takes the form much is ­little or upper is lower and for meiosis uppermost is lowermost. In both cases, the speaker presents the hearer with a (moderately or extremely) understated formulation of a scalar feature of a state of affairs; for interpretation, the hearer needs to move the formulation up the scale to a point that is compatible with his perception of the state of affairs. The moderately understated expression (e.g., “a little distance”) can be interpreted by the hearer either as a way of minimizing the seriousness of a situation (here “a little distance” would mean ‘not as far away as you seem to believe’) or as a way of asserting more emphatically the opposite of what is said (i.e., ‘a long distance, farther away than you seem to believe’). Meiosis, in contrast, can only have the effect of minimizing the impact of the real situation. But it cannot convey the opposite of what it literally says precisely because of the disproportionate way in which minimization is carried out, which, paralleling the communicative potential of auxesis, poses a risk on the speaker’s credibility. Finally, (7a) and (7b) illustrate the way contrasting works. In the correspondence, a is the opposite of some aspects of b and b contains a. This situation applies to paradox and oxymoron. In (7a), which is an example of paradox, a cruel action is in fact kind if seen from a different perspective. In (7b), an oxymoron, a person’s generally foolish behavior can be seen as occasionally having surprisingly positive results that would be considered wise by many. Paradox has a propositional dimension, but oxymoron is restricted to a limited number of predicates. In both the hearer is required to construct a source domain where the opposed terms can be reconciled on the basis of different perspectives. We can describe this situation in terms of the mapping a form of behavior a is the opposite form of behavior b, where, as mentioned above, a is the opposite of some aspects of b and b actually contains a (e.g. a cruel action is in fact kind if seen from a different perspective). Interestingly, in this kind of mapping, two separate (in fact opposite) domains are conflated into one single domain that reconciles their ­discrepancies,



Metonymy and cognitive operations 115

which ultimately converts the a is b operation into a domain-internal one, which is the canonical relationship in a for b operations. Note that metaphorical correlation operations are also based on experiential conflation. The difference is that conflation in the case of paradox and oxymoron is triggered by the specific nature of the linguistic expression, while in correlation metaphors conflation is an experiential issue. We now turn our attention to a for b operations. The essence of these operations is the “stand for” relationship between a and b, which is possible if a has the ability to afford access to b when cued to do so on conceptual grounds (i.e., on the basis of information provided by the linguistic expression, by the context, or by both) or as a syntactic requirement (see the discussion on saturation below). So far, only metonymy has been discussed as an a for b operation (e.g., Radden and Kövecses, 1999; Ruiz de Mendoza, 2000; Barcelona, 2002), perhaps because it provides us with clear-cut cases of the relationship, which neatly contrasts with metaphor as a straightforward case of a is b. But we need to be careful about defining metonymy as merely a “stand for” relationship between two conceptual domains, since this relationship also takes place in referential metaphors: (8) a. My husband is a rat; he abandoned me. b. There goes the damned rat that betrayed me.

In (8a) “rat” is used predicatively in order to point to some of the behavioral features of the speaker’s husband. In (8b) “rat” maps metaphorically onto the person that acted in a treacherous way. But “rat”, in being used referentially rather than predicatively, is part of a complex definite description (“the damned rat that betrayed me”) that not only singles out a clearly defined referent, but also, in so doing, stands for the referent (the speaker’s husband). This use of metaphor is not essentially different from the best known use of metonymy, which has often been said to be mainly referential (cf. Lakoff and Turner, 1989; Croft, 1993), as in making a musical instrument stand for the person that plays the instrument (e.g., The sax has the flu and won’t come to today’s rehearsal) or a company for its workers (e.g., Marlboro has decided to challenge the new anti-smoking campaign). But there are other forms of metonymy that have been recognized in the literature. Ruiz de Mendoza (2000) suggests that “brain” in He is a real brain is a special predicative use of a double metonymy where the brain stands for good intellectual abilities that in turn stand for the person that has the brain with such abilities. Brdar and Brdar-Szabó (2003) discuss I’ll be brief (‘I’ll speak briefly’) as a case of predicational metonymy where manner of performing an action stands for the action; Panther and Thornburg (2004) give another example of predicational metonymy in the sentence General Motors had to stop production, where the necessity or obligation to stop production stands for the actual occurrence of stopping production.

116 Francisco José Ruiz de Mendoza Ibáñez

Finally, illocutionary metonymies have been explored in some detail in ­ anther and Thornburg (1998), Pérez and Ruiz de Mendoza (2002), and Ruiz de P Mendoza and Baicchi (2007). A straightforward example is the expression of a need by the speaker that stands for a direction for the addressee to satisfy the need (e.g., I’m hungry meaning ‘Please, give something to eat’). In all of these cases, the “stand for” relationship is central, but it is not necessarily a distinctive feature of metonymy, since this relationship is also a side-effect of the referential use of language. This is so even when we are concerned with metaphor, as in the “rat” example above. What is distinctive about metonymy is the special domaininclusion relationship between a, the source domain, and b, the target domain. Domain inclusion can go in either of two directions: one where a is a subdomain of b (a source-in-target metonymy) and another where b is a subdomain of a (target-in-source metonymy). In the former case we have a domain expansion cognitive operation, while in the latter we have a domain reduction operation. And, as mentioned above in connection to cueing, domain reduction is in fact the result of either meaning facetization or domain highlighting. There are some special cases of metonymy that underlie one further content operation: parametrization. Here, the hearer is presented with a vague characterization that has to be pinned down in context. The metonymy that applies in these cases is generic for specific. This is a high-level metonymy whereby a set of items stands for one of its subsets. We shall distinguish three areas of activity of this metonymy and the resulting parametrization process: cases of lexical genericity, propositional truisms, and semantically underdetermined expressions. Lexical genericity is a familiar phenomenon that has drawn the attention of some discourse analysts (e.g., Hoey, 1991) because it allows language users to create discourse cohesiveness while avoiding repetition of a lexical item, as is the case with the use of animal in (9) below to stand for the ailing dog: (9) I saw a whining dog; the poor animal was gravely injured.

But there is another more subtle sense in which a generic lexical item can stand for a more specific one. For example, think of the multiple senses of the adjective

. Koskela (this volume) claims that such cases, which she calls “vertical polysemy”, are not instances of metonymy, as here we are dealing with the “contiguity” of vertically-related conceptual categories, and not the contiguity of entities. However, as mentioned above, conceptual metonymic operations are not restricted to relations between concepts designating entities, but range over a variety of cognitive models, whether they are high/low level, or situational/nonsituational (cf. Ruiz de Mendoza, 2007). For further insights into the generic for specific metonymy, see Ruiz de Mendoza and Pérez (2001).



Metonymy and cognitive operations 117

good, as in a good person (‘kind’), good feelings (‘tender’), a good computer (‘highquality’), a good time (‘pleasant’), and a good life (‘virtuous, admirable’). Each of these senses is a cued interpretation of a vague characterization; thus, ‘good’ holds of any entity, situation or event that is positively assessed by the speaker. Or consider the following examples: (10) a. What’s that bird? (‘What kind of bird is that?’) b. You do (‘clean’) the carpet and I’ll do (‘wash’) the dishes.

Example (10a) has been drawn from Panther and Thornburg (2000) and example (10b) from Ruiz de Mendoza and Pérez (2001). In both there is parametrization of meaning: “what’s” in (10a) is not a question about the genus but about the species; in turn, each of the objects in (10b) cues the specific interpretation of the generic verb do as either ‘clean’ or ‘wash’. Propositional truisms are expressions whose meaning, if taken at face value, is that of an obvious or self-evident truth, which in principle makes them communicatively superfluous. Sometimes truisms are solved on the basis of pragmatic implication based on the (culture-driven) selection of different aspects of the same cognitive model, as in Boys will be boys, Kids are kids, or Business is business (cf. ­Wierzbicka, 1991: 391). In these examples each of the nominal predicates denotes stereotypical attributes or behavior of their corresponding subjects. The interpretation in each case is that the subject is not to be understood in any manner that deviates from the stereotype with the added implication that the addressee was erroneously doing so. Other truisms involve a “stand for” relationship. Consider the following: (11) a. Something has happened. b. Your daughter has a temperature. c. My father drinks.

Each of the sentences in (11) is a truism: events happen; people who are alive have a temperature; people drink. However, each of these sentences has a standard culturally-driven non-truistic interpretation that is obtained through parametrization in application of the high-level metonymy generic for specific: “something” means ‘something bad’; “having a temperature” is ‘having a higher-than-normal body temperature’; “drinks” refers to ‘habitually drinks alcohol’. (11c) may be considered a special case of generic for specific, since “drinks” calls for a parametrization as to aspect, which takes place through the metonymy present for ­habitual (Radden and Kövecses, 1999: 33). Other aspectual parametrizations of the present tense are achieved through complementary lexical and grammatical cues that further exploit the present for habitual metonymy:

118 Francisco José Ruiz de Mendoza Ibáñez

(12) a. Your father drinks and drinks (iterative). b. Your father drinks non-stop (continuous). c. Your father seldom/often/always drinks (frequency of habit).

In (12a) the conjoined repetition of the verbal predicate drink cues for more specific parametrization of the metonymic target, in such a way that the action is construed as a very strong habit that takes place over and over again. The reason why conjoined repetition suggests iteration is a matter of iconicity (cf. Otal and Ruiz de Mendoza, 2007): literally, in the expression of habitual iteration in (12a) drinking is followed by further drinking to the exclusion of other possible actions (cf. Your father drinks and gambles). Of course, since we know that drinking is not the only possible activity of a human being, iteration expressed in this manner is taken as hyperbolic, ultimately mitigated into ‘Your father drinks too frequently’. In the mapping we use the impossible state of affairs in which a person does nothing but drink to refer to the actual state of affairs in which a person is too fond of drinking alcohol, thus suggesting that the situation is undesirable. Sentence (12b) is similarly critical of the protagonist’s regular habit. But the parametrization mechanism of the idea of habit is different: here, the adverbial expression “non-stop” is used to indicate continuity of the action every time it takes place. It brings to mind the implausible picture of a person drinking without even pausing to breathe much less to do something else. Again, we have a mapping from an implausible to a real situation and the accompanying implications about its undesirability. Finally, by way of contrast to what is the case with (12a) and (12b), in (12c) habit is directly parametrized as to frequency through explicit grammatical mechanisms, without the mediation of cued implications. The third area of activity of the generic for specific metonymy is semantically underdetermined expressions: (13) a. My sister has [exactly/more than] three children. b. There were [approximately/exactly/over] 3000 demonstrators. c. John and Mary went to the museum [together/separately].

Examples like these and also like those in (11) above have been treated by Carston (2002), within the framework of Relevance Theory, in terms of a form of pragmatic adjustment called free enrichment where the target meaning is a conceptually (i.e., not mandated linguistically) richer proposition than the source meaning. Interestingly, this view is consistent with our own since in free enrichment the denotation of the pragmatically adjusted expression is a subset of the initial denotation. This is another way of saying that meaning is parametrized as licensed by the generic for specific metonymy. But the relevance-theoretic account fails to recognize the existence of this and other cognitive operations that underlie



Metonymy and cognitive operations 119

and thus place constraints on pragmatic adjustment (cf. Ruiz de Mendoza, 2002, 2005, for further discussion on the limitations of a purely pragmatic explanation of phenomena like these). Saying that an expression is adjusted is not much if we are not told in what way adjustment is possible. For example, it would not explain why (12a) expresses iteration, which, remember, is prompted by the iconic value of the conjoined repetition of the verbal predicate. Nor would it account for the fact that speakers make use of lexical genericity, truisms, and semantically underdetermined expressions liberally because in their minds the vague expressions, given the right contextual conditions, can afford access and thereby stand for their “enriched” counterparts. The last operation that we have identified is saturation. Saturation or completion is the label used in relevance pragmatics (cf. Récanati, 1989; Bach, 1994; Sperber and Wilson, 1995; Carston, 2002) to identify a very well known linguistic phenomenon. Utterances may have incomplete and expanded versions. The context of situation provides language users with the conceptual material that is used to expand an utterance into a fully processable form. For any piece of conceptual material to be used in this function, it needs to be syntactically compatible and semantically coherent with the incomplete expression. Saturation occurs in two cases: (1) when we have a constructionally underdetermined expression, as is the case with constructions that can dispense with the prepositional object (e.g., be good for, be enough for, be ready for, be ready to, finish + ing), which has to be supplied from the context (cf. John’s not good enough [for an executive position]; (2) with minor clauses (subsentential utterances): [Have a] nice day!, [Good] morning!, [Have you] seen Fred? Evidently, saturation is different from parametrization in two ways. First, it is a constructional phenomenon, while parametrization is conceptual. Second, saturation is not supported by the generic for specific metonymy, which involves metonymic reduction. Instead, saturation involves expanding the initial schematic representation, that is, in the a for b relationship, a is a part of b. However, saturation is not to be confused with domain expansion, since saturation is cued constructionally.

3.

Conclusion

Our discussion of cognitive operations has two consequences for the topic of the present chapter. First, it has allowed us to see metonymy as a cover term for two more basic cognitive operations, domain expansion and domain reduction, the latter of which can be further broken down into other two processes. Domain ex-

120 Francisco José Ruiz de Mendoza Ibáñez

pansion occurs in cases of metonymy where a concept is just a point of access to a more complex conceptual representation that contains the concept and for which the concept stands. On the other hand, domain reduction is based on two kinds of highlighting process, viz. facetization and zone activation, which reduce the scope of the initial point of access. Second, our discussion has provided us with a paradigm of contrasts that has allowed us to define metonymy not only internally as a “stand for” (or a for b) conceptual relationship that takes place within a conceptual domain, but also externally, that is, in contrast to a for b content operations with constructional rather than conceptual grounding and with all content operations of the a is b kind. We have also been able to explore the way in which metonymic operations are supported by formal operations such as cueing, selection, and integration. These operations act as pre-requisites for content operations, but they affect metonymy in a different way. Thus, cueing underlies conceptually motivated expansion and reduction operations, which are typical of metonymy, and the selection of the right amount and kind of conceptual structure to make expansion and reduction communicatively relevant. Finally, metonymy is sometimes subservient to metaphor by integrating either with its source or with its target domain. In cases like this, the metonymy-based formal integration operation becomes a pre-requisite for the metaphorical mapping to take place.

References Anderson, John R. 1983. The Architecture of Cognition. Cambridge, MA: Harvard University Press. Bach, Kent. 1994. Conversational implicature. Mind & Language 9(2): 124–162. Barcelona, Antonio. 2002. Clarifying and applying the notions of metaphor and metonymy within Cognitive Linguistics: An update. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (207–277). Berlin & New York: Mouton de Gruyter. Brdar, Mario. 2007. Metonymy in Grammar: Towards Motivating Extensions of Grammatical Categories and Constructions. Osijek: Josip Juraj Strossmayer University. Brdar, Mario and Rita Brdar-Szabó. 2003. Metonymic coding of linguistic action in English, Croatian and Hungarian. In Panther, Klaus-Uwe and Linda Thornburg (eds.), Metonymy and Pragmatic Inferencing (241–266). Amsterdam & Philadelphia: John Benjamins. Carston, Robyn. 2002. Thoughts and Utterances: The Pragmatics of Explicit Communication. Oxford: Basil Blackwell. Croft, William. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4(4): 335–370. Fauconnier, Gilles and Mark Turner. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books.



Metonymy and cognitive operations 121

Goossens, Louis. 1990. Metaphtonymy: the interaction of metaphor and metonymy in expressions for linguistic action. Cognitive Linguistics 1(3): 323–340. Hoey, Michael. 1991. Patterns of Lexis in Text. Oxford: Oxford University Press. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: University of Chicago Press. Lakoff, George. 1993. The contemporary theory of metaphor. In Ortony, Andrew (ed.), Metaphor and Thought, 2nd ed. (202–251). Cambridge, MA: Cambridge University Press. Lakoff, George and Mark Johnson. 1999. Philosophy in the Flesh. New York: Basic Books. Lakoff, George and Mark Turner. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: University of Chicago Press. Langacker, Ronald W. 1984. Active Zones. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 12: 455–471. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford, CA: Stanford University Press. Langacker, Ronald W. 2000. Grammar and Conceptualization. Berlin & New York: Mouton de Gruyter. Langacker, Ronald W. 2009. Metonymic grammar. In Panther, Klaus-Uwe, Linda Thornburg and Antonio Barcelona (eds.), Metonymy and Metaphor in Grammar. Amsterdam & Philadelphia: John Benjamins. Mairal Usón, Ricardo and Francisco José Ruiz de Mendoza Ibáñez. 2009. Levels of description and explanation in meaning construction. In Butler, Christopher and Javier Martín Arista (eds.), Deconstructing Constructions (153–198). Amsterdam & Philadelphia: John Benjamins. Michaelis, Laura A. 2003. Word meaning, sentence meaning, and syntactic meaning. In ­Cuyckens, Hubert, René Dirven and John R. Taylor (eds.), Cognitive Approaches to Lexical Semantics (93–122). Berlin & New York: Mouton de Gruyter. Nunberg, Geoffrey and Annie Zaenen. 1992. Systematic polysemy in lexicology and lexicography. In Tommola, H., K. Varantola, T. Tolonen and J. Schopp (eds.), Proceedings of Euralex 2 (387–398). University of Tampere. Otal Campo, José Luis and Francisco José Ruiz de Mendoza Ibáñez. 2007. Modeling thought in language use: At the crossroads between discourse, pragmatics, and cognition. Jezikoslovlje 8(2): 115–167. Panther, Klaus-Uwe and Linda L. Thornburg. 1998. A cognitive approach to inferencing in conversation. Journal of Pragmatics 30: 755–769. Panther, Klaus-Uwe and Linda L. Thornburg. 2000. The effect for cause metonymy in English grammar. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (215–232). Berlin & New York: Mouton de Gruyter. Panther, Klaus-Uwe and Linda L. Thornburg. 2004. The role of conceptual metonymy in meaning construction. Metaphorik (electronic journal) http://www.metaphorik.de/ Journal/. Paradis, Carita. 2004. Where does metonymy stop? Senses, facets and active zones. Metaphor and Symbol 19(4): 245–264. Peña Cervel, Sandra. 2003. Topology and Cognition: What Image-Schemas Reveal about the Metaphorical Language of Emotions. Münich: Lincom Europa. Peña Cervel, Sandra. 2008. Dependency systems for image-schematic patterns in a usage-based approach to language. Journal of Pragmatics 40(6): 1041–1066.

122 Francisco José Ruiz de Mendoza Ibáñez

Pérez Hernández, Lorena and Francisco José Ruiz de Mendoza Ibáñez. 2002. Grounding, semantic motivation, and conceptual interaction in Indirect Directive Speech Acts. Journal of Pragmatics 34(3): 259–284. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought (17–59). Amsterdam & Philadelphia: John Benjamins. Récanati, François. 1989. The pragmatics of what is said. Mind and Language 4: 294–328. Ruiz de Mendoza Ibáñez, Francisco José. 1998. On the nature of blending as a cognitive phenomenon. Journal of Pragmatics 30(3): 259–274. Ruiz de Mendoza Ibáñez, Francisco José. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads (109–132). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José. 2002. From semantic underdetermination, via metaphor and metonymy to conceptual interaction. Theoria et Historia Scientiarum, An International Journal for Interdisciplinary Studies 6(1): 107–143. Ruiz de Mendoza Ibáñez, Francisco José. 2005. Linguistic interpretation and cognition. In Croitoru, E., D. Tuchel, and M. Praisler (eds.), Cultural Matrix Reloaded: Romanian Society for English and American Studies. Seventh International Conference (36–64). Bucarest: Didactica Si Pedagogica. Ruiz de Mendoza, Francisco José. 2007. High level cognitive models: in search of a unified framework for inferential and grammatical behavior. In Kosecki, Krzysztof (ed.), Perspectives on Metonymy (11–30). Frankfurt am Main: Peter Lang. Ruiz de Mendoza Ibáñez, Francisco José and Annalisa Baicchi. 2007. Illocutionary Constructions: Cognitive motivation and linguistic realization. In Kecskés, István and Laurence R. Horn (eds.), Explorations in Pragmatics:  Linguistic, Cognitive, and Intercultural Aspects (95–128). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José and Olga I. Díez. 2002. Patterns of conceptual interaction. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (489–532). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José and Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Radden, Günter, Klaus-Michael Köpcke, ­Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (33–51). ­Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco José and Ricardo Mairal Usón. 2008. Levels of description and constraining factors in meaning construction: An introduction to the Lexical Constructional Model. Folia Linguistica 42(2): 355–400. Ruiz de Mendoza Ibáñez, Francisco José and Sandra Peña Cervel. 2005. Conceptual interaction, cognitive operations and projection spaces. In Ruiz de Mendoza Ibáñez, Francisco José and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (254–280). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza, Francisco José and Lorena Pérez Hernández. 2001. Metonymy and the grammar: motivation, constraints, and interaction. Language and Communication 21: 321–357. Ruiz de Mendoza Ibáñez, Francisco José and Lorena Pérez Hernández. 2003. Cognitive operations and pragmatic implication. In Panther, Klaus-Uwe and Linda Thornburg (eds.), Metonymy and Pragmatic Inferencing (23–50). Amsterdam & Philadelphia: John Benjamins.



Metonymy and cognitive operations 123

Ruiz de Mendoza Ibáñez, Francisco José and Francisco Santibáñez Sáenz. 2003. Content and formal cognitive operations in construing meaning. Italian Journal of Linguistics 15(2): 293–320. Sperber, Dan and Deirdre Wilson. 1995. Relevance. Communication and Cognition, 2nd ed. Oxford: Basil Blackwell. Talmy, Leonard. 2000. Toward a Cognitive Semantics. Vol 1. Cambridge: The MIT Press. Turner, Mark and Gilles Fauconnier. 1995. Conceptual integration and formal expression. Metaphor and Symbolic Activity 10(3): 183–204. Turner, Mark and Gilles Fauconnier. 2000. Metaphor, metonymy and binding. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads (133–145). Berlin & New York: Mouton de Gruyter. Wierzbicka, Anna. 1991. Cross-Cultural Pragmatics: The Semantics of Human Interaction. Berlin & New York: Mouton de Gruyter.

Metonymy, category broadening and narrowing, and vertical polysemy* Anu Koskela University of Sussex

This chapter examines the relationship between metonymy and cases of category broadening and narrowing and the resulting state of vertical polysemy (e.g., cat ‘domestic cat’ > ‘any feline’ and drink ‘consume liquid’ > ‘consume alcohol’). Broadening and narrowing have been argued to be motivated by metonymic processes where a category member stands for the whole category or vice versa (Radden and Kövecses, 1999; cf. also Lakoff, 1987). Here, I show that there is a crucial difference between the domain structures involved in metonymy and in vertical polysemy. Unlike metonymies, broadening and narrowing do not involve a shift in the salience of domains (see Croft, 1993). Instead, I argue that there are four possible domain configurations that may underlie vertically related meanings. Keywords: broadening, domain, narrowing, vertical polysemy

1.

Introduction

In Cognitive Linguistics metonymy is generally seen as a fundamental cognitive process where one conceptual entity affords access to another closely associated one. However, the precise definition of metonymy remains in some ways elusive: not only in terms of issues such as whether it is better defined in terms of mappings (e.g., Lakoff and Johnson, 1980; Lakoff and Turner, 1989), domain highlighting

* I would like to thank the audience and discussants of the “What do we want to call a metonymy” theme session at the ICLC 2007 conference in Krakow for their comments on an earlier version of this chapter. Aspects of this chapter have also benefited from comments and discussion with Lynne Murphy, Vyvyan Evans, and Alan Cruse. My thanks also go to the editors of this volume for their helpful feedback and comments. All errors are, of course, my own.

126 Anu Koskela

(Croft, 1993, 2002) or conceptual contiguity (e.g., Feyaerts, 1999; Peirsman and Geeraerts, 2006), but also in terms of the phenomena that may be considered instances of metonymy. For example, the meaning alternation of book, whereby it can designate either the physical entity or its textual contents, is treated as metonymic by some authors, but not all (compare, e.g., Geeraerts and Peirsman, this volume; Croft, 1993; Koch, 2001; Paradis, 2004; Ruiz de Mendoza, 2000). The present chapter, for its part, aims to contribute to the definitional discussion by considering how metonymy relates to semantic broadening and narrowing and the resulting state of multiplicity of meaning – what I here call vertical polysemy. While cases of vertical polysemy have often been treated as being motivated by or as instances of metonymy (e.g., Kövecses and Radden, 1998), other authors have regarded categorial shift as a fundamentally different kind of phenomenon (e.g., Koch, 2001; Nerlich and Clarke, 1999; Seto, 1999, 2003). In this chapter I argue in favor of the latter view, by firstly noting that the entities involved in metonymic and categorial shifts are of ontologically different types, and consequently the contiguity or associative relationship between the elements in the two phenomena is not of the same type. In this I concur in particular with Seto (1999, 2003), who has argued that viewing the taxonomic relationship between categories as metonymic is based on a metaphorical understanding of categories either in terms of containers or part–whole relations. But in addition, I consider cases of vertical polysemy from the perspective of Langacker’s (1987) domain-based encyclopedic model of meaning and show that there is a crucial difference between the conceptual configurations involved in metonymy and in vertical polysemy. In his domain-based account of metaphor and metonymy, Croft (1993) argues that metonymy typically involves a shift in the salience of two domains that form part of the domain matrix that the “literal” meaning is profiled against. In contrast, I argue that category broadening and narrowing do not involve a shift in the salience of domains. Instead, a broader meaning arises as a result of a process of abstraction, while a narrower meaning is construed through elaboration. To see how these processes relate to profile–base . Croft’s approach has been criticised in the literature, particularly with respect to its ability to adequately distinguish between metonymy and metaphor (see, e.g., Barcelona, 2000; ­Feyaerts, 1999; and Haser, 2005). One of Feyaerts’ (1999) criticisms centres on the difficulties in determining objectively the structure of domain matrices. While such difficulties are acknowledged and discussed by Croft (1993, 2002; Croft and Cruse, 2004), he refutes the idea that the structures of domain matrices are completely subjective and can vary significantly from one person to another (Croft, 2006). With this in mind, while the purpose of the current chapter is not to give a detailed explanation or defence of the domain-matrix approach to semantic description and metonymy and metaphor, I suggest below some criteria that may be used to determine their structure.



Metonymy and vertical polysemy 127

organization and domain structures, I examine how the vertical relationship between a broader and a narrower meaning may be modeled in general. I propose that the theoretical constructs of the Langackerian domain-based view of meaning point to four possible domain configurations that may, in principle, underlie vertically-related meanings. For instance, in some cases the broader and narrower meanings are profiled against the same domains, while in others the narrower meaning presupposes an additional domain that is not essentially presupposed by the broader meaning. I begin in Section 2 by providing a brief definition and examples of vertical polysemy. Section 3 then reviews some of the literature on the relationship of metonymy and vertical polysemy. In Section 4 I develop a domain-based account of vertical polysemy, while Section 5 provides some more discussion of the relationship between vertical polysemy and metonymy, and the extent to which they involve similar conceptual processes.

2.

Vertical polysemy

The term vertical polysemy is used here to refer to cases where a single word form has the potential to designate multiple categories that are in a relationship of categorial inclusion. Such a word is therefore polysemous with a broader and a narrower sense that occupy different levels in a taxonomic hierarchy. In taxonomic diagrams, inclusion is commonly depicted on the vertical axis, which motivates the label vertical for this type of polysemy. Another term for vertical polysemy, autohyponymy (Horn, 1984), alludes to the fact that a vertically polysemous word is effectively its own hyponym. An oft-mentioned example of vertical polysemy is dog, which can refer to domesticated canines in general, or more specifically to male canines, in contrast with the meaning of bitch (see, e.g., Kempson, 1980). Cat is also vertically polysemous: in addition to its narrower ‘domestic cat’ sense, it also has a broader ‘any felid’ sense: (1) All cats except the tiger and cheetah can climb.

Other vertical polysemes include the verb drink, ‘consume liquid’/‘consume alcohol’ and finger, whose meaning can either include or exclude thumbs. In a similar fashion, the meaning of shoe can either include or contrast with that of boot (see also Rohdenburg, 1985 for more similar examples). Generic uses of brand names (e.g., hoover, aspirin) can also be treated as leading to vertical polysemy. . Source: http://www.lookd.com/cats/family.html.

128 Anu Koskela

Although this kind of meaning variation is frequently treated as amounting to vagueness or univocality (see, e.g., Becker, 2002; Wilson and Carston, 2007), all of the above examples are regarded here as cases of polysemy. This position stems from a view of meaning where word forms are seen as being associated with “meaning potentials”, rather than fixed sets of senses. As such, senses are construed in the context of word use (see, e.g., Allwood, 2003; Croft and Cruse, 2004; Geeraerts, 1993; Moore and Carling, 1982 and also Paradis, this volume, for similar proposals). Under this view, polysemy may be defined as involving a situation where a single lexical form has the potential to designate different readings in a way that may lead to ambiguity. This can be seen in (2), where shoe in A’s question is potentially ambiguous (depending on the season and the weather): between a broader (“boot-including”) and a narrower (“boot-excluding”) sense, as B’s two possible responses illustrate. This is here taken to indicate that shoe is contextually vertically polysemous. (2) A: Are you going to wear shoes? B1: Yes, I’m not going out barefoot! B2: No, I’ll wear my boots.

Although senses are considered to be construed in the context of word use, some senses can still be said to be conventional or established to the extent that the knowledge activation patterns involved in their construal are entrenched (in the sense of Langacker, 1987) in the meaning potential of the word form. It is plausible that this is the case for both the broader and narrower senses of shoe and for those of dog and drink, insofar as the senses are used frequently enough to become entrenched. Notably, however, the contextual definition of polysemy adopted here means that the term vertical polysemy covers not only more established cases, but also what Koch (2001) calls ad-hoc innovations. Consider, for instance, the use of book in (3) to designate ‘hardback books’, in contrast with its more conventional broader sense that also includes paperback books:

. Although these readings of shoe are viewed as distinct senses, under the current approach not all word senses are equally distinct or autonomous. Indeed, many vertically-related senses also exhibit some symptoms of unity, and are therefore less-than-fully-distinct senses (compare Croft and Cruse’s [2004] discussion of “facets” and “microsenses” – other types of sense units lacking full autonomy). One symptom of the unity of the senses of shoe is the fact that the narrower sense represents a prototypical subset of items of footwear in general (as designated by the broader sense). As such, it is possible to unite the boot-excluding and boot-including readings of shoe under a single prototype category construal of the category shoe (see Lehrer, 1990, for a similar point about the relationship of cup and mug).



Metonymy and vertical polysemy 129

(3) Binding Books and Paperbacks

We can speak of vertical polysemy as arising through broadening to the extent that the narrower of the senses is the more established one – or initially was so, in cases of conventional vertical polysemes whose senses are now equally established. An example of this might be the case of cat ‘domestic cat’ > ‘any felid’. In narrowing, the opposite applies: the broader of the senses is (or initially was) more strongly established (e.g., the case of drink). Meaning shifts between broader and narrower categories have sometimes been viewed as being motivated by metonymy (or as being a type of metonymy). The rationale behind this view is explored in the next section.

3.

Metonymy

Traditional definitions of metonymy generally characterize it as involving some kind of stand-for relationship between contiguous entities, such as the container and its contents in I drank the whole bottle, where the bottle stands for the liquid in the bottle. In cognitive linguistic accounts of metonymy, the notion of contiguity or association in metonymy is understood in conceptual terms. Thus many definitions of metonymy view it as involving conceptual elements that form parts of some coherent conceptual complex and that are associated with each other within that complex. This view is represented, for example, by Lakoff and Turner’s (1989) definition of metonymy as a conceptual mapping that links two entities within a single domain in a stand-for relationship. Radden and Kövecses (1999) likewise regard metonymy as involving conceptual entities within a single idealized cognitive model (although they do not view it as involving a mapping as such). And as noted above (and discussed in more detail in Section 4.1 below), within a Langackerian domain-based view, metonymy operates between two conceptual domains that are part of the same domain matrix (Croft, 1993, 2002). There is also a long history of seeing vertical polysemy as being motivated by metonymy. This can be related to the classical treatment of meaning transfers between a broader and a narrower category as a subtype of synecdoche, together with meaning transfers between a whole and its part (e.g., hand for human . A heading from a bookbinder’s website. Source: http://www.acmebook.com/bindery/ library/paperbacks. . Peirsman and Geeraerts (2006), on the other hand, avoid defining metonymy in terms of conceptual complexes, and instead see metonymy as involving a relationship of conceptual contiguity.

130 Anu Koskela

being­ in All hands on deck!). As Nerlich and Clarke (1999) note, while classical rhetoric distinguished synecdoche from metonymy, in later research (e.g., ­Ullmann, 1957) synecdoche was subsumed under metonymy. This makes intuitive sense to the extent that the relationship between a part and a whole can be seen as a kind of metonymic or associative relationship. For example, a hand, which is a part of a human body, is associated with the body and spatially contiguous with it. Cognitive linguistic work has also often followed this line of thinking – Lakoff and Johnson (1980) and Kövecses and Radden (1998), for example, classify meaning transfers between parts and wholes under metonymy, while Peirsman and Geeraerts (2006) even view the part–whole relationship between bounded spatial entities as the most prototypical kind of metonymic contiguity. But perhaps as a result of subsuming part–whole synecdoches under metonymy, metonymy has sometimes also been extended to cover cases of categorial shift. Kövecses and Radden (1998), for example, assert that the metonymy category for a member of the category motivates the narrowing of the pill to refer specifically to the contraceptive pill. Presumably the same metonymy could also then be said to motivate cases such as the narrowing of drink ‘consume liquid’ > ‘consume alcohol’ and even dog ‘canine’ > ‘male canine’. The reverse metonymy, member of a category for the category is viewed as motivating cases of broadening, such as aspirin ‘an Aspirin painkiller’ > ‘any painkiller’ (and, I assume, also e.g., cat ‘domestic cat’ > ‘any felid’ or cow ‘female bovine’ > ‘any bovine’). A slightly different metonymic account of categorial shifts is suggested by Peirsman and Geeraerts (2006). They propose that broadening, for instance, can either involve the metonymy individual for countable collection or entity for characteristic, depending on whether categories are viewed from an extensional or intensional perspective. In a somewhat similar vein, Lakoff ’s (1987) account of prototype effects also relies on viewing the relationship between a more specific and a more general category as a metonymic one. Lakoff argues that one source of prototype effects is metonymically structured cognitive . See also Benczes (this volume), who argues that the member for category metonymy plays a role in noun–noun compounds such as couch potato. According to her, in this compound potato (the main ingredient of potato chips) stands for any junk food that may be eaten while watching TV. . Note, however, that Peirsman and Geeraerts (2006: 308) see categorial shifts as being “metonymical only in a more or less derived sense”, insofar the metonymic contiguity is only apparent if you first conceptualise a category either as a set of entities or a set of properties – cf. the discussion of metaphorical views of categories below.



Metonymy and vertical polysemy 131

models. Thus robins are prototypical birds and housewife-mothers stereotypical mothers because the cognitive models that underlie the categories bird and mother are structured metonymically, with a specific subcategory standing for the whole category. Lakoff (1987) is naturally right in noting that certain properties are more expected of a category (see, e.g., Cruse, 1986) and some category members are more salient than others. But as Murphy (1996: 193) notes, it is not entirely clear whether this relationship between categories and their prototypical or stereotypical members really is an instance of a metonymic relationship; whether it involves the same kinds of cognitive process and conceptual entities as metonymies such as container for contents or producer for product. One problem with viewing the relationship between broader and narrower categories as metonymic is that such a move blurs the distinction between part– whole and taxonomic relations. That is, it amounts to the view that a subcategory is contiguous with and may stand for a superordinate one (or vice versa) in the same way as a container is contiguous with and may stand for its contents. However, as Seto (1999, 2003) points out, this view is based on a metaphorical conceptualization of categories: either as containers for the category members (see Lakoff, 1987) or in terms of wholes, with the members constituting the parts that make up the category (see Kövecses and Radden, 1998). Seto (2003), however, argues that although these conceptualizations are very appropriate and useful for a folk understanding of the nature of categories and taxonomies, they are harmful for a theoretical treatment of metonymy and the kind of categorial shifts that may result in vertical polysemy. As Seto (1999) observes, there is a crucial ontological difference between the entities involved in metonymy and those involved in categorial shift. Metonymy operates between two real-world entities – or rather, concepts thereof – while shifts in denotation between more inclusive and more restrictive categories operate between conceptual categories (see also Koch, 2001 and Nerlich and Clarke, 1999). Therefore, although categorial relationships can be conceived metaphorically in terms of containers and their contents or wholes and their parts, the “contiguity” of vertically-related conceptual categories is not the same as the contiguity of entities (relative to experiential domains in our conceptualization). This fundamental difference between conceptualized entities and conceptual categories can also be expected to be reflected in the conceptual structures involved in metonymy and vertical polysemy. This is the point I turn to in the next section, which considers metonymy and vertical polysemy from the perspective of the domainbased encyclopedic view of meaning.

132 Anu Koskela

4.

Metonymy and vertical polysemy in encyclopedic semantics

Central to Langackerian encyclopedic semantics is the idea that the meaning of a linguistic form consists of a profile and a base – the profile being the entity designated by the linguistic form and the base the conceptual material necessarily presupposed by the profile. Thus the profile of the meaning of banana, for instance, is a conceptualization of a kind of fruit, with particular physical characteristics, uses for human consumption, tropical origin and so on. The base of a profile typically includes specifications in multiple domains: for banana, domains such as shape, size, colour, eating and tropical plants. The multiple domains evoked by a profile collectively form a domain matrix. Within a domain matrix, each domain that provides the base for the characterization of some conceptualization may itself presuppose some other conceptual domain(s). Domains are therefore organized into hierarchies on the basis of chains of presuppositional relations, which culminate in basic domains that embody primitive ideas that are not defined in terms of any other knowledge. Thus, for instance, the domain of shape in the domain matrix for banana presupposes the basic domain of space. Furthermore, the domains that are presupposed by a profile also vary in their prominence within the domain matrix so that we may speak of primary and secondary domains. This notion of more central and more peripheral domains within a domain matrix forms the basis of Croft’s (1993) domain-based account of metonymy.

4.1 Metonymy in encyclopedic semantics According to Croft (1993), metonymy can be characterized in terms of domain highlighting insofar as many metonymies involve a shift in the salience of domains within the domain matrix against which the “literal” meaning of the form is profiled. For example, in (4) the name of the author metonymically stands for her works. (4) I got the students to read Wierzbicka for the seminar.

The domain matrix for Wierzbicka the person includes a domain characterizing her as a human being, but, given the encyclopedic view of meaning, it also includes a secondary domain against which our knowledge that she writes academic books and articles on linguistics is profiled. As Figure 1 shows, the metonymy operates by backgrounding the domain of human being and highlighting the domain of academic writing.



Metonymy and vertical polysemy 133

Wierzbicka the person

Human being

A paper by Wierzbicka

Academic writing

Human being

Academic writing

Figure 1.  Simplified domain matrix structures for the readings of Wierzbicka. The filledin ovals represent the profiles and the unfilled ovals the domains presupposed by the profile. Domains outlined in grey are secondary or backgrounded

Croft (1993) acknowledges that in some metonymies the highlighting of domains can be quite subtle. For example, in metonymies where a body part stands for a person, the body part presupposes and is therefore profiled against the whole human being. Although such metonymies appear to operate between a profile and its base, Croft argues that body part for human being metonymies do also involve domain highlighting. He notes that the domain structure of any body part term not only includes human being as a salient domain, but also another domain that characterizes the function of the body part in question. In the case of the hand for human being metonymy, this would mean that the domain matrix of hand includes a domain characterizing the salient role that hands have in human activities and the metonymy would then highlight this domain of manual activity.

4.2 Vertical relations in encyclopedic semantics Metonymic meaning shifts can be contrasted with cases of broadening and narrowing. Broadening essentially involves the construal of a more inclusive sense that lacks some of the knowledge specifications of a narrower sense associated with the same linguistic form. In Langacker’s (1987) terms, this process may be

. Compare Kövecses (2002: 208–209), who posits that the hand for person metonymy is based on another metonymy, hand for activity (e.g. put one’s hands in one’s pockets, ‘deliberately do nothing’). The hand for activity metonymy provides the motivation for the hand for person metonymy insofar as a prototypical person is an active one.

134 Anu Koskela

called abstraction, as it involves focusing on and abstracting certain properties in order to construe the broader category (e.g., cat ‘domestic cat’ > ‘any felid’). Conversely, in narrowing, the more restrictive sense encodes additional, richer specifications compared to the broader sense; in Langacker’s (1987) terms, it elaborates on its superordinate (e.g., dog ‘canine’ > ‘male canine’). To get an idea of how broadening and narrowing relate to domain matrices, we must consider more generally how word meanings at different hierarchical levels are represented. A key point stressed by Croft (1993) is that a more restrictive concept is not profiled against a more inclusive one. Instead, there is a principled distinction between taxonomic (i.e., vertical) relationships and the presuppositional relationship of a profile and a base. Thus domestic cat, for instance, would not be profiled against felid, its superordinate in a biological taxonomy, or even the more distant superordinate animal. This is because the profile–base relation is defined in terms of presupposition: the understanding of the profiled entity presupposes the knowledge structures of the base. But it is not necessary to understand the more abstract biological notion of the family Felidae or even animal in order to understand the concept domestic cat. Indeed, to argue that domestic cat is profiled against its superordinate would amount to the claim that the understanding of a basic-level concept presupposes a comprehension of a superordinate category. But this claim would be problematic given that basiclevel concepts appear to be the first to be acquired by children and the first to be lexicalized in a language (Berlin et al., 1973; Rosch et al., 1976). In fact, as Clausner and Croft (1999) point out, the canonical profile–base relation is the part–whole relation, rather than the type–token relation. This is illustrated by Langacker’s (1987: 183–184) example of arc, which is profiled against circle – the whole that an arc is a part of. In other cases the profile–base relation is grounded in some other relationship of contiguity – island, for example, necessarily presupposes the notion of the surrounding water (Taylor, 2002: 198). But if a concept doesn’t designate something that is a part of a whole or doesn’t take part in some other contiguity relationship, identifying its base is somewhat less straightforward. A biological kind like domestic cat is not part of a whole, nor is there a clear contiguous entity against which it could be profiled. In such cases the base may be viewed as consisting of general knowledge structures that afford the profiled concept its coherence. For example, Clausner and Croft (1999: 7) argue that [o]ur commonsense knowledge about birds for example includes their shape, the fact that they are made of physical material, their activities such as flying and eating, the avian lifecycle from egg to death, etc. These aspects of the concept bird are specified in a variety of different domains such as space, physical objects, life, time, etc.



Metonymy and vertical polysemy 135

Such domains may be viewed as contributing to a complex abstract domain of animate beings that broadly corresponds to the notion of a folk theory of animate life-forms in the so-called Theory Theory of concepts (see, e.g., Medin, 1989; Medin and Ortony, 1989; Murphy and Medin, 1985). However, if bird is regarded as being profiled against the domain animate beings, this may appear to contradict the earlier point about concepts not being profiled against their superordinates, given that a bird clearly is an animate being. However, the label animate beings should not be regarded as referring to the conceptual category of animate beings, but rather as shorthand for a complex domain that encompasses a basic and even inchoate understanding of the nature of animate beings as physical objects endowed with life. A similar point is made by Croft (1993), who notes that shape, as the base for circle, is not the taxonomic superordinate concept “a shape”, but rather a conceptualization of the property of shape. If more specific concepts are not profiled against more general ones, what kinds of domain structures can then underlie vertically-related senses? Croft (1993) argues that more general concepts are profiled against the same domains as their subordinates. In such cases the superordinate represents certain properties more schematically. For example, triangle and polygon are both profiled against the shape domain, but while polygon includes the notion of sides, it does not specify their number, unlike its subordinate triangle (Langacker, 1987: 133–135). There are also other ways in which a subordinate concept may encode more specific properties. According to Langacker (1987: 133), a subordinate concept may be specified for additional domains that are not essentially evoked by its superordinate. It also appears that in some cases the super- and sub-ordinate are profiled against different domains. Croft (1993) considers two meanings of the phrasal verb fill up: the general ‘make something full’ and the more specific ‘make the fuel tank of a vehicle full of fuel’ sense. He argues that the base domains of the two senses are presuppositionally related, as the base of the narrower sense itself presupposes the base of the broader sense (see Section 4.2.3 below for details). ­ Vertically-related senses may also presuppose different domains that are

. Having said that, it can sometimes be difficult to distinguish between profile–base and taxonomic relationships. For example, Croft (1993) maintains that the domain of writing presupposes the domain of human communication. However, he notes that while he prefers to view writing as something that can only be understood relative to human communication, it could alternatively be seen as a subtype of human communication. But although the distinction between taxonomic and profile–base relations can often be difficult to draw, it is nevertheless an important distinction to make in principle.

136 Anu Koskela

t­ hemselves in a vertical relationship. For example, Clausner and Croft (1999) argue that the domains of eating and drinking (liquid) may be viewed as ­instances of a more general domain consuming. This suggests that the meaning of eat is profiled against eating (a domain that incorporates the knowledge of the consumption of food and nutrition by animate beings), while the meaning of consume presupposes the more schematic consuming. In summary, a review of the literature suggests that any of the following four configurations of domains and profiling can underlie a vertical relationship between senses: 1. The broader and narrower senses presuppose the same domains, but the broader sense embodies more schematic specifications than the narrower sense. 2. The broader and narrower senses presuppose the same domains but the narrower sense includes a core specification in an additional domain. 3. The broader and narrower senses are profiled against different domains, but the base domain of the narrower sense presupposes the base domain of the broader sense. 4. The broader and narrower senses are profiled against different domains, but the base domain of the narrower sense is a subtype of the base domain of the broader sense. Each of these types is illustrated in the following subsections, with cases of vertical polysemy. However, it is worth first noting that one of the challenges for domain-based encyclopedic semantics is the difficulty of obtaining direct evidence of the domain structures that underlie linguistic meaning. While some linguistic reflexes of particular profile–base relationships have been suggested in the literature, such diagnostics cannot be consistently used for all types of concepts. For instance, ­Taylor (2002: 196) argues that the possibility of linking hypotenuse and triangle with of or have in (5) indicates that triangle functions as the immediate base of hypotenuse. In contrast, Taylor considers the oddity of the sentences in (6) to be due to the fact that triangle is only indirectly profiled against twodimensional­ space. (5) a. the hypotenuse of the triangle b. the triangle has a hypotenuse (6) a. #the triangle of two-dimensional space b. #two-dimensional space has a triangle

However, these of and have “tests” for profile–base relations and similar proposals by Langacker (1987: 119) are only appropriate in cases where the profile and its base are in the canonical part–whole relationship. For instance, if triangle is in fact directly profiled against shape, we would expect #the triangle of shape and #Shape has a triangle to be acceptable, but they are not – because triangle is



Metonymy and vertical polysemy 137

not part of shape. Given this, I suggest other criteria, such as the significance of cultural factors, to justify particular analyses below.10

4.2.1 Narrower and broader sense profiled against the same domains The first type of domain configuration is illustrated by finger, which can either mean ‘one of the hand digits’ or, more narrowly, ‘one of the four non-opposable hand digits (exclusive of the thumb)’. Given that a finger (in either sense) is part of a hand, both senses are profiled against hand. Because the profile–base relationship is in this case based on the part–whole relation, the fact that both senses of finger share the same base can also be demonstrated by the acceptability of both of the sentences in (7) (cf. (5b) above): (7) a. A hand has five fingers. [‘fingers’ in the broader sense] b. A hand has a thumb and four fingers. [‘fingers’ in the narrower sense]

The broader sense of finger is schematic with respect to the position of the digit relative to the hand, but the narrower sense specifies that the digit is not positioned at the side of the hand (like the thumb). The domain configurations of the two senses are shown schematically in Figure 2.

Hand digit

Non-opposable hand digit

Hand

Hand

Figure 2.  Both the broader and narrower senses of finger are profiled against the same base domain, hand 10. I have also suggested elsewhere (Koskela, 2007), that the different domain configurations may correlate with different types of inclusion relationships, in particular taxonymy and nontaxonymic simple hyponymy (Cruse, 2002). For example, Siamese and kitten are both subordinates of cat, but while Siamese is a specific breed of cat, kitten is not a breed but a subclass defined by maturity. These inclusion relations are different: cat > Siamese is a taxonymic relation, while cat > kitten is a simple hyponymous relation. I argue that in taxonymy, the superordinate and subordinate are typically profiled against the same domain(s), while one of the other domain configurations is likely to be involved in non-taxonymic simple hyponymy. The details of this suggestion are, however, beyond the scope of the present chapter.

138 Anu Koskela

The case of cat ‘domestic cat’/‘any felid’ can also be viewed as an instance of this type. Both senses of cat designate a biological kind and therefore necessarily presuppose the domain of animate beings, a complex domain that presupposes the domains of life and physical objects. According to Croft (1993), the domain physical objects itself presupposes basic domains such as matter, location, shape and size. The narrower sense of cat is quite richly specified for characteristics such as size, shape, coloration and habitat, while the broader one is more schematic in these respects. 4.2.2 Narrower sense profiled against an additional domain Langacker’s model of meaning posits that any aspect of encyclopedic knowledge may be included as part of a linguistic form’s semantic value. Consequently, more specific concepts will often presuppose domains that are not part of the domain matrix of their superordinates. For instance, the narrower ‘domestic cat’ sense of cat may include as part of its semantic value the idea that cats like playing with balls of wool. This idea would presuppose at least the domain of wool, which would be unlikely to form part of the domain matrix of the broader ‘any felid’ sense of cat. But wool is in fact also a rather marginal part of the domain matrix of the ‘domestic cat’ sense. On the other hand, there are cases where a subordinate sense saliently and necessarily presupposes an additional domain. An example of this is dog. Both the ‘canine’ and ‘male canine’ senses are again profiled against the domain of animate beings, but the ‘male canine’ sense is additionally specified for the sex of the animal. The property of maleness requires an understanding of the notion of sex and reproduction, and therefore the ‘male canine’ sense of dog is additionally profiled against the domain of sex, as Figure 3 illustrates.11 Although the specification of sex may also be marginally evoked by the broader ‘canine’ sense, it does not form a salient or necessary part of its characterization. The idea that the ‘male canine’ sense of dog involves an additional domain, as opposed to profiling more specific properties in the domain of animate ­beings, may need justifying. Croft (1993) suggests that a conceptual structure may be characterized as a distinct domain to the extent that it acts as the base for a substantial number of concepts. The domain of sex seems to fulfill this criterion, as it would be presupposed by any sex-specific terms and terms relating to sexual relations and reproduction.

11. Note that because sex is a biological property of animals, sex in fact also presupposes an understanding of animate beings (W. Croft, p.c.).



Metonymy and vertical polysemy 139

Male canine

Canine

Animate beings

Physical objects

Animate beings

Life

Physical objects

Sex

Life

Figure 3.  Partial domain matrices for the vertically-related senses of dog

4.2.3 Profiling in different domains, one presupposing another Croft’s (1993) example of the vertically-related senses of fill up, ‘make something full’ and ‘make the fuel tank of a vehicle full of fuel’ illustrates the third type of domain configuration. According to Croft, the two senses are profiled in different domains: the broader sense against the domain of substances and containers, and the narrower one against the domain of fuelling. The fuelling domain presupposes the domain of substances and containers, as well as the domain of fuel-requiring mechanical objects, as illustrated in Figure 4. Cow provides another case where the base domain of the narrower sense presupposes that of the broader one. In addition to its specific ‘mature female bovine’ sense, cow can in some contexts be used in a broader, less technical sense that includes any bovine animal, regardless of sex or maturity. This broader ‘bovine’ sense designates a biological kind and is therefore again profiled against animate beings. The narrower sense of cow, on the other hand, contrasts with bull, calf, and heifer and therefore specifies the properties of sex, age, and whether the animal has given birth or not. One possibility would be to treat the polysemy of cow as that of dog, by assuming that the ‘mature female bovine’ sense is profiled against animate beings plus the additional domains of sex, age, and birth giving. However, such an analysis would overlook the fact that understanding the ‘mature female bovine’ sense of cow and its contrast with the other bovine terms presupposes knowledge of the differing roles mature and young animals of different sexes have biologically and agriculturally. This sense of cow – and all the

140 Anu Koskela

Make the fuel tank of a vehicle full of fuel

Make something full

Substances and containers

Shape

Fuelling

Substances and containers

Fuel-requiring mechanical objects

Shape

Figure 4.  The domain matrices of the vertically-related senses of fill up: the narrower sense is profiled against a domain that presupposes the base domain of the broader sense

other terms for domesticated bovines – can therefore be viewed as being profiled against a complex domain of animal husbandry, which presupposes the domain of animate beings as well as the domains of sex, age, and birth giving and notions such as meat, milk, and other concepts relevant to agriculture. In contrast with the bovine terminology, the distinction between male and female dogs is mainly about biological sex-subclasses, and so there is less justification for assuming that the ‘male canine’ sense of dog is profiled against a specific domain that characterizes the role of male dogs in cultural activities. Another reason for regarding animal husbandry as a distinct domain is that it acts as the base for all the bovine terms (among many other agricultural concepts). In contrast, a knowledge structure relating to the roles of male and female dogs would seemingly only characterize the ‘male canine’ sense of dog and the meaning of bitch.



Metonymy and vertical polysemy 141

4.2.4  Profiling in different domains, one a subtype of another In the final type of domain configuration the vertically-related senses are also profiled against different domains but, unlike in the previous type, one of the domains is a more specific subtype of the other and therefore the domains are themselves in a vertical relationship. One such vertical polyseme is love, which can either refer to strong affection or to strong romantic affection.12 The broader sense is profiled against the domain of human relations, which presupposes domains such as human being and emotion. The ‘romantic affection’ sense of love, however, is profiled against a more specific subtype of human relations: romantic relations (which would, for its part, presuppose the domain of sex, among other things). The general domain configurations of the senses of love are shown in Figure 5. Strong romantic affection

Strong affection

Human relations

Emotions

Romantic relations

Emotions

Subtype of Human relations

Figure 5.  Partial domain matrices for the senses of love

12. It is also possible to regard the different kinds of love as contrasting categories – i.e., ‘romantic love’/‘non-romantic love’. But love does also have a broader, superordinate meaning, e.g., in There are many people who love you – your boyfriend, your parents and all your friends.

142 Anu Koskela

One issue for this type of domain configuration is how to justify the existence of the more specific domain, as distinct from its superordinate domain. An alternative analysis for love would be to say that the broader and narrower senses profile more and less schematic properties, respectively, relative to the same domain of human relations. However, what motivates the distinctness of the subordinate domain romantic relations is the experiential and cultural significance of romantic and sexual relationships for us as humans. Furthermore, a number of other terms, such as marriage, girlfriend/boyfriend, and dating, specifically refer to romantic or sexual relations and their meanings would therefore plausibly presuppose a specific domain of romantic relations.

5.

Vertical polysemy and metonymy: Discussion

The preceding sections considered the domain configurations that may underlie the broader and narrower senses of vertical polysemes. The aim of this discussion was to allow for a comparison between the domain structures that underlie the senses of vertical polysemes and those of metonymically-related senses. While metonymy involves a shift in the salience of two domains within a domain matrix, vertical polysemy emerges as a result of elaboration or abstraction, which may involve any of the four domain configurations discussed above. That is, the narrower sense may elaborate on the broader one relative to the same base domain (e.g., the narrowing of finger to contrast with thumb), introduce specifications that are profiled against an additional domain (as in the case of dog ‘male canine’) or require profiling against a different domain (e.g., fill up). The reverse applies in cases of broadening – some specifications are schematicized, either in relation to the same base domain (e.g., cat ‘any felid’), or else the broader sense makes reference to fewer domains or to a different domain that is either more general or more basic (e.g., the broadening of cow to ‘any bovine’). Vertical polysemy therefore involves different kinds of domain configurations from metonymy and should not be treated simply as a type of metonymy. However, it is possible that more general conceptual highlighting (and backgrounding) processes similar to those involved in full metonymy play a role in abstraction and elaboration. Croft (1993) argues that domain highlighting also operates in other semantic processes that are not instances of metonymy proper. For him, metonymy must be accompanied by a shift of reference to a different entity, and thus the meaning alternation of book (‘physical tome’/‘textual contents’) does not amount to metonymy, even though it does involve a metonymic-type conceptual highlighting process. This is consistent with the view proposed by Cruse (e.g., 2000), who argues that meaning of book has two facets, ‘tome’ and ‘text’, which



Metonymy and vertical polysemy 143

constitute less than fully distinct sense units. The idea that metonymic-type processes range from full metonymy to more subtle conceptual highlighting processes is also put forward by Paradis (2004; see also Paradis, this volume). She argues that fully-fledged metonymies operate between two different concepts, while facet phenomena involve the highlighting of a part of a concept (one qualia structure within a concept).13 Even more subtle and fine-grained conceptual highlighting occurs in active zone phenomena. For example, in I need to sharpen my pencil, the tip of the pencil is the active zone, the part of the pencil that is focused on. The facet-alternation of book and the zone activation of pencil thus involve the highlighting and backgrounding of certain conceptual parts. Vertical polysemy differs from these phenomena in that it involves conceptual categories rather than concepts for contiguous objects or parts of objects. But it is also possible to characterize the processes of elaboration and abstraction in terms of highlighting and backgrounding specific properties – that is, even more fine-grained conceptual structures than the intra-conceptual elements involved in facet-alternation or zone activation. For instance, abstraction essentially involves the drawing of attention to some characteristics of the category while ignoring others. Thus in construing the generic ‘any pain killer’ sense of the brand name Aspirin, the medical use of the branded substance is highlighted, and other properties that relate specifically to the branded medicine (including the fact that it is manufactured by a specific company) are backgrounded.14 Similarly, elaboration involves highlighting particular attributes that are to be specified more richly. This can in particular be seen in cases where vertical polysemy emerges through the narrowing of a superordinate term to contrast with a subordinate term. In such cases the narrowed sense is specifically defined by highlighting the properties that create the contrast with the other lexical item. Thus, for example, in narrowing the meaning of dog to contrast with bitch, the property of sex, rather than any other possible property associated with canines, is highlighted and elaborated on, against the

13. The view that facets fall short of full metonymy is disputed by Geeraerts and Peirsman (this volume), on the grounds that facets and metonymies exhibit similar grammatical, pragmatic and semantic properties. Space constraints unfortunately prevent a detailed discussion of these points here, but suffice to say that I would in any case consider facets to form a rather restricted class of phenomena. For instance, unlike the ‘tome’ and ‘text’ readings of book, the ‘building’ and ‘people’ readings of department are less representative of facets and instead closer to full metonymy (cf. the discussion in Croft and Cruse, 2004 of “near neighbours of facets”). 14. A somewhat similar account of “paragon” names such as That young playwright is a real Shakespeare is given by Barcelona (2004), although he treats the highlighting of specific properties of paragons as a characteristic properties for individual metonymy, rather than a metonymic-type conceptual highlighting.

144 Anu Koskela

additional domain of sex. Similarly, in the narrowing of book to contrast with paperback in (3) above, particular physical properties of books are highlighted to be elaborated, to create the contrast with paperbacks. A similar account could be given for the case of shoe and boot. Conceptual highlighting and backgrounding processes that are similar to those involved in metonymy, but do not amount to metonymy proper, can therefore be viewed as being of vital importance in the flexibility of meaning in general, including the development of vertical polysemy. But the subtlety of such processes may be lost if categorial shifts and vertical polysemy are simply considered a subtype of metonymy.

6.

Conclusion

In the preceding I argued that not only are the kinds of entities related by metonymy different from those involved in vertical polysemy but, viewed from the perspective of Langackerian encyclopedic semantics, the domain structures that underlie the two types of phenomena are also different. By recognizing that vertical polysemy is not straightforwardly a type of metonymy, we can provide more sophisticated analyses of the nature of metonymy and the related conceptual highlighting processes that are involved in sense broadening and narrowing.

References Allwood, Jens. 2003. Meaning potentials and context: Some consequences for the analysis of variation in meaning. In Cuyckens, Hubert, René Dirven, and John R. Taylor (eds.), Cognitive Approaches to Lexical Semantics (29–66). Berlin: Mouton de Gruyter. Barcelona, Antonio. 2000. Introduction: The cognitive theory of metaphor and metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (1–28). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2004. Metonymy behind grammar: The motivation of the seemingly “irregular” grammatical behavior of English paragon names. In Radden, Günter and Klaus-Uwe Panther (eds.), Studies in Linguistic Motivation (357–374). Berlin: Mouton de Gruyter. Becker, Thomas. 2002. Autohyponymy: Implicature in lexical semantics, word formation, and grammar. Journal of Germanic Linguistics 14: 105–136. Berlin, Brent, Dennis E. Breedlove and Peter H. Raven. 1973. General principles of classification and nomenclature in folk biology. American Anthropologist 7: 214–242. Clausner, Timothy C. and William Croft. 1999. Domains and image schemas. Cognitive Linguistics 10: 1–31. Croft, William. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4: 335–370.



Metonymy and vertical polysemy 145

Croft, William. 2002. The role of domains in the interpretation of metaphors and metonymies. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (161–205). Berlin: Mouton de Gruyter. Croft, William. 2006. On explaining metonymy: Comment on Peirsman and Geeraerts, “Metonymy as a prototypical category”. Cognitive Linguistics 17: 317–326. Croft, William and D. Alan Cruse. 2004. Cognitive Linguistics. Cambridge: Cambridge University Press. Cruse, D. Alan. 1986. Lexical Semantics. Cambridge: Cambridge University Press. Cruse, D. Alan. 2000. Aspects of the microstructure of word meanings. In Ravin, Yael and Claudia Leacock (eds.), Polysemy: Theoretical and Computational Approaches (30–51). Oxford: Oxford University Press. Cruse, D. Alan. 2002. Hyponymy and its varieties. In Green, Rebecca, Carol A. Bean, and Sung Hyon Myaeng (eds.), The Semantics of Relationships: An Interdisciplinary Perspective (3–21). London: Kluwer. Feyaerts, Kurt. 1999. Metonymic hierarchies: The conceptualization of stupidity in German idiomatic expressions. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (307–332). Amsterdam: John Benjamins. Geeraerts, Dirk. 1993. Vagueness’s puzzles, polysemy’s vagaries. Cognitive Linguistics 4: 223–272. Haser, Verena. 2005. Metaphor, Metonymy, and Experientialist Philosophy: Challenging Cognitive Semantics. Berlin: Mouton de Gruyter. Horn, Laurence R. 1984. Toward a new taxonomy for pragmatic inference: Q-based and R-based implicature. In Schiffrin, Deborah (ed.), Meaning, Form, and Use in Context: Linguistic Applications (11–42). Washington: Georgetown University Press. Kempson, Ruth M. 1980. Ambiguity and word meaning. In Greenbaum, Sidney, Geoffrey Leech, and Jan Svartvik (eds.), Studies in English Linguistics (7–16). London: Longman. Koch, Peter. 2001. Metonymy: Unity in diversity. Journal of Historical Pragmatics 2: 201–244. Kövecses, Zoltán. 2002. Metaphor: A Practical Introduction. Oxford: Oxford University Press. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9: 37–77. Koskela, Anu. 2007. Vertical polysemy: A Study in the Dynamicity of Meaning. PhD thesis, University of Sussex, UK. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago: Chicago University Press. Lakoff, George and Mark Turner. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. Lehrer, Adrienne. 1990. Prototype theory and its implications for lexical analysis. In Tsohatzidis­, Savas L. (ed.), Meanings and Prototypes: Studies in Linguistic Categorization (368–381). London: Routledge. Medin, Douglas L. 1989. Concepts and conceptual structure. American Psychologist 44: 1469– 1481. Medin, Douglas L. and Andrew Ortony. 1989. Psychological essentialism. In Vosniadou, Stella and Andrew Ortony (eds.), Similarity and Analogical Reasoning (179–195). Cambridge: Cambridge University Press.

146 Anu Koskela

Moore, Terence and Christine Carling. 1982. Understanding Language: Towards a PostChomskyan Linguistics. London: Macmillan. Murphy, Gregory L. and Douglas L. Medin. 1985. The role of theories in conceptual coherence. Psychological Review 92: 289–316. Murphy, Gregory L. 1996. On metaphoric representation. Cognition 60: 173–204. Nerlich, Brigitte and David D. Clarke. 1999. Synecdoche as a cognitive and communicative strategy. In Blank, Andreas and Peter Koch (eds.), Historical Semantics and Cognition (197–213). Berlin & New York: Mouton de Gruyter. Paradis, Carita. 2004. Where does metonymy stop? Metaphor and Symbol 19: 245–264. Peirsman, Yves and Dirk Geeraerts. 2006. Metonymy as a prototypical category. Cognitive Linguistics 17: 269–316. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought (17–59). Amsterdam: John Benjamins. Rohdenburg, Günther. 1985. Dogs, bitches and other creatures. Journal of Semantics 4: 117–135. Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray, David M. Johnson, and Penny BoyesBraem. 1976. Basic objects in natural categories. Cognitive Psychology 8: 382–439. Ruiz de Mendoza, Francisco. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (109–132). Berlin & New York: Mouton de Gruyter. Seto, Ken-ichi. 1999. Distinguishing metonymy from synecdoche. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (91–120). Amsterdam: John Benjamins. Seto, Ken-ichi. 2003. Metonymic polysemy and its place in meaning extension. In Nerlich, Brigitte, Zazie Todd, Vimala Herman, and David D. Clarke (eds.), Polysemy: Flexible Patterns of Meaning in Mind and Language (195–214). Berlin & New York: Mouton de Gruyter. Taylor, John R. 2002. Cognitive Grammar. Oxford: Oxford University Press. Ullmann, Stephen. 1957. The principles of semantics. Oxford: Blackwell. Wilson, Deirdre and Robyn Carston. 2007. A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts. In Burton-Roberts, Noel (ed.), Pragmatics (230–259). London: Palgrave.

Metonymy at the crossroads A case of euphemisms and dysphemisms* Tanja Gradečak-Erdeljić and Goran Milić Josip Juraj Strossmayer University of Osijek

The chapter uses two small-scale studies to examine the effects of metonymy in political discourse. We re-examine some theoretical constructs proposed as definitional of metonymy, particularly pragmatic function (Barcelona, 2003a) and inclusion of the source and target in the same functional domain (ibid.). By analyzing the dysphemism chickenhawk as an instantiation of the metaphoric mapping humans are animals and the metonymic mapping part of a scenario for the whole scenario in the euphemism body count, we delineate the possibilities and limitations of metonymic inferencing (Panther and Thornburg­, 2003). The observed differences in the pragmatic effect of euphemistic and dysphemistic expressions ascribed to metonymy reveal its differing foregrounding and backgrounding force. Keywords: dysphemism, euphemism, functional domain, metonymic inferencing, pragmatic function

1.

Introduction

The common traits of euphemisms and dysphemisms are their obvious reliance on a cognitively and pragmatically conditioned network of functioning roles, and * The authors gratefully acknowledge the support of the Croatian Ministry of Science, Education and Sports in funding the research reported in this chapter (Grant no. 122-130149-0606, Cognitive Linguistic Approach to Polysemy in Croatian and Other Languages). . Euphemism or “sweet talking” (Greek eu “good, well”+ phēmē “speaking”) is recognised if “…the interpreter perceives the use of some word or expression as evidence of a wish on the part of the speaker to denote some sensitive phenomenon in a tactful and/or veiled manner.” Warren (1992: 135). Dysphemisms (Greek dys- “bad, unfavourable”+ phēmē “speaking”) which are, as pointed out by Allan and Burridge (2006: 24) motivated by fear and distaste, hatred or contempt, include curses, name-calling and derogatory comments, and are quite frequently used by different political groups in order to disparage their political opponents.

148 Tanja Gradečak-Erdeljić and Goran Milić

their obvious directionality in pointing at certain features in the target domain via the source domain. This is done with the clear intention of creating the following pragmatic effects: the hiding of less favorable elements in the target concept in euphemisms and the clear exposition of these in dysphemisms. The nature of the euphemistic and dysphemistic taxonomic organization of concepts they designate is inevitably heterogeneous, that is, heavily dependent on the social structure and cultural influences from different sources and along historical lines. The subject matter they conceptualize is primarily emotionally marked, and thus even more emphatically imprinted on the conceptual system. As Kövecses (2000) suggests, an extremely complex interplay of body, emotions, language and culture must be thus assumed in order to understand why the language of emotions succeeds in entering all types of discourse in different cultural backgrounds. Furthermore, the link between language and ideology as “integrated systems of values and beliefs” (Dirven et al., 2001: 2) and thus an intrinsic object of study in linguistics, can hardly be overlooked. A number of volumes (Dirven et al., 2001, 2003) have argued for the implementation of conceptual tools of analysis from previous studies in Cognitive Linguistics for the purpose of studying ideological workings in language. In this vein, our study attempts to examine the workings and effects of metonymy by taking a broader view at political discourse as a particular discourse type. There is ample evidence (Dirven, 2005; Morgan, 2001) for the claim that a discourse type both creates and hinges on background information organized as a specific Idealized Cognitive Model (icm). The shared pool of conceptually relevant data makes up a pragmatically shaped (usage- and context-based) network (Fauconnier’s [1997] shared frame of experience), which serves as a basis for extracting contextually relevant metonymical or pragmatic function mappings. Even on the level of the shared frame of experience, we find evidence of bidirectionality. Each time a mapping is established, a check is made against the discourse background and a positive or negative image is created in order to evaluate the possible outcome of the pragmatic effect the metaphorical or metonymical links may create. It will be argued that these tactics are particularly prominent in political discourse, where the import of linguistic material is a predecessor to all possible future actions for which it sets the appropriate scene. The chapter is organized as follows. After a brief introduction into the notion of metonymy adopted for the present purposes, Section 2 focuses firstly on the backgrounding force metonymy has in the process of encoding euphemisms . E.g., gay, which retrograded from its early Middle English sense of “cheerful, full of fun” to one with sexual allusions, occasionally even dysphemistic ones, in the Modern English period (cf. Burridge [2005: 58ff.]).



Metonymy at the crossroads 149

and its foregrounding influence in dysphemistic expressions. These relationships are linked to the notion of the so-called conceptual iconicity (Gradečak-Erdeljić, 2005), which may be operative in the process of structuring meaning. Putting aside the obvious play on words referring to the collection of seminal papers on the relationship between metaphor and metonymy edited by Barcelona (2000), we shall offer a slightly adjusted perspective on the opposite directions, and sometimes opposing effects metonymy may produce by using the above-mentioned foregrounding and backgrounding force. Metonymy is thus at the crossroads of a subjectively assessed state of affairs, and how its role will be judged depends on the participants. Section 3 will present our two representative case studies, which include a quick glance at some of the euphemisms found in the discourse of politics, as well as a more detailed dissection of the dysphemism chickenhawk in U.S. political discourse. The specific formal nature of the latter and its conceptual underpinnings also make it a fruitful ground for a re-examination of a number of theoretical constructs proposed as definitional of metonymy, the most notable being that of the pragmatic function (Barcelona, 2003a) and inclusion of the source and target in the same functional domain (ibid.). Finally, in Section 4 we briefly readdress certain central issues proposed in the previous section by presenting some comparative findings from the Croatian National Corpus with particular reference to journalistic, i.e., political discourse.

2.

Metonymy in euphemisms and dysphemisms

In modern political discourse, in what Gibbs (1994: 140) calls “a figurative nature of political thought”, metaphor and metonymy are an indispensable cognitive tool for making abstract and complex concepts, that language-wise tend to be quite elusive, more accessible to the general public and sometimes even to the politicians themselves. With the abstract contents of political discourse, the ratio of communicational precision is always at stake: the implicature and inference link is very rarely direct and straightforward, and the processing of messages transferred is highly susceptible to metonymies as natural inference schemas (Thornburg and Panther, 1997). We see those inference schemas as a network of pragmatically triggered links between the source and target concepts, which activate particular functions necessary to establish the domain shared by the source and the target. All of these characteristics are commonly considered to take place in what ­Barcelona (2002) terms the functional cognitive domain (or icm, or frame Barcelona­ equates it to). Besides the taxonomic ordering of elements in the domain, both/all interlocutors should be aware of all the factors influencing the

150 Tanja Gradečak-Erdeljić and Goran Milić

meaning ­ structuring and restructuring where, as Barcelona (2005: 97) further established: “Metonymy thus seems to constitute the very skeleton of pragmatic inferencing” (emphasis as in the original). In the process of meaning structuring and restructuring, it is assumed that both the speaker and the listener share at least some of the structure of the icm, which should facilitate the inferential process of the communication event. In the process of construing the tabooed concepts, the icm created seems to be structured in such a way that the taboo elements of the icm are placed at the peripheral part of the construed event scenario. The euphemistically encoded elements in the event are metonymically highlighted but with a clear backgrounding intention, i.e., the inferencing process is prolonged in order to achieve a longer period of a face-saving situation. Barcelona (2003a: 233) states the following: “The metonymic source projects its conceptual structure onto that of the target, not by means of a systematic matching of the counterparts, but by conceptually foregrounding the source and by backgrounding the target” (emphasis as in the original). We find that the notions of foregrounding and backgrounding compared to that of highlighting (Croft, 1993) or metonymic zooming in and out of the domain matrix suggested by Brdar and Brdar-Szabó (this volume) are not at all incompatible. Thus, when considering the “lightning speed” (Barcelona, 2003b) of cognitive processing, we find that simultaneous activation of several domains may happen, so that both the source and target domains may be activated and reactivated in a very short period of time. We also find that different, pragmatically motivated elements serve as switch points for a particular direction of the metonymic mapping that may ensue. These elements are either functions of source and domain concepts in a metonymic mapping, e.g., the author in the author for work metonymy, or a purely contextual basis, e.g., awareness of the political jargon as a discourse type. Thus “mapping” may be preserved as a term in a sense of guideposts for the thinking process with its pragmatic function, which is very frequently shaped by specific metonymic chains (cf. Barcelona, 2005). To further elaborate on the workings of metonymy in the case of euphemistic and dysphemistic expressions, we can also resort to the notion of conceptual iconicity (Gradečak-Erdeljić, 2005), which as a term may cover the processes triggered by the use of metonymic expressions within certain euphemistic and dysphemistic structures (see Figure 1):

. For a very detailed discussion on domains, domain matrices and domain networks, see Benczes (this volume).



Metonymy at the crossroads 151

Content

Distance

Speaker

Distance

Listener

Figure 1.  The relationship of conceptual iconicity

In Figure 1, there is a bidirectional relationship between the speaker and the content of the message, as well as between the speaker and the listener. In both of these relationships a certain degree of conceptual iconicity arises, where the distance between the peripheral parts of the whole scenario temporally postpones the moment when the idea is grasped and reflects the intended situation. Due to the fact that the moment of understanding the message was postponed, the listener is spared from the suddenness of the shock upon hearing bad news. Simultaneously, the speaker is spared the proverbial messenger’s faith by implying his/her distancing from the content of the negative message import. On the other hand, as we will argue below, dysphemisms aim at the most obvious readings of metaphorical and metonymic mappings. There is no intention on the part of the speaker to distance himself/herself from the possible negative impact and the level of conceptual iconicity is preserved.

3.

Case studies

3.1

The euphemism body count

As pointed out in the introduction, we approach the process of meaning encoding and decoding in terms of Lakoff ’s (2006) framing or establishing the socalled deep frames which both determine and reflect the underlying conceptual structures connected to one’s moral worldview or political philosophy. They are structures that employ certain metaphorical models in order to derive the surface or lexical frames as working metaphors that are activated in particular situations (issue-defining­ frames). Two such major models, the strict father and the nurturant parent model, will prove crucial in our analysis of the dysphemism

. With euphemisms and dysphemisms the notion of face is the strongest force guiding their use since face-saving strategies revolve around the universal notion of politeness which follows the universal conception of self-esteem and positive light in which the self can be displayed (cf. Brown and Levinson, 1998). . E.g., Bush’s concept of war on terror in the aftermath of September 11, 2001.

152 Tanja Gradečak-Erdeljić and Goran Milić

chickenhawk below. These models were proposed by Lakoff (1996) in his analysis of the U.S. political system and stem from different understandings of the concept of family (underlying the metaphor society is a family) As for euphemisms, one of the obvious deep frames in both politics and warfare is the conceptual metaphor humans are objects with a politician or a military official being in charge of or having absolute power over his/her dehumanized underlings. This conceptual metaphor deeply grasps into the personal and moral values of not only politicians as creators of those euphemisms, entrenching thus a new line of thought which allows them to actually treat people as objects, but also of the general public, which is more ready to accept the consequences of such treatment (cf. population transfer or ethnic cleansing, two very “popular” euphemisms in the political jargon during the war for independence in Croatia in the early 1990s). A similar process is validated in a metonymic mapping which entered the political discourse from military jargon, where the euphemism body count reflects a less salient part of the scenario. It refers to the survivors counting the dead and clouds the more relevant elements of the possible source events preceding the actual killing of people or the moment of the killing itself. A part for part metonymy is at play (cf. Panther’s [2005] notion of source meaning elaboration), but then a second level of metonymic mapping kicks in and the part of the scenario for the whole scenario metonymy is activated where the focused body counting stands for the whole target domain of killed people: If we focus on the part presenting the scenario of killed people as the central part of the scenario (white, full arrow), we could claim that the basic metonymic mapping happens at its referential level. There, instead of the objectively more detailed and precise lexical item corpse, a more neutral or euphemistically styled item body is included in the euphemistic construction. Thereby the politeness effect of the euphemism prevents notions of death as a classic example of the socially recognized taboo from entering the scene. The time consumed represents a longer temporal distance, an iconic representation of the temporal postponement by which a part of the concept employed in the process of this metonymic mapping is separated from the mental grasp of the central part or a default meaning of the concept. As we elaborated above, this is the process of establishing the conceptual iconicity relationship between the source and target domains. The step-by-step procedure that a listener must go through in order to arrive at the conclusion that body count in fact means . We have found many further examples motivated by the metaphor politics is war, but the discussion of these remains out of the scope of this chapter.



Metonymy at the crossroads 153

T: Killed people Survivors count the dead

Number of dead is reported

S

People get killed

Figure 2.  Metonymic mapping in the euphemism body count

“number of the dead people” involves an intricate chain of metonymies, which, to a large degree, resembles Barcelona’s (2005) notion of inferential metonymic chaining. Both (all) participants in the communication process should be aware of the implicatures this euphemism may trigger. Also worth mentioning are those implicatures that can be lost precisely because of the lack of awareness of the mechanisms of political discourse or the lack of the common deep frame. The transparent arrows in Figure 2 denote the actual moment of people being killed, or the subsequent count of the corpses lying on the ground when the number of the dead is reported. The subjectivity element involved in this kind of metonymical mapping is the fact that a speaker may never know which part of the scenario the listener will eventually arrive at. The imagery that served as a conceptual structure for these event schemas, in our cases predominately related to the topic of war, is highly individualized and tied to one’s personal experience and understanding. Thus, for some body count would necessarily include gruesome images of dead human bodies seen on the frontlines, while for others, just a line in a newspaper article which would be vaguely connected with certain negative events. But, clearly, the pragmatic link is established, whereby both the speaker’s intention of covering the unpleasant truth and the listener’s (presupposed) readiness not to be exposed to certain negative feelings are merged in the shared experience domain of political discourse where certain rules are set for the participants who are supposed to understand and abide by them.

154 Tanja Gradečak-Erdeljić and Goran Milić

3.2 The dysphemism chickenhawk To further illustrate the crucial function of metonymy as a natural inference schema, serving as a basis for the other two major functions, the motivational and referential, we turn to our second example. This is the dysphemistically used compound chickenhawk, defined in the Hampshire Gazette in 2002 as: “public persons – generally male – who (1) tend to advocate military solutions to political problems, and who have personally (2) declined to take advantage of significant opportunities to serve in uniform during wartime.” Originally used for a slightly different purpose (cf. Table 1 along with the etymology and context for other terms relevant for our analyses), the term gained both in popularity and use in the debates on the involvement of US troops in Iraq under George W. Bush’s mandate. Table 1.  Origin and historical development of relevant figurative expressions related to chickenhawk a.  War Hawk was originally used to describe a member of the House of Representatives of the Twelfth Congress of the United States who advocated going to war against Great Britain in the War of 1812. The term has evolved into an informal Americanism used to describe a political stance of preparedness for aggression, by diplomatic and ultimately military means, against others to improve the standing of their own government, country, or organization. The War Hawks in the Twelfth Congress were mostly young Republicans. (Source: Wikipedia) b.  hawks & doves: Hawkish “militaristic” first attested 1965; hawk in this sense is attested from 1962; dove: political meaning “person who advocates peace” first attested in 1962, during the Cuban Missile Crisis. Source: http://www.etymonline.com c.  chickenhawk (politics): An association between the word chickenhawk and war was popularized several years earlier in the 1983 bestselling book Chickenhawk, a memoir by ­Robert Mason about his service in the Vietnam War, in which he was a helicopter pilot. ­Mason used the word as a compound oxymoron to describe both his fear of combat (“chicken”) and his attraction to it (“hawk”), a slightly different use of the term which nonetheless might have inspired the current usage. (Source: Wikipedia)

It is used primarily by Democrats as a political slur against Republicans like George W. Bush or Dick Cheney, who meet the above criteria, as illustrated by the following example:

. Barcelona (2005: 317) argues that “[a] metonymy has a motivational role when it crucially guides the development of a constructional meaning or form, whose entrenchment (Langacker, 1987) then leads it to acquire unit status (Langacker, 1987), that is, to be used in largely automatic fashion.” (emphasis as in the original)



Metonymy at the crossroads 155

(1) Heston may fancy himself an American eagle, but he’s really just a hypocritical chicken hawk.

Apart from its dysphemistic effect, the term is interesting both for its particular formal nature as a compound and as an instance of the general metaphorical mapping humans are animals. The nature and status of the latter has been a controversial issue in the debates on the delimitation and interaction of conceptual ­metaphor and metonymy. Thus, from a formal point of view, Ruiz de Mendoza and Díez Velasco (2001) use instantiations of this general mapping as an argument in favor of acknowledging a continuum between metaphor and metonymy, with NP vehicles as instantiations eligible for use in the referential position, preferred for referential metonymies. Furthermore, Barcelona (2002) sees instantiations of this mapping (John is a lion) as a potential problem for postulating the construct of pragmatic function and inclusion in the same functional domain (icm, frame) as a foolproof criterion for defining metonymy and distinguishing it from metaphor, based on the within-domain vs. cross-domain mapping. Its status as a compound brings further issues. Benczes (2006, this volume) acknowledges the intricate workings of both metaphor and metonymy on the constituents of a compound and studies the possible types of semantic relations underlying them. Benczes (2006) handles chickenhawk as a case of double metaphorical processing, where both the head, or the profile determinant, and the modifying element which specifies the relevant semantic relation, are both previously understood metaphorically, i.e., “serve as source domains of conceptual metaphors through which we are able to understand the (phonologically) unnamed target entities, w and z.” (p. 102). Furthermore, chickenhawk is classified as an appositional compound in traditional terminology, making the construction in question a hyponym of both constituents. In its literal use the compound functions as a varietal term, according to an underlying paradigm (cf. sparrowhawk, goshawk). The compound as a whole can be thus seen as triggering Lakoff ’s (1987) metonymic model of (­metaphorbased) social stereotype, ultimately based on the high-level metonymy category for defining property. In our case, Benczes highlights another problem: Although the compound inherits its profile from hawk, I believe that the relationship between the modifier and the head is very difficult to determine: neither of the constituents can be assigned a semantic role such as Patient, Agent, etc. (similarly to girlfriend for instance). A chicken hawk is a person who is both a chicken and a hawk at the same time – i.e. while advocating military intervention, they are nonetheless considered as cowards by society.  (Benczes, 2006: 103)

Both constituents thus rely on conventionalized conceptual mappings and serve as reference points for activating multiple domains within their respective ­domain matrices (or domain networks, cf. Benczes, this volume). The N2, hawk, as the

156 Tanja Gradečak-Erdeljić and Goran Milić

profile determinant, has acquired a particular conventionalized meaning of “one who takes a militant attitude and advocates immediate vigorous action; especially: a supporter of a war or warlike policy” (Merriam-Webster’s Online Dictionary) on the basis of its association with the domain involvement in war, as activated by the first constituent of the originally coined compound war hawk (see Table 1). On the other hand, the “versatility” of chicken in its metonymic and/or metaphorical senses stems from the fact that the concept satisfies a number of ­Radden and Kövecses’s (1999) principles for metonymic vehicles (whole over part, good gestalt over bad, concrete over abstract, visible over invisible, stereotypical over non-stereotypical). Now although we tacitly acknowledge the advantages of meaning construction analysis in terms of blending theory (Fauconnier and Turner, 2002) especially in terms of their emergent meaning hypothesis, we continue to embrace Barcelona’s (2005: 339) broad view of metonymic chains as “sets of general patterns identified in terms of the criteria of function, directness and crossing of analytical levels” guiding or triggering inferences. Our theoretically driven objective here is to address the importance of pragmatic function as an important factor in the process. Relying on Fauconnier’s (1997) ideas, Barcelona (2007: 105) sees the pragmatic function as “a strong built-in connection between roles in a frame or icm (cause–effect, author–work, agent–action)” and consequently “a fundamental property of metonymy, namely the fact that the source maps onto and activates the target in virtue of the experiential (hence pragmatic) link between the roles each of them performs in the same functional domain” (emphasis added). The importance of the italicized parts of Barcelona’s definition emerges clearly in our case. The agentive role of hawk as a predator ascribed to the bird was based either on the lack of experience with the bird other than the visual one (having to do with its interaction with other birds/animals due to the bird’s predominant habitat [air], or to the limited functional interaction in the scenario of hawks killing poultry, i.e., in the relevant functional domain). The negative effect this kind of behavior (has) had on people’s well-being has made the bird’s active, agentive role in the scenario eligible for a negative evaluation and interpretation as aggression (cf. the positive image evoked by the use of eagle limited to the former type of interaction). The ascribed role thus crucially determines, or imposes, a perspective on the source domain. Similarly, in the case of chicken, the . Barcelona (2003b) acknowledges the compatibility of his approach with the Blending Theory. Brdar and Brdar-Szabó (2007: 135) “also share with Barcelona the view that metonymy has a crucial role in blending, i.e. that considerable portions of such webs in blends can be unpacked and linearized into a series of metonymies. These come to be presented in a more compressed and holistic form of blending.”



Metonymy at the crossroads 157

cultural and necessarily anthropocentric interpretation results in the interpretation of the bird’s behavioral characteristic as easily scared. When applied to a human being, the role inherent to the bird and essentially determining its defining property is interpreted as passive and therefore negative. The effect is reinforced by the logic of the hierarchical cultural model of the Great Chain of Being (Lakoff and Turner, 1989), a folk theory of how things are related to one another in the world. By being equated with an animal, a human’s behavior is generically understood as a lack of, or deviance from, the desired quintessential property of humans, such as rational thought, moral and aesthetic values. Being glossed as cowardly, by default, vouches for a negative evaluation of the person being referred to and a strong emotional, dysphemistic effect. Our interim observation is that the influence of experiential motivation should be inevitably approached as twofold. On the one hand, one should distinguish between the types of relevant experience. Opposed to the more perceptual, there is the more functional or “anthropocentric” experience, which is [moreor-less] directly related to the actual interaction with the animal, in terms of its effect on, or purpose for, the well-being of humans. In other words, both the lack of direct experience, i.e., the limitation to perceptual experience (visual, audio), and the extent of direct experience of humans with objects should definitely play a role in any similar analysis. We believe such an approach would contribute to the refinement and application of both the criterion of pragmatic function and functional domain, as well as help resolve the cases of the metaphor humans are animals (e.g., John is a lion), emphasized by Barcelona (2002) as problematic in terms of its status.

. In a contrastive study of figurative uses of animal names in English and Croatian, Milić (2007) notices a considerable degree of cross-linguistic overlap and a marked tendency for animals closer to humans and serving a particular purpose (as sources of meat, [edible] products and in the domain of labour) to serve as more productive vehicles for metaphorical use. The author ascribes this to the workings of Grady’s (2005) primary metaphor similarity is closeness and the satisfaction of a number of Radden and Kövecses’s (1999) principles for the selection of metonymic vehicles. The findings mesh well with Ohnuki-Tierney (1990), who argues for the acknowledgement of a crucial difference between two types of metaphors, i.e., linguistic instantiations of humans are animals mapping. Comparing the monkey as a culturally important vehicle in Japanese culture to the classic example of a lion (George is a lion), the author makes it the role of the latter “to bring to the fore an inherent metonymic relationship – sometimes uncomfortable to articulate in the mind – between the two semantic domains by emphasizing similarities. Humans and monkeys (or dogs) are placed side by side in a metonymic relationship through the metaphorical emphasis on the likeness. (…) Put another way, ‘the lion’ leaves George and all other people alone, except for a particular feature of similarity, whereas the ‘monkey’ is ready to lump humans and monkeys in a single category.” (ibid., pp. 98–99)

158 Tanja Gradečak-Erdeljić and Goran Milić

On the specific level of the actual type of discourse, the dysphemistic meaning and its implications ensue from a clash between the actual behavior of the referent, i.e., on the real world facts that he has never served, or has dodged the draft, and the metaphorical implications of chickenhawk triggered in the context. According to his (metonymically activated) defining characteristic of an active, aggressive viewpoint, the hawk/politician is supposed to “practice what he preaches”, that is, as an agent lead by example. This clashes with past actions on which the characteristics of cowardice, triggered by chicken, is based. As a result, and based on an underlying frame determining moral values, the politician is not only seen as a coward, but also as a hypocrite, which in our opinion is at the core of the dysphemistic effect to which we return below. According to Barcelona (2005), there is a vast number of routes available by using metonymies as “ready-made” templates leading to the adjustment of domains activated by their chaining. However, similar to the euphemistic element we analyzed, we keep on arguing for the existence of the defaults we have already suggested above, which crucially guide and/or trigger the achieved effect. Lakoff ’s (2006) deep, surface and issue frames seem to be useful constructs mediating between the workings of the generic level cultural model of the Great Chain of ­Being, as a general, cosmovision model and the constructs responsible for the subjective use and desired effect of the term in actual discourse (type). The perspectives inherent in the already mentioned nurturant parent model (Lakoff 1996), embraced (unconsciously) by the Progressives, imply the following ways of thinking and acting: Empathy and responsibility for oneself and others are the core values of the family. Respectful, loving, and supportive parenting promotes healthy families. Health care, education, food on the table, and social systems are essential to the wellbeing of the family. Loving, committed, and supportive individuals define the family, not gender roles.  (Lakoff, 2006: 46)

As such, the model largely determines the overall ascription of roles underlying the evaluation of the metaphorical hawk’s metonymically foregrounded defining property and (presumably) guides the user’s intentions in the way described above, with the aim of achieving the desired effect on the hearer. Again, the speaker is certainly helped by the surface frames, i.e., structures/models more directly associated with the actual linguistic expression. The humans are animals mapping comprising of experientially-based factors presented above serves as a very powerful default. Coupled with Lakoff and Johnson’s (1999: 290) understanding of morality as being “about well-being”, the Democrats’ ascription of roles in the relevant scenario and the line of reasoning guiding the meaning construction, do provide powerful motivation for the desired effect.



Metonymy at the crossroads 159

However, guided by the same underlying type of conceptual mechanisms, the adoption of a different deep frame, such as the Strict Father model – one ascribed to the Republicans by Lakoff (1996), hinging on obedience and discipline as the core values of the family – may lead to a re-examination of the possible effect for both the hearer and the speaker. In the functional domain of hunting it is possible to construe the hawk as a useful bird whose active behavior has a desirable effect with regard to humans. Thus, by adopting the latter model, the “active” role of the hawk can be construed not as the inherently negative aggression, but as positive assertiveness. It is this possibility of reframing (Lakoff, 2006) that has given rise to constant media debates between Democrats and Republicans on the use and effect of the compound, as well as on the rationale for its use as a political slur. As a telling example for the importance of (re)framing and its repercussions on the conceptual base underlying a constructional form, consider a group of Republican bloggers calling themselves the 101st Fighting Keyboardists, using the chicken hawk as their mascot and sporting the motto “We eat chickens for breakfast”.10 Both the motto and Example (2) below clearly reflect the possibility of using the aforementioned productivity of the vehicle to evoke the intended reading. The latter ranges from literal through metonymic (by virtue of the conventionalized object (animal) for material constituting the object (meat) metonymy) to the metaphorical readings as an instance of the humans are animals metaphor to simultaneously achieve a number of effects: (2) Q: So why use the chicken hawk as the symbol for the 101st Fighting Keyboardists? A: Because the chicken hawk kills and devours chickens and rats, and the anti-war crowd are really just a bunch of chickens and rats whom we must kill and devour. I mean that symbolically, of course, as no cannibalism will be involved.

By adopting the particular framing in terms of the strict father model, the Fighting Keyboardists’ figurative reading makes use of the compound and the inferences of the humans are animals mapping for their own purposes, not unlike the Republicans’ rationale we proposed above. The example also demonstrates the position advocated by Barcelona (2002), of viewing metonymy as a mapping by means of which the source does not merely substitute the target, but activates it from a certain perspective. However, we concur with Brdar and Brdar-Szabó’s (this volume) doubts on metonymy as a simple case of unidirectional traffic. The above uses of both linguistic vehicles as separate units, parts of the compound and 10. For further reference, visit http://www.captainsquartersblog.com/mt/archives/006857.php.

160 Tanja Gradečak-Erdeljić and Goran Milić

the compound as a whole suggest that the intended reading is a function of the context, co-text and discourse type. The source (the bird[s]) remains conceptually present to a considerable degree, with its ascribed role as a crucial conceptual component contributing to its target meaning.11 The perspective it affords, open to manipulation, turns the figurative use from a slur into a desirable epithet. Finally, the interplay of the previously mentioned experientially motivated but culturally determined factors is further evidenced by the now entrenched dichotomy in political discourse between the already mentioned hawks and doves (politicians with aggressive and moderate political views, respectively), which is in turn motivated by a host of potential analogies. The proposed factors on the generic level could prove to be relevant cross-linguistically as factors determining the universality/variation in conceptual metaphors and metonymies. We now turn to this issue.

4.

A glance at conceptual and linguistic borrowing

Prompted by the universal notion of framing we explored the possibility of linguistic and conceptual borrowing by trying to find a potential equivalent for chicken­ hawk in Croatian, and failed to find a suitable one in the form of a compound, let alone one involving similar concepts. The varietal term (jastreb kokošar) did not prove to be a suitable metaphorical vehicle, due to the non-existence of the specific metaphorical meaning of chicken found in English, whose role and effect in the process of meaning development has been discussed above. Its absence in Croatian clearly renders impossible, or at least not well motivated, the possibilities available in English based on the relevant combinations of semantic relations, and dependent on pragmatic function mappings. Furthermore, it is interesting to note that Croatian uses the unit kukavica in the meaning of both “coward” and “cuckoo”, which could lead us to adopt it as a potential source domain concept, that is, an instance of humans are animals metaphor. However, it seems that the motivation for the name of the bird is an iconic one, to be found in the characteristic sound the bird makes, a phenomenon also found in many other European languages (German: Kuckuck, Italian: cuculo). The lexeme with the “coward” meaning stems from the verb kukati – “to wail” 11. Kövecses (2005: 226) notices in a similar vain that “once we have a source domain that conventionally constitutes a target, we can use any component of this source that fits elements of the target. Notice that there is a reversal here. In a dynamic discourse situation the activated target domain in the discourse can indeed select components of the source that fit a particular target idea or purpose.”



Metonymy at the crossroads 161

and presumably uses morphological means to produce the meaning of “one who wails”. The motivation behind this is also essentially metonymic, the most generic motivation being the person for characteristic behaviour metonymy, with the existence of the bird name and the general mapping humans are animals presumably only helping the entrenchment of this onomatopoeic expression. Finally, a survey of the newspaper section of the Croatian National Corpus has revealed a quite widespread use of generic level terms used as metonymic models of social stereotype (Lakoff, 1987). These are realized in the above-established­ dichotomy between hawks and doves typical of political discourse and the press, but also markedly limited to the matching type of discourse and journalistic style, where they are occasionally still marked by quotation marks, indicating their lesser status of conventionalization (see Example (3) below). We put the borrowing process and the easy conceptual transfer down to the metonymic entrenchment of the concepts as social stereotypes based on the defining properties mentioned earlier, as being predators, i.e., attackers vs. a symbol of peace, tolerance. The concept was easily adopted, the crucial motivating factor being again the universal tendency of conceptualizing people in terms of animals and the limited type of experience with both types of birds and their mutual relationship as predator and potential pray, i.e., metaphorical “enemies” in folk understanding Moreover, their use in the Croatian journalistic discourse seems to pertain to international and domestic political issues in general and is not constrained exclusively to U.S.-related issues: (3) Važno je znati da je kurdska samouprava u sjevernom Iraku, predvođena lokalnim vođama, “jastrebom” Massoudom Barzanijem i “golubom” Jalalom Talabanijem, spremna braniti se s oko 50.000 naoružanih boraca. “It is crucial to know that the Kurd home rule, headed by the local leaders, the ‘hawk’ Massoud Barzani and the ‘dove’ Jalal Talabani, stands ready to be defended with about 50,000 armed men.” (4) Pobij, a usput u privatnom aranžmanu, opljačkaj (jer nitko te neće pitati uoči glavnog sukoba), bila je slavonska taktika hadezeovskih jastrebova, koji su politički program temeljili na dinamitu. “Kill, and while you are at it, privately, plunder (because you won’t be asked any questions in the wake of the decisive clash); those were the tactics of HDZ hawks who based their political programme on dynamite.”

Apart from these conceptually and formally/lexically determined differences and similarities, to account for the lack of a direct equivalent we must also grant crucial importance to extralinguistic factors such as differences in the organization of political systems in the two countries, the lack of a clear division into Democrats

162 Tanja Gradečak-Erdeljić and Goran Milić

and Republicans, with the typical deep and surface frames underlying their views, talk and action, as well as the different issues and concrete events which all make clear the non-existence and/or use of different imagery in Croatian. The tendencies, possibilities and motivating factors, only hinted at here due to lack of space, invite further micro studies demanding a carefully developed methodology and corpus.

5.

Conclusion

The small-scale studies of euphemistic and dysphemistic political jargon have been presented in this chapter with the aim of pointing at the numerous possibilities of investigating metonymy at work, by exploring the network of meanings and their effects established through the interplay of metonymy, metaphor (and ultimately blending) in the light of some pragmatic universals and individual and idiosyncratic interpretations. Even in as limited an area of research as political discourse, the proposed status and function of metonymies as inference schemas providing (a chain of) pragmatically triggered links between the source and target concepts made us argue for their status as being at somewhat of a “crossroads” in terms of function (and effect). By concentrating on its foregrounding and backgrounding role in directing the interlocutors’ attention towards subjectively and intentionally chosen target meanings, we have argued for two constructs potentially relevant in the process: conceptual iconicity as understood by Gradečak-Erdeljić (2005) and the workings of frames. The latter were seen as simultaneously constituted by metaphorical and/or metonymic conceptual mappings, shaping particular types of discourses in conscious or subconscious ways by providing powerful defaults. Their differing degrees of specificity and interrelatedness were captured by ­ Lakoff ’s (2006) theoretical constructs of deep, surface and issue-frames, particularly tailored for examining political discourse. The choice of the euphemism body count and the dysphemism chickenhawk was in no way accidental, since they quite aptly point out the advantages and disadvantages of metonymy being treated in terms of its pragmatic function. We suggest that euphemistic expressions very frequently use pragmatic switches activated by various conceptual metonymies, one very productive being part of the scenario for the whole scenario or part of the scenario for the part of the scenario. In order to preserve the necessary social and emotional equilibrium, both the creators of such expressions (politicians in the case of the euphemism body count) and the general public as the target ­audience,



Metonymy at the crossroads 163

­ epend on the subjective selection of different conceptual sequences of the sced nario, which has been coded by the euphemism, in order to allow different inferential schemas to play a role. In theoretical terms, the study of chickenhawk allowed us to address and re-examine the proposals related to critical issues pertaining to metonymy and addressed in this volume, viz. its definitional criteria and its ambiguous status particularly when defined in contrast to metaphor. Both examined constructs proved to contain powerful elements of construal, determined, or at least guided by ultimately experiential, yet socially-shaped criteria, particularly when used in real instances of communication. The nature and importance of the pragmatic function criterion as a strong motivating factor was thus emphasized by addressing its perspective-defining function as a guidepost for determining the ensuing pragmatic effect of the metonymic mappings. The study stressed the strong anthropocentric, experiential bias both in the workings of the former construct and in the related one of functional domain, making Barcelona’s (2003a) specification of the term an effective and relevant one. The proposed factors, on the generic level, were found to be plausible as cross-linguistically relevant potential factors determining the universality/variation in conceptual metaphors and metonymies. However, they are still culturally determined in a number of ways, as illustrated by the possibility of reframing in actual contexts of use and ultimately determined by the discourse participants’ subjective construal. On a final note, although we acknowledge that the fine-grained analyses of the latter cases are traditionally thought of as the provenance of Fauconnier and Turner’s (2002) Blending Theory, we feel that other attempts at analysis would yield valuable new insights. Based on the ideas proposed in this study, we find suitable and promising both Barcelona’s (2005, 2007) analyses of inferential chains of conceptual metonymies used in discourse as well as the ways Grady’s (2005) primary metaphors combine to form compound metaphors. Needless to say, these two theoretical approaches, differing in the granularity of approach, would further benefit from the inclusion and analysis of a number of elements commonly associated with sociolinguistic and discourse-analytic approaches.

References Allan, Keith and Kate Burridge. 2006. Forbidden Words: Taboo and the Censoring of Language. Cambridge: Cambridge University Press. Barcelona, Antonio (ed.). 2000. Metaphor and Metonymy at the Crossroads. Berlin & New York: Mouton de Gruyter.

164 Tanja Gradečak-Erdeljić and Goran Milić

Barcelona, Antonio. 2002. Clarifying and applying the notions of metaphor and metonymy within cognitive linguistics: An update. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (207–279). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2003a. Metonymy in Cognitive Linguistics. An analysis and a few modest proposals. In Cuyckens, Hubert, Klaus-Uwe Panther, and Thomas Berg (eds.), Motivation in Language: Studies in Honor of Günter Radden (223–255). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2003b. The case for a metonymic basis of pragmatic inferencing: Evidence from jokes and funny anecdotes. In Panther, Klaus-Uwe and Linda Thornburg (eds.), Metonymy and Pragmatic Inferencing [Pragmatics and Beyond New Series 113] (81–102). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2005. The multilevel operation of metonymy in grammar and discourse, with particular attention to metonymic chains. In Ruiz de Mendoza, Francisco and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (313–352). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2007. The multi-level role of metonymy in grammar and discourse: A case study. In Kosecki, Krzysztof (ed.), Perspectives on Metonymy. Proceedings of the International Conference Perspectives on Metonymy, Held in Lodz, Poland, May 6–7 (103–133). Frankfurt am Main: Peter Lang. Benczes, Réka. 2006. Creative Compounding in English. The Semantics of Metaphorical and Metonymical Noun-Noun Combinations [Human Cognitive Processing 19]. Amsterdam & Philadelphia: John Benjamins. Brdar, Mario and Rita Brdar-Szabó. 2007. When Zidane is not simply Zidane, and Bill Gates is not just Bill Gates: Or, Some thoughts on online construction of metaphtonymic meanings of proper names. In Radden, Günter, Klaus-Michael Köpcke, Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (125–142). Amsterdam & Philadelphia: John Benjamins. Brown, Penelope and Stephen C. Levinson. 1998. Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press. Burridge, Kate. 2005.Weeds in the Garden of Words: Further observations on the tangled history of the English language. Cambridge: Cambridge University Press. Croft, William. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4: 335–370. Dirven, René. 2005. Major strands in Cognitive Linguistics. In Ruiz de Mendoza, Francisco and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (17–68). Berlin & New York: Mouton de Gruyter. Dirven, René, Roslyn Frank, and Cornelia Ilie (eds.). 2001. Language and Ideology. Vol. II: Descriptive Cognitive Approaches. Amsterdam & Philadelphia: John Benjamins. Dirven, René, Roslyn Frank and Martin Pütz (eds.). 2003. Cognitive Models in Language and Thought: Ideology, Metaphors and Meanings [Cognitive Linguistics Research 24]. Berlin: Walter de Gruyter. Fauconnier, Gilles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Fauconnier, Gilles and Mark Turner. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books.



Metonymy at the crossroads 165

Gibbs, Raymond W., Jr. 1994. The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Gradečak-Erdeljić, Tanja. 2005. Euphemisms in the language of politics or how metonymy opens one door but closes the other. In Cap, Piotr (ed.), Pragmatics Today (287–299). Frankfurt am Main: Peter Lang. Grady, Joseph. 2005. Primary metaphors as inputs to conceptual integration. Journal of Pragmatics 37: 595–614. Kövecses, Zoltán. 2000. Metaphor and Emotion. Cambridge: Cambridge University Press. Kövecses, Zoltán. 2005. Metaphor in Culture: Universality and Variation. Cambridge: Cambridge University Press. Lakoff, George. 1987. Women, Fire, and Dangerous Things. Chicago: University of Chicago Press. Lakoff, George. 1996. Moral Politics: What Conservatives Know That Liberals Don’t. Chicago: University of Chicago Press. Lakoff, George. 2006. Thinking Points: Communicating Our American Values and Vision. New York: Farrar, Straus and Giroux. Lakoff, George and Mark Johnson. 1999. Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. New York: Basic Books. Lakoff, George and Mark Turner. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: The University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Vol. I: Theoretical Prerequisites. Stanford: Stanford University Press. Milić, Goran. 2007. Metaforički jezični izrazi tipa čovjek je životinja: Međujezična i među­ kulturna analiza. Unpublished MA thesis equivalent. University of Zagreb. Morgan, Pamela S. 2001. The semantics of an impeachment. In Dirven, René, Roslyn Frank, and Cornelia Ilie (eds.), Language and Ideology. Vol. II: Descriptive Cognitive Approaches [Current Issues in Linguistic Theory 205] (77–106). Amsterdam & Philadelphia: John Benjamins. Ohnuki-Tierney, Emiko. 1990. Monkey as metaphor? Transformations of a polytropic symbol in Japanese culture. Man, New Series 25(1): 89–107. Panther, Klaus-Uwe. 2005. The role of conceptual metonymy in meaning construction. In Ruiz de Mendoza, Francisco and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (353–386). Berlin & New York: Mouton de Gruyter. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought [Human Cognitive Processing 4] (17–61). Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco José and Olga Díez Velasco. 2001. High-level metonymy and linguistic structure. Available at http://www.sincronia.cucsh.udg.mx/metonymy.htm. Thornburg, Linda L. and Klaus-Uwe Panther. 1997. Speech Act Metonymies. In Liebert, WolfAndrea, Gisela Redeker, and Linda Waugh (eds.), Discourse and Perspective in Cognitive Linguistics (205–219). Amsterdam & Philadelphia: John Benjamins. Warren, Beatrice. 1992. What euphemisms tell us about the interpretation of words. Studia Linguistica 46(2): 128–172.

The role of metonymy in complex tropes Cognitive operations and pragmatic implications Javier Herrero Ruiz University of La Rioja, Spain

This chapter, following preliminary work in Herrero (2009), adds to Lakoff ’s (1987) taxonomy of icms other so-called figures of speech or tropes (often disregarded within Cognitive Linguistics) such as irony, oxymoron, overstatement, understatement, euphemism, and dysphemism. These figures are icms refined to create specific meaning effects, and can be divided into: (i) simple (models in which the generation of additional meanings is built upon propositional icms), and (ii) complex (tropes in which the creation of extra meaning is based on a representational, non-propositional, model). The present chapter studies the role of metonymy in producing the extra meaning effects conveyed by some complex tropes, in terms of both the cognitive operations associated with metonymy and its pragmatic implications. Keywords: cognitive operations, figure of speech, Idealized Cognitive Model (icm), pragmatic implication, trope

1.

Introduction

Since the emergence of Cognitive Linguistics at the end of the 70s, both metaphor and metonymy have been promoted from being mere rhetorical figures or tropes to the status of cognitive strategies of knowledge organization and mental mechanisms underlying many aspects of human conceptualization. Similarly, in this chapter we attempt to show how some other traditionally considered tropes (in particular, irony, oxymoron, overstatement, understatement, euphemism, and dysphemism) can be actually considered idealized ­cognitive models (icms) of the kind proposed by Lakoff (1987, 1989), with a specific emphasis on how some of these figures are grounded in metonymy. The

168 Javier Herrero Ruiz

examples have been extracted from a multi-faceted corpus compiled from The British ­National Corpus (BNC), The Corpus of Contemporary American English (COCA), and selected Google searches. We first devote a section to the study of idealized cognitive models, tropes, and cognitive operations in the literature. Then we address the issue of the role of metonymy in constructing and understanding such tropes.

2.

icms and tropes

One of the main facets of Cognitive Linguistics research has been the analysis of Idealized Cognitive Models. icms can be defined as cognitive structures that depict reality from a certain angle, leading to a process of idealization of reality (Lakoff, 1987, 1989). More specifically, icms are essentially the way whereby we arrange our knowledge of the world. According to Lakoff (1987: 68), they are structured by four principles: 1. Propositional structure, as in Fillmore’s Frame Semantics (1982, 1985). Frames are sets of semantic conditions described propositionally, i.e. in terms of predicates and their related arguments. 2. Image-schematic structure. Image-schemas, as discussed by Johnson (1987), are abstract, topological, and pre-conceptual patterns of experience, such as spatial orientations (“up,” “down”), or the “container,” “path,” and “part– whole” configurations. 3. Metaphoric and metonymic mappings, as discussed by Lakoff and Johnson­ (1980, 1999). A metaphor is a set of correspondences across different conceptual domains whereas a metonymy is a domain-internal conceptual mapping. Of the four different types of icm proposed by Lakoff and his followers, the majority of the literature on the topic to date has dealt with metaphor and metonymy. For nearly two decades, the bulk of research was on metaphor (e.g., Lakoff and Johnson, 1980, 1999; Lakoff and Turner, 1989; Kövecses, 1990; and Lakoff, 1987, 1993, 1996), with metonymy relegated to a secondary place. Then, some cognitive linguists focused their attention on metonymy, which was seen as a highly pervasive cognitive strategy of knowledge organization (e.g., Panther and Thornburg, 1999, 2003; Dirven, 2002; Barcelona, 2002b). Initially, the differences between metaphor and metonymy had been sketched out by Lakoff and Johnson (1980: 35–40) and Lakoff and Turner (1989: 103–104) as follows:



The role of metonymy in complex tropes 169

1. Metaphor involves two conceptual domains (one is understood in terms of the other), while there is only one conceptual domain involved in metonymy, i.e., the mapping takes place within a single domain and not across domains. 2. On the one hand, the source domain of metaphor is mapped onto the target domain, and hence it is mainly used for understanding, e.g., I have control over him (having control/force is up). On the other hand, metonymy is mainly used for reference, as it is possible to signal an entity in a schema by referring to a different entity in the same schema, e.g., Wall Street is in crisis (the street stands for the institution that can be found in it). 3. Consequently, the relationship between the source and target domains in metaphor is of the “is a” type; in metonymy there is a “stands for” relationship, as one entity in a schema stands for another entity in the same schema or for the schema as a whole. However, as some scholars have noted (especially Ruiz de Mendoza, 1997a, 1997b, 1999a, 1999b), the only essential distinction between metaphor and metonymy is the domain-external nature of metaphor versus the domain-internal nature of metonymy. This accounts for the fact that both metaphor and metonymy can be employed either referentially or non-referentially (i.e., predicatively): metaphor: – Used referentially: The pig is waiting for his bill (the pig is the customer). – Used non-referentially: I have control over him (having control/force is up). metonymy: – Used referentially: Wall Street is in crisis (the street stands for the institution). – Used non-referentially: He is a real brain (he is very intelligent). Moreover, Ruiz de Mendoza (1997a, 1997b) has postulated two types of metaphor from the viewpoint of the nature of the mapping process. Thus, we can find one-correspondence metaphors (there is just one correspondence between the source and target domains; e.g., people are animals, in which animal behavior is mapped onto human behavior) and many-correspondence metaphors (there are several correspondences between the source and target domains; e.g., love is a journey, in which the lovers are the travelers, the relationship is the vehicle, the lovers’ shared goal is the destination, etc.; Lakoff, 1993). On the other hand, metonymies have been divided (Ruiz de Mendoza, 2000; Ruiz de Mendoza and Pérez, 2001) into (1) target-in-source (the source domain stands for a target subdomain) and (2) source-in-target (a source sub-domain stands for a target domain). See Figure 1.

170 Javier Herrero Ruiz

Domain

Domain

Wall St. (source)

Customer (target)

Stock exchange (target)

Sandwich (source)

Subdomains

Olives Bank

Water

Restaurant Target-in-source metonymy

Source-in-target metonymy

Figure 1.  Examples of target-in-source and source-in-target metonymies: Wall Street is in crisis and The ham sandwich is waiting for his bill

These metonymic mappings can be linked to the two basic functions of referential metonymy: (1) source-in-target metonymies, which grant access to the reference domain (called matrix domain by Ruiz de Mendoza, 2000) via one of its subdomains, are based on domain expansion; (2) target-in-source metonymies, which lead to the highlighting of a relevant part of a domain, involve domain reduction. The adequacy of this proposal has been evidenced by its ability to explain, in an elegant manner, some conceptually motivated grammatical phenomena. Examples include metonymic anaphora (Ruiz de Mendoza and Díez, 2004) and the occurrence of zeugma in connection to metonymy (Peirsman and Geeraerts, this volume). It has also proven useful in our own analysis of some complex tropes, as demonstrated below. The realization that metonymy is ubiquitous in language and thought has led some scholars to investigate to what extent metonymy lies at the core of metaphoric thinking (cf. Goossens, 1990; Radden, 1993; Taylor, 1995/1989; ­Barcelona, 2000a, 2000b; Niemeier, 2000; Ruiz de Mendoza and Díez, 2002). Moreover, metonymy has recently been argued to have a more important status in our conceptual systems than had initially been granted by Lakoff and his collaborators (cf. Gibbs, 1994; Thornburg and Panther, 1997; Panther and Thornburg, 1998; Ruiz de Mendoza, 1999a, 1999b). In order to continue, a clear-cut distinction must be made between basic cognitive models (such as frames, image-schemas, metaphor, and metonymy) and what has traditionally been termed “tropes”, which, in our view, are but cognitive



The role of metonymy in complex tropes 171

models whose basic meaning can be extended in order to produce certain contextual effects which transcend the meaning of the icm as such. This is the case of irony, paradox, oxymoron, overstatement, understatement, euphemism, and dysphemism (cf. Herrero, 2009). Let us consider the following examples: (1) (2) (3)

His claims aren’t clear. It’s just a scratch. [referring to a sizeable wound] A: Did you get some with Anne?? She’s really gorgeous! B: Come on, don’t exaggerate! She’s just tasty.

In (1), by means of the concrete domain of sight we are able to understand the abstract domain of knowledge via the metaphor knowing is seeing. This conceptual metaphor stems from the early childhood experience of receiving cognitive input through seeing (cf. Sweetser, 1990: 37–40). However, whereas perception and cognition are seen as being together in the first years of one’s life because of the tight interrelationship between the intellectual input and vision (i.e., they are the object of conflation in terms of Grady, 1997, and Grady and Johnson, 2002), these two domains move apart later on as there is a deconflation process (cf. Johnson, 1997). This explains why the metaphor knowing is seeing can be employed simply to refer to “awareness” without any direct connection to vision. Thus, what is important here is the shift from visual clarity to intellectual manifestness, which allows for correct interpretation and reasoning with no extra meanings being created. Example (2) is different. From a communicative perspective, it is a case of understatement where the speaker attempts to mitigate the actual seriousness of a situation. From a cognitive perspective, we can analyze this process as a projection between two conceptual domains that is mediated by a mitigation operation. The entire process (from the speaker’s point of view) can be represented as shown in Figure 2. Input 1/context

A sizeable wound

Figure 2.  It’s just a scratch

Input 2/linguistic expression

Mitigation

It’s just a scratch

172 Javier Herrero Ruiz

Figure 2 makes partial use of the terminology used by Fauconnier and Turner (2002) in mental space theory. Mental spaces are the parts of conceptual domains that are used in meaning construction (note that not all we know about scratches and wounds is relevant for the interpretation of our example). In any interpretation task mental spaces are “input spaces” for the process in question since they supply the information needed for the task. In the case of metaphor and metonymy we have source and target input spaces and a mapping relationship between them. But in the case of other tropes, as with Example (2), there is no conceptual mapping but a relationship of contrast. This is discussed further below. B’s reply in (3) also makes use of a mitigation operation but mitigation is achieved through the metaphor having sex is eating (cf. Alarcón, 2002; ­Herrero, 2005). This process is depicted in Figure 3. Real context Input 1

An extremely attractive girl

Mitigation (via metaphor)

Linguistic expression Input 2 Source

A tasty dish Metaphor Target A fairly attractive girl

Figure 3.  She’s just tasty

In this case we have a projection from “extremely” to “fairly” attractive, carried out on the basis of the conceptual metaphor having sex is eating, which here maps the pleasure derived from taste perception onto the pleasure derived from physical attraction. In other words, it is by means of the conceptual metaphor that the mitigation between A’s literal use of “gorgeous” and B’s metaphoric use of “tasty” is achieved. The distinction between the phenomena underlying these examples leads us to further divide tropes into:



The role of metonymy in complex tropes 173

1. Simple: models in which meaning is produced on the basis of propositional icms. Thus, simple models only include one mental operation, as in This suitcase weighs a ton, which can be accomplished by means of either a reinforcement operation on the part of the speaker, or a mitigation operation on the part of the listener. 2. Complex: the creation of extra meanings is achieved via a non-propositional model, i.e., image-schemas, metaphor, or metonymy. So, complex tropes involve more than one operation, as in I’m dead! meaning “exhausted, knackered,” in which an operation (strengthening) results from the activity of a non-propositional model based on a correlation operation. In this case, we are dealing with the metaphor EXHAUSTION/TIREDNESS IS DEATH, whereby the image of a dead person is mapped onto the weary body because of its inactive appearance. The correspondences stem from the fact that when we are very tired (almost falling asleep), we are motionless and tend to close our eyes, almost like a dead person. The metaphor is, of course, an exaggeration that has to be mitigated by the hearer into “extremely tired”; in this combination of cognitive operations some additional meaning effects arise (e.g., the idea that the speaker can hardly speak or move). We should also emphasize the existence of another crucial difference between tropes and basic icms, which lies in the very essence of the correspondences of these tropes. Metaphor and metonymy are conceptual mappings. According to Grady (1999), metaphorical mappings can work on the basis of either resemblance or correlation. Resemblance takes place when the metaphor finds similarities between source and target (e.g., physical appearance between an animal and a person); correlation arises when the metaphor is the result of experientially grounded conceptual conflation (e.g., goals are destinations conflates a traveler’s goal of getting somewhere and reaching the destination). Finding similarities and correlating pieces of our experience are mental operations. In turn, metonymy, as pointed out above, works either by having a matrix domain stand for one of its subdomains or by a having a subdomain stand for its corresponding matrix domain. Ruiz de Mendoza (2000) referred to these two cognitive operations as domain reduction and domain expansion respectively. In the rest of the tropes, such as irony, paradox, oxymoron, overstatement, understatement, euphemism, and dysphemism there are no source and target relationships but there are other cognitive operations based on contrast and opposition: one input space is related to the real context of situation of the communicative exchange or to the interpreter’s knowledge or expectations, and another input space creates a contrast with the previous input space. Following this line of thinking, Herrero (2009: 137) has defined tropes as:

174 Javier Herrero Ruiz

Meaning-derivation processes whereby the hearer reinterprets the conceptual structure of an utterance or action (the trigger, i.e., the action or linguistic expression that departs from the interpreter’s expectations, encyclopedic knowledge, or context) to fit the requirements of that cognitive environment (in terms of Relevance Theory, all the non-linguistic information that is relevant to interpret an utterance as subjectively perceived and interpreted by the language user) and in order to produce certain contextual effects.

This reinterpretation can be performed by means of several cognitive operations that Herrero (2009) has also refined. In order to better understand how the operations underlying the linguistic phenomena under study work, it may prove useful to consider the notions of enrichment and loosening, which have been explored in detail by Carston (2002) within the framework of Sperber and Wilson’s (1995) Relevance Theory. Enrichment operations are related to adapting a concept to its contextual requirements (like linguistic decoding processes); these operations permit language users to derive pragmatic inferences called explicatures or “explicitly communicated assumptions” (cf. Ruiz de Mendoza and Peña, 2005). By “loosening” Carston (2002) refers to the fact that on certain occasions a speaker chooses to express an utterance which is a “less-than-literal” or “loose” interpretation of the thought she intends to transmit. However, these concepts have recently been re-developed in terms of Cognitive Linguistics. One of the latest proposals, found in Ruiz de Mendoza (2005), develops ideas from Ruiz de Mendoza and Santibáñez (2003) and Ruiz de Mendoza and Peña (2005). According to Ruiz de Mendoza (2005), the contrast between mitigation and strengthening concerns the scalar nature of the concept to which either process refers. This is in contrast to loosening and strengthening, which are respectively related to broadening and narrowing. In light of this connection, Ruiz de Mendoza and Peña (2005) have suggested that mitigation involves cases of hyperbole whereas strengthening appears in understatement. Nevertheless, as argued by Herrero (2009), both strengthening and mitigation operations work in instances of overstatement and understatement but in different ways. In this respect, Herrero (2009) claims that the fact that Ruiz de Mendoza and Peña (2005) have connected mitigation with hyperbole and strengthening with understatement is a consequence of the fact that they have exclusively adopted a hearer-oriented perspective. Surprisingly enough, in a previous work, Ruiz de Mendoza and Díez (2002) link mitigation to understatement and strengthening to hyperbole, albeit with the default perspective as the speaker’s. Herrero (2009) solves this problem by discussing the different role of the two cognitive operations (strengthening and mitigation) in producing and processing both overstatement and understatement. The role changes according to the speaker and hearer’s stance.



The role of metonymy in complex tropes 175

Recent insights into cognitive operations have been provided by Ruiz de Mendoza (this volume). Specifically, he distinguishes between formal and content operations. Whereas the former (cueing, selection, integration, and abstraction) are high-level operations which have no direct impact on inference making but are still essential in making the conceptual material available for inferential activity, the latter (see the lists below) have been defined in the literature (Ruiz de Mendoza and Pérez, 2003; Ruiz de Mendoza and Santibáñez, 2003; Ruiz de Mendoza and Peña, 2005) as basic cognitive mechanisms that regulate inferential activity and result in different types of metaphor, metonymy, hyperbole, and other rhetorical figures and their combinations. Furthermore, Ruiz de Mendoza (this volume) observes that content operations put conceptual domains in two different kinds of correspondence: (1) in a is b operations (correlation, resemblance, strengthening, mitigation, echoing, contrasting) a conceptual domain is employed to understand certain aspects of a different conceptual domain; (2) in a for b operations (domain expansion, domain reduction, parametrization, and saturation) a “stands for” relationship is possible if a can access b when cued to do so. Herrero (2009: 138–141) distinguishes the following cognitive operations in order to explain tropes: 1. Reversion: or assigning a given expression the opposite meaning that it literally has. It underlies some cases of irony, as in the utterance Yeah, it’s really bad food! used to refer to a really tasty dish. Irony has been explained by Ruiz de Mendoza (this volume) as a matter of echoing. The ironical expression, in stating the opposite of what is actually the case, imitates one of the addressee’s thoughts. In the case of Yeah, it’s really bad food, the speaker is echoing the addressee’s assumption that the food was bad, when really the reverse is the case. But note that the echoing operation is complementary of reversion, which is necessary to make an adequate interpretation of the utterance – see Figure 4.

A



VS.

A

Figure 4.  Reversion

2. Integration and accommodation: the former is defined as the balanced combination of conceptual structure derived from different conceptual domains; the latter is the development of the basic conceptual layout provided by one or more input spaces in order to satisfy the requirements of a given cognitive environment. These operations apply to some instances of irony, oxymoron,

176 Javier Herrero Ruiz

and paradox. Let us take into consideration the case of the oxymoron sweet sorrow in the following situation: imagine that you are at a party and you flirt with a pretty girl. When it comes time to say farewell, until seeing her the next day, that very moment could be considered to be full of “sweet sorrow.” At first sight, “sorrow” contains a given gamut of features (sadness, bitterness, unhappiness, etc.) that clearly antagonize “sweet” (which is rather related to happiness, pleasure, joy…). However, in this instance, the two feelings present no conflict but rather a juxtaposition of contradictory emotions that the speaker compares/contrasts in his mind. In this sense, “sweet” has part of the scenario within its scope and “sorrow” applies to another part of the scenario. In terms of Cognitive Linguistics, a mental space which is generated by the term “sweet” and another one corresponding to “sorrow” stand in opposition. The resolution of this figure is also based upon the incorporation of the structure of both input spaces and their accommodation in a given context of situation, as shown in Figure 5. Input 2

Input 1

Sweet

VS.

Sorrow

Happy

Sadness

Pleasure

Grief

Integration + accommodation Case in which sorrow can lead to a certain degree of happiness





Figure 5.  Sweet sorrow

Remembering that paradox follows the same interpretation procedure as oxymoron, the famous I must be cruel only to be kind from Shakespeare’s Hamlet could be interpreted as depicted in Figure 6. Sometimes doing harm is a way of setting the stage for something good to happen. Hence, the idea is not necessarily that one is cruel and kind at the same time, but rather that being cruel will help you to behave kindly later. We are dealing with two contrasting yet interdependent mental spaces, which are also linked by a cause–effect relationship.



The role of metonymy in complex tropes 177

Input 1

Be cruel

Input 2

VS.

Be kind

Integration + accommodation Case in which being cruel will help to be kind



Figure 6.  I must be cruel only to be kind

3. Strengthening: strengthening or reinforcement underlies overstatement, understatement, overstated and understated irony, euphemism, and dysphemism. This mental operation can be considered a maximization by the ­linguistic expression of the extent to which a scalar concept applies in reality. For example, in This suitcase weighs a ton, there is an intentional maximization of the weight attributed to a real suitcase. 4. Mitigation: this cognitive process implies a minimization of the linguistic expression of the extent to which a scalar concept applies in reality. It is found in the same cases as overstatement. By way of illustration, imagine that it is very hot in summer but someone says It’s a bit hot today; in this case, the speaker has scaled down “very hot” into “a bit hot.” In this sense, strengthening and mitigation work on the basis of a scalar continuum in which we can move either from a lower value to a higher value in the scale (augmentation), or vice-versa, i.e., from a higher value to a lower value (diminution). Following Ruiz de Mendoza and Díez’s claim (2002) that an idealized cognitive model can be the direct consequence of the activity of one or more mental operations, since tropes are the result of several such operations, they can be given the status of cognitive models, as pointed out by Herrero (2009). If more than one operation is at work, the result is a complex trope. In the next section, we shall see how metonymy can perform the operations described above in such a way that it becomes essential in producing the extra meanings that characterize tropes.

178 Javier Herrero Ruiz

3.

The role of metonymy in complex tropes

3.1

Irony

First of all, following Herrero (2009: 159–160), we shall divide irony into two categories: simple and complex. The former includes two domains whose conceptual structure stands in opposition. To be reinterpreted, a reversion operation is needed over the first input space (which stems from the linguistic utterance) so that it means the same as the second, which is derived from the cognitive environment (e.g., Yeah, right, the omelet was terrible! – meaning “It was really tasty”). Complex irony is also characterized by two mental spaces but the second is not necessarily the result of reversing the first. In fact, in complex irony the structure of both input spaces must be integrated and accommodated into a third space in order to meet contextual requirements, as in How old did you say you were? addressed to an adult who is acting as a child. Here the linguistic expression is a question about the addressee’s age and the situation is one where the addressee knows that he has to behave as a mature person but is not doing so. The speaker’s question draws the addressee’s attention to the implicit contrast between what the addressee is doing and his actual age. However, there is no contrast between what the speaker explicitly says (first input) and the actual situation (second input), but rather between what the speaker’s question implicates (that the addressee is too grown up to act like a child) and the addressee’s behavior. As Barcelona (2002a) has noted, irony can co-occur with conceptual analytical phenomena such as metonymy and metaphor. As far as metonymy is concerned, since irony mainly works on a predicative basis it calls for the co-activation of non-referential types of metonymy. We can find two types of operation, namely: 1. Target-in-source metonymies within the ironic space directly arising from the designatum of the linguistic expression. 2. Source-in-target metonymies within the ironic space directly arising from the designatum of the linguistic expression. The ironic space that contains the information related to the linguistic expression is the only one that may include some other conceptual tools and operations; the other domain contains conceptual structure retrieved from the cognitive environment­. Target-in-source metonymies within the ironic domain that corresponds to the utterance are characterized by domain reduction and highlighting operations by means of which a most relevant part for the interpretation of the expression is brought into focus, hence acquiring a more central status. Let us analyze the case of He is a real brain!!! uttered about someone who is not too intelligent – see Figure 7.



The role of metonymy in complex tropes 179

Linguistic expression

Input 1

Irony

Input 2/context

Brain (source)

He is silly

Intelligence (target)

Figure 7.  He is a real brain!!!

As may be observed, the literal reading of the linguistic expression gives rise to the first input space in which, via a metonymy, the central inference of the expression (“he is intelligent”) is reached. Then, by means of a reversal operation that final inference is reversed to mean the opposite. The whole process can be illustrated as in Figure 8 below. Linguistic expression Input 1

Irony

Input 2/context

A (source)

A2 (target)

A2

Figure 8.  Target-in-source metonymy within the ironic source

As regards source-in-target metonymies within the first domain, consider the expression To be someone’s head used by the leader of a gang after learning that one of his henchmen is boasting that he is the real head of the gang – see Figure 9.

180 Javier Herrero Ruiz

Linguistic expression Input 1

Irony

Input 2/context

Leader (target)

Head (source)

Low-level servant

Figure 9.  He is the real head!

As can be observed, the linguistic expression is used to construct a cognitive domain that includes a source-in-target metonymy. The fact that the source of the metonymy is a subdomain of the target makes it possible to generate the ironic mapping via a domain expansion operation. More specifically, the final output of the metonymic mapping constitutes the input for the irony since its conceptual structure is then reversed. Cases of irony containing an operation of interaction between metaphor and metonymy have also been identified. As noted above, the interaction operation must occur within the first input domain and its central inference is turned in the second input domain. Imagine that someone notices an open lie and ironically says I am going to stomach your claim completely. In this example, the linguistic expression gives rise to a cognitive process whereby one of the correspondences of the source domain of the metaphor ideas are food (contained in the ironic first domain) is expanded by means of a metonymy. The central inference of the interaction (i.e., “agree with”) is then reversed in the second input domain of the ironic operation – see Figure 10. According to Herrero (2009), all of these cases are instances of simple irony. With regard to complex irony, we should bear in mind that the only domain that may include metaphoric and metonymic operations is the ironic domain that contains the information derived from the linguistic expression. This is so because the other mental space contains the conceptual structure retrieved from the cognitive environment. We can illustrate this point by means of a variant of an example of irony given above, i.e., How old did you say you were?, which can be metonymically expressed via How much grey hair do you have?



The role of metonymy in complex tropes 181

Linguistic expression Input 1 Metaphor Person that eats food Stomach Digest

Input 2

Person that thinks out ideas Agree with

Irony Reject

Figure 10.  I am going to stomach your claim completely

In this case, a source-in-target metonymy is under focus (i.e., a part of an adult body, the grey hair, stands for both the person’s age and her expected behavior) whereby the most prominent aspect of the utterance is accessed to be interpreted correctly, i.e., ironically – see Figure 11. How much grey hair do you have?

Context (acting like a child)

Input 1/expected behavior

Input 2

Adult Metonymy

VS.

Grey hair

Integration + accommodation A person should behave in a certain way that corresponds to her age; as she is not behaving in that way, the speaker shows annoyance

Figure 11.  How much grey hair do you have?

Actual behavior: childish

182 Javier Herrero Ruiz

3.2 Oxymoron An utterance contains a case of oxymoron when it includes a feature or attribute of an entity that contradicts the hearer’s default expectations (cf. Herrero, 2009: 203). As is the case with irony, oxymoron can also co-occur with other cognitive models such as metonymy and metaphor. In this case, representational models show up in the conceptual domain corresponding to the term that contains a metaphor or a metonymy. This can be illustrated by means of the expression He has a stupid brain to refer to a person who generally acts foolishly. This sentence can be roughly paraphrased as “he has a dull mind” and works on the basis of a target-in-source metonymy generated within the domain that corresponds to the term “brain.” In this sense, from a Cognitive Linguistics standpoint, two mental spaces are generated linguistically: one by the term stupid and another one corresponding to brain. The metonymy involves domain reduction and highlighting operations whereby the most relevant part for the interpretation of the expression is brought into focus (i.e., “intelligence”). Once the metonymy is developed to interpret the expression correctly, there is a clash, which has to be solved, between both input domains. The resolution of this figure is based on the incorporation of the structure of both input spaces and their accommodation in a given context of situation. Figure 12 represents this process. Input 1

Input 2

VS.

Brain Metonymy

Stupid

Intelligence

Integration + accommodation Someone acts with a dull intelligence, i.e. in a foolish way

Figure 12.  He has a stupid brain

More complex constructions can also be found, such as the expression He has a stupid big brain, characterized by an overstated oxymoron based on the previous example and which means “he is very intelligent but acts stupidly.” Also, cases of



The role of metonymy in complex tropes 183

understated oxymoron based on representational models are possible, as in He’s a bit of a stupid brain.

3.3 Overstatement Overstatement is “the process whereby we can represent (prototypically scalar) concepts that are actually at a lower level on a scale by referring to higher-level magnitudes on the same scale, with the objective of maximizing certain contextual effects.” (Herrero, 2009: 215). It is characterized by a reinforcement or strengthening operation on the part of the speaker (e.g., This suitcase weighs too much for me, say, 50 kilos, gives rise to This suitcase weighs a ton) and by a mitigation operation on the part of the listener (e.g., This suitcase weighs a ton is interpreted as “This suitcase weighs a lot for a single person to lift”). The strengthening operation can be accomplished via metonymy. Let us consider the following case: there is a man eating a sirloin steak in a restaurant and he says I’m so hungry I could eat the entire cow, which obviously contains an exaggeration and a target-in-source metonymy. The metonymy strengthens the effect that eating the meat has on him. The utterance can be roughly paraphrased into “I’m so hungry I could eat a lot, certainly more than anyone would expect” – see Figure 13. Real context/target

Metonymy Source

Part of animal: Eg. A lot of meat

Cow Strengthening

Figure 13.  I’m so hungry I could eat the entire cow

In our corpus of study, a case of an overstatement that works on the basis of the metonymy place for people living there can be found in I think the whole town of Las Lunas is coming. In this example, we come across a target-in source metonymy whereby a strengthening operation is performed – see Figure 14. Real context/target

Metonymy

Some people

Las lunas Strengthening

Figure 14.  The whole town of Las Lunas is coming

Source

184 Javier Herrero Ruiz

3.4 Understatement To start with, we shall stick to the notion of understatement put forward in ­Herrero (2009) (i.e., the model whereby it is possible to lessen certain contextual effects) with a threefold division into: 1. Meiosis: the process whereby higher level dimensions can be referred to by means of lower-level magnitudes (e.g., It’s just a scratch where scratch refers to a “sizeable wound”). Meiosis is characterized by a mitigation operation on the part of the speaker (e.g., “there is a long queue at the theatre” > There seems to be a bit of a queue), and a strengthening operation on the part of the listener (e.g., There seems to be a bit of a queue > “there is a long queue at the theatre”). 2. Paradiastole: a case of euphemism (or dysphemism), which will be addressed in a later section. 3. Litotes: the negation of a certain word or expression to refer to any point within the semantic space which remains after the denial (e.g., It is not bad meaning “it is very good”). As for (1), Herrero (2009) has argued that metonymy may affect cases of meiosis, as in the example below where a mother wants to find out if her son has done something wrong: (4) A: Did you eat half the cake? B: It was just a spoon.

B’s reply can be analyzed as a case of mitigation achieved via a container for content metonymy (i.e., the spoon stands for its contents). The underlying cognitive process is diagrammed in Figure 15. Real context Half a cake (metonymic target)

Mitigation (metonymic source) A spoon Metonymy

Figure 15.  It was just a spoon

More specifically, B’s answer, trying to soften the impact of the action of eating so much cake, contains a mitigation operation which works on the expression related to the amount of food ingested; this is produced by a source-in-target metonymy. At this point, if we consider the fact that A should perform the opposite operation (i.e., reinforcement) to make the correct interpretation of the utterance,



The role of metonymy in complex tropes 185

Herrero’s distinction between speaker and hearer’s models is particularly useful as the operations undergone by either of them may be reversible. This has led Herrero (2009), following Janssen (2007), to claim that pragmatic theory should pay equal attention to both the speaker and hearer’s perspectives when dealing with certain linguistic phenomena. In fact, a speaker/hearer-based division is essential in order to clarify how interlocutors are able to produce and understand utterances. Mitigation also works in cases of high-level metonymy. First consider Lakoff ’s (1987: 78–79) well-known example of situational metonymy included below: (5) A: How did you get to the airport? B: I waved down a taxi.

According to Lakoff, we have a metonymic operation on a structured scenario (or icm) with the following elements:

Precondition: having access to a vehicle. Embarkation: getting into the vehicle and starting it up. Center: driving the vehicle (or having someone drive it) to the destination. Finish: parking and getting out. End-point: being at the destination.

Lakoff points out that either the precondition (e.g., I have a car), the embarkation (e.g., I hopped on a bus), or the center parts (e.g., I drove) can be employed to refer to the whole scenario. The fact that metonymy appears in some cases of answers to questions has further implications, especially when considering that the principles that constrain exchanges of this type are studied in inferential pragmatics as cases of implicature. As Ruiz de Mendoza (2005) has suggested, relevance theorists (cf. Sperber and Wilson, 1995) would regard B’s response as an implicated conclusion resulting from the following reasoning pattern:

implicit premises a. Taxis stop when potential customers wave their hands as a signal to stop. b. A customer stops a taxi in order to hire a ride to a certain place.



explicit meaning c. Speaker B stopped a taxi by waving his hands at it.



implicated conclusion d. Therefore, B hired the taxi to get to the airport.

In fact, this schema does flesh out the various elements of Lakoff ’s scenario and applies them to the more specific “taking a taxi” icm: the implicit premises relate to

186 Javier Herrero Ruiz

the precondition element, the explicit meaning to the embarkation component, and the implicated conclusion to the other elements. In this connection, “taking a taxi” or “driving to a destination” are pervasive low-level icms of a situational kind, since they involve the interaction between different entities within a certain context. Such situational icms can be exploited metonymically to generate implicatures. Let us consider the following example: imagine that you go to a bingo site and meet a friend of yours there, and after asking how she is, she answers I’m OK! Just crossing out some numbers! Here we are dealing with a metonymic operation on a structured scenario (or icm) that is characterized by these elements: 1. Before: at the very least buying a card. 2. Core: playing a bingo game (at least), by means of crossing out the numbers in the card. 3. After: being able to win some money. So far, this expression can be explained on the basis of a situational metonymy whereby the whole action of playing bingo is called upon by invoking one part of the scenario. Now imagine that when you go to the bingo site on some other day, you meet the same friend once more, and you say Here again? Consider this reply your friend may give: Oh, I was just crossing out a couple of numbers! In her reply, your friend is mitigating “a large amount of numbers” into only “two numbers” – see Figure 16. Input 1 Linguistic expression Many numbers in different games

Mitigation 2 numbers

Figure 16.  I was just crossing out a couple of numbers!

By stating that the amount of numbers is smaller than it actually is, which in turn stands for the whole action of playing bingo via a source-in-target metonymy, the contextual effects – and consequent pragmatic significance – of the fact that someone spends a lot of time in bingo halls is mitigated. This case can be interpreted as a projection between two different conceptual spaces mediated by a mitigation



The role of metonymy in complex tropes 187

operation. This operation has to be reversed by the hearer (via strengthening) in order to get the real interpretation of the expression – see Figure 17. Real context

Mitigation

Playing many bingo games (metonymic target)

Metonymy

(metonymic source) Crossing out 2 numbers

Figure 17.  I was just crossing out a couple of numbers!

A well-known example of this type appears in the following example: (6) [Some friends come across a girl they know. She has bought lots of things and is carrying them in many bags; she produces the following remark when they ask her how she is doing:] Oh, I was just window shopping.

In this utterance, “window shopping” is a constituent of the “shopping” scenario (or icm) that includes the following elements: 1. Before: observing and studying the items as displayed in the shop window. 2. Core: buying the selected items. 3. After: leaving the shop with the items in bags. Via the sentence I was just window shopping the girl understates the fact that she has purchased many items; this is done by referring only to the “before” part of the scenario. The “core” part could also have been mitigated: I just bought a few items, but with different pragmatic implications. By using the “before” element, it is implied that the speaker has been studying her intended purchases not on the basis of their intrinsic value or usefulness but, rather, of their appeal as displayed. This implication is missing in a mitigated version of the “core” part of the scenario.

3.5 Euphemism and dysphemism As defined by Herrero (2009: 267), euphemism is “the process whereby we can speak about actual higher-level dimensions by means of mentioning lower-level­ magnitudes in an axiological scale related to affective and emotional values,” whereas dysphemism is “the process whereby we are allowed to represent concepts that are actually at a lower level by means of referring to the higher-level magnitudes in an axiological scale based on emotional and affective load.” These tropes are derived in the same way as overstatement and meiosis, explained in Sections 3.3 and 3.4 above.

188 Javier Herrero Ruiz

Euphemism and dysphemism also interact with metonymy, as attested by the following examples. Consider the following examples of euphemism: (7) Vile condition. (8) Take a complete rest from football.

In (7) we find a target-in-source metonymy, which allows us to refer to a “criminal” by simply mentioning his or her most prototypical trait. In turn (8), where take a complete rest from football means “stop being involved in football indefinitely,” is further supported by the effect for cause metonymy. Thus, “taking some rest after an activity” is the effect of “being tired or exhausted”; at the same time taking a rest is a mitigated way of referring to stopping indefinitely. Since speaker-based models of euphemism are based on mitigation operations according to which the speaker scales down the overall meaning impact of a given utterance in terms of its affective load, these examples can be represented as in Figures 18 and 19. Input 1/real context Linguistic expression

Criminal

Mitigation Vile condition

Figure 18.  Vile condition Input 1/real context Linguistic expression

Be finished

Mitigation

Figure 19.  Take a complete rest from football

Need a complete rest



The role of metonymy in complex tropes 189

As for the case of dysphemism, let us consider the expression He died, which is neutral, and its dysphemistic counterpart He is worm food. The encoder strengthens the emotional load of He died and uses the impolite expression He is worm food, which in turn stems from an underlying effect for cause metonymy (i.e., death causes a body to decay and become worm food) – see Figure 20. Input 2/expression Input 1/context He is worm food He died Strengthening

Figure 20.  He is worm food

To conclude, it should be noted that in this section we have only addressed the speaker’s perspective. In this sense, as regards Examples (7) and (8), the hearer’s standpoint is subject to strengthening or reinforcement cognitive operations since there is a shift from a lower to a higher value of the affective content of the concept represented. In so doing, the emotional nuances that the speaker wanted to minimize are recovered. In other words, the hearer’s role is based on first noticing the contrast existing between the speaker’s utterance and the hearer’s cognitive environment, and second, strengthening the affective scalar content of the linguistic utterance in order to derive the value that is consistent with her cognitive environment. On the contrary, in the expression to be worm food a mitigation operation applies, as the listener goes from a higher to a lower value along the emotional scale of the concept that is represented.

4.

Conclusion

In this chapter, we have accomplished a twofold task. Primarily, we have provided further support for the cognitive-pragmatic understanding of the tropes (i.e., icms whose meaning can be extended in order to bring about certain contextual

190 Javier Herrero Ruiz

effects) put forward in Herrero (2009) and for his distinction between simple and complex cognitive models (based on propositional and non-propositional models respectively). Furthermore, we have been able to explain how the basic meaning of some specific tropes (irony, oxymoron, overstatement, understatement, euphemism, and dysphemism) can be derived via metonymy, a fact which once again proves the importance of conceptual metonymy (and metaphor) as tools to help us understand – and in fact constrain – the basic meaning and interpretation of the expressions of our language. More specifically, we have shown how metonymy provides the main operation (via domain reduction and domain expansion, which serve to highlight a relevant part of a domain and grant access to the reference domain respectively) whereby we can get the most prominent aspect of the tropes analyzed to be interpreted correctly. In so doing, we have also identified which domains of the tropes under scrutiny are likely to include the non-propositional models which interact in them, with a focus on both the extra meaning effects and the pragmatic implications produced. For example, in irony, the space that contains the information related to the linguistic expression is the only one that may have some other conceptual tools and operations (as the other domain contains conceptual structure retrieved from the cognitive environment), whereas in oxymoron representational models appear in the conceptual domain corresponding to the term that contains a metaphor or a metonymy. Additionally, we have studied the way metonymy may accomplish reinforcement and mitigation operations by itself, as in the case of I’m so hungry I could eat the entire cow and It was just a spoon or, conveying emotional innuendoes, He is worm food and Vile condition. Furthermore, in our study not only have we come across a case based upon the interaction between metaphor and metonymy (I am going to stomach your claim completely) which confirms that conceptual interaction is fully constrained by a limited set of interactional patterns (cf. Ruiz de Mendoza and Díez, 2002), but we have also encountered cases of high-level metonymies underlying complex tropes (e.g., I was just window shopping). To end with, as we have found that there is reversibility in terms of the operations undergone by the speaker and hearer in some of these models (overstatement, understatement, euphemism, and dysphemism), we firmly believe that pragmatic theory should pay equal attention to both the speaker and hearer’s perspectives when dealing with certain linguistic phenomena.



The role of metonymy in complex tropes 191

References Alarcón, Paola. 2002. El acto sexual es comer: Descripción lingüístico-cognitiva. Revista de Lingüística Teórica y Aplicada (RLA) 40: 7–24. Barcelona, Antonio (ed.). 2000a. Metaphor and Metonymy at the Crossroads: A Cognitive Perspective. Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2000b. On the plausibility of claiming a metonymic motivation for conceptual metaphor. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (31–58). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2002a. On the ubiquity and multiple-level operation of metonymy. In Lewandowska-Tomaszczyk, Barbara and Kamila Turewicz (eds.), Cognitive Linguistics Today (209–224). Frankfurt: Peter Lang. Barcelona, Antonio. 2002b. Clarifying and applying the notions of metaphor and metonymy within cognitive linguistics: an update. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (207–277). Berlin & New York: Mouton de Gruyter. Carston, Robyn. 2002. Linguistic meaning, communicated meaning and cognitive pragmatics. Mind & Language 17: 127–148. Dirven, René. 2002. Metonymy and metaphor: Different mental strategies of conceptualization. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (75–112). Berlin & New York: Mouton de Gruyter. Fauconnier, Gilles and Mark Turner. 2002. The Way We Think: Conceptual Blending and the Mind’s Hidden Complexities. New York: Basic Books. Fillmore, Charles. 1982. Frame semantics. In Linguistic Society of Korea (ed.), Linguistics in the Morning Calm (111–138). Seoul: Hanshin. Fillmore, Charles. 1985. Frames and the semantics of understanding. Quaderni di Semantica 6(2): 222–254. Gibbs, Raymond. 1994. The Poetics of Mind: Figurative Thought, Language and Understanding. Cambridge: Cambridge University Press. Goossens, Louis. 1990. Metaphtonymy: The interaction of metaphor and metonymy in expressions for linguistic action. Cognitive Linguistics 1(3): 323–340. Grady, Joseph. 1997. theories are buildings revisited. Cognitive Linguistics 8: 261–290. Grady, Joseph. 1999. A typology of motivation for conceptual metaphor: correlation vs. resemblance. In Gibbs, Ray and Gerard Steen (eds.), Metaphor in Cognitive Linguistics (79–100). Amsterdam & Philadelphia: John Benjamins. Grady, Joseph and Christopher Johnson. 2002. Converging evidence for the notions of subscene and primary scene. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (533–554). Berlin & New York: Mouton de Gruyter. Herrero, Javier. 2005. The role of metaphor and metonymy as tools of lexical creation: The case of Spanish informal speech. ODISEA 6: 97–118. Herrero, Javier. 2009. Understanding Tropes: At the Crossroads between Pragmatics and Cognition. [Duisburg Papers on Research in Language and Culture 75]. Frankfurt: Peter Lang. Janssen, Theo. 2007. A speaker/hearer-based grammar: the case of possessives and compounds. In Hannay, Mike and Gerard J. Steen (eds.), Structural-Functional Studies in English Grammar (353–387). Amsterdam & Philadelphia: John Benjamins.

192 Javier Herrero Ruiz

Johnson, Christopher. 1997. Metaphor vs. conflation in the acquisition of polysemy: The case of see. In Hiraga, Masako K., Chris Sinha, and Sherman Wilcox (eds.), Cultural, Typological, and Psychological Issues in Cognitive Linguistics (155–169). Amsterdam: John Benjamins. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination and Reason. Chicago: University of Chicago Press. Kövecses, Zoltán. 1990. Emotion Concepts. Philadelphia: Springer-Verlag. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: University of Chicago Press. Lakoff, George. 1989. Some empirical results about the nature of concepts. Mind & Language 4(1–2): 103–129. Lakoff, George. 1993. The contemporary theory of metaphor. In Ortony, Andrew (ed.), Metaphor and Thought (202–251). Cambridge: Cambridge University Press. Lakoff, George. 1996. Moral Politics. Chicago: University of Chicago Press. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago & London: The University of Chicago Press. Lakoff, George and Mark Johnson. 1999. Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. New York: Basic Books. Lakoff, George and Mark Turner. 1989. More Than Cool Reason: A Field Guide to Poetic Metaphor. Chicago: The University of Chicago Press. Niemeier, Susanne. 2000. Straight from the heart – metonymic and metaphorical explorations. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (185–213). Berlin & New York: Mouton de Gruyter. Panther, Klaus-Uwe and Linda L. Thornburg. 1998. A cognitive approach to inferencing in conversation. Journal of Pragmatics 30: 755–769. Panther, Klaus-Uwe and Linda L. Thornburg. 1999. The potential for actuality metonymy in English and Hungarian. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (333–359). Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg. 2003. Metonymy and Pragmatic Inferencing. Amsterdam & Philadelphia: John Benjamins. Radden, Günter. 1993. How metonymic are metaphors? In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (407–434). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza, Francisco José. 1997a. Cognitive and pragmatic aspects of metonymy. Cuadernos de Filología Inglesa 6(2): 161–178. Ruiz de Mendoza, Francisco José. 1997b. Metaphor, metonymy and conceptual interaction. ATLANTIS – Revista de la Asociación Española de Estudios Anglonorteamericanos 19(1): 281–295. Ruiz de Mendoza, Francisco José. 1999a. Implicatures, explicatures, and conceptual mappings. In Cifuentes, J. L. (ed.), Estudios de Lingüística Cognitiva (429–440). Universidad de Alicante: Servicio de Publicaciones. Ruiz de Mendoza, Francisco José. 1999b. The role of cognitive mechanisms in making inferences. Journal of English Studies 1: 237–255. Ruiz de Mendoza, Francisco José. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (109–132). Berlin & New York: Mouton de Gruyter.



The role of metonymy in complex tropes 193

Ruiz de Mendoza, Francisco José. 2005. Linguistic interpretation and cognition. In Croitoru, E., D. Tuchel, and M. Praisler (eds.), Cultural Matrix Reloaded: Romanian Society for English and American Studies (36–64). Bucarest: Didactica Si Pedagogica. Ruiz de Mendoza, Francisco José and Francisco Santibáñez. 2003. Content and formal cognitive operations in construing meaning. Italian Journal of Linguistics 15(2): 293–320. Ruiz de Mendoza, Francisco José and Lorena Pérez. 2001. Towards a pragmatic cognitive functional grammar. Revista Canaria de Estudios Ingleses 42: 187–214. Ruiz de Mendoza, Francisco José and Lorena Pérez. 2003. Cognitive operations and pragmatic implication. In Panther, Klaus-Uwe and L. Thornburg (eds.), Metonymy and Pragmatic Inferencing (23–50). Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza, Francisco José and Olga Díez. 2002. Patterns of conceptual interaction. In Dirven, René and Ralph Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (489–532). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza, Francisco José and Olga Díez. 2004. Metonymic motivation in anaphoric reference. In Radden, Günter and Klaus-Uwe Panther (eds.), Studies in Linguistic Motivation (293–320). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza, Francisco José and Sandra Peña. 2005. Conceptual interaction, cognitive operations and projection spaces. In Ruiz de Mendoza, Francisco José and Sandra Peña (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (254–280). Berlin & New York: Mouton de Gruyter. Sperber, Dan and Deirdre Wilson. 1995. Postface to the second edition of Relevance: Communication and Cognition. In Sperber, Dan and Deirdre Wilson, Relevance: Communication and Cognition. Oxford: Blackwell. Sweetser, Eve. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Taylor, John R. 1995/1989. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press. Thornburg, Linda and Klaus-Uwe Panther. 1997. Speech Act Metonymies. In Liebert, WolfAndreas­, Gisela Redeker, and Linda R. Waugh (eds.), Discourse and Perspectives in Cognitive Linguistics (205–219). Amsterdam: John Benjamins.

part ii

Metonymy and metonymic chains as mappings or processes within domain matrices/networks

Putting the notion of “domain” back into metonymy Evidence from compounds* Réka Benczes Eötvös Loránd University, Budapest

This chapter explores the role that domains play in conceptual metonymy by examining the semantics of metonymical (and metaphorical) noun–noun compounds. It argues that the concept of “domain” is a necessary feature of any definition of metonymy (irrespective of the fact whether “domain” is referred to as a domain matrix, frame, or icm). The analysis of noun–noun compounds, such as couch potato and scarlet-collar worker, imply that the domains are best understood as networks of semantic associations, with links to further semantic domains or even other grammatical constructions. Therefore, the chapter proposes that metonymy operates within a domain network, where the domains form web-like semantic links of associations. Keywords: domain, domain network, metaphor, noun–noun compound

1.

The single domain-based approach to metonymy

According to Radden and Kövecses (1999: 21), “metonymy is a cognitive process where one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same idealized cognitive model”. As the authors point out, the idea of metonymy operating within a single idealized cognitive model (icm) goes back to Lakoff (1987). Although a number of cognitive

* I wish to thank my audience at the 10th International Cognitive Linguistics Conference (held in 2007 in Cracow, Poland), especially Günter Radden and Linda L. Thornburg, for the very helpful remarks. I am also indebted to Antonio Barcelona and Francisco José Ruiz de Mendoza Ibáñez for their meticulous editing and insightful comments. . For a similar view, see also Kövecses and Radden (1998).

198 Réka Benczes

linguists have also emphasized this feature, they have done so with different terminologies: for instance, Croft (1993) claims that metonymy operates within a single domain matrix, Barcelona (this volume) argues for a functional domain, while Panther and Thornburg (1999) use the notions of frame (for a similar view, see also Barcelona, 2002; and Kövecses, 2006) and scenario. Nevertheless, heavy criticism has been voiced with regard to the single domain-based approach to metonymy. As pointed out by Peirsman and Geeraerts (2006a), domains seem to be too elusive constructs to build a definition on them. However, is it really necessary to reject a domain-based account just because the notion of domain is “vague” or that the boundaries of domains are not distinct and clear? I wish to argue that this should not be the case. Within cognitive linguistics it is commonplace to claim that word meanings are encyclopedic (for a discussion, see Langacker, 1987: 155–158). As Langacker points out, “concepts presuppose other concepts and cannot be adequately defined except by reference to them” (Langacker, 1987: 147) – this referential context is the domain. Every single word (concept) needs to be specified in more than one domain for its characterization (this is the domain matrix; ibid., p. 154), and every single concept can become a domain for another concept. Such a view is able to accommodate the various semantic extensions (based on similarity or association) between the original sense of the word and its (metonymical and metaphorical) extended senses – although with the obvious implications that the various connections between concepts and domains will vary to some extent from individual to individual – resulting in perhaps a degree of “vagueness”. Langacker (1987: 156) aptly summarizes this state of affairs: “since meaning is, in the last analysis, a matter of conceptualization (what else could it possibly be?), it strikes . A functional domain is essentially the same thing as a frame or icm, provided that the source and target are linked by what Fauconnier (1997) calls a “pragmatic function mapping” – for a detailed discussion, see Barcelona (this volume). . By the “single domain-based approach” I am referring to those definitions of metonymy that claim it to take place within one domain (matrix), frame or idealized cognitive model. . “It is gradually becoming clear, however, that this [domain (matrix) based] definition relies too much on the vague notion of ‘domain (matrix)’ to be fully operational” (Peirsman and Geeraerts, 2006a: 269–270). See also Feyaerts (1999: 318), who argues that since domains are “to a large extent individually determined”, it is rather problematic for linguists to describe them accurately. . Langacker’s (1987: 147) definition of what a domain is is perfectly succinct to operate with: “All linguistic units are context-dependent to some degree. A context for the characterization of a semantic unit is referred to as a domain. Domains are necessarily cognitive entities: mental experiences, representational spaces, concepts, or conceptual complexes.”



Putting the notion of “domain” back into metonymy 199

me as pointless to avoid the challenge of describing it in these terms, whatever the limitations on our present ability to do so.” A composite structure, such as a noun–noun compound, is based on the integration of two or more components, and this integration relies on correspondences established between the various domains of the component structures (that is, the constituent nouns). Some of these domains are more prominent, some are less so (Langacker, 1987: 156). Noun–noun compounds illustrate the remarkable ease with which we are able to combine concepts based on both prototypical features (as in the case of apple tree, ‘a tree that bears apples’), and on more marginalized ones (as in the case of family tree, ‘chart that shows all the people in a family over many generations and their relationships to one another’). One of the huge assets of domains is that by virtue of their inherent flexibility (see also Geeraerts, 1993), we are able to account for the creativity that speakers make use of when coining new words. In fact, Taylor (2002: 441) points out that incidental – peripheral – knowledge does also have linguistic manifestations (therefore what counts as central or peripheral knowledge with regard to a concept is indeed relative), as in the case of the much-quoted apple-juice seat (Downing, 1977), referring to the seat in front of which a glass of apple juice had been placed. Nevertheless, the flexibility of domains does not necessarily mean that words cannot have stable meanings: as Langacker emphasizes (1987: 163), more central pieces of knowledge will be activated each time the concept is invoked, while less peripheral ones will be “primed” (Croft, 1993: 344) by the context. Croft (1993) claims that metonymy is domain highlighting (when metonymy makes a domain primary that is not the primary one in the literal meaning of the word), as opposed to metaphor, which is domain mapping. However, I do agree with Barcelona (this volume), who maintains that domain highlighting can in fact be considered as a kind of mapping, though in this case the mapping is asymmetrical (as opposed to metaphor, where the mapping is symmetrical): “The metonymic source projects its conceptual structure onto that of the target, not by means of a systematic matching of counterparts (as in metaphor), but by imposing a conceptual (and linguistic) perspective from which the target is activated; this activation entails a shift in (conscious or unconscious) attention from source to target” (emphasis as in original). In fact, Barcelona’s (2003, this volume) and Croft’s (1993) views are compatible, since Croft (p. 348) does explicitly state . The definition comes from Collins Cobuild Dictionary on CD-Rom, henceforth CCCD. . Downing referred to such coinages as “deictic compounds”. It should be noted, however, that while apple-juice seat might be used to distinguish one seat from others in its vicinity, the lexicalization potential of such a compound is quite low because it is based on a very temporary relationship.

200 Réka Benczes

that metonymy is a “mapping” that takes place within a domain matrix. Kövecses and Radden (1998), Radden and Kövecses (1999) and Kövecses (2010/2002) also agree with the view that metonymy can be considered as a “kind of mapping”, where a more salient entity (the vehicle) provides mental access to a less salient entity (the target) within the same domain – although by doing so the vehicle may evoke other parts of the domain (matrix) as well. I agree with the view that metonymy provides mental access to a less salient domain or entity through a more salient one, and in the following, I will use the terms highlighting, mapping, and foregrounding synonymously.

2. Why domains are indeed necessary: Evidence from noun–noun compounds Croft (1993, 2006) – after Langacker (1987) – points out that one of the significant features of domains is that concepts may be profiled against multiple domains (although still within the same domain matrix), and a concept can function as a profile for another base, or as a base for another concept profile. What this implies is that multiple metonymies may be at work within a single domain matrix. Usually, in the various scholarly studies on metonymy (including cognitive linguistic literature), the linguistic examples provided by the authors bear evidence for a single metonymic mapping. For instance, in the sentence Proust is tough to read (Croft, 1993), the underlying conceptual metonymy, producer for product, enables us to understand Proust as Proust’s works. However, the situation is not so straightforward in compounds, where multiple metonymies (and metaphors) can be at work in the meaning of the composite expression. Such a case is provided by couch potato (‘a lazy person whose recreation consists chiefly of watching television and videos’ – CCCD). Both constituents are metonymical and involve the multiple profiling of domains: couch, for instance, profiles the place where this piece of furniture is found (the living room, typically in front of the television) and its function (sitting on it, relaxing at home) – while keeping other pieces of information – that is, domains – in the background (such as its shape or appearance) – see Figure 1.

. In the cognitive linguistic literature there is no consensus either on the issue of whether metonymy takes place between entities or domains (for an overview, see Barcelona, 2002, who suggests using the term subdomain). I will nevertheless use the notion of domain, because it is able to capture the sometimes very general, abstract representational spaces – contexts – that can play a role in a compound’s meaning (as it will be shown below in the case of the noun–noun compound couch potato).

Television

Family friends

Function D: to sit on (with others) to lie on

Living room D: most important piece of furniture in living room

Function D: room for socializing, watching television

Location D: in living room

Figure 1.  The domains underlying couch potato

Peripheral domain

Central domain

D domain

Feature D: can have cushions

Shape D: big, comfortable

Shape D: long, soft seat with a back and arms

Furniture

Shopping D: can be bought in furniture shops

Couch Potato

Home

Health problems

Eating D: people eat it

Healthy living

Obesity D: people get obese from too much fat

Chips D: chips are made from them

Cooking D: can be cooked, fried

Plant D: a type of plant

Fast food

Dieting

Diet D: for losing excess weight

Fat D: result of eating fatty food

French fries D: french fries made from them

Vegetable D: a kind of vegetable

202 Réka Benczes

The location and function domain provides the location for action metonymy, whereby the place of the activity (couch) stands for the activity itself (watching television). Potato, on the other hand, profiles the food that is consumed in front of the television (potato-based snack, such as chips), resulting in the material for object metonymy (potato for chips), which then becomes the profile for the base of typical foodstuff consumed in front of the television, that is, the member of a category for the category metonymy (chips for any kind of snack eaten before the television). At the same time, potato is also profiled against the person who is eating it – object involved in the action for the agent of the action –, but the real beauty of this profile–base relation is that it also evokes a further metonymy, namely cause for effect, since eating too much fatty food results in surplus fat. The cause for effect metonymy raises a very acute question with regard to the distinction between metaphor and metonymy on the basis of the number of domain matrices involved. It can be argued that a further meaning extension is involved in the interpretation of the compound, whereby the person on the couch – the agent of the action – is likened to a potato. The motivation behind this cognitive process is evident: an overweight person with excess fat resembles the sturdy, round shape of a potato. The problem with this conceptual shift is the following: if this is a metaphor, then it should comprise a mapping between two separate domain matrices – as, based on the above, this is a decisive feature with regard to metaphor (two domain matrices) and metonymy (single domain matrix). Nevertheless, it sounds odd and counterintuitive to claim that the person who spends too much time before the television snacking on chips on the one hand, and the shape of a potato to which this person is likened to on the other hand, belong to two different domain matrices. Especially so if one considers the metonymic relationships alluded to above, existing between the agent of the action and the object involved in the action, and between the cause (eating) and effect (excess weight). Therefore, it seems more natural to conclude that a conceptual metaphor – unfit, overweight person is a potato – is also at work within the same ­domain matrix. Such a conclusion, however, implies that metaphors can also operate within a single domain matrix, just as metonymies do. This observation is of course by no means novel: it has been pointed out by a number of cognitive linguists as well, such as Feyaerts (1999: 319), who for this reason rejects the idea of domain matrices serving as a criterion for distinguishing between metaphor and metonymy: “As a matter of fact, one cannot exclude metaphoric mappings taking place within the boundaries of a domain matrix. Consequently, my claim . It can be argued that the metaphor in this compound is licensed by the metonymies object involved in the action for the agent of the action and cause for effect.



Putting the notion of “domain” back into metonymy 203

is that metaphor primarily depends on something different than a crossing of domain boundaries”. Actually, metaphors and metonymies can and do interact rather often in linguistic expressions; Goossens (2002/1990) for instance created the term “metaphtonymy” to refer to the process when both metaphor and metonymy act upon an idiomatic expression. For example, to beat one’s breast (‘to make an open noisy show of sorrow that may be partly pretence’) represents a case of metaphor from metonymy: the metonymic basis is the religious practice of beating one’s breast while one confesses one’s sins publicly. This image is then mapped through metaphor onto non-religious situations as well. In a more sophisticated analysis, Ruiz de Mendoza and Díez Velasco (2002: 518–521) argue that the expression is based upon a blend, where one of the input spaces (the source of the metaphoric mapping) is created by a metonymy: the action of beating one’s breast to express sorrow. The other input space (the target of the metaphoric mapping) is the situation to which the expression refers. The blended space inherits the metonymy-induced relationship between the action of beating one’s breast and expressing sorrow,10 as well as the corresponding elements of the specific situation (from the other input space). As the authors claim, the expression is, therefore, a case of a metonymic expansion of a metaphoric source.11 Reference should also be made to Geeraerts (2002), who analyzed the interaction of metaphor and metonymy in composite expressions, such as idioms and compounds. He claims that there are plenty of compounds that are neither purely metaphorical, nor purely metonymical, but involve both types of meaning extension. According to Geeraerts, such cases can be classified into three main categories: metaphor and metonymy can occur in a compound expression (1) consecutively; (2) in parallel; and (3) interchangeably.12 10. Referred to as “highlighting” by the authors. 11. Ruiz de Mendoza and Díez Velasco’s (2002) work on the interaction between metaphor and metonymy for a metonymy-based multimodal metaphor analysis has been very productively used, among others, by Urios-Aparisi (2009) and Hidalgo-Downing and Kraljevic Mujic (to appear, 2011), since it goes well beyond the initial metaphtonymy proposal as put forth by ­Goossens (2002/1990). 12. An example of the first category is schapenkop (“sheep’s head”, i.e., ‘dumb person’), where the analysis of the compound’s meaning involves two steps: first, “sheep’s head” is metaphorically extended towards the reading ‘a human head like that of a sheep, a stupid head’, and second, a metonymical step leads to ‘a person with a head like that of a sheep, a stupid person’. The compound droogkloot (“dry testicle”, i.e., ‘boring person, bore’) exemplifies the second case, when metaphor and metonymy act upon the meaning in parallel. The compositional, literal reading “dry testicle” metonymically stands for “a person with dry testicles”. This reading, however, serves as the input for a further metaphorical extension, leading to the boring person sense. Badmuts (“swimming cap”, i.e., ‘bald person’) represents the third category, i.e., the interchangeability of metaphor and

204 Réka Benczes

More recently, in Benczes (2006) I have analyzed a number of noun–noun compound expressions where an interaction of metaphors and metonymies was at work in the meaning of the composite expression, and in the majority of cases it was difficult to work out the sequence of the extensions (i.e., whether the metaphorical extensions came from the metonymies, or vice versa).13 My conclusion there was that the sequence of metaphors and metonymies was not as relevant for the understanding of the compound’s meaning as the fact that we can rely on a number of “interpretation routes” to access the meaning of the compound. Which route of interpretation is chosen depends upon a number of factors, such as the individual differences that might exist between among the domains (matrices) of speakers, or the context itself (as also alluded to by Croft, 1993). This is where we can go back to couch potato once again. Although I agree with Feyaerts (1999), that perhaps the domain-based approach is not enough in itself to distinguish between metaphor and metonymy, I do wish to maintain that domains play a very significant role in these cognitive processes and therefore should not be left out of a definition of metonymy (or metaphor for that matter). Figure 1 clearly shows that both constituents of the compound, couch and potato, have links to a number of domains – semantic associations – which do not play a role in the meaning of the compound expression (these are the peripheral domains, signified by the dotted ellipses), as opposed to those domains that are necessarily invoked (the central domains, signified by the full lines). Langacker (2006: 119) calls attention to the fact that there are many words that do not actually refer to anything in the real world but exist “at a higher level of conceptual organization”.14 Nevertheless, when we refer to things as village, flock or galaxy, we conceptualize a “non-random distribution of elements within a continuous field” (ibid.). This process, leading to the creation of domains, is probably natural and at

metonymy. According to Geeraerts, the compound can be analyzed in two ways: either “swimming cap” leads metonymically to ‘a person who looks as if he was wearing a swimming cap, a bald person’, or “swimming cap” is metaphorized as ‘a head that looks as if it is covered by a swimming cap, a bald head’, and from there metonymically extended to ‘a bald-headed person’. As the author points out, the semantic explanation can go either way, there is no principle by which one line of analysis can be favored over the other. 13. Note that both Goossens (2002/1990) and Barcelona (this volume) claim that the conceptual motivation of metonymy by metaphor (“metonymy from metaphor”) is extremely rare, if it exists at all. 14. Consider archipelago, for example, which describes a number of islands in a large body of water. However, by specifying the location of every individual island we are able to give a description of the scene – there is actually nothing distinctive about those islands in a large body of water.



Putting the notion of “domain” back into metonymy 205

the same time unavoidable, as it helps us to cope with the world. Langacker (2006) also makes note of the fact that within a domain every element is equivalent with the other; however, by the process of relation not only do the elements become distinguishable, but also enjoy varying degrees of importance. The implications of this idea for a noun–noun compound such as couch potato is that the two constituent nouns serve as reference points (Langacker, 1993)15 through which we are able to access the whole domain, and, at the same time, the unit that the two constituent nouns form also influences what domains become foregrounded or backgrounded within the compound’s meaning as a whole. The backgrounded domains, however, can serve as further reference points to access other domains within the network. This reasoning is in full agreement with Langacker’s (1987) views on compositionality: he maintains that linguistic phenomena are more likely to show partial compositionality than to be fully compositional. Composite structures (such as compounds) do follow conventional patterns of composition, that is, the relation that they bear to their components is not random, nor arbitrary. Yet composite structures are not constructed out of their components, nor are they “consistently or fully predictable” (Langacker, 2000: 16): “Rather than constituting a composite structure, the component structures correspond to certain facets of it, offering some degree of motivation for expressing the composite conception in the manner chosen” (ibid., italics as in original) – that is, the overall meaning of the compound evokes an elaborate semantic network that is partially motivated by the component elements. In fact, Croft (1993) calls attention to the fact that in the majority of cases, figurative language requires the “adjustment” of the domains of the conceptual elements in order to satisfy what he refers to as the “conceptual unity of domain” (which is a coherent semantic unit, and can range from a compound to a phrase or a sentence). Therefore, the couch potato conceptual unit forces us to adjust the domains of couch and potato – highlighting some domains and backgrounding others – so as to create a possible composite meaning. These adjustments are possible because we are able to rely on the domains and domain matrices that surround our concepts – yet as Langacker (2006: 108) points out, how we access domains varies, depending on the conceptual unit (whether we are talking about a couch potato, a couch cushion or a couch shop) – therefore “no two uses of a lexeme are precisely equivalent semantically”. 15. As Langacker (1993) claims, metonymy is the ability to invoke the conception of one entity for purposes of establishing mental contact with another. The process is so ubiquitous that we do not even realize it: for example, the computer’s on-switch can only be conceptualized through the whole computer. Langacker terms the former as “target” and the latter as “reference point”.

206 Réka Benczes

3.

Metonymy as a domain network of conceptual relationships

The adjustment of domains – that is, which domains become more salient within the domain matrix – is also complemented by our knowledge of other compound expressions. As Ryder (1994: 85–93) points out, when a novel compound is coined, it then needs to be interpreted by the listener, who makes use of a number of strategies when coming across with it. Although the listener cannot be sure whether the speaker based the novel compound on an already existing compound pattern, the listener will nevertheless use his own repository of “templates” when trying to interpret the novel expression. In the case of a noun–noun compound, the templates specify the semantic relationship between the two constituent elements, such as cause–effect (as in hayfever), part–whole (armchair) or location (fridge magnet). Very often there is a whole range of templates to choose from. In such a circumstance the listener checks what information the context can give about the compound’s meaning. These various compound patterns – schemas – may also be evoked when coming across a compound expression. Schemas vary in their strength: a schema gains strength, or becomes “established” or “entrenched”, in proportion to the number of instances which elaborate it. Schemas that are elaborated by many instances will tend to be highly entrenched; while those that have a small number of instances will be less salient; that is, they will be weakly entrenched (Taylor, 2002). Consider, for instance, compound expressions that have an [adjective denoting a color+collar] combination in their modifier position, such as blue-collar worker (‘an employee who performs manual or technical labor, such as in a factory or in technical maintenance trades’) and white-collar worker (‘an employee who performs clerical or intellectual work, such as those in clerical, professional, managerial or administrative positions, as opposed to people who do manual work’). In Benczes (2006) I maintain that in both of these compounds there is a part–whole metonymy operating on the first, modifying constituent, whereby the collar stands for the whole outfit (more precisely, the shirt worn by the employees). According to Webster’s, the origin of both blue-collar and white-collar worker comes from the dress codes found at their respective workplaces. Industrial blue-collar workers formerly and to a large extent today still do wear “work clothes”, where the shirts are of a navy blue color. The clothes are from a durable material and are intended to get dirty while working. However, white-collar workers wore traditional, white, button-down shirts originally. As white shirts are easily soiled, white-collar workers did not do physical labor. While menial jobs did not require formal education, the white-collar ones did: only those were employed in such positions who had a good secondary or even higher-level education.



Putting the notion of “domain” back into metonymy 207

However, what is intriguing about both blue-collar and white-collar worker is that the color of their shirts is not restricted to blue and white, respectively; instead, they have come to represent a category of individuals in the work force. What this implies is that a further metonymy, member of a category for the category (a very common metonymy in creating social stereotypes, as has been pointed out by Lakoff, 1987)16 acted upon the two constructions as a whole, whereby blue-collar worker and white-collar worker came to stand for the categories of physical and clerical professions, respectively. However, in order to explain the generalization of the meaning of the two compounds, it can be also argued that the salient property for category conceptual metonymy could just as well have operated on the compounds as wholes, whereby the salience of the shirt color (blue or white) came to stand for the occupational field (category). It is unclear which of these is the more likely candidate; it might even be the case that both metonymies had a role in the meaning generalization of the compounds. What is remarkable about this [color adjective + collar] schema (or template) is that it has served as the basis for a number of other compounds as well, such as pink-collar worker (‘a woman in a low-paid job, especially in an office or restaurant’), scarlet-collar worker (‘a woman who owns or operates an Internet pornographic site’) or green-collar worker (‘environmentalist’).17 While in the case of pink-collar worker it can be argued that waitresses, hairdressers, and beauticians typically wear pink outfits in America, and therefore similar metonymic processes are involved in the compound’s meaning as in white or blue-collar worker, no such thing can be claimed in the case of scarlet or greencollar worker. It seems rather that both compounds have “inherited” to some degree the metonymical relationships of the schema (member of a category for the category and salient property for category).18 The schema has also undergone semantic bleaching – collar does not have a role anymore within the meaning of the compound as whole. Nevertheless, collar does play an important role in scarlet and green-collar worker, as it preserves the ties to the

16. See, however, Koskela (this volume), who notes that the category–member relationship is not on a par with contiguity relationships involving entities (such as container for contents), and therefore should rather be treated as a case of vertical polysemy. 17. Source: http://www.wordspy.com 18. The color scarlet stands for sexuality (due to arousal, the sexual organs fill with blood), while green is motivated by the color of vegetation (which metonymically activates the domains of nature and nature preservation.

208 Réka Benczes

original coinages – blue and white-collar worker – and therefore maintains the metonymical relationships, too.19 What couch potato or the various worker compounds imply is that the domains are networks of semantic associations, with links to further semantic domains (as signified by Figure 1, where the domains provide links to other representational spaces, indicated by capital letters) or even other grammatical constructions (such as the [color adjective + collar] schema). Therefore, I propose that metonymy operates within a domain network, instead of a domain matrix or icm, where the domains constitute an intricate, web-like set of semantic associations. The term “domain network” reflects the nature of domains much better than previous concepts (such as domain matrix, icm or frame) because it concentrates on the links that can be exploited in meaning comprehension. Needless to say, the idea of networks is nothing new; Langacker (1987) explicitly states that “cognitive domains” are networks of intersecting categories. Also, Radden and Kövecses (1999: 20) make note of the fact that “icms and the network of conceptual relationships characterizing them give rise to associations which may be exploited in metonymic transfer” (emphasis added).20 Moreover, at the end of their article, Peirsman and Geeraerts (2006: 311) propose a “network-like expansion of the schematic domain matrix account”. The term domain network aptly captures two of the main features of domains: that they do not have fixed boundaries and overlap with one another, and that there are no fixed associations – these depend on the lexeme, the context, individual differences, etc. These networks of conceptual relationships lie at the heart of understanding the meaning of metonymical compounds, where meaning processing is enhanced by the associative relationships evoked by the component nouns. Such a view is in harmony with Langacker’s (2000: 174) characterization of reference-point constructions, which are “inherently dynamic”: a target within a domain can be invoked as a further reference point to reach another target. As Lamb (1998: 190) underlines, a speaker’s choice of a lexeme or the choice of one among a set of appropriate conceptualizations is influenced by the context. Lamb 19. In fact, Taylor (2004) claims that every structure in language – such as the [color adjective + collar] schema is – motivated to some extent, as no linguistic structure exists in isolation: “it is not the case that each structure occupies a self contained pigeon-hole. Each unit has pointers to other units and can therefore be ‘filed’ under a number of different ‘addresses’” (p. 58). It is this characteristic of linguistic structures that Taylor refers to as “ecological motivation”, a feature that is fundamental to language, because the “learning, storage, and retrieval of linguistic units is facilitated – by a dense network of relations which structures the inventory” (p. 71). 20. However, according to Barcelona (this volume), the fact that two domains are connected by the same network does not necessarily mean that they will be connected metonymically. Other factors are also needed, such as the pre-existence of a privileged pragmatic connection between the roles of the domains in the network, i.e., the pragmatic function.



Putting the notion of “domain” back into metonymy 209

cites the following example: “Are you ready to zoom to the camera store?” (emphasis as in original). There are plenty of alternatives in English to represent the action described by zoom, all of which – as the author indicates – would seem to be a more likely choice (such as go, go over, get going, etc.). According to Lamb, the reason for the selection of zoom over all the other possibilities is that the concept of camera or camera store influences the activation of connections within the semantic network to other concepts related to cameras, such as zoom lens. From zoom lens a further connection is made to zoom – and as this concept is already being activated (due to the connection from camera (store) to zoom lens to zoom), “it wins out in the competition among lexemes” to express the desired motion (ibid.).

4.

Contiguity or association?

In the previous sections I have been talking about “semantic associations” that give rise to metonymy, deliberately being vague about the nature of these associations. Yet within the literature on metonymy what these relationships look like exactly has been open to much debate – namely, whether they are based on contiguity, a stand-for relationship or something else (for an overview, see Barcelona, 2003). As Radden and Kövecses (1999: 18–19) argue, metonymy cannot be taken as a stand-for relationship, because it is not a simple substitution of meaning, but rather an “interrelation” of entities to create a new complex meaning.21 Croft (2006), however, argues that contiguity is not adequate either to describe metonymic relationships, because spatial nearness does not necessarily license a metonymy and therefore “association” is a better term for the metonymic semantic relations operating within a domain network.22 Peirsman and Geeraerts (2006b), on the other hand, maintain that contiguity should not be considered as spatial nearness only, but rather as conceptual contiguity. The authors quote ­Dirven (2002: 90–91), according to whom “contiguity cannot in all cases be based on a form of objective or ‘natural’ contiguity. This has the implication that contiguity must be taken to mean ‘conceptual contiguity’ and that we can also have contiguity in those cases where we just ‘see’ contiguity between domains”. 21. Therefore, according to the authors, metonymic relationships are better described by “an additive notation such as x plus y” instead of x for y. 22. See also Paradis (2004: 255; also this volume), who favors the view that metonymy is restricted to those cases where a lexical item is used to express a “nonconventional, although contextually motivated meaning”. Conventionalized meaning relationships (readings within the senses of a word) are referred to by the author as examples of “zone activation”.

210 Réka Benczes

What this implies is that Croft’s “association”23 and contiguity are compatible, synonymous terms – provided that the latter is taken as conceptual contiguity. Conceptual contiguity allows for the flexibility with which domains are foregrounded within a network, depending on context, etc.; and is also able to explain why the meaning of compound nouns can be analyzed (processed) via a number of interpretation routes: which concepts within a domain network are considered as more contiguous varies from individual to individual. Nevertheless, using a broad terminology such as “conceptual contiguity” or association to describe metonymical relationships does also have its pitfalls. If we go back to Figure 1 again, there is the problem that the metaphor that is also at work within the domain network (unfit, overweight person is a potato) is based upon two conceptually contiguous domains24 (since these have actually licensed metonymies in the network): the agent of the action and the object involved in the action. While such a state of affairs does not necessarily imply that we have to refuse conceptual contiguity as the relationship licensing metonymy within a domain network, it does, however, point to the possibility that the distinction between metaphor and metonymy is far from straightforward in all cases. As Barcelona (2003: 239) provisionally suggested, a continuum approach towards metaphor and metonymy is, in all likelihood, more “realistic”.

5.

Domains: Tentative evidence from psycholinguistics

Although direct evidence for the existence of domain networks cannot be cited, results from a series of psycholinguistic experiments concerning compound processing do suggest that composite expressions involving metaphors and metonymies are decomposed to some extent into their constituents. Libben et al. (2003) have shown that both constituents of a noun–noun compound show priming effects, regardless of the degree of transparency involved, or the status of the constituent in question, that is, whether it functions as a modifier or as the head of the construction. Libben et al. have presented four types of English compounds to native speakers: those with transparent constituents, such as bedroom; those with a non-transparent modifier element but transparent head, such as godchild; those with a transparent modifier element but non-transparent head, such as jailbird;

23. According to Croft (2006: 322), association is “a relationship between concepts occurring in two different domains in a domain matrix”. 24. This observation does refute the hypothesis that metaphor involves a relationship between entities that are conceptually “distant” (see Dirven, 2002).



Putting the notion of “domain” back into metonymy 211

and those where both constituents are non-transparent, such as fleabag. According to their data, both constituents showed priming effects, and priming was observable in all four types of compounds. The results of the experiment are rather important with regard to the age-old storage versus computation issue, that is, how composite expressions are stored in the mental lexicon: as wholes or as decomposed items? As Libben (2006) elaborates, if the human mind sought to maximize computational efficiency, then compound words would be represented as whole lexical forms as soon as possible. Such an approach “would limit the need for morphological composition to the times when a speaker constructs a novel compound and would limit the need for morphological decomposition to the times when a listener encounters a novel compound” (p. 5). In all other cases, the compound could be extracted from the mental lexicon. However, if the human mind was created so as to maximize storage efficiency, then compounds would never be entered as wholes in the mental lexicon (so that there would be “no additional storage cost” – ibid.). In this scenario, every time a compound was encountered, it would have to be “decomposed into an ordered arrangement of morphemes and interpreted anew in terms of those constituents and that ordering” (ibid.). However, psycholinguistic research suggests that the human mind seeks neither of these two extreme positions with regard to the representation of compounds. In fact, evidence indicates that the mind stores and computes as much as possible. A frequently encountered compound (and which – consequently – has undergone lexicalization to various degrees) can be stored as a whole in the mental lexicon, but – as Libben (2006: 6) emphasizes – this does not mean that the process of morphological decomposition is “shut down” (ibid.): the links between the constituent morphemes and the compound as a whole are still preserved. Why does, however, the human mind adopt such a complicated architecture? According to Libben, the main answer is that such a makeup “does not require anything to be decided […] all representations that can be activated, will be activated” (p. 9). Therefore, novel words will be processed in terms of their constituents (as there is no whole-word representation to activate), while already existing words will vary to the degree to which they are processed by their constituents and as a whole – depending on their semantic transparency and the level of lexicalization they have gone through. As Libben points out, such an architecture favors “a connectionist rather than a symbolic approach to modeling” (p. 13). Moreover, compound production experiments (Costello, 2002) have shown that when a novel compound is coined, it is not the communicative precision that influences the selection of the constituent nouns, but the defining properties of the categories themselves – that is, which noun is able to evoke the semantically most productive domain network.

212 Réka Benczes

The connectionist approach advocated by Libben ties in with the encyclopedic view of word meaning, as well as the idea that every single concept needs a bunch of referential contexts for its specification (i.e., the domain). In fact, the relational network model of Lamb (1998) also runs along similar lines: in Lamb’s view, the lexical system differs from the traditional treatment of the lexicon in that it does not incorporate meanings, but connections to the meanings. Concepts that are closely related semantically are also connected relatively close within the system. The mind is not seen as a device for storing and rewriting symbols but as a network system, whose information lies in its connectivity. Understanding a compound on the basis of its components and understanding it as a unit (that is, without recourse to the meaning of the components) is not an either-or question: as Lamb argues, these two processes may operate in parallel.25

6.

Conclusion

In this chapter I have investigated the notion of domain and its relationship to metonymy. There have been plenty of criticisms raised against defining metonymy as operating within a single domain (matrix), icm, frame, etc. by a number of linguists, based on the objection that domains are too vague and are subject to individual differences. Yet in my detailed analysis of a noun–noun compound,

25. One of the consequences of the approach offered by Libben and Lamb is that the ambiguous issue of motivation and remotivation can be finally settled – that is, even though very often we have absolutely no idea how the original word formation process took place, it is nevertheless possible for speakers (including the analyzing linguist) to make sense of the products, and remotivate them. However, if the meaning of a complex lexicalized form (which, however, is compositional – that is, the speaker is aware of the complexity of the term) can be accessed both as a whole and by its constituents, then the speaker necessarily – and quite automatically – remotivates the meaning of the whole, based on the meaning of the constituent parts. Remotivation as an input to a productive process is illustrated by hamburger (Taylor, 2004: 60–61). According to the OED, the expression Hamburger steak was in use around the turn of the nineteenth century, and denoted a piece of meat made out of minced beef. Hamburger was derived by -er suffixation from Hamburg, as the steak was associated with the city of Hamburg. However, at a certain point in time, speakers began to reanalyze the meaning of Hamburger as ham + burger (which Taylor accounts to the phonological properties of the word), after which all sorts of variations sprang up containing burger (cheeseburger, chickenburger, vegetable burger, etc.), which acquired the status of a meaning bearing unit. Nevertheless, speakers’ awareness towards the compositionality of an item (and hence its availability for morphological analyzability) varies. As Lamb (1998: 171) argues, a linguist might be conscious of the non-productive -th affix in words such as breadth and width, which at the same time might not be evident for speakers outside the linguistic community.



Putting the notion of “domain” back into metonymy 213

couch potato, I have wished to demonstrate that domains are indeed necessary for any definition of metonymy, since it is actually a domain matrix, icm, frame, etc. that is able to license the rich semantic associations that arise within a composite expression. What is more, a domain-based account is able to explain why compounds are open to multiple interpretations and is also illustrative of how metaphors and metonymies can interact with one another in the comprehension of a single compound word. What couch potato or the various worker compounds imply is that domains are networks of semantic associations, with links to further semantic domains or even other grammatical constructions. For this reason I have proposed that metonymy operates within a domain network, instead of a domain matrix or icm, where the domains constitute an intricate, web-like set of semantic associations. Bearing in mind that the network-idea is nothing new in the cognitive linguistic literature, the term “domain network” nevertheless is able to reflect the nature of domains much better than previous concepts (such as domain matrix, icm or frame) because it concentrates on the links that can be exploited in meaning comprehension. I have also offered some tentative evidence for the notion of domain networks arising from psycholinguistic experiments from compound processing. The results imply that when encountering a compound word, even an often heard one, we both process it as a single lexical item and decompose it into its constituents – with the aim of extracting as much information out of the lexical item as possible. Compound production experiments (Costello, 2002) have also shown that the creation of a novel compound is influenced more by the defining properties of categories than by the communicative precision of the compounds. What this implies is that we rely on a dynamic nature of interpretation, based on semantic associations, taking place within the domain network, to access and process the meaning of such expressions. This view also ties in with the relational network model of Lamb (1998), which claims that the lexicon should be viewed as a network of connections to meanings (as opposed to a static list of meanings), where concepts that bear a close semantic relation to each other are also located relatively close to each other. Therefore, the mind is not seen as a device for storing and rewriting symbols but as a network system, whose information lies in its connectivity.

References Barcelona, Antonio. 2002. Clarifying and applying the notions of metaphor and metonymy within cognitive linguistics: an update. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (207–277). Amsterdam & Philadelphia: John Benjamins.

214 Réka Benczes

Barcelona, Antonio. 2003. Metonymy in cognitive linguistics: An analysis and a few modest proposals. In Cuyckens, Hubert, Thomas Berg, René Dirven, and Klaus-Uwe Panther (eds.), Motivation in Language. Studies in Honor of Günter Radden (223–255). Amsterdam & Philadelphia: John Benjamins. Benczes, Réka. 2006. Creative Compounding in English. The Semantics of Metaphorical and Metonymical Noun–Noun Combinations. Amsterdam & Philadelphia: John Benjamins. Collins Cobuild Dictionary on CD-Rom, The (CCCD). Costello, Fintan J. 2002. Investigating creative language: People’s choice of words in the production of novel noun–noun compounds. In Proceedings of the 24th Annual Conference of the Cognitive Science Society, 2002. http://inismor.ucd.ie/~fintanc/postscript/Costello2002a.ps; accessed: 21 November, 2006. Croft, William. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4: 335–370. Croft, William. 2006. On explaining metonymy: Comment on Peirsman and Geeraerts, “Metonymy as a prototypical category”. Cognitive Linguistics 17(3): 317–326. Dirven, René. 2002. Metonymy and metaphor: Different mental strategies of conceptualisation. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (75–111). Berlin: Mouton de Gruyter. Downing, Pamela. 1977. On the creation and use of English compound nouns. Language 53(4): 810–842. Fauconnier, Gilles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Feyaerts, Kurt. 1999. Metonymic hierarchies: The conceptualization of stupidity in German idiomatic expressions. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (309–332). Amsterdam & Philadelphia: John Benjamins. Geeraerts, Dirk. 1993. Vagueness’s puzzles, polysemy’s vagaries. Cognitive Linguistics 4(3): 223–272. Geeraerts, Dirk. 2002. The interaction of metaphor and metonymy in composite expressions. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (435–465). Berlin: Mouton de Gruyter. Goossens, Louis. 2002/1990. Metaphtonymy. The interaction of metaphor and metonymy in expressions for linguistic action. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (349–377). Berlin: Mouton de Gruyter. Hidalgo-Downing, Laura and Blanca Kraljevic Mujic (to appear, 2011). The interaction between multimodal metonymy and metaphor in ICT advertising discourse: Meaning creation as a complex discourse process. Review of Cognitive Linguistics 9(1). Kövecses, Zoltán. 2006. Language, Mind, and Culture: A Practical Introduction. Oxford: Oxford University Press. Kövecses, Zoltán. 2010/2002. Metaphor: A Practical Introduction. 2nd, revised edition. Oxford: Oxford University Press. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9(1): 37–77. Lamb, Sydney M. 1998. Pathways of the Brain: The Neurocognitive Basis of Language. Amsterdam & Philadelphia: John Benjamins. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: The University of Chicago Press.



Putting the notion of “domain” back into metonymy 215

Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Volume I: Theoretical Prerequisites. Stanford: Stanford University Press. Langacker, Ronald W. 1993. Reference-point constructions. Cognitive Linguistics 4(1): 1–38. Langacker, Ronald W. 2000. Grammar and Conceptualization. Berlin: Walter de Gruyter. Langacker, Ronald W. 2006. On the continuous debate about discreteness. Cognitive Linguistics 17(1): 107–151. Libben, Gary. 2006. Why Study Compound Processing? An overview of the issues. In ­Libben, Gary and Gonia Jarema (eds.), The Representation and Processing of Compound Words (1–22). Oxford: Oxford University Press. Libben, Gary, Martha Gibson, Yeo Bom Yoon, and Dominiek Sandra (2003). Compound fracture: The role of semantic transparency and morphological headedness. Brain and Language 84: 50–64. Oxford English Dictionary, The (OED). 1989. 2nd edition. Clarendon Press: Oxford. Panther, Klaus-Uwe and Linda L. Thornburg. 1999. The potentiality for actuality metonymy in English and Hungarian. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (333–357). Amsterdam & Philadelphia: John Benjamins. Paradis, Carita. 2004. Where does metonymy stop? Senses, facets and active zones. Metaphor and symbol 19(4): 245–264. Peirsman, Yves and Dirk Geeraerts. 2006a. Metonymy as a prototypical category. Cognitive Linguistics 17(3): 269–316. Peirsman, Yves and Dirk Geeraerts. 2006b. Don’t let metonymy be misunderstood: An answer to Croft. Cognitive Linguistics 17(3): 327–335. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought (17–59). Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza, Francisco José and Olga Isabel Díez Velasco. 2002. Patterns of conceptual interaction. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (489–532). Berlin: Mouton de Gruyter. Ryder, Mary Ellen. 1994. Ordered Chaos: The interpretation of English noun–noun compounds. Berkeley & Los Angeles: University of California Press. Taylor, John R. 2002. Cognitive Grammar. Oxford: Oxford University Press. Taylor, John R. 2004. The ecology of constructions. In Radden, Günter and Klaus-Uwe Panther (eds.), Studies in Linguistic Motivation (49–73). Berlin & New York: Mouton de Gruyter. Urios-Aparisi, Eduardo. 2009. Interaction of multimodal metaphor and metonymy in TV commercials: Four case studies. In Forceville, Charles and Eduardo Urios-Aparisi (eds.), Multimodal Metaphor (95–118). Berlin & New York: Mouton de Gruyter.

What do metonymic chains reveal about the nature of metonymy? Rita Brdar-Szabó and Mario Brdar Eötvös Loránd University, Budapest / University of Osijek

The central issue that concerns us in this chapter is whether metonymy should be conceived as a mapping. The way metonymies function in authentic discourse indicates that we have two-way traffic. The initial conceptual substrate is designated by the source concept, but it is plastic and flexible enough to allow considerable customizing, often within complex metonymic networks. The inferences that steer the customizing are guided by the information based on text (i.e., cotext) and context (circumstances). It is argued that metonymy should best be approached as an inference-based domain elaboration (either expansion or reduction) of the metonymic source, in the course of which domains are tailored to an optimal conceptual measure with regard to their function. Keywords: domain elaboration, inference, discourse, mapping, metonymic chain

1.

Introduction

The general question that this volume raises, i.e., What do we want to call metonymy?, as well as some related questions, such as Is metonymy a mapping, and if so, what type of mapping? and Is the distinction between metaphor and metonymy in terms of the same/different domain criterion really adequate?, may best be answered by studying metonymy in action, i.e., as it evolves and functions in actual usage. This means that it should be studied as being embedded in authentic discourse. Two complementary enterprises are the central parts of such a research programme: large-scale corpus-based research on metonymy, and small-scale, detailed study of metonymy in context, along the lines of Barcelona (2005, 2007a). The aim of this contribution is to corroborate the above claim by studying in detail authentic uses of metonymy in context, primarily in light of the issue as to whether metonymy should be conceived as a mapping. Specifically, we concentrate on a well-known type of low-level metonymy, capital for government, in the discourse of media.

218 Rita Brdar-Szabó and Mario Brdar

Our central hypothesis is that in deciding whether it is appropriate to call metonymy a mapping we should be guided by the function(s) of metonymy in discourse, viz. by its role in the online construction of meaning. If metonymy should be modeled in terms of mapping, it is a type of mapping that is on several counts very different from the type of mappings we are used to seeing in the mainstream cognitive linguistic approach to metaphor. It will be claimed that the evidence from the study of the way metonymies work in discourse indicates that it should best be approached as an inference-based elaboration (either expansion or reduction) of the metonymic source. The organization of the chapter is as follows: Section 2 is an overview of several points of difference between metonymy and metaphor, as discussed in literature. In Section 3 we consider the metonymy-as-a-mapping position as found in recent literature. In Section 4 we then turn to metonymic chains and to the functions of metonymy in discourse and consider what these tell us about the nature of metonymy, while pointing out some theoretical and descriptive difficulties that the mapping assumption is bound to give rise to when metonymy is studied in action, i.e., in actual usage in authentic discourse. We also discuss the possibility that some prepositional phrases with names of capitals may also function as parts of such metonymic chains (thus expanding on the sense of the central question of the present volume, and producing a more specific question about the realization of metonymies, i.e., What categories do we want to call metonymy? Section 5 returns to the inadequacy of the metonymy-as-a-mapping approach and presents some tentative conclusions as well as suggestions for further research.

2.

Metaphor vs. metonymy

Within the cognitive linguistic framework, both metaphor and metonymy are considered to be cognitive operations involving domains and mappings. They have been contrasted with respect to five marked points of difference, although it has been repeatedly claimed that the borderline between the two is blurred (cf. Barcelona, 2000a, 2000b; Ruiz de Mendoza, 2000; Radden, 2002). As for the first point of difference, metonymy is in most traditional approaches usually contrasted with metaphor (and occasionally with synecdoche) concerning the nature of the relationship existing between the vehicle and the target. Firstly, metonymy is claimed to be based on contiguity, whereas metaphor is seen as resting on similarity (cf. Ullmann, 1962: 212; Taylor, 1989: 122). Contiguity is here taken to encompass all associative relations except similarity. Metaphors are in fact often considered to be shortened similes, i.e., two entities are brought into correlation as exhibiting some similarity, but there are no



Metonymic chains and the nature of metonymy 219

function words that would make this comparison explicit. In other words, something is described by mentioning another thing with which it is assumed to implicitly share some features: (1) Don’t worry, for the grandparents there is a health spa on-site with a Jacuzzi, solarium, indoor pool and gym which is a real oasis of relaxation as well as golf nearby and the Brookes Restaurant which offers award-winning cuisine and superb service.

In the above example, oasis is not used in its literal sense, ‘a place with water and trees in a desert’ but rather to denote a place that is different from its surroundings in a pleasant way, i.e., it is offers comfort and relaxation. The rationale for its metaphoric use is the fact that literal oases are places that are very different from their desert surroundings. They provide something that is missing in the desert, but craved by the people and animals traveling through the desert, viz. water, shade, and vegetation. In its metaphoric use, oasis very often implies a calm, quiet rural retreat with the promise of peace and relaxation (as opposed to the bustling cityscape). Note that similes are considered in traditional rhetoric as a figure on their own, and that some stereotyped (and often semantically opaque) similes are recognized as a type of phraseological units: (2) After a good night’s sleep he woke up feeling as fresh as a daisy.

While phraseological and literal similes always spell out the common ground (real or fictitious) between the two entities compared, and bring these entities explicitly into an equating relationship, metaphors may remain quite silent on this common ground or leave one of the entities unmentioned, in which case the rationale for using a metaphor must be figured out from the context. There are, however, metaphorical expressions that happen to suggest this common ground, i.e., the target domain, by mentioning the other entity within the phrasal units containing the metaphor, one particular type being appositive of-phrases: (3) The Valium pill of Mr Arafat’s release was also designed to soothe this dangerously overheated talk.

The release of Mr Arafat by Israeli forces is presented in the above example as an anxiety-reducing drug because of its temporarily soothing effect on the Middle East peace process. Metonymy, on the other hand, is traditionally approached as a stand-for relationship that is, unlike metaphor, not based on similarity but on contiguity or proximity. This means that metonyms are expressions that are used instead of some other expressions in such a way that the latter expressions are associated with, or suggested by, the former ones:

220 Rita Brdar-Szabó and Mario Brdar

(4) He was testifying on the Hill earlier in the week. (5) London is determined to grab a share of Iraq’s oil once Saddam is overthrown.

In Example (4), the Hill, short for Capitol Hill, is not used to denote this particular location in Washington, i.e., the hill where the Capitol building stands, or not even so much this particular building, as the institution of the US Congress which meets in this building. Similarly, in (5) London, collocating with the adjective determined which presupposes an animate subject that can be attributed will/intention, is not meant to simply refer to the capital of the UK but to the government of the UK, London being its official seat. The notion of contiguity is taken in its broadest sense to cover all associative relations except similarity, and may thus be too vague a notion, making metonymy almost a cognitive wastebasket. Panther and Thornburg (2002: 282) attempt to constrain the scope of metonymy by submitting that the relation between the metonymic source and the target is contingent, i.e., it does not exist by conceptual necessity. When a nurse, for example, refers to a patient as the ulcer in room 506, it is not conceptually necessary for the ulcer to belong to the patient in room 506. Metonymic relation is thus in principle defeasible or cancelable, because the source concept is still usually retrievable (though backgrounded), even if the target concept is conventionalized in the lexicon. The second important point of difference between metaphor and metonymy observed by cognitive linguists has to do with whether the mapping takes place across distinct conceptual domains or within a single domain (or idealized cognitive model, also called script, scenario, or frame in cognitive linguistic literature). The standard view is that a metonymic mapping occurs within a single domain, while metaphoric mappings take place across two discrete domains. The differences between the two types of mappings can be presented schematically, as depicted in Figure 1. As for the directionality of the two types of mappings as a third point of difference, metaphors typically employ a more concrete concept or domain as the source in order to structure a more abstract concept or domain as target. In the majority of cases, elements from the physical world are mapped onto the social and mental world. Metaphorical mappings are thus normally unidirectional, and the source and target are not reversible. The situation with metonymies is quite different. Metonymic mappings can proceed in either direction, from the more concrete part of the domain (i.e., the subdomain) to the more abstract one, and the other way round, as shown in Figure 1 below. According to Radden and Kövecses­ (1999: 22), “[i]n principle, either of the two conceptual entities related may stand for the other, i.e., unlike metaphor, metonymy is basically a reversible process.” This is quite obvious when we consider pairs of metonymies such as



Metonymic chains and the nature of metonymy 221

Metaphor

Metonymy

Figure 1.  Metaphor and metonymy distinguished on the basis of domain inclusion

cause for effect (e.g., healthy complexion) and effect for cause (e.g., slow road), ­generic for specific (e.g., Boys don’t cry, Radden and Kövecses, 1999: 34), and specific for generic, (e.g., The/A spider has eight legs, ibid.) etc. Another crucial point of difference between metaphor and metonymy has to do with the number of mappings taking place. Conceptual metaphors are characterized by whole sets of systematic conceptual correspondences between certain elements of the source domain on the one hand and their counterparts that are elements of the target domain, on the other. Technically, these conceptual correspondences are often referred to as mappings. Metaphorical expressions may work on the basis of a set of correspondences, though some may exploit only one, while metonymic mappings are based on a single correspondence (cf. Ruiz de Mendoza and Peña, 2002). The differences between the two types of mappings can now be presented schematically as shown in Figure 2. Finally, metaphor and metonymy are said to have different functions. According to Lakoff and Johnson (1980: 36f) metaphor is “principally a way of conceiving of one thing in terms of another, and its primary function is understanding” while metonymy “has primarily a referential function, that is, it allows us to use one entity to stand for another.” This difference is typically reflected in their realizations. Metaphors often function as predicative expressions (with an explicit or implicit verbal element), whereas metonymies function as nominal expressions, most commonly functioning as subjects and objects.

222 Rita Brdar-Szabó and Mario Brdar

Metaphorical mappings

Metonymic mapping

Figure 2.  Metaphor and metonymy distinguished on the basis of the number of conceptual mappings (correspondences) involved

However, both of the above statements have to be relativized. While Lakoff and Johnson (1980: 36f) see metonymy as primarily having a referential function, they are aware of its additional functions. They point out not only that metonymy is “naturally suited for focusing” (Lakoff and Johnson, 1980: 37ff), but that, just like metaphor, it can have a role in the construal, in other words in making it possible for us to see and understand things in alternative ways. Furthermore, both metaphor and metonymy can be used predicatively and referentially. We will return to the functions of metonymy in Section 4.

3.

Is metonymy a mapping?

As indicated above, one of the main differences between metaphor and metonymy is the number of correspondences or mappings. In the case of metaphors, we typically have more than one, while metonymies involve a single mapping. However, in view of the differing natures of the two cognitive operations, the different number of domains involved, and their different function, it is somewhat strange that the terms “mapping” and “correspondence” should be applied to both. When using terms such as “correspondence” and “mapping”, we must be aware of their metaphorical nature and their implications. Lakoff and Johnson (2003: 253f) themselves acknowledge in retrospect the use of (partially inadequate) metaphors in their discussion of metaphor:



Metonymic chains and the nature of metonymy 223

Our first metaphor for conceptual metaphor came from mathematics. We first saw conceptual metaphors as mappings in the mathematical sense, that is, as mappings across conceptual domains. This metaphor proved useful in several respects. It was precise. It specified exact, systematic correspondences. It allowed for the use of source domain inference patterns to reason about the target domain. Finally, it allowed for partial mappings. In short, it was a good first approximation. However, the Mathematical Mapping metaphor proved to be inadequate in an important respect. Mathematical mappings do not create target entities, while conceptual metaphors often do. For example, time doesn’t necessarily have a use and isn’t necessarily a resource. Many people in cultures around the world simply live their lives without being concerned about whether they are using their time efficiently. However, other cultures conceptualize time metaphorically as though it were a limited resource. The time is money metaphor imposes on the time domain various aspects of resources. In doing so, it adds elements to the time domain, creating a new understanding of time. In order to account for this creative aspect, it was necessary to find a more adequate metaphor for conceptual metaphor. We needed a way to think about metaphors so that they could not just be mappings but also could add elements to a domain. We accordingly adopted the Projection Metaphor, based on the image of an overhead projector. We saw a target domain as an initial slide on the projector and metaphorical projection as the process of laying another slide on top of the first one, adding the structure of the source to that of the target. This metaphor for metaphor allowed us to conceptualize the idea that metaphors add extra entities and relations to the target domain.

It is advisable at this point to consider how these metaphors have actually been used and understood in the Lakovian tradition. Although mapping and correspondence have been equated, as for example in: The love-as-journey mapping is a set of ontological correspondences that characterize epistemic correspondences by mapping knowledge about journeys onto knowledge about love. Such correspondences permit us to reason about love using the knowledge we use to reason about journeys.  (Lakoff, 1993: 207)

we cannot fail to realize that there are certain important differences between the two. Correspondence is the more static term of the two, and appears to be neutral as far as direction is concerned. Mapping, on the other hand, normally implies direction. It is easy to see that talking about mappings and source and target domains, as well as using preposition from and onto, just goes to ­further reinforce the idea of a directed process. Finally, the notion of direction is certainly­ not ­discouraged by our saying that the primary function of metaphor is understanding one thing in terms of another. Consider the following quotations from Lakoff (1987):

224 Rita Brdar-Szabó and Mario Brdar

A metaphoric mapping involves a source domain and a target domain. The source domain is assumed to be structured by a propositional or image-schematic model. The mapping is typically partial; it maps the structure of the icm in the source domain onto a corresponding structure in the target domain. … This is an example of an inference pattern that is mapped from one domain to another. It is via such mappings that we apply knowledge about travel to love relationships.  (Lakoff, 1987: 208) When those fixed correspondences are activated, mappings can project source domain inference patterns onto target domain inference patterns.  (Lakoff, 1987: 245) Metaphoric models are mappings from a propositional or image-schematic model in one domain to a corresponding structure in another domain. The conduit metaphor for communication maps our knowledge about conveying objects in containers onto an understanding of communication as conveying ideas in words.  (Lakoff, 1987: 114)

It is small wonder then that metaphors and mappings have occasionally been misunderstood in computer sciences, information processing psychology and cognitive science and treated as if they were directed, algorithmic processes. Mappings are static correspondences in mathematics, but in computer sciences and information processing psychology, mathematical mappings are often represented by algorithmic processes taking place in real time (cf. Lakoff, 1993: 203f). We are therefore warned that: [m]ore technically, the metaphor can be understood as a mapping (in the mathematical sense) from a source domain (in this case, journeys) to a target domain (in this case, love). The mapping is tightly structured. There are ontological correspondences, according to which entities in the domain of love (e.g., the lovers, their common goals, their difficulties, the love relationship, etc.) correspond systematically to entities in the domain of a journey (the travelers, the vehicle, destinations, etc.).  (Lakoff, 1993: 206) Mappings should not be thought of as processes, or as algorithms that mechanically take source domain inputs and produce target domain outputs. Each mapping should be seen instead as a fixed pattern of ontological correspondences across domains that may, or may not, be applied to a source domain knowledge structure or a source domain lexical item.  (Lakoff, 1993: 210)

Now what about metonymy? Can we also model metonymy in terms of a directed process? Can we also present it in terms of mappings that are correspondences?



Metonymic chains and the nature of metonymy 225

It is interesting that, to the best of our knowledge, the question of how far the analogy between metaphorical and metonymic mappings can go, has hardly ever been explicitly asked. Well, metonymy has routinely been said to be a single mapping between a metonymic source and a metonymic target within a single domain (cf. Lakoff, 1987; Lakoff and Turner, 1989: 103; Taylor, 1989: 123f; Gibbs, 1994: 13). In other words, once again we have the idea of directed conceptual transfer. Unlike metaphor where mappings typically go from more concrete domains towards more abstract domains, metonymy has been shown to be able to proceed in either direction, without any concreteness/abstractness differential. Namely, there are some concepts that can serve either as metonymic source or as metonymic target, e.g., cause and effect can be metonymically related in two ways, cause for effect and ­effect for cause. Of course, these are two distinct metonymies: metonymic mappings are not thought to be able to proceed in either direction simultaneously. As for the other question, we realize that it is one thing to talk about correspondences or mappings if the relationship obtaining between the vehicle and the target is based on similarity, and it is another if we talk about contiguity. Furthermore, it is far easier to talk about mappings if more than one domain is involved. Typically, in the case of metaphor, the domains are relatively distant and distinct (otherwise they would be the same, and this would preclude the relationship of similarity). Correspondences or mappings are only to be expected if the main function of metaphor is to present one thing in terms of another, i.e., understanding. Without such correspondences or mappings we would not know what is understood in terms of what. Pressing the metaphor-metonymy analogy further, we realize that the nature of metonymic mapping has hardly ever been explicitly spelled out, unlike in the case of metaphors. Specifically, we do not get to know what is mapped onto what. If the analogy were complete, and if we are right in assuming that in the case of metaphor the mappings go from the source domain to the target domain, i.e., that the former structures the latter, then we might as well think that in the case of metonymy, a single mapping operation takes place, proceeding in our specific case, from capital as the matrix subdomain, which is the metonymic source, towards government as the subdomain, which is the metonymic target. There are also some examples in relevant literature that more or less explicitly describe metonymic mappings in this way. Note thus that Gibbs (1994: 13), when talking about mappings between elements within the same domain, mentions that a salient property of the metonymic source is mapped, but stops short of actually stating what it maps onto:

226 Rita Brdar-Szabó and Mario Brdar

Unlike metaphorical mappings, metonymies involve only one conceptual domain, in that the mapping or connection between elements occurs within the same domain. For instance in “Hollywood is putting out terrible movies”, the movie industry is referred to by the place where movies are made, which maps a salient characteristic of one domain (its location) as representing the entire location (the movie industry).

As another example of exploiting the analogy, consider Yu (2007: 30): The metonymy style of dance stands for culture, realized visually, is an important component of that complex metaphor. In the process of mapping, we can trace the following steps of metonymic mapping governed by the principle of contiguity: ballroom dance → Western culture → developed countries → modernization and globalization. However, if we omit and ignore the two intermediate steps, we have a cross-domain mapping that is metaphorical: ballroom dance → process of modernization and globalization.

It is obvious, however, that the analogy does not work in this way. It would be odd to claim that the essence of the metonymy in (5) is that an element of London as a source domain, say the fact that it is the capital of the United Kingdom, maps onto the domain of British government, etc. It is actually the conceptual content associated with the lexical item London that is affected by metonymy and not the other way round. It appears at this point, paradoxically, that, if there were some sort of mapping, it is more likely to proceed from the concept or domain of government towards London, etc., i.e., from the metonymic target towards the metonymic source. Alternatively, we could stipulate (as we are actually going to do in the rest of this chapter) that the domain associated with the metonymic vehicle undergoes conceptual restructuring, either reduction or extension, in the sense of Ruiz de Mendoza and Díez (2002). In our particular example, the domain appears to be reduced to fit the metonymic target. Although Lakoff and his collaborators have devoted much less time to the study of metonymy, as compared to the study of metaphor, one cannot but notice a certain amount of vacillation in the way metonymy is treated. Although it is often said to be mapping, this qualification is occasionally dropped in favor of some other ways of presenting it. It is thus also claimed to be a model (i.e., just a model) with a function from one element of the model to another. Notice that the following quotation from Lakoff (1987: 114) immediately follows the quotation on page 8 of this chapter, where metaphoric models are explicitly said to be mappings. Metonymic models are models of one or more of the above types, together with a function from one element of the model to another. Thus, in a model that represents a part-whole structure, there may be a function from a part to the whole that enables the part to stand for the whole.



Metonymic chains and the nature of metonymy 227

Similar uneasiness about the analogy between metonymic and metaphoric mappings can be found in other authors, too. Barcelona (2007b: 105) thus warns against a simplified view of metonymic mappings: “Mappings refer to the fact that the source domain is connected to the target domain by imposing a perspective on it, not by projecting its structure onto it, as in metaphor.” Barcelona (2002, 2003: 99, and 2005) also points out that the mapping in metonymy is unidirectional and asymmetrical, whereas the one in metaphor is unidirectional and symmetrical. Unfortunately, Lakoff (1993: 245) remarks that metaphoric “mappings are asymmetric and partial.” Croft (1993: 364) remarks that instead of mapping, which he finds appropriate as far as metaphor is concerned, “in the case of metonymy, it is particularly appropriate to choose a different term to describe the domain adjustment involved.” What he suggests is the notion of domain highlighting, which “appears to be a necessary though not sufficient condition for metonymy, which also involves shift of reference, at least in the most typical occurrences thereof ” (1993: 349). It is our contention that the evidence in favor of the claim that metonymy is better served if it is modeled in terms of inferences coupled with a domain reduction or extension, rather than in terms of mapping, comes from the study of the way that metonymy functions in discourse, viz. of its role in the online construction of meaning.

4.

Function(s) of metonymy in discourse

4.1

Metonymic patterning: Networks and chains

The phenomenon of simultaneously activating more than one topical concept, viz. a metonymic source, and one or more metonymic targets, is ubiquitous. It can also be illustrated on the metonymic use of geographical proper names. It is well-known that in addition to their primary, literal use to refer to locations, e.g., in utterances specifying where a state of affairs takes place, geographical names can be used metonymically in a number of ways. For example, names of capitals are frequently used to refer to various political institutions, most notably to the institution invested with the executive political power, viz. the government of the country in question: (6) a. At a recent Politburo-level meeting, according to versions reaching Washington, President Jiang Zemin counseled a low-key, cautious approach toward the new administration.

228 Rita Brdar-Szabó and Mario Brdar

b. If Beijing doesn’t get anything substantial from Bush at the summit in Shanghai, such as a reaffirmation of U.S. support for the one-China policy; it might harden its policy toward Washington and Taiwan. c. Dublin denies accusations that it allowed the three men back into the country in return for an IRA pledge last month to end its armed campaign against British rule in Northern Ireland. d. Poland’s foreign minister has denied reports that Warsaw wanted to renegotiate its 1990 border treaty with Germany in a dispute over World War II property claims.

Names of capitals can be found in metonymic uses that are also characteristic of other place names, i.e., to refer to branches of companies and museums, tribunals, stock exchanges, events, etc., as shown by the following set of examples: (7) a. Arusha depends mostly on witnesses for evidence, many of them illiterate farmers who could not record their impressions at the time. The Hague enjoys intelligence intercepts from western armies, satellite photographs and other high-tech methods of collecting more durable evidence. b. Kuala Lumpur rose by 6.1% on optimism about prospects for Malaysia. (8) a. The risk is that scattered American units would find themselves fighting multiple mini-Mogadishus. b. So what sort of NATO will emerge after Prague?

Metonymically used names of capitals, just like other names, can be used repeatedly, i.e., they can be used several times in a running text. Cf. the following example: (9) Japanese government spokesman Shinzo Abe called for calm Wednesday, but says Tokyo will not give in to South Korea’s demand that it call off the survey. Abe says Japan is within its legal rights to conduct the survey, and will do so without making a big deal of it. He says Tokyo expects that both countries will deal with the matter in a level-headed manner.

This phenomenon of metonymic patterning in discourse has been studied in some detail by Barcelona (2005, 2007b), who uses the term metonymic chains to refer to “direct or indirect series of conceptual metonymies guiding a series of pragmatic inferences (Barcelona, 2005: 328). From this wording, one might get the impression that Barcelona is interested primarily in purely tracking a series of metonymic expressions as they occur linearly in a running text. However, it is apparent in the article that what Barcelona has in mind is a more complex system of interaction involving both textual and conceptual dimensions when



Metonymic chains and the nature of metonymy 229

he talks about “two, often more, metonymies regularly occurring at the same or different analytical levels in the same utterance, even in the same sentence” (Barcelona, 2005: 316). This term has also been used in a different, more specialized sense in metonymy research from Reddy (1979), Fass (1991, 1997), Nerlich and Clarke (2001), Radden and Kövecses (1999: 36), Ruiz de Mendoza and Díez (2002), Panther and Thornburg (2003: 6), to Ruiz de Mendoza and Mairal (2007), Ruiz de Mendoza (2008) and Hilpert (2007). These authors are primarily concerned with metonymies involving multiple conceptual shifts, breaking down “complex conceptual mappings into simple, well-motivated mappings with a strong experiential basis” (Hilpert, 2007: 80). These are cases of metonymic operations stacked onto each other, producing double or even triple metonymies (Ruiz de Mendoza and Mairal, 2007). Such metonymic chains were referred to as metonymic tiers in Brdar and Brdar-Szabó (2007), a neutral term allowing for metonymy’s interaction with metaphoric tiers. In our opinion the two dimensions are essential and inseparable. Both the textual (horizontal or linear) dimension and the conceptual (vertical) dimension should be integrated into a comprehensive study of how metonymy works in discourse, i.e., in the study of metonymic networks, similar to what has been shown for metaphors in a series of studies by Goatly (1997), Koller (2003), ­Cameron and Stelma (2004), Cameron and Low (2004), and Semino (2008). Note that these authors use the term metaphorical chain to refer to the phenomenon of “the occurrence of several related metaphorical expression throughout a text” (Semino, 2008: 226). Since we are obviously dealing in discourse with linguistically manifest metonymies, we take the threefold distinction between linguistic vehicle, metonymic source and metonymic target, as our starting point, as in Panther (2005: 358). This makes it possible to distinguish between linguistic or textual metonymic chains, on the one hand, and conceptual metonymic chains, on the other. The former are a series of linguistic metonymic vehicles. We may consider the totality of all metonymic vehicles within a text or discourse, regardless of whether they share metonymic sources and targets, as a metonymic textual macro-chain. It can contain a number of metonymic textual micro-chains, consisting of metonymic vehicles sharing the same metonymic source. As can be seen in Examples (10a–f), and Figures 3a–e, textual metonymic chains may be fairly simple, or quite complex in that they may include overlapping micro-chains, or micro-chains that are interrupted by other micro-chains, etc. In fact, complex chains are far more frequent in actual usage than simple ones. Metonymic micro-chains regularly interlace rather than neatly follow each other.

230 Rita Brdar-Szabó and Mario Brdar

Let us now look at some examples. For the sake of exposition, we concentrate here only on occurrences of low-level metonymy capital for government metonymies, but also take the related metonymy country for government into consideration (but not other possible metonymies). In (10a), we note two occurrences of Pyongyang (followed by Washington) in a single sentence (the network is schematically presented in Figure 3a). In (10b), we have four instances of Pyongyang, with one occurrence of North (short for North Korea) in the middle (see Figure 3b): (10) a. The Bush administration tried squeezing Pyongyang in this way in the former president’s first term, only for Pyongyang’s first successful nuclear test back in 2006 to force Washington back to the negotiating table.  (The Independent, London, 26 May 2009, p. 26)

V1′

V1′′

Figure 3a.  A simple textual metonymic chain with two metonymies

(10) b. But that has certainly not stopped Pyongyang from claiming it now has a satellite hurtling around the Earth broadcasting revolutionary songs in praise of “Dear Leader” Kim Jong and his even more illustrious father, “Great Leader” Kim Il Sung. Nor will it deter Pyongyang from dedicating the launch to the Dear Leader on the opening of the country’s newlyelected Parliament on Thursday. After all, the launch has confirmed that the North has made tremendous advances in missile technology. Sunday’s Taepodong-2 fired by Pyongyang travelled twice as far as the rocket in 1998. So if there is no conclusive evidence to show otherwise, Pyongyang will have to be taken at its word – that it had indeed tried to shoot a satellite into space.  (The Straits Times, Singapore, 7 April 2009)

V1′

V1′′

V2

V1′′′

V1′′′′

Figure 3b.  A simple textual metonymic chain with four metonymies, interrupted by another metonymy

The two occurrences of Moscow and Washington, respectively, follow each other neatly and appear to be two separate micro-chains, but they are actually part of a larger chain with Moscow and Washington interlacing:



Metonymic chains and the nature of metonymy 231

(10) c. Although Moscow was going through the theatrics of calling off the Shultz-Shevardnadze meeting, there were a lot of business-as-usual signals coming out of Moscow for Washington. Vladimir Horowitz has just made a spectacular appearance in Moscow, receiving widespread Soviet publicity. Many prominent Soviets attended. Had they wished to signal real Soviet disaffection with Washington, the Soviets could have played down the recital by the Soviet-born American citizen.  (Christian Science Monitor, 23 April 1986, p.13)

V1′

V1′′

V2′

V1′′

Figure 3c.  A textual metonymic chain with two non-overlapping micro-chains

Metonymic micro-chains can also span over another embedded chain or interlace, as seen in the following two examples. In the second example, we insert tags into the original texts so as to facilitate the tracking of metonymic vehicles. (10) d. Washington bows to Moscow’s demands to soften its tone on the gulf war; in return, Moscow heeds Washington’s warning to ease off on the use of force against the independence-minded Baltic republics.  (USA Today, January 30, 1991, p. 3A)

V1′

V2′

V2′′

V1′′

Figure 3d.  A textual metonymic chain with one micro-chain interrupting the other

(10) e. If, as some western analysts hold, Moscow [V1’] is using the summit as a bargaining chip, it is a shrewd assessment of Washington’s [V2’] deep interest in holding the meeting on American soil. According to one view, the summit is politically more valuable to Washington [V2”] than the INF treaty itself. Therefore, in withholding agreement on the place and time of a third Reagan-Gorbachev meeting, Moscow [V1”] is deliberately playing on Washington’s [V2”’] weak point. According to this view, Moscow [V1”’] this week simply played out its hand after raising expectations during the Shevardnadze visit to Washington [V2””] in September. But if it is a bluff, it is not clear how Washington

232 Rita Brdar-Szabó and Mario Brdar

[V2””’] can call it – particularly since Shultz today stressed again that the Reagan administration’s commitment to SDI is not negotiable. Some analysts here tonight argued that Moscow [V1””] this week was not pursuing a grand strategy but executing an about-face, shifting suddenly from optimistic predictions of a Gorbachev tour of the United States to artful dodging on who would sign an INF treaty, when and where.  (The Washington Post, 24 October 1987, A30)

V1′

V2′

V2′′

V1′′

V2′′′

V1′′′

V2′′′′

V1′′′′

Figure 3e.  A textual metonymic chain with interlacing micro-chains

Such chains often involve more than two metonymic vehicles exhibiting various arrangements, as shown by the following example: (10) f. Kyrgyzstan’s close relations with the United States have long unsettled Russia and China, which both have military interests in the region. In 2005, the country appeared to move further into Washington’s orbit after a popular uprising, supported in part by the United States, toppled the corrupt and increasingly authoritarian government of Askar A. Akayev, sending the president fleeing across the border. The bloodless coup was part of a wave of popular revolts, known as colored revolutions, that remain a source of anger and suspicion among Russian officials, who consider them Washington-hatched schemes meant to undermine Russia’s influence in the region. Similar uprisings in Georgia and Ukraine ushered in governments that quickly sought to shut out Moscow’s influence in favor of stronger ties with the West. Kyrgyzstan, however, has often sought to strike a balance among Washington, Moscow and Beijing.  (The New York Times, 4 February 2009)

Conceptual metonymic chains are series of metonymic sources unified by common metonymic targets. In Figure 4, the concepts [S1] and [S2] which serve as metonymic sources are associated with [V1] and [V3], but have the same metonymic target, [T1]. (11) Beijing [V1’] and Moscow [V2’] also have concluded numerous bilateral agreements on trade and investment, military affairs, nuclear weapons, energy cooperation, science and technology, cultural exchanges and international policy.



Metonymic chains and the nature of metonymy 233

Moscow [V2”] and Beijing [V1”] regularly vote in tandem on the United Nations Security Council. Their solidarity has reflected common world views. As a senior Russian Foreign Ministry official put it: “We have either shared or identical views of all international issues.” But the fact that China’s [V3’] global equities and responsibilities are growing while Russia’s [V4’] are minimal and declining may impinge on their solidarity. Prior to their meeting, Presidents Hu and Medvedev convened two other multilateral forums in the Ural city of Yekaterinburg, forums that China and Russia have introduced to counterbalance the United States in regional and world affairs.  (The International Herald Tribune, 16 June 2009, p. 6) T1

S1

V1′

V2′

T2

S2

V2′′

T3

S3

V1′′

V3′

S4

V4′

S5

V3′′

V4′′

V5

Figure 4.  Conceptual metonymic chains

Metonymic networks can often exhibit considerable complexities, as shown by the example that follows, in which we apparently keep the same metonymic target while using more than one metonymic source/vehicle. In other words, a number of metonymic vehicles and sources converge on a single metonymic target. (12) “Moscow is playing on the contradictions between Europe and the US, aiming to show that Sarkozy’s pragmatic and respectful approach, rather than Washington’s hard-line rhetoric, is the way to achieve concrete results with Russia,” says Sergei Strokan, a foreign-policy expert with the liberal Moscow daily Kommersant. “You can’t help but notice that the harsher the Kremlin’s tone toward the US becomes, the gentler and more subtle becomes its approach to Europe.”

234 Rita Brdar-Szabó and Mario Brdar

If Moscow fulfills the agreement, Sarkozy said, the EU may lift its main punitive measure taken against Russia during the war, which was to suspend ongoing talks on a new strategic partnership deal. “There is no reason why meetings between Russia and Europe … cannot be resumed in October,” he said. “We want partnership and peace, and hardly anyone wants a confrontation between Europe and Russia.” Russian analysts insist that the commitment of EU observers to hold the security zones between Georgia and the rebel statelets of South Ossetia and Abkhazia is exactly what the Kremlin wanted all along. “It’s no problem for Russia to withdraw its military forces from Georgia, because that decision was taken weeks ago,” says Alexei Mukhin, director of the independent Center for Political Information in Moscow. “Now the responsibility for what happens there, and Saakashvili’s behavior, will rest on the Europeans and not on us.”

Here we find two metonymic instances of Moscow, accompanied by five instances of Russia, and two instances of the Kremlin, all sharing the same metonymic target, i.e., the Russian government. Note that after the first mention of Moscow we get to Russia, thus moving out of the first source domain (of course assuming that russia as a domain is superordinate to the Moscow domain). This is followed by moving to the kremlin, to a domain presumably even one level below moscow. After that we return back to Moscow and further to russia, and then finally back to the kremlin, all the time apparently keeping more or less the same ­metonymic target. Double and triple metonymies are in this descriptive framework seen as special cases of conceptual metonymic chains. They are also unified by common metonymic targets because the metonymic target of one tier serves as the metonymic source for the next higher metonymic tier. In Example (10b) above, in addition to Pyongyang we also analyzed the expression the North as a metonymy standing for North Korea. The four cardinal directions or points have conventionally been used in metonymic reference to certain geo-political units. We assume that this part for whole metonymy is based on a map icm, so that the extreme point (i.e., the cardinal point) of the cardinal direction stands for the upper section of the direction line, and when this is translated onto a map, a geographic point stands for the whole region or area around that point, either globally or locally, which means that the same model can be applied to geographical areas of variable size. Witness the use of pairs such as the South and the North, and the East and the West, in reference to some global political, economic and cultural oppositions. Furthermore, notice the use of the South and North, with all their metonymic implications in the US context. In addition to the use of the West and the East to refer to the two



Metonymic chains and the nature of metonymy 235

political and military blocks (cf. Examples (14) below), there is also an opposition in the history of Germany (Osten ‘East’ vs. Westen ‘West’), in part comparable to that in the history of the USA. Now that we have seen that cardinal points or directions can be metonymically used to refer to countries, we realize that the names of capitals (as in (10b) above), or the names of countries (13), can appear in the same conceptual metonymic chain with a cardinal point, when they share the same metonymic target. In these specific examples the shared metonymic target is the respective national government. (13) North Korea accused the South of tampering with a border marker on Wednesday and warned of a possible clash along the heavily fortified frontier. The warning comes as South Korea’s relations with the North are at one of their lowest points in years.  (The New York Times, 29 April 2009, p. 13)

However, the case of the metonymic use of the names of capitals and countries is simpler than the metonymic use of cardinal points because the latter metonymy exhibits two metonymic tiers (i.e., qualifies as a double metonymy in the framework worked out by Ruiz de Mendoza and his collaborators: Ruiz de Mendoza and Díez, 2002; Ruiz de Mendoza and Mairal, 2007; Ruiz de Mendoza, 2008). The metonymic chains found in (12) can be presented schematically as in Figure 5. T1

S1

T4

T2 = S5

S3

T3 = S6

S2

V1

V2

S4

V3

V4

Figure 5.  A network with double(-tiered) metonymic chains

236 Rita Brdar-Szabó and Mario Brdar

Similarly, in order to properly analyze a text like (14), at least from a historical point of view, we would again need to assume such double(-tiered) metonymies in the case of Russia. In texts prior to 1991, i.e., before the breakdown of the Soviet Union, instances of Russia, metonymically intended as referring to the Soviet government, must be assumed to have a metonymic target (the Soviet Union) that is at the same the source for the metonymy that has the Soviet government as its target. Russia as the most dominant part of the Soviet Union has been used metonymically to stand for the whole of the Union, just like England is often used to refer to Great Britain or the UK (cf. Radden and Kövecses, 1999). (14) In fact, Moscow’s personal attack on Mrs Thatcher goes far beyond what we have just mentioned. As the British newspapers have pointed out, Mrs Thatcher is an “Iron Lady” on Russia’s list of enemies. The “Iron Lady” was a nickname Moscow bestowed on her before she became Prime Minister. With Europe facing the threat of war, Mrs Thatcher gave her warning explicitly: “The key lies in whether the Soviet Union really wants detente or uses detente as a false sense of security to deceive the West so that it will relax its vigilance and easily become the victim of Soviet expansion or blackmail…”  (BBC Summary of World Broadcasts, 14 November 1979)

The descriptive framework sketched above is open in at least two senses. Firstly, it also makes it possible to take the occurrence of metonymic anaphors and explicit mention of metonymic targets by means of lexical items normally associated with metonymic targets into consideration. Secondly, it is open in the sense that it can be also used to study the interaction between metonymy and metaphor. In other words, it is compatible with the research on metaphorical chains, as reviewed in Semino (2008).

4.2 Metonymy at work in discourse Metonymy can probably be described most succinctly as an efficient mental shortcut, since metonymies have the advantage of communicating additional information: one concept stands for another, but both concepts are actually activated, to at least some degree. Informally speaking, metonymy is an efficient means of saying two things for the price of one, i.e., two concepts are activated while only one is explicitly mentioned (cf. Radden and Kövecses, 1999: 19). Metonymy is therefore often used for understanding in a slightly different sense, i.e., not so much for highlighting, as in a more radical sense of providing a conceptual handle or window on a concept that would be otherwise quite difficult to conceptualize and lexicalize. A fairly drastic example of this comes from



Metonymic chains and the nature of metonymy 237

the special language of the military, where according to Oxford English Dictionary one of the (now obsolete) meanings of horse is given under 3 in (15) below: (15) 3. Mil. A horse and his rider; hence a cavalry soldier.  a. In sing., with pl. horses. Obs. rare. 1548 HALL Chron., Edw. IV 231 The Duke..came in no small hast.. onely accompaignied with sixtene horses. Ibid., Hen. VIII 32 The kyng contynually sent foorth his light horses to seke the country. b. Collective pl. horse: Horse soldiers, cavalry. light horse: see quot. 1853, and LIGHT-HORSE. 1548 HALL Chron., Hen. IV 13 King Henry..with a fewe horse in the night, came to the Tower of London. 1549 Compl. Scot. xi. 89 He furnest..tua hundretht lycht horse. 1597 SHAKES. 2 Hen. IV, II. i. 186 Fifteene hundred Foot, fiue hundred Horse Are march’d vp. 1698 Lond. Gaz. No. 3445/1 First marched an Alai Beg with about 50 Horse. 1777 ROBERTSON Hist. Amer. (1783) I. 157 The body..consisted only of two hundred foot, twenty horse, and twenty..Indians. 1853 STOCQUELER Milit. Encycl., Light horse, all mounted soldiers that are lightly armed and accoutred, for active and desultory service. Thus light dragoons, fencible cavalry, mounted yeomanry, etc. are, strictly speaking, light horse.

However, there are many less exotic examples where the presence of a metonymy is hardly ever noticed. In the following set of examples we have descriptions of a person placing a candle on a table: (16) a. Magdalen lit a candle. “What notice must I give you,” she asked, as she put the candle on the table, “before I leave?” b. “Can I put the candle on the table?” “Will you light it up for me?” c. She put the candle on the table, and taking her head between her hands sat down to think.

According to standard dictionary definitions, a candle is a source of artificial light made of wax, tallow or similar solid fat formed in a cylindrical shape around a wick of cotton or flax. Dictionary definitions do not mention not any candlestick or candle holder as part of a candle, i.e., they do not record any metonymic uses of the part for whole type, where candle would refer to the whole of the functional unit consisting of a candle and its holder. Note that the above examples could be actually interpreted as being about a candle without any holder, but it is more likely that both a candle and a holder are involved. In the latter case, we then have a part for whole metonymy that goes virtually unnoticed because it is subconscious, but is very useful indeed. The truth is that there is simply no ready-made label for the functional unit consisting of two parts, and it would be

238 Rita Brdar-Szabó and Mario Brdar

too cumbersome to specify the parts, say by means of a coordinating construction, on every occasion when we want to designate the unit. In both (15) and (16), saying that the metonymic mapping proceeds from the metonymic source, horse and candle, respectively, and that it helps us construct the metonymically targeted meaning would actually belie the truth. We can at best say that we pretend, for the reasons mentioned above, that the metonymic targets can be properly referred to by the lexical items serving as metonymic sources. The actually targeted metonymic meanings can only be constructed if we invoke information from the context and cotext. The same sort of uncertainty as to what is the metonymic target, may be observed with place names too, as shown by the following example with Berlin: (17) Chicagoans marveled at an order for May rye telephoned to the Chicago Board of Trade by Henry Hentz & Co. of Manhattan. The amount was not so stupendous – 30,000 bushels. The bid price was not so unusual – $1.19.* But when Chicago said, “All right, we’ll fill your order,” Manhattan replied, “Please execute it at once and confirm. It’s an order from Berlin, and Berlin is holding the wire.” Chicago executed and confirmed. Manhattan told Berlin. Berlin hung up, pleased by efficiency. The entire transaction had occupied only three minutes, cost $53.25. Henry Hentz & Co. reported that this transaction was but typical of other German orders similarly placed, executed and confirmed by telephone in recent weeks. Observers marveled as much at the German enterprise as at the rapid communication. During last fortnight’s furor on the New York Stock Exchange, a busy floor man was asked by his office for a quotation on General Electric. “What the –!” he roared. “I can’t be bothered for quotations at a time like this!” “But Berlin wants to know. They’re holding the wire.” Abashed, the floorman dove into the nearest drift of ticker-tape. “138½ and still going up!” he reported. Berlin bought 3,000 shares.

Another important discourse-pragmatic function of metonymy is to enhance cohesion and coherence of the utterance. It is something that is already at the very heart of metonymy as a conceptual operation whereby one content stands for another while both are activated. In other words, metonymy is an efficient mental shortcut making it possible to say two things for the price of one, i.e., two concepts are activated while only one is explicitly mentioned (cf. Radden and ­Kövecses, 1999: 19). This subsequently enhances the cohesion of the utterance because two or more topical concepts may be referred to by means of a single label. There is, however, another sense in which metonymy can enhance the text cohesion. A single metonymic source can be repeatedly used to refer to more than one metonymic target within a single text. In other words, we can shift between



Metonymic chains and the nature of metonymy 239

subdomains within a single domain matrix, picking different target meanings at different points in a text, while using a single lexeme as a metonymic source. There is consequently, at least nominally, less shifting or switching between topics, as far as the lexical realization of the metonymic source is concerned.

4.3 Keeping the same metonymic source with more than one metonymic target in play Place names can also be used to switch among more than one metonymic target. The following excerpt – part of a newspaper article – contains two occurrences of Beijing. The first appears to be an unambiguous instance of the capital for government metonymy. Notice that later in the text there is an explicit mention of the metonymic target – the Chinese government, which is in the same sentence juxtaposed to the US government. The second occurrence of Beijing is, however, somewhat less clear. It is possible to merely interpret it again in the same way, but it is far more likely that it is used to refer indirectly to the upcoming Olympics, to be held in Beijing in 2008, which is thus an instance of the place for event metonymy: (18) After months of detention amid widespread condemnation from Capitol Hill and US academics, Li’s swift court proceedings and promised release just hours after the Olympics vote leave an appearance of tit-for-tat justice, raising questions about whether Li and other detainees with US ties are being used as bargaining chips by Beijing, observers said. Li was “a hostage in the Olympics bid,” said Frank Lu, director of the Hong Kong-based Information Center for Human Rights and Democratic Movement in China, which tracks arrests and harassment of dissidents and activists. “We know that just two weeks ago the Chinese government told the US government that if the US voted against Beijing, they wouldn’t release him.”  (Boston Globe, 15 July 2001, A1)

Such shifting of metonymic targets may occasionally be more difficult to discover, as illustrated in the following example: (19) Serbia’s privatisation authority has confirmed that Morgan Stanley and a consortium led by JP Morgan will compete to become the financial advisor for the initial public offering (IPO) of fixed line telecoms monopoly Telekom Srbija. “The final decision will be made next week,” Nebojsa Ciric, the president of the tender commission, said on Friday. In a financial bid that was opened on Friday, Morgan Stanley requested a 1.19% fee after the whole transaction is completed, in addition to a EUR 250,000 fee. JP Morgan,

240 Rita Brdar-Szabó and Mario Brdar

meanwhile, is reported to have demanded a 1.35% fee. The tender for the financial advisor to Telekom Srbija was announced in January this year and Belgrade hopes to launch the IPO on the local stock market by mid-2009.

A default interpretation of Belgrade above would be that it is the Serbian government. However, notice that the paragraphs opens with a reference to Serbia’s privatization authority, and in fact the whole of the rest of the text is about its activities, as presented in direct speech and free indirect speech by the president of the tender commission of the privatization authority. This even continues into the first half of the very last sentence, implying that Belgrade may actually refer to Serbia’s privatization authority. However, basing our reasoning on general knowledge of world affairs, we know that it is the Serbian government and its budget that will benefit from the deal, and that the government therefore also has reason to hope. In other words, the metonymic target may be slightly indeterminate here. Both of these examples demonstrate a serious problem stemming from the mapping assumption. A metaphor-like mapping would result in a fixed targeted meaning that cannot be rewritten so easily. Metonymy, however, seems to be a dynamic meaning construction process. Note that multiple mappings or correspondences that are typical of metaphors effectively narrow down and cement their figurative meaning. It is as if in the case of metaphor we are working with several parameters along several dimensions, or with coordinates, thus fixing the position of the target unanimously, whereas in the case of metonymy, working with a simple parameter along one dimension, we cannot determine its exact position. With metonymy there is room for corrections, moving forward or backward, i.e., hopping from one metonymic tier to another metonymic tier in a conceptual chain.

4.4 Locatives as metonymies in a conceptual metonymic chain Some preliminary studies on the availability of metonymically used names of capitals show (Brdar-Szabó and Brdar, 2003) that this particular type of metonymy is ubiquitous in English and German, while less so in Hungarian and Croatian. Part of the difference between the two pairs of languages can be attributed to the fact that English and German metonymic place names that function as subjects often find their counterparts in adpositional phrases in Hungarian, as in (15), and in prepositional phrases in Croatian, as shown in (16). These are apparently used as adverbials of place: (20) Moszkvában most úgy látják,… Moscow-in now thus consider-3pl “Moscow now thinks…”



Metonymic chains and the nature of metonymy 241

(21) Odluka je nezakonita, smatraju u Sarajevu,… decision is illegal consider-3pl in Sarajevo-loc “It is considered in Sarajevo that the decision is illegal”

Comparable constructions, while not impossible, are relatively infrequent in English and German: (22) South Korea’s President Roh Moo-hyun’s first meeting with George W. Bush, his US counterpart, in Washington on Wednesday was hailed in Seoul yesterday as a happy ending to a period of turmoil in relations between the military allies. (23)

Nach mehr als zwanzig Jahren Krieg sei es äußerst after more than twenty years war is-subj it extremelly schwierig, einen Neuanfang zu finden, heißt es in difficult a new start to find said-is-3sg it in Islamabad. Islamabad “It said in Islamabad that after more than twenty years of war it is extremely difficult to make a new start.”

The replacement pattern of subjects that we observed above is one of the five strategies available to solve the problem of the preservation or maintenance of the topic-continuity in the flowing discourse in the case of metonymic subjects. Let us point out that the same pattern of replacement of metonymically used names of capitals is found in some other pro-drop languages, e.g., Slavic and Romance languages that have elaborate agreement systems as well as productive impersonal constructions, e.g., in Polish, Russian, Slovenian, Spanish, and Italian. Concerning the status of the adverbial replacements observed above, it has been argued in Brdar-Szabó and Brdar (2003) that such prepositional phrases, so far overlooked in the mostly English-biased research on metonymy, also count as referential metonymies. The locatives in question appear to be straightforward adjunct/adverbial structures with literal local meaning only before they are subjected to closer scrutiny. That they do not have the straightforward literal meaning is shown by the oddity of the sentences in which they are combined with other adverbials that have a genuine literal locative meaning (and even with some PPs with temporal meaning): (24)

?U

Sarajevu na trgovima/ u sjevernom dijelu grada in Sarajevo in squares in northern part city-gen smatraju ovu izjavu nezgodnom. consider-3pl this statement awkward “In Sarajevo in squares and in the northern part of the city this statement is considered awkward.”

242 Rita Brdar-Szabó and Mario Brdar

Rather than being literal locatives, we claim that these PPs are metonymies. There is an initial basic projection from a mental space that is opened by the particular discourse type and topic, i.e., a mental space is set up on the basis of our realization that the utterance in question is in terms of its text type a newspaper article dealing with politics, specifically with international relations, or with sports, etc. This mental space also contains elements of encyclopedic knowledge that are projected into it. These projections trigger the first layer of metonymic meaning. Sarajevo, Washington, and other such names of capitals in our examples, are not used to refer to the locality as a whole, i.e., not just everywhere or anywhere in Sarajevo is meant here. What is intended is not the whole domain but just a part or parts of it; specifically it is just the sphere of political life, more precisely its foreign affairs aspects. However, it appears that the metonymic target is not necessarily a well-circumscribed subdomain, in fact it is quite likely that it is a medley of more or less close domains, e.g., politicians-cum-journalists covering politics, perhaps even including sectors of the general public interested in these affairs. If the context is different, i.e., if the expression in question is found in a different type of newspapers (e.g., a paper focusing on business vs. a sports daily), or if it is found in a different type of article, the same place name can be used to refer to different things. E.g., it may be used to refer to some other aspects of political life, or it may be used to refer to the press/media that are associated with this locality and cover its political, sporting scene, and so on. We claim that this metonymic projection of the type part for whole is formed in Croatian and Hungarian in both straightforward examples of metonymies and locative expressions alike. The difference between them concerns the target domain. In the former case, we reduce the metonymic source, focusing on a more-or less well-defined part of it, in the latter, the focus is not so sharp. The context and the contents of the article in the second step then determine the entity that the capital name refers to, i.e., whether we actually mean the whole government, or focus our attention onto just a single ministry, or some other institution, legal, economic, or otherwise. Whenever a well-defined subdomain emerges, we assume that a metaphorical mapping can apply and fix the specific low-level metonymy. If a capital name stands for an institution that is a collective body, such as the government, it becomes personalized by the conceptual metaphor organizations are humans, which confers a certain amount of agency properties, such as control and responsibility, onto the capital name. If a less sharply delineated part of the matrix domain emerges on the basis of the information projected from the context and the contents of the article, the metaphorical mappings of the above type cannot apply. A capital name used in a weakly metonymic sense in a relatively poor context lends itself to a whole range of interpretations, like any other place name. It could refer to a salient event tak-



Metonymic chains and the nature of metonymy 243

ing place in the location specified, e.g., Paris or a prepositional phrase with this name, such as after Paris or in Paris (used in 2003 or around that time in the sports context), could be used to refer to the 2003 World Athletics Championships. In a different context, Paris might be used to refer to the domain of fashion. In that context, a sentence such as Paris was really appalling will hardly be understood as referring to designers only. It will also include the reference to the fashion industry in general, including shows, clothes, audience, tailors etc. But it may also be used to refer to just clothes. This seems to indicate that metaphorical personalization does not take place at this stage. Secondly, the organizationsare-humans metaphor can hardly apply to just any assembly of entities, even if they involve people. What seems to be necessary in our opinion for the metaphor to apply is that the entity in question should really emerge as a clearly defined one, i.e., as a genuine organization, with internal structure, and with more or less clear boundaries. In other words, it should be a well-defined domain, not a mere aggregate of two or more adjoined domains within a subdomain, each of which could have been a metonymic target on its own. This means that the locative expressions in question may function as a means of conveniently stopping or retarding the source reduction, so that we may end up at a tier with a fairly vaguely delineated subdomain, if they are used somewhere in the middle of a metonymic chain. Brdar-Szabó (2002) has found that native speakers of Hungarian and German try to assign the default metonymic interpretation to place names (i.e., to metonymic NPs) if possible, while resisting other possible interpretations (although they are prepared to revise their initial assumptions at a later date), whereas English speakers are more likely to gradually determine the ultimate reference of the topic on-line in a step-by-step fashion, i.e., by picking it up from the larger context or from one of the consecutively activated knowledge domains or icms, effectively allowing more switching between topical concepts as the discourse unfolds. This is why the range of possible referents of metonymic NP expressions based on place names seems to be broader in English than in Hungarian or Croatian. There appears to be a division of labor between bare NP metonymies and PP metonymies. The latter are a naturally suited answer to the functional pressure of providing relatively vague ways of referring to a subdomain within a domain matrix without specifying one among several of its potential metonymic targets. This is once again a means of keeping several topical concepts in play. With respect to our central question in this chapter – Is metonymy a mapping? – we now realize that we have another problem with the mapping assumption. If we assume that all that is happening is a single mapping proceeding from the metonymic source it is hardly possible to arrive at the range of metonymically targeted meanings that we have uncovered. What we need is an account of how alternative meanings become available at various points in the metonymic chain.

244 Rita Brdar-Szabó and Mario Brdar

5.

By way of conclusion: Is metonymy just a mapping after all?

Now that we have seen that metonymic target domains can range from wellstructured and -circumscribed ones to those that are conceptually more difficult to handle, as shown by the last examples, we should consider the consequences of this with respect to the second question in the introductory section: Is metonymy a mapping? It appears, paradoxically, that, if there were some sort of mapping, we get the impression that it would be more likely to proceed from the concept or domain of government, or some other domain activated by the context, as the case may be, towards the domain with the capital, i.e., from the metonymic target towards the metonymic source. After all, it is the conceptual contents of the place name that are subject to meaning reconstruction. This is precisely what makes a delayed decision about the metonymic target possible once enough information has been amassed from context, and also allows for alternative targets (as in the case of Beijing in (18)), as discussed in Section 4.3 above. In Brdar and Brdar-Szabó (2007), analyzing examples such as: (25) a. Steven has a bag of tricks, a good passer, can operate in confined areas and is the Zidane of Villa whose left foot is nearly as good as his right. b. Sarkozy is described by MEDEF, French CBI, as “the Zidane of Finance.”

we demonstrate how the metonymic target in the course of the source domain extension may come close to almost totally rewriting the structure of the metonymic source, obliterating, or defocusing most other elements of knowledge. The proper name Zidane as a metonymic vehicle, is normally used to cover all our knowledge about Zidane. However, in examples like (25) the name actually comes to mean something much more precise, i.e., it is reduced to ‘Zidane the footballer’ as the metonymic target, almost totally obliterating other elements of knowledge about him, i.e., defocusing them. The highlighted portions of the domain become so prominent as to overwrite the whole structure of the original knowledge base. A similar analysis of paragons is offered by Ruiz de Mendoza (this volume). We may also refer to Barcelona’s (2004: 364ff) analysis of Shakespeare as a paragon: Box (a) represents the mapping of Shakespeare as endowed with his characteristic properties onto the rest of our knowledge network about him. This mapping results in a stereotypical model of Shakespeare as primarily, in fact exclusively, a writer with immense literary talent. Box (b) represents the mapping of this stereotypical view of Shakespeare onto the class of writers with immense literary talent. Both are metonymic mappings and are symbolized by the arrows.



Metonymic chains and the nature of metonymy 245

An apparent problem, however, is that some metonymic targets/sources, as we have seen above in the introductory part of Section 4, are obviously quite vague, and therefore not very likely to have anything close to the structuring role that we attributed to the metonymic target/source in the case of capital for government metonymy above. The problem disappears if we recognize two points. First, in some, we should say limiting, cases the information associated with the target domain (or global inferences about it) can do no more than just unstructure the conceptual contents associated by default with the metonymic source in the sense of preparing ground for some later decision on the basis of context. Secondly, we would like to submit that the metonymy is not a simple case of unidirectional traffic. According to Panther (2005: 358), the target meaning is more prominent than the source meaning but the source meaning is not obliterated by the metonymic target meaning. The latter is still, to some degree, conceptually salient or activated. Ruiz de Mendoza (1999, 2000) makes a similar point. Panther also claims that the target meaning is an elaboration of the source meaning, with the source meaning being one conceptual component of the target meaning. From all we have seen so far, we might as well assume that in metonymy we witness not so much mappings, as two-way projections leading to mutual accommodation of source and target (so that we might get tempted to consider metonymy to be an instance of conceptual integration or compression, just like metaphor, as suggested by Radden, 2005). We do not consider that not much hinges on the notion of mapping in the case of metonymy. We have shown in Section 3 above that drawing a parallel with metaphor does not seem to solve any problems. On the contrary, it may actually compound them. We therefore suggest that the notion may be dropped altogether, and that we should instead treat metonymy as a discourse-driven inference or pragmatic function (Fauconnier, 1997), arising in the course of domain expansion or reduction (in the sense of Ruiz de Mendoza, 1999, 2000). In any case, there is no doubt that we witness two-way traffic: the initial conceptual substrate is no doubt designated by the source concept, but it is plastic enough to allow considerable customizing. The inferences that steer the customizing are crucially guided by the information based on text (i.e., cotext) and context (circumstances). The ultimate outcome of this domain elaboration (reduction or expansion) is that domains are tailored to an optimal measure with regard to their function, as discussed in Section 4. On the one hand, metonymy provides a conceptual handle or window on a concept that would be otherwise quite difficult to conceptualize and lexicalize. On the other hand, the fact that they often appear in networks, i.e., chains along the two dimensions discussed in Section 4.1, makes them an extremely useful device in the organization (i.e., construction and management) of discourse. They can make texts more cohesive and coherent, while allowing

246 Rita Brdar-Szabó and Mario Brdar

us plenty of ­conceptual maneuvering room. In these chains, different tiers may function as stepping stones: we do not have to “go down” to the metonymic source concept associated with the metonymic vehicle and effectively start the inferencing process anew (the way that mapping assumption would imply). We can just continue from the last activated tier, hopping up or down from it in the course of the meaning construction as domains are expanded or reduced, which we consider to contribute considerably to the ease of the information flow and to the speed of processing of discourse containing fairly complex metonymic networks.

References Barcelona, Antonio. 2000a. Introduction: The cognitive theory of metaphor and metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (1–28). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2000b. On the plausibility of claiming a metonymic motivation for conceptual metaphor. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (31–58). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2002. On the ubiquity and multiple-level operation of metonymy. In Lewandowska-Tomaszczyk, Barbara and Kamila Turewicz (eds.), Cognitive Linguistics Today (207–224). Frankfurt am Main: Peter Lang. Barcelona, Antonio. 2003. The case for a metonymic basis of pragmatic inferencing: Evidence from jokes and funny anecdotes. In Panther, Klaus-Uwe and Linda L. Thornburg (eds.), Metonymy and Pragmatic Inferencing (81–102). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2004. Metonymy behind grammar: The motivation of the seemingly “irregular” grammatical behavior of English paragon names. In Radden, Günter and KlausUwe Panther (eds.), Studies in Linguistic Motivation (321–355). Berlin & New York: ­Mouton de Gruyter. Barcelona, Antonio. 2005. The multilevel operation of metonymy in grammar and discourse, with particular attention to metonymic chains. In Ruiz de Mendoza Ibáñez, Francisco José and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (313–352). Berlin & New York: Mouton de Gruyter. Barcelona, Antonio. 2007a. The role of metonymy in meaning construction at discourse level: A case study. In Radden, Günter, Klaus-Michael Köpcke, Thomas Berg, and Peter ­Siemund (eds.), Aspects of Meaning Construction (51–75). Amsterdam & Philadelphia: John Benjamins. Barcelona, Antonio. 2007b. The multi-level role of metonymy in grammar and discourse: A case study. In Kosecki, Krzysztof (ed.), Perspectives on Metonymy: Proceedings of the International Conference “Perspectives on Metonymy,” held in Łódź, Poland, May 6–7, 2005 (103–131). Frankfurt am Main: Peter Lang. Brdar, Mario and Rita Brdar-Szabó. 2007. When Zidane is not simply Zidane, and Bill Gates is not just Bill Gates: Or, Some thoughts on online construction of metaphtonymic meanings of proper names. In Radden, Günter, Klaus-Michael Köpcke, Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (125–142). Amsterdam & Philadelphia: John Benjamins.



Metonymic chains and the nature of metonymy 247

Brdar-Szabó, Rita. 2002. Referentielle Metonymie im Sprachvergleich. In Barota, M., P. Szatmári­, J. Tóth and A Zsigmond (eds.), Sprache(n) und Literatur(en) im Kontakt (53–65). Szombathely: Pädagogische Hochschule Berzsenyi Dániel. Brdar-Szabó, Rita and Mario Brdar. 2003. Referential metonymy across languages: What can cognitive linguistics and contrastive linguistics learn from each other? IJES, International Journal of English Studies 3(2): 85–105. Cameron, Lynne and Graham Low. 2004. Figurative variation in episodes of educational talk and text. European Journal of English Studies 8(3): 355–373. Cameron, Lynne J. and Juurd H. Stelma. 2004. Metaphor clusters in discourse. Journal of Applied Linguistics 1(2): 107–136. Croft, William. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4(4): 335–370. Fass, Dan. 1991. met*: A method for discriminating metonymy and metaphor by computer. Computational Linguistics 17(1): 49–90. Fass, Dan. 1997. Processing Metonymy and Metaphor. Greenwich & London: Ablex Publishing Corporation. Fauconnier, Gilles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press. Gibbs, Raymond W., Jr. 1994. The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Goatly, Andrew. 1997. The Language of Metaphors. London & New York: Routledge. Hilpert, Martin. 2007. Chained metonymies in lexicon and grammar: A cross-linguistic perspective on body-part terms. In Radden, Günter, Klaus-Michael Köpcke, Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (77–98). Amsterdam & Philadelphia: John Benjamins. Koller, Veronika. 2003. Metaphor clusters, metaphor chains: Analyzing the multifunctionality of metaphor in text. metaphorik.de 5: 115–134. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago & London: The University of Chicago Press. Lakoff, George. 1993. The contemporary theory of metaphor. In Ortony, Andrew (ed.), Metaphor and Thought (202–251). Cambridge: Cambridge University Press. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago & London: The University of Chicago Press. Lakoff, George and Mark Johnson. 2003. Afterword. In Lakoff, George and Mark Johnson, ­Metaphors We Live By (243–274). Chicago & London: The University of Chicago Press. Lakoff, George and Mark Turner. 1989. More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago & London: The University of Chicago Press. Nerlich, Brigitte and David D. Clarke. 2001. Serial metonymy: A study of reference-based polysemisation. Journal of Historical Pragmatics 2(2): 245–272. Panther, Klaus-Uwe. 2005. The role of conceptual metonymy in meaning construction. In Ruiz de Mendoza Ibáñez, Francisco José and Sandra Peña Cervel (eds.), Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (353–386). Berlin & New York: ­Mouton de Gruyter. Panther, Klaus-Uwe and Linda L. Thornburg. 2002. The roles of metaphor and metonymy in English -er nominals. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (279–319). Berlin & New York: Mouton de Gruyter.

248 Rita Brdar-Szabó and Mario Brdar

Panther, Klaus-Uwe and Linda L. Thornburg. 2003. Introduction: On the nature of conceptual metonymy. In Panther, Klaus-Uwe and Linda L. Thornburg (eds.), Metonymy and Pragmatic Inferencing (1–20). Amsterdam & Philadelphia: John Benjamins. Radden, Günter. 2002. How metonymic are metaphors? In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (407–433). Berlin & New York: Mouton de Gruyter. Radden, Günter. 2005. Metonymic blending and the construction of meaning. Paper read at the International Cognitive Linguistic Conference “Converging and Diverging Tendencies in Cognitive Linguistics”, 17–18 October 2005, Dubrovnik, Croatia. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Radden, ­Günter and Klaus-Uwe Panther (eds.), Metonymy in Language and Thought (17–59). Amsterdam & Philadelphia: John Benjamins. Reddy, Michael. 1979. The conduit metaphor: A case of frame conflict in our language about language. In Ortony, Andrew (ed.), Metaphor and Thought (284–324). Cambridge: Cambridge University Press. Ruiz de Mendoza Ibáñez, Francisco José. 1999. From semantic underdetermination, via metaphor and metonymy to conceptual interaction. LAUD Series A: General and Theoretical Papers 492. Essen: Linguistic Agency, University of Duisburg. Ruiz de Mendoza Ibáñez, Francisco José. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (109–132). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José. 2008. Cross-linguistic analysis, second language teaching and cognitive semantics: The case of Spanish diminutives and reflexive constructions. In De Knop, Sabine and Teun De Rycker (eds.), Cognitive Approaches to Pedagogical Grammar: Volume in honor of René Dirven (121–152). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José and Olga Díez Velasco. 2002. Patterns of conceptual interaction. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (501–546). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza, Ibáñez, Francisco José and Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Radden, Günter, Klaus-Michael Köpcke, ­Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (33–49). ­Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco José and Sandra Peña Cervel. 2002. Cognitive operations and projection spaces. Jezikoslovlje 3(1–2): 131–158. Semino, Elena. 2008. Metaphor in Discourse. Cambridge: Cambridge University Press. Taylor, John R. 1989. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press. Ullmann, Stephen. 1962. Semantics: An Introduction to the Science of Meaning. Oxford: Blackwell. Yu, Ning. 2007. Cultural identity and globalization: Multimodal metaphors in a Chinese educational advertisement. China Media Research 3(2): 25–32.

Metonymic matrix domains and multiple formations in indirect speech acts* Xianglan Chen University of International Business and Economics, Beijing

The notion of matrix domain in metonymy needs further exploration in connection to its role in the unfolding of contextual elements in discourse. On the basis of different dialogues in Chinese, I examine the way shifts in indirect speech act meaning relate to changes in the metonymic source and target domains underlying such meaning. I claim that matrix domains can have different configurations within the same discourse: in some situations, one source domain can map onto more than one target domain, while in others several source domains can map onto only one target domain. I propose that speech acts sometimes change from one (sub)type to another as the speaker–hearer interaction develops, remodeling the initial configuration of the matrix domain. Keywords: matrix domain, matrix domain formation, ISA-type change

1.

Introduction

It is well accepted in Cognitive Linguistics that metonymy is a conceptual mechanism, called a mapping, by means of which part of a conceptual domain affords access either to other elements within the same domain or to the whole domain, or whereby the whole domain gives access to a relevant part of it (Kövecses and ­Radden, 1998). Thus, when we mention metonymy, we actually refer to a conceptual mechanism, i.e., a form of thinking or even an inferential schema (Panther and Thornburg, 2004), rather than simply to a rhetorical device. In the literature

* This chapter has been financially supported by the Research Center of Business English and Cross-Culture at the University of Business and Economics, China.

250 Xianglan Chen

on metonymy in Chinese, metonymy is regarded as the equivalent of so-called jiedai (借代) (Shen, 1999). As a matter of fact, jiedai is a term from rhetoric referring to cases of metonymy that have a literalness dimension. This notion comes very close to what Panther and Thornburg (1999) have called referential metonymy. But this is only one among several metonymy types discussed in the literature. Panther and Thornburg (1999) have further distinguished referential metonymy from predicational metonymy (e.g., I’ll be brief for “I will speak briefly”) and illocutionary metonymy (I’m thirsty for “give me something to drink”). Other scholars (e.g., Ruiz de Mendoza and Pérez, 2001; Ruiz de Mendoza and Mairal, 2007) have added another category of metonymy, high-level or grammatical metonymy, which is characterized by having an impact on grammar (e.g., mentioning a process to make it stand for an action, as in The door opened, meaning “Someone or something opened the door”). Figure 1 below captures the interrelationship among the different categories of metonymic thinking generally recognized in the literature. Recognized as literal (typically, referential metonymy, ‘jiedai’) Illoc utionary metonymy

Metonymic thinking (metonymy)

Predicational metonymy Unrecognized as literal (typically, non-referential metonymy)

Grammatical metonymy

…

Figure 1.  Categories of metonymic thinking

Through entrenchment (cf. Langacker, 1999), metonymy may acquire a literalness dimension. This can happen not only with referential metonymy (which is lexical), as in (1) below, but also with other kinds of metonymy, as in (2), discussed further below: (1) 他 是 这里 的 新手。 He is here mod new hand. “He is a new hand here.”

In Chinese, new hand (新手), which stands for ‘new worker’ (新职员), has a strong literalness dimension. Chinese speakers do not perceive new hand to be a figurative use of language. Consider also Example (2) below, borrowed from Ruiz de Mendoza and Otal (2002: 83):



Matrix domain formation in ISAs 251

(2) He hammered the nail into the wall.

This example is different from (1) above since there is no referential operation associated to hammer. Ruiz de Mendoza and Pérez (2001) have argued that in cases like this there is an underlying high-level instrument for action metonymy, which can be regarded as a grammatical metonymy since it involves the conversion of a noun with an instrumental function into a verbal predicate. This is evidenced by the paraphrase of hammer (a nail) as ‘use a hammer to drive (a nail) in’. Also, consider the sentence pattern in which the speaker mentions that he/she has a certain problem that is not beyond the hearer’s ability to solve: I have a headache, I am hungry, I am thirsty, It’s too hot in here, etc. Each of these sentences is likely to be understood as standing for a request to help the speaker out of his/her problem. This means that in the right context the expression of a situation that affects the speaker (or someone else) negatively may stand for a demand for the hearer to act in such a way that situation does not affect the speaker any longer. This is another case of non-referential metonymic thinking that underlies illocutionary interpretation, an idea that has been explored in Thornburg and Panther (1997), Panther and Thornburg (1998), Pérez and Ruiz de Mendoza (2002), Panther (2004), and Ruiz de Mendoza and Baicchi (2007), among others. However, these studies focus on single utterances that have either one illocutionary interpretation or several alternative interpretations depending on contextual variables. The present chapter intends to develop the seminal ideas in this previous work on metonymy and speech act meaning by looking into two phenomena which show in discourse: how one single metonymic source can have different targets and how successive metonymic shifts license the possibility of creating speech act transitions. Since how discourse unfolds is grounded in socio-cultural convention, we shall illustrate these phenomena with reference to a conversational sequence taken from the Chinese play Thunder Storm by Cao Yu.

2.

Ruiz de Mendoza’s double metonymy

Cognitive linguists generally insist on a (domain-internal) two-domain view of metonymy (Barcelona, 2000; Lakoff, 1987; Langacker, 2000; Panther and ­Thornburg, 1998, 1999, 2003, 2004; Radden and Kövecses, 1999). However, Ruiz de Mendoza (2000), Ruiz de Mendoza and Otal (2002), and Ruiz de Mendoza and Díez (2002) have drawn attention to some communicative and grammatical phenomena, especially in the field of metonymic anaphora, which cannot be explained from the point of view of a two-domain approach. For example, in Shakespeare­

252 Xianglan Chen

is on the top shelf, where in a default interpretation ‘Shakespeare’ stands for a book that contains (at least part of) Shakespeare’s literary work (author­ for work for medium­), Ruiz de Mendoza (2000) distinguishes two matrix domains, ‘Shakespeare­ as a writer’ and ‘the book that contains Shakespeare’s writings’, and a common subdomain, ‘Shakespeare’s writings’. The subdomain is the target of the first mapping of a double metonymy (author for work) and the source for the second mapping (work for medium). The existence of two matrix domains allows Ruiz de Mendoza to explain the appropriateness of these two anaphoric operations, i.e., Shakespeare is on the top shelf; he is not easy to read a first / it is very thick in terms of a principle, called the Domain Availability Principle, according to which only the matrix domain of a metonymy is available for anaphoric reference. In application of this principle, reference to the subdomain (i.e., the contents of Shakespeare’s book) is rather unnatural: *Shakespeare is on the top shelf and I can’t understand it. There are other cases of metonymy that involve more than one mapping and consequently more than one matrix domain and where, in the present author’s view, metonymic anaphora does not really play any special role to distinguish matrix domains. Take the following examples drawn from Ruiz de Mendoza and Díez (2002: 512–513, 514): (3) a. I want to buy an apartment in Wall Street (= the place). b. Wall Street (= the institution) will never lose its well-deserved prestige. c. Wall Street (= the people in the institution) is in panic. (4) His sister heads (carries out the action of leading) the policy unit.

Place

Institution

People

Figure 2.  place for institution for people



Matrix domain formation in ISAs 253

Head

Leader/ agent

Action of leading

Figure 3.  head for leader for action of leading

In (3a) Wall Street is used in a non-figurative way. It refers to a street in the southern section of Manhattan in New York. In (3b) Wall Street metonymically stands for one of the most important financial institutions in the United States. This is a target-in-source metonymy (place for institution), since the financial institution is located in Wall Street. In (3c) there is a double metonymic operation. There are two target-in-source metonymies, where the target of the first process refers to the institution that is the source in the second one. In (4) a part of the body, the head, metonymically refers to ‘the person carrying out the action of leading’. As the body part has no direct relationship with the action of leading, two metonymic mappings are needed, as illustrated in Figure 3. If we compare Figures 2 and 3, it is clear that they are just opposite in terms of how their source and target domains interact. In the second model, there are two cases of domain expansion. The target of the first metonymy is the source for the second metonymic shift. Through double domain expansion, ‘head’ in the first metonymy and ‘leader’ in the second metonymy are both highlighted, while the ultimate meaning interpretation rests mainly on the second matrix domain, i.e., the idea of leading. We agree with Ruiz de Mendoza and his collaborators that double metonymic operations are also necessary to explain examples like (3c) and (4) above. It is clear that Wall Street in (3b) only takes one metonymic shift, which would be insufficient to explain the meaning of (3c). In a similar fashion, the interpretation of ‘head’ as the action of leading is only possible if we understand ‘head’ as ‘leader’; that is, we define the verbal predicate ‘head’ as ‘act as a head (meaning “leader”) does’. However, in Examples (3c) and (4), there is no way anaphora can help us determine the two matrix domains, simply because in both cases one of the matrix domains works as a source domain for the final target and is thus an intermediate step in the double metonymic process. In cases like this, anaphoric operations

254 Xianglan Chen

are only possible with reference to the broadest matrix domain, whether it is the initial source or final target domain: Wall Street was in panic at first but it (*they) finally found something to cheer about when some key stocks rose; His sister heads the new unit; it (i.e., leading the unit) means a lot to her. Two more observations are relevant at this point. First, in order to understand the full meaning potential of a double metonymic mapping, it is necessary to take into account the existence of two matrix domains, especially since, as we have noted above, matrix domains seem to have a more prominent role in metonymic interpretation. The importance of matrix domains has not been fully addressed by Ruiz de Mendoza. Second, the double metonymies Ruiz de Mendoza has identified take place on the referential level. We believe that on the non-referential level, as in the case of indirect speech acts, there can also be double metonymies – and consequently more than one matrix domain – playing a relevant role in interpretation. We will explore this issue in the following section.

3.

Matrix domains and multiple formations

Thornburg and Panther (1997) and Panther and Thornburg (1998) have made the interesting proposal that speech act meaning is a matter of affording relevant access, through a part–whole metonymic operation, to special cognitive models that they label illocutionary scenarios. This proposal is a very important development of the traditional position held by Lakoff (1987) with respect to the role of metonymy in inferencing and this is also the first time that indirect speech acts have been discussed in terms of idealized cognitive models. At the same time, the proposal involves a drastic departure from other perspectives on illocution within the domain of inferential pragmatics. In this respect, Panther and Thornburg­ (1998) observe that one important shortcoming of traditional approaches to speech acts is that they cannot tell us why people can – given the appropriate context – derive indirect speech act meaning straightforwardly from utterances without any extra effort (cf. Gibbs, 1994: 86). Panther and Thornburg believe this deficiency in previous accounts can be solved by their approach, which will be explored in more detail in the following subsections.

3.1

Matrix domains with one source and various possible target domains

Let us now consider Example (5), taken from Radden (2005): (5) Let’s go to bed now.



Matrix domain formation in ISAs 255

Radden (2005) observes that go to bed is usually understood metonymically as ‘go to sleep’, but that we may have access to more than one metonymic target from the same reference point ‘go to bed’. In this sense, ‘go to bed’ can be part of a ‘sex’ idealized cognitive model or icm, which in some special situations is understood as ‘(go) have sex’. Let us put Radden’s observation in a slightly different perspective in connection to a specific context. That is, in different situations, go to bed can refer to different subdomains of the same icm. But it is argued in this chapter that even in the same situation sometimes the expression ‘go to bed’ can refer to both these target domains: ‘go to sleep’ and ‘have sex’. Think of a situation in which a middle-aged couple have been married for a long time and are not very active in love-making although they love each other deeply. They are more like friends. When the husband asks his wife to “go to sleep” with him without much enthusiasm, the wife may have the following two guesses: either ‘really go to bed to sleep’ or ‘go and have sex’. When the wife hears what the husband says without looking at her husband’s facial expression, the wife actually does not have enough information. These two guesses are possible. She may feel that her husband is tired and really wants to go to bed to rest or that her husband is only asking her to take some rest. The other possible guess is that her husband wants to have sex and is therefore asking for her opinion. Usually, as the amount of information increases, the target becomes more specific. In the whole process one source domain and the two target domains are involved, as shown in Figure 4. Immediate situation

Have sex Go to bed

Go to sleep

Figure 4.  ‘Go to bed’ matrix domain development

In this chapter we mainly talk about two kinds of metonymic operations involving more than two domains. The first kind consists of one source domain and more than one target domain. In this case, the activation is usually not so specific that

256 Xianglan Chen

more than two target domains are possibly available in the process. Example (5) is such a case. Here is another example. (6) 老师: 小王 快要 答辩了吧? Teacher: XiaoWang soon defend his PhD dissertation-per-pre “Is XiaoWang going to defend his PhD dissertation soon?” 学生: 他 在 北图 查 资料。 Student: He now at National Library look for reference materials “He is looking for some reference materials at National Library.”

In this example the reply is an indirect speech act. According to Panther and Thornburg’s (1998) scenario approach, the whole event can be divided into three parts: “before”, “core” and “after”. The “before” part includes the preliminary elements of ‘writing a dissertation’ such as ‘looking for reference materials’, ‘the ability to write the dissertation’, ‘reading the reference materials’, ‘absorbing the ideas’, and so on. The “core” part of the event refers to ‘writing the dissertation’. The “after” part includes ‘checking’, ‘revising’, ‘defending’, ‘making the final decision’ and so on. The reply “He is looking for some reference materials in the National Library” (他在北图查资料) metonymically activates the whole scenario where a person is still in the process of writing his PhD dissertation, which underlies the inference that ‘He has not finished his dissertation’. It should be noted that the reply in Example (6) is based on the before part of the event. Why does the speaker choose the “before” part and not the “after” part to activate the whole? This is a question of pragmatic factors such as good manners and face-saving strategies. On the basis of his background knowledge the speaker believes that it is not good to say directly “He has not finished his dissertation” or “He is just going to write”. He chooses an indirect way to express this. It is the background knowledge that guides the speaker to express the idea indirectly. The immediate situation in this example is that the two communicators are teacher and student. The student takes the immediate situation into consideration and believes that the indirect expression is better. In this way the immediate situation also works in guiding the choice of an adequate expression. The final expression comes from the speaker’s total belief about the environment together with other personal factors. That means that pragmatic factors do influence the choice of expressions. There is no doubt that pragmatic factors play a role in inferencing processes. In pragmatics, a lot of research has been carried out on the function of pragmatic factors. But in cognitive linguistics, the function of pragmatic factors has largely been ignored up to very recently. In this chapter it is believed that metonymic thinking – as part of human thinking – is also guided and controlled by (i) general background knowledge, (ii) the interpreter’s construal of the immediate situation,



Matrix domain formation in ISAs 257

and (iii) individual pragmatic factors, the three of which influence the meaning achievement process in a significant way. Background knowledge is the conceptualized knowledge structure in long-term memory. Together, our background knowledge and the information derived from the immediate situation make up the cognitive context in which metonymy operates. Our background knowledge is not unchangeable, since it combines with our construal of the immediate situation in the real environment and changes as the immediate situation is accessed and built into the language user’s set of background assumptions. If the immediate situation is consistent with the background knowledge, the background knowledge will become strengthened. If the immediate situation is not consistent with the background knowledge, the background knowledge will be changed with it. The immediate situation refers to the physical environment or situation. But not all the physical environment has something to do with inferencing. Only the part of the immediate environment that draws the speaker and hearer’s attention can influence the process. The individual pragmatic factors refer to the factors about individual psychology, personality and personal intention influencing the metonymic operation. Readers acquainted with Sperber and Wilson’s (1995) relevance pragmatics will easily identify the three factors distinguished here with the relevance-theoretic notion of cognitive environment. This notion refers to the speaker’s knowledge of the world, combined with their assumptions about the context of situation, their assumptions about the addressee’s sets of beliefs and their ability to construct mental representations about any new information that is presented to them). Our proposal, however, refines Sperber and Wilson’s since it highlights the mutual adaptive feedback between background knowledge and the immediate situation as perceived and interpreted by the speaker and adds individual pragmatic factors that may also affect interpretation. Now let us see how metonymic thinking works and how the matrix domain comes into being as a composite of different sources of information. In the above example the source domain is ‘He is looking for some reference materials in the National Library’. The possible targets activated in the hearer’s mind may be ‘He wants to write the dissertation’, ‘He has not begun to write the dissertation’, or possibly ‘He is going to write the dissertation’. These possible target domains may be activated at the same time. In this sense the whole metonymic thinking process from the source to the target involves more target domains than just one. Each indirect expression can be regarded as a subdomain in this event scenario that acts as a matrix domain. Note that all the subdomains together form the matrix domain. The whole operation is illustrated in Figure 5 below, where CBK stands for cognitive background knowledge, IC for the immediate context and IPF for individual pragmatic factors.

258 Xianglan Chen

IC

CBK

3 1

IPF

4

2

Figure 5.  Information inputs for an indirect speech act matrix domain

In this indirect speech act event, there is one source domain and more than one target domain. These domains together form a complex matrix domain that we call Event Matrix Domain. The Event Matrix Domain proposal made in this chapter surely absorbs some of the ideas from the scenario theory, such as the division of the domain into a “before”, a “core” and an “after”. However, it differs from ­Panther and Thornburg’s (1998) scenarios in two ways: (i) Panther and ­Thornburg (1998) place emphasis on the nature of metonymic thinking and do not consider the role of the matrix domain in indirect speech acts, but we argue that the indirect speech act works metonymically on the basis of a composite matrix domain; (ii) the function of our background knowledge, the immediate situation, and individual pragmatic factors in the conceptual operation are not the focus in scenario theory but play a central role in our own proposal.

3.2 Matrix domain with various sources and one target 3.2.1 Various sources with different illocutionary force Examples (5) and (6) feature a matrix domain with one source domain and more than one target domain. In (7) below (see Appendix 1 for the original text in Chinese), we will examine the case of part of a Chinese play, Thunder Storm by Cao Yu, where there is a matrix domain with one target domain and more than one source domain and where the illocutionary force of each subdomain within the same matrix domain is not the same in terms of its directive impact. In the play, the characters Zhouping (a young man) and Sifeng (a young woman who is Zhouping’s servant) love each other secretly and decide to go out to live together, in spite of their social distance. But Sifeng’s mother is strongly against their relationship, since Zhouping is also her child. Zhouping and Sifeng are unaware of the fact that they are brother and sister before they fall in love with each other. At this time Sifeng is pregnant already. At first, Sifeng is unwilling to disclose this fact to her mother. When she asks her mother to permit them to go away and live together, Sifeng uses indirect forms of approach:



Matrix domain formation in ISAs 259

(7) Sifeng (head bent, holding her mother’s hand tightly and helplessly): Mum. I have to leave you. Mother Lu (can’t help crying): You two cannot be together! (Stands up.) No, not at all! Sifeng (retreating emotionally): Mum, you cannot make such a decision. Mother Lu: No, not at all! (Dully.) Let’s go and leave this place, let’s go. Sifeng (begging): Mum. Would you like your daughter to be too urgent to die in front of you? Mother Lu (strictly): Let’s go! (Towards Sifeng) Feng, listen, I prefer not having you as my daughter. I cannot allow you to live together with him. Let’s go. (Pulls Sifeng towards the door.) Sifeng: Oh. Mum! (Passes out in her mother’s arm.) Mother Lu (holding Sifeng in her arm): My poor child! You… Zhuping (anxiously): She fell in a faint. Mother Lu (getting some cold water to pour in): Feng, my child. Come back to life. Come back to life… my poor child. Sifeng (her mouth opens, eyes open, takes a deep breath): Oh, Mum. Mother Lu (satisfied): Child. Don’t complain about my behavior. I cannot speak out my bitter experience. Sifeng (taking another deep breath): Mum. Mother Lu: What? Feng. Sifeng (towards Zhouping): I, I shouldn’t choose not to tell you, Ping. Zhouping: Feng, are you better? Sifeng: Ping, I always hide the truth from you. (Looks at her mother beggingly.) Not dare to tell you. Mum, You … Mother Lu: What? My child. Speak out. Sifeng: (sobbing): I, I (bravely) I have already had… with him. (Cries loudly.) Mother Lu (anxiously): What? You said that you have already had… (Shocked, standing still.) Zhouping (pulling her hand): Sifeng! What, Is it true? You… Sifeng (crying): Yes. Zhouping (happy with bitterness): When? When? Sifeng (head bent): About three months. Zhouping (satisfied): Oh, Sifeng, Why not tell me? I, mine… Mother Lu (in a low voice): My God! Zhouping (talking towards Mother Lu): Grandmother Lu, You should not be stubborn any more. It is all my fault. I beg you. (Kneels down.) I beg you to allow her to go out to live together with me. I promise that I will be always good to her in the future and also good to you. Sifeng (stands up at once, walks towards her mother and kneels down too): Mum. Have pity for us. Allow us to go.

260 Xianglan Chen

Mother Lu (keeping silent, seated dully): I am in a dream. My son and my daughter. I gave birth to them. 30 years… oh, my God. (Covers her face, waves her hand.) Go away. I will not recognize you. (Turns her head to other side.) Zhouping: Thank you! (Stands up at once.) Feng! (Sifeng stands up and walks two steps.) Mother Lu (turns back unconsciously): No, you can’t. (Kneels down again.) Sifeng (begging): Mum. You, what’s wrong with you? I have made up my mind already. No matter how rich he is or how poor he is, no matter what he is, I am his. I’ve given my heart to him and he is the only person in my eyes. Mum, up to this step, wherever he is, I will be with him. Whatever he is, I will be his. Mum. Can’t you understand?

From the very beginning, Sifeng uses indirect requests. According to the scenario theory, each indirect request can metonymically activate the whole, whose meaning is ‘Please allow us to go away to live together’. All the indirect requests work in the whole event and form the whole request matrix domain. The distance between each subdomain and the core part of the matrix domain is different. The subdomains in this matrix domain are different in their degree of indirectness. The matrix domain formed by all the subdomains is a special kind of Event ­Matrix Domain, with no unitary illocutionary interpretation, which we call the Illocutionary Force Matrix Domain. Since there are some other indirect speech acts, such as refusals, these also contribute to the configuration of the Event ­ Matrix Domain (Chen, 2006, 2009). The Illocutionary Force Matrix Domain is the part of the Event Matrix Domain that refers to the whole indirect speech act event scenario. Example (6) above is a common Event Matrix Domain, but Example (7) is a special one with different kinds of illocutionary force subdomains. In this example the subdomains are the following:

(1) (2) (3) (4) (5) (6)

Mum. I have to leave you. Mum, you cannot make such a decision. Mum, would you like your daughter to be too urgent to die in front of you? Oh. Mum! (Passes out in her mother’s arm.) I, I (bravely) I have already had… with him. (Stands up at once, walks towards her mother and kneels down too.) Mum. Have pity for us. Allow us to go away. (7) (Begging.) Mum. You, what’s wrong with you? I have made up my mind already. No matter how rich he is or how poor he is, no matter what he is, I am his. I’ve given my heart to him and he is the only person in my eyes. Mum, up to this step, wherever he is, I will be with him. Whatever he is, I will be his. Mum. Can’t you understand?



Matrix domain formation in ISAs 261

In order to express herself indirectly, the speaker at first chooses expressions that are far from the core meaning. When she fails, the speaker makes some changes to use less indirect expressions that are closer to the core meaning. So it is easier for the hearer to know the intention. In the matrix domain, the closer the expression is to the core meaning, the easier the hearer can understand the speaker’s intention. Although there is no direct metonymic relationship among the subdomains, there is a relationship among the subdomains in the whole meaning achievement process. In this way the whole operation involves more subdomains that form a metonymy inferencing chain within the matrix domain. In this request matrix domain, all the subdomains (1)–(7) above belong to the “before” part. The distances from the subdomains to the core meaning are different. With the development of the event, the distance between the subdomain and the core meaning is shorter with a smaller degree of indirectness, as shown in Figure 6. CBK

1

IC

2

3

IPF

8 Go together

Figure 6.  Illocutionary Force Matrix Domain in an indirect request

Why choose indirect expressions and why choose the before part to activate the whole? The reason comes from the influence of background knowledge, the immediate context and individual pragmatic factors. Let us see how these factors function. Sifeng is a servant, but she is pregnant from her young master. This fact could not be accepted by the society of the time. Even to her mother, she dared not speak out. Her idea of keeping the love relationship secret is based on the background knowledge of her culture. With this cognitive background knowledge, Sifeng chooses indirect ways to express herself. The immediate situation also plays a role in the whole process. The immediate situation is that Sifeng is pregnant already and, since her mother is not aware of this problem, she will not allow Sifeng to live together with Zhouping. When her communicative intention is not achieved on the basis of increasing indirectness, the daughter is increasingly compelled to tell the truth openly. During the whole process in the matrix

262 Xianglan Chen

domain, the individual pragmatic factors also function. The intention to leave with Zhouping acts as the decisive individual factor which constraints the choice of each indirect request and also of the different degrees of indirectness. Under this constraint the request develops from indirect to direct expressions step by step. The three kinds of factors all influence the choice of expressions and the development of the whole request act.

3.2.2 Matrix domains and speech act type transitions Within an indirect speech act event, sometimes speech acts change from one type or subtype to another as the interaction between the speaker and the hearer develops. In (8) below what we have is an example of transition from ordering to requesting and then to begging (see Appendix 2 for the original Chinese text): (8) Sifeng (turns back, back against the window): Go! The person outside: No! (The person outside pulls the window with strength. Sifeng blocks the window as tightly as possible.) Sifeng (anxiously): No. No. You should not come in tonight. The person outside (in a low voice): Sifeng, beg you! Open the window. Sifeng: No! No! It is already midnight. I have taken off my clothes. The person outside (anxiously): What? Sifeng: I am already lying in bed. The person outside (with trembling voice): Then, then, I… I (taking a deep breath). Sifeng (persuading): You’d better go, OK? The person outside (making a change): OK. OK. I will go. (anxiously again) But you’d better open the window first and let me… Sifeng: No. No. Go at once. The person outside (anxiously): No, Sifeng, only let me… Only let me give you a kiss. […] Sifeng (bitterly): Oh, the young master. This is not your residence. Forgive me. The person outside (complaining): You have forgotten me. You don’t miss… Sifeng: Who has changed? The person outside (anxiously): But you don’t open the window and let me in. You know that I love you truly. Sifeng (saying sadly): Don’t pester me again, will you? Today, the whole day, how many troubles do you bring to my family? Isn’t that enough? The person outside (sincerely): I know that I am wrong. But I want to see you now. Want to see you.



Matrix domain formation in ISAs 263

Sifeng (taking a breath): That is all right. Every thing is put off till tomorrow. Tomorrow I will obey you. You can do whatever you like! The person outside (sincerely): Tomorrow? Sifeng: Please go now. Beg you. Please go at once.

In the discourse context that gradually unfolds in Example (8), there are three speech acts each of which is a subdomain of the next. This situation is captured in Figure 7 below. Ordering: (1) (turns back, back against the window): Go! (2) (anxiously): No. No. You should not come in tonight. (3) No! No! It is already midnight. I have taken off my clothes. Requesting: (4) I am already lying in bed. (5) (persuading): You’d better go, OK? (6) No. No. Please go at once. Begging: (7) (bitterly): Oh, The young master. This is not your residence. Forgive me. (8) (saying sadly): Don’t pester me again, will you? Today, the whole day, how many troubles do you bring to my family? Isn’t that enough? (9) (taking a breath): That is all right. Every thing is put off till tomorrow. Tomorrow I obey you. You can do whatever you like! (10) Please go now. Beg you. Please go at once.

This exchange takes place between Sifeng and Zhouping. Sifeng is worried that her mother will come into her room at any moment. She dares not let Zhouping come in through the window secretly, although she loves him. At first she strongly rejects Zhouping with the speech act of ordering as in (1), (2) and (3). Facing Zhouping’s entreaty again and again, she changes the type of speech act from ordering to requesting as in (4), (5) and (6). Finally it is Sifeng who begs Zhouping to go away. Let us see why one speech act can shade into another. Sifeng does not want her mother to know about the love relationship between Zhouping and her since it will cause more trouble to the whole family because of their difference in social position. This thought is based on their culture and becomes part of her cognitive background knowledge. The immediate situation is that her mother will come into her room any time while Zhouping is urging her from outside the house. She loves Zhouping so she changes her way of speaking. The individual pragmatic factor is that she does not want to bring any trouble to the family. Under the influence of these three kinds of factors, the speech act type changes from

264 Xianglan Chen

ordering to requesting and then from requesting to begging. Among the three kinds of speech act, the degree of the willingness of the speaker and the hearer is changing from time to time in this example. In the speech act of ordering, the hearer has no choice but to obey. The speaker’s willingness is fully expressed. In the speech act of requesting such as “(persuading): You’d better go, OK?”, this expression is quite different from the expression in the act of ordering in the sense that the hearer has more choice and the willingness of the speaker is much lower. In the speech act of begging such as “(bitterly): Oh, The young master. This is not your residence. Forgive me”, the hearer’s degree of choice is extremely high. The willingness of the speaker is much lower. We can say that the transition of speech act type is the transition of the relative parameters between the hearer and the speaker. The transition is caused by the influence of cognitive background knowledge (CBK), the immediate context of situation (IC) and the individual pragmatic factors (IPF). CBK

1

IC

2

3

IPF

8 Go together

Figure 7.  Speech act type transition and metonymic thinking

According to the scenario theory, the speech act of ordering – such as “go!”–, the speech act of requesting – as in “I am lying in bed”–, and the speech act of begging – as is the case of “Oh, the young master. This is not your residence. Forgive me”– can all activate the whole separately. They are different speech act types and, in our view, they are also the different subdomains of a more complex event scenario. In turn, the cognitive background knowledge, the immediate situation and individual pragmatic factors influence the transition from one speech act type to another. Although there is no direct metonymic relationship among the subdomains, the understanding of the whole event involves metonymic thinking on the basis of a composite matrix domain. It may be interesting to note that the situation of successive embedding of each speech act type into the next, which we have just described, appears to run



Matrix domain formation in ISAs 265

parallel to what the grammatical system of some languages, including English, does to conventionally convey the different illocutions mentioned. A request can be constructed on the basis of an order through mitigating adverbial formulas and tags: Open the window (order) > Open the window, please/will you?/can you?

And a request can be transformed into begging through mechanisms that indicate insistence, such as repetition (cf. Ruiz de Mendoza and Baicchi, 2007): Open the window, please, please, do!

The possibility to transform one illocutionary type into another with more conceptual ingredients suggests that illocutionary scenarios can be arranged in terms of subdomain-matrix domain relationships. Begging contains more conceptual material than requesting, which is richer than ordering. In the case of illocutionary meaning conveyed through purely inferential clues, as in (8) above, the easiness with which the speaker creates transitions from one speech act to another is probably supported by (or at least is consistent with) the prior subdomain-matrix domain relationships.

4.

Conclusion

In cognitive linguistics, there is a lot of recent research on metonymy. But there is still little work on the notion of matrix domain – including the way it is created and communicatively exploited – and on the formation of indirect speech acts. The present chapter is a first attempt in this direction. Through the analysis above, we have tried to show that the matrix domain is the more reasonable candidate for an operational domain for indirect speech acts. Our analysis has focused on three aspects of indirect speech acts: different degrees of illocutionary force, the transition between speech act types, and the notion of event matrix domain. We have found that in discourse there are situations of matrix-domain embedding, which can be exploited metonymically when one matrix domain gives access to another, a situation which results in double metonymies. The research on such phenomena is just at an incipient stage. There are probably many other kinds of metonymic complexes waiting for further research.

266 Xianglan Chen

References Barcelona, Antonio. 2000. Introduction. The cognitive theory of metaphor and metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads (1–15). Berlin & New York: Mouton de Gruyter. Chen, Xianglan. 2006. 转喻与语用推理 (Metonymy and pragmatic inferencing). Unpublished PhD thesis. Beijing University. Chen, Xianglan. 2009. 转喻思维的多样性 (Multiple formations in metonymic thinking operations). Foreign Language Research 1: 40–44. Gibbs, Raymond W., Jr. 1994. The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Kövecses, Zoltán and Günter Radden. 1998. Metonymy: Developing a cognitive linguistic view. Cognitive Linguistics 9(1): 37–77. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: University of Chicago Press. Langacker, Ronald. 2000. A dynamic usage-based model. In Barlow, Michael and Suzanne ­Kemmer (eds.), Usage-Based Models of Language (1–63). Stanford, Cal.: CSLI Publications. Panther, Klaus-Uwe and Linda L. Thornburg. 1998. A cognitive approach to inferencing in conversation. Journal of Pragmatics 30: 755–776. Panther, Klaus-Uwe and Linda L. Thornburg. 1999. The potentiality for actuality metonymy in English and Hungarian. In Panther, Klaus-Uwe and Günter Radden (eds.), Metonymy in Language and Thought (333–357). Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg (eds.). 2003. Metonymy and Pragmatic Inferencing. Amsterdam & Philadelphia: John Benjamins. Panther, Klaus-Uwe and Linda L. Thornburg. 2004. The role of conceptual metonymy in meaning construction. metaphorik.de 6: 91–111. Pérez Hernández, Lorena and Francisco José Ruiz de Mendoza Ibáñez. 2002. Grounding, semantic motivation, and conceptual interaction in indirective speech acts. Journal of Pragmatics 34: 259–284. Radden, Günter. 2005. The ubiquity of metonymy. In Otal Campo, José Luis, Ignasi Navarro I Ferrando, and Begoña Bellés Fortuño (eds.), Cognitive and Discourse Approaches to Metaphor and Metonymy (11–28). Castellón: Jaume I University. Radden, Günter and Zoltán Kövecses. 1999. Towards a theory of metonymy. In Panther, KlausUwe and Günter Radden (eds.), Metonymy in Language and Thought (17–59). Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco José. 2000. The role of mappings and domains in understanding metonymy. In Barcelona, Antonio (ed.), Metaphor and Metonymy at the Crossroads (109–132). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco J. and Annalisa Baicchi. 2007. Illocutionary Constructions: Cognitive motivation and linguistic realization. In Kecskés, I. and L. Horn (eds.), Explorations in Pragmatics: Linguistic, Cognitive, and Intercultural Aspects (95–128). Berlin & New York: Mouton de Gruyter. Ruiz de Mendoza Ibáñez, Francisco José and Olga I. Díez. 2002. Patterns of conceptual interaction. In Dirven, René and Ralf Pörings (eds.), Metaphor and Metonymy in Comparison and Contrast (489–532). Berlin & New York: Mouton de Gruyter.



Matrix domain formation in ISAs 267

Ruiz de Mendoza Ibáñez, Francisco José and Ricardo Mairal Usón. 2007. High-level metaphor and metonymy in meaning construction. In Radden, Günter, Klaus-Michael Köpcke, ­Thomas Berg, and Peter Siemund (eds.), Aspects of Meaning Construction (33–51). Amsterdam & Philadelphia: John Benjamins. Ruiz de Mendoza Ibáñez, Francisco José and José Luis Otal Campo. 2002. Metonymy, Grammar, and Communication. Albolote, Granada: Comares. Ruiz de Mendoza Ibáñez, Francisco José and Lorena Pérez Hernández. 2001. Metonymy and the grammar: motivation, constraints, and interaction. Language and Communication 21: 321–357. Shen, Jiaxuan. 1999. 转指和转喻 (Zhuanzhi and metonymy). Contemporary Linguistics 1: 3–15. Sperber, Dan and Deirdre Wilson. 1995. Relevance: Communication and Cognition. 2nd edition. Oxford: Basil Blackwell. Thornburg, Linda L. and Klaus-Uwe Panther. 1997. Speech act metonymies. In Liebert, WolfAndreas, Gisela Redeker, and Linda R. Waugh (eds.), Discourse and Perspective in Cognitive Linguistics (205–219). Amsterdam & Philadelphia: John Benjamins.

Appendix 1 四凤：(低头，不得已紧握着鲁妈的手）妈，我只好先离开您了。 鲁妈：(忍不住） 你们不能在一块！ （站起） 不，不成！ 四凤：(死命地退缩） 妈，您不能这样做。 鲁妈：不，不成！（呆滞地，单调地） 走，走。 四凤：(哀求）妈，您愿您的女儿急得要死在你的面前么？ 鲁妈：(不得已，严厉地） 走！（向四凤）凤儿， 你听着，我情愿没有你，我不能叫 你跟他在一块儿。 ————走吧（立刻拉住四凤，向门口走） 四凤：啊，妈妈！ （晕倒在母亲怀里） 鲁妈：(抱着四凤）我苦命的孩子， 你———— 周萍：(急）她晕过去了。 鲁妈：(拿凉水灌四凤）凤儿，好孩子，你回来，你回来。————我的苦命的孩子。 四凤：(口渐张，眼睁开，喘出一口气）啊，妈！ 鲁妈：(安慰她）孩子，你不要怪妈心狠，妈的苦说不出。 四凤：(叹出一口气）妈！ 鲁妈：什么？凤儿。 四凤：(向周萍）我，我不能不告诉你，萍！ 周萍：凤，你好点了没有？ 四凤：萍，我，总是瞒着你，（乞怜地望着鲁妈）也不肯告诉您。妈，您— 鲁妈：什么，孩子，快说。 四凤：(抽咽）我，我——（放胆）我跟他现在已经有……（大哭） 鲁妈：(切迫地）怎么，你说你有——（过受打击，不动） 周萍：(拉动四凤的手）四凤！怎么，真的，你—— 四凤：(哭）嗯。 周萍：(悲喜交集）什么时候？什么时候？ 四凤：(低头）大概已经三个月。 周萍：(快慰地）哦，四凤，你为什么不告诉我，我，我的—— 鲁妈：(低声）天哪。

268 Xianglan Chen

周萍：(走向鲁）鲁奶奶，你无论如何不要再固执哪，都是我错了，我求您！（跪 下）我求您放了她吧。我敢保我以后对得起她，对得起您。 四凤：(立起，走到鲁妈面前跪下）妈，你可怜可怜我们，答应我们，让我们走吧。 鲁妈：(不做声，坐着，发痴）我是在做梦。我的儿女，我自己生的儿女，三十年 功夫——哦，天哪，（掩面哭，挥手）你们走吧，我不认得你们！（转过头 去）。 周萍：谢谢您！（立起）我们走吧。凤！ (四凤起，走了两步。) 鲁妈：(回头，不自主地) 不，不能够！ (四凤又跪下。) 四凤：(哀求）妈，您，您是怎么？我的心定了。不管他是富，是穷，不管他是谁， 我是他的了。我心里第一个许了他，我看得见的只有他。妈，我现在到了这一 步。他到哪儿，我也到哪儿。他是什么，我也跟他是什么。妈，你难道不明 白， 我—

Appendix 2 四凤：(转身，背倚着窗户）你走！ (外面的人：我不！) (外面向里用力推窗门，四凤用力挡住) 四凤：(焦急）不，不，你不要进来。 (外面的人：（低声）四凤，我求你，你开开！) 四凤：不！不！已经到了半夜，我的衣服都脱了。 (外面的人：（急迫地）什么？） 四凤：我已经在床上睡了！ (外面的人：（颤声）那……那……我就……我（叹一口长气）) 四凤：(恳求地）那你走吧，好不好？ (外面的人：（转了口气）好，也好，我就走，（又急切地）可是你先打开窗 户叫我……) 四凤：不，不，你就走！ (外面的人：（急切地）不，四凤，你只叫我……只叫我亲亲你。) 四凤：(苦痛地）啊，大少爷，这不是你的公馆，你饶了我吧。 (外面的人：（怨恨地）那么你忘了我了，你不再想……) […] 四凤：谁变了心？ (外面的人：（急躁地）那你为什么不打开窗户，让我进来，你不知道我是真 爱你么？) 四凤：(哀诉地）你别再缠我好不好？今天一天你给我们闹出多少事，你还不够么？ (外面的人：（真挚地）那我知道错了，不过，现在我要见你，要见你。) 四凤：(叹一口气）好，那明天说吧！明天我依你，什么都成！ (外面的人：（恳切地）明天？) 四凤：你走吧。我求你，你走吧。

Authors’ biodata

Antonio Barcelona is Full Professor of English Language and Linguistics at the Department of English and German, University of Córdoba, Spain. His research has covered such areas as the pragmatic motivation of syntax (especially constituent order), conceptual metaphor and metonymy and cognitive linguistics in general. His early published work (from 1976 until about 1993) was concerned with the pragmatic motivation of certain constituent order structures in English and Spanish: inversion, theme/topic selection, various types of raising, existential-presentative and other presentative constructions, converses, passive and others. Since 1986 onwards his publications have focused on the clarification of the cognitive-linguistic notions of metaphor and metonymy and on their interaction, on the role of these two cognitive mechanisms in various types of cognitive models and grammatical structures, and most recently, on the role of conceptual metonymy in cognition and in linguistic meaning and form, among other topics. Among his many international publications in these areas: the edited book Metaphor and Metonymy at the Crossroads: A Cognitive Perspective (Mouton de ­Gruyter, 2000), and articles like “Metaphorical Models of Romantic Love in Romeo and ­Juliet” (Journal of Pragmatics 25 (1995), 667–688), “The case for a metonymic basis of pragmatic inferencing: Evidence from jokes and funny anecdotes” (in K. Panther and L. Thornburg (eds.) (2003) Metonymy and Pragmatic Inferencing. Amsterdam: Benjamins, 81–102), “Metonymy behind grammar: The motivation of the seemingly ‘irregular’ grammatical behavior of English paragon names” (in G. Radden and K. Panther (eds.) (2004) Studies in Linguistic Motivation. Berlin: Mouton de Gruyter, 357–374), or “The role of metonymy in meaning at discourse level: A case study” (in G. Radden, K. Köpcke, T. Berg and P. Siemund­ (eds.) (2007), Aspects of Meaning Construction. Amsterdam: Benjamins, 51–75). One of his most recent publications is the co-edited book (with Klaus-Uwe Panther­ and Linda Thornburg), Metonymy and Metaphor in Grammar (Amsterdam: ­Benjamins, 2009). Réka Benczes (PhD 2005) is Assistant Professor at the Department of American Studies, Eötvös Loránd University, Budapest. Her main fields of interest centre around linguistic creativity, word formation and applied cognitive linguistics. Her first book, Creative Compounding in English, appeared in 2006, published in the Human Cognitive Processing series of John Benjamins. Her second book, an ­ introduction to cognitive linguistics in Hungarian (Kognitív nyelvészet)

270 Defining Metonymy in Cognitive Linguistics

co-authored­ with Zoltán Kövecses, was published in 2010 (Budapest: Akadémiai­ Kiadó). Réka has also done occasional review work for the journals Cognitive Linguistics, Journal of Pragmatics, Journal of Pragmatics and Cognition, Linguistics and Lingua, as well as for the Human Cognitive Processing book series (John Benjamins). Mario Brdar is Professor of English Linguistics in the Department of English Language and Literature at Josip Juraj Strossmayer University, Osijek. He is the editor of Jezikoslovlje and a member of the editorial board of Review of Cognitive Linguistics, Atlantis, Bosanski jezik, and The Open Applied Linguistics Journal. He is currently the vice-president of the Croatian Applied Linguistics Society. His main areas of research interests include cognitive linguistics, syntax, and lexical semantics. His publications mostly deal with contrastive and cognitive-functional approach to grammatical constructions and basic cognitive processes such as metonymy and metaphor. Rita Brdar-Szabó is Associate Professor of German Linguistics in the School of Germanic Studies at Eötvös Loránd University (Budapest, Hungary). Her main research interests include cognitive linguistics, morphology (in particular word formation), lexical semantics, contrastive linguistics and contact phenomena involving English, German, Hungarian, Croatian and Russian. She has published on the cognitive theory of metaphor and metonymy, grammaticalization, blending, prototype theory, and usage-based models. Xianglan Chen, PhD, has worked as an Associate Professor at the University of International Business and Economics after finishing her post-doctoral program at Peking University in 2008. Her main research area is cognitive linguistics and inferential pragmatics. She has been the head of nine research projects, five of which are on metonymic mechanisms. Her projects have been financed – among others – by the National Post-Doctoral Association (no. 20070420267) and the Ministry of Education (no. 09YJA740022). She has published 40 papers, mainly in the core journals in China, most of which are on metonymy. She has co-authored or co-edited five books. She has presented her research papers at various conferences, such as the 10th International Cognitive Linguistic Conference in Krakow (Poland), the 40th Annual Meeting of the Societas Linguistica Europaea in Joensuu (Finland), the Communication, Cognition and Media: Communication Sciences International Congress, the International Conference on Figurative Language Learning and Figurative Language Use: Theory and Applications (CRAL 2009) and the Eighth International Conference on Researching and Applying Metaphor (RaAM 8). She has been a keynote speaker in China on several occasions.



Authors’ biodata 271

Dirk Geeraerts (1955, PhD 1981) holds the chair of theoretical linguistics at the University of Leuven. He is the head of the research unit Quantitative Lexicology and Variational Linguistics (see http://wwwling.arts.kuleuven.be/qlvl/ for more information). His main research interests involve the overlapping fields of lexical semantics and lexicology, from the theoretical point of view of cognitive linguistics, with a specific descriptive interest in social variation and a strong methodological commitment to corpus analysis. His involvement with cognitive linguistics dates from the 1980s, when his PhD was one of the first in Europe to explore the possibilities of a prototype-theoretical model of categorization. As the founding editor of the journal Cognitive Linguistics, he played an important role in the international expansion of cognitive linguistics. At present, he is the managing editor of the book series Cognitive Linguistics Research (published by Mouton de Gruyter). His publications include the following authored and edited books: Paradigm and Paradox (1985), Woordbetekenis (1986), Wat er in een woord zit (1989), The Structure of Lexical Variation (1994), Diachronic Prototype Semantics (1997), Convergentie en divergentie in de Nederlandse woordenschat (1999), Words and Other Wonders (2006), Cognitive Linguistics: Basic Readings (2006), The Oxford Handbook of Cognitive Linguistics (2007), Theories of Lexical Semantics (2009). Tanja Gradečak-Erdeljić holds a PhD degree in Linguistics (University of Zagreb) and is currently positioned as an Assistant Professor and researcher at the Josip Juraj Strossmayer University of Osijek, Department of English. She teaches Introduction to Linguistics, Morphosyntax and Semantics of the Verb Phrase, Syntax of the Multiple Sentence, Pragmatics and History of English. Her interests include semantics, cognitive linguistics and pragmatics, which served as a linguistic background for more than a dozen of the original scientific papers she authored or co-authored. She is also the Editorial Secretary of the journal Jezikoslovlje and a member of the ICLA and DGKL. Javier Herrero Ruiz, PhD, graduated in English Philology with a specialization in linguistics at the University of La Rioja (Spain) in 2002. In 2003 he started his doctoral courses with Prof. F.J. Ruiz de Mendoza as supervisor and took part in a research project related to cognition and the compilation of lexical databases. In 2004 he was a visiting scholar at UC Berkeley, under the tuition of Prof. E. Sweetser; he also attended some NLING meetings, chaired by Prof. G. Lakoff. He defended his PhD thesis in 2008. His publications focus on pragmatics, discourse analysis, and Cognitive Linguistics. He has also presented his research at international conferences such as the International Conference on Cognitive Linguistics or AESLA (the Spanish Association of Applied Linguistics).

272 Defining Metonymy in Cognitive Linguistics

Anu Koskela works at the University of Sussex, where she teaches courses on semantics and cognitive linguistics. She received her doctorate from Sussex in 2007. Her research concerns cognitive approaches to lexical semantics and categorisation – her interests include polysemy and sense demarcation, lexical relations (in particular hyponymy), variation in category boundaries and the nature of conceptual metonymy. Goran Milić is a Teaching Assistant at the Department of English Language and Linguistics, University of Osijek, where he teaches courses in Discourse Analysis, Word Formation and Multi-Word Verbs. He got his MA in Linguistics from the University of Zagreb and is finishing his PhD thesis on the zoosemy of domestic animals in English and Croatian from a cognitive linguistic point of view. His research interests focus on the analysis of the impact of discoursive elements on Conceptual Metaphor and Metonymy Theory. He is a member of the ICLA. Carita Paradis is Professor of English Linguistics at the Centre of Language and Literature at Lund University in Sweden. Her main research interests concern meaning in the broad sense. She specializes in lexical semantics and the modelling of meaning within the Cognitive Linguistics framework. She combines corpus methods and experimental methods in her work. A presentation of her research interests and a list of selected publications are available at http://www. sol.lu.se/person/CaritaParadis. Yves Peirsman (1982) is a postdoctoral fellow of the Research Foundation − Flanders (FWO) at the University of Leuven, and a visiting postdoctoral scholar at Stanford University. His PhD, which was supervised by Dirk Geeraerts, focused on distributional, corpus-based models of lexical semantics. Yves has published on a variety of lexical-semantic topics, both from a cognitive-linguistic and a computational-linguistic perspective. Francisco José Ruiz de Mendoza Ibáñez has been a full professor of English Linguistics at the University of La Rioja since 1999. He is the author of the book ­Introducción a la teoría cognitiva de la metonimia (An introduction to the cognitive theory of metonymy, Granada: Método, 1999), and co-author of Metonymy, Grammar, and Communication (Granada: Comares, 2002) and Teoría Lingüística: Métodos, herramientas y paradigmas (Linguistic theory: Methods, tools and paradigms, Madrid: Ramón Areces & UNED, 2010). He is also co-editor of the books Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction (Mouton de Gruyter, 2005) and Cognitive Linguistics: Current Applications and Future Perspectives (Mouton de Gruyter, 2006). Ruiz de Mendoza has authored numerous articles and book chapters which have appeared in journals such as the Journal of Pragmatics, Language and Communication, Folia Linguistica, ­Italian Journal of



Authors’ biodata 273

­ inguistics, Jezikoslovlje, LAUD-Essen, and in book series such as Cognitive LinL guistics Research (Mouton), Topics in English Linguistics (Mouton), Pragmatics and Beyond (John Benjamins), Lodz Studies in Language (Peter Lang). He has also been a keynote speaker at international conferences such as the International Contrastive Linguistics Conference (Santiago, 2005), Converging and Diverging Tendencies in Cognitive Linguistics (Dubrovnik, 2005, 2008), Perspectives on Metonymy (Lodz, 2005), the 10th International Cognitive Linguistics Conference (Krakow, 2007), the 7th International Conference on Researching and Applying Metaphor (Cáceres, 2008), and the 3rd Conference on Metaphor in Language and Thought (Fortaleza, 2008). He serves on the scientific board of a number of journals, among them Jezikoslovlje, Estudios Ingleses de la Universidad Complutense, Revista Española de Lingüística Aplicada, Revue Romane and Cognitive Linguistics. He is the editor of the Review of Cognitive Linguistics (John Benjamins) and co-editor of the series Applications of Cognitive Linguistics (Mouton de Gruyter). He is the current president of the Spanish Association of Applied Linguistics and the founder and head of the Research Center in the Applications of Language (CRAL, University of La Rioja). 

Metaphor and metonymy index

A action/event/process for state  28 affection is warmth  112 anger is (the) heat (of a fluid in a container)  37, 39, 45, 47 angry human behavior is aggressive animal behavior  47 artist for work  33 attention is a (typically moving) physical entity 46 author for work  35, 97, 150, 252 B bad is good  113 body heat for anger  39, 45 body part for person  10, 20 C capital for government  217, 230, 239, 245 category for a member of the category  130 category for defining property  155 cause for effect  202, 221, 225 characteristic properties for individual  143 container for contents  94, 131, 207 country for government 230 D down for sadness  42 downward bodily orientation for sadness  42

E ear for attention  46 effect for cause  38, 121, 188–189, 221, 225 entity for characteristic 130 F fact for one of its conventional explanations  24–25 food for customer  47 form of behavior a is the opposite form of behavior b  114 G generic for specific  111, 116–119, 221 goals are destinations  173 H hand for human being  129, 133 hand for worker  106–107 having control/force is up 169 having sex is eating  172 humans are animals  147, 155, 157–161 humans are objects  152 I ideas are food  180 individual for countable collection  130 instrument for agent  24 K knowing is seeing  171

L life is a journey  12 less is down  39 location for action  202 location for institution  21 love is a journey  169 lower is upper  114 lowest is uppermost  114 M material for object  27, 202 member of a category for the category  22, 130, 202, 207 more is up  39, 47 much is little  114 O object for material constituting the object 159 object involved in the action for the agent of the action  202 P part for part  17, 20–21, 24, 47, 50, 152 part for whole  17, 20, 24, 27, 34, 38, 50, 96, 234, 237, 242 part of the scenario for the part of the scenario  162 part of the scenario for the whole scenario  152, 162 part of a thing for the whole thing  15 participant for action  94 path for goal  34 people are animals  47, 112–113, 169 person for characteristic behaviour  161

276 Defining Metonymy in Cognitive Linguistics

Picasso for his work  35 place for event  239 place for people (living there)  94, 97, 183 politics is war  152 potentiality for actuality 23, 56, 215, 266 present for habitual  117 process for action for result  77 producer for product  34, 131, 200 R result for condition  24 S sadness is down  39–40, 42, 44, 46

salient body part for animate being  94 salient body part for person  20 salient event for a salient successive (or co-occurring) event  43 salient property for category  207 similarity is closeness  157 society is a family  152 specific for generic  111, 221 style of dance stands for culture  226 subevent for complex event  27 T time is money  223

U unfit, overweight person is a potato  202, 210 upper is lower  114 uppermost is lowermost  114 V verticality scale for quantity scale  39 W White House for U.S. government  42 whole for part  2, 9, 15–18, 20–21, 34, 43, 50, 94 whole thing for a part of the thing  15

Name index

A Aijmer, K.  77, 82, 85 Alarcón, P.  172, 191 Allwood, J.  128, 144 B Bach, K.  119–120 Barcelona, A.  1–8, 10, 12, 16, 18–21, 23–25, 27–29, 33–34, 37, 39, 42, 45–47, 53–57, 67–68, 82, 86, 90, 93, 102–103, 115, 120–123, 126, 143–144, 146– 147, 149–150, 153–159, 163–164, 168, 170, 178, 191–192, 197– 200, 204, 208–210, 213–214, 217–218, 227–229, 244, 246, 248, 251, 266, 269 Barnden, J.  50–51, 54 Bartsch, R.  43, 54 Becker, T.  128, 144 Bellés Fortuño, B.  57, 266 Benbow, T.  44, 46, 54 Benczes, R.  1, 5, 7, 43, 63–64, 71, 82, 103, 130, 150, 155, 164, 197, 204, 206, 214, 269 Berg, T.  6, 54, 86, 122, 164, 214, 246–248, 267, 269 Berlin, B.  5–6, 53–57, 82–87, 102, 120–123, 134, 144–146, 163–165, 191–193, 214–215, 238, 246–248, 266, 269 Blank, A.  70, 82, 146 Bloomfield, L.  70, 82 Borst, E.  80, 82 Boye, K.  77, 82 Brdar, M.  5, 10, 12, 14–15, 110, 115, 120, 150, 156, 159, 164, 217, 229, 240–241, 244, 246–247, 270 Brdar-Szabó, R.  5, 12, 14–15, 115, 120, 150, 156, 159, 164, 217, 229, 240–241, 243–244, 246–247, 270 Brinton, L.  78, 82

Brown, P.  151, 164 Brugman, C.  80, 82 Buchstaller, I.  78, 82 Bultinck, B.  43, 55 Burridge, K.  147–148, 163–164 Bybee, J.  62, 82 C Cameron, L. J.  229, 247 Carpenter, M.  67, 86 Carston, R.  118–120, 128, 146, 174, 191 Chen, X.  5, 99, 108, 249, 260, 266, 270 Clarke, D. D.  70, 84, 87, 126, 130–131, 146, 229, 247 Clausner, T. C.  134, 136, 144 Costello, F.  211, 213–214 Croft, W.  8–11, 16–19, 22, 28, 31, 49, 55–56, 62, 65–67, 82, 90, 95, 102, 106, 115, 120, 125–126, 128–129, 132–136, 138–139, 142–145, 150, 164, 198–200, 204–205, 209–210, 214–215, 227, 247 Cruse, A.  31, 55, 62, 64–65, 82, 90, 102, 125–126, 128, 131, 137, 142–143, 145 Cuyckens, H.  54, 56, 62, 64, 82–83, 85, 121, 144, 164, 214 D Dasher, R.  62, 70–71, 78, 80, 87 Denison, D.  76, 82, 85 Díez Velasco, O.  155, 165, 203, 215, 248 Dirven, R.  1, 3, 5–6, 19, 22, 31, 38, 43, 54–57, 67, 82–84, 102, 120–122, 144–145, 148, 164– 165, 168, 191–193, 209–210, 213–215, 247–248, 266 Downing, P.  199, 203, 214

E Eckardt, R.  63, 79, 82, 86 Ekberg, L.  61, 78–80, 82–83 F Fass, D.  229, 247 Fauconnier, G.  12, 14, 29, 41, 48, 55, 104, 108, 110–111, 120, 123, 148, 156, 163–164, 172, 191, 198, 214, 245, 247 Feyaerts, K.  12, 37, 55, 126, 145, 198, 202, 204, 214 Fillmore, Ch.  41, 55, 64, 83, 168, 191 Fisher, O.  62, 83 G Gärdenfors, P.  67, 69, 83 Geeraerts, D.  4, 16, 26–27, 32–33, 50–51, 55–56, 62, 64, 67–68, 76, 82–83, 85, 89, 93, 100, 102, 106, 126, 128–130, 143, 145–146, 170, 198–199, 203, 208–209, 214–215, 271–272 Gibbs, R.  1, 6, 24, 55, 67, 69, 83, 149, 165, 170, 191, 225, 247, 254, 266 Goatly, A.  229, 247 Goldin-Medow, S.  67, 83 González Díaz, V.  83 Goossens, L.  39, 43, 45–47, 55, 109, 121, 170, 191, 203–204, 214 Gradečak-Erdeljić, T.  4, 41, 147, 149–150, 162, 165, 271 Grady, J.  157, 163, 165, 171, 173, 191 Gries, S. Th.  73, 83

278 Defining Metonymy in Cognitive Linguistics

H Halle, M.  55, 85 Harder , P.  77, 82 Haser, V.  67, 83, 126, 145 Heine, B.  79, 83 Herrero, J.  5, 167, 171–175, 177–178, 180, 182–185, 187, 190–191, 271 Hilpert, M.  229, 247 Hoey, M.  116, 121 Hopper, P.  62–63, 70, 78, 83 Horn, L.  122, 127, 145, 266 J Jakobson, R.  19, 55 Janssen, Th.  185, 191 Johnson, Ch.  1, 6, 8, 10, 14, 20–21, 31, 47, 56, 66, 84, 112, 121, 125, 130, 145–146, 158, 165, 168, 171, 191–192, 221–222, 247 Johnson, M.  1, 6, 8, 10, 14, 20–21, 31, 47, 56, 66, 84, 112, 121, 125, 130, 145–146, 158, 165, 168, 171, 191–192, 221–222, 247 Jucker, A.  78, 83 K Kemmer, S.  77, 84, 266 Kempson, R.  127, 145 Kitis, E.  76–77, 84 Koch, P.  70, 82, 126, 128, 131, 145–146 Koskela, A.  4, 22, 67, 69, 116, 125, 137, 145, 207, 272 Kövecses, Z.  1, 3, 6, 8–15, 20–21, 25, 34–35, 37–39, 41, 45, 47, 55–57, 67, 84–85, 94, 96, 102, 115, 117, 122, 125–126, 129–131, 133, 145–146, 148, 156–157, 160, 165, 168, 192, 197–198, 200, 208–209, 214–215, 220–221, 229, 236, 238, 248–249, 251, 266, 270 L Lakoff, G.  1, 6, 8–10, 12–15, 20–21, 31, 40–41, 45, 47, 56, 64, 66, 84, 104, 112, 115, 121, 125, 129–131, 145, 151–152, 155, 157–159, 161–162, 165, 167–170,

185, 192, 197, 207, 214, 221–227, 247, 251, 254, 266, 271 Lamb, S.  208–209, 212–214 Langacker, R.  9, 11–14, 16–19, 22, 25–26, 29–33, 35–36, 43, 49–50, 56, 64, 66–68, 84, 89–93, 102, 104, 107, 121, 126, 128, 133–136, 138, 145, 154, 165, 198–200, 204–205, 208, 215, 250–251, 266 Lapata, M.  67, 84 Lascarides, A.  67, 84 Lehrer, A.  128, 145 Lenker, U.  78, 84 Levin, M.  61, 69, 84 Levinson, S. C.  151, 164 Lewandowska-Tomaszczyk, B. 54, 191, 246 Libben, G.  210–212, 215 Liebert, W.  57, 165, 193, 267 Lightfoot, D.  62, 84 Liszkowski, U.  67, 86 M Mairal Usón, R.  86, 121–122, 248, 267 McKechnie, J. L.  36, 56 Medin, D. L.  135, 145–146 Méndez Naya, B.  78, 80, 84 Michaelis, L.A.  109, 121 Milić, G.  5, 41, 147, 157, 165, 272 Moore, T.  128, 146 Morgan, P. S.  148, 165, 239 Murphy, G.  125, 131, 135, 146 N Navarro i Ferrando, I.  57, 266 Nerlich, B.  70, 84, 87, 126, 130–131, 146, 229, 247 Nevalainen, T.  78, 80, 84 Niemeier, S.  170, 192 Nunberg, G. D.  96, 102, 105, 121 O Ohnuki-Tierney, E.  157, 165 Otal Campo, J. L.  24, 57, 121, 266–267 P Panther, K.-U.  1, 3, 6–8, 10, 23–24, 28, 33, 52, 54–57, 67,

69, 75, 83–85, 94, 99, 102, 115–117, 120–122, 144–147, 149, 152, 164–165, 168, 170, 192–193, 198, 214–215, 220, 229, 245–251, 254, 256, 258, 266–267, 269 Papafragou, A.  67, 85 Paradis, C.  4, 7, 32–34, 51, 56, 61–65, 67–69, 76–80, 82–83, 85, 89–91, 94–95, 97, 102, 106, 121, 126, 128, 143, 146, 209, 215, 272 Paul, H.  55, 70, 83, 85 Pauwels, P.  55 Peirsman, Y.  4, 16, 26–27, 32–33, 50–51, 55–56, 67–68, 85, 89, 93, 102, 106, 126, 129–130, 143, 145–146, 170, 198, 208–209, 214–215, 272 Peña Cervel, S.  54, 102, 121–122, 164–165, 246–248 Pérez, L.  111, 116–117, 122, 169, 175, 193, 250–251, 266–267 Peters, H.  78, 80, 85 Pörings, R.  1, 5, 54–55, 57, 67, 82–84, 102, 120, 122, 145, 164, 191–193, 213–215, 247–248, 266 Pustejovsky, J.  67, 85, 92, 102 R Radden, G.  1, 3, 6, 8–15, 19–22, 25, 34–35, 37, 39, 41–42, 46–47, 54–57, 67, 69, 83–86, 94, 96, 102, 115, 117, 122, 125–126, 129–131, 144–146, 156–157, 164–165, 170, 192–193, 197, 200, 208–209, 214–215, 218, 220–221, 229, 236, 238, 245–249, 251, 254–255, 266–267, 269 Récanati, F.  119, 122 Reddy, M.  229, 248 Redeker, G.  57, 165, 193, 267 Riemer, N.  39, 57 Rissanen, M.  78, 80, 84–86 Ryder, M. E.  206, 215 Roberts, I.  62, 85, 146 Rohdenburg, G.  127, 146 Rosch, E.  19, 57, 134, 146 Rosenkvist, H.  70, 85 Roussou, A.  62, 85



Rudzka-Ostyn, B.  55 Ruiz de Mendoza, F.  1–4, 6, 10, 12, 16–18, 24, 33, 53–54, 57, 67, 77, 86, 90, 95–96, 100, 102–106, 108–109, 111, 115–119, 121–123, 126, 146, 155, 164–165, 169–170, 173–175, 177, 185, 190, 192–193, 197, 203, 215, 218, 221, 226, 229, 235, 244–248, 250–254, 265–267, 271–272 S Santibáñez, F.  105, 108, 111, 123, 174–175, 193 Semino, E.  229, 236, 248 Seto, K.  126, 131, 146 Siemund, P.  6, 54, 86, 122, 164, 246–248, 267, 269 Simon-Vanderbergen, A.  55 Sperber, D.  119, 123, 174, 185, 193, 257, 267 Stern, G.  70, 86 Stoffel, C.  80, 86 Sweetser, E.  70, 77, 86, 171, 193, 271

Name index 279

T Tagliamonte, S.  78, 86 Talmy, L.  62, 66, 86, 104, 123 Taylor, J. R.  10, 31, 33–35, 57, 68, 77, 87, 121, 134, 136, 144, 146, 170, 193, 199, 206, 208, 212, 215, 218, 225, 248 Thornburg, L.  1, 6–8, 10, 23–24, 28, 33, 52, 54, 56–57, 67, 69, 75, 84–85, 94, 99, 102, 115–117, 120–122, 147, 149, 164–165, 168, 170, 192–193, 197–198, 215, 220, 229, 246–251, 254, 256, 258, 266–267, 269 Tissari, H.  77, 86 Tomasello, M.  67, 86 Traugott, E.  61–63, 70–71, 78–80, 82–83, 86–87 Turewicz, B.  54, 191, 246 Turner, M.  8, 10, 14–15, 41, 45, 48, 55–56, 104, 108, 110–111, 115, 120–121, 123, 125, 129, 145, 156–157, 163–165, 168, 172, 191–192, 225, 247

U Ullmann, S.  70, 87, 130, 146, 218, 248 Ungerer, F.  78, 87 V Vanparys, J.  55 Verhagen, A.  77, 87 W Warglien, M.  67, 83 Warren, B.  12, 57, 67, 69, 87, 147, 165 Waugh, L.  57, 165, 193, 267 Wilson, D.  119, 123, 128, 146, 174, 185, 193, 257, 267 Wierzbicka, A.  117, 123, 132–133 Wood, E.  66–67, 82 Y Yoshimura, K.  68, 77, 87 Yu, N.  226, 248, 251, 258 Z Zaenen, A.  105, 121

Subject index A a bit  80, 177, 183–184 abductive reasoning  70 absolute  50, 78–80, 82, 152 absolutely  80, 212 abstraction  22, 66, 105, 110–111, 126, 134, 142–143, 175 active zone  26, 31–35, 43, 51, 100, 107–108, 143 afraid  10, 72, 76–77, 86 association  43, 76, 86, 129, 154, 156, 198, 209–210, 270–271, 273 autohyponymy  127, 144 see also vertical polysemy awful  78–80 awfully  80 B backgrounding  65, 132, 142–144, 147–150, 162, 205 base  9, 43, 64, 107–108, 126, 132–140, 142, 159, 200, 202, 244 see also profile boundedness  65, 68, 71, 78 broadening  4, 22, 67, 125–126, 129–130, 133–134, 142, 144, 174 C coercion  92, 109 cognitive background knowledge (CBK)  5, 257, 261, 263–264 cognitive operations  4, 103–105, 107–108, 118–119, 122–123, 167–168, 173–175, 189, 193, 218, 222, 248 content operations  103–105, 111, 120, 175 formal operations  103–105, 120 complete  78–80, 106–107, 188

completely  80, 180–181, 190 compound  130, 154–156, 159–160, 163, 197, 199–200, 202–207, 210–215, 245 concept  9, 17, 34, 37, 53, 69, 89, 96–97, 101, 103, 105–107, 109–110, 120, 134–135, 143, 148, 151–152, 156, 160–161, 174, 177, 189, 197–200, 209, 212, 217, 220, 226–227, 236, 244–246 conceptual distance  74, 99 conceptual iconicity  149–152, 162 conceptual integration  2, 108, 123, 165, 245 integration by combination 108–109 integration by enrichment 108, 110 conceptual link  39, 93–94, 98 conceptual structures  4, 65, 68, 131, 143, 151 construals  61–62, 64–67, 71, 75, 80–81, 85 contrasting  111–112, 114, 141, 175–176 contiguity  4, 12, 27–29, 68, 89, 93–94, 116, 126, 129–131, 134, 207, 209–210, 218–220, 225–226 conventionality  17–18, 22, 34, 51, 94 conventionalization  2, 4, 9, 34–35, 62–63, 68–71, 76, 81, 161 see also properties of metonymy correlation  104, 108–109, 111–112, 115, 173, 175, 191, 218 correspondence  4, 12, 104, 111–114, 169, 175, 221–223

crawl  72–76 crawler  72, 75 Croatian  120, 147, 149, 157, 160–162, 240, 242–243, 270, 272 cueing  4, 23, 105, 107–108, 113, 116, 120, 175 D dead  80, 173 deep frame  153, 159 degree  65–66, 78–80, 85 discourse  5, 20, 46, 48, 53–55, 57, 65, 71–73, 75, 77, 82, 85, 87, 102, 108, 116, 121, 147–150, 152–153, 158, 160–165, 193, 214, 217–218, 227–229, 236, 238, 241–243, 245–249, 251, 263, 265–267, 269, 271–272 domain  2–5, 7–9, 11–21, 25–33, 35–44, 48–49, 51–53, 64, 68, 77–78, 89, 91, 93–94, 96, 103–104, 106–108, 110–112, 114–116, 119–120, 125–127, 129, 131–142, 144, 147–150, 152–153, 155–157, 159–160, 163, 168–171, 173, 175, 178–180, 182, 190, 197–200, 202–206, 208–213, 217, 219–221, 223– 227, 234, 239, 242–245, 249, 251–255, 257–258, 260–262, 264–265 and categorial inclusion  127 criteria for  32 distinct 138, 140 functional 3, 12, 14, 20, 32, 41, 42, 43, 51, 52, 53, 147, 149, 155, 156, 157, 159, 163, 198 taxonomic  22, 40–43, 53, 126–127, 131, 134–135, 148–149

282 Defining Metonymy in Cognitive Linguistics

domain expansion  4, 12, 103– 104, 106–107, 111–112, 116, 119, 170, 173, 175, 180, 190, 245, 253 domain highlighting  4, 7, 9, 11, 16, 91, 106, 116, 125, 132–133, 142, 199, 227 see also highlighting domain matrix  9, 11–12, 16–18, 49, 126, 129, 132–133, 138, 142, 150, 197–198, 200, 202, 206, 208, 210, 213, 239, 243 domain network  5, 43, 197, 206, 208–211, 213 domain reduction  4, 103–104, 106–107, 110, 112, 116, 119–120, 170, 173, 175, 178, 182, 190, 227 draft  72, 76 dreadful  78–80 dysphemism  5, 147, 149, 151, 154, 162, 167, 171, 173, 177, 184, 187–190 E echoing  111–113, 175 emotion  73, 76–77, 141 euphemism  5, 147, 151–153, 162–163, 167, 171, 173, 177, 184, 187–188, 190 Event Matrix Domain (EMD) 258, 260, 265 F facet  33, 89, 97, 100, 107, 143 see also properties of metonymy, problematic properties facetization  4, 32–33, 65, 68, 89–91, 94–95, 97–101, 103, 105–107, 110, 116, 120 fairly  80 fear  72, 76–77, 84, 86, 147 figure of speech  167 foregrounding  65, 71, 78–80, 147, 149–150, 162, 200 fovame  72, 76 frames  12, 41, 43, 55, 64, 71, 81, 151–152, 158, 162, 168, 170, 191 frightfully  80 functional cognitive domain 149 see also domain, functional

G gradient  71, 81 gradual  4, 27, 70, 81 grammaticalization  62, 70–71, 79, 81–87, 270 Great Chain of Being  157–158 H highlighting  2, 4, 7, 9, 11, 13, 16, 22, 49, 91, 95, 106–108, 110, 116, 120, 125, 132–133, 142–144, 150, 170, 178, 182, 199–200, 203, 205, 227, 236 see also properties of metonymy, problematic properties horrible  78–80 horribly  80 hyperbole  70, 105, 111, 113–114, 174–175 I idealized cognitive model (icm) 3, 11–12, 14, 17, 20, 24, 40–43, 51–52, 64, 129, 148–150, 155–156, 167–168, 171, 177, 185–187, 197–198, 208, 212–213, 220, 224, 234, 255 Illocutionary Force Matrix Domain (IFD)  260–261 illocutionary metonymy  24, 250 immediate context of situation (IC)  257, 264 implicit target  31, 51 indirect speech act  249, 254, 256, 258, 260, 262 individual pragmatic factors (IPF)  5, 257–258, 261–262, 264 inference  5, 13, 20, 24–25, 38, 53–54, 99, 105, 145–146, 149, 154, 162, 175, 179–180, 217– 218, 223–224, 245, 256 integration  2, 4, 105, 108–110, 120, 123, 165, 175, 199, 245 interaction of metaphor and metonymy  45, 55, 83, 121, 191, 203, 214 conceptual motivation of metaphor by metonymy 45

conceptual motivation of metonymy by metaphor 204 irony  5, 113, 167, 171, 173, 175, 177–178, 180, 182, 190 J jolly  80 K key  72, 76, 254 L lexical meaning as ontologies and construals (LOC)  62, 64, 81 location  21, 41, 72–75, 138, 202 M mapping  2–5, 7, 9–15, 19, 21, 26, 29–32, 37, 39–41, 44–46, 48–49, 51–53, 70, 104, 113–114, 118, 120, 129, 147–148, 150, 152–153, 155, 157–159, 161, 168–169, 172, 180, 198–200, 202–203, 217–218, 220, 222–227, 238, 240, 242–246, 249, 252, 254 see also properties of metonymy, problematic properties matrix domain 5, 94, 170, 173, 242, 249, 252, 253, 254, 255, 257, 258, 260, 261, 262, 264, 265 meaning effects  167, 173, 190 mechanisms  61–63, 100, 111, 118, 153, 159, 167, 175, 192, 265, 269–270 meiosis  105, 114, 184, 187 mental space  22, 73, 76, 108, 172, 176, 180, 242 metaphor  1–2, 4–8, 11–13, 15, 19, 29, 36–37, 39–40, 42–48, 50–57, 64, 66–67, 70, 82–86, 93, 95, 100, 102–103, 105, 109–113, 115–116, 120–123, 126, 144–146, 149, 152, 155, 157, 159–160, 162–165, 167–173, 175, 178, 180, 182, 190–193, 197, 199, 202–204, 210, 213–215, 217–227, 236, 240, 242–243,



245–248, 266–267, 269–270, 272–273 definitions of metaphor  36, 50 metaphtonymy  55, 83, 121, 191, 203, 214 metonymic chain  217, 230–232, 235, 240, 243 metonymic expansion  109, 203 metonymic reduction  110, 119 metonymic source  5, 13–14, 106, 150, 199, 217–218, 220, 225– 227, 229, 233–234, 238–239, 242–246, 249, 251 metonymic vehicle  226, 244, 246 metonymization  61–65, 67–71, 76–81 metonymy  1–17, 19–57, 62–63, 67–68, 70–71, 81–87, 89–111, 115–123, 125–127, 129–133, 142–150, 152, 154–156, 159, 161–165, 167–170, 172–173, 175, 177–186, 188–193, 197–200, 202–210, 212–215, 217–222, 224–227, 229–230, 234–254, 257, 261, 265–267, 269–270, 272–273 activation  of 2–4, 7, 9, 11–13, 16, 22, 26, 29, 31–34, 44, 49, 51, 61–63, 65, 67–71, 77, 79–81, 89–91, 93, 99–100, 103, 107–108, 120, 128, 143, 150, 178, 199, 209, 255 as asymmetrical mapping  12, 29 definitions 7, 15, 129, 198 expressed 19, 21 perspectivization  13, 32, 46 pragmatic-function 32 unexpressed  20, 31 see also properties of metonymy mitigation  111–113, 171–175, 177, 183–186, 188–190 motion  72–76 motivation  2, 10, 39, 45–47, 52, 54, 62, 83, 122, 133, 144, 157–158, 160–161, 164, 191, 193, 202, 204–205, 208, 212, 214–215, 246, 266–267, 269

Subject index 283

multiple formation  249, 254 N narrowing  4, 22, 67, 70, 125–126, 129–130, 133–134, 142–144, 174 natural inference schemas  149 Nurturant Parent model  151, 158 O ontologies  62, 64–66, 81, 85 orientation  42, 72–75 overstatement  5, 167, 171, 173– 174, 177, 183, 187, 190 oxymoron  5, 105, 114–115, 154, 167, 171, 173, 175–176, 182–183, 190 P paradox  114–115, 171, 173, 176, 271 paragon  20, 54, 110, 143–144, 244, 246, 269 parametrization  111–112, 116–119, 175 participant  72–75, 77, 94 part–whole configurations  61, 81 perfect  78–80 perfectly  80 politeness effect  152 political discourse  5, 147–149, 152–153, 160–162 polysemy  4, 22, 26, 31, 35, 51, 54, 63–64, 67, 69, 71, 80–83, 87, 102, 116, 121, 125–129, 131–132, 136, 139, 142–147, 192, 207, 214, 272 polysemy vs. vagueness  128, 145, 198, 214 see also vertical polysemy pragmatic effect  5, 147–148, 163 pragmatic function  3, 5, 14, 19, 26, 28–29, 41–43, 49, 51–53, 95, 98, 101, 147–150, 155–157, 160, 162–163, 198, 208, 238, 245 pragmatic implications  167, 187, 190 pragmatics  54, 56, 83–84, 86, 94–95, 102, 119–123, 145–146, 164–165, 185, 191–193, 247, 254, 256–257, 266, 269–273

predicational metonymy  24, 115, 250 pre-meaning  65, 73–74 pretty  33, 80, 86 profile  2, 4, 9, 26, 31–32, 61, 81, 89, 93, 101, 107, 126, 132–137, 142, 155–156, 200, 202 see also base profile/zone discrepancy  4, 89, 93, 101 profile discrepancy  2, 9, 26, 31–32 zone discrepancy  4, 89–90, 93, 101 properties of metonymy  8, 11–12, 29, 49, 51 and conventionalization  2, 4, 9, 34–35, 62–63, 68–71, 76, 81, 161 and modulation  2, 7, 9, 26, 31, 51 and zone activation  4, 31–34, 51, 61–63, 65, 67–71, 77, 79–81, 89–91, 93, 99–100, 103, 107, 120, 143, 209 problematic properties  7–8, 36, 48–49, 51 directionality of the mapping  15 distinction from metaphor 51 facets  2, 4, 7, 9, 20, 26, 31–32, 49, 51, 56, 66, 85, 89–90, 97, 102, 107–108, 121, 128, 142–143, 168, 205, 215 metonymy as a “stand-for” relationship  2, 7, 9–10, 49, 129, 209, 219 metonymy as domain highlighting  4, 7, 9, 11, 16, 91, 106, 116, 125, 132–133, 142, 199, 227 metonymy as reference point  50 problems derived from the “same/different domain criterion”  37, 51, 217 prototypicality  25–26, 28, 50, 68

284 Defining Metonymy in Cognitive Linguistics

target in whole for part metonymies  2, 9 primary domain  11, 16–18, 108 secondary domain  11, 16–18, 49, 108, 132 prototypical metonymy  21, 24 purely inferential metonymy  10, 15, 20, 23, 25, 33, 34, 50 Q quite  62, 76, 80 R rather  80, 85 really  68, 80, 175, 217, 243 reference  2, 4–5, 9–10, 13–14, 17, 20–23, 25, 29–33, 41, 49–51, 56–57, 62, 67–69, 74, 76, 81, 84, 86, 91–93, 97, 100–102, 142, 149, 155, 159, 169–170, 190, 193, 198, 203, 205, 208, 215, 227, 234, 240, 243, 247, 251–252, 254–257 see also properties of metonymy reframing  159, 163 reinforcement  173, 177, 183–184, 189–190 relation  65, 71–76, 80 resemblance  104, 108–109, 111–113, 173, 175, 191 right  80, 84, 178

S saturation  111–112, 115, 119, 175 scale  39, 65–66, 71, 78–80 schematic metonymy  19–20, 25–26, 29–30, 35, 52 purely schematic metonymy 35 selection  4, 13–14, 22, 24, 31, 66, 68, 86, 105, 108, 117, 120, 157, 163, 175, 209, 211, 269 space  38, 72–76, 132, 136 speech act-type transition speed  72–75 strengthening  111–113, 173–175, 177, 183–184, 187, 189 Strict Father Model  159 synecdoche  70, 129–130, 146, 218 T taxonomy  40–41, 134, 145, 167 terrible  78–80 terribly  80 thing  15, 31–32, 35, 43, 65, 71–72, 74–76, 79–80 total  78–80 totally  66, 80 trope  167, 177 two-domain view  251 typical metonymy  20–21, 24, 26 simply typical metonymy  22, 25, 27, 35, 50, 51 U understatement  5, 114, 167, 171, 173–174, 177, 184, 190

V vertical polyseme  141 aspirin  22, 127, 130, 143 book  16, 18–19, 31–32, 36, 49, 90–91, 93–101, 126, 128, 142–144 cat  32, 51, 90–93, 101, 107, 125, 127, 129–130, 134, 137–138, 142 cow  112, 130, 139, 142, 183, 190 dog  32, 51, 90–93, 101, 116, 127–128, 130, 134, 138–140, 142–143 drink  22, 68, 118, 125, 127–130 fill up  93, 135, 139–140, 142 finger  127, 137, 142 hoover  127 love  109, 141–142, 169 shoe  127–128, 144 the pill  20, 130 vertical polysemy  4, 22, 67, 116, 125–129, 131–132, 136, 142–145, 207 very  31, 49, 80, 82–83, 252 Z zeugma  32–33, 51, 73, 90, 95–98, 101, 106, 170 zone activation  4, 31–34, 51, 61–63, 65, 67–71, 77, 79–81, 89–91, 93, 99–100, 103, 107, 120, 143, 209 see also properties of metonymy