Motivation in Language: Studies in Honor of Günter Radden

MOTIVATION IN LANGUAGE

AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE General Editor E. F. KONRAD KOERNER (Zentrum für Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin) Series IV – CURRENT ISSUES IN LINGUISTIC THEORY

Advisory Editorial Board Raimo Anttila (Los Angeles); Lyle Campbell (Christchurch, N.Z.) Sheila Embleton (Toronto); John E. Joseph (Edinburgh) Manfred Krifka (Berlin); Hans-Heinrich Lieb (Berlin) E. Wyn Roberts (Vancouver, B.C.); Hans-Jürgen Sasse (Köln)

Volume 243

Hubert Cuyckens, Thomas Berg, René Dirven and Klaus-Uwe Panther (eds.) Motivation in Language Studies in honor of Günter Radden

MOTIVATION IN LANGUAGE STUDIES IN HONOR OF GÜNTER RADDEN

Edited by

HUBERT CUYCKENS University of Leuven

THOMAS BERG University of Hamburg

RENÉ DIRVEN University of Duisburg-Essen

KLAUS-UWE PANTHER University of Hamburg

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 Motivation in Language : Studies in honor of Günter Radden / Edited by Hubert Cuyckens... [et al.] p. cm. -- (Amsterdam studies in the theory and history of linguistic science. Series IV, Current issues in linguistic theory, ISSN 0304-0763 ; v. 243) Includes bibliographical references and index. 1. Linguistics. I. Radden, Günter. II. Cuyckens, H. III. Series. P26 .R24M68 2003 415--dc21 2003052107 ISBN 90 272 4755 2 (Eur.) / 1 58811 426 0 (US) (Hb; alk. paper) © 2003 – 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

CONTENTS

Editors' Foreword

vii

Acknowledgments

ix

In Search of Conceptual Structure: Five Milestones in the Work of Günter Radden René Dirven Section 1: Motivation in Lexico-Grammar

xiii 1

Extreme Subjectification: English Tense and Modals Ronald W. Langacker

3

Meaning and Context John R. Taylor

27

Lexical Rules vs. Constructions: A False Dichotomy William Croft

49

Schemas and Lexical Blends Suzanne Kemmer

69

Valency and Diathesis Heinz Vater

99

To Get or to Bel Use and Acquisition of Get- versus Be-Passives: Evidence from Children and Adults Kerstin Meints

123

Section 2: Motivation in the Lexicon

151

Space and Time in Polish: The Preposition za and the verbal Prefix zaElžbieta Tabakowska

153

VI

TABLE OF CONTENTS

Functions of the Preposition kuom in Dholuo Mechthild Reh

179

Grammaticalization of Postpositions in German Claudio di Meola

203

Metonymy in Cognitive Linguistics: An Analysis and a Few Modest Proposals Antonio Barcelona

223

Section 3: Motivation in Socio-Cultural Conceptualizations

257

How Language Is Conceptualized and Metaphorized in Japanese: An Essay in Linguistic Ideology Yosh ih iko Regarni

259

The Ever-Stifling Essentialism: Language and Conflict in Poland (1991-1993) Karol Janicki

273

Motion Metaphorized in the Economic Domain Olaf Jäkel

297

Section 4: Motivation in Applied Linguistics

319

English in the World and English in the School Willis Edmondson & Juliane House

321

Attitudes towards Luganda, Kiswahili, English, and Mother Tongue as Media of Instruction in Uganda Meike Sprenger-Tasch

1

Style Labels in Monolingual English Learner's Dictionaries Wolfgang Hünig

367

Name Index

391

Subject Index

397

EDITORS' FOREWORD

This volume contains a collection of papers in honor of Günter Radden, who is currently Professor Emeritus of English Linguistics at the University of Hamburg. The thematic perspective reflected by the title of the volume, Motivation in Language, has been chosen to reflect two interconnected aspects of Günter Radden's career as a linguist. First, as surveyed in the introduction "In Search of Conceptual Structure" (Dirven), Günter Radden has never ceased to be captivated by the conceptual intricacies of language. Throughout his career, he has not only shown a boundless enthusiasm for things linguistic himself, but he has also been able to instill this enthusiasm in students and colleagues. It is safe to say, then, that this double motivation for Günter Radden's scholarly pursuits resides in language. Second, as a psychologically highly-motivated linguist, Günter Radden has always looked for the conceptual factors underlying or motivating language use. Early on in his career, he focused on psychological factors motivating learners of English as a Second/ Foreign Language, and on societal motivational factors in sociolinguistic variation. Ever since the advent of Cognitive Linguistics (and even before), Günter Radden's research has been characterized by the belief that the link between the form and meaning of linguistic expressions is often not arbitrary, but cognitively motivated, thus reflecting his skepticism about the autonomy of linguistic form. Motivation in Language tries to do justice to these general two-pronged research interests. As such, the greater part of the papers (Sections 1-3) analyze various linguistic phenomena in which not arbitrary, but cognitively motivated links between an expression's form and its meaning play a role; a smaller set of papers in Section 4 (at least indirectly) point to the role learner and attitudinal motivation may play in applied linguistics domains. At the same time, due care has been taken to include papers on specific topics that are at the heart of Günter Radden's research interests (either because he has published on them himself or because he has taught courses on them). The papers in the first section "Motivation in Lexico-Grammar" present a discussion of some highly subjectified uses of tense and modal markers

Viii

EDITORS' FOREWORD

(Langacker), the interplay between a verb's meaning—including its participant roles—and the syntactic and semantic context in which the verb occurs (Taylor, Croft, Vater), the morphological device 'blending' in terms of the cognitive grammar notion 'schema' (Kemmer), and the acquisition of get- vs. be- passives within a prototype approach to the passive (Meints). The papers in the second section "Motivation in the Lexicon" study the various factors moti­ vating or underlying the range of (semantic) variants that attach to a particular lexical item. As such, the Polish perfectivizing prefix za- is shown to be systematically related to the goal/purpose preposition za (Tabakowska); the superficially heterogenous looking uses of the preposition kuom in Dholuo can be understood as deriving from one basic spatial image schema CONTACT-CUMSOLID (Reh); and the coexistence, in German, of lexical and grammatical vari­ ants of a single word form is accounted for in terms of grammaticalization (Di Meola). A final paper in this section reviews the cognitive linguistic approach to metonymy, which is a conceptual factor par excellence in motivating lexical change (Barcelona). The papers in the third section "Motivation in SocioCultural Conceptualizations" look at motivating factors in the conceptual content or meaning of linguistic items that can be situated in the socio-cultural domain (Ikegami on the concept of 'language', and Janicki on the concept of 'abortion'), or illustrate how (aspects of) one particular socio-cultural domain (the economic domain) can be understood in terms of the domain of motion. The final set of papers "Motivation in Applied Linguistics" honors Günter Radden's (mostly earlier) work in the domain of Applied Linguistics. Edmon­ son and House's paper examines, among other things, how the 'relevance' factor contributes to long-term language learning motivation; Sprenger-Tasch's paper looks at the attitudinal factors determining the choice of the language of instruction in the Ugandan school system; Hünig's paper, finally, presents a comparative study of the use of style labels in learner's dictionaries. A more detailed description of how the papers reflect the volume's title Motivation in Language can be found in the introductions to each of the sections.

Leuven, Hamburg, and Duisburg, May 2003 Hubert Cuyckens, Thomas Berg, René Dirven, and Klaus-Uwe Panther

ACKNOWLEDGMENTS

It is only befitting to express our gratitude to the many people who helped us realize this project. First of all, we would like to express our thanks to all the authors for their contributions, for their speediness at every stage of the reviewing and editorial process, and for their patience. We are also very grate­ ful to the many colleagues who acted as reviewers of the papers. We would also like to thank Anke de Looper of John Benjamins Publishing Company and Konrad Koerner, the General Editor of the Current Issues in Linguistic Theory series, for accepting this volume. Finally, a special word of thanks goes to Günter Radden for being the wonderful scholar and person he is, and thus for having, unwittingly, occasioned this volume.

IN SEARCH OF CONCEPTUAL STRUCTURE FIVE MILESTONES IN THE WORK OF GÜNTER RADDEN

RENÉ DIRVEN University of Duisburg-Essen

This survey can by no means offer a complete picture of Günter Radden's research, but it will highlight some of his most remarkable achievements in the pre-cognitive and cognitive climate of the past twenty-five some years. Everything that Günter Radden has approached, and not only his academic enterprises, has received his full devotion, energy, and love. His main academic point of attraction has always been the conceptual challenge, first in the conceptual dimension of semantic roles, and later in the full primacy of the conceptual over the formallinguistic, as sought for and realized in cognitive linguistics.

1.

Towards a cognitive sociolinguistics

Günter Radden's contribution to sociolinguistics is that he enriched the quantitative-statistical methods developed in the sixties and seventies with a qualitative conceptual analysis. This analysis employs case grammar categories or semantic roles, role configurations, and valence types. Importantly, it assumes that the class-specific, differentiated use of certain conceptual-gram­ matical categories such as Agent or Patient may point to the existence of substantial conceptual preferences in the speech of different social classes. Radden establishes that speakers of higher classes display a significantly higher number of the following construction elements: Patients in general, one-place predicates, adjectives used as predicates or as nominal modifiers, diversified semantic roles in subject functions, that/whether/if clauses, relative clauses, and passive constructions. Speakers of lower social classes, who, in contrast with middle class speakers, use a more informal type of language, use other construction elements: more Agent roles, typically associated with two-place predicates (or transitive verbs), less differentiated semantic roles in subject function, many more grammatically correct and complete sentences (which

XiV

RENÉ DIRVEN

may be due to the strong concentration on one type of semantic role), and many more phatic interpolations (you know, you see). These results lead Radden to the conclusion that higher class speakers manifest "a greater representational interest", and lower class speakers a "greater action interest" (Radden 1979:375; my translation, R.D.). The former group also exploits the thematic variation in subject position a great deal more, and thus, in cognitive linguistic terms, it could be hypothesized that they draw upon all possible event schemata (such as the ESSIVE schema, the PROCESS schema, the ACTION schema, the EXPERIENCING schema, the POSSESSION schema, the MOTION schema, the TRANSFER schema, and the LOCATION schema), whereas lower class speakers prefer the specification of concrete aspects of information such as place, time, etc. (Radden 1979: 376), and therefore draw more strongly upon the action and location schemata. In sum, then, the conceptual and perceptual orientations found with speakers of different social classes mark a clear cognitive differ­ entiation, which vindicates Bernstein's (1958) biased and much contested thesis of elaborated and restricted codes, which are mainly based on superficial formal criteria. From their presence or absence Bernstein concluded, taking a rather big leap, that the elaborated code is predominant in middle class children and the restricted code in working class children. In strong contrast with Bern­ stein's work, Günter Radden's first work is really conceptually oriented and can even be seen as a precursor of a cognitive sociolinguistics, in which socialclass differences become manifest in the preferential use of different concep­ tual categories.

2.

The conceptual elaboration of Case Grammar

Although Fillmore, the father of Case Grammar, had expressed his discour­ agement with his spiritual child in the late seventies (1987:104), several linguists kept making use of his insights. Some incorporated case roles into their overall grammar theory; e.g. theta roles in Chomsky (1981) or role archetypes such as agent, patient, instrument, experiencer, and mover in Langacker (1991:285). It is one of Günter Radden's most remarkable empirical accomplishments that he tested Fillmore's cases on a large data-set of spoken sentences and gradually elaborated the case grammar repertory in greater detail. Table 1 gives a survey of Radden's achievements in this respect. From this table, it can readily be seen that, aside from having first employed case grammar categories in a 'cognitive sociolinguistics' (cf. Section 1), Günter Radden has, over the years, made several contributions to the development of Case Grammar theory itself.

RADDEN'S SEARCH FOR CONCEPTUAL STRUCTURE

xv

Archetypal role Radden category (1979:336)

Radden & Dirven Radden (1989a: 421,1989b: 552) (1981:237)

Radden & Dirven (in preparation)

Human roles

Agent Experiencer Possessive Beneficiary Comitative

Agent Experiencer Recipient 560 Possessor Beneficiary Companionship

Agent Experiencer Receiver/Recipient Possessor Beneficiary

Cause Reason Instrumental Means Patient Area Partitive/Entitive

Cause Reason Instrument Means Patient Area: 558 Partitive

Cause/Natural Force Reason/Purpose Instrument Means Patient Area Wholes/Parts

Representative

Identification Class Change of state Place Goal+ Fig. 564 Source+Fig.564 Path Target Destination

Essive

Agent Dative

-

Causal roles

Cause Instrumental

Object roles

Patient Area Partitive

Essive roles

Spatial roles

Locative Goal Source Path Direction

Temporal roles

Time

Circumstantial roles

Manner Measure Result

Condition

Place Goal Source Path Target Direction Destination Time

Manner Circumstance Measure Result Purpose Modality Comparative Concessive

Physical space Place Close/Distant Goal Source Path Target Destination

Time point: 571 Time duration: 572 Relational time: 575 Temporal space Point/Stretch/Span Duration/Time Path Manner Accompaniment Measure Result Purpose Consequence Modal Comparison Concession

Table 1 : Radden 's specification of semantic roles

Abstract space/settings Circumstantial sur-rounding:/ se­ quence/connection /accompaniment Causal/conditional /concessice /contrastive settings

XVi

RENÉ DIRVEN

The first column presents the general types of semantic roles—setting off human roles from all other roles—according to which the many detailed case categories can be grouped. The second column offers a further specification of the seven archetypal categories of roles in the first column: Agent, Dative, Cause, Patient, Locative, Time, and Circumstance. Alongside the object role Patient, Radden (1979) also introduces Area and Partitive; the spatial role Locative is broken up into Locative, Goal, Source, and Path, and is comple­ mented by Direction; finally, Circumstance is further specified as Manner, Measure, Result, and Condition. The third column offers a selection from a wider collection of case categories in Radden and Dirven (1981:237) used by the editors in their own contributions. The new list expands Dative into Experiencer, Possessive, and Beneficiary, integrates the many diverse semantic relations in figuratively used prepositional phrases, and introduces Comitative as in agree with; it adds Reason to Cause and Means to Instrument, and intro­ duces the object role Entitive (the opposite of Partitive) for the leg of the table; furthermore, it brings in the essive role Representative (as in regard as a loss), it adds the spatial roles Target (shoot at) and Destination (leave for London), and introduces Modality, Comparative, and Concessive as circumstantial roles. By this time, the set of semantic roles has doubled from fifteen to thirty, whereby the 'new' roles are in fact subcategories of the archetypal categories of roles. The fourth column is taken from Radden (1989a, 1989b), which adds the role Recipient and expands the essive role Representative into Identification, Class, and Change of State, the temporal role Time into Time Point (at three), Time Duration (during three hours), and Relational Time (three hours ago). As these last examples show, the analysis of all possible prepositions in their literal and figurative uses always increases the number of semantic relations. The set of labels in the last column has not been finalized yet: it reflects the presently growing distinction between predicate linked semantic roles and sentence-nucleus linked settings (Radden & Dirven, in prep.), whereby tem­ poral and circumstantial categories are seen as metaphorical mappings of physical space concepts into temporal and abstract space. 3.

Rescued from the wastebasket: The role of Area

One of the strengths of Günter Radden's work on Case Grammar is probably his data-driven approach. In working through these data, and struggling with many single problem cases, he must no doubt have been tempted, like many other case grammar workers, to throw the unanalyzable examples in the wastebasket. However, his persistence has certainly paid off: especially his system-

RADDEN'S SEARCH FOR CONCEPTUAL STRUCTURE

xvii

atic concentration on prepositional phrases (Radden 1981, 1985, 1989b, 1997, 1998a) has opened all kinds of new insights. One important new insight is his distinction between 'Immediate and ultimate recipients' (Radden 1998b). Another is the role of 'Area', which was introduced under the original title "Can Area Be Taken out of the Waste-basket?" (Radden 1978). In Radden's view, the notion of Area, as in He told me about his secret, is related to that of Patient, as in He told me his secret, but differs fundamentally from it, in that the predicate's action does not 'affect' the participant in the Area role. As such, from He told me about his secret, which includes the Area phrase about his secret, it may be inferred that he told me how difficult it was to live with his secret or how he had come to have it, but it does not necessarily mean that he revealed his secret. Radden's insights into the Area role show gradual refinement. His first definition of Area runs as follows: "Obviously, these preposition phrases ... express the overall theme or area within the frame of which certain actions, states, or events are located" (Radden 1978:328). Clearly, by referring to the three categories 'actions, states and events', Radden still limits the discussion to verbs only. These 'area verbs' comprise verbs of knowing and verbs of communicating. Importantly, he also tries to distinguish between the Patient role (as in ?Forget cognitive semantics) and the Area role (as in Forget about cognitive semantics) in terms of the scope of the field denoted by the Area participant. As such, the paradigm of Cognitive Semantics is a wide field which can serve as Area, but it is too wide a field to become a Patient with forget. In comparing forget with teach, Radden finds that the verb itself imposes conceptual constraints, too. With teach wide fields such as 'music', 'phonetics', or cognitive semantics' are manageable and can therefore assume the Patient role (indeed, normally one teaches something—e.g. music/ phonetics/cognitive semantics—not about something). This difference between wider and narrower fields is also felt in compounds with teacher, where one of the composite expressions can only denote a wider field: the music/ phonetics/cognitive semantics teacher, and not a smaller topic, as shown by the unacceptable form *the gerund teacher. In his 1981 paper on the figurative use of prepositions, Radden diver­ sifies the various subtypes of Area. In particular, he shows that (i) the prepo­ sition about denotes 'motion in any possible direction' and as such is the prototypical candidate for the general notion of Area (e.g. talk about); (ii) the prepositions at, on, in all designate more specific Area notions: a 'focused Area' {be good at something), a 'specialization Area' (to lecture on), and an 'encompassing Area' (be good in classics); (iii) the preposition of (weakened

XViii

RENÉ DIRVEN

from off) signals an 'origin Area': when one speaks or talks of something, it is suggested that one has only limited information—this difference is also mani­ fest in think of versus think about. Nearly a decade later, Radden's view of Area has, on the one hand, been further specified, and, on the other hand, it has been widened to encompass various syntactic categories other than prepositional phrases (Radden 1989b:571). For one, the preposition about denoting 'general' Area (e.g. in talk about) now also includes about signaling Area 'causing5 emotional reactions (in emotive predicates such as be angry about) and Area 'of intent' (e.g. in be serious about). The Area preposition in—in comparison with about far less prototypical—is now specified as: "The spatial sense of enclosure gives rise to figurative meanings which relate to some kind of delimitation in some area or other... or to a state or circumstance in which an entity is seemingly contained (Radden 1989b:558). Some examples are: (i) general Area {reasonable in price), (ii) material Area (abound in fish), (iii) partial Area (I hit him in the face), and (iv) Area of activity (take part in, specialize in). As well, when widening the syntactic categories in which Area can be realized, Radden showed a very great sense for conceptual similarities, with the Area role now also being spotted (Radden 1989b) in idiomatic adverbial phrases such as as far as X is concerned, as for X, re X, concerning X) and in derived adjectives (an economic adviser, a religious fanatic). 4.

Cognitive linguistics in a first synthesis

Günter Radden's 1992 article "The cognitive approach to natural language" was for a decade probably his best known and certainly his most quoted paper. In this paper, he synthesizes the inroads Cognitive Linguistics had made by the early nineties into the tranquil waters of century-old objectivist assumptions in linguistics and philosophy. Radden puts forward five innovative lines of thought which together constitute a revolutionary break with traditional thought: (i) iconicity, (ii) categorization, (iii) metaphor, (iv) cultural models, and (v) grammar as a conceptual system. (i) Iconicity. The principle of iconicity (Haiman 1985; Givón 1985) stipulates that "our conception of reality is mirrored in the structure of language" (Radden 1992:514) and is thus indicative of the primacy of the perceptual over the linguistic. This manifests itself in the three iconic principles of quantity (more sound, or more form in general, is more meaning), of prox­ imity (what belongs together conceptually is put together linguistically), and of sequential order (linguistic order reflects the temporal order of events). These

RADDEN'S SEARCH FOR CONCEPTUAL STRUCTURE

XiX

findings are also a breakthrough relativizing the Saussurean view that language is almost purely symbolic, and hence also predominantly arbitrary. (ii) Categorization. The importance of Rosch's insights (1977, 1978), according to Radden, is that she put an end to the unshaken, Aristotelian belief in fixed categories based on the 'essential' features of things in the world, thereby revolutionizing our Western views on categorization. Similarly, in her psycholinguistic experiments, Aitchinson (1987:44) clearly illustrated how categories are not fixed, but may show great individual differences during a person's development. (iii) Metaphor. In Radden's view, Lakoff and Johnson's (1980) brilliant new insight was that the mind itself operates in terms of conceptual metaphors, which in turn may underlie all kinds of linguistic metaphors. So metaphor is not in the first place a matter of words, of language, but a matter of mental images and of thought. In creating new metaphors, the human conceptualizer builds new concepts, and ultimately new meanings for words. The mapping of one domain of experience, e.g. warfare, to another domain, e.g. conversation or arguing, is a conceptual process allowing the mind to build up ever more abstract relations. Günter Radden was not only well-read in the conceptual theory of metaphor, but through a number of sabbaticals at the University of California, he also had first-hand access to the newest developments in meta­ phor theory, some of which are reflected in his 1992 survey article. As such, for instance, he was able to attend Lakoff s lecture on "Metaphor and Metonymy" in San Diego in 1990, where Lakoff posited that mappings between domains in specific-level structural metaphors (such as ARGUMENT IS WAR) tend to be culture-specific, whereas mappings between domains in generic-level metaphors (such as ACTIONS ARE SELF-PROPELLED MOTIONS) may be found universally (cf. Radden 1992: 525, 537 fn.23). (iv) Cultural models. As the preceding findings suggest, man's linguistic knowledge is intertwined with his encyclopedic knowledge. Since such a large segment of encyclopedic knowledge is culture-specific, we need 'cultural models' for the interpretation of our linguistic categories. Thus, according to Radden, the linguistic item Friday may receive different interpretations depending on the cultural model it is part of: as the day between Thursday and Saturday in a purely 'calendric' cultural model, as the day of Christ's suffering and death in the Christian cultural model, as the day of bad luck and accidents in a cultural model of superstition, and as the last day of the working week and the transition to a long weekend in the cultural model of organized labor. In the same vein, the cultural anthropologist D'Andrade (1989) showed that students

RENE DIRVEN

XX

performed significantly better on syllogism tests involving culturally realistic content (or cultural models) than arbitrary material. (v) Grammar as a conceptual system. Radden describes the important achievement of cognitive linguists such as Talmy, Sweetser, Langacker as having shown that grammar operates as a conceptual organizing system. Talmy (1985) discovered that we use 'force dynamic schemata' in our understanding of human interaction, which we understand as the interaction of physical forces. Sweetser (1990) explained the two types of epistemic and deontic modality in these terms of force dynamics. Talmy (1988) and Langacker (1987, 1988) applied the organizing principles of perception to the grammar of a language. As such, each sentence, be it simple or complex, is organized in terms of the gestalt perception principle of figure-ground alignment. And just as an observer looks at a scene from a certain vantage point, a speaker may construe a situation from a certain perspective; a typical instance is mental scanning, whereby the same hill can be seen as gently rising from a river or as gently falling to the bank of a river. Radden's own most innovative contribution to the new cognitive para­ digm is his theory of metonymy. 5.

It is all metonymy

The past few years have seen a very strong current of research on metonymy: Croft (1993), Dirven (1993), Thornburg and Panther (1997), Kövecses and Radden (1998), Radden and Kövecses (1999), the collective volumes of Panther and Radden (1999) and Barcelona (2000), and Radden (2000). Espe­ cially, Radden and Kövecses's (1999) paper "Towards a theory of metonymy" has become a trendsetter, in which Radden pursues his innovative intuition that, rather than metaphor, "it is all metonymy". In this paper, Radden exam­ ines three basic aspects of metonymy: (i) the various ontological realms in which metonymy operates; (ii) the basic relationships that are possible between the two elements of a metonymy, i.e. the vehicle (= actual linguistic expres­ sion) and the target (= intended concept or referent); and (iii) the cognitive principles motivating the choice of one vehicle over another. An ontological realm in which metonymy operates, almost invisibly, is semiotics itself. The structure of the sign is itself metonymic, as the sign's form stands for the sign's meaning or the concept associated with it. Furthermore, the function of the sign is metonymic: although, in actual fact, the sign denotes a conceptual category, i.e. a representation of things or events of the world in the speaker's mind, speakers use signs, i.e. words, to refer to reality itself. In

RADDEN'S SEARCH FOR CONCEPTUAL STRUCTURE

xxi

addition to describing these two semiotic types of metonymy, i.e. 'sign metonymy' and 'reference metonymy', Radden characterizes our traditional understanding of metonymy as 'concept metonymy': a given sign is used, not to denote its own concept, but to denote another concept, closely linked to the first; put differently, concept metonymy may denote a concept that can be expressed by another sign (as in The busses are on strike, where BUSSES stands for BUS-DRIVERS), or it may denote a concept for which no sign exists (e.g. the more general concept MOTHER may be, and usually is, used to stand for a stereotypical and prototypical subtype of this category, i.e. THE HOUSEWIFE MOTHER. Generalizing, Radden concludes that even a concept's structure is metonymic, since the concept usually stands for its most prototypical member. As for the relationships between the elements in a metonymy, Radden distinguishes two basic types: part-whole relationships (or vice versa) or partpart relationships. The part-whole or whole-part configuration typically occurs with things, the part-part configuration with events or states. Since a thing has internal structure, the whole can conceptually be selected for any of its parts, or any relevant part for the whole. This relationship not only applies to concrete objects, but also to mental entities such as present time, which may not only stand for past and future time, but also for genericness (She speaks Spanish), and for potentiality {She is an angry person in the sense of 'She can be angry'). The part-part configuration typically occurs with events or states, which conceptually consist of a relation and one or more participants. Frequently occurring relations in metonymies are action, perception, causation, possession, containment, and location. Thus, in the relation of causation, the cause (good health) stands for the visible effect in the expression a healthy face. Radden notices a large number of cognitive principles triggering the choice of a metonymy's vehicle, which he groups under three main factors. The human experience factor makes us select human over non-human vehicles, subjective over objective ones, concrete over abstract ones, etc. The perceptual factor favors the selection of, for instance, the immediate over the non-imme­ diate, the good gestalt over the poor gestalt, the bounded over the unbounded, etc. Cultural factors favor stereotypical vehicles over non-stereotypical ones, initial or final over middle ones, basic over non-basic ones, etc. In addition to these cognitive principles, communicative principles such as the principles of clarity and of relevance may also influence the choice of the vehicle. These cognitive and communicative factors and motivations may also be in compe­ tition; for instance, the selection of the non-human vehicle busses (standing for the bus-drivers in The busses are on strike) reflects that people interact more relevantly with the mode of transport than with the bus-drivers. Finally, there

XXii

RENÉ DIRVEN

are often overriding factors such as euphemism which cancel a cognitive or communicative principle: thus go to the bathroom is less clear than urinate, but the principle of clarity submits to that of politeness. Radden (2000) extends the scope of metonymy to its interaction with metaphor and elaborates Taylor's (1989) and Goossens's (1990) thesis that a number of metaphors are based on metonymy. In Radden's view, there are three bases common to metonymy and metaphor, (i) i.e. common experience, (ii) implicature, and (iii) cultural models. First, an experiential basis common to metaphor and metonymy may be provided by phenomena in correlation, in complementarity, or by phenomena viewed in comparison. Two phenomena in correlation are, for instance, the simultaneously increasing height and quantity of a liquid as the liquid is poured into a container. When we (physically) observe these two situations, we can literally 'see' a link between the two. When we merely 'imagine' this correlation, e.g. when we 'fill up' the petrol tank of our car, we have a metonymic link. When we speak of 'rising' or 'high prices', we have a metaphoric link based on metonymy. And in 'sky-rocketing' or 'soaring prices', we have a pure metaphor in which the underlying concep­ tual metaphor could be PRICES ARE FLYING OBJECTS. The theory of a metonymy-metaphor continuum assumes that the two conceptual processes are not opposed, but that they constitute two extremes on a continuum of concep­ tual blending, ranging from mere correlation, via metonymy and metonymybased metaphor to pure metaphor. Two phenomena in a complementary relation at the conceptual level are, e.g., husband and wife; this relationship constitutes the basis for metaphors such as that of partners 'bound together' or 'tied to each other' in marriage. Similar pairs are 'mind and body' or 'body and soul'. The comparison of similar elements leads to conceptual metaphors such as: SIMILARITY IS CLOSENESS (i.e. the elements are experienced together) and DIFFERENCE IS REMOTENESS.

Two phenomena/elements may not only have a common experiential basis, they may also be linked purely mentally by the process of implicature, a typical instance being the case of sequential events (what comes first tends to be seen as the cause) and resultative events (what you have 'taken', becomes your possession). As well, the naming of a place may carry the implicature of the standard activity associated with that place, e.g. Children are going to the playground is interpreted as 'Children are going to play'. This metonymic link of PLACE FOR ACTIVITY may become the input for a metonymy-based metaphor, such as PURPOSES ARE DESTINATIONS, as in We have reached the end to be interpreted against literal destinations as in We have reached the station.

RADDEN'S SEARCH FOR CONCEPTUAL STRUCTURE

xxiii

Finally, cultural models constituting the metonymic basis for metaphors are found, for instance, in our Western folk theories of motion: we see an 'internal force' or impetus in a moving object as its propelling cause, which leads to the conceptual metaphor FORCE IS A SUBSTANCE CONTAINED IN AFFECTED CAUSES. Furthermore, the folk theory of communication, which is summarized in Reddy's (1979 [1993]) CONDUIT metaphor, conceptualizes communication as the transmission of information through a channel, and meaning as something which is to be taken out of the word containers. The container image schema also strives to conceptualize both the mind, which is a container for ideas, and the body, which is a container for emotions. Given this huge body of metonymy-based metaphors (in fact, especially conceptual metaphors), it is very likely that many more metonymy-based metaphors will be found. In this respect, Radden's first findings have made major inroads into a further and deeper understanding of conceptual structure. The best tribute the linguistic community can make to Günter Radden is to respond massively to his invitation to delve deeper in this huge conceptual quarry. References Aitchison, Jean. 1987. Words in the Mind: An Introduction to the Mental Lexicon. London: Basil Blackwell. Barcelona Sánchez, Antonio, ed. 2000. Metaphor and Metonymy at the Cross­ roads. (= Topics in English Linguistics, 30). Berlin: Mouton de Gruyter. Bernstein, Basil. 1958. "Some Sociological Determinants of Perception: An Inquiry into Subcultural Differences". British Journal of Sociology 9.159-174. (Reprinted in Basil Bernstein. 1971. Class, Codes and Control, Vol. 1, Theoretical Studies Towards a Sociology of Language, 23-41. London: Routledge and Kegan Paul,. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris. Croft, William. 1993. "The Role of Domains in the Interpretation of Metaphors and Metonymies". Cognitive Linguistics 4.335-370. (Reprinted in René Dirven & Ralf Pörings, eds. 2002. Metaphor and Metonymy in Com­ parison and Contrast, 161-204. Berlin: Mouton de Gruyter.) D'Andrade, Roy. 1989. "Culturally Based Reasoning". Cognition and Social Worlds ed. by A. Gellatly & D. Rogers, 132-143. Oxford: Oxford University Press.

XXiV

RENE DIRVEN

Dirven, René. 1993. "Metonymy and Metaphor: Different Mental Strategies of Conceptualisation". Leuvense Bijdragen 82.1-28. (Strongly revised version published in René Dirven & Ralf Pörings, eds. 2002. Metaphor and Metonymy in Comparison and Contrast, 75-111. Berlin: Mouton de Gruyter.) Fillmore, Charles J. 1987. "Schemata and Prototypes". Fillmore's Case Grammar: A Reader ed. by René Dirven & Günter Radden, 99-105. Heidelberg: Julius Groos. Givón, Talmy. 1985. "Iconicity, Isomorphism and Non-Arbitrary Coding in Syntax". Iconicity in Syntax ed. by John Haiman, 187-219. Amsterdam: John Benjamins. Goossens, Louis. 1990. "Metaphtonymy: The Interaction of Metaphor and Metonymy in Expressions for Linguistic Action". Cognitive Linguistics 1.323-340. (Reprinted in René Dirven & Ralf Pörings, eds. 2002. Meta­ phor and Metonymy in Comparison and Contrast, 349-377. Berlin: Mouton de Gruyter.) Haiman, John. 1985. Natural Syntax: Iconicity and Erosion. Cambridge: Cambridge University Press. Kövecses, Zoltán & Günter Radden. 1998. "Metonymy: Developing a Cogni­ tive Linguistic View". Cognitive Linguistics 9.37-77. Lakoff, George & Mark Johnson. 1980. Metaphors We Live By. Chicago, I11.: The University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1, Theoretical Prerequisites. Stanford, Calif.: Stanford University Press. . 1988. "A View of Linguistic Semantics". Topics in Cognitive Linguistics ed. by Brygida Rudzka-Ostyn, 49-90. Amsterdam: John Benjamins. . 1991. Foundations of Cognitive Grammar, Vol. 2, Descriptive Appli­ cation. Stanford, Calif.: Stanford University Press. Panther, Klaus & Günter Radden, eds. 1999. Metonymy in Language and Thought. Amsterdam: John Benjamins Radden, Günter. 1978. "Can 'Area' Be Taken out of the Waste-Basket?" Valence, Semantic Case and Grammatical Relations ed. by Werner Abraham, 327-338. Amsterdam: John Benjamins. 1979. Ein Profil soziolinguistischer Variation in einer amerikanischen Kleinstadt. Frankfurt a.M.: Peter Lang. . 1981. "Die übertragenen Bedeutungen der englischen Raumpräposi­ tionen". Radden & Dirven 1981. 133-179.

RADDEN'S SEARCH FOR CONCEPTUAL STRUCTURE

xxv

— . 1985. "Spatial Metaphors Underlying Prepositions of Causality". The Ubiquity of Metaphor: Metaphor in Language and Thought ed. by Wolf Paprotté & René Dirven, 177-207. Amsterdam: John Benjamins. — . 1989a. "Semantic Roles". A User's Grammar of English: Word, Sentence, Text, Interaction ed. by René Dirven, 421-472. Frankfurt a.M.: Peter Lang. .1989b. Figurative Use of Prepositions. A User's Grammar of English: Word, Sentence, Text, Interaction ed. by René Dirven, 551-576. Frankfurt a.M.: Peter Lang. —. 1992. "The Cognitive Approach to Natural Language". Thirty Years of Linguistic Evolution ed. by Martin Pütz, 513-541. Amsterdam: John Benjamins. . 1996. "Motion Metaphorized: The Case of 'Coming' and 'Going'". Cognitive Linguistics in the Redwoods: The Expansion o f a New Para­ digm in Linguistics ed. by Eugene H. Casad, 423-458. Berlin: Mouton de Gruyter. — . 1997. "Time Is Space". Human Contact through Language and Linguistics ed. by Birgit Smieja & Meike Tasch, 147-166. Frankfurt a.M.: Peter Lang. . 1998a. "The Conceptualisation of Emotional Causality by means of Prepositional Phrases". Speaking of Emotions: Conceptualisation and Expression ed. by Angeliki Athanasiadou & Elžbieta Tabakowska, 273294. Berlin: Mouton de Gruyter. . 1998b. "Immediate and Ultimate Recipients". The Diversity of Linguistic Description: Studies in Honour of Béla Korponay ed. by Joszef Andor, 265-278. Debrecen: Angol-Amerikai Intézete. . 2000. "How Metonymic are Metaphors?". Barcelona 2000. 93-108. (Strongly revised version published in René Dirven & Ralf Pörings, eds. 2002. Metaphor and Metonymy in Comparison and Contrast, 407-434. Berlin: Mouton de Gruyter.) & René Dirven, eds. 1981. Kasusgrammatik und Fremdsprachendidaktik. (= Anglistik und Englischunterricht, 14). Trier: Wissenschaftlicher Verlag. — - & René Dirven. In preparation. Cognitive English Grammar. Amster­ dam : John enjamins. and Zoltan Kövecses. 1999. "Towards a Theory of Metonymy". Panther & Radden 1999. 17-59.

XXVi

RENÉ DIRVEN

Reddy, Michael J. 1979. "The Conduit Metaphor: A Case of Frame Conflict in Our Language about Language". Metaphor and Thought ed. by Andrew Ortony, 284-324. Cambridge: Cambridge University Press. Rosch, Eleanor. 1977. "Human Categorization". Studies in Cross-Cultural Psychology ed. by Neil Warren, Vol. I, 1-49. London: Academic Press. . 1978. "Principles of Categorization". Cognition and Categorization ed. by Eleanor H. Rosch & Barbara B. Lloyd, 27-48. Hillsdale, N.J.: Lawrence Erlbaum. Sweetser, Eve E. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge Univer­ sity Press. Talmy, Leonard. 1985. "Force Dynamics in Language and Thought". Papers from the Twenty-First Regional Meeting of the Chicago Linguistic Soci­ ety, 293-337. Chicago, I11.: Chicago Linguistic Society. . 1988. "The Relation of Grammar to Cognition". Topics in Cognitive Linguistics ed. by Brygida Rudzka-Ostyn, 165-205. Amsterdam: John Benjamins. Taylor, John R. 1989 [1995, 2nd revised edition; 2003, 3rd revised edition]. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press. (Chapter 7 reprinted in René Dirven & Ralf Pörings, eds. 2002. Metaphor and Metonymy in Comparison and Contrast, 323347. Berlin: Mouton de Gruyter.) Thornburg, Linda & Klaus-Uwe Panther. 1997. "Speech Act Metonymies". Discourse and Perspective in Cognitive Linguistics ed. by Wolf-Andreas Liebert, Gisela Redeker & Linda Waugh, 205-219. Amsterdam: John Benjamins.

SECTION 1 MOTIVATION IN LEXICO-GRAMMAR The six papers in this first section examine selected aspects of the motivated, or non-arbitrary character of linguistic units in the continuum between lexicon and grammar. Ron Langacker's paper discusses the highly subjectified nature of English tense and modal markers. As subjectively construed grounding predications, tense and modal markers reflect language users' underlying cognitive ability to take a particular perspective on a scene. The markers specifically examined in this paper (present-time epistemic uses and uses of the present tense that do not refer to present time) are further motivated by the fact that they constitute a development (or extension) from the more basic modal and tense uses, in that they eliminate any direct concern with the occurrence of actual events. John Taylor argues that there exists a close association between linguistic (syntactic and lexical) context and word meaning, requiring us to explicate a word's meaning by reference to the linguistic contexts in which the word is used, and thus excluding a view of words as isolated, arbitrary formmeaning units. Addressing a similar concern, Bill Croft deals with the question whether to derive the meaning of a verb occurring in a particular construction (e.g. bake in the ditransitive construction bake someone a cake) by lexical rule or to have the verb's meaning fall out from the semantics of the construction. He shows, on the basis of the ditransitive construction, that both analyses are partly right. Further, it is his contention that the verbs or verb classes occurring in the ditransitive construction do not constitute an arbitrary set, but are pecu­ liar to that construction and are motivated through family resemblance to the meaning of other verb classes found in the construction. Suzanne Kemmer discusses the morphological device 'blending' in terms of the independently motivated and cognitively realistic notion of 'schema'. The non-arbitrary char­ acter of blends derives from the fact that, while they are difficult to incorporate in 'building-block' style theories of morphology, they are, as complex forms, to a large extent transparent because the combination of their constituent parts (in themselves arbitrary) leads to a transparent meaning. Heinz Vater's paper, which, unlike the other papers, is not to be situtated in the cognitive linguistic

2

SECTION 1: MOTIVATION IN LEXICO-GRAMMAR

framework, shares Taylor's and Croft's interest in the interplay between a verb's meaning—including its participant roles—and the syntactic and seman­ tic context in which the verb occurs. In particular, Vater suggests differen­ tiating between a verb's valency potential in the lexicon and its various valency realizations, which are conditioned, or motivated, by the syntactic environ­ ments in which the verb occurs. Finally, Kerstin Meints's paper discusses the use and acquisition of get- vs. -passives within a prototype framework. Meints considers get- and be-passives as interrelated members of the radial category 'passive' displaying different degrees of prototypicality. Significantly, the prototype approach predicts children's acquisition of be-passives, with getpassives being more peripheral members.

EXTREME SUBJECTIFICATION ENGLISH TENSE AND MODALS

RONALD W. LANGACKER University of California, San Diego

English tense and modals are grammaticized grounding elements, in which the ground—the speech event and its participants—are 'offstage' and subjectively construed. In each case, particular uses pose special problems of description and analysis: present-time epistemic uses of the modals, and myriad uses of the present tense that do not refer to present time. Characterizations are proposed in which these uses carry the subjectification process one step further, affecting the status of the grounded entity.

1.

Introduction

One of the more subtle notions of Cognitive Grammar, and apparently one of the most difficult to grasp and apply, is that of subjectivity (Langacker 1985, 1987a: section 3.3.2.4, 1990, 1993, 1997a, 1998, 1999a; Carey 1994). It is, however, essential to the proper characterization of semantic and grammatical structure. For example, it is pivotal to the description of grounding elements, the grammaticized means of relating nominal and clausal referents to the speech situation. In the case of English finite clauses, the grounding elements are tense and the modals. While these have been analyzed from the standpoint of Cognitive Grammar (notably in Langacker 1991: chap. 6), their preliminary descriptions naturally leave a lot of room for elaboration and refinement. I will be concerned here with one facet of their characterization that bears on the nature and the possible extent of subjectification. 2.

Grounding and subjectivity

In this context, the terms subjective and objective pertain to the roles of subject and object of conception. An entity is said to be objectively construed to the extent that it is put 'onstage' as an explicit, focused target of conception.

4

RONALD W. LANGACKER

Conversely, it is said to be subjectively construed to the extent that it remains implicit as an aspect of the conceptualizer or the conceptualizing activity; an entity construed with maximal subjectivity is wholly 'offstage', having only a tacit presence at the locus of consciousness. Let us take a visual example, on the assumption that conception and perception are analogous in this respect (Langacker 1995b; Talmy 1996). Suppose I am looking at something, e.g. a bird perched on a branch, from a moderate distance. If I see the bird in sharp detail, and it stands out clearly against the background, I am construing it visu­ ally with a high degree of objectivity (it is optimal as an object of perception). In this same situation, on the other hand, my own eyes are construed with full subjectivity, since they are part of the perceptual apparatus, at the locus of perception, but are not themselves perceived. If I should examine my eyes in the mirror, however, they would have a dual role, as both the subject and object of perception. My eyes themselves would then be construed subjectively, and their image objectively. It makes no sense to say that linguistic expressions or their meanings as a whole are subjectively or objectively construed, but only specific entities within the overall conception evoked. Consider the subtle difference between the woman next door to us and the woman next door. There is certainly a contrast in subjectivity, since the former explicitly mentions a group including the speaker (us) and puts it onstage as an object of conception, whereas the latter leaves the group implicit as part of the offstage locus of viewing. Yet this contrast pertains to just one facet of the overall situation evoked by the two expressions. Both expressions construe the woman with a high degree of objectivity, as their profile, i.e. the entity they specifically refer to within the overall conceptual base. Via the phrase next door, both expressions evoke a domicile, but it remains implicit and at the margins of awareness; as neither the conceptualizer nor the focused object of conception, it is intermediate in regard to subjectivity/objectivity. Moreover, in their role as conceptualizers appre­ hending the meanings of the two expressions, the speaker and addressee are tacitly construed with full subjectivity (even when they also have an objective role and in that guise are coded by us). The term ground refers to the speech event, its participants, and its immediate circumstances. Its main participants, the speaker and hearer, are the primary conceptualizers with respect to the meanings of linguistic expressions. In this role, they function as implicit, offstage loci of consciousness construed with maximal subjectivity even though, in principle, they can be regarded as peripheral facets of every expression's conceptual import. The speech event participants, as well as other ground elements, can also be invoked with

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS varying degrees of salience as objects of conception, and thus construed with varying degrees of objectivity. Consider the different types of imperatives, as in(l): (1) a. Shred those documents right now! b. You shred those documents right now!  I order you to shred those documents right now! Here, in addition to being conceptualizers, the speech event participants are engaged in a force-dynamic communicative interaction wherein the speaker intends to get the hearer to carry out a certain action. Depending on such factors as the degree of cooperation expected, and hence the amount that needs to be put onstage for focused awareness, different portions of the 'interactive frame' are made explicit and objectively construed. In (la), only the envisaged action is spelled out in this way. In (lb), which tends to be viewed as more forceful, the hearer is overtly singled out by a use of the pronoun you, which I suspect combines the functions of subject, topic, and vocative (cf. Lambrecht 1996). Finally, in (1c), the interaction itself is put onstage as the focused object of conception, specifically as the process profiled by the main clause. A type of semantic change reasonably called subjectification (though not in the same sense that Traugott (e.g. 1989) uses the term) appears to be quite common. As I incline to analyze it (see especially 1999a), subjectification is the loss of an objectively construed entity, leaving behind a subjective coun­ terpart consisting in certain cognitive operations inherent in its conception. A case in point is the phenomenon variously known as 'subjective', 'fictive', or 'virtual' motion (Langacker 1986, 1999b; Matsumoto 1996a, 1996b, 1997; Talmy 1996). The examples in (2) illustrate subjectification involving the verb run (as well as the entire clause-level construction X run along Yfor Z). (2) a. The triathletes ran along the coast for several miles. b. This freeway runs along the coast for several miles. c. These mountains run along the coast for several miles. Sentence (2a) profiles an objectively construed event of running on the part of the subject. The triathletes are conceived as moving rapidly along a spatial path, and thus as occupying a continuous series of distinct spatial locations as the process unfolds. It should be apparent that the conceptualizer, in appre­ hending the meaning of (2a), has to successively invoke this series of locations in order to conceive of the mover progressing through them. In so doing, the

5

6

RONALD W. LANGACKER

conceptualizer effectively scans mentally along the same path that the subject traverses physically. This cognitive operation of scanning mentally along a spatial path is inherent in the conception of physically moving along it. The conceptualizer's mental scanning is a subjective counterpart of the subject's objectively construed motion. Sentences (2b-c) profile static configurations that continue through time unchanged (this makes them imperfective, which enables them to take the present tense, as discussed below). Their evident dynamicity, the intuitive sense of movement they impart, does not reside at the level of the objective situation being described. Rather, it is a matter of how the conceptualizer mentally accesses the profiled configuration in 'building up' to its full conception. The same mental scanning used to conceptualize the subject's objective path of motion in (2a) is instead used to dynamically evoke the subject's static spatial extension. All that remains of the subject's objectively construed motion is mental scanning along the same spatial path on the part of the conceptualizer, who thus moves subjectively along it. The conceptualizer's movement is subjectively construed because it inheres in the conceptualizing activity but is not itself a target of conception. In apprehending a sentence like (2c), the speaker does not think of himself as moving along a path—the speaker's conception is merely that of the mountains having a certain spatial alignment. The semantic contrast between run (or the whole construction) in (2a) and (2c) is thus a case of subjectification, as previously characterized: the loss of an objectively construed entity (in this case movement through space) leav­ ing behind a subjective counterpart consisting in certain cognitive operations (mental scanning through space) inherent in its conception. This loss can be a matter of degree, and need not occur in a single step. For instance, (2b) is intermediate between the fully objective construal of the spatial motion in (2a) and its subjective construal in (2c). Like the latter, (2b) profiles a stable situa­ tion in which the focused participant (the freeway) does not move. Yet the notion of a freeway tends strongly to evoke the conception of travelers moving along it. To varying degrees, therefore, an imagined or generalized mover might figure in the understanding of (2b) on particular occasions. To the extent that the mover and the motion remain implicit, fictive, and unindividuated, their construal is largely subjective. Pure subjective motion, in which there is no longer any vestige of physical motion and only mental scanning remains, represents the limiting case where subjectification has fully run its course. Let us turn now to nominal and clausal grounding. I use the term quite narrowly, for grammaticized means of specifying the status of nominal and

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

7

clausal referents vis-à-vis the ground. A grounding element specifies the rela­ tionship that the thing profiled by a noun phrase, or the process profiled by a finite clause, bears to the ground with respect to fundamental notions such as time, reality, and referential identity. In the case of English, nominal grounding elements include the articles, demonstratives, and certain quantifiers. The grounding of English finite clauses is effected by tense and the modals. Pivotal to the analysis of these grounding elements is the claim that they construe the grounded thing or process with a high degree of objectivity, while the ground itself is construed with a high degree of subjectivity. The grounded entity is put onstage as the specific focus of attention, i.e. it is profiled, by both the grounding element and the noun phrase or finite clause as a whole. The ground, on the other hand, is offstage and non-salient despite its essential role in the grounding predication. It is implicit and subjectively construed, serving almost literally as the locus of 'viewing', from which the speaker and hearer direct their attention to the profiled entity onstage. Although the distinctive semantic value of a grounding element resides in the profile's status and loca­ tion with respect to the ground, this grounding relationship is not per se the focused object of conception. Rather, it is inherent in the 'viewing arrange­ ment' in terms of which the speaker and hearer conceptualize the nominal or clausal referent. There is considerable grammatical evidence for the claim that grounding predications profile only the grounded thing or process and construe the ground itself subjectively. A first bit of evidence is that the ground itself is left implicit, which is indicative of subjective construal. For instance, articles and demonstratives specify (in)definiteness, a matter of whether the speaker and hearer are able to single out the intended nominal referent. Yet they provide no intrinsic grammatical means (such as an argument slot) allowing the speech event participants to be mentioned overtly; even though this indicates identification by the speaker and hearer as well as proximity to the speaker, we merely say this rock, not *us this rock or *this me rock. Another piece of evidence is that many grounding elements are able to stand alone as nominal or clausal pro forms. Demonstratives as well as the quantifiers all, most, some, any, and each can function as full noun phrases (I prefer those, Some were damaged). Similarly, the modals can serve as pro forms for finite clauses (as is usual in English, a subject is required as well): They should, I may. This suggests quite strongly that grounding elements have the same referent (i.e. profile) as the structures they ground. An expression's profile determines its grammatical class. The fact that a grounding element functions as a noun phrase or a finite clause when used as a

8

RONALD W. LANGACKER

pro form indicates that it profiles a thing or a process, respectively. We thus find further evidence for the claim that grounding predications profile the grounded entity, and not the grounding relationship, by observing that expres­ sions which do profile relations arguably analogous to grounding relations are classed grammatically neither as nouns (which profile things) nor as verbs or clauses (which profile processes). If the past tense relation is reasonably para­ phrased as before now, we can observe that this latter is neither verbal nor clausal, but adverbial. May can often be paraphrased by possible, but the latter is adjectival. The proximity and definiteness conveyed by this are roughly approximated by near me and identified to us, which are prepositional and participial phrases rather than nouns. A grounding element therefore has the properties shown in Figure 1(a). The entity it profiles (indicated by heavy lines) is either a thing or a process. Within the grounding predication itself, the profiled entity is characterized only schematically (i.e. beyond THING or PROCESS, abstractly defined, it has few if any semantic specifications). The 'stage' can be described informally as the general locus of attention; technically it is referred to as the immediate scope of predication (IS). As a matter of definition, the profile is onstage and objectively construed as the primary focus of attention (the focal point within the imme­ diate scope). By contrast, the ground (G) and the grounding relationship are offstage and construed subjectively. The ground is centered on the speaker (S) and hearer (H) in their role as 'viewers' or conceptualizers. The dashed arrows represent their 'viewing' of the focused entity onstage. The specific nature of that viewing constitutes the grounding relationship, which is what distinguishes one grounding predication from another.

Figure 1: Grounding

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS Figure 1(b) represents a grammatical construction in which a grounding element combines with a nominal structure (consisting of the head noun and its modifiers) to form a full noun phrase. The profiled entity is a thing, indicated diagrammatically by a circle. The grounding predication is shown in abbre­ viated form, with a solid line standing for the grounding relationship. A dotted correspondence line specifies that the schematic profile of the grounding element is equated with the profile of the nominal structure, whose more elabo­ rate semantic specifications are labeled X. From the two component structures (the grounding predication and the nominal structure it combines with), the composite structure is obtained by superimposing corresponding elements and merging their specifications. Hence the full noun phrase profiles a thing which has a certain semantic characterization (X) and bears a certain relationship to the ground. The diagram for a finite clause would be the same, except that the profiled entity would be a process rather than a thing. 3.

English modals

In terms of their diachronic evolution, at least some grounding elements achieve that status by a process of grammaticization involving subjectification. A well-known example is the English modals (will, can, may, shall, and must), whose treatment in Cognitive Linguistics has been comprehensively and insightfully surveyed in Radden (1999). Historically, the modals derive from Old English main verbs with meanings like 'want (to)', 'know how (to)', and 'have the power (to)'. These are force-dynamic notions designating the existence of a kind of 'potency' which, once unleashed, tends to induce the occurrence of the activity indicated by the complement clause. Synchronically as well, the modals are forcedynamic in nature (Talmy 1988; Sweetser 1990; Langacker 1991: section 6.3). However, since they now profile the same process (schematically charac­ terized) as the one designated by the complement verb, the potency tending to induce its occurrence is no longer itself the onstage focus of attention. The force-dynamic aspects of modal meanings are thus less objectively construed than they were at the main-verb stage (when they were profiled), and have therefore been subjectified to some degree. It is the extent and nature of their subjectification that concerns us. We can approach this by considering the locus of the potency expressed by modals. At the main-verb stage, the potency was localized in the subject, who had the power, ability, or desire to realize the event coded by the comple­ ment clause, and who was also the understood agent with respect to the

9

10

RONALD W. LANGACKER

complement event. This potency on the part of the subject referent was profiled and objectively construed; it itself constituted the situation 'out there, in the world' that was being described. That is not the case with the grammaticized modals of modern English. In line with the monoclausal nature of the overall construction, the situation 'out there in the world'—the profiled target of description—is now the one that the modal force tends to realize; it is the process designated by the modal complement, which has thus become the main verb (cf. Langacker 1991: section 5.1). The modal profiles that same (sche­ matically characterized) process, rather than the force-dynamic relationship designated by its historical antecedent. The potency conveyed by the modal is offstage and subjectively construed. Moreover, the grammatical subject is no longer its locus; in That spy must be caught!, for instance, it is not the spy who gives the order. The semantic role of the subject is now confined to whatever role it has in the complement (e.g. be caught). The subject's loss of potency is responsible for the modals being transparent, in the sense that anything which can be the subject of the complement can also be the subject of the modal and of the clause as a whole (Langacker 1995a, 1999a). Where, then, is the locus of potency? Consider first the root (or deontic) modal uses. I follow Sweetser (1990) in saying that the root modals indicate force-dynamic relationships in the realm of social interaction. Typically, they reside in the interaction between the speaker and hearer, the latter tending to be expressed overtly as the grammatical subject. Thus, with the different modal choices in (3a) the speaker conveys to the hearer varying degrees of permission or obligation with respect to the complement process. (3) a. You {may, might, can, could, will, should, must} change the cat box. b. This cat box must be changed immediately! [You had better do it right away!]  This cat box must be changed immediately! [Mother insists, so you'd better tell Billy!] d. There must be clean litter in the cat box before the guests arrive! Observe that the hearer, as grammatical subject, is not the source but the target of the modal force. The locus of this potency is the speaker, offstage and subjectively construed. But the hearer can also be implicit, as in (3b), in which case the force-dynamic interaction is wholly offstage. It is not even necessary that the speaker and hearer be the ultimate source and target of the modal force. In (3c), the further context makes it evident that the speech event participants

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

11

are acting only in a communicative capacity, the real source and target of potency being the mother and Billy. Since the grammatical subject does not necessarily function in either capacity, root modals impose no restrictions on their subject. This transparency is seen in (3d), where the subject there is clearly incapable of either exerting or responding to social force. The epistemic modals are also transparent, the different choices indi­ cating varying degrees of likelihood that the complement process will be real­ ized: (4) There {may, might, could, will, should} be some spirited discussion at the meeting tomorrow. What can we say about the nature and the locus of the modal force in cases like these? Sweetser (1990) has described it in terms of the concept of physical force being projected metaphorically onto the domain of reasoning. With may, for example, the import of (4) would be something like 'I am not barred by my premises and the evidence from concluding that there will be some spirited discussion at the meeting tomorrow'. I agree that the conceptualizer's mental activity is ultimately identifiable as the locus of the force. However, I am not sure that reasoning in the narrow sense suggested by notions like 'premise' and 'conclusion' is necessarily involved. I would further question the claim that the epistemic modals are instances of metaphor. Certainly the suggested paraphrase ('I am not barred...') is inappropriate because it explicitly mentions the speaker and the speaker's thoughts and thus construes them objectively, while we have every reason to believe that the modals construe them subjectively. I myself (1991: section 6.3.2) have described the modal force in terms of reality and its evolutionary momentum. Reality, for a particular conceptualizer, is characterized as the history of what has happened up through the moment of speaking. The evolution of reality is seen as having a certain 'momentum' determined by the course it has already followed together with the nature of the world in which it unfolds. Given its current momentum, some future paths are excluded altogether, some are at least potential, and others can be projected with varying degrees of confidence. This account is not unfairly regarded as ambivalent with respect to the locus of potency. On the one hand, evolutionary momentum is seen as being 'out there in the world', construed objectively as inhering in the flow of events within the supporting and constraining matrix of the world's structure. On the other hand, reality is specifically defined as what a given speaker accepts as having occurred, so its evolutionary momentum must also reside in the speaker's assessment.

12

RONALD W. LANGACKER

The two interpretations can be reconciled in the following manner. Although 'reality' pertains to (is a conception of) what happens 'out there in the world', it is still a mental construct. For linguistic purposes, what counts as reality is what a given conceptualizer accepts as having happened up through the present. Evolutionary momentum can thus be localized in the speaker's assessment of reality and its evolution into the future. It resides in the forcedynamic experience engendered by the speaker's mental extrapolation from present reality to an imagined future event. In assessing the flow of reality up through the present, the speaker experiences a certain degree of ease or diffi­ culty in projecting its continuation on a course such that it comes to encompass the envisaged event. This experience is not the focused object of conception (the event is), nor does the conceptualizer explicitly think of him-/herself as having it. Instead the experience and the experiencer are offstage and subjec­ tively construed. In an English finite clause, the absence of a modal indicates that the profiled process is located in reality. In this case the general proximal/distal contrast (whose schematic meaning is epistemic immediacy/non-immediacy; cf. Brisard 1999) takes on its prototypical value of 'present' vs. 'past' tense (proximity to the ground—in particular the time of speaking—vs. distance from the ground). A modal indicates that the profiled process is not accepted as part of reality. It is, however, assessed in relation to reality, specifically by the mental extrapolation of reality to encompass it. In this context the proximal/ distal contrast bears on the ease or difficulty of the extrapolation, interpretable as the likelihood of the process occurring. For instance, It may rain conveys that there is no barrier to extrapolating from reality as it is presently constituted to a future event of raining. The corresponding distal form, It might rain, portrays the likelihood of the event as being more tenuous. It suggests that a comparably unproblematic extrapolation to an event of raining would have to start from a situation that is somehow removed from current reality. Of course, the proximal and distal forms of the modals (will vs. would, shall vs. should, etc.) involve numerous subtleties that we cannot pursue here. What does concern us here is the well-known fact that the process grounded by a modal is not always located at a future time. The processes profiled in (3) and (4) are typical in being subsequent to the time of speaking. With epistemic modals, however, it is also possible for the profiled situation to temporally coincide with the ground. In (5a), the situation of her being finished has not yet been accepted as part of reality, but in purely temporal terms it is present rather than future.

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

13

(5) a. She may be finished already. b. They must have left the dog at home. From examples like (5b) it might appear that the process in question can also lie in the past; if indeed they left the dog behind, they did so prior to the time of speaking. But it is not leave that is profiled and grounded in (5b), but rather have. The sentence designates an ongoing situation characterized by the prior occurrence and continuing relevance of the canine's abandonment. Hence the modal uses of interest consistently profile a present situation. How do we then describe these uses? What is it that unites or distin­ guishes the future and the present epistemic modals, as in (4) and (5)? They are basically alike in that the profiled process is not yet accepted as part of reality. The reason of course is different in the two cases. With a future modal, accep­ tance in reality is not yet an issue, since by definition reality extends no farther than the present. By contrast, a present modal hinges on the fact that a given conceptualizer has only imperfect knowledge of what may have actually hap­ pened up through the time of speaking. Whether the profiled situation exists has already been determined 'out there in the world', but the speaker does not yet know one way or the other. If we want a single account of the English epistemic modals, we need to extract what is common to the future and present uses. Their abstract common­ ality resides in the force-dynamic experience engendered by the speaker's extrapolation of currently accepted reality to encompass the profiled process. This schematic characterization neutralizes the contrast between present and future time, as well as the concomitant difference in whether or not the issue of the occurrence of the designated process has already been decided 'out there in the world'. Although we do want a single account, I have no real doubt that the present and future modal constructions also exist as established cognitive units. Presumably, the future modal construction is a prototype with respect to which the present modal construction constitutes an extension. The schema they both instantiate embodies the abstract similarity which motivates the extension, whereby the same form (e.g. She may have a headache) can often have either a present or a future interpretation. That brings us to the crux of our immediate concern with modals, which lies precisely in the relationship between the prototypical and extended values of the epistemic modal constructions. The shift from a future to a present modal interpretation involves a subtle kind of subjectification, above and beyond the high degree of subjectification already implied in a future use. For this reason, I describe the present modal construction as a case of extreme

14

RONALD W. LANGACKER

subjectification. I will go on to suggest that an analogous instance of extreme subjectification is the key to understanding various 'special' uses of the English present tense. Let us quickly review the degrees of subjectification exhibited by the English modals. In the main verbs from which the modals descend historically, the locus of potency was the grammatical subject, and that potency was onstage as the verbal and clausal profile, hence construed with maximal objectivity. In modern English, the modals are no longer 'main verbs' (or 'content verbs', as defined in Langacker 1991: section 5.1) but rather part of the grounding appa­ ratus for finite clauses. They are still verbs in a broad sense of the term, in that they profile a (schematic) process. However their essential semantic contri­ bution now resides in a force-dynamic relationship that in several respects is construed more subjectively than the potency profiled at the main-verb stage. One factor in its lesser objectivity is that it is non-physical, and consequently less salient. Another factor is that the source of the potency is no longer the subject (the primary focal participant onstage), but rather some entity that is offstage, implicit, and thus subjectively construed. This results in the third factor, namely that the force-dynamic relationship is itself offstage and unprofiled. In the case of root modals, the force inheres in a social interaction between the speaker and hearer. To the extent that they themselves function as its source and target, they also have the mental experience of exerting or absorbing it. There are also uses, like (3c), where the speaker's and/or hearer's involvement is largely limited to their communicative role. If they are not themselves the source and target of the social force, but merely convey it, they may nonetheless experience it vicariously through empathy. At the very least, they entertain the conception of social force by virtue of apprehending the message they convey. We can plausibly assume that the conceptualization of force, even on an intellectual plane, involves some abstracted vestige of its physical or social experience—an embodied image schema in the sense of Johnson (1987) and Lakoff (1987). Be that as it may, the force implied by a root modal use is meant to be effective: it is aimed at actually bringing about the occurrence of the onstage process designated by the grounded verb. By contrast, the epistemic modals are merely predictive with respect to the profiled process. The modal force is not directed outward, but is rather an aspect of the conceptualizer's experience in mentally extrapolating reality to encompass that process. The epistemic modal force is thus construed even more subjectively than the root modal force, inhering solely in the conceptualizer and the conceptualizing activity. This is

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

15

true for both the future and the present epistemic modals. But if they both construe the modal force with a high, almost maximal degree of subjectivity, there is still a subtle sense in which its subjectification in the present-time modals is even more extreme. With the present modals, the evolution of reality—as a mental construct —is no longer tied to occurrences 'in the world out there'. With future-time modals, the speaker's mental extrapolation at least pertains to how the world out there can be expected to evolve. There is still the notion that previous events out there have an evolutionary momentum that will carry over and influ­ ence what will happen in the future. Although this later course of events has yet to be determined, the future modals purport to predict it to the extent this momentum allows. The conceptualizer's force-dynamic mental experience can perhaps be taken as an internal representation of the force ascribed to the exter­ nal flow of events. With present-time modals, on the other hand, evolution of the world out there is not at issue. The profiled process is a stable situation whose existence or non-existence out there, as a factual matter, has already been determined. The conceptualizer's mental extrapolation does not pertain to how the world might evolve, nor is the force-dynamic experience a repre­ sentation of force ascribed to the external flow of events. The only thing conceived as evolving is what the speaker supposedly knows, i.e. reality as a mental construct. The modal force inherent in its evolution is therefore subjec­ tive in the extreme. 4.

The English present tense

Let us turn now to the English present tense, for which I posit an analogous case of extreme subjectification. I have previously argued for a 'naive' charac­ terization of the English present, whereby "a full instantiation of the profiled process occurs and precisely coincides with the time of speaking" (1987b:82). The key factor is what it means for a process to 'occur'. In the uses we think of as canonical, the occurrence is 'out there in the world'. Of course, in describing the occurrence of a process, a speaker must conceptualize it, mentally running through the continuous series of profiled relationships (distributed through a span of conceived time) that constitute it. This mental scanning through the constitutive phases of a process amounts to its conceptual occurrence, residing in the speaker's conceptualizing activity. Now a speaker can certainly carry out these operations autonomously, without their being driven by immediate obser­ vation of the process happening 'out there'. I will be suggesting that, for certain uses of the present, what counts as the occurrence of a process is precisely an

16

RONALD W. LANGACKER

autonomous conceptual occurrence of this sort. Not pertaining to external developments, but residing solely in the conceptualizer's mental activity, the temporal unfolding of the process is construed with extreme subjectivity. The proposed characterization of the present tense accounts directly for some basic facts about its canonical use, which I take to be the description of actual processes occurring at the time of speaking. In this canonical situation, a perfective verb normally cannot be used in the present, but must instead be made progressive. On the other hand, an imperfective verb does take the present but cannot be made progressive. Moreover, the verb in a true perfor­ mative clause occurs in the present tense (as a matter of definition) despite being perfective. These facts are respectively exemplified in (6). (6) a. b.  d. e.

*Cher cleans her room right now. [perfective] Cher is cleaning her room right now. [progressive of perfective] This cat has fleas. [imperfective] *This cat is having fleas. [progressive of imperfective] I promise that I will behave. [performative (perfective)]

A perfective verb is bounded, so a full instantiation of the profiled process includes its endpoints. A present-tense perfective thus implies the configuration sketched in Figure 2(a), where the arrow represents conceived time (t), the box with squiggly lines stands for the speech event, and the profiled process is given by a heavy line segment with a boundary shown at each end. An imper­ fective verb profiles the continuation through time of a stable relationship not conceived as being inherently bounded. It is thus depicted in Figure 2(b) as a line that potentially extends indefinitely. An imperfective process is mass-like, in that any portion of an instance is itself a valid instance of the same process type (Langacker 1987b). In 2(b), only that portion of the overall process which coincides with the time of speaking is shown as being profiled (and thus as constituting the grounded process). This is because the time of speaking defines the temporal onstage region for the present tense, and by definition an expression's profile is the onstage focus of attention. Finally, a performative verb is perfective, and a performative sentence is one that profiles the very speech act effected by its utterance. Therefore, as shown in Figure 2(c), the profiled process and the speech event are one and the same. The data in (6) can then be explained as follows. An imperfective process that extends through the time of speaking can always be expressed in the present tense, since that portion of it which coincides with the time of speaking is selected as the processual profile (which has to be onstage) and counts as a

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

17

full instantiation of the process type (owing to the mass-like nature of imperfectives).

Figure 2: Present tense configurations

The progressive construction serves to imperfectivize a perfective process. A progressive can therefore occur in the present tense, and only perfectives are made progressive (imperfectivizing an imperfective would be vacuous). A performative clause takes present tense, despite its perfective nature, because it has the special property that the profiled process and the speech event are one and the same, hence temporally coincident. But what about non-performative perfectives? Why are they infelicitous in canonical uses of the present? Observe that there is nothing inherently anomalous about the configuration diagrammed in Figure 2(a). The difficulty does not reside in conceptual incoherence but rather in pragmatic unlikelihood. There is both a durational problem and an epistemic one. The durational problem is that the length of time it takes for a process to occur is usually not the same as the length of time required to describe it—cleaning a room takes longer than uttering (6a). As for the epistemic problem, if one observes an event in order to identify and then describe it, the description has to lag behind the event itself, so temporally they cannot coincide. I take this explanatory force as strong evidence for the correctness of the proposed characterization of the English present tense. But what about the many uses where the designated process does not occur at the time of speak­ ing? These are the so-called 'special' uses, of which I will single out just three as representative examples: singular generics, the scheduled future, and the historical present: (7) a. A kitten is born with blue eyes. b. We leave for France next Tuesday.  A man comes into the store last night and starts complaining.

[singular generic] [scheduled future] [historical present]

18

RONALD W. LANGACKER

The crucial property of such expressions, I suggest, is that they do not refer directly to actual events. They represent different kinds of departures from the canonical viewing arrangement in which a static observer directly describes objectively construed events and situations involving specific individuals. There are many such departures. Often they involve some kind of 'fictive' or 'virtual' entity, such as the fictive motion exemplified in (2). Of particular rele­ vance here are cases of fictivity in the tacit viewing arrangement an expression presupposes (Talmy 1996; Langacker 1999b). A sentence like (8), for example, can be analyzed as an accurate description of what the speaker observes while driving along, but construed with respect to a fictive (albeit canonical) arrange­ ment in which the observer is static. (8) The trees are rushing past me at 90 miles per hour. It is helpful to draw a terminological distinction between actual and virtual (or fictive) entities, and to think of them metaphorically as occupying different 'planes'. The actual plane comprises reality as well as occurrences whose reality is at least at issue (hence negation or a modal does not remove them from actuality). Occurrences that are non-actual are said to occupy a virtual plane of some kind. The process described in (8), for instance, is in a virtual plane created by interpreting actuality in terms of a fictive viewing arrange­ ment. Let us then consider a singular generic, such as (7a) (for an analysis of plural generics, see Langacker 1996, 1997b). It is a truism that generic state­ ments do not refer directly to actual events involving specific individuals. Thus, (7a) does not mention any actual kitten nor any actual instance of being born. Yet it consists of a finite clause, implying that it profiles a grounded instance of a process type, and its subject is a full noun phrase, implying that it profiles a grounded instance of a thing type (Langacker 1991). But not every instance of a thing or process type is an actual one. We very often deal with arbitrary instances, which are 'conjured up' for some special local purpose and have no existence or status outside that context. A familiar example is the referent of a non-specific noun phrase in the context of a verb like hope, e.g. the car in the complement of (9), on the interpretation where Monica has no specific car in mind. (9) Monica hopes to own a fancy car.

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

19

Being conjured up just for a special local purpose, arbitrary instances are fictive entities rather than actual individuals. In (7a), then, the kitten and the birth are fictive instances of their respec­ tive types, conjured up for purposes of expressing a generalization about the nature of our world. I will say that the profiled thing and process occupy the structural plane, since the statement is meant to characterize a constant general aspect of the world's 'structure', as opposed to specific phenomena which happen to occur within that stable framework (Goldsmith & Woisetschlaeger 1982). Of course, as a generalization about what the world is thought to be like, a process in the structural plane is understood as projecting to indefinitely many actual occurrences of the same process type, as sketched in Figure 3.

Figure 3: Singular generic

What we accept as the world's stable structure is not immutable. Like a policy manual, our conception of the structural plane needs updating from time to time, as new generalizations are added and others cease to hold. At a given moment, however, it is like a policy manual in that a certain number of gener­ alizations are now in force and represent a stable characterization of how the world is currently structured. We can usefully push the metaphor even further, and think of the structural plane as a kind of 'virtual document' available to be 'read' at any time. Inscribed on this document are a large number of virtual events and situations representing particular facets of the world's structure, each of which is normally expected to be instantiated in actuality under the appropriate circumstances. In making a generic statement, one is 'reading' from this virtual document. The reading of an inscription on the virtual document involves a con­ ceptual occurrence of the process in question. In making a generic statement, the speaker apprehends and describes an event in the structural plane in a way

20

RONALD W. LANGACKER

quite analogous to the apprehension and description of events in actuality, except that the viewing arrangement is different. It is different because, in the canonical viewing arrangement, the speaker conceptualizes the event by virtue of observing its actual occurrence, so the occurrence and its initial concep­ tualization (although not necessarily its description) are temporally coincident. With a generic statement, on the other hand, the profiled event does not actu­ ally occur at all and thus has no specific temporal location. Being inscribed on a virtual document, it is available for 'reading' at any time, and whenever the speaker reads it off, it enjoys a virtual occurrence residing in the very activity of the speaker conceptualizing it. When the speaker reads it aloud in making a generic statement, its conceptual occurrence—implied by the speaker appre­ hending what he/she is saying—coincides with the time of speaking, as shown in Figure 3. This special viewing arrangement avoids the durational and epistemic problems that normally preclude the felicitous use of present-tense perfectives. Since the profiled event is only virtual, the duration of its conceptual occur­ rence is independent of how long it takes for such an event to actually occur. Moreover, since the speaker can scan the virtual document at leisure, and knows what he/she wants to say before reading it aloud (thus inducing its conceptual occurrence), the epistemic problem also disappears. It turns out that these problems are not inherent in perfectives or the present tense per se, but rather in their application to the canonical viewing arrangement, where the speaker has no control over the timing, duration, or identity of the outside events being reported. When the speaker does control these matters, as in the case of performatives (or the reading of virtual documents), the problems do not arise and a present tense perfective is completely well-formed. For special viewing arrangements involving the reading of virtual docu­ ments, we need to distinguish two kinds of temporal location: the time to which a document pertains, and the time when it is available for viewing. In the case of English generics, it appears that these two time spans are the same, as argued below. We find a plausible rationale for their coincidence in the very nature of generic statements. The virtual document they invoke is abstracted from actu­ ality as a representation of its stable features—a compendium of constant situations and recurrent events. This document is a characterization of what the world is currently like, a listing of the generalizations currently in force. If the world changes, so that certain generalizations are no longer in force, by defi­ nition they are no longer part of the structural plane, hence no longer available for viewing. Inscriptions describing what the world is currently like are only available for viewing during the time span they pertain to.

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

21

Generic statements can, however, pertain to the past or the future, not just the present: (10) a. At that stage of feline evolution, a kitten was born with green eyes. b. In the coming evolutionary era, a kitten will be born with purple eyes. I analyze such expressions as implying a shift in vantage point. Although generalizations not currently in force are not available for immediate viewing, in the manner of Figure 3, we can still rely on our general ability to invoke a vantage point other than our actual one and imagine what things look like from there. The sentences in (10) invoke a viewpoint located at the time specified by the initial adverbial phrase, and state a generalization in force at that time. The implication is that a viewer at that location would be able to read it off the virtual document then available for immediate viewing. The effect of the past tense in (10a), as well as the future will in (10b), is thus to indicate the position vis-à-vis the ground of the viewpoint and the tacit viewer. These notions are more evident in examples like the following, where the viewer is made explicit: (11) a. Julius Caesar believed that a kitten was born with green eyes. b. For zoologists of the coming evolutionary era, a kitten will be born with purple eyes. The distinction between the time to which a document pertains and the time when it is available for viewing may not be fully evident in the case of generics, precisely because the two time spans coincide. It is more apparent in other kinds of examples, involving other sorts of viewing arrangments. In particular, there is an obvious discrepancy between the two time spans in both the scheduled future, as in (7b) We leave for France next Tuesday, and the historical present, as in (7c) A man comes into the store last night and starts complaining. In both these uses of the present tense, a virtual document is available for viewing during an interval that includes the time of speaking. Yet the scheduled future clearly pertains to the future, and the historical present, to the past. I am positing a basic difference in the tacit viewing arrangement presup­ posed by sentences like (12a) and (12b):

22

RONALD W. LANGACKER (12) a. My mother will arrive next week. b. My mother arrives next week.

Despite the modal will, (12a) is a direct description of an actual event. It presupposes the canonical viewing arrangement in which the speaker is observing and describing what happens. The import of a future modal is that the event in question has not yet happened, so that the speaker's conception of reality has to be mentally extrapolated (with a certain amount of felt effort) to encompass it. It is nonetheless the future event itself—actual though not yet real—that is profiled and grounded. By contrast, I maintain that (12b) is not the description of an actual event but a virtual one. The profiled process is the not the arrival per se, but rather its representation on a virtual schedule, i.e. a plan or projection concerning the anticipated occurrence and timing of events in future actuality. That some kind of schedule is involved is suggested by data like the following: (13) a. ??My mother arrives. b. ??An earthquake strikes next year. This use of the present tends to be infelicitous unless there is some indication of when the event is expected to occur. It is also problematic for events not amenable to scheduling.

Figure 4: Scheduled future

While pertaining to the future, the virtual document is available for reading at any earlier time when the schedule is in effect. If its availability includes the time of speaking, the speaker can read off an entry to produce a sentence like (12b). And since the document is available for immediate scanning, reading off

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

23

an entry induces a conceptual occurrence of the profiled process coincident with the time of speaking. This viewing arrangement is sketched in Figure 4. Note that rectangles delimit spans of time during which events occur. Although the times indicated on the schedule project to future times in actuality, the events inscribed on the schedule are only virtual. Hence the conceptual occurrence of the profiled event coincides with the time of speaking, even though its actual occurrence is projected for a future time. I will not attempt to examine the historical present in any detail. Suffice it to say that the virtual document being read is a recollection of past occurrences. In this conventional mode of speech, the speaker reads off the inscribed events in the order of their previous occurrence, and in so doing produces their conceptual occurrence (a kind of 'mental replay'). The immediacy of this (re)viewing contrasts with the 'distance' of a past-tense narrative, where the profiled events are actual rather than virtual, but viewed from a temporally later vantage point (the ground). Observe that this analysis of the historical present is different in detail (though comparable in spirit) to the more usual account in which the speaker, by mental transfer, adopts a vantage point in the past. Immediacy of fictive viewing is instead accounted for by positing a virtual document available for reading in the present. A significant advantage of this account is that the historical present is handled in the same basic way as other 'special' uses of the present. (For these, too, the account offered here differs in detail from the one I previously proposed (1991: section 6.2.3). I have profited from discussions with Mariko Higuchi Goto on these matters.) 5.

Conclusion

To summarize, I have tried to draw a connection between two well-known phenomena, each involving the grounding of English finite clauses: the use of epistemic modals for present situations, and 'special' uses of the present tense. I have analyzed them both as cases of extreme subjectification, a step beyond the extensive subjectification resulting in more basic uses of these grounding elements. The further step they both exemplify is to eliminate any direct concern with the occurrence of actual events. In the case of modals, the mental extrapolation inducing a force-dynamic experience no longer pertains to events 'out there in the world', but rather to what the speaker knows about a situation whose possible occurrence has already been decided. In the case of present tense, the occurrence of a process precisely coincident with the time of speak­ ing is no longer an actual occurrence, but rather a conceptual occurrence created by the 'reading' of a 'virtual document'. I am not claiming that these

24

RONALD W. LANGACKER

developments are analogous in all respects, nor do I know whether connecting them has any real consequences. I do, however, believe that the parallel is suggestive and worthy of further investigation. References Brisard, Frank. 1999. A Critique of Localism in and about Tense Theory. Ph.D. dissertation, University of Antwerp. Carey, Kathleen. 1994. Pragmatics, Subjectivity and the Grammaticalization oj the English Perfect. Ph.D. dissertation, University of California, San Diego. Goldsmith, John & Erich Woisetschlaeger. 1982. "The Logic of the English Progressive". Linguistic Inquiry 13.79-89. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago,I l l . :The University of Chicago Press. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago, Ill.: The University of Chicago Press. Lambrecht, Knud. 1996. "On the Formal and Functional Relationship between Topics and Vocatives: Evidence from French". Conceptual Structure, Discourse, and Language ed. by Adele Goldberg, 267-288. Stanford, Calif: CSDL Publications. Langacker, Ronald W. 1985. "Observations and Speculations on Subjectivity". Iconicity in Syntax ed. by John Haiman, 109-150. Amsterdam: John Benjamins. . 1986. "Abstract Motion". Proceedings of the Twelfth Annual Meeting of the Berkeley Linguistics Society ed. by Vassiliki Nikiforidou, Mary VanClay, Mary Niepokuj & Deborah Feder, 455-471. Berkeley, Calif: Berkeley Linguistics Society. . 1987a. Foundations of Cognitive Grammar, Vol. 1, Theoretical Pre­ requisites. Stanford, Calif: Stanford University Press. . 1987b. "Nouns and Verbs". Language 63.53-94. . 1990. "Subjedification". Cognitive Linguistics 1.5-38. . 1991. Foundations of Cognitive Grammar, Vol. 2, Descriptive Appli­ cation. Stanford, Calif: Stanford University Press. . 1993. "Deixis and Subjectivity". New Horizons in Functional Linguistics ed. by S. K. Verma & V. Prakasam, 43-58. Hyderabad: Booklinks Corporation. . 1995a. "Raising and Transparency". Language 71.1-62.

EXTREME SUBJECTIFICATION: ENGLISH TENSE AND MODALS

25

. 1995b. "Viewing in Cognition and Grammar". Alternative Linguis­ tics: Descriptive and Theoretical Modes ed. by Philip W. Davis, 153— 212. Amsterdam: John Benjamins. . 1996. "A Constraint on Progressive Generics". Conceptual Structure, Discourse, and Language ed. by Adele Goldberg, 289-302. Stanford, Calif. : CSDL Publications. . 1997a. "Consciousness, Construal, and Subjectivity". Language Structure, Discourse and the Access to Consciousness ed. by Maxim I. Stamenov, 49-75. Amsterdam: John Benjamins. . 1997b. "Generics and Habituais". On Conditionals Again ed. by Angeliki Athanasiadou & René Dirven, 191-222. Amsterdam: John Benjamins. . 1998. "On Subjectification and Grammaticization". Discourse and Cognition: Bridging the Gap ed. by Jean-Pierre Koenig, 71-89. Stanford, Calif.: CSLI Publications. . 1999a. "Losing Control: Grammaticization, Subjectification, and Transparency". Historical Semantics and Cognition ed. by Andreas Blank & Peter Koch, 147-175. Berlin: Mouton de Gruyter. . 1999b. "Virtual Reality". Studies in the Linguistic Sciences 29.77103. Matsumoto, Yo. 1996a. "How Abstract is Subjective Motion? A Comparison of Coverage Path Expressions and Access Path Expressions". Conceptual Structure, Discourse, and Language ed. by Adele Goldberg, 359-373. Stanford, Calif: CSDL Publications. . 1996b. "Subjective Motion and English and Japanese Verbs". Cogni­ tive Linguistics 7.183-226. . 1997. "Linguistic Evidence for Subjective (Fictive) Motion". The Locus of Meaning: Papers in Honor of Yoshihiko Ikegami ed. by Kei Yamanaka & Toshio Ohori, 209-220. Tokyo: Kuroshio. Radden, Günter. 1999. "Modalverben in der Kognitiven Linguistik". Grammatik und Mentale Prozesse ed. by Angelika Redder & Jochen Rehbein, 261-294. Tübingen: Stauffenberg Verlag. Sweetser, Eve E. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Talmy, Leonard. 1988. "Force Dynamics in Language and Cognition". Cognitive Science 12.49-100.

26

RONALD W. LANGACKER

, 1996. "Fictive Motion in Language and 'Ception'". Language and Space ed. by Paul Bloom, Mary A. Peterson, Lynn Nadel & Merrill F. Garrett, 211-276. Cambridge, Mass.: MIT Press. Traugott, Elizabeth Closs. 1989. "On the Rise of Epistemic Meanings in English: An Example of Subjectification in Semantic Change". Language 65.31-55.

MEANING AND CONTEXT JOHN R. TAYLOR University of Otago, New Zealand

In this contribution, I address some aspects of what it means to know the meaning of a word. The contribution was triggered by my reading of Gleitman (1990) and Wilkins (1993/4). Gleitman argued that verb meanings cannot be inferred solely from the extralinguistic context of their use; crucial information on a verb's meaning can only come from the kinds of linguistic contexts in which the verb is used. In fact, the two aspects are so closely intertwined that it might be fruitless to contemplate the one except in terms of the other. Wilkins, on the other hand, presents the case of a word—the English verb put—whose syntactic distribution, she claims, is highly idiosyncratic, given its semantics; accordingly, the syntax of put ought to present the learner with a major learning problem. I suggest that Wilkins greatly exaggerated the problematic nature of this verb. In fact, its syntax closely matches its semantics—something that child learners appear to grasp at a very early age, and with relatively few errors along the way.

1.

On word meanings

Suppose you were asked to explicate the difference in meaning between the three nouns chair, table, and cupboard. The task seems fairly straightforward, even trivial. A not unreasonable approach might be to say that the difference in meaning between the three nouns is simply the difference between chairs, tables, and cupboards. Chairs, tables, and cupboards are different kinds of things, they are structured in different ways and they serve different functions, and the English words name these different categories of things. Ostension (pointing to objects of the relevant kinds) will surely play a critical role, at least for a gross differentiation of the categories. There are, to be sure, some well-known problems with ostension as a means of elucidating word meanings, even for words which designate cate­ gories of concrete objects (Quine 1953; Wittgenstein 1958). If I point to a chair, and say, "That is called chair in English", how do you, the hearer, infer

28

JOHN R. TAYLOR

that chair means 'chair', and not, say, 'antique armchair', 'furniture', 'some­ thing that is not a cupboard', 'bits and pieces assembled in a chair-like way', 'firewood', 'something on my left', or whatever? For ostension to work, we need to assume that certain cognitive capacities are in place. Perhaps the most basic of these, underlying the very capacity for language itself, is symbolic ability, i.e. the ability to establish arbitrary associations between sounds and concepts.1 In order to form concepts, the learner must also be able to generalize over instances.2 For the concepts CHAIR, TABLE, and CUPBOARD to emerge, in contrast to some of the more outlandish possibilities suggested above, there has to be a predisposition to pick out individuated things as the objects of categori­ zation.3 There also has to be a predisposition to categorize at the right level of abstraction, i.e. at the so-called basic level (Rosch 1978) rather than at subor­ dinate or superordinate levels.4 Given these preconditions, ostension need not be problematic at all, at least for the initial 'bootstrapping' of meanings; on the contrary, its role will be crucial. However, for many words in the lexicon, perhaps the majority, ostension clearly will not work at all. Lila Gleitman (1990) discussed this issue with reference to verbs, their meanings, and the learning of their meanings. The problem, as Gleitman persuasively argued, is that verb meanings are not 'out there', clearly individuated and marked off, one from another, such that language users can unambiguously pick them out and point to them. With the English verbs (to) see, (to) look, and (to) watch, for instance, it is evident that ostension can play hardly any role at all in differentiating their meanings. It seems hardly possible at all to point to situations and to sort them into exam­ ples of seeing, looking, and watching. And if we include other verbs from the same semantic field, such as notice, observe, sight, etc., the problem becomes yet more impervious to ostension.5 This state of affairs raises interesting methodological questions. If osten­ sion is excluded as an option, how does one, in fact, go about elucidating the meanings of these verbs? What range of facts does one appeal to? In what way do these facts have a bearing on the explications that one gives? And how is a semantic explication to be evaluated? For example, if someone were to come up with definitions of see, look, and watch that other speakers of English felt were clearly inadequate, what criteria would these speakers apply in order to argue that the definitions are indeed inadequate or wrong? Gleitman situated this problem in the context of language acquisition. Given that ostension is inadequate as a means of differentiating verb meaning, how can a child, learning English as its mother tongue, end up with the correct knowledge of the three verbs? I suspect that the problem of definition, and the

MEANING AND CONTEXT

29

acquisition problem as discussed by Gleitman, are rather intimately connected. The criteria which a reflective linguist appeals to in elucidating word meanings are plausibly not dissimilar to the evidence that the child learner must pay attention to. Even though the differences between see, look, and watch are impervious to ostension, speakers of English manifestly do acquire the three words, and their differentiated meanings, with complete success. (We all end up using the words with complete appropriateness, with little variation from one speaker to the next.) If learners cannot rely on ostension, they must be relying on some­ thing else. And when mature native speakers—as well as linguists and lexi­ cographers—attempt to describe the differences between the words, they too must appeal to this 'something else'. The 'something else' is, of course, and can only be the 'context' in which the words are used. 'Context', as used here, may certainly include the extralin­ guistic context, but extralinguistic context alone is insufficient, to the extent that configurations of extralinguistic reality cannot be identified as examples of a 'seeing' kind of event, in contrast to, say, examples of a 'looking' kind of event. The context has to include the linguistic context, broadly understood, that is, the syntactic and lexical environments in which the words are used. Let us see how this might work. If we examine the ways in which the three verbs see, look, and watch are used, we discover that they are associated with quite different linguistic contexts. I review some of the more obvious differences. (i) Use with a prepositional complement. A salient difference between the three verbs is that if the thing perceived is mentioned, this participant in the situation appears as the direct object of see and watch. With look, a preposition is required, typically, but not exclusively, at. (1) a. I saw the man. / *I saw at the man. b. I watched the man. / *I watched at the man.  *I looked the man. / I looked at the man. (2) a. b.  d. e.

look into the room look out of the window look over your shoulder look under the bed look through the keyhole

30

JOHN R. TAYLOR

What these prepositions have in common is that they are all directional prepo­ sitions, and contribute the idea of directed motion. They profile not a location, but a path; typically, the path is characterized with respect to its end point. This is true even of at. Although at typically designates a location (e.g. the man at the bus stop), its use in (1) parallels its directional use after certain contact verbs (3a), certain verbs of caused motion (3b), some verbs of 'aggressive' verbal activity (3c), and with a verb which inherently designates directed activity (3d). (3) a. b.  d.

kick (at), hit (out) (at), grasp (at), snatch (at) shoot (at), throw (at) swear at, shout at, scream at, yell at aim at

At, then, contributes the idea that the subject entity performs some activity that proceeds in the direction of a target entity. Some verbs, such as hit and shoot, allow a choice between using at and omitting it, with, though, a significant difference. If you 'shoot a person', the bullet makes contact with that person (and thus injures and, perhaps, kills them). But if you 'shoot at a person', you merely shoot in their direction; you may well have the intention of hitting them with the bullet, but may not actually do so. Note also the contrast between throwing a ball 'to' someone (with the intention that they catch it), and throw­ ing a rock 'at' someone (with the intention of injuring them). What does this tell us about look vs. see and watch? Clearly, it tells us that look (at) construes visual perception in terms of directed motion. It is as if a person 'sends their gaze' along a path which terminates 'at' the perceived object.6 Hence the idea of a person actively investigating something with their gaze, rather than just passively registering its presence in the visual field. The point to emphasize, of course, is that this insight into the semantics of look follows from observations about the syntax of the verb. It is not some­ thing that could be inferred from correlating the use of see with certain kinds of situation. (ii) Use in imperatives. See tends not be acceptable as an imperative, irrespec­ tive of whether the thing to be seen is mentioned or not. Equally impossible is the use of see with certain adverbials, such as carefully, intentionally, on purpose.

MEANING AND CONTEXT (4) *See! Look! Watch!

*See this!7 Look at this! Watch this!

31

*S this carefully! Look at this carefully! Watch this carefully!

Prototypically, an imperative instructs a person to do something, and is there­ fore only acceptable if a person has a choice between carrying out the instruc­ tion or not. The impossibility of the imperative indicates that see designates a process over which a person has no real control. Seeing is something that just happens, given appropriate preconditions (that the person is sighted, that there is illumination, etc.). This also accounts for the impossibility of using adverbials like carefully: one cannot possibly exercise care on a process which just happens. (iii) Use with progressive. Related to the above is the difficulty of using see with progressive aspect. Progressive aspect, in English, presents a momentary 'snapshot' of a situation in progress at a given time; moreover, the situation is construed as internally homogeneous, often with the suggestion that the situa­ tion is of relatively short duration, or, at least, that it is temporally demarcated, with a definite starting-point and finishing point. That is to say, even though progressive aspect does not 'profile' the beginning and end points of the situa­ tion (the progressive is therefore incompatible with 'perfective' events), the temporal demarcation of the situation is subsumed within the 'base of predi­ cation' of the progressive. For these reasons, the progressive is in principle also incompatible with stative situations.8 With respect to the progressive, see behaves very much like a stative verb; it designates an on-going process that simply 'happens', with no assumptions about its beginning or end points.9 Watch and look (at), in contrast, can designate situations that are construed both as internally homogeneous and as temporally circumscribed. The latter aspect is consistent with the 'volitional' character of watching and looking. (5) a. I saw the Taj Mahal. / *I was seeing the Taj Mahal. b. I looked at the Taj Mahal. / I was looking at the Taj Mahal.  I watched television. / I was watching television. The compatibility of the three verbs with the progressive can also be studied with respect to their complements. Compare: (6) a. I saw the sun go down. b. I saw the sun going down.

32

JOHN R. TAYLOR (6) c. *I was seeing the sun go down. d. *I was seeing the sun going down. (7) a. b. c. d.

*I looked at the sun go down. I looked at the sun going down. *I was looking at the sun go down. I was looking at the sun going down.

(8) a. I watched the sun go down. b. I watched the sun going down.  I was watching the sun go down. d. I was watching the sun going down. Be seeing in (6c) and (6d) is as bad as ever. Look (at) and watch, however, show an interesting contrast, which is possibly related to another aspect of these verbs, i.e. their choice of nominal complements. Watch tends to take different kinds of nominal complements than see and look at. Only things that move or change, or that can be expected to move or change, can be watched. To watch the Taj Mahal would be an odd thing to do—unless one expected the building to change in some way, during the period of time it was being watched. Similarly, a sentence such as The private eye watched the house for two days would imply that the detective was waiting for some change to occur with respect to the house, e.g. that someone entered, or left. Watch is remark¬ able, in that it can take a complement denoting both a perfective, completed event (I watched the sun go down) and an imperfective process (I watched the sun going down); and the watching itself can be both perfective and imper­ fective (I watched ... vs. I was watching ...). This suggests that watching is an 'elastic' process, whose duration and aspectual properties are contingent on the perceived situation. Look, with its sense of 'inspect with the eyes', is directed either at things or at things undergoing change, but not at all at perfective events. Hence, one can 'look at something happening', but one cannot 'look at something happen'. (iv) Extended uses. The three verbs of visual perception have a range of 'extended' uses having to do, not with visual perception per se, but with nonvisual cognitive processes. (9) a. I must look into this matter. (= examine) b. Watch the situation as it develops. (= pay attention to)  I do not see what the problem is. (= know, understand, be aware)

MEANING AND CONTEXT

33

As a visual word, look has the sense 'direct one's gaze at, in order to find out about'. The extended use of the verb perspectivizes the component 'find out about' (by whatever means). Watch, in its basic sense, seems to have the sense 'attach one's gaze to', 'track changes to an object with one's gaze'. Not surprisingly, watch comes to be used in the more general sense of 'keep track of how some situation changes over time'. Note that these extensions do not bring with them changes in the verbs' syntax; they concern only selection restrictions. Especially interesting is the wholesale extension of see to a verb of cognition. As such, its syntax overlaps with that of other verbs of cognition, such as know, believe, understand, realize. The extension is plausibly moti­ vated by the fact that much—perhaps most—of our knowledge of the outside world (for sighted people!) comes from vision. For most people, 'seeing something with their own eyes' is good enough evidence that such is the case. Compare: (10) a. I saw the cat on the mat. b. I saw that the cat was on the mat. From (10a), it is reasonable to infer the truth of the proposition 'the cat is on the mat'; it would be incoherent to say, "I saw the cat on the mat, but the cat was not on the mat". It would be wrong to conclude, however, that perception is always and necessarily propositional. On the one hand, it could be argued that in (10a) and (10b), I see exactly the same thing in both cases—a cat, a mat, with the cat on the mat. But the syntax tells a different story. In (10a), I see a thing in a location; in (10b), I become aware, through the evidence of my eyes, of a fact. Importantly, I can become aware, through the evidence of my eyes, that something is not the case; hence the contrast in (11): (11) a. *I saw the cat not on the mat. b. I saw that the cat was not on the mat. Sentence (11b) is not synonymous with I did not see the cat on the mat, which could signify that perhaps the cat was there all along, only that I did not see it. What (11b) conveys is that I 'looked' (in the visual and/or extended sense of look) at the appropriate evidence, and drew the—for me—reasonable inference that the proposition 'the cat is on the mat' was not true.

34

JOHN R. TAYLOR

Incidentally, perception verbs differ with respect to their suitability as 'cognitive' verbs. If the difference between the sentences in (10) is rather subtle, the contrast in (12) is more transparent. (12) a. I heard you {play/playing} the piano. b. I heard that you played the piano. Example (12a) reports an auditory sensation, while (12b) reports an inference drawn (probably) from a heard linguistic utterance, rather than from heard music. The extension plausibly reflects the fact that a second important channel for gathering information about the world is other people's verbal utterances. In contrast, taste and smell can hardly be used at all as cognitive verbs, possibly reflecting the fact that these sense modalities are relatively unimportant for information gathering. (13) a. ?I smelt that you were cooking in the kitchen. b. *I tasted that you had put too much salt in the food. 2.

'Bootstrapping'

There is obviously much more that could be said about the syntax and the distributional characteristics of the three verbs discussed above.10 The picture that emerges, even from this minimal discussion, is that the meaning differ­ ences between the verbs are correlated with their linguistic properties; in other words, their meaning differences cannot be accessed except through their linguistic properties. We know that look, for instance, involves actively direct­ ing one's gaze at a scene because of its syntax: look requires a directional preposition, it can occur in the progressive, and it can co-occur with inten­ tionally. A word has the syntax that it has because of its meaning; conversely, the meaning of a word can be inferred on the basis of its syntax. This is the essence of Gleitman's (1990) 'syntactic bootstrapping' hypothesis. It is linguistic information which enables the learner to gain access to the meaning of a verb; the learner uses the syntax to 'bootstrap' the semantics. As I showed above, the linguist who attempts to explicate word meanings needs to appeal to essentially the same kind of information. We might push the argument a little further, and claim that linguistic context does not just provide a clue to word meaning, rather that linguistic context and word meaning are but two sides of the same coin. On this account, the learner acquires the meanings of words by learning the repertoire of their

MEANING AND CONTEXT

35

accepted uses, while the mature native speaker's knowledge of word meanings resides in knowledge which ranges over a word's syntactic and distributional characteristics. To be sure, this position may seem rather outrageous, because it flouts the well-entrenched role of the dictionary/lexicon in both folk and expert theories of language. Words have meanings (we believe), and these meanings are specified in the dictionary; that, after all, is what dictionaries are for. Complex expressions have meanings in virtue of the fact that words contribute their meanings to the meanings of complex expressions in which they occur. In order for compositionality to work, words have got to have meanings, and fairly rigid meanings at that, which are specifiable independently of the contexts in which they are used. To claim that word meanings emerge from their contexts of use, and can only be stated in terms of their contexts of use, undermines the very possibility of semantic compositionality. Two things should give us pause for thought, however. First, it may be queried whether full compositionality is indeed a typical, let alone essential, feature of complex expressions (see Langacker 1987: chap. 12). Secondly, native speakers of a language are generally quite at a loss to state what the meanings of words are (Johnson-Laird 1987:191). This, actually, is rather strange, if compositionality is to be a design feature of language. Bear in mind that it is only after considerable training in analytical and introspective tech­ niques that a linguist/lexicographer is able to come up with a statement of a word meaning at all (particularly if the word's meaning is opaque to definition by ostension). Ask a linguistically naive English speaker to state the difference between look and watch, or between speak and talk, and they will probably be quite baffled. (Yet they 'know' what the difference is—otherwise, they would not be able to use the words look and watch, speak and talk correctly.) In trying to characterize the meaning of verbs such as think (cf. note 5) or the difference between think, know, and believe, it is doubtful whether verbal definitions can take us very far at all here—so much so that Wierzbicka (1996) includes know and think in her list of (presumably innate) semantic primitives, which simply cannot be explicated.11 And what, after all, does it mean, to know the meaning of a word? What are word definitions, and what are they for? Goddard, in his introduction to Wierzbicka's program of definitional semantics, has this to say: We will assume that we are entitled to expect that an accurate definition will pre­ dict the appropriate range of usage of a word. Putting it another way, the reader [= the user o f a definition? JRT] should be able to trust that the definition is a reliable guide to how to use the word. (Goddard 1998:31)

36

JOHN R. TAYLOR

Quite so. A definition is a device for generating and evaluating uses. But language learners do not learn definitions; rather, they learn uses. And mature speakers of a language are able to use words appropriately, not because they have learned the definitions and how to apply them, but because they have learned how to use the words appropriately. Definitions, in short, are linguistic abstractions. Like the linguist's rules of syntax, definitions may purport to describe the linguistic data, and may even predict (more or less adequately) the range of potential uses. But it would be an erroneous move to suppose that syntactic rules and semantic definitions are necessarily implicated in language use. Gleitman's syntactic bootstrapping hypothesis is controversial. In a rebuttal of Gleitman's hypothesis, Pinker (1994) argues that Gleitman had overstated her case that meaning can be inferred solely from the linguistic envi­ ronment. Word learning—at least, early word learning—must be supported by non-linguistic context. If this point were to be denied, then a language comes to be regarded as nothing other than a vast calculus of system-internal relations, which makes no contact at all with (speakers' construals of) extralinguistic reality. Gleitman indeed acknowledged the role of extralinguistic context: Of course, one cannot converge on the unique construal of a verb from syntactic properties alone. Because the subcategorization properties are the syntactic expressions of their arguments, it is only those aspects of a verb's meaning that have consequences for its argument structure that could be represented syntac­ tically. Many - most - semantic distinctions are not formally expressed with this machinery. The role of the syntax is only to narrow the search-space for the meaning, as this latter is revealed by extralinguistic context. (Gleitman & Gillette 1995:424)

Note that in this passage, the authors are assuming a distinction between two facets of a verb's meaning: those aspects that determine a verb's argument structure (cf. infra) and a residue. Their idea is that the former aspect can be reliably inferred from syntactic properties, and that the latter must be filled out from extralinguistic context of use (cf. Fisher, Gleitman & Gleitman 1991). Elsewhere, I have argued against the validity of this bifurcation of word mean­ ing (Taylor 1996a). I would divide the labor differently. The extralinguistic context certainly gives crucial information about the general conceptual domain of a verb's semantics, e.g. that see and watch have to do with visual perception (in their 'basic' senses, at least), that talk and speak have to do with verbal activity, etc. But it is the linguistic context (which I would want to

MEANING AND CONTEXT

37

understand more broadly than syntactic subcategorization frames), which enables the learner to home in on the specifics of a verb's semantics.12 Pinker's alternative account is not without problems either. Pinker's theory has it that children, on the basis of an innate Universal Lexical Seman­ tics, somehow infer the meanings of words from the situations of their use, and then map the words' semantics onto the syntax, by (once again, innate) univer­ sal linking rules. Pinker is therefore proposing semantic bootstrapping, whereby the learner somehow gets hold of the meaning and then predicts the syntax. This might indeed be an option, if it were the case that all languages lexicalize much the same range of (innately available?) meanings, and that all languages encode these meanings by much the same syntactic devices. But, as Saussure pointed out long ago, and as even the most cursory cross-linguistic comparison confirms, a language's lexicon cannot be regarded as a nomen­ clature for a pre-existing list of concepts; neither can a syntactic category (such as 'Subject-of', 'Direct Object') be unproblematically transferred from one language to another. The learning of word meanings, in most cases, is a funda­ mentally linguistic process (Regier 1996:197-198). One problem with syntactic bootstrapping, as Gleitman (1990: fh. 10) acknowledges, might be the question of 'quirky' usage. Consider the following: (14) a. John saw his brother out of the room. b. John looked his uncle in the eye. If syntax is a clue to semantics, it might be inferred from (14a) that see is a verb of caused motion (cf. He pushed his brother out of the room), and from (14b) that look is a verb of contact (cf. He punched his uncle in the eye). Or take Goldberg's (1995:9) well-known example: (15) He sneezed the napkin off the table. Should we infer that sneeze in (15) is also a caused motion verb? Surely not. The caused motion component in (15) comes from the syntactic construction itself, not from the choice of verb. Still, not all verbs are eligible to occur in the caused motion construction, or in the contact construction (14b). The fact that look, but not see or watch, can be used as a contact verb (cf. the impossibility of *He saw his uncle in the eye, *He watched his uncle in the eye) is surely not unrelated to properties of look that were discussed earlier. The possibility of (14b) is therefore not quite as uninformative with respect to the verb's meaning

38

JOHN R. TAYLOR

as Gleitman suggests. Although I have no ready explanation to hand, I suspect that the availability of see, but not watch or look, in the caused motion construction (cf. the impossibility of *He looked his brother out of the room, *He watched his brother out of the room) may also cast light on the semantics of the three verbs. A more general point is, if linguistic context 'bootstraps' semantics, it is necessary to know the semantic value of the linguistic context. The question is relevant to my earlier discussion of see, look, and watch, which appealed to such things as the special 'directional' value of at, the value of the progressive and imperative, and so on. Concretely, in order to appreciate the directional value of look, one has to be aware of the directional value of at (in certain contexts of use, the most salient of which, probably, is ... with look)', and in order to draw appropriate inferences from the use of a verb with the progres­ sive, one has to have some feel for the special semantic value of the progres­ sive. If it is inappropriate to regard a language as a 'vast calculus of systeminternal relations' (to cite my earlier wording), it also behoves us to emphasize the status of a language as an interconnected network, in which the value of any particular bit is dependent in multiple ways on the values of many other bits. This, of course, was the message of structuralism, with its characterization of a language not just as a collection of signs but as a system of signs (cf. Saussure 1964:32). 3.

'Subcategorization', 'argument structure', and 'linking': The case of put

The 'standard' approach to many of the issues discussed above recognizes syntax and semantics as separate, autonomous components of the grammar, with linking rules mediating between them. Sub categorization—the kinds of complements that a certain word (typically a verb) may, or must occur with— is a syntactic phenomenon. Argument structure is a semantic notion, having to do with semantic relations between terms and a relational item (again, typically a verb). Linking rules interface between the two levels, by associating argu­ ments with syntactic constituents. Selection restrictions are of ambiguous status vis-à-vis syntax and semantics, having to do with the semantic type of syntactic constituents, for example, the fact that a verb takes only certain kinds of NP complements. A general assumption, in much mainstream work on lexical semantics has been that argument structure and subcategorization frames, although they belong to different 'levels' of the grammar, are nevertheless interdependent,

MEANING AND CONTEXT

39

otherwise 'bootstrapping' (either from the syntax to the semantics, à la Gleitman, or from the semantics to the syntax, à la Pinker) could not work. But most scholars have insisted on the need to keep the two levels distinct. Support for two distinct levels could come from words whose syntactic and semantic properties are out of synch. Such words would clearly present a severe problem for the thesis that meaning and linguistic context are but two sides of the same coin. How many such words there might be in any given language is not clear. It is also not clear how many such cases are required in order to convincingly substantiate the autonomy of syntax vis-à-vis semantics; would one sole example be enough? At the same time, it must also be recog­ nized that alleged cases of divergence between syntax and semantics may simply reflect an inadequate semantic and/or syntactic analysis. One case that I want to look at is the English verb put, in contrast to other verbs of caused motion, such as move, throw, push, etc. I have selected this example because of the extended discussion it received in Wilkins (1993/4). As with the verbs of visual perception, it is difficult to imagine how the concepts of PUTTING and MOVING could be explicated by ostension. Moving something necessarily means putting that thing in a new location; conversely, putting something somewhere necessarily implies moving it.13 Yet, put and move are felt to be very different in meaning. Corroborating this intuition is the fact that the two verbs are used very differently, so much so, in fact, that it turns out to be quite difficult to find a sentence in which move and put are interchangeable. A possible example is the following: (16) a. Put the bookcase into my study. b. Move the bookcase into my study. Even here, there is a subtle difference: in (16a), into could be replaced by in (indeed, to my mind, (16a) sounds better with in than with into), whereas in (16b), into is not easily replaced by in. Let us look, first, at the subcategorization frames of the verbs. The verb put absolutely requires both a direct object NP and a locative phrase. (17) Put the book {on the table/here/away/somewhere else}. Neither put the book (omitting the locative) nor put on the table (without speci­ fying what is to be put on the table) are grammatical in English, and no amount of contextualization can legitimate these expressions. Put therefore has the following subcategorization frame:

40

JOHN R. TAYLOR (18) put [

NPThemePPGoal]

The NP designates the thing that changes location (the Theme); the PP desig­ nates the location where it ends up (the Goal). As I mentioned, both constitu­ ents are absolutely essential to syntactic well-formedness. Another remarkable fact is that put does not allow specification of other aspects of the Theme's trajectory, such as Source (where the Theme was moved from) or Path. With from the floor and up the stairs specifying a Source and a Path, respectively, the following are ungrammatical.14 (19) a. *Put the book from the floor onto the shelf. b. *Put the bookshelf up the stairs into the study. Other caused motion verbs, such as throw, push, and move, are much less restrictive, in that they can appear without a locative expression {Throw the ball! Push the cart! Move the piano!). Some, such as push, can be used without any complement at all (Push!). Moreover, all these verbs are fully compatible with an (optional) expression of Source and Path, and even of Location (i.e. the Place where the designated activity occurs), as well as any combination of these. Move, then, has the subcategorization frame in (20). (Parentheses indi­ cate optionality; the subscripts mark the semantic roles of the constituents.) (20) move [

N P T h e m e (PP S o u r c e ) (PP p a t h ) (PP G o a l ) ( P P L o c a t i o n ) ]

Wilkins (1993/4) sees here a major problem for language acquisition, in that the syntactic behavior of put presents us with a severe case of the 'negative evidence problem' (Bowerman 1987). Put is a caused motion verb, but its syntax is much more constrained than that of all the other caused motion verbs. Why does the learner, then, not generalize from these other verbs, and construe put with expressions of Source, Path, and even Location? In this connection, Wilkins emphasizes the 'language-dependent lexicalization fact' (1993/4:40) about English put, pointing out that the restriction against stating the Source does not apply in Hungarian. (Parentheses in the following example indicate that the constituent in question is optional.) (21) Ben (a polcról) az asztalra tette a kulcsot. Ben (the shelf:from) the desk:onto put:PAST the key:ACC "Ben put the key (from the shelf) onto the desk."

MEANING AND CONTEXT

41

The observation is valid, only to the extent that Hungarian tesz is semantically equivalent to English put. My browsings in Hungarian dictionaries suggest that tesz may be a rather general change-of-state verb, not at all equivalent to English put. I think that Wilkins greatly exaggerated the learning problem of put. The puzzle largely dissipates once we consider the linguistic contexts in which put is used and the semantic inferences that can be drawn from these. Since put invariably occurs with both Theme and Goal, it is apparent that the semantic particularity of the verb is a focus on getting the Theme into a location, and not at all on getting the Theme away from some place, or simply causing it to move along some trajectory. Naturally enough, the Goal will tend to be construed as the place where the Theme 'belongs', 'where it should be', or 'where the speaker wants it to be'; otherwise, why focus on getting the Theme into that place at all? This aspect of the verb's meaning will surely be apparent to any child who has been subjected to innumerable requests 'to put your things away!' Consider the following contrast: (22) a. Move your toys away! b. Put your toys away! Example (22a) is a request to get the toys someplace else—where they go is immaterial, as long as they go from where they are now; away is largely redun­ dant, and could easily be omitted. Example (22b) very strongly conveys that the toys should go to where they belong (e.g. in the closet, or wherever), not just that the toys should go from where they are now. The idea that put denotes that things should go to where they belong (either to their 'canonical' location, or to a place contingent on present concerns) manifests itself in innumerable turns of phrase. The canonical loca­ tion of an article of clothing is on the body of the wearer—that is what clothes are for. Hence Put your clothes on is acceptable, but Put your clothes off is not. Similarly, Put the stamp on the envelope (again, that is what stamps are for) or Put the letter in the envelope (that is what envelopes are for) are acceptable, but Put the stamp off the envelope or Put the letter out of the envelope are not. Put it down strongly conveys 'don't handle it'; the thing 'belongs' on the floor, on the table, or wherever its proper resting place may be. One can 'put the pieces together' (in the sense 'assemble'), but one cannot 'put the jigsaw apart'. One can 'put a tent up' (in the sense 'erect'), but one cannot 'put a tent down'. The same pattern underlies extended uses of put which do not necessarily involve

42

JOHN R. TAYLOR

movement at all, as in Put the situation right, in contrast to the impossibility of Put things wrong. Put is amongst the most frequent verbs in child speech (Clark 1978), also in child-directed speech.15 From the very beginning, children seem to have a feel for the semantic specifics of the verb. Here is 'Eve', at age 1;7 (23), and at 1;9(24):16 (23) EVE: MOTHER:

(24) EVE:

I sock put on. Oh you're putting on your sock? That's the dollie's sock. Put sugar in.

Here, the child has clearly grasped that put requires both a Theme complement, and a locative Goal complement. Subcategorization 'errors', v/hich are actually rather few in the CHILDES corpora that I have surveyed, consist mainly in the child omitting one of the two obligatory complements. In (25)—Eve at 1;8— the Goal argument in missing (Mummy appears to be a Beneficiary, not a Goal). In (26)—Shem (Clark 1978) at 2;2,16—the Theme is missing. (But notice in both cases how the adult immediately 'corrects' the child's utterance, by supplying the missing argument.) (25) EVE: MOTHER:

(26) SHEM:

Eve put blocks Mommy. Yes. You help me put the blocks away.

INTERVIEWER:

I wantuh put away. Want to put it away?

SHEM:

Yeah.

A more drastic error is the following (cited, without source, in Peccei 1994:87): (27) [Child has been tidying up] I putted my room all clean. Here, it looks as if the child has got both the subcategorization frame and the argument structure all wrong (not to mention the morphology). The direct object my room is not an entity that is caused to move, and all clean does not designate a locative Goal. But the child certainly has grasped the idea that put involves arranging things 'as they should be'. Goldberg (1998:207) takes put to be the prototype of a caused motion verb; she claims that the semantics of this high frequency 'light' verb matches the schematic meaning of the caused motion construction [NP V NP PP] (see

MEANING AND CONTEXT

43

also Goldberg 1995). This may be true to the extent that caused motion expres­ sions prototypically focus on getting things into a place where they belong. But most caused motion verbs do not have this property; they merely designate that a thing is made to move, often in a particular manner, but are generally rather indifferent as to whether motion is specified with respect to Source, Path, or Goal, or even at all. Considering the broad class of caused motion verbs {throw, push, pull, move, etc.), put may be somewhat untypical. To conclude this section, I return to an issue raised by (16). Compare: (28) a. Put the books {on/onto} the bookshelf. b. Move the books onto the bookshelf.  Move the books on the bookshelf The use of put vs. move interacts with a curious (and, as far as I am aware, little studied) aspect of English. This is the contrast between on and onto. (There is a parallel contrast between in and into). As is well-known, on (and in) can be ambiguous between a Place and a Goal interpretation, whereas onto (and into) are exclusively Goal prepositions. If a person 'fell in the water', this could mean that the person was already 'in the water', and while in the water, they fell. Alternatively, it could mean that they fell, and ended up 'in the water'. This latter meaning could be made explicit by the use of into. In and on are not always ambiguous in this way. (Equally, not all uses of into and onto can be replaced by the simple prepositions.) On in (28c) desig­ nates a Location; the books are to be moved around 'on the bookshelf. To get the Goal reading, it is necessary to use directional onto, as in (28b). On the other hand, with put, either on or onto may be used; both have the Goal read­ ing, though on is no doubt preferred. The reason for this, presumably, has to do with the fact that put requires a directional, Goal-oriented PP; on the bookshelf (28a) gets a directional reading by coercion from the verb. The contrast between on and onto is effectively neutralized, such that the 'unmarked' term can be used. With move, the contrast is a real one, and with real semantic consequences. 4.

Conclusion

In this homage to Günter Radden, I have taken a number of speculative liber­ ties. (I should like to think that this is a feature of the genre, and I should hope that the congratulee will be sympathetic.) Certainly, my claim of an intimate association between linguistic context and word meaning needs much more

44

JOHN R. TAYLOR

substantiation than was possible within the confines of this contribution. But if the association is as close as I believe it is, there are important implications for the theory and practice of semantic analysis (and, indeed, linguistic analysis more generally). On the one hand, a close association requires us to explicate a word's meaning by reference to the linguistic contexts in which a word is used; on the other hand, it entitles us to analyze the range of contexts in which a word is used as a reflex of a word's meaning. Moreover, linguistic context will be a crucial source of semantic information for the language learner. In asserting the close association between meaning and linguistic context, I am not denying the role of extralinguistic context in determining the appro­ priateness of a word's use, nor as an aspect that needs to be considered in elucidating word meaning, nor, indeed, as a factor that facilitates word learn­ ing. Even so, the role of extralinguistic context in the learning and explication of word meaning is easily exaggerated. In fact, once a learner has gained even a minimal proficiency in a language, it is surely the language itself that, increas­ ingly, creates the context for further language learning. On the view put forward here, linguistic context is understood rather broadly, to include much more than simply the subcategorization frames in which words (e.g. verbs) occur. Likewise, meaning is understood much more broadly than knowledge of argument structure. The association between linguistic context and meaning therefore comprises much more than the 'link­ ing' of argument structure to subcategorization frames. Linguistic context, to my mind, must comprise collocational preferences, occurrence in construction types, compatibility with various moods and modalities, as well as (though this was not addressed in the paper) availability for various derivational morpho­ logical processes. Likewise, 'meaning' is understood as encyclopedic in scope, in the manner argued by Langacker (1987) and many others. We need look no further than to the contrast between put your clothes on and *put your clothes off, to substantiate this claim. The relation between meaning and linguistic context is a two-way street. To be sure, the linguistic context bootstraps the semantics; but for this to be possible, the semantic value of the linguistic context has to be appreciated. Equally, the semantics predicts the linguistic context; but the range of available linguistic contexts, in any given language, is itself circumscribed by the language system, i.e. the language-specific conventions for the association of forms and meanings. Paradoxical as it might seem, word meanings are ency­ clopedic in scope, but at the same time they are quintessentially linguistic enti­ ties, and the learning of word meanings is a quintessentially linguistic process.

MEANING AND CONTEXT

45

Notes 1 Cf. Noble and Davidson (1996), who argue that the emergence of symbolic ability was the key event in the evolution of language. 2 'Concepts' are tricky entities to define. Although the point would need to be elaborated, I take 'concepts' to be principles of categorization (cf. Komatsu 1992). See the recent statement by Fodor and Lepore (1996:255-256): "It's common ground that concepts are the kinds of things that apply to things (that is, they're the kinds of things that can function as categories) and that it is constitutive of the identity of a concept that it applies to the things that it does." 3 Langacker (1993:3) speaks of the 'conceptual archetype' of a 'thing'. 4 See Markman (1990) for empirical support that such a predisposition is in fact the case. 5 In case the reader is not convinced by the examples see, look, and watch, consider the verb think, especially in comparison and contrast with believe and know. There is no way in which one could elucidate (let alone teach) the meaning of think—even less, distinguish think from believe and know—by pointing to thinking-kinds of events. This raises the intriguing question how language users do learn the meaning of think. Come to that, what does think mean, any­ way? 6 Curiously, the one directional preposition that is problematic with look is to. You can 'look to your left', but you cannot 'look to the picture'. 7 See and See this are certainly possible utterances, but such cases are not imperatives, but rather elliptical questions: (Do you) see? (Do you) see this? 8 There are, of course, certain circumstances in which even stative verbs can be used with the progressive: I'm seeing more and more detail in this painting each time I look at it; You 'll be seeing better when you get your new glasses, etc. These uses imply a change in state, and thus give a temporal contour to the designated situations. Also, in the extended sense of 'meet', see is quite compatible with certain uses of the progressive: I'm seeing her tomorrow. On the semantics of the English progressive, see Taylor (1996b: 268). 9 As mentioned in the following footnote, see can also function as a verb of perceptual achievement: Suddenly, I saw it! As such, the verb is also incompatible with the progressive. 10 Points for further discussion include the extension of look to a verb of appearance, whereby look takes as its grammatical subject the perceived entity (It looked red to me), or even 'imper­ sonal' it (It looks like we're in trouble); the use of see in the sense 'meet with'; and a certain ambivalence in perceptual see, between a verb of perceptual ability, and a verb of perceptual achievement. On the latter, see Jackendoff (1983:150-151), and Fisher, Gleitman and Gleitman (1991: 360). 11 Note that an appeal to innateness—whether in the shape of semantic primitives (Wierzbicka 1996), or as determined by a 'Universal Lexical Semantics' (Pinker 1994), or even as complex semantic gestalts (Fodor 1980)—does not solve the learnability problem, as Gleitman addressed it. Even if meanings were somehow available, prelinguistically, to the learner, the learner still has the task of finding out which meaning, out of those that are available, is to be associated with which lexical item. It can only be the context of a word's use that gives a clue as to which of the innately available meanings is to be matched up with which word. 12 For a fine example of how verb meanings can (must?) be differentiated on the basis of their distributions, see Dirven et al. (1982), on speak, talk, say, and tell. 13 I assume only the transitive uses of the two verbs. Move, of course, has intransitive uses (The animal didn 't move, The box won't move) that put lacks.

46

JOHN R. TAYLOR

14 The sentences are, of course, acceptable, if from the floor and up the stairs are modifying phrases. 15 Drawing on the resources of the CHILDES database (MacWhinney 1995), I discovered that for 'Adam' (Brown 1973), put ranked as his fourth most frequent verb, after want, do, and have. In the mother's speech directed to Adam, put came in after do, don't, have, did, are, and think. For 'Eve' (Brown 1973), put ranked as the third most frequent verb in her speech, com­ ing after go and have, while in the mother's speech directed to Eve, put logged in after is, are, do, don't, and have. 16 Examples (23) to (26) are cited from the CHILDES database. Orthography and punctuation have been slightly regularized.

References Bowerman, Melissa. 1987. "The 'No Negative Evidence' Problem: How Do Children Avoid Constructing an Overly General Grammar?" Explaining Language Universals ed. by John A. Hawkins, 73-101. Oxford: Blackwell. Brown, Roger. 1973. A First Language: The Early Stages. Cambridge, Mass.: Harvard University Press. Clark, Eve. 1978. "Discovering What Words Can Do". Papers from the Parasession on the Lexicon ed. by Donka Farkas, Wesley M. Jacobsen & Karol W. Todrys, 34-57. Chicago, Ill.: Chicago Linguistic Society. Dirven, René, Louis Goossens, Yvan Putseys & Emma Vorlat. 1982. The Scene of Linguistic Action and its Perspectivization by Speak, Talk, Say and Tell. Amsterdam: John Benjamins. Fisher, Cynthia, Henry Gleitman & Lila Gleitman. 1991. "On the Semantic Content of Subcategorization Frames". Cognitive Psychology 23.331-392. Fodor, Jerry. 1980. "The Present Status of the Innateness Controversy". Repre­ sentations: Philosophical Essays on the Foundations of Cognitive Science, 257-316. Cambridge, Mass.: The MIT Press. & Ernest Lepore. 1996. "The Red Herring and the Pet Fish: Why Concepts Still Can't Be Prototypes". Cognition 58.253-270. Gleitman, Lila. 1990. "The Structural Sources of Verb Meaning". Language Acquisition 1.3-55. & Jane Gillette. 1995. "The Role of Syntax is Verb Learning". The Handbook of Child Language ed. by Paul Fletcher and Brian MacWhinney, 413-427. Oxford: Blackwell. Goddard, Cliff. 1998. Semantic Analysis: A Practical Introduction. Oxford: Oxford University Press.

MEANING AND CONTEXT

47

Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago, Ill.: The University of Chicago Press. . 1998. "Patterns of Experience in Patterns of Language". The New Psychology of Language: Cognitive and Functional Approaches to Language Structure ed. by Michael Tomasello, 203-219. Mahwah, N.J.: Lawrence Erlbaum. Jackendoff, Ray. 1983. Semantics and Cognition. Cambridge, Mass.: The MIT Press. Johnson-Laird, Philip N. 1987. "The Mental Representation of the Meaning of Words". Cognition 25.189-211. Komatsu, Lloyd. 1992. "Recent Vews of Conceptual Structure". Psychological Bulletin 112.500-526. Langacker, Ronald, W. 1987. Foundations of Cognitive Grammar, Vol. 1, Theoretical Prerequisites. Stanford, Calif: Stanford University Press. . 1993. "Reference-point Constructions". Cognitive Linguistics 4..1-38. Markman, Ellen. 1990. "Constraints Children Place on Word Meanings". Cognitive Science 14.57—77. MacWhinney, Brian. 1995. The CHILDES Project: Tools for Analyzing Talk. Hillsdale, N.J.: Lawrence Erlbaum. Noble, William & Iain Davidson. 1996. Human Evolution, Language and Mind: A Psychological and Archaeological Inquiry. Cambridge: Cambridge University Press. Peccei, Jean Stilwell. 1994. Child Language. London: Routledge. Pinker, Steven. 1994. "How Could a Child Use Verb Syntax to Learn Verb Semantics?" Lingua 94.377-410. Quine, Willard Van Orman. 1953. "Two Dogmas of Empiricism". From a Logical Point of View, 20-46. Cambridge, Mass.: Harvard University Press. Regier, Terry. 1996. The Human Semantic Potential: Spatial Language and Constrained Connectionism. Cambridge, Mass: The MIT Press. Rosch, Eleanor. 1978. "Principles of Categorization". Cognition and Categori­ zation ed. by Eleanor Rosch and Barbara. B. Lloyd, 27-48. Hillsdale, N.J.: Lawrence Erlbaum. Saussure, Ferdinand de. 1964. Cours de linguistique générale. Paris: Payot. Taylor, John R. 1996a. "On Running and Jogging". Cognitive Linguistics 7.2134. . 1996b. Possessives in English: An Exploration in Cognitive Gram­ mar. Oxford: Clarendon Press.

48

JOHN R. TAYLOR

Wierzbicka, Anna. 1996. Semantics: Primes and Universals. Oxford: Oxford University Press. Wilkins, Wendy. 1993/4. "Lexical Learning by Error Detection". Language Acquisition 3.121-157. Wittgenstein, Ludwig. 1958. The Blue and Brown Books. Oxford: Blackwell.

LEXICAL RULES VS. CONSTRUCTIONS A FALSE DICHOTOMY WILLIAM CROFT University of Manchester

Verbs appear to change their meaning when put into particular constructions. For example, bake means 'bake x and give x to someone' in the ditransitive construction. Two alternative analyses have been proposed: to derive the meaning of bake in the ditransitive construction by a lexical rule, or to have the meaning of bake in the ditransitive construction be predictable from the semantics of the construction. An analysis of the behavior of different verb classes in the ditransitive construction indicates that both analyses are partly right. The ability to enter the ditransitive construction is verb-class-specific, or even verb-specific, but the meaning of the verb (class) found in the ditransitive is peculiar to the ditransitive construction and clearly involves a family resemblance to the meaning of other verb classes found in the construction. I argue that the simplest way to capture these facts is to represent verb-class-specific construc­ tions and verb-specific constructions in the grammar of English.

1.

Introduction1

Consider the following two examples: (1) Tess baked a cake. (2) Tess baked Bill a cake. Example (1) describes an act of creating a food item through a certain process. This is usually analyzed as the meaning of the verb bake in that sentence. Example (2) describes the same act of creation but also describes an intention to subsequently transfer possession of the cake to Bill. The question is, where does this additional component of meaning come from? How should example (2) be analyzed? The traditional approach—traditional in generative grammar and its off­ shoots, that is—is the lexical rule approach. The additional component of

50

WILLIAM CROFT

meaning in (2) is analyzed as part of the meaning of the verb bake and other verbs of creation like it. Thus, bake in (2) does not mean the same thing as bake in (1), and so there must be a rule deriving bake in (2) from bake in (1), which allows it to be used with the appropriate meaning in the construction in (2). The lexical rule approach has been developed in greatest detail by Levin and Rappaport Hovav (see, for example, Levin & Rappaport Hovav 1995). A more recent approach, associated with the cognitive linguistic tradition, is the constructional approach. The additional component of meaning in (2) is analyzed not as a part of the verb meaning, but as a part of the meaning of the ditransitive construction. The ditransitive construction is an independently represented syntactic unit in its own right, and possesses its own semantics as well as its syntax. In particular, the ditransitive construction implies the presence of transfer of possession in some way or another in (almost) all of its occurrences, as in (3)-(5): (3) She gave him a book. (4) She bequeathed him her art collection. (5) She threw him the ball. The constructional approach is central to cognitive linguistic models of syntax (Langacker 1987, 1991; Lakoff 1987; Fillmore & Kay 1993), but has been most vigorously defended by Goldberg (1995). Which approach gives us the right analysis of (2)? In this paper, I will argue that it is neither, or perhaps both. The lexical rule-construction debate revolves around a false dichotomy. Although constructions exist, as cognitive linguists have argued, the relation between constructions (and their meanings) and their component words (and their meanings) is not as neat or as clear as the lexical rule-construction debate implies. Language once again successfully resists the attempt of linguists to make it neat and clean. 2.

Arguments against lexical rules

Goldberg presents a number of arguments against a lexical rule approach to argument structure alternations (Goldberg 1995:9-23). Goldberg's arguments are briefly summarized in the remainder of this section. (i) Lexical rules require the positing of 'implausible' verb meanings. For example, the sentences in (6) and (7) would require a derived meaning of sneeze as 'X CAUSES Y to MOVE Z by sneezing' and bake as 'X INTENDS to CAUSE Y to HAVE Z' (Goldberg 1995:9).

LEXICAL RULES vs. CONSTRUCTIONS

51

(6) He sneezed the napkin off the table. (7) She baked Bill a cake. [= (2)] However, we find no language in which there is a verb that means 'X INTENDS to CAUSE Y to HAVE Z by BAKING' distinct from the verb meaning 'X BAKES Y' (Goldberg 1995:10). (ii) Texical rules are required because of the assumption in many syn­ tactic theories that "the verb determines how many and which kinds of complements will co-occur with it" (Goldberg 1995:11; cf. Jackendoff 1997: 557). This argument for lexical rules is circular. (iii) It is more parsimonious to attribute to constructions the semantic properties that would have to be assigned to different verb senses in the lexical rule approach. (iv) Semantic compositionality is affected by the constructions in which verbs (and other words) occur. Hence a version of semantic compositionality can be preserved by assuming that verb meanings combine with constructional meanings in a regular fashion. (v) A psycholinguistic experiment yielded different reaction times for verbs with different senses vs. verbs with the same sense (in the constructional analysis) but occurring in different constructions (Carlson & Tanenhaus 1988). That is, there were significant differences in reaction times for the pair of sentences in (8a-b) (distinct senses) than for the pair in (9a-b) (same sense, dif­ ferent constructions): (8) a. Bill set the alarm clock onto the shelf. b. Bill set the alarm clock for six. (9) a. Bill loaded the truck onto the ship. b. Bill loaded the truck with bricks. (vi) Syntactic bootstrapping in language acquisition is really acquisition of constructional meaning. The syntactic bootstrapping hypothesis is that chil­ dren acquire verb meanings by exploiting their occurrence in syntactic constructions (Landau & Gleitman 1985). Goldberg argues that the reason that children can acquire verb meanings from the syntactic contexts in which they occur is that those syntactic contexts are constructions with meanings in their own right; the constructional meanings are what children are using in iden­ tifying the word meanings.

52

WILLIAM CROFT

Arguments (i)-(vi) are arguments in favor of a constructional analysis over a lexical rule analysis. Arguments (vii)-(x), given below, are based on arguments in favor of a lexical rule analysis over a syntactic transformational analysis. Goldberg argues in each case that the reasons for choosing a lexical rule analysis over a transformational analysis do not rule out a constructional analysis, and indeed there is sometimes evidence favoring a constructional analysis. (vii) What appear to be lexical restrictions on verbs' argument structure are better described as constraints conferred by the construction itself. For example, in (10a-b) the same verb can be either acceptable or unacceptable in the ditransitive construction (Goldberg 1995:22, taken from Langacker 1991): (10) a. Joe cleared Sam a place on the floor. b. *Joe cleared Sam the floor. Thus, one cannot say that there is a lexical rule (dis)allowing clear. Instead, the ditransitive construction requires that the first object 'receive' the second object; this is acceptable for a place on the floor, but not for the floor itself. (viii) Lexical rules allow for the restriction of idiosyncratic information to the lexicon. However, there is evidence that idiosyncratic information is associated with phrases and other syntactic units; in fact, this is the primary argument for the existence of constructions as independent units of grammatical knowledge. Hence the idiosyncrasy of argument structure alternations is not an argument against the constructional analysis. (ix) In other languages, there are morphological affixes on verbs that differentiate verbs that occur in the two argument structure constructions, although in English the 'same' verb occurs in both contexts. For example, compare the Russian verbs to the English translations in the following sentences (Croft 1991:201, from Comrie 1985:314): (11) Iva n se -jet pšenic -u v pol -e. John sow.IMPF -3SG.PRES wheat -ACC in field -LOC2 "John sows wheat in the field." (12) Ivan zase -iva -jet pol pšenic John DERIV-SOW-IMPF-3SG.PRES field(ACC) wheat "John sows the field with wheat."

-ej. -INST

LEXICAL RULES vs. CONSTRUCTIONS

53

The Russian sentences have an argument structure alternation equivalent to that found in their English translations, but "sow" in (12) is morphologically derived. Goldberg argues that a morphological affix plus the verb class it combines with is itself a construction (e.g. for the causative; cf. Goldberg 1995:61), so such examples can be handled in a constructional account as well. (x) Verb forms that are derived from argument structure alternations, such as the passive, can be the input to further word derivational processes, such as that which yields the adjectival passive. However, there is much evidence (see references in Goldberg 1995:23) that the line between lexical and syntactic rules is not at all clearcut, and so this does not count as an argument against a constructional analysis. On the whole, these arguments (along with arguments in the other construction grammar sources cited above) convincingly demonstrate that constructions are a necessary part of the model of syntactic representation for a speaker's knowledge of their language, and that constructions have meanings. What is at issue here is the relationship between constructional meaning and verb meaning in argument structure constructions. In order to explore this issue, we turn to Goldberg's analysis of a specific construction, the ditransitive construction. 3.

The grammatical representation of the ditransitive construction

The English ditransitive construction is usually described as [SBJ VERB OBJ1 OBJ2], and is illustrated in its most common form in (13): (13) Julie gave Simon a book. Goldberg argues that the ditransitive construction contributes the meaning that "the agent ... acts to cause transfer of an object to a recipient" (Goldberg 1995:32). This meaning is not necessarily part of the meaning of the verb. As we saw with example (2) above, the basic meaning of bake does not involve transfer of possession; this semantic component is present only when bake occurs in the ditransitive construction. Goldberg notes, however, that in the case of bake and other verbs of creation, it is in fact intended, not actual, transfer of possession that is contrib­ uted by the ditransitive construction. Whereas it is an entailment of (13) that Simon comes into possession of the book, it is possible in (2) that Bill does not receive the cake. For example, Tess could bake Bill the cake, but then get

54

WILLIAM CROFT

angry at him and throw the cake away. Hence the contribution of the ditransitive construction is slightly different for give and for bake. Goldberg resolves this problem by proposing that constructions can be polysemous. Polysemy is the phenomenon by which a single form has two meanings that are semantically related. Typically, one meaning is a historical extension of the other meaning. Synchronic polysemy, however, also implies that the speaker internally represents the two meanings as semantically related (Croft 1998b). For example, the ditransitive construction has two related constructional meanings: actual transfer of possession and intended transfer of possession. The former is found with give, the latter with bake. There is good motivation for constructional polysemy. Other construc­ tions have been found to be polysemous (e.g. Michaelis & Lambrecht 1996; Jackendoff 1997). Since constructions are essentially schematic, multi-word 'lexical items', one would expect to find similar properties among construc­ tions as among lexical items, and lexical items are frequently, perhaps univer­ sally, polysemous. There seems to be no reason to posit two independent ditransitive constructions, one with a meaning of actual transfer of possession, the other with a meaning of intended transfer of possession. Instead, it seems quite plausible to posit a single ditransitive construction with two related constructional meanings. In fact, Goldberg lays out an analysis in which there are not two but six related constructional meanings for the ditransitive, five of which are exten­ sions of the first, central sense (Goldberg 1995:38, Figure 2.2). These meanings are associated with verb classes as given in (14): (14) A. Central sense: agent successfully causes recipient to receive pa­ tient 1. Verbs that inherently signify acts of giving: give, pass, hand, serve, feed,... 2. Verbs of instantaneous causation of ballistic motion: throw, toss, slap, kick, poke, fling, shoot,... 3. Verbs of continuous causation in a deictically specified direc­ tion: bring, take,... B. Conditions of satisfaction imply that agent causes recipient to re­ ceive patient 1. Verbs of giving with associated satisfaction conditions: guarantee, promise, owe,...

LEXICAL RULES vs. CONSTRUCTIONS

55

. Agent causes recipient not to receive patient 1. Verbs of refusal: refuse, deny D.Agent acts to cause recipient to receive patient at some future point in time 1. Verbs of future transfer: leave, bequeath, allocate, reserve, grant,... E. Agent enables recipient to receive patient 1. Verbs of permission: permit, allow F. Agent intends to cause recipient to receive patient 1. Verbs involved in scenes of creation: bake, make, build, cook, sew, knit,... 2. Verbs of obtaining: get, grab, win, earn,... If we examine the ditransitive construction more closely, it appears that the variation in the ditransitive construction's meaning is not true polysemy. Each verbal semantic class is associated with only one sense of the ditransitive construction. It is not an accident that the verbs found with ditransitive sense E, 'agent enables recipient to receive patient', are verbs of permission, or that the verbs found with sense C, 'agent causes recipient not to receive patient', are verbs of refusal. That is, the modulation of the possessive relation specified by each constructional sense—actual, enabling, and negative transfer of posses­ sion—matches a semantic component of these verbs. Further evidence for this is the existence of at least one other sense of the ditransitive construction, required by one class of verbs: (15) a. That vase cost him $300. b. That painting set him back $500. (16) G. Acquisition of goods causes recipient to no longer have posses­ sion of money 1. Verbs of costing: cost, set back,... There may be other senses of the ditransitive construction lurking in the lexicon of English.

56

WILLIAM CROFT

Moreover, even the verb classes that do not inherently specify a modula­ tion of the transfer of possession relationship occur with only one 'sense' of the polysemous ditransitive construction: for example, verbs of creation are found only with the intended transfer of possession 'sense'. Goldberg points out that this is evidence against a monosemous analysis of the ditransitive construction, that is, an analysis such that the ditransitive construction simply means 'trans­ fer of possession' without specifying the modulation of this event (Goldberg 1995:36-37). If the ditransitive construction were truly polysemous, one might expect that the verb bring, for example, would be found with ditransitive sense F, resulting in a meaning like 'X brings Z with the intention of causing Y to receive Z', or kick could also occur with ditransitive sense C, resulting in a meaning like 'X kicks Z causing Y not to receive Z'. But we do not. Instead, it seems that the different 'senses' of the ditransitive construction are very closely tied to the verb classes that each 'sense' occurs with. In other words, a proper representation of the construction schema for each 'sense' of the ditransitive construction must specify which verb classes occur with it. That is, the proper representation of an English speaker's knowl­ edge of the ditransitive construction is not the polysemous representation given in (17): (17) A. [[SBJ VERB B. [[SBJ VERB C. [[SBJ VERB D. [[SBJ VERB

OBJ1 OBJ1 OBJ1 OBJ1

OBJ2]/[actual XPoss]] OBJ2]/[conditional XPoss]] OBJ2]/[negative XPoss]] OBJ2]/[future XPoss]]

E. [[SBJ VERB OBJ1 OBJ2]/[enabling XPOSS]]

F. [[SBJ VERB OBJ1 OBJ2]/[intended XPoss]] G. [[SBJ VERB OBJ1 OBJ2]/[depriving XPoss]] In this representation, there is a single syntactic schema that has six (or more) different but related meanings associated with it (I use the format of Langacker 1987, where [[form]/[meaning]] represents a conventional symbolic unit; and I abbreviate 'transfer of possession' as 'XPoss'). Instead, there is a distinct syntactic schema for each constructional 'sense' specifying the verb classes found with each meaning, with correspon­ ding specific meaning. The syntactic structures and (informal) semantic struc­ tures are given in (18):

LEXICAL RULES vs. CONSTRUCTIONS

57

(18) A. [[SBJ GIVING.VERB OBJ1 OBJ2]/actual XPoss]] [[SBJ BALL.MOT.VERB OBJ2 OBJ2]/[actual XPoss via ballistic motion]] [[SBJ DEIC.CAUS.VERB OBJ1 OBJ2]/[actual XPoss via deictic caused motion]] B. [[SBJ COND.GIVING.VERB OBJ1 OBJ2]/[conditional XPoss]]  [[SBJ REFUSE.VERB OBJ1 OBJ2]/[negative XPoss]] D. [[SBJ FUT.GIVING.VERB OBJ1 OBJ2]/[future XPoss]] E. [[SBJ PERMIT.VERB OBJ1 OBJ2]/[enabling XPoss]] F. [[SBJ CREATE.VERB OBJ1 OBJ2]/[intended XPoss after creation]] [[SBJ OBTAIN.VERB OBJ1 OBJ2]/[intended XPoss after obtaining]] G. [[SBJ COST.VERB OBJ1 OBJ2]/[depriving XPoss via paying]] Of course, there is a family resemblance in both the form and the meaning of the ditransitive constructions in (18): all involve a modulated transfer of possession, and all have two object phrases and a verb. But the verb classes used for each construction are mutually exclusive, and the modulation of the transfer of possession differs for each construction type. The semantic type of the second object NP also varies significantly: the latter two constructions of sense A require a mobile physical object NP, while senses C-E do not; sense G requires a price NP, etc. The subject and first object NPs are human NPs, of course. Just as semantically based verb classes are involved in constructional representations, so are semantically based noun classes. Adopting properly specified argument types will make even clearer how distinct each of the constructions in (18) actually are. It is not clear from Goldberg (1995) whether she adopts the construc­ tional representations in (17) or in (18). On the one hand, Goldberg argues forcefully for constructional polysemy (Goldberg 1995:31-39), and accepts the need to account for which verb classes go with which 'sense' of the ditransitive construction (Goldberg, pers. comm.); these imply the representations in (18). On the other hand, the syntactic representations in Goldberg (1995) indicate only the general verb category V for the construction (Goldberg 1995:3, 53, 54, 55, 77, 91, 93, 96, 142, 145). In many of these representations, the meaning is expressed simply as 'CAUSE-RECEIVE—PRED'. Goldberg represents both sense A (actual transfer) and sense F (intended transfer) using only the general verb category V and its semantic counterpart PRED (Goldberg 1995:77), and in fact

58

WILLIAM CROFT

indicates that the representation [[V]/[PRED]] in sense F is inherited from the semantic representation of sense A. These descriptions imply the represen­ tations in (17). However, for the reasons given above, the representations in (18) are a more precise and accurate representation of an English speaker's knowledge than those in (17). I will call the representations in (18) verb-classspecific constructions. In fact, other evidence that Goldberg presents indicates that even more specific representations than those in (18) are necessary. She points out that not every permission and refusal verb occurs in the ditransitive construction (Gold­ berg 1995:130): (19) Sally permitted/allowed/*let/*enabled Bob a kiss. (20) Sally refused/denied/*prevented/*disallowed/*forbade him a kiss. A representation of these two 'classes' would in fact have to specify each verb that occurs in the ditransitive construction: (21) E. [[SBJ permit OBJ1 OBJ2]/[enabling XPoss by permitting]] [[SBJ allow OBJ1 OBJ2]/[enabling XPoss by allowing]] (22)  [[SBJ refuse OBJ1 OBJ2]/[negative XPoss by refusing]] [[SBJ deny OBJ1 OBJ2]/[negative XPoss by denying]] I will call the constructions in (21)-(22) verb-specific constructions. There is also evidence that even the classes for which there appears to be a good case for only a schematic verb-class-specific construction must also include representations of verb-specific constructions. Goldberg points out that there are at least two other anomalous examples of verbs used in the ditran­ sitive construction (Goldberg 1995:132), which would be represented as verbspecific constructions, as in (23b) and (24b): (23) a. He forgave her her sins. b. [[SBJ forgive OBJ1 OBJ2]/[forgive someone for something they did]] (24) a. He envied the prince his fortune. b. [[SBJ envy OBJ1 OBJ2]/[envy someone for something they possess]]

LEXICAL RULES vs. CONSTRUCTIONS

59

Goldberg points out that historically forgive meant 'to give or grant', and envy meant 'to give grudgingly' or 'to refuse to give' (1995:132); hence they began as members of verb classes normally found in the ditransitive construc­ tion. Their meanings changed, yet they are still allowed in the ditransitive construction (although Goldberg suggests that they have become archaic and are being lost; 1995:132). If sense A of the ditransitive construction were represented only by a verb class schema as in (18). then the occurrence of the original forgive 'give, grant' and envy 'give grudgingly, refuse to give' in the ditransitive construction could be predicted from the semantics of the basic verb entry and the semantics of the appropriate verb-class-specific construction in (18). That is, there would be no need for verb-specific constructions for these verbs. If this were true, then one would predict that once the verb meanings changed, the verbs could no longer occur in the ditransitive construction. Instead, the verbs did continue to occur in the ditransitive construction. This implies that for at least these two verbs, and probably many other verbs, the verb-specific constructions are autonomous (Bybee 1985), that is, independently represented in the mind, even though their syntax and semantics are predictable from the verb-class-specific construction. Their subsequent semantic divergence from the ditransitive construction meaning is the constructional analog to lexical split, where for example "something can be dirty without involving real dirt at all" (Bybee 1985:88). 4.

Constructions, lexical rules, and verb-specific constructions

In the preceding section, I argued that the facts about the ditransitive construc­ tion in English requires the postulation of verb-class-specific constructions and verb-specific constructions as part of an English speaker's knowledge of his/her language. What is the consequence of this argument for the debate over lexical rules vs. constructions? Verb-specific constructions are simply more specific types of construc­ tions. Constructions are popularly represented as abstract syntactic schemas without specific lexical content (except for obligatory inflections and function words). But they need not be that abstract. Verb-specific constructions are constructions but their schemas have specific lexical content for the verb. Verbclass-specific constructions are slightly more abstract, but still much less specific than a construction schema such as [SBJ VERB OBJ1 OBJ2]. The exis­ tence of verb-specific constructions and verb-class-specific constructions does

60

WILLIAM CROFT

not preclude the existence of a more schematic construction, if the speaker has made such a generalization in his/her mental representation (see below). The existence and necessity of verb-specific constructions indicates that there is a false dichotomy underlying the lexical rule analysis and the abstract constructional analysis. One can account for lexical idiosyncrasy and construc­ tional generality with verb-specific and verb-class-specific constructions. We now turn to the arguments given by Goldberg about the verbal rule vs. construction controversy, in order to support this point. Some of Goldberg's arguments are in favor of constructions in general, not critiques of a lexical rule analysis of argument structure patterns. Argument (viii) demonstrates the need for idiosyncratic information with respect to constructions, not just simple lexical items, and argument (x) cites the lack of a sharp boundary between lexical and syntactic processes. Although a model relying exclusively on lexical rules will find these facts problematic, the postulation of verb-specific constructions does not, since verb-specific constructions are constructions. Argument (vii) identifies a semantic constraint associated with the ditransitive construction; but the grammar must still sanc­ tion the use of clear in (10a) to mean 'clear with intention to transfer (meta­ phorical) possession' with (at least) a verb-class-specific construction. Argument (vi) makes the case that what is called syntactic bootstrapping, where a child learns verb meanings via their syntax, is in fact constructional bootstrapping: the child learns syntactic constructions with their meanings, and uses the constructions to figure out the meanings of verbs, or to match verbs of a certain meaning with the appropriate syntactic constructions. Goldberg's interpretation of so-called syntactic bootstrapping is correct, I believe. Further evidence from language acquisition supports the initial learning of verbspecific constructions. Tomasello (1992) argues for the Verb Island Hypoth­ esis: that children acquire verbs first in a single argument structure construc­ tion, and then learn to apply that verb to other constructions. More generally, I would propose that the acquisition of constructions proceeds inductively, with the child learning specific word-construction combinations and gradually generalizing. However, in the generalization process, it is not necessarily the case that the more specific instances are then dropped as independent grammatical representations, even when they are completely predictable from the more general constructions (cf. Langacker 1987:374). The historical evolu­ tion of forgive and envy also implies that redundant representation of verbspecific and verb-class-specific constructions is necessary. Goldberg argues that parsimony dictates assigning the meaning to the (schematic) construction. Goldberg now withdraws this argument (pers.

LEXICAL RULES vs. CONSTRUCTIONS

61

comm.); however, since it is a widely accepted argument, I will discuss it here. Parsimony is not a criterion for mental representations; only psychological reality is (Croft 1998b). Moreover, Goldberg argued for only one sort of parsi­ mony, storage parsimony. Storage parsimony involves minimizing redundant storage of grammatical information. The result of storage parsimony is computational complexity: language users must compute online the meaning of a combination of a verb and a construction in which it occurs (this requires some preservation of compositionality, hence argument (iv)). Storage parsi­ mony has long been the sine qua non of generative (and structuralist) analyses of grammar. But one could argue for computing parsimony instead: minimizing the computation involved in language use. Computation cannot be eliminated, of course; but nor can storage of knowledge in the storage parsimony approach. The truth is, though, that on a priori grounds one cannot argue for storage parsimony or computing parsimony (Croft 1998b). There is likely to be some storage redundancy and some computational complexity, though some cogni­ tive psychologists suggest that there is probably more of the former (see refer­ ences cited in Barsalou 1992:180). Linguistic evidence can place an upper bound on how abstract or general mental representations must be; but they allow for less abstract and potentially redundantly represented storage of linguistic knowledge (Croft 1998b). In fact, as demonstrated above, closer examination of the linguistic facts almost always reveals idiosyncrasies that show that more specific representations are required than is usually thought. Goldberg argues in favor of redundancy in stored information for more schematic constructions. In comparing more schematic constructions to more specific instantiations, she allows for full-entry representations of the infor­ mation 'inherited' by the more specific construction from the more schematic construction (Goldberg 1995:74). Verb-specific constructions simply extend the full-entry principle down to constructions specific enough to specify particular verbs (or narrow semantic verb classes). Thus, verb-specific con­ structions are compatible with Goldberg's general approach to the represen­ tation of constructional knowledge. Most of Goldberg's remaining arguments turn on the nature of verb entries: she argues against the necessity or desirability of multiple verb entries linked by lexical rule. In her argument (i), she argues that the derived verb meanings are implausible as verb meanings; one would not expect to find a distinct verb in another language with the derived verb meaning. This argument is slightly too strong. One does not find basic verb forms with the derived verb meanings. One does find morphologically derived verb forms with the derived verb meaning, derived by the addition of an applicative affix to the verb.

62

WILLIAM CROFT

Goldberg argues that the applicative affix should be represented as part of the construction, not as a derived verbal lexical entry with a derived meaning (argument (ix) above; cf. Goldberg 1995:61). This analysis will work for abstract constructions in some cases, but not in others. For example, the Russian verbs that participate in the locative alternation illustrated in examples (11)—(12) above require different derivational prefixes that must be lexically specified; that is, they must be represented as at least verb-specific construc­ tions (see examples in Schupbach 1978). 5.

Verb-specific constructions, verb entries, and analyzability

As a matter of fact, verb-specific constructions do not appear to be that dif­ ferent from multiple verb entries, each associated with the argument structure construction it occurs with, and linked together by lexical rules. What is the relationship between verb entries and verb-specific constructions? The chief difference is that a verb entry must have associated with it its specific contri­ bution to the meaning of the constructional whole. Yet that contribution differs from verb to verb for the ditransitive construction, even if we assume a singleentry model of verbal semantic representation, as Goldberg does. The semantics of the combination of verb + ditransitive construction can be divided into three components. The first is what Rappaport Hovav and Levin (1998:107) call the verbal constant: the core of meaning that differentiates verbs of the same semantic class such as throw, toss, kick, etc. The second is the transfer of possession meaning that is associated with the verbs when they occur in the ditransitive construction. The third component is what I called the modulation in the preceding section: whether the transfer of possession is actual, conditional, intended, etc. (see Table 1) The verbs that occur in the ditransitive construction can be put into three groups depending on how many of the above semantic components the basic verbal meaning would include (see Table 1). When any verb from any of these classes occurs in the ditransitive construction, all three semantic components are present. For Group I verb classes, occurrence in the ditransitive construc­ tion does not 'add' anything over and above the verbal meaning. For Group II verb classes, occurrence in the ditransitive construction adds only transfer of possession. For Group III verb classes, occurrence in the ditransitive construc­ tion adds both a modulation (actual or intended) as well as transfer of posses­ sion. The semantics of Group I verbs is a simple relation of specific meaning subsumed under a more schematic meaning (Goldberg 1995:60). The question

63

LEXICAL RULES vs. CONSTRUCTIONS

then, is, do verbs of Groups II and III include the additional semantic compo­ nents as part of derived verbal lexical entries, so that the process of semantic composition of verb + construction is simply subsumption? Or are they contributed only by the (verb-class-specific) ditransitive construction, in which case a more complex semantic composition process is necessary? Verbal class

Verbal constant

Modulation

Group I inherent giving (A1) conditional giving (Bl) future transfer (Dl)

√ √ √

√ [actual] √ [conditional] √ [future]

Group II refusal (C1) permission (E1) costing (Gl)

√ √ √

√ [negative] √ [enabling] √ [depriving]

Group III instantaneous ballistic motion (A2) deictic continuous causation (A3) creation (F1) obtaining (F2)

√ √ √ √

[actual] [actual] [intended] [intended]

Transfer of possession

√ √ √

Table 1: Componential analysis of ditransitive verb class basic meanings

I would like to suggest that this question is not decidable by purely linguistic evidence. Speakers may or may not induce more schematic constructions, and they may analyze their component units in different ways. Consider the options available to an English speaker for the array of constructions in (18) and (21)(24) with respect to both modulation and transfer of possession. An English speaker could abstract a single schematic ditransitive construction [[SBJ DITR.VERB OBJ1 OBJ2]/[XPOSS]] and analyze out fully derived verb meanings, along the lines of the Group I verbs. The construction could include transfer of possession meaning, except for forgive and envy, but the modulation would come from the derived verb entries. Or a speaker could abstract a set of ditransitive constructions like the ones given in (17), each construction speci­ fying transfer of possession and its modulation, combining with underived verbs, which would have to match the modulation of the particular construction for Groups I and II.

64

WILLIAM CROFT

The speaker need not induce any construction more abstract than the verb-class-specific constructions in (18), or the verb-specific constructions in (21)-(24). Even in that case, the same options for analyzing out a meaning for the verb are available. The speaker may analyze out the verb form with a single basic meaning; or the speaker may analyze out the verb form as possessing a meaning including the modulation and also the transfer of possession meaning that are specific to the argument structure construction. One might want to invoke a principle of syntagmatic parsimony—avoid overlap in meaning between the verb and the construction. The same arguments against storage and computing parsimony apply to syntagmatic parsimony, of course. Even if we set these aside, however, storage parsimony fails to simplify the analysis. No matter how abstract or how specific is the meaning of the ditransitive construction induced by the speaker, there will be one or more group of verbs whose meaning will overlap with that of the construction even in a single verb entry model. (I find the syntagmatically most parsimonious solution implausible, namely that the ditransitive construction has no meaning with Group I verbs, only transfer of possession meaning with Group II verbs, and both transfer of possession and modulation meaning with Group III verbs.) In other words, the data does not unambiguously point to a single optimal solu­ tion for analyzing the form-meaning mapping for the family of ditransitive constructions. Finally, we must ask ourselves, what is the verb meaning taken as 'basic' by advocates of lexical rules or taken as the 'single' verb meaning argued for by Goldberg? The 'basic' or 'single' verb meaning is actually the meaning of the verb when it occurs in some other argument structure construction such as the transitive construction. Verb meanings cannot be defined in pure isolation; they are only definable with respect to the construction(s) they occur in. The verb-specific and verb-class-specific constructions such as those in (18) and (21)—(25) are much closer to what a speaker actually hears and uses. Linguists cannot second-guess the sort of generalizations speakers make beyond these constructions. (Nor should we expect all speakers to form the same generalizations.) Only psycholinguistic experimentation might be able to establish the generalizations formed by individual speakers (Croft 1998b). Unfortunately, the psycholinguistic experiment conducted by Carlson and Tanenhaus (1988), discussed by Goldberg (argument (v) above), does not decide the issue. One need not assume that verb senses are either the same (leading to a shorter reaction time) or different (leading to a longer reaction time). It could be that there are degrees of difference between senses leading to differences in reaction times (cf. Goldberg 1995:230, fn. 13). For example, the

LEXICAL RULES vs. CONSTRUCTIONS

65

difference between set [something somewhere] and set [a clock to a time] is greater than the difference between load [something somewhere] and load [something with something else]. The latter two share a semantic frame that the former two do not (see the analysis of spray/load verbs in Croft 1998a). Thus, the experimental results are compatible with autonomous derived verb senses or construction senses. 6.

The mapping of form to meaning in language

In the preceding section, I suggested that the meaning of the verb-specific construction as a whole cannot be unambiguously factored out into the meaning of the more schematic construction and/or the meaning of the verb taken as an isolated lexical item. In fact, the relationship between (basic) verbal meaning and constructional meaning varies from class to class (see Table 1). For example, it follows from neither the semantics of verbs of creation nor the semantics of the ditransitive construction (in all of its uses) that She baked Bill a cake involves only intended and not actual transfer of possession. In other cases there is semantic overlap. For example, it follows from both the semantics of verbs of inherent transfer of possession and the semantics of the ditransitive construction that She gave Bill a book involves (actual) transfer of possession. What conclusions can we draw from these facts? Speakers are not exposed to verbs in isolation, nor are they exposed to schematic argument structure constructions without verbs in them. Actually occurring utterances are closest to verb-specific constructions. The process of language learning and (re-)using is a process of analyzing the structure of utter­ ances into its component parts. The parts will possess recurring meanings in most of their various contexts: bake will involve creating a food item, and the ditransitive construction will involve transfer of possession. The recurrent aspects of meanings will sanction the use of bake and of the ditransitive construction in new situations that the speaker wants to describe. But those recurrent meaning aspects do not give the full description of the meaning of the verb or the construction in each and every context of use (see Croft 2000: chap. 4). They do not even give the full description of the meaning in other conven­ tionalized contexts, such as the ditransitive construction with refuse, bequeath, ox forgive. This is in fact the great asset of language, its flexibility in commu­ nication—although it seems the great bane of natural language processing and of formal models of syntax and semantics. Linguists spill much ink arguing where the components of the meaning of a verb-specific construction lie. Do they belong in the verb in isolation,

66

WILLIAM CROFT

analyzed out from the construction? Or only in a more schematic construction? Or neatly divided between the two? But another perspective is more interesting. If linguists have so much difficulty analyzing the semantic contribution of each element of a fully specified construction, don't speakers have the same problem? In fact, there is evidence that they do. Consider first some simple morphological examples. The plural of child is children. It contains two etymological markers of plurality, -r and -en. Speakers apparently stopped identifying -r as a marker of plurality and added -en (this common process is called reinforcement in language change). That is, speakers reanalyzed the mapping between form and meaning, in part because the original form-mean­ ing mapping is somewhat opaque. Similar cases are found in innovations such as feets and PFI initiative [PFI = 'Private Finance Initiative'] (see Croft 2000: chap. 5). Moving closer to argument structure, we may consider the loss of governed oblique cases in Germanic languages (Croft 2000: chap. 5). That is, a verb that occurred in an argument structure construction with an oblique case argument, such as the dative or genitive, is found with an accusative (object) case argument instead. The speakers no longer identified the semantic contri­ bution of the oblique case, which overlapped with that of the verb anyway. So they reanalyzed the form-function mapping so that the oblique case made no semantic contribution, and replaced the oblique case with the object case form. In other words, the form-function mapping of the component parts of constructions is not necessarily fully specified. Another way of putting it is that multiple analyses of the form-function mapping may be available to the speaker (cf. Hankamer 1977). Part of the reason for this is that verb-specific and verb-class-specific constructions exist as part of a speaker's grammatical knowledge, so the form-meaning mapping does not have to be fully specified. Verb-specific and verb-class-specific constructions are semantically partially analyzable but need not be completely so. Most of the time this is not a prob­ lem in communication, since the verb-specific construction specifies the meaning of the whole as well as its syntax. But the analysis of the form-func­ tion mapping can vary across speakers, and can result in variation and change across time. As linguists, we should accommodate this variation and indeter­ minacy in our analysis of argument structure constructions.

LEXICAL RULES vs. CONSTRUCTIONS

67

Notes 1 I would like to thank Adele Goldberg, Hubert Cuyckens, and an anonymous referee for their comments on an earlier version of this paper; they are hereby absolved of any defects in the final product. 2 The following abbreviations are used in the glosses: ACC accusative; INST instrumental; LOC locative; DERIV derivation; IMPF imperfective; 3SG 3rd person singular; PRES present tense.

References Barsalou, Lawrence R. 1992. Cognitive Psychology: An Overview for Cogni­ tive Scientists. Hillsdale, N.J.: Lawrence Erlbaum. Bybee, Joan L. 1985. Morphology: A Study into the Relation between Meaning and Form. Amsterdam: John Benjamins. Carlson, Gregory N. & Michael K. Tanenhaus. 1988. "Thematic Roles and Language Comprehension". Thematic relations (= Syntax and Semantics, 21) ed. by Wendy Wilkins, 263-288. New York, N.Y.: Academic Press. Comrie, Bernard. 1985. "Causative Verb Formation and Other Verb-Deriving Morphology". Language Typology and Syntactic Description, Vol. 3, Grammatical Categories and the Lexicon ed. by Timothy Shopen, 309348. Cambridge: Cambridge University Press. Croft, William. 1991. Syntactic Categories and Grammatical Relations: The Cognitive Organization of Information. Chicago, Ill.: The University of Chicago Press. . 1998a. "Event Structure in Argument Linking". The Projection of Arguments: Lexical and Compositional Factors ed. by Miriam Butt & Wilhelm Geuder, 1-43. Stanford, Calif.: CSLI Publications. . 1998b. "Linguistic Evidence and Mental Representations". Cognitive Linguistics 9.151-173. . 2000. Explaining Language Change: An Evolutionary Approach. London: Longman. Fillmore, Charles J. & Paul Kay. 1993. Construction Grammar Coursebook, Chapters 1-11. {= Reading Materials for Ling. X20). University of Cali­ fornia, Berkeley. Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago, Ill.: The University of Chicago Press. Hankamer, Jorge. 1977. "Multiple Analyses". Mechanisms of Syntactic Change ed. by Charles Li, 583-607. Austin, Tex.: University of Texas Press. Jackendoff, Ray. 1997. "Twistin' the Night Away". Language 73.534-559.

68

WILLIAM CROFT

Lakoff, George. 1987. Women, Fire and Dangerous Things: What Categories Reveal about the Mind. Chicago, Ill.: The University of Chicago Press. Landau, Barbara & Lila R. Gleitman. 1985. Language and Experience: Evidence from the Blind Child. Cambridge, Mass.: Harvard University Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1, Theoretical Prerequisites. Stanford, Calif: Stanford University Press. ----. 1991. Foundations of Cognitive Grammar, Vol. 2, Descriptive Appli­ cation. Stanford, Calif: Stanford University Press. Levin, Beth. 1993. English Verb Classes. Chicago, Ill.: The University of Chicago Press. & Malka Rappaport Hovav. 1995. Unaccusativity: At the SyntaxLexical Semantics Interface. Cambridge, Mass.: The MIT Press. Michaelis, Laura & Knud Lambrecht. 1996. "Toward a Construction-Based Theory of Language Function: The Case of Nominal Extraposition". Language 72.215-247. Rappaport Hovav, Malka & Beth Levin. 1998. "Building Verb Meanings". The Projection of Arguments: Lexical and Compositional Factors ed. by Miriam Butt & Wilhelm Geuder, 97-134. Stanford, Calif: CSLI Publications. Schupbach, Richard D. 1978. "Semantic Features and Russian Verb Prefixation". American Contributions to the VIIIth International Congress of Slavists, Vol. 1, Linguistic Contributions ed. by Henrik Birnbaum, 616636. Columbus, Ohio: Slavica. Tomasello, Michael. 1992. First Verbs: A Case Study of Early Grammatical Development. Cambridge: Cambridge University Press.

SCHEMAS AND LEXICAL BLENDS SUZANNE KEMMER Rice University

This paper presents a schema-based and usage-based analysis of lexical blends such as glitterati "celebrities, Hollywood royalty" and carjacking "forcible takeover of a car from its driver". Lexical blends are difficult to incorporate in 'building-block' style theories of morphology, in which morphemes are sharply-bounded, primitive units that cannot by defini­ tion overlap within a word. It will be shown that blends fit naturally into a Cognitive Grammar mode of description in which the possible range of lexical and sublexical form-meaning mappings is much greater and includes elements of all degrees of entrenchment and all degrees of separability from other morphological elements. The analysis uses independently motivated CG constructs such as composition and entrenchment, as well as the notion of conceptual blending from Turner and Fauconnier (1995), to sketch a coherent theory of lexical blends and their place in morphology. Various types of blends are discussed, as well as dynamic processes in which a particular blend can form the nucleus of a new, analogical word formation process that can generate a whole family of words, and ultimately, a newly entrenched bound morpheme.

1.

Introduction1

Lexical blends have long been of interest to those studying English linguistics. There is something fascinating about a word in which different ideas are brought together into a new, integrated concept by simply fusing the corre­ sponding words into a single lexical item. Words like affluenza, Chunnel, glit­ terati, swooshtika, and hundreds of other examples could be cited, some quite familiar, others recent neologisms beginning to diffuse through the language, still others nonce words. In English, lexical blending is an important source of neologisms, although it is probably true that derivation and compounding are much more frequent processes. Starting in the last few centuries, the creation of new lexi­ cal items by blending has become a productive morphological device in its own right, both in written and spoken language, and in recent years its popularity

70

SUZANNE KEMMER

appears to have skyrocketed. This situation is in sharp contrast to many, if not most other languages, where lexical blending of the English type is rare to non­ existent. One might think that blending would be common in languages closely related to English, but in other Germanic languages compounding appears to be the dominant word-formation process for neologisms at least, and lexical blends are extremely rare.2 Iam aware of no other language in which blending plays anywhere near the role that it does in English. A lexical blend can be generally defined as "a new lexeme formed from parts of two or more other lexemes" (Bauer 1988:238). Lexical blends comprise the class of 'portmanteau' words referred to in Lewis Carroll's Through the Looking Glass, as described by the character Humpty Dumpty: "You see it's like a portmanteau—there are two meanings packed up into one word." The term at that time was a transparent metaphor: A portmanteau was a nineteenth century piece of luggage consisting of two equal parts connected by a hinge. When closed, the two parts formed a functional unity designed for compactness and portability. Lexical blends are a difficult conceptual problem for many theories of language. I argue in this paper that what is needed for an account of blends is the notion of schemas, a theoretical construct which is not only cognitively well-founded but provides the requisite mechanisms for handling the peculiar properties of lexical and other blends. The main aim of this paper is to sketch the basics of an explicit analysis of lexical blends in terms of schemas. It is not designed to argue against possi­ ble alternatives to a schema-based analysis, but rather to show how the inde­ pendently motivated and cognitively realistic notion of schemas (Langacker 1987; Barlow & Kemmer 1994) can be applied to the difficult problem of lexi­ cal blends. Although the analysis is preliminary, it points to a more farreaching view of the importance of blending in language, going beyond the lexicon, as discussed in Turner and Fauconnier (1995) and Fauconnier and Turner (1996), inter alia. In addition, I hope to make a contribution to English linguistics by empirically documenting the high degree of productivity of lexi­ cal blending in Present-Day English and showing the wonderful creativity that underlies it. The inspiration for this paper is the work of Günter Radden, who not only has contributed greatly to Cognitive Linguistics and English Linguis­ tics, but who also has shown a most keen appreciation for all of the innovative and creative aspects of language.

SCHEMAS AND LEXICAL BLENDS 2.

71

Some characteristics of lexical blends

There is a large literature on lexical blends, much of which consists of copious examples and proposed typologies, with particular reference to the form of blends (see Hansen 1963; Algeo 1977; Soudanek 1978 and the many refer­ ences cited therein; Cannon 1986; cf. also Lehrer 1996, a psycholinguistic study). Lexical blends are so varied in form that no neat taxonomy can do justice to the full range of the phenomenon. In what follows, I will distinguish several different subtypes defined by particular properties. These types have been distinguished before in the literature by various authors, and further they do not define mutually exclusive categories. My main purpose in identifying the properties of blends defining these types is not to propose a new or improved typology, but to allow us to begin to consider what would be mini­ mally required by any theory of morphology that would aspire to include them. Although, as stated above, the literature contains large numbers of ex­ amples of blends, I chose to observe the blends occurring in two databases of neologisms whose collection I organized, about 150 examples in all.3 I believe they are quite representative of the blending processes that have been discussed in the literature; and they have the advantage of being more recent than those in many published collections. Almost all blends encountered in my data (and in the literature) are formed from two lexical items that provide what is often thought of as 'input' to a morphological process of blending. The analytical model I will use is not a process model, so I will avoid the 'input-output' metaphor as much as pos­ sible; we can instead think of blends as words that are cognitively linked to pre-existing words which are co-activated when the blend is used.4 The seman­ tics of a lexical blend is a coherent cognitive structure that selectively incor­ porates and integrates aspects of the semantics of the activated words (cf. Turner & Fauconnier 1995). I will call the co-activated words source lexemes, since they are diachronically prior and also synchronically more primary as well (at least when the blend is first formed), being more cognitively entrenched (i.e. routinized) than the resulting blend. We can show the source elements for blends using the kind of formula illustrated in (1), in which source lexemes are given in brackets; the parts of the source appearing in the blend are indicated in small caps; and the symbol x stands for 'is blended with'. (1) a. skyjacking "hijacking of a commercial aircraft" b. Chunnel "the Channel tunnel"

[SKY x hiJACKING] [CHannel x tUNNEL]

72

SUZANNE KEMMER

Although two source lexemes is the norm for a blend, I did encounter two cases with three source lexemes, given in (2): (2) a. turducken "a meat dish of turkey stuffed with duck and chicken" b. motormobilia "automotive memorabilia"

[TURkey x DUCK x chickEN] [MOTOR x autoMOBILE x memorABILIA]

Some of Lewis Carroll's original blends from "Jabberwocky" in Through the Looking Glass illustrate what we might call intercalative blends: formations in which the two words involved in the blend are so tightly integrated in the blended word that the sounds of one source lexeme are interspersed between the sounds of the other. The famous examples in (3) illustrate this type. (3) a. chortle "laugh softly with sharp, repeated expulsions of air" b. slithy "slimy and lithe"

[CHuckLE x snORT] [SLImY x LITHE]

In chortle, the elements ch...le of chuckle do not appear contiguously in the blend, but are separated by the ort of snort. Similarly, in slithy, the elements sli...y from slimy are separated by the th of lithe. There are no intercalative blends in my data that do not also have a possible non-intercalative analysis. For example, in (4a), fantabulous could either be f...abulous merged with the (fiant oí fantastic, or simply a sequential blend of fant(a) from fantastic and (a)bulous from fabulous, with an over­ lapping a. Similarly with (4b), in which the initial segment of the blend might or might not be identified with the /s/ sound in psychedelic, and in (4c) and (4d), in which both intercalative and sequential analyses are also possible. (4) a. fantabulous "extremely wonderful" b. Celtadelic "Celtic hard rock music"  clownselor "idiot high school counselor" d. dumbsizing "ineffective, poorlythought out downsizing"

[(F)ANTAStic X (FA)BULOUS] [CELT(I)C X (PS)ychEDELIC]

[CLOWN x (COUN)SELOR] [DUMB X (D)ownSIZING]

The lack of unambiguous examples, and the relative infrequency of even ambiguous ones, suggest that as in the case of blending of three source

73

SCHEMAS AND LEXICAL BLENDS

lexemes, intercalation is a minimally productive pattern, although the creative resources of the language certainly allow it.6 Another possible characteristic of blends is that the two source lexemes share some phonological material. This is very common, but by no means inevitable in a blend. Some examples of this type, which we can call overlap blends, include, in addition to at least some of the examples in (4), all of the cases in (5). Here, the overlapping elements are given in boldface in the source lexemes. (5) a. glitterati "the glamorous elite" b. coffnoscenti "coffee connoisseurs" c. stalkerazzi "tabloid photographers who dog celebrities like stalkers" d. swooshtika "derogatory reference to Nike swoosh logo" e. affluenza "malaise/lack of ambition common among children of the affluent"

f. imagineer "to envision and design innovative inventions" g. gayborhood "gay neighborhood"

[GLITTER x LITERATI] [COFFee x COgNOSCENTI] [STALKER X

papARAZZI]

[SWOOSH X

SWasTIKA]

[AFFLUENt X inFLUENZA] [IMAGINE X enGINEER] [GAY x nEIGHBORHOOD]

Overlapping of phonological segments is not a unique characteristic dividing this type from the other subtypes described; note that among the intercalative blends, slithy has source lexemes that share phonological material (li) but chortle does not. The other types discussed below also include instances with and without segmental phonological overlap. In fact, overlap seems to be a special case of a more general phenomenon found with blends: they typically share phonological similarities, and only incidentally morphological structure, with one or both of their source lexemes. Overlap blends are cases in which the blend contains a phonological string that is present in both source items. Since the amount of overlap can range from a single segment (as in dumbsizing) to strings larger than a syllable (as in glit­ terati), the degree of similarity with the sources is variable. But similarity goes beyond shared segments; sometimes the phonological string is not identical to the source, but deviates from it in the features of one or more segments. For example, in swooshtika, the sw is certainly part of both source lexemes, but the sh is identical only to the corresponding sibilant in one source lexeme, swoosh,

74

SUZANNE KEMMER

and is merely phonetically similar to the sibilant in corresponding position in swastika. Furthermore, it seems that similarity of syllable structure is a factor in the felicity or likelihood of a blend: the resulting blend tends to share syllable structure with one or both of its source words. This factor is visible in virtually all of my data. For example, dumbsizing is a good blend because it echoes not only the d and final nasal of down, but also its CVC syllabic structure; more­ over, the resulting blend is identical in syllable structure to one of the source words, downsizing. Affluenza is a very similar case: the blend has the syllabic structure of influenza and shares with its source words the string fluen plus a sibilant. In an interesting case of variation, both digiterati and digerati are attested blends meaning 'computer elite'; the first shares the string it with digit(al) and literati, but the second shares the syllable and rhythmic structure of literati. (Very often, one source lexeme seems to be more primary in the formation of the blend than the other, in that, for example, its syllable structure is main­ tained in the blend; this has consequences for the extension of blends as discussed in Section 5). I would suggest that recognizability of the source words, on the one hand, and similarity of syllable structure (including stress), on the other, are competing pressures that give rise to these two forms. If syl­ lable structure is a genuine guiding factor in blends as claimed here, we might predict that with increased frequency and thus familiarity (which essentially removes recognizability as a factor), blends with clipped roots like digerati would become the preferred forms over competing blends with deviating syl­ lable structure. Many blends involve a substitution of part of one source lexeme with another whole lexeme, a type we may call substitution blends. Some examples include the following: (6) a. carjacking "theft in which a car is forcibly taken from its rightful driver" b. stoolgazing "examination of a baby's stools to gauge its health"  hipoisie "the class of sophisticated, hip yuppies" d. nom de net "pen name for use on the internet" e. goodie-two-skates "skater who is overly well-behaved and conventionally moral"

[CAR x hi JACKING] [STOOL

x starGAZING]

[HIP x bourgeOISIE] [NOM DE plume x (inter)NET] [GOODIE-TWO-shoes x SKATES]

SCHEMAS AND LEXICAL BLENDS

75

This property is independent of whether there are phonologically overlapping source elements. Carjacking, for example, has no such overlap, but stoolgazing does. From the perspective of overlap, we can think of examples like car­ jacking as cases in which the degree of segmental overlap between the source lexemes is zero. Phonological similarity of the parts with source lexemes is still present, of course, and again the tendency towards 'echoed' syllable structure is evident in all of the examples, in which the blend is identical in syllabic structure to one of the source items. Substitution blends, although they are not an empirically separate class from the overlap blends, are nevertheless useful to distinguish as a concep­ tually separate class. Substitution blends are defined as having a clear substi­ tution of part of a lexeme with another whole source lexeme (as opposed to merely a part of a source, as in coffnoscenti). In substitution blends, the whole source lexeme that is used as a 'replacement part' (e.g. car in carjacking) seems to serve a contrasting function and to bring its full conceptual semantics to bear on the blend (cf. Section 4 below). These properties make substitution blends a perfect locus for the creation of new word families, as discussed in Section 5. Given the cross-cutting and rather heterogeneous set of types described above, we might now ask, are there any properties which characterize blends in general? We can extract from the above discussion the following charac­ teristics, which are not necessary and sufficient conditions for defining a blend, but rather properties that tend to cluster in instances of blending. a. b.



Blends combine parts of lexical source words, rather than whole source words; this distinguishes them from compounds. Morphological structure is not particularly relevant to blends. Source lexemes are combined without regard to their morphological boundaries, and the internal structure of the resultant blend is not necessarily morphologically analyzable concatenatively. It is usual to find morphological overlap, residue, and/or clipped forms, rather than exhaustive divisibility of the blend into sequential, whole morphemes. Phonological properties are highly relevant to blending; phonological similarity of the blend with part or whole source lexemes increases the likelihood or felicity (the 'goodness') of a blend. Similarity can range from segmental identity through segmental similarity to same or similar syllable structure; and the

76

SUZANNE KEMMER similarity can range from identity/similarity of the blend with both source lexemes, to one source lexeme, or to parts of these.

3.

Lexical blends in a theory of morphology

Despite the interest they attract due to their creative nature and their typ­ ological unusualness, lexical blends have been little more than a footnote in the study of morphology in modern linguistics. Bauer (1988:39) sums up their peripheral status as follows: "It is ... extremely doubtful whether such words can be analyzed into morphs, and thus whether they form a real part of morphology." In standard, building-block theories of morphology, morphemes are concatenated into larger units, words, which are in turn concatenated into syntactic phrases. (See Rubba 1993 for a critique of 'building-block' models of morphology, and a schema-based alternative applied to non-concatenative morphology.) Such theories seem to work reasonably well for the typologically most widespread morphological processes (i.e. inflection, derivation, and compounding), since in these processes the source elements contribute morphological elements, either roots or affixes, which are then concatenated in the target word. Although there are standard difficulties, such as the wide­ spread phonological fusion found in inflection and derivation, at least the source elements are generally recognizable morphemes whose boundaries line up in the target, or at worst, are blurred together in the same phonological stretch. Blending, however, is a different story. It poses a very difficult problem for building-block theories for a number of reasons. Most obviously, although blending seems to require some kind of compositionality, morpheme bounda­ ries found in the source elements not only do not necessarily coincide in the target, but in fact are not even relevant to the structure of the blend. The extreme case of this is found in intercalative blends, in which source elements are split up and distributed in such a way as to generally echo the sound of the source word(s), but without remotely reproducing their morphological struc­ ture. These we might be tempted to disregard due to their rarity. However, overlap blends, which are far more frequent, illustrate the same problem to a different degree. For example, the final boundary of the source lexeme glitter is not at all important in the blend glitterati, which has no boundary after glitter. The shared phonological string is what gives the blend its main structure or scaffolding, and the rest of the word is made up by other phonological elements contributed by the source lexemes in a non-predictable way, guided by global similarities of the blend with one or both source items.

SCHEMAS AND LEXICAL BLENDS

77

Thus, the blend is not really composed of morphemes in the sense of recurrent minimal meaningful parts (this is the lack of morphological analyzability referred to by Bauer, above), but of phonological strings that trigger meanings.7 Although in some cases these elements may have the form of a complete word (e.g. car in carjacking), often they only suggest the lexical source words. In hipoisie, for example, the oisie is not a segmentable part of a word like bourgeoisie that recurrently and conventionally combines with other roots (in fact, bourge is not even a root). The string merely evokes bourgeoisie as a recognizable lexical model, from which some of the formal and semantic properties can be extracted for the analogical creation of the new form. Recognizability is a matter of degree, of course, and does not lend itself to formu­ lation in a rule. Furthermore, it is affected by such factors as familiarity (correlated with frequency) which are generally considered extragrammatical in rule-based theories. The general effect of phonological similarities on blends, including the apparently powerful effect of syllable structure, makes them very hard to handle with traditional grammatical machinery. The amount of similar struc­ ture can vary a great deal, so it is impossible to state a general formal rule that will license some blends and exclude others. Moreover, the effects of similarity at what are usually treated as different levels of sound structure (features, phonemes, and syllables) is extremely difficult to capture in a set of rules; rules are generally stated in terms of one specific kind of information. For example, phonotactic generalizations are traditionally treated differently from individual segmental representations; segmental phonology is treated separately from syllabic structure. Given the general effect of similarity of different kinds in blends, it seems as though, instead of rules, speakers are operating with a facility for global pattern-matching that allows similarities on many different dimensions to count, as long as there are enough of them (with 'enough' not being pre­ cisely quantifiable). It is clear from the above that the requirements for handling blends in a theory of morphology are rather daunting. The requisite theory must have a number of properties that are at odds with building-block theories and rulebased theories in general. First, it must have sufficient flexibility to accom­ modate non-concatenative structure of various sorts. It must allow for forms in which there is overlap, that is, in which the 'components' of the blend do not have coinciding morphological boundaries. If it is to be truly general, it should also allow the possibility of intercalative structure. It must allow for the state­ ment of phonological factors at the level of syllable as well as segmental

78

SUZANNE KEMMER

structure, and must have the delicacy required to accommodate degrees of similarity. It must integrally incorporate the possibility of variation and of multiple possible analyses sanctioned by the system. It must take into account linguistic experience, which observably affects linguistic usage (Bybee 2000). Finally, the theory should have a realistic level of predictivity: predictions should be cast in the form of likelihoods of occurrence, rather than as predic­ tions of 'possible' and 'impossible' forms. Degrees of licensing and higher and lower probabilities of occurrence are the kinds of predictive generalizations that befit a biological, cognitive, and cultural phenomenon like language. I claim that schemas are an ideal theoretical tool for a theory of lexical blends. Schemas in language are generalizations extracted from linguistic forms and meanings. A schema is a cognitive representation consisting of perceived similarities across many instances of usage. Schemas are essentially routinized, or cognitively entrenched, patterns of experience. They arise via repeated activation of a set of co-occurring properties; once sufficiently entrenched, they can be used to produce and understand linguistic expressions. Linguistic expressions are categorized by schemas in production and compre­ hension; in other words, they are licensed to occur by those schemas. In this way expressions are linked to the knowledge structures that produce them and make them interpretable.8 A simple example of a schema would be schemas for specific phonotactic patterns in English, for example, the syllable-initial sequence /$str/. Knowledge of this pattern is part of the grammar of English, which speakers of English have abstracted through repeated linguistic experience of words like strength and strip. Newly-heard words containing /$str/ would match that schema by virtue of having all the relevant properties, and, assuming the rest of the word conformed to other available phonological schemas, would thus be recognized as possible words of the language. Other kinds of linguistic schemas include those for syntactic constructions (e.g. passive constructions, transitive clauses, there-constructions, etc.) and morphological forms and processes (e.g. the -able suffix, the past tense, and schemas for relating derived morphological forms with vowel and stress changes to their stem forms like derive-derivation, etc.). The use of schemas in modeling linguistic knowledge has been exten­ sively discussed and developed by Langacker (1987, 2000), Bybee (1988, 2000), Bates and MacWhinney (1987), and MacWhinney (2000); see also Barlow and Kemmer (1994) and Kemmer and Israel (1994) for discussion of the advantages of schemas over rule-based systems in accounting for linguistic

SCHEMAS AND LEXICAL BLENDS

79

data of various kinds. Here, I will summarize only the aspects of schemas that will be relevant for the account of blends in this paper. In a schema-based theory, probabilities of occurrence of an expression relate to the availability and entrenchment of schemas that can be used to cate­ gorize the expression, and the degree to which it matches the schema, thus accommodating the effects of local and global similarities on the creation of blends. Schemas, being simply what two or more forms or meanings have in common, are not restricted by the kinds of sequencing required by rules, nor are they restricted to information of particular types or levels. Schemas allow for competing factors in licensing expressions, and for multiple possible sanc­ tioning structures, thus giving rise to variation and to possible non-unique analyses for linguistic expressions. And, importantly, schemas are sensitive to frequency: frequency measures repetition of instances, which leads to conven­ tionalization (cognitive entrenchment and degree of dispersal in a group of speakers) which itself affects the possibilities licensed by the system: instances of use, over time, shape the very system that gives rise to them. A schema-based model of language is constraint-based, rather than rulebased, in that schemas can be seen as effectively operating as constraints on the linguistic expressions that can be produced and understood by means of them. Recently, theories of language within the generative linguistic tradition have emerged which are also constraint-based. Any constraint-based theory, even one without schemas and the mechanisms for schema selection described by the schema theories mentioned above, will have a number of the advantages enumerated in the previous paragraph. For example, Optimality Theory (), currently the principal constraint-based phonological theory, has ranked, weighted constraints, which allows for variable output, given the appropriate assumptions about differences across speakers in weighting and constraint ranking. Constraints can also come into conflict and compete, just as we can have schemas that compete for activation in language use. Representations, being the result of simultaneous constraint satisfaction, would permit the inte­ gration of different kinds of information within them, just as in a schema repre­ sentation; and blended representations can relate non-concatenatively to their morphological sources. (See Pineros 1999 for an  account of one type of blending in Spanish.) Nevertheless, the schema-based framework presented here is in my view preferable to an Optimality Theoretic approach on theoretical grounds. For one thing, it is not presently clear what the cognitive underpinnings of the opti­ mality system are intended to be. The generative assumptions out of which the theory arose require a maximal role for innate structures and a minimal role for

80

SUZANNE KEMMER

learning and experience; the cognitive structures assumed are thus apparently independent of such factors as frequency. The importance of frequency in learning in general is beyond question, and psycholinguists have generally found it an important factor in language processing (cf. Marslen-Wilson 1990). A usage-based model of language (Langacker 2000) predicts frequency to be directly relevant to the grammatical system, and findings such as those in Bybee (2000) support this. Secondly, the representations are, like other generative representations, categorical, rather than allowing for degrees of fit between a linguistic expression and a cognitive representation. And third, there is no general optimality theory of language which includes all the levels of language (although beginnings have been made on treating syntax with an optimality approach). In particular, aspects of language that relate to meaning and conceptual structure have not been treated in . The focus on form and on formal explanations (i.e. explanations derived from the structure of language) have left correspondingly little attention to the role of meaning in language, even in language phenomena involving meaningful units such as morphemes and syntactic constructions. In contrast, the schema-based theory sketched above has semantics as an integral aspect of it, as described in Langacker (1991). As will be shown in the next section, this view of semantics, in which domains of experience play a key role, allows us to understand the semantic side of lexical blending—a topic virtually untouched in the literature on blends. Thus, while the schema-based approach is perhaps not the only possible framework for accounting for the formal structure of blends, I believe it has the best chance of providing an integrated and cognitively well-grounded account of this multi-faceted linguistic phenomenon. In the following section, I will show how schemas can be applied to the analysis of overlap blends, as an illustration of their utility and their potential for accounting for lexical blends more generally. 4.

Schemas in the analysis of lexical blends

In this section, I analyze some aspects of blends using schemas, in order to show how, with independently motivated theoretical devices and minimal innovation, these constructs can capture the unusual morphological properties of lexical blends. The examples studied here are overlap blends. Of all blends, overlap blends are the type that exhibits perhaps the most striking creativity of usage. In order to make a felicitous overlap blend, the speaker has to select two words that evoke concepts he or she wishes to relate, and which also share phono-

SCHEMAS AND LEXICAL BLENDS

81

logical similarities—a formal aspect that makes the blend formation a kind of linguistic puzzle similar to punning. In fact, we could even consider these overlap blends a subtype of pun, in which the homophony involved does not pertain to a lexeme or morpheme in either component, but simply to a phonological string. I believe that it is this 'word game' aspect of blends that has been a prime impetus for their enormous popularity in English, once the pattern became sufficiently established.9 Schemas, as we saw above, are not restricted as to types or level of information; in particular, they can contain linguistic substance of all different degrees of specificity. Thus, a phonological string in a schema might contain some syllable structure as well as segmental material of different degrees of specificity. This capability allows us to account for cases of phonological similarity of different degrees. For example, to capture the commonalities in phonological structure in 'quasi-overlap' blends such as swooshtika, we can posit a schematic phonological form like /swVS/, where V stands for any vowel and S for a voiceless sibilant. This schematic form is shared by the source lexemes as well as by the blend. We might hypothesize that the speaker who first thought of the conceptual relation between the Nike swoosh10 and the Nazi swastika also abstracted this phonological similarity, thus creating an adhoc schema, and used this schema in the construction of the blend. Subsequent users of this form can make the same phonological connection, consciously or unconsciously; such connections are what makes blends with close phonological similarities with their sources especially felicitous and appre­ ciated by speakers for their creativity. Figure 1 illustrates for swooshtika the relevant connections between the phonological elements of the source lexemes, the blend, and schemas abstract­ ing the similarities between them. It also shows the special relation of the blend to the source lexeme swastika, the one from which it gets its overall syllable structure. (This figure borrows from the basic mode of representation used in Bybee 1988, 2000, which is extremely useful for showing relations among phonological segments.) From this illustration, it can be seen that schemas easily allow the repre­ sentation of phonological overlap as well as similarity more generally: effec­ tively, the overlapping segments sw are treated the same way as other simi­ larities shared by the blend with its source elements. The parts of the blend which relate to each source element happen to be sequential in this case, but the basic mechanism equally allows for intercalation of the elements of the source, as long as they can be related to a shared syllabic schema.

82

SUZANNE KEMMER

Figure 1 : Phonological relations in an English blend

I will now proceed to examine another overlap blend, focusing this time on the semantics of blending. The blend treated here is the word glitterati, which will not only illustrate the use of constructional schemas for another dimension of information (the symbolic), but will also lead to the discussion in the following section regarding the formation of new word families by analogical extension from blends. Schemas, which are analogical in nature and not rule-based, are eminently well-suited to modeling this empirical phenomenon.11 Glitterati, like all words, is a symbolic unit consisting of a semantic pole and a formal pole. The semantic aspect can be summed up in shorthand fashion as something like 'glamorous elite group of people'. The formal pole includes its phonological and morphological structure. Symbolic units are conventionalized form-meaning pairings. They are linked to other units in a network organized in terms of relations of similarity.

SCHEMAS AND LEXICAL BLENDS

83

of specific types. Complex units in syntagmatic relations relate to the smaller units that are invoked by the complex form by means of the two cognitive operations of integration and composition, as described in Langacker (1987). Glitterati thus relates to its two source lexemes, glitter and literati (also symbolic units) via integration and composition. However, the integration is slightly different from that found with concatenative morphology such as compounds. The overlap of the source elements suggests an especially tight formal integration in which morphological boundaries do not separate them. The two forms are linked intimately in the new unit. Furthermore, the concep­ tual integration is also more extreme in blends than in concatenative units like compounds (which in turn are more tightly integrated conceptually than similar syntactic phrases).12 Conceptual integration in this example will be discussed further below in this section. Figure 2 represents the analysis sketched above in terms of the relations of the relevant symbolic units, in the mode of representation used in Cognitive Grammar. In the center, glitterati is shown as a composite symbolic structure integrating its source lexical items. The especially intimate integration of the two source units is represented with a zigzag line; otherwise the integration and composition of the units are interpreted standardly as in Cognitive Grammar, with semantic poles represented on top in each unit (in small caps) and phonological poles at the bottom (in plain type; here, given in ordinary orthog­ raphy). The crucial aspect of the formal structure of glitterati is shown by means of the overlapping boxes inside the phonological pole of the blend: instead of a sequence of morphological forms as in concatenative morphology, there is an overlap of the forms from the source units (the string lit(t)er comes from both). Now let us turn to the semantic side. Symbolic units are viewed in Cognitive Grammar as access points to knowledge structures, called conceptual domains. The two source lexemes in our example provide access points to the specific domains evoked by glitterati. These domains are suggested by the related concepts presented underneath each source lexeme in the large ovals. The domains represented, it should be emphasized, are integrated, non-sharply bounded knowledge networks whose complexity can only be hinted at in such a figure. Links among the concepts (represented, again, in small caps) found within a domain are not shown to avoid undue complexity in the repre­ sentation. This simplified and partial representation will, I hope, serve to give a sufficient idea of the relevant conceptual structures for the present purpose. (The figure borrows from the mode of representation used in Turner and Fauconnier 1995; the domains correspond to what they term 'mental spaces'.)

84

SUZANNE KEMMER

Figure 2: Cognitive domains in the blend glitterati

SCHEMAS AND LEXICAL BLENDS

85

Looking at the lower right side of Figure 2, we can see that literati accesses knowledge of the world of literary intellectuals, in which there is a group deemed (perhaps self-deemed) as the elite within this world. This group is characterized by knowledge and expertise in literary matters; the members are cultivated and conversant with literary productions; they value language and its expressive possibilities and hence are (ideally) particularly articulate with language themselves. Looking to the lower left for the semantic domain suggested by glitter, we see that this word invokes (via metonymy) a world of rich socialites and high society. Glitter suggests of course jewelry: diamonds, gold, perhaps sequins on rich dress; and also the general notion of visual salience (meta­ phorical brightness) that goes with overall elegance and polished appearance. Within this glittering world there is also an elite group—this time defined specifically with respect to the social domain. There can, of course, be refer­ ential overlap of these two worlds—for example, non-wealthy, but respected, literary types might possibly be sought after to attend the parties of rich boors in high society. A few relevant linguistic units are included in the two domains: in the one for GLITTER, the unit glittering, as found in the attested expression the glittering Mrs. [X] and similar, and in the domain for LITERATI, the word illu­ minati, which is typically specifically associated with a more learned domain rather than the entertainment or socialite world, and which also bears some formal resemblance to literati. These conceptual connections are likely to have been relevant to the creation of the blend in the first place, and possibly (in the case of illuminati) also to the further development of a pattern of -erati words (Section 5). The oval on top represents the conceptual structure of the resulting blend (what Turner and Fauconnier term the 'blended space'.) The concepts invoked in the blend are taken from the two source domains; the connecting links repre­ sent the conceptual connections of the blend with its source lexemes. It is important to note that the result of the conceptual blending of these two domains is not the sum of the source domains. Instead, it is very specific, with aspects selectively taken from the source domains and merged into a functional whole in the blended concept. Glitterati does not refer to a group of literary people who wear glitter eye-shadow, say, or glinting sunglasses; in fact, in its current usage, the idea of 'literary' is absent from the blend. Instead, the word refers to a particular kind of elite group, socially sought after, who meta­ phorically and metonymically 'glitter': in clothes, jewelry, and more abstractly, beauty, wealth, fame and prospects. In fact, it generally refers to 'Hollywood

86

SUZANNE KEMMER

royalty' or similar celebrated megastars in some creative domain, typically movies but perhaps also including fashion (supermodels) or musical enter­ tainment. In the blend, all idea of learning, intellectuals, and books is completely absent. The notion of expertise does seem to be a connecting link between the concept LITERATI and the blended concept; but in the blend the notion is real­ ized as the idea of talent as defined by the relevant industry, and perhaps, by the mass market. Other related concepts imported include creativity and professional success; and the idea of illustriousness appears in the new domain as celebrity, generally mass-market celebrity. From the 'glittering' domain of high society, some imported notions are wealth and ostentation, parties, and the focus on dress and ornament, which appears in the blend in relation to visual appearance more generally, including beauty. Absent in the blend is the idea of social aristocracy in the traditional sense of 'high society': the world of rich socialites, formal evening wear, benefits and galas to benefit institutions of high culture (museums, opera houses), and so forth. Thus, although glitterati profiles a concept that contains some elements that occur in both source domains, it has additional, emergent conceptual structure that goes beyond what is found in either source domain. Presumably, much of this structure comes from the typical contexts in which the word is currently used. Finally, let us consider the way the two source words are conceptually integrated in the blend in regard to what is usually thought of as semantic compositionality. The first element in glitterati is not simply integrated like a standard modifier of a head noun, in which the modifier describes the type of the head category. Instead, glitter names an abstract thing that gets its specific interpretation in the integrated unit from a metonymic association with the other source lexeme. Its interpretation depends, even more extremely than with simple compounds, on the other conceptual structures evoked in the expres­ sion—whether evoked by the source lexeme literati (e.g. 'illustrious people', 'parties', 'elites') or by the blend as a whole (the emergent structure referred to above). Similarly, literati does not itself exactly name a head category in the blend, but instead suggests a group like literati—a well-known elite, but not with literary inclinations or expertise; just talented in the context of the blended domain. The meanings of the parts, then, do not contribute independently to the whole, but depend on each other to an extreme extent. Hence the claim of especially intimate conceptual integration made above.

SCHEMAS AND LEXICAL BLENDS 5.

87

The creation of a new lexical schema

Now let us consider some words related to literati that have recently come into English. The schema apparatus of Cognitive Grammar allows us to model how such words arise, and how the linguistic system can change over time in response to changing patterns of usage. Glitterati, the first new formation found, appears in the Oxford English Dictionary in the following citation from 1956: (7) Bobbing and weaving about the premises are a passel of New York glitterati. There is a highbrow editor of a popular magazine who is keen on starting a new literary journal [etc.]. {Time Magazine) The original coiner of the word (who may or may not have been this writer) brought together two semantic domains, presumably under the influence of the sound similarity/overlap between the word literati (whose meaning may have been the rough target, if the context in the above example is representative) and the evocative word glitter. The next stage is the spread of the word to new speakers, who presumably on the basis of the form of the word and contextual cues, are able to interpret the meaning and recognize the clever word-play involved. The word gradually spreads throughout at least a subportion of the linguistic community, becoming more frequent and undergoing some semantic change on the way, in which it largely loses its 'literary' associations and picks up its glamorous Hollywood overtones from frequent use in such contexts. (The semantic developments of glitterati are fascinating in their own right, but will not be further pursued here.) Interestingly, glitterati is not the only formation based on literati that we find in the language. Let us look at some data for other neologisms that are quite similar to these words, and whose existence is plausibly related to them. The data in (8)-(11) below are the results of a search of the London Times/Sunday Times corpus for words ending in the string rati for 1990 and the three-year sequence of 1995-1997. All of the occurrences of words in rati are listed for each year searched, followed by the numbers of citations of each word for each year. These data show a wide variety of words that are reminiscent of literati. They are almost all interpretable out of context; further, when examined, the contexts are consistent with meanings involving 'elite groups' of various kinds. This leads me to claim that these words are based, directly or indirectly, on the form and meaning of literati. Speakers originally created these new words by using literati as a source word, and blending it with source words from other

88

SUZANNE KEMMER

domains to achieve the desired integrated concept. Note that this process, as analyzed here, is analogical in nature: an instance was used as a basis from which similar words with similar meanings, mutatis mutandis, could be created. (8) Times/ST 1990 literati glitterati chatterati designerati liberati numerati

43 31 1 1 1 1

(10) Times/ST 1996 glitterati 41 40 literati culturati 3 2 botherati Britpoperati 2 chatterati 2 1 flitterati

(9) Times/ST 1995 43 literati 36 glitterati 2 Briterati 2 digerati 1 chatterati culturati 1 1 luncherati 1 operati (11) Times/ST 1997 glitterati 47 literati 27 2 chatterati 2 digiterati 2 soccerati digerati 1 1 lefterati

Notice that in 1990 and 1995, instances of literati outnumber those of glitterati. This is in fact the general pattern in the first half of the 1990s (and no doubt earlier). In 1996, however, for the first time glitterati outnumbers literati—and in 1997, it does so by a significant margin. These observations suggest that literati was the original model for all words of this type; but that with the growing frequency (indicating conventionalization, or cognitive entrenchment) of glitterati, the latter has now become a more entrenched unit and hence a possible model in its own right for further extensions. However, another important innovation between 1990 and 1995 is suggested by these data. In 1990, all of the source lexemes that blended with literati had the string er as part of the root lexeme. Thus, all the instances in that year were cases of overlap blends, just like glitterati. In some of these cases, the root of the source lexeme was truncated (liberati "liberal intelli­ gentsia" from [liberal x literati]; numerati "the mathematical elite" from [numer- or numerate x literati]), but the er was nonetheless part of both roots.

SCHEMAS AND LEXICAL BLENDS

89

But in 1995, we find Briterati "ruling British [pop artists]", digerati "computer elite", and even luncherati "[powerful publishers] who do business/wield power over lunch", in which the er certainly is not part of the first root. It could—and presumably must have originally—come from literati, i.e. in each instance of a new word based on literati the er was part of the model source lexeme and was simply taken into the blend. On the other hand, given the entrenchment of a second word, namely, glitterati, already strong in 1990 and ever more frequent since then, there is another hypothesis for the occurrence of er in individual blends. It is plausible that, given the numerous instances of overlap blends containing er, speakers have extracted a general schema for a new morpheme -erati from instances of this type. If this is the case, then the recurrent string erati could be said to have been reanalyzed as a morphemic unit, given its recurrent form and the recurrent meanings associated with it. The new unit thus formed could then be attached concatenatively to new roots or parts of roots of any form. The latter scenario is essentially the process by which a new morpheme emerges diachronically via reanalysis. The emergence of a morpheme, I would suggest, is a gradual process, proceeding usage event by usage event, both within and across speakers. We call the new unit a morpheme once it has reached a certain arbitrary threshold of entrenchment in individual speakers and once it has diffused through the speech community to some arbitrary degree, often defined by some particular behavioral or distributional criteria set up by the linguist which is symptomatic (after the fact) of its changed status. On this view, there is no sharp dividing line between morphemes and expres­ sions on the way to becoming morphemes; we call them morphemes when they have become as conventionalized as other morphemes and behave accord­ ingly.13 The data also suggest that a few other instances have attained some degree of conventionalization, e.g. digerati and chatterati "the chattering classes; the intellectual elite who have access to the media to express their views". The more conventionalized units of this type there are, the more likely speakers are to be forming new instances from a generalized pattern rather than directly from individual instances. It should be stressed, however, that speakers may have multiple competing sanctioning structures that might license the same output in some instances. For example, given that instances in which er is present in both source lexemes are still very frequent, it is likely that the schema involving an overlap blend is still reasonably strong, and may coexist with a schema containing a suffix -erati. Similarly, the phonological shape of the blend is still highly influenced by the forms of the original source words;

90

SUZANNE KEMMER

hence the relatively large number of instances that end in terati, e.g. Briterati, flitterati, chatterati, lefterati, digiterati. To the extent that specific phonological shapes are conventionalized (hence entrenched) as guides for forming new words of this type, we can posit in the network schemas invoking these specific forms.

Figure 3: Some schemas licensing-erati

blends

Figure 3 shows a possible partial network structure for schemas sanctioning words in this lexical family. The phonological poles of each schema show the general form that the instances must take to fit the schema. Instances that fit the schema exactly are linked to it via solid arrows pointing to the instance—they are related to the schemas via a specificity relation, in which the instance is a more specific version of the schema. Instances that do not quite fit a schema relate to it via extension. This relation is represented by means of dotted arrows in the figure.

SCHEMAS AND LEXICAL BLENDS

91

Schemas can also be related to other schemas via specificity or extension. The figure shows a relation between two of the schemas involving specific syllable structures. The top left-hand schema, with its final consonant specified as a /t/, is simply a more specific instance of the schema below it, which has a schematic consonant. Thus, a solid arrow links the more schematic and more specific schemas. Entrenchment is shown in the figure by means of the thickness of the lines of the boxes representing units. Literati and glitterati, being the most frequent and hence heavily entrenched units, are shown in heavy-lined boxes. Chatterati is shown as having intermediate entrenchment, and the rest are shown as having the ordinary degree of entrenchment of lexical items. The lexical family spawned from literati/glitterati originally involved an over­ lap blend. However, probably the most common type of blend generating such word families is the substitution blend, in which one source lexeme replaces part of the other. This replacement brings to bear a new conceptual domain in the blend, which then selects aspects of each domain to highlight in the coher­ ent overall conceptualization. An example of a substitution blend giving rise to a new family of words is the famous case of -gate. Watergate was the original source lexeme that was then blended with various other lexemes, in each case linking the relevant domains. -Gate took its domain (which included such notions as '[presidential] scandal; illegal maneuvers by the executive branch') from the whole source lexeme Watergate. Once -gate was attached to one or two other instances, it could be seen as a general pattern and thus further exten­ sion would be even more likely (because the general pattern would have achieved a certain level of entrenchment). Thus, we now have a word family that includes Watergate, Contragate, travelgate, Monicagate, Camillagate, Notting Hillgate,14 etc. A similar example, but this time without accompanying reanalysis, is -ware as in hardware, software, wetware, freeware, shareware, fatware, vaporware, etc. 6.

Conclusion

I hope to have shown that lexical blends, although apparently a rather unusual phenomenon, are nevertheless not so extremely different as to remain outside of a general theory of language. Blends can be integrated into a theory of morphology and lexical structure, despite their heterogeneity and special formal characteristics, provided we use a theory that is not restricted to the building-block model. Using a schema-based theory, in particular Cognitive Grammar, no additional mechanisms are required beyond independently moti­ vated constructs. Even the modification of the process of syntagmatic inte-

92

SUZANNE KEMMER

gration suggested in the analysis in Section 3 is not an unusual step; some notion of different degrees of conceptual integration are required anyway for capturing the difference between compounds and non-conventionalized lexical phrases. Thus, the use of schemas makes it possible for blends to be naturally subsumed into the family of morphological processes that occur in the languages of the world. The building-block model tends to downplay the extent of non-concatenative morphological phenomena, which comprise a wide range of phenomena including not only fusion at morpheme boundaries, but also 'cranberry morphs', stem-changing morphology such as umlaut and ablaut, and Semitic-type root pattern morphology. Rubba (1993) shows how schemas handle these types of phenomena, particularly the last. The present account adds lexical blends to this list, providing one more step towards a linguistic theory comprehensive enough to handle the complete range of empirically observed processes in human language.15 The present study is in many ways preliminary and programmatic. Many questions remain about lexical blending, and many of the theoretical claims made here about processing (specifically, the activation of domains) that occurs when a blend is produced or comprehended are hypotheses that require empirical testing, and indeed, the development of methods that would make them even susceptible to testing. The claims regarding frequency made on the basis of the model also require testing; many predictions are generated by the usage-based account that are in principle testable.16 Also worthy of future research is the question of the relation of blend neologisms to speech error blends, given that, on the account presented here, both involve online proc­ essing. Lehrer (1996), an investigation of the factors that influence the interpretability of blends that are novel to speakers, has made a valuable beginning towards the experimental study of lexical blends (including some discussion of their relation to speech errors). A next step might be to study the processing of blends already known to the speaker, using, for example, priming method­ ology. Moving from lexical blends to a broader picture, I will conclude by noting that the analysis presented here fits very well into the general theory of blending developed in Turner and Fauconnier (1995). Turner and Fauconnier's account goes beyond lexicon and morphology to embrace the general relation of conceptual structure and formal expression.17 Lexical blends in this larger view are just one type of blending, in which form happens to be blended as well as concepts. They apply their general blending apparatus to syntactic constructions in Fauconnier and Turner (1996). Barlow (2000) investigates

SCHEMAS AND LEXICAL BLENDS

93

blending in syntax in greater depth, using schemas to account for syntactic blends involving idioms and in general showing that syntax is thoroughly permeated by blending processes. I hope with this paper to have awakened further interest in the general process of blending, whose role in linguistic structure—and cognition more generally—is hard to overestimate. Notes 1 I am grateful to Ada Rohde and Anatol Stefanowitsch for very helpful comments and discus­ sion as well as information on blends in German, as well as to two anonymous reviewers who provided useful comments. A small portion of this paper was presented in Barlow and Kemmer (1998), and thanks are due to Ronald Langacker, Gilles Fauconnier, and Eve Sweetser for valuable discussion. Special thanks to Carlos Nash for his help in collating blend examples from my collections of neologisms and for maintaining the associated websites (cf. note 3); and to Anatol Stefanowitsch for computerizing the figures. 2 German has a few examples of blends constructed of native material, e.g. Verschlimm­ besserung "attempted improvement that only makes things worse" [Verschlimmerung "worsening"' x Verbesserung "improvement"]. In recent years, blending has become popular in advertising and product names (e.g. Elephantastisch, an ad for Carlsberg Elephant beer; Chocolaccino "chocolaty cappuccino drink"); these often include recently borrowed lexical material. Given its relatively restricted realm of application and its affinity towards the borrowed (and less integrated) 'edges' of the lexical system, we can say that lexical blending is probably a recently adopted device that is at present unquestionably far more marginal in German than in English. 3 Data for this paper are taken from New Words in English and New Word Journal, two collec­ tions of neologisms assembled by students in Linguistics 215, Words in English, taught at Rice University, 1996-1998 (see Data Sources for websites containing these collections), supple­ mented by my own blends collection, in total more than 100 blends; and also, for the data in Section 4, the London Times and Sunday Times, from which families of specific blends were collected using the text analysis program MonoConc 1.5 (Barlow 1998). 4 I adopt the view of the lexicon proposed by Langacker (1987 and elsewhere), in which lexical items are access points to knowledge structures. Lexical items are linked not only to conceptual information but to each other in a network of relationships. See also Lamb (1998) for a fundamentally similar view with a somewhat different notion of the nature and types of relations in the linguistic/conceptual network. 5 The quantity of phonological overlap differs depending on the dialect of English spoken. In my Midwestern American dialect the first part of -abilia is pronounced identically to the part of automobile following the /m/. 6 The possibility of intercalative analyses for examples like those in (4), and the occasional production of intercalative blends as speech errors, perhaps served as an inspiration (and licensing factor) for examples like (3) as a creative extension of the basic blend formation mechanism. 7 Since phonological strings can in other cases also trigger meaning (e.g. phonaesthemes, clipped lexemes), it is reasonable to suppose that morphemes are simply a special case of the more general phenomenon of the association of sounds with meanings, namely, the case in which the connection of a sound string with a meaning is very well-entrenched and the unit is

94

SUZANNE KEMMER

recombinable. On this view, there is no clear dividing line between 'pure' sound strings and stretches of sound that are symbolic. 8 The term 'schema' is also used for a semantic category that abstracts over the properties of its instances or exemplars. This is in contrast to a 'prototype' semantic category which contains information only about the 'best examples' of a given superordinate category (see Langacker 1991). The two senses of schema are related in that they are both extractions of commonalities by speakers. The sense used in this paper, however, is more general in that it is not limited to semantic categories in particular. 9 It is no coincidence that lexical blends are particularly endemic to journalese, in which writers are searching for new and striking means of expression. They are not, however, restricted to that genre, nor do they occur exclusively in written language (cf. Algeo 1977). An investigation of the relation of genre to blends remains to be done. 10 The logo of the Nike Corporation, which resembles a rounded-off check-mark, has come to be referred to in English as the 'swoosh'. Comparison with the swastika makes reference to the hegemony of Nike in the world of sportswear and marketing, the blind logo-worship of unthinking masses of people, and Nike's treatment of its third world workers. 11 Analogy-driven models of language have been developed by a number of other linguists besides Bybee and Langacker, including Esa Itkonen, Sydney Lamb, Royal Skousen, and connectionists such as Jeffrey Elman. 12 In blends, the specially integrated formal structure seems to mirror the well-integrated conceptual structure. It seems likely that this iconicity is a motivating factor in the choice of a blend over a similar compound or phrase, where there is a choice (e.g. affluenza vs. influenza of the affluent). 13 Rohde and Stefanowitsch (1998) describe a similar process for the emergence of borrowed morphemes from instances of loanwords containing them, e.g. the development of the deriva­ tional morpheme -ry as in bravery, bakery in the history of English. Further, the development of bound morphemes from lexical elements via the well-known process of grammaticalization is hypothesized on this model to follow a similar gradual trajectory proceeding usage event by usage event. 14 Notting Hillgate refers to a scandal in which British cabinet minister Peter Mandelson secretly accepted a real estate loan from a political colleague. This example is especially inter­ esting and amusing because the name of the real estate development containing Mandelson's house was Notting Hill Gate—making one of the source lexemes actually identical to the blend. 15 w h y certain types of morphology such as concatenative morphology are more frequent than others is a separate question whose answer most probably lies in cognitive preferences or biases that relate to processing and are not necessarily specific to language (e.g. preference for wholes over parts, continuous over non-continuous entities, etc.). 16 For instance, for known blends, the more frequent the blend, the less strongly activated its relevant semantic domains should be (just as in general, more frequent words should be less susceptible to morphological analysis in processing). Practical difficulties may arise from the fact that like all neologisms (and indeed, all non-frequent words), frequencies of blends will be extremely low. 17 See http://www.wam.umd.edu/~mturn/WWW/lending.html#ARTICLES for a bibliography of articles on blending that relate to Turner and Fauconnier's general model of blending.

SCHEMAS AND LEXICAL BLENDS

95

Data Sources New Words in English, http://www.owlnet.rice.edu/~ling215/NewWords/ index.html, an annotated collection of neologisms. Produced by Linguis­ tics 215 at Rice University, taught by Suzanne Kemmer, 1996 and 1998. New Word Journal, http://dacnet.rice.edu/projects/ling215/wordjournal/list, a similar annotated collection. Produced by Linguistics 215 at Rice University, taught by Suzanne Kemmer, 1997-1998. The Times and Sunday Times corpus on CD-ROM, 1990, 1995-1997. London: Times Network Systems Ltd. References Algeo, John. 1977. "Blends: A Structural and Systematic View". American Speech 52.47-64. Barlow, Michael. 1998. MonoConc 1.5. Houston: Athelstan Publications. . 2000. "Usage, Blends, and Grammar". Usage-Based Models of Language ed. by Michael Barlow & Suzanne Kemmer, 315-345. - Stanford, Calif. : CSLI Publications. & Suzanne Kemmer. 1994. "A Schema-Based Approach to Grammatical Description". The Reality of Linguistic Rules ed. by Roberta Corrigan, Gregory Iverson & Susan Lima, 19-42. Amsterdam: John Benjamins. & Suzanne Kemmer. 1998. "Idioms and Blending". Paper presented at the Fourth Conference on Conceptual Structures, Discourse, and Language (CSDL), Emory University, Atlanta. Bates, Elizabeth & Brian MacWhinney. 1987. "Competition, Variation, and Language Learning". Mechanisms of Language Acquisition ed. by Brian MacWhinney, 157-193. Hillsdale, N.J.: Lawrence Erlbaum. Bauer, Laurie. 1988. Introducing Linguistic Morphology. Edinburgh: Edinburgh University Press. Bybee, Joan. 1988. "Morphology as Lexical Organization". Theoretical Morphology ed. by Michael Hammond & Michael Noonan, 119-141. San Diego, Calif.: Academic Press. . 2000. "The Phonology of the Lexicon: Evidence from Lexical Diffu­ sion". Usage-based Models of Language ed. by Michael Barlow & Suzanne Kemmer, 65-85. Stanford, Calif: CSLI Publications. Cannon, Garland. 1986. "Blends in English Word Formation". Linguistics 24. 725-753.

96

SUZANNE KEMMER

Carroll, Lewis (pseud. of Charles L. Dodgson). 1871. Through the Looking Glass and What Alice Found There. Oxford: The Millenium Fulcrum Edition. Fauconnier, Gilles & Mark Turner. 1996. "Blending as a Central Process of Grammar". Conceptual Structure, Discourse, and Language ed. by Adele Goldberg, 113-130. Stanford, Calif: CSLI Publications. Hansen, Klaus. 1963. "Wortverschmelzung". Zeitschrift für Anglistik und Amerikanistik 11.117-142. Kemmer, Suzanne & Michael Israel. 1994. "Variation and the Usage-Based Model". Papers from the Thirtieth Regional Meeting of the Chicago Linguistic Society, Vol. 2, Parasession on Variation and Linguistic Theory ed. by Katharine Beals, Jeannette Denton, Robert Knippen, Lynette Meinar, Hisami Suzuki & Erica Zeinfeld, 165-179. Chicago, Ill.: Chicago Linguistic Society. Lamb, Sydney. 1998. Pathways of the Brain. Amsterdam: John Benjamins. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1, Theoretical Prerequisites. Stanford, Calif: Stanford University Press. - - - - - - - - - - . 1991. Concept, Image, and Symbol. Berlin: Mouton de Gruyter. ----------. 2000. "A Dynamic Usage-Based Model". Usage-based Models of Language ed. by Michael Barlow & Suzanne Kemmer, 1-63. Stanford, Calif: CSLI Publications. Lehrer, Adrienne. 1996. "Identifying and Interpreting Blends: An Experimental Approach". Cognitive Linguistics 7.359-390. MacWhinney, Brian. 2000. "Connectionism and Language Learning". Usagebased Models of Language ed. by Michael Barlow & Suzanne Kemmer, 121-149. Stanford, Calif.: CSLI Publications. Marslen-Wilson, William. 1990. "Activation, Competition, and Frequency in Lexical Access". In Cognitive Models of Speech Processing: Psycholinguistic and Computational Perspectives ed. by Gerry T.M. Altmann, 148-172. Cambridge, Mass.: The MIT Press. Pineros, Carlos-Eduardo. 1999. "Word-Blending as a Case of Non-Concatenative Morphology in Spanish". Rutgers Optimality Archive ROA-3430999. Rohde, Ada & Anatol Stefanowitsch. 1998. "Complex Morphemes: The Case of English -ery". Ms., Rice University. Rubba, Johanna. 1993. Discontinuous Morphology in Modern Aramaic. Ph.D. dissertation, University of California, San Diego. Soudanek, Lev. I. 1978. "The Relation of Blending to English Word-Forma­ tion: Theory, Structure, and Typological Attempts". Proceedings of the

SCHEMAS AND LEXICAL BLENDS

97

Twelfth International Congress of Linguists: Vienna, August 28September 2, 1977 ed. by Wolfgang Dressier & Wolfgang Meid, 462466. Innsbruck: Institut für Sprachwissenschañ der Universität Innsbruck. Turner, Mark & Gilles Fauconnier. 1995. "Conceptual Integration and Formal Expression." Journal of Metaphor and Symbolic Activity 10.183-203.

VALENCY AND DIATHESIS HEINZ VATER University of Cologne

Several linguists have pointed out severe discrepancies in existing valency descriptions. As a way out of this 'valency misery, Ágel (1995) proposes a differentiation between 'valency potential', inherent in a verb's lexicon entry, and 'valency realization', determined by the interaction of valency potential, sentence structure, and communicative factors. In the present paper, this approach will be applied to diathesis alternations of German verbs. A two level valency dictionary is proposed which attempts to unite a lexical entry's valency potential as well as its valency realizations in different diatheses. This account is exemplified by the verb

öffnen.

1.

Introduction

As Storrer (1992:118) and Jacobs (1994a:5) have pointed out, practically all of the available valency descriptions are characterized by severe discrepancies and inconsistencies, a situation which has been labeled 'valency misery'. As such, Jacobs (1994a:4) has claimed that valency should best be viewed as a collective term covering several different concepts and is thus of low value for linguists. Agel (1995), however, has tried to rescue the concept of valency by proposing a differentiation between 'valency potential' and 'valency realization'. In this paper, I will try to show the usefulness of Ágel's distinction and its appli­ cability to the analysis of a verb's valency as it manifests itself in the verb's diathesis alternations. My examples will be mainly taken from German (with English translations).1 Before embarking on this paper, I should shortly dwell on the notion 'diathesis' itself, as it is less commonly used in the English than in the German linguistic literature. Diatheses can be defined as "grammatische Zustände des Verbs (bzw. Verbcomplexes) wie z. B. Aktiv, Passiv, Antipassiv, Medium, Reflexiv, Kausativ ... u.a. mehr" [grammatical dispositions of the verb (or verbal complex), such as active, passive, antipassive, middle, reflexive, causa-

HEINZ VATER

100

tive, etc.]. The diathesis alternations of a verb differ in the number of comple­ ments (or arguments positions, in Wunderlich's 1993 terminology) and the way these complements are realized. As such, for instance, the 'normal' active diathesis of the verb to open, as in (la), requires three complements, which are realized as agent, patient, and instrument and whose syntactic functions are subject, direct object, and prepositional object, respectively. Then again, the deagentivized active diathesis of to open, as in (lb), only requires two complements, a patient and an instrument, which are realized syntactically as direct object and subject, respectively. (1) a. He opens the door with a key. b. The key opens the door. 2.

Valency

2.1

Valency in dependency and constituency frameworks

Tesnière (1969) used the term 'valency', which he had borrowed from chem­ istry, to describe relationships between a verb and a set of elements filling slots attached to this verb. These relationships are characterized as dependency relations, where the verb plays the role of 'governor' ('régissant') and the governed elements (usually nouns) play the role of 'governees' ('actants'). Valency, then, is defined as the number of 'actants' a verb is able to govern.2 Later, Tesnière extended valency to include other kinds of governors as well (especially nouns and adjectives). Valency was originally considered a purely syntactic relationship (cf. Tesnière 1969:12, who calls valency a lien syntaxique, and Heringer 1968:426, who describes Tesniere's theory as a syntactic valency theory).3 Similarly, Erben (1972:§487) tries to connect the syntactic valency of verbs with corre­ sponding primary sentence types.4 Later on, it was realized that semantic and pragmatic factors play a role in valency as well. Helbig and Schenkel (1978:60), for instance, emphasize that they restrict their representation to syntactic descriptions in specifying the valencies of the separate entries in their dictionary, but that there is a 'semantic valency' {semantische Valenz) connected with syntactic relationships, although not in a one-to-one corre­ spondence. Whereas many of Tesnière's followers (cf. Engel 1994; Erben 1972; Tarvainen 1986; Welke 1988) used a dependency framework for the descrip­ tion of valency, other linguists developed valency descriptions within other grammatical frameworks like constituent grammar (including Generative

VALENCY AND DIATHESIS

101

Transformational Grammar; cf. Chomsky 1965; Vater 1981)5 and categorial grammar (cf. Günther 1978; Zifonun 1992). In the following, I will describe valency relations within a constituent framework (cf. Vater 1981). Constituents are sequences of words resulting from the progressive segmentation of sentences. Constituents, Bierwisch (1970:16) emphasizes, are not characterized by segmentation alone; they have, at the same time, to be associated with syntactic categories like noun phrase (NP) or verb phrase (VP).6 An advantage of a grammar working with constituents and their syn­ tactic functions (like subject, objects, etc.) is that it can handle the placement of complements, their syntactic relationships within a sentence, and their connec­ tions with semantic roles (like agent, patient, recipient of an action, etc.) more easily and convincingly than a dependency grammar working exclusively with final elements, all of which are dependent of the verb in an equal fashion. Chomsky (1965), for instance, defines syntactic functions as relationships between dominated constituents and their dominating nodes, 'subject-of thus being defined as the relation between an NP and the dominating S node and 'object-of as the relation between an NP and the dominating VP. Constituents, their syntactic functions, and their semantic roles will turn out to be relevant tools in the following analyses. Finally, it should be kept in mind that the constituent structure of a sentence and verbal valency are closely interconnected (actually, this had already been observed by structural grammarians who thought that 'sentence types' ('Satzbaupläne') had to do with valency; cf. the quotation above in Erben 1972: §487). A phrase structure rule generating the different kinds of VP, for instance, would run into problems because verbs differ as to their valency (cf. Vater 1996); more specifically, a general VP rule generating a verb and one or two objects could not predict the precise number and possible forms of the complements connected with each special verb and would furnish ungrammatical sequences like *Paul schläft ein Bett "*Paul sleeps a bed". Fanselow and Felix (1987:46) solve this problem by positing a very abstract type of rule where possible complements are indicated by dots. (2) VP → ...V ... 2.2

Diathesis alternations and subject

In describing valency, account will need to be taken of the subject's prominent role in the sentence. As is well known, the subject plays a prominent role in a sentence, as it is the only NP complement in an agreement relation with the

102

HEINZ VATER

finite verb.7 More important, though—at least in the context of this paper—is the fact that the subject of a sentence derives its special role from its being 'diathesis-specific'; in other words, a verb's diathesis alternations usually differ in the choice of the complement serving as subject (cf. example (1); see also Wunderlich 1993:730). 3.

Valency potential and valency realizations

The two main problems in valency analysis concern (i) the distinction between semantic and syntactic valency and (ii) the distinction between complements and adjuncts (cf. Vater 1978). On several occasions, it has been pointed out that existing studies of verb valency are largely unsatisfactory. As such, Storrer (1992) finds huge discrepancies when comparing several valency analyses of the same verbs, which leads her to think that valency theory has landed in a blind alley.8 Jacobs (1994a: 110) attributes the 'valency misery' to the fact that valency has been defined according to seven different criteria, which are partially incompatible (for criticism of his view, see Agel 1993). Finally, Jacobs (1994b:288) notes that the realization of complements follows from mechanisms of sentence grammar rather than directly mirroring valency prop­ erties of lexical entries. In answer to this 'valency misery', Ágel (1995) suggests that a distinction should be made between a verb's ' Valenzpotenz' ('valency potential'), which is inherent in a verb entry in the lexicon, and a verb's 'Valenzrealisierung' ('valency realization'), which is determined by the interaction of valency potential, sentence structure, and communicative factors (like the speaker's intentions and his/her assumptions concerning the recipient's knowledge).9 In line with Agel's distinction, I would like to show that an adequate description of a verb's valency requires making a clear distinction between the constant part of the verb's valency (including the semantic functions or 'theta roles') and some variable properties dependent on the type of syntactic construction in which the verb occurs (including the number of actually occurring comple­ ments and their syntactic status). In Section 4, I will describe different valency realizations of verbs used in different syntactic constructions, concentrating on constructions in diathesis relations. Before turning to the description of these valency realizations, two prob­ lems need to be tackled: (i) It has to be decided whether (logico-)semantic factors or syntactic factors determine a verb's valency; (ii) It has to be decided how many variants of a verb have to be assumed in the lexicon.

VALENCY AND DIATHESIS

103

(i) Evidence has been offered that syntactic rather than semantic features of a verb determine the number and kind of its complements (cf. Höhle 1978; Vater 1981). Consider the following sentences: (3) a. Peter dined in a Chinese restaurant. a' *Peter dined chicken with rice in a Chinese restaurant. (4) a. There arose rebellions. a' *By the soldiers there arose rebellions. Although on the semantic (or pragmatic) level, the verb dine in (3a) implies something to be eaten, this element cannot be realized on the syntactic level (cf. '). This constraint does not depend on specific syntactic constructions; it has to be stated in the lexicon as a matter of the verb's valency potential. Similarly, in the arise-construction in (4), the agent cannot be expressed, although it is necessarily implied. Conversely, the subject it in cases like (5) fills a necessary syntactic position but does not correspond to any element in the verb's logico-semantic or pragmatic structure: the speaker does not really want to name an agent of the corresponding meteorological processes. (5) It is raining/snowing/hailing/getting dark. (ii) The second problem (the number of variants of a verb's lexical entry in the lexicon) is closely connected with the assumption of obligatory and optional complements. 'Complement X of Y' is defined as follows by Jacobs (1994b:284): X is a complement of Y iff one of the facts a)-d) is true: a) X is obligatory for Y, i.e. it fulfills a realization requirement by Y; b) X is subject to a requirement by Y as to its formal features (e.g. concerning its case); c) X is subject to a require­ ment by Y as to its semantic features (e.g. concerning the feature [animate]); d) X is an argument of Y, e.g. it fills a slot in the meaning of Y. [my translation]

On this basis, Jacobs (1994b:285) provides the following definitions of 'obligatory' and 'optional' complement: X is an obligatory complement of Y iff X is a complement of Y and obligatory for Y; X is an optional complement of Y iff X is not obligatory for Y but a complement of Y (i.e. form specific or semantically specific for Y or an argu­ ment of Y). [my translation]

104

HEINZ VATER

Since 'obligatory for Y' is already included in the definition of 'complement', the definition of 'obligatory complement' should read: X is an obligatory complement of Y iff X is obligatory for Y and at least one of the other deconditions of 'complement' is valid.10 Thus, in the variant illustrated in (6), the German polysemous verb gehen "to go" occurs with an optional prepositional phrase denoting the (local) goal; at the same time, its subject denotes an (animate) agent. In the variant in (7), gehen occurs with an obligatory (non-omittable) directional phrase, whereby the subject denotes an (inanimate) patient. (6) a. Paul geht zum Bahnhof. "Paul is going to the station." b. Paul geht. "Paul is going/walking." (7) a. Der Brief geht nach London. "The letter is going to London." b. *Der Brief geht. "*The letter is going." Importantly, Jacobs (1994b:286) maintains that the issue of obligatory vs. optional complements correlates with the number of variants of verb valency that need to be distinguished in the lexicon. As such, the occurrence of an optional complement with a verb presupposes an alternative valency for a lexi­ cal entry without the complement in question. Thus, to write is a verb for which bivalency as well as monovalency is predicted in the lexicon. To admire does not enjoy a corresponding alternative. 4.

Valency realizations of verbs in various diatheses

4.1

Personal active constructions

'Personal' syntactic constructions are those containing a VP which can be used in the first, second, and third person (singular or plural); cf. Latin amo, amas, amat, etc. The verb in (8) is monovalent, the verb in (9) is bivalent with an optional direct object, and the verb in (10) is bivalent with an obligatory direct object. Exmple (11) contains a trivalent verb with an obligatory direct and an optional indirect object. (8) a. Der Winter kommt. "Winter is coming."

VALENCY AND DIATHESIS

105

(8) b. *Der Winter kommt das Jahr. "*Winter is coming the year." (9) a. Paul sang ein Lied. "Paul sang/was singing a song." b. Paul sang. "Paul sang/was singing." (10) a. Paul beschrieb sein Abenteuer. "Paul described his adventure." b. *Paul beschrieb. "*Paul described." (11) a. Paul gab mir den Brief. "Paul gave me the letter." b. Paul gab den Brief. "Paul gave the letter." c. *Paul gab mir. "*Paul gave me." d. *Paul gab. "*Paul gave." 4.2

Impersonal active constructions

Impersonal constructions occur only in the third person singular, as in Latin oportet "it is necessary". In English and French, they are usually accompanied by a subject pronoun (it and il, respectively); in English, there are also imper­ sonal constructions with there and a non-agent subject (in the singular or plu­ ral). In German, impersonal verbs usually require the subject es "it".11 (12) Es regnet/schneit/hagelt/wird dunkel. "It is raining/snowing/hailing/getting dark." (13) a. Jemand/der Mann klopfte. "Somebody/the man knocked (at the door)." b. Es klopfte. "*It knocked." (= "There was a knock at the door.") The subject es "it" can be omitted if the first position of the sentence is filled by another constituent; this structure occurs especially with verbs expressing an emotion:

HEINZ VATER

106

(14) a. Es friert/hungert dürstet/ekelt ihn.12 "*It freezes/hungers/thirsts/disgusts him." b Ihn friert/hungert/dürstet/ekelt. "*Him freezes/hungers/thirsts/disgusts."

(Eisenberg 1994:194)

In English, these examples are rendered by personal constructions or predicates containing a copula and an adjective: "He is freezing/starving (hungry)/thirsty/ disgusted". In PRO-drop languages, impersonal constructions usually occur without a subject (cf. Italian piove "it is raining", Latin oportet "it is neces­ sary"; Polish trzeba "it is necessary"). In many cases, there is no corresponding personal active construction, although German offers examples in which the same situation can be expressed either by a personal or by an impersonal construction: ich schaudere/mich schaudert "I shudder". 4.3

Personal passive constructions

4.3.1 Werden-passive Passive constructions are the most prominent instance of grammatical converses (cf. Zifonun 1992:254). They are 'agent absorbing', in that they imply an agent even if it is not expressed.13 According to Hansen (1967:48), only ten percent of all passive sentences in Danish contain an agent; still, many types of passive allow for an explicit agent (e.g. expressed by a prepositional phrase). Morphosyntactically, the passive is realized in different ways: -

-

In many languages, an auxiliary verb is added (in English be, in German sein "to be" or werden "to become") and the main verb is realized as a non-finite form (usually past participle). In other languages like Latin, the passive is formed synthetically, with special endings: amamus "we love", amamur "we are loved". Some languages (e.g. the Scandinavian ones) make use of both possibilities. In Danish, the choice depends on morphological and semantic conditions: the past tense of strong verbs is in most cases not compatible with the synthetic passive (cf. Hansen 1967:46). On the other hand, the synthetic passive is preferred if a habitual state is expressed, as in her slibes sakse "here, scissors are sharpened" (Hansen 1967:51).

In passive constructions, syntactic functions are associated with different theta roles than in active constructions: the subject NP in active sentences usually

VALENCY AND DIATHESIS

107

refers to the agent of an action, while the object NP refers to the patient; the subject NP in passive sentences is associated with the patient role, while the agent is optionally realized as a PP. In German, only the prototypical passive constructed with werden is characterized by all of these features; cf. (15).14 (15) a. Die Gäste tranken den Wein. 'The guests drank the wine." b Der Wein wurde von den Gästen getrunken. "The wine was drunk by the guests." . Der Wein wurde getrunken. "The wine was drunk." 4.3.2 The sein-passive The German 'stative' passive constructed with sein never allows for an agent to be expressed: (16) a. Das Theater ist geschlossen. 'The theater is closed (= is no longer open)." b. *Das Theater ist von der Stadtverwaltung geschlossen. "*The theater is closed (= is no longer open) by the city council." 4.3.3 The bekommen-passive A third type of passive in German, constructed with bekommen "to get", asso­ ciates the recipient role (connected with the indirect object in the active construction) with the subject NP (cf. Leirbukt 1977, 1997; Abraham 1985; Wegener 1985a, 1985b). Since this kind of passive is also formed with verbs expressing a removal (cf. 17), bekommen obviously has lost its original meaning and has become an auxiliary (cf. Reis 1976). Hentschel and Weydt (1995:181) point to the fact that the bekommen-passivo obeys more regularities than the sein-passive. (17) Er bekam den Führerschein entzogen. "He had his driver's license taken away."

(Leirbukt 1997:209)

By way of summary, Table 1 shows the different valency realizations (includ­ ing syntactic functions and theta roles) of the complements of trivalent verbs like geben in four different diatheses.

108

HEINZ VATER

AGENT PATIENT RECIPIENT

ACTIVE

WERDEN-PASSIVE

SEIN-PASSIVE

BEKOMMEN-PASSIVE

subject direct object indirect object

(prep. object) subject indirect object

Ø subject indirect object

(prep, object) direct object subject

Table 1: Complements of trivalent verbs in different diatheses

4.4

Impersonal passive constructions

Monovalent verbs (cf. 18) as well as bivalent verbs (cf. 19) allow for an impersonal passive in German (the English equivalents are not grammatical). As can be seen, it is not necessary for the topic position to be filled with a nominal (like es "it") or prepositional constituent (like den ganzen Tag "the whole day"), as the participle is an adequate filler as well (cf. 18c). Bivalent verbs usually do not allow for agent or theme realizations (cf. 19): (18) a. Es wurde gearbeitet. "*It was worked." ("= People were working/at work.") b. Den ganzen Tag wurde gearbeitet. "*The whole day was worked." (= The whole day people were working.")  Gearbeitet wurde. "*Worked was." (= "People were working.") (19) a. Es wurde gegessen/getrunken/gelesen. "It was eaten/drunk/read." (= "People were eating/drinking/ reading.") b. *Es wurde von den Gästen gegessen/getrunken/gelesen. "*It was eaten/drunk/read by the guests." (= "The guests were eat­ ing/drinking/reading.") c. *Von den Gästen wurde gegessen/getrunken/gelesen. "*By the guests was eaten/drunk/read." (= "The guests were eat­ ing/drinking/reading.") Fixed syntagms like Karten spielen "to play cards" can form an impersonal passive keeping the theme (which is probably to be interpreted as an accusative object); note that example (20) does not show agreement: (20) Heute wird Karten gespielt... (Plank 1993:153) "*Today is played cards." (= "Today we are playing cards ...")

VALENCY AND DIATHESIS 4.5

109

Impersonal reflexive passive

While many speakers of German still consider the impersonal reflexive to be ungrammatical (or rather, not to be standard), numerous examples of this construction can be found throughout various periods of German; cf. Andresen (1854), Behaghel (1924:214f), and Erben (1980:85). Today, the impersonal reflexive passive occurs quite frequently in everyday speech as well as in newspaper articles, on the radio and on TV; cf. (21).15 In everyday speech, this construction is found especially frequently in sentences intended as requests like (22), an example that is found in almost all grammars of German: (21) Da wird sich ein Urteil angemaßt über Dinge, von denen man nichts versteht. "*There a judgement is claimed about things people do not know anything about." (= "There people claim to be able to judge some­ thing they do not know anything about.") (WDR III, Themen der Zeit, 12.12.97, 1245) (22) Jetzt wird sich gewaschen. "*Now it is washed oneself." (= "Now you have to wash yourself/ yourselves.") (Duden-Grammatik 19955:178) It has to be mentioned that this kind of impersonal passive is not confined to spoken language. Behaghel (1924:214f), Erben (1980:85), and Duden-Gram­ matik (1995:178) quote examples from literary language (e.g. Da wurde...in zitternder Angst sich verkrochen, Clara Viebig). The famous German linguist Jacob Grimm seems to have liked this construction, as can be seen from the following example quoted in Behaghel (1924:215): (23)... konnte sich nicht auf die einzelnen Drucks eingelassen werden. "*... could not be involved itself with the separate printings." ("= ... could not get involved with the separate printings.") (J. Grimm, preface to Deutsches Wörterbuch, LXXXV) 4.6

Middle constructions

4.6.1 Agentless middle constructions in the active form According to Abraham (1987:3ff.), middle constructions can be defined as syntactic constructions lacking a surface (agentive) subject but implying it.

110

HEINZ VATER

Usually, a further verbal complement—theme, locative, or instrumental—has to be realized. Like the passive constructions, middle constructions belong typically to the agent-absorbing diatheses but, unlike them, they share the formal characteristics of actives (including reflexive constructions). In some cases, they may involve a non-animate (or even abstract) cause as well as an agent (cf. 24b).16 There are verbs that occur in two different kinds of active construction: (i) 'real' or 'normal' actives with an agent subject and a thematic object (24a, 25a), and (ii) 'deagentivized actives' with a thematic or instrumental subject (and possible further complements like goal and locative) (24b, 25b). Although the relationship between 'real active' and 'deagentivized active' constructions is not formally marked, it behaves as a diathesis relationship (comparable to the active-passive relation). Note that in generative grammar, verbs occurring in the second type of construction are called 'ergative' or 'unaccusative'. As it appears that both terms are inadequate, I have chosen to call this phenomenon 'deagentivization'. While deagentivized actives are typically intransitive, there is a transitive subtype with an instrumental subject and a thematic object (cf. 26b). In cases such as (26), then, the patient as well as the instrument can be made the subject of a deagentivized middle construction: (24) a. Maria verbrannte den Brief. "Maria burnt the letter." b. Der Brief verbrannte. "The letter burnt." (25) a. Der Mann/der Wind zerbrach das Glas. "The man/the wind broke the glass." b. Das Glas zerbrach. "The glass broke." c. *Das Glas zerbrach von dem Mann/von dem Wind. "*The glass broke by the man/by the wind." d. Das Glas zerbrach dem Mann/*dem Wind.17 "*The glass broke to the man/*to the wind." (26) a. Das Geschäft öffnet. "The shop opens." b. Dieser Schlüssel öffnet jede Tür. "This key opens every door."

(Helbig & Schenkel 1978:381)

VALENCY AND DIATHESIS

111

It has to be pointed out that with öffnen patient subjects seem to be restricted to noun phrases designating certain kinds of institutions: e.g. Das Theater /Museum/Kaufhaus öffnet "The theater/museum/department store opens" vs. *das Haus/*Tor/*Fenster öffnet; the corresponding English constructions are grammatical, cf. The house/gate/window opens. In these cases, the reflexive middle diathesis has to be applied in German: Das Haus/Tor/Fenster öffnet sich "The house/gate/window opens itself' (cf. 4.6.2). There do not seem to be corresponding restrictions for instrumental subjects: Der Dosenöffner öffnet die Dose "The can opener opens the can"; Peters Messer öffnete die Auster "Peter's knife opened the oyster", etc. 4.6.2 Reflexive middle constructions In German, reflexivization is a frequently occurring alternative to detransitivization (cf. 27b). In Polish, reflexivization is actually the only way to absorb the agent in these cases: Waza zbiła się "*The vase broke itself' (= "The vase broke"). (27) a. Das Mädchen öffnete die Tür. "The girl opened the door." b. Die Tür öffnete sich. "The door opened itself." (= "The door opened.") In German (as well as in French) there is a special type of reflexive middle construction having a modal component (i.e. designating potentiality); the cor­ responding active and passive diatheses contain the modal verb können "can, to be able to": e.g. Man kann das Rätsel leicht lösen "You can easily solve this riddle" / Das Rätsel kann leicht gelöst werden "The riddle can be solved easily". Typically, this construction needs a complement (an obligatory adjunct, according to Steinitz 1969) functioning as a modal adverbial, usually in the form of an adjective—especially leicht "easily" or gut "well" (cf. Abraham 1987; Vater 1988)—or a prepositional phrase like mit Genuß "with pleasure". With instrumental subjects of middle constructions, the verb occurs in its intransitive (deagentivized) rather than its reflexive form (cf.29). (28) a. Der Wein trinkt sich gut. "*The wine drinks itself well." (= "The wine is palatable.")

112

HEINZ VATER (28) b. Das Rätsel löst sich leicht. "*The riddle solves itself easily." (= "The riddle can be easily solved.") (29) a. Der Bleistift schreibt (*sich) gut. "The pencil writes (*itself) well." b. Das Messer schneidet (*sich) gut. "The knife cuts (*itself) well."

4.6.3 Reflexive middle constructions with lassen "to cause" Reflexive middle constructions containing the verb lassen in causative use, as in (30), are an alternative to reflexive middle constructions implying poten­ tiality (cf. 28-29). Instrument expressions are not possible as subjects but can occur as adverbials (in the form of prepositional phrases), as in (31); the subject es is optional. (30) De Wein läßt sich gut trinken. "*The wine lets itself drink well." (= "The wine is agreeable to the palate.") (31) Mit dem Bleistift läßt (es) sich gut schreiben. "*With the pencil (it) lets itself write well." (= "The pencil is perfect for writing.") Whereas simple reflexive middle constructions do not seem to allow for an instrument expression, instruments can be added with lassen-constructions. Manner adverbials like leicht "easily" are optional. (32) a. *Die Tür öffnete sich (leicht) mit diesem Schlüssel. "*The door opened itself (easily) with this key." b. Die Tür ließ sich mit diesem Schlüssel (leicht) öffnen. "*The door let itself (easily) open with this key. (= "The door could be opened (easily) with this key.") 4.6.4 Middle constructions with a copula, zu + infinitive Like the reflexive middle constructions treated in Sections 4.6.2 and 4.6.3, this diathesis implies potentiality. Manner adverbials such as leicht "easily" or gut "well" seem to be optional in German but obligatory in English. The construc­ tion without the manner adverbial is ambiguous between a potential and a

VALENCY AND DIATHESIS

113

deontic reading (expressing the necessity to do something), which is rendered in English as the riddle has to be solved. (33) Das Rätsel ist (leicht) zu lösen. "The riddle is easy to solve." There are still more types of middle constructions occurring in German (and English), e.g. a construction containing a copula, an optional adjective in adverbial function and an adjective with the suffix -bar/-ble; cf. Das Rätsel ist leicht lösbar "The riddle is easily solvable".18 In the conclusion, I will give a synopsis of verbs occurring in several middle diatheses. 5.

Conclusion

The fact that every verb allows for a range of different valencies depending on the syntactic constructions it participates in has troubled all linguists dealing with valency. In the first period of valency grammar, linguists believed in a fixed valency schema associated with each verb. Thus, Tesnière (1969:108) posited that the subject is "le prime actant" [the first actant], expressing "celui qui fait l'action" [the one who takes the action]. But in dealing with a verb's diathesis alternations, he has admit that facts are quite different in the passive voice, where the prime actant "subit l'action" [undergoes the action] (1969:242). These statements are quite unconnected. It is even harder to deter­ mine the valency of a verb if we have to do with examples like (25a) Der Mann zerbrach das Glas "The man broke the glass" and (25b, d) Das Glas zerbrach (dem Mann) "The glass broke (*for the man)", both being in the active voice but with the subject assuming different roles (agent vs. patient). It is problems such as these that have given rise to the 'valency misery' as described by Jacobs (1994a:l 1). A way out of this 'misery' has been offered by Ágel (1995), who proposes differentiating between valency potential and valence realization, thus allowing for a description of the different realizations of a verb's valency due to structural differences of the syntactic environments. As such, the verb geben "to give" (requiring three complements) is realized with its agent surfacing as the subject of an active construction, whereas the theme is made the subject in the werden -passive, and the recipient surfaces as the subject of the bekommen-passive (cf. Table 1). Whereas the agent comple­ ment is obligatorily realized (as the subject) in an active sentence (cf. 34a vs. 34b), it is suppressed in the German sein-passive and in middle constructions (cf. 35), and it is optional in the German werden -passive (cf. 36):

114

HEINZ VATER

(34) a. Ein berühmter Architekt baute das Haus. "A famous architect built the house." b. *baute das Haus "*built the house" (35) a. Das Haus ist (*von einem berühmten Architekten) gebaut. "The house is finished (*by a famous architect)." / "The building of the house is ready (*by a famous architect)." b. Das Buch liest sich gut (*von den Studenten). 'The book reads easily (*by the students)." (36) Das Haus wird (von einem berühmten Architekten) gebaut. "The house is being built by a famous architect." The middle constructions of German offer good examples of different valency realizations of the same verb. For instance, the three complements of the verb öffnen (each of them associated with a special thematic role) show up with different syntactic functions in the 'normal' active and in the middle diatheses. The middle diatheses may have an actual (non-potential) or a potential meaning. 'Deagentivized active' middles allow for an actual as well as for a potential reading; in the latter case, however, the 'potentiality' needs to be expressed by the modal verb können (cf. 37a), as in the normal active (cf. 37b). In the reflexive and copulative middle constructions, the potentiality element is not expressed, since it is a component of the meaning of these diatheses (cf. Sections 4.6.2-4.6.4). When the middle diatheses express potentiality, a fourth complement—whose thematic role is here called 'modalizer'—comes into play. It is obligatory in the reflexive middle diathesis (cf. 38a) but optional in all other diatheses;19 in other words, reflexive middles without a modalizer simply get a non-potential reading (cf. 38b). (37) a. Der Schlüssel kann die Tür öffnen. "The key can open the door." b. Man kann die Tür mit diesem Schlüssel öffnen. "One can open the door with this key." (38) a. Die Tür öffnet sich leicht. "*The door opens itself easily." (= "The door opens easily.") b. Die Tür öffnet sich. "*The door opens itself." (= "The door opens.")

115

VALENCY AND DIATHESIS

Table 2 shows the valency realizations of öffnen expressing non-potentiality, while Table 3 shows the valency realizations of öffnen expressing potentiality. 'NORMAL' ACTIVE AGENT

PATIENT INSTRUMENT

DEAGENTIVIZED ACTIVE

subject

Ø

direct object prepositional object

direct object subject

Table 2: Complements of öffnen in the 'normal' active and deagentivized active diatheses expressing non-potentiality

AGENT PATIENT INSTRUMENT MODALIZER

'NORMAL' ACTIVE

REFLEXIVE MIDDLE

LASSENMIDDLE

COPULA + ZU + INFINITIVE

Subject direct object prep, object (adverbial)

Ø subject Ø adverbial

Ø Subject (prep, object) (adverbial)

Ø subject (prep.object (adverbial)

Table 3: Complements of öffnen in the 'normal' active and m iddle dia th es es expressing potentiality

Valency descriptions of the traditional kind which do not take different valency realizations into consideration run into problems or give a defective picture. Thus, the description of öffnen "to open" in Helbig and Schenkel's (1978) valency dictionary runs as follows: (39)

öffnen öffhen2(Vl = aufmachen) öffnen → Sn, Sa Sn →Hum (Der Junge öffnet die Tür.) Sa →-Anim (Er öffnet die Büchse.) I. öffnen1 (V2 = aufgemacht werden, aufmachen) IL öffnen → Sn III. Sn→-Anim (Das Kaufhaus öffnet). (Helbig & Schenkel 1978:381)

I. IL III.

In this notation, Roman numbers refer to three levels of valency description. Level I. indicates the number of obligatory and optional complements of each variant of the verb and its meaning; two obligatory complements are registered with the first variant of öffnen and one obligatory complement with the second

HEINZ VATER

116

variant. Level II. specifies the morphological form of the complements, 'Sn' designating a noun (substantive) in the nominative case and 'Sa' designating a noun in the accusative. Level III. represents the semantic makeup of the complements (by means of semantic features and examples). The authors point out that the two variants are in a transformational relationship: Er öffnet das Geschäft "He opens the shop" (VI) → Das Geschäft öffnet "The shop opens" (V2). A more adequate valency description has to consider the valency reali­ zations of the complements in different diathesis alternations, as shown in Tables 2 and 3. To that effect, the valency description I propose distinguishes between at least two levels (cf. 40). Level 1 lists the number of potential complements and their thematic roles. Contrary to Helbig and Schenkel (1978), Level 1 does not indicate the number of optional complements, since this cannot be decided at this level. Level 2, then, indicates the syntactic reali­ zations in different diathesis alternations as well as whether the complements are obligatory or optional. (40) LEVEL 1

öffnen4: LEVEL 2 ACTIVE

DEAG. ACT. REF. MIDDLE LASS-MIDDLE COP. MIDDLE

C1: agent, C2: patient, C3: instrument, C4 modalizer C1: subject, C2: direct object (C3: prep. object, C4 adverbial) C3: subject, C2: direct object (C4: adverbial) C2: subject, C4: adverbial C2: subject (C3: prep. object, C4: adverbial) C2: subject (C3: prep. object, C4: adverbial)

On a third level, further semantic features of the single complements could be specified (cf. Vater 1975). Valency description (40) constitutes an attempt to unite both valency potential (Level 1) and valency realizations (Level 2) in the specification of a lexical entry. The description on Level 2 corresponds roughly to the syntactic information contained in a lexical entry as presented by Levelt (1989) and Handke (1995). It has to be pointed out that in my view only one variant of öffnen has to be postulated, since the relationship between Jemand öffnet das Geschäft and Das Geschäft öffnet is a difference between diathesis alternations (and thus can be derived by transformations, as indicated by Helbig and Schenkel 1978:381 themselves). The differentiation between the valency potential associated with a verb in the lexicon and the various valency realizations conditioned by the syntactic

VALENCY AND DIATHESIS

117

environments in which the verb occurs helps to solve at least part of the prob­ lems that were described as the 'valency misery' by Storrer (1992), Jacobs (1994a), and others. The suggested distinction between a verb's valency potential and its realizations can also be usefully employed in describing constructions such as the AcI-construction, e.g. I saw him open the door (vs. I saw that he opened the door),20 or in imperative constructions characteristically suppressing an agen­ tive subject, e.g. Open the door (vs. He opens the door); cf. Vater (1996) for a treatment of these constructions as different valency realizations of the same valency potential. It goes without saying that a lot of work still has to be done in this field and that the theoretical framework employed plays an important role in the description of the relationship between valency potential and valency realizations as well as of the relationships between the single valency realizations.21 Notes 1

English translations of German sentences will stay as closely as possible to the German original. If the translation of a grammatical sentence in German is ungrammatical in English, it will be starred and a semantically equivalent paraphrase will be added. The starred English translation thus signals that there is no English equivalent for the German construction. An ungrammatical sentence in German will be translated as an ungrammatical sentence in English, signaling that a particular structure is not present in either German or English. 2 "On peut ainsi comparer le verbe à une sorte d'atome crochu susceptible d'exercer son attraction sur un nombre ... d'actants, selon qu'il comporte un nombre ... de crochets pour les maintenir dans sa dépendance. Le nombre de crochets que présente un verbe et par conséquent le nombre d'actants qu'il est susceptible de régir, constitue ce que nous appellerons la valence du verbe" (Tesnière 1969:328). 3 Heringer (1968:426) remarks, "Die Theorie der syntaktischen Wertigkeit wurde entwickelt von L. Tesnière" [The theory of syntactic valency was developed by L. Tesnière]. 4 Erben (1972:§487) states "daß eine begrenzte Zahl einfacher - mit Verbklassen verschie­ dener syntaktischer Valenz verbundener - Grundmuster die Primärstruktur der Sätze bestimmt" [that a limited number of simple basic patterns - each linked up with a particular class of verbs having a particular syntactic valency - determines the primary structure of sentences]. 5 Valency is located within the 'strict subcategorization' part of the syntax proposed by Chomsky (1965); it is confined to the VP, with the (semantic) relation between a verb and its subject being accounted for by 'selectional rules'. For a dependency-based generative grammar, cf. Vater (1975). 6 It is a matter of controversy whether German has a VP constituent. Haider (1983) thinks that German is a non-configurational language having a flat sentence structure without VPs. Grewendorf (1988:293ff), on the other hand, argues that German is a configurational language and has a hierarchical sentence structure including VPs. Cases such as (i) furnish an important

118

HEINZ VATER

argument in favor of the VP assumption: here, the sequence den K i n d e r n eine Geschichte erzählt "told a story to the children" must be a constituent, since it has been moved to the ' Vorfeld' (initial field, top position in the German sentence, preceding the position of the finite verb in main sentences); this constituent must be a VP, since it contains all elements of the sentences except the subject and the finite verb (which is usually assumed to have been moved out of its original position within the VP). (i) a. Den Kindern eine Geschichte erzählt hat die Großmutter (Dürscheid 1989) *"The children told a story has the grandmother." b. *Die Großmutter erzählt hat den Kindern eine Geschichte *"The grandmother told has the children a story." In English, structures of this kind are not possible. Grewendorf (1988:294) presents more evidence for the constituent status of the German VP including pronominalization and coor­ dination tests. His arguments were confirmed by Rahmig (1998). 7 The subject's special role is even admitted by Helbig and Schenkel (1978:27), who advocate the primacy of the verb in the sentence, all its complements (including the subject) filling slots provided by the verb. 8 See also Welke (1988) and Ágel (1993) for a similar view. 9 This distinction corresponds by and large to the distinction between static and dynamic valency made by Sadzinski (1989). 10 It would be even better to exclude the feature 'obligatory' altogether from the definition of 'complement' in order to avoid circularity. 11 Helbig and Schenkel (1978:443ff.) treat meteorological verbs like regnen "to rain" and schneien "to snow" as 'avalent' ('nullwertig'), thus forgetting their principle to restrict their descriptions to syntactic valency (cf. Section 2.1); after all, it cannot be omitted from any constructions in which meteorological verbs occur, i.e. it is a complement of these verbs. 12 Many of these impersonal constructions are considered to be archaic, especially by young people. 13 Passive and impersonal constructions are united by Rytel-Kuc (1988:348), who bases her statement on Weisgerber (1963), under the heading of 'agensabgewandte Diathesen' ('agent suppressing diatheses'). Heidolph et al. (1981:541) call these diatheses 'nicht-täterbezogen' ('non-agent-related') or 'agensunabhängig' ('agent-independent'). I prefer to call them 'agent absorbing diatheses' since they imply an agent role also in cases where they do not permit its realization. 14 The werden -passive can designate a state in cases like Alle Häuser werden vom Dom über­ ragt "All houses are dominated by the cathedral/The cathedral rises high above the houses"; thus, it should not be equated with a 'processual passive' (cf. Helbig & Heinrich 1972:9). In these cases, the agent phrase cannot be suppressed: *Alle Häuser werden überragt. 15 For recent studies of the reflexive passive, see Plank (1993), Vater 1995, and Ágel (1997) 16 For detailed analyses of middle constructions in English, German, and other Germanic languages, see Wagner (1977), Abraham (1987), and Vater (1988). 17 Sentence (25d) shows that an animate being responsible for the process can be added as a dativus incommodi. 18 In keeping with recent grammars and monographs (e.g. Steinitz 1969), I make a distinction between 'adverb' as a word class and 'adverbial' as a syntactic function which can be fulfilled by phrases with heads of different word classes (e.g. prepositional phrases and adjective phrases).

VALENCY AND DIATHESIS

119

19 'Obligatory' and 'optional' refer to syntactic rather than to semantic relationships. In all cases, there is a difference in meaning between the constructions with and without the adver­ bial; cf. Man kann die Tür öffnen "You can open the door" vs. Man kann die Tür leicht/mit großer Kraft öffnen "You can open the door easily/with great force") or Die Tür ist leicht zu öffnen "The door is easy to open" vs. Die Tür ist zu öffnen "The door has to be opened". 20 Quirk et al. (1972:841) call this a 'bare infinitive with subject'. 21 I would like to thank my friend Glenn Levine for proofreading the text and checking my English examples.

References Abraham, Werner. 1985. "Grammatik von kriegen und bekommen". Osnabrücker Beiträge zur Sprachtheorie 30.142-165. . 1987. "Zur Typologie des Mediums in der Westgermania". Linguistik in Deutschland: Akten des 21. Ling. Kolloquiums, Groningen 1986 (= Linguistische Arbeiten, 182) ed. by Werner Abraham & Ritva Ǻrhammer, 3-23. Tübingen: Max Niemeyer. Ágel, Vilmos. 1993. "Ist die Dependenzgrammatik wirklich am Ende?" Zeitschrift für germanistische Linguistik 21.20-70. 1995. "Valenzrealisierung, Grammatik und Valenz". Zeitschrift für germa­ nistische Linguistik 23.2-32. . 1997. "Reflexiv-Passiv, das (im Deutschen) keines ist. Überlegungen zu Reflexivität, Medialität, Passiv und Subjekt". Sprache im Fokus. Fest­ schrift für Heinz Vater zum 65. Geburtstag ed. by . Dürscheid, K.H. Ramers & Monika Schwarz, 147-187. Tübingen: Max Niemeyer. Andresen, Karl Gustaf. 1854. "Passiv des Reflexivs". Allgemeines Nassaui­ sches Schulblatt 48.763-764. Behaghel, Otto. 1924. Deutsche Syntax: Eine geschichtliche Darstellung, Vol. IL Heidelberg: Winter. Bierwisch, Manfred. 1970. "Aufgaben und Form der Grammatik". Vorschläge für eine strukturale Grammatik des Deutschen (= Wege der Forschung, 146) ed. by Jugo Steger, 1-51. Darmstadt: Wissenschaftliche Buch­ gesellschaft. Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, Mass.: The MIT Press. Duden. Grammatik der deutschen Gegenwartssprache, 5th edition. 1995. Mannheim: Dudenverlag. Dürscheid, Christa. 1989. Zur Vorfeldbesetzung in Verbzweit-Strukturen. (= Fokus, 1).Trier: Wissenschaftlicher Verlag. Eisenberg, Peter. 1994. Grundriß deutschen Grammatik. Stuttgart: Metzler.

120

HEINZ VATER

Engel, Ulrich. 1994 [1977]. Syntax der deutschen Gegenwartssprache, 3rd edition. (= Grundlagen der Germanistik, 22). Berlin: Erich SchmidtVerlag. Erben, Johannes. 1972. Abriß der deutschen Grammatik, 11th edition. München: Hueber. Fanselow, Gisbert & Sascha Felix. 1987. Sprachtheorie (= UTB, 1441/1442), 2 Vols. Tübingen: Francke. Grewendorf, Günther. 1988. Aspekte der deutschen Syntax. (= Studien zur deutschen Grammatik, 33).Tübingen: Gunter Narr Günther, Hartmut. 1978. "Valence in Categorial Syntax". Valence, Semantic Case and Grammatical Relations ed. by Werner Abraham, 127-156. Amsterdam: John Benjamins. Haider, Hubert. 1983. "Connectedness Effects in German". Groninger Arbeiten zur Germanistischen Linguistik 23.82-119. Handke, Jürgen. 1995. The Structure of the Lexicon: Human vs. Machine. Berlin: Mouton de Gruyter. Hansen, Aage. 1967. Moderne Dansk, vol III. Copenhagen: Grafisk Forlag. Heidolph, Karl Erich, Walter Flämig, Wolfgang Motsch et al. 1981. Grundzüge einer deutschen Grammatik. Berlin: Akademie-Verlag. Helbig, Gerhard 1982. Valenz - Satzglieder - semantische Kasus - Satz­ modelle. Leipzig: VEB Verlag Enzyklopädie. Helbig, Gerhard & Gertraud Heinrich. 1972. Das Vorgangspassiv. Leipzig: VEB Verlag Enzyklopädie. --& Wolfgang Schenkel. 1978. Wörterbuch zur Valenz und Distribution deutscher Verben, 4th edition. Leipzig: VEB Verlag Enzyklopädie. Hentschel, Elke & Harald Weydt. 1995. "Das leidige bekommen-Passiv". Deutsch als Fremdsprache: Festschrift für Gerhard Helbig zum 65. Geburtstag ed. by Heidrun Popp, 165-183. München: Iudicium. Heringer, Hans-Jürgen. 1968. "Präpositionale Ergänzungsbestimmungen". Zeitschrift für deutsche Philologie 87.426-457. Höhle, Tilman N. 1978. Lexikalische Syntax: Die Aktiv-Passiv-Relation und andere Infinitkonstruktionen im Deutschen. Tübingen: Max Niemeyer. Jacobs, Joachim. 1994a. Kontra Valenz. (= Fokus, 30). Trier: Wissen­ schaftlicher Verlag. . 1994b. "Das lexikalische Fundament der Unterscheidung von obligatorischen und fakultativen Ergänzungen". Zeitschrift für germanis­ tische Linguistik 22.284-319.

VALENCY AND DIATHESIS

121

Leirbukt, Oddleif. 1977. "Über passivische Fügungen der Struktur bekommen/ kriegen/erhalten + Partizip II im heutigen Deutsch". Språk og språk­ undervisning 10.47-55. . 1997. Untersuchungen zum bekommen-Passiv im heutigen Deutsch, (= Reihe Germanistische Linguistik 177). Tübingen: Max Niemeyer. Levelt, Willem J.M. 1989. Speaking: From Intention to Articulation. Cambridge, Mass.: The MIT Press. Plank, Frans. 1993. "Peculiarities of Passives of Reflexives in German". Studies in Language 17.135-167. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1972. A Grammar of Contemporary English. London: Longman. Rahmig, Martina. 1998. Deutsch als konfigurationeile Sprache? Unpublished paper, University of Cologne. Reis, Marga. 1976. "Zum grammatischen Status der Hilfsverben". Beiträge zur Geschichte der deutschen Sprache und Literatur 98.64—82. Rytel-Kuc, Danuta 1988. "Die täterabgewandten -no/-to-Konstruktionen im Polnischen und ihre Entsprechungen im Deutschen". Valenzen im Kontrast: Ulrich Engel zum 60. Geburtstag ed. by Pavica Mrazoviæ & Wolfgang Teubert, 347-356. Heidelberg: Groos. Sadzinski, Roman. 1989. Statische und dynamische Valenz: Probleme einer kontrastiven Valenzgrammatik Deutsch-Polnisch. Hamburg: Buske. Steinitz, Renate, 1969. Adverbialsyntax (= Studia grammatica, X). Berlin: Akademie-Verlag. Storrer, Angelika. 1992. Verbvalen:. Theoretische und methodische Grund­ lagen ihrer Beschreibung in Grammatikographie und Lexikographie (= Reihe Germanistische Linguistik, 126). Tübingen: Max Niemeyer. Tarvainen, Kalevi 1986. Deutsche Satzstruktur und ihre Entwicklung. Dependenzgrammatik des Deutschen mit historischen Erläuterungen. Jyväskylä: Jyväskylän yliopiston monistuskeskus. Tesnière, Lucien 1969 [1959]. Eléments de syntaxe structurale, 2nd edition. Paris: Klincksieck. Translated into German by U. Engel, 1980. Grund­ züge der strukturalen Syntax. Stuttgart: Klett-Cotta. Vater, Heinz 1975. "Toward a Generative Dependency Grammar". Lingua 36.121-145. . 1978. "On the Possibility of Distinguishing Complements and Adjuncts". Valence, Semantic Case and Grammatical Relations ed. by Werner Abraham, 21-45. Amsterdam: John Benjamins.

122

HEINZ VATER

. 1981. "Valenz". Kasusgrammatik und Fremdsprachenunterricht (= Anglistik und Englischunterricht, 14) ed. by Günter Radden & René Dirven, 217-235. Trier: Wisscnschaftlicher Verlag. . 1988. "Mittelkonstruktionen im Englischen, Dänischen und Deut­ schen". Valenzen im Kontrast: Ulrich Engel zum 60. Geburtstag ed. by Pavica Mrazoviæ & Wolfgang Teubert, 398-417. Heidelberg: Groos. . 1995. "Zum Reflexiv-Passiv im Deutschen". Deutsch als Fremd­ sprache - An der Quellen eines Faches: Festschrift für Gerhard Helbig zum 65. Geburtstag ed. by Heidrun Popp, 185-192. München: Iudicum. . 1996. "Valenzpotenz und Valenzrealisierung im Deutschen". Jahr­ buch der ungarischen Germanistik, 143-165. Budapest/Bonn: Gesell­ schaft ungarischer Germanisten/DAAD. Wagner, Fritz 1977. Untersuchungen zu Reflexivkonstruktionen im Deutschen (= Regensburger Beiträge zur deutschen Sprach- und Literatur­ wissenschaft, 10). Frankfurt a.M.: Peter Lang. Wegener, Heide. 1985a. Der Dativ im heutigen Deutsch (= Studien zur deut­ schen Grammatik, 28). Tübingen: Gunter Narr. . 1985b. '"Er bekommt widersprochen' - Argumente für die Existenz eines Dativpassivss im Deutschen". Linguistische Berichte 96.127-139. Weisgerber, Leo. 1963. Die vier Stufen in der Erforschung der Sprachen. Düsseldorf: Schwann. Welke, Klaus. 1988. Einführung in die Valenz- und Kasustheorie. Leipzig: VEB Bibliographisches Institut. Wunderlich, Dieter. 1993. "Diathesen". Syntax: Ein internationales Handbuch zeitgenössischer Forschung ed. by Joachim Jacobs, Arnim von Stechow, Wolfgang Sternefeld & Theo Vennemann, vol I, 730-747. Berlin: Mouton de Gruyter. Zifonun, Gisela. 1992. "Das Passiv im Deutschen: Agenten, Blockaden und (De-)Gradierungen". Deutsche Syntax: Ansichten und Aussichten (= IDSJahrbuch 1991) ed. by Ludger Hoffmann, 94-127. Berlin: De Gruyter

TO GET OR TOBE? USE AND ACQUISITION OF GET- VERSUS BE-PASSIVES: EVIDENCE FROM CHILDREN AND ADULTS KERSTIN MEINTS Lincoln University, UK

The use and acquisition of the get-passive has so far yielded a variety of accounts and sugges­ tions. This paper presents new experimental evidence concerning the use and the acquisition of the get-passive by children, as well as adult judgments of get- and be-passives. Within a prototype approach to the passive, experiments investigated when 2-4-year-old British chil­ dren produce get- as opposed to be-passives. The role of direct affectedness of the patient on get-passive production was investigated further in a follow-up experiment. In addition to the child data, ratings of get- and be-passives were obtained from British English adult speakers to investigate the acceptability of these passives and their relationship to developmental data. The first experiment showed that the chosen prototype approach clearly predicts children's acqui­ sition of be-passives with get-passives being more peripheral members of the category 'passive' than be-passives. The second study shows that even if the child herself is the affected patient in the play action, get-passives are only rarely produced. In contrast to American chil­ dren, direct affectedness did not induce British children to produce a significant amount of getpassives. Last, adult ratings confirm that British English speakers rate be-passives consistently as better examples of passive sentences than get-passives. The evidence suggests that getpassives are more peripheral for British than for American children and adults. Implications for the possible role of parental input and the validity of existing accounts of the get-passive are discussed.

1.

Get- vs. be-passives: An overview1

Why do speakers of English in some situations prefer the get-passive over the be-passive? What are the developmental differences in the acquisition of get­ and be-passives? Great uncertainty prevails as satisfactory explanations for the use of the get-passive have neither been found by investigating adult passives, nor have they been detected through the study of child language. In this paper, I will present new experimental evidence on the use and the acquisition of the

124

KERSTIN MEINTS

get-passive. Within a protoype approach to the passive, production experiments were carried out investigating children's use of the get- as opposed to the bepassive. In addition, typicality judgments of get- and be-passives were obtained from adults. In order to clarify the positions held on the get-passive, I will start by giving a brief overview of the positions held on the get-passive per se (largely in the adult speaker) and then move on to describe how researchers find more evidence by watching the get-passive's coming into existence in children's language. 1.1

The get- over the be-passive in adult speakers

A first set of factors trying to explain language users' choice of the get-passive over the -passive is of a semantic and pragmatic nature. a.

b.

c.

Dynamism. It has been suggested that the get-passive has more 'active' or more causative qualities than the be-passive (Svartvik 1966; Lakoff 1971; Gee 1974; Romaine 1984) and receives a more dynamic interpretation (Keenan 1975; Quirk et al. 1985). Diachronic studies (Langacker & Munro 1975; Shibatani 1985; Givón & Yang 1994) support this view as they suggest that the get-passive has grown out of the more active transitive meaning of to get, i.e. 'to take/grab something', and has only established itself since the middle of the eighteenth century after detransitivization processes. Responsibility. Events described by the get-passive have been described by some linguists as rather random, unexpected, or even rare (e.g. Standwell 1981); others (e.g. Schibsbye 1970; Lakoff 1971), however, have suggested that the get-passive emphasizes that the patient draws the action intentionally towards himself, thereby gaining a degree of control and responsibility over the action while at the same time being affected by it (as in The politician got arrested to prove his point/got arrested on purpose) (Dirven & Radden 1977; Givón & Yang 1994). Speaker viewpoint and adversity. It has also been claimed that the get-passive can be used to express attitudes, personal viewpoints, and emotional states, as in I got promoted (vs. I was promoted) (Lakoff 1971; Dirven & Radden 1977; Chappell 1980). It is seen to especially reflect negative attitudes of the speaker, as in Charlene got fired (Banks 1986; Budwig 1990 in a study on children's use of the passive). Herold (1986, cited in Givón 1994) has found that 85%

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

d.

125

of get-passives have adversative character and negative conno­ tations; similarly, Collins (1996) has observed that 67.4% of getpassives2 imply adversative behavior (while only 23.4% gave rise to beneficial conversational implicatures and 9.3% were neutral). Some of the get-passives studied diachronically by Givón and Yang (1994) also describe a state that is evoked by an adversary agent and is usually not desired by the affected patient, as in get hurt, get killed, get run over. Agent and patient roles. The adversative character of the get-passive brings about a stronger focus on the patient than the -passive, which may explain why get-passives are used more often without an agent than b-passives (Svartvik 1966; Quirk et al. 1985; Johansson & Oksefjell 1996). Givón and Yang (1994) have also found a differ­ ence in the type of entities in the patient role: primarily, get-passives occur with animate, human patients, while -passives occur with human as well non-human patients. The ideal agent is seen as being either obvious or at least easy to recover, with human agents being better examples of agents than animal or non-human agents (Collins 1996).

In an attempt to integrate most of the above factors, Collins (1996) suggests a description in terms of a fuzzy set approach, with the array of get-passives ranging from central instances (dynamic, with traceable agents, and the ability to alternate with active sentences) to more peripheral adjectival and formulaic instances.3 However, although Collins divides get-passives into five groups with varying centrality, it is not possible to predict clearly which get-passive belongs to which subgroup because he does not clarify the weight of the getpassive components. As he has not carried out typicality ratings with native speakers, his centrality groups so far stand without any empirical foundation. In addition to semantic and pragmatic factors influencing the get-passive, a number of stylistic, social, and regional factors also seem to play a role in the use of the get-passive. a.

Stylistic variety. In more formal contexts (Collins 1996) and in written English, the get-passive tends to be avoided whereas it is observed more frequently in informal and in spoken English (Svartik 1966; Standwell 1981; Quirk et al. 1985; Collins 1996; Johansson & Oksefjell 1996). Collins (1996) shows that central get-passives occur with an average frequency of 96 per million words in corpora of

126

KERSTIN MEINTS

b.



spoken English (Australian and British English),4 whereas they only occur with 43 per million words in written corpora. Nevertheless, in spoken British English, be-passives are used with a proportion of 30:1 times more often than get-passives (Johansson & Oksefjell 1996). Social variety. Making use of data in Herold (1986), Givón and Yang (1994) investigated the distribution of get- and be-passives on social class background. Their results show that speakers belonging to the American working class use get- and be-passives almost equally frequently (49% and 51%, respectively). In contrast, Ameri­ can upper class speakers used the get-passive only 17% of the time whereas their use of the be-passive exceeded the get-passives clearly with an 83%) preference. Regional variety. In different varieties of English, e.g. British and American English, the get-passive appears roughly equally often in written British and American English, while it has been claimed to occur more frequently in spoken American than in spoken British English (Dirven & Radden 1977). In contrast to this, it has been stated that overall, the be-passive occurs more frequently in both spoken American and British English. Thus, American speakers use the get-passive more than British speakers, but both use the bepassive most of the time (Svartvik 1966; Sussex 1982). However, there are to date no large corpora on spoken American English avail­ able to sustain these claims with the same amount of empirical data as provided by Collins (1996) and Johansson and Oksefjell (1996) for written and spoken British English.5

In explaining the choice of the get-passive over the -passive in adult speak­ ers, I have drawn attention to (i) semantic and pragmatic factors, in particular the degree to which the get- or -passive displays 'active' or causative quali­ ties (cf. dynamism) and the role and focus of the patient and the agent in the event described by the get-/be-passive (cf. responsibility, (speaker) viewpoint, and adversity), and (ii) stylistic, social, and regional factors. In order to get a better grasp of how this complex interplay of factors works, it is useful to investigate its development as we can observe how the different components of the get- and be-passive start to work and interact. Looking at child language as well as at the target language helps us to get a more complete picture of the underlying principles and mechanisms at work.

USE AND ACQUISITION OF GET- VS. BE-PASSIVES 1.2

127

Development of the get-passive

Much as the various aspects of the use and frequency of get-passives have been discussed for adult speakers of English, the use and frequency of get- as opposed to be-passives in children is still an open question, especially as conflicting evidence concerning the acquisition of get-passives needs to be resolved. Attention to auxiliary choice in the acquisition of the passive voice has been rather meager and systematic empirical investigation of get- versus be-passives in young children is still sparse. To find a connection between assumptions and theories on the get-passive in the target language and in child language, experiments with children will be reviewed here in more detail. In the child language literature, it has been assumed that the get-passive is a form of early 'children's passive' (Turner & Rommetveit 1967) from which the -passive might evolve (see also Marchman et al. 1991, for age differences in the production of be- and get-passives). In a similar vein, Harris and Flora (1982) hold that get-passives rather than be-passives are a forerunner in imitation, comprehension, and production. Fraser, Bellugi and Brown (1963) hold the view that children use the get-passive in order to differentiate passive sentences more strongly from active sentences. Budwig's (1990) analyses spontaneous language data of two American children (Melissa Bowerman's two daughters; a longitudinal study from ages 3 to 10) and she also carries out an imitation and sentence repair experiment with children and adults in order to find functional differences in children's and adults' use of the get- versus the be-passive. She shows that these two children produce fewer get-passives than be-passives at all ages, with less get-passives being produced as they get older.6 This stands in sharp contrast to the assump­ tions of Turner and Rommetveit (1967), Fraser, Bellugi and Brown (1963), and Marchman et al. (1991), and is also in conflict with reports on spoken Ameri­ can English (Svartvik 1966; Dirven & Radden 1977). Furthermore, Budwig (1990) finds that get-passives occur more often in combination with a byphrase than be-passives, which, once more, conflicts with expectations derived from studies on adult language (cf. Svartvik 1966; Quirk et al. 1985; Givón & Yang 1994). Then again, Budwig's (1990) results are consonant with adult language studies in the following two respects. First, for agentless passives, Budwig observes that get-passives are often used in combination with animate patients while agentless -passives combine mostly with inanimate patients. This finding is in accordance with Givón and Yang (1994). Second, as far as adversity is concerned, Budwig (1990) finds that get-passives are more often combined with an adversary action and a negatively affected patient than be-

128

KERSTIN MEINTS

passives. This could be seen as evidence in favor of the theoretical position above that get-passives express the adversary action strongėr than be-passives and thus focus on the affected patient (cf. Lakoff 1971; Dirven & Radden 1977; Quirk et al. 1985; Herold 1986; Givón & Yang 1994). In her imitation and sentence repair task with American children and adults, Budwig (1990) finds further evidence for her claim that children use the ė-passive more frequently to focus on the non-agentive perspective in contexts expressing an action that negatively affects the patient. Adults, however, when presented with -passives, show a preference for be-passives over get-passives as they retain 70% of be-passives while changing 30%. When presented with adversative get-passives, they retain only 53% of the tested getpassives while 48% are changed to -passives or actives.7 Budwig concludes that adults and children might still possess the same underlying functional distinctions (favoring get-passives if expressing the negative affectedness of the patient), but that adults are influenced by tuition to favor be-passives. Further research with adults needs to be carried out as Budwig's studies cannot resolve this matter. Experiment 3 in this paper will take a first step towards addressing the status of get-passives in British adults. A final outcome of Budwig's study is that children produce more getpassives than be-passives when they themselves are affected by the action. This is noteworthy as it fits in with the view that get-passives express the affect­ edness of the patient more strongly than be-passives. However, systematic testing of this assumption has not been carried out. Experiment 2 will focus on testing children using direct affectedness. Child language experiments by Marchman et al. (1991) show preferences for the production of get- vs. be-passives which differ from Budwig's (1990) results. Making use of earlier findings that children produce more passives if they encode prototypically transitive verbs like to kick (Maratsos et al. 1985), Marchman et al. studied the production of get- and -passives in adults and children using prototypically transitive (e.g. biting) and non-prototypically transitive (e.g. giving) elicitation scenes.8 While American adults produced mostly be-passives (only 8% get-passives), with prototypically transitive elici­ tation scenes, American children between 3 and 11 years of age over­ whelmingly preferred get- over -passives, with 5-6-year-olds producing 100%) get-passives. Overall, prototypically transitive scenes were more often encoded with get-passives than with -passives (e.g. simple scenes: horse bites goat; complex scenes: snake bites a horse and then a sheep). Be-passives were produced to a greater extent with non-prototypical dative and locative actions (e.g. man gives flower to the lady; baby crawls under a table) than with

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

129

prototypical transitive scenes. However, younger children (3-4 years) showed a different pattern of reaction as they commented upon all non-prototypical scenes with get-passives only (if they used passives at all) and used the bepassive in 15% of the cases to describe prototypical transitive scenes (with the remaining 85% of produced passives being get-passives). Marchman et al. conclude that -passives reflect a functional relation to ongoing states while get-passives seem to be used more frequently to express change of state result­ ing from prototypically transitive actions. Although this interpretation is supported by the results of the 5-6-year-olds, Marchman et al. emphasize that further investigation is necessary to account for the deviating pattern of results of the 3-4-year-olds. Furthermore, they have left the question unanswered of what makes an action prototypical and they have not systematically inves­ tigated the interplay of the role of patient, agent, action and the degree of prototypicality of the whole action scene. Meints (1996, 1999a) attempts to overcome these problems by systematic variation of the different factors involved in comprehension and production of the get- and be-passive.9 Within a prototype approach, Meints systematically varies patient and agent animacy together with a scale of five differing degrees of prototypical transitivity of the action (based on a previous investigation of the conceptual structure of verbs). Thus, the above-mentioned factors (e.g. dynamism, agent and patient, affectedness) are subsumed under this prototype approach. Moreover, factors like semantic reversibility, focus, and event-prob­ ability are taken into account as well. Therefore, the effects of the single factors can be scrutinized as well as possible interactions. For this paper, especially the production results (reported in Meints 1996, 1999a) are of interest as children were free to produce get- or -passives. The results indicate that the prototypicality degree of the passive construction can successfully predict the acquisition order of the passive (for more detailed explanations see Experiment 1). Typical passives (e.g. The postman was bitten by the dog) are indeed acquired earlier than less typical passives (e.g. The boy was seen by the girl). In view of these results, the aim of Experiments 1, 2, and 3 (cf. infra) is to find empirical evidence for some of the proposed theoretical explanations for the production of get- as opposed to be-passives and to inves­ tigate whether a prototype approach (building mainly on semantic-pragmatic factors) can contribute to solving some of the still open questions concerning the use and acquisition of the get- as opposed to the be-passive. In addition, the category status of get-passives for British adults is investigated and then compared with Budwig's and Marchman et al.'s data on American adults. Differences between American and British participants—children and adults—

130

KERSTIN MEINTS

are especially of interest as the get- and be-passive differences might stem more strongly from differences in category status and/or regional variation than from underlying semantic and pragmatic factors. First, experiments on the production of get- and be-passives was carried out with 2-, 3-, and 4-year-olds to investigate the influence of different prototypicality levels on the production of get- and be-passives (Experiment 1 reported in Meints 1996, 1999a). 2-year-olds were included to capture the earliest possible time of production of either get- or be-passives; 3- and 4-yearolds were selected to replicate the findings of Marchman et al. (1991) with British children. In addition to the results of Meints (1996, 1999a), a separate data analysis with respect to get-passives was carried out in Experiment 1. Will British children show a production of about 15% be-passives to encode proto­ typical transitive events with the remaining produced passives being getpassives, as the American children did? If the get-passive is generally produced more often than the -passive whenever a prototypical transitive scene is encoded, as the data by Marchman et al. (1991) suggest, then we should be able to find this effect with British children, too. If, on the other hand, the choice of get- or b-passives is independent of the underlying prototypically transitive scene, then the question of to get or to be must be due to other effects (e.g. status of get-passive category membership, regional variation, frequency of input) and there should be no increase in the production of get-passives for prototypical passive constructions (i.e. those passives with an underlying prototypical transitive scene like he got bitten). 2.

Experiment 1: Children 's production of get- and he-passives

2.1

Introduction

Experiment 1 looks at the following questions: (i) When do British children use the passive voice? (ii) Which type of passive (be- or get-passive) do they acquire first? (iii) Which factors determine the acquisition of passives? In order to address these questions, the English passive was investigated as a category displaying typicality effects, with the canonical action event as its prototype (e.g. The postman was bitten by the dog) and other kinds of passives as more peripheral members (e.g. The postman was seen by the cat). In particular, the aim of this experiment was twofold: (i) to examine the effect of different prototypicality levels on the production of get- and be-passives; (ii) to investigate whether British children show a production of about 15% bepassives to encode prototypical transitive events, with the remaining produced passives being get-passives, as the American children did. As a preliminary, it

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

131

needed to be established what constitutes a prototypical passive. The various factors contributing to a prototypical passive were studied by Meints (1996, 1999a, 1999b), who investigated the prototypicality of agent, patient, and action, along with the role of focus, semantic reversibility, and likelihood of the action scene. (In order to determine the underlying conceptual structure of the different actions/verbs, a study had been carried out with British adults and adolescents). The results of these investigations and of further research into the character of patient, agent, action and transitivity, focus, reversibility and event-probability led to the following predictions about a passive prototype (Meints 1996, 1999a, 1999b). Prototypical passives should comprise a patientfocused, prototypical transitive scene as described below: Prototypical transitive scene The postman patient

was bitten action

by the dog agent

is -highly affected -animate or inanimate -focused

-incorporates high degree of -actionality -punctuality -direct physical contact - a visible result

is -acting -animate -defocused

Table 1 : Prototypical transitive scene with factors contributing to the prototypicality of passives

In prototypical passives, the patient should be the focus of attention and the sentence should be semantically irreversible.11 The sentences the postman was bitten by the dog and the cheese was eaten up by the mouse are examples for prototypical passives as defined in this study. Examples of atypical passives are The postman was seen by the dog and the cheese was smelled by the mouse. In these cases, (i) the patient is not affected; (ii) as far as the action is concerned, the degrees of actionality and punctuality are much lower, no physical contact is involved, and there is no visible result; (iii) the agent is not a highly active one.12 In other words, the degree of prototypicality of a passive sentence is predicted to increase (i) as the affectedness of the patient (which is focused and animate or inanimate) is higher, (ii) as the scores on the action group factors (intentionality, result, duration, degree of activity of the action, and direct physical contact during the action, cf. Meints 1996) are higher, (iii) and as the affecting force of the agent increases. The passive should be acquired in the following order:

132

KERSTIN MEINTS

Prototypicality Hierarchy

Patient (focused)

Action

Agent (defocused)

high

highly affected

action group 1 (high typicality of action)

highly affecting

high to medium

medium affected

action group 2 (medium typicality of action)

medium affecting

medium to low

medium affected

action group 3 (adj.) (medium-low typicality of action)

medium affecting

low

least affected

action group 4 (low typicality of action)

least affecting

Table 2: Prototypicality Hierarchyfor passive acquisition

All in all, the passive is treated as a graded category with fuzzy boundaries and prototypical passives at its center. Prototypical passives comprise a patientfocused, prototypical transitive scene. As prototypical members of a category are acquired earlier than non-prototypical members (Rosch 1973, 1978; Anglin 1976), a gradual acquisition of the passive was predicted with children acquir­ ing prototypical examples of the passive earlier than more peripheral ones. 2.2

Method

Participants. Participants were thirty-five English children of three age groups (2-, 3-, and 4-year-olds). All children were monolingual speakers of English from a middle class social background. The experiments were carried out in three different nurseries in Fulham, London, UK. The nurseries offered sepa­ rate rooms so that the study could be carried out without any disturbances. The sessions took between ten and twenty minutes and were documented by videoand DAT-recordings. Materials. 110 stimuli scenes were constructed in which the factors prototypicality of agent, patient, and action, and semantic reversibility were varied systematically from high to low prototypicality. Toys were used to enact the scenes for the children (animals/humans, vehicles, everyday objects). All children were able to name the toys. Procedure. All children in the production study had to comment on the 110 scenes that were acted out by the experimenter while the children were watching. The study was split up into different sessions to avoid tiring or boring the children. In order to elicit either get- or be-passives, it was necessary to use a patient-focused question (e.g. What happened to the X?) in these

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

133

production experiments, as children would not normally respond with a passive sentence but use the active voice instead. Table 3 shows the type of passive constructions the children could produce (e.g. a full, reversible, verbal be- or get-passive like The postman was/got hit by the burglar):

Possible passive constructions Full passives/ passives without by-phrase Reversible/irreversible passives Verbal/adjectival passives Get/be-passives Table 3: Overview of passive constructions investigated in production experiments (Meints 1996)13

The dependent variable was the child's response (utterance of a passive—with be or get, an active sentence, or other utterances). To avoid training effects, filler items were used, with a 2:3 ratio between passives and filler items. The above mentioned toys were used as agent as well as patient items. It is impor­ tant to note that, in order to avoid confounding prototypicality and frequency effects, frequency of occurrence of objects (to be expressed by nouns) and actions (to be expressed by verbs) in the passive sentences was controlled by using only those items that were highly frequent in child language (data obtained from MacArthur GDI (Fenson et al. 1994) and Bristol Child Language Corpus (Wells 1971-1985)). 2.3

Results and discussion

Three 3-factorial MANOVAs were calculated (Age group x Action group x Agent; Age group x Action group x Patient; Age group x Action group x Reversibility).14 Analysis of variance showed that the predicted prototypical passives were produced earlier and in larger numbers than more peripheral passives (age group main effect: [F (2,24) = 5.41; p < 0.05]; action group main effect: [F (2,48) = 7.93; p < 0.011]).15As expected, children do not produce all sorts of possible passives equally. 2-year-olds produce a few passives of medium typicality and slightly more highly prototypical passive sentences (while they are able to produce active sentences). They do not produce atypical passives at all. 3-year-olds produce more passives overall than 2-year-olds and their correct responses increase in accordance with the predicted passive hier­ archy. 4-year-olds produce the highest number of passive sentences, including the more peripheral ones. Figure 1 illustrates the action group effect for the three age groups (for all produced passives together):

134

KERSTIN MEINTS

Results for Action group, plotted by age group

Action group Figure 1: Correctly produced passives in % for action group, divided into age groups

The prototype approach successfully predicts the course of acquisition of the passive, and thus confirms the existence of a hierarchy in the acquisition of the passive. At the same time, this approach provides a coherent image of how children acquire the English passive, in that it has successfully integrated, and systematically tested, a variety of factors influencing passive acquisition into a category 'passive'. However, if we take a closer look at the experimental results, we find that children's production of passives mainly consists of be-passives. While of all elicited utterances (100%), 60.6% were passives, out of these 60.6%, only 183 (4.8%) were get-passives. The vast majority of the children in this study (91.4%) preferred be-passives over get-passives, if they produced passives at all. Only three (8.6%) of the children produced the get-passive as their predominant passive form. Additionally, out of the 91.4%, one child used a get-passive once, three children twice, one child four times and one child produced 10 out of possible 110 get-passives. Focusing on the get-passive only, we find that only the above mentioned three children produce get-passives more often in reply to the elicitation scenes than be-passives. These three 4-year-old children produced 46%, 45%, and 57% get-passives (which is comparable with the rate of be-passives produced

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

135

by other children) and only 3.8%o, 7.7%, and 2.9% be-passives. Their strategy seems to be to use to get instead of a form of to be whenever the patient is talked about. This way, he got melled (= smelled) is produced alongside with got seened, got forgotten (atypical passives), got followed, got drawed (medium typicality passives), got troddened on, got fallten over, and even got dead (their places in the hierarchy would be higher as the actionality involved is higher), and more prototypical passives like got chased, got pushed, got hitten, got eatened (only verb forms listed). A further six children produced between 1 and 10 (0.9% - 9%) get-passives out of the maximum of 110 pos­ sible passive sentences. One of them produced only one get-passive (he [the ring] got kissed, age group 3) and three of them produced only two getpassives each (got eaten all, got called, age group 3; he got left behind, he got eatened and the little spider got stolen (...), the cup got hurt, age group 4); the rest of their utterances consists mostly of -passives and some active voice utterances. One other 4-year-old child produced three get-passives (got eaten up, got left behind, she got pushed by the car), again with the rest of the utter­ ances being mainly be-passives and a few patient-focused actives of the type The postman was sad. The sixth 4-year-old produced ten get-passives alto­ gether: got hitten, got kissten, got drawed, got chased, got pushed, got eaten, got fallten over, got caught, got hurt, and got dead alongside with been pushed, been eat up, been killed, etc. Again, there seems to be no evident criterion for when the get- or the be-passive is selected. The results of Experiment 1 do not replicate the results obtained for American children (cf. Budwig 1990; Marchman et al. 1991). The claim that get-passives occur more frequently with animate patients (cf. Budwig 1990; Givón & Yang 1994) could not be shown with British children. Nor do the results show that the get-passives that did occur have a by-phrase more often than be-passives do (cf. Budwig 1990). British children do not show prefer­ ences for get-combinations with particular types of verbs or patients or agents either. In the few cases where get-passives were produced, get simply replaced be. The scarcity of elicited get-passives as opposed to the elicited be-passives shows a clear tendency towards the use of the be-passive among British children: an overall higher production of get- than be-passives cannot be found. More specifically, the claim that get-passives should be used more when expressing dynamic or prototypical transitive actions (Collins 1996; Marchman et al. 1991; Quirk et al. 1985; Romaine 1984; Keenan 1975; Gee 1974; Lakoff 1971; Svartvik 1966) cannot be supported for British children. Higher frequency of get-passives with prototypical transitive scenes could not be found in the age groups 2-4; neither was a gradual increase found in get-passive

136

KERSTIN MEINTS

production in accordance with the prototypicality hierarchy. Therefore, it can be stated that within the subcategory of get-passives, British children do not differentiate get-passives according to their degree of prototypicality. However, looking at the overall category of passives (be- and get-passives) and at the occasional use of get-passives, it could be argued that get-passives are periph­ eral members of the category; in other words, that be- and get-passives are not equally good category members in British English, with get-passives more central and get-passives at the periphery of the passive category. As such, getpassives might be accessible for typicality effects only to a later time in the development when the range of the passive category has been fully developed; in other words, at the beginning of passive acquisition process, the scope of the get-passive might not be fully developed yet. Thus, it can be hypothesized that with older children and adults, gradations within get-passives should occur as well (see Experiment 3 for adults). The experimental results support Marchman et al.'s findings that 3-year-olds use b-passives 15% more often than 4-5- or 5-6-year-olds to encode prototypically transitive events insofar as the British 3-4-year-olds produced even more -passives (between an average of 30% at the age of three and an average of 47% at the age of four) than their American peers. Another factor that might contribute to the use of the passive—and which has not been empirically tested yet—is patient affectedness (cf. Section 1.1). In order to test the claim that get-passives focus more on the patient and are used more if patients themselves are directly (and for the most part negatively) affected by the action (Svartvik 1966; Quirk et al. 1985; Givón & Yang 1994), the following experiment was set up to evoke elicitations of get-passives directly involving the children as patients. Using this new method also allowed the investigation of Budwig's more specific claim that children produce more get-passives if they are directly involved in the action themselves (Budwig 1990). Note that this approach is consonant with the above prototype approach. The verbs are taken from the group of highly typical action verbs, the patient is the child herself (animate, affected, and focused), and the agent is the experi­ menter (animate, acting, and defocused at the time of the question). The differ­ ence to Experiment 1 lies in the direct personal involvement of the child as the affected patient herself. Thus, this group of passives is a subgroup of the tran­ sitive prototype with an even stronger focus on personal and direct affect­ edness.

USE AND ACQUISITION OF GET- VS. BE-PASSIVES 3.

Experiment 2: Elicitation of children's get-passives

3.1

Introduction

137

In this experiment, the children themselves were directly involved in the action and directly affected by the action. The hypothesis was tested that children produce more get-passives if they themselves are the patient of the action and if they are directly and physically involved in and affected by the action, as was found by Budwig (1990). This study is at the same time an attempt to explore a new way of eliciting the get-passive as the usual elicitation procedure was only successful to elicit be- but not get-passives in British children. Highly prototypical action scenes were used in order to facilitate production as much as possible. It is possible that Experiment 1 may not have been focusing enough on the patient's affectedness to make the typicality hierarchy in the getpassive production visible. If it is the case that get-passives are inherently structured in the same way as be-passives are, children's get-passive produc­ tion could be enhanced by testing prototypically transitive scenes and at the same time adding direct affectedness (see Budwig 1990). The findings of Marchman et al. (1991) also suggest that it should be tested whether the transi­ tive prototype makes especially the production of get-passives easier. It is hypothesized that, if get-passives do express more direct affectedness of the patient by the action, then they should be produced in this task. 3.2

Method

Participants. Twenty-six children participated in this investigation, six 2-yearolds, ten 3-year-olds, and ten 4-year-olds. Nine children were visited in nurs­ eries in London and seventeen children were seen in nurseries in Oxford. Materials. The following passive sentences were to be elicited: (1) (2) (3) (4) (5) (6) (7)

I got lifted up (by you). I got scratched (by you). I got kissed (by you). I got stroked (by you). I got bitten (by you). I got tickled (by you). I got scared (by you).

Note that, except for (7), these sentences mostly encode highly prototypical transitive scenes. All actions/verbs were known to the children and the enacting of the sentences was carried out in random order.

138

KERSTIN MEINTS

Procedure. All children were in a testing room they were familiar with. They were asked if they would like to play a game with the experimenter. No toys were used; instead, a scene was enacted involving interaction between the experimenter and the child, whereby the experimenter assumed the role of the agent and the child was assigned the role of the patient. If, for example, the passive sentence I got lifted up by you was the target utterance, then the experimenter lifted the child up and afterwards seated the child again on the chair the child had been sitting in before the action. Directly after the scene had been enacted, the experimenter asked the child: 'What happened to you?', requiring the chiid to comment on the action. In the case of actions like to scratch, to bite, etc., the experimenter only pretended to carry out the action and this was explained to the child beforehand. The nursery's headmistress was present at all times. 3.3

Results and discussion

Even in this personalized experimental set-up, a rise in the production of getpassives could not be found. The results of the investigation are as follows: Age group

Be-passives

Get-passives

2 years (6) 3 years (10) 4 years 10)

8 (11.4%) 2 (2.9%)

1 (1.4%) 1 (1.4%)

Past participles only Active voice/others 13(18.6%) 8 (11.4%)

42 (100%) 48 (68.6%) 59 (84.3%)

Table 4: Number of elicited utterances in the get-passive investigation with specific emphasis on affectedness (Experiment 2)

Only two get-passives were produced by two different children with one of these being an adjectival get-passive I got scared (age 4) and the other one being a single verbal passive I got tickled (age 3). In contrast, ten be-passives were produced. These results show that get-passives are produced extremely rarely by British children, even under linguistic and contextual conditions strongly favorable to their production. This finding contrasts with the results for the American children studied by Budwig (1990) and Marchman et al. (1991). The differences in results between Experiments 1 and 2 on the one hand and the American studies on the other hand might be explained by the difference in British and American English, as emphasized by Dirven and Radden (1977) and Quirk et al. (1985). As we could not replicate the results of Budwig's and Marchman et al.'s studies on American English with British children, it seems plausible to argue that a different usage pattern of the get-

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

139

passive in American and British English (as Svartvik 1966 and Quirk et al. 1985 stated for adults) could account for the more frequent use of get-passives in American children and its infrequent use in British children. To sum up, British children only very infrequently produce get-passives. In Experiment 2, which was specifically designed to focus on a highly proto­ typical transitive scene while at the same time emphasizing the direct involve­ ment and affectedness of the patient by putting the child into the role of the patient of the action, only two children produce one get-passive each. It can be concluded that the get-passive cannot be found as often in British children's language production as in American children's utterances.16 Thus, the getpassive cannot be seen as a precursor of all children's -passives or as a prototypical children's passive. Instead, these results restrict possible gener­ alizations of Turner and Rommetveit (1967) and Marchman et al. (1991) to American children only. As this raises the question if regional variation, status of the get-passive within the category of passives, and parental usage patterns and input to the child play a role in the acquisition of the get-passive, the following study set out to investigate possible differences in parents' input to their children resulting from differences in the status of the get-passive for British English and American English adults. In view of Budwig's and Marchman et al.'s results with American adults, it is necessary to find out whether for British native speakers the get-passive is a peripheral member of the category of passives. Thus, the next step is to investigate whether British adults rate get- and -passives differently in terms of 'goodness of member­ ship' in the category of passives. 4.

Experiment 3: Adults ' ratings of be- and get-passives

4.1

Introduction

In order to gain more insight into the status of the different get- and be-passives in British children, and in particular to explain their preference of the bepassive, adults' judgments were obtained using ratings of get- and -passives. Here, the aim was to investigate the status of get- and -passives for British English speakers and to compare the results with data from American English speakers. If get-passives are in general judged by British adults as poorer examples of passive sentences, then this more peripheral member status could predict the infrequent use of get-passives by adults and subsequently also explain a lower frequency of use by children using British English. Analyzing which get- and -passives obtain high and low ratings will also enable us to determine whether there is a difference between prototypical and atypical get-

140

KERSTIN MEINTS

or be-passives; in addition, it will enable us to decide whether British adults judge highly prototypical get-passives as more acceptable than peripheral getox be-passives or whether British English speakers reject the get-passive in general, in which case the difference between get- and be-passive could be viewed as mainly a regional phenomenon. Thus, be- and get-passives might turn out not to be equally good examples of category members within the cate­ gory 'passive'. While gradations can occur within be- and within get-passives, be-passives could be more central in British English with get-passives at the periphery of the passive category. Research by Dirven and Radden (1977), Svartvik (1966), and Sussex (1982) also suggests a higher occurrence of getpassives in American English. Note that Budwig's participants accepted the relatively high number of 53% of get-passives in the sentence repair task and rejected 48%, while accepting 70% of be-passives and rejecting only 30%. In contrast, Marchman et al.'s production elicitation task revealed that adults produced only 8% of get-passives. 4.2

Method

Participants. 29 undergraduates and graduates at Oxford provided stimulus ratings; 18 were female, 11 male. They were aged between 18 and 30 years and were all monolingual British nationals. Materials. Each stimulus sentence was presented to the adult informants on a computer monitor and a rating scale that ranged from 1 (very good example) to 7 (very bad example) was displayed directly under the passive sentence. Informants made their ratings using the number keyboard on the computer. All stimuli were presented against a white background. First, the sentence was presented; after 500 ms, the rating scale appeared. The full display only disappeared after informants had made their judgment. The passive sentence and the scale were centered horizontally, with the sentence in the vertical center of the screen and the rating scale just below. Procedure. The informants received written instructions to rate prototypicality similar to those used by Rosch (1973). Thus, to clarify the difference between good exemplars of a category and less good exemplars, they were asked to think of a 'true red' as compared to an 'orangish red' or 'purple red', and they were reminded that some dogs (such as Chihuahuas) are less good examples of the category 'dog'. Participants were then asked to judge passive sentences according to the written instructions and to make their decision as quickly as possible. Informants had no difficulty in understanding the instruc­ tions. The experimenter started the first trial and all other trials were launched

USE AND ACQUISITION OF GET- VS. BE-PASSIVES

141

500 ms after the participants had pressed the number key indicating their rating. They were presented with a total of 226 sentences, in random order; half of the sentences were get-passives and the other half be-passives. Adults rated all passives the children had been tested on before (get- and be-passives), enabling us to compare 'goodness of membership' ratings in the category 'passive' of get- versus be-passives. 4.3

Results and discussion

The results of the ratings confirmed that indeed, get-passives received higher ratings than be-passives. The overall mean rating for be-passives was 3.06 (SD=0.73) and 4.09 (SD=0.66) for the get-passives. A dependent t-test showed that the differences between get- and be-passives were highly significant (p < 0.001), with get-passives being rated as rather poor examples of the passive construction. To test the extent to which the rating difference was stable across all get- and be-passives, each action group (4 groups: passives designating a highly prototypical, medium and low prototypical transitive scene; plus adjec­ tival passives; see Table 2) was coded in terms of whether the rating difference was in the direction expected (i.e. a lower rating result for the be-passive groups as compared to the get-passive groups) or reversed. For each of the four verb groups, the rating difference was in the direction expected and differed markedly from chance expectation (p < 0.001). A further 2x2 analysis of variance with only highly typical and highly atypical get- and be-passives reveals a highly significant main effect for the factor 'Auxiliary' (get versus be) with p < 0.001 and a highly significant interaction between the factors 'Auxil­ iary' and 'Typicality' (p < 0.001). Figures 2 and 3 illustrate the effects: The interaction shows strong differences between typical and atypical get-passives (planned comparisons: p < 0.001), between typical get- and bepassives (planned comparisons: p < 0.001) and between atypical get- and passives (planned comparisons: p < 0.001). There is no significant difference between typical and atypical be-passives. From the data, two important results emerge. First, the main effect for the factor 'Auxiliary' shows that British adults judge be-passives generally as sig­ nificantly better examples of passive sentences than get-passives. As predicted, sentences like The postman was bitten by the dog are rated to be better examples of the category 'passive' than get-passives like The postman got bitten by the dog; or conversely, get-passives seem to be generally less good examples of the passive category than be-passives. As the get-passive's more peripheral status as a category member suggests a less frequent use of get-

142

KERSTIN MEINTS

passives by adults, this could subsequently also explain a lower production frequency by British English children (Experiment 1). Main effect of Auxiliary F(1,18)=223.18;p dative shift has generally been described as a feature of spoken language, here we have evidence that the dative is advancing even in highly controlled, written forms of language that can be safely considered typical instances of 'standard German'. In all other cases, a dative-postposition (in pre-posed position) comes to govern a 'new' genitive (Table 4 below):

214

gegenüber gleich entgegen ähnlich nahe/nah entsprechend gemäß entlang zum Nutzen zu Ehren

CLAUDIO DI MEOLA

Dative

Genitive

725 65 55 25 60 79 49 17

1 1 3 2 5 10 16 52 5 25

Table 4: Genitive-government for

Genitive (%) dative (cf. Müller 1990; Ágel 1992; Petig 1997): e.g. kraft "by virtue of', mittels "by means of', während "during". Still, our corpus contains clear evidence that quite a few dative prepositions can change to the genitive: the prepositional genitive proves to be highly productive (cf. Di Meola 1999, 2002). Finally, a number of postpositions do not change government. First of all, there is the group of accusative adpositions: betreffend, eingeschlossen, ausge­ nommen. Then there are three genitive adpositions (the extremely rare bar and eingedenk; the rather obsolete ungeachtet) and two dative adpositions (näher, fern) with very few preposed occurrences. In conclusion, all common dative and genitive postpositions tend to acquire a new case: a genitive and a dative respectively. As far as the distribution of the 'new case' over the five text types is concerned, non-fictional texts again prove to be more innovative: out of 90 occurrences of the 'new' genitive, 82 can be found in non-fictional texts and 8 in fictional ones. Only the dative adposition wegen shows the opposite tendency: the 'new' dative occurs 35 times in non-fictional texts and 112 in fictional ones.4

GRAMMATICALIZATION OF POSTPOSITIONS IN GERMAN

215

5.4. Change of semantics It is well-known that semantic bleaching is one of the key features of grammaticalization. It also characterizes the semantic development of various German postpositions, in that their original meaning has been subject to gener­ alization. For example, wegen has lost its original meaning of 'path', Trotz "contrariness" in zum Trotz "despite" has taken on a general meaning of 'oppo­ sition', and so on. In this section, I will focus on another aspect of adpositional semantics; in particular, I will consider those adpositions which show both preposing and postposing and examine whether there are semantic differences between the two positional variants. Two semantic groups of adpositions can be distinguished: (i) adpositions with spatial as well as non-spatial meaning (gegenüber, entgegen, entlang', nahe/nah, näher, fern); (ii) adpositions with exclusively non-spatial meaning (gemäß, gleich, ähnlich, bar, eingedenk; entsprechend, betreffend, einge­ schlossen, ausgenommen, ungeachtet; zu Ehren, zum Nutzen, wegen). The six adpositions of the first group have a prototypical spatial meaning. The following two pairs of sentences illustrate the differences between spatial and non-spatial uses: (15) a. Sie lief dem Auto entgegen. "She ran towards the car." b. Entgegen allen Erwartungen gewann sie das Spiel. "Against all expectations she won the match." (16) a. Sie steht nahe dem Baum. "She stands close to the tree." b. Sie steht ihrem Bruder nahe. "She is close to her brother." Interestingly, the two types of usage tend to occur in different positions. As Table 5 shows,5 the positional pattern of the spatial and non-spatial uses of the adpositions nahelnah, fern, and entgegen shows a clear tendency towards complementary distribution. Both entlang and genenuber fail to show signifi­ cant positional differences for the two types of semantic use; actually, postposing seems to be ruled out altogether for entlang,s non-spatial uses. Moreover, entlang shows a substantial semantic difference between its postposed and preposed spatial uses: postposed entlang (governing the accusative) expresses a goal-oriented, complete movement, while preposed entlang

216

CLAUDIO DI MEOLA

(governing the dative or the genitive) denotes other types of movement or even a static configuration:6 (17) a. Erfährt die Straße entlang bis zur Kirche. "He drives along the road up to the church." b. Entlang der Straße liegen schöne Häuser. "There are beautiful houses alongside the road."

nahe/nah fern entgegen entlang gegenüber

Position

Spatial

preposed postposed preposed postposed preposed postposed preposed postposed preposed postposed

129 1 3 6 6 62 184 92 1009 236

Non-spatial 9 32 7 3 120 44 10

81 26

Table 5: Position and semantics

The second semantic group comprises thirteen adpositions with exclusively abstract meaning. In what follows, I will present some of their semantic differ­ ences correlating with the two positional variants. Adpositions have a stronger tendency to be postposed with proper names than with other governed phrases. Adpositions occurring with proper names in our corpus {gleich, ähnlich, betreffend, eingeschlossen, ausgenommen, zu Ehren, zum Nutzen, wegen) show the following distribution: 28% of postposing in the case of proper names (22 out of 80), but only 7% (148 out of 2,051) with other governed phrases. The adpositions gleich, entsprechend, and gemäß constitute an interesting subgroup. In a broad sense they are all connected to a semantic element of 'correspondence', but there seem to be slight differences between preposing and postposing. Gleich has two different meanings: (i) similarity {gleich einem Hund "like a dog"); (ii) exact correspondence or identity {die Exporte sind gleich den Importen "the exports equal the imports"; die Summe ist gleich Null "the sum equals zero"). While the first semantic variant can occur both with preposing and with postposing {gleich einem Hund/einem Hund gleich), the second is only to be found with preposing {"die Exporte sind den Importen

GRAMMATICALIZATION OF POSTPOSITIONS IN GERMAN

217

gleich) *die Summe ist Null gleich). Entsprechend can express rather vague correspondence {entsprechend den Regeln "in correspondence with the rules") or exact accordance {entsprechend Absatz 5 "in accordance with Section 5"). While the more 'general' semantic variant can also be postposed {den Regeln entsprechend), postposition is not allowed with the more 'specific' variant (??Absatz 5 entsprechend). Similarly, gemäß describes vague correspondence {der Verfassung gemäß "in correspon-dence with the constitution") as well as exact accordance {gemäß Absatz 5 "under Section 5"). Finally, for both semantic groups—prototypically spatial and exclusively non-spatial—there is a general tendency for postposing in the case of idiomatic expressions or in the case of other more or less fixed units: des Geldes wegen "for money", den Umständen entsprechend "under the circumstances", den Tränen nahe "near to tears", Frauen gegenüber "with regard to women", etc. In conclusion, the adpositions analyzed here seem to develop semantic differences correlating with the two positional variants. My intent here was to give a brief outline of main tendencies. These need to be investigated in a more specific, semantically-centered study. 6.

A diachronic outlook on position and government

So far, I have examined the synchronic result of grammaticalization processes: the coexistence, in contemporary German, of differently grammaticalized vari­ ants of one and the same form. I will now examine the processes from a diachronic perspective. As a basis of comparison, I will use the electronic corpus edited by Bertram et al. (1997), which comprises the work of fifty-eight eminent literary authors in the period between 1750 and 1920. I will consider two syntactic aspects: change of position and change of government. Consider first those adpositions which today show an alternation between postposing and preposing. Table 6 below presents the proportion of the two positional variants in older texts. A comparison with the figures for contem­ porary German (though including non-literary texts) shows that the percentage of preposed uses has considerably increased for nearly all adpositions. As expected, in the course of time the 'new', more grammaticalized variant (i.e. preposing) has been gradually replacing the 'old', less grammaticalized one (i.e. postposing).

218

CLAUDIO DI MEOLA

Postposed betreffend entgegen

gemäß näher bar Eingeschlossen gegenüber entlang ähnlich nahe/nah Ausgenommen zu Ehren Entsprechend eingedenk wegen zum Nutzen gleich ungeachtet fem

66 1383 532 342 13 9 1246 291 429 328 276 151 30 35 1166 9 729 108 77

Preposed 2 55 26 23 1 1 172 44 75 76 94 58 12 36 1911 16 1479 234 190

Preposed(%) 3% 4% 5% 6% 7% 10% 12% 13% 15% 19% 25% 28% 29% 51% 62% 64% 67% 68% 71%

Preposed today (%) 26% 45% 97% 3% 88% 16% 73% 68% 56% 80% 79% 83% 75% 50% 96% 71% 82% 98% 36%

Table 6: Postpositions with alternation postpsing/prep sing (1750-1920)

Let us now consider the historical development of government. The only postposition characterized by the direction of change genitive > dative (wegen) does not show a clear diachronic trend. I will therefore concentrate on the dative-postpositions acquiring a genitive. Table 7 below illustrates—for preposed occurrences—the proportion of 'old' dative and 'new' genitive. Dative ähnlich nahe /nah entsprechend gemäß gleich gegenüber entgegen entlang zum Nutzen zu Ehren

58 56 6 18 1173 117 39 30

-

Genitive

2 1 1 8 12 34

Genitive (%)

5%

0% 0% 0% dative-government), the new function comes to be signaled by a new construc­ tion. The reanalyzed form thus progressively departs from its structural origin: a dative-postposition ends up as a genitive-preposition and a genitive-postposition ends up as a dative-preposition. As a result of this differentiation process, a clear boundary between content word and function word has been established. Syntactic differentiation can be reinforced by loss of morpho­ logical transparency (through word formation) and semantic transparency (through generalization or transfer of meaning). (iii) Interrelation of grammaticalization parameters. The various aspects of grammaticalization are strictly interrelated: a change of position is almost invariably followed by a change of government; conversely, a change of government is not possible without a prior change of position. In other words, a mew' case always arises in pre-position. Thus, a clear chronological sequence can be established: change of position seems to be a necessary precondition for change of government. (iv) Layering. In the case of syntactic and semantic change, older variants often persist in language use: postposing beside preposing, 'old' genitive beside 'new' dative, 'old' dative beside 'new' genitive, spatial semantics beside non-spatial semantics. As a result, differently grammaticalized stages of one and the same form may coexist (cf. Hopper 1991).

220

CLAUDIO DI MEOLA

One aspect that certainly deserves further investigation is the inter­ connection between two potentially contrasting grammaticalization principles of German postpositions: differentiation from the original structure and approximation to the prototype. In some cases, this differentiation results in an alignment with prototypical characteristics of German adpositions; consider, for instance, the transition from postposing to preposing and the transition from genitive to dative. On the other hand, there is a substantial group of dativepostpositions which tend to transition to the (non-prototypical) genitive, reviving a case generally considered unproductive in contemporary German. Notes 1It should be mentioned that the adjectives gemäß "appropriate" and eingedenk "bearing in mind" can only occur predicatively but not attributively. The adjectival use of these two forms, however, is extremely rare in contemporary German. 2 Abbreviated forms such as gem. for gemäß and einschl. for einschließlich have been included. 3 Zum Nutzen and zu Ehren are not generally considered adpositions and the question of government is thus completely disregarded. 4 As far as the structure of the governed NP is concerned, no relevant distributional tendency could be found (the overall number of occurrences of preposed pronominal NPs was too small for statistical exploration). 5 In Table 5, pronominal NPs are not considered, since they have the strong tendency to be postposed regardless of their semantics. Furthermore, Table 5 does not list näher, as it only occurs once in preposed position in the present corpus 6 Cf. Di Meola (1998); for the semantics of the dative and accusative in German, see, among others, Smith (1987) and Leys (1995).

References Abraham, Werner. 1990. "Postposition, Präpositionaladverb oder Präfix? Zum Status der sog. 'trennbaren Verbpräfixe' im Deutschen". Studien zum Deutschen aus kontrastiver Sicht ed. by Andrzej Katny, 7-34. Frankfurt a.M.: Peter Lang. Abraham, Werner, ed. 1993. Grammatikalisierung und Reanalyse: Konfron­ tation. (= Folia Linguistica Historica, 13). Berlin: Mouton de Gruyter. Agel, Vilmos. 1992. "Die deutschen Genitivpräpositionen: Prinzipien ihrer Verwendung". Texte, Sätze, Wörter und Moneme: Festschrift für Klaus Heger zum 65. Geburtstag ed. by Susanne R. Anschütz, 17-33. Heidel­ berg: Orientverlag.

GRAMMATICALIZATION OF POSTPOSITIONS IN GERMAN

221

Bartels, Gerhard & Birgit Tarnow. 1993. Von 'à' bis 'zwischen': Das Beziehungs-wort der deutschen Gegenwartssprache. Frankfurt a.M.: Peter Lang. Bertram, Mathias et al., eds. 1997. Deutsche Literatur von Lessing bis Kafka. Berlin: Directmedia. Di Meola, Claudio. 1998. "Semantisch relevante und semantisch irrelevante Kasusalternation am Beispiel von 'entlang'". Zeitschrift für Sprach­ wissenschaft 17.204-235. 1999. "'Entgegen', 'nahe', 'entsprechend' und 'gemäß': Dativpräpo­ ------. sitionen mit Genitivrektion". Zeitschrift für germanistische Linguistik 27.344-351. ------. 2000. Die Grammatikalisierung deutscher Präpositionen. (= Studien zur deutschen Grammatik, 62). Tübingen: Stauffenburg. -------. 2002. "Präpositionaler Rektionswechsel unter dem Gesichtpunkt der Grammatikalisierung: Das Prinzip der 'maximalen Differenzierung'". Perspectives on Prepositions ed. by Hubert Cuyckens & Günter Radden, 101-129. Tübingen: Max Niemeyer. Diewald, Gabriele. 1997. Grammatikalisierung: Eine Einführung in Sein und Werden grammatischer Formen. Tübingen: Max Niemeyer. Engel, Ulrich. 1996. Deutsche Grammatik: Vollständige Darstellung der deutschen Gegenwartssprache, 3rd edition. Heidelberg: Groos. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. 1991. Grammaticalization: A Conceptual Framework. Chicago, I11.: The University of Chicago Press. Helbig, Gerhard & Joachim Buscha. 1986. Deutsche Grammatik: Ein Hand­ buchfür den Ausländerunterricht, 9th edition. Leipzig: Enyklopädie. Hopper, Paul J. 1991. "On Some Principles of Grammaticization". Approaches to Grammaticalization ed. by Elizabeth.  Traugott & Bernd Heine, Vol. II, 17-35. Amsterdam: John Benjamins. ----& Elizabeth C. Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. König, Ekkehard & Bernd Kortmann. 1991. "On the Reanalysis of Verbs as Prepositions". Approaches to Prepositions ed. by Gisa Rauh, 109-125. Tübingen: Gunter Narr. Kortmann, Bernd & Ekkehard König. 1992. "Categorial Reanalysis: The Case of Deverbal Prepositions". Linguistics 30.671-697.

222

CLAUDIO DI MEOLA

Lang, Jürgen & Ingrid Neumann-Holzschuh, eds. 1999. Reanalyse und Grammatikalisierung in romanischen Sprachen. (= Linguistische Arbeiten, 410). Tübingen: Max Niemeyer. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1, Theoretical Prerequisites. Stanford, Calif: Stanford University Press. Lehmann, Christian. 1995. Thoughts on Grammaticalization. München: LINCOM Europa. Leys, Odon. 1995. "Dativ und Akkusativ in der deutschen Sprache der Gegenwart". Leuvense Bijdragen 84.39-62. Lindqvist, Christer. 1994. Zur Entstehung von Präpositionen im Deutschen und Schwedischen. (= Linguistische Arbeiten, 311.). Tübingen: Max Niemeyer. Müller, Wolfgang. 1990. "Mittels neuen Kleinbusses oder mittels neuem Kleinbus?" Muttersprache 100.53-59. Petig, William E. 1997. "Genitive Prepositions Used with the Dative in Spoken German". Unterrichtspraxis 30.36-39. Schröder, Jochen. 1986. Lexikon deutscher Präpositionen. Leipzig: Enzyklopädie. Smith, Michael B. 1987. The Semantics of Dative and Accusative in German: An Investigation in Cognitive Grammar. Ph.D. dissertation, University of California, San Diego. Taylor, John R. 1995. Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Oxford University Press. Weinrich, Harald. 1993. Textgrammatik der deutschen Sprache. Mannheim: Dudenverlag. Willems, Klaas 1997. Kasus, grammatische Bedeutung und kognitive Linguistik: Ein Beitrag zur allgemeinen Sprachwissenschaft. Tübingen: Gunter Narr. Zifonun, Gisela, Ludger Hoffman, Bruno Strecker et al. 1997. Grammatik der deutschen Sprache. Berlin: Mouton de Gruyter.

METONYMY IN COGNITIVE LINGUISTICS AN ANALYSIS AND A FEW MODEST PROPOSALS

ANTONIO BARCELONA Universidad de Murcia

The paper is devoted to the discussion of some of the main issues raised by the standard theory of metonymy in Cognitive Linguistics—which views metonymy as a mapping within one domain—and to the presentation of some new proposals. Both the discussion and the proposals are aimed at refining the theory, which the author accepts in most of its basic tenets. Particular attention is devoted to the entity vs. domain issue, to the single-domain vs. two-domain issue, and to the questions of mapping and activation, subdomain centrality, and referentiality. A number of suggestions are offered with respect to each of these issues. On the basis of these suggestions, four tightly interrelated basic kinds of conceptual metonymy are proposed: 'sche­ matic' metonymies, 'prototypical' metonymies, 'typical' metonymies, and 'conventional' metonymies. It is argued that schematic metonymies are the most general type of conceptual metonymy and that the first three types represent three degrees of metonymicity.

1.

Introduction1

The goal of this article is to address some of the problems that affect the stan­ dard cognitive linguistic account of metonymy, an account which, for the most part, I endorse. By the 'standard' or 'dominant' account of metonymy is meant the conception developed mainly by George Lakoff, Mark Johnson, and Mark Turner in three fundamental publications: Lakoff and Johnson (1980), Lakoff (1987), and Lakoff and Turner (1989). I will begin by specifying the main 'standard' properties of metonymy, and provide a definition of metonymy that can be deduced from them (Section 2). I will then examine those properties, pointing out the problems they create for the theory, and I will offer some answers (Section 3). Section 4 contains the main proposals of the article. In the light of the conclusions reached in Section 3, three basic 'degrees' of metonymicity are postulated: 'schematic' metony­ mies, 'typical' metonymies, and 'prototypical' metonymies. The fourth kind of

224

ANTONIO BARCELONA

metonymy proposed, 'conventional' metonymies, is not set up on the basis of its degree of metonymicity (i.e. conventional metonymies can be 'schematic', 'typical', or 'prototypical'), but on the basis of its likelihood to become cultur­ ally entrenched and to enter a linguistic system. The ideas presented in this paper constitute a thorough reworking and an extension of those presented in Barcelona (2000a) and Barcelona (1999). 2.

The 'standard' view of metonymy in Cognitive Linguistics

This section summarizes, without any attempt at critical analysis, what I call the 'standard' or 'dominant' account of metonymy in Cognitive Linguistics. In the short history of Cognitive Linguistics, ideas on metonymy have undergone several changes, from Lakoff and Johnson's (1980) initial emphasis on its referential function, to Lakoff s (1987) stress on its role in categorization, and to its characterization in Lakoff and Turner (1989) as a form of mapping. In the course of this development, these cognitive linguists have added several other properties, resulting in the so-called 'standard' view of metonymy. The larger part of this section, therefore, will be devoted to a description of these proper­ ties (Section 2.1). Section 2.2 will be concerned with the claims made by the proponents of the standard view on the differences and similarities between metaphor and metonymy. Finally, Section 2.3 will present the notion of me­ tonymy that results from the properties spelled out in 2.1 and 2.2. 2.1

The properties

Most of the 'standard' properties can already be found in Lakoff and Johnson (1980); other properties were added in the two other studies cited above. a.

b.

Metonymy is a conceptual process, not merely a figure of speech (Lakoff & Johnson 1980:36). In fact, metonymy need not even be expressed linguistically. Gestural metonymies constitute a good case in point: in a situation where an office clerk answers the question 'Who has just come in?' by simply covering his/her ears with two large newspapers, all of his/her colleagues will immediately under­ stand that the clerk is referring to a customer with uncommonly large ears. Metonymy involves two conceptual entities which are closely asso­ ciated in experience (Lakoff & Johnson 1980:35). In example (1), we refer to the Yugoslavian government by virtue of mentioning its location (Belgrade), with which it is closely linked in experience:

METONYMY IN COGNITIVE LINGUISTICS

225

(1) Belgrade did not sign the Paris agreement. The experiential contiguity of the entities linked by metonymy allows Lakoff and Johnson (1980:36) to include synecdoche in me­ tonymy, since a physical or a functional part of an entity is closely linked in experience to the entity as a whole: (2) The factory has taken on two hundred extra hands, (A BODY PART FOR THE WHOLE PERSON) [PART FOR WHOLE]

c.

d.

e.

Property (b) implies that metonymy is experientially motivated (Lakoff & Johnson 1980:39). It arises on the basis of human bodily and spatial experience and on the basis of cultural and cognitive models. Metonymy is a cognitive model. In addition to its usual referential role, metonymy is also attributed with an important cognitive func­ tion (Lakoff & Johnson 1980:36-40). Metonymy can be used to construct conceptual categories, particularly radial categories built around metonymic prototypes. Metonymic models for categories or for individuals often govern reasoning, discourse inferences, and grammar (Lakoff 1987:77-90; Gibbs 1994:324-333; Panther & Thornburg 1999,2000). Metonymy is systematic. Metonymies respond to various patterns (Lakoff & Johnson 1980:37-40), usually represented in small capi­ tals. In the examples below, they have been enclosed in round brack­ ets. These patterns or types can normally be grouped within other more general types, indicated below in square brackets.2 (3) The ham sandwich is waiting for his check. (FOOD ITEM FOR CUSTOMER) [PART FOR PART]

(Lakoff & Johnson 1980:35) (4) She's just a pretty face, (A BODY PART FOR THE WHOLE PERSON) [PART FOR WHOLE]

(Lakoff & Johnson 1980:37) (5) I'll just order a Löwenbräu. (PRODUCER FOR PRODUCT) [PART FOR PART]

(cf. Lakoff & Johnson 1980:38)

226

ANTONIO BARCELONA

(6) Washington is insensitive to the needs of the people. (PLACE FOR POLITICAL INSTITUTIONS LOCATED IN IT) [PART FOR WHOLE]

(Lakoff & Johnson 1980:38) (7) There have not been many Shakespeares lately. (STEREOTYPICAL MEMBER OF A CATEGORY—THAT OF TALENTED WRITERS—FOR THE CATEGORY)3 [PART FOR WHOLE]

(8) He walked with drooping shoulders. He had lost his wife. (EFFECT FOR CAUSE) [PART FOR PART]

(9) John has a long face, (EFFECTFORCAUSE)[PARTFORPART]

f.

g.

h.

i.

Examples (8) and (9) manifest the metonymic pattern EFFECT FOR CAUSE. According to Kövecses and Radden (1998: 56), a cause (an emotion in this case) and its effect are two conceptual 'parts' in what they call the 'causation ICM'. The CAUSE in (8) is an emotion (sadness), and the EFFECT is a behavioral effect of this emotion (a drooping bodily posture, namely drooping shoulders). The cause in (9) is the same emotion, and the effect another behavioral effect of this emotion (again, a drooping bodily posture, in this case a sagging jaw). In these two examples, then, the metonymie pattern EFFECT FOR CAUSE is manifested in the lower-level metonymy EFFECTS OF EMOTION FOR EMOTION. The effects of emotion that are usually metonymie sources are either physiological or behavioral (see Kövecses 1990:209-211, and Barcelona 1986:21). Metonymy involves two conceptual entities. 'Entity' is the term used by Lakoff and Johnson (1980:36) and Lakoff and Turner (1989:103). Metonymy occurs in a single domain. This property is also intro­ duced in Lakoff and Turner (1989:103), and seems a refinement of Lakoff and Johnson's (1980) and Lakoff s (1987) earlier characteri­ zation of metonymy as involving two entities closely associated in experience; if both are part of the same cognitive domain, this is because they are experientially close. In Cognitive Linguistics, cognitive domains are fundamentally experience-based domains. Metonymy is a stand-far relationship, which "allows us to use one entity to stand for another" (Lakoff & Johnson 1980:36). In (1), an entity (Belgrade) stands mentally for another entity (the Yugoslavian government). Metonymy is a type of mapping. This property, which is introduced in Lakoff and Turner (1989:103), seems to account for the cognitive

METONYMY IN COGNITIVE LINGUISTICS

j.

k. 1.

2.2

227

effect of the 'stand-for' relationship. As I indicated above, if entity A stands mentally for entity B, then, A is also somehow projected onto B. As is well known, the element that is mapped is called the 'source', and the element onto which the source is mapped is called the 'target'. Metonymy is primarily referential (Lakoff & Johnson 1980:36). This seems to be true of nominal metonymies, which are typically used for the purpose of referring to one entity by means of another. Metonymy can be conventionalized, i.e. "made part of our everyday conceptual system" (Lakoff & Turner, 1989:104). Metonymy is used for an immediate pragmatic purpose. It is often used in discourse "for some limited and immediate purpose" (Lakoff 1987:79): it may make processing easier or faster, or it may save the speaker and listener extra effort.

Similarities and differences between metaphor and metonymy

Metonymy is said to share with metaphor several of the properties discussed above (Lakoff & Turner 1989:103-104): -

both both both both

are conceptual, often conventional, processes; are experientially motivated and systematic; are fundamental types of cognitive models; are conceptual mappings.

Then again, metonymy (on the standard view) differs from metaphor in the following respects (cf. also Lakoff & Turner 1989:103): -

-

-

Metonymy involves a link between entities, whereas metaphor links domains. The two terms in the relationship are experientially and conceptually close in metonymy and distant in metaphor. That is, the entities connected by metonymy are included in the same domain, whereas metaphor establishes a link between two separate domains. Metonymy is a 'stand-for' relationship, whereas metaphor is princi­ pally a way of "conceiving of one thing in terms of another" (Lakoff & Johnson 1980:36; my emphasis). Metonymy is primarily referential; no such claim has been made about metaphor in the three studies above.

228

ANTONIO BARCELONA

-

2.3

Metonymy is used for an immediate limited pragmatic purpose; no such claim has been made about metaphor in the three studies above. Metaphor is unidirectional (Lakoff & Turner 1989:132-133), where­ as nothing is said in this respect about metonymy—at least not in the standard view.

A 'standard' definition

Sections 2.1 and 2.2 allow us to provide the following preliminary definition summarizing the standard view of metonymy: Metonymy is a mapping with a primarily referential purpose, in which the source and the target are entities in the same domain. 3.

The 'standard' view of metonymy: Problematic issues

The following characteristics of the dominant view of metonymy are uncontroversial and do not need further clarification, as they have been convincingly justified in Lakoff and Johnson (1980), Lakoff (1987), and Lakoff and Turner (1989): -

Metonymy is fundamentally conceptual in nature (like metaphor)— cf. property (a). Metonymy is experientally grounded (like metaphor)—cf. property (b). Metonymy involves elements which are experientially (thus concep­ tually) contiguous (unlike metaphor)—cf. property (c). Metonymy can be the root of certain cognitive models (like meta­ phor)—cf. property (d).

The remaining properties, to my mind, are either decidedly controversial (at least in the way they have been formulated), or in need of further clarification, as they raise a number of unanswered questions. In what follows, I will discuss each of these properties, pointing out the problems that affect them and suggesting some possible answers.

METONYMY IN COGNITIVE LINGUISTICS

3.1

229

Metonymy as a relationship between '(sub)domains' rather than 'enti­ ties '

Metonymy is usually described as a relationship between two entities, whereby the term 'entity' does not merely refer to physical entities, but rather to conceptual entities in a 'schema' or 'domain'. As such, it may cover objects, animate beings, and even well-defined abstract objects such as 'agents', 'actions', 'institutions', 'emotions', 'categories', 'causes, 'effects' (cf. exam­ ples (6)-(9) above, for illustrations of metonymic relationships involving at least an abstract entity).4 However, these abstract entities have an intricate conceptual structure and as such constitute 'domains' themselves (either abstract domains or basic domains; cf. Langacker 1987:147-152); even concepts correlating with physi­ cal entities (e.g. FINGER) may constitute abstract domains. Following Croft (1993), I would therefore like to suggest that the elements linked by metonymy are domains rather than entities.5 Now since, according to the standard view, metonymy occurs within one single domain (cf. infra), I suggest that a domain linked by metonymy be called a subdomain, if it is included in the overall experiential domain within which metonymy operates. For example, in a PART FOR WHOLE metonymy (cf. 4), a subdomain (BODY PART) stands for the whole

domain (PERSON). In a PART FOR PART metonymy (cf. 3), a RESTAURANT domain, namely FOOD ITEM, stands for another RESTAURANT domain, namely the CUSTOMER that ordered it. 3.2

subdomain of the subdomain in the

Same or different domain?

On the standard view, metonymy links entities/(sub)domains within the same common domain; this criterion is essential in distinguishing metonymy from metaphor, which links two separate domains. Put differently, in metonymy, a source entity/domain is mapped onto a target entity/domain, both of which are in the same domain, whereas in metaphor, the source and target are said to be totally different domains, which do not include each other. In the metonymic expression (4) She's just a pretty face, both source and target can be said to be in the same domain, in that the face (the source) is included in the PERSON domain (the target). In the metaphorical expression Our love is starting to wither, on the other hand, the domain of LIVING BEINGS, specifically plants, is projected onto that of EMOTIONS, specifically love; but love is not literally a type of plant, nor are plants a type of emotion. Even though the 'same domain' property does work in many cases, it is not a completely reliable criterion to distinguish metaphor from metonymy; as

230

ANTONIO BARCELONA

such, a neat distinction between metaphor and metonymy cannot always be maintained. First, a cognitive domain is understood by Langacker (1987:154-158), Taylor (1995:83-87), and most other cognitive linguists as an 'encyclopedic' domain (i.e. including all the entrenched knowledge about the domain that a speaker may have). Thus, it will normally vary in breadth from person to person, and may not have precise boundaries. If that is the case, i.e. if no neat distinction between two domains can be established, how, then, can metonymy reliably be distinguished from metaphor? This problem might be illustrated with the following example. Most English speakers would agree that the cause-effect link between an emotion such as sadness and certain behavioral effects of sadness is included in the overall domain of SADNESS. One of these behavioral effects is a drooping bodily posture (drooping head, shoulders, or facial muscles), which then constitutes a subdomain of SADNESS. NOW, this subdomain presupposes the domain of VERTICALITY; this, in turn, might lead us to conclude that VERTICALITY (and possibly SPACE) is also part of the cognitive domain of SADNESS, and that therefore the mappings, as illustrated in (10), would be metonymic and not metaphorical. (10) a. b.  d. e.

She is in the pits. Mike is in low spirits, I am prostrated with grief. Mary is down. She has been a bit droopy lately.

Indeed, it could be argued that in (10), a subdomain within the overall domain of SADNESS (viz. VERTICALITY, as manifested by one of the behavioral effects of sadness, viz. 'drooping bodily posture') would be mapped onto another part of the overall domain (SADNESS as a cause); or that a part of the SADNESS domain (VERTICALITY) is mapped onto the whole domain. That is, as in every metonymy, the source and the target would be in the same superordinate domain. However, most cognitive semanticists would be inclined to regard these examples as manifestations of the metaphor SADNESS IS DOWN, and not of the metonymy DOWN (VERTICALITY) FOR SADNESS. Note, in this respect, that the sentences in (10) can easily be used to describe the emotional state of someone who may not actually display the conventional drooping posture behavior; in other words, there certainly exists a conventional metaphorical mapping of DOWN onto SADNESS which is independent of actual bodily behav-

METONYMY IN COGNITIVE LINGUISTICS ior.6 The problem is that the source

(VERTICALITY)

231

is apparently included in the

target (SADNESS).

A possible solution is offered in Barcelona (1997, 2002a:233-236). While the domain of VERTICALITY may be indirectly included in the domain of SADNESS via speakers' unconscious experiential knowledge of the behavioral effects of sadness, no English speaker is likely to categorize VERTICALITY (or SPACE) consciously as a subdomain of SADNESS. In other words, speakers would normally reject the proposition that the domain of SADNESS includes that of VERTICALITY (or SPACE in general), if it were explicitly submitted to them. These observations, according to Barcelona (1997, 2002a:235), suggest that speakers make use of a conscious folk taxonomy of domains (a culture-specific Idealized Cogntive Model of the taxonomy of domains), in which certain (sub)domains are excluded from others, and in particular, in which the domains of VERTICALITY and SADNESS are separated. Two domains, then, which by this conscious taxonomy are not included in the same overall domain may give rise to metaphorical mapping; conversely, two domains which by this conscious classification are included in a single overarching domain may be linked metonymically.7 These remarks should not be interpreted as claiming that every time the SADNESSISDOWNmetaphor is activated, people are conscious of the fact that verticality belongs to an altogether separate domain. As is wellknown, both metaphor and metonymy are typically unconscious mental proc­ esses. While these observations may help us to distinguish metonymy from metaphor in most cases, they would not be of great help in examples such as John is a lion. No cognitive linguist would probably regard this example as metonymic, given that the folk taxonomy of domains does not include ANIMALS and PEOPLE into each other; 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 domain of LIVING BEINGS.8 Perhaps a solution would be, then, to require that the metaphorical source and target be regarded by the conscious taxonomy of domains, not only as not included in each other, but also as being 'conceptually distant' (in metonymy, source and target would be conceptually 'close').9 However, conceptual distance is a relative concept, and it is not easy to provide non-ad hoc criteria to decide how 'distant' two domains have to be from each other to be metaphorical. A second factor attesting to the actual difficulty of determining whether two domains belong to a single overarching domain or whether they are inde­ pendent from each other is contextual in nature and is illustrated by cases such

232

ANTONIO BARCELONA

as He fell in the war (cf. Bultinck 1998:65), which can be interpreted either metonymically or metaphorically. Depending on the context of interpretation, this sentence could describe the conventional image of a soldier who is seri­ ously wounded as he advances through enemy lines, who falls, and subse­ quently dies; in this case, the meaning could be claimed to be metonymic: A SALIENT EVENT (FALLING) FOR A SALIENT SUBSEQUENT EVENT (DYING). Then

again, the speaker could simply refer to the fact that the soldier died when he was in the war, not taking into consideration whether or not he actually fell down before dying; on this interpretation, the domain of FALLING is used as a generalized metaphorical source for the domain of DYING AT WAR. Knowledge of the context is decisive to know whether the domains of FALLING, or down­ ward movement in general, and that of DYING AT WAR are in the same domain (metonymy), or in two different experiential domains (metaphor). In these cases, there is only a thin basis for distinguishing between metaphor and metonymy. Examples such as He fell in the war point to a further complicating factor. Doubtless, the generalized metaphorical mapping between separate domains (expressed, e.g., in He fell in the war) is grounded in the prototypical metonymie connection between two subdomains (e.g. actual FALLING and DYING AT WAR) of a single, overarching domain. What this suggests, then, is that a large number of metaphors are based on or motivated by metonymy (Barcelona 2000b; Radden 2000). Given that it is often not possible to neatly distinguish between different domains, i.e. that the distinction between domains is often gradual, the distinc­ tion between metaphor and metonymy in terms of within- or cross-domain connections is often gradual as well. In other words, there appears to be a smooth transition, rather than a sharp, discrete distinction, between meta­ phorical and metonymie mappings; clear cases of metaphor and metonymy should probably be regarded as points on a continuum of mapping processes (cf. Radden 2000). While the 'same domain' property, then, is often a reliable criterion to distinguish clear cases of metonymy from clear cases of metaphor, it might have to be supplemented with other criteria, such as the type of mapping (see Section 3.3), for less clear cases. 3.3

Mapping, domain highlighting, and mental activation

What does it mean for metonymy to be understood as a 'mapping' relation? And is it the same kind of mapping as metaphor?

METONYMY IN COGNITIVE LINGUISTICS

233

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). It can also be understood, in a narrower sense, as the projection or superimposition of one structure onto another, so that the projected structure imposes (some of) its internal elements, traits, and proper­ ties onto their counterparts in the other. Mapping is facilitated if both structures have a basic degree of structural 'match'. This is what happens in metaphor, in which source and target share at least the most abstract level of their imageschematic structure, so that sources are mapped onto sources, goals onto goals, etc. (Lakoff 1993). In CAREERS ARE JOURNEYS, for instance, the beginning of the journey maps onto that of the career, the obstacles in the journey onto the difficulties in the career, etc. This structural match does not seem to occur 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. The only counterparts in metonymy are normally the domains and the subdomains which are linked by means of this mapping. The metonymically linked (sub)domains do not normally exhibit any degree of structural match. Wholes do not exhibit the same abstract struc­ ture as parts, and when the mapping relates two parts of the same ICM (e.g. a food item and a customer in the 'restaurant ICM'; cf. 3), those parts are seldom, if ever, equivalent functionally or relationally. In view of this asymmetry, then, are we still entitled to regard metonymy as a mapping? While some authors (e.g. Kövecses & Radden 1998; Feyaerts 2000) treat metonymy as a 'relationship' based on conceptual contiguity and thus refrain from regarding metonymy as a mapping at all, I maintain that there is no problem in treating metonymy as a mapping, provided that metonymic mapping is viewed as asymmetrical, whereas that occurring in metaphor is symmetrical. The metonymic source projects its conceptual structure onto that of the target, not by means of a systematic matching of counterparts, but by conceptually foregrounding the source and by backgrounding the target. In (11), the subdomain of Proust's LITERARY WORK (the target) is 'backgrounded' and the whole domain matrix of their author (in which PERSON is a primary subdomain) becomes prominent. (11) Proust is tough to read. A metonymie mapping affects the conceptualization of the target domain, which is understood 'in a new light': the mapping invites viewing Proust's works as an extension of Proust's personality. This re-conceptualization can be relatively transient and limited, as in (11), or it can leave an extensive perma-

234

ANTONIO BARCELONA

nent trace in cognitive domains and in inference patterns. The latter situation is typically encountered in metonymic models (cf. Section 3.5 for an example of reasoning in terms of the 'housewife-mother' model). The different nature of metonymic and metaphorical mapping can be used as a supplementary criterion to distinguish metaphor from metonymy. Some authors, like Cuenca and Hilferty (1999:113), seem to favor this crite­ rion. However, detailed empirical work testing its efficiency with a large number of examples is still needed. At first sight, Croft (1993) does not regard metonymy as a kind of mapping either, as he reserves the term 'domain mapping' for metaphor and characterizes metonymy as involving domain highlighting, which consists in highlighting a secondary (sub)domain10 within the domain matrix constituted by a speaker's encyclopedic knowledge of the meaning of a linguistic expres­ sion. For instance, in (11), the subject NP metonymically refers to Proust's literary output, which, within the domain matrix PROUST, constitutes an impor­ tant subdomain, but one that is secondary to 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 PROUST. Although Croft, in characterizing metaphor and metonymy as domain mapping and domain highlighting, respectively, may be misconstrued as implying that metonymy is not a mapping, he explicitly associates metonymy with mapping: "in metonymy ... the mapping occurs only within a domain matrix" (1993:348). Furthermore, it is my contention that Croft's notion of highlighting inseparably involves mapping. Indeed, Croft's notion of domain highlighting, as I understand it, 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. In my view, this mental activation of the target—an essential ingredient of metonymy—is a different type of activation than the conceptual foregrounding of the source in metonymy: the target is accessed mentally (activated) via the source, which is its 'reference point' (Langacker 1993), but the choice of the source constrains the way in which the target is activated (cf. above). In (11), the literary works are activated mainly as produced by a given author, not in terms of their actual content, which is less prominent in our conceptualization of these works. In other words, the author is mapped onto his works. This is the reason why metonymy is not only a matter of highlighting or activation, but also of mapping. In sum, then, the notions 'mapping' and 'highlighting' should be seen as inseparable rather than as competitive notions in metonymy; in other words, while examples such as (11)

METONYMY IN COGNITIVE LINGUISTICS

235

undeniably involve highlighting of a relatively secondary subdomain, this should not detract from the fundamental fact that metonymy is a mapping. 3.4

The centrality of sub domains

As I pointed out above, Croñ (1993) characterizes metonymy in terms of domain highlighting, which consists in highlighting or mentally activating a secondary (sub)domain, the target, through another (sub)domain. The question then arises, What are the factors determining the ranking of subdomains that make a subdomain primary or secondary in the domain matrix? According to Croft (1993:349-351), there is only one determining factor: extrinsicness. Secondary subdomains, he claims, are more extrinsic than primary subdomains within the relevant domain matrix. Extrinsic subdomains make reference to entities which are external to the domain matrix concept (cf. Langacker 1987:160-161); in contrast, intrinsic subdomains make no reference to such external entities.11 In (11), for instance, Proust's LITERARY WORK is relatively extrinsic in the domain matrix for PROUST, whereas the PERSON subdomain is quite intrinsic. Accordingly, the LITERARY WORK subdomain qualifies as a secondary subdomain that can be highlighted and thus become the target of the metonymy in (11). But domain extrinsicness is not the fundamental factor characterizing secondary subdomains. As Ruiz de Mendoza (2000:127-128) points out, the really important factor, according to Langacker (1987:158-161), is the fact that secondary subdomains tend to be relatively less central (and not simply more extrinsic) than primary subdomains in the conceptualization of a domain matrix; extrinsicness is just one of the factors in centrality. Additional factors making a particular subdomain less central are (i) conventionality, i.e. not being conventionally regarded as part of the source (e.g. LITERARY WORK is conven­ tionally a part of the overall PROUST domain, but less so than the PERSON subdomain); (ii) characteristicness, i.e. not being exclusive to the source (e.g. LITERARY WORK is not exclusively part of PROUST); (iii) genericness, i.e. it not being generic knowledge that the target belongs to the source (e.g. it is generic knowledge that LITERARY WORK is a part of PROUST, but less so than PERSON).12 As we will see in Section 4.1, the fact that a subdomain is not central, hence secondary, in a domain matrix is one of the ways in which it can be neatly distinct from it.

236

3.5

ANTONIO BARCELONA

Metonymy's primarily referential purpose

The chief pragmatic purpose underlying the use of most metonymies for indi­ viduals (i.e. (groups of) individual entities, including not only physical entities such as living beings, but also abstract entities such as institutions or literary works) is to refer to these (groups of) individuals; see examples (l)-(3), (5)(7), and (11). Although metonymy is primarily referential for individuals (cf. also Lakoff 1987:85)—Croft (1993:349) even seems to require every meton­ ymy to be referential—it appears that referentiality is not a necessary property for every metonymy. For one, the referential use of metonymy is not necessarily relevant in conceptual categories, particularly radial categories, built around metonymie prototypes (cf. Lakoff 1987: chap. 5). The HOUSEWIFE-MOTHER FOR MOTHER metonymy, for instance, motivates a cultural model for the whole 'mother' category and can operate in reasoning without involving a referential shift (as reflected, e.g., in Mary is an excellent mother, even though she has a demand­ ing job). Another instance of non-referential metonymy is what Ruiz de Mendoza (2000) calls 'predicative metonymies', as in (12): (12) Oh, your kid is a Pelé. In (12), the speaker indicates that the hearer's child is an excellent soccer player. The NP a Pelé does not refer to an individual, that is, it cannot be a referring expression because this phrase expresses the predicate in the propo­ sition. However, the NP a Pelé is metonymic; in particular, it is an example of what Lakoff (1987:87-88) calls paragons: an individual widely agreed to have been an ideal soccer player is used to conceptualize the whole category of 'exceptional soccer players'.13 Referentiality is not a sufficient criterion for metonymy, either. While it is true that metonymies for individuals are primarily used for reference, it has been observed that metaphorical expressions are often referential as well; see Ruiz de Mendoza (2000:114), who provides the example The pig is waiting for his check, in which the metaphorical expression the pig is used by a waitress to refer to a particularly unpleasant customer. 3.6

Metonymy as a 'stand-for' relationship

Although there is no harm in considering metonymy as a 'stand-for' relation­ ship in informal descriptions of metonymie examples, the 'stand-for' relation­ ship can safely be dropped as a necessary property of metonymy. First, as

METONYMY IN COGNITIVE LINGUISTICS

237

Kövecses and Radden (1998:40) have shown, even though many 'clear' metonymies involve full substitution of the target by the source (as in The kettle is boiling—the kettle substitutes unambiguously for the liquid contained in it), this substitution (i.e. this stand-for connection) is not so obvious in other metonymic examples like The car stopped, "where the object car is contingent on a person in control of the car" (1998:40). Second, the 'stand-for' property can, in my view, be subsumed by the notion of mapping: the metonymic source is 'projected' onto the target, so that it 'takes its place' (see Section 3.3 above). 3.7

The remaining properties

The 'standard' account of metonymy does not systematically treat the factors that favor the conventionalization of metonymy. These factors have been dealt with extensively by Kövecses and Radden (1998); additional insights have also been provided by Taylor (1995). Since the issue of the conventionalization of metonymy is not easy to discuss in isolation from the definition of conven­ tional metonymies and from the definitions of the other types of metonymy (cf. Section 4), I will defer my brief discussion of conventionalization factors to Section 4. Another shortcoming of the standard account is that it fails to highlight the fact that metonymy, just like metaphor, serves the immediate pragmatic purpose of achieving a maximum range of implications at the expense of zero or minimal cognitive effort. Examples of low-cost maximization of implica­ tions in metonymy and metaphor are provided by Ruiz de Mendoza (1997:171176). An example like Wall Street is in a panic (Ruiz de Mendoza 1997:174) automatically brings about the implications that it is the brokers and investors in the New York Stock Exchange that are afraid, and that the cause may have to do with the falling price of their stocks, or any other negative circumstances affecting their activities. Similarly, metaphorical expressions may give rise to various implications with relatively little cognitive effort. An example provided by Ruiz de Mendoza (1997:173) is the metaphorical expression He crashed right after take-off, meant to refer to "the fate of a basketball player's initially promising but eventually short-lived career" due to alcohol abuse. As a manifestation of the metaphorical schema A CAREER IS A JOURNEY, this meta­ phorical expression strongly implies that "the player ruined his career not long after it had started" (1997:173).14 Just like metaphor is unidirectional (cf. Lakoff & Turner 1989:131-133), that is, the mapping follows one direction, from source to target but not simul­ taneously from target to source, I believe that metonymy can be said to be

238

ANTONIO BARCELONA

unidirectional as well. The difference between metaphor and metonymy in this respect is that, unlike metaphor, a large number of metonymies are reversible, as Kövecses and Radden (1998:46) have noticed. That is, we can often find the opposite type of metonymy. See example (13), adapted from Kövecses and Radden (1998:50): (13) a. WHOLE THING FOR A PART OF THE THING: America helped Europe in World War II ('America' for 'United States') b. PART OF A THING FOR THE WHOLE THING: England has had a large empire ('England' for 'Great Britain' or 'the United Kingdom') Reversibility is not the same thing as bidirectionality. The mapping in (13a), whereby a whole continent is mapped onto a political unit, is very different from the mapping in (13b), where a political-geographical unit is mapped onto the whole political-geographical unit in which it is integrated, which means that we are not dealing with two variants of the same metonymy. Furthermore, in neither metonymy do the source and the target simultaneously map onto each other. 3.8

Reassessing the relation between metaphor and metonymy

In discussing the properties and problems of the 'standard' account of meton­ ymy, I have also been able to reassess the differences between metaphor and metonymy. While none of the similarities listed in Section 2.2 is contro­ versial—except that the conventionalization of metonymy has not been studied in any detail in Lakoff and Johnson (1980), Lakoff (1987), or Lakoff and Turner (1989)—none of the differences listed in 2.2 have been found to be fully adequate for distinguishing metaphor from metonymy in a principled way. As I suggested above, the nature of the mapping involved may be a useful property to distinguish metaphor from metonymy. Metaphor and me­ tonymy are two specific manifestations of a general mapping ability (cf. Fauconnier 1997:9-13). In metaphor, the elements of the abstract imageschematic structure of the source are mapped onto their counterparts in the target (Lakoff 1990, 1993). In metonymy, the mapping consists basically in the perspectivization of the target from the source, without a systematic matching of structural counterparts. But this hypothetical difference requires careful further exploration. If found to be correct, it could replace or at least supple­ ment the 'within/across domain' criterion.15 In the meantime, especially in the

METONYMY IN COGNITIVE LINGUISTICS

239

light of the arguments offered in Section 3.3, it seems more realistic to regard metaphor and metonymy as two points on a continuum of mapping processes. 4.

A network of cognitive notions of metonymy

A number of general conclusions can be drawn from the foregoing analysis of the standard account of metonymy. I have claimed that metonymy involves (sub)domains, rather than enti­ ties, and I have demonstrated the ultimate unreliability of the 'single domain' criterion to distinguish metonymy from metaphor, at least in the way this crite­ rion is normally presented. I have lent support to the view of metonymy as a mapping, and attempted to clarify the nature of this mapping. In this respect, I have emphasized the activation or highlighting of the target as characteristic of metonymy, and introduced the criterion of 'subdomain non-centrality' as a factor favoring metonymicity; in this respect as well, I have pointed out that metonymy, as a type of mapping, can easily subsume its standard 'stand-for' property. I have explained why referentiality cannot be a necessary nor a suffi­ cient property of metonymy, and I have attempted to clarify such properties of metonymy as its pragmatic purpose and its directionality. In the light of the above, it has become clear that an absolute distinction between metonymy and metaphor is probably impossible to maintain, and that a 'continuum approach' seems more realistic. The standard view of metonymy has evolved towards a fairly abstract notion of metonymy as a mapping which is not necessarily tied up to a refer­ ential function. This evolution has greatly extended the initial notion of me­ tonymy. However, as we have noted in the preceding sections, the term 'metonymy' is used in Cognitive Linguistics to cover very different phe­ nomena, including 'classical' examples of linguistic referential metonymies for individuals, 'clear' non-referential metonymies, and relatively 'peripheral' instances of metonymy. The present section is an attempt at providing both a broad definition of metonymy capable of capturing what all of these different phenomena have in common, and a set of more specific definitions that will account systematically for each major kind. Each specific definition at the same time indicates how it is linked to the broad definition and to the other specific definitions. 4.1

Prototypical, typical, and schematic metonymies

As I have mentioned earlier (Section 3.4), Ruiz de Mendoza (2000) charac­ terizes metonymy as an intra-domain mapping whose target is a secondary or

240

ANTONIO BARCELONA

non-central domain. Ruiz de Mendoza furthermore holds that this definition is only of relevance in those cases where the mapping proceeds from a domain as a whole to one of its subdomains, that is, in those cases where the target has to be a clearly secondary subdomain of the source.16 This criterion, from my point of view, can be reformulated as requiring that in WHOLEFORPART metonymies the target be sufficiently distinct from the source for the mapping to be easily felt as a semantic shift, i.e. as a shift from the standard meaning of the source linguistic expression. In (13a) above, UNITED STATES is a clearly secondary subdomain in the overall conceptual domain of AMERICA (as a continent). And if (11) is regarded as a WHOLE FOR PART metonymy, LITERARY WORK is secon­ dary in the overall PROUST domain. I would like to argue that the distinctness criterion can be generalized to PART FOR WHOLE metonymies and to PART FOR PART metonymies. In PART FOR WHOLE, the source is necessarily distinct from the target, because the target is

conventionally understood as having a part-whole structure and the source is indisputably understood as a distinct part of it. In (4), for example, (having a) FACE (source) is a primary subdomain in PERSON, but this subdomain is clearly distinct from PERSON, with which it is clearly understood to have a part-whole connection. Metonymies like (4) can, then, be easily felt as semantic shifts. And in PART FOR PART metonymies, the source and the target are not even included as subdomains in each other, for which reason they are regarded as clearly distinct (e.g. the FOOD ITEM is quite distinct from the CUSTOMER, though both are subdomains in the same overall domain or ICM of RESTAURANTS).17 These observations could be summed up as follows: in metonymy, the target is, typically, clearly distinct from the source, either because it is a noncentral, hence secondary, subdomain of the source (WHOLE FOR PART metony­ mies), or because it is not included in the source as a part of it (PART FOR WHOLE or PART FOR PART metonymies). In metaphor, the two domains are also different, but this is due to the fact that they are not experientially 'contiguous', and to the fact that neither can include the other. So the causes for the distinct­ ness of source versus target in metonymy are not the same as in metaphor. The distinctness criterion can be used to distinguish 'typical' and 'proto­ typical' metonymies from 'schematic' metonymies. 'Prototypical' metonymies comprise referential metonymies for indi­ viduals, where the target domain must be a clearly distinct entity from the source for reference to be possible. What justifies regarding these metonymies as prototypical? First, reference is the primary pragmatic purpose of metony­ mies for individuals; in fact, it is on the basis of referential metonymies for individuals that the very notion of metonymy arose in rhetoric and linguistics.

METONYMY IN COGNITIVE LINGUISTICS

241

Second, their referential role is readily perceived by the average speaker. The usual question to be expected from a hearer, 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. The relative distinctness of source and target also helps to distinguish 'typical' from 'peripheral' metonymies (or 'relatively central' metonymies from metonymies in a 'broad sense'), and is also useful as one of the factors that facilitate the conventionalization of a given metonymy. 'Typical' metony­ mies are the ones closest to the basic-level prototype on the metonymicity continuum: like prototypical metonymies, their target must be clearly distin­ guishable from their source. However, typical metonymies are not restricted to metonymies for individuals, nor do they have to be referential. Let us examine examples (14) and (15): (14) This book weighs two kilograms. (15) This book is highly instructive. On the basis of the distinctness criterion, example (14) would not be a typical metonymy, since the verb form weighs activates the PHYSICAL OBJECT subdomain, which is a primary subdomain in the source domain BOOK. Example (15), on the other hand, might be regarded as a typical metonymy on the distinctness criterion: in (15), the SEMANTIC CONTENT subdomain of BOOK is activated, which is relatively more secondary in BOOK than PHYSICAL OBJECT is. Indeed, a book is first and foremost a physical object; no matter how promi­ nently a book's semantic content may figure in language users' conceptu­ alization of books, BOOK is 'profiled' (cf. Langacker 1987) primarily in the domain of physical objects. Books are objects consisting of a number of sheets of paper, typically printed, and bound together in a cover. A typical book, then, can be described as a kind of object, but it is more difficult to say that it is a kind of semantic content.18 As I pointed out in Section 3.4, secondary subdomains tend to be rela­ tively less central than primary subdomains in the conceptualization of a domain matrix. When we apply the criteria for centrality (intrinsicness, conventionality, genericness, and characteristicness), it becomes clear that the subdomain SEMANTIC CONTENT scores lower than the subdomain PHYSICAL OBJECT in terms of its degree of centrality in the domain BOOK. The interpre­ tation of BOOK as a physical object is intrinsic because it does not make essen-

242

ANTONIO BARCELONA

tial reference to external entities or domains—it is conventional and generic knowledge that books are physical objects; however, it is not a unique charac­ teristic of books to be physical objects: many other entities are physical objects. On the other hand, having semantic content is a relatively extrinsic property of books, as it necessarily makes reference to external domains like the MIND, MEANING, or LANGUAGE; it is conventional and generic knowledge that books have semantic content, but less so than that they are physical objects; and having semantic content is not unique to books (cf. linguistic utterances, signboards, films, songs, etc.). Given, then, that SEMANTIC CONTENT is a less central subdomain than PHYSICAL OBJECT in the domain matrix BOOK, SEMANTIC CONTENT is more distinct from BOOK than PHYSICAL OBJECT is; as such, (15) is a more typical metonymy than (14). A more typical metonymy than (15) is (13a) above (AMERICA FOR UNITED STATES): UNITED STATES is an extrinsic subdomain in AMERICA, as it makes reference to external domains (STATES and POLITICAL GEOGRAPHY) whereas AMERICA is profiled primarily in the domains of CONTINENTS and PHYSICAL GEOGRAPHY. The UNITED STATES subdomain is less conventionally and less generically acknowledged as a specification in the domain AMERICA than CONTINENT is. And AMERICA does not uniquely include UNITED STATES as a subdomain (on the other hand, the U.S. includes Hawaii, which is not located on the American continent). Example (15), thus, bears a degree of similarity to (11), (12), or (13). Examples (11), (12), and (13) are clear metonymies in many languages (AUTHOR FOR WORK; PARAGON FOR CATEGORY; CONTINENT FOR COUNTRY; REGION FOR COUNTRY), and in them the metonymie target is clearly distinct

from the metonymie source. It is for this reason that (15), but not (14), is treated as a metonymy by Ruiz de Mendoza (2000:127-128).19 Examples (11), (12), (13), and (15) would, then, be instances of typical metonymies, with (15) displaying the lowest degree of typicality. Of them, however, only (11) and (13) would be prototypical metonymies, since they have an individual entity (or, as in (11), a well-defined group of individuals, namely Proust's works) as target and since they involve a referential shift. Example (12) is neither refer­ ential nor does it have an individual as target; the target in (15) is not an individual either, although the NP this book in it is referential. It appears, then, that only example (14) would be neither a typical nor a prototypical me­ tonymy.20 But is there any sense in which it could be said to be metonymie? Both in (14) and in (15), the source domain (the whole of the domain matrix BOOK) is mapped onto a subdomain in this matrix (PHYSICAL OBJECT in one case, and SEMANTIC CONTENT in the other). In both examples, as a result of the

METONYMY IN COGNITIVE LINGUISTICS

243

mapping, these subdomains are mentally activated via the global domain matrix (the source). As I pointed out earlier, the mental activation of a (sub)domain via mapping from another (sub)domain is an essential property of conceptual metonymy. Therefore, to the extent that they feature both mapping and mental activation, examples such as (14) can also be regarded, at a schematic level, as a conceptual metonymy, though not as a typical or proto­ typical conceptual metonymy. 4.2

Conventional versus non-conventional metonymies

As I stated in Section 4.1, the conventionalization of metonymies is facilitated, first of all, by the distinctness of their metonymic target vis-à-vis their source, that is, by their being 'typical' metonymies. Schematic non-typical metonymies are less likely to become conventional metonymies. But neither typical nor schematic metonymies automatically become conventional metonymies, as a series of additional social factors/requirements are crucially involved (cf. Taylor 1995: 122-123). First, a conventional metonymy must follow one of the 'natural' patterns or types of metonymic relationships (types such as PART FOR WHOLE, WHOLE FOR PART, PRODUCER FOR PRODUCT, PATH FOR GOAL, etc.). For

a long, fairly representative list of natural metonymie patterns, the reader is referred to Kövecses and Radden (1998) ('default' metonymies, in their termi­ nology). Secondly, and more importantly, a convential metonymy must be socially sanctioned. This is, obviously, the most important of these additional requirements. Social sanction, in turn, depends on: a.

b.

the number of general cognitive and communicative principles favoring the conventionalization of 'natural' cases of metonymy that apply in the case in question; the higher the number of these princi­ ples licensing a specific metonymy, the more motivated it will be and the higher the likelihood that it will become conventional (again, the reader is referred to Kövecses and Radden 1998:62-71, where a large number of such principles is listed); the existence of a specific cultural-cognitive principle favoring the conventionalization of that metonymy (Taylor 1995:123), or conversely, the lack of a specific principle blocking the conven­ tionalization.

I will now briefly turn to requirements (a) and (b).

244

ANTONIO BARCELONA

One of the general types of metonymy is AUTHOR FOR WORK, as in (16): (16) I have just bought a Picasso The expression a Picasso highlights the subdomain of Picasso's artistic work, and in this example, it refers to a particular painting by this artist. Kövecses and Radden (1998:71) suggest that, in keeping with requirement (a), metony­ mies such as the one in (16) are motivated by the general cognitive principles HUMAN OVER NON-HUMAN, CONCRETE OVER ABSTRACT, a n d GOOD GESTALT OVER POOR GESTALT.21 A principle that could be added to their list of general principles and that also seems to motivate this metonymy is SALIENT OVER NON-SALIENT, because artists' works constitute a naturally salient subdomain in

the matrix domain of artists. In line with requirement (b), the metonymy in (16) also gets conventionalized because there exists a specific cultural principle whereby works of art are regarded as unique products of the creative genius of well-known artists (Taylor 1995:123). Now, if my sister Mary makes wonderful pineapple pies, the metonymy intended in (17) may be claimed to respond to the general AUTHORFORWORK pattern, and to be motivated by the same general cognitive principles, but it is not socially conventionalized according to requirement (b), except perhaps within my small family circle, provided that pies are regarded as works of art in that circle. (17) We ate an excellent Mary yesterday. Still, (17) constitutes a typical metonymy because the meals mixed or cooked are a clearly distinct subdomain in the COOK domain. However, this typical metonymy is not conventional in society at large. For a waitress, the kind of food consumed by her customers is a naturally relevant domain in her 'restaurant ICM', which explains the conventionality of example (3), in which the food consumed is metonymically used to highlight and refer to the customer. The general communicative principle at work here is RELEVANT OVER IRRELEVANT (requirement (a)), which overrides the cognitive principle HUMAN OVER NON-HUMAN (Kövecses & Radden 1998:70). As far as I can see, no additional specific cognitive-cultural principle contributes to or, conversely, blocks the conventionalization of this metonymy (requirement (b)). It could be argued that prototypicality should be added to the list of requirements for conventionalization. But facts seem to disprove this claim, because a great many typical metonymies that have to be categorized as non-

METONYMY IN COGNITIVE LINGUISTICS

245

prototypical on the basis of the requirements set above are actually quite conventional and productive. Take example (12), which is fully conventional but non-prototypical (it is not a mapping onto an individual but a mapping onto a category, and it is not referential). Or take the non-conventional metonymy in (17), which is both typical and prototypical (it is referential and its target is an individual entity). An important point to note is that any of the three types or degrees of metonymy may become conventionalized; this normally happens in typical and prototypical metonymies, but it can also happen with (purely) schematic metonymies. The targets of BOOK in (14) and (15) have become essential parts of several established separate senses of the lexeme book. It should be noticed that standard dictionaries (e.g. Webster's) often register in separate subentries both of the metonymic senses displayed in (14) and (15), i.e. as just a physical object (as a mere collection of sheets fastened together) or as just a semantic unit (as in 'the books of the Bible', i.e. the various divisions in the same volume; and today we talk of 'electronic books'). This shows that even these subtle metonymic shifts have been conventionalized in the polysemy of book. 4.3

The definitions

4.3.1 Sch em a tic metonymy The definition I propose is based upon Kövecses and Radden's (1998:39) general definition of conceptual metonymy: "Metonymy is a cognitive process in which one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same domain, or ICM" (1998:39). My definition is: A schematic metonymy is a mapping, within one cognitive domain, o f a cognitive (sub)domain, the source, onto another cognitive (sub)domain, the target, so that the target is mentally activated. I prefer to use the term 'source' rather than 'vehicle', because in Cognitive Linguistics it is normally the former term that contrasts with 'target'. (But I admit that this is just a matter of personal preference.) The choice of the term 'mapping' rather than 'process' is more significant, because I believe that mapping is an essential property of metonymy. The use of the notion 'mental activation' instead of 'mental access' is, again, a matter of personal preference. Finally, my choice of the terms 'domain' and 'subdomain' instead of 'concep­ tual entity' was explained in Section 3.1 above.22

246

ANTONIO BARCELONA

4.3.2 Typical metonymy This definition presupposes the previous definition, and runs as follows: A typical metonymy is a schematic metonymy whose target is distinct from the source, either because it is a non-central subdomain. of the source or because it is not included in it. 4.3.3 Prototypical metonymy This definition presupposes the definition of typical metonymy, which in turn presupposes that of schematic metonymy: A prototypical metonymy is a typical metonymy with individuals as targets and as referents. 4.3.4 Conventional metonymy This definition simply presupposes the first definition: A conventional metonymy is a metonymy that responds to a natural type of metonymic relationship and which is socially sanctioned. 4.4

Comments on the definitions

The preceding classification constitutes a basic typology of metonymy in terms of degrees of metonymicity (schematic - typical - prototypical), and in terms of likelihood of conventionalization (from the lowest degree of likelihood, represented by (purely) schematic metonymies, to the highest degree, repre­ sented by fully conventional metonymies). It is a different typology from the traditional three-class typology (cf. note 17), which attempts to capture the type of the relationship between the (sub)domains involved. The four types of metonymy constitute a network in which schematic metonymies are represented as underlying typical metonymies, and the latter as underlying prototypical metonymies. Purely schematic metonymies may some­ times become conventionalized, but it is normally typical and prototypical metonymies that become conventional. And, although a prototypical me­ tonymy normally also conventionalizes, it does not necessarily do so; see (17) as an example of a prototypical metonymy that has failed to achieve social conventionalization. The interconnections between the definitions are shown in Figure 1. The thick solid arrows point towards a type of metonymy which is

METONYMY IN COGNITIVE LINGUISTICS

247

presupposed by the type where the arrow originates. The thin solid arrows simply indicate that both typical and prototypical metonymies tend to become conventional; and the broken arrow signals the low likelihood for purely schematic metonymies to become conventional. Note the central position of prototypical metonymies, the relatively central position of typical and conventional metonymies, and the peripheral position of exclusively schematic metonymies. As can be seen, conventional metonymies are firmly linked to the network.

Figure 1 : The network of metonymic kinds

The first definition (schematic metonymy; cf. 4.3.1) is by far the most interesting one from the cognitive point of view, as it underscores the funda­ mental similarity between examples like (11), which are normally categorized as indisputable instances of metonymy, and examples like (14), whose meto­ nymic status is questionable for many linguists. Schematic metonymies, more­ over, have the fundamental advantage of making us aware of the extent to which intra-domain conceptual mapping and mental activation of the target— which, in my view, characterize every metonymy—are ubiquitous in cognition and language, thus making metonymy particularly useful for the study of cate­ gorization and inference patterns. However, the definition cannot sufficiently explain why native speakers do not as readily perceive (14) as a semantic shift as they do an obvious PART FOR WHOLE example such as (2). Furthermore, the general cognitive nature of this definition has the disadvantage of not providing a principled criterion for

248

ANTONIO BARCELONA

distinguishing examples like (11) from (14) or (15), or from (16) and (17), which despite their schematic similarity, are fairly different in other respects. Therefore we need to supplement this definition with other definitions that will handle these differences. A further disadvantage of this definition, from the practical point of view of linguistic description, is that it would have to regard a very large number of linguistic expressions as metonymie, even those that are hardly ever regarded as metonymie.23 This situation should not be particularly surprising, anyway, given that active zone metonymies (Langacker 1999:62-67) are the rule rather than the exception. An illustration of this sort of problem is represented by (18), an example taken from Croft (1993:362): (18) I won't buy that idea As Croft points out, BUY is metaphorically mapped onto the domain of mental activity, and the domain expressed by the pronoun I, i.e. the whole PERSON domain, undergoes highlighting of its MIND subdomain thanks to the presence of the NP that idea. The use of I in (18) might be argued to qualify as a sche­ matic metonymy if one accepts that, in addition to the activation of MIND (which is the active zone in I), the whole PERSON domain is mapped onto its primary subdomain MIND. In the most general sense of 'mapping' (i.e. as a connection between counterparts), the shift from PERSON to MIND certainly qualifies as a mapping. The source is foregrounded in the mapping, and the target backgrounded, though much less obviously than in (11), where the liter­ ary works are perspectivized from the author.24 The second definition (typical metonymy; cf. 4.3.2) is less revealing from the cognitive point of view, but it offers a criterion to distinguish typical metonymies, i.e. those in which the semantic shift is relatively obvious to native speakers, from less typical or 'peripheral' ones (i.e. purely schematic metonymies), in which the shift is very subtle. WHOLE FOR PART typical metonymies, in particular, are apt to display varying degrees of typicality. The criterion of centrality, which is a major criterion in determining the primary or secondary status of the target of these metonymies, and as such its distinctness from the source, is itself a conjunction of criteria (intrinsicness, genericness, etc.). In other words, in order to determine the target's distinctness from the source in these metonymies, a set of criteria is available which allow the target to be a subdomain in the source domain matrix to various degrees, depending on the target's relative score on the criteria intrinsicness, conventionality, genericness, and characteristicness. These criteria, then, can help us decide the

METONYMY IN COGNITIVE LINGUISTICS

249

place of a given WHOLE FOR PART metonymy on the continuum leading from prototypical to purely schematic metonymies. The third definition (prototypical metonymy; cf. 4.3.3) is also less reveal­ ing from the cognitive point of view, but it offers a criterion to distinguish 'typical' metonymies from the type of mappings which are most likely to be perceived as metonymic, that is, mappings used in an act of reference whose target is an individual entity clearly distinguishable from the source. Prototypical metonymies are thus more likely to be noticed as semantic shifts or mappings than merely typical metonymies, and the latter are in turn more likely to be noticed as semantic shifts than (purely) schematic metony­ mies, both by linguists and by speakers. We may then say that we have a scale or continuum of metonymicity, with prototypical metonymies displaying the highest degree of metonymicity at one end, and purely schematic metonymies displaying the lowest degree at the other end. The fourth definition (conven­ tional metonymy; cf. 4.3.4) is particularly useful for deciding whether a metonymy is conventional, and to identify the factors determining its conven­ tionalization. The fact that a metonymy has become conventional is often reflected in grammar, polysemy, reasoning patterns, or non-linguistic social symbolism.25 5.

Closing remarks

This article started out with an analysis of the key properties characterizing the 'standard' account of metonymy in Cognitive Linguistics. While some of these properties are relatively uncontroversial, a number of them are problematic. These controversial properties were then discussed at length. Two of the major claims that were made are the following: (i) every metonymy is an intradomain mapping that involves two (sub)domains and results in the activation of one of them; (ii) an absolute distinction between metaphor and metonymy is very difficult to maintain in some cases, and viewing both phenomena as points on a gradient or continuum may be more realistic. The claims made in the analysis of the properties of metonymy have led to proposing four basic kinds of metonymy constituting a closely-knit network. The definitions of these metonymies can serve different purposes. A cognitive scientist would tend to be interested in exploring the connections between schematic metonymy and other fundamental cognitive processes and abilities such as reference-point ability, figure/ground organization, etc., which are pervasive in perception, cognition, and language. For the scholar more directly concerned with the study of the actual description, classification, and evolution

250

ANTONIO BARCELONA

of metonymic expressions (i.e. symbolic expressions of metonymy, linguistic or non-linguistic), the distinction between typical, prototypical, and conven­ tional metonymies may prove quite helpful. Notes 1 This paper has greatly benefited from the insightful remarks and suggestions of two anony­ mous reviewers, and from the advice of the editors, Hubert Cuyckens in particular. I am grate­ ful to them for their careful, professional work. The research reported in this article is part of a research project funded by the Spanish Government, Secretaría de Estado de Universidades, Investigación y Desarrollo (State Secretary for Universities, Research and Development), Project no. PB-98-0375. 2 Kövecses and Radden (1998) is an excellent attempt at setting up a systematic typology of metonymy. The classification of the metonymies illustrated in this article has been done on the basis of that typology. 3 Kövecses and Radden (1998:54) contend that examples like (7) illustrate a different me­ tonymy, CATEGORY FOR DEFINING PROPERTY, whereby "well-known individuals ... are metonymically reclassified as a class on the basis of their defining, stereotypical property". In my view, CATEGORY FOR DEFINING PROPERTY as well as STEREOTYPICAL MEMBER FOR THE

CATEGORY could apply here, depending on the point of view adopted. CATEGORY FOR DEFINING PROPERTY highlights a salient, stereotypical property in Shakespeare—his immense literary talent—and this property is the basis for the figurative reclassification. Alternatively, we could argue that Shakespeare in the English speaking world, Cervantes in the Hispanic world, or Dante in Italian culture are each stereotypical members of a class of uncommonly talented writers. From this point of view, example (7) would then be an instantiation of the STEREOTYPICAL MEMBER FOR THE CATEGORY m e t o n y m y .

4 Kövecses and Radden (1998) offer numerous examples of metonymy in which the entities connected are well-defined abstract 'entities' of an ICM or frame: FORM FOR CONCEPT, ACTION FOR AGENT, CONTROLLER FOR CONTROLLED, etc. 5 When describing social stereotypes as cases of metonymy, "where a subcategory ... [stands] for the category as a whole", Lakoff (1987:79) also seems to suggest that the elements linked by metonymy are two domains rather than two entities. 6 This does not mean that in these, as in a great many other cases, metaphor is not ultimately based upon metonymy. In (10), the metaphorical readings actually arise through generalization of the EFFECT FOR CAUSE metonymy. I have argued elsewhere that the metonymic motivation of metaphor is in fact the rule rather than the exception (see Barcelona 2000a, n.d.); see also Radden (2000) for a similar view. But the fact that metaphor is often based on metonymy does not mean that it can be totally reduced to metonymy. 7 The preceding remarks are fully consonant with Langacker's notion of 'unit status' (1987:161-166), which would offer a different way of formulating a solution to this puzzle. If we consider that the notion of verticality presupposed by the notion of a drooping bodily posture is not automatically activated whenever we use a conventional non-metaphorical and non-metonymic linguistic expression of the domain of SADNESS, this means that the notion of verticality has not reached 'unit status' within the domain of SADNESS; in other words, that it is not an entrenched subdomain in it. This amounts, in fact, to saying that VERTICALITY is not

METONYMY IN COGNITIVE LINGUISTICS

251

automatically regarded by the conscious folk model of the taxonomy of domains as a subdomain of SADNESS. 8 I owe the example to one of my anonymous reviewers. 9 Dirven (1993) believes that maximal conceptual distance between two domains (which is a prerequisite for metaphor) is, in fact, a matter of cultural construal. If two domains are regarded as separate, then metaphor obtains. If the same two domains are regarded as forming a continuum, then metonymy would arise. 10 The technical terms 'secondary' (or, rather, 'subsidiary') and 'primary' domain are borrowed by Croft from Langacker (1987:165, 222). 11 "Shape, for example, is a highly intrinsic property of physical objects, as it reduces to rela­ tions 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). 12 Of course, characterizing Proust's literary activity as a secondary subdomain 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, charac­ teristically, and generically regarded as central a subdomain in PROUST as the fact that Proust was a PERSON, which is a primary subdomain in our conceptualization of any human being. 13 Note that not all predicative metonymies involve a paragon; cf. (4) She is just a pretty face. Moreover, metonymized paragons can also be used referentially, as in (7). 14 This example actually suggests an even wider range of implications. One of them is that, just as much more energy is consumed during take-off than during the flight, the beginnings of a career as a professional player are very hard and require much time, effort, and sacrifice. Another implication is that the player's efforts could have been highly rewarding but they turn out to be useless, since, just as the doomed flight comes to a sudden, unexpected end, the player's activities come to an abrupt end. 15 A refinement of the 'within/across domain' criterion that seems to work in most cases consists of noting that metonymy links two domains included by the same overall 'functional' domain, that is, by the same frame, while the two domains linked by metaphor are not included in the same frame. Furthermore, in metonymy the source has a role which is linked to the target role by a "pragmatic function" (Fauconnier 1997:11); none of this occurs in metaphor. Frames can connect domains quite flexibly. The White House is a building (with the role of LOCATION) in the same frame as the U.S. government (with the role of LOCATUM); thus, we can have metonymies like The White House did not intervene; locations are pragmatically linked to locata. But buildings and people are normally regarded as being in different frames, which licenses the metaphorical mapping of a building onto a person as in Romeo and Juliet (III.3.26), where Juliet is said to be 'a mansion of a love'. Since people and animals are also normally in different frames, John is a lion is metaphorical and not metonymic. For details, see Barcelona (2002a:236-239), which was written after the present paper, but appeared before. Dirven (1993) makes a similar proposal. 16 "In metonymy, if we have a case where the target is a subdomain of the source, the target must be a secondary or non-central domain; on the other hand, if we have a source-in-target metonymy, it is immaterial whether the source is deemed to be a primary or a secondary domain" (Ruiz de Mendoza: 2000:129-130). 17 In this paper, I employ the traditional typology of metonymy: WHOLE FOR PART, PART FOR WHOLE, and PART FOR PART. This is also the general typology adopted in Kövecses and

252

ANTONIO BARCELONA

Radden (1998), a rigorous attempt at laying the foundation for a systematic cognitive theory of metonymy. Ruiz de Mendoza (2000), however, suggests that a general typology should only include WHOLE FOR PART or PART FOR WHOLE metonymies. His two-class typology is quite intriguing and has the advantage of simplifying the range of metonymic relationships, but it is not free of problems. I therefore feel that, at this point, it is safer to maintain the three-class typology, despite its imperfections 18 At the same time, it should be admitted that SEMANTIC CONTENT is a very prominent secon­ dary subdomain in BOOK; so much so that Croft (1993:349) considers it a primary subdomain, together with PHYSICAL OBJECT. 19 It would be difficult to decide the exact metonymic type of (15). The notion of 'active zone/profile discrepancy' might be useful in this respect. It seems that the 'active zone' of 'book' in this example is SEMANTIC CONTENT. Kövecses and Radden (1998:50) treat these cases of active zone involvement as WHOLE FOR PART metonymies, as the semantic content of a book can be regarded as one of its 'parts', together with its physical components. 20 Croft (1993:349), who comments on two similar examples, does not treat (14) and (15) as manifesting two different kinds of metonymy, because he simply operates with the criterion of intrinsicness; as such, 'being a physical object' and 'having semantic content' are both intrinsic properties of a book. In addition, in keeping with his referential view of metonymy, he main­ tains that it "is not clear that there are in fact two different entities being referred to" in (14) and (15). As a result, he denies metonymie status to both of them. 21 Kövecses and Radden (1998:67) argue that human beings are particularly good gestalts and tend to be used as 'vehicles' (sources) for their active zones. Here, PICASSO is used as a gestalt source to activate the subdomain constituted by his works of art. 22 Note that two seemingly important properties of metonymy mentioned above (the pragmatic function linking source and target into one functional domain or frame mentioned in an earlier note, and the asymmetric nature of the mapping mentioned in Section 3.3) are not included in the definition. The first property is implicit in the ability of the source to activate the target. which is a consequence of both being pragmatically linked in a frame; it is incorporated into the definition of metonymy in Barcelona (2002a:246). The second property still has to be tested carefully. 23 In fact, every sign, including linguistic expressions, can be said to be metonymie, in that it stands for the global 'concept' it expresses (Kövecses & Radden 1998:42). But the 'disad­ vantage' I am trying to point out is that the notion of schematic metonymy would force us to recognize that, additionally, a metonymie mapping highlighting a conceptual (sub)domain within the global 'concept' tends to take place in the interpretation of almost every linguistic expression in context. 24 The preceding example raises an important issue. The source in (18) is also the reference point for its active zone. As Langacker (1993) emphasizes, metonymy is a kind of 'referencepoint construction', which involves a connection between two elements, a 'reference point' and a 'target' (which is often an active zone). A natural question to be asked is how these constructions can be systematically distinguished from metonymy. 25 Barcelona (1998, 2002b) are two brief surveys of the range of linguistic and cultural areas in which metonymy is clearly relevant.

METONYMY IN COGNITIVE LINGUISTICS

253

References Barcelona, Antonio. 1986. "On the Concept of Depression in American English: A Cognitive Approach". Revista Canaria de Estudios Ingleses 12.7-35. . 1997. "Clarifying and Applying the Notions of Metaphor and Metonymy within Cognitive Linguistics". Atlantis 19.21-48. . 1998. "El poder de la metonimia". Estudios de lingüística cognitiva ed. by José Luis Cifuentes, 365-380. Alicante: Departamento de Filología Española, Lingüística General y Teoría de la Literatura, Univer­ sidad de Alicante, Spain. . 1999. "Notas sobre la teoría cognitiva de la metonimia y su poder explicativo". Perspectivas aplicadas de la organización del cono­ cimiento, procesamiento y uso del lenguaje ed. by Franciso Ruiz de Mendoza, 913-938. Logroño: Universidad de La Rioja, Spain. . 2000a. "On the Plausibility of Claiming a Metonymic Motivation for Conceptual Metaphor". Barcelona 2000b. 31-58. , ed. 2000b. Metaphor and Metonymy at the Crossroads: A Cognitive Perspective. Berlin: Mouton de Gruyter. . 2002a. "Clarifying and Applying the Notions of Metaphor and Metonymy within Cognitive Linguistics: An Update". Metaphor and Metonymy in Comparison and Contrast ed. by René Dirven & Ralf Pörings, 207-277. Berlin: Mouton de Gruyter. ----------. 2002b. "On the Ubiquity and Multiple-Level Operation of Metonymy". Cognitive Linguistics Today (Lodz Studies in Language) ed. by Barbara Lewandowska-Tomaszczyk & Kamila Turewicz, 207-224. Frankfurt a.M.: Peter Lang. . n.d. "Types of Argument for the Metonymic Motivation of Concep­ tual Metaphor". Unpublished manuscript, Department of English, University of Murcia, Spain. Bultinck, Bert. 1998. "Metaphors We Die By: Conceptualizations of Death in English and their Implications for the Theory of Metaphor". APIL Paper no. 94, University of Antwerp, Department of Linguistics. Croft, William. 1993. "The Role of Domains in the Interpretation of Metaphors and Metonymies". Cognitive Linguistics 4.335-371. Cuenca, Maria Josep & Joseph Hilferty. 1999. Introducción a la lingüística cognitiva. Barcelona: Ariel.

254

ANTONIO BARCELONA

Dirven, René. 1993. "Metaphor and Metonymy: Different Mental Strategies of Conceptualisation". Leuvense Bijdragen 82.1-28. Fauconnier, Gilles. 1997. Mappings in Thought and Language, Cambridge: Cambridge University Press. Feyaerts, Kurt. 2000. "Refining the Inheritance Hypothesis: Interaction between Metaphoric and Metonymic Hierarchies". Barcelona 2000b. 5978. Gibbs, Raymond W., Jr. 1994. The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Kövecses, Zoltán. 1990. Emotion Concepts. New York: Springer Verlag. Kövecses, Zoltán & Günter Radden. 1998. "Metonymy: Developing a Cogni­ tive Linguistic View". Cognitive Linguistics 9.37-77. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago, I11.: The University of Chicago Press. . 1990. "The Invariance Hypothesis: Is Abstract Reason Based on Image-Schemas?" Cognitive Linguistics 1.39-74. . 1993. "The Contemporary Theory of Metaphor". Metaphor and Thought, 2nd edition ed. by Andrew Ortony, 202-251. Cambridge: Cambridge University Press. & Mark Johnson. 1980. Metaphors We Live By. Chicago, I11.: The University of Chicago Press. --------- & Mark Turner. 1989. More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago, I11 .: The University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. 1, Theoretical Prerequisites. Stanford, Calif: Stanford University Press. . 1993. "Reference-point constructions". Cognitive Linguistics 4.1-38. —. 1999. Grammar and conceptualization. Berlin: Mouton de Gruyter. Panther, Klaus-Uwe & Linda Thornburg. 1999. "The POTENTIALITY FOR ACTUALITY Metonymy in English and Hungarian". Metonymy in Cogni­ tion and Language ed. by Günter Radden & Klaus-Uwe Panther, 333— 357. Amsterdam: John Benjamins,. . 2000. "The EFFECT FOR CAUSE Metonymy in English Grammar". Barcelona 2000b. 215-231. Radden, Günter. 2000b. "How Metonymie are Metaphors?". Barcelona 2000b. 93-108. Ruiz de Mendoza, Francisco. 1997. "Cognitive and Pragmatic Aspects of Metonymy". Cognitive Linguistics in the Study of the English language

METONYMY IN COGNITIVE LINGUISTICS

255

and Literature in English (= Monograph issue of Cuadernos de Filología Inglesa) ed. by Antonio Barcelona, 161-178. Murcia: University of Murcia, Spain. ---------.2000. "The Role of Mappings and Domains in Understanding Metonymy". Barcelona 2000b. 109-132. Taylor, John. 1995. Linguistic Categorization: Prototypes in Linguistic Theory, 2nd edition. Oxford: Clarendon Press. Webster's New Twentieth Century Dictionary of the English Language, 2nd edition. 1978. Collins-World.

SECTION 3 MOTIVATION IN SOCIO-CULTURAL CONCEPTUALIZATIONS

The first two papers in this section look at motivating factors in the conceptual content or meaning of linguistic items that can be situated in the socio-cultural domain; the third paper illustrates how several aspects of one particular sociocultural domain (the economic domain) can be metaphorically understood in terms of the domain of motion. Yoshihiko Ikegami explains that the concep­ tualization of 'language' may vary with the language (and, more generally, the cultural model) in which the item is used; as such, 'language' is conceptualized as an object, bounded and stable, in English, and as an event, unbounded and in permanent flux, in Japanese. What underlies, or motivates, these dissimilar conceptualizations, Ikegami argues, is the CONTAINER or CONDUIT metaphor of language in English and the LANGUAGE IS A FLUID METAPHOR in Japanese. Karol Janicki claims that speakers' understanding and use of language may be conditioned by their set of beliefs about the world. He illustrates this claim with a discussion of two heated debates on socio-political issues in Poland— one on abortion, the other on Christian values and censorhip—which stem from the conflicting parties' essentialist view on word meaning and definitions. As such, it is made clear that essentialist understanding and use of language is conducive to conflict. Olaf Jäkel's study amply illustrates how the Cognitive Theory of Metaphor underlies speakers' understanding of (aspects of) the domain of the economy in terms of the domain of spatial motion. Based on linguistic evidence from German and English, Jäkel shows how metaphorical mappings can account for the polysemy in lexical items such as move ahead, outpace, drag behind, etc.

HOW LANGUAGE IS CONCEPTUALIZED AND METAPHORIZED IN JAPANESE AN ESSAY IN LANGUGE IDEOLOGY

YOSHIHIKO IKEGAMI Showa Women's University, Tokyo

This paper discusses the traditional conceptualization of language in Japan. The discussion covers such points as the historical development of the terms used to refer to 'saying, speech, language' in Japanese, some characteristic features of the traditional language-ideological view of (i.e. the common-sense notions about the nature of) the Japanese language entertained by its speakers, and the way language is metaphorically referred to in Japanese. It will be shown that language is not so much conceived of as an object but as the speaking subject's activity and that it is predominantly metaphorized as a fluid—reflecting the underlying idea that what the speaker has in mind/heart (without being 'poured' into some kind of receptacle) transforms itself into language. The picture which emerges will serve to relativize the one offered by Reddy (1993) for English, suggesting that it may be useful to take a typological perspective into account.

1.

Introduction

One of the current common practices in cognitive semantics is the study of conceptualization, especially in terms of metaphorization, as instigated by Lakoff and Johnson (1980). The conceptualization of space and emotions has, among others, been a popular field of investigation (cf. Herskovits 1986; Lakoff 1987:380-415; Kövecses 1988; Pütz & Dirven 1996; Niemeier & Dirven 1997). The present essay is intended as a contribution broadly in the same line, but it is targeted at a particular notional field, namely 'language', a topic in which my esteemed colleague Günter Radden is highly interested. Still, I would like this essay to be more than simply a study of conceptualization and consider also issues of 'language ideology'. By 'language ideology' here is meant "[a set] of beliefs about language articulated by users as a rationalisation

260

YOSHIHIKO IKEGAMI

or justification of perceived language structure and use", as the term was origi­ nally defined by Silverstein (1979:193). This notion has since undergone further ramifications (cf. Woolard & Schieffelin 1994; Schieffelin, Woolard & Kroskrity 1998). I would, however, like to adhere to a fairly neutral linguistic definition of 'language ideology', such as Silverstein's definition quoted above or such as the following, less technical one: "shared bodies of common-sense notions about the nature of language in the world" (Rumsey 1990; quoted in Woolard & Schieffelin 1994:57). My discussion is thus differently oriented from the kind of discussion on 'language ideology' which centers around the question of 'authority' and 'power' in language, as found, for example, in Lippi-Green(1994). By relating the question of conceptualization to language-ideological issues, I am hoping to place it in a yet wider perspective, going beyond the linguistic description of metaphorical transfer and further into the speaker's cognitive processes which motivate the transfer. 2.

The historical developm en t of words for 'language'

In modern Japanese, there are two words commonly used to represent the notion of 'language/speech': kotoba and gengo. Both words are recorded in the most ancient writings in Japanese of the eighth century. Their etymological derivation, however, is different and they are used in different registers. Kotoba, composed of koto "saying, speech" and ba (from ha) "leaf (of a tree)" belongs to the native language stock and has generally been part of the common vocabulary to this day. Gengo, composed of gen "saying, speech" and go "talk(ing), word", by contrast, is a borrowing from Chinese.1 The word gengo, unlike the word kotoba, is—and has always been— a formal word, and it sometimes has a technical connotation. Thus, linguistics is rendered as gengogaku in Japanese, where the second part, gaku means 'science, -ology'. Semantically, both kotoba and gengo have apparently undergone a parallel development. In their earliest recorded usage, both words refer predominantly to 'an act of saying, speaking or expressing oneself. Next comes the stage where they refer to 'a product of saying/speaking, a linguistic expression', followed later by the stage where they refer to a particular language, a 'langue' (Kindaichi 1944:7-9). Although all three senses occur in contemporary Japa­ nese, the word kotoba is primarily associated with the second meaning above, while the formal word gengo is predominantly used in the third sense. For our present purposes, kotoba, the word of native stock, has far more interesting data to offer and we will be concentrating on this word in the following sections.

THE CONCEPTUALIZATION OF LANGUAGE IN JAPANESE

261

2.1 The word kotoba, as I pointed out, consists of koto "saying, speech" and ba (from ha) "leaf (of a tree)". This is the way the contemporary speaker of Japa­ nese feels about the word. Etymologically, however, many more things are involved in the meaning of the word, thus justifying a fuller account. The first part of the word kotoba, namely koto, is itself a form that can be used as an independent word. In ancient Japanese (eighth century), this form had two related meanings: 'what happens (in time), event' and 'what one says, speech'. While koto's polysemy may at first sight appear to be puzzling to the modern mind, it is in fact a familiar one commonly observed in the early stages of human culture. It is based on the familiar belief in the 'magical' power of language, namely what one utters in words will be realized in time. A wellknown instantiation of this belief is that if one knows the name of a person or thing, one is in control of the person or the thing referred to by that name. The same belief is also involved in certain ritual uses of language, either benedictive or maledictive, through which either benefit or harm is supposed to be brought on the person for whom the words are intended. Concrete instances of this sort are fully attested in the ancient Japanese writings (Ikegami 1998). What one actually finds with the word koto in ancient Japanese is not so much polysemy in a strict sense as one very general, vague meaning in which the notions of 'saying' and 'happening' are merged with each other. This is attested by the way in which the native word koto was transcribed in terms of imported Chinese characters. Chinese had different characters for 'saying' and 'happening'. In the documents of the period, both characters are used to tran­ scribe the native word koto. It is conceivable that the transcribers tried to choose between the two Chinese characters depending on whether the intended sense was contextually 'saying' or 'happening'. Actually, the choice between the two characters was not as consistently made as might be imagined. An investigation (reported in Toyoda 1990) reveals that the word koto used in the sense of 'saying, speech' is transcribed in terms of the Chinese character for 'happening, event' just as frequently as in terms of the Chinese character for 'say(ing), speech', while the word koto used in the sense of 'happening, event' is transcribed overwhelmingly in terms of the Chinese character for 'hap­ pening, event' and rarely in terms of the Chinese character for 'saying, speech'. The statistics seems to show, if anything, that the primary meaning of the word koto was 'happening, event' rather than 'saying, speech'. The latter meaning was peripheral and this situation interestingly foreshadows the later morpho­ logical and semantic development of the word.

262

YOSHIHIKO IKEGAMI

2.2 The ensuing centuries saw both the morphological and the semantic differ­ entiation of the word, as the contrast between koto "saying, speech" and koto "happening, event" came to make itself felt. The transcription of the word in terms of the two different Chinese characters became ever more consistent with the intended meaning—a sure indication that language users were now aware of two homonymous words rather than one polysemous word or a word with a very general, vague meaning. In addition, there was a growing tendency to distinguish between the two meanings, or rather two homonymous words, in terms of different word forms. Thus the word koto for 'happening, event' retained its original form, while the word koto for 'saying, speech' was given a somewhat more elaborated form, namely kotonoha or kotoba. The form kotonoha consisted of the word koto, followed by the particle no "of, -'s" and the noun ha "edge, tip, end"; so its literal meaning was 'koto's edge'. The other form, kotoba was a compression of the form kotonoha into a compound: the word koto followed by the noun ba, a phonetic variant of the noun ha "edge, tip, end". Two views can be offered to account for the meaning of koto in kotonoha "koto's edge" and kotoba "koto-edge". According to the traditional view, the meaning of the word koto had already been specialized to 'saying, speech' when the word koto was integrated into the phrasal word kotonoha and the compound word kotoba. Either of the resulting words is therefore presumed to have meant 'the edge of saying/speech', by which it was implied that uttered words represented only a tiny (and spurious) portion of what was to be said, or that, like the 'tip' of an iceberg, uttered words did not convey the whole story. Thus the two words, which originally connoted somewhat pejoratively the essential inadequacy of language, were to serve the purpose of specifically encoding the meaning 'saying, speech, language' in order to resolve the homo­ nymic clash between koto "happening, event" and koto "saying, speech". Yet another possible view to account for the meaning of koto in kotonoha "koto's edge" and kotoba "koto-edge" is to assume (contrary to the traditional view) that the word koto still retained its original semantic ambiguity between 'happening, event' and 'saying, speech' and that the word was still understood to mean 'happening, event' (rather than 'saying, speech') when it became incorporated into the phrasal word kotonoha and the compound word kotoba. Either word is therefore presumed to have meant 'the edge of a(n) hap­ pening/event'. Notice that this view tallies rather nicely with the native belief already mentioned, namely that 'saying something' would lead to the reali­ zation of what had been said. As such, 'saying, speech' constituted the edge of an event in the sense that it caused an event to happen. Pending further philol-

THE CONCEPTUALIZATION OF LANGUAGE IN JAPANESE

263

ogical investigation, it is not possible to definitely choose between the two alternative views. It may also be suggested that the two views are not really exclusive of each other. It may very well have been the case that kotoba's/kotonoha's initial meaning 'the edge of a(n) happening/event' was later reinterpreted as 'the edge of saying/speech'. 2.3 The newly emerged synonymous words for 'saying, speech', kotonoha and kotoba, soon started to diverge stylistically. The use of the phrasal word kotonoha came to be restricted to the literary, elevated register. And it was probably at this stage that a further modification was introduced. The word ha "edge" in the phrasal word kotonoha "koto's edge" became associated with the homonymous word ha "leaf (of a tree)" and was reinterpreted as such. Kotonoha "koto's edge" was now understood as meaning 'koto's leaf or more precisely, 'speech's leaf/leaves', since by this time the word koto in kotonoha came to be interpreted exclusively as referring to 'speech, language'.2 The reinterpretation from 'edge' to 'leaf is generally considered to have been trig­ gered by Chinese phraseology, in which an anthology of poems was compared to a wood of trees. Similarly, ba in the compound word kotoba, a variant of ha "edge", came to be reinterpreted as a variant of the homonymous word ha "leaf. The character for 'edge' in either word was replaced by the character for 'leaf. In this way, the two words for 'speech, language' arose in contem­ porary Japanese: kotoba (literally, "speech-leaf), used in daily speech, and kotonoha (literally, "speech's leaf), its formal, literary counterpart. 3.

Characteristics of the Japanese speaker's view on language

The conceptualization of the notion 'language' can also be characterized by referring to the term with which the term for language is preferentially both associated and contrasted. 3.1 It seems to me that the mainstream of the Western interest in language has been dominated by the question of reference, whereby the primary function of language is that it 'refers to' something. Depending on the linguistic level at which the discussion is held, the question of reference can be formulated as the relationship between the word and the object it applies to or as the relationship between the sentence and the situation it encodes. In either case, the basic idea is that language is opposed to something extralinguistic—some entity in the 'real' world. From this orientation has sprung a natural interest in the possible match or mismatch between the word and the object it refers to or between the sentence and the situation it encodes. Not unexpectedly, mismatching has been

264

YOSHIHIKO IKEGAMI

found to predominate by far over matching. Language, then, has been branded as 'imperfect'—an imperfect instrument for unmistakable reference and hence for communication. As is well known, this has caused some scholars to start a painstaking search for the 'perfect language' (Eco 1995), often motivated by missionary zeal. As Lyons writes in his article on subjectivity in language, Western preoccupation with language seems to have been largely "dominated by the intellectual prejudice that language is essentially, if not solely, an instrument for the expression of propositional thought" (Lyons 1982:103). It will not be necessary to point out that this concern with propositional thought goes hand in hand with the interest in correct reference and the search for the perfect language. 3.2 The line of thinking which has dominated the Western tradition is rather different from the native conceptualization of language in Japan, where the general orientation is much more concerned with the speaking subject. If, in the Western tradition, the basic contrast is between 'language' and 'referent' (either an object or a situation) and the relationship between the two terms is characterized as the former 'referring to' the latter, the basic contrast in Japa­ nese tradition is between 'language' and 'heart/mind' (i.e. something which the speaking subject has to say) and the relationship between the two terms is characterized as the former 'expressing' (i.e. externalizing) the latter.3 One important point that derives from the Japanese conceptualization of 'language' in relation to 'heart/mind' is that these two notions are, unlike the corre­ sponding two notions 'language' and 'referent' in Western conceptualization, not so much contrasted with each other as they are closely integrated. There is, in fact, an essential continuity between the two. They differ only in that one ('mind/heart') is something internalized (and hence not directly perceivable) and the other ('language') is something externalized (and hence perceivable): the former is 'expressed' (in the etymological sense of the word) as the latter. A Japanese way of capturing this process is to say that the former ('mind/ heart') 'spontaneously becomes' the latter ('speech/language'). The Japanese speaker, like the Western philosopher, is fully aware of the inadequacy of the language he uses, but reacts differently. Unlike the Western philosopher, the Japanese speaker does not address him-/herself to rectifying the imperfect language. Instead, he/she is encouraged to rectify his/her mind/heart. The idea is that proper language derives from a proper mind/heart; if the mind/heart is right, language will follow suit. The latter is no more than the outward mani­ festation of the former. Quite characteristically, rhetoric was not a popular

THE CONCEPTUALIZATION OF LANGUAGE IN JAPANESE

265

theme in Japanese tradition; the emphasis was always on spiritual discipline— what the proper disposition of one's mind/heart should be. It is also to be noticed that in a conceptualization of language as described above, language is viewed not as a tool that the speaker has at his/her disposal and manipulates, but as something that happens, i.e. as an event. The potential ambiguity of the word koto in ancient Japanese, which was discussed in the previous section, may be recalled here: koto could mean either 'saying, speech' or 'happening, event' (or perhaps still better, both at the same time). Language is an event in which the speaker is directly involved and not some­ thing to be hypostatized as an object apart from the speaking subject (Tokieda 1940:17-21). Another interesting corollary that derives from conceptualizing language as coterminous with the mind/heart rather than as contrasting with (extralinguistic) objects and situations is a conceptualization of language according to which the primary function of language is for the speaker to express him/herself properly rather than for the hearer to be properly communicated with. This may sound like too egocentric a view of language (and for that reason, rather unlikely). But if one takes into account the obvious fact that language serves 'expressive' (i.e. 'speaker-oriented') purposes as well as 'commu­ nicative' (i.e. 'hearer-oriented') purposes, the 'egocentric' view is just as acceptable an alternative from a language ideological point of view (with language ideology to be taken in the sense of 'common-sense notions about the nature of language') as the communication-oriented one. It can easily be imagined that if the communicative purposes are to be fulfilled through an egocentrically oriented use of language, this will require active participation on the part of the hearer; in other words, the burden of successful communication will be on the hearer as much as, and sometimes even more than, on the speaker. Hinds (1987), in his contribution to a volume on contrastive rhetoric, proposes a typology of the modes of communication in terms of the contrast between 'speaker-/writer-responsibility' and 'listener-/reader-responsibility', and offers the following characterization: "In some languages, such as in English, the person primarily responsible for effective communication is the speaker, while in other languages, such as Japanese, the person primarily responsible for effective communication is the listener" (Hinds 1987:43). For those who have some knowledge of Japanese, it may be added that the speaker-centered orien­ tation of the language, complemented by the hearer's active cooperation, has much to do with the remarkably high degree of ellipsis commonly tolerated in Japanese discourse (Ikegami 1997). The pragmatic principle allowing the speaker to leave unmentioned those points which can be fully recovered by the

266

YOSHIHIKO IKEGAMI

hearer will universally apply in all linguistic communities. The extent of ellip­ sis in Japanese discourse, however, goes far beyond what such a pragmatic principle (which, incidentally, is based on the assumption that the primary function of language is communication and is hence 'hearer-/addresseeoriented') can account for. We need not be concerned here with the question of which is the cause and which is the effect. Suffice it to say that there is a moti­ vated relationship involved here, in that the speaker adapts his behavior to the way his/her language functions and that the way the speaker's language func­ tions causes him/her to modify his/her behavior accordingly. 4.

The metaphorization of language in Japanese

In Section 3, I have pointed that, among other things, language in the Japanese tradition is conceptualized not as a tool that is waiting to be used by the speaker, but as something that arises spontaneously from the speaker's mind/heart. It may well be expected that such a conceptualization is reflected in the way language is metaphorized. 4.1 One very common metaphorization of language that immediately comes to mind is the metaphorical transfer whereby the word for 'tongue' (as organ of speech) also comes to include the sense 'language' (see Radden 2001). In this respect, Ullmann (1962:226) writes: "The use of the word for 'tongue', organ of speech, in the sense of 'language' occurs not only in many European idioms which may have influenced each other, but is also found in various nonEuropean languages."4 The Japanese word for 'tongue' (as organ of speech), s hita, has been attested since ancient writings, but it has not developed the sense 'a (particular) language' (as in the English tongue)—and it is certainly not likely to develop that sense. Intuitively, such a development would be rather alien for Japanese speakers. This does not mean that the Japanese word for 'tongue' is not used at all in idiomatic expressions referring to certain aspects of language and language use. On the contrary, there are a good number of such idioms in contemporary Japanese; e.g. shita ga (yoku) mawaru "have a glib tongue, speak a lot, speak eloquently (literally, the tongue turns (well))"; shita o furuu "speak eloquently (literally, swing the tongue)"; and shita no saki de marumeru "talk/wheedle (someone) into doing (something) (literally, wad up with the tip of the tongue)". Note that while the word shita in these examples refers to the movement of the tongue in the act of speaking, it does not specifically refer to a language (in the sense of a 'langue') in any of these examples. It is impossible in Japanese to replace the word shita "tongue"

THE CONCEPTUALIZATION OF LANGUAGE IN JAPANESE

267

in these idioms with the word kotoba "language", unlike the word tongue, which can be replaced by language in the phrase the English tongue. The difficulty of replacing shita "tongue" with kotoba "language" here is probably well-motivated. As we have noted in Section 3, the traditional Japa­ nese conceptualization of language is primarily concerned with language as an event—a process in which kokoro "mind/heart" emerges as kotoba "speech, language". In no way is language primarily conceptualized as a (hypostatized) object—a 'langue', or an abstract system of signs with its associated set of rules. 4.2 An extensive investiagation of the way language is metaphorized in Japa­ nese is found in Nomura (1996), a study undertaken at my instigation, and I can do no better here than present a summarized statement of his findings. Nomura's findings on Japanese can be summarized by the following general methaphor: LANGUAGE IS A FLUID, which involves the following four metaphors: a.

PRODUCTION OF WORDS IS LETTING OUT A FLUID: e.g. shinratsuna

b.

kotoba o abiseru "shower (someone) with biting words (literally, pour biting words)"; kotoba o morasu "utter words in spite of one­ self (literally, let leak words)"; FLUENCY OF SPEECH IS THE SPEED OF FLOW OF A FLUID: e.g. yodominaku hanasu "speak fluently (literally, speak without stagnation)"; tateita ni mizu o nagasu yo ni hanasu "speak fast and fluently (liter­ ally, speak as if letting water flow down on an upright board)";

C.

INTELLIGIBILITY OF WORDS IS TRANSPARENCY OF A FLUID: e.g.

d.

kotoba

o nigosu "speak ambiguously (literally, make words turbid)"; RECEPTION OF WORDS IS TAKING A FLUID: e.g. kotoba o kumu "take (someone's) words into consideration (literally, draw/scoop up words)"; kotoba ga kokoro ni simiru "be impressed by (someone's) words, deeply appreciate (someone's) words (literally, words soak into the heart)".

While it is true that there are examples of non-fluid metaphorization of language in Japanese and that English also has examples of fluid metapho­ rization of language, Nomura (1996) has convincingly shown that, while English tends to metaphorize language as some kind of receptacle (cf. Reddy 1993), fluid metaphors of language are dominant in Japanese. As is apparent from the examples above, 'speaking' is very often described in terms of verbs,

268

YOSHIHIKO IKEGAMI

adjectives, and adverbs which normally collocate with words referring to a fluid or flowing. Nomura furthermore points out that 'thoughts' as well as 'feelings' (in fact, the two categories tend not to be as clearly distinguished conceptually in Japanese as in English; of. Section 3) are most commonly metaphorized as fluids, whereby the relationship between 'thoughts/feelings' and 'words' is not one of 'content' and 'container', as one may find in English (cf. Reddy 1993), but one of thoughts/feelings 'becoming' words (see also Section 3). 5.

Conclusion

In concluding the present short essay, let me consider some general impli­ cations from the above discussion of the Japanese conceptualization of 'language'. Reddy's discussion of the metaphorization of the communicative process in English suggests that the basic schema is based on the contrast between 'content' and 'container'. The content is 'unbounded', while the container is 'bounded'; by being put into the 'bounded' container, the 'unbounded' content is articulated and given the status of an individuated entity. 'Content' and 'container' clearly stand opposed to each other in a dichotomous relationship— a relationship which one may detect in such pairs of notions as 'mind' vs. 'body' and 'material' vs. 'form'. While the conceptualization described by Reddy is characterized in terms of 'boundedness' and 'dichotomy', the Japanese conceptualization is charac­ terized by neither. In Japanese, language is metaphorized as a fluid, just as the speaker's thoughts and feelings are. All of these are conceptualized as 'unbounded' and the relationship is conceived of in terms of thoughts and feelings (kokoro "mind/heart") being turned into, or becoming, language {kotoba "speech/language"). Here is the world of '(ewiges) Werden'. One may also recall the rejection of language in Zen Buddhism as shackles to one's thinking and of the dichotomy between 'mind' and 'body' in the ascetic prac­ tices to be undergone on the way to enlightenment. The two different approaches to the conceptualization of language suggest that the word for 'speech/language' may differ across cultures depending on whether it primarily refers to the process of the speaker putting his thoughts and feelings {kokoro) into words {kotoba) or to the system of linguistic signs associated with a set of rules for their use. These two concep­ tualizations are very different things. One conceptualizes 'language' as an object, bounded and stable; the other as an event, unbounded and in permanent flux.5

THE CONCEPTUALIZATION OF LANGUAGE IN JAPANESE

269

Finally, it must be emphasized that the difference between the two conceptualizations I have discussed in the present paper should not be taken as an absolute one. As I have already pointed out, the conceptualization of language as a fluid is not totally absent in English. At the same time, the conceptualization of language in terms of content and container is now also quite familiar in Japanese, after it has been been exposed to Western ideologies for centuries. What we find here is, after all, nothing but a familiar situation in which one culture opts for one cognitive alternative and the other for another possible alternative, both alternatives, however, being well-motivated in terms of human cognition (and, therefore, fully understandable by a member of a culture who conventionally opts for a different alternative). Notes 1 Note that the Chinese writing system was introduced into Japan in the fifth century. Besides being adapted to transcribe the native language stock in writing, Chinese was accorded for some centuries a prestigious position analogous to Latin in medieval Europe, as it was used in the officially edited history books and in the literary compositions practiced among the aristo­ crats. 2 Notice also that the Japanese noun makes no morphological distinction between singular and plural. 3 It may be added in passing that the contrast in question here is rendered in Japanese as the one between kotoba "speech/language"—the word discussed in Section 2—and kokoro "heart/mind", and that the word kokoro in Japanese covers both the intellectual and emotive aspects of the psychological process, which is a symbolic indication that the distinction between the subjective and the objective is not as rigidly insisted on in the Japanese tradition as it is in Western thought. 4 Ullmann refers to Révész (1956:56-57), where, sure enough, a number of examples are adduced, ranging from European languages (Gr. glossa, L. lingua, Fr. langue, Eng. tongue, G. Zunge, Rus. jazik, etc.) to 'primitive languages' including Malayan, Samoan, Bantu, Sudanese, etc. 5 I suspect that in Révész's description referred to in note 4, the distinction is not properly taken into account and that at least some of his alleged examples of metaphorization of 'language' as 'tongue' do not involve the notion of 'a particular language' but simply of 'speaking, uttering words'.

References Eco, Umberto. 1995. The Search for the Perfect Language. Oxford: Blackwell. Herskovits, Annette. 1986. Language and Spatial Cognition. Cambridge: Cambridge University Press. Hinds, John. 1987. "Reader Versus Writer Responsibility: A New Typology". Writing across Languages: Analysis of L2 Text ed. by Ulla Connor & Robert B. Kaplan, 141-152. Reading, Mass.: Addison-Wesley.

270

YOSHIHIKO IKEGAMI

Ikegami, Yoshihiko. 1997. "Subjectivity, Ego-Orientation and the So-Called Subject-Object Merger. Paper read at the Eighth Conference of the Euro­ pean Association for Japanese Studies, Budapest. . 1998. "Sign Conceptions in Japan". Semiotik: Ein Handbuch zu den zeichentheorischen Grundlagen von Natur und Kultur ed. by Roland Posner, Klaus Robering & Thomas A. Sebeok, Vol. II, 1898-1910. Berlin: Walter de Gruyter. Kindaichi, Kyosuke. 1944. Kotodama  Megurite [On Word-Spirit]. Tokyo: Yoshima Shoho. Kövecses, Zoltán. 1988. The Language of Love: The Semantics of Passion in Conversational English. Lewisburg, Pa.: Bucknell University Press. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago, I11.: The University of Chicago Press. & Mark Johnson. 1980. Metaphors We Live By. Chicago, Ill.: The University of Chicago Press. Lippi-Green, Rosina. 1994. Language Ideology and Language Change in Early Modern German: A Sociolinguistic Study of the Consonantal System of Nuremberg. Amsterdam: John Benjamins. Lyons, John. 1982. "Deixis and Subjectivity: Loquor, ergo sum?" Speech, Place, and Action ed. by Robert J. Jarvella and Wolfgang Klein, 101— 124. New York, N.Y.: John Wiley & Sons. Niemeier, Susanne & René Dirven, eds. 1997. The Language of Emotions. Amsterdam: John Benjamins. Nomura, Masuhiro. 1996. "The Ubiquity of the Fluid Metaphor in Japanese: A Case Study". Poetica (Tokyo) 46.41-75. Pütz, Martin & René Dirven, eds. 1996. The Construal of Space in Language and Thought. Berlin: Mouton de Gruyter. Radden, Günter. 2001. "The folk model of language". Metaphorik 01/2001 (PDF), via www.Metaphorik.de. Reddy, Michael J. 1993 [1979]. "The Conduit Metaphor: A Case of Frame Conflict in Our Language about Language". Metaphor and Thought, 2nd edition ed. by Andrew Ortony, 164-201. Cambridge: Cambridge Uni­ versity Press. Révész, G. 1956. The Origins and Prehistory of Language, translated by J. Butter. London: Longmans, Green and Co. Rumsey, Alan. 1990. "Wording, Meaning and Linguistic Ideology". American Anthropologist 92.346-361. Schieffelin, Bambi . , Kathryn A. Woolard & Paul V. Kroskrity, eds. 1998. Language Ideologies: Practice and Theory. Oxford: Oxford University Press.

THE CONCEPTUALIZATION OF LANGUAGE IN JAPANESE

271

Silverstein, Michael. 1979. "Language Structure and Linguistic Ideology". The Elements: A Parassession on Linguistic Units and Levels ed. by Paul R. Clyne, William F. Hanks & Carol F. Hofbauer, 193-247. Chicago, Ill.: Chicago Linguistic Society. Tokieda, Motoki. 1966 [1940]. Kokugogakushi [History of Japanese Linguis­ tics], 14th revised edition. Tokyo: Iwanami Shoten. Toyoda, Kunio. 1990. Nihonjin no Kotodama Shiso [The Japanese Idea of Word-Spirit]. Tokyo: Kodansha. Ullmann, Stephen. 1962. Semantics: An Introduction to the Science of Meaning. London: Blackwell. Woolard, Kathryn A. & Bambi . Shieffelin. 1994. "Language Ideology". Annual Review of Anthropology 23.55-82.

THE EVER-STIFLING ESSENTIALISM LANGUAGE AND CONFLICT IN POLAND (1991-1993)

KAROL JANICKI University of Bergen

This paper consists of a theoretical/philosophical part and an empirical one. The theoretical part first discusses conceptual essentialism and the view that one correct definition of a word/concept is in principle possible. Subsequently, it provides some fundamental arguments for a contrary, non-essentialist view of meaning and definitions. According to this non-essentialist view, language users, on the whole, define words differently (and thus, there is no one correct definition). This view allows for individual differences as well as for a significant degree of word meaning overlap, dictated by the experience shared by the speakers of a particular language or language community. It also stresses that words have no axiological load. The paper provides fourteen statements which characterize the non-essentialist position and which constitute the theoretical framework for the analysis of the data. The empirical part of this paper addresses two conflicts that erupted in Poland after 1989, that is, after the fall of communism. They revolve around abortion and the Christian values and censorship. After analyzing the data, the claim is discussed in some detail that essentialist understanding and use of language is conducive to conflict.

1.

Introduction1

Conflicts are of many kinds and they have many sources. While there are many factors that may contribute to the development of a conflict, one very important factor is of a linguistic nature. Obviously, many, if not most, conflicts involve the use of language, but it is not this trivial observation that I intend to discuss. Rather, the linguistic factor at issue in this paper is the conflicting parties' essentialist understanding and use of language, which, in turn, reflects their essentialist set of beliefs about the world.2 It is my contention, then, that a speaker's/hearer's set of fundamental beliefs about language will markedly contribute to alleviating or deepening conflicts. In this paper, I will try to show

274

KAROL JANICKI

how essentialist beliefs about language may be seen as significantly contrib­ uting to the emergence of conflicts in present-day Poland. Although this paper discusses data from conflicts in Poland, these conflicts are not restricted to the Polish context. It is in Poland, however, that conflicts couched in essentialist language use (i.e. reflecting an essentialist view on language) abound. The fact that Poland is a former communist country is of relevance here. Indeed, the communist totalitarian ideologists, like any totalitarian ideologists, advocated, countenanced, and sometimes ruthlessly demanded a black and white, good guys and bad guys (i.e. essentialist) percep­ tion of the world. As a result, a good ten years after the fall of the communist regime, this essentialist mindset is still widely present in Polish society, which, interestingly, has never been communist in any politically truthful way. Although the bipolar perception of the world engendered by essentialism is by no means endemic to the societies of totalitarian political systems, it is espe­ cially in those systems that it becomes officially promulgated and required. Like the other former communist countries, present-day Poland sees a plethora of conflicts. They are largely due to the fall of communism in Eastern Europe, which necessitated major political, social, and psychological reorien­ tations and changes in value judgments. Some of the questions posed, or posed anew, around which conflicts evolve are: Is Poland really a part of Europe? Should Poland join or continue to be a member of a variety of European organizations (such as the European Parliament, the European Union)? Is Poland a predominantly Catholic country? What are Christian values? What is natural law? Is censorship necessary? Is communism utter evil? I intend to show that such questions are often asked as essentialist ques­ tions; furthermore, answers given to these questions are also often given according to the essentialist canon. An essentialist question generating an essentialist answer is like two people shooting and trying to kill each other, both parties believing to be shooting, killing, or dying for the right cause. When essentialist reasoning built into language use is imposed on an emerging or an existing conflict, prospects for a compromise or agreement are extremely bleak. 2.

Essentialism

In the philosophical tradition, essentialism goes back to Plato and Aristotle (cf. Popper 1945; Hallet 1991). Aristotle (1983) claimed that all things have their essences, which Lakoff (1987:161) defines as "those properties that make the thing what it is, and without which it would not be that kind of thing". Essen-

THE EVER-STIFLING ESSENTIALISM

275

tialist beliefs have a direct reflection on assumptions about language. Aspects of the essentialist view of language which are of interest to me here concern meaning. According to Aristotle, "by describing the essence to which [a] term points which is to be defined, we determine or explain the meaning of the term also" (Popper 1945:13). In other words, the essentialist definition of a term, or word, constitutes its meaning. Aristotelian essentialism has flourished for the last two thousand years. It has taken a variety of forms, the most important of which is the common belief that words have or should have one meaning. This observation is true for many academics and laypersons alike (cf. Janicki 1990 for examples from sociolinguistics, and Janicki 1999 for a fuller discussion of the issue). Essentialism built into language use may take the form of an explicit intellectual commitment, a salient practice, a covert realization, or of any combination of these three. The first form typically occurs in religiously motivated discourse, in which the belief is conveyed that 'one must call things by their names', that good and evil, as well as many other concepts with a heavy axiological load, are clearly distinguishable and unequivocally definable, and that 'morality' has only one undebatable definition. Explicit essentialism hardly allows compro­ mises about the meanings of words. Words, it is believed, do not really mean what they mean for the thousands or millions of their users in a particular language; words really mean what they mean to the selected few philosophers, thinkers, ideologists, or leaders. The tacit or explicit argument here is that the right, real, ultimate, correct meaning of a particular word can only be known to the enlightened insider, and not to the unsophisticated ignorant outsider. It is crucial to realize that the opinions concerning meaning referred to above are derived from the belief in the existence of objective truth and from the related belief that the objective truth not only exists but that it can also be known, be it to the selected few only. The following is an example of essentialism as an explicit intellectual commitment: I am not interested in what the opinion of the majority is. This is not about poli­ tics—as in the Mazowiecki government case—but about the fundamental distinction between good and evil. Abortion is evil, without any doubt. There is nothing to talk about. (Polish Radio Broadcast, January 25, 1991)

When essentialism takes the form of a salient practice, it most often occurs as explicit What-is-auestions (cf. Popper 1945), which require one unambiguous answer. Examples include: 'What is science?', 'What is sociolinguistics?', 'What is a conflict?', 'What is ethical?', 'What is an explanation?', and 'What

276

KAROL JANICKI

is morality?'. Wittgenstein's early work (1961 [1921]) is another case in point, in that Wittgenstein authoritatively states what propositions are and what language is, and that he is not ready for compromises, which would have to take into account other speakers' use of the terms. Answers to What-is-auestions may not always be readily available, but they are believed to be available in principle, independent of how hard it may be to find them. Consider the following example: Few problems continue to generate so much endeavor and so much conflict as the problem of style. Even conferences are called to attempt to answer the questions: 'what is style?', 'how to study it?' ... Marozean began by admitting that the question of style is as open today as it was two millennia ago. Some progress was made in the nineteenth century when scholars began 'traiter de style comme un objet de science', but consequently the question of what style is became more pressing. (Gray 1969:7-8)

Thirdly, essentialism is also very common as a covert phenomenon, whereby authors may even overtly reject it, but their essentialist mindset moves them back to it. Essentialistic thinking has been far commoner than the formulation of explicit, general essentialistic doctrines ... These doctrines are but surface waves; the sands beneath are drenched with essentialistic practice—with piecemeal essen­ tialistic reasoning. More extensive, more diffuse, and less obvious, hence less exposed to effective criticism, this underlying essentialism may long survive the surface variety. (Hallett 1991:28)

Hallett provides excellent examples of implicit essentialism: Observe this argument of Waismann's: 'What I have to say is simply this. Philosophic arguments are not deductive; therefore they are not rigorous; and therefore they don't prove anything. Yet, they have force.' Here two explicit premises beget an explicit conclusion with the aid of two tacit premises. Whereas the spoken premises are not essentialistic, the unspoken ones are. The first therefore suggests that only deductive arguments are or can be rigorous; the second therefore suggests that only rigorous arguments do or can prove anything. Such is the essence of rigor; such is the essence of proof. The essentialism looks multiple. (Hallett 1991:70)

Philosophers such as Popper (1945), Wittgenstein (1967), and Hallett (1991), among others, have shown that essentialism is wrong. The two main reasons are stated in Popper (1945:9-21). First, it is not at all beyond argument that intuitive definitions, extolled by Aristotle, let us in on the secret of the essence

THE EVER-STIFLING ESSENTIALISM

277

of things designated by the words defined. Second, defining words will always require further defining of the words in the definition, thus giving rise to an infinite regression of definitions. Indeed, anyone adhering to the Aristotelian method of definition reads definitions 'from the left to the right'; in a definition such as 'A hill is a small mountain', 'a small mountain' is believed to tell us what 'hill' is. The danger of definitional infinite regression can easily be seen, as the questions that immediately come to the fore, 'What is a mountain?' and 'What is small?', elicit answers that will bring forth an array of further ques­ tions, thereby continuing the regression. In Popper's view, the sciences that have discarded this 'left-to-right' way of handling definitions, which hardly leads anywhere, have been moving forward. Those sciences are reputed to read definitions 'right-to-left'; the definition 'A hill is a small mountain', then, should be read as 'A small mountain is (called) a hill'. Truly, such definitions do not explain anything (but Aristotelian definitions did not either). They only introduce one short label for a somewhat longish set of labels; "they cut a long story short" (Popper 1945:14). It follows that 'left-to-right definitions' should not be taken at all seriously. Rather than make things simpler or more precise—as is commonly believed—they make things more complicated and less precise. And this is primarily because any definition entails the necessity of further definitions with a continuously growing number of terms to be further defined. The 'right-to-left definitions', though not explaining anything, are nevertheless very handy because they save discussion time. 3.

A non-essentialist's view

My own non-essentialist position may be traced back to the work of the philosophers mentioned above, to that of general semanticists (e.g. Korzybski 1933; Hayakawa 1939) as well as to that of cognitive linguists (e.g. Lakoff 1982, 1987; Johnson 1987; Lakoff & Johnson 1980; Gibbs 1994; Ungerer & Schmid 1996). Importantly, though, its details do not reflect any one single theory advanced by any one of these authors mentioned, and there may be no other cognitive linguist that will endorse all the theoretical assumptions that I make in (a) through (n) below. The most important aspects of my non-essen­ tialist position, which are of relevance to the present discussion of conflicts, are the following: a.

All concepts expressed in natural language are fuzzy. There is no such thing as a precise concept.

278

KAROL JANICKI

b.

The meanings of words are best understood in terms of prototypically structured concepts, with instances gradually moving from the prototypical center to the periphery.  Concepts expressed by words include a borderline area where instances are variably assignable to one concept or another; in other words, conceptual category membership is variable at conceptual boundaries. d. The variable assignment in (c) is intersubjective; that is, language users who are socially close to one another assign examples to concepts in similar ways. e. Concepts are grounded in the individual's experience. Thus, mean­ ings of words, which stand for concepts, reflect that individual's experience, in addition to the experience which is shared with other individuals around him/her. f. Concepts involve an axiological aspect; one of the basic axiological continua is that of good-bad. g. It is a futile task to try and give unambiguous definitions, as the defining words will incessantly require further definitions.3 h. The existence of only one definition of a word may be thought of as a myth, or an impossibility. i. Definitions can only be treated as working definitions and they should be sought only to the extent that they may help clarify a point or solve a problem. j. Definitions should always be read 'right-to-left' and not 'left-toright'. That is, for instance, Measles is an acute contagious viral disease, usually involving the eruption of red spots should not be understood as saying that measles really is an acute contagious... (reading the definition from left to right), but as saying merely that an acute contagious viral disease, usually involving the eruption of red spots may be called, or is called, measles (reading the definition from right to left). k. Attempts at finding precise, ultimate, correct, right, unambiguous definitions are futile. 1. Words may best be conceived of as mere labels which allow us to talk about reality. We may therefore freely play around with words to the extent that we can communicate with other language users, provided communication is our goal. m. Words are not things, just as maps are not the territory.

THE EVER-STIFLING ESSENTIALISM

n.

279

We never know exactly what we are talking about. Low degree mis­ understanding is built into language due to the experiential ground­ ing of the individual's concepts.

In order to preempt the criticism that may be leveled at my own position, I wish to stress that sometimes the reader may suspect that I have fallen victim to the essentialist trap myself, as in (e) above. My formulations may be suggestive of essentialism, but only because of their stylistic form (see Janicki 1999 on this point), which is hardly avoidable, at least in the English language. Formulation (e) above should thus be read as saying 'concepts may best be thought of as grounded in the individual's experience', and not as 'definitely, indisputably grounded in the individual's experience'. Thus, still further in (e) above, 'meanings of words, which may be thought of (within a theoretical framework) as standing for concepts, may be thought of as reflecting that individual experience'. 4.

Two conflicts in present-day Poland

In the present section, I briefly report on two conflicts in current Poland. The data is meant to illustrate the claim that the essentialist understanding and use of language is conducive to conflict. I chose the two conflicts below because they are crucial to the understanding of the socio-political situation of the country, and, most importantly, because any conclusions about the linguistic aspects of these conflicts are also valid outside out of Poland. 4.1

Abortion

Between 1956 and 1993, Polish law allowed abortion virtually on request. In 1989, a heated debate began over what constitutes legal abortion, a debate which is in fact still going on even though a highly restrictive law has recently been passed by the parliament. The abortion conflict involves die-hard oppo­ nents of legal abortion (primarily centered around the Polish Catholic Church); liberals, who are in favor of legal abortion on request; and a relatively quiet group of semi-liberals, who accept abortion with some restrictions. The first group is against abortion even when the pregnancy is a result of rape or when the fetus is genetically impaired. The third group accepts abortion, for instance, when the pregnancy results from rape. The aspect of the conflict of interest for this paper is in what terms arguments for the various positions are couched. As might be expected, the various issues raised revolve around the followig two fundamental questions: (i) 'What is life? and (ii) 'What is the

280

KAROL JANICKI

human being?' Related questions such as (l)-(3) below can be answered only as a function of the answers to questions (i) and (ii). (1) Czy lekarz dokonujący przerwania ciąžy jest zabojcą? "Is the doctor performing an abortion a murderer?" (Staniewicz 1990:4)4 (2) Co jest sprawa fundamentalna dla ludzi? "What is the matter of fundamental concern for the human being?" (question raised during the parliamentary debate on abortion on May 16,1991) (3) Czy w istocie, pod pozorem rozwazania problemu aborcji, nie rozważamy problemu prawa do žycia? "Pretending to be discussing the question of abortion, aren't we in fact debating the question of the right to live?" (statement formu­ lated during the parliamentary debate on abortion on May 16, 1991) In (1), the underlying What-is-question is 'What is a murderer?' The way (1) is formulated suggests that only one of the following two answers is possible, and thus true: a doctor performing an abortion is a murderer, or a doctor performing an abortion is not a murderer. No other answers are invited here. In (2), the What-is-question is formulated in a straightforward way, with only one answer being invited here as well. In (3), essentialist reasoning is reflected much less explicitly than in (1) and (2). Statement (3) says that the real problem under discussion is the problem of the right to live, and that the question of abortion actually only masks the discussion of the real problem. The What-is-question involved, then, is 'What is the real problem under discussion?' The sole essen­ tialist answer to this question is: the real problem under discussion is the right to live (and not abortion). A study of the ways in which questions such as (l)-(3), and especially questions (i) and (ii) above, are answered enables us to identify at least the following three categories of conflict participants: -

those that claim that they definitely know the answers via religious illumination or teachings (e.g. life begins at conception); those that claim that biological science has given the definitive defi­ nition of what life is; and

THE EVER-STIFLING ESSENTIALISM

-

281

those that realize that biology has not yet given any final absolute definition of life, but assume that it will, or that biologists are in principle capable of formulating such definitions; cf. (4):

(4) Nauka nie wypowiedziala jeszcze ostatniego slowa kiedy rozpoczyna się žycie. "Science has not yet pronounced its last word as to when life begins." (Stawicka 1990:4) Statement (4) is a fairly typical example of covert essentialism. Its essentiahst character can be inferred from kiedy rozpoczyna się życie "when life begins", which is a paraphrase of the What-is-question 'What is the beginning of life?'. Furthermore, the fact that science has not yet pronounced its last word on this matter suggests that it can in principle do so. Hence, the reasoning runs further, the question 'What is the beginning of life?' is ultimately answerable. Fierce discussions around the central issues of the abortion debate— expressed in (i) and (ii) as well as in (1) and (2) above—lead to a multitude of questions, highly arguable statements, emotional assertions, and allegations. In what follows, I will present a selection of these, and some brief comments. Consider (5): (5) Nauka już wie na pewno  to jest życie. Niektórzy zadają jednak pytanie: ale kiedy zaczyna się czlowiek? To jest tak jakby mnie zapytano czy jestem mokry kiedy wyjdę na deszcz. "Science has given the final answer as to what life is. Some people ask the question, however, of when the human being begins. This is as if I were asked if I get wet when I go out in the rain." (statement from the parliamentary debate on abortion on May 16, 1991) In (5), science is given credit for having provided the definitive answer to the question 'What is life?', which is obviously a What-is-question. The other question in (5), 'When does the human being begin?', easily translates into the essentiahst question 'What is a human being?'. The jocular answer in (5) to the latter question may be taken as essentiahst in that, just like rain implies getting wet, life implies human beings. In other words, if we know what life is, we know what human beings are. This essentiahst reasoning indicates that the definitive (scientific) definition of life provides an answer to the question what a human being is. Definitions that do not do so are wrong.

282

KAROL JANICKI

In the same parliamentary debate on abortion, the following opinions were voiced: (6) Musimy się nauczyć odróżniać obłudę i kłamstwo od prawdy. "We must learn to distinguish hypocrisy and lies from truth." (state­ ment from the parliamentary debate on abortion on May 16, 1991) (7) Przemilczenie jest mówieniem nieprawdy. "Wilful silence is lying." (statement from the parliamentary debate on abortion on May 16, 1991) Statement (6) implies that hypocrisy, lies, and truth are clear-cut concepts which do not allow for any degree of overlap. In line with essentialist tenets, (6) conveys that no statement can be true for one reason and untrue for another. Moreover, it rules out truth as a relative notion, or any individual perceptions and judgments as to what is truth and what is not. Statement (7) is another example of an explicitly essentialist position, couched in terms of a straight­ forward answer to a What-is-question. Statement (8) concerns the referendum on abortion: (8) Głosowanie nad zalegalizowaniem niszczenia yżcia ludzkiego nie tylko tamie prawo człowieka, ale godzi wręcz w caiy porządek naturalny. Ewentualne dopuszczenie do takiego głosowania - czytamy dalej - mogloby w konsekwencji doprowadzic do innych czynów niegodziwych, np. eutanazji, czyli pozbawienia życia ludzi chorych jako nieużytecznych dla spoleczeństwa. Najbardziej radykalną formą kwestionowania prawa do žycia w oparciu o powyższe kryteria były -jak wiemy - obozy koncentracyjne. "Voting on the legal destruction of human life not only violates human rights but is an attack on the whole natural order. If the refer­ endum is permitted—we read later—it may consequently lead to other vicious deeds, for example, euthanasia, i.e. killing sick people who are perceived as useless for society. As we know, the most radi­ cal form of questioning the right to live, based on the above criteria, was concentration camps." (Rakowiecki 1991:3) In (8), the essentialist reasoning is more covert. It springs from the following chain of deductive steps. As voting on the legal destruction of human life violates human rights, any correct definition of human rights rules out

THE EVER-STIFLING ESSENTIALISM

283

reference to the legal destruction of human life. Further, as the legal destruction of human life also constitutes an attack on the whole natural order, the correct definition of natural order rules out reference to the legal destruction of human life as well. Further still, euthanasia is defined as killing sick people who are perceived as useless for the society, and as such provides a clear answer to the essentialist question 'What is euthanasia?'. To end with, this passage tells us that euthanasia equals questioning the right to live, just like concentration camps questioned the right to live. So, euthanasia is like concentration camps. Ultimately, the reasoning leads us to the conclusion that voting on legal destruction of human life equals (is) voting on the legitimacy of concentration camps. Statement (9) provides essentialist answers to the questions 'What is a pro-choice person?' and 'What is a pro-life person?': a pro-choice person is an advocate of murdering innocent children, while a pro-life person is an advocate of murdering and mutilating women as well as of depriving them of civil rights. (9) W Sejmie, jak zanotowała dziennikarka 'Gazety Wyborczej', zwolennicy mordowania niewinnych dzieci walczą ze zwolennikami mordowania, kaleczenia kobiet oraz pozbawiania ich praw obywatelskich. "In the Parliament, as a journalist from Gazeta Wyborcza reports, advocates of murdering innocent children fight with the advocates of murdering and mutilating women as well as of depriving them of civil rights." (Passent 1991:3) The following question asked by an MP of another MP may be taken as the epitome of the abortion conflict exchange of insults: ( 1 0 ) - Szanowny Pan Poseł..., kończąc wywiad dla Życia Warszawy, wypowiedział się o osobach, które nie dają przyzwolenia na [ustawę o ochronie žycia poczętego] w następujący sposób: j eželi Bóg chce ukarać to rozum odbiera. - Panie Pośle! A czym Pan sobie zaslużył na taką karą boską? - "You, sir, toward the end of your interview for Życie Warszawy, expressed your opinion of the people who are against antiabortion law, in the following manner: if God wants to punish someone, he deprives him/her of reason."

284

KAROL JANICKI

- "Sir! And how about yourself; what made God punish you in the same way?" (statement from a parliamentary debate, January 1993) In exchange (10), a pro-choice person is referred to as a person who has been deprived by God of his/her reason. Some people have reason (those that are against legal abortion), and others do not (those that are pro-choice). As such, the correct definition of a pro-choice person excludes reference to reason; that is, a pro-choice person has no reason. Conversely, any definition that attributes reason to a pro-choice person will be treated as utterly and ultimately wrong. 4.2

Christian values and censorship

The lifting of communist censorship in Poland in 1989 led to freedom in the mass media. As a result, a very wide range of views has since then been reflected in the Polish press, on the radio, and on television. Shortly after the communist censor's office was disbanded, however, opinions began to be voiced covertly arguing for some form of censorship. The mass media began to be accused of promulgating pseudo-values, pornography, lascivious behavior and promiscuity, as well as unjust criticism of the government and the Catholic Church. Toward the end of 1992, after a very long and fierce public and parliamentary debate, the Polish Parliament passed a constitutional amendment requiring that the mass media in Poland comply with the fundamental set of Christian values. In the line of this amendment, statement (11) explicitly commends the return to formal censorship: (11) Nie cenzura polityczna, lecz stojąca na strazy humanistycznych wartosci, przede wszystkim godności cziowieka - taka cenzura jest konieczna. "Not political censorship, but one that protects humanistic values, first of all human dignity—that kind of censorship is indispensable." (Mastalska 1993:9) In (11), essentialism materializes in two different ways. First, (11) expresses the belief that political censorship and censorship protecting humanistic values can clearly be distinguished from each other. Second, it implies that these two types of censorship are also axiologically different, with political censorship being bad and the other one being good (indispensable). The way the statement is formulated rules out the possible interpretation that 'censorship that protects

THE EVER-STIFLING ESSENTIALISM

285

humanistic values' might be bad (dispensable). In sum, what is good is clearly demarcated from what is bad, and no third possibility is allowed. The conflict around censorship is reflected linguistically in ways similar to those of the abortion conflict. In what follows, I will present a selection of opinions and some brief comments. In a TV interview, a Polish MP, questioned about the meaning of 'Chris­ tian values', responded: (12) [Respektowanie wartości chrześcijańskich oznacza], it nie powinno się pokazywać negatywnych przykladów z žycia sług Kościoła. "Respecting Christian values means that no examples of negative aspects of the life of the vicars of the Church should be shown." (Pietrzak 1993:1) Statement (12) provides an (at least partial) answer to the question 'What are Christian values?'. According to this MP, 'Christian values' means refraining from showing examples of negative aspects of the life of the vicars of the Church, or more generally, refraining from criticizing the Church. In other words, any criticism of the Church is a violation of Christian values. In an official statement on Christian values and censorship, the Polish bishops issued the following declarations (cf. 13-14): (13) Przyjęcie ostatniego postulatu nie ma na ceïu ani wprowadzenia żadnego systemu cenzury, ani tez ograniczania czyichkolwiek praw. Rozwój naszej kultury oraz sposób przepływu informacji nie mogą być jednak podporządkowane zasadom wolnego rynku, które mogłyby łatwo schlebiać prymitywnym gustom, wulgarnym formom, pogonii za brutalnością czy sensacją. Troszcząc się o dobro wspólne i o integralny rozwój osoby ludzkiej, opowiadamy się za regulacjami prawnymi, które nie pozwalałyby propagować cynizmu jako postawy życiowej równie dobrej jak humanizm. "Accepting the last postulate aims neither at the introduction of any censorship nor at restrictions of anyone's rights. The development of our culture and the way in which information is conveyed cannot, however, be subordinated to the free market principles, which could easily pay lipservice to primitive predilections, vulgar forms, chasing brutality, and sensationalism. Caring for the common good and for an integral development of the human being, we opt for legal regulations which would not allow the propagation of cynicism as a

286

KAROL JANICKI

world view which is equally good as humanism." (excerpt from a statement made by the Polish bishops, published in Tygodnik Powszechny 19, 1993, p. 2) What follows from (13) is the claim that legal regulations which keep cynicism from being put on a par with humanism cannot be called censorship; or conversely, censorship on this matter does not count as legal regulations. More generally, this passage reveals the frequently held view that things need to be called by their right name (which may be paraphrased as the metaphorical 'the map is the territory'), clearly an essentialist belief. Thus, the reasoning goes, legal regulations precluding the propagation of cynicism cannot be called 'censorship'. If the term censorship is used with reference to these legal regu­ lations, it would be misused or used inappropriately; something which is not really censorship should not be called censorship. The reasoning, then, is that the word censorship has one correct definition (which excludes legal regula­ tions precluding the propopagation of cynicism) and that things should be called by their correct name. (14) Troska Kościoła o respektowanie sygnalizowanych praw [wartosci chrześcijańskich - K.J.], zarówno w ustaleniach prawnych, jak i w praktyce życia, znajdowała konsekwentny wyraz w nauczaniu spolecznym papiežy ostatniego stulecia. Troski tej nie można uważać za ingerowanie wladzy duchownej w sprawy polityczne. "The Church's concern over the question of respecting the mentioned laws [Christian values, K.J.], in the written law as well as in the practice of everyday life, found its firm expression in the papal social teachings of the last century. This concern cannot be consid­ ered as the clergy's interference in political matters." (excerpt from a statement made by the Polish bishops, published in Tygodnik Powszechny 19, 1993, p. 2) In (14), the bishops state that the Church's concern over the question of respecting Christian values cannot be equated with the clergy's interference in political matters. In other words, the one correct definition of interference in political matters (whatever it is) cannot include the Church's concern over the question of respecting Christian values. Any definition which would include such concern would be thought to be undeniably and indisputably wrong. Finally, statement (15) may be treated as the epitome of the urge for censorship in the Poland of the 1990s:

THE EVER-STIFLING ESSENTIALISM

287

(15) Czyż ta 'polska wieża Babel' nie domaga się. zdrowej, humanistycznej i patriotycznej cenzury? "Doesn't this Polish Tower of Babel demand a healthy, humanistic, and patriotic censorship?" (Mastalska 1993:9). In (15), as in (11), axiology is saliently involved, in that the author explicitly makes reference to the essentialist claim that censorship protecting humanistic values (cf. 11) is healthy, humanistic, patriotic, and thus good. As a corollary, other kinds of censorship, e.g. political censorship, are unhealthy, non-human­ istic, non-patriotic, and thus bad. Any fuzzy case of censorship that protects humanistic values but which may at the same time be unhealthy in some respects, and thus to some extent bad, is ruled out. Again, the binary opposition of good and bad censorship is overtly present here. 5.

Discussion

The conflicts to which the above excerpts pertain involve people of various political orientations and personal beliefs. While in both conflicts a large spec­ trum of positions can easily be identified, it is the extreme positions which are very frequent and are most interesting for the purposes of this paper. With reference to the abortion conflict, for example, one extreme position is that abortion is indisputably murder; the other is that abortion is indisputably not murder. Similarly, one extreme position is that the human embryo is definitely a human being; the opposite extreme is that the human embryo is definitely not a human being. The above statements may be taken as responses to the Whatis-questions 'What is abortion?' and 'What is a human embryo?' The answers 'Abortion is murder'/'Abortion is not murder' and 'The human embryo is a human being'/'The human embryo is not a human being' instantiate essentialist positions because they are believed to be definitely correct, right, indisputable, and final; in addition, they express these positions in a salient fashion. (In other cases, however, essentialism might be expressed in a more covert fashion; cf. example (6)).5 Some of the statements in (1)—(15) look stylistically as essentialist, but may actually be non-essentialist in nature. In other words, some of these state­ ments may, somewhat misleadingly, be couched in essentialist terms so that they easily lend themselves to a left-to-right reading, whereas the author may in fact have meant them to be read from right to left. Statement (7), for instance, raises the question whether 'wilful silence is lying' should be taken in some absolute sense as 'wilful silence definitely, indisputably is lying', or whether

288

KAROL JANICKI

the sentence simply means something like 'I think that wilful silence may be called lying', or 'I want to call wilful silence lying', or 'I believe that wilful silence may be called lying'. In the former case, the definition in (7) should be regarded as an instance of essentialist reasoning; as such, it is read from left to right, and it is taken very seriously, as being true independent of what other people may think. In the latter case, (7) is merely a stylistic formulation which may invoke essentialism, but which in fact is not the result of any essentialist philosophical commitment; the definition is read from right to left, and it is seen as a working definition that may be useful for some purposes but that may be adapted according to circumstances. There are no safeguards against misinterpreting a non-essentialist state­ ment (i.e. a statement which looks stylistically like an essentialist one) as an essentialist one. Philosophical essentialism (not stylistic) should be seen as a set of beliefs access to which is certainly very limited for the linguist.6'7 It fol­ lows that purportedly essentialist statements should never be qualified as essentialist when considered out of context; authors of such statements should be given the benefit of the doubt. With this provision in mind, I have presented examples (1)—(15) as more or less typical cases of essentialist reasoning for which larger linguistic and/or extensive extra-linguistic context could easily be provided. Indeed, the authors of statements (1)—(15), or the people referred to therein, repeatedly provide evidence or make statements revealing their essen­ tialist philosophical commitment, even to the biggest skeptic. It is exactly this larger linguistic context and the extensive extralinguistic context (not discussed in this paper, except for the two cases below) that allows me to claim that my analysis is correct. The fact, then, that (7) can be qualified as an essentialist statement (which conveys that being silent on the abortion question must be called lying, or that lying is the only word appropriate to name such silence) does not merely follow from analyzing the statement as such; it also follows largely from the analyst's knowledge of the extensive extralinguistic context in which statements such as (7) are produced. I will now further develop the notion extensive extra-linguistic context, as it relates to quotations (6) and (9), whose English translations have been repeated here for ease of reference: (6) "We must learn to distinguish hypocrisy and lies from truth." (state­ ment from the parliamentary debate on abortion on May 16, 1991)

THE EVER-STIFLING ESSENTIALISM

289

(9) "In the Parliament, as a journalist from Gazeta Wyborcza reports, advocates of murdering innocent children fight with the advocates of murdering and mutilating women as well as of depriving them of civil rights." (Passent 1991:3). Quotation (6) exemplifies the type of statement often formulated by represen­ tatives of the political right in the parliament. These MPs, inspired by their Catholic faith, do not cease making claims in the mass media that only they are exponents of truth, justice, and morality. Consistently adhering to their reli­ gious commitment, they fiercely attack any form of moral relativism. Addi­ tionally, some of them give the impression of belonging to a small group of the selected few who 'know better' than those other Catholics outside the Parlia­ ment. One of the MPs is reported to have said (during a debate on the possible referendum on abortion law) that przypadkowe spoleczeństwo "accidental society" should not be allowed to decide about important matters. In view of all this, there can be little doubt that (6) is an example of essentialist thinking. The notions 'hypocrisy', 'lies', and 'truth', then, can only be given one single defi­ nition, which the person formulating the statement believes to be the correct one; these notions, the essentialist reasons, cannot be freely defined. On this view, something which is not really hypocrisy cannot be called hypocrisy; and something which is definitely hypocrisy must be called hypocrisy. Similarly, on the basis of extra-linguistic context, there can be little doubt that (9) is an essentialist statement. In (9), it is actually the two feuding sides which have fallen victim to the essentialist trap. The pro-life advocates are well known for uncompromisingly defying abortion and for calling it murder; in this respect, 'Murdering innocent children' is a phrase they frequently use in public statements, interviews, and newspaper articles. They base their point of view on both religious and scientific authorities, which are reported to have infallibly defined the essence of 'life' and to have identified it with 'the human being'. In their view, any attempt at interrupting life should be labeled murder. On the other hand, the pro-choice advocates argue that banning abortion equals murdering or mutilating women as well as depriving them of human rights. Many of them maintain an uncompromising position in viewing anti-abortion laws only as an act directed against women. While pro-life activists claim that legalizing abortion violates human rights (unborn children's rights), pro-choice advocates hold that anti-abortion laws violate human rights (women's rights). In my view, the way the case is argued in daily exchanges shows that both sides are equally essentialist. In other words, to claim that legal abortion is definitely a violation of human rights, or to claim that banning legal abortion is

290

KAROL JANICKI

definitely a violation of human rights, is to make equally essentialist contentions. Clearly, many participants in these two conflicts (independent of the ideology they adhere to) exhibit an essentialist frame of mind. They do not understand, or do not want to accept, that words have fuzzy meanings and that there are no clearly delineated concepts whose definition everyone can agree on. Secondly, they do not understand, or do not want to accept, that word meanings can best be defined in terms of more or less typical cases. Thirdly, they do not understand, or do not want to accept, that there is no such a thing as one final, ultimate (in some absolute sense) definition of any concept, including such highly disputed concepts as 'life', 'morality', 'ethical', 'killing', etc. It is certainly not true that science (or even physics) can provide any final answer as to the nature (ultimate definition) of any concepts that it uses (for extensive support of this claim, see, e.g., Heisenberg 1971; Kitcher 1984; Popper 1957). Finally, the participants in the conflicts do not understand, or do not want to accept, that words are mere labels which can be replaced and about whose meaning speakers/hearers can find agreement in the process of language acqui­ sition and language use. In other words, most participants in the conflicts take words very seriously; that is, they claim that a particular phenomenon must, or should not be, referred to in a particular way (e.g. the practice of abortion must be called murder). In the paragraph above, I have repeatedly used the expressions do not understand or do not want to accept. These expressions should not be put on a par. Given the 2000-year-old Aristotelian tradition and given the extensive context referred to above, one can only assume that in the overwhelming majority of cases, it is the former alternative {do not understand) that applies; that is, it seems safe to assume that the majory of politicians exhibiting essentialism are in fact die-hard essentialists, who believe that words have only one meaning, and that theirs is the correct one, or who believe that a word can only have one value (e.g. 'good'), and that theirs is the correct one. It cannot be ruled out, however, that there is a particular group of politi­ cians who realize that language does not work the way the essentialist reasons, but who nevertheless stick to the essentialist way of argumentation, in order to sound decisive, convincing, and forceful (cf. the Afterthought below)—a strat­ egy which might eventually lead to political victory. The question of how much argumentation on the part of the conflict participants is genuine and how much of it may perhaps be sheer political manipulation remains to be investigated. Our data, including the extensive extra-linguistic contexts, do not allow us to take even a preliminary position on this issue.

THE EVER-STIFLING ESSENTIALISM

291

As non-essentialists argue, concepts are best understood as evolving in the mind of human beings, not outside of it. If one accepts the non-essentialist position that all concepts are grounded in (i.e. evolve from) human experience, it follows that concepts reflect the experiences of individual human beings, and thus may differ to the extent that individual human beings are different. Keep­ ing in mind that, for the non-essentialist, the meaning of a word is best under­ stood as the concept (mental image, mental description) associated with that word, we are led to conclude that the meaning of a word (or its definition) must be equated with what the word means to an individual on a particular occasion. There are absolutely no grounds for believing that some people's understanding of a word (or concept) is, in any absolute sense, better than that of other people's. Meanings of words are open; new definitions of words may be proposed, and new words may be coined. As the experience of an individual changes, meanings of words get redefined when new incoming experience dictates such redefinition. The belief in one fixed and untouchable meaning of words may be seen as belief in something which does not exist or as an attempt at the subordination of language to human wishful thinking. For the conflicts discussed above, this implies that there is no final, unequivocal answer to questions such as 'Is the doctor performing an abortion a murderer?' (cf. 1), or 'What are hypocrisy, lies, and truth?' (cf. 6), or 'Who is an advocate of murdering and mutilating women?' (cf. 9). With reference to the question in (1), for instance, there needs to be a social agreement (i.e. between people) about the meaning of the notions/words life, human being, and murder before it can be decided whether or not to assign the label 'murder' to abortion. Stating that abortion is definitely murder (in some absolute sense) or stating that it is definitely not murder (again, in some absolute sense) are both essentialist positions. Politicians (as conflict participants) should realize that the meaning of words needs to be agreed upon by real people. Obviously, in a discussion leading toward such an agreement, people will be allowed to bring in their various convictions, beliefs, religions, and ideologies. Importantly, all conflict participants should realize that the definition of a concept/word agreed upon is nothing but 'a definition agreed upon' and that no one definition of any word exists. An awareness of these simple facts should be extremely conducive to alleviating conflict; furthermore, it should block conceitedness, self-righteous­ ness, and, most importantly, the feeling of infallibility. How do conflicts among essentialists or with essentialists come about? Essentialists believe that their understanding of concepts is the right one, that there is only one correct definition of a concept (in addition to many others

292

KAROL JANICKI

which are used and which, however, are incorrect), that words need to be understood as what they really mean. Essentialism is rampant: "People instinc­ tively feel that their ways of expressing things and of being polite and rude are 'natural' and 'logical'" (Tannen 1986:188). As such, our readiness to attribute good intentions to ourselves, to believe there is only one way and ours is it, and to extrapolate from frustrations with an individual in order to generalize to a group [contribute to] the global animosities that are causing so much suffering. (extract from Deborah Tannen's speech delivered during Commencement ceremonies at Binghamton University, May 16, 1993; as reported in the Binghamton University paper Inside 14: 32, May 20, 1993)

If a person A and a person each believe that their meanings are the correct ones, conflicts are very likely to come about, even in cases where the experi­ ential discrepancies between A and are very small. In cases where the discrepancies between A and are significant, there is obviously a lot of room for conflict. The sides—usually two—involved fiercely argue, as if each had the exclusive right to using words the proper way. The two conflicts in Poland discussed in this paper amply testify to this. Finally, one might wonder whether and/or to what extent arguments about 'abortion is murder' and 'abortion is not murder' are arguments about language or arguments about exo-linguistic axiological or moral systems. An answer that a non-essentialist will come up with is: 'You may take it as a moral issue, as an axiological, or as a linguistic one'. There is no one answer to the question (any one answer commitment would move us directly into essen­ tialism). It follows that on the non-essentialist view, the conflicts listed above can be studied as linguistic ones or as partly linguistic ones, but not necessarily so.8 6.

Afterthought

As the proceedings of the numerous sessions of the Polish Parliament between 1989 and 2002 illustrate, when two essentialists disagree on the definition of concepts, they usually end up in a ferocious battle (linguistic or physical). In a situation like that, one of the two essentialist sides gives in or is swept aside. In a real war (which I take to be an extreme example of a conflict), one of the two essentialist sides simply needs to die. If, in turn, an essentialist fights a nonessentialist, it is usually the non-essentialist who is in more trouble. The essen­ tialist is certain, determined, and convinced of his/her being right. The nonessentialist is in doubt, allows alternative interpretations, is willing to discuss alternative solutions, and is willing to compromise. The non-essentialist, as

THE EVER-STIFLING ESSENTIALISM

293

opposed to the essentialist, is much more open-minded and tolerant. In an intellectual battle with an essentialist, the non-essentialist cannot even attempt to impose his/her viewpoint on the adversary. That would be betraying the nonessentialist philosophy. In a physical battle, it is thus very likely that the nonessentialist is going to have to die! Notes 1 Although the title of the present paper refers specifically to the period from 1991-1993 (due primarily to the data which pertains to that period), all conclusions are definitely valid for the subsequent years up to the publication of this article. 2 As what is at play is basic beliefs and assumptions, one may view the conflicts discussed below as philosophically grounded. 3 Note that my position on definitions is in disagreement with Wierzbicka's theory of Natural Semantic Metalanguage (cf., e.g., Wierzbicka 1996), which states that complex concepts can be defined in terms of simple ones (e.g. people, think, feel, know) that are self-explanatory and which are understood unambiguously. My position is definitely not in agreement with what seems to be one of Wierzbicka's basic assumptions, namely that "one of the main tenets of this book is that words do have meanings, and that these meanings can be articulated. If they haven't been succesfully articulated in the past, for example, by the proponents of semantic 'features' and 'markers', it is not because words do not have any constant meanings but because the methodology was inappropriate" (Wierzbicka 1996:24). My own position on this question is that meanings are not in words; they are assigned to words by people (cf. also Lakoff & Turner 1989). 4 Note that some of quotations in Polish below do not reflect the opinions of the authors, who are journalists simply reporting on what others have said. I have chosen this way of presenting the necessary bibliographical information because it seemed the simplest, even though I realize that there is the risk of a journalist's being associated with opinions he/she does not share. 5 Whether authors of essentialist statements are dedicated and consciously committed essentialists or whether they have fallen victim to essentialism while being totally unaware of it is an entirely different matter and will not be discussed here. 5 It is interesting to note, that many essentialists, when approached explicitly about the matter, strongly object to being classified as essentialists, whereas what they do and say clearly allows for such classification. Many essentialists are like racists who tell extremely unpleasant jokes about Blacks, Jews, Arabs, and Gypsies, for example, and when directly asked the question whether they are racists, their definite answer is negative. 7 Linguists, or anybody else for that matter, have very restricted access to what people believe. Beliefs are mental entities; our access to people's mental life (their ideas, beliefs, memories, etc.) is restricted. What we have access to is only what people do (write, jump, smile, etc.) and not what they think or believe. 8 For a socio-political discussion of the conflicts discussed in this paper, cf. Flis 1999.

294

KAROL JANICKI

References Aristotle 1983. Metafizyka [Metaphysics], translated from the original Greek by Kazimierz Leśzniak. Warszawa: Polish Scientific Publishers. Flis, Andrzej. 1999. "The Polish Church as an Enemy of the Open Society: Some Reflections on Post-Communist Social-Political Transformations in Central Europe". Popper's Open Society after 50 Years: The Con­ tinuing Relevance of Karl Popper ed. by Ian Jarvie & Sandra Pralong, 159-169. London: Routledge. Gibbs, Raymond. 1994. The Poetics of Mind. Cambridge: Cambridge Univer­ sity Press. Gray, Bennison. 1969. Style: The Problem and its Solution. The Hague: Mouton. Hallett, Garth. 1991. Essentialism: A Wittgensteinian Critique. Albany, N.Y.: State University of New York Press. Hayakawa, S.I. 1939. Language in Thought and Action. London: George Allen and Unwin Ltd. Heisenberg, Werner. 1971. Schritte über Grenzen. München: R. Piper and Co. Janicki, Karol. 1990. Toward Non-Essentialist Sociolinguistics. Berlin: Mouton de Gruyter. ----------. 1999. Against Essentialism: Toward Language Awareness. München: LINCOM Europe. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago, Ill.: The University of Chicago Press. Kitcher, Philip. 1984. The Nature of Mathematical Knowledge. Oxford: Oxford University Press. Korzybski, Alfred. 1933. Science and Sanity. An Introduction to Non-Aristo­ telian Systems and General Semantics. Garden City, N.Y.: Country Life Press Corporation. Lakoff, George. 1982. "Categories and Cognitive Models". Trier: LAUT Series A, no. 96. . 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago, Ill.: The University of Chicago Press. --------- & Mark Johnson. 1980. Metaphors we Live By. Chicago, Ill.: The University of Chicago Press. --------- & Mark Turner. 1989. More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago, Ill.: The University of Chicago Press.

THE EVER-STIFLING ESSENTIALISM

295

Mastalska, Danuta. 1993. "Polska wieża Babel [The Polish Babel]". Niedziela 3.9. Passent, Daniel. 1991. "Partia nienawisci [A party of hatred]". Polityka 18.3. Pietrzak, Barbara. 1993. "Wartosci chrzescijanskie wedlug ZChN [Christian values according to ZChN-Christian National Party]". Wprost 6.1. Popper, Karl. 1945. The Open Society and Its Enemies, 2 Vols. London: Routledge and Kegan Paul. ---------. 1957. The Poverty of Historicism. London: Routledge and Kegan Paul. Rakowiecki, Jacek. 1991. "Biskupi przeciw referendum [The bishops against the referendum]". Gazeta Wyborcza 103, May 4-5, p. 3. Staniewicz, Ewa. 1990. "Dziecko w drodze [A child on the way]". Glos Wielkopolski 7-8 July. 4. Stawicka, Joanna. 1990. "Bez wyboru [With no choice]". Glos Wielkopolski 7 8 July. 4. Tannen, Deborah. 1986. That's Not What I Meant. New York, N.Y.: Ballantine Books. Ungerer, Friedrich & Hans-Jörg Schmid. 1996. Introduction to Cognitive Linguistics. London: Longman. Wierzbicka, Anna. 1996. Semantics: Primes and Universals. Oxford: Oxford University Press. Wittgenstein, Ludwig. 1961 [1921]. Tractatus Logico-Philosophicus. London: Routledge. . 1967. Philosophical investigations, 2nd edition. Oxford: Blackwell.

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

OLAF JÄKEL Martin-Luther-University, Halle-Wittenberg

This study applies the Cognitive Theory of Metaphor in the investigation of the domain of the economy as a target domain for metaphorical projections. It focuses on the concrete source domain of spatial motion, which provides a number of alternative metaphorical models for the economic domain. Based on linguistic evidence from English and German, the account shows which kinds of motion are metaphorized, what the mappings look like in detail, and what func­ tions each particular model fulfills. Finally, probable reasons for the use of exactly those particular models established by the investigation are discussed.

1.

Introduction

According to the Cognitive Theory of Metaphor, human cognition is funda­ mentally structured by means of conceptual metaphors.1 These help us to understand and handle abstract and complex concepts in terms of concrete and basic domains of experience. On the language level, these fundamental cogni­ tive structures surface in the shape of conventional metaphors, which in many cases have long been 'dead', congealed into idiomatic phrases, and lexicalized. Seen from the perspective of this cognitive approach to metaphor, the conceptualization of the economic domain and its highly complex processes constitutes a target domain par excellence for a number of systematic meta­ phorical projections. The following investigation focuses on the concrete source domain of spatial motion, which provides a number of alternative meta­ phorical models for the economic domain.2 My account will show which kinds of motion are metaphorized, what the mappings look like in detail, and what functions each particular model fulfills. The investigation is based on linguistic evidence from English and German.3 The German language material comes from a corpus covering the business sections of all issues of the Hamburger Abendblatt newspaper of the

298

OLAF JÄKEL

month of April 1991. It contains more than 3,000 sentences exemplifying the conceptual metaphors postulated. For the comparative investigation of English, I used a mixed corpus consisting of just under 400 example sentences taken from three sources: the monolingual dictionaries Longman Dictionary of Con­ temporary English (1987) and Collins Cobuild English Language Dictionary (1987), with their corpora of everyday English, and the stock market reports of all issues of the Financial Times of March 1991. Section 2 constitutes the main section of this paper and presents the linguistic data and analyses. Section 3 summarizes the results, and concludes with a discussion of probable reasons for the use of exactly those particular metaphorical models established by my investigation. 2.

Alternative models of metaphorical motion for the dynamic quantifica­ tion of economic processes

The economic domain is inhabited by abstract entities such as numbers, percentages, prices, and market quotations. All of these are reified as physical entities and endowed metaphorically with diverse kinds of MOVEMENT. This enables a kind of refined quantification of economic processes, which is clearly a principal concern of economic metaphors. Two major subgroups to be distin­ guished within these metaphorical MOVEMENT models are HORIZONTAL MOVEMENT (2.1) and VERTICAL MOVEMENT (2.2). Within VERTICAL MOVEMENT, we can differentiate between ORGANIC GROWTH (2.2.1), ACTIVE VERTICAL MOVEMENT (2.2.2), and PASSIVE VERTICAL TRANSPORT (2.2.3). Finally (2.3), we will deal with what I call the MOUNTAINSIDE-TOPOLOGY. All of these meta­ phorical models play their role in shaping our folk theoretical understanding of economic processes.4 2.1

HORIZONTAL MO VEMENT

This account of metaphorical MOVEMENT models begins with the Idealized Cognitive Model (ICM) of HORIZONTAL MOVEMENT: The conceptual metaphor ECONOMIC DEVELOPMENT IS HORIZONTAL MOVEMENT is based on the PATH

schema,5 which is employed in all its details for the conceptualization of eco­ nomic processes. In my presentation of linguistic data, I go by the following method: English-German equivalents are listed as variants (a, b) of the same example number. In each case, the 'equivalence' relates to the conceptual metaphors exemplified, or to some of their particular aspects analyzed in the commentary. As far as syntax, morphology, or lexical details are concerned, example sentences may differ greatly. Examples in English or German only are

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

299

given in cases where one of the two corpora did not yield suitable examples. However, this should not be taken to imply the lack of an equivalent expression in the other language. If this exactly is what I claim, it will be made explicit in the commentary. To begin with, there is simple FORWARD MOVEMENT (1) denoting positive 'progress'. Quite logically, then, negative development is understood as BACK­ WARD MOVEMENT (2):

(1) a. Oil shares continued to move ahead... b. XYZ-Aktien rückten vor. (2) a. In London sind die Kurse ... zurückgegangen. b. The market eased back from its recent advance. The speed of movement serves as a measure of success (3), whereas its slowing down is regarded as alarming (4): (3) a. Umsatzrenner. XYZ-Aktien. b. ... fund managers, who are still very afraid of being left behind if the widely-expected base rate cuts send the market racing ahead again. (4) a. Yesterday's slower performance from equities ... b. ... die Anzeichen einer Verlangsamung der Konjunktur ... On the stock exchange, where the dynamics of prices get set over time, daily quotations can even run a sort of 'race' (5), with some frontrunners outpacing the others (5a) as well as some slower participants dragging behind (5b). As in sports, where certain races can last several days, competitors can catch up on their backlog of the day before (5c): (5) a. The UK stock market outpaced other leading world markets. b. Among the few sectors to drag somewhat behind the market...  Der Markt habe knapp die Hälfte der Vortagesverluste aufholen können ... Thus, examples (3)-(5) can be seen as linguistic instances of a structural meta­ phor (see Lakoff & Johnson 1980:14) ECONOMIC COMPETITION IS A RACE, featuring the competing market quotations as runners.

300

OLAF JÄKEL

Metaphorical CHANGES OF DIRECTION also have a role to play in the description of economic trends, as example (6) proves: (6) a. K. erwartet eine Konjunkturwende in Ostdeutschland noch in diesem Jahr, b. There was a remarkable turn-round from an $828,000 loss to a $175,000 surplus. The profits aimed at by industrialists are destinations to be reached (7). Besides such profitable target regions (8), there are real danger zones (9) on the cogni­ tive metaphorical map, which ought to be avoided: (7) a. Bereits 1990 erreichte Deutschlands drittgrößter Anbieter ... ein Umsatzplus von 40 Prozent. b. We have reached the point at which we are beginning to make a profit. (8) a. ... bewegte sich Tokio wieder in die Gewinnzone ... b. Both shares moved to significant premiums. (9) a. Firma X ist in Konkurs gegangen. b. The company has gone bankrupt. For comparative purposes, it is worth noticing that in German this HORIZONTAL MOVEMENT is always an ACTIVE one. While English also makes frequent use of 6 PASSIVE HORIZONTAL TRANSPORT (10a, 11a, 12a-c), the German corpus yields not one single instance of transitive constructions within this metaphorical model. In German, no company is *brought into profit (10b) or *lead/thrown into bancruptcy (11b), and no price is *sent/taken forward or driven ahead (12d): (10) a. ... hoping to bring the system into profit this year. b. * ... wurde Tokio wieder in die Gewinnzone gebracht/bewegt.7 (11) a. A really big strike will throw the firm into bankruptcy.8 b. * Firma X ist in Konkurs geführt/geworfen worden. (12) a. Modest buying support was enough to send the market forward again. b. The squeeze on their positions helped to drive prices ahead.

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

301

(12) c. What will the increase take your salary to? d. *XYZ-Aktien wurden vorgerückt/vorangeschickt/vorangetrieben/ vorwärtsgebracht. In the English sentences in (10)—(12), the ACTIVITY OF TRANSPORTING is ascribed to persons responsible (10a) or even to abstract factors such as strikes (11a), buying support (12a), some unspecified squeeze (12b), or an increase (12c), while the economic entities to which some change happens remain totally PASSIVE. In German, however, those reified entities move on their own resources, without being transported by other actors. They must therefore be conceptualized metaphorically as LIVING BEINGS or at least as AUTOMOBILES.9 On the other hand, no difference whatsoever will be found between English and German concerning the issue of this model's focus. All of these instances of HORIZONTAL MOVEMENT focus metaphorically on a straightforward orientation towards progress: economic developments are meant to go ahead as fast as possible, and, of course, in the direction of increased profits. 2.2

VERTICAL MOVEMENT

The remaining metaphors of motion make use of various kinds of VERTICAL In order to understand these metaphorical models, we have to take recourse to one more fundamental image schema: VERTICAL ORIENTATION.10 AS Lakoff and Johnson (1980:14ff.) have shown, the concept HEIGHT in general functions as a schematic measure of quantities, and consequently is endowed with positive associations. In the context of economics, the vertical measure is utilized to compare amounts of money as being relatively high-higher-highest (13)-(15a). The same applies to import and export shares (15b) as well as to price quotations (16): MOVEMENT.

(13) a. Die Zinsen sind hoch. b. Europeans are now complaining of high American interest rates. (14) a. Indes habe das Angebot der Firma X um 30 Prozent höher gelegen. b. Home ownership involves higher indebtedness than renting a house does.

302

OLAF JÄKEL

(15) a. ... die USA, die ... die mit Abstand höchste Privatverschuldung verzeichnen ... b. Die Einfuhr lag um 14 Prozent höher als vor einem Jahr, die Aus­ fuhr um 3,9 Prozent niedriger. (16) a. XYZ-Aktien schlossen ... niedriger ... b. Bass closed sharply lower although above the worst of the session. What is measured here are sums of money (15a), or percentages (13, 14, 15b), or quotations on the stock exchange (16). In describing economic affairs by statically pointing at some relative height, examples (13)—(16) instantiate the image schema VERTICAL ORIEN­ TATION. While these examples do not explicitly include any idea of motion, the following models are dynamic in nature; yet, they presuppose the same schematic VERTICAL ORIENTATION, which therefore remains prevalent in the remaining sections. 2.2.1 ORGANIC GROWTH

One type of VERTICAL MOVEMENT manifests itself as the the metaphorical model ORGANIC GROWTH.11 Thus, we find the conceptual metaphor ECONOMIC DEVELOPMENT IS ORGANIC GROWTH—in English and German alike. Concep­ tualized as plantlike organisms are in particular a country's gross national product (17), an industry's/company's turnover (18b), and surpluses (19b, 20). Thus, either a more or less healthy growth (17, 18) or, antonymically, shrinking processes (19, 20) can be attributed to such abstract entities: (17) a. Economic growth has slowed down dramatically. b. Für 1992 wird mit einem Wachstum des Bruttosozialprodukts um 2,5 Prozent... gerechnet. (18) a. The economy showed a slow but steady growth. . ... ein kräftiges Wachstum der Medienumsätze ... (19) a. There has been a twenty per cent shrinkage in the steel industry in the past few years, b. Der Überschuß in der Handelsbilanz ... wird 1991 noch stärker schrumpfen...

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

303

(20) Der Überschuß im ... Außenhandel ... dürfte sich nach Einschätzung der Hamburger Wirtschaftsforscher 1991 wieder weitgehend zurück­ bilden. Example (21) presents the three verbs increase, decrease, and decline, which are frequently found in English economic discourse. That even these expres­ sions are based on the metaphorical GROWTH model can be shown by looking at their etymology. Increase goes back to the Latin verb increscere: in + crescere "grow"; the antonym decrease goes back to Latin decrescere: de "down from" + crescere "grow"; its synonym decline goes back to Latin declinare: de "down from" + clinare "bend". (21) a. World energy demand is increasing at a rate of about 3% per year. b. Our sales are decreasing.  There has been a sharp decline in profits this year. This metaphorical model of GROWTH focuses on the vertical distance which is covered by means of a continuous movement, not on intermediate stages or on a fixed goal. The growth is steady (18a) or strong (18b), or is expressed as Wachstum um/growth by a certain value (17b, 19a, 21a); hardly ever would we find in German * Wachstum auf einen bestimmten Endbetrag, or in English *growth to a fixed sum total. If an economic upward movement is conceptualized as GROWTH, it becomes quasi-natural in character: a mysterious intrinsic vital force, hardly controllable, seems to be responsible for the developments, and not a conscious actor proceeding intentionally. The possibility of such a naturalist charac­ terization is most welcome with highly complex processes, whose causal connections can be so multifarious and incalculable that even experts run into doctrinal arguments about the effectiveness of regulative interventions in networks such as these. How can a whole national economy be developed successfully? And who actually does push up the gross national product (17b), or manage to bring about higher turnover for a whole industry (18b)? We do not know for sure, but we nevertheless hope for growth. Another case of naturally vague knowledge can be seen in predictions about the future. As a matter of fact, the German corpus (notwithstanding a few exceptions) shows a clear tendency towards the use of the GROWTH metaphor in economic forecasts (cf. 17b, 19b, 20). However, no such tendency could be established for English usage.

304

OLAF JÄKEL

2.2.2 ACTIVE VERTICAL MOVEMENT

Besides GROWTH, which is not exactly a prototypical kind of movement, there are some other metaphorical conceptualizations of active, unaided movement along the vertical: ECONOMIC DEVELOPMENT IS ACTIVE VERTICAL MOVEMENT. Sums of money (22, 23, 24a, 25a) rise and climb, and even more unfathomable entities such as inflation (24b) or the whole market (25b) toil up that strenuous ascent: (22) a. Prices rose by more than 10% per annum, b. Die Preise erhöhten sich zum Teil drastisch. (23) a. Prices have gone up again. b. Die Baupreise sind gestiegen und steigen weiter. (24) a. The project's estimated cost has climbed to a staggering £35 billion, b. ... die auf rund 250 Prozent gekletterte Inflation. (25) a. Costs are moving upwards. b. Der Markt tue sich auf seinem Weg nach oben schwer ... Things are getting athletic when these entities, which must be conceived of as living beings, have a go at jumps (26, 27), or perform upswings (28): (26) a. Their profits jumped from £3.5 million to £22 million in a single year. b. Mit einem Kurssprung um zwei Pfennig reagierte der Dollar ... (27) a. Im ... Bereich Rohbau habe sich ein Kostensprung von 18 Prozent ergeben, b. He blamed the leap in oil prices for our present economic problems. (28) a. The sudden upswing in share prices ... b. Andererseits fehlen die für einen raschen deutlichen Kurs­ aufschwung erforderlichen positiven Meldungen. That those jumps in examples (26, 27) must be regarded as high jumps and not as horizontal long jumps is due to the general VERTICAL conceptualization of differences between sums of money that was introduced above (cf. Section 2.2). These 'athletic' expressions are used to denote upward movements in

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

305

economic affairs that either extend relatively far (26, 27) or that are extremely fast (28), and sometimes even movement that both extend far and are fast at the same time (26a, 28b). Sometimes monetary values are not only conceptualized as living beings, but they seem to take on birdlike qualities, as in example (29): (29) a. They watched the value of their shares soar from less than £50 to £87. b. XYZ-Aktien fielen nach ihrem vorübergehenden Höhenflug um 10 auf 410 Mark zurück. Monetary values may also have the dynamic potential for motion along the vertical that is typically associated with inanimate things such as spring mecha­ nisms (30a), projectiles (30b), rockets (30c), or explosives (31): (30) a. Im Konzern schnellte die Bilanzsumme um 13,5 Prozent auf 283,3 Milliarden Mark hoch. b. In a single year the inflation rate shot up from 30% to 48%.  The price of sugar has suddenly rocketed up. (31) a. Leipziger Unternehmen ... befürchten eine Mietenexplosion. b. ... an explosion in demand for telephones. The expressions in these examples (30, 31) all denote upward movements which happen fast and suddenly. The metaphorical model ACTIVE VERTICAL MOVEMENT not only comprises UPWARD MOVEMENT, but also DOWNWARD MOVEMENT. Before going DOWNWARD, however, economic entities may also remain on a particular level: (32) a. Bei ausgeglichenen Nachfrage- und Angebotsverhältnissen konn­ ten sich die Preise auf dem erreichten Niveau behaupten. b. House prices seem to have reached a plateau ...  Inflation has begun to level off. But even this holding out at a particular level requires some struggle with gravity, which inevitably prevails in the end: The entities lose height, as quite a number of examples (33)-(38a) show. One way this DOWNWARD MOVEMENT can be expressed is as a kind of slumping (33) or sinking (34):

306

OLAF JÄKEL

(33) a. Während im Vorjahr die Umsätze zwar ... anstiegen, sackten die Ergebnisse auf einen Konzernverlust von 8,9 Millionen Mark ... ab. b. Profits last year slumped from $ 40 million to $ 26 million. (34) a. Die Umsätze sanken auf den tiefsten Stand seit zwei Monaten, b. The pound has sunk to its lowest level against the dollar.  The Bank of England took action to prevent the pound from sinking any further. Then there is falling motion (35, 36), which may be rendered more dramatically by adverbs like sharply (36a) or steeply (36b): (35) a. The value of the dollar has fallen. b. Der Financial-Times-Index ...fiel um 41,8 Punkte auf 1938,3. (36) a. Interest rates fell sharply last week. b. Die durchschnittliche Umlaufrendite fiel steil auf 8,47 Prozent zu­ rück. Furthermore, economic entities can start to slide (37, 38a): (37) a. Profits have slipped slightly this year. b. With prices sliding fast, small computers are becoming popular,  Der Deutsche Aktienindex rutschte wieder leicht unter die 1600Punkte-Marke, bei deren nachhaltiger Überwindung Börsianer mit relativ freier Kursfahrt nach oben rechnen. (38) a. Er warnte vor Verlusten bei der Telekom, durch die die Post insgesamt in die roten Zahlen rutschen könnte b. Your account is in the red. Whereas the German example (38a) opens up another 'danger zone' (cf. 9) in which the DOWNWARD MOVEMENT might terminate, the English example (38b) uses the red {numbers) only for the static localization of a financial balance without mention of a sliding into that area.12 In German, after slipping and sliding, the economy can also fall into step again: (39) Nach der schärfsten Wirtschaftskrise ...faßt die ... Wirtschaft wieder Tritt.

MOTION M E T A P H O R I Z E D IN THE ECONOMIC DOMAIN

307

This expression (39) as well as the sliding movement in itself and the whole of example (37c) already suggest that the metaphorical UPWARD and DOWNWARD MOVEMENTS are not pictured on an abstract measure or scale, but are located in a particular conceptual 'landscape', which we will inspect closely below (in Section 2.3). In contrast to the GROWTH metaphor, this model of ACTIVE VERTICAL MOVEMENT focuses not just on the distance covered (e.g. 22a, 26, 27a), but it can also set a temporary endpoint of movement (24, 29, 30, 33, 34, 35b, 36b) or other salient landmarks (32, 37c, 38) on the metaphorical path. Our comparative study of the English and German language material reveals only a marginal difference: while the English corpus evidence frequently combines starting point and end point of movement within one syntagm (26a, 29a, 30b, 33b), the German evidence favors the combination of distance and goal within one sentence (29b, 30a, 35b). Furthermore, in both languages, the ACTIVE VERTICAL MOVEMENT model permits the description of processes and activities of reified abstract entities, thus avoiding the need to ascribe responsibility (for economic developments) to causal relationships or human actors. Unlike the vague GROWTH model, though, ACTIVE VERTICAL MOVEMENT is capable of the highest precision in the numeri­ cal quantification of economic developments. For that reason, it will hardly ever be found in forecasts, but very commonly in reports (cf. examples (22)(38), where, with only few exceptions, verbs are in past tense), in which exact­ ness concerning numerical facts is possible and desired. 2.2.3 PASSIVE VERTICAL TRANSPORT

Unlike HORIZONTAL MOVEMENT (cf. Section 2.1), VERTICAL MOVEMENT in the economic domain can be conceptualized as PASSIVE in English as well as in German: ECONOMIC DEVELOPMENT IS PASSIVE VERTICAL TRANSPORT.13 Sums of money and exchange rates, those reified abstract entities already familiar to us, do not have to surmount the differences in metaphorical altitude on their own resources. Instead, they are given transport by other forces lifting them upward (40)-(45a) or bringing them down again (45a)-(47). To start with, UPWARD TRANSPORT may proceed by means of lifting (40), raising (41), or putting up (42):

308

OLAF JÄKEL

(40) a. Firma X will die Preise in diesem Jahr um bis zu 20 Prozent anheben. b. They expect that this will lift the rate of inflation back to 13 per cent. (41) a. A country in deficit abroad should raise its interest rates to attract foreign money, b. ... wegen der Mehrwertsteuererhöhung ... (42) a. Einem Berliner Automechanikermeister setzte ... die Reichsbahn die jährliche Grundstücksmiete von 560 auf 186732 Mark herauf. b. They've put the price up. While in English, UPWARD TRANSPORT may be lexicalized by a simple (i.e. transparent) bringing up (43a), the German equivalent steigern (43b) can only be related to the same metaphorical model via its etymology: it has its origin in the Middle High German causative steigen "make climb, raise". (43) a. The Equal Pay Act has failed to bring women's earnings up to the same level, b. Firma X hat den Umsatz 1990 um 5,6 Prozent auf 101,4 Millionen Mark gesteigert. can also proceed in a rougher manner, as both languages make use of the expression push up (44):

UPWARD TRANSPORT

(44) a. Die Meldungen ... trieben auch die Notierung der XYZ-Titel in die Höhe. b. This oil boom can only push the basic inflation rate up to higher levels. The metaphorical model PASSIVE VERTICAL TRANSPORT also motivates various options for the DOWNWARD TRANSPORT of sums of money or exchange rates. These entities are lowered (45, 46a), brought down (46b), or again in a rougher manner, pushed down (47): (45) a. Während einige Großröster ihre Fabrikabgabepreise erhöhen, senkt Firma X die Preise, b. They've lowered the price from £15 to £10.

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

309

(46) a. W., der eine Senkung der deutschen Leitzinsen nach wie vor ablehnt, b. The good harvest brought down the price of strawberries. (47) a. International confidence in the mark has eroded, pushing it down more than 25 per cent against the dollar, b. Die Niederlage der CDU ... hat am Montag den Kurs der Mark gedrückt. The actors held responsible for all these TRANSPORT activities are the persons directly in charge (45b, 46a), conceptual personifications of companies (40a, 42, 43b, 45a), or other abstract entities (41a, 43a, 44, 46b, 47). It may be note­ worthy that the more violent methods of 'transport' (44, 47) are ascribed neither to real persons nor to personified companies, but only to impersonal abstract entities such as certain news reports (44a), an oil boom (44b), an ero­ sion of confidence (47a), or the election defeat of a political party (47b). In focusing on the achievement of these actors/carriers, the model of PASSIVE VERTICAL TRANSPORT on the one hand allows the attribution of causal responsibility (42-45, 46b, 47) for economic processes. On the other hand, it serves to express certain intentions to control particular economic develop­ ments (40, 46a). The language comparison reveals a general congruence of English and German with regard to this motion metaphor. 2.3

MOUNTAINSIDE-TOPOLOGY

To conclude this study of motion metaphors in the economic domain, an essential component of metaphorical orientation in space must be discussed, one without which the conceptualization of economic affairs could not quite be accounted for: the MOUNTAINSIDE-TOPOLOGY. The following English example (48) names, with its peaks and troughs, the main features of a complex meta­ phorical 'mountainscape', whose further details will be explored below: (48) The business cycle is a series of peaks and troughs. Against this background, economic development can be conceptualized meta­ phorically as a kind of mountaineering activity: ECONOMIC DEVELOPMENT IS MOUNTAINEERING/MOVEMENT ON A MOUNTAINSIDE.

The lowest point in this MOUNTAINSIDE-TOPOLOGY is called Talsohle "valley floor" or rock bottom (49). A negative development of the economy is

310

OLAF JÀKEL

described as a downward movement towards this lower region; e.g., the econ­ omy is sliding down the slippery slope (50a) or going downhill (50b): (49) a. Allerdings scheine das Unternehmen die Talsohle erreicht zu haben, b. They bought their house when prices were at rock bottom. (50) a. The economy is steadily sliding down the slippery slope. b. ... doch sei bisher nicht damit zu rechnen, daß ... die Talfahrt der ostdeutschen Wirtschaft schon zu Ende ist. Afer reaching the bottom of the valley, the journey can lead upwards again (51), with the rise being characterized as more or less steep (52): (51) a. Auch die Chemieindustrie kommt erst 1992 aus ihrem Ertragstal heraus. b. The price of oil bottomed out at S 12 a barrel and has now started to rise again. (52) a. Der steile Anstieg des US-Dollar ... b. ... a steep rise in prices. An enduring climb goes uphill (53a) to the highest heights (53b), reaching proper peaks (54). As in real physical moutaineering, these peaks can be chal­ lenged and surpassed (54b); after that, the direction of movement is again more or less steeply downhill (54c): (53) a. Nach Einschätzung der deutschen Pelzbranche geht es bei den Geschäften mit Fellen wieder bergauf. b. Your pay will never rise to the same heights. (54) a. Sales have reached a new peak. b. Market peaks were quickly challenged and despite some brief hesitation, the previous closing peak ... and then the trading peak were surpassed.  Der Stand von 2722 am 25. August 1987 war der vorläufige Gipfel. Wenig später ging es mit dem Crash vom 19. Oktober 1987 steil bergab. Apart from peaks and troughs with the respective uphill and downhill move­ ments, the economic MOUNTAINSIDE-TOPOLOGY features real plateaus (55), on

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

311

which mountaineering entities such as prices can remain a while before continuing on their way up: (55) House prices seem to have reached a plateau, but they may start rising again soon. The remaining examples confirm that the MOUNTAINSIDE-TOPOLOGY is really conceptualized as a very concrete mountainscape, which features not only peaks, valleys, and plateaus, but even overhangs, caves, and caverns. Thus, economically fatal collapses (56a) may occur where the ground does not stay firm (56b). And it is equally dangerous when parts of the mountain range crumble (57a-b) and drop off (57c): (56) a. ... wegen des Ertragseinbruchs im 1. Quartal 1990 ... b. The pound stayed firm against the dollar in London but fell a little in New York. (57) a. ... obgleich die höheren Anfangsnotierungen großenteils abgebröckelt sind. b. The upswing soon crumbled. c. Sales have dropped off this winter.

im

Verlauf

How can we account for the clearly negative denotation of Talsohle "valley floor" (49a) or Talfahrt "downhill drive" (50b) and for the positive denotation of bergauf "uphill" (53 a), both of which are so commonly found in the economic context? Indeed, reaching the bottom of a valley for a mountain hiker or going downhill for a tobogganist can hardly be called negative; nor is walk­ ing uphill a cause for mounting joy. I believe that a satisfactory languageinternal explanation is not in the offing. Instead, one useful explanation supported by the corpus phenomenology seems to lie in the hypothesis that the curve diagrams commonly used in economic presentations have been hypostatized to form a conceptual MOUNTAINSIDE-TOPOLOGY. In these more or less schematic curve diagrams, as shown in Figure 1, the horizontal axis represents the units of time, while the vertical axis shows the sums of money or quotation points.14 Though, initially, graphs like these were meant to simply illustrate trends in economic development, their jagged 'curves' easily tempt us into taking them for more than what they really are, and interpreting them as a mountain range. Graphic designers have long taken notice of this overinterpretation, and are toying with it, as Figure 2 shows.15

312

OLAF JÄKEL

Figure 1 : Curve diagram with Talfahrt

This hypostatizing would be the reason for the development of a conceptual metaphor in which the schematic MOUNTAINSIDE-TOPOLOGY serves to locate many of the VERTICAL MOVEMENTS described above. This metaphor motivates such expressions as trough (48)—(51), peak (48, 53b, 54), steeply uphill (53a), and downhill (54c). It is as a result of this metaphor introducing a quasi-concrete mountain range that even Einbrechen "breaking through" (56a) or crumbling off (57) have become regular expressions. Furthermore, this conceptual metaphor can be seen as providing a backdrop to the motivation of sentences (13)—(16) as well as (22)-(47) exemplifying VERTICAL MOVEMENT. Though itself not a motion metaphor, the MOUNTAINSIDE-TOPOLOGY thus supplies a firm basis for the metaphorization of motion in the economic domain.

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

313

Figure 2: Hypostatized representation reflected in a curve diagram: Mountain eering p ercentages

3.

Summary and conclusion

Before we turn to a discussion of probable reasons for the metaphorical models established above, a conclusion summarizing the most important results of our investigation is due. By means of a linguistic corpus study, a number of conceptual metaphors of motion in the economic domain were analyzed. These metaphorical models shape much of the discourse on economic issues both in English and in German. The principal function of these motion metaphors was seen to lie in the dynamic quantification of economic processes, with alternative models focusing on different aspects. The main result of this comparative study is that English as well as German economic discourse are determined to a large extent by the same meta­ phorical models. Most of the conceptual metaphors that I studied commonly exist in both languages, based on a shared core inventory of metaphorical expressions. While the models may have individual realizations in either

314

OLAF JÄKEL

English or German for which the other language lacks an equivalent, such differences are clearly secondary to the similarities both in linguistic expres­ sions and, more importantly, in the basic conceptualizations. One of the most marked differences noted was that in German economic discourse, metaphorical HORIZONTAL MOVEMENT is only realized as an ACTIVE one, whereas English has linguistic expressions instantiating ACTIVE HORIZONTAL MOVEMENT as well as PASSIVE HORIZONTAL TRANSPORT. Further­ more, a number of usage tendencies could be established which are specific to one of the two languages, e.g. the German preference for GROWTH metaphors to express economic forecasts. I will start the last part of this conclusion with a loose remark on the frequency, and therefore also on the explanatory value, of the economic meta­ phors treated in this investigation. It seems pretty safe to state that there exist no newspaper reports on economic issues of more than three sentences in length which do not contain instances of at least one of these most important conceptual metaphors. Many articles in the corpus material investigated abound with the linguistic expressions in question. Furthermore, it seems equally safe to state that the metaphorical models presented above account for approxi­ mately 90 percent of all linguistic metaphors to be found in economic reports. What remains is a tiny fraction of metaphorical expressions which are idiosyn­ cratic and do not display any systematicity. Why, then, are metaphors of motion so prominent in economic discourse? One could come up with theoretically possible alternatives, e.g. metaphorical conceptualizations of ECONOMIC DEVELOPMENT as PUTTING ON WEIGHT, R as PAINTING A PICTURE, DOING NEEDLEWORK, r LEARNING SOMETHING NEW. Yet apparently none of these have been selected by economic journalists/writers. What makes the general conceptualization of ECONOMIC DEVELOPMENT as MOVEMENT and its various specifications discussed above the preferred choice? There are probably a number of reasons. First of all, each motion meta­ phor employed is grounded in some of the most central of image schemata. As I mentioned above, the model of horizontal motion is based on the PATH schema, while the various models of vertical motion are based on the UP-DOWN schema of VERTICAL ORIENTATION. Not only are these two schemata pervasive in our everyday experience and therefore familiar to everybody, but they are at the same time both rich in structure and simple enough to supply the expe­ riential grounding for an abstract target domain such as the economic domain.16 Moreover, both schemata may be said to be experientially correlated with basic economic issues. Having more of any substance or objects evokes the UP-DOWN

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

315

schema in the shape of a rising pile. And competing with others in reaching for shared goals evokes the PATH schema. The theoretical alternatives mentioned above lack such schematic grounding. They are either too abstract and not at all schematic, such as LEARNING SOMETHING NEW; and if schemata are involved, they may be too poor in structure, such as PUTTING ON WEIGHT, or much too complicated and not pervasive, such as PAINTING PICTURES or DOING NEEDLEWORK. In addition to their being so well-grounded in image schemata, the meta­ phorical motion models discussed above provide enough variety to concep­ tualize different aspects of economic development, always with a strong element of dynamic quantification. It was shown in Section 2 how the meta­ phors of HORIZONTAL MOVEMENT, ORGANIC GROWTH, ACTIVE VERTICAL MOVEMENT, and PASSIVE VERTICAL TRANSPORT are each employed with a different focus. They give speakers every freedom to be just as precise or as vague as intended, to highlight or hide the responsibility of particular actors, and to give reports or make forecasts concerning economic development. It is this combination of firm schematic grounding, direct experiential correlation between schematic structures in the source domain and basic target domain issues, and their versatility of metaphorical focus that probably accounts for the fact that motion metaphors figure so prominently in the economic domain. Notes 1 Cf. Lakoff and Johnson (1980), Johnson (1987), and Lakoff (1987, 1990, 1993). For a comprehensive discussion of the cognitive approach to metaphor, see Jäkel (1997). 2 The notion of model is used here in the sense of Lakoff s (1987:68 passim) idealized cognitive models (ICMs), which may be taken to represent folk theories of certain domains (cf. Lakoff 1987:118). For a more comprehensive account of metaphors of economy beyond mere motion metaphors, see the case study in Jäkel (1997:205-245). 3 This study goes back to my own research on German metaphors of economy; see Jäkel (1993, 1994). The results of other recent investigations into metaphors of economy found in English as well as in other European languages are very much in keeping with mine. Thus, most of the linguistic material from economic reports presented in Hennet and Gil (1992) for Spanish, in Schmitt (1988) for French, as well as in Hübler (1989), Langfeldt (1992), and Boers (1995) for English can be understood as evidence for the conceptual metaphors presented and discussed below on the basis of English and German language data. The reasons for slight deviations of analysis lie in the individual hermeneutics of interpretation as well as in different theories of metaphor that are applied, and with different facets of economy focused upon. Both Hübler (1989) and Langfeldt (1992), for example, confine themselves to reports on the stock market. On the other hand, Boers (1995) treats the target domains of economy and general politics together in one go without caring to differentiate. In general, though, it seems

316

OLAF JÄKEL

to be the case that, with only minor exceptions, in English, German, French, and Spanish, the same metaphorical source domains are employed in the description of economic processes. 4 The following analysis is a translation of the German account given in Jäkel (1997:224-241). 5 The term image schema is used to designate recurrent dynamic patterns of our sensorimotor bodily experience. These non-propositional schemata constitute the preconceptual basis of all our conceptualizations; see Johnson (1987:XIV) and Lakoff (1987:267) as well as Jäkel (1997:29-30, 286-292). For discussions of the PATH schema in particular, see Lakoff (1987:275ff.), Johnson (1987:113ff.), and Radden (1989:230f., 1995). 6 Here and in the following sections, the term PASSIVE does not refer to syntactic voice, but to the distribution of conceptual or semantic roles; as such, PASSIVE is similar to the PATIENT of Case Grammar; cf. Radden (1985). 7 The unacceptable passive examples (10b, 11b, and 12d) were formed as semantic equivalents to the active German sentences (8a, 9a, and lb) in order to exclude other thinkable, but irrele­ vant reasons for their unacceptability as far as possible. 8 Example (11) may also contain an additional vertical aspect. However, this is irrelevant for the particular difference between German and English discussed here. 9 Compare the evidence provided for the PERSONIFICATION model and the MACHINE model in the economic domain in Jäkel (1997:211-223). 10 This schema is also referred to as the UP-DOWN-schema (Lakoff 1987:275) or the UPDOWN-orientation (cf. Krzeszowski 1993:321). For a general discussion of this schema, see Lakoff and Johnson (1980:14ff.). 11 For an analysis of the ontological metaphor ECONOMY IS A PLANT, see Jäkel (1997:214216). 12 If its historical origin is taken into account, the metaphorical expression in die roten Zahlen geraten "be in the red" has a metonymic motivation. Traditionally, merchants wrote their defi­ cit budgets in red ink, and profits in black. The metonymy in its pure form can still be seen in the German expression rote Zahlen schreiben "making a loss". In our example (38), though, we find metaphorical motion as well as metaphorical orientations of in/out plus vertical. 13 As in Section 2.1 above, the term PASSIVE is meant to denote not a syntactic, but a concep­ tual category (cf. note 6). 14 This general custom can be regarded as a non-linguistic instance of the metaphorical VERTICAL ORIENTATION.

15 Figures 1 and 2 have also been taken from the business section of the Hamburger Abendblatt. A comparative analysis of linguistic expressions and graphic representations, both treated as instances of conceptual metaphors, would make a worthwhile and comprehensive semiotic investigation. A sample of further graphic representations of economic metaphors from the Abendblatt corpus can be found in Jäkel (1997:302-307). 16 Cf. the profiles given for optimal schemata in Lakoff (1987:278) as well as in Johnson (1987:116). The simplicity of the schemata employed can also be detected in the graphic reali­ zations of the MOUNTAINSIDE-TOPOLOGY that were presented above (in Section 2.3).

References Boers, Frank. 1995. Crafty Creatures on the Move: A Few Metaphorical Models in Economic Discourse. Universität Duisburg: L.A.U.D.  30.

MOTION METAPHORIZED IN THE ECONOMIC DOMAIN

317

Hennet, Heidi & Alberto Gil. 1992. "Kreative und konventionelle Metaphern in der spanischen Wirtschaftssprache der Tagespresse". Lebende Sprachen: Zeitschrift für fremde Sprachen in Wissenschaft und Praxis 1/92.30-32. Hübler, Axel. 1989. On Metaphors Related to the Stock Market: Who Lives by Them? Universität Duisburg: L.A.U.D.  19. Jäkel, Olaf. 1993. 'Economic Growth' versus 'Pushing up the GNP': Meta­ phors of Quantity from the Economic Domain. Universität Duisburg: L.A.U.D.  24. . 1994. "'Wirtschaftswachstum' oder 'Wir steigern das Bruttosozial­ produkt': Quantitäts-Metaphern aus der Ökonomie-Domäne". Unter­ nehmenskommunikation: Linguistische Analysen und Beschreibungen ed. by Theo Bungarten, 84-101. Tostedt: Attikon. . 1997. Metaphern in abstrakten Diskurs-Domänen: Eine kognitiv­ linguistische Untersuchung anhand der Bereiche Geistestätigkeit, Wirtschaft und Wissenschaft. Frankfurt a.M: Peter Lang. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason, Chicago, 111.: The University of Chicago Press. Krzeszowski, Tomasz. 1993. "The Axiological Parameter in Preconceptual Image Schemata". Conceptualizations and Mental Processing in Language ed. by Richard A. Geiger & Brygida Rudzka-Ostyn, 307-329. Berlin: Mouton de Gruyter. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago, 111.: The University of Chicago Press. . 1990. "The Invariance Hypothesis: Is Abstract Reason Based on Image-Schemas?" Cognitive Linguistics 1.39-74. — . 1993. "The Contemporary Theory of Metaphor". Metaphor and Thought, 2nd edition ed. by Andrew Ortony, 202-251. Cambridge: Cambridge University Press. & Mark Johnson. 1980. Metaphors We Live By. Chicago, I11.: The University of Chicago Press. Langfeldt, Birgit. 1992. "Metaphern in der Wirtschaftssprache: Eine Unter­ suchung anhand von Artikeln aus der FINANCIAL TIMES mit Schwer­ punkt Börse". M.A. thesis, University of Hamburg. Radden, Günter. 1985. "Looking Back at Case Grammar". LT.L. Review of Applied Linguistics 67/68.185-199. . 1989. "Das Bewegungskonzept: 'to come' und 'to go'". Raumkonzepte in Verstehensprozessen: Interdisziplinäre Beiträge zu Sprache

318

OLAF JÄKEL

und Raum ed. by Christopher Habel, Michael Herweg & Klaus Rehkämper, 228-248. Tübingen: Max Niemeyer. . 1995. "Motion Metaphorized: The Case of: 'Coming' and 'Going'". Cognitive Linguistics in the Redwoods: The Expansion of a New Para­ digm in Linguistics ed. by Eugene H. Casad, 423-458. Berlin: Mouton de Gruyter. Sclimitt, Christian. 1988. "Gemeinsprache und Fachsprache im heutigen Fran­ zösisch: Formen und Funktionen der Metaphorik in wirtschafts­ sprachlichen Texten". Fachsprachen in der Romania ed. by Hartwig Kalverkämper, 113-129. Tübingen: Gunter Narr.

SECTION 4 MOTIVATION IN APPLIED LINGUISTICS

This final set of papers deals with various aspects of motivation in the domain of Applied Linguistics. Willis Edmonson's and Juliane House's paper exam­ ines, among other things, what factors motivate long-term language learning, and in particular the learning of English as a Foreign Language inside the German secondary school system. Edmondson and House's data seem to support an 'instrumental relevance hypothesis', which leads them to the ques­ tion whether teaching programs might become more effective by reducing or indeed eliminating cultural content from the English language classes, in favor of a focus on intercultural awareness, and ELF-oriented communicative strate­ gies. Meike Sprenger-Tasch looks at the attitudinal factors determining the choice of the language of instruction in the Ugandan school system. In particular, she concentrates on attitudes among the Ugandan population in general, and among teachers in particular, towards Luganda, Kiswahili, English, and mother tongue as media of instruction. The results show (i) that the role of English remains unchallenged, and (ii) that Kiswahili is gaining acceptance as a medium of instruction and as such may present an alternative to the highly controversial Luganda. Wolfgang Hünig's paper, finally, pres­ ents a comparative study of the use of style labels in learner's dictionaries. As such, it is less obviously concerned with issues of motivation. Still, learner's dictionaries can be an important, and therefore potentially motivating, factor in English as a Foreign Language if learners are made sufficiently aware that style labels should only be seen as hints, which have to be fine-tuned in different social and discourse contexts.

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

WILLIS EDMONDSON & JULIANE HOUSE University of Hamburg

My mother was of the sky My father was of the earth But I am of the universe And you know what it's worth. (Lennon & McCartney) The paper argues that the emergence of English as a lingua franca (ELF) should impact on the goals and content of English-as-a-Foreign-Language teaching programs, specifically inside the German secondary school system. More particularly, it is suggested that the role of English literature teaching, and the teaching of English Cultural Studies need to be radically re­ assessed. Intercultural awareness regarding the use of English is put forward as a more relevant goal. The argumentation is supported by empirical findings derived from the analysis of ELF interaction, and the contents of language learning autobiographies produced by students reporting on their school experience of English language teaching programs.

1.

Introduction

Over sixty years ago, Edward Sapir (1933:169) wrote: The future alone will tell whether the logical advantages and theoretical necessity of an international language can overcome the largely symbolic opposition which it has to meet. In any event it is at least conceivable that one of the great national languages of modern times, such as English or Spanish or Russian, may in due course find itself in the position of a de facto international language, without any conscious attempt having been made to put it there.

At the end of this millennium it is clear that the 'great national language', which is de facto an international or global language, is English. A very large part of international communication today is conducted in a lingua franca, and

322

WILLIS EDMONDSON & JULIANE HOUSE

the majority of lingua franca interactions worldwide take place in English. Over the past decades, English has consistently grown in international importance and is now established in areas such as politics, business negotiations, cultural and scientific congresses, workshops and symposia, international student meetings, and wherever members of smaller, 'exotic', 'less commonly taught' languages have occasion to talk to members of larger language communities, who see no reason to acquire, and converse in, the 'smaller language'. English has further been spread by unprecedented developments in information and communication technologies, and by the recent economic developments towards globalization and internationalization. For example, although in business mergers an incoming company may not have its head­ quarters in an English-speaking country, joint ventures typically adopt English as their working language, which promotes a local need for English (Graddol 1997). It is against this background that we wish to contribute to the ongoing debate as to what impact the development of English as a lingua franca (ELF) should have on the goals and content of English-as-a-Foreign-Language (EFL) teaching programs inside educational systems. We shall be focusing specifi­ cally on English language teaching in German secondary schools, but our remarks are of sufficient generality to be relevant to the teaching of English in other educational contexts. The observation that school teaching programs, specifically in Germany, need to take account of the changing role of English in the world is not new: it is voiced inter alia by Hüllen (1982), Knapp (1987), and Beneke (1991). The critical issue raised by the spread of English as lingua franca is, it seems to us, that of 'cultural content', i.e. how far and in what ways the specifically English cultural load traditionally bound up with English teaching programs, and made tangible in the relevance of cultural studies, Landeskunde, and literature, continues to be relevant, and in what sense.1 As any instance of the use of ELF will necessarily constitute an instance of intercultural communication, we are necessarily also concerned with the field of intercultural communication, inside which the issue of cultural orientation or norms is necessarily raised once more. The question as to what consequences, if any, are to be drawn for the teaching of English inside the German school from the spread of ELF is of course an issue of some complexity. We shall seek to bring forward relevant empirical data, without however wishing to claim thereby that empirical evidence alone can decide issues of school language policy.

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

323

The paper is in four parts. The starting point is an investigation of the nature of communication between persons of different non-English-speaking linguistico-cultural backgrounds who communicate in English as a lingua franca. Our brief review of the literature, together with some further data culled from ongoing research, leads to establishing features of ELF communication, which make it distinct from and independent of native-speaker behavioral norms. ELF-communication is, we repeat, by definition intercultural communi­ cation. Given this, we turn in Section 3 of the paper from the practice of ELF communication to the nature of the English language training transmitted in schools. Here, we consider data taken from students' language learning autobi­ ographies, and report on these subjects' views on and experience of learning and using English inside and outside of school settings. The critical questions focused upon are whether and how far English is perceived by such learners as a means of ELF communication, and whether and how far the teaching and training received is seen as relevant to ELF. The role of cultural content— cultural studies, the study of literary texts—is also addressed. The data inside this section of the paper supports an 'instrumental relevance hypothesis', which posits perceived functional use as the engine driving the effectively motivated learning of English. This perceived functional use is moreover overwhelmingly focused on English as a lingua franca. We bring Sections 2 and 3 together in Section 4, in which we put forward what we provocatively term the culturalirrelevance hypothesis. It says that the rich cultural values commonly perceived to be embedded in the English language are essentially of no relevance in the context of the ELF communication and acquisition. We therefore go on to ask whether teaching programs might become more effective by reducing or indeed eliminating cultural content from the English language classes, in favor of a focus on intercultural awareness, and ELF-oriented communicative strategies. A brief resumée follows this section and concludes the paper. 2.

English as lingua franca in intercultural communication

Interactions in English as lingua franca can be defined as interactions in English between members of two or more different languages and cultures, for none of whom English is the mother tongue. It is interesting to ask whether and in what ways such interactions differ from interactions both between English native speakers, and between native speakers and non-native speakers of English. Very few empirical studies have been conducted in this field. We will review these, before presenting some of our own data.

324

WILLIS EDMONDSON & JULIANE HOUSE

Meierkord (1996) conducted an empirical study of ELF interactions, in which she audiotaped dinner table conversations in a student dormitory in Great Britain. In her analysis of this data, Meierkord, relying largely on Edmondson's (1981) categorial framework for the analysis of spoken discourse as well as on the continuation of this work in Blum-Kulka, House and Kasper (1989), concentrates on analyzing routinization in opening and closing phases, gambits, topic management, politeness, requests, and misunderstandings. Her findings include the following: a.

b. 

d. e.

There are surprisingly few misunderstandings/communication break­ downs. Meierkord accounts for this by claiming that her ELF data reflects a highly consensual interactional style, largely achieved through heavy backchanneling and the use of interpersonally active gambits such as appealers. The few misunderstandings that do occur are not overcome by negotiations, but rather bypassed by topic changes. Misunderstandings/communication breakdowns seem not to be caused by interactants' lack of English linguistic competence, but rather by gaps in their knowledge of the world, in particular knowl­ edge of the subject being addressed in talk. ELF interactants have a noticeably limited repertoire of tokens at their disposal, especially for use in ritualized phases of ELF talk. Transfer and concomitant interference from L1 interactional norms is almost completely absent.

Another project examining ELF talk is Firth's and Wagner's work on interna­ tional business communication, in which telephone conversations between employees of different Danish companies and their foreign trade partners were audiotaped and transcribed (Firth 1996a, 1996b; Wagner & Firth 1997). The analysis is based on conversation analysis, supplemented by an examination of relevant written documents, and ethnographic information collected at the workplace settings. Some of the findings of this project relate to the 'fleeting' nature of ELF talk, i.e. the fluidity of norms, which in turn reflect participants' insecurity regarding which norms are operative, and interactants' explicit, often strained attempts at conversational attuning, which—on the surface—result in cooperative, supportive, and consensus-oriented conver-sational behavior. Wagner and Firth (1997) further find in their data a clear tendency on the part of ELF interactants to seek to 'normalize' potential trouble sources by skating over them, rather than attending to them explicitly, via repair initiation, refor-

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

325

mulation, or other negotiating behaviors. As long as a certain threshold of understanding is achieved, such that current conversational purposes can be met, ELF participants appear to adopt a principle of 'Let-it-Pass'. This is a major characteristic of ELF interactions in the Firth/Wagner data. It can, however, the authors suggest, endanger effective communication: the superfi­ cial mutual intelligibility established in ELF interactions may break down in communicative contexts requiring precise understanding. Thus, in intercultural business transactions, a general, relatively superficial consensus may mask deeper sources of trouble, as when for example details discussed have to be acted upon. These findings from the literature suggest then that in ELF commu­ nication, -

overt misunderstandings scarcely occur; repair is scarcely initiated and achieved, rather is potential unclarity passed over ('Let-it-Pass'); in technical or business communication, superficial 'robustness' may disguise the fact that the discourse outcomes arrived at are impre­ cise, if not nebulous.

The ELF data for the small empirical pilot study to be reported on here is consistent with these findings. It is taken from a larger corpus of ELF, EFL, and baseline data which forms part of ongoing Ph.D. work being conducted by Agnes Lesnyak at the University of Hamburg. The analysis offered here is exploratory in nature, and is to be understood as a preliminary attempt at char­ acterizing the nature of this type of talk. In other words, no claims will be made for the representativeness of the sample used, nor for the exhaustiveness of the analysis offered (see Lesnyak, forthcoming). The data used here was elicited during an international student confer­ ence in the Netherlands, in which a simulation formed part of the conference program. The task given to participants in the simulation was to decide which of fifteen European states qualify on various criteria to take part in the third phase of the European Economic and Monetary Union. Participants included speakers of Dutch, Hungarian, and German. The group discussion was taped and transcribed. The data is quasi-authentic and was elicited via participant observation. As reported by Lesnyak, and as can be gleaned from the tapes, participants appeared very quickly to forget that a taped simulation was being enacted.

326

WILLIS EDMONDSON & JULIANE HOUSE

A 30-minute segment of this data was analyzed by one of us (House). Lesnyak's transcripts have been simplified for the purposes of the present analysis.2 The analysis conducted followed the analytic model of spoken discourse provided by Edmondson (1981) and Edmondson and House (1981), as well as the categorial framework developed by Blum-Kulka, House and Kasper (1989). The analysis focused on misunderstandings, following previous work by House (1993a, 1993b, 1996a, 1997, 1998, 2000). These earlier studies demonstrate that talk between members of different linguo-cultural communities is rife with intercultural miscommunication, deriving from differ­ ent culturally-conditioned knowledge schemata. All of House's previous studies were based, however, on interactions between native and non-native speakers. The current analysis was therefore conducted inter alia to find out if it was also the case that differences in culture-conditioned schemata caused misunderstandings in ELF. Further, this small empirical pilot study was moti­ vated by the desire to find out whether Meierkord's and Firth's and Wagner's findings were in fact confirmed by Lesnyak's data. Consider Dl: Dl (1) d1 : (2) g: (3) d2: (4) d5: (5) g: (6) d1: (7) g: (8) d4: (9) d5 : (10) d l : (11) d5: (12) h2: (13) d l : (14) h2: (15) h2: (16) d6: (17) dl: (18) (19) (20) (21)

d2: h2: dl : h2:

What do we say about Germany I mean there's some Germany should be in (?) no question about that at the do we(?) [vocalizations trying to turn-take] we all agree the yeah I but we have to talk about that like if you have a debt yeah yeh okay and it's (?) the right direction yeah but if you look at France for example it's the okay but(?) same phase that's the same phase because (?) so and it's just the question of time yeah but it's still (?) okay wait (?) maybe it's smart and (?) useful now to discuss if there is a trend towards the there's (?) one direction of the debt no but no but in general (?) because you've also mentioned France but (?) we can also say of Germany...

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

327

We can see in this short sequence that there are no overt communication break­ downs. Rather there seem to be minor 'local routine disruptions' (e.g. in lines 2, 4, 7, 13, 19). These are possibly recognized as such at some low level of awareness by the participants, and are thus in theory locally repairable. Repairs, however, are not initiated by interactants at all in this data stretch. Apparently, repairs are not felt to be necessary—the interactants seem to be attuned towards achieving a level of 'normality' and consensus. In the 30-minute sequence of talk studied in detail no 'on record' repairables or misunderstandings are to be found. Putting this another way, even if there are communicative inadequacies in this ELF interaction, these are never foregrounded, but rather passed over. The talk is 'robust', to use Firth's (1996b) term; much is taken for granted, much is just left as it occurs, in the belief that meanings will be revealed in due course. The findings reported on by Meierkord, Firth, and Wagner are therefore upheld by the data looked at, and exemplified in D1. The contrast with the results of House's studies on intercultural misunderstanding between native and non-native speakers using English is marked. The above features seem therefore to derive from the fact that we are handling ELF communication. In ELF talk, interactants are not oriented towards some real or imaginary L2 norms, nor is there any evidence of participants' attempts to impose their own culturally-conditioned discourse norms and values. The interactants arrive at a superficial consensus: diplomatic and optimistic let-it-pass behavior cloaks misunderstandings, and appears to compensate for the absence of a common culture in the sense of a Hofstedian 'collective programming of the mind'. Levinas's theory of conversation is of possible relevance in this connection, in particular his notion of the conversationalist as self-oriented speaker (Levinas 1961, cf. Jaszczolt 1994). A further possibly useful frame of reference in seek­ ing to come to grips with these features of ELF is Taj fel's (1981) social theory. On this view, persons align themselves along a continuum of interpersonal and group identities, whereby considerations of group influence and group stereo­ typing are set off against interpersonal and individual feelings and behavior. Inside this theory, then, we might posit a focus on interpersonal and individual concerns rather than on group identity in this ELF data. However, this opposi­ tion is too simplistic: the group identity dimension is realized in participants' willingness to allow other participants equal individual freedom. Instead of acting as speakers of Hungarian, German, or Dutch using a foreign language, interactants are individual ELF speakers belonging to something we may call 'the community of ELF speakers'. For this community, base-rules may operate, but their applicability and interpretation have to be constituted anew in ongoing

328

WILLIS EDMONDSON & JULIANE HOUSE

talk. Using Hüllen's (1992) distinction between language as means of identifi­ cation and language as a means of communication, we might suggest that ELF interactants are using English as a means of communicating, without neces­ sarily identifying with English as a cultural symbol. ELF, we may say, does not provide a cultural identity, though it may, presumably, be used by the indi­ vidual to establish his or her cultural identity in ELF communication. Before leaving this section of the paper, let us return briefly to the data reproduced in D1. The fact that this talk does not exhibit misunderstanding, is not characterized by repair sequences, is, in fact, 'robust', does not of course in itself imply or entail that the interaction can be characterized as smoothly coherent and communicatively effective. We have already suggested that the interactants seem frequently to be engaged in non-aligned, 'parallel talk'; they are 'playing their own game', without seeming to overtly register what their interlocutors are actually saying. Briefly, we may suggest that the talk exhibits inefficient turn-taking, relatively short turns, most of which fail to carry discourse 'weight', and little mutual orientation, affiliation, or reciprocal facework. The speakers, in a nutshell, do not exhibit advanced pragmatic fluency, in the sense of House (1996b). Clearly, this may be a matter of inadequate exposure, learning or teaching, and we shall raise such issues at the end of the paper. Here, we wish to suggest that pragmatic fluency, embracing issues such as face-work, turn-change, and mutual alignment, is a relevant parameter in all communication, and constitutes further a 'problem' for non-native speakers regardless of who they might be talking to; that is, the issue is not specific to ELF discourse. Further, the features which do distinguish ELF talk do not in themselves constitute a different-order 'pragmatic fluency'. Thus, 'letting things pass' may help surmount difficulties of organizing turns at talk, but does not in itself constitute a convention for organizing turns at talk. 3.

English as a foreign language in the context of school instruction

We turn in this section of the paper from the communicative use of English in lingua franca settings to the instructed acquisition of English as a 'foreign language' in schools. The general purpose of this switch of focus is to seek to establish in what sense and to what degree the goals and procedures commonly found in teaching practice map on to the ways in which English is used as lingua franca outside of school settings. We do not wish, however, to claim a priori that a one-to-one mapping should necessarily obtain. Whether and in what sense this might be the case will be discussed in Section 4 of the paper. Here, we seek to address the descriptive issue—what, if any, seems to be the

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

329

impact of the status of English as lingua franca in the world on English language teaching and learning in schools. To attempt to provide an adequate answer with some generality of appli­ cation would be a daunting task. It might for example involve analyzing cur­ rently operative Teaching Guidelines {Richtlinien) for different types of school in different parts of Germany, analyzing current teacher training programs, commonly used English textbooks, and so forth. Further, widespread classroom observation would be called for, in order to seek to establish how far the didac­ tic practices encouraged or required in guidelines, training programs, or prescribed texts are actually followed, and how far the syllabus goals laid out in these various sources are in fact achieved. Instead of pursuing such a macro­ scopic documentation of didactic goals and practices, we shall here take a microscopic perspective, by seeking insight into the goals and practices of English language instruction, as these have been experienced by the consumers themselves, i.e. products of the school system. Our data is then subjective data, in the sense that it consists essentially of subjects' self-reports: the focus is on learners' subjective theories, in the sense of Groeben et al. (1988) (on the role of subjective theories in language learning research, see Grotjahn 1991 and Kallenbach 1996; on the validity and relevance of the data used here for language learning research in general, see Edmondson 1996, 1997a). Inside an obligatory introductory course, students of Sprachlehr­ forschung "Language Learning/Teaching Research" at the University of Hamburg are required to write a 'foreign language learning autobiography' in their first term of study. The brief is to record what has influenced their language learning positively, and/or negatively, and to try to say why. It is stressed that there are no correct answers, and that the resultant texts will be handled anonymously (an original identification is, however, exacted, as a safeguard against trivial responses). The resulting statements (with the authors' names deleted) are actually used for teaching purposes, and also—given the authors' consent—incorporated into an annotated data bank, which is the data source for this small study.3 In an earlier paper (Edmondson 1996), a central finding that emerged from a preliminary analysis of the data then available was that the major source of positive long-term motivation is the realisation that the English language is needed for purposes outside the classroom. The general point is that English is as it were perceived not as something 'foreign', but as directly relevant for the learners, both for their current interests and needs in Germany and also as a World Language. (Edmondson 1996:461-464). The paper proposed finally that this relevance might be exploited didactically: "as regards English, learners

330

WILLIS EDMONDSON & JULIANE HOUSE

may be prepared better to accept the rigors and indeed the inbuilt unevenness of classroom teaching, if they perceive a use for English outside the classroom for themselves here and now" (1996:464). In general, this finding remains valid, and therefore justifies an 'instru­ mental relevance hypothesis', which posits that perceived practical relevance for life outside of the classroom constitutes the major motive for English language learning. Such a hypothesis impinges therefore on motivational theory. The hypothesis requires, however, a more differentiated analysis, as the nature of the perceived 'needs outside the classroom' referred to differs from learner to learner, and equally obviously differs for the individual learner over time. When we consider that English instruction inside the German school system commonly extends over nine years(!), then clearly attitudes to, and experience of, the English language is unlikely to remain a constant. Amongst other things, we may expect increasing life experience and cognitive matu­ ration, together with developing L2 competence and relative learning success to influence learning set, drive, or orientation. Some data will be brought forward, then, to illustrate such inter-individual and intra-individual variation. The analysis will show that ELF-oriented purposes are in fact the most important long-term goals positively affecting attitude and motivation. Not surprisingly, at the time of writing their learning autobiographies, almost all students acting as subjects are well aware of the relevance of English for general professional, educational, and communicative purposes—"a bit of English never harmed anybody", as one student put it (see D4 below). Of inter­ est for the purposes of this paper is how far and in what forms this awareness was present and pedagogically exploited when English was being learnt at school. Before turning to the data, however, it is necessary to suggest a theo­ retical framework inside which the 'instrumental relevance hypothesis' can be placed. In other words, we need a working theory of motivation. The term 'instrumental relevance' evokes the original distinction made by Gardner and Lambert (1972) between an integrative and an instrumental motivation. Since that time, two major developmental steps have been taken inside research into motivation and L2 acquisition. Firstly, as more studies were carried out, the original distinction made by Gardner and Lambert was relativized and placed in a much more broadly-based concept of motivation, which embraced inter alia both instrumental and integrative 'orientations' or goals, willingness to invest time and energy towards attaining those goals, and attitudes towards language learning in general and the target language in particular. The theory developed in Gardner (1985) captures one influential result of this develop-

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

331

ment. Such a theory can be refined ad infinitum, as newer and more refined statistical analyses of questionnaire material can always discover factors or bundles of factors which carry heavier weighting in specific learning circumstances. The second major development has been to incorporate classroom-specific motivational factors into such models, and to consider further 'motivation' not as something static, located in a learner's cognitive and affective disposition, but as a variable interacting inter alia with learning experience, and the nature of any contact with the target language and/or its culture (Skehan 1991; Crookes & Schmidt 1991; Oxford & Shearin 1994). 'Motivation' has then, following this development, been coupled with 'motivating', for example via pedagogic strategy (Heckhausen 1989; Dörnyei 1994). On this view, 'motivation' becomes a highly complex phenomenon, containing both relatively stable elements, and more unstable factors, some of which may be personality-bound, others of which are contextually-determined (cf. Heckhausen 1989:424). On the one hand, therefore, it is to be borne in mind that 'instrumental relevance' is only one possible factor inside the motivational complex. On the other hand, it is self-evident that the case of English as opposed to other foreign languages taught at school is a special one, precisely because of the status of English as a world language. Contact with English is to all intents and purposes nowadays almost unavoidable for young persons of secondary school age in Germany, and in many other countries also, both inside and outside of Europe. Students commonly claim that they were positively oriented towards English as a subject before encountering it in school. Time and time again, subjects refer to looking forward to learning the language, though this initial enthusiasm wears off over the first two or three years (cf. Solmecke 1976:10, in which the same pattern was noted). Consider the data extracts in D2 below:4 D2 a.

b.



Certainly, by the time we were old enough to attend Primary school, listening to pop music—usually English pop music—on the radio became more and more a matter of course. We loudly joined in singing our favorites, more or less inaccu­ rately, and without having at the time any idea whatsoever as to the meaning. [SS9802] When I started to learn English in grade 5, I was highly motivated and actively involved. Now I would be able to learn the language at the core of the culture of young people of my generation!... My motivation sank to an alltime low, however, in the following years (grades 6 through 10). [SS9525] I was looking forward to learning English in grade 5 at secondary school. This pleasurable anticipation was however rapidly dispelled by my very first English teacher, who I can remember all too well to this day. [WS9314]

332

WILLIS EDMONDSON & JULIANE HOUSE

d.

e.

I started learning English, like almost everybody else, in grade 5 at secondary school. I was highly motivated at that time, as my father subscribed to Time maga­ zine, and more importantly National Geographic, and I was very keen to learn to read the exciting articles on my own. [SS9707] At the age of 13, I started with a second foreign language—English. As I knew many words and expressions already from everyday contact, I set out to learn English with enthusiasm, looking forward to being able to express myself in this language as well. [SS9706]

The grounds for a diminishing or disappearance of an initial enthusiasm for English are complex. Three contributory factors are: -

-

the development of different interests, outside of media-oriented youth culture; the discovery that learning a foreign language actually involves a good deal of effort and persistence, and not simply role-plays and conversational routines ('fun and games'); exposure to demotivating/demotivated teachers, and/or uninspiring teaching materials.

Some variations on these themes are indicated in the data segments collated in D3: D3 a.

b.



At first, English was a lot of fun, but as time went on, the language lost its novelty and attraction, and I was not interested in English music or literature, which might have enabled me to rediscover my initial enthusiasm. [WS9516] At this time I confused the English language and American foreign policy, and because of this, my teacher's good intentions were totally ineffective. I did not dislike this teacher, and her lessons weren't too bad either, but I just did not want to learn English! [WS9516] ...I was just indifferent to the English language—I really wanted to learn French... Now, I'm learning Italian, a language I have always wanted to learn... My own motivation to learn a language is much higher than when someone forces me to learn it... As English is a World Language, and I'm constantly confronted with the language, I'm forced to improve my proficiency by attending further courses. [SS9902]

The data in (a) suggests that an initial excitement and interest based on novelty— which can characterize first contacts with any foreign language and not just English, of course—is likely to diminish over time. It further estab­ lished the obvious point that the English-language mediated youth culture

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

333

prevalent on disk, television, and on the Internet is not to everybody's taste. The data in (b) is of interest, in that a negative cultural attitude apparently blocked any school-internal motivational drive, while (c) suggests that attempts to impose learning imperatives can be counter-productive. This student seems in fact to be arguing that the fact of instrumental relevance can lead to affective resentment, and negative attitudes. Note, however, that this student continues to learn English, despite negative attitudes: in this sense, the instrumental motive seems in fact to be effective. In the case study presented in truncated form in D4 below, the student had no initial sense of curiosity when confronted with English, and had appar­ ently little or no contact with English music or other cultural artefacts.5 Conse­ quently, he or she was unable to see any point to learning the language, and consequently did so less than effectively: D4 I started learning a foreign language in 1986 in grade 5. As might be expected, my first foreign language was English. And I think that right from the beginning, my teachers made a mistake which is in large part responsible for my current—totally inadequate— mastery of English. The point being that nobody told us at that time why we were supposed to learn the language. ... Nobody thought it was necessary to explain to us what English is and what one can do with it. ...by 1992 I had in fact realized that "a bit of English never harmed anybody". But by this time it was too late—we were wrapped up in different interpretations of literature, instead of grammar and vocabulary, i.e. learning the language. And this was no use to me at all. We didn't have language lessons any more, just literature lessons, and as I already knew for sure that I certainly wouldn't go on to study English at university, I was back to the situation I was faced with in 1986—I was learning stuff, without knowing why. [WS9705]

What emerges from such data as this is that the perception of instrumental rele­ vance is indeed most effective when it derived from the recognition that English has a broad range of potential applications outside of the classroom. A further source of positive orientation, and therefore another aspect of the instrumental relevance hypothesis, is the communicative experience of using English in the course of school trips, exchange visits, tourist travels, or during a longer time spent in an English-speaking environment (the au-pair syndrome). Such experiences can be viewed as more 'integrative' than 'instrumental', or rather can be seen in terms of a 'friendship' or a 'travel' orientation, which may accompany or supplement a more obviously 'instrumental' orientation (see the discussion of 'friendship' and 'travel' orientations in Belmechri & Hummel 1998). A detailed analysis of students' attitudes to and experience of 'the stay abroad' is conducted in Edmondson (2000). It transpires, firstly, that it is not

334

WILLIS EDMONDSON & JULIANE HOUSE

always the case that the stay abroad has positive affective/motivational conse­ quences. Secondly, when this does in fact happen, the gains are 'instrumentalized', in that the contact is evaluated highly in large part because it translates into better marks and easier learning success upon returning to school. In other instances, the consequence of the stay abroad is not increased motivation for school learning, but either dismissal of or at best simply tolerance of the language teaching program, which is seen to be at odds with the learners' own communicative experience (for details, see Edmondson 2000). In terms of effective orientation, we posit a motivational hierarchy, map­ ping on to a widening domain of relevance (Figure 1). The first circle embraces initial excitement caused by language learning experience in a classroom, and covers also the 'fun-and-games' syndrome, which is often characteristic of the early stages of communicatively-oriented textbooks for younger school learners. Secondly, we move into localized cultural experience —the overlap between English language use and current hobbies, habits, sources of excite­ ment, and friendships (Iggy Pop, pen friends, trips to Bath or Birmingham). This type of orientation is then extended to the recognition that English is the major world language, and that investments in learning it will probably pay off in the future. On the basis of the data examined, we claim it is overwhelmingly the case that the third level orientation is most effective in terms of long-term learning motivation.

Figure 1: Learning orientation: The widening circle

Surprising is the fact that the data gives very little support for an intrinsic interest in and positive attitude towards British/American culture (the lure of the language, the sweep of the syllables, the hipness of Hemingway, or the bounty of Blair), as a motivational factor. There are, for example, only five positive references to the excitement of reading English works of literature in the original in the data, to be balanced against some fifteen negative references to classroom literary experience (cf. D4 above), and, further, circa forty posi­ tive references to English pop music. It is plausible to suggest that the learning diaries of university students studying Anglistik "English Language and

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

335

Literature" might reveal different profiling in this respect, but we have no evidence to suggest this would be the case.6 Even were this so, however, it would be necessary to point out that English language teaching in schools should not be primarily catering for learners who will go on to be students of English language and literature at university. 4.

Discussion

The term 'intercultural communication', of which ELF communication is a special but primordial instance, obliges us as a matter of logical necessity to say that the cultural norms associated with or indeed embedded in the English language are not valid for such discourse. The claim is supported by the nature of ELF communication, as examined in Section 2 of the paper above. Section 3 of the paper has further produced evidence to suggest that the major long-term motive for learning English has little or nothing to do with the culture or cultures associated with that language. Further, there was some evidence that a focus on 'high' culture in the form of the teaching of English literature for example is perceived by learners as unrelated to the goal of international communicative proficiency. On these bases, then, we put forward a 'culturalirrelevance' hypothesis, which says that the rich cultural values commonly seen to lie at the core of the English language, and claimed, moreover, to be an inherent part of what is to be taught, are irrelevant for the practice of English as a lingua franca. How far, and in what sense, such cultural values might none­ theless be relevant inside English language teaching programs is the question to be addressed now. As intimated at the beginning of Section 3 of this paper, linking the nature of ELF communication to the content and goals of EFL teaching in schools is a complex undertaking. We have assumed that there should be a relationship. It is, however, naive to assume that the former can be simply taken over inside the latter, such that we can derive a syllabus from a descrip­ tion of the use of English as a lingua franca. Such a shopping-list concept of syllabus design was in fact quite widespread in the early days of functional communicative language teaching, and Edmondson and House (1981) may well have contributed to it. However, there is no necessary one-to-one mapping between a target behavioral specification, and a specification of the means whereby such targets are to be achieved didactically (Widdowson 1983). Thus, there may be a role for activities such as dictation in language learning, even though the taking of dictation in L2 is neither a forseeable nor a sensible activ­ ity for the learners concerned. We accept this, of course (cf. the distinction

336

WILLIS EDMONDSON & JULIANE HOUSE

between 'external' and 'internal' curricula in Edmondson & House 1993:294307). Thus, one might for example postulate a linear progression in terms of goals, attitudes and motives, as captured in Figure 1, and seek to match didactic goals and materials onto such a developmental pattern, thereby justifying, for example, the early 'fun-and-games' approach, and, indeed, the use of literary narrative texts at some juncture in teaching programs. Therefore, while we believe the cultural-irrelevance hypothesis holds, it is not a ground for excluding British or American cultural content in a teaching program (assuming for the sake of the argument that this would, in fact, be pos­ sible). For example, to deliberately avoid contact with stimulating and engag­ ing narrative texts, simply because such materials might be seen as highly laden culturally, would be totally absurd. However, that one would not actively avoid English/American cultural content does not imply that one should actively seek to transmit, present, or 'teach' it in a systematic fashion. We wish to argue that the role of English as lingua franca and indeed the increasing use and status of the English language inside German language norms has radical consequences for the goals and content of secondary school English teaching programs. We shall focus on two specific points, involving consideration of intercultural goals, and the validity of native-speaker norms respectively.7 Precisely what is involved in intercultural language teaching is less than clear (cf. Edmondson & House 1998). Here, we confine ourselves to the literal meaning of the term, whereby an 'intercultural' standpoint is one which lies as it were, between L1 and L2 (cf., e.g., Kramsch 1993). In setting up intercultural goals for language teaching therefore, the comparison of L1 and L2 cultural norms is of importance—and indeed many different cultures and languages may be involved, given the multicultural composition of most classrooms. It is worth stressing here too that intercultural competence is not simply something that non-native speakers should aspire to: on the contrary, there are good sociological reasons for stressing the desirability of intercultural skills on the part of the native speaker. What needs to be stressed here is, then, that English language classes in Germany can usefully pay more attention to the role of English in Europe, and other parts of the world, and different norms of usage in different contexts.8 The point being, of course, that the British or American cultural inheritance (and how much time has been spent in the past musing over the interpretation of that or!) is of only peripheral or accidental importance for the purpose of developing a reasonable basis for proficiency in English. The second and related issue to be raised is the status of native speaker norms in general. For ELF purposes, native-speaker competence is simply an inappropriate goal. Native-speaker norms (assuming we think we know what

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

337

they are) are by definition not appropriate for non-native speakers! Indeed, the very term 'non-native speaker' reeks of a deficiency hypothesis. Thinking on this issue in the context of English language teaching is at times lacking in clarity. Thus, the widespread claim that native speakers make better language teachers by virtue of their having a mastery of the target language that a nonnative can never match is odd, as it both presupposes that non-natives will not master some aspects of L2, and at the same time suggests that teachers are required who will in fact teach precisely these aspects of the target language.9 The same logic is evidenced in the 'flog-a-dead-horse' didactic syndrome, which says roughly that the more evidence accrues that trying to get one's learners to fully master some difficult feature of L2 (for example, present perfect versus past simple tense for English) will be unsuccessful, the more time and effort teachers put into teaching precisely this feature. Advanced learners can be quite aware of this problem of unattainable norms: D5: In general I always get totally demotivated, when I realize—as I do from time to time— that what I have to learn via intensive detailed study is mastered absolutely effortlessly by native speakers. There is for example a distinction in English between present perfect and past simple. Every English teacher knows how hard it is to teach a speaker of German when to use the one, and when to use the other. ... When I just imagine how many boring sentences exhibiting the use of these two forms we had to work through, just to avoid really blatant mistakes, and when I then realize that every child with L1 English acquires all this and much more, together with amongst other things a much better pronunciation as well, then at times I can see no sense in learning foreign languages in the classroom at all (although I learnt enough to be able to start studying Sprachlehrforschung, I suppose!) [WS9813]

To deny the relevance of native-speaker standards does not of course take us far, unless something is put in place in their stead. It is not the case that all interlanguages are equal. Some will be more equal than others. General purpose English language programs in secondary schools clearly have to center on some norm or target. Though Crystal (1994:113) claims that moves are afoot to establish a 'World Standard English', the issue is an extremely complex one. We can do no more than offer some rough guidelines, which in practice may well be commonly accepted. One will distinguish for example between different linguistic domains, and different degrees of communicative importance. Teaching programs have always done this implicitly, in selecting syllabus contents, deciding thereby to teach some things, but not others. There is further, of course, an important distinction to be made between receptive and productive skills. Again, we may

338

WILLIS EDMONDSON & JULIANE HOUSE

distinguish between norms and standards, between awareness and prescription. In other words, for some domains (for example in syntax or pronunciation), one may teach certain structures and a 'standard pronunciation', but sensibly accept some communicatively viable approximation to or indeed deviation from these norms. The distinction between awareness and prescription can be further illustrated by reference to the remarks on pragmatic fluency offered in Section 2 of this paper. Here, we see an important role for an awareness of cultural (and individual) variation in terms of socio-pragmatic norms. The issues of turntaking and face-work are critical in terms of coherent and effective discourse communication: the norms holding for English are, however, socio-cultural norms, and thus both culture-specific, and indeed open to abuse, flouting, and strategic exploitation inside the linguo-culture where these norms operate. From a pedagogic perspective, it is clearly important to aim at increasing ELF interactants' behavioral repertoire, and to empower them to try out different, 'smoother', more partner-oriented and potentially more successful interactional styles than those illustrated in D1 above. In order to survive in the intercultural give and take of ELF oral interaction, it may well be that learners of English need to be taught a number of 'survival skills' such as repair strategies, communication strategies (in particular compensatory strategies), turn-gaining, turn-taking and turn-keeping strategies, and a general attitude of interpretative openness and flexibility. 5.

Summary

We have shown in this paper that in intercultural interactions between young people using English as lingua franca, various cooperative strategies are employed to avoid and/or overcome potential misunderstandings. The different cultural backgrounds of participants do not seem to offer difficulties of communication, and there is no convergence on cultural or social norms deriv­ able from English-native-speaking cultures. We have further shown that consumers of English language teaching programs inside the German secon­ dary school system develop, in the course of their language instruction, an awareness of the practical utility of English as a lingua franca, which contrib­ utes positively to long-term language learning motivation, despite a hiatus between the perceived instrumental relevance of English and the nature of school instruction. We then suggest that teaching programs should be geared precisely towards cultural openness, and the use of appropriate interpretive and negotiation strategies necessary for successful ELF communication. In the context of these goals, we suggest that the teaching of 'Culture', as embodied

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

339

in the appreciation of literary texts, and gaining insight into other cultural aspects of a country or countries where English is L1 can have no central role to play. The point is not that young people should be discouraged from enjoy­ ing reading English literature, if they wish to do so, or discouraged from aspir­ ing to as unmarked an English accent as possible. The point is rather that these things are peripheral to the central target of English language teaching, which is to equip the customers for international communication in English over as wide a range of relevant topics as possible, encompassing an appropriate variety of relevant skills. Notes 1

Edmondson (1991, 1997b) has already argued that foreign language literature has no specific role to play in foreign language teaching. 2 d, g, h = Dutch, German, and Hungarian native speakers respectively; (?) = difficult to iden­ tify; overlap and turn-taking is indicated through spatial display; [ ] = description of non-verbal actions. 3 This data bank currently holds annotated extracts from some 360 texts, comprising ca. 35,000 words. 4 Extracts from learning biographies will be presented in translation in the body of the paper; the original German texts are given in an Appendix at the end. The translations are functional; i.e. they aim at giving the sense of the original only. 5 It is possible that this student attended schools in what was East Germany prior to German reunification. 6 Some students of Sprachlehrforschung are also students of Anglistik, but we have no data on how high this percentage might be. 7 Cf. the arguments and opinions in Gnutzmann (1997); in this article, Gnutzmann focuses on the educational role of Language Awareness, but his conclusions regarding the goals and content of English language instruction converge with our own. 8 Though work on intercultural communication has practically ignored ELF interactions, a number of socio-political and critical discourse studies have interpreted the spread of English throughout the world as linguistic colonialization, or worse (e.g. Pennycook 1993; Fairclough 1995; Phillipson & Skutnabb-Kangas 1996; Phillipson 1998). Others strike a more balanced note, arguing in favor of a world English, while not supporting the imposition of non-linguistic cultural norms from the English-native-speaking world (cf., e.g., Y. Kachru 1985; B. Kachru 1992, 1997; Crystal 1997; Widdowson 1997). 9 Of interest here is the discussion in the pages of Anglistik, the official newsletter of the German Association of University Teachers of English {Anglistenverband) initiated by Bonheim (1998). In that publication, Bonheim showed that that German university teachers of English often have deficient English language skills. In a later contribution to the ensuing discussion (Bonheim 1999), the author further claims that native-speaker lecturers—particu­ larly such as have an EFL qualification—also often do not speak good English.

340

WILLIS EDMONDSON & JULIANE HOUSE

References Beimechri, Faïza & Kirsten Hummel. 1998. "Orientations and Motivation in the Acquisition of English as a Second Language among High School Students in Quebec City". Language Learning 48.219-244. Beneke, Jürgen. 1991. "Englisch als lingua franca oder als Medium interkul­ tureller Kommunikation". Grenzenloses Sprachenlernen ed. by R. Grebing, 54-66. Berlin: Cornelsen. Blum-Kulka, Shoshana, Juliane House & Gabriele Kasper, eds. 1989. CrossCultural Pragmatics: Requests and Apologies. Norwood, N.J.: Ablex. Bonheim, Helmut. 1998. "Problems of English in German Letters of Recom­ mendation". Anglistik 9.119-123. . 1999. "A Reply". Anglistik 10.234-235. Crookes, Graham & Richard W. Schmidt. 1991. "Motivation: Reopening the Research Agenda". Language Learning 41.469-512. Crystal, David. 1997. English as a Global Language. Cambridge: Cambridge University Press. . . 1994. "Which English - or English Which?". Who Owns English? ed. by Mike Hayhoe & Stephen Parker, 108-114. Buckingham: Open University Press. Dörnyei, Zoltán. 1994. "Motivation and Motivating in the Foreign Language Classroom". Modern Language Journal 78.273-284. Edmondson, Willis. 1981. Spoken Discourse: A Model for Analysis, London: Longman. . 1991. "Sind literarische Texte für den fremdsprachlichen LehrLernprozeß besonders geeignet?" Texte im Fremdsprachenunterricht als Forschungsgegenstand ed. by Karl-Richard Bausch, Herbert Christ & Hans-Jürgen Krumm, 53-60. Bochum: Brockmeyer. — 1996. "Subjective Theories of Second Language Acquisition", Anglistentag 1995 Greifswald: Proceedings ed. by Jürgen Klein & Dirk Vanderbeke, 453-464. Tübingen: Max Niemeyer. . 1997a. "Sprachlernbewußtheit und Motivation beim Fremdsprachen­ lernen". Fremdsprachen Lehren und Lernen 26.88-110. — —. 1997b. "The Role of Literature in Foreign Language Learning and Teaching: Some Valid Assumptions and Invalid Arguments". AILA Review 12.42-55.

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

341

. 2000. "Grammatik und Kommunikation bei Auslandsaufenthalte". Sprachlehrforschung im Wandel: Beiträge zur Erforschung des Lehrens und Lernens von Fremdsprachen - Festschrift für Karl-Richard Bausch zum 60. Geburtstag ed. by Beate Helbig, Karin Kleppin & Frank Könings, 363-382. Tübingen: Stauffenburg. & Juliane House. 1981. Let's Talk and Talk about It: A Pedagogic Interactional Grammar of English. München: Urban & Schwarzenberg. & Juliane House. 1993. Einführung in die Sprachlehrforschung. (= UTB, 1697.) Tübingen: Francke. & Juliane House. 1998. "Interkulturelles Lernen: Ein überflüssiger Begriff". Zeitschrift für Fremdsprachenforschung 9.161-188. Fairclough, Norman. 1995. Critical Discourse Analysis. London: Longman. Firth, Alan. 1996a. "The Discursive Accomplishment of Normality: On 'Lingua Franca' English and Conversation Analysis". Journal of Prag­ matics 26.237-260. . 1996b. "Averting Miscommunication: On 'Robustness' and 'Fra­ gility' in Spoken Discourse". Paper given at the ALLA Congress 1996, Jyväskylä. Gardner, Robert. 1985. Social Psychology and Second Language Acquisition: The Role of Attitudes and Motivation. London: Arnold. & William E. Lambert. 1972. Attitudes and Motivation in Second Language Learning. Rowley, Mass.: Newbury House. Gnutzmann, Claus. 1997. "Normen im Englischunterricht: Zur Rolle des Sprachbewußtseins ('Language Awareness') und seiner kognitiven, sozialen und affektiven Vernetzung". Zeitschrift für Fremdsprachen­ forschung 8.213-226. Graddol, David. 1997. The Future of English? London: The British Council. Groeben, Norbert, Diethelm Wahl, Jörg Schlee & Brigitte Scheele. 1988. Forschungsprogramm Subjektive Theorien: Eine Einführung in die Psy­ chologie des reflexiven Subjekts. Tübingen: Francke. Grotjahn, Rüdiger. 1991. "The Research Programme Subjective Theories: A New Approach in Second Language Research". Studies in Second Language Acquisition 13.187-214. Heckhausen, Heinz. 1989. Motivation und Handeln, 2nd edition. Berlin: Springer.

342

WILLIS EDMONDSON & JULIANE HOUSE

House, Juliane. 1993a. "Toward a Model for the Analysis of Inappropriate Learner Responses". Interlanguage Pragmatics ed. by Gabriele Kasper, Juliane House & Shoshana Blum-Kulka, 163-184. New York, N.Y.: Oxford University Press. -. 1993b. "Mißverstehen im interkulturellen Diskurs". Kontroversen in der Fremdsprachenforschung ed. by Johannes Timm & Helmut Vollmer, 178-193. Bochum: Brockmeyer. . 1996a. "Contrastive Discourse Analysis and Misunderstanding: The Case of German and English". Contrastive Sociolinguistics ed. by Mariis Hellinger & Ulrich Ammon, 345-361. Berlin: Mouton de Gruyter. . 1996b. "Developing Pragmatic Fluency in English as a Foreign Language: Routines and Metapragmatic Awareness". Studies in Second Language Acquisition 18.225-252. . 1997. "Kommunikative Bewußtheit und Fremdsprachenlernen". Fremdsprachenlehren und Lernen 26.68-87. — . 1998. "Kontrastive Pragmatik und interkulturelle Kompetenz im Fremdsprachenunterricht". Kontrast und Äquivalenz: Beiträge zu Sprach­ vergleich und Übersetzung ed. by Wolfgang Börner & Klaus Vogel, 6288. Tübingen: Gunter Narr. ————. 2000. "Understanding Misunderstanding: A Pragmatic-Discourse Approach to Analysing Mismanaged Rapport in Talk Across Cultures". Managing Rapport Through Talk Across Cultures ed. by Helen SpencerOatey, 145-164. London: Continuum. Hüllen, Werner. 1982. "Teaching a Foreign Language as 'Lingua Franca'". Grazer Linguistische Studien 16.83-88. . 1992. "Identifikationssprachen und Kommunikations sprachen". Zeitschrift für germanistische Linguistik 20.298-317. Jaszczolt, Katarzyna. 1994. "Relevance and Infinity: Implications for Discourse Analysis". Journal of Pragmatics 25.703-722. Kachru, Braj, ed. 1992. The Other Tongue. Urbana, I11.: University of Illinois Press. . 1997. "World Englishes and English-Using Communities". Annual Review of Applied Linguistics 17.66-90. Kachru, Yamuna. 1985. "Discourse Analysis, Non-Native Englishes, and Second Language Acquisition Research". World Englishes 4.223-232. Kallenbach, Christiane. 1996. Subjektive Theorien - Was Schüler und Schülerinnen über Fremdsprachenlernen denken. Tübingen: Gunter Narr.

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

343

Knapp, Karlfried. 1987. "English as an International Lingua Franca and the Teaching of Intercultural Communication". Perspectives on Language in Performance ed. by Wolfgang Lörscher & Rainer Schulze, 1022-1039. Tübingen: Gunter Narr. Kramsch, Claire. 1993. Context and Culture in Language Teaching. Oxford: Oxford University Press. Lesnyak, Agnes, (forthcoming). Untersuchungen zu Besonderheiten der Kom­ munikation im Englischen als Lingua Franca. Ph.D. dissertation, Univer­ sität Hamburg. Levinas, Emmanuel. 1961. Totalite et infini. The Hague: Nijhoff. Meierkord, Christiane. 1996. Englisch als Medium der interkulturellen Kom­ munikation. Frankfurt a.M.: Lang. Oxford, Rebecca L. & Jill Shearin. 1994. "Language Learning Motivation: Expanding the Theoretical Framework". Modern Language Journal 78.12-28. Pennycook, Alastair. 1993. The Cultural Politics of English as an International Language. London: Longman. Philhpson, Robert. 1998. "Globalizing English: Are Linguistic Human Rights an Alternative to Linguistic Imperialism?" Language Sciences 20.111112. & Tove Skutnabb-Kangas. 1996. "English Only Worldwide or Language Ecology?" TESOL Quarterly 30.429-452. Sapir, Edward. 1933. "Language". Encyclopedia of the Social Sciences, Vol. 9, 155-169. New York: Macmillan. Skehan, Peter. 1991. "Individual Differences in Second Language Learning". Sudies in Second Language Acquisition 13.275-298. Solmecke, Gert. 1976. Motivation im Fremdsprachenunterricht. Paderborn: Schöningh. Tajfel, Henri. 1981. Human Groups and Social Categories. Cambridge: Cambridge University Press. Wagner, Johannes & Alan Firth. 1997. "Communication Strategies at Work". Communication Strategies ed. by Gabriele Kasper & Eric Kellerman, 323-344. London: Longman. Widdowson, Henry. 1983. Learning Purpose and Language Use. Oxford: Oxford University Press. 1997. "EIL, ESL, EFL: Global Issues and Local Interests". World Englishes 16.135-146.

344

WILLIS EDMONDSON & JULIANE HOUSE

Appendix D2 a.

b.



d.

e.

D3 a.

b.



Spätestens im Grundschulalter wurde das Hören vornehmlich englischsprachiger Musik aus dem Radio immer selbstverständlicher, die Texte der Lieblingsstücke wurden dabei lautstark (und wenig sinngemäß) mitgesungen, ohne daß zu diesem Zeitpunkt auch nur eine Silbe davon verstanden wurde. [SS9802] Als der Englischunterricht dann in der 5.Klasse begann, war ich höchst motiviert am Unterricht beteiligt. Endlich durfte ich die Sprache lernen, die das Zentrum der Jugend­ kultur meiner Generation darstellte... In den darauf folgenden Jahren (6.-1 O.Klasse) stellte sich bei mir ein Motivationstief ein. [SS9525] Ich freute mich sehr darauf, Englisch in der fünften Klasse am Gymnasium zu erlernen. Diese Freude wurde mir jedoch von meiner ersten Englischlehrerin, an die ich mich auch heute noch besonders gut erinnern kann, genommen. [WS9314] Englisch fing ich an, wie die meisten auch, in der fünften Klasse auf dem Gymnasium, zu lernen. Meine Motivation war damals sehr groß, da mein Vater regelmäßig das Time Magazin, aber besonders auch National Geographic bezog.. und ich brannte darauf, selbst die spannenden Artikel lesen zu können. [SS9707]. Mit 13 Jahren kam dann meine zweite Fremdsprache dazu - Englisch. Viele Wörter, die man im Alltag hört, kannte ich schon, und daher ging ich auch mit großer Erwartung an das Erlernen des Englischen heran, da ich mich auch endlich selbst darin ausdrücken wollte. [SS9706]

Anfanglich machte mir Englisch auch Spaß, aber mit der Zeit verlor Englisch für mich den Reiz des Neuen, und ich hatte auch kein Interesse an englischsprachiger Musik oder Literatur, mittels derer ich wieder ein Interesse an der englischen Sprache hätte finden können. [WS9516] Zu diesem Zeitpunkt begann ich die englische Sprache mit der amerikanischen Außen­ politik zu verwechseln, und alle Anstrengungen seitens meiner Lehrerin schlugen fehl. Ich hatte anonsten nichts gegen die Englischlehrerin, und ihr Unterricht war auch nicht schlecht, aber ich wollte partout kein Englisch lernen. [WS9516] ...die englische Sprache lag mir nicht, ich wollte Französisch lernen... Zur Zeit lerne ich Italienisch, eine Sprache die ich schon immer lernen wollte... Meine Motivation ist viel größer, als wenn mir eine Sprache aufgezwungen wird... Da Englisch eine Weltsprache ist and ich ständig damit konfrontiert werde, bin ich gezwungen, meine Sprachkennt­ nisse durch Kurse zu erweiten. [SS9902]

D4 Mein erster Fremdsprachenunterricht begann 1986, als ich in die 5. Klasse kam. Wie es nicht anders zu erwarten ist, war diese erste Fremdsprache Englisch. Und ich glaube, daß schon damals der Fehler gemacht worden ist, der zum Teil für meine heutigen— eigentlich inakzeptablen—Kenntnisse in dieser Sprache verantwortlich ist. Es wurde uns damals nämlich mit keinem Wort gesagt, wozu wir diese Sprache denn lernen müßten. ... Niemand hat es für nötig befunden, zu erklären, was Englisch ist und was man damit machen kann. ...um 1992 hätte ich wohl eingesehen, daß "ein bißchen

ENGLISH IN THE WORLD AND ENGLISH IN THE SCHOOL

345

Englisch" nicht verkehrt sein kann. Aber da war es auch schon vorbei—statt der Grammatik und den Vokabeln, also dem Erlernen der Sprache, beschäftigten wir und nur noch mit englischer Literatur und diversen Interpretationen. Und das konnte mir nun gar nichts mehr nutzen. Es war das eben kein Sprachunterricht mehr, sondern nur noch Literaturunterricht, und da ich schon damals wußte, daß ich ganz bestimmt nicht Anglistik studieren würde, hatte ich eben wieder den Punkt erreicht, an dem ich schon 1986 angelangt war—lernen, ohne zu wissen, warum. [WS9705] D5 Insgesamt demotiviert es mich immer wieder, wenn ich mir (was von Zeit zu Zeit vor­ kommt) klarmache, daß alles das, was ich in mühsamer Kleinarbeit lerne, von den ent­ sprechenden Muttersprachlern ohne jede Mühe täglich benutzt wird. Z.B. gibt es im Englischen eine genaue Trennung von Simple Past und Present Perfect. Jeder Englisch­ lehrer weiß, wie mühsam es ist, einem Deutschen beizubringen, wann er welche Form zu benutzen hat. ... Wenn ich mir nun vorstelle, wieviele langweilige Beispiel-Sätze wir diskutieren mußten, um die gröbsten Mängel zu vermeiden, und dann kurz überlege, daß jedes Kind mit Englisch als Muttersprache das und noch viel mehr ohne Nachdenken noch präziser hinbekommt, dazu noch mit einer wesentlich besseren Aussprache usw., dann zweifle ich, zumindest zeitweise, am Sinn des Fremdsprachenunterrichts über-haupt. (Aber immerhin reicht es noch, um Sprachlehrforschung zu studieren!) [WS9813]

ATTITUDES TOWARDS LUGANDA, KISWAHILI, ENGLISH, AND MOTHER TONGUE AS MEDIA OF INSTRUCTION IN UGANDA

MEIKE SPRENGER-TASCH ABIT AG, Meerbusch

This paper focuses on attitudes towards language use in schools among the Ugandan population in general, and among teachers. In particular, it concentrates on attitudes towards Luganda, Kiswahili, English, and mother tongue as media of instruction. The data for this study were collected by means of a quantitative survey, along with participant observation and/or informal interviews. The results showed that the role of English remains unchallenged, Many people fear that, if African languags are used as media of instruction, Uganda will lose its status and students will not be able to learn English properly. The data also suggest that Kiswahili is gaining acceptance as a medium of instruction, and as such may present an alternative to the highly controversial Luganda.

1.

Introduction

Today, it is generally acknowledged that a child's mother tongue, and con­ comitantly, the culture it is embedded in, is essential for his/her cognitive and emotional development.1 As such, cultural barriers to language learning may substantially hinder the child's cognitive, emotional, and social development, especialy when the culture of the child and that of the target language differ greatly—as is the case with the cultures of the Ugandan languages and the English culture (see Kiwanuka 1967; Mukama 1986). These facts are readily acknowledged in Uganda, as in many other African countries. However, as Mukama (1989:193) puts it, education is an area which exhibits an abysmal rift between theory and practice with regard to the indigenous languages. In all the memoranda and Education Reports issued between 1927 and 1965 the indigenous languages are given their rightful role as media of instruction in the early years of education while nothing whatsoever is

348

MEIKE SPRENGER-TASCH

done to ensure or facilitate the success of those languages as the most effective means of teaching our pupils subjects and other things.

Instead, English is introduced as a medium of instruction as early as possible, especially in urban areas, where it is usually used as a medium of instruction from the first year of primary education and even in pre-primary education (Matovu & Kiguli 1998). The reasons why many African countries retain their former colonial language are numerous. Maintaining existing power relations, fear of ethnic rivalries, and the lack of technical vocabulary and teaching materials in the indigenous languages are but a few. In examining the highly emotional lan­ guage issue, this paper will focus on attitudes towards language use in schools (particularly in an urban school setting) among the Ugandan population in general, and among teachers. The data were collected in two steps. First, a quantitative survey based on orally administered questionnaires was conducted in various parts of Uganda in 1996. Second, a qualitative study was carried out at a Teacher Training College close to Kampala, Uganda's capital, which included participant observation in the school setting and informal interviews. I would like to stress here that the focus of this paper will be on the quantitative part of the project. It is com­ plemented by qualitative data gathered during informal interviews in the school setting. This qualitative data serves to confirm or explain some of the obser­ vations made and some of the quantitative data obtained. Before being able to proceed to the data analysis, a brief description of the language situation in Uganda, an overview of the Ugandan school system, and a short account of the role of Luganda and Kiswahili are in order. 2.

The language situation in Uganda

Recent linguistic data on Uganda are not widely available. The only large-scale survey, which was carried out in Uganda by Ladefoged, Glick & Griper (1972), is already thirty years old. Even more recent publications draw on this dated material (i.e. Wald 1994; Walusimbi 1994; Mukama & Ndoleriire 1997). Lade­ foged, Glick & Griper (1972:31) managed to trace sixty-three languages and dialects, thirty of which they considered as languages as opposed to dialects. Twenty-four of those classified as languages were considered major languages. Figure 1 shows some of those major Ugandan languages.

MEDIA OF INSTRUCTION IN UGANDA

349

Figure 1 : Some major Ugandan languages

Genetically, Uganda's indigenous languages belong to four different language groups: Bantu, Central Sudanic, Eastern Nilotic and Western Nilotic.2 Accord­ ing to the generally acknowledged classification of Greenberg (1963), these four groups belong to two distinct language families; the Bantu languages belong to the Niger-Kordofanian language family, and the Nilotic and Central Sudanic languages are incorporated in the Nilo-Saharan language family. The south of Uganda has a majority of Bantu speakers; they constitute nearly two thirds of Uganda's population, the largest single nation being that of the Baganda (about 18.8% of the population), who speak Luganda. The north­ east of the country is largely made up of speakers of the Eastern Nilotic lan­ guages Ateso3 and Karamojong, the extreme north-west predominantly has

350

MEIKE SPRENGER-TASCH

speakers of the Central Sudanic languages Lugbara and Madi, and the central region has a majority of speakers of with the Western Nilotic language Lwo (Ladefoged, Glick & Griper 1972:18). In addition to the major indigenous languages of Uganda, two other languages play an important role: English and the East African lingua franca Kiswahili. The role of English as Uganda's official language has never been challenged. In fact, its importance has increased since Uganda's inde­ pendence—in the school system and beyond.4 However, the position of Kiswahili has changed considerably throughout Uganda's history. Since the 1930s, its role in the school system has been insignificant and present attempts to re-integrate Kiswahili into the school curriculum have so far been without much import. Still, Kiswahili is being used as a lingua franca by a substantial number of Ugandans: it functions as a medium of horizontal, interethnic communication among the uneducated in Ugandan society; in contrast, English is reserved for vertical communication and mobility among the educated (Mazrui & Mazrui 1995:91). 3.

Language use in the school system

The first missionaries who arrived in 1877 started to teach in Kiswahili, as traders from the East African coast had already established the language at the Buganda court. Before the turn of the twentieth century, attitudes towards Kiswahili changed: missionaries and traditional rulers, especially the Kabaka (King) of Buganda, developed a dislike of Kiswahili. Two reasons can be cited here: (i) the Baganda feared that their language, Luganda, could lose its supre­ macy; (ii) a number of missionaries claimed that Kiswahili was associated with Islam and that it was a language which belonged to no cultural group and, therefore, was alien to the people (Ssekamwa 1997:130). Aside from their sup­ port of Luganda, the general attitude among a large section of the missionaries, the so-called Livingstonians, was that a child should be taught in his/her mother tongue as long as possible, an attitude which encouraged them to develop Uganda's indigenous languages in order to make them viable media of instruction in primary schools (Mazrui & Zirimu 1978:432). In the 1920s, the political climate worked in favor of an East African Union and a common lingua franca for its proposed member states. Governor Gowers blamed his predecessors for having encouraged Luganda since 1912 and declared that in Uganda's south, Kiswahili should replace Luganda as a medium of instruction in primary schools (Ssekamwa 1997:128-129). The sup­ port of Kiswahili was vehemently opposed by large sections of the missionaries

MEDIA OF INSTRUCTION IN UGANDA

351

and by the kingdom of Buganda. In the 1930s, plans for an East African Union and the establishment of Kiswahili as a medium of instruction were shelved. Instead, the teaching of English was intensified. Until then, English had only been taught in a few schools and only from the fifth year onwards. Unlike in Kenya, where a large white settler community wanted to preserve the dis­ tance between white master and black servant by means of language, there were no such reservations about the use of English in Uganda. For parents as well as pupils, formal education was a means of acquiring literacy, learning a little bit of English, and later becoming clerks in offices. The colonial situation had introduced new values in which the knowledge of English and a clerical job were associated with a high status and success in life. The missionaries' claim that a child should first of all master concepts in his/her indigenous language before learning a foreign language was regarded with suspicion. The parents' suspicion, coupled with the administrative demands for more clerks competent in the English language, eventually forced the missionaries to put more emphasis on English in their schools (Ssekamwa 1997:105). Even stronger emphasis was put on the teaching and use of English after independence, when the new government claimed that English had to take over the role of a national language and help in the task of uniting Ugandans to form a modern nation-state (Mazrui 1967:12). In addition, it has been the policy to use six area languages5 (Ateso/ Akarimojong, Luganda, Lugbara, Lwo, Runyankore/Rukiga, Runyoro/ Rutooro)6 in the first years of primary school up to the fourth year (P4), and to have English take over from P5 to P7 (fifth to seventh year) onward and use it in all types of higher education. The introduction of only six languages has never been unproblematic. This holds especially true for Luganda which has not only been introduced in areas where Luganda is the dominant language. In this respect, Mukama (1989:193) states that in the Luganda educational area, for example, the Jopadhola and Iteso of Tororo have never actually accepted Luganda. The Sebei have also always resented the use of Luganda. And in the north-east, the trial marriage between Ateso and Ngakarimojong has never actually worked. And for the Ik (or Teso) people, Ngakarimojong is as different from their own language as English is. ... In the Western region the Banyarwanda have never accepted Runyankole/Ruhororo; just as the Bakonjo are visibly hostile to Rutooro. 7

In spite of the emphasis on English as a medium of instruction, it has been regular practice in the Ugandan school system to use the mother tongue in the first four years of primary education. It is an interesting fact about the Ugandan

352

MEIKE SPRENGER-TASCH

school system that the official language policy is not always adhered to in practice. Walusimbi (1994:4826) states that "although the medium of instruc­ tion in primary schools is supposed to be the local language, in practice many schools use English or one of the regional languages (Lugbara, Acholi, Ateso, Luganda, Runyoro, or Runyankore) even when the local language is different".8 Not much research seems to have been done on the extent to which English is really used—either in the countryside or in towns. There is, however, a common belief that there is a tendency for town schools to start with English straightaway, while schools in the countryside start with the local language. According to Bukenya (1997:2), some urban schools even insist on the use of English outside class or school boundaries, while in mral areas the use of English is handled in a much more relaxed way, in that, in class, a mixture of the mother tongue and English is often used. After primary eduction, the language of instruction, all over the country, is English. This means that literacy programs are also supposed to be carried out in English; in practice, however, they are usually carried out in the language of the catchment area because these programs are targeted at illiterate people who, obviously, are usually not competent in English. Recently, a new language policy was adopted for primary education {Primary School Syllabus 1997-2000) which, in theory, confirms the use of the mother tongue as the medium of instruction for the first four years and which states that English and Kiswahili are to be taught as subjects. English is to take over as a medium of instruction in Primary 5, while mother tongue and Kiswa­ hili are supposed to be continued as subjects up to the Primary Leaving Exam­ ination, which concludes seven years of primary education. The qualitative study has shown that the syllabus has not been implemented, as teachers have not been at all prepared for the teaching of Kiswahili; in fact, most of the teachers did not know enough Kiswahili to be able to teach it or seemed not to take the plans of the Government seriously.9 Other problems with the intro­ duction of Kiswahili in primary schools are the following: first, it is not clear how Uganda recruits the numbers of Kiswahili teachers needed to cover all the primary schools in Uganda;10 and secondly, it is not obvious where the text­ books will come from. According to officers of the the Curriculum Develop­ ment Center, these problems have up to now not been seriously addressed. Finally, there is the problem of the old rivalry between Luganda and Kiswahili, which may hinder the successful implementation of Kiswahili in the Ugandan school system. To illustrate the problem, the following two sections deal with the respective roles of Luganda and Kiswahili in Uganda.

MEDIA OF INSTRUCTION IN UGANDA

4.

The role of Luganda and Kiswahili in Uganda

4.1

Luganda and the Baganda

353

Due to missionary activities and Buganda's collaboration with the British Protectorate government, the Baganda had many political, educational, social, and economic privileges. Their language enjoyed high prestige, especially among the colonizers but also among some of the Bantu speaking people in the south. Luganda was used officially for administrative purposes all over the protectorate. Ssekamwa (1997: 27) notes that Luganda[,] which by government policy of 1912 had been encouraged to be the official language in Buganda in government business besides English, had tipped the scales because of the large number of people who were speaking it in Buganda and outside. Since it was spoken in the area where all the Protectorate administrative headquarters had been established, many other people beside the Baganda had to be in contact with that language. And since the first African teachers and the British colonial collaborators in terms of chiefs were Baganda, they had helped to spread Luganda outside Buganda, making it a language for both long distance trade and education.

However, attitudes towards Luganda outside Buganda remain controversial. On the one hand, Luganda continues to maintain high prestige among many Ugan­ dans, as it is the language of the political and economic capital of the country— knowing Luganda these days is considered equivalent to having been to the city and having 'made it'. On the other hand, many people reject Luganda because it was imposed on them and because it is connected with the Baganda, who are rejected by many because of their former collaboration and their privileges.11 The situation is not helped by the unequal distribution of education in the various parts of Uganda. As the Baganda have always enjoyed a priviliged status with the protectorate governments, their area has the highest density of schools, with most secondary schools as well as Makarere University being found there. In this respect, it is telling that, in a 1991 study by Mukama, the student body at Makarere's Department of Languages shows the following ethnic distribution: over 60% were Baganda, 30% speakers of the western Bantu languages, and only 5-10%o of the students were speakers of the other Ugandan languages (Mukama 1991:345). This educational and social inequal­ ity makes the language of the Baganda very unpopular. Many fear that the inequality will increase if Luganda gains even greater importance at the expense of other African languages. In sum, then, although Luganda is the indi­ genous language with the largest group of mother tongue speakers and the

354

MEIKE SPRENGER-TASCH

second largest of second language speakers after Kiswahili, it would be prob­ lematic if even more importance was accorded to it in the school system nationwide. 4.2

Kiswahili

In Uganda, Kiswahili does not have the status or prestige it has in the neighboring countries. The comments Mazrui and Mazrui (1995:68) made about Kiswahili in the newly independent Uganda are still valid today. The social prestige which Kiswahili enjoyed in Tanzania, with all the associations of a complex culture and political society, was conspicuously absent in Uganda. Many of those who did speak the language did not speak it well. The Kiswahili of Uganda was indeed basically a language for the workers, functionally specific and non-versatile, and, for those reasons, more limited in scope.

In Uganda, as elsewhere in East Africa, Kiswahili has often been called a 'neutral language' as it is connected with no particular ethnic group. While it is true that it is not the language of any specific ethnic group in Uganda, it is by no means true that it is neutral. If it ever had been neutral, it stopped being so when Kiswahili began to be considered a threat to Luganda. Basically, it was the Baganda and their king (the Kabaka) who had always fought against Kiswahili, fearing to lose their supremacy in Uganda. The problems of the Baganda with Kiswahili intensified in 1966, when the then Prime Minister Milton Obote removed the Kabaka from presidency. Baganda suspects were treated with 'Kiswahili language tests' by the predominantly 'swahiliphile' Northern soldiery in order to to humiliate them even further (Mazrui & Mazrui 1995:80). In addition, Kiswahili was reinforced as the language of the police and the army, especially in the seventies under the dictator Idi Amin, when it became their official language. At that time, the language was also used to discriminate against some groups of people, especially against the Baganda. In the following years, despotism in the army and arbitrary acts of humiliation at roadblocks, and other malpractices, instilled fear and hatred in many Ugandans against Kiswahili. An informant reported that, even today, criminals use Kiswahili when they break into houses at night, even if they are Baganda. On the other hand, Kiswahili was the language of liberation when the Tanzanian army supported Ugandan rebels to dismantle the Idi Amin dictatorship. Besides, Kiswahili has a very strong hold over the north of Uganda. The nonBantu northerners love it "partly because it is not Luganda" (Mazrui & Mazrui 1995:81). Finally, Kiswahili is the language of horizontal communication

MEDIA OF INSTRUCTION IN UGANDA

355

among the lower classes of Ugandan society and the language of trade, especially with neighboring countries like Kenya, Tanzania, and parts of the Congo. Without going into more detail, it will have become clear that neither Luganda nor Kiswahili are neutral. 5.

The quantitative study on linguistic attitudes

5.1

Methodology: Participants, procedure, and scope

The quantitative study made use of a questionnaire consisting of questions on various sociolinguistic topics. The aim of this questionnaire was to establish people's linguistic attitudes. The questions were presented orally, most of them being closed questions which were highly standardized to facilitate compara­ bility of answers. Three of the questions dealt with participants' linguistic atti­ tudes towards the use of Luganda (for non-native speakers), Kiswahili, English, and mother tongue as media of instruction (and those are the ones that we will focus on the next section). The interviewers were Ugandans who were either native speakers of the region's dominant language or had sufficient command of the language to carry out the interview. Twelve Ugandan districts were incorporated into the survey. These were Mbarara, Kabarole, and Hoima for the western part of Uganda; Kapchorwa, Mbale, and Iganga for the eastern part; Masaka, Mukono, and Mubende for the central region; and Moroto, Lira, and Arua for the north. The researcher was present at most of the interviews observing the scene and taking part in dis­ cussions which occurred after some of the interviews.12 The number of participants in this study was fifty per district, adding up to a total of six hundred questionnaires. Participants represented different linguistic, social, and cultural backgrounds as well as different gender and age groups. Half of them originated from the rural areas and half of them from the muni-cipality. Table 1 presents the number of participants per language (ethnic affi-liation) in my 1996 Attidude Survey, in comparison with the 1991 Population Census. Due to the relatively small number of informants in the attitude survey sample, some groups are over- or underrepresented; for instance, minority groups such as the Sebei and the Kakwa are relatively overrepresented, while the Kiga are underrepresented. In the analysis of the survey, special care has been taken to minimize bias due to the relative over- or underrepresentation of these groups.

356

MEIKE SPRENGER-TASCH

1991 Census total % Ganda Lwo (Acholi/Langi/Adhola) Alur Gisu Gwere Kiga Konjo Nkore Nyarawanda Nyoro/Gungu Soga Toro (Toro/Tuku/Songora) Yole Karamojong (Karamoj ong/Dodoth/Tepeth) Ateso (Iteso) Kumam Lugbara (Lugbara/Aringa) Madi Kiswahili (Nubian) Samia Sebei Kakwa Other Ugandan

Attitude survey 1996 total %

3,015,980 1,959,964 395,553 751,253 275,608 1,391,442 361,709 1,643,193 329,662 495,443 1,370,845 488,024 228,918 346,166

18.8 12.2 2.5 4.7 1.7 8.7 2.3 10.2 2.1 3.1 8.5 3.0 1.4 2.2

131 62 5 38 4 8 3 45 4 48 25 43 5 30

21.8% 10.3% 0.8% 6.3% 0.7% 1.3% 0.5% 7.5% 0.7% 8.0% 4.2% 7.2% 0.8% 5.0%

999,537 112,629 588,830 178,558 14,739 185,304 109,939 86,472 742,780

6.2 0.7 3.7 1.1 0.1 1.2 0.7 0.5 4.6

27 3 24 11 3 6 43 12 20

4.5% 0.4% 4.0% 1.8% 0.5% 1.0% 7.2% 2.0% 3.3%

16,072,548

600

Table 1: Mother tongues (ethnic affiliation) in the 1991 Population Census compared to the 1996 Attitude Survey 5.2

Findings and discussion

As this paper focuses on linguistic attitudes towards Luganda (for non-native speakers), Kiswahili, English, and mother tongue as media of instruction in Uganda, only the following three questions in the questionnaire will be dealt with here: 1. 2. 3.

What languages would you want your children to learn? As a medium of instruction for my children, I would prefer... Do you think students/pupils learn more effectively if they are taught in their mother tongue?

MEDIA OF INSTRUCTION IN UGANDA

357

The first question was formulated in such a way as to give a general picture of informants' attitudes towards languages which are relevant to the Ugandan context. The second and third question relate to the use of these languages as media of instruction in schools. 5.2.1 Question 1: 'What languages would you want your children to learn? ' All the informants (100%) who answered this question (591 respondents) mentioned their mother tongue. From this information, and from other information gathered throughout the quantitative study, it can be inferred that the mother tongue plays a major role in the social and cultural life of the parti­ cipants, especially in all monolingual settings. In theory, everybody wants their children to learn and use their mother tongue. In practice, some parents find it difficult to have their children learn their parents' language. Especially in situations where the parents not only belong to different ethnic groups/speak different languages, but, at the same time, live in an area in which their mother tongue(s) are not prevalent, the children often use the dominant language of that area, that of their peers. This tendency became very clear through informal interviewing at the Teacher Training College, where Luganda is the dominant area language. It appeared that Luganda makes it very difficult here for the parents to ensure that their children learn and use their parents' language. Quite a few parents try to force their children to learn and use their language by sending them to their relatives in the countryside for a while, where Luganda is not understood or at least not tolerated. One Iteso informant, who lived in the Luganda language area and who was in a mixed Iteso-Mugwere marriage, made the following comment: I even married my wife—she is an Iteso. I made sure that at least ... my wife ... will have the courage to teach this child Ateso. But again it is worse. The child only knows Luganda because of the other children. So my child doesn't know any Ateso word. Now, I am really in trouble. ... The only solution now is to tell my wife—because my wife comes from Kumi District, which is ... Iteso. So my child when it goes there it has no other language to talk apart from hearing Ateso. So I have encouraged now my wife to be going ... with the child there. Possibly with time the child can grasp something.

Respondents not only showed 100% mother tongue loyalty, but almost all of them (96.8%) also mentioned English as a language they wanted their children to learn. This high percentage suggests that in today's Ugandan society, upward mobility is thought to be impossible without a fair command of English. Nineteen respondents, however, did not display this double loyalty as

358

 SPRENGER-TASCH

they did not mention English. These were all informants from the southern part of the country, especially Banyankore (especially Bahima) and Baganda. The questionnaires showed that all of these informants had very strong sentiments towards their mother tongues. Nine of them had never attended school, and an additional five had only finished primary school. All of them had jobs which did not require competence in English; they were market vendors, shopkeepers and other small-scale businessmen, peasants, pastoralists, and housewives. A relatively high number of respondents (79.5%) indicated that Kiswahili was a language they also wanted their children to learn. These respondents either displayed a triple loyalty towards English, Kiswahili, and their mother tongue, or—to a minor extent—a double loyalty towards mother tongue and Kiswahili. Furthermore, it was found that the acceptance rate of Kiswahili varied with their home district and with their ethnic background. In Buganda, acceptance of Kiswahili was much lower than nationwide. Even in Buganda, two thirds of the informants liked their children to learn Kiswahili. These results are promising for Kiswahili, especially if we keep in mind that the informants were not prompted.13 Respondents were not nearly as loyal towards Luganda as they were towards their mother tongue, English, and Kiswahili. Nationwide, only 40.4%) of them (including Baganda informants) indicated that they wanted their child­ ren to learn Luganda. In Buganda, acceptance of Luganda was, of course, much higher (96.6%o). Outside Buganda, only 21.3% of respondents were in favor of their children learning Luganda; the majority of them live in the East (especially Bagishu, Basoga, Bagwere, and Jopadhola), where Luganda has been used as a medium of instruction in the first years of primary education. From these figures, it can be deduced that the informatants fully acknow­ ledge the importance of mother tongue and English, and to a somewhat lesser extent, the importance of Kiswahili. Furthermore, it appears that attitudes towards Luganda (and to some degree also towards Kiswahili) differ according to home district as well as ethnic background. Table 2 summarizes the frequency of responses to Question 1. 5.2.2 Question 2: 'As a medium of instruction for my children, I would prefer.. ' This question covers four school levels: lower primary school, upper primary school, secondary school, and higher education. While asking the question, the languages were prompted, with respondents being able to choose between English, Kiswahili, Luganda, mother tongue, or a combination. Only very few informants insisted on other languages, in particular French and Arabic.

359

MEDIA OF INSTRUCTION IN UGANDA

Frequency Mother tongue English Kiswahili Luganda

Percentage

591 572 470 109 (237)*

100.0% 96.8% 79.5% 18.4% (40.1%)

Frequency Percentage 591 100.0% 572 96.8% 470 79.5% 109(237)* 18.4% (40.1%) Table 2: Frequencies of responses to Question 1: 'What languages would you want your children to learn? ' (591 respondents) *numbers in brackets include Baganda

Mother tongue English Kiswahili Luganda

Table 3 shows the frequencies of responses for each of the languages, either as Mother English Luganda Kiswahili the sole medium of instruction or intongue combination with other languages.

(iil)

in the lower classes of Primary School (PI-P4) in the upper classes of Primary School (P5-P7) in Secondary School

(IV)

in Higher Education

(i) (ii)

(i) (ii) (iii) (iv)

in the lower classes of Primary School (PI-P4) in the upper classes of Primary School (P5-P7) in Secondary School

79.2%

7.7%

23.7%

59.7%

28.3%

6.3%

39.8%

95.8%

15.3%

3.5% (12.7%)* 2.5% Luganda

42.5%

98.2%

33.7% Kiswahili

English

7.7%

23.7%

59.7%

6.3%

39.8%

95.8%

12.3% Mother tongue 79.2% 28.3% 15.3%

3.5% 42.5% (12.7%)* in Higher Education 12.3% 2.5% 33.7% Table 3: Frequencies of responses to Question 2: 'As a medium of instruction for my children, I would prefer ... ' *the number in brackets includes Baganda

98.2%

98.2% 98.2%

As can be seen from Table 3, about 80% of the informants prefer their mother tongue to be used, either alone or in combination with other languages, in the lower primary years. This confirms the trend observed in the results of the first question. The second most preferred language is English, at about 60%. Kiswahili comes third with 24%, while Luganda only has a preference rate of 8%) with non-Baganda informants (in total, Luganda scores nearly 28%, which is 4%o more than Kiswahili). Preference for the language of instruction changes at upper primary school level. Far fewer respondents (than at the lower primary level) show a preference for mother tongue instruction (only 28%), while almost all (96%) prefer English. As well, more respondents would like to see Kiswahili as a language of instruction, while still fewer favor Luganda. This favorable attitude towards English and Kiswahili continues at secondary school level. Interestingly, mother tongue instruction receives a higher

360

MEIKE SPRENGER-TASCH

score (15.3%) than instruction in Luganda, even the score which is inclusive of the Baganda respondents (12.7%). This means that the Baganda tend to opt for English only—without the mother tongue—in secondary school, more so than the other ethnic groups in Ugandan society. University education is almost exclusively the domain of English. Still, every third informant would like to see Kiswahili used as a medium of instruc­ tion. And 10% of the informants would like their mother tongue to be used. These 10% are largely made up of Baganda (59.9%) and Banyankole (13.6%). Noteworthy here is the high number of Banyankole, especially the Bahima, who are the ethnic group of President Museveni. Further research may reveal whether there is a connection between these two factors. Let us now look at the preference rates of each of the languages at district level. Kiswahili scores best in the Eastern Ugandan districts, especially in Kap­ chorwa, but also in Mbale and Iganga. In Kapchorwa, many people have stronger ties with their Kenyan neighbors than with fellow Ugandans. It should come as no surprise, then, that 78% of the Sebei, which is the major ethnic group in Kapchorwa, even indicated Kiswahili as a preferred medium of instruction at university level. This does not mean that English scores low in Kapchorwa, Mbale, and Iganga. On the contrary, the East Ugandans inter­ viewed want both languages to be used—in primary, secondary as well as in higher education. Obviously, Luganda (also when the Baganda are not included) scores well in Buganda, but high percentages can also be found in Iganga and Mbale, which are districts in the East where Luganda is used as a medium of instruc­ tion in lower primary. English attains a high score in most districts, except in Arua and Hoima for the lower primary level. Hoima is a more or less monolingual district, being the core land of the old kingdom of Bunyoro. Then again, Arua informants firmly believe that English should not be introduced too early. They insist on mother tongue instruction for the first years of primary schooling and later opt for English in combination with Kiswahili. Mother tongue instruction scores lowest in Iganga, which is under­ standable as Iganga is the most multilingual district of the survey. In addition, the low score can be explained as the effect from Luganda being used in this district at early primary level (cf. above). Significantly, Mbale, the other East Ugandan district where Luganda is used at lower primary level as the medium of instruction, shows the second lowest approval of mother tongue education. As we have seen, Luganda scores very high in these two districts. On the whole, from the upper primary level onwards, it appears that districts with

361

MEDIA OF INSTRUCTION IN UGANDA

dominant languages or where the dominant language is also a major language of Uganda approve of their mother tongue much more strongly than multi­ lingual districts or districts in which the dominant language is still one of the less influential and numerically strong languages of Uganda (e.g. Kapchorwa, Iganga, Moroto, Arua). 5.2.3 Question 3: 'Do you think pupils/students in general learn more effectively if they are taught in their mother tongue? ' While teachers and highly-educated people are generally believed to be more language-conscious than, for instance, people who have dropped out of school, responses to Question 3 show that attitudes of teachers and highly educated Ugandans do not differ too much from those of the general population; see Table 3.

all respondents teachers university graduates

Strongly agree

Agree

Disagree

Strongly disagree

50.1% 46.6% 38.9%

22.1% 28.8% 36.1%

16.0% 12.3% 16.7%

11.8% 12.3% 8.3%

Table 4: Frequencies of responses to Question 4: 'Pupils in general will learn more effectively if they are taught in their mother tongue. '

Interestingly, even people who want English to be used at the beginning of a school career think that mother tongue instruction will increase the learning effect; most of them even admit that it had a positive effect on their own learn­ ing while they were still at school. In other words, although informants want English to be used from P1, or at least from P5, they realize that they them­ selves and their children experience(d) a higher learning effect with mother tongue instruction. This reflects a practice which has been going on for decades. English has been the language of education—no matter how hard it is for pupils to cope with. English has been considered the key to success. People are somehow acknowledging the problem which the use of English poses for their children and are lamenting over it, but the step to opt for mother tongue, or at least an African language such as Kiswahili or Luganda, as a medium of instruction is too big for many. The attitude that English is the only efficient medium—be it for economic, political, or historical reasons—is rooted too deeply.

362

6.

MEIKE SPRENGER-TASCH

Conclusions

Attitudes towards English, the mother tongue, Luganda, and Kiswahili as media of instruction are mainly subject to ethnic origin and personal back­ ground. Variables like social status, educational background, and profession do play a role in people's attitudes towards some languages, but to a much lesser degree. These variables do, however, play an important role in the degree of competence in these languages. The role of English in education remains unchallenged. Except for the lower primary years, English is very much wanted as a medium or at least a comedium of instruction. This preference for English is not so much due to the conviction that it is intrinsically the best language to use. Rather, it stems from a long tradition of half a century of colonial rule, prestige of the language, the so-called chances of upward mobility it offers, and sometimes lack of an alternative.14 To quote Walusimbi (personal communication, 1997), "you won't get a job because of good performance in your vernacular lessons; you get it because of English". So the motivation for using English in Uganda, as well as in all other African states, is purely extrinsic. It is the result of the 'colonialization of the mind' (cf. Thiong'o 1986) in policy makers as well as in the population at large. Many informants do not believe that it is impossible to use African languages as media of instruction. Rather, they fear that English may lose its significance, that students will not be able to learn English properly. In many interviews, it has become apparent that people believe that English can only be learned if it is used as a medium of instruction. Besides, as I mentioned before, because of the efforts and costs necessarily involved in developing African languages for use in formal domains, it is not widely believed that any steps will be taken. This is probably also the reason why education in the mother tongue is seen as indispensable by most interviewees at lower primary level but not beyond. Mother tongue instruction has been used for decades in the years P1-P4 and has shown to be well equipped for use at that level. However, introducing African languages beyond P4 would require equipping them with enough scientific vocabulary and literature. It seems, then, that the costs and efforts of more sustained mother tongue instruction are not yet weighed against the benefits of a better cognitive and emotional development of the children (see also Nsibambi 1971); yet, these benefits are accepted by most of the inter­ viewees and a strong loyalty towards the mother tongue as a home language and for primary education is evident—in the results of the quantitative as well as the qualitative study.

MEDIA OF INSTRUCTION IN UGANDA

363

Luganda possesses a few strongholds, which can be found mainly among the people living in Kampala and the rest of Buganda, as well as among some of the people living in the Eastern part of Uganda—an area in which Luganda has a long-standing tradition as a medium of instruction in the lower classes of primary school and an area in which the knowledge of Luganda signals or connotes 'being educated' and 'having been to school'. However, as the people showing loyalty towards Luganda are very strong in their attitudes and belong to the educational and economic elite of the country, the weight of Luganda in Uganda should not be underestimated. The Odoki Report {Report of the Con­ stitutional Commission 1993) and the debate on the national language qustion in the Constitutional Assembly Debates {Proceedings of the Constituent Assembly 1995) make this clear. There is a general commitment in the people's views to having a national language as an instrument for nation-building but views differ over which language. We believe that this is an issue that may not be solved by statistics but should be cautiously handled. Although views were expressed widely, the debate over the national language was not controversial. But premature attempts to select a national language may well turn the issue into a major controversy. {Report of the Constitutional Commission 1993:80).

Still, there seems to be a growing loyalty towards Kiswahili, which is sug­ gested by the remark "Although views were expressed widely, the debate over the national language was not controversial". And the voting was strongly in favor of Kiswahili. A significant number of Ugandans believe that Kiswahili should be used as a medium of national communication. Although the sample of the present survey is too small to speak for all Ugandans, it nevertheless suggests that many Ugandans also feel that Kiswahili should be used as a medium of instruction. Kiswahili has gained acceptance although it has not played a significant role in education for about half a century. Kiswahili is an African language, an East African lingua franca, and as such may present an alternative to the highly controversial Luganda. It is a language which can be easily acquired and is already used by a number of Ugandans for horizontal communication. If taught at school or even used as a medium of instruction, Kiswahili could extend this horizontal function and, in the long run, also serve the function of vertical communication. Its use at school could thus be seen as one step towards its development as a national language. For the time being, the Odoki Commission (1993:11) has acknowledged the sensitive LugandaKiswahili rivalry and has "left the choice [for a national language] to future

364

 SPRENGER-TASCH

generations, when all possible languages will have competed and the most preferred emerged naturally". Notes 1 Although there is some controversy over the degree of mutual influence between language and culture, few people doubt that language and culture do affect one another and that it is difficult to separate the two. 2 Western Nilotic actually also belong to the Sudanic language group, but they are Eastern Sudanic languages. 3 The paper makes use of the common nomenclature of people, languages and kingdoms in Uganda. The Baganda speak the language Luganda and their kingdom is called Buganda. Similarly, the Nilotic Jopadhola speak Dhopadhola and the Iteso speak Ateso. 4 Uganda has a large number of English speakers compared to other anglophone African countries. English has therefore become a regular means of communication for some (edu­ cated) Ugandans and its importance is growing. 5 The term 'area language' was not defined. It seems that the term referred to those languages which were dominant in the area and most developed in European terms. According to Ladefoged, Glick and Griper (1972: 22) the six area languages were spoken by 6 1 % of the popu­ lation, leaving 39% of Uganda's primary school pupils without mother tongue instruction. 6 The languages combined by the slashes are very similar and considered by the Ugandan policy makers as more or less the same languages (which they are not). 7 Ruyankole and Ruyankore are spelling variants. 8 Regional language = area language; local language = mother tongue or mother tongue of the majority of the area in which the school is situated. 9 So far only the first volume of the Syllabus has been published. The volume mentions Kiswahili as a compulsory subject for three hours per week throughout the whole primary cycle. Details on the content are yet to be published in the second volume. 10 Researchers have been told by various authorities that there are a few 'open secrets'; e.g. that there is a Teacher Training College in the West of Uganda where Kiswahili teachers are trained. However, people seem to be afraid to disperse this information too openly because of fear of protests from Buganda. 11 The rejection of the Baganda has brought with it considerable ethnic stereotyping. The major stereotypes describe the Baganda as proud and arrogant, with a superior attitude towards other Ugandans. Obbo (1988:212) makes reference to this superior attitude, which is said to be common among the Baganda, in stating that in fact, the word Munyarwanda is not only applied to the Banyarwanda but to all Westerners and that it "often had unstated meanings other than a group name. It meant poverty and non-personhood". (For more ethnic stereotypes in the Ugandan population, see Nganda 1996:49f). 12 For a more detailed account on experiences of the researcher during the quantitative survey, see Tasch (1997). 13 Prompting in this case refers to the mentioning of alternative answers when the question is asked. 14 Reasons given are the conflict between Luganda and Kiswahili, finances that would be necessary to develop vernacular languages, and political as well as economic necessities.

MEDIA OF INSTRUCTION IN UGANDA

365

References: Bukenya, Gerold. 1997. "Linguistic Situation Country by Country". Ms., Uganda. Greenberg, Joseph H. 1963. The Languages of Africa. The Hague: Mouton de Gruyter. Kiwanuka, P. 1967. "Bilingualism in Education: The Role of the Vernacular Languages". East Africa Journal 21-24. Ladefoged, Peter, Ruth Glick & Clive Criper. 1972. Language in Uganda. Oxford: Oxford University Press. Matovu, Kasalina & Susan Kiguli. 1998. "The Role of English in Pre-Primary Education in Uganda: A Curse or a Blessing?" Paper presented at the sixth International Pragmatics Conference, Reims, July 1998. Mazrui, Ali A. 1967. "The National Language Question in East Africa". East Africa Journal 12-19. & Alamin M. Mazrui. 1995. Swahili State and Society: The Political Economy of an African Language. Nairobi: East African Educational Publishers & London: James Currey. & Pia Zirimu. 1978. "Church, State, and Marketplace in the Spread of Kiswahili: Comparative Educational Implications". Case Studies in Bilingual Education ed. by Bernard Spolsky & Robert L. Cooper, 427543. Rowley, Mass.: Newbury House. Ministry of Education. 1989. Education for National Integration and Development: Report of the Education Policy Review Commission. Kampala. Mukama, Ruth. 1986. "The Viability of the Indigenous Languages in the Ugandan Context. Mawazo 6.49-60. . 1989. "The Linguistic Dimension of Ethnic Conflict". Conflict Resolution in Uganda ed. by Kumar Rupesinghe, 178-206. London: James Currey. . 1991. "Recent Developments in the Language Situation and Prospects for the Future". Changing Uganda ed. by Holger . Hansen & Michael Twaddle, 334-349. London: James Currey & Kampala: Fountain Press. & Oswald Ndoleriire. 1997. "Nationality and Languages in Uganda: Complementary or Conflict?" Proceedings of the LiCCA Workshop in Dar es Salaam ed. by Birgit Smieja, 66-76. Duisburg: LICCAP.

366

MEIKE SPRENGER-TASCH

Nganda, Cecilia Namulondo. 1996. Primary Education and Social Integration: Ethnic Stereotypes in the Uganda Basic Text Books. (= Bayreuth African Studies). Bayreuth: University of Bayreuth. Nsibambi, Apolo. 1971. Language Policy in Uganda: An Investigation into Costs and Politics. African Affairs 70.62-71. Obbo, Christine. 1988. "What Went Wrong in Uganda?" Uganda Now ed. by Holger . Hansen & Michael Twaddle, 205-223. London: James Currey & Kampala: Fountain Publishers,. Primary School Syllabus 1997-2000. 1997. Kampala: Ministry of Education. Ssekamwa, J.C. 1997. History and Development of Education in Uganda. Kampala: Fountain Publishers. Tasch, Meike. 1997. "Language Attitude Survey in Uganda: Work in Progress Report". Proceedings of the LiCCA Workshop in Dar es Salaam ed. by Birgit Smieja, 137-144. Duisburg: LICCAP. Thiong'o, Ngugi wa. 1986 [1981]. Decolonising the Mind: The Politics of Language in African Literature, 2nd edition. London: James Currey & Nairobi: Heinemann. Report of the Constitutional Commission (Odoki Report). 1993. Kampala: Uganda Constitutional Commission. Proceedings of the Constituent Assembly (on the national language question). 1995. January 31 and February 2, 1995. Kampala. Wald, Benji. 1994. "Sub-Saharan Africa". Atlas of the World's Languages ed. by Christopher Moseley & R.E. Asher, 287-346. London: Routledge. Walusimbi, Livingstone. 1994. "Uganda: Language Situation". The Encyclopedia of Language and Linguistics ed. by R.E. Asher, 48254826. Oxford: Pergamon.

STYLE LABELS IN MONOLINGUAL ENGLISH LEARNER'S DICTIONARIES

WOLFGANG K. HÜNIG University of Duisburg-Essen

This paper deals with the issue whether learners can rely on the stylistic classifications used in dictionaries. The four most important British monolingual learner's dictionaries were compared and the result of this research is that the information which the dictionaries convey through their stylistic labels such as 'derogatory', 'formal', etc. is not very reliable. Learners who expect hard and fast rules have to be made aware of the fact that they should take into account the whole communicative context as well as human face wants. Most of the stylistic labels can only be taken as hints, which have to be fine-tuned in different social and discourse contexts. Exchanges in such contexts may have very different functions and may reflect differ­ ent social relationships.The learners also have to be made aware of the fact that meanings change with time and that such changes may or may not enter the linguistic system perma­ nently.

1.

Introduction

Whoever has seen the film Die Hard with a Vengeance will remember the remarkable initial scene. The main character, Bruce Willis, is walking down a street in Harlem as a plain-clothes policeman; he is carrying a sandwich board which reads, "I hate niggers!" He is soon vigorously attacked by a group of black youths. From the point of view of linguistic politeness (Brown & Levinson 1987), the sandwich man performs a severely face-threatening act towards black people. In addition, the term nigger in itself is a taboo term,1 which is shown by the dictionary entries in Table 1. The four dictionaries agree in defining nigger as a very or extremely offensive word used for a black person; they do, however, differ with respect to the style labels characterizing the word's usage: the word nigger is character­ ized as a taboo word in one dictionary (LDOCE), as a derogatory, offensive word in another dictionary (OALD), both as a taboo and slang word in a third

368

WOLFGANG K. HÜNIG

dictionary (CIDE), and it is not given a style label in a fourth dictionary (COBUILD). Dictionary

Dictionary entry for nigger

Collins Cobuild English Dictionaiy (COBUILD)2

nigger is an extremely offensive word for a black person.

Dictionary of Contemporary English (LDOCE)

nigger taboo an extremely offensive word for a black person.

Oxford Advanced Learner's Dictionary of Current English (OALD)

nigger (!) [= taboo] derogatory offensive a black person.

Cambridge International Dictionary of English (CIDE)

nigger taboo slang a black person. This is a very offensive word used to or about a black person by a white person.

Table 1: Dictionary entries for nigger

This paper wants to investigate the use of stylistic or situational labels such as formal, informal, offensive, derogatory, slang, humorous, etc. in these four important, widely used British monolingual dictionaries. It will, in particular, address the questions what these style labels mean, whether the dictionaries apply identical or at least similar labels in characterizing the same word and, since all four dictionaries are also conceived of as 'learner's dictionaries', whether the labels are of any help to the language learner. Special attention will be devoted to the learner of English as a Foreign Language, for whom style labels may be a valuable help in developing a feeling for the stylistic appropri­ ateness of a word.3 Dictionary

Size

Year

Edition

Collins Cobuild English Dictionaiy

over 75,000 references 80,000 words and phrases 155,000 definitions and examples 100,000 words and phrases

1995

2nd

1995

3rd

2000

6th

1995

1st

Longman Dictionary of Contemporary English Oxford Advanced Learner's Dictionaiy of Current English Cambridge International Dictionary of English

Table 2: The four learner's dictionaries

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

369

In order to guarantee comparability of the style labels within a variety of English, the investigation is restricted to the four dictionaries of British English mentioned above.4 As Table 2 shows, the editions of the four dictionaries considered for this study are comparable with respect to their target group, their size, and year of publication (1995):5,6 2.

Using a dictionary

An analysis of dictionaries and their use has to take into account two different perspectives: that of the lexicographer and that of the user. The perspective of the lexicographer includes, amongst others, the following aspects, which may influence his/her choice of a particular dictionary format: a. b.  d. e. f. g. h.

the dictionary as a reference tool for language usage; the dictionary as a databank; the dictionary as a means of improving communication; the dictionary as a means of strengthening (codifying) a language; the dictionary as a stimulus for the reflection on language; the dictionary as a help for the language learner; the dictionary as an ideological weapon; the dictionary as a commodity.

The issue of the lexicographer's motivation for a particular format will not be explored here. The user's perspective deserves much more attention and will, or should, ultimately also have repercussions on the lexicographer's perspec­ tive. Studies on the perspective of the user published in seven papers between 1955 and 1984 and summarized in Hartmann (1987:125) present the following results: a. b.  d. e. f.

Learners do not use all the information offered. The meaning and spelling of the words are regarded as most im­ portant. How a dictionary is used depends on the goal of the users and their level of proficiency. Students prefer bilingual dictionaries. Using a dictionary does not improve test results. Monolingual dictionaries cannot help in solving all problems of grammatical correctness.

370

WOLFGANG K. HÜNIG

These results suggest that foreign language learners need more practice in using their dictionaries efficiently. More recently, Harvey and Yuill (1997) investigated the way learners use monolingual dictionaries. Since their study is pertinent to this paper, I will summarize its results. A total of 211 students in various fields of study participated in this investigation. Only students who were used to working with monolingual dictionaries were selected. Their degree of linguistic proficiency in English (as a foreign language) was, however, ignored because this aspect was regarded as too complex and transient to measure. Students were categorized according to five criteria: (i) gender, (ii) mother tongue, (iii) subject of study, (iv) degree of experience in using the COBUILD dictionary, and (v) learning context. All participants were asked to look up a word in COBUILD and then fill in a flow chart questionnaire. The results show that the students looked for information in the following areas: (i) spelling, (ii) meaning, (iii) existence,7 (iv) synonymy, (v) grammar, (vi) register, (vii) collocation, and (viii) inflection. The frequencies with which each of these areas was searched are presented in Figure 1.

Figure 1: Areas searched in dictionary lookups

Apart from spelling and meaning, which are looked up most frequently (24.4% and 18.3%, respectively), lookups in the other types of information that are presented in a dictionary entry occur about equally frequently (with frequencies ranging between 12.8% and 5.9%). It appears that information on register is

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

371

also regarded by the users as fairly important.8 The overall success rate of the searches was 92.1%. In about 57% of the cases, the students found the information they searched for in the definitions, and in about 40% of the cases, they found it in the examples provided by the dictionary. This leads Harvey and Yuill (1997:268) to draw the following conclusion: "These results emphasize the importance and salience of examples, and suggest that lexicographers need to consider carefully the appropriateness of their examples, and perhaps to give the reader the source of examples where they are typical of a given genre." To test the value of examples, I checked the collocations of the word destitute, which Birmingham University and Harper Collins Publishers provide on the internet. Let us consider four of the twenty-five predicative uses of destitute in of-constructions: (1) (2) (3) (4)

... despite being nearly destitute of natural resources, ... Conversely, Japan, virtually destitute of all natural resources ... Are your wife and children destitute of'a bed to lie on, or bread to... ... situation is deplorable, but not destitute of spiritual consolation.

Examples (l)-(3) belong to the technical, rather than formal, register; only example (4) sounds genuinely formal. But the reason for seeing this usage of destitute as more formal than the others is to be found in its collocation with spiritual consolation, which conjures up a religious domain. This means that, if the last example was used as an illustration of the word's usage in a dictionary, students would most likely associate it with a higher degree of formality than if one of the other examples was quoted. Learners obviously derive hypotheses about a word's meaning and contextual usage from examples as a whole. Important as examples are as additional illustrations of a word's meaning, they may be problematic to the language learner, who, of necessity, generalizes the word's usage from the contextual associations evoked by the sentence given. Harvey and Yuill (1997) restricted their investigation to only one diction­ ary. Admittedly, COBUILD is a very popular dictionary, not least because it was the first corpus-based learner's dictionary. The issue which I want to address is, however, the extent to which different dictionaries agree on stylistic classifications of words and expressions. With regard to the word destitute, for example, the 1995 edition of COBUILD labels it as 'formal', whereas the earlier 1987 edition does not. The 1995 edition of LDOCE and OALD does not label the item destitute itself either, but it labels destitute of as 'formal'. CIDE also does not label destitute.

372

3.

WOLFGANG K. HÜNIG

The practice of stylistic labeling

In order to be able to compare the practice of stylistic labeling in the four dictionaries under investigation, I present in Table 3 all the labels which are used. Equivalent labels are co-aligned.9,10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

COBUILD

OALD

LDOCE

CIDE

literary

rhetorical

literary poetic

formal informal

formal informal slang derogatory (!) [= taboo] offensive sexist approving euphemistic jocular ironic dated archaic

literary poetic biblical formal informal slang

offensive

old-fashioned spoken written

formal informal slang disapproving

taboo

taboo

approving

approving

humorous

humorous

old-fashioned old use spoken

dated old use

Table 3: Stylistic labels used in four monolingual dictionaries

As this synopsis shows, the number of labels used in a dictionary varies considerably: COBUILD uses seven labels, CIDE eleven labels, LDOCE twelve labels, and OALD fourteen labels. A comparison of the dictionaries has to take this imbalance into account. In order to be able to assess the practice of labeling, it is worthwhile having a closer look at the way labels are defined and used in the different dictionaries. I will, by way of illustration, compare the use of one of the labels, i.e. 'informal', which occurs in all four dictionaries. The definitions are given below: COBUILD

Used mainly in informal situations, conversations, and personal letters; e.g. decaf elbow room.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

373

OALD

Informal expressions are used between friends or people who know each other well, in a relaxed or unofficial context. They are not appropriate for formal situations. Examples are brainy, dad, umpteen, wallop.

LDOCE

A word or phrase that is used in normal conversation, but may not be suit­ able for use in more formal contexts, particularly in writing e.g. essays or business letters.

CIDE

Used with friends or family or people you know in relaxed situations. Informal words are more common in speech than in writing. She works as an admin assistant (= administrative assistant). She always stops for a coppa (= cup of coffee) at 11  'clock. It might be considered not polite to use very informal language in formal situations.

All four definitions are to some extent circular: informal use of language is defined as occurring in informal situations (COBUILD), as denoting informal expressions (OALD), as not suitable in more formal contexts (LDOCE), or as used in relaxed situations (CIDE). Three of the definitions relate informality to a personal relationship (COBUILD, OALD, LDOCE), three characterize infor­ mality as mainly occurring in conversation or speech (COBUILD, LDOCE, CIDE), and two of the definitions mention contexts in which informal use of language is inappropriate (LDOCE, CIDE), one of which relates it to politeness (CIDE). Although the definitions given by the four dictionaries do not contra­ dict each other, they are based on different defining criteria and obviously lead to different notions of 'informal language'. The defining criteria in fact relate to different types of variation, as described, e.g., by Quirk et al. (1985:15-27), whose categories ultimately go back to Gregory's (1967) three types of variation, i.Q. field of discourse, mode of discourse, and tenor of discourse. Gregory regards these three types as situationally related "in a general sense to the role being played by the user in the language event. This role may be sub-categorised into the USER'S PURPOSIVE ROLE relating to the field of discourse, MEDIUM RELATIONSHIP relating to mode of discourse and ADDRESSEE RELATIONSHIP relating to tenor of discourse" (Gregory 1967:184). Quirk et al. also use the term 'field of discourse', but instead of 'mode of discourse' and 'tenor of discourse', they adopt the broadly equivalent terms medium and attitude and introduce two additional types of variation, namely according to social group and region (cf. Quirk et al. 1985). Table 4 summarizes these two approaches to types of variation in English:

374

WOLFGANG K. HÜNIG

Gregory (1967)

Quirk et al. (1985)

1. field of discourse 2. medium 3. attitude 4. social group 5. region Gregory (1967) Quirk et al. (1985) 1. field of discourse (user's purposive role) 1. field of discourse 2. mode of discourse (medium relationship) 2. medium 3. tenor of discourse (addressee relationship) 3. attitude 4. social group 5. region Table 4: Types of variation in English 1. field of discourse (user's purposive role) 2. mode of discourse (medium relationship) 3. tenor of discourse (addressee relationship)

Quirk et al. (1985:16) define their types of variation as follows: [Types 1-3] relate to language use. People select the varieties according to the situation and the purpose of communication. The field of discourse relates to the activity in which they are engaged; the medium may be spoken or written ... and the attitude expressed through language is conditioned by the relationship of the participants in the particular situation. [Types 4-5] relate primarily to the language user. People use a regional variety because they live in a region ... Similarly, people use a social variety because of their affiliation with a social group.

Since variation according to region is not relevant to the style labels considered here, only the first four of Quirk et al.'s (1985:16) five varieties will be used as criteria in examining the commonalities and differences in the definitions of the label 'informal'. COBUILD, LDOCE, and CIDE associate the label 'informal' with the spoken medium; COBUILD, OALD, and CIDE also categorize 'informal' language according to attitude, in that they define it as personal and relaxed; COBUILD and LDOCE define 'informal' according to the field of discourse (personal letters); CIDE, finally, connects 'informal' to the social criterion of politeness. The most important criterion on which the definitions of the label 'informal' seem to hinge, however, involves the concepts 'scenario', 'situation', or 'context', which can be gathered from wordings such as "personal letters" (COBUILD), "unofficial occasion" (OALD), "normal conversation" (LDOCE), and "friends or family or people you know" (CIDE). Such concepts are used neither by Gregory (1967) nor by Quirk et al. (1985). Given that an analysis of the other labels yields similar results, it is neither quite clear what the methodological background of the labels is nor how they function precisely for the language user. Dictionaries seems to suggest that there exists a one-to-one relationship between lexical items (such as nigger, wireless, or beatify) and varieties, that is, that lexical items signal, or are associated with, a specific variety or style.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

375

This would be in line with Amnion's (1995:61) description of how varieties are constituted. Ammon defines varieties as dependent on the occurrence of differ­ ent variants. Thus, the terms wireless and radio, which may lexicalize the same concept (or, which may refer to the same referent), would be regarded as two onomasiologically constituted variants. The use of wireless might be seen as constituting a variety which could be called 'Old-fashioned English', the use of radio a variety which might be called 'Common Core English' or 'Nuclear English'.11 The term 'variety' is understood here in a broad sense and includes regional or national varieties as well as registers or styles. Figure 2 represents the two variants radio and wireless within their varieties:

variety: 'Common Core English'

variety: 'Old-fashioned English'

Figure 2: Linguistic variants and within varieties of English

If one accepts that single linguistic items can constitute a specific variety, then there would, in theory, be an unlimited number of varieties; but this would render the term 'variety' useless. As Biber's (1988) statistical analyses have shown, single linguistic items do not constitute a specific style, i.e. they do not belong to a specific variety. Instead, styles are constituted by the co-occurrence of several different linguistic phenomena such as passive constructions, types of noun modification, and the use of certain lexical items. In dictionaries, however, stylistic information is often presented such that "single lexical items are taken as signs for particular one-dimensional varieties which are expressed with the help of labels like AmE (American English), sl (slang), finl (formal)

376

WOLFGANG K. HÜNIG

and law" (Esser 1993:10). These one-dimensional varieties are "methodo­ logical shortcut[s] to refer to an unspecified group of texts where the focus is on only one of the factors of the communication situation and referential content" (1993:10). This way of presenting information may lead to a prolif­ eration of varieties: apart from varieties which are based on general usage situations such as formal, informal, literary, and biblical English, we would also have to distinguish varieties such as 'offensive English', 'taboo English', 'euphemistic English', etc., which represent the speaker's personal attitude in a given situation and which therefore may be subject to change. More impor­ tantly, though, dictionary users will, in all likelihood, not share the scholarly view that the labels are only to be understood as methodological shortcuts. Instead, because of their limited command of English, they are likely to take these labels at their face value, assuming that the words and expressions labeled in a certain way represent different varieties of English, e.g. that the use of any word labeled 'informal' is sufficient to render a text or situation infor­ mal. 4.

Statistical comparison

The final section will look in more detail at the labeling practices utilized in the four dictionaries and compare them statistically. To allow for comparison, the eighteen labels listed in Table 3 had to be regrouped. Labels which occurred in all four dictionaries were retained as such (i.e. were not put together): this goes for labels 4 and 5 ('formal' and 'informal', respectively). Labels which did not occur in all four dictionaries were grouped together. As such, 'literary', 'rhe­ torical', and 'poetic' (labels 1-2) were combined. 'Biblical' (label 3) had to be left out because it only occurs in one of the dictionaries. Labels 6-10, i.e. 'slang', 'derogatory', 'disapproving', 'offensive', '(!)', 'taboo', and 'sexist' were subsumed under one and the same overall category: it includes words to which a certain social stigma is attached. Two examples may illustrate this point: the noun queer is labeled 'informal' and 'offensive' in COBUILD and LDOCE, 'slang' and 'derogatory' in OALD, and 'informal' in CIDE. The expression a bit of crumpet is labeled 'informal' and 'offensive' in COBUILD, 'sexist' in OALD, and 'slang' in LDOCE and CIDE. The labels listed under 1516 were also joined because COBUILD does not differentiate between two temporal stages, as do the others. The last two labels, 'spoken' and 'written', were not taken into account because OALD and CIDE do not use them at all, LDOCE only employs the label 'spoken', and only COBUILD uses both labels. Thus, the following five groups of labels were tested for comparison.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

a. b.  d. e.

377

literary, rhetorical, or poetic; formal; informal; old-fashioned, dated, archaic, or old use; slang, derogatory, disapproving, offensive, (!), taboo, or sexist.

Since the frequency of occurrence of these labels varies widely, the number of words tested differ. Thus, the lemmas marked 'formal' and 'informal' were collected at different, randomly chosen pages in the dictionaries. Each word had to be listed in all four of the dictionaries with an equivalent meaning. In this way, 227 words labeled 'formal' and 169 words labeled 'informal' in any one of the dictionaries were collected. The other three groups of labels appear less frequently and required a more intensive search. This resulted in the following selections: 65 words labeled 'literary', 'rhetorical', or 'poetic'; 50 words labeled 'old-fashioned', 'dated', 'archaic', or 'old use'; and 110 words labeled 'slang', 'derogatory', 'offensive', '(!)', 'taboo', or 'sexist'. The dictionaries were judged to be in agreement in labeling a word when the same style label or an equivalent style label belonging to the same group of labels occurred in all four dictionaries. Whether further labels were added and in which order they occurred was not taken into consideration. Overall agree­ ment on the label 'informal', for instance, required that a particular word was marked 'informal' across the four dictionaries; agreement on the label 'offen­ sive' required that a particular word carried this label or an equivalent one ('slang', 'derogatory', 'disapproving', 'offensive', '(!)', 'taboo', or 'sexist') across the four dictionaries. Table 5 presents the results of this investigation: Style labels

1. literary, rhetorical, and poetic 2. formal 3. informal 4. old-fashioned, dated, archaic, and old use 5. slang, derogatory, disapproving, offensive, (!), taboo, and sexist

Number of lemmas

Number of words identically labeled in all four dictionaries

Percentages

65 227 196 50

4 59 63 7

6.15% 25.99% 32.14% 14.00 %

110

34

31.00%

Table 5: Overall agreement on labels in the four dictionaries

378

WOLFGANG K. HÜNIG

Agreement between dictionaries on all labels falls well below 50%. The highest rate of agreement is found with the labels 'informal' (32.14%) and the group of labels 'slang, etc.' (31%). Agreement on the label 'formal' follows closely at almost 26%, while the values for the groups 'old-fashioned, etc.' and 'literary, etc.' stand only at 14% and 6.15%, respectively, i.e. dictionaries hardly agree on whether an expression is to be classified as 'old-fashioned' or 'literary'. In order to test the degree to which two dictionaries agree on the labels given to words, I made a number of pair-wise comparisons. Pair-wise comparison CIDE-COBUILD CIDE-LDOCE CIDE-OALD OALD-COBUILD OALD - LDOCE COBUILD - LDOCE

Number of lemmas

Number of identically labeled words

Percentages

65 65 65 65 65 65

26 29 8 8 10 33

40.00 % 44.61 % 12.31% 12.31% 15.40% 50.80 %

Table 6: Words labeled 'literary', 'rhetorical', and 'poetic' Pair-wise comparisons CIDE-COBUILD CIDE-LDOCE CIDE-OALD OALD-COBUILD OALD-LDOCE COBUILD - LDOCE

Number of lemmas

Number of identically labeled words

227 227 227 227 227 227

105 99 94 92 96 100

Percentages 46.25 43.61 41.41 40.53 42.30 44.05

% % % % % %

Table 7: Words labeled 'formal' Pair-wise comparisons CIDE-COBUILD CIDE-LDOCE CIDE-OALD OALD-COBUILD OALD-LDOCE COBUILD - LDOCE

Number of lemmas 196 196 196 196 196 196

Number of identically labeled words 100 102 100 89 100 100

Table 8: Words labeled 'informal'

Percentage 51.02% 52.04 % 51.02% 45.41 % 51.02% 51.02%

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

Pair-wise comparisons CIDE- COBUILD CIDE-LDOCE CIDE-OALD OALD- COBUILD OALD-LDOCE COBUILD - LDOCE

379

Number of lemmas

Number of identically labeled words

Percentage

110 110 110 110 110 110

48 47 68 54 53 37

43.36% 42.72 % 61.82% 49.09 % 48.18% 33.64 %

Table 9: Words labeled 'slang', 'derogatory', 'disapproving, 'offensive', '(!)', 'taboo', and 'sexist' Pair-wise comparisons CIDE-COBUILD CIDE-LDOCE CIDE-OALD OALD-COBUILD OALD-LDOCE COBUILD - LDOCE

Number of lemmas

Number of identically labeled words

Percentages

50 50 50 50 50 50

18 21 17 17 18 30

36% 42% 34% 34% 36% 60%

Table 10: Words labeled 'old-fashioned', 'dated', 'archaic', and 'old use'

Tables 6-10 show the results of the pair-wise comparisons for each of the five groups of labels.Most agreement percentages of the pair-wise comparisons are higher than those of the overall comparisons, but, apart from two agreement values at 60% and nearly 62%, the agreements stay at or below the 50% mark. Any two dictionaries agree least with regard to labeling words as 'literary', 'rhetorical', or 'poetic', i.e. they range from an agreement level of 12.31% to one of 50.8%). In this group, COBUILD's and LDOCE's labeling practice is the most similar. COBUILD and LDOCE also display the highest level of agreement on labeling 'old-fashioned', 'dated', 'archaic', or 'old use'. On the other hand, these two dictionaries disagree most with regard to their labeling of words as 'slang', 'derogatory', 'disapproving', 'offensive', '(!)', 'taboo', or 'sexist'. There is, however, not one pair of dictionaries which stands out among the other five pairs by having high values of agreement for all five groups of labels. Agreement between any two dictionaries is highest on judging words as 'informal': with the exception of the OALD - COBUILD pair (45.41%), dictionaries agree on the label 'informal' in about 51% of the cases. The same holds for words labeled 'formal', where agreement hovers around 43%o.

380

WOLFGANG K. HÜNIG

The comparison of words labeled 'slang', 'derogatory', 'disapproving', 'offensive', '(!)', 'taboo', and 'sexist' shows that dictionaries agree on an aver­ age of 46.5% of the lemmas, with apeak of 61.82% and a low of 33.64%. The agreement level for words labeled 'old-fashioned', 'dated', 'archaic', and 'old use' lies below 42%, except for the COBUILD - LDOCE pair, with agreement in 60% of the cases. Examples of disagreements are, for instance, the word bard, which the dictionaries label as 'old-fashioned', 'archaic', or 'literary', and the word whore, which they label as 'dated', 'offensive', and 'taboo'. How can these results be interpreted? Since even the examples which the dictionaries provide to illustrate their labels in their introductions are partly controversial, it has to be concluded that only a very small number of words can only truly prototypically be characterized by a stylistic label. Such words obviously retain their stylistic meaning irrespective of their context. They may even constitute such a context. Among these are terms which are racially insulting such as nigger or which belong to the class of so-called four-letter words. Other examples are words with occur fairly rarely, such as sagacity, to imbibe, to imperil, impecunious, and unctuous.12 The vast majority of words, on the other hand, acquire their labeled meaning only within their linguistic context. Eggins and Slade (1997) present a telling example of the contextual function of small talk between employees in Australia, one of whose main functions is to construct cohesive relationships. In the situations of small talk analyzed by Eggins and Slade (1997:123), cohesive relationships are estab­ lished by means of lexical items from the following domains: a. b.  d.

attitudinal vocabulary to appraise and evaluate the other's behavior; in-group words or anti-language (terms such as blackfella)13 to indi­ cate group boundaries; slang, swearing, and drawing on shared past history to suggest group cohesion; abbreviated and Australianized terms of address to indicate levels of intimacy.

Rather than a single word labeled in a certain way, it is the situational and linguistic context as a whole that characterizes the language of small talk as 'informal'.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

5.

381

Conclusion

It has been shown that the stylistic information in dictionaries is not very reli­ able. Language learners who still adhere to a naive all-or-nothing principle and expect hard and fast rules have to be made aware of the fact that they should take into account the whole communicative context as well as human face wants. They will have to learn that most of the stylistic labels given in diction­ aries are only taken to be hints of limited value, which have to be fine-tuned in different social and discourse contexts. Exchanges in such contexts may have very different functions and may reflect different social relationships, which may be crucial for the meaning of the lexical items used. Lexical irony or sarcasm is a case in point. A statement such as Paula simply is a genius said when Paula has done something silly or a remark such as Bloody impeccable, these Americans! uttered by a clerk in a London Underground ticket office who is annoyed at the rude behavior of some American customers have to be assessed according to their situational and communicative value. The learner also has to be made aware of the fact that meanings change with time and that such semantic changes are brought about by speakers in social and discourse contexts and may or may not enter the linguistic system more or less permanently. This also means that the linguistic considerations and analyses which the learner should engage in could and should be regarded as part of his or her language learning activities. Notes 1 There are contexts in which a black person may address another black person as 'nigger' as a signal of solidarity, racial pride, etc. 2 COBUILD is an acronym for Collins Birmingham University International Language Data­ base. 3 Vorlat emphasizes that style is important in learning a foreign language (1989:691): "While learning English one must develop a sense of style, i.e. a semi-instinctive knowledge of what can be said and also what cannot be said. This is as important as using the right tenses, distin­ guishing between some and any, etc." 4 At the time this paper was written, the MacMillan English Dictionary for Advanced Learners (2002) had not yet been published. 5 As it happened, 1995 saw the publication of revised editions of COBUILD, LDOCE, and OALD and saw the publication of the first edition of CIDE. 6 A more detailed characterization of the four dictionaries can be found in Appendix 1. 7 The term existence refers to the issue of whether a word belongs to the English language or not, e.g. whether there is a word such as beweep analogous to the German word beweinen 'weep over'. 8 The term register covers all style labels in COBUILD.

382

WOLFGANG K. HÜNIG

9 The symbol '(!)' for a taboo word in OALD was given a separate line because COBUILD only has the label offensive which in OALD is used as well. This does not make any difference because for the statistic analysis I have included these labels alongside others in one category; cf. Section 5. 10 A definition of these labels can be found in Appendix 2. 11 According to Quirk et al. (1985:16) "a common core or nucleus is present in all the varieties so that, however esoteric a variety may be, it has running through it a set of grammatical and other characteristics that are present in all the others." 12 The more an expression is used the more it will acquire an everyday quality and lose ist marked stylistic value. 13 The term blackfella used to be a taboo word but is now an accepted term for naturalized Australians.

Dictionaries Procter, Paul et al., eds. 1995. Cambridge International Dictionary of English. Cambridge: Cambridge University Press. Sinclair, John et al., eds. 1995. Collins COBUILD English Language Diction­ ary, 2nd edition. London: Harper Collins. Summers, Delia et al., eds. 1995. Longman Dictionary of Contemporary English, 3rd edition. London: Longman. Hornby, Albert Sydney et al, eds. 1995. Oxford Advanced Learner's Dictiona?y, 5th edition. Oxford: Oxford University Press. References Ammon, Ulrich. 1995. Die deutsche Sprache in Deutschland, Österreich und der Schweiz. Berlin: Mouton de Gruyter. Biber, Douglas. 1988. Variation Across Speech and Writing. Cambridge: Cambridge University Press. Brown, Penelope & Stephen C. Levinson. 1987. Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press. Eggins, Suzanne and Diana Slade. 1997. Analyzing casual conversation. London: Cassell. Esser, Jürgen. 1993. English Linguistic Stylistics. Tübingen: Max Niemeyer. Gregory, Michael. 1967. "Aspects of Varieties Differentiation". Journal of Linguistics 3.177-198. Hartmann, R.R.K. 1987. "Dictionaries of English: The User's Perspective". Dictionaries of English ed. by Richard W. Bailey, 121-135. Cambridge: Cambridge University Press.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

383

Harvey, Keith & Deborah Yuill. 1997. "A Study of the Use of a Monolingual Pedagogical Dictionary by Learners of English Engaged in Writing". Applied Linguistics 18.253-278. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Vorlat, Emma. 1989. "Style". A User's Grammar of English ed. by René Dirven, 687-722. Frankfurt a.M.: Peter Lang. Appendix 1: Detailed characterization of COBUILD, OALD, LDOCE, and CIDE. Appendix 2: Definition of style labels in COBUILD, OALD, LDOCE, and CIDE.

3,500 words

70% BrE 25% AmE 5% other national varieties of English

no strictly scientific or technical texts; most texts produced after 1990

-

-

'one word, one entry' principle; superheadwords to find right definition;

. National varieties

d. Text types

e. Learner corpus

2. Defining vocabulary

3. Defini­ tions and subsenses

definitions followed by corpus-based examples

not specified

13.3% spoken-86.7% written

b. Spoken/ written

headwords with same spelling separated by numbers or, when closely related, by letters; definitions followed by corpus-based examples

not given

10% s p o k e n - 9 0 % written

100 million words

200 million words

a. Size

British National Corpus

OALD

Bank of English

COBUILD

1. Corpus

Appendixl

find

right

most definitions followed by corpus-based examples

signposts to definition;

2,000 words

-

not specified

'good coverage' of BrE and AmE

ratio of spoken and written not given

140 million words

The Longman Corpus Network and British National Corpus

LDOCE

guide words to find right definition; marked core meaning given

2,000 words

learner corpus with codified errors

natural conversation, television and radio, literature and non-fiction

major varieties of English; BrE and AmE equally represented

ratio of spoken and written not given

100 million words

Cambridge Language Survey Corpus

CIDE





>





3

384

usage notes, in which words with related meanings are differ­ entiated

definition text in British English

pictorial illustrations colored geographical maps

contains study pages on compounds, phrasal verbs, idioms, collocations, etc.

no pictorial illustrations

abbreviations and prefixes are headwords

7. Special features

special symbols for idioms and phrasal verbs

simple explanation of parts of speech at the beginning; grammatical information given in brackets in definition text

verb codes before examples

grammatical classification (word class) and often structural note as well

6. Grammatical information

pictorial illustrations

collocations shown in frequency order

additional explanations on study pages

statistically relevant, corpus-based collocations

5. Collocations

3,000 most frequently spoken and written words marked; frequency graphs for certain words

different senses arranged acording to frequency of use, with most common meaning given first

according to five frequency bands

LDOCE

OALD

4. Frequency markings

COBUILD

contains lists of false friends

every idiomatic phrase to be found under every word

pictorial illustrations

simple explanation of parts of speech at the beginning; simple grammar codes, attached to an example sentence

statistically relevant, corpus-based collocations

CIDE

STYLE LABELS IN ENGLISH LEA

386

WOLFGANG K. HÜNIG

Appendix 2 No.

Label

Dictionary

Definition

1.

literary

COBUILD

rhetorical

OALD

literary

LDOCE

literary

CIDE

Used mainly in novels, poetry, and other forms of litera­ ture; e.g. beatific, forsaken. "Rhetorical" items are associated with writing or speech on serious or elevated themes, especially on very formal occa­ sions (e.g. public meetings, state ceremonies). The use of such words elsewhere suggests a self-consciously pompous speaker or writer, e.g. tribulation ("event that causes suffering"), alas (expression of sorrow). A word or phrase used mainly in English literature, and not in normal speech or writing. Words and phrases which are mainly used in literature, (for example in a novel or a play), or when writing in a literary way: The valley was enshrouded (=covered) in thick mist.

2.

poetic poetic

LDOCE CIDE

A word that is used mostly in poetry. Literary words usually found only in poetry: The zephyr (=light wind) wafted a sweet perfume of flowers  'er (= over) the fields.

3.

biblical

LDOCE

A word that is used in the language of the Bible, and would sound old-fashioned to a modern speaker.

4.

formal

COBUILD

formal

OALD

formal

LDOCE

formal

CIDE

Used mainly in official situations, or by political and busi­ ness organizations, or when speaking or writing to people in authority; e.g. belated, demonstrable. Words chosen when speaking or writing in a serious or official context to someone who is not a close friend or relation. A word that is suitable for formal speech or writing, but would not normally be used in ordinary conversation. Words and phrases used in a serious way, for example in business documents, serious newspapers and books, lectures, news broadcasts. Formal language is often used when people want to appear polite. Example: He is anxiously awaiting (^waiting for) the result of the medical tests. Guests are requested to comply with (= obey) all the fire and safety rules.

informal

COBUILD

5.

Used mainly in informal situations, conversations, and personal letters; e.g. decaf elbow room.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

6.

7.

8.

informal

OALD

informal

LDOCE

informal

CIDE

slang

OALD

slang

LDOCE

slang

CIDE

derogatory

OALD

disapproving

CIDE

(!) [=taboo] OALD

387

Informal expressions are used between friends or people who know each other well, in a relaxed or unofficial context. They are not appropriate for formal situations. Examples are brainy, dad, umpteen, wallop. A word or phrase that is used in normal conversation, but may not be suitable for use in more formal contexts, particularly in writing e.g. essays or business letters. Used with friends or family or people you know in relaxed situations. Informal words are more common in speech than in writing. She works as an admin assistant (= adminis­ trative assistant). She always stops for a coppa (= cup of coffeej at 11 o'clock. It might be considered not polite to use very informal language in formal situations. indicates "slang" words and senses, i.e. inventive and often colourful items generally used in a very informal spoken context. Such items usually belong to, or originate in, the language of a particular social or occupational group (e.g. soldiers, nurses, prisoners), e.g.: nick ("prison"), boozer ("pub"), scarper ("go away"). A word or phrase that is used by a particular group of people, but is not normally used by most people. Informal language which might include words which are not polite. Slang is often used between members of a particular group when speaking together, and might stay in use only for a short time. That's a lot of bullshit (= non­ sense). The fuzz (= The police) have searched my flat four times this year. "Derogatory" words, etc. imply that one disapproves of or scorns the person or thing referred to or described by those words, e.g. puerile, skulk, suspect (adjective). Words usually used in a negative way, showing that the speaker has a bad opinion or feelings: She seems to have a very clinical (= emotionless) attitude towards her children. She's a bit fuzzy (= not easily satisfied) about what she '11 eat and what she won't. (!) denotes words or senses likely to be thought offensive or shocking or indecent (though not necessarily by everyone or on every occasion), e.g. wop, nigger, Christ!; fuck, prick, shit, piss. Foreign learners should exercise great care in using these words. They should also note that words such as wop and nigger are generally used with the deliberate aim of giving offence.

388

9.

WOLFGANG K. HÜNIG

offensive

COBUILD

Likely to offend people, or to insult them; words labelled offensive should therefore usually be avoided; e.g. berk, commie. This label denotes words used to address or refer to people, usually with the deliberate intention of offending them, especially on account of their race or religion. Words such as spick, wop, nigger are almost always used offensively in this way; words such as arsehole and prick are found shocking, but they need not be used as terms of abuse. A word that should not be used because it is very rude or offensive. Words which are likely to offend someone and are not used in formal situations. Certain words referring to sex or sexual organs, excretion and people's nationality or race can be particularly offensive. Strong swear words are marked taboo.

offensive

OALD

taboo

LDOCE

taboo

CIDE

10.

sexist

OALD

This label denotes words and phrases that express a (some­ times unconscious) discriminatory or patronizing attitude towards someone of the opposite sex. They are almost always words, used by men about or to women, and can be used to express approval in a "man-to-man" context, e.g. dolly, dumb blonde, a bit of skirt/crump et/all right, an easy lay.

11.

approving

OALD

approving

LDOCE

approving

CIDE

"Approving" words, etc imply the opposite of derogatory ones; they suggest approval or admiration of the thing or person referred to or described, e.g. petite, slender,, spotless. A word that is used to praise things or people, although this may not be clear from its meaning. Words usually used in a positive way, showing that the speaker has a good opinion of someone or something, or good feelings towards them: Her designs were always original and imaginative. This steak is beautifully tender (= easy to cut).

12.

euphemistic

OALD

"Euphemistic" words, etc are ones chosen to refer to something unpleasant or painful in a pleasant (because more indirect) way, e.g. pass away ("die"), senior citizen ("old age pensioner").

13.

jocular

OALD

humorous

LDOCE

Jocular words and phrases are intended to be funny, whether grim or innocent humour is meant, e.g. push up the daisies, Alma Mater, have a plum in one's mouth, put one's foot in it. A word that is normally used in a joking way.

STYLE LABELS IN ENGLISH LEARNER'S DICTIONARIES

389

humorous

CIDE

Words or phrases which are often intended to make people laugh or smile. Some are informal or slang words: With a pot (= big stomach) like this, do I look as though I get any exercise? Often a dated or formal word is used humorously in an informal situation: Let me take you to my humble abode (= home).

14.

ironic

OALD

This label denotes words , often used within a longer phrase, that are intended to convey a sense opposite to the apparent sense, e.g. fine (as in a fine mess), lovely (as in lovely black eye).

15.

old-fashioned dated

COBUILD

Generally considered to be old-fashioned, and no longer in common use; e.g. bawdy, frock. Certain words (e.g. court (verb), gramophone) are still used by older speakers but not by the majority of younger ones. These are words passing out of use and they are labelled dated, A word that was used earlier in this century, but would sound old-fashioned today. Words or phrases which sound old-fashioned, not modern: My grandmother still calls the radio "the wireless ".

16.

17.

18.

OALD

old-fashioned dated

LDOCE

archaic

OALD

old use old use

LDOCE CIDE

spoken

COBUILD

spoken

LDOCE:

written

COBUILD

CIDE

Other words (e.g. thou for "you" knave in the sense "scoundrel"), though found in books written in the first half of this century or earlier , have now passed out of use alto­ gether. These are labelled arch ("archaic"). A word used in earlier centuries. Used before the 20th century but now rare: She suffered from melancholia (= a condition of great sadness) after the untimely death of her father. Used mainly in speech rather than in writing; e.g. school kid, whoops, - not defined Used mainly in writing rather than in speech; e.g. animus, bespectacled.

NAME INDEX

A Abraham, Werner, 107-109, 111, 118, 203-204 Ágel, Vilmos, 99, 102, 113, 118, 214 Aitchison, Jean, xix Akwey, Sam A., 186 Algeo, John, 71,94 Alowo, Jane, 186 Ammon, Ulrich, 375 Andresen, Karl Gustaf, 109 Anglin, Jeremy M., 132 Aristotle, 274-276

Banks, David, 124 Barcelona, Antonio, xx, 224, 226, 2 3 1 232,250-252 Barlow, Michael, 70, 78, 92-93 Barsalou, Lawrence R., 61, 147 Bartels, Gerhard, 203 Bates, Elizabeth, 78 Bauer, Laurie, 70, 76-77 Behaghel, Otto, 109 Bellugi, K., 127 Belmechri, Fa'iza, 333 Beneke, Jürgen, 322 Bernstein, Basil, xiv Bertram, Mathias, 217 Biber, Douglas, 375 Bierwisch, Manfred, 101 Blum-Kulka, Shoshana, 324, 326 Boers, Frank, 315 Boguslawski, Andrzej, 155 Bonheim, Helmut, 339 Bowerman, Melissa, 40, 127, 146

Brisard, Frank, 12 Brooks, Lee, 147 Brown, Penelope, 367 Brown, Roger, 46, 127 Budwig, Nancy, 124, 127-129, 135-140, 143, 146-147 Bukenya, Gerold, 352 Bultinck, Bert, 231 Buscha, Joachim, 203 Bybee, Joan L., 59, 78-81,94  Cannon, Garland, 71 Carey, Kathleen, 3 Carlson, Gregory, 51, 64 Carroll, Lewis, 70, 72 Cervantes, Miguel, 250 Chappell, Hilary, 124 Chomsky, Noam, xiv, 101, 117 Cienki, alan J., 156 Clark, Eve, 42 Claudi, Ulrike, 182,210 Collins, Peter , 125-126, 135, 143, 146 Comrie, Bernard, 52 Crazzolara, J. P., 186 Creider, Chet, 199 Croft, William, xx, 52, 54, 61, 64-66, 229, 234-236,248,251-252 Crookes, Graham, 331 Crystal, David, 337, 339 Cuenca, Maria Josep, 234 Cuyckens, Hubert, 67, 250 D Dabrowska, Ewa, 157, 170, 174

392

NAME INDEX

D'Andrade, Roy, xix Dante, 250 Davidson, Iain, 45 Di Meóla, Claudia, 204, 214, 220 Diewald, Gabriele, 210 Dirven, René, xv-xvi, xx, 45, 124, 126— 128, 138, 140, 143,251,259 Dörnyei, Zoltán, 331 Dunaj, Boguslaw, 156 Dürscheid, Christa, 118 E Eco, Umberto, 264 Edmondson, Willis, 321, 324, 326, 329, 333-336, 339 Eggins, Suzanne, 380 Eisenberg, Peter, 106 Engel, Ulrich, 100,203 Erben, Johannes, 100-101, 109, 117 Esa Itkonen, 94 Esser, Jürgen, 376 F Fairclough, Norman, 339 Fanselow, Gisbert, 101 Fauconnier, Gilles, 69-71, 83, 85, 92-94, 232,238,251 Felix, Sascha, 101 Fenson, Larry, 133 Feyaerts, Kurt, 233 Fillmore, Charles, xiv, 50 Firth, Alan, 324-327 Fisher, Cynthia, 36, 45 Flis, Andrzej, 293 Flora, June A., 127 Fodor, Jerry, 45 Fraser, C , 127 G Gardner, Robert, 330 Gee, J.P., 124, 135 Gibbs, Raymond W., 225, 277 Gil, Alberto, 315 Gillette, Jane, 36

Givón, Talmy, xviii, 124-128, 135-136 Gleitman, Henry, 36, 45 Gleitman, Lila, 27-29, 34, 36-39, 45, 51 Glick, Ruth, 348, 350, 364 Gnutzmann, Claus, 339 Goddard, Cliff, 35 Goldberg, Adele, 37, 4 2 ^ 3 , 50-54, 5 6 62, 64, 67 Goldsmith, John, 19 Goossens, Louis, xxii Graddol, David, 322 Gray, Bennison, 276 Greenberg, Joseph H., 349 Gregory, Michael, 373-374 Grewendorf, Günther, 117-118 Grimm, Jacob, 109 Griper, Clive, 348, 350, 364 Groeben, Norbert, 329 Grotjahn, Rüdiger, 329 Grzegorczykowa, Renata, 155 Günther, Hartmut, 101 H Haider, Hubert, 117 Haiman, John, xviii Hallet, Garth, 274 Hallett, Garth, 276 Halliday, M.A.K., 189 Handke, Jürgen, 116 Hankamer, Jorge, 66 Hansen, Klaus, 71, 106 Harris, Frances, 127, 145 Hartmann, R.R.K., 369 Harvey, Keith, 370, 371 Hayakawa, S.L, 277 Heckhausen, Heinz, 331 Heidolph, Karl Erich, 118 Heine, Bernd, 182, 193, 199, 210 Heinrich, Gertraud, 118 Heisenberg, Werner, 290 Helbig, Gerhard, 100, 110, 115-116, 118, 203 Hermet, Heidi, 315 Hentschel, Elke, 107 Heringer, Hans-Jürgen, 100, 117

NAME INDEX

Herold, R., 124, 126, 128 Herskovits, Annette, 259 Higuchi Goto, Mariko, 23 Hilferty, Joseph, 234 Hinds, John, 265 Hofsted, 327 Höhle, Turnan N., 103 Hopper, 146,210,219 House, Juliane, 305, 311, 321, 324, 326328,335,336 Hübler, Axel, 315 Hüllen, Werner, 322, 328 Hummel, Kirsten, 333 Hünnemeyer, Friederike, 182, 210 I Ikegami, Yoshihiko, 261, 265 Israel, Michael, 78 J Jackendoff, Ray, 45, 51, 54 Jacobs, Joachim, 99, 102-104, 113, 117 Jäkel, Olaf, 315-316 Jakowicka, Elžbieta, 156 Janda, Laura, 154, 156-157, 161, 166, 169-170 Janicki, Karol, 275, 279 Jaszczolt, 327 Jeffrey Elman, 94 Johansson, Stig, 125-126 Johnson, Mark, 14, 35, 182-183, 190, 193, 223-228, 238, 259, 277, 299, 301, 3 1 5 316  Kachru, Braj, 339 Kachru, Yamuna, 339 Kalėta, Zofia, 154 Kalisz, Roman, 169 Kallenbach, Christiane, 329 Kasper, Gabriele, 324, 326 Kay, Paul, 50 Keenan, Edward L., 124, 135 Kernmer, Suzanne, 70, 78, 93

393

Kiguli, Susan, 348 Kindaichi, Kyosuke, 260 Kitcher, Philip, 290 Kiwanuka, P., 347 Knapp, Karlfried, 322 Knappert, Jan, 186 Komatsu, Lloyd, 45 König, Ekkehard, 157, 163, 209 Kortmann, Bernd, 209 Korzybski, Alfred, 277 Kövecses, Zoltán, xx, 226, 233, 236-238, 243-245, 250-252, 259 Kramsch, Ciaire, 336 Kroskrity, Paul V., 260 Krzeszowski, Tomasz, 316 Kubinski, Wojciech, 169 Kurzowa, Zofia, 154, 155 L Ladefoged, Peter, 348, 350, 364 Lakoff, George, xix, 14, 50, 147, 182-183, 190, 193, 200, 223-228, 233, 236-238, 250, 259, 274, 277, 293, 299, 301, 3 1 5 316 Lakoff, Robin, 124, 128, 135 Lamb, Sydney, 93-94 Lambert, William E., 330 Lambrecht, Knud, 5, 54 Landau, Barbara, 51 Lang, Jürgen, 204 Langacker, Ronald W., xiv, xx, 3-5, 9-10, 14, 16, 18, 35, 4 4 ^ 5 , 50, 52, 56, 60, 70, 78, 80, 83, 93-94, 124, 159, 166, 175, 182, 187, 208, 229, 230, 234-235, 241,248,250-252 Langfeldt, Birgit, 315 Lehmann, Christian, 210 Lehrer, Adrienne, 71, 92 Leirbukt, Oddleif, 107 Lepore, Ernest, 45 Lesnyak, Agnes, 325-326 Levelt, Willem J. M., 116 Levin, Beth, 50, 62 Levinas, Emmanuel, 327 Levinson, Stephen C , 367

394

NAME INDEX

Levinson, Stephen C , 367 Leys, Odon, 220 Lindqvist, Christer, 203, 214 Lippi-Green, Rosina, 260 Lyons, John, 264 M MacWhinney, Brian, 46, 78 Maratsos, Michael, 128 Marchman, Virginia, 127-130, 135-140, 143,147 Markman, Ellen, 45 Marslen-Wilson, William, 80 Mastalska, Danuta, 284, 287 Matovu, Kasalina, 348 Matsumoto, Yo, 5 Mazrui, Alamain M., 354 Mazrai, Alamin M., 350 Mazrui, Ali A., 350-351,354 Medin, Doug L.5 147 Meierkord, Christiane, 324, 326-327 Meints, Kerstin, 129-131, 133, 145, 147 Michaelis, Laura, 54 Mukama, Ruth, 347-348, 351, 353 Müller, Wolfgang, 214 Munro, Pamela, 124 N Ndoleriire, Oswald, 348 Neisser, 147 Neumann-Holzschuh, Ingrid, 204 Nganda, Cecilia Namulondo, 364 Niemeier, Susanne, 259 Nikiforidou, Kiki, 156 Noble, William, 45 Nomura, Masuhiro, 267-268 Noonan, Michael, 186 Nsibambi, Apolo, 362  Obbo, Christine, 364 Ogot, Grace, 183, 199-200 Okoth Okombo, Duncan, 179, 184, 187188

Okoth-Okombo, Duncan, 199 Oksefjell, Signe, 125-126 Oxford, Rebecca, 87, 137, 140, 331 P Panther, Klaus-Uwe, xx, 225 Passent, Daniel, 283, 289 Peccei, Jean Stilwell, 42 Pennycook, Alastair, 339 Petig, William E., 214 Phillipson, Robert, 339 Piernikarski, Cezar, 156 Pietrzak, Barbara, 285 Pineros, Carlos-Eduardo, 79 Pinker, Steven, 36-37, 39, 45 Plank, Frans, 108, 118 Plato, 274 Popper, Karl, 274-277, 290 Pütz, Martin, 259 Q Quine, Willar Van Orman, 27 Quirk, Randolph, 119, 124-128, 135-136, 138-139,373-374,382 R Radden, Günter, xiii-xxiii, 9, 43, 70, 124, 126-128, 138, 140, 143, 182, 226, 232233, 236-238, 243-245, 250-252, 259, 266,316 Rahmig, Martina, 118 Rakowiecki, Jacek, 282 Rappaport Hovav, Malka 50, 62 Reddy, Michael J., xxiii, 259, 267-268 Regier, Terry, 37 Reh, Mechthild, 186, 199 Reis, Marga, 107 Révész, G., 269 Rohde, Ada, 93-94 Romaine, Suzanne, 124, 135 Rommetveit, Ragnar, 127, 139 Rosch, Eleanor, xix, 28, 132, 140 Royal Skousen, 94 Rubba, Johanna, 76, 92

NAME INDEX

Ruiz de Mendoza, Francisco, 235-237, 239,242,251 Rumsey, Alan, 260 Rytel-Kuc, Danuta, 118 S Sadzinski, Roman, 118 S apir, Edward, 321 Saussure, Ferdinand de, 37-38 S chaffer, Marguerite M., 147 Schenkel, Wolfgang, 100, 110, 115-116, 118 Schibsbye, Knud, 124 Schieffelin, Bambi . , 260 Schmid, Hans-Jörg, 277 Schmidt, Richard W., 331 Schmitt, Christian, 315 Schröder, Jochen, 203 Schupbach, Richard D., 62 Shakespeare, William, 250 Shearin, Jill, 331 Shibatani, Masayoshi, 124 Silverstein, Michael, 260 Skehan, Peter, 331 Skutnabb-Kangas, Tove, 339 Slade, Diana, 380 Šmiech, Witold, 153-158, 168, 170, 174175 Smith, Michael . , 220 Solmecke, Gert, 331 Soudanek, Lev. I., 71 Ssekamwa, J.C., 350-351, 353 Stafford, R. L., 179, 182-186, 188, 195, 199 Standwell, G., 124-125 Staniewicz, Ewa, 280 Stawicka, Joanna, 281 Stef ano witsch, Anatol, 93-94 Steinitz, Renate, 111, 118 Storrer, Angelika, 99, 102, 117 Sussex, R., 126, 140, 143 Svartik, Jan, 125 Svartvik, Jan, 124-127, 135-136, 139140, 143 Svorou, Soteria, 182-193, 197

395

Sweetser, Eve E., xx, 9-11, 93 T Tabakowska, Elžbieta, 161, 164, 175 Tajfel, Henri, 327 Talmy, Leonard, xx, 4-5, 9, 18 Tanenhaus, Michael K., 51, 64 Tannen, Deborah, 292 Tarno w, Birgit, 203 Tasch, Meike, 347, 364 Taylor, John R., xxii, 36, 45, 208, 230, 237, 243, 244 Tesnière, Lucien, 100, 113, 117 Thompson, 146 Thornburg, Linda, xx, 225 Tokieda, Motoki, 265 Tomasello, Michael, 60 Toyoda, Kunio, 261 Traugott, Elizabeth Closs, 5, 157, 163, 210 Tucker, Archibald N , 180, 182, 184-186, 188-189, 195, 199 Turner, Mark, 69-71, 83, 85, 92, 94, 127, 139, 223-224, 226-228, 237-293 U Ukoko, Joseph, 186 Ullmann, Stephen, 266, 269 Ungerer, Friedrich, 277 Uriat, ChamU., 186 V Van Spaandonck, Marcel, 186 Vater, Heinz, 101-103, 111, 116-118 Vorlat, Emma, 381 W Wagner, Fritz, 118, 324-327 Wald, Benji, 348 Walusimbi, Livingstone, 348, 352, 362 Wegener, Heide, 107 Weinrich, Harald, 203 Weinsberg, Adam, 158-159, 161-162, 164, 174-175 Weisgerber, Leo, 118

396

Welke, Klaus, 100, 118 Wells, G., 133 Westermann, Diedrich, 186 Weydt, Harald, 107 Widdowson, Henry, 335, 339 Wierzbicka, Anna, 35, 45, 293 Wilkins, Wendy, 27, 39, 40-41 Willems, Klaas, 214 Wittgenstein, Ludwig, 27, 276 Woisetschlaeger, 19 Woolard, Kathryn, 260 Woolard, Kathryn A., 260

NAME INDEX

Wróbel, Henryk, 155-156, 169-170 Wunderlich, Dieter, 100, 102 Y Yang, Lynne, 124-128, 135-136 Yuill, Deborah, 370-371 Z Zifonun, Gisela, 101, 106, 203 Zirimu, Pia, 350

SUBJECT INDEX

A acquisition, 29, 51, 60, 123, 127, 130, 132, 134, 136, 139, 144-146, 323, 328, 330 activation, 78, 86 active zone, 248, 252 adlative motion/sense, 160, 163 adposition, 182, 206, 209, 211-217, 219220 diachronic evolution, 9, 157-158, 203, 217-218 spatial, 217 adpositional semantics, 215 adpositional system, 203 affix, 52-53, 61-62, 76 agent, 9, 53-55, 100-101, 103-108, 110— 111, 113, 115-116, 118, 125-126, 129, 131-133, 135-136, 138, 144, 147, 169170, 229, 250 American English, 126-138, 139, 143, 145-146, 375 arbitrary, 18-19,28,89, 174 argument structure, 36, 38, 42, 44, 50, 5 2 53, 60, 62, 64-66 aspect marker, 153-154, 157 atemporal relation, 161-166 Australian English, 126, 146 autonomy of syntax, 39 axiological load, 273, 275, 278

base, 4, 31,289, 299, 327 basic level, 28 ^-passive, 123-127, 129-130, 135, 140141, 143, 145 bleaching, 210, 215

blend. See lexical blend blended domain, 86 blended space, 85 bootstrapping, 28, 34, 36-39, 44, 51, 60 British English, 123, 126, 136, 139, 142146, 369 building-block model, 91-92 morphology, 76-77 business communication, 324-325  case marking, 156, 159, 172, 183 categorization, 28, 45, 224, 247 category conceptual, 225, 236, 278, 316 semantic, 94, 155 syntactic, 37, 101 caused motion verb, 37, 40, 43 centrality, 125, 223, 234-235, 239, 241, 248 change of position, 211, 217, 219 cognitive domain, 84, 226, 230, 233, 245 Cognitive Grammar, 3, 69, 83, 87, 91, 158 Cognitive Linguistics, 9, 50, 70, 153, 156158, 172, 174, 179, 182, 193, 200, 223, 224, 226, 230, 239, 245, 249, 259, 277 cognitive verb, 33-34 collocation, 168, 370-371 common core meaning, 154 communicative context, 367, 381 complement vs. adjuncts, 102 component structure, 9 composite structure, 9 composition, 63, 69, 83, 336 compositionality, 35, 51, 61, 76, 86

398

SUBJECT INDEX

compounding, 69, 76 computing parsimony, 61 conceptual essentialism. See essentialism conceptual metaphor. See metaphor, conceptual conceptual metonymy. See metonymy, conceptual conceptualization, 14, 20, 91, 162, 163— 165, 169-170, 175, 187, 189, 193-194, 197, 233, 234-235, 241, 251, 260, 2 6 3 269, 297, 298, 301, 303-305, 307, 309, 311,314 conceptualization of language, 259, 263, 267, 269 conceptualizer, 4-5, 6, 8, 11-16 conceptualizing activity, 4, 6, 14-15 conflict, 79, 127, 274-277, 293, 364 abortion, 273, 279, 280-292 christian values and censorship, 273, 284-287 constituent, 38, 40, 100-101, 105, 108, 117-118, 183,204-205 constructional polysemy, 54, 57 constructional schema, 56, 59, 82 constructions, 13, 44, 49-54, 57-66, 78, 80,92, 102-106, 109, 111, 113, 117— 119, 130, 133, 144-145, 206, 252, 300, 371,375 content word, 206-208, 219 context, 3, 10, 12, 18, 28, 45, 65, 86-87, 102, 156, 166, 188, 208, 231, 252, 274, 290, 301, 311, 323, 328, 337, 338, 357, 370, 373, 374 extralinguistic, 27, 29, 36, 44, 288 linguistic, 29, 34, 36, 38-39, 43-44, 288-289, 380 social, 381 of use, 35, 38, 65 cultural-irrelevance hypothesis, 323, 336 D definition, 8, 13, 16, 20, 28, 35-36, 69, 103-104, 118, 146, 158, 223, 228, 237, 239, 245-249, 252, 260, 273, 275-278,

280-282, 284, 286, 288-293, 323, 337, 368, 371-374, 382. See also ostension dependency framework, 100-101, 117 derivation, 67, 69, 76, 78, 155, 175 diathesis, 110-112, 114 alternation, 99-100, 102, 113, 116 dictionary lexicographer's perspective, 369 user's perspective, 369 entry, 370 monolingual, 298, 368, 370, 372 directed motion, 30 discourse context, 367, 381 distal, 12, 161-162 ditransitive construction, 49-50, 52-60, 62-65 domain, 11, 80, 84-88, 92, 94, 182, 191192, 195, 198, 223, 226-235, 238-242, 244-252, 298, 307, 314-316, 334, 337338,360,362,371,380 conceptual, 36, 83, 91, 200, 240 economic, 297-307, 309-316, 322 domain highlighting, 196, 232, 234-235, 239, 248, 252 E EFL, 322-323, 325, 328, 330, 335, 339, 368 ELF, 321-328, 330, 335-336, 338-339 communication, 323, 325, 327-328, 335,338 -oriented communicative strategies, 323 encyclopedic, 44, 230, 234 English as (a) lingua franca. See EFL English-as-a-Foreign-Language. See EFL entrenchment, 69, 79, 88-89, 91 epistemic, 3, 11-14, 17, 20, 23 essentialism, 273-277, 279-284, 286-293 experience, 12-15, 23, 78, 80, 190, 191, 224-226, 273, 278-279, 291, 314, 316, 321, 323, 330, 331, 333-334, 361, 370 experiencer participant, 190 extension, 6, 13, 33-34, 45, 54, 74, 82, 88, 90-91, 93, 158, 160, 162-166, 168171, 173, 183, 188, 194, 197, 224, 233

SUBJECT INDEX

F family resemblance, 49, 57 force-dynamic, 5, 9, 10, 12-15, 23 foregrounding vs. backgrounding, 161 form-function/meaning mapping, 66 frequency, 42, 74, 77, 79-80, 88, 92, 125, 127, 130, 133, 135, 139, 142-143, 146, 165,213,314,358,377 function word, 207-208, 219 fuzzy categories/concepts, 125, 132,208, 277, 287, 290 G generalization, 19-21, 64, 77-78, 155, 174 generative grammar, 49, 79-80, 110, 117 get-passive, 123-127, 130, 133-135, 137, 139-141, 143, 145, 146 adversative character, 125, 127-128 Goal, 40-43, 183, 187-198 governee, 100, 117 governor, 100 grammatical meaning, 154, 158, 171 grammaticalization, 3, 6, 10, 94, 157-158, 163, 174, 179, 182, 186-187, 193, 2 0 3 204, 207-211, 213, 215, 217, 219-220 grammaticization. See grammaticalization ground, 3-4, 7-9, 12, 21, 23, 45, 219, 249, 311,336,362 grounding, 3, 6-9, 14, 23, 279 grounding predication, 7-9 H historical present, 17, 21, 23 homonym, 262-263 I Idealized Cognitve Model, 233, 240, 244245, 250, 298 ICM. See Idealized Cognitive Model idioms, 297 image schema, 14, 179, 182-183, 188189, 197, 233, 238, 301-302, 314-316 imperative, 5, 30-31, 38, 45, 117, 199, 333

399

imperfective process, 16, 32 imperfective verb, 16, 153 inflection, 76, 210, 370 intercultural awareness, 323 intercultural communication, 322, 323, 335,339 intercultural miscommunication. See intercultural communication intersubjectivity, 278 L Landeskunde, 322 landmark, 159, 160-166, 168-171, 174175, 179, 183, 186-193, 197,307 language acquisition, 28, 29, 40, 51, 60, 290 language and reference, 263-264 language and the speaking subject, 259, 264-265 language ideology, 259-260, 265 language vs. heart/mind, 259, 264- 268 layering, 219 lexical blend, 69-83, 85, 87-94 intercalative, 72-73, 76-77, 81, 93 overlap, 75, 80-82, 89, 91 substitution, 74-75, 91 lexical rule, 49-52, 60-61 approach, 49, 50-51 lexical rule analysis, 52, 60 lexical source, 75, 77, 183, 186, 197, 199 lexicographer, 35, 369 lexicographers, 29, 156, 174, 371 lexicon, 28, 35, 37, 52, 55, 70, 92, 93, 99, 102-103, 104, 116 lingua franca, 321-323, 328, 335-336, 338,350,363 location, 7, 20-21, 30, 33, 39, 40-41, 160163, 173, 182, 187, 189, 211, 224, 251 Location, 40, 43 locus of potency, 10-11, 14 M mapping, 64-66, 69, 195, 200, 223-224, 226-234, 237-239, 242, 244-245, 247249, 251-252, 297, 328, 334-335

400

SUBJECT INDEX

medium of instruction English, 347-348, 350-352, 355- 356, 358-362, 364 indigenous language, 351 Kiswahili, 347-348, 350-352, 3 5 5 356, 358-364 Luganda, 347-352, 355-356, 358-364 mother tongue, 323, 347, 350-352, 355-356, 358-362, 364, 370 mental activation, 223, 232, 234, 241-242, 245, 247 mental scanning, 6, 15 metaphor, 11, 19, 60, 70, 71, 85, 160-161, 163, 165, 171, 189, 190-191, 193, 195197, 224, 228, 233, 236-240, 249, 250251, 260, 266, 286, 300, 303, 305, 307308, 309 cognitive approach to, 297 conceptual, 179, 182-183, 200, 297, 298,302,312-316 conventional, 230, 297 metaphor vs. metonymy, 227, 229, 231-232,234 structural metaphor, 183, 189-193, 197, 200, 299 metaphorical conceptualization, 304, 314 expressions, 189, 236-237, 313-314 extension, 161, 165, 168, 171 model, 298, 300-302, 313-314 motion, 298, 315 transfer, 182, 189-190, 193, 196, 260, 266 metaphorization, 259, 266, 268-269, 312 of language, 267 metonymic relationship, 188 source, 233, 237, 242 target, 242-243 transfer, 182 metonymy, 85, 224, 226-241, 244- 252, 316 and contiguity, 225 and experiential motivation, 225 as a conceptual process, 224 as a stand-for relationship, 226

conceptual, 223, 242-243 conventional, 223-224, 227, 230, 232, 237, 241-242, 246-247, 249-250 conventional vs. non-conventional 2 4 3 245 occurring in a single domain, 226 peripheral, 239, 241, 247-248, prototypical, 223-224, 232, 240-249 propositional metaphor, 190,200 referential, 224-225, 227-228, 2 3 5 236, 239, 240-242, 244-245, 252 schematic, 223-224, 239-240, 243, 245-249, 252 standard account of, 223-224, 228 typical, 223-224, 239-249 middle construction, 109-110, 113 reflexive, 111, 114 with a copula, 112, 114 modals, 3, 7, 9, 12-13, 15,23 deontic, 10, 113 root, 10-11, 14, 77, 88, 89, 92, 228 modulation, 55-57, 62-64 monosemy, 56 morphological productivity, 69, 73, 153, 175 morphology, 42, 71, 83, 91, 94, 174, 298 non-concatenative, 76-77, 92 motivation, 13, 33, 54, 69, 70, 80, 91, 154, 158, 169-170, 173-174, 182, 197, 199, 227, 232, 236, 243-244, 250, 260, 264, 266-267, 269, 275, 308, 312, 316, 323, 326, 329, 330-334, 338, 362, 369 N neologism, 69-71, 87, 92, 93-94 network analysis, 158, 174 non-congruent expressions, 189, 197 non-spatial meaning, 187, 191,215 norms, 214, 322-324, 327, 335-336, 3 3 8 339

 objective construal, 3-6, 8-10, 18 objective truth, 275 objectivity, 4-5, 7, 14

SUBJECT INDEX

obligatory vs. optional complements, 104 ontological metaphor, 179, 183, 193, 195196,200,316 Optimality Theory, 79-80 ostension, 27-29, 35, 39 p paragon, 236, 242, 251 parsimonious, 51, 64 parsimony, 51, 60, 64 computing, 61 storage, 61, 64 passive, 53, 78, 99, 106, 110-111, 118, 125-126, 138, 141, 309, 314, 316, 375 acquisition of, 123, 127, 129, 130, 132, 134, 136, 139, 144, 145, 146 atypical, 129, 131, 133, 135, 142 bekommen-, 107, 113 impersonal, 108 prototype approach to, 124, 130, 131 prototypical, 107, 123, 128-137, 139140, 142-145 sein-, 107, 113 stative, 107 elicitation of, 128, 132, 134, 137, 140, 143,147,277 fuzzy-set approach, 125 impersonal, 108 impersonal reflexive, 109 path, 5-6, 30, 40, 45, 175, 215, 243, 298, 307,314,316 patient, 54-55, 100-101, 104, 107-108, 110, 111, 113, 115-116, 123-129, 131133, 135-139, 143-145, 147,316 perception verb, 34 perfective event, 31-32 perfective verb, 16, 155, 169 perfectivizing prefix, 166 peripheral, 4, 76, 123, 125, 130, 132-133, 136, 139, 140-141, 145, 147, 163, 165166,336,339 Place, 40, 43, 183, 187, 198 polysemy, 54-56, 104, 156, 245, 249, 261-262

401

postposition/postposing, 203-214, 215, 217-220 potency, 9, 11, 14 pragmatic fluency, 328, 338 prefix, 62, 153-158, 166, 169, 171-172, 174-175 preposing, 203, 206-220 preposition, 29, 30, 34, 43, 45, 153, 156162, 166, 168, 172, 174-175, 179, 183189, 194-195, 197, 199, 208, 210, 214, 219 directional, 30, 34 non-spatial, 179 spatial uses of, 39-40, 42, 62, 67, 110, 128, 179, 183-184, 186, 189, 199 preposition-prefix correspondence, 157 present tense, 3, 6, 14-17, 20-21, 23, 67 primacy of space, 156, 158, 182 Primary Leaving Examination, 352 primary subdomain, 233, 240-241, 248, 251-252 profile, 4, 6-9, 13-14, 16, 30, 31, 252 progressive, 16-17, 31, 34, 38, 45, 101, 219 prototype, 13, 42, 94, 162, 168, 171, 220, 241, 278 prototypical, 12-13, 107, 162, 188, 208, 214-215, 220, 304. See also passive, prototypical transitive event/scene, 128-132, 135, 137, 139, 141, 144 prototypical passives, 140 proximal, 12, 161-162 R radial category, 144 reanalysis, 89, 91, 203-207, 219 redundancy, 61, 174 reference-point construction, 166, 252 register, 245, 263, 328, 370-371, 381 S scheduled future, 17, 21

402

SUBJECT INDEX

schema, 13, 56, 59, 69, 70, 76, 7 8 - 81, 87, 89-94, 179, 199, 229, 237, 268, 298, 314,316 schematic construction, 60-61, 63, 65, 66 search domain, 161-162, 165 secondary schools, 322, 337, 353 semantic change, 5, 87, 219, 381 semantic domain, 190 semantic network, 159-163, 166, 172, 175 semantic relatedness, 157-158, 172, 174, 179, 182 semantic reversibility, 129, 131, 132 semantic role, 10, 40, 101, 179, 183, 316 semantic/conceptual overlap, 64-65, 69, 73, 75, 76-77, 80-83, 85, 87-89, 91, 93, 163, 174-175, 273, 282, 334, 339 semantically empty, 153-154, 158, 172 situational label. See style label source, 10, 14, 40, 42-44, 69, 71-76, 81, 83, 85-89, 91, 94, 183-184, 186-190, 194, 197-198, 208-209, 227-243, 2 4 5 246, 248-249, 251-252, 297, 315-316, 329,333,371 source domain, 85-86, 190, 297 source lexeme/word, 71-77, 81, 83, 85, 86,88-89,91,94 spatial meaning, 159, 183, 187, 189, 191, 209, 215 spatial orientation, 161-162 spatial usage, 157, 197 spelling, 200, 364, 369, 370 static relation, 6, 18, 118, 160, 197, 216, 306,331 stative verb, 31, 45 storage redundancy, 61 structural match, 233 structuralism, 38, 61, 153, 156, 157, 158 style label, 190, 277, 290-291, 367-369, 371-372,374-382 subcategorization, 36-40, 42, 44, 117 subdomain extrinsic vs. intrinsic, 235 primary, 235, 241 secondary, 234-235, 239, 241, 251-252 subjectification, 3, 5-6, 9, 13-15, 23

subjective construal, 3—4,6-7, 10, 12, 14 subjectivity, 3-4, 7, 15-16, 264 symbolic ability, 28, 45 symbolic unit, 56, 82-83 synecdoche, 225 synonymy, 33, 263, 370 syntactic function, 100-101, 106-107, 114, 118 syntactic schema, 56 T taboo word, 367, 382 target, 3, 6, 10, 14, 30, 76, 87, 126-127, 138, 166, 227-231, 234-242, 245-249 251-252, 300, 314, 330, 335, 337, 339, 347, 369 target domain, 233, 297, 315 thematic role, 114 theme, 108, 110, 113,265 theta role, 102, 106-107 trajector, 160-165, 169-171, 175, 179, 183,187-190, 197 transformational analysis, 52, 116 Transformational Grammar, 101 typicality effect, 130, 136, 145-147 U Uganda's indigenous languages, 349, 350 upward movement, 303, 305 usage-based, 69, 80, 92 V valency, 102, 104, 172 description, 99-100, 115-116 dictionary, 99, 105 misery, 99, 102, 113, 117 potential, 99, 102, 103, 113, 116-117 realization, 99, 102, 107, 114-117 valency schema, 113 semantic, 100 syntactic, 100, 102, 117, 118 variety or style, 87, 123, 125-126, 134, 144, 146, 186, 274-276, 324, 334, 3 3 8 339, 367-369, 374-377, 381-382

SUBJECT INDEX

verb meaning, 27-28, 34-35, 45, 50-51, 53,59-61,63-65 verbal prefix, 153-158, 166, 168, 172, 174-175 adnominal origin, 153, 157 verb-class-specific construction, 58-60, 64,66 verbs of motion, 160, 163

403

verb-specific construction, 49, 58, 59-60, 62, 64-66 viewing arrangement, 7, 18, 20-23 W Werden-passive, 106 word meaning, 27, 29, 34-37, 43-44, 51, 273,290-291 words as labels, 278