Discourse and Perspective in Cognitive Linguistics (Current Issues in Linguistic Theory)

DISCOURSE AND PERSPECTIVE IN COGNITIVE LINGUISTICS

AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE General Editor E.F. KONRAD KOERNER (University of Ottawa) Series IV - CURRENT ISSUES IN LINGUISTIC THEORY

Advisory Editorial Board Henning Andersen (Los Angeles); Raimo Anttila (Los Angeles) Thomas V. Gamkrelidze (Tbilisi); John E. Joseph (Edinburgh) Hans-Heinrich Lieb (Berlin); Ernst Pulgram (Ann Arbor, Mich.) E. Wyn Roberts (Vancouver, B.C.); Danny Steinberg (Tokyo)

Volume 151

Wolf-Andreas Liebert, Gisela Redeker and Linda Waugh (eds) Discourse and Perspective in Cognitive Linguistics

DISCOURSE AND PERSPECTIVE IN COGNITIVE LINGUISTICS Edited by

WOLF-ANDREAS LIEBERT University of Trier, Germany

GISELA REDEKER University of Groningen, The Netherlands

LINDA WAUGH Cornell University, U.S.A.

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA

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 Discourse and perspective in cognitive linguistics / edited by Wolf-Andreas Liebert, Gisela Redeker, Linda Waugh. p. cm. ~ (Amsterdam studies in the theory and history of linguistic science. Series IV, Current issues in linguistic theory, ISSN 0304-0763 ; v. 151) Selected rev. papers read at the 4th bi-annual International Cognitive Linguistics Confer­ ence held in Albuquerque at the University of New Mexico, July 16-21, 1995. Includes bibliographical references and index. 1. Cognitive grammar. 2. Discourse analysis. 3. Modality (Linguistics) 4. Metaphor. I. Liebert, Wolf-Andreas, 1959- II. Redeker, Gisela. III. Waugh, Linda R. IV. International Cognitive Linguistics Conference (4th : 1995 : Albuquerque, N.M.) V. Series. P165.D57 1997 415~dc21 97-39411 ISBN 90 272 3655 0 (Eur.) / 1-55619-866-3 (US) (alk. paper) CIP © Copyright 1997 - 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 75577 · 1070 AN Amsterdam · The Netherlands John Benjamins North America · P.O.Box 27519 · Philadelphia PA 19118-0519 · USA

Table of Contents

Acknowledgements

vii

Introduction Wolf-Andreas Lieben, Gisela Redeker, and Linda Waugh I. Cognition, Perspective, and Modality in Discourse Form-Use Mappings for Tag Questions Maria Josep Cuenca

3

The Social Dimension of a Cognitive Grammar Bruce Hawkins

21

Even, Sae/Sura/Mo as Constraints on Contextual Assumptions Isao Higashimori

37

Semantic Content and Depth of Intention: A Study in Cognitive Semantics Richard Hirsch

61

Perspective, Subjectivity, and Modality from a Cognitive Linguistic Point of View José Sanders and Wilben Spooren

85

II. Metaphors and Metonymy in Discourse A few Metaphorical Models in (Western) Economic Discourse Frank Boers and Murielle Demecheleer

115

The Spatial ization of Judgement Michele Emanation

131

vi

Stop Making Sense! Metaphor and Perspective in Creative Thinking Sessions of Scientists and Scientific Radio Broadcasts Wolf-Andreas Liebert

149

Conceptual Blending on the Information Highway: How Metaphorical Inferences Work Tim Rohrer

185

Speech Act Metonymies Linda Thornburg and Klaus Panther

205

III. Correlates of Discourse Structure Focus Movements and the Internal Images of Spoken Discourse Kenneth Holmqvist and Jana Holsánová

223

Pauses, Cognitive Rhythms and Discourse Structure: An Empirical Study of Discourse Production Joost Schilperoord and Ted Sanders

247

Subject Index

269

List of Contributor

Acknowledgements All papers in this publication were read and positively evaluated by two anonymous referees. We would like to thank these colleagues for the time they spent and for the careful analyses of the papers they reviewed. We would also like to thank the series editor, E.F. Konrad Koerner, and John Benjamins Publishers for the opportunity to publish our book. Our greatest debt of thanks is to our formatter, Anastasia Moraitis, for the large amount of work she did and the patience she had with us throughout. Finally, we want to thank Anke de Looper, our contact person at Benjamins, and René Dirven, president of the International Cognitive Linguistics Association (ICLA) from 1995 to 1997, who is responsible for the idea and the structure of the three volumes. Marjolijn Verspoor took on the difficult task of coordinating all three volumes—thank you, Marjolijn, for doing a terrific job. Finally, special thanks goes to Katharina Oppitz, our assistant at the Department of German Language, University of Trier, and to Lucie and Kater ina Piro from the Institute of German Language, Mannheim.

Introduction Wolf-Andreas Liebert, Gisela Redeker, and Linda Waugh University of Trier, Germany, Free University of Amsterdam, The Netherlands, Cornell University, U.S.A.

While cognitive models, perspective, and the construction of situated meaning have always been core concepts in Cognitive Linguistics, genuinely pragmatic approaches have long been lacking. Cognitive Linguistics research focused on grammar and semantics during the last two decades, but recently, pragmatic approaches have gained more and more interest, and "Discourse" and "Perspective" have become pivotal terms in today's Cognitive Linguistics. This volume is a compilation of current research in this area. A wide vari­ ety of discourse phenomena is covered from a Cognitive Linguistics' point of view, including the use and interpretation of metaphors, modal expressions, focus particles, tag questions, indirect speech acts, and iconographie textual ref­ erences, but also cognitive processes involved in discourse production. We have organized the contributions to this volume in the three clusters "Cognition, Per­ spective, and Modality in Discourse", "Metaphors and Metonymy in Dis­ course", and "Correlates of Discourse Structure". The contributions are based on papers read at the 4th bi-annual International Cognitive Linguistics Confer­ ence held in Albuquerque, University of New Mexico, from July 16 to 21, 1995. Together with its companion volumes "Lexicon and Grammar in Cogni­ tive Linguistics" and "Cultural, Psycholinguistic, and Typological Approaches in Cognitive Linguistics," this collection provides an overview of the broad range of cognitive linguistic approaches in the mid-nineties.

Cognition, Perspective, and Modality in Discourse "Form-Use Mappings for Tag Questions", by Maria Josep Cuenca, demon­ strates that the forms and meanings of tag questions are closely interrelated with each other. More specifically, she shows that tag questions combine assertion and interrogation syntactically, intonationally, and pragmatically (i.e., they have mixed modality); and that they are evidentials which manifest speakers' assess­ ment of the truth value of what they are saying, as well as their discursive atti­ tudes. She then goes on to show, on the basis of a cross-linguistic typology of form-use mapping for tag questions and in the light of grammaticalization the­ ory, that there is a gradient of tag questions, from less grammatical ized to more

χ

INTRODUCTION

grammaticalized ones, in conformity with the defining properties of tag ques­ tions and with the hypothesis that grammatical ization tends in the direction of more subjectification of meaning. In other words, tag questions exemplify grammatical ization of a conversational meaning by means of pragmatic inferencing and subjectification. In "The Social Dimension of a Cognitive Grammar", Bruce Hawkins ar­ gues that Cognitive Grammar has the potential to attend both to cognitive dy­ namics and to the socio-political/ideological aspects of language use. He then goes on to argue that as language learners we acquire not only the categories and expressions of our native language, but also the ideology of the collective or ruling entity of the particular speech community of which we are a part. All of this is done, according to his account of Cognitive Grammar, through experi­ ence, selection, and grounding (or deixis). On the basis of an examination of iconographie textual references, which construct an image of the referent which is intended to evoke a strong emotional response to that referent, he explores both icons (revered images) and caricatures (reviled images) and relates them to interpellation, the socio-political process through which the individual becomes subject to the world view of the prevailing ideology. In "Even, Sae/Sura/Mo as Constraints on Contextual Assumptions", Isao Higashimori explores the cognitive similarities and differences between the par­ ticle even and its Japanese counterparts sae, sura, and mo. He claims that they have in common a procedural meaning, namely the denial by the speaker of a contextual assumption believed to be present in the hearer. What differentiates them are the ways of processing to pick up a focused constituent (e.g., compari­ son of elements in the same category, extraction of only one exceptional case) and their different constraints on implications. For this analysis, he uses the framework of relevance theory. He explores problematic cases of the use of the particles and shows how these cases can be handled, with the use of relevance theoretic elements such as contextual assumptions, processing effort, contextual effects, and the nature of the focused elements (or the lack of focused elements). He concludes that even, sae/sura/mo are like traffic signals meant to affect the hearer's mental processings. "Semantic Content and Depth of Intention: A Study in Cognitive Seman­ tics" by Richard Hirsch contains an explication of the fundamental theoretical principles of the Nordic-Scandinavian school of semantic analysis, which was founded by the Norwegian philosopher Arne Naess. Hirsch illustrates how se­ mantic analysis within this tradition can be carried out on empirical linguistic data. The basic tenet of this approach is that speakers more often than not, are not completely clear on what they mean with what they say when first formulat­ ing their thoughts and feelings in linguistic expressions. Particularly, misunder­ standings can arise from differences in the cognitive content associated with the use of an expression. Naess called the number of cognitive distinctions speakers and addressees associated with a particular use of a linguistic formulation their

INTRODUCTION

xi

"depth of intention". The participants have to coordinate on the depth of inten­ tion and cognitive content in order to achieve a ratified agreement on what is meant by what is being said. The paper illustrates this coordination by showing how speakers "precisify", "specify", "paraphrase", and "elaborate on" contri­ butions in a construction task dialogue. In "Perspective, Subjectivity, and Modality from a Cognitive Linguistic Point of View" José Sanders and Wilbert Spooren combine the concepts of "subjectification", where the information in an utterance is connected to the speaker by the use of expressive predicates, and "perspectivization", where it is connected to concrete or abstract cognisant entities other than the speaker as, for instance, in quotation. This distinction allows the authors to represent the differ­ ence between 'objective' and 'subjective' uses of deontic modal expressions like "Jan can stay home": If an authority different from the current speaker is in­ voked as the source of (in this example) permission, this utterance involves per­ spectivization; if it is the speaker, it involves subjectivization (of course it can also simply be a factual statement). Epistemic modal expressions, by contrast, always involve subjectivity; in past tense uses, this is not the current but an em­ bedded speaker's subjectivity. That is, such utterances are both, subjectivized and perspectivized. Utterances with first-person mental state predicates, finally, involve a combination of perspectivization and the current speaker's own sub­ jectivity.

Metaphors and Metonymy in Discourse Frank Boers and Murielle Demecheleer investigate "A Few Metaphorical Mod­ els in (Western) Economic Discourse". The models they discuss include the source domains of physical motion (specifically, PATH, HEALTH and WAR). It is shown how a wide variety of figurative expressions can systematically be traced back to a limited number of conceptual metaphors. In this context, atten­ tion is also given to the relation between metaphor and metonymy. Samples of popular economic discourse were acquired from The Economist, The Financial Times, and equivalent newspapers in Dutch and French (a total corpus of over 130,000 words). A frequency count of the occurrence of these metaphors across the three languages points to a number of interesting cross-cultural differences. In her contribution "The Spatialization of Judgement" Michele Emanatian explores a small group of spatial metaphors in English, specifically those which elicit the concept of distance in the expression of judgement. Specific metaphors examined are the 'distance' of a proposition from the (location of) truth, and the 'distance' of a thinker from the subject matter. The first is interesting for what it reveals about how we understand truth: Basically, there seem to be two different conceptualizations of truth operation. First comes the truth we 'aim for' in our judgements, a truth whose location is absolute and can be known. Second, there

xii

INTRODUCTION

is the truth our judgements have, which may be measured in degrees of distance from absolute truth. The second metaphor is interesting for what it reveals about how we conceive of human understanding and involvement. We assess compe­ tence in making judgements partly by assessing degree of involvement, meas­ ured in distance terms. Our everyday view of the relationships between exper­ tise, bias, neutrality, and the likelihood of making accurate judgements seems to be equivocal. The 'closer' a person is to an issue, the more likely their judge­ ments about it will be true by virtue of their expertise. However, if someone is 'too close' to an issue, their interests may interfere with their ability to judge. Also implicated in these metaphors are our cultural models of how intellect and emotion interrelate. Wolf-Andreas Liebert discusses basic discourse patterns to show the emer­ gence of metaphor models in verbal interaction in his paper "Stop Making Sense! Metaphor and Perspective in Creative Thinking Sessions of Scientists and Scientific Radio Broadcasts". After discussing theories of metaphors, problem solving and perspective, he outlines the basic units necessary for the analysis of metaphors in verbal interaction. These include a two-turn dialogue unit consist­ ing of pragmatic operations of initiative and response, and semantic operations of expansion and adaptation of metaphors. These operations are applied to the talk of scientists who are discussing the metaphors in their own research topics, and to radio talk shows on science. The paper concludes that in-vivo studies of metaphorical reasoning have on the one hand a practical impact on creative problem solving and on improved comprehension of popular science texts or radio broadcasts. On the other hand, they provide new insight into the process of metaphor model emergence in verbal interaction, when viewing each speaker's playful and interactive linking and extending of their cognitive spaces. Are metaphors constitutive of social policy? How can cognitive semantics account for the differences between domains which seem to motivate revisions to social policy? Interpreting data collected from the news media and from ac­ tual policy statements by political figures such as U.S. Vice-President Al Gore, Tim Rohrer claims in his paper "Conceptual Blending on the Information High­ way: How Metaphorical Inferences Work" that metaphorical inferences produce a kind of conceptual blending which takes place between the domains. This then motivates changes in the world to fit our metaphorical understanding. Rohrer illustrates this point by drawing a diagram of how metaphorical inferences pro­ duce parallel knowledge structures. He concludes that the conceptual blending proposed by Gore and criticized by his adversaries is constitutive of contempo­ rary U.S. information policy and is transforming the future of how we see social policy. Linda Thornburg and Klaus Panther ("Speech Act Metonymies") propose that conventionalized indirect speech acts can be analyzed as pragmatic me­ tonymies. They use the notions of "speech act scenario" and "action scenario" to show how certain attributes of a speech act are used to represent a whole see-

INTRODUCTION

xiii

nario or particular components. The analysis concentrates on directive speech acts and includes hedged performatives and sentences containing perception verbs and mental activity verbs. On the basis of this analyses, the authors de­ velop a tentative taxonomy of metonymie relationships such as part-whole, cause-effect, ability-action, reason-action, and so forth, which are systematically exploited in the performance of indirect speech acts.

Correlates of Discourse Structure In their paper "Focus Movements and the Internal Images of Spoken Dis­ course", Kenneth Holmqvist and Jana Holsánová present the results of two proj­ ects on focus movements. Starting from Chafe's observations about the parallels between visual perception, memory, and verbalization processes, the authors investigate how speakers illustrate and support ongoing discourse with drawings. They claim that focus movements will leave traces in the drawings, which can tell us how the focus of attention moves across the underlying internal images of speaker and listener. The correlation between spoken discourse and focus movements is illustrated by analyzing the transcript of a conversation in which a speaker used an abstract drawing in support of a descriptive explanation. Holm­ qvist and Holsánová conclude that traces of the continuous movement of a speaker's attention across a reconstructable internal image appear in all of lan­ guage, even if the speaker denies having conscious access to those images. The last paper in this collection is "Pauses, Cognitive Rhythms and Dis­ course Structure: an Empirical Study of Discourse Production" by Joost Schilperoord and Ted Sanders. Using a corpus of dictated judicial letters, they show that pause times in dictation directly reflect major and minor transitions in dis­ course structure, with longer pauses occurring at major breaks. Discourse structure is described with the PISA-procedure, a hierarchical analysis that takes into account the genre-specific 'superstructure' of a text, its global thematic structure and local coherence relations. This study extends earlier findings with spontaneous spoken discourse in two important ways: It shows that cognitive rhythms can also be found in partially preplanned speech; and it adds to the growing body of evidence supporting the validity of hierarchical discourse analysis as a tool for studying cognitive processes in discourse production.

I. COGNITION, PERSPECTIVE, AND MODALITY IN DISCOURSE

Form-Use Mappings for Tag Questions* Maria Josep Cuenca University of Valencia, Spain

1. Introduction Little attention has been paid to tag questions in theoretical linguistics, at least after the transformational debate held in the late 60s and the early 70s about their deep structure (cf. Nässlin 1984: 11-38; see also Section 3.1.)· This situa­ tion probably has to do with the fact that, because tag questions appear mainly in oral exchanges, they seem to show an "anomalous structure" (as compared to the general patterns of a language). To complicate matters even further, tag questions adopt very different forms from one language to another and even within a language. The consequence of these characteristics is that tag questions have often been relegated to the huge "circular file" of peripheral grammar. Tag questions, and more precisely reversed polarity tag questions1, can be looked upon as a specific kind of linguistic construction that display a pragmatic function, typically associated with conversation. Their function is that of re­ questing a confirmation about what has been previously said. Take, for exam­ ple, classical tag questions such as (1). (1) You will help me, won't you? There are two different parts: the assertive one (You will help me) and the interrogative one, the tag (won't you?), which formulates an interrogation about the modality of the former and requests for a confirmation on the part of the hearer. Because of the specificity of this kind of structure, it can be considered a construction in the sense described by Fillmore, Kay & O'Connor (1988: 501): Constructions in our view are much like the nuclear family (mother plus daugh­ ter) subtrees admitted by phrase structure rules, except that (1) constructions need not be limited to a mother and her daughters, but may span wider ranges of the sentential tree; (2) constructions may specify, not only syntactic, but also lexical, semantic and pragmatic information; (3) lexical items, being mentionable in syntactic constructions, may be viewed, in many cases at least, as construc­ tions themselves; and (4) constructions may be idiomatic in the sense that a large construction may specify a semantics (and/or pragmatics) that is distinct from what might be calculated from the associated semantics of the set of smaller con­ structions that could be used to build the same morphosyntactic object.

4

MARIA JOSEP CUENCA

The concept of grammatical construction is based on the hypothesis of form-use interdependence, which is one of the basic postulates of Cognitive Linguistics (Lakoff 1987: 463, 465). As a matter of fact, it is generally possible to discover some kind of correlation between the form and meaning of a certain construction at some moment of its diachronic evolution, even though from a synchronic point of view form and meaning could be understood as being arbitrarily re­ lated. In any event, tag questions can be seen as a holistic grammatical construc­ tion in which form and meaning are closely related and the meaning of the whole cannot be derived by simply adding up the meaning of its parts. This will be my central assumption. Henceforth, an attempt will be made to demonstrate that the form and meaning of tag questions are not independent of one another. After establishing the characteristics of tag questions (section 2), I will suggest a cross-linguistic typology of form-use mappings for tag questions (section 3), using the approach of grammaticalization as a theoretical base. I will analyze the main forms of tag questions on the basis of some data I have col­ lected from some Western languages, namely, English, German, Catalan, Spanish, Italian and French (section 4). Specifically, this contrastive study will deal with several forms associated with a single communicative use, that of re­ questing a confirmation.

2. Theoretical Framework: Grammaticalization Theory Grammatical ization theory has developed considerably in recent years. I will focus on two of the main lines of research on this subject: the discourse per­ spective and the subjectification hypothesis. I will now summarize their main assumptions in order to explain my data. The discourse perspective describes grammaticalization as the fixing of discourse strategies. Givón (1979: 209) as­ sumes that "today's morphology is yesterday's syntax," which can be extended in the following way: "Today's syntax is yesterday's pragmatic discourse" (Heine, Claudi & Hunnemeyer 1991: 12). Both Givón (1979) and Hopper (1987) claim that grammaticalization is a process that tends to grammatically encode relations that were previously not encoded or were encoded in a different way. Hopper has insisted that grammar, as an emergent process, tends to fix structurally what originally could be understood as a communicative strategy. Because grammar is always emergent but never present, it could be said that it never exists as such, but is always coming into being. There is, in other words, no 'grammar' but only 'grammaticization' - movements toward structure which are often char acter izab le in typical ways (Hopper 1987: 147).

FORM-USE MAPPINGS FOR TAG QUESTIONS

5

The structural fixation of discourse mechanisms is achieved by means of routinization, semantic loss and frequency (Hopper 1991). On the other hand, the subjectification hypothesis focuses on the speaker as the origin of linguistic change. The theory has been largely propounded by Traugott (1980, 1989, 1990; Traugott & König 1991). In general terms, it can be defined as a common type of semantic change that is often involved in the process of grammaticalization. In the words of Langacker (1990: 5); "Subjecti­ fication is a recurrent and highly important type of semantic extension and is often a central factor in the evolution leading from 'lexical' to 'grammatical' elements." The evolution from objective to subjective construal refers to the relation­ ship between the perceiving individual and the entity perceived, which, concep­ tually, are relatively separated in objective constructions and fairly interrelated in subjective ones. Hence, subjectification implies some degree of integration of the perceiver in the description of an object or a process. It leads, therefore, to a greater pragmaticization of meaning. Traugott's main claim is that one of the essential pathways for linguistic changes goes from propositional meanings to textual or even expressive ones. In other words, there is a tendency to change "from meanings grounded in more or less objectively identifiable extralinguistic situations to meanings grounded in the speaker's attitude to or belief about what is said (...)" (Traugott & König 1991: 198). In addition, Traugott (1989: 34-35) has formulated three general tendencies in grammaticalization: (i) from the outside world (external situation) to the in­ side world (internal situation); (ii) from the described situation to the textual one, and (iii) from objectivity to subjectivity. These three points, and to some extent also the discourse perspective, can be integrated in the subjectification hypothesis as a whole, considering that rather than semantic attenuation or bleaching there is an increase of pragmatic meaning, mainly in the early stages of grammaticalization (cf. Sweetser 1988 & Traugott 1988). Speakers encode their attitude in a new grammatical form, so that the result is more relevant, more informative, with regard to the speaker's attitudes and beliefs and also to conversational strategies. The subjectification perspective has been applied especially to connectives (see Traugott & König 1991). Here I will apply the hypothesis to tag questions, following the research line defined by Hopper (1987: 148): The major descriptive project of Emergent Grammar is to identify recurrent strategies for building discourses—strategies which have intra-linguistic or interlinguistic generality (or both) and which move toward grammaticalization along parallel lines.

6

MARIA JOSEP CUENCA

3. The Characteristics of Tag Questions As a first step in the discussion, it is necessary to establish the characteristics of tag questions, considering both their form (intonational and syntactic aspects) and their meaning (pragmatic use). 3.1. Intonational and syntactic characteristics Intonationally and syntactically, the construction as a whole is formed by two different elements: the first one is assertive and typically sentential; the second one is interrogative and exhibits different categorial structures depending on the language and the construction considered. In English, the latter can correspond either to the canonical tag (2a), which changes according to the subject and the verbal tense of the assertive part, or to the invariant tag (2b). (2) a. Mary is coming along, isn't she? b. Mary is coming along, huh? The syntactic status of the tag is controversial. In the framework of Chom­ sky's Standard Model, a deep discussion arose between scholars who thought that tags were reduced sentences, and hence the result of a deletion transforma­ tion, and those who were against this postulate, most of which proposed an un­ derlying simple sentence approach (see Huddleston 1970 & Nässlin 1984 for discussion and references). As Norrick (1993; 1995: 689)2 has pointed out, the reduction hypothesis could be accepted when applied to tags like okay, right, y'know, which can be made into full sentences, as in (3): (3) a. Is that okay? b. Isn't that right? c. Do you know that? However, the same is not true for tags such as huh, for no corresponding sentence can be developed (4). (4) a. *Isn't that huh? b. *Does it huh? Norrick (1993: 4) concludes: [....] I do not think we hear right, okay, y'know as reduced versions of complete sentences; I do not think canonical tags work that way either, though Hudson shows such an analysis is possible.

FORM-USE MAPPINGS FOR TAG QUESTIONS

7

I fully agree with Norrick. Nevertheless, in my opinion it is possible to give an alternative which brings together the insights of both the "reduced-sentence hy­ pothesis" and what we can call the "idiosyncratic-structure hypothesis". Invari­ ant tags are not reduced versions of complete sentences but nevertheless they have a sentential nature: intonationally, semantically and pragmatically they behave like sentences.3 So they can be considered peripheral members of the category 'sentence'. They do not have the prototypical sentential structure (roughly subject plus predicate) but they are essentially equivalent in all other respects. Canonical tags, on the contrary, still show sentential structure, though schematically (auxiliary + subject pronoun). My point is that tags are emphatic, focal elements that highlight the most important and contrastive information of a sentence, so that they do not need the presence of any other explicit constituent. They implement the modal value of the sentence, similar to the responses Yes or No. Their focal character also helps to explain why they appear in peripheral positions, mainly in final position, though sometimes initially too, as can be seen in (5), the Catalan translation of (i). 4 (5) a. M'ajudaràs, ventati cl(me)-help-FUT- 2sg., truth? 'You will help me, won't you?' b. Verität que m'ajudaràs! Truth that cl(me)-help-FUT-2sg? 'You will help me, won't you?' The initial position version (5b) displays an integrated sentence structure in which veritat subordinates the sentence. The result of this inversion is a con­ struction similar to WH-interrogation, except for the emergency of the general complementizer que ('that'). Brucart (1993) states that this requisite, also ob­ servable for Spanish in other structures with an emphatic phrase, such as those in (6), is duc to the lack of the WH-feature. (6) a. Juan nos explicó lo difícil que es entender ese artículo. John- cl(us)-expained the-neuter difficult that is to understand this paper. 'John explained to us how difficult it is to understand this paper.' b Es increíble los disparates que dice. Is incredible the nonsense that says-3sg. 'It is incredible what silly things he/she says.'

8

MARIA JOSEP CUENCA

The tags I am considering are also emphatic and behave similarly to the con­ structions in (6) where the basic structure is modified either by means of preposing, as in (5b) above, or by means of reduplication, as in (7): (7) M'ajudaràs, ventat que si? cl(me)-help-FUT- 2sg., truth that yes? 'You will help me, won't you?' These ideas allow us to reconcile the reduced-sentence hypothesis with the idiosyncratic-structure hypothesis. The tag is thought to be sentential in nature but not an actual reduced sentence, and is treated as an emphatic utterance which is syntactically, intonationally, and informatively identifiable as a focus. 3.2. Pragmatic characteristics From a pragmatic point of view, the tag questions I am considering here mani­ fest polarity reversal that can be overt, as in canonical tags, or covert, as in in­ variant tags. They result in a question that tends towards affirmation (8) or to­ wards negation (9), depending on the modality (affirmative or negative) of the sentence preceding to the tag (cf. Leech & Svartvik 1975: 112). (8) - Mary is coming along, isn't she? Expected answer: -Yes (9) - Mary isn't coming along, is she? Expected answer: -No As a matter of fact, it is crucial to point out that tag questions are condu­ cive in that the speaker expects either an affirmative or a negative response to his or her question. After Weber (1993: 24), we can consider that the term 'question' applied to this kind of tags refers to the pragmatic function rather than to the form of the construction. The interrogative character is, nevertheless, preserved, since a counter-expectative answer is also possible (10). (10)

a. - Mary is coming along, isn't she? -No, she isn't. b. - Mary isn't coming along, is she? -Yes, of course she is.

It seems to me that tag questions should therefore be located between asser­ tion and interrogation in the modality "continuum" from YES to NO, as shown in Table (l). 5

FORM-USE MAPPINGS FOR TAG QUESTIONS

9

Table 1. Modality continuum

YES Mary is coming along. Mary is coming along, isn't she? Is Mary coming along? Isn't Mary coming along? Mary is not coming along, is she? NO Mary is not coming along.

affirmation DECLARATIVE affirmative hypothesis affirmative question INTERROGATIVE negative question negative hypothesis negation

DECLARATIVE

R. Lakoff (1972: 917-918) expresses quite the same opinion: ... a tag-question is really intermediate between a statement and a question: a statement assumes that the addressee will agree, and a question leaves the re­ sponse of the addressee up to him, but a tag-question implies that, while the speaker expects a certain sort of response, the hearer may not provide it ... The effect of the tag, then, is to soften the declaration from an expression of cer­ tainty, demanding belief to an expression of likelihood, merely requesting it ... Considering in more detail the pragmatic value of tag questions, Norrick (1993: 1) has pointed out that they convey presuppositions about the speaker's and the hearer's knowledge and attitude. That is, ... invariant tags typically signal a perception of concurrence or difference in knowledge or attitude between the speaker and another participant in the conver­ sation. In particular, tags tend to mark an assumption: (1) that the speaker and hearer share a belief or attitude, (2) that the hearer knows better the speaker does, or (3) that the speaker knows better than the hearer, and hence challenges the hearer's incorrect belief. These pragmatic interpretations are shown in (11) to (13), corresponding to Norrick's examples (1993; 1995: 690-691). (11) (12)

(13)

Bob: Great vegie dip, huh? Brad: Yeah. Awesome. Shelley: So Paul can get it over with. Vera: Then he- then he's got to go look for a job, hunh? Shelley: Yeah. And he ... Pat: And they raped and privileged. Suzy: Privileged? Jack: Privileged? Huh huh huh huh. Suzy: So you went to college, hunh Pat?

10

MARIA JOSEP CUENCA

In (11), the tag marks a presupposition that the speaker and hearer share the same belief. In (12), the speaker manifests doubt about a certain fact and pre­ supposes that the hearer has superior knowledge about it. Conversely, in (13), the speaker possesses better knowledge and challenges the hearer. In other words, whenever the speaker uses this type of tag, she or he for­ mulates a hypothesis and asks for a confirmation, so that some kind of opposi­ tion between the speaker's and hearer's knowledge or attitude about a certain fact shows up. Hence the speaker modalizes the utterance by introducing im­ plicit information denoting her or his attitude about the communicative ex­ change. Hudson (1975: 23) proposed a similar idea: the declarative sentence and the tag define the same proposition, that is, they have the same propositional con­ tent (apart from the polarity reversal), but "they give different information about the speaker's relation to the same proposition." To summarize, it can be said that tag questions exhibit the following char­ acteristics: (a) They are nonprototypical instances of interrogatives. Tag questions syntactically, intonationally, and pragmatically combine assertion (the first part is typically declarative but can also be exclamatory or imperative) and interro­ gation. The final result of this combination is a metalinguistic question about the modality value of the sentence. (b) Tag questions are interactive constructions that must be interpreted in a communicative exchange where the speaker and the hearer are explicit and ac­ tive. (c) Formally, they consist of a sentence (or a sentential element) plus a pragmatic marker (the tag), which expresses the illocutionary force of the whole structure, that of requesting a confirmation or often a kind of directive function. The tag, consequently, functions as an overt manifestation of modalization; in other words, it shows evidence of the speaker's attitude with respect to the pre­ vious utterance. In light of these facts, I would like to suggest that the form and use of tag questions are interdependent.

4. A Cross-Linguistic Typology of Tag Questions The hypotheses formulated in the previous section will now be applied to some examples from English, German, Catalan, Spanish, Italian and French. 4.1. English Tag questions have been mostly studied with reference to English. As I have stated before, in English, there are two kind of tags: inflective or 'canonical'

FORM-USE MAPPINGS FOR TAG QUESTIONS

11

tags, as in (14a, b), and invariant tags, as in (14c-e) (cf. Holmes 1982 & Quirk et al. 1985: 810-814). (14)

a. b. c. d. e.

He had come with you, hadn't he? He hadn't come with you, had he?  had/hadn't come with you, huh? He had/hadn't come with you, eh? He had/hadn't come with you, right?

The first tags (14a, b) manifest explicit polarity reversal, since they are formed by the postposed subject and an auxiliary verb in the affirmative or the negative form, according to the modality of the previous sentence (Quirk et al. 1985: 810). The tag is syntactically and intonationally a question and displays a modality opposite to that of the sentence. Therefore, the tag varies with regard to verbal morphology and polarity. Besides this structure, there are invariable markers for the same pragmatic use, i.e. huh and right, mainly used in the United States and parts of Canada, and also eh used in England, Australia and northern Canada, according to Nor­ rick (1995: 689).6 Some of them, being general interjections, have a range of contextually determined pragmatic values. As Norrick (1995: 688) states, most languages have invariant markers of this kind, while the inflective forms are rarer, although Danish and Finnish, for instance, also use inflected tags. 4.2. German German shows a great variety of forms associated to the pragmatic meaning of confirmation, some of which are shown in (15): (15)

a. Er ist mit dir gekommen, nicht wahr? b. Er ist mit dir gekommen, nicht?  Er ist mit dir gekommen, ja? d. Er ist (nicht) mit dir gekommen, hm? e. Er ist (nicht) mit dir gekommen, oder? f. Er ist (nicht) mit dir gekommen, gell?1

The first tag (15a) corresponds to the negation of the word wahr ('true'). We can also find the negative (nicht) or the affirmative (ja) marker in this posi­ tion (15b,c). These forms, although fixed, can be associated with the inflec­ tional forms of English. Conversely, the other three (15d-f) are invariant and function as interjective markers, which can combine either with affirmative or negative sentences. Hm is similar to Eng. huh, while oder and gell(e)/gelt are grammaticalized forms.

12

MARIA JOSEP CUENCA

Oder is the German disjunctive conjunction ('or'). The evolution from conjunc­ tion to interjection is based on reanalysis and metonymy. Firstly, the conjunc­ tion loses its capacity of connecting two elements and appears without them. Secondly, the meaning of the word is reinterpreted: the new pragmatic value derives metonymically from the notion of choice (disjunction) between affirma­ tive and negative modality. Instead of grammatically fixing one of the modali­ ties as in (15b,c), speakers have fixed the idea of alternation, which is repre­ sented by the conjunction. As for gell (and its variants gelle/gelt), it is a form restricted to the South­ ern varieties of German (especially those spoken in Austria), and has evolved from the verb gelten ('being worth, valuable', mainly referring to a contract or a game). The decategorialization - from verb to interjection - is based on a metaphorical process by which the objective notion of value is reinterpreted as value in discourse by means of pragmatic inferencing. 4.3. Romance languages In Romance languages different tags implement the same pragmatic use that we are dealing with. Catalan uses the word Verität ('truth') and also its phonetic variants vritat, vitat, itat, tat (16a), the negative adverb no (16b), and two inter­ jections: a specific one, oi (16c), whose use is almost restricted to the confirma­ tion value, and a general one, eh (16d). (16)

a. b. c. d.

(No) ha vingut amb tu, Verität? Ha vìngut amb tu, no? (No) ha vingut amb tu, oi? (No) ha vingut amò tu, eh?

The situation in Spanish (17) and Italian (18) is very similar, except for the absence of a specific marker such as Cat. oi. The tags in both languages are based on the word 'truth/true' (17a, 18a), on the negative marker (17b, 18b) which can also combine with 'truth/true' (18c), and sometimes also on a general interjection, such as Sp. eh (17c). (17) (18)

a. (No) ha venido contigo, ¿verdad? b. Ha venido contigo, ¿no?  (No) ha venido contigo, ¿eh? a. (Non) è venuto con te, vero? b. È venuto con te, no? e. È venuto con te, non è vero?

FORM-USE MAPPINGS FOR TAG QUESTIONS

13

As for French, the word vrai ('truth') must be negated (19a), and alternates with the negative sentence n'est-ce pas (19b), similar to English canonical tags, the negative adverb non (19c), and the interjection hein (19d). (19)

a. Il (n') est (pas) venu avec toi, pas vrai? b. // (n') est (pas) venu avec toi, n'est-ce pas? c. Il est venu avec toi, non? d. // (n') est (pas) venu avec toi, hein?

5. Grammaticalization in Tag Questions Considering the data at hand, which is not intended to be exhaustive but illus­ trative, we can establish different sources of tag questions: (a) the canonical tag in English, which shows syntactic variation according to polarity (such as Eng. will you?/won't you?); (b) structures with a modality marker, generally a negative one (Ger. nicht?, Cat./Sp./It. no?, Fr. n'est-ce-pas?, non?) and seldom an af­ firmative one (Ger. ja?); (c) structures containing an element related to the concept of truth (Eng. right?, Ger. nicht wahr?, Cat. ventât?, Sp. ¿verdad?, It. vero?, Fr. pas vrai?); (d) the use of some interjections whose origin can still be derived by the speaker (as Ger. oder?); (e) the use of grammatical ized elements that have reduced their phonetic form, though their origin is still interpretable (e.g., Cat. vitat/tat, which can be easily recognized as reduced forms of ventât); (g) completely grammatical ized elements, whose origin is generally un­ known to the speakers (Ger. gelll and Cat. oí?); (h) the use of proper interjections, that is, meaningless forms from a lexical point of view (Eng. huh?/ehi, Ger. hm?, Cat. ehi, Sp. ¿ehi, Fr. hein?). Some of the tags include lexical items that have turned into grammatical ones by reanalysis: the constructions containing the noun truth or the adjectives true and right, the German conjunction oder, modality markers (mainly adver­ bial), and also Cat. oi, which apparently derives from an ancient affirmative adverb (the same that has evolved into oc 'yes' in Occitan).8 Category shift in these cases implies increase in the pragmatic meaning. The tags considered in­ stantiate the two major motivations for change described by Hopper & Traugott (1993: cap. 4), that is metaphor and metonymy. Metaphor is involved in the change from the propositional notion of truth to the pragmatic use that tags

14

MARIA JOSEP CUENCA

show. The path leading from Ger. conjunction oder to interjection is me­ tonymie, as disjunction represents the choice between affirmation and negation. Finally, the modality markers in a tag are located in an interrogative context, so that their illocutionary force is to a certain extent neutralized and they simply convey polarity reversal. This case can be interpreted as the fixation of a dis­ course strategy, that of asking for the hearer's confirmation, which is a kind of metonymy according to Traugott (1988). On the other hand, the various forms of tag questions follow within their derivational history the general points in the continuum of grammaticalization established by Bybee (1993): Table 2. The continuum of grammaticalization

(0 (ii) (iii) (iv) (v) (vi)

LEXICAL stems polysyllabic large open classes free position relatively infrequent semantically rich

> > > > > >

GRAMMATICAL auxiliaries, particles > affixes monosyllabic > unique segment large closed classes > small closed classes > rigidly fixed relatively fixed > obligatory rather frequent more general > reduced or empty

Most of these characteristics are also present in tags, as I will show now. Firstly, a number of them include a content item that has been reanalyzed as a particle, or even as an interjective element (i). Secondly, they tend to be simplified phonological forms (ii). Some of them, like Cat. veritat, show different points of phonetic weakening (vritat, vitat, itat, tat); another example is that canonical tags in English are generally used with the verb plus a contracted not, which implies a certain degree of phonetic re­ duction.9 However, it is important to bear in mind that they will never become affixes because they occur in a focus position, which prevents them from a complete loss of tonicity. Moreover, generalisation of meaning -i.e., (vi)-, or more specifically pragmatic strengthening, as proposed by Sweetser (1988) and Traugott (1988), is also associated with changes in morphology, viz, a tendency towards invari­ ability, in lexical nature: i.e., reduced paradigmatic variability (iii), and in syn­ tactic distribution (iv): they are located in the final position and only in certain languages can some of them be sentence initial, as I have illustrated above (see also (20)). (20)

a. Veritat que vindràs amb mi? b. Oi que vindràs amb mi?

FORM-USE MAPPINGS FOR TAG QUESTIONS

15

Finally, the items of a tag become obligatory (v) to turn an assertion (or even­ tually an exclamative or an imperative) into a question. Tags illustrate grammaticalization in more respects. They follow the grammaticalization principles described by Hopper (1991): layering, divergence, specialization, persistence and decategorialization. In fact, the broadening of the functional domain produces the emergence of new layers (e.g., the phonetic variants of veritat) and the divergence of forms: the new item is understood as an extension of the lexical one, and they both co-exist in a language. However, because polysemy does not cancel all the semantic characteristics of the original item, there is some kind of persistence. Finally, the process of grammaticaliza­ tion involves decategorialization from noun, adjective, adverb or conjunction to interjection.

6. Form-use Mappings for Tag-Questions At this point a direct connection can be established between the tag forms con­ sidered and the defining properties of tag questions. For concreteness, there are four general mechanisms in the languages analyzed: (a) Tags indicating polarity reversal (such as canonical tags in English or negative and affirmative markers). (b) Tags including the word truth and the like. (c) Tags formed by an interjection focusing on different kind of subjective meanings. (d) Specific markers, etymologically related to one of the mechanisms in (a) - such as Cat. oi, whose origin is likely to be an affirmative word or (b) - e.g., Ger. gell/gelt, that comes from the verb gelten t cost' 'to have a value', metaphorically reinterpreted as 'truth value'. The various forms of this construction can easily be related to at least one of the defining properties of tag questions: (a) Tag questions display a discursive mechanism opposing both assertion and interrogation, on the one hand, and affirmation and negation, on the other. First, the construction containing these elements includes a statement - or rather a non-interrogative part, which is usually declara­ tive but sometimes is exclamatory or imperative -, corresponding to the propositional content, and an interrogative part, corresponding to the tag. Second, they are nonprototypical idiomatic constructions lo­ cated in the modality continuum between affirmation and negation. Therefore, they are defined by their mixed modality.

16

MARIA JOSEP CUENCA

(b) Tag questions are related to "truth value" (not in the logical sense but rather from a communicative perspective), since they evidence the speaker's and the hearer's subjective belief state about the facts they are commenting on. (c) They are evidential in that they help to manifest the opinions and dis­ cursive attitudes of the speaker. Their meaning is more pragmatic than lexical; therefore, they can be described only in terms of subjective values.

7. Conclusion In this paper I have shown that Cognitive Linguistics, and especially grammati­ cal ization theory, constitutes an optimal framework to account for the character­ istics of tag questions, since it allows for a pragmatic and syntactic approach which is necessary to account for the formation of tag questions. Specifically, it is possible to uncover the form-use mappings of this construction, as the forms that display the pragmatic function of confirmation are linked to at least one of the characteristics that define this kind of structure. Although the forms chosen to become confirmation tags are to a certain extent idiomatic, there is a common basis that accounts for the fact that different Western languages use similar markers. Linguistic form is then not independent of meaning and use. Moreover, there are several theoretical aspects of tag questions that can be associated with grammaticalization. On the one hand, they exemplify the direc­ tionality in grammaticalization proposed by Traugott, in that the process from the source structures to tags can be seen as a movement towards the subjectification of meaning. On the other hand, tag questions can be treated as the con­ ventionalization of a discourse strategy that involves the speaker and the hearer, as well as their beliefs. In other words, there is a grammaticalization of a con­ versational meaning by means of pragmatic inferencing and subjectification. Finally, the different kinds of tags illustrate various stages of fixation and allow us to understand how the grammaticalization process takes place. Endnotes * This paper has been partially supported by the Research Grant Ps 91-0134 (DGICYT). I am indebted to Eve Sweetser for her comments on this paper and to Joseph Hilferty for his sug­ gestions and the linguistic correction. 1. It should be noted that I will restrict my study to reversal tags, although for the sake of simplicity, I will use the denomination tag question henceforth. 2. Norrick (1995) is a version of Norrick (1993). Since both have some different information along with some common information, I will refer to both papers separately.

FORM-USE MAPPINGS FOR TAG QUESTIONS

17

3. This is roughly the opinion expressed by Huddleston (1970: 215). 4. In this paper, I will not deal with the semantic-pragmatic differences associated with the change in position or in marker. 5. Pragmatic complexity of interrogation and negation cannot be captured in a simple scheme such as Figure 1. Nonetheless, I believe that it is a useful picture of the general location of tag questions with respect to assertion and interrogation. Givón (1984: 250-251) proposes a similar cline from declarative to Yes-No questions, though he does not differentiate the negative and the affirmative ones. This is the continuum space between these two major pro­ totypical speech-acts: (i) FROM DECLARATIVE TO YES-NO QUESTIONS a. Joe is at home. [most prototypical DECLARATIVE] b. Joe is at home, I think. c. Joe is at home, right? d. Joe is at home, isn't he? e. Is Joe at home? [most prototypical INTERROGATIVE]

6. 7. 8.

9.

Givón claims that there is empirical evidence of this continuum, which has been tested by Tsuchihashi (1982) in a work about Japanese speech-act verbal suffixes. Quirk et al (1985: 814) consider right "informar' and etti "casual and maybe impolite." The meaning of the examples, irrespective to the language, will always be the same, 'You will come with me, won'tyou?' 'You won't come with me, willyou?\ unless otherwise indi­ cated. Therefore, I will not include a translation. This is the hypothesis maintained by Alcover-Moll (1926-1968). The etymon proposed is the Latin neuter pronoun hoc. However, Coromines (1980) does not agree and says that oi is simply a combination of vowels as in other interjections, such as ai, au, ei, etc. In my opin­ ion, the first theory is much more convincing. Quirk et al. (1985: 129) in fact explain the forms aren't Hain't /, which deviate from the general pattern of tag formation, as due to the impossibility of creating a contraction of Am I not? This fact indicates the importance of phonetic reduction in this construction.

References Abraham, Werner. 1991. "The grammaticization of the German modal particles". In Elisabeth C. Traugott and Bernd Heine (eds), Approaches to Grammaticalization. Amsterdam: Benjamins, vol. 2, 331-380. Alcover, Antoni M. and Francese de . Moll. 1926-1968. Diccionari Català-ValenciàBalear. Palma de Mallorca: Moll. 1983. Brucart, José M. 1993. "Sobre la estructura de SComp en español". In Amadeu Viana, (ed.) Sintaxi. Teoria i Perspectives. Lleida: Pages eds, 59-102. Bybee, Joan. 1993. "Mechanisms of semantic change in grammaticization", Third In­ ternational Cognitive Linguistic Conference, Leuven, Belgium, 18-23, July, 1993. Coromines, Joan. 1980. Diccionari etimologie i complementari de la llengua catalana. Barcelona: Curial. Cuenca, M. Josep and Josep M. Castella. 1995. "Una caracterització cognitiva de les preguntes confìrmatòries (question tags)", Caplletra 18: 65-84.

18

MARIA JOSEP CUENCA

Givón, Talmy. 1979. On Understanding Grammar. New York: Academic Press. Givón, Talmy. 1984. Syntax: a Functional-typological Introduction, vol. 1. Amster­ dam: Benjamins. Heine, Bernd, Ulrike Claudi and Friederike Hünnemeyer. 1991. Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press. Holmes, Janet. 1982. "The functions of question tags", English Language Research Journal 3: 40-65. Hopper, Paul. 1987. "Emergent grammar", Berkeley Linguistic Society, Papers of the Thirteenth Annual Meeting, 139-157. Hopper, Paul. 1991. "On some principles of grammaticization". In Elisabeth  Traugott and Bernd Heine (eds), Approaches to Grammaticalization. Vol. 1. Amsterdam: Benjamins, 17-35. Hopper, Paul and Elisabeth  Traugott. 1993. Grammaticalization. Cambridge: Cam­ bridge University Press. Hudson, Richard A. 1975. "The meaning of questions", Language 51: 1-31. Langacker, Ronald W. 1990. "Subjectification", Cognitive Linguistics 1: 5-38. (Re­ printed in Langacker 1991): Concept, Image and Symbol. The Cognitive Basis of Grammar, Berlin: Mouton de Gruyter. Lakoff, George. 1987. Women, Fire and Dangerous Things. What Categories Reveal about the Mind. Chicago: The University of Chicago Press. Lakoff, Robin. 1972. "Language in context", Language 48.4: 907-927. Leech, Geoffrey and Jan Svartvik. 1975. A Communicative Grammar of English. Lon­ don: Longman. Huddleston, Rodney. 1970. "Two approaches to the analysis of tags", Journal of Lin­ guistics 6: 215-222. Nässlin, Siv. 1984. The English Tag Question: A Study of Sentences Containing Tags of the Type Isn't it? Is it?. Stockholm: Almqvist & Wiksell International. Norrick, Neal R. 1993. "Invariant tags and evidentiality in conversation", Third Inter­ national Cognitive Linguistic Conference, Leuven, Belgium, 18-23, July, 1993. Norrick, Neal R. 1995. "Huhn-tags and evidentiality in conversation", Journal of Pragmatics 23: 687-692. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A Com­ prehensive Grammar of English Usage. New York: Longman. Sweetser, Eve. 1988. "Grammaticalization and semantic bleaching", Proceedings of the Annual Meeting of the Berkeley Linguistic Society 14: 389-404. Traugott, Elisabeth  1980. "Meaning-change in the development of grammatical markers", Language Science 2: 44-61. Traugott, Elisabeth  1988. "Pragmatic strengthening and grammaticalization", Pro­ ceedings of the Annual Meeting of the Berkeley Linguistic Society 14: 406-404. Traugott, Elisabeth  1989. "On the rise of epistemic meanings in English: an exam­ ple of subjectification in semantic change", Language 65: 31-55. Traugott, Elisabeth  1990. "From less to more situated in language: the unidirectionality of semantic change". In Sylvia Adamson et al. (eds), Papers from the Fifth International Conference on English Historical Linguistics. Amsterdam: Benjamins, 496-517.

FORM-USE MAPPINGS FOR TAG QUESTIONS

19

Traugott, Elisabeth  and Bernd Heine (eds). 1991. Approaches to Grammaticalization. Amsterdam: Benjamins. Traugott, Elisabeth  and Ekkehard König. 1991. "The semantic-pragmatics of grammaticalization revisited". In Elisabeth Traugott and Bernd Heine (eds), Ap­ proaches to Grammaticalization. Vol. 1. Amsterdam: Benjamins, 189-218. Tsuchihashi, M. 1983. The speech-act continuum: an investigation of Japanese sen­ tence-final particles, Journal of Pragmatics: 7, 4. Weber, Elisabeth G. 1993. Varieties of Questions in English Conversations. Amster­ dam: Benjamins.

The Social Dimension of a Cognitive Grammar Bruce W. Hawkins Illinois State University, U.S.A. 0. Introduction The purpose of this paper is to demonstrate that a cognitive grammar can and should attend to the sociopolitical aspects of language use. To that end, the pa­ per examines the phenomenon of iconographie textual reference. A textual ref­ erence can be anything from a simple nominal, like jerk, for example, to an extended text like Joe McGinniss's recent biography of Teddy Kennedy, The Last Brother: the rise and fall of Teddy Kennedy. An iconographie reference is one that constructs an image of the referent which is intended to evoke a strong emotional response to that referent. In this paper, I will focus on two specific types of iconographie reference: icons and caricatures. An icon is an image to be revered, while a caricature is an image to be reviled. The contrast between icon and caricature is evident in the examples below. (1) President Richard Nixon will be remembered for reasserting U.S. lead­ ership in world affairs at a crucial period in the Cold War. He was a statesman with an ability to view global affairs in broad strategic terms and, once having done that, locate U.S. interests precisely and pursue them tenaciously no matter how difficult the terrain got... In international terms, Watergate would, given the space of time and reflection, turn out to be only a footnote in history. Richard Nixon, by contrast, will be remembered as one of the great statesmen of this century. (2) As the 37th president, (Richard) Nixon's achievements in reopening relations with China and forging historic arms agreements with the Soviet Union helped end the Cold War and transformed geopolitics. But his abuses of power in the White House and the Watergate scandal breached the Constitution, undermined public faith in govern­ ment and ensured that he will be remembered as the only president forced to resign the nation's highest trust. These texts were juxtaposed in a column titled "What others are saying" on the editorial page of my local newspaper just a few days after Nixon's death in April 1994. The text in (1) was attributed to The Strait Times of Singapore while that in (2) was attributed to the San Francisco Chronicle. In (1), The

22

BRUCE W. HAWKINS

Strait Times projects a positive image of Richard Nixon as one of the great statesmen of this century. In (2), on the other hand, the San Francisco Chroni­ cle reveals its negative assessment of Richard Nixon by reducing him to the only President forced to resign the nation's highest trust. Data like these exhibit the impact of ideology on the semantics of textual reference. The role of ideology in our textual experiences is an issue of consid­ erable interest to scholars in literary critical theory, rhetoric and pedagogy. Many of these scholars lament the perceived fact that current linguistic theories grounded in cognitive science do not provide a site for considerations of ideol­ ogy. This has led to a situation in which some of these scholars with interests in the power of linguistic texts ignore the insights of contemporary theoretical lin­ guistics and appeal, instead, to selected insights from the icon of an earlier gen­ eration of theoretical linguistics: Saussure. This appeal, in and of itself, is not problematic; contemporary scholars in linguistics can still profit from a disciplined, informed, critical reading of Saus­ sure. However, as Jackson (1991) points out clearly, problems do arise in the critical theorists' appeal to Saussure. The problems Jackson identifies stem from the critical methodology known as rereading (or misreading), through which principles of Saussurean structuralism have been appropriated from the context of the socially-sensitive structuralist linguistic theory which gives them sub­ stance and re-interpreted for the specific purpose of constructing a critical theory that can provide sophisticated insights into the manipulation of language as an instrument of socio-political power. Jackson (1991: 3) succinctly summarizes his response to this critical methodology in noting that "whenever I read Derrida's book Of Grammatology, I hear the ghost of Saussure howling 'Liar'!" One fundamental goal of the research program from which this paper emerges is to demonstrate that Langacker's Cognitive Grammar (hereafter CG), with its foundation in what Lakoff (1987) calls 'experiential realism' and its explicit attention to the pragmatic grounding of linguistic structures (cf. Lan­ gacker 1991), defines the domain of inquiry for a linguistic theory in such a way that the social dimension of human language need not be left out, and indeed, that it can ultimately be examined with as much rigor as the cognitive dimen­ sion. The paper is presented in five sections. Section 1 reviews briefly the role of such language dynamics as acquisition and processing in defining the domain of inquiry for cognitive theories of language. Section 2 considers the dynamic of interpellation, which is the grounding dynamic that helps define the critical theorists' social-constructionist perspective on language. Section 3 provides a summary interpretation of CG structured strategically to demonstrate that the cognitive concerns highlighted in the dynamics of acquisition and processing are easily reconciled with the social concerns highlighted in the dynamic of interpel­ lation. Section 4 demonstrates the utility of this integrative socio-cognitive per-

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

23

spective in the analysis of a number of iconographie textual references. Section 5 draws conclusions that emerge from the discussion in the preceding sections.

1. The Grounding Dynamics of Cognitive Theories of Language Most contemporary linguists would be quite categorical in identifying linguistics as a cognitive science. This is, at least in part, a direct result of the fact that the dynamic considerations of acquisition and processing have had profound effects on how the domain of inquiry for linguistic theory is defined. I have no problem with this. Language is undeniably a cognitive phenomenon and linguistics should, therefore, be considered a cognitive science. My point is that it is inap­ propriate to conclude that language is only a cognitive phenomenon and, there­ fore, linguistics is only a cognitive science. The critical theorists' attraction to Saussurean structuralism is driven, at least in part, by the perception that Chomskyan generativism is irreconcilable with the assertion that language is a social phenomenon. Chomsky's (1965) ar­ gument for the innateness of certain cognitive universals which profoundly im­ pact on language acquisition has been misinterpreted by many of his critics out­ side the cognitive sciences as a claim that human social interactions have no significant impact on language acquisition. This is clearly a problematic mis­ reading of Chomsky, who has been both consistent and explicit in acknowledg­ ing that language acquisition cannot take place in the absence of human social interaction. While Chomsky cannot be justifiably criticized for denying the le­ gitimacy of the social dimension of language, his approach to defining the do­ main of inquiry for linguistic theory by grounding it in considerations of acqui­ sition of a cognitive system can rightfully be ascribed responsibility for leaving the social dimension of language out of the picture drawn by most modern lin­ guistic theories. Acquisition is not the only language dynamic that will focus the theoretical linguist's attention on the cognitive dimension of language. Another is process­ ing. If a theory of linguistic structure is profoundly affected by considerations of language acquisition, so too will considerations of language processing direct (and, crucially, constrain) developments in a linguistic theory. Not only must the structures of language be acquirable, but they must also be such that they can be manipulated successfully under the severe temporal limitations of normal language performance. One of the characteristics that distinguishes cognitive linguistics (at least in the Langacker and Lakoff mold that I work within) from other approaches to linguistics as a cognitive science is the refusal to endorse categorical distinctions which are either artifacts of a particular mode of reasoning or artificially im­ posed on the data and/or the domain of inquiry for the precise purpose of facili­ tating the process of constructing the linguistic theory (cf. Lakoff 1987, Lan-

24

BRUCE W.HAWKINS

gacker 1987). This is precisely what happens when a linguistic theory is grounded in dynamics like acquisition and processing which bring out the cog­ nitive dimension of language without also being grounded in whatever dynamics would bring out the social dimension of language. As I interpret CG, it is a linguistic theory that is ultimately grounded in both cognitive and social dynamics. The cognitive dynamic that most perva­ sively influences the developments in CG appears to be processing, although considerations of language acquisition (or, more generally, cognitive develop­ ment) clearly do influence certain aspects of the theory. Attention to the social dimension of language within CG is invited, I believe, by the assertion that CG is a "usage-based model" (Langacker 1988). Usage brings into the picture social variables along with those cognitive variables brought out by the attention to processing. In this paper, I am suggesting that any modern cognitive linguistic theory such as CG can benefit significantly by being grounded in the dynamic of interpellation at the same time that it is also grounded in cognitive dynamics such as acquisition and processing.

2. Interpellation Mowitt (1988) characterizes interpellation as "the process through which the human being is constituted as a subject through its relations to the ideological practices of society." This characterization is basically a rewording of Althusser's thesis: "Ideology interpellates individuals as subjects" (cf. Althusser 1971). In effect, interpellation is the process any individual goes through in any socio-cultural environment to adapt behaviors and responses so that they will be in harmony with the behaviors and responses typical of and expected from the average member of that socio-cultural environment. To make the characterization somewhat more explicit, we can appeal to some of the basic descriptive constructs of CG. Interpellation is a processual relation in the experiential domain of human socio-cultural-political interactions. Evident in the wording of Althusser's thesis is the fact that the trajector in this socio-political relation is ideology, however we ultimately conceptualize it. The landmark is the individual human being. As is characteristic of the processual relations Langacker calls "perfective", the relationship between ideology as tra­ jector and the individual as landmark is not constant. This relationship changes crucially over time. This change is important in understanding the nature of the process of interpellation. Interestingly, the entity changed (at least most overtly and directly) is the individual. This is reflected explicitly in the assertion that the individual, through interpellation, becomes a subject. The relevant sense of subject here is that of a socio-political subordinate in relation to a ruling power of some sort. A subject, in this sense, is so-called because s/he is subjected to a life which must follow modes of reasoning and decision-making which are en-

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

25

dorsed and supported by some entity viewed as dominant. The change that oc­ curs in the process of interpellation is that in which the individual becomes sub­ ject to a particular way of categorizing experiences which is viewed simply as common sense within that particular community. The typical cognitivist take on this relation between ideology and the indi­ vidual would probably be to claim that the individual interpellated has simply acquired the ideology of the collective or ruling entity. Social constructionists would find this reinterpretation problematic in that it appears to ascribe a sig­ nificant measure of agency to the individual in his/her relation to ideology. Where the cognitivist will see a mind acquiring a particular conceptual system, the social constructionist sees a powerful social organization exerting its power over an individual in the process of gaining a new adherent to the "common sense" ways of that social organization. Both perspectives on the relation are legitimate. Despite the figure/ground discrepancy that seems to distinguish the per­ spective of the cognitivist from that of the social constructionist, cognitive lin­ guists should nonetheless be able to recognize that interpellation, as viewed by the social constructionist, is undeniably involved in the socio-cognitive process of acquiring the grammar of a particular language. We may begin this process with the innate linguistic competence Chomsky refers to as 'universal grammar', but we emerge from this process as native speakers of a particular language. In the process, we acquire the categorizations and modes of expression of our na­ tive language. We learn to conceptualize time, space, and other experiential domains in the way that has become conventional within the community in which we are acquiring language. We may not become permanent prisoners of these modes of conceptualizing our experience, as the Sapir-Whorf Hypothesis suggests we do, but we remain comfortable operating within this set of conven­ tions as long as we invest ourselves in the particular speech community. It is this perspective on the relationship between language and ideology that informs the central thesis of Hodge & Kress (1993).

3. An Overview of Cognitive Grammar In my treatment of textual reference, I take the verb refer quite literally to mean the process of directing attention or turning to some fundamental resource for information. As a result, I take a dynamic processing perspective in interpreting and applying CG to the analysis of particular cases of textual reference. From this perspective, CG comprises three interactive processes: experience, selec­ tion, and grounding (or deixis). The process of experience provides the basic substantive resources from which any meaningful text is derived. As no text can ever include all relevant information pertaining to any experience, a textual ref­ erence must be selective in the information it presents. This process of selection

26

BRUCE W.HAWKINS

appeals to sets of conceptual tools such as those discussed in Langacker (1987). In most cases, and especially in cases of iconographie reference, it is clear that the information selected for presentation in a particular textual reference is cho­ sen for a strategic rhetorical purpose. Selection and grounding, then, are seen as interactive processes; the particular conceptual tools accessed depend on the nature of the information needed to construct the image that the speaker/writer intends to convey. Experience is the process which provides the fundamental resource we turn to individually and collectively in particular acts of reference. As characterized by Lakoff 1987, experience is an embodied process. That is, the nature of ex­ perience is profoundly affected by the bodily nature of the experiencer. How­ ever, experience is not strictly an individual process. Indeed, it is a process that is, at once, both individual and collective. These two aspects can be reconciled by conceiving of experience as a process with a center-periphery structure that results from the fundamental egocentrism of the human experiencer. At the center of experience is embodied ontogeny. Removed from this center but still clearly relevant is situated (i.e., egocentric, ethnocentric, geocentric, ...) phylogeny. At an extreme so distant from the center of experience that it may exist only as an abstract ideal is a non-situated phylogeny which would produce abso­ lute truth. In this respect, experience can be considered co-extensive with phy­ logeny but focussed most directly on ontogeny. Experience comprises a complex set of dynamic symbiotic relationships in­ volving at least individual sentient and creative beings, collectives of such be­ ings, and a surrounding ecosystem. Absolute knowledge of any of these entities is impossible. All understandings of these entities are necessarily artifactual and contingent in that they are by-products of the symbiosis that characterizes expe­ rience. Selection is the cognitive process of accessing particular conceptual tools to construct textual images which re-present experiences in a digitalized form (cf. Dretske 1983, Kittay 1987) recognizable to interlocutors. One useful way to introduce the cognitive process of selection is through metaphor, which is, it­ self, among the conceptual tools of selection. In a very real sense, we become artists whenever we confront the problem of communicating meaningful mes­ sages to another person. In the same way that a sculptor begins with raw mate­ rial in the form of a block of stone or a lump of clay, we approach any expres­ sive task with the raw material of experience. The sculptor uses some set of tools to cut away unwanted portions of stone or clay and eventually to reshape it into a particular image. Similarly, we make strategic use of sets of conceptual tools to form particular textual images. Where sculpting is the artistic process of working a block of stone or mass of clay into some chosen visual form using specially-designed sculpting instruments, selection is the cognitive process of working experience into carefully structured (digitalized) mental images using certain conceptual tools. Langacker (1987) provides an extremely detailed ac-

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

27

count of certain sets of conceptual tools which play a crucial role in the process of selection. I take grounding and deixis to be essentially synonymous. Grounding/deixis is the process of relating, on the one hand, a symbiosis of experience which includes a particular speech act (or, more generally, a particular act of languagebased communication in either the spoken or written mode) and, on the other hand, the particular digitalized image of an experience that that communicative act selectively calls attention to. Space, time, and discourse are significant in understanding grounding/deixis precisely because these are prominent domains in the symbiosis of experience that subsumes the particular communicative act. The participants in any given communicative act exist in a particular place, at a particular time, and are involved in a particular discourse. Traditional accounts of deixis recognize that these domains of the interlocutors' on-going experience play a significant role in determining the meaning that the interlocutors derive from the discourse that occurs in that experiential context. The only point on which I would differ with traditional accounts of deixis is with any suggestion that space, time and discourse are the only domains of the interlocutors' on­ going experience that will have this impact on the meaning-making process in which these interlocutors are engaged. I suggest that other domains of experi­ ence can and do, at least in certain cases, impact on the meaning-making process in essentially the same way that space, time and discourse are recognized as do­ ing. In the next section, I present and discuss evidence which suggests quite strongly that ideology is one such grounding experiential domain.

4. Toward a Socio-Cognitive Theory of Reference Let us return to the iconographie references to Richard Nixon in (1) and (2). (1) President Richard Nixon will be remembered for reasserting U.S. lead­ ership in world affairs at a crucial period in the Cold War. He was a statesman with an ability to view global affairs in broad strategic terms and, once having done that, locate U.S. interests precisely and pursue them tenaciously no matter how difficult the terrain got... In international terms, Watergate would, given the space of time and reflection, turn out to be only a footnote in history. Richard Nixon, by contrast, will be remembered as one of the great statesmen of this century. (2) As the 37th president, (Richard) Nixon's achievements in reopening relations with China and forging historic arms agreements with the Soviet Union helped end the Cold War and transformed geopolitics.

28

BRUCE W.HAWKINS

But his abuses of power in the White House and the Watergate scandal breached the Constitution, undermined public faith in govern­ ment and ensured that he will be remembered as the only president forced to resign the nation's highest trust. These two texts project diametrically opposed images of Nixon, but they use essentially the same conceptual tools. Both texts are constructed by identify­ ing a significant category to which Richard Nixon belongs and then highlighting factors which distinguish him from other members of the category identified. The text in (1) categorizes Richard Nixon primarily as a statesman. It acknowl­ edges globally the negative factors that brought an end to the Nixon Presidency, but highlights the positive factors which distinguish Nixon "as one of the great statesmen of this century". In contrast, the text in (2) focuses attention on the categorization of Richard Nixon as a President of the United States. It mentions briefly a few of the major positive achievements of President Richard Nixon, but highlights the negative factors which distinguish Nixon from other U.S. Presidents. Foremost among these, of course, is the fact that, to date, Nixon remains the only president forced to resign the nation's highest trust. In effect, then, the construction of each of these texts involves the same two basic tools of selection, which we can refer to as categorization and individua­ tion (via distinguishing characteristics). However, these texts differ significantly in the rhetorical purposes to which these cognitive resources are put. The text in (1) strategically selects particular information to create a positive attitude toward Richard Nixon. In contrast, the text in (2) strategically accesses different infor­ mation which reveals a much more negative attitude toward Nixon. Such texts can be used to lead interlocutors to a similar assessment of the late President. Cooperative readers will find themselves accepting the positive or negative evaluation of Nixon proffered by a given text. In contrast, resistant readers will probably counter the selection of information in such a text because of a nega­ tive response to the ideology underlying that selection. Let us turn our attention to another set of textual references to Richard Nixon. (3) a. Richard Nixon's first conscious memory was of falling—falling and then running. He was three years old, and his mother had taken him and his brother out riding in a horse-drawn buggy, and the horse turned a corner too fast on the way home. The boy fell out. A buggy wheel ran over his head and inflicted a deep cut. "I must have been in shock," Nixon recalled later, "but I managed to get up and run after the buggy while my mother tried to make the horse stop." The only aftereffect, Nixon said, was a scar, and that was why he combed his hair straight back instead of parting it on the side.

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

29

In a sense, Nixon spent his whole life falling and running and falling again. A symbol of the politics of anger, he was one of the most hated figures of his time, and he was also the only man in U.S. history ever to be elected twice as Vice President and twice as President. In the White House, he achieved many major goals: the U.S. withdrawal from Vietnam, restored relations with China, the first major arms agreement with the Soviet Union and much more. But he will always be remembered, as he was at his death last week at 81, as the chief perpetrator—and chief victim-of the Watergate scandal, the only President ever to resign in disgrace. b. His first public appearance [after resigning the Presidency] came in 1978, and then the long, slow process of self-rehabilitation. Per­ haps, in his last years, having regained a certain amount of public respect and even some grudging admiration, having acquired four grandchildren and all the comforts of leisurely wealth, Nixon finally found a little peace, finally got over that mysterious anger that had fueled his ambition throughout his long life. Perhaps. Taken independently, each can be recognized as presenting a caricature of Richard Nixon. However, these texts are not, in fact, independent of each other. The text in (3a) comprises the first two paragraphs of a nine-page article (half of which is written text, the other half photos). The text in (3b) is the final para­ graph of that same article. One need not have any further information about the rest of the article to conclude that it constructs an extended caricature of Richard Nixon. The writer of this article made strategic use of what psychologists studying human memory call the primacy and recency effects to guarantee that, no matter how objective the body of the article was, the reader would be left with a negative assessment of Richard Nixon. The same conceptual tools of primacy and recency are used in (4) to con­ struct an icon of Richard Nixon. As with (3), the two texts in (4) could each be treated independently as icons. However, they are textually linked in virtually the same way as are (3a) and (3b). That is, (4a) is the first piece of text that follows the title of a four-page article (less than half of which is written text), and (4b) is the final paragraph in the same article. (4)

a. Richard Nixon failed more spectacularly than any other U.S. President, yet by sheer endurance he rebuilt his standing as the most important figure of the postwar era. b. History will judge Richard Nixon as much more than the Water­ gate man. And he leaves another, brighter monument: his own

30

BRUCE W. HAWKINS

superhuman determination and stamina. It seems almost impos­ sible that he has finally been defeated. Given these two "bookends", it is clear that this particular article constructs a very positive image of Richard Nixon, i.e., an icon. Like Richard Nixon, heavyweight boxer George Foreman has had his pro­ fessional life documented by numerous icons and caricatures. The earliest George Foreman icon was constructed in the visual medium of television during the 1968 Summer Olympics in Mexico City. The text in (5) attempts to capture some of the significant features of that visually constructed George Foreman icon. (5) We first met him during the troubled summer of 1968, a year of inva­ sive images, most of which had strong racial overtones. A whirlwind media tour of some of America's "greatest" cities had pummeled us for weeks with images of some of the most violent race riots in our nation's history. We were almost counted out after a murderous com­ bination of bloody images: a civil rights leader shot dead on a motel balconey in Memphis and the presidential candidate who was the dar­ ling of the civil rights movement slain in similar fashion in the kitchen of a ritzy LA hotel. Reeling from these images, we began looking for­ ward to a two-week media vacation in Mexico City. Our dream quickly faded when two black-gloved African American fists, shining with the luster of Olympic medals and the spotlight of a nation, in­ sisted with an eloquent, deafening silence that we could not run away from our social problems. Just then, along came a towering young heavyweight who strode into the ring and gave us an image to restore our faith that these problems would ultimately work themselves out. After earning a gold medal for disposing of the Soviet threat in the ring, this huge, powerful man with a cute smile wrapped his brilliantly glowing melanin around a tiny American flag and bowed. This George Foreman icon is constructed from a series of powerful con­ trasts: the violence of one form of protest against racial inequality in contrast with the eloquent non-violence of another; the overpowering strength of George Foreman, the boxer, in contrast with the gentle submissiveness of George Foreman, the American patriot; and finally the social resistance of both forms of protest in contrast with the benevolent collegiality of a powerful African Ameri­ can champion. By the early 70s, this particular image of George Foreman had faded into history along with other images of the 60s. In its place has emerged a long and not yet completed series of contrasting images of George Foreman. One of the more recent of these is an icon constructed around a single punch—the one that

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

31

felled Michael Moorer, a well-conditioned 26-year-old champion, allowing Foreman to reclaim the heavyweight crown in November 1994. The text in (6) takes a very direct approach to constructing this icon. (6) Twenty years and five days after he lost the title on an eighth-round knockout by Muhammed Ali in Zaire, Big George [Foreman] deliv­ ered probably the most dramatic punch in the history of the sport, a crushing, overhand right that put Moorer down and out at 2:03 of the 10th round Saturday night at the MGM Grand. The blow made Fore­ man the oldest heavyweight champion in history — by an incredible eight years . "Given the circumstances, that probably was the greatest punch in boxing history," said Foreman's promoter Bob Arum. "For him to lose the title 20 years ago in Zaire, to come back and to throw a punch like that to regain it... it's amazing. It's beyond amazing." Foreman's bolt out of the blue came with him trailing on all three scorecards and seemingly headed to the postboxing life to which dis­ believers had assigned him. But, by virtue of that one huge punch, Foreman won not only the World Boxing Association and International Boxing Federation ver­ sions of the title, but the rarest of distinctions. Like AH, Babe Ruth and Michael Jordan, he now has transcended his sport and become an icon. This text constructs the Big Punch George Foreman icon with a strategic use of redundancy. Notice, in this regard, that there are no less than seven dif­ ferent references to the punch in (6). A different report of the fight which ap­ peared in the 14 November 1994 issue of Sports Illustrated takes a very differ­ ent approach to constructing the same Big Punch icon. This Sports Illustrated report can be summarized with reasonable success as A Tale of Two Punches. Taking strategic advantage of the primacy effect, the report opens with a twopage picture of Foreman's floored opponent being counted out by the referee, a much smaller inset picture of Foreman standing over his opponent, a caption appropriate to these two pictures, the title "KO'd" in large, bold type, and only the text in (7): (7) With one blockbuster right to the chin of Michael Moorer, George Foreman stunningly regained the heavyweight crown. Taking some license with the recency effect, the four-page article, half of which is text and half photographs, does not mention the punch again until the fourth paragraph from the end. That text appears in (8).

32

BRUCE W.HAWKINS

(8) Late in the 10th round, Foreman hit Moorer with a left-right com­ bination to the forehead that had the effect of disgusting the champion. He momentarily dropped his hands, as if he was mad at himself. Then Foreman hit him with that chopping right hand and bedlam ensued, Foreman crouching at a neutral ring post in prayer, even as Moorer reclined on the canvas, peering into the dark lights. Much of the rest of this article constructs a narrative that focusses strategi­ cally on the pre-fight preparations in both camps. In the process, the text effec­ tively uses contrast to highlight the Big Punch by calling attention to a very dif­ ferent punch that felled a young, imposing George Foreman twenty years ear­ lier. Interestingly, the account of that earlier punch mentions explicitly the role it played in the making and unmaking of different George Foreman images: (9) "Foreman is a con artist," said Atlas [Teddy Atlas, Moorer's trainer] in the days before the fight. "In the old days, he had the bully con. But in the ring in Zaire that night AH shredded that con. He destroyed him with the Rope-a-Dope, destroyed him because for the first time in his life, Foreman's bully con did not work." In 1974, just a year after he overwhelmed Joe Frazier to win the title, Foreman was indeed re­ garded as the meanest man alive. But on that strange night in Zaire, AH neutralized the hate, absorbing it with his arms until Foreman's fury was spent. "Once that happened," Atlas continued, "there was nothing left in George but character, and George had very little there. You look at the film of that fight, that pirouette of his across the ring. That was not a knockout punch, he wasn't seeing the dark lights. That punch was just a convenient reason to pirouette. He submitted to AH ... And you know why he's dangerous? Because, 20 years later, he knows he could have taken that punch. He knows he quit. And he knows that if he had taken the punch, it wouldn't have hurt as much as what he's lived with." We have seen in the cases of Richard Nixon and George Foreman that the amount of text used to construct an icon or a caricature can vary greatly. The caricature in (2), for example, was constructed with less than 80 words, while a number of others we have considered developed over a few pages of written and photographic text. Despite this considerable difference, we haven't even begun to approach either extreme. In this regard, it is relevant to note that the August 2, 1993 issue of Time magazine characterized McGinniss' depiction of Teddy Kennedy in The Last Brother as a "stony caricature". McGinniss dedicates more than 600 pages of text to the development of this caricature. Near the other ex­ treme of textual quantity is the caricature in (10) of Susan Smith, the young

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

33

woman from South Carolina who killed her two young sons by driving her car into a lake with the boys still strapped into their car seats. (10)

She's slime, just slime.

Although this caricature comprises only five morphemes (actually four, one of which is reduplicated), it has a sophisticated structure integrating a number of significant tools of selection. The caricature is built on the foundation of a cate­ gorization of Susan Smith as slime. While this categorization is overtly meta­ phorical, the reader need not dig deeply to see a serious attempt to associate this child-murdering young mother with the lowest forms of life on earth. The writer emphasizes the perceived appropriateness of this categorization with strategic redundancy. Finally, the use of just reflects the conceptual strategy of selective inattention. Selective inattention is criterial for categorization in that it allows us to ignore certain significant differences between experiences so that we can at­ tend primarily to the similarities between those experiences. In this instance, however, the selective inattention has the effect of a refusal on the part of the speaker to recognize any experiences in the life of Susan Smith other than her role in drowning her two sons. I will close this exposition on iconographie textual reference with the cari­ cature of Louis Farrakhan in (11). (11)

Louis Farrakhan is a problem. He is a problem for the Rev. Benjamin Chavis of the N. A. A.C.P. and Abraham Foxman of the Jewish Anti-Defamation League of B'nai Brith, who met last week to discuss what to do about him in a meeting so sensitive they would not even confirm he was the topic under discussion. On Saturday, the N.A.A.C.P. said it would con­ vene a national summit of black leaders and would pointedly in­ clude Farrakhan as a gesture of support, despite expected Jewish condemnation. "We have every right to convene African-American leadership," said Chavis. "There's a deep hunger in our commu­ nity." He is a problem for the Congressional Black Caucus, whose chairman, Representative Kweisi Mfume of Maryland, has em­ broiled himself in controversy by pledging a "covenant" of coop­ eration—since disavowed—with both Farrakhan and mainstream black leaders. He is a problem for a broad range of American blacks, who rightly fear that his anti-Semitic rhetoric erodes the moral authority of his appeals against racism and who are chagrined that his Nation of Islam, long an angry voice of the underclass, now enjoys a fol­ lowing among college students.

34

BRUCE W.HAWKINS

He is a problem for American Jews, who want to ensure that his brand of racism means automatic disqualification from national de­ bate. He is a problem for the vast majority of Islamic Americans, who already suffer from having their religion equated with hostage tak­ ing and terrorism and who mostly reject Farrakhan's racial isolation and abuse of other faiths. He is a problem for some of his adherents, who hear in his speeches black self-love and self-help and who see the Nation of Islam as a force against crime and drugs, bringing order and disci­ pline to neighborhoods with almost none-yet who know that many of their associates hear only hatred in his preachments. And Farrakhan, still impetuous at 60, is a problem for himself. In private a calm, seemingly rational man yearning for a place among trusted elders of his race, he is apt in public to get carried away on a wave of rhetoric and say things so intemperate, so easily misun­ derstood—and sometimes not misunderstood—that he thwarts his ambition. Above all, he is a problem for an America that is increasingly multiracial and multicultural and is consequently in growing need of tolerance and mutual respect. His success underscores two ugly truths of American life. A great many black Americans view their fellow white citizens with anger. And a great many white Ameri­ cans view their black fellow citizens with fear. As with a number of other caricatures we have seen, including (10), this one is built on the foundation of a categorization. However, this categorizing statement is different from others analyzed earlier in that it constitutes an ex­ plicit judgment: Louis Farrakhan is a problem. The text uses strategic redun­ dancy to emphasize this judgmental categorization. Most interesting, however, is the way the text also uses generalization across distinct communities to under­ line the appropriateness of the judgment that Louis Farrakhan is a problem. By calling attention to the variety of communities which might share the negative assessment of Farrakhan, the text is effectively suggesting that these distinct communities may differ significantly in the ideological positions they take on other issues. The power of the caricature is derived, in a large part, from the text's suggestion that the negative judgment of Farrakhan can be arrived at while operating within a wide variety of ideological systems.

THE SOCIAL DIMENSION OF A COGNITIVE GRAMMAR

35

5. Conclusions We have examined a number of cases of iconographie textual reference. In the process, we have seen the powerful role that iconographie reference can play in determining how a particular referent will be interpreted. In this regard, we have seen that the same referent can be constructed either as an icon or a carica­ ture, depending upon the ideological position of the person constructing the text. In such cases, the referent can become effectively imprisoned by a textual battle of icon and caricature raging on the battlefield of that referent's life experiences. We have also seen that the amount of text dedicated to constructing an ico­ nographie reference can range from a matter of morphemes to an entire volume. Before closing, it is useful to redirect attention to the sociopolitical dynamic of interpellation in order to call attention to the close relationship between inter­ pellation and iconographie textual reference. Recall that interpellation is the so­ ciopolitical process through which the individual becomes subject to the worldview of the prevailing ideology. Iconographie textual reference is among the discursive practices through which interpellation is accomplished. Through ico­ nographie reference, an individual is defined relative to a particular conventional image from a culturally-based system of images. This system constitutes a cognitive iconography. There is a close relationship between ideology and ico­ nography. Ideology is ubiquitous in our lives, but its common-sensical nature makes it something that we are generally not equipped to articulate effectively. That is, we live by it, but we generally don't think about it and, therefore, few of us have any ability to describe it. However, we do periodically find ourselves engaged in discourse in which our ideology is challenged. This is where the re­ lationship between ideology and iconography becomes apparent. Our ideology enters our conscious awareness most directly in the form of our iconography; our icons are concrete symbols of our abstract ideological values. Since few of us are equipped linguistically to debate ideologies, we fall back on our icons. In a discursive battle of ideologies, we invoke our icons who stand metonymically for the ideology we embrace and defend.

References Althusser, Louis. 1971. Lenin and Philosophy. New York: Monthly Review Press. Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, MA.: The MIT Press. Derrida, Jacques. 1976. Of Granimatology. Chicago: The University of Chicago Press. Dretske, Fred I. 1983. Knowledge and the Flow of Information. Cambridge, MA: The MIT Press. Hodge, Robert and Günther Kress. 1993. Language as Ideology (2nd edition). London: Routledge.

36

BRUCE W.HAWKINS

Jackson, Leonard. 1991. The Poverty of Structuralism. London: Longman Group UK. Kittay, Eva Feder. 1987. Metaphor: its Cognitive Force and Linguistic Structure. Ox­ ford: Oxford University Press/Clarendon Press. Lakoff, George. 1987. Women, Fire, and Dangerous Things. Chicago: The University of Chicago Press. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, vol. I: Theoretical Prerequisites. Palo Alto: Stanford University Press. Langacker, Ronald W. 1988 "A Usage-Based Model." In Brygida Rudzka-Ostyn (ed.), Topics in Cognitive Linguistics (Current Issues in Linguistic Theory 50.). Am­ sterdam: Benjamins. Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, vol. II: Descriptive Application. Palo Alto: Stanford University Press. Mowitt, John. 1988. uThe Resistance in Theory", Foreword to Paul Smith, Discerning the subject, Minneapolis: University of Minnesota Press, ix-xxiii.

Data sources Friedrich, Otto. 1994. "I have never been a quitter." In Time, 2 May 1994, 43-51. Gibbs, Nancy. 1994. "Death and deceit." In Time, 14 November 1994, 43-48. Henry, William Α., III. 1994. "Pride and Prejudice." In Time, 28 February 1994, 2127. Hoffer, Richard. 1994. "KO'd". In Sports Illustrated, 14 November 1994, 18-23. Knight-Ridder/Tribune Service. 1994. "King George." In The Pantagraph (Bloomington, Illinois), 8 November 1994, C-l. McGinniss, Joe. 1993. The Last Brother: the rise and f all of Teddy Kennedy. New York: Simon and Schuster. Stacks, John F. 1994. "Victory in defeat." In Time, 2 May 1994, 26-29. The Pantagraph (Bloomington, Illinois). 1994. "What others are saying," 30 April 1994.

Even, sae/sura/mo as Constraints on Contextual Assumptions Isao Higashimori* Kobe College, Japan

1. Introduction In this paper, I would like to show that all the previous analyses concerning even are insufficient in some respects.1 Relevance Theory (henceforth RT) proposed by Sperber & Wilson (1986, 19952), Blass (1990), Higashimori & Wilson (1996) and Lavin (to appear) can capture even not as a 'conceptual', but as a 'procedural' meaning. This newly developed cognitive analysis can be extended to provide the explanations for (i) some problematic data containing even; (ii) some Japanese counterparts sae/ sura/mo (cf. 'particle typology' Blass 1990); (iii) the relation between even in English and sae/sura/mo in Japanese, and the co-occurrence relation between saeI'sura and mo in Japanese. Theoretically, I'd like to assert the superiority of RT over other cognitive approaches like mental space theory, construction grammar, cognitive grammar and so on. So the purpose of this paper is to clarify the cognitive similarity be­ tween these particles and also the cognitive differences between them. Each particle needs a lot of detailed analysis, but the present paper is a first step to capture their cognitive similaries and differences in the framework of RT. 1.1. Even: 'procedure' and 'non-truth-conditional' In RT, even can be characterised as 'procedural', since (a) even denotes infor­ mation about computations, and (b) procedural representations with even cannot be brought to consciousness. Also, even is 'no-truth-conditional', that is, even does not contribute to the truth conditions of the utterance in which it appears, since the negative operator not cannot negate even as is shown in Example (1): (1) It is not true that {even) Max tried on the trousers.

38

ISAOHIGASHIMORI

1.2. Some problems with previous analyses of even Now let us observe example (2). Even, in this example, can be translated into Japanese by using a lot of different particles like saelsuralmolmadeldattelkatte (in some Kyoto dialect). Because of limitation of space, in this paper I will limit myself to the Japa­ nese particles such as saelsuralmo as counterparts of English even. Traditionally, (2) can be analysed to imply (2a) and (2b). (cf. Kartunnen & Peters 1979: 12) (2) (2a) (2b)

Even <Max> tried on the trousers.2 Existential presupposition: Other people besides Max tried on the trousers. Scalar proposition: Of the people under consideration, Max was the least likely to try on the trousers.

So the first problem to be considered is the existence of examples which can be used without implying existential presuppositions (e.g. 3,4) and scalar propositions (e.g. 5). (3) Since she could not find anything else to eat, Mary ate even < snakes > . (4) All the kids tried on something. Mary tried on a pair of trousers, Sue a long shawl. Even <Max> tried on a fancy tie. But then it's not really surprising. Now that his mother has married again, he joins in with things much more. (Kempson 1975: 201) (5) The administration was so bewildered that they even had lieutenant colonels making major policy decisions. (Kay 1987: 24) According to Fauconnier (1975), the semantic function of even is to mark the existence of a pragmatic probability scale, where the constituent that is the focus of even is the low point on the scale, and the scale is ordered with respect to the propositional schema which results from substituting a variable for the focus of even. But the focus 'lieutenant colonels' marked by even in (5) is not end-of-scale; in British Army ranking (e.g. LDOCE, COBUILD) the hierarchy is: general, lieutenant colonel, major, captain and sergent. The lieutenant colo­ nel is ranked under a general.3

EVEN, SAE/SURA/MO

39

1.3. Some theoretical problems 1.3.1. Where do scales come from ? The previous analyses, such as Kay (1987) and Fauconnier (1975), considered scales to be ready-made and given in advance. In order to solve problems like (5) (cf. (45)), we have to think of (i) scales not as ready-made, but as having to be constructed on the spot as part of the interpretation process; (ii) in the proc­ essing of even sentences as in the processing of any other type of sentence, we rely on a variety of assumptions retrieved from (long-term and short-term) memory or constructed as part of the interpretation process: assumptions about the relevant general knowledge, and knowledge about the situation and the peo­ ple involved in that particular case; (iii) scales may be taken to be representa­ tions of the outcome of inferences (Lavin, to appear: 7). 1.3.2. Problems with 'informativeness' Kay (1987: 9) defines 'informativeness' as a relation holding between two propositions relative to a scalar model SM, in which a more informative one unilaterally entails the less informative one in SM. The use of even is character­ ized as follows: (i)

(ii)

even does not affect the truth conditions of a sentence in which it oc­ curs—having the effect merely of expressing the speaker's opinion (or someone else's) that the proposition expressed is more informa­ tive than another taken to be present in the context. (Kay 1987: 31) even possesses direct pragmatic interpretation in that it denotes (or evokes) a relation (superior informativeness) between the proposition expressed (text proposition) and one taken to be already in the con­ text (context proposition). (Kay 1987: 10)

There are two problems to be considered here. One is the definition for context as 'to be already' there in advance. In RT, a context is a psychological construct, a subset of the hearer's be­ liefs and assumptions about the world, including beliefs and assumptions about the speaker. That is, a set of assumptions retrieved or derived from memory or acquired by perception and used in the interpretation process (Blass 1990: 9). And the mechanism of context selection or context construction must be in­ corporated in choosing among a range of potential contexts in which to process a given item of information, that is, the aim will be to choose the context in which that information will yield the greatest possible effects for the smallest possible effort. The other problem is that a pragmatic criterion based on informativeness would rule out reminders/admission/acknowledgements/repetitions/paraphrases/

40

ISAO HIGASHMORI

summing up/reiterations of necessary truth. So this criterion seems to be inade­ quate. In RT, human cognition is relevance-oriented, and we pay attention to in­ formation that seems relevant to us. Relevance is a cognitive principle which is defined in terms of contextual effects and processing effort.4

2. A Relevance-Theoretic Account of even/sae/sura/mo In RT, the hearer's cognitive environments can be changed into the following three ways: utterance Ρ Ρ Ρ

context Ρ -Ρ P->Q

contextual effects strengthening contradicting & eliminating Q=a contextual implication

Contextual effects: (a) strengthening an existing assumption (b) contradicting and eliminating an existing assumption (c) combining with an existing assumption to yield a contextual implication. Processing effort depends on: (a) the form in which information is presented (b) the accessibility of the context. Relevance: (a) the greater the contextual effects, the greater the relevance (b) the smaller the processing effort, the greater the relevance. Optimal relevance: An utterance, on a given interpretation, is optimally relevant if: (a) it achieves enough effects to be worth the hearer's attention (b) it puts the hearer to no gratuitous effort in achieving those effects. In my analysis, the cognitive similarity between even, saelsuralmo comes from the fact that these particles can be used as a denial of a contextual assump­ tion as is shown in Table 1. Table 1 deals with cognitive similarity between even and saelsuralmo'. the same 'procedural', meaning, i.e. a denial of a contextual assumption.

EVEN, SAE/SURA/MO

41

Table 1. Even, saelsuralmo as constraints on context

Utterances - > Context Selection > ('EXPLICATURE') (a 'contextual assumption') Even Ρ Ρ sae Ρ sura Ρ mo

Ρ

The hearer already knows or expects -P

Contextual Effects Denial of a contextual assumption

Now we shall look into individual cases as below. 2.1. Even as a constraint on contextual assumptions5 (6) Even < Max > tried on the trousers, (cf. Lavin, to appear: 11) utterance with — > context even selection (a) explicature

(b) contextual assumptions

> contextual effect (c) Denial of a contextual assumption

Explicature: (6a) χ tried on the trousers. Context selected by even: (6b) the hearer already knows or expects that χ is not Max (i.e. χ = -P) Focus processed by even: (6c) χ is Max (i.e. χ = P) Contextual effect: (6d) Denial of a contextual assumption (i)

Even utterances are usually uttered in a context where some such assump­ tions already exist (Lavin, to appear: 10). (ii) The contribution of even is to signal the existence of an assumption in the context "Max did not try on the trousers", which contradicts the main ex­ plicature of the utterance, and probably leads to its elimination. (iii) Processing an utterance with even should give rise to two logical forms. One of them will be developed into a fully propositional form (=main explicature), and the other is an incomplete logical form or assumption schema designated by the function of contrastive stress. The incomplete logical form (or assumption schema) with an empty slot in the place of highlighted constituent for the utterance (6) is as follows: (6a) χ tried on the trousers.

42

ISAO HIGASHIMORI

The reason that <Max> is chosen as the focused constituent is because pre­ cisely this assumption will have the biggest impact as far as the derivation of the contextual effects goes, it is the most relevant one in the context, the one that brings about the greatest modification of our set of assumptions (Lavin, to ap­ pear: 11). 2.1.2. Japanese particles: sae/sura/mo Next, let us look at the Japanese counterparts. The same kind of computation we have shown in (6) is applicable to Japanese counterparts. Observe the examples (7) to (9): (7) kodomo de sae kono mondai wa child prt this problem TOP 'Even < a child> can solve this problem.'

tokeru can-solve

Explicature: (7a) χ (wa)kono mondai wa tokeru Context selected by sae: (7b) The hearer already knows or expects that χ is not 'kodomo' Focus processed by sae: (7c) χ is 'kodomo' Contextual effect: (7d) denial of a contextual assumption Examples (8) and (9) can be analyzed in the same way as in (7): (8) kare wa nihongo sura hanase nai he TOP Japanese prt can speak not 'He cannot even speak < Japanese > . ' Explicature: (8a) kare wa χ ga hanase nai. Context selected by sura: (8b) The hearer already knows or expects that χ is not 'nihongo.' Focus processed by sura: (8c) χ is 'nihongo' Contextual effect: (8d) denial of a contextual assumption (9) ima de wa watashitachi wa tsuki ni mo now we TOP moon to prt 'Now we can even go to < the moon>.'

ikeru can go

Explicature: (9a) ima de wa watashitachi wa χ ni ikeru. Context selected by mo: (9b) The hearer already knows or expects that χ is not 'tsuki.' Focus processed by mo: (9c) χ is 'tsuki.' Contextual effect: (9d) denial of a contextual assumption

EVEN, SAE/SURA/MO

43

2.2. Cognitive differences between even, sae/sura/mo Now let us turn to the cognitive differences between these particles. Table 2 designates that the cognitive differences between even, saelsural mo come from the different way of processing in which one focused element is picked up. Table 2. Different processing for a focused constituent selection

Etymologically, the English even was used to show "an emphatic marker of identity" by comparing two values like Japanese mo in Table 2, but from the 16th century onwards, it has been used as a focus particle. The Japanese particle sae was originally related to 'soe' and was an additive marker like Japanese mo, but in present-day Japanese, it is used as an emphatic/unexpected marker. From a diachronic point of view, even and sae show some similarity in their develop­ ment.7 Let us now investigate each case in some detail. 2.2.7. English particle: even Procedural meaning of even: Process < x > as the focused element, since it will have the biggest impact as far as the derivation of the contextual effects goes. It is important to bear in mind that the above characterisation of the proce­ dural meaning of even is defined not by using the notion of 'scale' but by basic notions such as 'relevance' and 'contextual effects'. In (10), the whole domain, or the members of the same category, are linguistically expressed. In (10a), for instance, everything is a domain, and a swimming pool is picked up as a focus of even. (10)

a. The hotel had everything. There was even < a swimming pool>. (COBUILD)2

44

ISAO HIGASHIMORI

b. Their parents disapproved of many of their friends. Even . (Longman Written Corpus (=LWC))  The British, the Americans, even have their ar­ eas, called concessions. (LWC) d. Americans go to a far-off city for an evening of theater music or even < a movie > . (Lanier, 9) In examples like (11) and (12), the members of the same category are not explicitly expressed but they are computed by our general knowledge or the su­ perlative forms. In (12c), the whole domain can be calculated as the things that can be sold (e.g. my car, my house and my blood). (11)

a. I cannot ignore even < the slightest chances > that she may have survived. (LWC) b. The main problem with rubber, even < 100% virgin rubber > , is that it oxidizes quickly. (LWC)

(12)

a. You couldn't even trust < a policeman > . {Hiragana Times (=HT) July 1994: 66) b. Now, even can freely enjoy dressing up and being fashionable. (HT, July 1994: 50)  I even had to sell <my blood > . (LWC)

Let us now go on to some Japanese counterparts. 2.2.2.

Japanese particles: sae/sura/mo

2.2.2.7. Sae: Up-/Downwards processing. Procedural meaning of sae: Process < χ > as a focused element by processing other members in the same category. (13)

boku wa 10 en sae I TOP 10 yen prt 'I don't even have < 10 yen>.'

nai. not

(13) seems to imply 'I don't have < 100 yen, 1000 yen, and 10,000 yen> (i.e. upwards processing). (14)

kono mondai wa daigakusei sae toke nai this problem TOP university student prt can solve not 'Even < a university student> cannot solve this problem.'

(14) seems to imply < A high school student, a junior high school student, a schoolboy/girl > cannot solve this problem, (i.e. downwards processing)

45

EVEN, SAE/SURA/MO

2.2.2.2. Sura: an exceptional case processing. Procedural meaning of sura: Process < x > as an exceptional case (i.e. two different domains (positive and negative) can be processed; one member has been picked up in a positive (or negative) domain, and then it is extracted (or inserted) as an exceptional case.). (15)

a. 10 en sura nai. 10 yen prt not 'I don't even have < 10 yen>.' b. konna mondai kodomo de sura dekiru. this problem child prt can solve 'Even a child can solve this problem.'

2.2.2.3. Mo; a contrastive processing. Procedural meaning of mo: Process < x > as a contrastive element with < y > (i.e. χ and y are contrastive elements in the same category). (16) cf.

tsuki ni mo ikemasu moon LOC prt can go 4 You can even go to < the moon > . ' Are mo kore mo umaku iku. that prt. this prt well go. 'everything goes well.'

In (16), the hearer contrasts two situations like (You can go abroad) and (you can go to < the moon>). 2.3. Even, sae/sura/mo and constraints on implicatures, and parallel processing Table 3. Cognitive similarities and differences between even, sae/sura/mo: con­ straints on implicatures and parallel processing (cf. Blass 1990).

Particles Ways of processing Parallel confirmation Parallel premises Backwards confirmation Backwards contradiction

English too also even + * + +

+ +

* *

? * * + +

Japanese sae sura ?

* ?* + +

?

?* + (?) +

German auch

mo +

+

+

+

+ +

+ *

Let me summarize the main points of Table 3. With respect to the direction of processing in which relevance is to be sought, mo can be used both for paral-

46

ISAO HIGASKMORI

lel and for backwards processings, but even, sae, sura seem to be used only for backwards processing. Needless to say, there are some discrepancies in accept­ ability especially in parallel processing as is shown by?* In Table 3 note that unlike mo, English also is to indicate to the hearer that relevance is not to be achieved by either backwards confirmation or backwards contradiction. Each processing category will be illustrated with an example (17)-(24). 2.3.1. English particle: even Even works as a constraint on computations. It introduces a parallel implication. 2.3.1.1. Parallel confirmation (17)

A: John is living in Kyoto. B: l*Even <Mary> is living in Kyoto. (17A): If he is living in Kyoto (=Premise 1), then John can visit a lot of old temples (=Conclusion 1). (17B'): If she is living in Kyoto (=Premise 1), then Mary can visit a lot of old temples (=Conclusion 2). The use of even doesn't seem to encourage the hearer to make the parallel processing as is shown in (17B). (17B) does not achieve relevance by providing additional evidence for a contextual implication already derived from (17A). 2.3.1.2. Parallel premises (18) (18A): (18B'):

A: John is coughing. B: *John has even < a slight fever > . If he is coughing (=Premise 1), then John has caught a cold (=Conclusion 1). If he has a slight fever (=Premise 2), then John has caught a cold (=Conclusion 1).

(18) shows that the use of even also doesn't seem to encourage the hearer to reach the same Conclusion 1 by adding another Premise 2 like (18B). (18B) with even cannot be understood as providing additional support for the conclu­ sion which has been derived from (18A): Even as a constraint on implicatures or contextual effects. 2.3.1.3. Backwards confirmation Even strengthens an existing assumption: (19)

A: John knows Japanese very well. B: He can read even < difficult Chinese characters. >

EVEN, SAE/SURA/MO

47

To understand (19B), the hearer has to have access to the assumptions such as (19B·). (19B'): If somebody can read difficult Chinese characters, then he knows Japanese very well. (19B) is used to confirm, or strengthen (19A), by providing evidence for its truth. The proposition expressed in (19B) is understood as providing additional support for the conclusion which has been derived from (19A). 2.3.1.4. Backwards contradiction Even contradicts and eliminates an existing assumption. (20)

A: John is rich. B: But he didn't even have < 10 yen> yesterday.

To understand (20B), the hearer has to have access to assumptions such as (20B'). (20B'): If somebody doesn't have 10 yen, then he is not rich. (20B) is used to contradict and eliminate (20A), with the implication that John is not rich. Cf. A contextual implication: Even combines with an existing assumption to yield a contextual implication (20B + 20B')~>a contextual implication: John is not rich (cf. Blass 1990, Higashimori 1994a). 2.3.2. Japanese particles: sae/sura/mo Sae/sura/mo and constraints on implicatures, and parallel processings. Parallel confirmation (21) A: John wa Kyoto ni sundeimasu John TOP Kyoto LOC is living 'John is living in Kyoto.' B: Mary ?*sae/?sura/mo Kyoto ni sundeiru Mary prt Kyoto LOC is living 'Even <Mary> is living in Kyoto > . ' Parallel premises (22) A: John wa seki John TOP cough 'John is coughing.'

o ACC

shiteimasu is ... doing

48

ISAOHIGASHIMORI

: sukoshi netsu ?*sae/?*sura/mo arimasu slight fever prt has 'He has even < a slight fever > . ' Backwards confirmation (23) A: John wa nihongo  yoku shitteimasu John TOP Japanese ACC very well know 'John knows Japanese very well.' B: muzukashii kanji sae/(?)sura/mo yomemasu difficult Chinese characters prt can-read 'He can read even < difficult Chinese characters. > ' Backwards contradiction (24) A: John wa okane-mochi des  John TOP rich is 'John is rich.' B: Demo kino wa 10 en sae/sura/mo but yesterday TOP 10 yen prt motte nakatta. have not-PAST 'But he didn't even have < 10 yen> yesterday.' It is interesting to note that there is some discrepancy in acceptability in parallel processing of saelsura. As I have mentioned above, these particles en­ code procedural information which lacks direct access in our brain. Now let's take a brief look at Diagram 1. Diagram 1 is my rough idea of the meaning relation from core procedural meaning of even to various uses such as surprise, unexpectedness, emphasis, agreement, and disagreement. My point is that emotions like surprise and unexpectedness are not primitives, but they are derived from the cognitive gap as was discussed above, and that agreement and disagreement are related to some social assumptions such as interpersonal rela­ tions.

Diagram 1. Core procedural meaning of even and its various uses

EVEN, SAE/SURA/MO

49

3. Some Problems 3.1. Problems with even All the problematic examples containing even can be explained by applying the same processing mentioned in Table 1. 3.1.1. Examples like (3) and (5) Example (3) can be analysed as follows: Explicature: (3a) Since she could not find anything else to eat, Mary ate x. Context selected by even: (3b) χ is not 'snakes'. < According to our general knowledge, we normally don't eat snakes. > Focus processed by even: (3c) χ is 'snakes' Contextual effect: (3d) Denial of a contextual assumption. Next, example (5) can be analysed as follows: Explicature: (5a) The administration was so bewildered that they had χ making major policy decisions. Context selected by even: (5b) χ is not 'lieutenant colonels' < According to our general knowledge of army ranking, lieutenant colonels normally do not make major policy decisions. > Focus processed by even: (5c) χ is 'lieutenant colonels' Contextual effect: (5d) Denial of a contextual assumption. 3.1.2. (25)

Case with more then one even8 John: How did your class do on the quiz? Mary: Fantastic, even <my slowest student > even got . (cf. Kay 1987)

Explicature: (25a) Fantastic, χ got y. Context selected by even: (25b) χ is not 'my slowest student'; y is not 'the hard­ est problem' Focus processed by even: (25c) χ is 'my slowest student'; y is 'the hardest problem' Contextual effect: (25d) Denial of contextual assumptions. 3.1.3. Other uses of even Adjective-focus use of even:9

50

ISAOHIGASHIMORI

(26)

One is aware of strong, even < violent > , emotion behind what is actually said. (LWC) (26a) One is aware of strong, χ emotion behind what is actually said. (26b) χ is not 'violent'.

Focus processed by even: (26c) χ is 'violent'. Contextual effect: (26d) Denial of a contextual assumption. Transitive verb-focus:10 (27) Mary even < admires > Bill. (Karttunen & Peters 1979: 24) (27a) Mary χ Bill. (27b) χ is not 'admire' Focus processed by even: (27c) χ is 'admire' Contextual effect: (27d) Denial of a contextual assumption. VP-focus: (28) Bill even < drinks beer>. (Karttunen & Peters 1979: 24) (28a) Bill x. (28b) χ is not 'drink beer' Focus processed by even: (28c) χ is 'drink beer' Contextual effect: (28d) Denial of a contextual assumption. Comparative intensifying use of even (cf. Berckmans 1993: 599). (29) Bill is even < taller > than John. Cf. Japanese 'zutto' (29a) Bill is χ than John. (29b) χ is not 'taller' (cf. more handsome/smarter/richer ...) Focus processed by even: (29c) Contextual effect: (29d)

χ is 'taller' Denial of a contextual assumption.

ADV-focus use of even11 (30) Even < if it rains > , the game will continue. (30a) if x, then the game will continue. (i.e. x-*Q) (30b) χ is not 'it rains' (i.e. χ is not P) Focus processed by even: (30c) χ is 'it rains' (i.e. χ is P). Contextual effect: (30d) Denial of a contextual assumption —Cf. (30a) + (30b) = -P -> Q.

51

EVEN, SAE/SURA/MO

(31) Even < if my wife smoked cigarettes > , he would fire me. (31a) if x, he would fire me. (i.e. χ -* Q) Explicature of counterfactual induced by if (of. Smith & Smith 1988: 348): (3Γ): The hearer does not think that 'my wife smoked cigarettes' (i.e. The hearer doesn't think that P.). (31b) χ is not 'my wife smoked cigarettes' (i.e. χ is not P). Focus processed by even: (31c) χ is 'my wife smoked cigarettes, (i.e. χ is P) Contextual effects: (31d) Denial of a contextual assumption. —Cf. (31a)+ (3 lb)+counterfactual explicature= Ρ -> Q. (32)

Even < though you dislike ancient monuments > , Warwick Castle is worth a visit. (Quirk et al. 1985: 1099)

(32a) Though x, Warwick Castle is worth a visit. Contextual assumption induced by though: (32') 'You dislike ancient monu­ ments'. Contextual assumption: (32b) χ is not 'you dislike ancient monuments.' Focus processed by even: (32c) χ is 'you dislike ancient monuments.' Contextual effects: (32d) Denial of a contextual assumption (32b) Strengthening of a contextual asumption (32'). 3.1.4.

Focus and relevance

3.1.4.1. Even and variables Information which has an unassigned variable does not yield sufficient rele­ vance, so the examples in (33) are unacceptable. (33)

a. *Even < someone > has already arrived. (Fraser 1971: 153) b. *Even <who> likes dogs? cf. Japanese: demo 'Whoever'

3.1.4.2. Even and contextual assumptions (34) *Mary took a swim even < since it was warm. > (Fraser 1971: 162) (34a) Mary took a swim since x. Contextual assumption induced by since: (34') Contextual assumption induced by even: (34b)

The hearer already knows that it was warm. The hearer already knows and expects that it was not warm.

52

ISAOHIGASHMORI

Focus processed by even: (34d) x is 'it was warm'. Because of the contradiction between these two contextual assumptions (34') and (34b), (34) is unacceptable. 3.1.4,3. Other 'procedural' uses of even. The advantage of my cognitive anlaysis is that it can account for some problematic data from (35) to (37) in which no focused element is explicitly described, or even can be repetitively used. The existence of these data can be easily explained by our procedural characterisa­ tion of even. For instance, the repetition of even in (37) can be analysed as a demand for additional processing effort. The speaker must expect it to achieve additional contextual effects such as a hesitating or intensifying function. Some­ times even can be used alone in initial position as in (36) without any focused element.12 Like the Japanese sae + mo, the English even and also can be used together as is shown in (38). Let us call these phenomena Cognitive Harmony. No focused element: (35)

Joseph: Yeah. And so I started out just, you know ... I didn't even <]6> -nobody had thought of anything. {Best of the English Journal Interviews (BEJ) 7: 34)

Even used alone: (36)

But you've got lots of Lego. Even, I've never seen it all set up at the same time. (Longman Spoken Corpus (LSC))13

Repetitive use of even: (37)

and in fact, even, even < lions > do this. (LSC)

Cognitive harmony case: (38) 3.1.5.

But now, of course, intellectuals, even < scientists > also have re­ sponsibility on the social problems. (BEJ 6: 42)

Processing effort and implicit even

(39)

a. John is very friendly, (even) if he is an aristocrat. b. (Even) though she hadn't eaten for days, she looked healthy (cf. Horn 1989: 399).

Although an utterance without even may on occasion ultimately give rise to the same contextual effects as the corresponding utterances with even, as is

53

EVEN, SAE/SURA/MO

shown in (39a, b), this will only happen at a higher processing effort (cf. Lavin, to appear: 8). 3.2. Problems 3.2.1. Sae/sura and negation As we have shown in Table 2, Japanese sura tends to be used in negative uttererances like (40b). The unacceptability of sura in (40a) is due to the fact that it is not easily inserted as an exceptional case from a negative into a positive do­ main. (40)

a. kare wa kono kuruma he TOP this car 'He can even buy .'

sae/?*sura prt

kaeru. can-buy

b. kara wa kono kuruma sae/sura kae nai he TOP this car prt can-buy not 'He cannot even buy ' 3.2.2.

Sae/sura and sufficient conditions

(41)

eigo sae/*sura dekire ba, gaikoku ni ikeru English prt can if abroad LOC can-go 'If you can even speak < English > , you can go abroad. '

The acceptability of sae in (41) is due to the fact that sae is a signal of up/downwards processing, and so the sufficient conditions can be assessed by comparing all the alternatives in the same class, while the unacceptability of sura in (41) comes from the fact that sura does not make the hearer compare all the alternatives but makes him choose only one exceptional member. 3.2.3.

Sae/sura/mo and pseudo-imperatives

(42)

mono  itte *sae/*sura/mo, John o korosu zo something ACC talk prt John ACC shoot 'Even talk and I'll shoot John.'

According to RT, a pseudo-imperative sentence like (42) can be analysed to describe a 'desirable and potential world', and so mo can be used to compare two different worlds such as a 'potential world' (e.g. you move) and a 'potential world' (e.g. you talk).

54

ISAO HIGASHIMORI

3.2.4.

Direction of processing and two particles sae, mo

(43)

saikin no kodomo-tachi wa, ofuro de sae mol*mo recent GEN child Plural TOP bath LOC prt prt mizugi  kiru swimsuit ACC wear 'Recently, children wear swimsuits even < in the public bath > . '

The particles sae + mo can be used in that order, since after establishing the domain of all the contrasting members in the same class by the use of sae, it's easy to compare just the two of them which were already mentioned by us­ ing mo: (a swimming pool) and (a public bath) are computed vertically by the use of sae and horizontally by the use of mo, and so the result of sae + mo seems to be much more emphatic than sae alone.

On the other hand, *mo+sae can be blocked by the following cognitive constraint: it takes a lot of processing efforts first to compare just the two dif­ ferent places, times, or worlds by using mo, and then other additional members of its domain must be calculated by the use of sae. In short, *mo+sae seems to be blocked by the difficulty in establishing the whole domain until the utterance of sae.

The same kind of explanation can be applicable to sura mol*nw sura and made mo/*mo made:

55

EVEN, SAE/SURA/MO

(44)

Kare wa nihongo sura mo/*mo sura hartase nai he TOP Japanese prt prt can-speak not 'He cannot even speak < Japanese > . '

In (44), sura picks up only one element 'nihongo' as an exceptional case and then compares this with another possible language like 'English' by using mo. But the opposite order *mo+sura seems to be blocked by picking up two contrasting elements by mo and then one of them should be considered as only one exceptional case by sura. (45)

umi de oyoida mono no, pukapuka ukabu, ocean LOC swam floating purasuchikku plastic botoru ya potetochippusu no fukuro, soreni bottle and potato chips GEN bag and kaimen  tadayou floating kudamono made mo/*mo made sake nakerebanaranakatta no desu fruit prt prt avoid had to 'I swam in the ocean but I had to avoid floating plastic bottles, potato chip bags, and even .'

3.3. Others: te+molde+mo Ini+ Ino+molto+mo/made+ + is used when that which is expressed in the main clause is not what is expected from the content of the dependent (te+mo) clause: (46)

watashi wa ame ga I TOP rain NOM 'I'll go even < if it rains > . '

futte mo fall prt

desu+mo— > demo: (47)

sensei demo machigaeru. teacher prt make mistakes 'Even < a teacher> makes mistakes.'

(48)

koko kara demo fujisan ga here from prt Mt. Fuji NOM 'Even we can see Mt. Fuji.'

mieru. can see

iku. go

56

ISAOHIGASHIMORI

ni +: (49)

kankoku ni wa "warikan" no shuukan Korea in TOP splitting the bill POSS custom ga arimasen. NOM not exist 'In Korea, there's no custom of splitting the bill'. guruupu no baai ni ,   motteiru group POSS case prt money ACC having danse ya, man and 'Even < when there's a group>, it's the guy with money and' sono hi kibun ga ii hito ga, zen-in that day mood NOM good person NOM everyone nobun  ogorimasu. POSS portion foot bill 'the one in a good mood that day that foots the bill.'

made is a particle to indicate a spatial, temporal or quantitative limit or an un­ expected animate/inanimate object. (Makino 1986: 225) (50)

Nihonjin wa, kurisumasu, barentain dee, harowiin, Japanese TOP Christmas Valentine's day Halloween kanshasai made tanoshinde iru. Thanksgiving prt is enjoying 'Japanese enthusiastically celebrate Christmas, Valentine's Day, Halloween, even < Thanksgiving > . ' (Hasegawa 1994: 129)

We don't have enough space here, so we have to leave a detailed discussion about teldelni/no/tolmade +mo.14

4. Conclusion In order to define the lexical items such as even, saelsuralmo, we have to clar­ ify their cognitive inferential constraints, that is, how these linguistic expres­ sions affect the hearer's mental processing. Their functions are like traffic sig­ nals. In this paper, we have argued that (i) the cognitive similarity between even, saelsuralmo can be explained by the same kind of procedural meaning (i.e. constraint on context as was discussed in Table 1), and (ii) their cognitive dif-

EVEN, SAE/SURA/MO

57

ferences can be captured by the different ways of processing to pick up a fo­ cused constituent (in Table 2), and also their different constraint on implicatures as was shown in Table 3. In short, I have shown that RT can suggest a new and interesting approach to particle studies and particle typology and that RT is superior to other cogni­ tive approaches in capturing the true nature of these particles. Endnotes * This is a revised and extended version of a paper that appeared in Higashimori (1994b) and was presented at ICLA (International Cognitive Linguistics Association), 4th International Conference in 1995. 1. Shanon (1978) defines even as a function upwards on the scale. In formal semantics, Lycan (1991: 135) uses the notion of 'the contextually indicated reference-class', and Berckmans (1993: 593, 598)'universal quantification over a contexually determined relevant comparison class' and 'the existence of a contrast class whose members share a relevant property with the object in focus of even'. 2. Angle brackets < > indicate a focused constituent. 3. See Sperber & Wilson (1986: 208) and Smith & Smith (1988: 344) for 'anticipatory scale'. 'The set of anticipatory hypotheses forms' is called a scale. 4. From this observation it follows that 'frame knowledge' or 'default knowledge' is not suffi­ cient for accounting for the examples containing even. 5. For further details of RT, please refer to Sperber & Wilson (1986, 1995) and Blakemore (1992). 6. In RT, 'the focally stressed constituent' is called focus. The following example (a) has the NP as focus (or foreground information) and the assumption 'Mary ate something' as back­ ground information. Sperber & Wilson (1986: 217) distinguishes foreground and background information as follows: Foreground information: information that is relevant in its own right by having contextual effects; Background information: information that contributes only indi­ rectly to relevance, by reducing the processing effort required. But there still remain some theoretical problems with focus which are illustrated from (i)-(iv). (i) Identification focus ( < > ) : foreground information/prosodie peak e.g. (a) Mary < snakes > . Contrastive focus (BOLD)foreground information/-prosodic peak e.g. (b) (ii) John ate < a BIG rat in Africa > . (iii) Emphatic focus (CAPITAL): e.g. (c) MARY ate < SNAKES > (but John ate nothing), (iv) Corrective focus: e.g. (d) I said Defensive, not Offensive. 7. See König (1991: 135) and Kindaichi et al. (1972) for the original meaning of English 'even' and Japanese 'sae'. English 'even': (1) an emphatic marker of identity: e.g. She that you gaze on so?- Even I meane. (Shakespeare, Two Gentlemen of Verona, II.i). (2) a focus particle (from 16th century onwards, even came to be used as a focus particle.) Japanese'sae' (1) an additive marker (i.e. etymologically, sae is related to soe) (2) an emphatic/unexpected marker (in present-day Japanese)

58

ISAO HIGASHIMORI

8. Similar examples can be found in COBUILD on CD-ROM (1995): (i) Tony talked about the scale of the traffic problem; the growth in demand; the limit on resources; the accessibility of the Forest; and even the air we breathe even in a place like this isn't fit to breathe by world health standards, (ephemera) 9. The following examples are interesting in that even can modify an adjective backwards: (i) I find some of his habits rather unpleasant, < disgusting > even.-Cambridge International Dictionary of English (1995). Cf. He is tall; he is even < towering >. (Shanon 1987: 36). Cf. It is a good way for one or even young players to learn. (LWC) 10. The problem is to show how an actual focus is chosen from a range of potential foci. (Sperber & Wilson 1986: 203). In order to simplify the discussion here, we shall consider example (27) as a transitive verb focus. 11. See Smith & Smith (1988) and Mori (1992) for the detailed difference between even if and even though. A lot of previous analyses on concessives and conditionals such as Dirven (1995) and Fujii (1995) seem to be inadequate in that they analyse 'even if (or Japanese 'temo') as a unit. My point in this paper is that it is extremely important to note that in order to explain various cognitive behaviours of these particles, we have to consider even + if to be two separate parts. 12. Mo and even can appear in an initial position of an advertisement: (i) Gaikokujin mo nattokusuru keshohin (HT 1995) Foreigners prt convinced cosmetics 'the cosmetics that even are convinced of.' Just like even Just can be used alone: (ii) (Hearing crash in the kitchen) A: Are you okay? B: Yeah; just, the frying pan fell. 13. In Japanese, made is a goal-oriented particle. For Japanese particles, see Ninnata (1992). 14. See Higashimori (1994b) for the detailed discussion on these cases.

References Berckmans, Paul. 1993. "The quantifier theory of EVEN." Linguistics and Philosophy 16:589-611. Blass, Regina. 1990. Relevance Relations in Discourse: With Special Reference to Sis­ sala. Cambridge: Cambridge University Press. Dirven, René. 1995. "The conceptual space between conditionals and so-called con­ cessions." Paper delivered at International Cognitive Linguistics Association 4th International Conference in 1995. Fauconnier, Gilles. 1975. "Pragmatic scales and logical structure." Linguistic Inquiry VI: 353-75. Fraser, Bruce. 1971. "An analysis of "even" in English." In Charles J. Fillmore and D.Terence Langendoen (eds) Studies in Linguistic Semantics: 150-78. Fujii, Seiko. 1995. "Additive and scalar operators in the context of concessive condi­ tionals." Paper delivered at ICLA 1995. Higashimori, Isao. 1994a. "Cognition, synonymy and definition." EURALEX 1994 Proceedings: 93-100.

EVEN, SAE/SURA/MO

59

Higashimori, Isao. 1994b. "A relevance-theoretic analysis of EVEN, SAE/SURA/MO/ TEMO/DEMO/DATTE/MADE." The English Literature Review 38: 51-80. Higashimori, Isao and Deirdre Wilson. 1996. "Questions on RELEVANCE." Uni­ versity College London Working Papers in Linguistics 8: 111-124. Horn, Larry R. 1989. A Natural History of Negation. Chicago: Chicago University Press. Kay, Paul. 1987. "EVEN". Berkeley Cognitive Science Report No.50. Karttunen, Lauri and Stanley Peters. 1979. "Conventional implicature." In Choon-Kyu Oh and David A. Dinneen (eds) Syntax and Semantics 11: 1-56. Kempson, Ruth. 1975. Presupposition and the Delimitation of Semantincs. Cambridge: Cambridge University Press. Kindaichi, Kyosuke et al. (eds). 1972. Nihonkokugo Daijiten (Shogakkan's Japanese Dictionary). Tokyo: Shogakkan. König, Ekkehard. 1991. The Meaning of Focus Particles. London: Routledge. Lanier, Alison Raymond. 1973. Living in the U.S.A. New York: Charles Scribner's Sons. Lavin, Eva D. to appear. "Even." English and Philosophy Department, University of Basque County. Lycan, William G. 1991. "Even and Even if." Linguistics and Philosophy 14: 115-150. Makino, Seiichi and Michio Tsutsui. 1986. A Dictionary of Basic Japanese Grammar. Tokyo: Japan Times. Mori, Sadashi. 1992. "A cognitive linguistic analysis of concessive utterances: 'even though' is not equal to 'even if." Fukui Technical College. Numata, Yoshiko. 1992. MO, DAKE, SAE nado-Toritate. (Particles such as MO, DAKE, SAE) Tokyo: Kiiroshioshuppan. Quirk, Randolph, Sydney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Shanon, Benny. 1978. "Even, only and almost hardly." Studies in Language 2: 35-70. Smith, Neil and Amahl Smith. 1988. "A relevance-theoretic account of conditionals." In Larry M. Hyman and Charles N. Li (eds) Language, Speech, and Mind. Lon­ don: Routledge 322-52. Sperber, Dan and Deirdre Wilson. 1986, 19952. Relevance: Communication and Cog­ nition. Oxford: Blackwell.

Semantic Content and Depth of Intention: A Study in Cognitive Semantics Richard Hirsch Linköping University, Sweden

1. Introduction Since the late 1970's and early 1980's there has been increased interest and in­ sight into the Anglo-American tradition of linguistics and psychology, into the use of language in utilizing, establishing, and maintaining mutual understanding and knowledge among interlocutors. In linguistics and philosophy the work of Grice (Grice 1989) and his successors working in Relevance Theory (Sperber & Wilson 1986) reveals how interlocutors utilize presumed shared knowledge to work out what is meant by what is being said, especially when what is being said is in need of some complementary interpretative work. In psychology, Clark (Clark 1992; 1996) has moved from the study of individual psycholinguistics to the study of the psycholinguistics of interpersonal communication, in which the notion of mutual knowledge or common ground plays a central role. His latest studies investigate and show the various linguistic methods, among a variety of non-linguistic non-verbal methods, that interlocutors use to establish and maintain the common ground necessary for achieving mutual understanding. Here I give a presentation of a Nordic-Scandinavian school of semantic analysis which is less well-known, or perhaps totally unknown, to most linguists and psychologists outside of Scandinavia. This method of semantic analysis was initiated in the late 1940's and early 1950's by the Norwegian philosopher Arne Naess, perhaps more well-known today for his engagement in ecological issues, and can be seen, I think, as a forerunner and complement to current develop­ ments in Anglo-American linguistic semantic analysis. The Naess tradition works with a number of notions that are vital for an explication of the some­ times intricate linguistic communicative work that interlocutors carry out in their attempts to achieve mutual understanding. This paper contains a presentation of the fundamental theoretical notions of the Naess semantic tradition and an illus­ tration of how semantic analysis within this tradition can be carried out on em­ pirical experimental linguistic data. In the Naess tradition of semantics, the point of semantic analysis is to gain insight into how people go about dealing with communication problems involv­ ing the achievement of cognitive understanding1 or the avoidance of cognitive misunderstanding, i.e. how people arrive at mutually confirmed agreement about what is meant by what is said even though parties may disagree as to whether what is meant is true or false, practical or impractical, etc. This way of

62

RICHARD HIRSCH

viewing semantics can be formulated in terms of a basic postulate which has an intrasubjective and an intersubjective version. Intrasubjective version: There is a difference in cognitive meaning for a person between two ex­ pressions/formulations if, and only if, it is the case that there is at least one situation in which the person would assent to, or affirm one expres­ sion/formulation but reject or deny the other. Intersubjective version: Every aspect of cognitive meaning of an expression/formulation is paral­ leled by a corresponding possibility of intersubjective misunderstanding and/or disagreement. In any act of communication there are a number of types or levels of un­ derstanding which can be identified in relation to what is said by the persons in the interaction. Figure 1 contains a rough typology of these different levels of understanding. Understanding in the sense of type A in Figure 1 is on the bor­ derline to being noncognitive. Acts of understanding of types B, C, and D in­ volve different kinds of more or less complicated and differentiated cognitive semantic analyses by the participants in communication. Under normal circum­ stances where interlocutors are speaking the same language, an act of under­ standing of type  usually entails an act of understanding of type  but not nec­ essarily the reverse. The same is true of the relation between acts of understand­ ing of types  and D, i.e. type D usually entails type  but not necessarily vice versa. Because there are different ways and levels of understanding what is be­ ing said in communication, there is much room for uncertainty concerning what is meant by what is being said. 1.1. Depth of intention Naess introduced the notion of depth (sometimes called definiteness) of intention in relation to expressions or formulations in order to give an analysis of the variation in cognitive understanding present among interlocutors in communica­ tion. Naess's notion of depth of intention is meant to capture the insight that speakers desire to be understood in a certain way and have a certain, more or less definite, cognitive content in mind when expressing themselves. More pre­ cisely, the notion of depth of intention can be characterized as follows: Depth/definiteness of intention: The number of cognitive distinctions that a speaker or hearer is aware of and prepared to take into consideration in a particular situation in connec­ tion with the use of a particular expression or formulation.

SEMANTIC CONTENT AND DEPTH OF INTENTION

63

Figure 1. Typology of levels of understanding If, for example, a speaker in a communication were to use the expression 'four-sided' when speaking of geometrical shapes, it may be uncertain whether the speaker is aware of many distinctions that could be made by a hearer. Figure 2 shows a number of cognitive distinctions that could be made in relation to this expression. This network of distinctions is something that a speaker and hearer may be mutually aware of and be relating to in the same way. However, it is probably equally likely the case that the speaker and hearer differ to a greater or lesser extent in their awareness of these distinctions in the actual on-going com­ munication. The speaker's attention may be focussed on shapes that fall on the left branch of the diagram, e.g. squares or rectangles, whereas the hearer is fo­ cussing on shapes somewhere on the right branch of the diagram, e.g. parallelo­ grams, rhombuses.

Figure 2. Types offour-sided figures

64

RICHARD HIRSCH

Naess's notion of depth of intention can be illustrated by referring to Figure 2 and saying that movement downward in the network of distinctions constitutes greater depth (or definiteness) of intention by a speaker/hearer in relation to the expression 'four-sided'. For a speaker and hearer to accomplish their project of achieving cognitive understanding and avoiding cognitive misunderstanding each of them must find out where the other is (if they are there at all) in the space of cognitive distinctions. 1.2. Agreement and disagreement In order to achieve cognitive understanding, speakers and hearers must work out some mutually confirmed agreement or disagreement concerning what is meant by what is said, i.e. what depth of intention is connected to the expressions and formulations they use. The mutually confirmed agreement or disagreement is referred to by Naess as real agreement or disagreement and can be defined as follows: Real Agreement: Two persons hold one and the same cognitive content of a formulation to be true in the same situation or same set of situations. Real Disagreement: For one and the same cognitive content of a formulation there is one per­ son for whom the cognitive content is true and another person for whom the cognitive content is false in the same situation or same set of situa­ tions. These definitions invoke the notion of truth (or falseness) as a criterion for agreement or disagreement. This notion of truth, however, is not restricted to the specialized sense used in truth-conditional semantics. Rather truth is used in a vague and intuitive sense and may incorporate many different theories of truth, e.g. coherence, correspondence, pragmatic. In their effort to achieve cognitive understanding, it is not certain that in­ terlocutors will always arrive at real agreement or real disagreement. There is always the danger that they will come to some pseudo-agreement or pseudodisagreement. Pseudo-agreement (verbal agreement but real disagreement): Two persons hold a formulation to be true and are unaware of the fact that for them the formulation expresses two different cognitive contents.

SEMANTIC CONTENT AND DEPTH OF ΙΝΤΕΝΉΟΝ

65

Pseudo-disagreement (verbal disagreement but real agreement): Two persons are unaware of the fact that for them two different formula­ tions actually express the same cognitive content and it is the case that one of the formulations is held to be true while the other formulation is held to be false. These situations can be described according to our typology of understand­ ing as a confusion between types  and  of Figure 1. In the case of pseudoagreement the speaker and hearer agree on what has been said and therefore assume that it means the same for both of them, whereas it is possible for both to have different or even incompatible understandings of what was meant by what was said. In the case of pseudo-disagreement we have the converse case where, because the speaker and hearer disagree as to an understanding of type  in relation to an expression or formulation, they assume that they also differ as to an understanding of type  They may, however, really be in agreement as to an understanding of type  without being aware of this or ever discovering it during the course of the communication. From a normative point of view, speakers and hearers should pursue a strategy in their striving for cognitive understanding that could be formulated as follows: Minimize pseudo-agreement and/or pseudo-disagreement and maximize real agreement and/or real disagreement. Both real agreement and real disagreement presuppose that the interlocutors have reached a cognitive understanding of what is meant by what is being said. That speakers and hearers, for various reasons, fail to live up to this maxim is a commonplace of everyday living. It is also what makes communication so diffi­ cult, risky, and in need of reflective analysis. Coming to focus on the same depth (or definiteness) of intention in relation to expressions or formulations used in a communication is obviously crucial to arriving at real agreement or real disagreement and thereby arriving at confirmed cognitive understandings. 1.3. Semantic uncertainty and situation-relative equivalence of cognitive content Given the risks for pseudo-agreement and pseudo-disagreement and misunder­ standings in connection with differing depths of intention in relation to expres­ sions and formulations, communication is a complicated and difficult business to do well in. Another of Naess's fundamental insights concerning the meaning of expressions and formulations used in communication does not make the situation any easier, rather the contrary. I often refer to this insight as Naess's semantic uncertainty principle. It comes in two versions, a strong one and a weak one, both of which lead to the same consequence. Naess's semantic uncertainty principle:

66

RICHARD HIRSCH

Weak version: It is not necessarily the case that a speaker has a definite determinate meaning (a particularly well determined definiteness of intention) in mind when speaking. Strong version: In many/most cases in everyday life, a speaker does not have a definite de­ terminate meaning in mind when speaking. Consequence: What a speaker has/had in mind (how well determined the speaker's definiteness of intention is) when speaking can, in certain (weak ver­ sion/most (strong version) cases, only be determined by a process of in­ terpretation. Naess's semantic uncertainty principle gives rise to an empirically testable research hypothesis. The cognitive content of expressions/formulations (an interlocutor's defi­ niteness of intention) varies according to situation of utterance and personal in­ terpretations of the expressions/formulations. This hypothesis calls for a very liberal nonprescriptive semantics and allows for a variety of idiosyncratic and nonoverlapping interpretations of expressions and formulations that are peculiar to the specific communication situation and circumstances of an interlocutor. Naess identifies six cases of situation-relative equivalence of cognitive con­ tent for expressions or formulations. Two expressions or formulations (U and T) can be cognitively equivalent for: 1. 2. 3. 4. 5. 6.

All persons in all situations In at least one situation for all persons For all persons in at least one situation For at least one person in all situations In all situations for at least one person For at least one person in at least one situation

For instance, for a person untrained in making geometrical distinctions and descriptions we might find the following expression U = 'a square block' to be interpretable in accordance with a type D understanding as equivalent with Τ = 'a cube'.

SEMANTIC CONTENT AND DEPTH OF INTENTION

67

This interpretation is, of course, not one that would be valid for all interlocutors in all situations. For some interlocutors (especially those schooled in making geometrical distinctions) the expression U might be pure nonsense in all con­ ceivable situations. 1.4. Terminology and definitions A Naess-inspired cognitive semantics works with a number of technical concepts that are defined and illustrated in this section. We begin with a definition of what is meant by an expression/formulation being an interpretation of another expression/formulation. An Interpretation: An expression/formulation Τ is an interpretation of an expres­ sion/formulation U if and only if there is at least one situation in which for at least one person, Τ and U have the same cognitive content (it is impos­ sible to imagine calling one expression or formulation true but the other one false in that particular situation). We next define what is meant by equivalence of cognitive content. Equivalence of cognitive content: A necessary but not sufficient condition for equivalence of cognitive content is that two expressions/formulations are not cognitively nonequivalent. This rather complicated formulation is aimed at avoiding the danger of calling formulations such as 'It's either a cube or it isn't' and 'Either the block is on the table or it isn't' equivalent in cognitive content simply because they are both trivially true (one cannot imagine calling one expression true but the other one false in any conceivable situation). The definition of equivalence of cognitive content invokes the notion of cognitive non-equivalence which is defined as follows: Cognitive non-equivalence: Two formulations are cognitively non-equivalent for at least one per­ son in at least one situation if and only if there is one imaginable set of circumstances for which the person(s) would claim that the one expres­ sion/formulation was true and the other formulation false. If we take, for example, the expression 'square block', we can well imag­ ine a block that has one or more square faces without necessarily being a cube.

68

RICHARD HIRSCH

It is conceivable that a person might be willing to call a rectangular block with two square faces a square block but unwilling to call it a 'cube'. The concept of equivalence of definiteness of cognitive content can be de­ fined by invoking the notion of interpretation as follows: Equivalence of definiteness of cognitive content: Two expressions/formulations (T and U) are equivalent in definite­ ness of cognitive content if and only if for at least one person in at least one situation all interpretations of Τ are interpretations of U and all inter­ pretations of U are interpretations of T. This means that, for instance, all cognitively equivalent interpretations of an expression such as 'cube', e.g. 'three dimensional figure with six square faces', 'three dimensional figure in which all sides are the same length and all angles are right angles', etc. are going to be interpretations of some other ex­ pression say 'die' for at least one person in at least one situation. Equivalence of depth/definiteness of intention is in turn defined in terms of equivalence of definiteness of cognitive content. Equivalence of depth/definiteness of intention: Two persons have the same depth/definiteness of intention in relation to an expression/formulation if and only if the expression/formulation is equivalent in definiteness of cognitive content for both persons, i.e. for these two persons all interpretations of Τ are interpretations of U and all interpretations of U are interpretations of Τ in at least one situation. This means that the two interlocutors maneuver in a similar fashion in a space of distinctions, as for instance in Figure 2 above. As long as they are aware of and share a conception of the 'looseness' or 'tightness' with which they are applying a term such as fyrkantig, 'four-sided', they will be using the term with the same depth/definiteness of intention. If one speaker is, however, em­ ploying the term in a loose sense, i.e. without an awareness that it applies nor­ mally only to two-dimensional figures, for instance in an expression such as fyrkantig kloss, 'four-sided/square block', and another speaker is using the term in a tight sense, i.e. with such an awareness, then the interlocutors do not have the same depth/definiteness of intention in relation to the expression/formu­ lation. These two speakers might have trouble reaching an agreement about what is meant by this expression, ending up in either pseudo-agreement or pseudodisagreement. Interlocutors may share a depth/definiteness of intention in relation to an expression/formulation which is restricted to the specific purpose of the com­ munication. This I call a sufficiently equivalent depth/definiteness of intention and define it as follows:

SEMANTIC CONTENT AND DEPTH OF INTENTION

69

Sufficient equivalence of depth/definiteness of intention: Two persons have sufficiently equivalent depth/definiteness of intention in relation to an expression/formulation if and only if they can use a set of interpretations of the expression/formulation to arrive at a mutually con­ firmed agreement as to the meaning of the expression/formulation (to solve at least one communication problem) in a least one situation. Interlocutors use expressions/formulations which are more or less determi­ nate in definiteness of intention. Those expressions/formulations that are more determinate in definiteness of intention are referred to as being more precise than certain other relevant expressions/formulations. Precision of expres­ sions/formulations is defined as follows: Precision: An expression/formulation Τ is more precise (more determinate in definiteness of intention) than an expression/formulation U if and only if all interpretations of Τ are interpretations of U and there is at least one interpretation of U which is not an interpretation of T. If we go back to our earlier examples and let Τ = 'a cube' and U = 'a square block' we can say that Τ is more precise than U because there are interpretations of U, for instance, Oblong rectangular block with a square face' which are not inter­ pretations of T, whereas all interpretations of Τ are also interpretations of U, i.e. all ways of describing cubes are going to be descriptions of some sorts of square blocks (blocks with square faces). Most of these research hypotheses are at odds with standard views of the cognitive semantic content of expressions and formulations, which is normally assumed to be stable over situations and interlocutors, even if, due to some sort of metaphoric extension, the expressions may be polysemous. The second part of the paper is devoted to an analysis of a short stretch of dialogue taken from a construction task experiment conducted at the Department of Linguistics at the University of Linköping. In this analysis we will see how our research hypothe­ sis (see section 1.3.) can be confirmed in the interpretive work that the partici­ pants carry out in the course of the dialogue.

2. The Experiment The experiment consisted of a two-party blind reconstruction task. The subjects were in the same room but placed on opposite sides of a table with a barrier in

70

RICHARD HIRSCH

the middle of the table which blocked mutual visual access. One of the subjects functioned as an instructor for the other participant, the instructee, whose job it was to attempt to rebuild á blocks' construction (the model) that the instructor had in front of him/her on his/her side of the table. The construction presented to the instructor resembled an aqueduct in which a small steel marble should roll first to the instructor's right along a trough, down a stairlike construction, and then to the left along another trough (see Figure 3).

Figure 3. Aqueduct as constructed during the experiment The blocks available on the instructee's side of the table were the same blocks contained in the instructor's model except for one cubic block which contained only two holes for the instructor, but three for the instructee. This meant that for the instructee the marble could take two different paths after en­ tering the block but only one path in the instructor's model. If this block was not turned so as to obstruct one of the exits, the marble could roll out of the block on to the table without passing through the rest of the construction. The participants in the experiment easily dealt with this complication by turning the block so as to obstruct one of the possible exits. All of the subjects, consisting of six pairs of participants, completed the task in about three to six minutes. The solutions are either exact reconstructions or mirror-image versions of the instruc­ tors' model. An excerpt from one of these six dialogues has been chosen here for in-depth annotated analysis to illustrate how a Naess-inspired semantics can deal with the participants' attempts to work out the depth of intention connected with expressions in one particular case.

SEMANTIC CONTENT AND DEPTH OF INTENTION

71

2.1. Annotated analysis of an excerpt The dialogue begins with the instructor  giving a short description of the func­ tion of the construction to be built by the instructee S (see note 2 for transcrip­ tion conventions). 1 :

jaha (....) det hör ska alisa (..) i slutändan bli som en. well this should then it ends up like a (.) öh (.) liten uh little (...) man ska alisa  rulla kulan uppifrán à . you can roll the marble from up to down

SI:

mm, jo, jagforstà're. ((fniss)) ((fniss)) mhm yeah i understand ((giggle)) ((giggle))

When S says "I understand" in SI it is not at all clear what type of under­ standing is being referred to in relation to our typology of understanding (see Figure 1). As we will see, it is pretty certain that S does not understand in the sense of type  in our typology.  continues with a typical 'bottom-up' descrip­ tion of the blocks' construction, i.e. describing the individual blocks that go into the construction, rather than continuing with the overall description that was attempted in CI. C2:

sá att, da ska vi se hör ((klossljud)) ((rullljud)) (.) so that well let's see ((block-knock)) ((marble-roll))  vi börjar med—du har en kloss ... if we start with you have a block

S2:

hur hög, hur hög kommer den va ungefär? how high how high is it going to be roughly

C3:

eh .. den kommer à börja mä den klossen som à .. uh it's going to start with the block that is eh tjock à avlâng. (.) me en ränna i= fat and long with a groove in it

S3:

=mm ganska kort ((klossljud)) mhm pretty short ((block-knock))

72

RICHARD HIRSCH

C4:

a-a = yeah

S4:

= ìforhàllande till= in comparison with

C5:

=den ä ungefär dubbelt sá lang som domfyrkantiga. it's about twice as long as the square ones

S5:

((hostning)) ((hostning)) okej (.) mm. ((cough)) ((cough)) okay mm

C6:

den ä lika lang som tvá som tvá sana härfyrkantìga. it's as long as two as two of these square ones alisa. — then

S6:

=á lika tjockja. and as fat yeah

In this stretch of the dialogue, C2 - S6, we find a number of expressions that are in need of determination in order for the instructee S to be able to carry out the task of identifying the block that is being referred to by the instructor C. The expression "the block that is fat and long with a groove in it" contains the expressions "fat and long" and "with a groove in it" which both need to be de­ termined more precisely (see Figure 4). the block that is fat and long with a groove in it

SEMANTIC CONTENT AND DEPTH OF INTENTION

73

Figure 5. Determination of "fat and long" in turns S3 & S6 C7 continues with a reformulation of the expression "with a groove in it" (see Figure 6). C7:

a-α, à de â en ba, en san hör rullbanapà den. yeah and there is a tra one of these tracks on it

S7:

mm. mhm

Figure 6. Determination of "with a groove in it" Assuming that S now knows what block  is talking about,  continues with a description of what S should do with the block referred to. C8:

den ställer du ρά högkant. (.) ((klosslud)) you stand it on end ((block-knock))

74

RICHARD HIRSCH

S8:

ja. yeah

C9:

rätt upp à  sadärja. (.) sen (.) ovanpà den (.) straight up and down like that and then on top of it

S9:

mm. mhm

C's expression "on end" in C8 is further determined in the first part of C9 "straight up and down like that". In the second part of C9,  initiates a compli­ cated description of a larger section of the construction of which this block is a part. CIO: vilandepà tvà (..) tvà sana härfyrkantiga klossar resting on two two of these square blocks S10: mm— mhm 11 : =ska en utav dorn här lànga (..) should one of these long Sil:

mm mhm

C12: eh .. (.) lànga, smala liggapá. uh long thin ones lie on  is trying to describe a structure that could possibly be called an arch or bridge. S obviously understands this formulation by  to be a description of a tower-like structure. S12: okej. det ska alisa va tvà klossar (.) ovanpà den har okay there should be two blocks on top of this först va? first right? But is corrected by C. C13: ΝΑΕ! NO!

SEMANTIC CONTENT AND DEPTH OF INTENTION

75

S then initiates a clarification sequence with a paraphrase of CIO. S13 : vitande pà tvâ fyrkantiga klossar sa du ? resting on two square blocks you said  contributes to the clarification sequence C14-C15 by paraphrasing C5 14: a du har alisa först den hör tje tje tjocka som är eh you have then first this f f fat one that is lika stor= as big S14:

=mm mhm

15: som tvâ smâ fyrkantiga. as two small square ones S continues S15-S16 with a paraphrase of C9. S15: den har jag stallt rakt upp à . that one I've stood upright C16: ja yeah S16: mm. mhm  follows C17-C18 with an elaboration on formulations C3, C8, and C10C12. 17: à (...)  ungefär va kan de röra sig  (.) tía and eh about what can it be ten centimeter if ran den inte rìktigt sa langt centimeters from it not really so far S17: mm mhm C18: ställer du tvâ (.) fyrkantiga klossar ovanpâ varan. you stand two square blocks on top of each other

76

RICHARD HIRSCH

This elaboration formulation by  (17-18) contains expressions that are themselves in need of determination (see Figure 7).

Figure 7. Determination of C17-CIS S18-C19 constitutes a determination of C's expression "two square blocks on top of each other" where it is established which of the square blocks (there are different kinds of square blocks) are being referred to in 18. S18: sana alisa tvá sty eken ma hall i the ones then with two holes in them 19: ((inandning)) en en ska vara hall i à en ska inte vara ((inhale)) one is with holes and another is without nat hall alls i any holes at all in it S19-C20 determine what definiteness of intention on C's part is to be re­ lated to the expression "on top of each other" in C18. S19: okejja har en utan hall ska den vara okay I have one without holes should it be underst eller överst underneath or on top C20: den utan hall ska ligga underst the one without holes should be underneath ((klossljud)) ((block-knock))

SEMANTIC CONTENT AND DEPTH OF INTENTION

77

S20-S23 constitute a determination of the expression "about ten centimeters from it (the upright block)" in C17. This expression is broken up into two parts that are dealt with more or less separately in the sequence (see Figure 8).

Figure 8. Determination of CI 7 S first focuses on the expression "from it (the upright block)" and suggests a series of possible determining interpretations (see Figure 9).

Figure 9. Determining interpretations of the expression "from it" in CI 7 Two interesting things occur in this sequence S20-S23. The first is the use, in S22, of a repetitive non-verbal block-knock with the upright block as a refer­ ring expression substituting for a verbal descriptive expression. The second is the choice of the interpretation "beside it" as a determination of the expression "from it". S20: ska de ska dem har klossarna sta (..) ((inandning))

78

RICHARD HIRSCH

alisa f ran mej eller mot mej det spelar ingen roll away from me or toward me does it make any eller= difference C21 : = nej dem kan du stalla= no you can put them S21 :

=den ska inte sta brevid utan dem ska staframför it should not stand beside but should stand in front eller bakom of or behind

C22: (....) ((fniss)) ((giggle)) S22: inte till höger e te vänster om mig utanframför eller not to the right or left of me but in front of or bakom den hör eh ((klossljudx 2)) (..) eh = behind this uh ((block-knock χ 2)) uh C23: -du kan stalla dem bredvid en liten bit ifrán. you can stand them beside it a little bit away S23: okej okay The use of objects as resources (letting objects in a sense 'speak for them­ selves') in communication is to my knowledge an unwritten chapter in cognitive linguistic semantics. C's expression "beside it" is less precise than a subphrase in one of the other alternatives "not to the right or left" proposed by S (the blocks should actually be placed to S's right from the upright block). It also appears to contradict one of the other alternatives "not stand beside". Despite these weaknesses it appears to do the job of relaying C's definiteness of inten­ tion in a manner appropriate to S's needs at this particular moment in the on­ going construction task. Both expressions in C23 "beside it" and "a little bit away" which function as a determination of "about ten centimeters" can be said to be some sort of vaguification operations (the opposite of making an expres­ sion more precise), but both seem to work fine in the communication situation.

SEMANTIC CONTENT AND DEPTH OF INTENTION

79

In answer to C's elaboration in C24 of 17 and 18, the excerpt under discus­ sion closes with a confirmed agreement in S24 as to how the completion of the arch or bridge construction described in 17 and 18 should be carried out. C24: for att ovanpä de hör ska de bli som ett tak (..) cause on top of this there is going to be like a roof med en sân hör läng with one of these long S24: okidàki okidoki C25: lang smal long thin ones 2.2. A network analysis of the excerpt Contributions by S and  following C's initial description of the blocks' con­ struction can be related to each other in terms of a network (more like a clinging vine really) of determination operations. A partial analysis of the contributions (up to and including contributions C17-C18) is presented in Figure 10 where each contribution is defined as an operation in relation to some previous contri­ bution or combination of contributions. The operations are presented in func­ tional notation where the operation is written in capital letters followed by the argument(s), i.e. the contribution(s) being determined. The relevant determination operations of precisification, specification, paraphrase, and elaboration are defined as follows: Where A and  are verbal expressions in a given context C;  is a precisification of A if and only if  is more clearly decidable in its application and non-application to any given entity or phenomenon within a domain than is A.  is a specification of A if and only if  denotes a class of entities or phe­ nomena that are included in the class of entities or phenomena denoted by A.  is a paraphrase of A if and only if  expresses the meaning of expres­ sion A on more or less the same level of specificity and/or preciseness as expression A, where  may be a repetition of A.

80

RICHARD HIRSCH

 is an elaboration of A if and only if  supports and develops to a greater degree of detail the information conveyed by A, where this information is not exhaustively characterizable in terms of specificity and/or preciseness.

Figure 10. Network of determination operations in expressions C1-C18

SEMANTIC CONTENT AND DEPTH OF INTENTION

81

3. Discussion and Conclusions The analysis of this excerpt from a construction task dialogue has hopefully made clear the empirical basis for the claims made in the first part of this paper concerning communication and the determination of depth/definiteness of inten­ tion which may have (often has) to be worked out in the course of a dialogue and may in fact constitute the major part of the body of the dialogue. This em­ pirical evidence supports Naess's semantic uncertainty principle and reveals the type of interactive situation-relative interpretative work that participants in dia­ logues (at least in the task-oriented dialogues of the type analysed above) carry out in their attempt to reach a confirmed real agreement as to whether they un­ derstand each other or not. What we see in the analysis above and in Figure 10, is how utterances and parts of utterances are (can be) interrelated in some sort of semantic refinement work that the interlocutors are carrying out in a step-by-step manner in the on­ going dialogue. This type of analysis is, I believe, entirely consonant with the style and spirit of analysis to be found in Clark's most recent work on the psycholinguistics of interpersonal communication (Clark 1996). What we gain in a Naess-inspired semantic analysis, which I miss in the Anglo-American tradition, is a set of conceptual tools and methods that allows us to give a detailed account of the, sometimes, minute semantic work that interlocutors can (often) become involved in in trying to reach a cognitive understanding. The type of semantic uncertainty and cognitive flexibility that is revealed in this study also calls into serious question any semantic theory, cognitive or oth­ erwise, that presupposes some sort of literal, i.e. transindividual and transsituational conservation of 'meaning' as a necessary prerequisite for the accomplish­ ment of understandings of types B, C, and D according to our rough typology. The expression fyrkantig kloss, 'square block', for instance, would appear from a normative literalistic semantic perspective to be at best strange and at worst anomalous. It is, however, an expression that is used repeatedly in this dialogue and many of the other construction task dialogues we have recorded. In this context and for these speakers it seems to be the appropriate expression for enabling the communicators to work out the cognitive distinctions that are rele­ vant for a completion of the construction task. Some readers might question the representativeness of this type of experi­ mental data for what speakers/hearers are doing in other, everyday settings. I would claim that there are many everyday settings, especially telephone conver­ sations with service institutions, such as hospitals, public utilities, etc. which call for the type of communicative competence exhibited by the participants in our experiment.

82

RICHARD HIRSCH

Acknowledgements I would like to thank Wolf-Andreas Liebert and two anonymous referees for their efforts, constructive criticism and suggestions in helping me get this paper in shape for publication. All weaknesses and errors that remain are entirely of my own doing.

Endnotes 1. Cognitive understanding is distinguished here from understanding in a wide undifferentiated sense to refer to an understanding that can be given a linguistic formulation. Understanding in the sense of being on the same wave-length or some sort of empathetic harmonic resonance that interlocutors might experience are therefore not included in our notion of cognitive un­ derstanding. This is, of course, not to say that such phenomena are of no interest to a full analysis of interpersonal linguistic communication. 2. Transcription conventions Symbol: , (comma) . (period) ? .. (two periods) ... (three periods) CAPITALS

= C) C) (...)

.) ((expression))

Meaning: continuing intonation falling intonation rising intonation exclamatory intonation intra-utterance extension prolonged intra-utterance extension emphasis expression in immediate continuation of other speaker's utterance, no pause very short but noticeable pause pause between 0.5-1.0 sec. pause between 1.0 - 2.0 sec. pause between 2.0-3.0 sec. non-verbal aspects

References Clark, Herbert H. 1992. Arenas of Language Use. Chicago: University of Chicago Press. Clark, Herbert H. 1996. Using Language. Cambridge: Cambridge University Press. Grice, Paul. 1989. Studies in the Way of Words. Cambridge, Mass.: Harvard Univer­ sity Press. Hirsch, Richard. 1989. Argumentation, Information, and Interaction. Gothenburg Monographs in Linguistics 7. Dept. of Linguistics. University of Göteborg. Hirsch, Richard. 1992. "Powerful Stuff - Discourse Processing: Social and Cognitive Aspects of Content Development in Conversation. Nordic Journal of Linguistics 15: 65-84.

SEMANTIC CONTENT AND DEPTH OF INTENTION

83

Naess, Arne. 1953. Interpretation and Preciseness: A Contribution to the Theory of Communication. Oslo: Norwegian Academy of Science. Naess, Arne. 1966. Communication and Argument: Elements of Applied Semantics. Oslo: Universitetsforlaget. Naess, Arne. 1989. En del elementaere logiske emner (Eleventh edition). Oslo: Uni­ versitetsforlaget. Sperber, Dan and Wilson, Deirdre. 1986. Relevance: Communication and Cognition. Oxford: Basil Blackwell.

Perspective, Subjectivity, and Modality from a Cognitive Linguistic Point of View* José Sanders and Wilbert Spooren Nijmegen University, Tilburg University, The Netherlands

1. Introduction The concepts of perspective and subjectivity are widely used with different meanings and in various contexts. In this paper, we discuss several approaches to perspective and subjectivity, using them as the building blocks of a cognitive linguistic model that covers both concepts. In their discourse, speakers and writers always indicate the point of view from which they are speaking or writing. Since people usually talk about events and states in which they themselves were participants, the point of view is usu­ ally that of the speaker (Chafe 1994: 132). The point of view can also shift from the speaker to another character. This means that events and situations are not necessarily represented from the same viewpoint all the time. In order to represent the semantics of a text, all the meanings expressed in that text have to be attributed to their personal or purported personal sources (Palacas 1993: 239). Because every utterance is attributed to some subject, that is, the current speaker, they are subjective in the most literal sense, that is, con­ nected to the point of view of a subject. Unfortunately, this use of the concept 'subjective' is not very informative. Therefore, we propose the following re­ striction: it only makes sense to call a report subjective if it is bound to a speaking or thinking subject of consciousness. In the following example, (1) is not subjective, whereas (2) and (3) are: (1) (2a) (2b) (3)

Jan is in Paris. Marie believes Jan is in Paris. I believe Jan is in Paris. Jan likes Paris.

In the case of example (1), there is no subjectivity. In examples (2a, b), by contrast, subjectivity is created because the information "Jan is in Paris" is ex­ plicitly bound to Marie/the speaker. Similarly, if a report carries a mental-state meaning with respect to a subject in the discourse other than the speaker, such as liking in (3), subjectivity is implicitly created by presupposing the presence of this character's active consciousness to which the mental state is attributed (see Sanders & Redeker 1996).

86

JOSÉ SANDERS AND WILBERT SPOOREN

In other words, point of view becomes subjective if a speaker connects it ex­ plicitly to himself or to another subject.1 Thus, subjectivity is brought about in two ways: as a connection to the speaker, which we call subjectification, or as a connection to concrete or abstract persons other than the speaker, which we call perspectivization. First, we shall discuss how the distinctions between perspectivization and subjectification can clarify the notions of perspective and subjec­ tivity as they are used in the literature. These notions will then be used in the analysis of deontic and epistemic modal verbs. It will be shown that the repre­ sentation of the different types of subjectivity in a framework of cognitive do­ mains is crucial to the analysis of deontic and epistemic modality.

2. Cognitive Representation of Subjectivity and Perspective in Discourse The idea of perspective or viewpoint has two connected and inseparable aspects: vantage point, that is, the viewpoint from which something is represented, and orientation, the resulting specific representation of objects once a particular vantage point is taken (Langacker 1987). Langacker describes vantage point as the cross of the axes of a visual field or a comparable coordinate system; orien­ tation is the alignment of construed objects with respect to these axes. The van­ tage point is a metaphorical specification of Langacker's general cognitive con­ cept of reference point: the conception of one entity for purposes of establishing mental contact with another (Langacker 1993: 1). In every utterance, except for certain elliptic utterances, a speaker /writer chooses one or more vantage points. A vantage point is the set of possible in­ stantiations of an T , a deictic center, an origo (see for instance Doron 1990: 80). Natural language has a wide variety of means to express vantage points. The speaker is relatively free to choose vantage points using various semantic and syntactic structures, such as converse predicates (buy - sell), ergative predi­ cates (rolling the ball - the ball is rolling), deictic distinctions (come - go) and anaphoric distinctions (Jan hit Marie - My neighbour hit Mane) (for an over­ view see Siewierska 1991; also DeLancey 1982; Fillmore 1982; Bolkestein & Risselada 1987; Kuno 1987; Zribi-Hertz 1989). In all cases, the choice of a particular vantage point expresses the speaker's empathy with one person/thing rather than with other people/things, like a camera that is placed at this per­ son/thing's viewpoint while looking at the other people/things (Kuno 1987). For instance, by saying Jan is going to Paris, the speaker expresses that Jan is moving from a point close to the speaker to Paris; thus, the speaker's empathy is positioned with Jan close to the starting point of the movement. The actual location and time of the speech act may have nothing to do with the vantage point that is expressed by an utterance; in this sense, the position of vantage points is neutral. The actual location and time of the speech act is a

PERSPECTIVE, SUBJECTTVTTY, AND MODALITY

87

special, non-neutral case of vantage point that is called the referential center (R). R is the vantage point of the current speaker, possibly realized as T . The '(current) speaker' is a conscious entity who can refer to himself as T , whose tense is present, and in whose reality the information expressed is factual by default. R is indicated by explicit signals that anchor the utterance, such as the choice of pronoun and verb tense. First and second person pronouns and the present tense explicitly indicate the referential center T . If those signs are ab­ sent, the Τ is the implicit subject whose verb tense might have been present, whose time and place are the 'here' and 'now' of the speech moment, and in whose discourse world events and objects are presented as factual. Another non-neutral vantage point we want to distinguish is the subject of consciousness (S). S is the subject, either the speaker or the character in the dis­ course, to whom the responsibility for the information is attributed; in other words, the subject is responsible for the propositional content of the utterance and possibly also for its form. S is often, but not necessarily always, established in the current speaker. The default positions of R, S, and other vantage points are one and the same position, namely, the current speaker, as in example (4): (4) I am going to Paris. R and S are located in /, which refers to the current speaker and subject of consciousness of the sentence; the verb going also indicates that the direction of the movement is from the speaker toward another entity and not the other way around. It is not necessarily the case that all signals for vantage points point towards one and the same position. If R and S are not explicitly instantiated as T , they are located with the implicit current speaker by default, while vantage point is, by default, located in the subject position (e.g. Kuno 1987). This is exemplified in (5): (5) Jan is going to Paris. In (5), R and S are located in the current speaker, whereas the vantage point is located with Jan, which is indicated by the subject position of Jan and by the verb going which expresses a movement from Jan's position to Paris. If S is moved from the speaker to another subject, linguistic elements in the discourse will indicate the presence of active consciousness in a subject. Exam­ ples of such indicators are verbs of cognition and perception, as exemplified in (6): (6) Jan wanted to go to Paris.

88

JOSÉ SANDERS AND WILBERT SPOOREN

In (6), R is with the current speaker, while S and vantage point are with the subject of the sentence. The verb wanted attributes volitionality to Jan, which establishes him as the S of the sentence. An Τ in an utterance does not imply that all vantage points are located with the referent of this 7, that is, the current speaker: the vantage point can be located outside the I, as exemplified in (7): (7) I am coming to Paris. In (7), R and S are located with the current speaker of the utterance, whereas the vantage point is located in Paris by the use of coming instead of going. The use of coming indicates that the current speaker has located a subject of consciousness in Paris, either in a person he is addressing with this utterance, for instance, in a letter or telephone conversation, or in a personified entity - the city itself - as in L.Á., here I come!. Likewise, an utterance such as (8) is best understood in a context in which the vantage point is firmly established in a given addressee, in this case, his mother: (8) Jan told his mother he was coming to her. 2.1. Perspectivization In this section we will try to relate the conceptions of perspective discussed above with the conception of subjectivity in discourse studies and narrative the­ ory. Several authors have dealt with the question whether choosing a particular vantage point means that the utterance will necessarily be subjective as well. Wiebe (1990, 1994) suggests this is not so. She proposes that sentences be regarded as subjective when they express the psychological point of view (Uspensky 1973) of a discourse character, such as examples (2) and (3), in con­ trast to sentences that objectively narrate events or describe the narrative (discourse) world, such as example (1). Note that spatio-temporal vantage points are chosen all the time in narrating events and describing narrative worlds. Con­ versely, a psychological point of view is taken solely by presenting a discourse character's private state: the states of a character holding a propositional atti­ tude, optionally towards an object. Such a private state can take two forms: it can be represented, in any form of representation of thought, beliefs, knowledge states, utterances, and so forth, as given in example (2), or it can be merely reported, as illustrated in example (3). Using quite different terms, Reinhart (1983: 170) distinguished similar phenomena within the concept of point of view: "To say of an expression E that it is from a certain person P's point of view is to say that E represents P's judgments (wishes, etc.) or that Ρ is responsible for E." The judgment element can be seen as Withe's private state report type, that is, a person's judgment or

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

89

attitude about an object (see examples 3 and 6); the responsibility part can be seen as Wiebe's represented thought type, that is, a piece of information that must be attributed to a person as his or her thought (see example 2). The two aspects of psychological point of view were linked to the concept of vantage point by Sells (1987) in his model of logophoricity. In a Discourse Representation Theory framework, he introduced the notion source, self, and point of view (pivot) as the fundamental aspects of logophoricity. Source refers to the intentional agent of the communication, for example, the speaker, and self refers to the individual whose mental state or attitude is described by the content of the proposition. What we called vantage point can be seen as Sells' category point of view/pivot, that is, the person/thing from whose point of view a report is being made. Source can be regarded as the person whose thought (Wiebe) is represented, and who is thus responsible (Reinhart), as is illustrated in example (8).2 Self cm be regarded as the person whose private state is reported (Wiebe) or whose judgment is revealed (Reinhart) as is illustrated in examples (2) and (3). Source/represented discourse and self/private state are special cases of van­ tage point which presuppose a conscious subject as the deictic center. We will aggregate these cases using the concept that we have defined as subject of con­ sciousness (S): a propositional content is attributed to some discourse character in source as well as in self; source and self differ only in the nature of the sub­ jective information, as indicated by their specific predicates. The positioning of an S other than the speaker connects the meaning of some information to this other subject. This connection results in what we shall call perspective. Perspective is established by various linguistic means. The most implicit manner is the representation of a person as an active subject of con­ sciousness without representing his inner or spoken discourse, as in (3). The most explicit type of perspective is direct quotation, in which a current speaker lends not only his 5, but even his R to another subject in the discourse, thus creating a new 'I' as the embedded current speaker: (9) Jan said: "At this moment a bear is coming towards my kitchen." In (9), my depends on an Τ which does not refer to the current speaker but to Jan. In general, whenever world-creating predicates such as verbs of utter­ ance (tell, say, etc.) and cognition (think, believe) attribute speech, thoughts, beliefs, perceptions, and so forth, to a subject in the discourse, perspective is created. The position of R and S and the vantage point(s) affect how directly the embedded speaker's utterances or thoughts are represented (Sanders & Redeker 1996). For instance, R may be positioned with the current speaker, while S and other vantage points are positioned with a person in the discourse as in (10), an example of indirect speech quotation: (10)

Jan said that at that moment a bear was coming towards his kitchen.

90

JOSÉ SANDERS AND WILBERT SPOOREN

In the subordinated sentence in (10), S is with Jan, to whom the represented discourse is attributed in the subordinated sentence by said. Likewise, the van­ tage point is positioned with Jan by coming. By contrast, R is positioned with the current speaker of the sentence as a whole, hence his instead of my. At this point, it is important to repeat that one utterance can have several vantage points instantiated at different positions. In (10), there are two vantage points, one with the current speaker as signalled by the deictics in 'at that moment' and the past tense, and one with Jan, signalled by coming. This situation is typical of indirect representation of speech and thought. It is also possible for R to be positioned with the current speaker, while all of the deictics in the embedded utterance point towards a vantage point that is positioned with another subject, as in (11): (11)

Jan was afraid. At this moment, a bear was coming towards his kitchen.

The situation in (11) is typical of free indirect representation of speech and thought: Deictic expressions (at this moment, coming) establish the vantage point very strongly in the subject of consciousness' position, whereas R remains located at the (implicit) current speaker's position, as is signalled by tense. What examples (9-11) have in common is that the content of Jan's utter­ ances/thoughts is valid for Jan, but not necessarily for the current speaker of the whole utterance. In (9), for instance, the current speaker reports that S is with Jan who says that there was a bear in his kitchen; this may mean that the current speaker shares Jan's report and regards it as true in his/her discourse reality, but this need not be the case. Generally speaking, a perspective expresses that the claim for validity with respect to something is restricted to somebody in some way. Thus, the conception of perspective does not, in itself, imply the truth or non-truth of the predicated information, but embeds it in the discourse reality by ascribing it to a subject that is responsible for its meaning. This embedding may have consequences for the validity of the information in the discourse reality. By default, information that is predicated as embedded in a perspective is taken to be true in the discourse reality; however, some perspective predicates indicate that, likewise by default, the perspectivized information should not be taken to be true in the discourse reality. Generally, what was presented from a perspec­ tive can be corrected afterwards by the current speaker. Thus, perspective has the characteristics of non-monotonic reasoning. Mayer (1990: 125) gives the following formal definition of perspectivization in truth-conditional terms: 'Let Τ = S(l) ... + S(n) be a text. Then Τ (...) perspectivizespif there is a text fragment T*(j) of Τ such that T*(j) (assumed to be the only information avail­ able) implies ρ but Τ does not.'

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

91

For instance, the information attributed to a subject as a belief can be followed by a denial of the embedded information, as in (12): (12)

Jan says there is a bear in the kitchen, but, in truth, it is Christopher Robin.

What is thought to be true by Jan may or may not be believed by the speaker. Table 1 presents an overview of how the concepts we have used are related to those used in the literature to characterize perspective. 2.2. Subjectification In the previous section we argued that, in principle, every utterance is subjective in the most literal sense, and we proposed that the concept of perspective be restricted to subjects other than the speaker. Thus we seem to have lost the use of the concept subjective altogether as it is used in daily life: frequently, 'subjective' is used for utterances or world views that are biased in one way or another. Subjectivity will not be dealt with in its ideological sense in this arti­ cle,3 but we do wish to elaborate on the expression of subjectivity by the subject of the discourse, the current speaker. Subjectivity is expressed when the current speaker displays an attitude towards the predicated information, as in (13): (13)

Surely, Jan is in Paris.

Rather than simply stating something, the speaker of (13) adds his degree of certainty to his statement. In general, expressions such as surely in (13) add a dimension to the discourse at the level of the speaker's consciousness.4 In this case, the speaker verbalizes the degree of certainty he feels with respect to his statement, but he could also add his wish that what he expresses is true or his evaluation of its necessity. The speaker's consciousness can be established by various linguistic means, such as modality (Jan must be in Pans), subjective Iembedding (/ think that Jan is in Paris), predictions {Jan will stay in Pans), conditionals {If Marie is well informed, Jan is staying in Paris), and evaluative reflections {Jan is staying in Paris, unfortunately). By the use of such expressive predicates, the speaker foregrounds himself to some extent. Such speaker-foregrounding can be seen as subjectification: the speaker himself is objectified, in the sense that he becomes part of the discourse object, that is, the utterance, while the discourse object is subjectified because of the speaker's subjective presentation (Langacker 1990). Subjectification does not have the particular characteristic of non-monotonic reasoning that perspectivization has—the possibility of immediate correction.

92

JOSÉ SANDERS AND WILBERT SPOOREN

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

93

For instance, the proposition "Jan is in Paris" expressed by (13) cannot be denied afterwards as is shown in (13'): (13') ?? Surely, Jan is in Paris, but that is not true. This utterance can only be denied in a context that indicates that the proposition is perspectivized, as shown in (14): (14)

Surely, Jan is in Paris. Ha, surely! Marie was always sure of her statements, but she was seldom right. This time was no exception: Jan was found to be in Prague.

In (14), the utterance "Surely, Jan is in Paris" must be attributed to Marie, who is responsible for its meaning. Therefore, it is possible for the current speaker of (14) to deny the truth of this utterance. The fact that subjectified utterances in contrast to perspectivized utterances are not immediately correctable, does not imply that they do not express subjec­ tivity. Subjectification expresses subjectivity in its own vitally different manner: it binds the meaning of the predicated information to the current speaker. Thus, it is indicated that the current speaker foregrounds his opinion with respect to his statement. For instance, surely in (14) expresses Marie's commitment to her statement "Jan is in Paris." Thus, a second linguistic manifestation of subjectiv­ ity can be distinguished: in addition to perspectivization, that is, the binding to a subject other than the speaker, there is subjectification, which is the explicit binding to the subject (i.e., the current speaker). This type of binding can be represented as the speaker's conceptual division of himself into a representing and a represented part, and the connection of the responsibility for the meaning of his statement's content to his represented self.6 The conceptualization of such a connection is quite complicated in comparison to the connection of meaning to another speaker in perspectivization. Therefore, subjectification can be regarded as more specific and more 'sophisticated' than perspectivization.7 2.3. Cognitive representation of perspective and subjectification in terms of dis­ course domains Both perspectivization and subjectification can be described in terms of dis­ course domains, or embedded subspaces, which entail a restricted claim of the validity or factuality of the embedded material (Fauconnier 1985; Dinsmore 1991). A domain is always set up as subordinate to a 'parent'; the outermost parent is that of the speaker's reality and is called the basic domain or base (B). In the case of narrative discourse, the base is the narrator's reality. Linguistic expressions that establish a subdomain (M) or refer back to one are domain indi­ cators (Fauconnier 1985: 'space builders'). They mark the meaning of the in-

94

JOSÉ SANDERS AND WILBERT SPOOREN

formation in the embedded domain as restricted. Depending on the type of do­ main indicator, the material is restricted to a particular temporal domain (In 1929), a spatial domain (In France), a hypothetical domain (If only ...), a pos­ sibility domain (Maybe ...), or to a particular perspective (Jan believes ...) (see Spooren 1989, Spooren & Jaspers 1990). The difference between perspectivization and subjectification in the ways in which they create subjectivity can be modeled as follows. Both create a subjectbound domain in which the predicated information is represented. Perspective is about the attribution of elements to an ' Γ other than the speaker in the discourse representation, for instance, by belief predicates such as in (2a) Marie believes Jan is in Paris, whereas subjectification is about the commitment of T , the cur­ rent speaker, to the value or validity of the information stated, for instance, evaluative predicates such as in (12) Surely/unfortunately, Jan is in Paris.* The concepts we have used thus far can be summarized in a single frame­ work: A speaker may conceptualize himself subjectively, depending on how much he presents himself on stage by expressing himself. This degree of sub­ jectification is represented by the position along the vertical axis in Figure 1: the higher the position chosen, the more subjectification is expressed (this will be explored more fully in section 3). Within the conceptualized segment, one or more neutral vantage point choices may be expressed. This is represented by the position(s) on the horizontal axis.

Figure 1. Conceptualization of vantage points

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

95

This process of conceptualization is recursive.9 Any speaker may represent an embedded S (subject of consciousness, Figure 2) or even give the R (referential center) to another subject. Thus, a perspective of another speaking, thinking, or perceiving person in the discourse is created. conceptualized

current speaker (R)

Figure 2. Conceptualization of embedded speaker The perspective of this embedded speaker may be created more subjectively or less subjectively, depending on the degree of subjectification of the embedded speaker. For instance, when a current speaker gives the R to an embedded quoted speaker, a new discourse is created in the form of a direct quote. The objectification of the embedded speaker subjectifies his/her quoted discourse, which results in a highly subjective type of perspective, because it places the embedded speaker's discourse entirely in the foreground, while the current, quoting speaker is completely backgrounded. In Sanders & Redeker (1996), arguments are given on the specifics of distinguishing various types of discourse representation in terms of the degree of the quoting speaker's and the quoted speaker's subjectivity. In the next section, we will investigate how the concepts of perspectivization and subjectification and their cognitive representations can help understand a complex type of subjectivity, namely modality.

96

JOSÉ SANDERS AND WTLBERT SPOOREN

3· Modality and Subjectivity Evidentials are a specific type of modal expression that indicate the amount and source of evidence a speaker has for a particular statement, as in example (15): (15)

Supposedly, Jan is in Paris.

As with the epistemic modal expression surely in example (13), supposedly expresses that the current speaker is committing himself to some extent to the truth of the statement "Jan is in Paris," but adds a dimension to this commit­ ment: the current speaker is using evidence from sources outside himself as the basis of his commitment. Evidential predicates such as supposedly combine per­ spectivization and subjectification. Firstly, they refer to a subject of conscious­ ness other than the present speaker, to whom the responsibility for the utterance must be attributed (perspectivization). Secondly, they express some degree of certainty with respect to the utterance in the current speaker (subjectification). The observation that evidentials are subjective is not new. Lyons (1977), for instance, states this explicitly. Generally, all expressions of modality such as modal verbs cause subjectivity in that they express a subject's (speaker's) opin­ ion or attitude towards the proposition that a sentence expresses or the situation that the proposition describes (Lyons 1977: 452; see also Mayer 1990). The relationship between evidentiality and perspective is less frequently noticed, but it can be inferred from the fact that several analyses of evidentiality make use of categories that reflect the relationship with perspectivization, like the category of source indication, which subsumes quotative conditional and hearsay (Givón 1982, Lyons 1982, Chafe 1986, Willett 1988, Mayer 1990).10 This implies that such evidentials are perspective indicators. Below, we explore in what sense subjectivity is characteristic of epistemic modality in general, as well as of deontic modality. In particular, the relation­ ship between deontic and epistemic modals and the concepts of perspectivization and subjectification are discussed. 3.1. Perspectivization and subjectification in deontic and epistemic modality The difference between epistemic modality and deontic or 'root' modality, the two major types of modality, has been the subject of numerous studies (see Ly­ ons 1977, Coates 1983, Palmer 1986, and various contributions in Bybee & Fleischman 1995). Sweetser (1990) discusses the relation of root to epistemic senses in modals, especially those in English. Here Dutch modals are used with English translations. Compare, for instance:

PERSPECTIVE, SUBJECTTVITY, AND MODALITY

(16)

Jan moet thuisblijven. Jan must home stay 4 Jan must stay at home. '

(17)

Jan moet thuisgebleven zijn. Jan must home stayed be 'Jan must have stayed at home. '

97

Sentence (16) is an example of root modality. Root modals express obliga­ tion, permission, or ability by an external socio-physical force ('real-world sense,' Sweetser 1990: 59-61). Thus, root modality includes the classical deontic modality, that is, the expression of moral obligation as in the meaning per­ mit/oblige, as well as the expression of physical necessity or ability (Sweetser 1990: 49 and 152). For instance, moet ('must') in (16) modifies the factual proposition (P) "Jan stays at home" by the extension that Jan has to/is forced to (deontic) achieve P. 11 The modal is used in its real-world sense, which can be analyzed as "the direct force of the person with authority compels you [Jan] to ... [stay at home]" (Sweetser 1990: 61). Sentence (17) is an example of epistemic modality. In its epistemic mean­ ing, the abstract force expressed by the modal verb is not manifested in the so­ cio-physical sphere of objective reality but metaphorically, in the epistemic do­ main of the speaker's reasoning (Sweetser 1990: 59). Epistemic modal predi­ cates indicate that the speaker is not stating a fact but is adding the degree of his commitment to what he is stating (Sweetser 1990, Palmer 1986). Moet ('must') in (17) indicates that the speaker, reasoning on the basis of the circumstances, is strongly committed to the factuality of proposition P, "Jan has stayed at home."12 Compared to the epistemic modal in (17), sentence (16), which contains the deontic modal, is an objective, factual statement, because the socio-physical force expressed in (16) must be represented in the social reality. This is shown by the fact that in Dutch, the socio-physical force can be presented explicitly, as is shown in example (18). Another indication for the presence of the force in the discourse's social reality is that it can be represented as active in the past with­ out a change of perspective. Example (19) shows that representation in the past does not change the meaning of the modal, but merely transposes it relative to the current speaker's temporal reference point, his here-and-now. With respect to the translation of the last example, it should be noted that English must, like may, does not have a past tense form (Palmer 1986: 60); must is replaced by had to (Coates 1983: 40 and 57): (18)

Jan moet van Klaas thuisblijven. Jan must by order of Klaas home stay 4 Jan must [by order of Klaas] stay at home. '

98

JOSÉ SANDERS AND WILBERT SPOOREN

(19)

Jan moest thuisblijven. Jan had to home stay 'Jan had to stay at home.'

Sentence (18) explicitly expresses the external force compelling Jan to stay at home, i.e., Klaas. Sentence (19) expresses the fact that there was a force which made Jan stay at home; this force may be either personal (some authority other than Jan himself), or impersonal (circumstances that made Jan realize that he could do nothing but ...). Likewise, the force expressed by volitive modals such as want can be transformed to the past as in Jan wanted to stay at home. The use of wanted indicates that Jan's desire to stay at home was active in the past and is no longer active in the current speaker's present. In a non-modalized utterance like (20), Ρ is simply presented as factual: in terms of the cognitive linguistic framework, Ρ is valid in the speaker's reality (the basic domain B). As a consequence, immediate correction of Ρ as in (20') is not possible: (20)

Jan stayed at home.

(20') ?Jan stayed at home, but he did not stay at home. The explanation is that the factuality of the proposition cannot be denied within the same domain: Ρ and not-P cannot both be true in one domain (Strategic Principle 1: "Avoid contradiction within a space," Fauconnier 1985: 86). Contrary to non-modified utterances, deontic modals allow for the modified proposition to be immediately corrected, because the modal predicate opens a subdomain (M) within the truth-domain or basic reality (B) of the discourse. What is represented in the embedded domain can be denied in the basic reality, as is illustrated in (16'): (16') Jan moet thuisblijven, maar hij blijft niet thuis. Jan must home stay, but he stays not home 'Jan must / ought to stay at home, but he does not stay at home.' The fact that Jan is or was obliged to do something (either by the speaker or by some other authority) does not entail that he will do it in future or has done it. In Dutch, as is shown in (16'), the proposition in the scope of moeten can therefore be denied.13 In the epistemic use of moet ('must'), as in (17'), how­ ever, immediate correction of Ρ is much less likely than in the deontic use (see note 20): (17') ? Jan moet thuisgebleven zijn, maar hij is niet thuisgebleven. ? 'Jan must have stayed at home, but he did not stay at home.'

PERSPECTIVE, SUBJECTTVITY, AND MODALITY

99

3.2. Subjectivity in deontic modals The possibility of immediate correction shows that the deontic modals can be regarded as perspective domain indicators: they create domains in which infor­ mation is bound to a subject in the text, and this information can be denied in the parent domain.14 The epistemic modals obviously do not create similar do­ mains. This implies that the deontic and epistemic modals express different types of subjectivity in discourse. A starting point for this line of thought is of­ fered by Lyons' (1977) distinction between objective and subjective modality. Lyons (1977) and, subsequently, Coates (1983), and Siewierska (1991) dis­ tinguished between objective and subjective interpretations of modality inde­ pendent of the distinction between deontic and epistemic modality. Lyons linked the objectivity /subjectivity distinction to /activity. Subjectively modalized state­ ments are non-factives in the sense that they are "(···) statements of opinion, or hearsay, or tentative inference, rather than statements of fact" (Lyons 1977: 799). Objectively modalized utterances are factual statements or 'acts of telling' that "contain an unqualified or categorical I-say-so component. The speaker is committed by the utterance of an objectively modalized utterance to the factuality of the information that he is giving to the addressee" (Lyons 1977: 799). Following this analysis, a sentence like (21) will sustain both objective and sub­ jective interpretations, provided the context is appropriate. (21)

Jan kan thuis blijven. Jan can home stay 'Jan can / may stay at home.'

Construed objectively, (21) would be held to be a factual assertion: its mo­ dality, whether deontic or epistemic, is part of the true or false proposition that is being asserted. In the subjective interpretation, (21) is either a directive (deontic) or a statement of opinion (epistemic) (Lyons 1982: 109-110). In other words, subjectively used modals foreground the speaker by expressing his authoritative or reasoning force. However, an objectively used deontic modal can still express subjectivity in the sense that its socio-physical force can presuppose that a subject is repre­ sented as active consciousness. Note that this subject is not the speaker, as in subjective deontic modals. Thus, these deontic modals implicitly open a subjec­ tive, that is, a person-bound subdomain, in the discourse representation and thus represent perspectivization. The modal creates a mental subdomain M embedded within the speaker's base B, in which the proposition Ρ is placed. In order to illustrate this, an example of a deontic modal in its objective reading is analyzed more closely below. The statement that "Jan can stay at home"(21) in its reading "Jan is al­ lowed to stay at home" attributes an opinion or evaluation to some higher

100

JOSÉ SANDERS AND WILBERT SPOOREN

authority (A), who is represented as an active consciousness. In other words, the domain indicator, that is, the deontic modal, is connected to (A). This is shown in Figure 3. Since this authority is not instantiated in the actual dis­ course, A is created implicitly and represented in the implicit, that is, nonencoded area of the base B.

Figure 3. Representation of deontically modified utterance "John can stay at home" (objective use) Likewise, in (16), deontic moet ('must'), in its objective use, constructs an embedded perspective domain that is connected to an authority. One might ask in what sense an objectively used deontic is subjective. The interpretation in which Jan's own judgment of the circumstances allows or obliges him to stay at home reaches beyond a strict interpretation of the deontic modal as moral permission/obligation, that is, actually permitted/obligated by a moral law or moral conscious entity. Yet, in our opinion, the categorization as a deontic modal is justifiable: when a deontic modal indicates someone's evalua­ tion of circumstances, i.e., that something is permitted or forced by circum­ stances, in the representation this evaluation must be attributed to this person's 'higher self (consciousness or moral judgment) which permits or obliges the 'acting self to do something. In this case, the self is the modal verb's subject.17 In sum: compared to epistemic modals, deontic modals are objective (see, for instance, Heine, 1995); compared to non-modalized utterances, deontic modals are subjective. With respect to what Lyons calls subjective uses of deontic modals, it is important to note that they have a performative function because, in these cases, the authoritative force is the foregrounded speaker him/herself. In their subjec­ tive use, these modals create directives, as in (22), and commissives, as in (23):

PERSPECTIVE, SUBJECTTVTTY, AND MODALITY

(16)

Jan moet thuisblijven. Jan must home stay 'Jan must stay at home.'

(22)

Jij moet thuisblijven. You must home stay 'You must stay at home.'

(23)

Jij kunt thuisblijven. You can home stay 'You can stay at home.'

101

In the subjective use of (22) and (23), the socio-physical force is the speaker himself, who is thus foregrounded. In (22), must is interpreted as /, speaker, as authority, oblige you to do so. The speaker opens an embedded do­ main that is bound to himself, which is represented as the connection of M to / in the implicit area of B. Thus, the subdomain represents subjectification: the subdomain is bound to the subject of the discourse, that is, the (implicit) speaker.

Figure 4. Representation of deontically modified utterance uJohn must stay at home" (subjective use) According to Coates, however, subjective use of deontic modals is infre­ quent in both spoken and written discourse (Coates 1983: 21). Generally, deon­ tic modals are used in their objective meaning. A subjective interpretation of the deontic modal in (16) is, of course, possible. In that case, it is a performative utterance in which the current speaker foregrounds his authority, for instance, as the response to the question Can Jan go out now? However, this subjective in­ terpretation is not made by default.

102

JOSE SANDERS AND WILBERT SPOOREN

In conclusion, deontic modal verbs can be used in various interpretations that range from subjective to objective, depending on the subject to whom the claim of validity of the information that is modified by the deontic modal is attributed: to the current speaker Τ (subjectification) or to another subject (perspectivization), who may be the implicit or an explicit authority or even, in certain cases, the person who is the subject of the deontic modal. 3.3. Subjectivity in epistemic modals According to Lyons, all three epistemic modals {must, may and can) can express both subjective and objective modality (Lyons 1977: 801). In the objective in­ terpretation, the necessity or possibility itself is presented ('told') as a fact which can be denied, questioned, or accepted as a fact, and which can be re­ ferred to by the complement of a factive predicator or a conditional predicator (Lyons 1977: 799-805). Thus, in (17b), must is only objectively interpreted in a dialogue as a response to a statement such as the one in (17a): (17a) Speaker A: Jan must have stayed at home. (17b) Speaker B: Yes, it has happened before that Jan must have stayed at home. If a speaker has access to knowledge of an objective possibility or necessity of some state of affairs, he may use an epistemic modifier in the objective inter­ pretation (Siewierska 1991: 127 and others). The objective use of epistemic mo­ dality occurs only peripherally in actual language use, for instance, when chance can be quantified (a craps game), or 'descriptively', that is, when a modal qualification is reported that is not the speaker's own (Nuyts 1993: 951), while subjective use of epistemic modals is considered the 'core' (Coates 1983: 245) or 'basic' (Lyons 1977: 805) interpretation. Consequently, the objective, factive interpretation of epistemic modality is non-default; in fact, we believe that the objective interpretation is established in context precisely by marking the modified sentence as a fact and by using em­ bedding clauses, as is done in (17b). In epistemic modality, subjectivity is cre­ ated in an essentially different manner than in deontic modality. Epistemic mo­ dality expresses the current speaker's commitment to the validity of the proposi­ tion on the basis of his estimation of the chance that the state of affairs ex­ pressed in the proposition is true. This is no less true in the case in which the speaker has objective and verifiable information than when he is merely guess­ ing (see Nuyts 1992: 304). Thus, epistemic modification necessarily represents subjectification: it reflects the active consciousness of the discourse's subject. Note that the discourse's subject refers to the current speaker, who may be ei­ ther the speaker of the discourse as a whole or a speaking person within the text.

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

103

The subjective (and performative, see Palmer 1986: 60; Nuyts 1993: 951) na­ ture of epistemic modals is emphasized by the fact that, unlike deontic modals (see previous section), they cannot be transposed to the past tense without also changing the perspective. This can be seen by comparing (17) to (24): (24)

Jan moest thuisgebleven zijn. Jan had to home stayed be 'Jan had to have stayed at home.'

In (17), moet... zijn ('must have') indicates a present commitment to a past state of affairs. In other words, must have indicates the judgment of a current speaker, in the here-and-now, who is not Jan, about a proposition in the past. By contrast, the past tense moest ('had to') in (24) suggests the presence of a person other than the current here-and-now speaker, and other than Jan, whose spoken or mental discourse is represented in the free indirect mode. The force existing in the speaker's domain of reasoning can only be expressed in the dis­ course's here-and-now reality at the moment of speaking. Therefore, transfor­ mation to past tense implies perspectivization, a shift to the perspective of an­ other, embedded current speaker, whose moment of speaking is positioned in the past. In other words, when the epistemic modification itself is situated in the past, as in had to have, this causes a double manifestation of subjectivity: the epistemically modified information (subjectification) is interpreted as bound to a subject in the text (perspectivization). Epistemic modification can also be represented in a domain representation. An epistemic modal verb such as moet ('must') in (17) indicates the speaker's commitment to the proposition "Jan has stayed at home." The speaker in (17) is expressing that he considers the proposition to be at least quite certainly factual. The epistemic modifier moet ('must') opens an embedded domain M within the narrator's base B, in which Ρ is placed. The domain indicator is not connected to a subject within the discourse (e.g., Jan), but to the speaker ( T ) , because the epistemic modal foregrounds the speaker's commitment to the proposition's va­ lidity. Since the speaker in (17) is not instantiated as Τ in the actual discourse, an Τ is created implicitly and represented in the implicit, that is, non-encoded area of B, as is shown in Figure 5. As in the case of subjectively used deontic modals (see Figure 4), the subdomain represents an implicit perspective, that is, a subdomain bound to the implicit speaker. Note that the structure of the discourse domain representation of this epis­ temically modified utterance does not differ crucially from the representation of a subjectively used deontic utterance as depicted in Figure 4. In both cases, the modal's force is connected to the implictly represented Ί ' . However, the nature of this force is different (socio-physical versus epistemic reasoning; the precise 'semantics' of the arrows must be derived from the lexico-semantic characteris­ tics of the modal verb).

104

JOSÉ SANDERS AND WILBERT SPOOREN

Figure 5. Representation of epistemically modified utterance "John must have stayed at home" If the representation of epistemically modified utterances resembles the rep­ resentation of subjectively used deontic modals so much, the question that arises is why the former resists immediate correction while the latter does not. This might be due to the very nature of epistemic modification, namely, to give an estimate of the relative truth of a particular state of affairs. By Gricean implicature, this implies that the state of affairs is not considered 'not-true.'19 If an epistemic modal is used within an embedded perspective, it is the em­ bedded speaker's degree of commitment that is expressed. This was already shown by the epistemic modification within free indirect thought in example (24) above. For the sake of clarity, we will analyze a more explicit example of epistemic modification within a perspective domain, sentence (25). The example is constructed by adding an embedding clause to (17): (25)

Klaas denkt dat Jan thuisgebleven moet zijn. Klaas thinks that Jan home stayed must be 'Klaas thinks that Jan must have stayed at home. '

In (25), the commitment with respect to the proposition is represented in the embedded domain Mx attributed to Klaas; thus, the embedded domain M2 created by the epistemic modal is connected to the implicit Τ of Klaas, not to the speaker of the entire utterance (Figure 6). Note that in the case of must, the distance between  and M will be shorter because must expresses a higher de­ gree of certainty than may.

PERSPECTIVE, SUBJECTTVTTY, AND MODALITY

105

Figure 6. Representation of epistemically modified utterance within an embed­ ded perspective "Klaas thinks that John must have stayed at home" In the literature on epistemic modality, the difference between epistemic necessity, as expressed by must, and epistemic possibility, as expressed by may, as well as the effect of denial in combination with epistemic modifiers, have been described by Lyons (1977), Coates (1983), Seuren (1985), Nuyts (1992), and others. Distinctions have also been made with respect to the type of evi­ dence that epistemic modals represent, e.g., Chafe's (1986) inductive and de­ ductive evidential devices and Palmer's (1986) two basic modifying possibili­ ties: judgmental modifiers, involving opinions and conclusions by the speaker, and evidential modifiers, involving the evidence a speaker has for what he is saying. Sanders & Spooren (1996) propose that the type of evidence expressed by epistemic modal verbs be described in terms of the degree of subjectivity of the evidence that the modals represent. Some epistemic modal verbs, such as Dutch blijken, 'apparently be', and schijnen, 'seem', have a relatively low sub­ jectivity in that they represent intersubjective, directly manifest evidence, that is, observations in the here-and-now. Other epistemic modal verbs, such as Dutch moeten, 'must', and kunnen, 'can/may', represent highly subjective evi­ dence which concerns the speaker's reasoning based on his own knowledge of a situation.20

4. Conclusions In this paper we have sketched a way of expressing subjectivity phenomena within a framework of discourse domains. Subjectivity is manifested in one of two ways: perspectivization—information is linked to a subject in the dis-

106

JOSÉ SANDERS AND WILBERT SPOOREN

course—and subjectification—information is linked to the subject of the dis­ course. In the reader's representation of a text, subjective information is as­ sumed to be situated in an embedded discourse domain which is either linked to a character (perspectivization) or to the subject Τ (subjectification). With the use of this terminology, the distinction between epistemic and deontic modality was characterized as follows. Deontic modals indicate perspec­ tivization: the modified information is linked to a socio-physical force, usually someone or something other than the discourse subject. If, in non-default cases, this subject is the current speaker, subjectification is created by the foreground­ ing of the speaker as part of the discourse object. Epistemic modals signal sub­ jectification: they are truly evidential in that they presuppose some evidence which is the basis for the speaker's claim, and thereby indicate the speaker's commitment. Epistemic modals can be categorized further in terms of degree of certainty, that is, the speaker's commitment, and degree of subjectivity, that is, the type of evidence that is presupposed (Sanders & Spooren 1996). Within epistemic modals, a gradation of subjectivity can be seen as the expression of the type of evidence that the speaker has. Epistemic modals are closely related to I-embeddings such as 'I think that', 'I believe that'. The latter represent sub­ jectification by definition: they indicate the current speaker's commitment. The analysis presented here fits in with other approaches to modal expres­ sions as subjective, that is, as markers of elements in discourse deviating from 'objective reality'. Our purpose was to describe modal expressions in terms of degrees of subjectivity, and not in terms of the objective - subjective dichotomy proposed by Lyons (1977). We showed that with respect to both 'objective de­ ontic' and 'objective epistemic' modification, the term 'objective' is not ade­ quate, since, in both cases, a subject's force is crucial to the modal's semantics. In cases of 'objective deontic modification,' the modal is bound to a subject, implicitly or explicitly present in discourse, who expresses a force, that is, a moral or physical necessity or possibility. The objective character of objectively used epistemic modals is of a quite different nature. Here, the probability of, or commitment to, a statement is bound to the discourse's subject, as in all epis­ temic modal constructions. The present model of subjectivity should be able to describe various ex­ pressions of subjectification and perspective. One such case is the linguistic phe­ nomenon of clauses that explicitly embed the speaker's thoughts, beliefs, or opinions: / think that, I suspect that, I believe that, and other mental state predicates (Nuyts 1992: 305) that we call I-embeddings. In I-embeddings, sub­ jectification and perspectivization seem to merge. An example is given in (26): (26)

Ik denk dat Jan thuis is gebleven. I think that Jan home is stayed 'I think Jan stayed at home. '

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

107

In (26), the speaker explicitly indicates that the validity is restricted to himself. Therefore, the validity of the claim is restricted to the embedded domain M bound to the speaker I, thus marking the modified proposition PM as subjective. In fact, the perspectivization element, i.e., the binding to a subject, is over­ ruled by subjectification, i.e., the binding to the subject. Like epistemic modifi­ ers, these expressions also foreground the speaker's commitment with respect to his claim, but, by the explicit reference to himself as Τ in subject position, they express subjectification more openly than epistemic modifiers. By referring to himself as 'I,' the speaker is explicitly encoded or 'on stage.' Compared to the unmodified statement "Jan stayed at home" in (26), the speaker objectifies him­ self, that is, makes himself part of the object of the statement. In comparison, epistemic modifiers such as must and may also create subjectivity by foreground­ ing the speaker's degree of commitment to his statement. But since both this commitment and the speaker remain implicit, subjectification is less strong. This is illustrated in Figure 7.

Figure 7. Degrees of subjectification in non-modified statement, epistemically modified statement, and I-embedding As in the cases of epistemic modifiers and subjectively used deontic modifi­ ers, the information modified by an I-embedding cannot be denied as a conse­ quence of the binding to T , the speaker of the sentence. This is shown in (26'): (26') ? Ik denk dat Jan thuis is gebleven, I think that Jan home is stayed, maar hij is niet thuisgebleven.

JOSÉ SANDERS AND WILBERT SPOOREN

108

but he is not home stayed ? Ί think that Jan stayed at home, but he did not stay at home.' Yet, it is clear that linguistic markers such as think, presume, be convinced that, etc., always open an embedded domain: when the I-speaker is replaced by another speaker, the modified utterance can be denied. Compare (26') to (27'): (27') Klaas Klaas maar but 'Klaas

denkt dat Jan thuis is gebleven, thinks that Jan home is stayed hij is niet thuisgebleven. he is not home stayed thinks that Jan stayed at home, but he did not stay at home.'

In (27), an embedded perspective domain is opened by thinks and connected to Klaas. The difference between (26') and (27') is that the subjective epistemic modifier in (26') explicitly marks information as bound to the speaker as T : in other words, the speaker is present in the domain indicator. Since the base is defined as the speaker's reality, the embedded domain is in fact automatically incorporated into the base. Thus, the information embedded within the domain M that is modified by a subjective modifier is immediately transferred to base  and therefore cannot be denied within the base. Epistemic modifiers, subjectively used deontic modifiers, and I-embeddings thus have in common that they create a special type of subjectivity, i.e., speaker-commitment, represented in speaker-bound subdomains. Such subjectification is absent in objectively used deontic modals: these modals, by contrast, create perspectivization by expressing a connection between the speaker and a subject other than the speaker.

Endnotes * Preparation of this article was supported by Grant NWO 300-173-031 of the Research Foun­ dation Language, Speech, and Logic (Netherlands Organization of Scientific Research) for José Sanders. An earlier version of this paper was presented at the Functional Approaches to Grammar Conference, July 1995, in Albuquerque. We are indebted to many colleagues for helpful comments on the ideas presented in the paper, especially Leo Noordman, Gisela Redeker, Eve Sweetser, and Arie Verhagen. Please address all correspondence to Wilbert Spooren, Discourse Studies Group, Tilburg University, P.O.Box 90153, NL-5000 LE Tilburg, e-mail: Wilbert.Spooren @kub.nl. 1. In referring generically to speakers and other subjects, we will use the generic male referen­ tial forms. Please read for he: she or he and for him: her or him. 2. Kuno's (1987) notion of Direct Discourse Perspective seems closely related to the concept of source, that is, the representation of speech and thought on the verge of direct and indirect representation. Direct Discourse Perspective occurs in utterances such as That hei was the best boxer in the world was claimed by 1 repeatedly·, Kuno analyzes such utterances as

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

109

having an underlying structure with a complement in a direct discourse ( 7 am the best boxer in the world') (Kuno 1987: 106). 3. The subjectivity of a statement should not be confused with its bias. A statement is biased if the information expressed by it is limited to a restricted set of propositions held to be true (i.e., fewer than all the relevant propositions are included). A statement is subjective when a statement is connected to a subject in the text, that is, perspective and/or subjectification. 4. Palacas (1993) describes such expressive language as linguistic world-creating at the level of reflection; Mayer (1990) uses the term 'subjective assessment.' 5. Immediate correction of example sentences is indicated by numbers with primes. 6. Sometimes, the subject to whom a perspectivized statement is bound is not responsible for a particular part of the perspectivized meaning. In an utterance such as Jan said that that idiot is not coming, the choice of the words 'that idiot' may or may not fall under the responsibil­ ity of Jan (see Sanders and Redeker, 1996). 7. The phenomenon oí progressive subjectification is discussed by Langacker (1990). See Sand­ ers (forthcoming) for a diachronic approach of this phenomenon in discourse representation devices (direct speech, indirect speech, free indirect speech and thought) and in modal verbs. 8. Note that 'I-embedding' predicates such as I believe that X and / think that X are special cases in that they combine both types of subjectivity: perspective and subjectification. This point will be discussed further in the last section. 9. Multiple embeddings occur often at the discourse level, for instance, in multi-layered frame stories. 10. Another more subtle manifestation of source indication is what might be called the 'quotative imperfect,' as in the following Dutch utterance: [Waar is Jan?] Jan kwam eraan [zei hij] (English: [Where is Jan?] Jan was coming [he said]). 11. In Dutch, the modal verbs moeten ('must') and kunnen ('can') can also express physical ne­ cessity or ability, which is called 'quasi' modality ([oneigenlijk modaal], Geerts et al. 1984: 551) apart from their deontic and epistemic meaning. Examples are Hij kan heel goed zwemmen (Eng. He can swim very well) and De hond moet eten hebben (Eng. The dog wants/needs/must have food). 12. A subtype of deontic modality is volitive (or boulomaic) modality, which expresses desire (Palmer 1986, Simpson 1993), indicated by volitional modal verbs like want as in Jan wants to stay at home. Volitive modality can also be considered as the non-factive end of the epistemic-volitional scale (see Seuren 1985). 13.Note that in English, instead of must, ought to/should is used if the speaker thinks the obli­ gation may not be fulfilled (Palmer 1986: 100; also Lyons 1977: 846). Thus, in the case of must, immediate correction is not very likely. Apparently, the socio-physical force expressed by deontic must/had to is, as a default, interpreted as decisive. In the case of a permissive deontic modal such as can, immediate correction is possible because the ability or permission to do something does not entail its realization. 14. Note that volitive modals such as hope, wish, and want by their nature also do not entail the predicated proposition's factivity. For instance, Jan wants to stay at home does not entail the proposition Jan stays at home, and thus Ρ can be denied, as in Jan wants to stay at home, but he does not stay at home. 15. An interpretation is possible in which the character (Jan) is aware of the permission or obli­ gation, either by his own judgment of the circumstances or by some other authority. In that case, the domain indicator (the deontic modal) can be connected to the modal verb's subject (Jan) in B. As such, it is expressed that Jan is present as an active consciousness, considering the obligation or necessity to achieve P. 16. In representing domain diagrams, domain indicators, and pragmatic (identity) relations, we follow the conventions introduced in Fauconnier (1985, chapter 1). Lines with arrows repre-

110

JOSÉ SANDERS AND WILBERT SPOOREN

sent connections between the parent domain (typically the speaker's/narrator's base) and subdomains that are established by domain indicators. Lines without arrows connect elements and their counterparts. 17. The phenomenon of 'acting' self versus 'higher' self in deontic modality can be compared to multiplications of Τ or 'self in perspectivization, where 'self can be positioned in the past, in a dream, or can even be the addressee of the utterance. 18. Note that in the case of a strictly physical ability /necessity interpretation of root modals, the embedded domain that is opened is not bound to a character in the text, and, hence, there is neither perspectivization nor subjectification. For instance, compare (21) to the following utterance: Toen zijn rusteloosheid en angst voorbij waren, werd Jan ontslagen uit de inrichting; hij kon eindelijk thuis blijven. 'When his restlessness and anxiety were over, Jan was released from the asylum; he could finally stay at home.' In this example, could can be paraphrased as in "(Jan) was able (to stay at home)". Could then is, a description of physical ability without any subjective consciousness (as in "Jan could walk again — his leg had healed"). 19. Only in particular contexts could an immediate correction of epistemic modals be acceptable, as in (16'). Question by A : Where is Jan ? Statement by B: . Jan must have stayed at home. Response by A: (16') Ja, Jan moet thuisgebleven zijn. Maar hij is niet thuisgebleven. 'Yes, Jan must have stayed at home. But he did not stay at home. The use of epistemic modals in such contexts is unusual in that the sentences can only be understood as (more or less) acceptable if they are imagined as responses to a statement in a conversation. The sentence part with the epistemic modal is, in fact, a repetition of the other speaker's utterance, and as such - in the terminology used here - a perspectivized element (to be attributed to a subject other than the current speaker). Thus, what is represented as factual in the embedded domain is modified as possibly factual (thus possibly non-factual) in the ba­ sic reality. 20. Experimental support for these categorizations is presented in Sanders and Spooren (1996), where we also provide empirical support for the claim that epistemic modals differ with re­ spect to the degree of certainty they express: we found that epistemic modals can be catego­ rized in terms of the lowest degree of evidential certainty they permit.

References Bolkestein, Machtelt and Rodie Risselada. 1987. "The pragmatic motivation of syntac­ tic and semantic perspective." In Jeff Verschueren and Marcella Bertucelli-Papi (eds), The pragmatic perspective. Amsterdam: Benjamins, 497-512. Bybee, Joan and Suzanne Fleischman (eds). 1995. Modality in Grammar and Dis­ course. Amsterdam etc.: Benjamins.

PERSPECTIVE, SUBJECTIVITY, AND MODALITY

111

Chafe, Wallace. 1986. "Evidentiality in English conversation and academic writing." In Wallace Chafe and Johanna Nichols (eds), Evidentiality: The linguistic coding of epistemology. Norwood, N.J.: Ablex, 261-272. Chafe, Wallace. 1994. Discourse, Consciousness, and Time. The Flow and Displace­ ment of Conscious Experience in Speaking and Writing. Chicago: University of Chicago Press. Coates, Jennifer. 1983. The Semantics of Modal Auxiliaries. Beckenham: Cr oom Helm. DeLancey, Scott. 1982. "Aspect, transitivity and viewpoint." In Paul Hopper (ed.), Between Semantics and Pragmatics. Typologial studies in language. Amsterdam: Benjamins, 167-183. Dinsmore, John. 1991. Partitioned Representations. Dordrecht etc.: Kluwer. Doron, Edith. 1990. Point of View. CSLI-paper nr. 90-143. Stanford: Leland Stanford Junior University. Fauconnier, Gilles. 1985. Mental Spaces: Aspect of Meaning Construction in Natural Language. Cambridge: Bradford. (2nd, revised edition: 1994). Cambridge: Cambridge University Press. Fillmore, Charles. 1982. "Frame semantics." In The Linguistic Society of Korea (ed.), Linguistics in the Morning Calm. Seoul: Hanshin, 111-137. Geerts, Guido, Walter Haeseryn, Jacobus de Rooij and Maarten van den Toorn (eds). 1984. Algemene Nederlandse spraakkunst [Universal grammar of Dutch]. Gron­ ingen: Wolters Noordhoff. Givón, Talmy. 1982. "Evidentiality and epistemic space." Studies in Language 6 (1): 23-49. Heine, Bernd. 1995. "Agent-oriented vs. epistemic modality: Some observations on German modals." In Joan Bybee and Suzanne Fleischman (eds), Modality in grammar and discourse. Amsterdam etc.: Benjamins, 17-53. Kuno, Susumo. 1987. Functional Syntax. Chicago: University of Chicago Press. Langacker, Ronald. 1987. Foundations of Cognitive Grammar, vol. 1. Stanford, CA.: Stanford University Press. Langacker, Ronald. 1990. "Subjectification." Cognitive Linguistics 1 (1): 5-38. Langacker, Ronald. 1993. "Reference-point constructions." Cognitive Linguistics 4 (1): 1-38. Lyons, John. 1977. Semantics (2 volumes). Cambridge: Cambridge University Press. Lyons, John. 1982. "Deixis and subjectivity: Loquor, ergo sum?" In R. Jarvella and W. Klein (eds), Speech, place and action. Studies in deixis and related subjects. New York: Wiley, 101-124. Mayer, Rolf. 1990. "Abstraction, context, and perspectivization: Evidentials in dis­ course semantics." Theoretical Linguistics 2: 101-163. Nuyts, Jan. 1992. Aspects of a Cognitive-Pragmatic Theory of Language: On Cogni­ tion, Functionalism, and Grammar. Amsterdam: Benjamins. Nuyts, Jan. 1993. "Epistemic modal adverbs and the layered representation of concep­ tual and linguistic structure." Linguistics 31: 933-969. Palacas, Arthur. 1993. "Attribution semantics: Linguistic worlds and point of view," Discourse Processes 16: 239-277. Palmer, Frank. 1986. Mood and Modality. Cambridge: Cambridge University Press.

112

JOSÉ SANDERS AND WILBERT SPOOREN

Reinhart, Tanya. 1983. "Point of view in language: The use of parentheticals." In G. Rauh (ed.), Essays on deixis. Tübingen: Narr, 169-194. Sanders, José, forthcoming. "Degrees of subjectivity in epistemic modals and perspec­ tive representation." In Leon de Stadler (ed.), Proceedings of the Third Interna­ tional Cognitive Linguistics Society. Sanders, José and Gisela Redeker. 1996. "Perspective and the representation of speech and thought in narrative discourse." In Gilles Fauconnier & Eve Sweetser (eds), Spaces, Worlds, and Grammar. Chicago: Chicago University Press, 290-317. Sanders, José and Wilbert Spooren. 1996. "Subjectivity and certainty in epistemic mo­ dality. A study of Dutch epistemic modifiers." Cognitive Linguistics: 7 (3), 241264. Sells, Peter. 1987. "Aspects of logophoricity." Linguistic Inquiry 18 (3): 445-479. Seuren, Pieter. 1985. Discourse Semantics. New York: Basil Blackwell. Siewierska, Anna. 1991. Functional Grammar. London: Routledge. Simpson, Paul. 1993. Language, ideology and point of view. London: Routledge. Spooren, Wilbert. 1989. Some aspects of the form and interpretation of global con­ trastive coherence relations. Doctoral dissertation. Nijmegen: University of Nijmegen. Spooren, Wilbert and Joep Jaspers. 1990. "Tekstoperaties en tekstperspectieven [Discourse operations and discourse perspectives]." Gramma 19: 195-219. Sweetser, Eve. 1990. From etymology to pragmatics: Metaphorical and cultural as­ pects of semantic structure. Cambridge: Cambridge University Press. Uspensky, Boris. 1973. A poetics of composition. Berkeley: University of California Press. Wiebe, Janyce. 1990. "Identifying subjective characters in narrative." Proceedings COLING 90: 401-406. Wiebe, Janyce. 1994. "Tracking point of view in narrative." Computational Linguistics 20 (2): 233-287. Willett, Thomas. 1988. "A cross-linguistic survey of the grammaticalization of evidentiality." Studies in Language 12 (1): 51-97. Zribi-Hertz, Anne. 1989. "Anaphor binding and narrative point of view: English re­ flexive pronouns in sentence and discourse." Language 65: 695-727.

IL METAPHORS AND METONYMY IN DISCOURSE

A Few Metaphorical Models in (Western) Economic Discourse Frank Boers and Murielle Demecheleer Free University of Brussels, Belgium

1. Introduction: Metaphor in Cognitive Semantics In Cognitive Semantics (Lakoff 1987; Johnson 1987) metaphor is considered as a basic cognitive process which allows us to talk and think about abstract phe­ nomena. We map the structure of (concrete) source domains onto abstract do­ mains of experience. Abstract reasoning is largely governed by these metaphori­ cal mappings. Since certain metaphors are so deeply entrenched in everyday human thought, Lakoff and Johnson (1980) have called these metaphors we live by or conceptual metaphors. Metaphors we live by appear to be of different degrees of abstraction. Some map image schémas onto abstract domains. These image schémas are gestalt structures that organize our experience of physical space and that serve as identifying patterns in a large number of experiences. The PATH schema, for instance, emerges from a multitude of physical activities involving motion from one location to another. From the time we can first crawl, we often (try to) traverse a certain distance in order to reach an inviting objective (a toy or another object that strikes our fancy). This everyday experi­ ence continues into adulthood, for instance when we (try to) cross the floor that separates our table from the bar, in order to get more drinks. The PATH schema arises from these recurring experiences, structures them (source-path-goal), and identifies them as analogous patterns. Other image schémas include LINK, CONTAINER, UP-DOWN, FRONT-BACK, CENTRE-PERIPHERY, COM­ PULSION, and so on (Johnson 1987: 126; Lakoff 1987: 271). Through meta­ phor, image schémas are also used to structure abstract experience. Some ex­ amples: categories are often understood in terms of CONTAINER schémas; hierarchies are often understood in terms of UP-DOWN schémas; relations are often understood in terms of LINK schémas; causation is often understood in terms of COMPULSION schémas. Other metaphors do not build directly on image schémas. Instead, they map the structure of more specific or elaborate source domains onto abstract ones: e.g. COMPETITION IS A RACE {the British economy was lagging behind), COMPETITION IS A FIGHT, {the British car industry was unable to fight their Japanese competitors), and so on. If our conception of abstract domains of experience is essentially a matter of metaphorical mappings, then this also holds for the domain of economics. Indeed, throughout the history of economic thought, various metaphors have

116

FRANK BOERS AND MURIELLE DEMECHELEER

been used to comprehend and to clarify economic concepts and processes. This already appears from the pioneering days of economic theory. Adam Smith (1776) introduced the term invisible hand to argue his point that a free market system always adjusts to new circumstances. He referred to tariffs as trade bar­ riers that should be taken down, even if they were meant to give a chance to beginning enterprises, so-called infant companies. There is an abundance of metaphors in current economic discourse too (e.g. McCloskey 1986: 74-86). Some examples: money transfers constitute cash-flow, economies need equilibrium, otherwise they will collapse, public debt is a bur­ den, currencies may be weak or strong, a company may take its products up­ market or downmarket, business moves in cycles, employees are human capital, demand may be elastic. In this paper we focus on three general, conventional metaphorical models: the PATH metaphor, which builds on an image schema (the PATH schema), and the HEALTH and WAR metaphors, which build on more specific and elaborate source domains. Mapping the structure of one domain of experience onto more abstract ones is not an arbitrary process. The experiential "logic" of the source domain (its inference patterns and associated value judgements) is generally preserved in the metaphorical mapping. This is the Invariance Hypothesis (Lakoff 1990: 54). According to the "logic" of the PATH schema, for instance, the goal of the path is the desired location that one wants to reach. Motion towards the goal is posi­ tively valued, while immobility or motion away from the goal are negatively valued. The shortest way to the goal is generally preferred. Mapped onto ab­ stract experience, this means that activities which serve a clear purpose and which yield results fast are valued positively. Inactivity and indecisiveness are valued negatively, and so on. As we shall see, the Invariance Hypothesis also applies to those metaphors that build on more specific and elaborate source do­ mains. Gentner & Gentner (1983), for example, showed that people's inferences about electricity differed, depending on the metaphorical model they used (fluid flow model versus moving crowd model). Metaphor is helpful (and even indispensable) as a vehicle to think about abstract phenomena, but one should be careful not to mistake the metaphors for the "reality" they try to describe. Two sources of bias are involved: (a) The source and target domain may be perceived/conceived as analogous, but they are by definition different. The logic of the source domain need not hold in the target domain. (b) Most metaphors have a metonymie basis: a typical example or substruc­ ture of the source domain is taken to stand for the entire domain (i.e. as a me­ tonymie model) and subsequently mapped onto abstract domains. AN ACTIVITY IS MOTION ALONG A PATH, for instance, builds on a typical example of physical activity: moving one's body from one location to another. This is taken as representative of the category of physical activities in general.

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

117

Then, it is mapped onto abstract activities: reaching the end of one's speech, going over the same old problem, thinking ahead, taking the right course of ac­ tion, getting over one's difficulties, etc. Of course, moving one's body from one location to another is not the only possible physical activity: other typical cases include direct manipulation (e.g. building or destroying things) and oral con­ sumption. Mapping the structure of those other physical activities yields other inference patterns. The metonymie basis of metaphor can also be recognized in more specific source domains. Physical competition, for instance, seems to dis­ play two typical examples: races and fights. The structure of races can be used to describe other kinds of physical competition, like soccer: e.g. Manchester were two points behind Liverpool. Alternatively, the structure of fights can be used: e.g. Manchester's counterattack broke through the Liverpool defence. Subsequently, both metonymie models can be selected for metaphorical mapping onto abstract kinds of competition, like economics: e.g. Domestic industry was lagging behind its foreign competitors, or Domestic industry couldn 't fight its foreign competitors. In short, metaphors highlight some aspects of the target domain, but they leave other aspects in the dark. The race metaphor, for instance, will probably highlight the importance of R & D and quick responses to market changes, whereas the fight metaphor may highlight the importance of power structures, etc. Conceptual metaphors need not all be universal (e.g. Lakoff 1987: 295). Different communities may have different metaphors to conceive of the same abstract concepts. The way a community conceives of an abstract phenomenon through metaphor is also subject to change over time (e.g. Johnson 1987: 129). There will always be room for the creation of new metaphors, which may in turn become conventional. Those conventional metaphorical models of a com­ munity are a reflection of the predominant ideology of that community at that time. In this context, the cross-linguistic study of conventional metaphors may offer support for the notion of linguistic relativity. However, one should be careful not to interpret this notion as linguistic determinism, because the con­ ventional metaphors of a given community do not prevent its language users from inventing new ones which offer new insights into abstract phenomena. Nevertheless, it is the conventional imagery that is most readily available and accepted as "true", of course. Obviously, conventionality is a matter of degree. One indication of a phe­ nomenon's conventionality in a given community is its frequency of occurrence. A certain metaphorical model may be instantiated more frequently in the dis­ course of one community than that of another. Being more typical of the former community, it may be more deeply entrenched in its conventional thought pat­ terns, too. This holds for different kinds of communities, whether defined so­ cially, economically, ideologically, or linguistically.

118

FRANK BOERS AND MURIELLE DEMECHELEER

As economics involve abstract concepts, we cannot avoid using metaphors in this domain either (e.g. Jäkel 1993). Once in a while, however, it may be sen­ sible to reconsider their validity. After all, our vision of economics may be blurred by the fashionable metaphorical models of our era. Describing an eco­ nomic recession as a tunnel, for example, leads us to expect better times ahead (the light at the end of the tunnel). Describing public debt as a burden may af­ fect our value judgement about new taxes raised to reduce the weight. Describ­ ing taxes as a burden weighing down on entrepreneurs may lead us to allow them tax reductions. And so on.

2. Focus and Sample Our study focussed on a few conventional metaphorical models only. These re­ late to the PATH source domains of WARFARE and HEALTH CARE. The analysis was further restricted to western economic (and political) discourse ad­ vocating a free-market ideology. At a first stage, we looked at the persuasive use of the metaphors in the leaders (opinion articles) of The Economist (referred to below as Έ ' ) . The sample here consisted of 58 articles, yielding a total of approximately 50,000 words. At a second stage, we counted the relative fre­ quencies of occurrence of the three metaphorical models across three languages: English, Dutch and French. Here the sample consisted of newspaper editorials and background articles taken from The Financial Times (referred to below as TT') and equivalent newspapers in Dutch and French {De Financieel Ekonomische Tijd; Le Quotidien de l'Economie et de la Finance). The contrastive part of the study was based on a corpus of approximately: 43,400 words for the English sample, 18,500 words for the Dutch sample, and 21,800 words for the French sample. The three samples were comparable in terms of ideology and subject matter.

3 . The P A T H Metaphor Various activities in the domain of economics are commonly conceived as mo­ tion of a company, an organization, or a country over a path towards a goal. This PATH metaphor has its proper experiential "logic", and its inferences yield additional metaphors with their own associated value judgements: PURPOSES ARE DESTINATIONS; DIFFICULTIES ARE OBSTACLES; DECISION-MAKING IS CHOOSING A DIRECTION; PROGRESS IS MOVING FORWARD; MOVING FORWARD IS GOOD; IMMOBILITY IS BAD; etc.

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

119

AN ACTIVITY IS MOTION OVER A PATH: The nature of the conceived path often remains implicit and only the motion is referred to: (1) The moves towards privatisation and liberalisation. (FT 9.3.94) When the path is made explicit, it appears in a variety of shapes. In the most common case it is conceived as a solid surface, like a road: (2) India this week altered course on the road to economic reform, aban­ doning the tough fiscal path. (FT 2.3.94) The moving entity may be conceived as some kind of vehicle, like a car: (3) The economy has shifted from neutral into first or second gear. (FT 28.3.94) A fair number of instances (especially in the English and Dutch sample) build on the nautical domain: (4) (5) (6) (7) (8)

bailing out a sinking bank. (E 8.1.94) Japan has been in the doldrums before. (E 18.12.93) keeping the company afloat. (FT 15.3.94) introducing the measure and then shifting tack. (FT 29.3.94) the government has been blown off fiscal course. (FT 26.3.94)

Other instances use the image of railways: (9) (10) (11) (12)

nor need the ensuing turbulence derail economic reform. (FT 21.3.94) Economic growth has ground to a halt. (FT 25.3.94) The programme of labour market de-regulation has been stopped in its tracks. (FT 7.3.94) as the economy picks up steam (E 5.2.94)

Various means of transport may be selected as source domains and each may incorporate slightly different inference patterns and associations. PURPOSES ARE DESTINATIONS : According to the experiential "logic" of the PATH schema, paths are typically goal-oriented. It is the desired goal that motivates motion towards it. If one does not move towards the desired goal, one's need will not be satisfied. Mapped

120

FRANK BOERS AND MURIELLE DEMECHELEER

onto abstract domains this means that activities that serve a clear purpose are positively valued. (13)

Korea's march towards market-driven prosperity. (FT 22.3.94)

Other instances exploit the idea of proximity to or distance from the desired goal. According to the metaphor, the goal's proximity or distance is a measure of the degree of success of an activity or policy. (14) (15)

it is not far from independence. (E 12.2.94) neither the treaty nor the fund come close to a solution. (FT 21.3.94)

DIFFICULTIES ARE OBSTACLES: In abstract domains, difficulties are often conceived as obstacles on one's goaloriented path: (16) (17)

Africa faces many formidable obstacles. (FT 18.3.94) while the obstacles remain mountainous. (FT 23.3.94)

According to the metaphor's logic, difficulties are bad, because they block the access to one's goal (they stand in the way of progress): (18) (19)

the image of Britain as a negative force, bent on blocking any kind of reform or advance. (FT 14.3.94) Britain's treasury has become the main obstacle to improving Lon­ don's public transport system. (FT 14.3.94)

To reach one's destination one has to go round the obstacle, climb or jump over it, or remove it. Metaphorically, then, difficulties should either be over­ come, or avoided, or removed: (20) (21) (22)

there is a chance for Brazil to surmount its economic and political difficulties. (FT 31.3.94) This was blocked by environmental objections. When this hurdle was overcome. ... (FT 23.3.94) It is this impasse that the commission is now trying to get round in talks with the steel firms. (E 19.2.94)

DECISION/POLICY-MAKING IS CHOOSING A DIRECTION: Different paths or directions represent different economic/political options:

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

(23) (24) (25) (26)

121

the country is at a turning point. (FT 31.3.94) to give the organization a new sense of direction. (FT 1.3.94) the Fed's change of monetary course. (FT 5.3.94) The country that was the catalyst for so much has lurched into the other direction. (E 18.12.93)

Good decision/policy-making is choosing the right direction: (27) (28) (29)

These are steps in the right direction. (FT 22.3.94) this is the only policy likely to put industrial countries on the path they are apparently seeking. (FT 16.3.94) The right way forward looks different. (E 29.1.94)

Bad decision/policy-making is choosing the wrong direction: (30) (31)

Greenspan's slow-moving Federal Reserve is more likely to err in that direction than in the opposite one. (FT 19.3.94) They appear to have steered companies in the wrong direction. (E 22.1.94)

Inability to choose or to pursue one's direction is a token of indecisiveness and is generally valued negatively, since it does not bring the desired goal any closer: (32)

If you stay the course, there is hope. If you waiver, the pain can only get worse. (FT 15.3.94)

In physical space, it is usually the shortest and quickest path to one's desti­ nation that is considered the best. This value judgement may be preserved in the target domain. Moreover, one cannot tread different paths at the same time in physical space. Preserved in the target domain, this may feed a belief in exclu­ sive choices. Choosing a path in physical space means abandoning other options. Mapped onto the domain of economics, this may preclude policies that combine aspects from different ideologies. The opinion articles under examination, for instance, generally advocate "pure" free market economics (without government interference, without incorporating any aspects of socialist models, etc.). PROGRESS IS MOVING FORWARD: (33) (34)

Convertibility of the rupee is an important step forward. (FT 4.3.94) America, a more advanced single market. (E 5.2.94)

122

FRANK BOERS AND MURIELLE DEMECHELEER

Progress (or lack of it) can be assessed by measuring the metaphorical approxi­ mation to the desired goal again: (35) (36)

The government is further distancing itself from the aim it espoused in 1991. (FT 2.3.94) if he is only half-way successful, he will have started the work of making a seventh of America's economy function properly. (E 19.1.94)

The faster one reaches the desired goal, the faster one's need is satisfied. Mapped onto abstract domains, this generally translates as fast forward motion is better than slow forward motion: (37) (38) (39)

the lack of progress, which threatens Brazil's hopes of rejoining Latin America's economic fast track. (FT 31.3.94) the faster the transition in made, the better. (E 19.2.94) too timid in the pace and extent of their economic changes. (E 29.1.94)

Standing still is a token of hesitation, indecisiveness or disorientation, and does not bring the desired goal any closer. Hence, immobility is usually valued negatively: (40) (41)

economic stagnation and rising unemployment. (FT 10.3.94) the penalty for so much earlier hesitation. (E 29.1.94)

According to the PATH schema, backward movement takes one away from the desired destination. Consequently, moving (or being) backward is consid­ ered to be bad: (42) (43)

the most backward southern region. (FT 23.3.94) A setback for China's economy. (8.1.94)

A great number and a great variety of figurative expressions can systemati­ cally be traced back to the PATH metaphor. In spite of the variety of expres­ sions, lexical cohesion of the articles is preserved by the underlying conceptual metaphor. At its most abstract (or schematic) level, the PATH metaphor builds on the source domain of physical space in general, but it also subsumes a num­ ber of more concrete instances (or elaborations) building on more specific source domains within physical space, like sailing, railways, etc. Moreover, at any level of abstraction, the PATH metaphor actually consists of a cluster of associated metaphors (PURPOSES ARE DESTINATIONS, etc.). In turn, this cluster can be linked up with other ones, which have not been included in this

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

123

study, like the COMPULSION schema (pressing on with reform, pushing through a new policy), the conception of time as a path (looking forward to the years ahead, the cold war is behind us). Does the PATH metaphor involve potential bias of our reasoning about economics? One of its inferences, namely "moving forward is good and stagna­ tion is bad," certainly fits in nicely with economic models that encourage eco­ nomic growth: while stagnation (immobility) is generally considered as bad, economic growth (progress) is widely considered as a necessity. In the opinion articles that we have analyzed, economic growth is presented as a worthwhile goal (destination). Moreover, investors and managers are presented as the peo­ ple that lead us towards that goal. If these premises are accepted, then it follows that investors and managers should not be impeded in their mobility by obstacles (e.g. trade barriers & trade unions), restnctions (e.g. regulations & labour legis­ lation) or burdens (e.g. taxes). It is hard to tell to what extent perceived eco­ nomic "truths" are grounded in metaphorical models, but questioning the valid­ ity of those models may be worthwhile from time to time.

4. The HEALTH Metaphor Health is associated with life expectancy and life quality, while illness is associ­ ated with pain and death. Those associations may be preserved when describing economics in terms of health care. (1) (2) (3) (4)

Brazil's chronic budget deficit. (FT 8.3.94) this business headache. (FT 28.3.94) the financial haemorrhaging. (FT 11.3.94) Britain's car industry now looks healthier. (E 5.2.94)

Healthy people tend to be active and energetic. Consequently, a high level of activity is usually valued positively. Illness or injury typically undermine the patient's energy and mobility. Hence, immobility or a low level of activity are considered as negative: (5) even greater paralysis in decision-making. (FT 14.3.94) (6) it left the economy stubbornly unstimulated. (E 18.12.93) As long as one is in good health, there is no need for medical treatment. When one's health breaks down, however, the illness has to be diagnosed and medical treatment prescribed. If necessary, the patient will have to undergo sur­ gery. In traditional western medical practice, this usually boils down to remov­ ing the infected or malfunctioning tissue. If the treatment is successful, the pa­ tient will recover and become active again. This "logic" is generally exploited

124

FRANK BOERS AND MURIELLE DEMECHELEER

in the leaders of The Economist and the editorials of the newspapers under ex­ amination as follows: (a) as long as companies or industries are "healthy" (i.e. active and making profits) there is no need for interference of any kind; (b) when the activities slow down and profits fall, however, some kind of diagnosis and treatment are required (auditing, subsidies, tax incentives, government par­ ticipation, etc.); (c) if necessary, malfunctioning branches have to be removed (e.g. closed down); (d) if the remaining body becomes active and profitable again, the treatment has proved a "success": (7) (8) (9) (10) (11) (12) (13) (14) (15)

symptoms of an over-centralized system. (FT 4.3.94) If France fails to recover fast, the OECD's alternative prescription of a change in policy mix ... (FT 10.3.94) to ease the pain of closing many steel mills. (E 19.2.94) to seek a negotiated economic remedy. (E 12.2.94) the market cure. (E 19.2.94) The former defect is now beginning to be healed ... The latter may not be wholly curable ... to cause a relapse. (FT 8.3.94) the right economic medicine. (FT 23.3.94) A worker-owned firm is likely to be bad at short-term surgery that costs jobs. (E 25.12.93) corporate resuscitation. (FT 15.3.94)

The underlying conceptual metaphor ensures lexical cohesion within the discourse, despite the apparent variety of expressions. Again, we might wonder, does the HEALTH metaphor involve potential bias of our reasoning about economics? By and large, the economic views ex­ pressed in The Economist and the three newspapers under examination corre­ spond to the views held by the employers' organizations. How can the HEALTH metaphor be used as a persuasive tool to support the employers' in­ terests? One's physical health is precious and consequently health care should be high on one's list of priorities in life. This inference may be preserved in the conception of economics under this metaphor. Similarly, the patient cannot usually be blamed for being ill. Preserved in the metaphorical mapping, this may serve as a very convenient excuse for enterprises asking for injections of public capital. Within the same metaphorical model, of course, organizations, enterprises, etc. should take care not to endanger their health. To convince them that certain policies are economically harmful, those policies can be likened to health hazards (e.g. alcohol): (16) (17)

Government intervention can be addictive. Unfortunately, kicking the addiction seems to be getting even harder. (E 19.2.94) This is a self-inflicted headache, with the Yen's rising adding to the headache. (E 18.12.93)

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

125

Having a slim and athletic body has become a token of physical health and fit­ ness. Being slim and athletic is valued positively, while being fat is generally valued negatively in our era. Mapped onto the domain of economics, these value-judgements can be exploited for reasons of persuasion again. In fact, ex­ pressions like slimming down have become common euphemisms for closing down subsidiaries, laying off employees, etc. In the same line of argument, overcentralized organizations are often described as too heavy and hence too slow and immobile. (18) (19) (20) (21)

let it slim capacity and costs. (FT 28.3.94) the American disease: an overweight monster. (E 15.1.94) slimmed-down American companies are invading weaker markets. (15.1.94) mass-firings left a meaner and leaner corporate America in their wake. (15.1.94)

The notions of good economic health and fitness also link up with the no­ tion of mobility that is associated with the path metaphor.

5. The WAR Metaphor As The Economist and the three newspapers under examination generally favour a free-market system, their opinion articles tend to argue against trade wars. And yet, their discourse shows a high frequency of instances where companies, organizations, national economies, etc. are conceived as animate entities en­ gaged in a struggle of some kind: (1) (2) (3) (4) (5)

The Commission may not have the stomach for a fight. (FT 11.3.94) The Bank of England is flexing its muscles. (FT 12.3.94) to combat fraud. (FT 23.3.94) getting to grips with the structural rigidities. (FT 29.3.94) If you wait until inflation punches you in the nose, the subsequent fight (ie. monetary tightening) will have to be nasty. (E 26.2.94) (6) economic arm-twisting. Π29.1.94)

Warfare is essentially an elaborate fight with greater numbers and heavier weapons. ECONOMIC/POLITICAL ACTIVITY IS WAR: (7) (8) (9) (10)

The budget for 1994 is a field of struggle. (FT 22.3.94) Brussels is on the war-path against subsidies. (FT 23.3.94) a battle the union could not have won. (FT 7.3.94) to launch a new assault on the budget deficit. (FT 10.3.94)

126

FRANK BOERS AND MURIELLE DEMECHELEER

(11) (12) (13) (14) (15) (16)

the recent epic battle for Paramount Communications. (FT 17.3.94) Chinese retaliation against US exports. (FT 17.3.94) tactical retreat in the face of special interests. (FT 10.3.94) the PRU broke ranks with the Association. (FT 25.3.94) a full-blown attack on youth unemployment. (FT 29.3.94) American companies are invading weaker markets. (E 15.1.94)

Fights and wars being what they are, their imagery leaves little room to conceive of economics in terms of joint problem solving. Instead, they depict a tough and relentless business world, calling for a win or lose approach or a sur­ vival of the fittest attitude. Describing one's company or economy as being en­ gaged in a war also creates an atmosphere of solidarity within a group: one stands together against a common enemy (foreign competition, the budget defi­ cit, etc.). The imagery is also used to justify extra effort and sacrifices on the part of the employees. For instance, if people view their domestic economy as being threatened by foreign competition, they may more easily accept measures to enhance domestic competitive strength (flexible working hours, extra incen­ tives for investors, reducing employers' labour costs, etc.). If employees view their company as being engaged in a competitive struggle, they may be more likely to accept the "necessity" of proposed anti-social measures (freezing wages, giving up weak branches, etc.), and so on: (17) (18) (19)

the crusading vision that kept western Europe and America shoul­ der to shoulder. (E 8.1.94) Britain too needs to remain on guard. (E 5.2.94) the sacrifices were made, victory was declared. (E 19.2.94)

The WAR metaphor is also commonly used to describe competing ideolo­ gies. Conceiving different ideologies as opponents in a battle may blur the op­ tion of viewing them as complementary. The metaphor clearly highlights the conflicting aspects of different ideologies, and it leaves potential common or compatible aspects in the dark: (20) (21)

The two great enemies of liberal democracy in the 20th century have been communism and fascism. (E 18.12.93) The two possibilities, freedom and protection, continuously battle it out. (E 29.1.94)

6· Frequencies and Contrastive Observations The three clusters of metaphors examined in this study (the PATH, HEALTH and WAR metaphors) are very common in all three samples (i.e. all three Ian-

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

127

guages). Considering the close relationship (both diachronically and synchronically) between English, Dutch and French, and the international nature of the subject matter (economics and related politics), it would be very odd indeed, if one of the above metaphorical models were completely absent from any of the samples. Nevertheless, the frequencies of occurrence listed below indicate dif­ ferences in degrees of conventionality of the given metaphors, even at the level of closely related languages. Table 1. Average frequencies of occurrence per 10,000 words

English Dutch French

PATH

HEALTH

WAR

61.01 47.03 57.87

27.64 9.72 30.27

39.40 47.30 68.35

Counting more specific subcases of the PATH metaphor could yield more contrastive observations. For instance, one might for historical reasons expect a higher frequency of PATH metaphors based on the nautical domain in English than in French. Indeed, this expectation seems to be confirmed by our data: 5.74 instances over 10,000 words in the English sample, compared to 3.77 in the French sample. Obviously, the corpus at our disposal does not allow us to draw anything but very tentative conclusions at this level of organization.

7. Conclusions and Perspectives Clearly, this study focussed on only a few metaphorical models in economic discourse. To give but one example, it did not include metaphors like AN ECONOMIC SYSTEM IS A MECHANISM: (1) a mechanism to reduce inflation; the exchange rate mechanism; an overhaul of the national health care system; etc. A full analysis of the metaphorical models behind economic thought would be a major undertaking. Yet, it appears that a great variety of figurative expres­ sions can be traced back to a limited set of clusters. A corpus-based study of frequencies and distribution of those clusters of metaphors would give us an indication of their degrees of conventionality. It would point up those metaphors that are central to a given community's conception of economics. Another approach could be to take national stereotypes as a starting point and to look at the ways these are reflected in figurative language. For example,

128

FRANK BOERS AND MURIELLE DEMECHELEER

the national stereotype of English includes a preoccupation with gardening, whereas the national stereotype of the French includes a preoccupation with food. Gardening and oral consumption also serve as source domains: (2) flourishing companies; branches of a company; a withering economy; pruning costs; down to the grassroots, etc. (3) to swallow subsidies; to gobble up a smaller company; salami-slicing tactics; to feed economic growth, etc. The national stereotypes shine through in our quantitative data: (i)

Figurative expressions related to gardening occur 4.38 times over stretches of 10,000 words in the English sample, while the French sample shows a frequency of only 1.38.

(ii)

Conversely, figurative expressions related to food (or oral consump­ tion) occur 7.34 times over 10,000 words in the French sample while the English sample shows a frequency of only 2.76.

Conventional metaphors of a linguistic community correspond to the con­ ventional patterns of thought of that community. Even if different communities share similar metaphorical models to conceive a given abstract domain, they may still differ in the relative preferences with which these models are applied. In other words, a degree of linguistic relativity may still be reflected by the dif­ ferent frequencies of occurrence of metaphorical models. Since metaphors highlight some aspects of a target domain but leave other aspects in the dark, one's conception of economics may partially be predisposed by the metaphors used. Fortunately, one can easily shift from one conventional metaphor to another. Furthermore, if established metaphors are not satisfactory anymore, one may create new ones that offer new perspectives on seemingly established "truths", or one may look for alternative metaphorical windows in other communities.

References Gentner, Dedre and D. R. Gentner. 1983. "Flowing Water or Teeming Crowds, Men­ tal Models of Electricity." In Dedre Gentner and Albert Stevens (eds), Mental Models. Hillsdale: Lawrence Erlbaum. Jäkel, Olaf. 1993. "Economic Growth versus Pushing up the GNP: Metaphors of Quantity from the Economic Domain. " In L.A. U.D. Series C, 24, 1-20. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagina­ tion, and Reason. Chicago/London: University of Chicago Press.

A FEW METAPHORICAL MODELS IN (WESTERN) ECONOMIC DISCOURSE

129

Lakoff, George. 1987. Women, Fire and Dangerous Things: What Categories Reveal about the Mind. Chicago/London: University of Chicago Press. Lakoff, George. 1990. "The invariance hypothesis: Is abstract reason based on imageschemas?" Cognitive Linguistics 1.1: 39-74. Lakoff, George and Mark Johnson. 1980. Metaphors We Live By. Chicago/London: University of Chicago Press. McCLoskey, Donald N. 1986. The Rhetoric of Economics. Brighton: Harvester. Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. Campbell, Skinner & Todd (eds), Oxford: Clarendon Press, 1976.

The Spatialization of Judgment Michele Emanatian University of' Massachusetts-Amherstf U.S.A.

1. Introduction Probably cognitive linguists have talked more about space than about any other topic. From body part term sources of adpositions, to the lexicalization of mo­ tion, from ideas about the container structure of categories, to the localist ori­ gins of possessives and tense/aspect markers, space has filled the pages of jour­ nals and the halls of conferences. Metaphor studies of English have yielded the very general bundle of meta­ phors, collectively labeled "Event Structure" (Lakoff 1993), which spatialize our understanding of events in time and in their causal context. The Event Structure metaphor(s) give us State as Location, Change of State as Motion, Purposeful Activity as Journey, etc. Such metaphors, along with metaphors for knowledge, ideas, the mind, and communication which draw on the location, transmission, and size/shape/orientation of objects in space, "work out different themes in our (non-physical) reflections about (non-physical) knowledge, but always in a metaphorical context of physical activity in a physical environment" (Salmond 1982: 67). A specific and robust instance of the Event Structure meta­ phor in English, discussed by Salmond (1982) and by Sweetser (1992), has to do with how we conceive of thought and of conversation as travel through ter­ rain. The participants in thinking, conversing, and more particularly, arguing, can trace paths through and/or "cover" intellectual terrain. Examples of this metaphor follow. (1) Intellectual & Conversational Travel Through Terrain (Salmond 1982; Sweetser 1992) You reach, or arrive at, a conclusion. From premises you started out at, you reason to conclusions. Do you follow me? Have I lost you? Are you having trouble keeping up with me? I can always go back over - or take you back over - the most important points. He interrupted my train of thought; I can't remember how far we'd got­ ten. Let's see, where were we? Anyway, ... I feel like we've gotten a little bit bogged down here. Let's move on. We've covered that already. This leads me to postulate ...

MICHELE EMANATIAN

132 Intellectual States

They keep shifting their position on the immigration issue, so it's hard to know where they stand on it. "But linguists and anthropologists follow their own paths independently." Relationships Between Thinker & Thoughts Think it over. I've got to think it through. the sheer range of his interests, breadth of vision He's a deep thinker. As Sweetser puts it, "linguistic interaction [is] a shared journey through a mental landscape". If States are Locations, then "thinking, or changing opin­ ions, is moving from one mental state (set of beliefs or thoughts) to another". When speaker and hearer understand each other, they are thus in a (partially) shared mental state, which is metaphorically mapped onto a shared location. When they continue to understand each other through a shifting sequence of mental states, they metaphorically travel from location to location together, re­ maining co-located throughout a shared journey (pp. 716-717). This paper is a contribution to a larger study of intellectual and conversa­ tional paths, of metaphorical location, orientation, place, and position. My focus here is on two spatial metaphors that enlist the concept of Distance in the ex­ pression of Judgment. In particular, I will look at: (1) the distance of a proposition from the location of truth, which I will call the Propositional Distance metaphor; and (2) the distance of a thinker (a potential judge or decision maker) from the subject matter to be thought about (and perhaps ultimately to be talked about), which I will call the Personal Distance metaphor. Taken together, the two metaphors give us a way to understand the rela­ tionship between, on the one hand, expertise, neutrality, and bias, and on the other hand, a person's likelihood of making accurate statements about the world, that is, making accurate judgments. Our everyday ("folk") views of such mat­ ters, as suggested by the metaphors, seem equivocal. For example, the closer a person is to an issue (or idea, event, etc.), the more likely their judgments about that issue will be true. On the other hand, if someone is too close to an issue, their ability to judge may be influenced by their interests. The metaphors are also interesting for what they reveal about how we un­ derstand truth, and how we conceive of human understanding and involvement. The Propositional Distance metaphor manifests two different conceptualizations of truth. There is the truth we aim for in our judgments, a kind of truth that is

THE SPATIALIZATION OF JUDGMENT

133

"out there", if only we could know it; and there is the true-ness our judgments have, trueness which falls off in degrees from absolute truth. The Personal Distance metaphor allows us to assess a person's competence in making judgments through assessing their degree of involvement in the matter at hand. As involvement is measured in distance terms, metaphorical proximity (that is, involvement) has implications for metaphorical viewing, that is to say, understanding. A person making a judgment should be neither too close nor too far from the matter to be judged, although there are situations in which extreme positions provide more favorable conditions for judgment. But the optimal viewing distance is not the only factor determining the appropriate degree of Personal Distance. Also implicated are our models of how emotion and intellec­ tion are related.

2. Distance in Metaphors of Judgment 2.1. Propositional Distance: Distance of a proposition from the truth I first want to look at the Propositional Distance metaphor and the verbal as­ sessment of truth. If you say, as speaker A does in (2) (2) A: Crime rates have increased dramatically as a direct result of cutting drug education programs. and your interlocutor replies as speaker  does: B: That's not even close to the truth. (Drug education has had no effect on crime rates. And, as a matter of fact, crime rates have actually gone down over the past 2 years.) together you will have provided me with the kind of case I am interested in here. The reply, That's not even close to the truth, is a comment on your asser­ tion. What is not even close is your proposition (the "content" of your state­ ment); what it's not even close to is the truth-of-the-matter; and the closeness it fails to manifest (for your interlocutor, anyway) is the degree of true-ness it fails to manifest. The assessor evaluates the first speaker's proposition with respect to its trueness, and offers this judgment by drawing on the spatial source domain of Distance.2 Propositions - true ones and not-so-true ones, even false ones - are located in space.3 Discourses can concern themselves with contesting the location of propositions. So, for example, the speaker's statement can be deemed

MICHELE EMANATIAN

134

(3) way off. Perhaps it is so inaccurate that (4) Nothing could be further from the truth! On the other hand, sometimes we deem a guess or generalization (5) Close! or, compared to previous claims (6) "actually closer to the truth" In shorthand then, Truth Is Location, and Relative Trueness Is Closeness to that Location.4 When a speaker makes a claim, ideally it will be a true claim. That is, it will manifest truth with a small "t": true-ness, accuracy, aptness, or reasonable­ ness. If truth (trueness) is the goal or destination, the act of making an assertion traces a trajectory over which the proposition-trajector travels, toward the desti­ nation, the truth about the matter at hand. Once we have a trajectory toward a destination, we can use the imagery of searching for, aiming for, chasing, and hitting or missing a target. There are variations on this theme for specific games, sports, races and other contests, and the language of winning and losing is also available. (7) Bingo! Bull's eye! Right on target with that one! You hit the nail on the head. That's way off target/the mark. Close, but no cigar. You're way off base there. That's not even in the ballpark, (from baseball) "I think we're closing in on it now",

(it = answer to the puzzle)

Notice that it is possible for the Distance predicate to have an animate sub­ ject, such that the location of the speaker, rather than the proposition, is as­ sessed. For example: (8) (A) You're not even close!

(B) You're way off!

(C) call-in radio show: M: Immigrants are a drain on the economy.

THE SPATIALIZATION OF JUDGMENT

135

[time passes] N:The last caller was way off; in fact, immigrants contribute much more to the Gross Domestic Product than they take from it. We can understand these cases if we acknowledge the various possibilities the Intellectual-Conversational Journey metaphor allows for its traveller ele­ ments: the thinker or speaker, the thoughts or statements themselves, or the en­ tire conversation or argument (see Emanatian, 1996ms). In the examples in (8) the speaker has journeyed with their proposition toward the intellectualconversational destination, the truth. These sorts of Distance metaphor judgments are found wherever the truth of what is being said is at issue. They are especially prevalent in academic or other problem-solving settings where people are making concerted efforts to get at the truth about something. In academic or other 'expert' settings higher regis­ ter versions of the Propositional Distance metaphor, such as (9), are common. (9) call-in radio show: A: Standardized tests clearly show that Bilingual Education is detrimen­ tal to student learning. B: That doesn't begin to approximate the truth.6 (cf. That's not remotely true.) As you may have noticed, there is considerable overlap between this Propositional Distance metaphor and the Intellectual/Conversational Journey metaphor reviewed in the introduction. While, in the Distance metaphor, the destination of the proposition-trajector is truth, in many of the Intellec­ tual/Conversational Journey cases there is no destination: the objective is simply to "cover" an area. In other Journey cases, though, the destination is a conclu­ sion, outcome, decision - that is, a judgment of some kind - which you try to reach. If an accurate judgment (the truth) is the final destination, then the lan­ guage of increasing or decreasing closeness coheres with that of journeys, and, in fact, you can arrive at the correct answer, or decide that someone's statement departs from the truth. Notice that Distance is part of what is brought along by the path and desti­ nation terms of the Event Structure metaphor. We can use distance notions even for purposeful behavior outside the realm of thought and discourse: "He applied for this job, and I came close to applying for the job he has. " Now, when we assess another person's statement by pronouncing it far from the truth or right on the mark, we seem to know - or think we know where truth is located, for that topic. How do we know this? Interestingly, the use of this Distance-from-the-Truth metaphor suggests the availability of two different models (or kinds) of truth: The assessor speaks, in such metaphorical expressions, as if she knows the absolute location of Truth,

MICHELE EMANATIAN

136

what all propositions are aimed at. (She needn't really believe she knows where the truth is, but may behave as though she does for rhetorical purposes.). An utterance like That's not even close, sets up as the standard of reference the Truth, in its fixed and knowable location, external and objective, speakerindependent. On the other hand, when we employ this metaphor, we use degrees of Dis­ tance to assess the true-ness of someone's statement. To the extent that a claim is felt to be inaccurate or unreasonable, it falls short of the Truth. A proposi­ tion, in being closer to or further from absolute Truth, has some measure or degree of Truth. Its trueness is relative in that sense. We've seen in the Propositional Distance metaphor, then, source concepts of Distance put to use in evaluative commentary on the trueness of a proposi­ tion. The closer to true a statement is, the truer we believe it to be. The Intellec­ tual/Conversational Travel metaphor overlies the Distance metaphor with the notion of Truth as destination. As seen in the examples in (7), the language of competitive games and sports supplies imagery to flesh out the schematic pro­ jected trajectory toward truth. The Propositional Distance metaphor reveals, or manifests, two kinds of truth: there's absolutely-located Truth, our destination when we make statements, and there's the truth (or "trueness") those statements have, a truth which comes in degrees. As Salmond points out, one of the values of using metaphors drawn from location and activity in the physical landscape is the "flexibility, ... capacity of ambiguity and rich resource for variation" (1982: 67). 2.2. Personal Distance: Distance of a thinker from the subject matter The second kind of spatial metaphor of judgment has to do with the location of a person with respect to some matter about which they have, or are expected to have, thoughts - and perhaps ultimately, speech. The assessor comments on the location of this person in terms of their Personal Distance from the matter at hand. For example: (10)

He's too close to the project; he doesn't even notice that it no longer has much support from the other guys.

Utterances like these are statements about the involvement of a person - call them "the thinker" - with something: an idea, a subject area, an event, a situa­ tion (here, a project). Such statements are instances of the general metaphor Involvement Is Physical Proximity. More specifically, in the expressions we are interested in here, being very close means being intellectually involved and being knowledgeable; thus:

THE SPATIALIZATION OF JUDGMENT

(11)

137

I've been in the middle of this stuff for years, and I know my way around, ("this stuff" being, say, ordering produce at a market, an area of expertise)

In shorthand, Intellectual Involvement Is Proximity. I mean "intellectual involvement" broadly: it entails familiarity with and knowledge of, and often, comprehensive understanding of. Familiarity usually means considerable time spent with, and sometimes commitment to. A full understanding of something can also engender appreciation for it. But metaphorical closeness can bring bias. As a person's distance from the matter at hand shrinks, so does their capacity for neutral judgment. In other words, (12)

Distance is required for objectivity.

Notice that, while this sentence can be "heard" as an instance of the use of the Involvement-Proximity metaphor, as a pronouncement on the nature of judgment, it is equally intelligible as an analytical statement about the concep­ tual metaphor. We use the Personal Distance metaphor to make assessments of bias quite routinely. For example, you bluntly urge your interlocutor to be more objective by saying (13)

Take a few steps back, buddy.

You notice that your own involvement in something is inhibiting good judgment: (14)

I've got to get some distance from this.

If you are apprehensive about a presentation you are to give, a friend might suggest: (15)

You're too close to the topic to be able to predict how it will be re­ ceived.

Part of the reason that Distance is required for objectivity is that this meta­ phor, linking Intellectual Involvement in something to Physical Proximity to that thing, coheres well with a pervasive visual metaphor in English. In this meta­ phor Mental Knowledge or Intellection maps onto Physical Sight (Sweetser 1990). Or, in the more familiar shorthand, Understanding (and/or Knowing) Is Seeing.

MICHELE EMANATIAN

138

Up to a point, closeness is necessary for sight, and distance inhibits it.. This is the gist of the metaphorical reading of (16), for example: (16)

I'm too far removed, these days, to see who the real players are.

On the other hand, if Knowledge (and/or Understanding) Is Seeing, then to get comprehensive knowledge you need a panoramic view. Objective knowing requires comprehensive seeing. By "comprehensive seeing" I mean seeing with appropriate scope and focus, to take in the totality of the object of vision. So, for instance, if you want to be able to speak confidently about a situation, you have to be able to see all of it, including where it borders on other situations. You might therefore have to take a few steps back. Too close, and you get only a partial view: (17)

You can't see the forest for the trees.

A partial view gives rise to partiality and bias. The objectivity needed for impartial judgment is absent. Sometimes the failure to get the whole picture is attributable to a metaphorical visual defect. If, for example, (18)

I think he's pretty myopic about it.

I am ascribing his lack of discernment in thinking or planning to his own traits. Suppose you are trying to revise a book-length manuscript. Typically you need to let a bit of time pass between the drafting and the revising because, as someone recently said to me, (19)

"You have to be able to see the thing in its entirety." (That is, you're too close to it to see its problems clearly.)

The basic idea, then, is that vision fosters knowledge, and therefore, good judgment. Sometimes you are too far away to make out the details, and some­ times you are too close to focus properly, or to get a comprehensive view. The relationship between Distance and understanding needs to be repre­ sented by a different cline for the Involvement metaphor than for the Under­ standing Is Seeing metaphor. The cline for Involvement is a straight line. The further away you are, the less involved you are.

THE SPATIALIZATION OF JUDGMENT

139

Figure 1. Involvement vs. Distance Accurate seeing, or optimal viewing, on the other hand, needs to be repre­ sented by a bell curve: for vision, you want to be neither too close nor too far.

Thus, despite the fact that they are largely mutually reinforcing, there seems to be an interesting non-fit of the Involvement-Proximity metaphor and the Understanding-Seeing metaphor. So far, excepting acknowledgment of the bias that comes with a partial view, I have presented a picture in which knowledge is the only prerequisite for good

MICHELE EMANATIAN

140

judgment. But that is not the case: a sufficiently high degree of Intellectual In­ volvement may implicate Emotional Involvement as well. This is the second reason Distance is required for objectivity.7 Emotional Involvement is understood partly in terms of a rather elaborate system of metaphors which includes Distance source concepts, a system de­ scribed by Jordan Howe (1993ms; cf. Sweetser 1993ms). Most generally, there is the now-familiar Involvement Is Physical Proximity. It is the application of this metaphor to the domain of emotion which is responsible for our knowing that the statement I feel really close to her. means 'I like her, am fond of her, love her, or feel intimate with her', and that Back off, will ya? can be a plea for more emotional "space". (Of course we also need to know the contextdependent valuation of involvement.) As may be seen in the examples discussed so far, degrees of Intellectual In­ volvement are expressible in terms of degrees of Distance. On the far end, there is infinity. Intellectually, you can be very far removed from something, or even light years away from it. You can be so distant that Boundary metaphors and In/Out metaphors become appropriate. Perhaps, for example, Your earlier pas­ sion, of Geomorphology, has become merely a peripheral interest. On the near end, the limiting case is Contact, or Co-location. Here there is overlap with Sense-of-Touch metaphors. You can be in touch with what's going on in your field, for example.9 If you couldn't be more involved, the Contact end of the Distance metaphor scale fades off into Sub-part and Identity metaphors. You can say of someone, for instance, that At that point in time, he was Mr. Free Trade Agreement or That guy lives and breathes A.I.10 Likewise, degrees of Emotional Involvement may be expressed in terms of degrees of Distance. The distant end of the scale may be associated with a lack of intimacy. For example, someone in the very same room with you may say (20)

I feel very far away from you right now.

articulating in Distance terms their sense that an important quality is missing in your relationship (e.g., good communication). On the close end of the scale, again, the limiting case is contact, or even merger (which may or may not be positively evaluated, as Howe points out). (21)

We're really in touch with each other's feelings lately.

(22)

This relationship is stifling, it's suffocating me. Give me room to breathe, I need a little space once in a while. Do you always have to be clinging to me? You've got to learn to let go, to back off a little. You don't always have to be on me about one thing or another. You can't have me all to yourself.

THE SPATIALIZATION OF JUDGMENT

141

You have to respect my personal space and not overstep your boundaries (Howe 1993ms: 4). 11 How are the two kinds of Involvement related to each other? We seem to understand Intellectual Involvement and Emotional Intimacy (Involvement) as being related to each other (not surprisingly) by degrees of Distance from what­ ever the object of thought or feeling is.12

Figure 3. Emotional and Intellectual Distance The system of Involvement-Proximity metaphors illustrates this connection between Intellection and Emotion. For example, for a person to be too close to something could mean that their Involvement with that thing is not merely Intel­ lectual, but verging on something Emotional. Similarly, too much Distance be­ tween a person and an event or idea may indicate a lack of interest or commit­ ment on the part of that person. If we remark that (23)

She's been at some remove from all the goings-on.

a listener might conclude that she has too little of herself invested in whatever it is to be counted on. If we begin in the Emotional realm of human relationships and increase the Distance (by breaking a connection), we move out of that realm: (24)

I can now look at her in a more detached way.

MICHELE EMANATIAN

142

This statement means you view her less emotionally, more rationally. Mo­ tion in the other direction is also possible, as (25) indicates. (25)

"... members of the dominant culture are unable to retain their sense of distance and separation from what they dominate. Instead over a period of generations, they increasingly develop direct ties to the 'new land' ..."

The following sentence (from a linguistics paper whose topic is not metaphor) speaks directly to the relationship between Emotional and Intellectual Involve­ ment, in metaphorical Distance terms: (26)

"Viewing one's own feelings objectively in order to verbalize them requires one to create distance from the feelings expressed."

Evidently then, we operate with a background model in which an Emotional relationship with something or someone puts you in greater Proximity to it or them than an Intellectual relationship does. Distance metaphors for Involvement suggest only a quantitative, not qualitative, difference between an Emotional relationship and an Intellectual one. Interestingly, this seems to be a case in which the language we use betrays one of our most culturally entrenched di­ chotomies.13 It is important to note, however, that it needn't be the case that greater and greater Intellectual Proximity becomes Emotional Involvement. Someone could insist, for example, (27)

Look, I've been in the middle of this stuff for years (but I'm not emotionally tied up in it); I could do the job.

Notice that a but clause could be included to explicitly counter the usual expectation.14 Being positioned at a large Distance from whatever matters concern you can be bad or good, as I've implied. It depends. In many cases, a minimal familiar­ ity is necessary to make a judgment. If you're too far removed to know enough about it, your ability to judge will be limited. But there are situations in which the optimal Distance between assessor and idea/event/etc. is quite large, in which the assessor has no Emotional Involvement whatsoever. Judges, for ex­ ample, in the ideal case are not involved emotionally, despite their intellectual expertise. (In fact, if they do have any personal stake in a case, they are re­ quired to excuse themselves.) The ideal juror is neither emotionally nor intellec­ tually involved in the matter or person being tried. He or she is intelligent and open-minded but ignorant of legal matters; the judge instructs on points of law. This need for Distance is expressed in the following sentence, about a man who

THE SPATIALIZATION OF JUDGMENT

143

served as an expert witness, despite the fact that he was one of the defendants in the case. (28)

"... Mr. Powell may well have been so close to the situation in all of its personally unpleasant aspects that he has to some degree lost his sense of perspective in the matter".

Likewise, ordinary people entering a contractual relationship with each other normally want to (29)

Keep it an arm's length agreement.,

that is, restrict their dealings with each other to business only. The Contact metaphor in this example has implications for judgment via the Understanding Is Seeing metaphor: non-contact suggests sufficient Distance for dispassionate re­ gard. In decision-making, whether in formal judgment or in everyday situations, objectivity - or impartiality - matters. In such cases the Closeness necessary for Seeing - that is, Understanding - may be at odds with the Distance necessary for Non-Involvement. If, in a particular situation, Involvement is desirable or necessary, and if Involvement is Proximity, then it behooves the person in question to attempt to move closer. Attempts to achieve Proximity are implicit in the way we talk about the difficulties encountered. In the Emotional realm, sometimes a person you would like to be close to is remote or unapproachable. In the Intellectual realm, while we don't have *unapproachable ideas or theories or plans, we do have inaccessible ones: Do you find Ingmar Bergman films to be accessible!15 To sum up, in the Personal Distance metaphor we have the notion of Intel­ lectual Involvement - of familiarity and understanding - couched in terms of a person's Distance from an object of intellection. We use this Proximity meta­ phor to make offhand evaluations of someone else's ability to judge, as a means of indirectly assessing their decision-making competence. We have seen that the Involvement-Proximity metaphor engages with the pervasive Knowing and/or Understanding Is Seeing metaphor, so that a position neither too close nor too far becomes the ideal location for Intellectual Involvement. On the other hand, because as English speakers we have available to us a pervasive understanding of Intellectual Distance in which it decreases at the risk of Emotional Involve­ ment, there are times when very close is the most appropriate location, and times when far away is the best place to be. The relative valuation of Proximity, then, depends on whether more value is attached to familiarity, comprehensive understanding, and commitment, or to impartial and objective judgment.

144

MICHELE EMANATIAN

3 . Conclusion I have described two different metaphors that provide us with Distance concepts for making judgments, and for making judgments about judgment.16 Though I have focused on Distance concepts as the source, equally implicated are loca­ tion, motion, orientation, and really, a whole "geometry" of the relations of thinkers to ideas, of talkers to other talkers, and of ideas to other ideas. But ex­ amining even this small piece of how we spatialize notions like truth and intel­ lectual involvement has helped to reveal some of the culturally entrenched models that situate the metaphors. It would seem that we hold two relatively inconsistent views of truth, at least as suggested by the Propositional Distance metaphor. That we function well enough with mutually conflicting conceptuali­ zations is not news (see for example Kay 1983), but it is still interesting. What, then, is truth for ordinary people? Analysis of conceptual metaphors has illumi­ nated this question, if not its answer. The Personal Distance metaphor is comparably revealing of folk views of emotion and intellection, and of expertise, objectivity, and judgment. Analysis suggests that in many situations of everyday life, a person's ability to judge hinges on their positioning themselves at some ideal distance at which their un­ derstanding is deep and comprehensive but unbiased and untainted by feeling. A tall order. Naturally, one question which arises concerns the relation between Propo­ sitional Distance and Personal Distance. Strictly speaking, I suppose, the Dis­ tance of a proposition from the Truth, and the Distance of a thinker from some subject matter are conceptually independent of one another. Yet we do seem to believe that the closer a person is to something, the closer their proclamations about that thing will be to the Truth. In other words, Intellectual Involvement has positive implications for the trueness of statements issued. That said, it appears that the Distance and Intellectual/Conversational Jour­ ney metaphors themselves indicate otherwise. To aim for the truth, to try to get at it, to attempt to approximate it, and so on, suggest that, at least at first, the thinker-speaker is separated from the Truth, no matter what their Involvement. Perhaps it is expertise - Intellectual Closeness - that closes the gap between the proposition's location and the location of Truth.

Acknowledgments I appreciate helpful comments on this work offered by Eve Sweetser, Joe Grady, Jordan Howe, George Lakoff, and Alan Cienki. I'm especially grateful to David Delaney for the long conversations, the good ideas, and for convincing me that this topic was worth pursu­ ing.

THE SPATIALIZATION OF JUDGMENT

145

Endnotes 1. Throughout this paper examples presented in quotes are attested; others are not. 2. I am using the term "speaker" to cover both speakers and writers. Likewise, "conversational" is here sloppily intended to include many cases of written English which seek to engage the reader in intellectual "dialogue". 3. In the metaphorical expressions under discussion propositions are objects located, as opposed to themselves being locations. We aim our prorx)sition-objects toward a destination, truth, and they follow a trajectory, (see below.) It should be noted that there is another metaphor in which propo­ sitions seem to be locations for truth (in particular, containers, as in That statement has a certain amount of truth (to it), or There's some truth to/in that.). These two metaphors are duals in the sense of Lakoff 1993. 4. Another metaphor, or set of metaphors, which draws on the source domain concept of distance (in this case on a scale) might be abbreviated Similar Is Close, as instantiated in: A: Is that what you were looking for? B:

It's close, but I wanted something a little dressier. "I'm not in anywhere near as good a mood as I sound".

A: Kronos is the best quartet on the planet! B: (They're) not even close! (cf. Propositional Distance: That's/You're not even close!) "No, we don't eat cookies for breakfast, but we eat the nearest possible thing". As to how to best capture the relationship between the Propositional Distance metaphor and this one, the jury is still out. For the purposes of this paper however, I will consider them to be sepa­ rate but related. Given the cultural importance of conceptions of truth for epistemology, under­ standing, and communication, and the elaborateness of ways of talking about judgment, the Propositional Distance metaphor will be credited with a Ufe of its own. 5. Alternatively, we could say that metonymy is at work in cases like (8): the speaker may stand for the proposition. I prefer the metaphor analysis because there is plenty of evidence outside Distance metaphors proper that we systematically see the thinker-speaker as journeying. See (1) for exam­ ples. 6. Approximate means 'get proximal, or close, to'. 7. In speaking as if Intellectual Involvment and Emotional Involvement are distinct kinds of things, I am merely reporting on the reigning dichotomization of the emotional (or more broadly, nonrational) and the intellectual (rational, cognitive) in American culture. Describing this construct should in no way imply that I share or endorse it. 8. By "emotional" involvement with X, I mean a relationship to X which agitates feelings other than those arising through conscious mental effort. 9. There are also Connection metaphors: connection, ties, links, bonds, to be wed to an idea, to stick firmly to an idea, etc. 10. Sometimes there are difficulties with being so close: He's just so wrapped up/tangled up/enmeshed /tied up in this metaphor stuff that he's lost sight of the point. Note coherence with the visual metaphor: too close and the view is blocked. 11. There are Weight and Boundary metaphors mixed in with Distance, Contact, and Merger meta­ phors here.

146

MICHELE EMANATIAN

12. There are terminological difficulties here. The target concept label "Intellectual Involvement", selected to describe metaphorical closeness to ideas, events, etc., itself carries overtones of emo­ tion. That is, to be intellectually involved is to have more than a dispassionate interest, more than a disinterested understanding, as the following example implies. "I am afraid to say what I say ... because of my reader. More likely than not, you are not a Palestinian woman. Are you situated close to ideas about an outpost of Western Democ­ racy and Civilization, that shining star in the midst of darkness - Israel?" It is often the case in analyses of this sort that a metaphor undermines a dichotomy (here, intellec­ tion vs. emotion) which nevertheless must be employed to explicate its demise. 13. The Sub-part and Identity metaphors exemplified earlier (Mr. Free Trade Agreement; lives and breathes A.I.) have no implication of Emotional Involvement, though we might expect that state­ ments about the close end of the scale would. Notice, though, that the metaphors involved are not Distance metaphors; they constitute an alternative. The Distance model is not always the one em­ ployed. 14. We also know of cases in which the implications for Proximity carried by the use of the Under­ standing Is Seeing metaphor seem to conflict with the model of Emotional and Intellectual relation­ ships presented above. We seem to feel that looking at something/someone too closely disables any Emotional relationship we may have with it/them. If we bring our analytical powers to bear on a person we love, or on our sex life, or on any other aspect of life in which non-rationality domi­ nates, it then becomes an object of intellection (almost by definition). In other words, if we scruti­ nize, we endanger our feelings. We move out of the Emotional realm, toward the Intellectual realm. Though this may be the outcome of a person looking closely, it seems to me unlikely that in such cases the assessor would be said to be too close to the object of thought or feeling. That is, looking closely at something is not the same as being close to it, despite the fact that the former should entail the latter. This remains a puzzle. In any case, we seem to have more than one cognitive model available in the culture for conceptualizing such matters. 15. The word access is from OF acces 'arrival'. It is not surprising to meet up with the Intellectual Journey metaphor again in this context. Notice that there is a difference in perspective afforded by the terms of the Distance metaphor vs. the Journey metaphor; compare We've been working on it for months, and we can only get so close with We've been working on it for months, and we can only get so far. The former expresses the limits of progress in terms of Distance from the goal, while the latter expresses progress in terms of Distance measured from the point of origin on the path (cf. Lakoff 1993: 221). 16. It would be interesting to examine, as Salmond (1982) does for the Intellectual Terrain and Journey metaphors in Anthropology, the 'expert use' of Distance metaphors in analyses of linguistic phenomena. Frawley, for example, in his deictic theorization of epistemic modal­ ity, enlists "remoteness" of the source of knowledge from the object of knowledge, in at­ tempting to provide a scalarization of such dimensions as speaker confidence, strength of evidence, and speaker commitment (1992: 413). To take another example, Lakoff seems to draw on everyday English uses of Emotional Distance in her typology of politeness systems (which includes "Distance Politeness", 1990: 35-36); in her earlier classification of discourse styles (1979); and in analyzing the pragmatics of subordination ("distancing cultures", 1984). But close scrutiny of linguistic ideologies is, as they say, beyond the scope of this pa­ per.

THE SPATIALIZATION OF JUDGMENT

147

References Emanatian, Michele. 1996ms. Intellectual and Conversational Paths. Talk presented at Conceptual Structure, Discourse, and Language II; Buffalo, NY. Frawley, William. 1992. Linguistic Semantics. Hillsdale, NJ: Lawrence Erlbaum. Howe, Jordan. 1993ms. Big Love Paper. Kay, Paul. 1983. "Linguistic competence and folk theories of language: two English hedges," BLS 9: 128-137. Lakoff, George. 1993. "The contemporary theory of metaphor." In A. Ortony (ed.), Metaphor and Thought, 2nd ed. Cambridge: Cambridge University Press, 202251. Lakoff, Robin. 1979. "Stylistic strategies within a grammar of style." In J. Orasanu et al (eds), Language, Sex, and Gender. New York: Academy of Science. Lakoff, Robin. 1984. "The pragmatics of subordination," BLS 10: 481-92. Lakoff, Robin. 1990. Talking Power. New York: Basic Books. S almond, Anne. 1982. "Theoretical landscapes: On cross-cultural conceptions of knowledge." In D. Parkin (ed.), Semantic Anthropology. (Association of Social Anthropologists Monograph 22) London/New York: Academic Press, 65-87. Sweetser, Eve E. 1990. From Etymology to Pragmatics: The Mind-as-Body Metaphor in Semantic Structure and Semantic Change. Cambridge: Cambridge University Press. Sweetser, Eve E. 1992. "English metaphors for language: Motivations, conventions, and creativity." Poetics Today 13.4: 705-724. Sweetser, Eve E. 1993. "Inside/outside, upside down" - Co-orientation of metaphori­ cal mappings in literary and everyday language.

Stop Making Sense! Metaphor and Perspective in Creative Thinking Sessions of Scientists and Scientific Radio Broadcasts Wolf-Andreas Liebert University of Trier, Germany

1. Introduction "A considerable amount of work needs to be done before we achieve a detailed understanding of these processes" (van Dijk & Kintsch 1983: 314-315). This statement concerning the study of metaphors in discourse is still valid today. Despite a large amount of literature on metaphor, detailed studies of metaphors in discourse or verbal interaction are yet outstanding (cf. Glucksberg 1989). Even recent publications (cf. Indurkhya 1992) discuss this lack of re­ searched material. This gap is quite surprising, since we might expect new insights into ongo­ ing cognitive processes while people are negotiating metaphoric models. Studies of cognitive semantics suggest that conceptual systems are already present, whereas studies of verbal interaction focus on the development of conceptual systems. Accordingly, participants in talk begin with a shared conceptual system which includes metaphors. However within a verbal interaction, such as a crea­ tive thinking session or a radio talkshow, a dynamic change and interchange of the participants' conceptual systems can be observed, appearing as on-line proc­ essing of the development of metaphor models, how speakers build them up, cross link or reject them. Are there structures of order within the situationally determined chaos of our talk? This paper discusses basic discourse patterns to show the emergence of metaphor models in verbal interaction. From a discussion of different theories of metaphors, problemsolving and perspective, basic units for the analysis of meta­ phors in verbal interaction are developed. These include a two-turn dialogue unit consisting of pragmatic operations of initiative and response, and semantic op­ erations of expansion and adaptation of metaphors. These operations will be applied to the talk of scientists who are discussing the metaphors in their own research topics, and to radio talkshows on science.

150

WOLF-ANDREAS LIEBERT

2 . Theoretical framework The introductory remarks show that there is no "ready to go" theoretical frame­ work for the analysis of metaphors in verbal interaction. From the point of view of cognitive semantics, metaphorical reasoning could be considered a process whereby speakers begin with a metaphor model which they expand while continuing to project metaphorical concepts from source domain to target domain. The process is complete when the goal of solving a problem or of finding a good explanation is reached. Indeed things are even more complex. At least the following operations have to be considered in the study of metaphors in verbal interaction: (1)

Pragmatic operations In verbal interaction, there is no such straight "thematic metaphor pro­ gression", as can be found in text of a single author. Instead, there are search procedures, changes of source domain, etc., by two or more speak­ ers. In some cases, speakers enter into a coherence-generating proc­ ess of give and take of metaphors, and thus reach a certain flow of emo­ tion (cf. Csikszentmihalyi 1990) as well as thematic progression (cf. Chafe 1994).

(2)

Semantic operations There are not only expansion principles at work in verbal interaction, such as the generation of new metaphors, but also, complementary to them, adaptation principles. While expansion principles govern the projection of new concepts from the source domain to the target domain, adaptation principles deal with the transformation of these concepts within the target domain (e.g. by specifying or limiting).

The many aspects which are involved in the investigation of metaphors in verbal interaction deem it necessary to consider approaches from a number of different disciplines. These include central concepts of cognitive and social psy­ chology (problemsolving, analogical reasoning, perspective), conversational analysis (turn-taking, aspect use), and cognitive linguistics (levels of metaphor models). 2.1. Problemsolving and analogical reasoning Scientific discussions may be regarded as processes of problemsolving. If meta­ phors are involved, these processes may be regarded as analogical problemsolv­ ing (cf. Falkenhainer 1987; Gentner & Gentner 1983; Collins & Gentner 1987; Gentner & Jeziorski 1993; Gilhooly & Keane & Logie & Erdos 1990; Keane 1990a, 1990b; Hesse 1991).

STOP MAKING SENSE!

151

According to analogical reasoning the whole process of metaphorical reasoning can be divided into three main phases: I. II. HI.

Commencement in the target domain either with a problem to be solved or a topic to be explained Entrance into the phase of metaphorical reasoning Development of solutions or explanations for the problem

When connecting these phases to verbal interaction, it must be considered that this is not a strict sequence. While some speakers are finding solutions or explanations by metaphorical reasoning, others may reject these solutions. An­ other speaker may then return to the definition of the problem, perhaps reformu­ lating it, etc. Nevertheless, each phase can be characterized by its own opera­ tions. For Phase I (problem setting in the target domain) the following processes may be assumed (Hesse 1991): -

Formulation of a problem in the target domain Reformulation of the targeted problem Focus on certain aspects and features Abstraction from these aspects and features

For Phase II, Hesse elaborates a model of analogical problemsolving, but neglects the investigation of metaphors: Metaphors represent an important special form but they shall not be considered here due to their linguistic components (Hesse 1991: 14; translation by W.-A. Liebert). Therefore a model for metaphors in verbal interaction, Phase II2 has yet to be developed. Hesse's statement indicates that we must link different disciplines to better understand the phenomenon. An integrated approach of psychology and linguistics is of great value and will be presented below. 2.2. Perspective, turn-taking and aspect use The following approach was developed within the Sonderforschungsbereich 245 (SFB 245) "Language in Social Context".3 The SFB 245, an interdisciplinary research program financed by the German Research Foundation, was working between 1989 and 1997. One of the projects, project Bl, developed an inte­ grated theory of linguistic perspective and put forward a qualitative framework of analysis for thematic aspect use in verbal interaction (cf. Graumann 1989,

152

WOLF-ANDREAS LIEBERT

1990; Graumann & Sommer 1988; Rettig, Sommer, Kiefer & Graumann 1993; Sommeretal. 1994, 1995). In their approach they refer to the "ego-logical" model of perspectivity which was previously elaborated by Graumann (1960). Graumann's model is called "ego-logical" because "in focusing on the relationship between individual and object world, it neglected the interindividual relationship" (Sommer et al. 1994: 4). [Graumann's model] starting from such philosophers as Husserland and MerleauPonty, presented a psychological theory of perspectivity that stressed the inten­ tional person/world relationship and the body-centered nature of experience. Graumann's theory states that experience always takes place in situations struc­ tured by a given perspective: from different viewpoints, individuals perceive dif­ ferent aspects of an object (or event); they structure the object differently, and they refer to different contexts (Sommer et al. 1994: 3). The authors also refer to the interactionist approach of Mead (1959, 1962) who introduced the term "perspective taking". According to Mead, perspective taking is a "precondition of communication" (Sommer et al. 1994: 4). Based on the approaches of Graumann and Mead, a theory of linguistic perspectivity was developed: The theory of linguistic perspectivity can be considered as a combination of the egological and the interactionist approaches to perspectivity. It is based on the egological assumption that perception, cognition, and behaviour always take place in situations structured by a given perspective. From different spatiotemporal or cognitive points of view persons perceive different aspects of an ob­ ject. They structure the object differently, and they refer to different contexts. This holds not only for visual but also for cognitive objects, i.e. for imagined objects or abstract issues. 'Point of view', here, can be every definable position in spatio-temporal as well as in cognitive space (Sommer et al. 1994: 6; cf. Graumann 1993: 159; Rettig et al. 1993). The perspective a person takes "is also reflected in the linguistic represen­ tation of the object. One can identify linguistic characteristics in speech that indicate a speaker's perspective" (Sommer et al. 1994: 6). Their task was to put forward the idea of verbal interaction as a "dynamic interaction of perspective setting and perspective taking" (Sommer et al. 1994: 15) through qualitative analyses. Negotiation on aspect relevance was analyzed because the theory of perspective assumes that the development of one perspective can be shown by the progression of aspects: Each speaker introduces those aspects of the theme he or she considers as rele­ vant from his or her point of view. Moreover he or she has to deal with the as­ pects brought forward by the opponent. If communicating is to work on a com­ mon perspective - as it is assumed here - both discussants have to negotiate

STOP MAKING SENSE!

153

which aspects relate to the overall topic and which do not (Sommer et al. 1994: 15-16). Sommer et al. (1994) distinguished four basic types of interactive aspect use; successfully applied to a corpus of adversarial discussions on immigration (Sommer et al. 1994: 17-23; cf. Rettig et al. 1993). The types were established as follows: 1. 2. 3. 4.

First introduction of a new aspect into the discussion Acceptance of an aspect Rejection of an aspect Re-introduction of an aspect

The focus of this approach on the embodiment of language and cognition fits perfectly into the approach of cognitive linguistics which also assumes the embodiment of cognition, especially the embodiment of metaphorical reasoning (Lakoff 1987, 1993; Johnson 1987). The approach to perspective may also be combined with metaphor theory: in cognitive linguistics the selection of one metaphor as opposed to another is seen as the highlighting of certain aspects of an object and the downplaying of others. This also fits into the approach of perspective as elaborated by Graumann (1960, 1989, 1990), Rettig et al. (1993) and Sommer et al. (1994): The selection of a metaphor would indicate the use of a specific perspective. In ver­ bal interaction the second speaker would have the choice of taking up this meta­ phor to develop it further, or of rejecting this metaphor and selecting another, thus trying to put forward another perspective. 2.3. Shared knowledge and metaphor The term "metaphor" is a complex notion with simultaneous meanings on dif­ ferent levels (cf. Liebert 1992, 1995a): - the lexical level (which includes single concepts) - the conceptual level (cognitive models) - the domain level These are important distinctions for the analysis of metaphors in verbal in­ teraction, because speakers are always negotiating at different levels of meta­ phor. When a speaker says, for example, "the cell is a tube roller factory", he introduces a cognitive model THE CELL IS A TUBE ROLLER FACTORY, and the source domain ECONOMIC SYSTEM. Another speaker may introduce another cognitive model, say THE CELL IS AN INFILTRATED COMPANY, while he takes up the source domain ECONOMIC SYSTEM introduced bv the

154

WOLF-ANDREAS LIEBERT

previous speaker. A metaphor is thus always discussed at all three levels, even if meta-linguistically we speak only of "a" metaphor. In this study's transcript analyses, all three metaphoric levels are considered. Finally, let us relate the term "shared knowledge" to metaphors. "Shared knowledge" was a key word of the theoretical discussion within the SFB 245. In verbal interaction "shared knowledge" refers to the tacit knowledge shared by speakers. Shared knowledge becomes observable when speakers refer to it, or in cases of misunderstanding. Therefore, the introduction of a metaphor may make explicit what was tacit knowledge in the conceptual system. Scientists live in their own world of discourse, but they share the same "Lebenswelt" with non-scientists, as well. They may thus not only refer to the conceptual metaphoric network of their science, but also to that of everyday life. For example political or social developments reported in the press, may be in­ troduced as a source domain for new ideas in virology. Case study 3 will pres­ ent this case. On the other hand, there are "left-overs" of metaphors within scientific language. Regarding these as isolated "fixed expressions" or even as "dead metaphors" is not appropriate because systems of metaphors were found at the core of many scientific theories (cf. Liebert 1995a, 1995b; Jäkel 1995, 1996; Knudsen 1996). Even abstract concepts such as TRANSPORT may be regarded as metaphors taken from everyday life. This kind of "protometaphor" (Liebert 1996b) becomes vivid at once when scientists are generating hypotheses for un­ solved problems, or when science is explained to non-scientists (see case studies 4 and 5). Within this interconnected network of discourse, metaphor models obtain a transdiscoursive function which goes beyond the emergence of social represen­ tation (Moscovici 1981), because metaphor models form an interconnected net­ work which crosses all text and discourse types (Liebert 1996b). 2.4. Basic discourse patterns of metaphorical reasoning in verbal interaction All the aspects that were discussed above shall now be integrated into an ap­ proach which shall be applied to analyzing metaphors in verbal interaction. 2.4. L Pragmatic operations: initiative and response Based on the model of Sommer et al. (1994) I suggest that there are two basic operations of metaphorical reasoning in verbal interaction: an initiative and re­ sponsive operation, each containing the following two suboperations: I. Initiative operations 1. introduction of a metaphor 2. re-introduction of a metaphor

155

STOP MAKING SENSE!

II. Responsive operations 1. taking up an introduced metaphor 2. ignoring an introduced metaphor These correspond with the model of Sommer et al. (1994) in the way shown in Table 1. Table 1. Pragmatic operations of aspects use applied to metaphors in discourse

Pragmatic operations of aspect use

Pragmatic operations of metaphor use

(Rettig et al. 1993; Sommer et al. 1994:)

(Liebert)

first introduction of a new aspect into the discussion re-introduction of an aspect 0

1. introduction of a metaphor 2. re-introduction of a metaphor 3. taking up a metaphor

acceptance of an aspect

3.1 acceptance of a metaphor

rejection of an aspect

3.2 rejection of a metaphor

ignoring an aspect (included under | "rejection")

1

1 1

4. ignoring a metaphor

These operations should be understood as follows: Introduction of a metaphor This refers to the first introduction of á metaphor into a discussion. Included is the first introduction of concepts and submodels of metaphoric models if the metaphor had not yet been introduced by another speaker at any level of the metaphor (see 2.3. 'Interacting at different levels of metaphor'). Like in Som­ mer & Rettig's approach it should be emphasized that this operation applies to the first introduction only. If a speaker introduces metaphorical concepts, mod­ els, or submodels to a metaphor which has already been introduced into the dis­ cussion this operation should be regarded as an operation of expansion (see be­ low). Re-introduction of a metaphor A metaphor is re-introduced by one speaker if it has been introduced by the same speaker before, but was rejected or ignored.

156

WOLF-ANDREAS LDEBERT

Taking up a metaphor A speaker takes up a metaphor if he refers to the metaphor uttered by another speaker. There are two choices: to accept or to reject the metaphor. Ignoring a metaphor A speaker ignores a metaphor if he does not take up the metaphor uttered in the previous turn. Obviously initiative and response operations form a two turn sequence by different speakers; an initiative action of a speaker Sj and a responsive turn by another speaker Sj can be represented in the following manner: <SI: initiative, Sj: response> Let's call this minimal sequence the "dialogue unit". The initiative turn contains the possibility of introducing or re-introducing a metaphorm whereas the responsive turn alows the other speaker to take up or ignore the introduced metaphor m : <SI: {introduce metaphorm, re-introduce metaphorm}, SJ: {take up metaphorm, ignore metaphorm} > If speaker Sj takes up the metaphorm introduced by speaker S. he has two choices. Sj may accept or reject metaphorm. <S[: {introduce metaphorm, re-introduce metaphorm}, Sj: {{accept metaphorm, reject metaphorm}, ignore metaphorm} > 2.4.2. Semantic operations: expansion and adaptation In addition to the turn oriented operations discussed above, some semantic op­ erations must also be considered. There are two basic semantic operations in metaphorical reasoning in verbal interaction: expansion and adaptation. At least the following semantic operations and suboperations must be con­ sidered: 1.

Expanding the metaphor 1.1. addition of a new metaphorical concept 1.2. addition of a new metaphorical submodel 1.3. generation of a scenario 1.4. superimposition of two or more metaphor models taken up from different turns

STOP MAKING SENSE!

2.

157

Adapting the metaphor 2.1. limiting of the metaphor to certain aspects 2.2. super imposition of different metaphor models 2.3. drawing conclusions in the target domain 2.4. specification of the metaphorical concepts with target elements 2.5. explanation of mappings (to profile component models) These operations shall now be explained briefly.

The expansion operations The first two cases cover the operations of adding one or more metaphorical concepts or submodels to an already introduced metaphor (see 1.1 and 1.2.). The suboperation of scenario generation (see 1.3.) is performed if a speaker not only introduces an additional concept but generates the whole sce­ nario which belongs to the metaphor introduced or taken up before. The case of taking up a model and superimposing it on an introduced model (see 1.4.) covers the merging of a metaphor from the previous turn with the metaphor from a former turn. The adaptation operations The complementary principle to the expansion of metaphors is the adaptation of metaphors. It is well known that metaphors only apply to a certain degree (cf. Lakoff & Johnson 1980). Speakers try to reach the most appropriate metaphor ization degree while adapting introduced metaphors. This can easily be observed in verbal interaction and analyzed in transcripts. Usually, a speaker limits an introduced metaphor to certain aspects (see 2.1.), in other words, he restricts the metaphor's range within the target do­ main. This may be reached also by superimposing different metaphor models (see 2.2.). Perhaps it seems confusing that superimposing different metaphors is found as a suboperation of both expansion and adaptation. However, as super­ imposing may be applied to broaden the imagination, it can also be used in combination with such limiting phrases as "yes, but only if ...". In conversa­ tional analysis this has been called "focus opposition" (cf. Kallmeyer, i.p.). One example may be "Yes, the DNA is a text, but only if the text of the DNA is seen as a software program". A speaker draws conclusions in the target domain (see 2.3.) to show the consequences of an introduced metaphor and hence to limit the metaphor's range. Adaptation of the metaphor can be reached if the speaker specifies intro­ duced or taken up metaphorical concepts with target elements (see 2.4.). In this case, new metaphorical lexemes are introduced, for example the multiword lex­ eme "gene taxi", an exocentric construction where the head "taxi" is an element of the source domain TRANSPORT which is determined by "gene", an element of the target domain GENETICS.

WOLF-ANDREAS LIEBERT

158

A similiar adaptation operation is the explanation of source-to-target mappings (see 2.5.): Which areas of the target object does the metaphor apply to? Do we need other metaphors for other areas of the target object? When answering these questions and explaining the mappings, speakers make hidden aspects of the target object explicit or they even might create new ones. According to Collins and Gentner (1987) these processes may be understood as profiling component models. 2.4.3.

Metaphor models in verbal interaction - units ofpragmatic and semantic operations After having introduced and discussed pragmatic operations (initiative and re­ sponse) and semantic operations (expansion and adaptation), these operations shall finally be connected. Can a speaker perform every semantic operation after any pragmatic operation? Obviously, they do not connect in all cases. No se­ mantic operation will be performed if an introduced metaphor of one speaker is ignored by another speaker. But in all other cases speakers may perform expan­ sions and adaptations to the metaphor introduced by the other speaker. Thus a speaker may perform the semantic operations of expanding or adapting metaphor m in either the initiative turn or in the responsive turn.4 The process of metaphorical reasoning in verbal interaction may hence be described as follows: The basic discourse pattern of metaphorical reasoning in verbal interaction is the dialogue unit. The dialogue unit consists of two turns, an initiative turn of a speaker S. and a responsive turn of a speaker Sj. In his initiative turn Sj may introduce or re-introduce a certain metaphor (metaphorm). After having introduced metaphorm he may expand or adapt it by operations of limitation, drawing inferences, etc. The initiative turn of S[ is followed by the responsive turn of Sj. Sj has the opportunity to take up or to ignore the metaphor introduced by Sj. Regardless if he accepts or rejects the introduced metaphorm he also may expand or adapt metaphorm by the operations of limitation, drawing inferences, etc. In summary, these pragmatic and semantic operations form an enhanced dialogue unit in metaphorical reasoning. The simple dialogue unit consists of a pair of two turns: <S¡: initiative, Sji response > The enhanced dialogue unit consists of a pair of two turns whereas each pragmatic operation is followed by a semantic operation. Thus the basic dis­ course pattern of metaphorical reasoning is a two turn unit of initiative and re­ sponse whereby a speaker may perform all semantic operations after the prag­ matic operations, if he choses not to ignore the uttered metaphor:

STOP MAKING SENSE!

159

< <S¡: initiative, semantic operation >,<Sjt response, semantic operation> > Finally, it should be emphasized that these categories are analytical categories, which may help to analyze metaphorical reasoning in verbal interaction. They do not provide a one-to-one representation of metaphorical reasoning in discourse. A pragmatic operation and a semantic operation are sometimes not performed in a sequence, one after the other, but simultaneously. However, the analytical categories only made it possible to analyze this congruence. The fol­ lowing analyses will demonstrate how these categories of turn-oriented prag­ matic operations and semantic operations apply to verbal interaction. 3. Case Studies of Metaphorical Reasoning in Verbal Interaction The following case studies analyze transcripts5 of creative thinking sessions of scientists who reflect on their own metaphors, and radio talk shows on science. Before perfoming the analyses there are some data acquisition problems which should be addressed. 3. 1. Strategies of data acquisition of metaphorical reasoning There are two main strategies for data acquisition when observing metaphors in verbal interaction. The first strategy deals with the elicitation of metaphors, i.e. by requesting the production of metaphors for a specific task. The second is to analyze the use of metaphors in authentic verbal interaction. Settings in which metaphors were elicited provide a great density of meta­ phorical reasoning processes which may be assumed to occur under natural conditions also (cf. Gordon 1961; Schoen 1979), whereas authentical settings can show if the analyses are valid. Thus, we should study both elicited metaphors and metaphors in "natural" interaction. Therefore, the following case studies contain analyses of interactions where metaphors were elicited (case studies 1-3) and analyses of interactions where metaphors were not elicited (case studies 4-5). 3.2. The structure of the case studies The case studies are each separated into four sections. First section: previously introduced metaphor models Two problems appear when linguistic analyses of verbal interactions are pre­ sented. First, no paper of this size can show full-length analyses. Therefore, it is necessary to select limited passages of the transcripts. When selecting excerpts, the second problem occurs: Explanations of which metaphor models had previ-

160

WOLF-ANDREAS LIEBERT

ously been introduced become necessary. Case studies 1-3 will begin with the appropriate metaphor models and explanations. Second section: transcript in German The following section of the case study is the transcript in German. All talks were transcribed according the guidelines of the SFB 245 (cf. Gutfleisch-Rieck étal. 1989). Some basic notation features are arrow up /down intonation up/down capital letters stressed syllables or other units underlined words overlapping turns of two or more speakers : acoustic expansion * pause of 1 to 1,5 seconds / re-starting the phrase after making a mistake (CAPITAL LETTERS) #...# Paralinguistic feature (for example LAUGH­ TER) followed by it's range Third section: English translation This section is followed by a translation into English. Here we have to deal with the problem of an accurate translation of verbal interaction. A whole different project would have to deal with the subject of translation of filler-words and mistakes. Therefore the English translation will not attempt to give a one-to-one translation containing intonation marks, etc., but will try to be as close as pos­ sible to the sense of what the speakers said. It leaves out parts of the German original and is more word-by-word than an idiomatic translation. Fourth section: linguistic analysis The last section focusses on the linguistic analysis of the metaphorical reasoning processes within the aforementioned framework. All analytical categories con­ sisting of the pragmatic and semantic operations are set in bold face. 3.3. Metaphorical reasoning in science (case studies 1-3) 3.3.1. Research in context The first three case studies are microscopic analyses of metaphorical reasoning in science. This work was done in the so called TLMSF-project.6 The TLMSF-Project raised the question of what results will be received when scientists reflect on their own metaphors (Liebert 1995b). This project analyzed the conceptual system of virologists working in the German Center for Cancer Research (Heidelberg). The virologists under investigation worked on the inhibition of the AIDS-virus HIV. Their cognitive metaphor system was analyzed (Liebert 1995a, 1995b) and assembled in a hypertext lexicon (Liebert

STOP MAKING SENSE!

161

1994, 1996a). In a final step, the virologists reflected on their own metaphor models at a workshop. During the project a corpus was built up which consists of the following components: (1) (2) (3) (4)

10 authentic work meetings of the virologists (~20 participants, 1.5 hours each) A guided discussion of a group of 20 virologists about the meaning of sci­ ence 1 hour) Interviews with 12 virologists (1.5 hours each) Three sessions of reflection on metaphor models with 6 virologists (1.5 hours each).

The corpora were transcribed and analyzed according to the methods of Lakoff & Johnson (1980) and Lakoff (1987). The metaphorical reasoning of the virologists at the workshop (subcorpus (4)) was elicited in the following way: After having reconstructed their meta­ phorical system, it was installed in the authoring software ToolBook as "Lascaux", a hypermedia lexicon of metaphor models. It was named "Lascaux" because it modeled after the famous Lascaux Cave near the Dordogne Valley in France. The user interface metaphor of the cave and the mosaic were selected because the analysis of the subcorpus (2) concluded that the scientists concep­ tualized science as the assembling of a mosaic, and as looking into the dark with a flashlight8 (a detailed presentation of the construction of the user interface metaphor "Lascaux" from subcorpus (2) is described in Liebert (1996a)). In the hypermedia version of "Lascaux", the user can click on a large number of but­ tons, visual objects, or hotwords. If he clicks for example on "research tasks", he finds a wall decorated with a visual version of the HIV-infected human cell. If he clicks on a rectangle he zooms in closer to the wall. After pressing the metaphor button he "walks" into another cave, the metaphor cave. The meta­ phor cave is central for the elicitation process: here, he finds a number of hotwords again. Every "hotword" represents a source domain which was found in the corpora. Clicking for example on the hotword "transport", the user again moves closer to the wall and sees an elaborated view of the selected source do­ main TRANSPORT. Here, he sees an interactive visualized transport frame with images of transport vehicles (taxi, airplane, train etc.), models of transport, maps for the different vehicles, or appropriate word fields. The user of "Lascaux" then combines his own transport model, say a swimming train or a computer-controlled shuttle, and projects the selected transport scenario onto an empty Lascaux wall, where he is free to arrange or edit the selected images and word fields as he likes. By constructing a mosaic of metaphorical pebbles the user can begin to understand from a new angle what goes on in an HIV-infected cell.

162

WOLF-ANDREAS LIEBERT

Lascaux was presented to a group of six of the scientists during a weekend workshop where they reflected on their own metaphors and built new ones. Two sessions with Lascaux, lasting 1 1/2 hours each, and a third session without Las­ caux, of equal length, guided each by a moderator, were recorded on videotape and transcribed. The results of these sessions are described in greater detail in Liebert (1995a, 1995c, 1996a). The following excerpts are drawn from the last session. Scientific speakers are abbreviated with "S + Number", for example S6, while the moderator is "M". 3.3.2.

Case study 1: Incorrect proteins - to repeat a grade or to be discarded in the garbage bin

Metaphor models previously introduced DIFFERENTIATION OF HUMAN CELLS IS LEARNING IN SCHOOL (school model) Transcript S11: du kannst eigentlich nur durch diese */durch dieses modifikation und instruktions(grundlagen) DURCHlaufen du kannst entweder richtig oder FALSCH ablaufen * und wenn du dann die richtige äh information und kodierung mitbringst * dann wirst du in die gruppe integriert werden wenn du sie nicht mitbringst dann wirst du in den MÜLLeimer wandern oder (-) sonst was wird passieren S7: oder sitzenbleiben und später kommen S11: oder sitzenbleiben und später kommen Translation S11: you can actually only go through these modification and instruction frameworks and you can either turn out right or WRONG * and if you ahm bring with you the correct information and encodings * then you will be integrated into the group if you don't bring it you will be thrown into the garbage bin or (—) something else will happen S7: or you will repeat a grade and be late S11: or you will repeat a grade and be late Analysis In previous turns the metaphor model DIFFERENTIATION OF HUMAN CELLS IS LEARNING IN SCHOOL with metaphorical concepts like SCHOOL, TEACHER, ELITE, HIGHSCHOOL DIPLOMA (ABITUR), EX-

STOP MAKING SENSE!

163

AMINATION, and INSTRUCTION had already been introduced and dis­ cussed. In the first turn of the transcript S11 takes up the metaphor model DIFFERENTIATION OF HUMAN CELLS IS LEARNING IN SCHOOL. Then S11 adapts the school model by drawing conclusions in the target do­ main: he states that the following proposition must be true in the target domain: 'a protein is either correct or incorrect'. This proposition could not be covered by the introduced school model. He introduces the new metaphorical concept 'garbage bin' for the incorrect proteins, implicitly introducing the new source domain EXACT PRODUCTION which includes the model of garbage removal. Then he adapts it by superimposing it onto the school model thus creating a "chimerical concept"9: The school as imagined by S11 seems to be a kind of automatic machine. The student has to run through this automatic instruction machine which encodes him/her either as right or wrong. Students with correct codes become integrated into the group, while wrong-coded ones are not stu­ dents at all, they are garbage. Thus S11 expresses the perspective of the cell as a fully deterministic system, whereas the simple school model would imply at least some degree of autonomy. S7 then ignores the metaphor model of the "garbage bin" and hence the source domain EXACT PRODUCTION/GARBAGE REMOVAL introduced in the previous turn by S11. He takes up the school model and expands it through adding two new metaphorical concepts 'repeating a grade' and 'being late' for incorrect proteins. In showing that the school model is able to include the proposition 'a protein is either correct or incorrect' as stated by S11, S7 dem­ onstrates that the school model is still appropriate as is the autonomy-perspective for the cell. S i l confirms this by echoing the phrase uttered by S7 ("or you will repeat a grade and be late"). 3.3.2.

Case study 2: Talking raft or tube roller factory - how viral proteins find each other in the cell

Metaphor models previously introduced DIFFERENTIATION OF HUMAN CELLS IS LEARNING IN SCHOOL THE CELL IS A STATE GOVERNED BY A DICTATOR THE CELL IS A PARLIAMENT THE CELL IS A HOUSE THE CELL IS AN ANTHILL THE CELL IS A BEEHIVE THE CELL IS AN ECONOMIC SYSTEM VIRAL PROTEINS TALK IN ORDER TO FIND EACH OTHER

164

WOLF-ANDREAS LIEBERT

Transcript S7: NAJA aber dann hatten wir mit * KOlokalisation als soche im Eigentlichen sinne des wortes/das verlangt ja nicht daß die miteinander SPRECHen sondern sie können gemeinsam * EINen gespRÀCHspartner sagen wir so ein RAFT ahm das env würde dahinfinden und das gag würde dahinfinden Sil: mhm(-) S7: und dann wären=sie da und würden partikel machen und alles friede freude eierkuchen aber sie hätten NIE miteinander gespROCHen * Sil: aber ich glaube daß dieses BIEnen- und ameisenbeispiel ist vielleicht auch GUT wenn man * die komplexität des ganzen zellsystems * ahm * (hört) wenn man jetzt aber so WEIter gehen würde in so einen individuELLen {ar/bestandteil glaub—ich is) * das proBLEM ist eigentlich für mich=η ANderes S8 * das is schon RICHtig was du sagst daß * die zelle zu komplex is um mit irgendeinem dieser analoGlEbeispiele * ahm * supergut beschrieben zu werden deswegen kommen=wir sof Ort immer wieder an die grenze und vielleicht deswegen weil wir zu VIEL damit beschreiben wollen immer mit JEdem dieser wir können glaub=ich die analoGIE dann immernur für einen TEILaspekt bringen *und dürfen dann nicht weitergehen und sagen aber jetzt müssen=wir das gesamte ZELLprogramm noch mit beschreiben wol­ lend das wo mir der BIEnenstaat gut gefiele um die * REgulation der zelle über * interaktive faktoren (transcriptionsfaktoren) oder irgend sowas zu beschreiben hat mir WEniger gut gefallen um SOLche phänomene wie wir Eben andiskutiert haben (zu geschreiben) und jetzt also * das mag genauso schlecht oder schlEchter sein * würde mir zum beispiel das ist jetzt gerade mal GANZ kurz angesprochen das war mit der (WIRTschaftsmöglichkeit) fast BESser gefallen also angebot und NA.CHfrage * son/(es lohnt sich) der bedArf an stAhlträgern oder kunststoffrohren * und die wer­ den dann so lange dorthin gellEfert bis du dort irgendetwas zusammen­ montieren kannst und dann wird das wieder ABgeschaltet die produktiom und du machst irgendwas das mag jetzt (LACHEND)#auch nicht besser sein aberti es fiel mir jetzt gerade so ein in dem Zusammen­ hang das würde mir das BESSer beschreiben als die frage wie ameisen sagen was sie haben M: gut ihr meint aho * auf das was der S7 gesagt hat es geht noch mal um diese FRAgestellung auch zu beginn der diskussion das noch mal (anzureichern) äh (alle anderen) wenn=s=n gesprächspartner (wie)/(für) BEIde (eben der * da ist) und diREden an sich gar nicht miteinander

STOP MAKING SENSE!

S11 :

165

JA du hast=η RÖHrenwalzwerk oder du hast deinen/ deinen */deinen äh * walzWERKSfritzen (LACHEN) und der braucht halt äh stAHlige walzen der/äh zum walzen und er braucht halt irgendwelche maSCHINen dazu und die müssen beide dahin kommen obwohl der stahl der gewalzt wird mit den maschinen nichts zu tun hat in dem sinne nichts

Translation S7: well, yes, then we would have co-localization in the true sense of the word/ this does not require that they talk to each other, instead they can share a discussion partner, let's say a raft, the env would find it so would the gag S11: hmm S7: then they'd be there and make particles and everything would be fine and dandy, and they would never have talked to each other S11: I find correct what you said about the cell being too complex to be described perfectly with one of these analogies. We always hit another limit, perhaps because we want to describe too much at once, I think we can only use the analogy for one aspect of the problem and not for the whole. I believe that this bee and ant example is perhaps even then good if you consider the whole complexity of the cell system. If you take things further to talk about in­ dividual phenomena such as were just described, the economic model seemed to be better, meaning supply and demand, like the demand for steel girders or plastic tubes which will be supplied as long as you can as­ semble something with them, then the shipments are stopped and you pro­ duce something else again I feel this model works better than the question of how ants say what they want M: to get back to the beginning, you mean that they don't actually talk to each other S11: say you have your tube roller factory and the tube roller worker needs this steel roller in order to roll and he needs machines and they both have to be shipped even though the steel that needs to be rolled and the machines don't have anything in common, per se. Analysis The topic covered in this transcript is the so called "co-localization": core pro­ teins ("gag") and envelope proteins ("env") of HIV10 emerge at different loca­ tions in the cell but "find" each other through a yet unknown process for the assembly of the whole virus. The discussion included many metaphor models mentioned above. The pivotal metaphors of former turns were THE CELL IS

166

WOLF-ANDREAS LIEBERT

AN ECONOMIC SYSTEM, THE CELL IS AN ANTHILL, THE CELL IS A BEEHIVE, PROTEINS TALK TO FIND EACH OTHER. S7 ignores the metaphor models ANTHILL and BEEHIVE introduced by S8 in a former turn and takes up the talk model. Then, he introduces the meta­ phor model of a RAFT. He expands the talk model through superimposing the raft model onto it, thus creating the chimeric model of a raft which talks to the viral proteins in order to lead them to the correct assembly location. Sil ignores the "talking raft" model which was introduced by S7 in the previous turn. He takes up the metaphor models ANTHILL and BEEHIVE in­ troduced by S8 in a former turn, adapts them by limiting these models to cer­ tain aspects (they are good for conceptualizing the complexitiy of the cell), and then rejects the two models. Finally he takes up the metaphor model THE CELL IS AN ECONOMIC SYSTEM and expands it by adding new meta­ phorical concepts (SUPPLY, DEMAND, STEEL GIRDERS, PLASTIC TUBES, SHIPMENT) and by adding a metaphorical submodel (TUBE ROLLER FACTORY). M ignores the metaphor THE CELL IS AN ECONOMIC SYSTEM and takes up the talk model referring to the utterance of S7. In his next turn, Sil ignores the talk model again, instead he re-introduces the metaphor model THE CELL IS A TUBE ROLLER FACTORY and expands the model by gen­ erating a scenario (tube roller factory containing machines, tube roller worker with goals and a typical procedure). Again there is a conceptual conflict between the autonomy-perspective rep­ resented in the talk model and the deterministic-perspective represented in the factory model. The excerpt shows how a perspective is made dominant by just rejecting or ignoring other models and a constant re-introduction and expansion of one's own point of view. 3.3.3.

Case study 3: The infected cell - an infiltrated company

Metaphor models previously introduced DIFFERENTIATION OF HUMAN CELLS IS LEARNING IN SCHOOL THE CELL IS A STATE GOVERNED BY A DICTATOR THE CELL IS A PARLIAMENT THE CELL IS A HOUSE THE CELL IS AN ANTHILL THE CELL IS A BEEHIVE THE CELL IS AN ECONOMIC SYSTEM VIRAL PROTEINS TALK IN ORDER TO FIND EACH OTHER

STOP MAKING SENSE!

167

Transcript S8 das könnte=ich mir vorstellen das/* die rhein-neckar-zeitung mit ihrer berichterstattung über B.M. * der/der find den Inhalt LINK * (_____) wenn ich das jetzt BÖSartig interpretier sag=ich halt daß das ein Virus ist also daß die STEUerzahler*/STEUerzentrale des konzerns * in der zelle REINgekommen * und möchte jetzt den konzern umkrempeln WEIL er * (meinetwegen) möchte daß B.M. IRgendwas ganz bestimmtes was (in ist) in GROßem maße (macht) Ψ und er steuert jetzt B.M. so um * daß sie HAUPTsächlich * also (vitamin ) produzierend und ER sitzt oben und sch/kann alle Schalthebel so beeinflussen daß die/ daß die vitamin-c-synTHEsemaschine angeht * das maximale aus (--) S11 er benutzt den normalen vertnebsweg also alle B.M. * Organisationen betriebsorganisationen geschäftsorganisationen und so weiter werden weiter geNUTZT in dem sinne also er muß sich keine NEUEN * wege in dem sinne schaffen son­ dern er kann auf (alten wegen) Translation S8: I could imagine this - the Rhein-Neckar newspaper with their coverage of B.M. if I interpret this in a mean way, I say that the virus has entered the executive office of the company and wants to restructure the company, he wants that B.M. now should produce mainly vitamin C, and he sits at the top and can influence the mechanisms so that the vitamin  synthesis ma chine is turned on S11: he uses B.M. in all its aspects and structures of or­ ganization and distribution, meaning he doesn't have to build anything new, just simply takes over to his own advantage Analysis These two turns show cooperative expansion by scenario generation. With his introductory remarks about the newspaper article on B.M., S8 contextualizes shared knowledge which was discussed in the local press: B.M. is a large Ger­ man firm, whose management was accused of being infiltrated by persons working to their own advantage. In a restructuring of the company these persons were fired. Thus S8 takes up the economic model and expands it through adding a new submodel of the economic model: THE CELL IS A COMPANY. This model is adapted by specifying: THE CELL IS AN INFILTRATED COM­ PANY. This metaphor model is expanded by scenario generation. The scenario includes the mappings of main participants, goals, and actions as shown in Table 2.

168

WOLF-ANDREAS LEEBERT

Table 2. The infected cell is an infiltrsated company

Projected Submodel: Infiltrated company COMPANY EXECUTIVE OFFICE INVADER INVASIVE PLAN INFILTRATION ENTERING THE EXECUTIVE OFFICE PRODUCING VITAMIN  INVADER SITS AT THE TOP INVADER COMMANDS INVADER TURNS MACHINE ON

Target model: Infected cell CELL DNA VIRUS VIRAL RNA INFECTION LOCKING INTO THE DNA PRODUCING VIRAL PROTEINS VIRAL DNA INTEGRATED IN THE HUMAN DNA VIRAL DNA IS READ BY HUMAN RNA VIRAL RNA IS READ BY HUMAN RIBOSOMES

Sil takes up this development of a metaphorical scenario in a cooperative way and expands it further through the addition of new metaphorical concepts to complete the scenario. The mappings are shown in Table 3. Turn for turn, a common perspective is thus developed through the coconstruction of a coherent metaphor model. Table 3. The infected cell is an infiltrated company (expansion)

Projected Submodel: Infiltrated company USE OF NORMAL PASSAGES USE OF AFFILIATES

Target model: Infected cell

DISTRIBUTION PROTEIN PATHWAYS IN "THEM CELL USE OF CELL COMPARTMENTS AND FUNCTIONS |

3.4. Metaphorical reasoning in scientific journalism (case studies 4-5) The following studies have come out of project C7 of the Sonderfor­ schungsbereich 245 on the transfer of scientific information to the public (cf.

STOP MAKING SENSE!

169

Biere & Liebert 1997). In this project a corpus containing radio broadcasts on science covering the time span of 1992-1996 was built up. The use of metaphors was analyzed and then compared with the metaphors found in the scientists' corpus of the TLMSF-project (see above). This was done in order to find an observable process of social cognition and the emergence of social representa­ tion via metaphor (Moscovici 1981; Liebert 1996b). In this paper some tran­ scripts of radio talks will be analyzed in order to sequence the metaphoric pro­ gression. However, some remarks on the journalists' interests in metaphors, and the "naturalness" of radio talks must be made first. Journalists have different tasks when dealing with metaphors, than scientists who are reflecting on their own metaphors (cf. Göpfert 1997). We must con­ sider the following idea. According to Gordon (1960: 104): It is essential to distinguish between metaphor which is decorative and after-thefact (in that it does not discover but describes the already-discovered) and meta­ phor which is generative, inductive, before-the-fact - the initial leap in the proc­ ess of discovery. Scientists are interested to find metaphors in areas where they have no so­ lutions or clear concepts of what is going on, say in the human cell. Gordon (1960: 104) refers to this as a "metaphor before-the-fact". Journalists are inter­ ested in new ways of describing what has already been discovered by the scien­ tists, in Gordon's words, they are trying to find a "metaphor after-the-fact". They use metaphors intentionally to get a connection between the scientific and the everyday world. They adjust scientists' metaphors to the public level. This has to be considered when we speak of "natural communication". An interview in itself is a somewhat artificial situation. Usually, the journalist receives a press release about scientific news and contacts the scientist. Then they have a short discussion about the topic before the broadcast. Of course not everything is ne­ gotiated in advance in order to preserve the live interview atmosphere, but to some extent an interview is always prepared in advance. Also some broadcasts only appear to be dialogues. Journalists refers to this as "built contribution" ("gebauter Beitrag"; cf. Bischl 1997). Its creation runs as follows: the journalist receives a press release about some interesting scien­ tific news. Then he interviews the scientists involved. Interviews are not broad­ cast in the original version, but are cut and arranged with extra comments from the journalist. The listener is lead to believe that the scientist is answering the journalist's questions directly, and that the journalist is also adding his com­ ments "live" (for a detailed analysis see Bischl 1997). This kind of discourse is completely different from any other sort of verbal interaction. It might not be regarded as interaction at all. Perhaps, it would be more appropriate to concep­ tualize it as a text produced by a single author with acoustic voices. The reason why a "gebauter Beitrag" is analyzed below is that the journalist tries to simu­ late authentic verbal interaction, and the analysis will show if he succeeds.

170

WOLF-ANDREAS LIEBERT

In the following transcripts scientists are abbreviated with "S + Number", for example S6, and journalists with "J + Number", for example "  " . 3.4.1.

Metaphorical reasoning in a live radio interview

3.4.1.1. Case study 4: Settle down - The pickaxes and shovels of the black fungi The following interview was broadcast on in the "West German Radio Sta­ tion 2" (WDR 2) on March, 3rd 1994, in the daily radio talk show, the "Mittagsmagazin", with a five minute slot for the presentation of scientific news. Transcript Jl: wir wollen * über die schwarzen löcher auf antiken bauten die stammen nämlich NICH vom säurefraß sondern auch von einem fressenden LEbewesen * einem schwarzen ' * heRAUSgefunden hat das ein WISsenschafilerteam unter der leitung von * professor w. k. der * an der Universität Oldenburg arbeitet * guten tag herr prof essor SI: ja guten tag f rau Jl: diese schwarzen * die sie da als verURsacher raus­ gefunden haben * wie kommen die überhaupt auf diese MARmorbauten^ * S1 : nun * PILze sind Mikroorganismen und können überall in der luft herum * * das wissen wir ja auch durch Infektionen * der menschen mit schnupfen und * * und wenn w sie * DORT einen geeigneten siedlungsgrund und geeignete nährstoffe fin­ den dann werden sie sich auch * und leider nich so schnell * ABsterben (...) Jl: wie: MAchen sie das denn an dem MARmor * enthält der MARmor nah/ nährstoff e die den pilz am leben halten und zum Wachstum Sl: nein der marmor bildet nur so etwas wie ein geHÄUse für die * auf dem sie sich EINrichten * und sie ernähren sich überwiegend aus luftverUNreinigungen von denen es ja nicht nur SÄUren gibt sondern auch sehr viele organische (...) J l : (PRÄGEND) # wie sind sie diesen kleinstlebewesen denn auf die spur gekommen wie konnten sie das geheimnis Sl: nun * wir haben * Einerseits gesehen daß diese löcher sich nicht VÖLlig PLÄCHIg über die gesteine zogen wie das die: äh REgel wäre wenn das durch sauren REgen * sondern selektiv sich ansiedelten so wie man halt die pilze im

STOP MAKING SENSE!

171

WALD * und dann ha=m wir mikroskopisch festgestellt daß das KLEINSTlebewesen Jl: und die haben OFFENbar ja/ * also * geRÄTWERKzeug das mag man zwar gar nicht SAgen aber=s Muß ja wohl so * damit sie dann ihre arbeit verrichten Sl ja das ist w ' * man kann im labor natürlich SEHR schnell testen ob diese: winzigen lebewesen säuren herstellen die dann: Ah mit säuren das gestein auflösen ,.. . . * das KÖNnen sie * da sie aber trotzdem Löcher machen müssen sie das SO: machen wie wir das mit=m bohrrneißel bei—ner erdölbohrung machen * oder wie wir=s mit hacke und schaufei '*  Jl: (FRAGEND) ff und wie machen sie das * aho die HAben wirklich kleine schaufeln Sl: ja sie HAben so etwas wie kleine schaufeln sie bilden spezielle * und sie LAgern in die zellWÄNde * melaNIN einen schwarzen farbstoff * der die zeilwände * und dann können sie etwas besser in hartes material ein­ dringen und DAS konnte man auch Translation Jl: Let's talk about the black holes in old buildings. They don't come from acids but from a a living being, a black fungus. This was discovered by a team of scientists under the leadership of professor W. K. at the university of Oldenburg, welcome Professor K. Sl: Yes, hello Ms. K. Jl: These black fungi - how do they get to be on these marble buildings? Sl: Fungi are micro­ organisms and they move all around in the air. We also know them through infections such as colds if they find a good living space with proper nutrients then they willl develop and not die very quickly (...) Jl: How do they do it on marble? Does marble contain nutrients that keep a fungus alive and stimulate its growth? Sl: No. The marble is only a sort of set­ tlement area for the fungi where they can get comfortable They feed mainly on dirty air particles which are not only acids but also organic substances. Jl: How did you find these little organisms? Sl: Well we noticed that the holes did not spread over complete surfaces as would be the case with acid rain. They settled selectively, like mushrooms in a forest, and then we discovered microscopically that they were tiny organisms.

172

WOLF-ANDREAS LIEBERT

-

And they have apparently so, thus/tools I don't even like to say it this way but it must be like this, in order to do their work with Si: Yes, this is sad. In the lab one can test very quickly if these tiny organisms produce acids in order to dis­ solve the rock surface. They can't. But since they make holes anyway, they do it like we do it with a drill in an oil drilling, or like we do, with a pickaxe and shovel. Jl: And how do they do that? So they really have something like small shovels, pickaxes? SI: Yes they have something like small shovels. They build special cells and they store melanin in the cell walls. Melanin is a black color that hardens in the cell wall, and then they can penetrate more easily into hard materials. This has been proven.

Analysis The interview is introduced by a scene of a German humorist which has been omitted in the transcript. The transcript begins with the journalist's first turn. She introduces the theme and her interview partner. She welcomes him and he greets her in return. Then she asks him how he discovered that fungi caused dark stains on old buildings. The scientist states that fungi are microorganisms that float in the air. They need nutrients and a place to live. For the concept of their living space he introduces the metaphoric concept SETTLEMENT AREA, implying the do­ main metaphor FUNGI ARE PEOPLE WHO SETTLE DOWN. The journalist asks, if the fungi need marble of antique buildings as nutrients. The scientist negates and re-introduces FUNGI ARE PEOPLE WHO SETTLE DOWN and expands this with the concepts MARBLE IS A HOUSING and MARBLE IS A HOUSING WHERE THE FUNGI INSTALL THEMSELVES. Their nutrients are organic substances in the air.The omitted two turns cover the nutrition problem. The journalist then asks how the scientist got the idea that the dark regions were caused by fungi. The scientist responds that they found no com­ plete but selective coverage. Then he re-introduces the metaphor model FUNGI ARE PEOPLE WHO SETTLE DOWN and expands it through the addition of the metaphorical concept TO SETTLE. Finally, the journalist takes up this metaphor and expands it by the addi­ tion of the metaphorical concept TOOL. This expansion of the metaphor model is embedded in the following metacommunicative phrases (cf. Dobrovol'skij 1997): (1) "... OFFENbar ja/ * also * ..." ("... apparently so, thus ...")

STOP MAKING SENSE!

173

(2) "das mag man zwar gar nicht S Agen aber=s MUß ja wohl so sein * damit..." ("I don't even like to say it this way but it must be like this, in order to...") In (1) the journalist is explicitly marking her reasoning process. In (2) the journalist first indicates the preliminary character of her expansion of the meta­ phor model, and on the other hand she indicates again that the inferred meta­ phors were forced by the process of metaphorical reasoning. In his next turn the scientist introduces the metaphor model FUNGI ARE SEEKING OIL and expands it by adding the metaphorical concept TOOLS OF THE FUNGI ARE DRILLS. In the same turn he takes up the metaphor model FUNGI ARE PEOPLE WHO SETTLE DOWN and expands it by adding the new metaphorical submodels TOOLS OF THE FUNGI ARE PICKAXES AND SHOVELS. Thus he makes concrete the journalist's TOOL metaphor through the hyponyms PICKAXE and SHOVEL. The journalist ignores the OIL model and takes up the PICKAXE AND SHOVEL model asking for more information about the tools of the fungi whereby she emphasizes this model by saying "So they really have something like small shovels, pickaxes?" By uttering and emphasizing the particle "really" the journalist is signalling that the scientist should now explicitly map the meta­ phor models to target models. Finally, the scientist takes up this model (agreeing on ignoring the OIL model), and adapts it through drawing conclusions in the target domain: the tools PICKAXE and SHOVEL are specific cells where Melanin is stored, a black substance which makes these cells harder. This transcript is interesting because it shows a successful interplay of metaphors for the transfer of scientific knowledge to the public. Some features of the succes of the interplay are the selection of an appropriate metaphor model by the scientist, a sensible re-introduction by the scientist when the journalist did not take up the offered metaphor model, and a cooperative phase of metaphori­ cal reasoning, where the metaphor ic fungus settlement scenario is developed and altered until the conflict between the adequacy of manner and the adequacy for the addressees (Biere 1989) has been solved. 3.4.2.

Artificial metaphorical reasoning in a radio feature

3.4.2.1. Case study 5: Stop making sense! - on textual medicine The following transcript is an excerpt from a feature on genetics which lasted about half an hour. It was broadcast by in the "Bavarian Radio Station 2" (BR 2) in May, 4th 1995.13

174

WOLF-ANDREAS LIEBERT

Transcript J2: verschiedene arten von zeilen * darunter auch die Stammzellen * tragen charakteristische strukturen auf ihrer *7,5* v bei bestimmten zeiltypen * ist es bereits gelungen die genfähre mit einer art antenne zu verseh= .. * die sie gezielt zu zeilen mit sol­ chen strukturen * damit soll erreicht werden * daß das fremde erbmaterial * nur in DIEjenigen zeilen eingeschleust wird die es erhalten *2* und auch um die verMEHrung des virus zu verhindern * gibt es inzwischen mehrere Sl: es gibt ver­ schiedene Strategien wie man gegen * Viren blockierende gene entwickelt * das sind die sogenannten RIbozymen * ahm das sind also rns molekäle * s die es schaffen selbst rns * und * die virusinformation bei hiv besteht ja aus * rns eine and=re strategie sind sogenannte ANtisense * moleküle also moleküle die * eine GEgenrichtung * ahm der leserichtung * des eigentlichen genoms dar- stell = * und * damit ver­ hindern * daß die erbinformation des virus berhaupt abgelesen werden kann^ * und eben dann auch partikel gebildet werden * dann gibt es eine ganze menge Strategien daß man * REgulatorische * eiweißbausteine des virus/ * also hiv is ja ein sehr kompliziert reguliertes virus * und hat ins­ gesamt fünf gene * die dazu dienen * es auf kosten * zellulärer * gene zu vermehrend * und wenn man eben * bei diesen genen bestimmte änderungen vornimmt kann man auch * äh es schaffen * zumindest zunel zum ge­ wissen Prozentsatz hiv/ * die hiv Vermehrung in der zelle zu unter­ drückend Jl: auch zu diesen behandungs-strategien der hiv infektion laufen zur zeit mehrere klinische Studien in den usa an * sie umfassen patiënten in unterschiedlichen Stadien der krankheit * auch solche bei denen der ausbruch des aidsvollbildes verhindert werden *7,5* und auch für an­ dere virusinfektionen gibt es plane für eine gentherapeutische behandlung J2: zu ihnen gehört auch eine bestimmte form der *1,5* oder hepatitis * wie sie die ärzte *7,5* in schweren f allen ist die Überfunktion so stark gestört * daß der volbmund der gesichtsfarbe wegen * von gelbsucht -r * doch gelbsucht ist nicht gleich * in den fernen Urlaubsländern spielt vor allem die hepatitis a eine rolle * die durch spuren vonfäkalien im wasser übertragen * * gegen sie kann man sich also schützen * indem man kein wasser trinkt und auffrischgewaschenen salat oder obst " ' *2* ANders bei den hepatitiserkrankungen mit der bezeichnung b und  *2* sie werden durch viren übertragen * die in den körperflüssigkeiten lauern * vor allem im blut *1,5* anstecken kann man sich also genauso wie mit dem aidsvirus * durch ungeschützten sexuellen kontakt * durch bluttransfusionen oder ge-

STOP MAKING SENSE!

175

brauchte *7,5* und genau wie bei aids ist eine therapie der hepatitis  nicht „ ' *7,5* grund genug für mehrere forscher * bei ihrer bekämpfung an die gentherapie zu * einer von ihnen ist der internist doktor wolfgang . *7,5* er ist an der medizinischen klinik zwei der Universität manchen im klinikum großhadern S2: wir setzen * gentherapeutische ansätze ein * um * die in deutschland häufigste * virale lebererkrankung nämlich die hepatitis  und wir haben in deutschland * circa zwanzigtausend neuerkrankungen zu therapier=η * die besondere bedeutung der hepatitis  liegt dann * daß sie * ganz im gegensatz zur b virushepatitis * zur Chronifizierung w * über die hälfte * aller patiën­ ten mit aKöter hepatitis  *7,5* zeigen einen CHROnischen verlauf * das heißt die erkrankung geht in ein chronisches stadium ■ * zwanzig prozent von diesen patiënten entwickeln durchschnittlich innerhalb von sieben jahren eine leberzirrhose einen bindegewebigen umbau der * und ein beträchtlicher wenn auch im detail * noch nicht bekannter * Prozentsatz dieser zirrhotischen patiënten ein . * DESwegen haben WIR jetzt versucht * unter einsatz gentherapeutischer meth­ oden * die * virusvermehrung zu Jl: die mediziner versuchen das genetische material mit der sogenannten antisense technik UNschädlich zu *1,5* dabei schleusen sie in die infizierte zelle eine kopie ein die genau das negativ des ursprünglichen textes darstellte *1,5* legt man eine solche negativkopie auf den Originaltext * so löschen sich beide gegenseitig *7,5* daß man dabei nur die erbinformation des virus erwischt liegt daran * daß man einen rund zwanzig buchstaben lan­ gen begriff verwendet * der nur in SEInem Wortschatz *7,5* versuche * leberzellen die mit hepatitis  infiziert sind zu behandeln * fin­ den derzeit erst im reagenzglas Translation J2: Different kinds of cells, including the stemcells, carry characteristic structures on their surface. For particular cell types we have already man­ aged to fit the gene ferry with an antenna which conducts them only to cells with these specific structures. This is done so that the alien hereditary material is only introduced into cells that are meant to receive it. To in­ hibit the spreading of the virus there are by now several possiblities. SI: There are different strategies on how to develop genes that block viruses, socalled ribozymes. These are RNA molecules that manage to cut other RNA. Virus information on HIV is made up of RNA. Another strategy are antisense molecules that are molecules representing the reverse reading di­ rection of the genom, which do the job that the virus' hereditary informa-

176

WOLF-ANDREAS LBEBERT

tion cannot be read in the first place. Then there are strategies of regula­ tive protein production. HIV is a very complexly regulated virus, contain­ ing a total of 5 genes, that serve to multiply it at the cost of cellular genes. If we can change these, than we can, at least to a certain extent, under­ mine the spreading of HIV within the cell. Jl: Several clinical studies are be­ ing conducted in the U.S. on these strategies. They research patients at different stages of the illness, even those who are infected by HIV but who do not have AIDS yet, for other diseases plans also exist for gene therapy. J2: These include liver infection or hepatitis. In certain cases the functions of the liver are disturbed to such an extent that we call this disease yellow fever. However, there are different forms of yellow fever. In exotic vacation spots there is mainly hepatitis A which spreads through traces of feces in the water. You can avoid this kind of hepatitis by not drink water, and not eat freshly washed fruits or vegetables. It's different for hepatitis  or  which are spread by a virus lurking in the blood-stream. You can be in­ fected in the same way as with HIV. Like AIDS epatitis  has no effective therapy. Scientists are thus thinking of the possibility of gene therapy. One of them is Dr. Wolfgang K. at the clinic of the University Munich. S2: We are using some gene therapy to deal with viral liver infections called hepatitis  In Germany there are app. 20.000 new cases. The important issue is that the symptoms are often chronic. 20% of the patients develop cirrhosis of the liver within seven years. We are attempting gene therapy for this reason. Jl: Doctors are attempting to control the virus with the antisense technique. They infect the infected cell with a copy which repre­ sents exactly a negative of the original text. The two copies laid one onto the other delete each other. You only delete the hereditary information of the virus because you use a concept containing 20 letters which only occur in its vocabulary. These therapies are so far only occuring as tests in labo­ ratories. Analysis In the first turns of the radio broadcast the topic is introduced. The journalists are speaking of gene technology medicine as MAGIC BALLS. Viruses like HIV are seen as STOWAWAYS who secretly make their way into the cell. They REPROGRAM THE CELL so that the cell produces virus proteins instead of cell proteins. Virology is conceptualized as the development of medicine which would reverse the viral programming. The transcript starts with a journalist's turn, J2, who introduces the meta­ phor model GENE TECHNOLOGICAL MEDICINE IS A FERRY by creating

STOP MAKING SENSE!

177

the new lexeme "gene ferry" while he takes up the source domain TRANSPORT and SHIPMENT which was first applied for the virus itself (STOWAWAY). The concept of the GENE FERRY is adapted by superimposing the con­ cept of an ANTENNA. A chimeric concept has now been created which concep­ tualizes the gene technological medicine as transported by a kind of REMOTE CONTROLLED FERRY which transports the healing substance to the correct location. SI ignores all these metaphor models and explains two strategies for the healing of the infected cell. The first strategy introduces so-called ribozymes where the metaphor of CUTTING is introduced, a metaphorical concept based on the metaphor model THE RNA IS A ROPE, the second strategy is to create antisense molecules. "Antisense" is a fixed expression in virology which is based on the metaphor system of molecular biology. This conceptual system includes the source domain TEXT/COMMUNICATION which is mapped onto specific target areas (cf. Liebert 1995a, 1995c, 1995c). Among others the fol­ lowing metaphor models are common in virology:

THE DNA IS A TEXT THE RNA IS A TEXT THE RNA IS A TEXT TRANSCRIBED FROM THE DNA THE RNA IS TRANSLATED INTO THE LANGUAGE OF PROTEINS These relationships hold for the human DNA and RNA and for the viral DNA and RNA as well. The concept of ANTISENSE as stated by SI, is based on the following re­ lations: IF THE RNA IS A TEXT IT MUST MAKE SENSE IF WE COULD MAKE THE VIRAL RNA STOP MAKING SENSE IT COULD NOT BE READ NOR BE TRANSLATED, THUS IT WOULD BE HARMLESS SI thus introduces the metaphorical concept of ANTISENSE and hence the source domain TEXT/COMMUNICATION. Then she expands the antisense model by adding the metaphorical concepts of REVERSE READING DIRECTION which ensures that the viral information may not be READ. The metaphors introduced in the turn of SI form part of the larger meta­ phorical system and are shared knowledge in virology. However it is doubtful that these metaphors are shared knowledge in everyday life. For this reason, this would represent a typical passage where the journalist could make tacit meta­ phorical knowledge explicit, by using the scientific metaphors as a bridge be­ tween science and public.

178

WOLF-ANDREAS LIEBERT

However, the journalist's next turn is astonishing. Jl completely ignores the antisense model. Jl gives an overview of this research, then shifts the focus on external processes, and on the side introduces the metaphorical concept OUTBREAK, which implies the metaphor model AIDS IS A DANGEROUS ANIMAL.14 The animal model highlights a completely different target area and shifts to another point of view. In the next turn J2 shifts to hepatitis infections and introduces the metaphor model A VIRUS IS A DANGEROUS ANIMAL and expands it by adding the concept of LURKING. The scientist S2 takes up the model A VIRUS IS A DANGEROUS ANIMAL and expands it through adding the concept of PROPAGATION. In the next turn Jl takes up the antisense model and expands it through scenario generation: a copy is locked into the infected cell this copy represents the negative of the original text original text and its negative delete each other viral texts are selected because they contain a concept the concept contains only 20 letters which exist exclusively in the vocabu­ lary of the virus This scenario presupposes the DNA AS TEXT metaphors as described above, as well as the following metaphorical relations: AN ANTISENSE MOLECULE IS A NEGATIVE OF THE RNA RNA IS A TEXT WHICH CAN BE COPIED A NEGATIVE IS A COPY OF A RNA IN REVERSE READING DI­ RECTION THE TEXT CONTAINS A CONCEPT THE CONCEPT CONTAINS LETTERS THE NEGATIVE TEXT AND THE ORIGINAL TEXT DELETE EACH OTHER A VIRUS HAS A VOCABULARY A VIRUS UNDERSTANDS ONLY THOSE WORDS THAT ARE IN ITS VOCABULARY These relations are neither explained, nor made explicit. The journalists continue to refer to tacit knowledge of molecular biology which they may have understood, but which are inaccessible to a larger audience. This transcript is very interesting because the metaphors uttered in the last turn are very close to those metaphors which virologists would use themselves when they are requested to explain their research topic (cf. Liebert 1995a, 1995c). Normally most texts of science journalism concerning AIDS are fo-

STOP MAKING SENSE!

179

cussed on the WAR metaphor (Sontag 1989; Liebert 1996b; Dobrovol'skij 1997). These journalists solved the conflict between the adequacy of manner and the adequacy for the addressees (Biere 1989) to the benefit of manner. They failed to reach adequacy for the addressees because they did not utilize underly­ ing scientific metaphors like THE DNA IS A TEXT for the mapping to target facts. There is no playful metaphorical progression from introduction and taking up of metaphors, as were seen in the creative thinking sessions or in the live interview discussed above. There are no responsive reactions to metaphors in­ troduced or expanded by a journalist. There is only one responsive reaction when S2 takes up the animal model. But while taking up this metaphor model he traces another area of the target domain. The result is that the metaphorical pro­ gression is disjointed and connected to different point of views. It is striking that the journalists waited three turns before taking up the antisense model which was probably not understood even when first introduced. In these three turns, the introduced animal model is connected to different points of view and thus inhibits the fluent development of a coherent perspective induced by a coherent metaphorical progression. Obviously, these passages were not natural dialogue or a sequence of authentic interviews, but rather a "gebauter Beitrag" which only simulates natu­ ral dialogue. Further comparison between natural and artificial dialogue might show how more "naturalness" could be achieved to improve comprehension.

4. Conclusion In-vivo studies of metaphorical reasoning do not only have a practical impact on creative problemsolving or improved comprehension of popular science texts or radio emissions. They also give us new insight into the process of the emergence of metaphor models in verbal interaction: we can look at this process as a nego­ tiation process or as a process of playful cooperative construction. The basic dialogue unit containing the pragmatic operations of initiative and response, each followed by semantic operations of expansion or adaptation are necessary concepts for the analysis of metaphors in verbal interaction. The combination of pragmatic and semantic operations within one turn shows that verbal interaction is more than "just talk". Based on shared knowledge, verbal interaction is each speaker's playful interactive linking and extending of their cognitive spaces.

180

WOLF-ANDREAS LIEBERT

Endnotes 1. "Metaphern stellen schließlich noch eine wichtige Sonderform dar, die aber schon aufgrund ihrer linguistischen Anteile hier vorläufig nicht weiter berücksichtigt wer­ den soll" (Hesse 1991: 14). 2. Phase III is not relevant to this paper. For an elaborated view of Phase III see Hesse (1991) and Liebert (1995c). 3. In its last five years the SFB 245 was labelled "Language and Situation". 4. These operations could all be integrated into the formula as well, but in my view this would not increase understanding. 5. All transcriptions according to Gutfleisch-Rieck et al. (1989). 6. TLMSF is a German acronym meaning "The lexicon of metaphor models as a tool for imagination in science". The project was financially supported by the German Research Foundation. 7. The analysis of the corpora showed that TRANSPORT, PRODUCTION and COM­ MUNICATION models play a central role in the metaphorical system of the vi­ rologists. This contradicts the common assumption as suggested by Sontag (1989) or Várela (1991) that the WAR-scenario is the most important metaphor model in AIDS research (for a detailed discussion see Liebert 1995a). 8. There is an interesting connection to metaphors in the history of science. We can assume that all metaphors for science which developed over the past two centuries form an inventory of concepts in contemporary science (cf. Jäkel 1995, 1996). 9. "Chimerical concepts" are close to the approach of conceptual blending introduced by Turner and Fauconnier (1995). The differences between chimerical and blended concepts shall not be discussed here. 10.This process is even more complex. For example the viral RNA reaches the correct assembly location as well. How it finds its way is unknown. ll."B.M." is an abbreviation used to preserve the anonymity of the company. 12.1 would like to thank Helga Kirchner and the WDR for the permission to quote from this broadcast. 13.1 would like to thank the BR for permission to quote from this broadcast. 14.This model is based on the metaphor HIV IS A DANGEROUS ANIMAL trans­ formed through the metonymie operation EFFECT STANDS FOR CAUSE.

References Biere, Bernd Ulrich. 1989. Verständlich-Machen. Tübingen: Niemeyer. Biere, Bernd Ulrich and Wolf-Andreas Liebert (eds). 1997. Metaphern, Medien, Wis­ senschaft. Opladen: Westdeutscher Verlag. Bischl, Katrin. 1997. "Vermittlungsdiskurse im Rundfunk". In B. U. Biere, WolfAndreas Liebert (eds), Metaphern, Medien, Wissenschaft. Opladen: Westdeut­ scher Verlag. Chafe, Wallace L. 1994. Discourse, Consciousness, and Time. Chicago: Chicago Uni­ versity Press. Collins, Α.. and D. Gentner. 1987. "How people construct mental models". In D. Holland, N. Quinn (eds), Cultural Models in Language and Thought. Cam­ bridge, 243-265.

STOP MAKING SENSE!

181

Csikszentmihalyi, Mihaly. 1990. FLOW - the Psycholgy of Optimal Experience. New York: Harper & Row. van Dijk, Teun A. and Walter Kintsch. 1983. Strategies of Discourse Comprehension. New York, London: Academic Press. Dobrovol'skij, Dmitrij. 1997. "Metaphern für AIDS in Wissenschaftssendungen im Hörfunk". In . U. Biere, Wolf-Andreas Liebert (eds), Metaphern, Medien, Wissenschaft. Opladen: Westdeutscher Verlag. Falkenhainer, Brian. 1987. "Scientific theory formation through analogical inference". In Pat Langley (ed.), Proceedings of the Fourth International Workshop on Ma­ chine Learning. June 22-25, 1987, University of California, Irvine. Los Altos: Morgan Kaufmann Publishers, 218-229. Gentner, Dedre and Michael Jeziorski. 1993. "The shift from metaphor to analogy in western science". In A. Orthony (ed.), Metaphor and Thought (2nd ed.) Cam­ bridge: Cambridge University Press. Gentner, Dedre and Donai Gentner. 1983. "Flowing water or teeming crowds: mental models of electricity". In Dedre Gentner, Albert Stevens (eds), Mental Models. Hillsdale, N.J. Gilhooly K. J., M. T. G. Keane, R. H. Logie and G. Erdos (eds). 1990. Lines of Thinking: Reflections on the Psychology of Thought. Volume 1: Representation, Reasoning, Analogy and Decision Making. Chichester, New York etc.: John Wiley and Sons. Glucksberg, Sam. 1989. "Metaphors in conversation: How are they understood? Why are they used", Metaphor and Symbolic Activity 4(1): 125-143. Göpfert, Winfried. 1997. "Verständigungskonflikte zwischen Wissenschaftlern und Wissenschaftsjournalisten". In B. U. Biere, Wolf-Andreas Liebert (eds), Meta­ phern, Medien, Wissenschaft. Opladen: Westdeutscher Verlag. Gordon, William J. J. 1960. Synectics. New York. Graumann, Carl Friederich. 1960. Grundlagen einer Phänomenologie und Psychologie der Perspektivität. Berlin: Mouton de Gruyter. Graumann, Carl Friederich. 1989. "Perspective setting and taking in verbal interac­ tion". In R. Dietrich, C.F. Graumann (eds), Language Processing in Social Context. Amsterdam: North-Holland, 95-122. Graumann, Carl Friederich. 1990. "Perspectival structure and dynamics in dialogues". In I. Markova, K. Foppa (eds): The Dynamics of Dialogue. New York: Springer, 105-127. Graumann. Carl Friedrich and Carlo Michael Sommer. 1988. "Perspective structure in language production and comprehension", Journal of Language and Social Psy­ chology 7: 35-54. Gutfleisch-Rieck, I., W. Klein, A. Speck, and Th. Spranz-Fogasy. 1989. Transkrip­ tionsvereinbarungen fir den Sonderforschungsbereich 245 "Sprechen und Sprachverstehen im sozialen Kontext". Heidelberg, Mannheim: Heidelberg Uni­ versity Press. (Arbeiten aus dem Sonderforschungsbereich 245 "Sprechen und Sprachverstehen im sozialen Kontext"; 14). Hesse, Friedrich W. 1991. Analoges Problemlösen: eine Analyse kognitiver Prozesse beim analogen Problemlösen. Weinheim: Psychologie Verl. Union.

182

WOLF-ANDREAS LIEBERT

Indurkhya, Bipin. 1992. Metaphor and Cognition. An Interactional Approach. Dor­ drecht, Boston, London: Kluwer. Jäkel, Olaf. 1995. "Metaphorical scenarios of 'Science"'. University of Hamburg. C.L.E.A.R. report 6, also to appear. In Réne Dirven and Martin Pütz (eds), The Construal of Space in Language and Thought. Berlin: Mouton de Gruyter. Jäkel, Olaf. 1996. Metaphern in abstrakten Diskurs-Domänen. Ph.D. thesis, Univ. of Hamburg. Johnson, Mark. 1987. The Body in the Mind. The Bodily Basis of Reason and Imagi­ nation. Chicago, London: University of Chicago Press. Kallmeyer, Werner (ed.) i.p. Formulieren im Gespräch (3 vols.). Berlin: Mouton de Gruyter. Keane, Mark T. G. 1990. "Analogy, metaphor and transfer". In Kenneth J. Gilhooly, Mark T. G. Keane, Robert H. Logie and George Erdos (eds), Lines of Thinking: Reflections on the Psychology of Thought. Volume 1: Representation, Reasoning, Analogy and Decision Making. Chichester, New York etc.: John Wiley and Sons, 210-203. Knudsen, Sanne. 1996. By the Grace of Gods - and Years and Years of Evolution. Analysis of the Development of Metaphors in Scientific Discourse. Ph.D. thesis, University of Roskilde, Denmark. Lakoff, George. 1987. Women, Fire, and Dangerous Things. What Categories Reveal About the Mind. Chicago: University of Chicago Press. Lakoff, George. 1993. "The contemporary view of metaphor". In A. Orthony (ed.), Metaphor and Thought (2nd ed.).Cambridge: Cambridge University. Lakoff, George and Mark Johnson. 1980. Metaphors we live by. Chicago, London: University of Chicago Press. Liebert, Wolf-Andreas. 1990. "Zu einem Wörterbuch der festen Metaphern der deut­ schen All-tagssprache im Spannungsfeld zwischen Kognitiver Linguistik und onomasiologischer, lexikographischer Datenbank". In B. Rieger, B. Schaeder (eds), Lexikon und Lexikographie. Maschinell - maschinell gestützt. Grundlagen Entwicklungen - Produkte. Hildesheim, Zürich, New York: Olms, 11-28. Liebert, Wolf-Andreas. 1992. Metaphernbereiche der deutschen Alltagssprache. Kog­ nitive Linguistik und die Perspektiven einer Kognitiven Lexikographie. Frank­ furt/Main, Berlin, Bern etc.: Lang. Liebert, Wolf-Andreas. 1994. "Lascaux - a hypermedia lexicon of metaphor models for scientific imagination". In W. Martin, W. Meijs, M. Moerland, E. ten Pas, P. Sterkenburg and P. Vossen (eds), Proceedings of the 6th EURALEX Congress on Lexicography in Amsterdam, The Netherlands. Free University of Amsterdam, 494-500. Liebert, Wolf-Andreas. 1995a. "Metaphernbereiche der virologischen Aidsforschung", Lexicology 1: 142-182. Liebert, Wolf-Andreas. 1995b. "The lexicon of metaphor models as a mental tool for analogical problemsolving in science". In René Dirven, Jan Vanparys (eds), Cur­ rent Approaches to the Lexicon. Bern, Frankfurt/M., Berlin, New York: Lang, 433-448.

STOP MAKING SENSE!

183

Liebert, Wolf-Andreas. 1995c. Metaphernreflexion in der Virologie. Abschlußbericht zum TLMSF-Projekts (Final report). Bonn, Mannheim: German Research Foun­ dation & Institute for German Language. Liebert, Wolf-Andreas. 1996a. "Hypertextdesign in der kognitiven Lexikographie". In H.-E. Wiegand (ed.), Wörterbücher in der Diskussion IL Tübingen: Niemeyer, 103-139. Liebert, Wolf-Andreas. 1996b. "Die transdiskursive Vorstellungwelt zum AIDS-Virus. Heterogenität und Einheit von Textsorten im Übergang von Fachlichkeit und Nicht-Fachlichkeit". In H. Kalverkämper, K.-D. Baumann (eds), Fachliche Textsorten. Tübingen: Narr, 690-712. Moscovici, Serge. 1981. "On social representation". In J. P. Forgas (ed.), Social Cognition. Perspectives on everyday understanding. London: Academic Press. Rettig, Heike, Lydia Kiefer, Carlo M. Sommer and Carl F. Graumann. 1993. "Persuasionsstrategien und Perspektivität", Sprechen. Zeitschrift für Sprechwis­ senschaft 1: 38-51. Schön, Donald A. 1979. "Generative metaphor: A perspective on problem-setting in social policy". In A. Orthony (ed.): Metaphor and Thought. Cambridge, Lon­ don, New York etc.: Cambridge University Press, 254-283. Sommer, Carlo M., Heike Rettig, Lydia Kiefer and Dieter Frankenhauser. 1994. Ger­ many will be one single block ... "Point of View and Reference to Topic Aspects in Adversarial Discussions on Immigration". Arbeiten aus dem Sonderfor­ schungsbereich 245 "Sprache und Situation", Bericht Nr. 78. Heidelberg/ Mannheim: Heidelberg University Press. A slighltly revised version appeared in: Journal of Language and Social Psychology 14: Nos. 1-2, 124-143. Sontag, Susan. 1989. AIDS und seine Metaphern. München, Wien: Hanser. Turner, Mark and Gilles Fauconnier. 1995. "Conceptual integration and formal ex­ pression", Metaphor and Symbolic Activity 10: 183-204. Várela, Francisco J. 1991. "Der Körper denkt. Das Immunsystem und der Prozeß der Körper-Individuierung". In H. U. Gumbrecht, K. L. Pfeiffer (eds): Paradoxien, Dissonanzen, Zusammenbrüche. Situationen offener Epistemologie. Frankurt/ Main: Suhrkamp, 727-743.

Conceptual Blending on the Information Highway: How Metaphorical Inferences Work Tim Rohrer University of Oregon, U.S.A.

1. Introduction: CYBERSPACE and the CYBERFUTURE I like to think of myself as someone who sits on the curb of the info-highway, bemusedly watching all the traffic whiz by—and also as someone who occa­ sionally darts out into the traffic to pick up interesting litter. I frame my paper in these terms because I believe we can learn a lot by examining the linguistic litter thrown off in the actual conversation about technology policy in the United States. In this paper I will begin by offering some of the rich and complex ex­ amples I have collected from the news media, then turn to some of the actual policy statements by political figures such as U.S. Vice-president Gore, paying particular attention to how metaphors shape both science policy and the critiques of science policy made by political rivals. The examples in this paper were all originally drawn from the internet, but are available in both electronic and print form. I will discuss these examples using some recent theoretical tools developed in cognitive semantics: conceptual metaphors and their mappings, embodied image schémas, metaphorical inference diagrams, and conceptual blending. First, I will explain what a conceptual (metaphor) mapping is in discussing the INTERNET IS A HIGHWAY (INFORMATION HIGHWAY) metaphor system. However, not every portion of the highway domain maps across to the informa­ tion domain, and some of the parts that do map across may not map across well. I argue that the conceptual mapping is not purely arbitrary but is constrained by embodied image schemas and by internal coherence with other elements of the conceptual mapping. Next, I will examine metaphorical inferences in narratives using several long quotes from Gore's speech. In the United States for example, the Clinton-Gore administration has argued that the U.S. economic boom of the 1950s and 1960s was fueled by the federal commitment to plan and build the interstate highway system, and a similar economic boom would result from a federal commitment to plan and build the information highway. I diagram how each step of this inference is metaphorically mapped from the source domain (transportation highway policy) to the target domain (information infrastructure policy), presenting the metaphorical inferences as parallel knowledge structures. The wide acceptance of these examples of metaphorical reasoning about U.S. information policy raise general questions about social policy: Are meta­ phors constitutive (as Lakoff and Johnson suggest) of social policy? If so, then

186

TIΜ ROHRER

how can cognitive semantics account for parts of the source domain which don't map well or at all? How can cognitive semantics account for the differences between domains which seem to motivate revisions to social policy? In this pa­ per, I claim that metaphorical inferences produce a kind of conceptual blending which takes place between the domains, motivating changes in the world to fit our metaphorical understanding. I conclude that the conceptual blending pro­ posed by Gore and criticized by his critics is constitutive of contemporary U.S. information policy and is transforming the future.

2. The Difference between CYBERSPACE and the CYBERFUTURE The INTERNET IS A HIGHWAY (INFORMATION HIGHWAY) metaphor is with­ out a doubt a wildly successful metaphor, so much so that the phrase informa­ tion highway is in danger of becoming a cliché. Here are a few initial examples of the metaphor, all of which are direct quotations from news reports and news headlines: (1) Prime minister rides the info-highway (2) White House counts two million cybertourists (3) Potholes along the information highway: if this is the highway, I'd rather hitchhike along a country road (4) AT&T stalled on the info-highway (5) Clearing the roadblocks on Japan's information highway (6) Congress suffers wreck on info highway (7) Billions parked on the info-highway—R&p. Jack Fields (R-Texas), Chairman of the Subcommittee on Telecommunications and Finance, thinks industry will respond to telecom reform with billions of dollars in investment: "I'm convinced that the people we talked with have tens of billions of dollars parked on the side of information superhighway waiting for us to pass a piece of legislation to give definition and cer­ tainty." (8) Ride sharing on the infobahn—A leading technology merger and ac­ quisition firm sees more convergence in the future. As big business seeks to enter the third wave, it will be looking for easy ways to hitch a ride on the information superhighway: "If Blockbuster can be acquired for $7.9 billion by Viacom, that says that anybody can be a target." (9) CELLULAR: It's still a bumpy highway for data Most of us typically think of the information highway as something which allows us to move in space, such as allows us to visit—albeit virtually—the White House. I call this the CYBERSPACE case of the INFORMATION

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

187

HIGHWAY metaphor because we envision ourselves and our computers as traveling through space to another destination. Sometimes we travel to obtain information along the fiber-optic roadways of the internet; at other times we take joyrides without a particular destination in mind. But making full sense of many examples of the metaphor requires a second, slightly different version of the metaphor. Few people realize that there are actually two distinct INFORMATION HIGHWAY metaphors, but when the subsequent article dis­ cusses both problems with cellular modems and problems with government regulation, we might realize that the word 'bumpy' in example (9) is a pun as it is meaningful in two distinct but related cases of the INFORMATION HIGHWAY metaphor system. The word 'bumpy' can refer either to the fact that the cellular transmission of data is still error-prone compared to cabled trans­ mission, or it can refer to the Federal Communications Commission regulatory hurdles companies face in building a cellular information highway in the United States. The former understanding rests on the CYBERSPACE case of the meta­ phor: errors in data transmission are technological obstacles (bumps) between us and our destinations on the information highway. In what I call the CYBERFUTURE case of the metaphor, the information highway is a road through time rather than through space. In this case the in­ formation highway is a road leading into the future down which we (as persons, as corporations, as nations) must travel. If a person, corporate or national entity does not build and drive on this information highway, then it, like AT&T in example (4), will remain hopelessly stalled on the way to the future and become a backward nation, corporation or person. At the end of the cyberfuture's in­ formation highway lie visions of technological utopias: for the U.S. Democrats, new job creation rivalling the new job creation during the construction of the interstate highway system; for the U.S. Republicans, the unregulated sale of information-as-goods which comprises the marketplace of ideas. Just as the bumps on the road through cyberspace are the technological glitches which slow our travel on the information highway between one computer and another, bumps on the road through the cyberfuture are the regulatory hurdles standing between us and information nirvana. As another example of these dual metaphor systems, consider the two dif­ ferent ways in which something can be just around the corner on the informa­ tion highway. In the CYBERSPACE version, this expression can mean that the right information is never much further than a few mouse clicks away (providing one knows where to look) but in the CYBERFUTURE version saying that something is just around the corner usually means that the something—e.g., digital movies-on-demand or interactive television—will be available soon. In the former usage we highlight the proximity of the goods (information), but in the latter usage we highlight the proximity (or obstacles) to new services in the near future.

188

TM ROHRER

These dual metaphors are so intertwined in our experience that they are often confounded. In part this confounding is due to the fact that although the two cases are distinct metaphor systems, they can be deliberately blended together by skillful politicians and other media voices because the two systems frequently cohere in their metaphorical entailments. However, the two cases do not always cohere in every entailment and, as I will argue in section 6, these differences shape much of the debate over U.S. information science policy.

3· Mapping the INFORMATION HIGHWAY From these examples, I can begin working out some preliminary conceptual mappings for the dual INFORMATION HIGHWAY metaphors. A conceptual mapping is given as a list of the correspondences operative between the source and target domains of a metaphor. However, the conceptual mapping of a meta­ phor is more than a mere list of corresponding words; it is also a claim about a dynamic and flexible set of activation patterns in the brain (Damasio 1995; Edelman 1992). The use of arrows to link the source domain elements to the target domain symbolizes the active and dynamic character of the conceptual mapping in virtue of which a metaphor is a process of thought, not a fixture of language. As Lakoff (1993: 208) puts it, metaphors are "not just a matter of language, but of thought and reason. The language is secondary. The mapping is primary in that it sanctions the use of source domain language and inference patterns for target domain concepts." In a conceptual mapping, the initial or ontological correspondences between domains serve as a springboard for map­ ping inferential relations and entailments between the domains (Gentner & Gentner 1983; Lakoff 1993). After mapping the major features of the terrain in the dual systems, I discuss further evidence for these mappings from the infor­ mation policy debate. But before proceeding, I want to consider the objection that the science of metaphors and mappings is fuzzy, arbitrary and so lacking in rigor as to be unfalsifiable. However, as cognitive semanticists are describing dynamic processes which change over time, some fuzziness is inevitable—because change is an intrinsic part of the phenomena. Human beings are continually learning, re­ evaluating and modifying their metaphors and their conceptual mappings at all times. Nonetheless, I think there are important strategies which minimize this degree of fuzziness. One important way is to acknowledge that some elements in the mapping are more stable than others; thus, mappings typically try to present the most stable portions of the metaphor first. Another source of fuzziness often results from the fact that some elements of the source domain map equally well to more than one item in the target (for example, vehicles may map to the com­ puter, the operating system, or a particular software application). However, we can describe the pressures which constrain how multivalent mappings are re-

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

189

solved; most dramatically, they are typically constrained by how other important elements of the mapping are instantiated in a particular case (Spellman, Ullman & Holyoak 1993). Thus if vehicle maps to computer, the engine will typically map to the operating system and the steering wheel to the software application. Thus, the important point to remember about conceptual mappings is that they are necessarily constraint-based descriptions of dynamic systems rather than formal logic-based descriptions of discrete-state systems. That is, metaphor sys­ tems are necessarily indeterminate but not necessarily unpredictable. As with the mathematics of chaotic systems, we may not predict exactly where the drop of water will fall or to what element of the target certain elements of the source will map, but we can offer a description of the range of possible targets. Fi­ nally, it is possible to get a conceptual mapping clearly wrong in at least two ways: first, the mapping between the elements may be unsupported by the evi­ dence—I have seen no evidence supporting a mapping between vehicles and the internet's cabling; second, the metaphor may be stated at the wrong level of generality. A mapping of the CYBERSPACE case, in which we envision ourselves as ^traveling through cyberspace along the cable-highways in search of informa­ tion-goods, is given in Figure 1. The most stable elements of the mapping are given first. For example, the strong visual similarities between highways and cables makes it a particularly stable element of the metaphor; visual similarities such as these are part of what Mark Johnson (1987: 29) means by his term "image-schematic structure." An image schema—defined as a recurrent motor or visual pattern common to the activities of bodily experience—typically un­ derlies each mapping (or cluster of mappings) in the metaphor. However, image schémas are not just visual, but embodied; they invoke bodily experiences which stretch across multiple sensory modalities. In fact, these other modalities can be so strong that the metaphor can be extended to cover cases where no visual in­ formation is present. As an example consider the case of the information sky­ way: not only do the snarl of cables connecting the various computers, periph­ erals, etc. which comprise the information highway system visually resemble the snarl of freeways, arterials, collector roads and residential streets which make up the asphalt highway system, but our experiential knowledge of the highways as that which carries goods is so strong that we find it relatively easy to con­ ceive of non-physical highways such as satellite transmission links and broadcast TV (i.e., the information skyway). This non-visual extension of the highway element seems intuitive because the first three elements of the mapping cohere tightly together. If cyberspace is the intangible space in which informationgoods travel, then the fleeting and ephemeral pathways of a satellite link can be a highway as real for information as an asphalt highway is for cars, trucks, and tangible goods. Thus I have used the more abstract term 'transmission pathways' rather than 'cables', because the mapping of highway is underlaid by the image schema of a path.

190

TIM ROHRER

Figure 1. A mapping of the INTERNET IS A HIGHWAY (CYBERSPACE case) A mapping of the CYBERFUTURE case is given in Figure 2. In this sys­ tem, the highway maps to a road stretching off into the horizon, where the road is the information highway and the horizon is the future. Though these informa­ tion highways can still lead us to places in space, their most important function is the destination to which they lead in time—a society transformed by informa­ tion technology. Traveling along this information highway is moving into the future; thus motion in space is refigured as motion in time. We are on a journey into the future in which various new technologies provide us with our vehicles and in which the quality and amount of information available fuels our rate of travel. The other elements of the mapping are refigured to accommodate these shifts, especially that of the role of information. Instead of transporting infor­ mation as goods, in this system the technologies transport human beings as goods—people, nations and corporations—into the future. Rather than focusing on the user's understanding of the internet, this case is centered on the telecom­ munication experts' understanding of the internet as fueling social change. In the cyberfuture the driver's seat is occupied by the expert instead of the user (though the journey will supposedly transform the passengers into drivers and the users into experts). Having set out the basic structure of the dual mappings, I now turn to dis­ cussing how metaphorical reasoning shapes social policy. I will begin by illus­ trating how Gore's use of the information highway metaphor shapes reasoning about science policy and then discuss how some critical responses are also shaped by the dual cases of the metaphor. I will conclude by examining what it means for metaphors to be constitutive of social policy.

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

191

4. Metaphorical Reasoning and Social Policy: How Metaphorical Inferences Work When U.S. Vice President Gore announced the Clinton administration's Na­ tional Information Infrastructure initiative in December of 1993, he exploited both versions of the INFORMATION HIGHWAY metaphor. Analyzing his speech illustrates how reasoning about social policy is being shaped by the information highway metaphor. Consider the following excerpts from his speech: (10)

It used to be that nations were more or less successful in their com­ petition with other nations depending upon the kind of transportation infrastructure they had. Nations with deep water ports did better than nations unable to exploit the technology of ocean transportation. Af­ ter World War II, when tens of millions of American families bought automobiles, we found our network of two-lane highways completely inadequate. We built a network of interstate highways. And that contributed enormously to our economic dominance around the world. (11) Today, commerce rolls not just on asphalt highways but along in­ formation highways. And tens of millions of American families and

192

I ROHRER

businesses now use computers and find that the 2-lane information pathways built for telephone service are no longer adequate [...] This kind of growth will create thousands of jobs in the commu­ nications industry. To understand what new systems we must create though, we must first understand how the information marketplace of the future will operate. One helpful way is to think of the National Information In­ frastructure as a network of highways—much like the Interstates be­ gun in the '50s. These are highways carrying information rather than people or goods. And I'm not talking about just one eight-lane turnpike. I mean a collection of Interstates and feeder roads made up of differ­ ent materials in the same way that roads can be concrete or macdam—or gravel. Some highways will be made up of fiber optics. Others will be built out of coaxial or wireless. But—a key point—they must be and will be two way roads. These highways will be wider than today's technology permits. This is important because a television program contains more infor­ mation than a telephone conversation; and because new uses of video and voice and computers will consist of even more information moving at even faster speeds. These are the computer equivalent of wide loads. They need wide roads. And these roads must go in both directions. I have quoted extensive passages from Gore's speech for two reasons. First, I argue in this section that Gore's use of the information highway metaphor ex­ tends the mappings of both the CYBERSPACE and the CYBERFUTURE meta­ phor systems in a manner which has meaningful (and controversial) implications for social policy. I aim to show how metaphorical extensions rely on the infer­ ential structure of the source domain to sanction the projection of a parallel in­ ference in the target domain. While such inferences often go a little awry due to slight but significant incongruities between the domains, a careful working out of the differences in the inferences engendered by the metaphor can still produce useful knowledge. This process of working out incongruities in domains in po­ litical rhetoric and legislative definitions is strong evidence for the claim that metaphors can come to constitute social policy. Second, I will argue in the next section that Gore's speech is a remarkably successful conceptual blend of the CYBERSPACE and CYBERFUTURE metaphor systems, and in generating a blend works many of the incongruous entailments between the two systems. Finally, this conceptual blend has been so successful as to provoke considerable criticism, which I will discuss in the penultimate section of this paper.

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

193

In example (10) Gore explains that in the future nations will be more or less economically successful according to how sophisticated their information infra­ structure is, arguing that contemporary "commerce rolls not just on asphalt highways but along information highways. " Consider how this metaphor draws on the CYBERSPACE version of the metaphor: information highways are con­ ceptualized as transmission pathways carrying commercial goods. Next, Gore extends the metaphor: "And tens of millions of American families and busi­ nesses now use computers and find that the 2-lane information pathways built for telephone service are no longer adequate ..." Why does Gore highlight the 'width' of a highway? Width is an element of the source domain which does not seem to map well to the target. After all, anyone familiar with cabling knows that there is no necessary connection between the width of a cable and the amount of information it carries; for example, thick-wire ethernet cable carries less information than thin-wire ethernet cable. However, in characterizing phone cables as both two-lane highways and inadequate to meet demand, Gore draws on our mundane understanding of how narrow two-lane asphalt highways be­ come choked with traffic as the demand for travel grows. Just as the width of an asphalt highway is proportional to the amount of traffic it can carry, the width of the information highway needs to be proportional to the amount of informa­ tional traffic it will carry. Presumably, if the existing highway (whether asphalt or phone cables) is too narrow to meet the demand, the highway must be wid­ ened to accommodate the traffic. That is precisely what Gore proceeds to argue. In the final paragraph of ex­ ample (11) Gore elaborates on the need for wider information highways. He spells out the source of the demand for wider highways: "This is important be­ cause a television program contains more information than a telephone conver­ sation; and because new uses of video and voice and computers will consist of even more information moving at even faster speeds." These new video and voice applications require so much more information to be transmitted quickly that he argues that they are "the computer equivalent of wide loads. They need wide roads." Wide loads, he points out, need wide roads; hence we need to plan to widen the information highway. I have developed a schema for diagramming metaphorical inferences which makes explaining how metaphorical reasoning works much more straightforward (Rohrer 1995). In Figure 3 I have both diagrammed Gore's inference and pre­ sented the general form of the schema. On the left side of the parallel lines is the source domain inference, and on the right side is the inference which paral­ lels it. The things known about the source domain are listed on the left side with their corresponding metaphorical projection listed on the right. The conclusions to the metaphorical inference—the non-metaphorically entailed understanding and the metaphorically entailed understandings—are listed below horizontal bar. In this particular case, in the source domain we know that conventional high­ ways carry goods, that wider highways can carry more and larger loads of

194

TIM ROHRER

goods, and that vehicles which transport larger loads of goods need wider roads. From those premises Gore implicitly draws a practical conclusion about needing to build wider highways in order to carry these goods, but his explicit conclu­ sion is that we should transfer this reasoning to information policy: namely, that we should build a wider information highway. While it is tempting to argue with Gore's reasoning (and I have many examples of people who do), the important point is that the metaphor is constitutive of social policy by providing the policy with the model within which the reasoning about the meaning of the policy takes place. The schema illustrates this point by diagramming how metaphorical infer­ ences produce parallel knowledge structures. "Information Highway" ("INTERNET IS A HIGHWAY") CYBERSPACE case

HIGHWAYS (source)

INTERNET (target)

Highways carry goods As we make highways wider they can carry more cars and bigger loads of goods Large tractor-trailer trucks and other wide loads need wider highways Therefore, we must build wider highways to accommodate the larger loads of goods

The internet carries information Technological advances (fiber-optics etc.) make it possible for information highways to carry more information New video and voice applications which use more infomation need more advanced cabling and more bandwidth Therefore, we must build wider information highways to accommodate the larger loads of information as goods

Metaphor in 'TARGETIS SOURCE" form Source domain

Target domain

things known about source do­ main non-metaphorically entailed under­ standing in source domain

projections of things known about the source domain onto target domain metaphorically entailed understanding in target domain

Figure 3. Example and schema of a metaphorical inference

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

195

5. Conceptual Blending on the Information Highway: Gore's Vision of the Cyberfuture The previous explanation of Gore's use of the INFORMATION HIGHWAY metaphor is, however, only half of the story, as I have been deliberately over­ simplifying matters by presenting an analysis of the speech solely in terms of the CYBERSPACE version of the metaphor. The CYBERSPACE dual is the static side of the information highway metaphor system, hiding the dynamic transfor­ mative power of the information highway as an agent of social change. In con­ trast, the CYBERFUTURE metaphor system foregrounds the power of informa­ tion technology to transform our lives as time passes and we proceed down the information highway into the cyberfuture. The CYBERFUTURE dual is the dynamic figure to the static ground of the CYBERSPACE dual. Gore uses the dual cases to complement and balance one another as he articulates his vision of the cyberfuture. This kind of combination of two metaphor systems is one type of conceptual blend as Turner and Fauconnier (1995) describe them. In a con­ ceptual blend a mapping of one mental space into another occurs for example, when we say that the first-place yacht in a transatlantic sailing match is currently fifteen hours ahead of last year's race, we superimpose the position of the yacht from previous years onto the current race, creating a conceptual blend of the two mental spaces which makes the comparison possible. Gore's speech blends the imagery of the CYBERSPACE system together with the imagery of the CYBERFUTURE system into a coherent information policy. Probably no part of Gore's speech has drawn more criticism than his vision of the cyberfuture, because Gore uses the language of traveling into the future on the information highway to emphasize that the federal government should play a significant policy-making role in developing the national information in­ frastructure, a position highly unpopular with the internet's largely antigovernment libertarian constituency. In the opening paragraph of (10) Gore makes four basic points about the source domain: first, that nations have histori­ cally been more or less successful depending on their transportation infrastruc­ ture; second, that postwar social forces fueled a rising consumer demand for automobiles in the U.S.; third, that the extant U.S. network of two-lane high­ ways was inadequate to meet the increased demand; and fourth, that the U.S. pursued a purposeful national transportation policy to build an interstate high­ way system which contributed to U.S. economic success around the world. In the second paragraph of (11) Gore refigures those four points when he intro­ duces the CYBERSPACE case by remarking, "Today, commerce rolls not just on asphalt highways but along information highways." This statement compels us to consider how the four points about highways have implications for the target domain. Gore even maps the third point across for us in his subsequent point about many Americans finding that the existing telecommunications net­ work is already inadequate to meet demand.

196

ΉΜ ROHRER

After a long (and omitted herein) litany of examples illustrating how information technology is improving lives and spurring economic growth, Gore uses the CYBERFUTURE case to articulate the impetus behind the National Information Infrastructure proposal: (12)

But the biggest impact may be in other industrial sectors where those technologies will help American companies compete better and smarter in the global economy. Today, more than ever, businesses run on information [...]. If we do not move decisively to ensure that America has the information infrastructure we need, every business and consumer in America will suffer. What obstacles lie ahead in the rush to the future!

Example (12) illustrates how Gore skillfully blends the CYBERSPACE to the CYBERFUTURE version of the metaphor. Gore has previously set up a mental space in which we imagine information as commercial goods traveling along information highways to us and to other people, the image established by examples (10) and (11). In a second mental space of the CYBERFUTURE we imagine that we are moving along the information highway and moving into the future, in vehicles (businesses) which are fueled by information. To conclude that governmental action is necessary to plan the information highway, Gore blends these two spaces together into a third integrated conceptual space. In example (12) Gore establishes a blended space in which moving along the information highway means acting to ensure that the infrastructure of the information highway is built to take us to the right place in the future, one which will ensure both that the U.S. economy will remain strong in a global economy and that U.S. consumers will maintain a high standard of living due to the ability of the information highway to bring us quality goods on demand. If we do not move along the cyberfuture's information highway, our cyberspace economy will suffer. To move decisively into the future would be to debate and enact a national information policy, which would presumably clear away some of the regulatory obstacles which lie ahead on the cyberfuture's information highway. In the blended space, movement on the highway is simultaneously both building the information highway's pathways (CYBERSPACE) and travel­ ling on the information highway's road into the future (CYBERFUTURE). Gore concludes in (13) that the federal government needs to lead the way in supplying the principles which shape the information marketplace of the future: (13)

That's what the future will look like-say, in ten or fifteen years. But how do we get from here to there! This is the key question for the government. It is during the transition period that the most com­ plexity exists and that government involvement is the most impor­ tant.

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

197

The 'here' and 'there' in example (13) are not simply spatializations of times, but refer to the status of currently ill-defined information legislation and an un­ built information infrastructure ('here'), and to a future of better-defined infor­ mation legislation and a built information infrastructure ('there'). Thus Gore blends the two versions of the INFORMATION HIGHWAY metaphor in order to argue for government involvement in planning both the building of the infor­ mation highway and the destination to which it will take us. Figure 4 diagrams the metaphorical inference in this blended space. "Information Highway" CYBERSPACE and CYBERFUTURE blended

HIGHWAYS (source)

INTERNET (target)

Nations are more or less successful depending on their transportation in­ frastructure Postwar social forces fueled a rising consumer demand for automobiles

Nations will be more or less successful de­ pending on their information infrastructure

The existing network of two-lane highways was inadequate to meet the increased demand As a nation, the U.S. pursued a pur­ poseful transportation policy to build an interstate highway system In pursuing a national transportation policy, we as a nation embarked on a journey to fulfill the goals in it The process of fulfilling the mission of that transportation policy by building highways was itself a transformative journey The roads of the interstate highway system built by transportation policy also embody the roads traveled in pursuing a deliberate transportation policy Thus, simultaneously building and traveling the interstate highway system transformed the U.S. into the major economic power of the post-war era

Contemporary social forces are fueling a rising demand for computers and other in­ formation "appliances " The present network of narrow information highways is inadequate to meet future de­ mand As a nation, the U.S. can now pursue a purposeful information policy to build an information highway system In pursuing a national information policy, we as a nation can embark on a journey to fulfill the goals in it The process of fulfilling the mission of that information policy by building information highways can itself be a transformative journey The roads of the information highway sys­ tem built by information policy also embody the roads traveled in pursuing a deliberate information policy Thus, simultaneously building and traveling the information highway system will trans­ form the U.S. into a major economic power in the information era

Figure 4. A blended metaphorical inference

198

TIM ROHRER

By introducing the proposed revisions to telecommunications law as embarking on a journey to implement a national information highway, Gore extends the metaphorical inference previously outlined into a vision of the future. This ex­ tended inference is diagrammed in Figure 4. In this conceptual blend, our cul­ tural understanding of how difficult journeys can transform the travelers is blended with our understanding of how our journey down the information high­ way into the future will transform our lives. In the blended space, successfully meeting the challenges of the cyberfuture becomes narratively mapped onto suc­ cessfully completing a spiritual quest. Setting out a proposal for information policy is to embark on the journey, and the transformative potential of informa­ tion is the potential of the journey to effect transformation of the national spirit. As the formation and the debate over information highway policy (the route of the journey) progresses, the society (the travelers) are transformed. However, since choosing the route involves both building the information highway and forming information policy, both the economy and the principles which govern it will be transformed. Building the information highway is both laying the con­ duits on which information travels and transforming the U.S. federal regulatory system to provide needed legislative definitions to adjudicate commercial trans­ actions in the future. Since the information highway is both the means to and the end of its own construction, it is a particularly rich example of how metaphors constitute social policy. In Gore's blended space, the information highway is both figure and ground; we must travel the highway even as we build it.

6. Policing the Infobahn: Cybercrime from Software Piracy to Criminal Bureaucrats Intellectual property, whether in printed or electronic form, is often subject to theft as it travels from author/publisher to end user. Such theft en route is typi­ cally called piracy. Protection from intellectual property piracy has grown sub­ stantially more difficult as reproducing information has become easier by vari­ ous technological developments from the printing press to the photocopier to the computer. The concept software piracy shaped and still shapes most of the con­ temporary controversies about copyright law and intellectual property. How­ ever, while concerns about piracy continue, consider the shift in the identity of the criminals in the following set of examples culled from the media: (14) (15) (16)

Finnish executives jailed for software piracy Info highway needs policing: Copyright recommendations for the digital age Trespassing on the internet: I think intellectual property is more like land, and copyright viola­ tion is more like trespass. Even though you don't take anything away from the landowner when you trespass, most people understand and

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

(17) (18)

199

respect the laws that make it illegal. The real crime in copyright violation is not the making of the copies, it's the expropriation of the creator's right to control the creation. (The founder of ClariNet Communication Corp). Highway cops: How federal regulation of the telecommunication in dustry limits potential growth for the information superhighway. Information highway robbery: Government regulation of cable indus­ try impedes development of information superhighway, derails Bell Atlantic/TCI merger.

In the examples (14) through (16) the criminals are those who steal infor­ mation (i.e. hackers), but in (17) and (18) the criminals are those who block progress (i.e. government and other biisybodies). This difference can be ex­ plained by considering how the CYBERSPACE and CYBERFUTURE meta­ phors have a figure-ground structure which engenders reversed metaphorical entailments about what constitutes cybercrimes. The concept of software piracy is entailed by the CYBERSPACE version of the INFORMATION HIGHWAY metaphor. The logic engendered by the meta­ phor is straightforward: If information is a good, and the goods are being shipped to a destination for purchase, then the theft of the goods while en route to purchasers is an act of piracy. The goods (typically software programs) are hijacked and illegally distributed, thus depriving the creator of potential sales. Since software programs are easily copied, enforcement of software copyright law has tended to focus on those who widely distribute the pirated goods, either when repackaged for sale or when distributed over telecommunications networks such as the internet or electronic bulletin board services. Such criminals are dubbed software pirates. The most important point about the CYBERSPACE system is not that soft­ ware theft is understood as piracy; instead it is the way in which we conceptual­ ize information which engenders the piracy entailment. In conceptualizing in­ formation as goods, information becomes an object which can be bought, sold, traded, and stolen. Like land, information becomes a kind of property —something which can be marked out, paced off, and bounded like land. Once information is defined as a property, the title to it can be traded as a commodity in much the same way as any other object. The CYBERSPACE understanding of information is precisely the one embedded in the tradition of intellectual property law. Eric Nee (1994) makes this issue plain in example (19): (19)

Similarly, just as the assigning of private property rights to land and capital were critical for the First Wave and Second Wave economies to take off, so is the extension of property rights to the intellectual world necessary for the Third Wave to take off. Some on the Utopian fringe believe that cyberspace will result in the end of intellectual

200

TEM ROHRER

property rights with all knowledge free and available, but all that would result from that approach would be the collapse of the system. The conceptualization of information as goods (and as properties) relies on an understanding of information as fundamentally static; that is, as something definite which can either be packaged for sale or parceled out for sale. No one denies that the peculiarly fluid quality of electronic information creates problems for understanding information as goods in the CYBERSPACE metaphor. Consider the situation of copyright law: The information available over the world wide web, for example, is copied every time it is accessed— posing peculiar problems for copyright laws geared to dealing with information encoded in tangible objects like books (or their CD-ROM editions). In fact, it is precisely problem areas such as this one which have motivated the intense de­ bate about the CYBERFUTURE case as an alternative. For example, radical opponents (e.g. Richard Stallman and the Free Software Foundation) of copy­ right law as currently applied to software have charged that since the software pirate takes nothing tangible away from the creator, the piracy entailment of the CYBERSPACE metaphor is not only inappropriate but downright misleading about the role of the software pirate in bringing about the cyberfuture. In char­ acterizing intellectual property crime as more like trespass on land than hijack­ ing goods, example (18) seeks to offer a resolution from within the CYBERSPACE framework to this difficulty about what is taken away in an act of software piracy. While it remains to be seen how workable a conception this retooled anal­ ogy will be, I believe this defense can only be successful as long as the terms of the debate are defined by the CYBERSPACE metaphor system. Furthermore, this resolution cannot be successful in the CYBERFUTURE system, because one cannot trespass on information if it is conceptualized as a dynamic substance (fuel) rather than as a static substance (land/goods). The static understanding of information in the CYBERSPACE system is anamethea to the radical Utopians; they propose re-envisioning our intellectual property system around precisely those fluid qualities of information which are problematic to the CYBERSPACE system. In the CYBERFUTURE metaphor system, software pirates are no longer criminals but instead are the heroes who are bringing technological utopia closer. The CYBERFUTURE system conceptualizes information as fundamentally a dynamic process undergoing constant change—that is, as a transformative force bringing about the future. In the CYBERFUTURE case of the INFORMATION HIGHWAY metaphor, information is conceptualized as the fuel of the future. Nee's Utopian fringe, advocating the abolition of intellectual prop­ erty, are simply being true to the entailments of the CYBERFUTURE metaphor. The logic engendered by the metaphor is simple: If information is what fuels the rapid process of technological change, and if we both desire to be (and in fact

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

201

are) on our way into the information age, then anyone or anything which im­ pedes our motion toward that destination is blocking our right-of-way and slowing our progress. Just as with the shift from a stationary to a moving ob­ server changes the entailments in metaphors for time (Lakoff 1993; Johnson 1993), the shift from a static to a dynamic conceptualization of information re­ verses the metaphorical entailments about what constitutes crime on the infor­ mation highway. Whereas the primary criminal activity in CYBERSPACE is software piracy, the CYBERFUTURE view identifies the primary criminal activity as blocking the road to the future. Two possible sets of culprits are usually fingered: gov­ ernment bureaucrats and intellectual property lawyers. Government bureaucrats are seen as interfering with and slowing down the rush toward information nir­ vana by parasitically saddling the entrepreneurs and technologists with regula­ tions designed to ensure public access and participation. Intellectual property law is seen as an attempt to privatize and make static the dynamic and transfor­ mative power of traveling the information highway by slowing down the rapid spread of information. For example, instead of envisioning intellectual property laws as protecting the property rights of the software programmer, in this con­ ceptualization they fence out software programmers, thereby depriving them of useful tools which they need to share as they build the information highway. Since both government regulations and intellectual property laws are perceived as unnecessary brakes on the rush to the technological utopia of the future, they are both criminal activities. In fact, instead of advocating the criminalization of copyright violation, some of the most radical techno-utopians argue that we should revere those hackers who "liberate quarantined information" (i.e., pirate software) when they do it in order to advance us on the road to techno-utopia. Like the gun-slinging outlaws who became town marshals in the mythology of the American West, in the mythology of the cyberfuture's electronic frontier the hackers become the law-givers, and the old law and order set is reconstituted as the criminals. The CYBERFUTURE system motivates different responses to the problems of intellectual property rights. While in example (16) the founder of Clarinet Corp. argued that copyright violation of electronic property was better under­ stood as trespass rather than piracy, John Perry Barlow (as quoted in Dyson, Gilder, Keyworth & Toffler 1994) insightfully argues that the central issue is the conceptualization of electronic information as objects or properties, and that the real problem—how creators are to continue being paid for their work—can be solved within the CYBERFUTURE system if we conceive informational goods as performances: (20)

One existing model for the future conveyance of intellectual property is real-time performance, a medium currently used only in theater, music, lectures, stand-up comedy and pedagogy. I believe the con-

202

TIM ROHRER

cept of performance will expand to include most of the information economy, from multi-casted soap operas to stock analysis. In these instances, commercial exchange will be more like ticket sales to a continuous show than the purchase of discrete bundles of that which is being shown. Barlow's imaginative suggestions hinge on a dynamic conceptualization of information. Information is valuable only insofar as it fuels transformation; hence what is needed is not better protection of the information from others but simply a way to pay information-creators based on the timeliness and usefulness of their information; presumably the more transformative the information, the more money the creator will make. Thus Barlow argues that if electronic infor­ mation is different in kind from information which must be encoded in physical objects (books, CD's, celluloid, etc.), then intellectual property law must be radically reconceived with real-time performance as the normative conception of what constitutes a good. The conceptualization of information as dynamic com­ pels him to conclude not only that electronic information is in fact different in kind, but that the vision of the CYBERFUTURE requires that we must allow this difference to transform our reasoning about intellectual property. Finally, proponents of the CYBERFUTURE system advocate that their vi­ sion of the cyberfuture is an historical inevitability, for the future will arrive and it will necessarily be different from the past. Like all Utopian visions, it seeks to impress us not only with the quality of life in this idealized world, but with the inevitability of this idealized world. Thus information is conceptualized as a tidal wave of change which sweeps us along, or as a revolution which funda­ mentally alters our governmental institutions. Moreover, since proponents of this view see themselves as scientists and technologists responsible to natural law rather than human law, they see our arrival at a technological utopia as in­ evitable—a position perhaps best summed up by the old cliché "one doesn't ar­ gue with the laws of physics." The painful human dislocations caused by the onslaught of progress are regrettable but inevitable.

1. Conclusion Different problems and solutions are figured prominently by each of the dual cases of the INFORMATION HIGHWAY metaphor system. Conceptual map­ pings are resolved in accordance with the valency of key elements in the source domain, resulting in the two dual cases. These dual cases drive various meta­ phorical inferences which both set problems for and drive solutions to problems in information technology. Gore's highly successful conceptual blend of the dual metaphor systems traded on blending the building of the information highway's nathwavs with the Dower of information to fuel future social change, setting the

CONCEPTUAL BLENDING ON THE INFORMATION HIGHWAY

203

stage for a complex metaphorical inference about the role and degree of gov­ ernmental involvement in constructing this highway. Critics of Gore's conclu­ sion argued that we should reject the blended space and the inferences engen­ dered by the CYBERSPACE metaphor in favor of re-imagining intellectual property law using the dynamic conception of information as understood in the CYBERFUTURE system. Finally, taken as a whole, these examples of the debate over information policy in the U.S. serve to illustrate how metaphors can come to be constitutive of a social policy. For example, both proposals concerning reimbursement for intellectual property illustrate how the debate over information policy is shaped by which version of the INFORMATION HIGHWAY system serves as the metaphorical ground. For Barlow and the many other critics of U.S. informa­ tion technology policy who are dedicated to re-envisioning the future in a way consonant with a dynamic conceptualization of information, the INFORMA­ TION HIGHWAY metaphor system is highly misleading—in no small part be­ cause they oppose Gore's use of a blended information highway metaphor to argue for a federal regulatory role in planning for the future. Successfully countering the success of Gore's blend requires offering an equally transforma­ tive view of their own. Such a project may even require abandoning the INFORMATION HIGHWAY metaphor entirely and proposing radically different metaphors to prefigure the transition that the dynamic view of information produces; for ex­ ample, Dyson et al. (1994) propose that the internet is better understood as a bioelectronic environment: (21)

More ecosystem than machine, cyberspace is a bioelectronic envi­ ronment that is literally universal: It exists everywhere there are telephone wires, coaxial cables, fiber-optic lines or electromagnetic waves. This environment is 'inhabited' by knowledge, including in­ correct ideas, existing in electronic form. It is connected to the physical environment by portals which allow people to see what's inside, to put knowledge in, to alter it, and to take knowledge out. Some of these portals are one-way (e.g. television receivers and television transmitters); others are two-way (e.g. telephones, com­ puter modems).

To highlight the dynamic character of information, the INTERNET AS ENVIRONMENT metaphor begins by postulating that electronic information is alive. It remains to be seen whether this metaphor or any other offering by ad­ ministration critics will be able to supplant Gore's blend of the INFORMATION HIGHWAY metaphor as the dominant metaphor guiding U.S. information pol­ icy.

204

ΤΊΜ ROHRER

References Damasio, Antonio. 1994. Descartes Error: Emotion, Reason and the Human Brain. New York: G.P. Putnam & Sons. Dyson, Esther, George Gilder, George Keyworth and Alvin Toffler. 1994. Cyberspace and the American Dream: A Magna Carta for the Knowledge Age. Internet document (release version 1.2) available from the Progress and Freedom Foun­ dation: http://www.pff.org Edelman, Gerald. 1992. Bright Air, Brilliant Fire: On the Matter of Mind. New York: Basic Books. Gentner, Dedre and Donald Gentner. 1983. "Flowing Waters or Teeming Crowds: Mental Models of Electricity." In Dedre Gentner & A.L. Stevens (eds), Mental Models. Hillsdale, NJ: Lawrence Erlbaum Associates, 99-129. Gore, Albert. 1993. Remarks by Vice President Gore at the National Press Club on December 21, Washington D.C.: The White House, Office of the Vice Presi­ dent, Press Release. Johnson, Mark. 1993. Moral Imagination: Implications of Cognitive Science for Ethics. Chicago: University of Chicago Press. Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagina­ tion and Reason. Chicago: University of Chicago Press. Lakoff, George. 1993. "The Contemporary Theory of Metaphor. In Andrew Ortony (ed.), Metaphor and Thought, second ed. New York: Cambridge University Press, 202-251. Nee, Eric. 1994. Eric Nee Reflects on the Social Revolution Wrought by Information Technology. Internet document available from the Progress and Freedom Foun­ dation: http://www.pff.org. Rohrer, Tim. 1995. "The Metaphorical Logic of (Political) Rape: The New Wor(l)d Order. " Metaphor and Symbolic Activity 10: 115-137. Spellman, Barbara, J. Ullman, and Keith Holyoak. 1993. "A Coherence Model of Cognitive Consistency: Dynamics of Attitude Change During The Persian Gulf War." Journal of Social Issues 49: 147-165. Turner, Mark and Giles Fauconnier. 1995. "Conceptual Integration and Formal Ex­ pression." Metaphor and Symbolic Activity 10: 183-204.

Speech Act Metonymies Linda Thornburg & Klaus Panther Eötvös Lor and University of Budapest, Hungary and University of Hamburg, Germany 1. Introduction Metonymy is a well-known rhetorical device that is used in everyday communi­ cation in order to refer in an indirect way to people, things, and events. Com­ mon examples are: (1) a. The White House announced today ... [the President or spokesper­ son] b. Congress passed a bill ... [the legislators]  I'm taking Shakespeare this semester, [course on Shakespeare's works] d. I've got five volumes of Dickens on my bookshelf, [books] e. The first violin injured her hand, [the violinist] The phenomenon of metonymie reference as an indirect means of referring to things has been noted in traditional and classical rhetoric and poetics for more than two thousand years. The most comprehensive dictionary of the English language, the OED (Compact Edition), provides the following etymology and definition: (2) Metonymy: "adaptation of late L. metonymi-a, adopted from Gr. μετονυμία, lit. 'change of name'. [...] A figure of speech which consists in substituting for the name of a thing the name of an attribute of it or of something closely related." The oldest citations the OED provides go back to the 16th century. 1547 Hooper. Men seyeth that they admyt metonymian, and say under the forme of breade is the trew bodye of Christ. 1562 Cooper. The figure is named Metomymia: when the name of the thynge is geuen vnto the signe. 1573 Cartwright. The Apostle by a metonimie Subiecti pro adiuncto, do the giue to understand from whence ye assured persuasion doth spring.

206

LINDA THORNBURG AND KLAUS PANTHER

1625 Gill. Shebet signifies either a staffe, a truncheon, or Scepter, ... and so by a metonymia it may signifie authority. 1657 J. Smith. A metonymie of the effect, is when the effect or thing caused, is put for its cause. 1668 H. Moore. Here is a double Metonymie, Christ is put for the Doc­ trine of Christ, and Hope for the Cause of Hope. 1676 W. Hubbard. By times we are to understand things done in those times, by a metonymy of the adjunct. 1868 Bain. By what is called 'metonymy', the fact intended to be ex­ pressed is denoted by one of the adjuncts. The term adjunct above is used in the sense of "associated property." As can be seen easily, the definitions given in these sources are applicable to the present-day examples in (1). When saying The first violin injured her hand, it is not the musical instrument but rather the musician herself who is referred to by mentioning an adjunct or associated property, that is, her position as first violin. The traditional approach to metonymy has mainly been limited to how people refer to things and events in the "real" world. Cognitive linguists assume that metonymy is a much broader cognitive principle than has been previously recognized. In particular, we hypothesize that metonymy is instrumental in indi­ rectly accomplishing linguistic actions such as promises, offers, requests, sug­ gestions, and so on. In other words, just as a speaker can refer to a person by mentioning an attribute of that person, we will show that a speaker can perform a speech act by mentioning an attribute of that speech act. In other words, we want to demonstrate that indirect speech acts can be described in terms of me­ tonymie models (in the sense of Lakoff 1987; see also Gibbs 1994). This enterprise is both theoretically and pedagogically rewarding. On the one hand, it is a step towards reconciling pragmatics and cognitive linguistics. On the other hand, it illuminates an important aspect of native speaker commu­ nicative competence that is not transparent to the non-native learner. Apart from speech acts in the traditional sense, we will briefly consider other data including "hedged performatives" (Fraser 1975), and sentences about sense perceptions and mental activities corroborating our hypothesis that speech act metonymies are just special cases of more general metonymie principles. In this study we focus on what are called "conventionalized" or "stand­ ardized" indirect speech acts (Searle 1975; Gordon & Lakoff 1975, and the collection of articles in Cole & Morgan 1975; see also Bach & Harnish 1979) of the types that Searle calls directive, commissive and expressive (see Searle 1976 for a standard classification of illocutionary acts). According to Searle (1975), an indirect speech act is a speech act which is performed by means of another speech act. For example, in stating The dishes aren't done a speaker can indi­ rectly request of the hearer Do the dishes. In this case, the speaker performs one speech act (a statement) to accomplish a different speech act (a request). Many

SPEECH ACT METONYMIES

207

indirect speech acts have become conventionalized to such an extent that, to the native speaker, their indirect force is hardly recognizable, as in the well-known example Can you pass the salt!, which in most contexts is understood as a re­ quest and not as a question. The majority of our data were collected from the Lancaster-Oslo-BergenCorpus, which allowed us to extract a number of contextualized naturalistic ex­ amples instantiating the different types of indirect speech acts we wanted to analyze. The remainder of the paper will be organized as follows: In Section 2 we will introduce the notion of the action scenario, deriving from it speech act scenarios of the types called directive, commissive, and expressive. In Section 3 we will concentrate on the directive speech act type, that is, on the indirect re­ alization of speech acts like requests, orders, suggestions, and recommenda­ tions. In Section 4 we will present a tentative taxonomy of metonymies we ob­ served in our data base. Finally, we will draw some conclusions in Section 5.

2. The Action Scenario Speech acts are actions, and therefore share certain properties with ordinary non-linguistic actions. Like other actions they have different "phases" or "stages" or "conditions" for their successful performance. We assume that ac­ tions have at least the following characteristic features: (3) The Action Scenario (i)

The BEFORE: preconditions which enable a physical action, le­ gitimize a social action or motivate an action (including speech acts);

(ii)

The CORE and its RESULT: properties which define the action as such and the immediate outcome of a successful performance of the action;

(iii) The AFTER: intended or unintended consequences of the action which are not its immediate result. As an example of a linguistic action1 consider the (requestive) speech act conveyed by the imperative in (4): (4) Hand me that superglue. The first phase of a directive like (4), that is, the BEFORE, is the condition that the hearer must be able to carry out the requested action. As is well-known

208

LINDA THORNBURG AND KLAUS PANTHER

from speech act theory, this speech act is infelicitous if the enablement condition does not hold. It is important to see that on one level this condition is a prag­ matic or discourse precondition that (among others) licenses the performance of the request as a linguistic action. But on the level of non-linguistic actions, the enablement precondition is a real-world condition for the performance of the action, here the handing of the glue. The BEFORE also contains a mental atti­ tude of the speaker, in this case, the want of the speaker that the action be car­ ried out by the hearer. We regard this particular precondition as motivation or reason for carrying out both the speech act and the associated non-linguistic ac­ tion. If the preconditions2 are given, the next stage is the CORE (traditionally called the "essential condition" or "illocutionary point") of the speech act. We assume, contrary to Searle (1975) and in accordance with Wunderlich (1976), that a directive speech act essentially aims at placing the hearer under some more or less strong obligation to carry out the action requested. If this attempt or intention is successful, then the hearer is under a (more or less strong) obli­ gation to carry out the action. In other words, the hearer must, should or ought to perform the action, where the choice of the modal depends on the strength of the force of the speech act.3 We call this stage the PRAGMATIC RESULT (briefly the RESULT) of the speech act. The final stage here is called the PRAGMATIC CONSEQUENCE of the speech act (briefly, the AFTER). It focusses on the realization of the propositional content, in this case, the carrying out of the action requested of the hearer. Again, the AFTER is not a logically necessary result of the speech act, that is, in uttering a sentence like (4) the speaker does not necessarily achieve the (usually desired) consequence, in this case that the hearer will hand over the superglue. In conclusion then, the different stages of the speech act type directive can be summarized as in (5) where S stands for the speaker, H for the hearer, and Λ for the action requested. In what follows, the AFTER will refer to the usually intended sequel of the speech act. (5) Scenario for Directive Speech Acts (i)

the BEFORE:

Η can do A. S wants H to do A.

(ii)

the CORE:

S puts H under a (more or less strong) obli­ gation to do A. H is under an obligation to do A (H must/ should/ought to do A)

the RESULT:

(iii) the AFTER:

H will do A.

SPEECH ACT METONYMIES

209

The speech act types commissive and expressive will be analyzed analogously. (6) Scenario for Commissive Speech Acts (i)

the BEFORE:

S can do A. A is desirable to H. S intends to do A.

(ii)

the CORE:

S puts self under an (more or less strong) ob­ ligation to do A. S is under an obligation to do A (S must/ should/ought to do A).

the RESULT: (iii) the AFTER:

S will do A.

(7) Scenario for Expressive Speech Acts (i)

the BEFORE:

There is a fact p. S has an emotional attitude (EA) regarding p.

(ii)

the CORE: the RESULT:

S expresses an EA regarding p. S is regarded as having EA.

(iii) the AFTER:

[There are no general consequences holding for this speech act type.]

In this paper we demonstrate the metonymie function of only the BEFORE component in directive speech acts to illustrate in some detail our method of analysis. In the concluding section we place our findings regarding indirect speech acts along with other data in a more general framework of metonymie models.

3· The Metonymie Function of the BEFORE Component in Directives It is well-known that indirect directive speech acts can be performed by means of constructions like You can VP/ you VP. Examples of these types are given in (8)-(12).4 You can VP (8)

"Now you're here you can give me a lift back. "

210

LINDA THORNBURG AND KLAUS PANTHER

(9)

"No," he replied, "there is nothing to be anxious about, but I shall be obliged if you can give me a few moments alone."

(10)

"Well if you've all had sufficient you can make a move to the fire."

(11)

I'll leave you there, Charlotte, and you can double back to Mark.

Can you VP (12)

It did not take Gaffer long to explain what he wanted the Italian to do. "Well," he concluded, "what about it? Can you do it?"

In (8), (9), and (12) the speaker talks about the hearer's ability to perform a certain action (give a lift back/give a few moments alone/do it). Thus, the speaker highlights the precondition of ability for a directive speech act and in so doing performs a request. Moreover, a closer look reveals that the context pro­ vides additional clues that suggest a directive interpretation. The additional clue in these cases is another BEFORE condition, "desirability for the speaker," which is either derivable from world knowledge (as in example (8) with the phrase give me a lift) or expressed directly as in (9) with the phrase / shall be obliged and in (12) with he wanted. Examples (10) and (11) will be discussed below. Given the speech act scenario above, a distinction must now be made be­ tween those parts of the scenario which can metonymically "stand for" the en­ tire scenario (as the precondition of ability does in the examples just given) as opposed to components which merely "point to" the scenario. The following data in (13) and (14) from Levinson (1983: 347) will be used to elaborate and illustrate this distinction: Can I VP (13)

: S: C: S:

Do you have hot chocolate? mmhmm Can I have hot chocolate with whipped cream? Sure ((leaves to get))

(14)

C: S: C: S:

Do you have the blackberry jam? Yes. Okay. Can I have half a pint then? Sure ((turns to get it))

SPEECH ACT METONYMIES

211

Unlike the requests (8), (9), and (12), examples (13) and (14) involve the trans­ fer of objects. This necessitates an elaboration of the request scenario in the following way: (15)

Elaborated Scenario for Directive Speech Acts (Type: S requests of H that H provide some X for S) (i)

the BEFORE:

1. Ή can provide X for S' presupposes Ή has X' presupposes 'X exists.' 2. Ή can provide X for S' entails 'S can have X from H.' 3. S wants H to provide X for S.

(ii)

the CORE:

S puts H under a (more or less strong) obli­ gation to provide X for S. H is under an obligation to provide X for S (H must/should/ought to provide X for S).

the RESULT:

(iii) the AFTER:

H is willing to provide X for S. Ή will provide X for S ' entails 'S will have X.'

In examples (13) and (14) the Do you have X utterances have a different pragmatic status from the Can I have X utterances. Whereas the latter are inter­ preted as requests, as indicated by the subsequent compliant action of the ad­ dressee, the former are much less likely to be understood as requests for provid­ ing the object stated. The reason seems to be that the Do you have X utterances highlight a condition that is at the "periphery" of the request scenario or even outside it.5 This analysis is supported by the possibility of using the request marker please (see e.g. Sadock 1974) with Can I have X, less so with Do you have X, and, we hypothesize, even less so with the logically prior existential presupposition Is there (any) X.6 On the basis of these data, and more to follow, we propose the metonymie principles in (16)a. and b.: (16)

a. The more a speech act component is located at the periphery of the speech act scenario, the less likely that component will be in a "stand-for" (metonymie) relation to the scenario. b. The more components of a scenario present in a piece of dis­ course, the easier to draw inferences from the parts to the whole.

212

LINDA THORNBURG AND KLAUS PANTHER

In (16)a. we discriminate a relation between part and whole in which the part signifies the whole from a relation in which a (peripheral) part merely evokes or "points to" the whole without actually standing for it. A possible objection to this analysis might be that the Can I have X cases belong to the periphery of the request scenario, just like the Do you have X ex­ amples. The elaborated scenario in (15) shows that this objection does not hold. The BEFORE conditions under 1 are ordered along a presuppositional scale. In this sense, the existential condition X exists is more peripheral than H has X which in turn is more peripheral than H can provide Xfor S. This presupposi­ tional scale is in fact reflected in the sequential order of the conversational turns in examples (13) and (14). Both conversations start out by questioning a more peripheral condition (H has X) and then proceed to questioning a more central condition (S can have X from H). It should be pointed out that the more central condition S can have X from H is actually an entailment of the BEFORE condition H can provide X for S, that is, the entailed proposition stands for the entailing proposition. Another way of saying this is that there is a metonymic principle at work, namely, THE RESULT STANDS FOR THE ACTION (i.e. the receiver's having stands for the agent's giving). Thus the Can I have X utterances exemplify a case where two parts of a scenario stand in a metonymie relation. Moreover, our me­ tonymie framework allows us to neatly account for the common exploitation of the Can I VP pattern for conveying indirect requests. In (16)b. we simply state that the more clues available for interpretation, the easier it will be to retrieve the intended speech act scenario. This principle will be further demonstrated in subsequent examples. Examples (10) and (11) also highlight BEFORE conditions for directives. Here they seem to stand for suggestions or recommendations rather than re­ quests, due to the fact that (especially in (11)) speaker desire is hardly detect­ able. However, in (10), given our background knowledge of dinner parties, we may conclude that moving to the fire is desirable to the addressees (guests) and in (11) that "doubling back" is beneficial to Charlotte. We now turn our attention to the BEFORE component S wants H to do A, which speech act theorists have called the sincerity condition of requests. Con­ sider the following data from the LOB-corpus: / want (NP) to VP/I want NP8 (17)

7 want to marry you,' he said. 'We will live for ever in a little house by the sea. '

(18)

The father said to his boys, There's a bird apiece for you. / want you to fatten them up, and when they are fit for eating we'll have a great party and Nelly will become your mother.

SPEECH ACT METONYMIES

213

As stated above, we regard "sincerity conditions" as reasons or motivations for performing the verbal action. A clear case in which the mental attitude of the speaker most likely stands for a request is in (18). The speaker explicitly says what he wants the addressees to do, and thus indirectly requests the action be done. In contrast, (17) is less transparently a request in that the wanted action of the addressee is not overt in the utterance. But since the verb marry is symmet­ ric, it is safe to conclude that the speaker wants the addressee to marry him. Because a requestive interpretation of (17) requires an additional inferential step in comparison to (18), the utterance in (17) is more indirect than its counterpart in (18). In both cases, however, other contextual clues—in the form of future consequences—contribute to and support the request interpretations. This concludes the analysis of the metonymie functioning of the BEFORE component of some directive speech acts. We hope to have shown that synecdo­ che is an important means of performing indirect speech acts; yet, depending on context, a part may be in a "stand-for" or a "pointing-to" relation to the sce­ nario.

4. A Taxonomy of Metonymies The evidence we have presented so far supports a metonymie analysis of indirect speech acts. For illustrative purposes we have focussed solely on the BEFORE component of directive speech acts. Drawing from these data as well as addi­ tional data from directives, other speech act types, hedged performatives and sentences containing perception and mental activity verbs, we would now like to present a tentative taxonomy of the metonymies which we have observed in our larger data base. We start out with a taxonomy of speech act metonymies based on the action scenario and the speech act scenarios given in Section 2. All of these metony­ mies can be regarded as synecdoches, that is as part-whole relationships. 4.1. Types of speech act metonymies 4.1.1.

The BEFORE component for whole scenario Ability to perform an action for a linguistic action

(19)

"Don't be absurd, darling, / can advance you any amount you ask for," said Caroline. "Don't you know I am a very wealthy woman?" (offer)

214

LINDA THORNBURG AND KLAUS PANTHER

Wish of H that S perform an action for a linguistic action (20)

"Try not to worry so much. Would you like me to have a word with Stuart?" (offer) The COKE/RESULT component for whole scenano9

4.1.2.

Obligation to perform an action for a linguistic action (21)

"Julie, you're wet. You must change, "(directive)

(22)

"Do sit down, both of you. You must certainly play for us, if you will, Mr. er-" (request)

4.1.3.

The AFTER component for whole scenario A future action for a linguistic action™

(23)

"Oh, Rachel, don't you see, I can ask you to marry me now," he said huskily. "You do care for me, dearest? You will say 'yes'?" (request)

(24)

"/ will come with you tomorrow, but more than that I cannot prom­ ise." (promise)

We assume that the speech act metonymies given above can ultimately be situated in a larger system of metonymie principles, some of which are given below and illustrated by sentences from our data base. For example, the speech act metonymy ability to perform an action for a linguistic action, amply illus­ trated in Section 3 and by (19) in Section 4, is a specific subcase of a more gen­ eral metonymie principle potentiality for actuality. The speech act metonymy wish of H that S perform an action for a linguistic action is a special case of the general metonymie principle motivation/reason for action. The speech act me­ tonymy obligation to perform an action for a linguistic action exemplifies the general metonymie principle effect for cause, another specific subcase of which is the result for action metonymy. 4.2. Some general metonymies While the speech act metonymies we have looked at lend themselves to an analysis as synecdoches, with part of the speech act scenario standing for the whole speech act, it is also often plausible to regard them as metonymies in which one cognitive domain or part of a cognitive domain stands for another

SPEECH ACT METONYMIES

215

cognitive domain or part of a cognitive domain. For example, in (25) below, a part of the cognitive domain of deontic or dynamic modality (cf. Palmer 1986), denoted by must, stands for a part of the domain of desires, motivations, and intentions, denoted by want. In contrast, in (26) the speaker remains within the speech act domain of "offering", metonymically linking one part of that do­ main, the ability to perform an action, to another part of the domain, the want to perform the action. 4.2.1.

Necessity for motivation

(25)

4.2.2.

He was passing her with an inclination of his head. "No," she said unsteadily. "/ must speak to you, please!" He followed her into the room near by. [/ must speak to you stands for / want to speak to you]

Potentiality for motivation

Ability for motivation (26)

"... / can lend you whatever money you require until we return [sic] then you can pay me back. ..." [/ can stands for / want]

4.2.3. Potentiality for actuality Potentiality for actuality, certainly one of the most general metonymie princi­ ples, is amply instantiated in three variations in our data base. The first subcase, the metonymy ability for action, which is extensively exploited for conveying indirect directives and commissives (see above), is also used in hedged perfor­ matives, as exemplified in (27) and (28) below, where the abilitiy or legitimiza­ tion to perform the linguistic action stands for the linguistic action per se. The second subcase, exemplified by (29) - (31), concerns the well-known phenome­ non attested in English and other languages that perceptual events are expressed as abilities, such as in / can hear you (for "I hear you") or / can see the Rockies from my window (for "I (actually) see the Rockies from my window").11 In the third subcase, the desire to do something (a mental attitude that motivates an action) stands for the actual future event. Ability for present/future action (27)

/ can testify that, seen from the surrounding heights, it is a fairyland of lights, that many shops are open and the windows of the others lit up. (hedged performative) [/ can testify stands for I testify]

216

LINDA THORNBURG AND KLAUS PANTHER

(28)

/ can recommend this type of cage, as it is impossible for the birds to throw out any seed husks, and I found that they very quickly get used to the idea of a glass front, (hedged performative) [I can rec­ ommend stands for / recommend]

Disposition for perception/feeling/mental activity (29)

I said, "Sure, honey. Sure I love you." I meant it, too. But another man lay on the bed beside us. I could hear his sneering laughter as her arms carried me off through the fire of oblivion. / can hear it yet. [/ can/could hear stands for I hear/heard]

(30)

Cuddy emitted a low soundless whistle. "That's a lot of change." "Yes, but I'll get it." "You'll be lucky." "You bet I am. I'm on the right streak tonight, I can feel it." [/ can feel it stands for I feel it]

(31)

"I can't let her down just like that, yet one day it will have to come. / can see that now." [/ can see (= understand) that now stands for / see (= understand) that now]

Motivation/reason for action (32)

"ƒ want to marry you," he said. "We will live for ever in a little house by the sea. " [/ want to marry you stands for / will marry you]

4.2.4. Effect for cause The general metonymy effect for cause is illustrated below by the more specific subcase result for action, which is all-pervasive in language. In Section 3, in the discussion of example (33), it was pointed out that in the transfer of some object X the receiver's having stands for the agent's giving. Example (34) differs slightly in that there is no transfer of an object; in this case, the understood subjects of the stative verbs stand and have have themselves performed prior actions. Yet in both examples it is the state that metonymically stands for the prior actions leading to that resultant state. Result for action (33)

Can / have hot chocolate with whipped cream? [have is the result of give]

SPEECH ACT METONYMIES

(34)

4.2.5.

217

Please stand behind the yellow line. Have documents ready. (U.S. Customs, Atlanta Airport) [stand and have are the result of prior actions]

Form for content

(35)

"It may be as long as seven or eight months, but you'll be married to me before that, June." "Oh, no!" Even his attraction was power­ less to affect her at that moment. "How can you mention such a thing?" (expressive: reproach) [In asking how the legitimacy of the speech act comes about, the speaker implicitly rejects the legitimacy of the speech act and therefore its content.]

5. Conclusion Speech act theorists assume that the identification of the intended (primary) communicative intention of an indirect speech act requires a certain amount of inferencing on the part of the hearer. In his classical analysis of indirect speech acts, Searle (1975) claims that the intended force of an indirect speech act has to be inferred on the basis of what is literally conveyed, the linguistic and extralinguistic context of the utterance, and Gricean principles (see Grice 1975) of what constitutes rational and cooperative behavior. We share this view, although it should be borne in mind that people usually draw the necessary inferences to arrive at the intended interpretation very quickly and without any noticeable effort, at least in the case of conventional­ ized indirect speech acts. The idea that metonymie principles are at work pro­ vides a natural explanation for this fact. Metonymie relationships such as partwhole, cause-effect, ability-action, reason-action, and so forth, are easily per­ ceived and processed. These general metonymies constitute natural inference schemata; their application enormously facilitates the inferencing needed to identify not only the intended force of indirect speech acts but other kinds of pragmatic implications, as shown in some detail in Section 4. Yet, pragmatics is but one cognitive domain in which such metonymie reasoning is at work. For example, in the domain of visual perception we find synecdoches such as face for person (as in passport photos) or static images for dynamic events (as in countless works of visual art and representation). We have shown that linguistic phenomena traditionally described in terms of pragmatics can be explained insightfully by and integrated into a cognitivist framework. Pragmatic theories (e.g. Bach & Harnish 1979, Leech 1983, Sper­ ber & Wilson 1986) usually assume that the interpretation of utterance or speaker meaning requires inferential capacities on the part of the addressee of the communicative act. However, these theories do not systematically specify

218

LINDA THORNBURG AND KLAUS PANTHER

the kinds of inference patterns needed for utterance interpretation. We have demonstrated that much of pragmatic inferencing is metonymic in nature.

Endnotes 1. As an example of a non-linguistic action consider the act of opening a window. A precondition for this action (a BEFORE condition) is the ability to perform it. Sup­ pose that Mary wants or intends to open the window (a motivation for the action), but is not strong enough or does not have the time to perform the action, then the action simply does not come about. The CORE of the action is, among other things, characterized by certain bodily movements. The immediate RESULT of a successful act of opening a window is that the window is open. An intended consequence of opening a window (the AFTER) might be that fresh air flows through the window. An unintended consequence might be that flies come through the open window. 2. In what follows we will ignore other possible preconditions, e.g. that H (or some­ body else) has not yet done A, which is highlighted in utterances like You haven't handed me that superglue (yet). 3. The obligation is very strong in the case of an order; it is weaker in cases of advis­ ing, recommending or suggesting. 4. Italics in the examples are ours and indicate the syntactic types that we extracted from the corpus. 5. These same (or analogous) preconditions stand at the periphery of, or outside other speech act scenarios as well, e.g. commissives. 6. The complexities of the distribution of please in indirect requests go beyond the scope of this paper. 7. This distinction should not be viewed as binary, but rather as describing the endpoints on a continuum. 8. Our corpus yielded no relevant data of the type I would like (NP) to VP/I would like NP. 9. In terms of speech act theory this metonymy could also be called illocutionary effect for illocutionary act. This is a specific instance of the more general metonymy effect for cause (see below). 10. After a presentation of this paper at Southern Illinois University, Carbondale, IL, it was pointed out to us that in Arabic, in certain contexts, a promise may be per­ formed by presenting a propositional content as a past event and it was suggested that in English utterances like It's done, Consider it done may function in a similar way, that is as commissives. Surprising as the use of a perfective aspect in these forms may be, we claim that they are unexceptional and can be accommodated in our framework. They are instantiations of the after component for the whole sce­ nario metonymy. First, they only occur as responses to requests. Second, their propositional content is not explicitly verbalized but referred to by the pro-forms it and done. Finally, these expressions are based on the metaphor the future is the past. The propositional content is presented as if it had already occurred, i.e., the future (the promised action) is represented as being as certain as a past event. This can be seen by comparing the degree of certainty and commitment of It will be done with that of It's done. 11. In some languages, such as Hungarian, only the non-modal option seems to be available.

SPEECH ACT METONYMIES

219

References Bach, Kent and Robert M. Harnish.1979. Linguistic Communication and Speech Acts. Cambridge, MA: MIT Press. Cole, Peter and Jerry Morgan (eds). 1975. Speech Acts (Syntax and Semantics 3). New York: Academic Press. Fraser, Bruce. 1975. "Hedged performatives. " In Peter Cole and Jerry Morgan (eds) Speech Acts (Syntax and Semantics 3). New York: Academic Press, 187-210. Gibbs, Raymond W., Jr. 1994. The Poetics of Mind: Figurative Thought, Language, and Understanding. Cambridge: Cambridge University Press. Gordon, David and George Lakoff. 1975. "Conversational postulates." In Peter Cole and Jerry Morgan (eds) Speech Acts (Syntax and Semantics 3). New York: Aca­ demic Press, 83-106. Grice, H. Paul. 1975. "Logic and conversation." In Peter Cole and Jerry Morgan (eds) Speech Acts (Syntax and Semantics 3). New York: Academic Press, 41-58. Lakoff, George. 1987. Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Leech, Geoffrey. 1983. Principles of Pragmatics. London: Longman. Levinson, Stephen. 1983. Pragmatics. Cambridge: Cambridge University Press. Palmer, Frank R. 1986. Mood and Modality. Cambridge: Cambridge University Press. Sadock, Jerrold M. 1974. Toward a Linguistic Theory of Speech Acts. New York: Academic Press. Searle, John. 1975. "Indirect speech acts." In Peter Cole and Jerry Morgan (eds) Speech Acts (Syntax and Semantics 3). New York: Academic Press, 59-82. Searle, John. 1976. "A classification of illocutionary acts," Language in Society, 5: 123. Sperber, Dan and Deirdre Wilson. 1986. Relevance: Communication and Cognition. Cambridge, MA: Harvard University Press. Wunderlich, Dieter. 1976. Studien zur Sprechakttheorie. Frankfurt/M.: Suhrkamp.

III. CORRELATES OF DISCOURSE STRUCTURE

Focus Movements and the Internal Images of Spoken Discourse Kenneth Holmqvist and Jana Holsánová Lund University, Sweden

1. Introduction Chafe (1980, 1994) formulated the hypothesis that similar mechanisms are used (a) when we perceive information visually, (b) when we recall it from memory, and (c) when we verbalize it. "All three processes may be guided by a single executive mechanism which determines what is focused on, for how long, and in what sequence" (Chafe 1980: 16). The analyses we present below stem from two projects studying the corre­ lations between units in vision and in discourse. The first of these projects (Holmqvist) attempts to describe focus movements to an extent that will allow a future computational model of discourse understanding, in terms of an evolving mental image. The other project (Holsánová) is an empirical investigation of the correlation between eye movements when watching a narrative film, and focus movements in the spontaneous retellings of such dynamic visual events. In this paper, we will concentrate on the reconstruction of the participants' internal images during spoken discourse. Our first question is: What happens if the speaker draws to illustrate something in the discourse? The focus movements should then leave traces in the drawing that could tell us how the focus of atten­ tion moves over the undrawn internal image of speaker and listener. The first part of this paper discusses the correlation between spoken discourse and focus movements, exemplified by an analysis where a speaker uses an abstract draw­ ing as a complement to a descriptive explanation. The second part of this paper discusses how we can construct a close repre­ sentation of the internal images and the focus movements over it. We use Langacker (1987) and Holmqvist (1993) for an authentic example of building a spa­ tial layout, which can then be filled in with complete lexical images. This research has been financed by the Swedish Council for Research in the Humanities and Social Sciences.

2. Focus of Attention and Intonation Units When we listen to spoken discourse in an authentic communicative situation, we filter out all interruptions and odd repetitions, and perceive it as a continuous stream of ideas. The discourse seems coherent and fluent. Not until we tran-

224

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

scribe and analyse a spoken text, do we realise that it consists of small units and includes small words (discourse markers) that reflect the planning and produc­ tion process of the speaker. Studies made on reports of an event, on spoken de­ scriptions, and on retellings of a film (Chafe 1980; Redeker 1990), show that we formulate in brief spurts, focus on small units, jump between different se­ quences, and add digressions and comments. How can we explain this normally imperceptible fragmentation of spoken language? Chafe (1980) decided to base the explanation on the properties of our attention: We can concentrate only on parts of the vast amount of information we get from our perceptual system, our emotions, and our memory. We choose a specific part at a time, according to our needs, interests, and current goals. Our attentional system can be conceived of as a limited capacity resource which selects limited information for further specialised processing (cf. Tomlin, in press). Therefore, when we retell something from memory, our attention moves sequentially from focus to focus as the retelling progresses. In spoken language, these foci are verbalised in what Chafe (1994) calls intonation units. In our material, intonation units vary in length between 0.3 and 2.5 seconds. There are often pauses between the units, but they are given a rather prosodie characterisation (Chafe 1994: 58): The features that characterise intonation units may involve any or all of the fol­ lowing: changes in fundamental frequency (perceived as pitch), changes in dura­ tion (perceived as the shortening or lengthening of syllables or words), changes in intensity (perceived as loudness), changes in voice quality of various kinds, and sometimes changes of turn. Intonation units may coincide with syntactic units (such as sentences), but more often there are mismatches between syntactic and intonation units in spo­ ken discourse. A sentence is often spread over a number of intonation units. Intonation units group to larger units in spoken discourse. According to Chafe, foci combine to centres of interest or super foci, i.e. cognitive units based on experience, intellect, and judgement, which intermediate between into­ nation units and basic-level topics, which are units at an even higher level (cf. Chafe 1994: 137-138).

3· Analyses of two Discourse Passages Below we present focus movement analyses from two discourse passages. Both have the same main speaker, a young Swedish-Canadian man, whom we call A. During his short visit to Sweden, he is having an evening of small talk with his friends. The other two participants,  and C, leave the initiative to A and usu­ ally speak only in the form of questions or support signals such as humming and laughter.

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

225

Focus in the drawings and in the figures is shown with a white spot-light. Those parts that are currently not focused are shadowed (Chafe uses the term periphery for these attentional outskirts). We show the English translations to the right of the corresponding drawings. The Swedish original is found in the appendix. The numbering indicates the intonation unit number in a larger transcript, from which both examples (312-369 and 1-9) are taken. The symbols used in the tran­ scriptions are explained in the appendix.

4. First Analysis: Focus Movements over an Abstract Drawing In the first extract, 1.13 minutes long, the speaker spontaneously produced an abstract drawing as a part of his explanation of the low road quality in Canada. As he moved through his explanation, A used a combination of physical point­ ing (with the pencil) and linguistic means, to help the listeners identify the cur­ rent focus within the picture. Our analysis will therefore mainly concern the correlation between visual focus movements over the drawing and focus move­ ments in the discourse. After having talked a while about his experiences of poor Canadian road quality, speaker A starts drawing the one-dimensional spectrum in Figure 1. From the preceding context, we can guess that the spectrum content is road quality or road builder quality or something similar.

226

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

In intonation units 313-337, the empty spectrum is filled in from discourse by the superimposition of at least three polar contrast dimensions onto it. The de­ scription of the contrast dimensions is given by focusing on their end points. In 313, focus is placed on the right hand side of the drawing. Intonation units 314 and 315 then superimpose the much money end point and the high quality end point of two contrast dimensions onto the focused pole in the drawing. Speaker  signals understanding in 316, and A then elaborates the relations between the two dimension end points, focusing back and forth between them. In 320,  expects a continuation, signalling understanding even before A has ended. Since contrast dimensions have two poles, a natural expectation is that the unfilled poles of the contrast dimensions (see Figure 2b) are to be elaborated next. Initially this also seems to happen. In 321, speaker A moves to the opposite pole of the drawing. However, he then unexpectedly introduces the negative pole of an ethnic contrast dimension and superimposes it onto the spectrum (322-324).

Figure 3b. Contrast dimensions, filling of left pole To complete the spectral contrast, the two unfilled dimension poles must still be filled in, and A does this by moving first to the money dimension in 325

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

227

and then to the quality dimension in 326. All these movements are internal. In the drawing, focus stays on the left pole. Figure 3b shows the filling state of the contrast dimensions at intonation unit 327. There now remains one dimension pole unfilled, and A immediately undertakes the task. Since it means a move in the drawing, A uses a clear atten­ tion mover we have ... up here, (contrasting against the down here mover of 321, in an implicit fourth contrast dimension, which is verbalised again by at the bottom in 336). 321(A) ...0 62 then we have down here 322(A) we have ..0.27 the fellows who 323(A) ..·1 50 come from italy 324(A) an all those countries 325(A) they spend quickly the money 326(A) an they . . 0 36 dónt care. 327(B) ••0.42 rnhm

Figure 3. Drawing focus in 321-327 328(A) 329(A)

so we have mòre or less Scandinavians and scots up here

330(A)

then we have the Italians and the port[gues] Pin] and other. . 0 23 [tr/] [haha]ha hafhaha] [silly] < LAUGHING > ... 064  other trash at the bottom < LAUGHING > ..0.3ohnhnhn

Figure 4. Drawing focus in 329

331(B) 332(A) 333(B) 334(B) 335(A) 336(A) 337(B) Figure 5. Drawing focus in 330-337

228

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

Continuing the summary announced by the marker so in 328, A uses the mover then to go back down to the negative pole of the ethnic dimension, and dwell a little on it in 330-336, thereby entertaining his partners. The three dimensions (money, quality, ethnicity) that have been superimposed onto the drawn spec­ trum, are by now well understood by all partners. The introductory description of the spectrum is complete. When looking at the drawing on the paper, they now see these contrasts and not the initially empty bar. The partners are thus prepared for the causal part of the explanation, in which the spectrum is to be the main participant. The change from spectrum description to causal description is marked by 338, a regulative intonation unit, including the mover an now, the deictic like this and a very clear voice-quality change from laughter to seriousness. In 339, the stressed deictic attention mover all these makes us focus on the entire spectrum. A continues by describing the process of the road builders giv­ ing offers to districts etc, and even quotes the offers in 342/343. In 345-350, A demonstrates the decision process of the authorities. There are several attention movers here. The an in 345 simply signals progression in time and discourse. The an then in 347 and 348 also signals movement of the (visual) attention of the authorities.

Figure 7. Drawing focus in 348-349

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

229

Figure 8. Drawing focus in 350 and 353 During 340-347, there are no focus changes in the drawing. The focused entities, perhaps especially the district, are judged to be efficiently communi­ cated by language, and would be messy if drawn. Our focus of attention reap­ pears in the drawing when the deictic attention mover like this in 348 coincides with A drawing the line across the spectrum. In 350, down here takes us to the cheap and low quality side of the spec­ trum (the ethnic dimension is now forgotten). The point being made is that authorities sacrifice quality for a low price. In the following, A will describe how this affects the spectrum of road builders in the long run. In 355, then signals causal and temporal progression, and these moves us to the good side of the spectrum. In 357, here moves the focus back to the cutting line. The an in 356 and 357 and the an then in 358 continue to signal progres­ sion. The stressed this takes us to the bad side of the spectrum.

Figure 10. Drawing focus in 357

230

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

Figure IL Drawing focus in 358 In 358-361, A gives the idea of the bad part of the spectrum growing over time, taking up the space from the bankrupt good side. The focused magnifica­ tion of the bad section of the spectrum, is drawn as two lines leading downward (ahead in time) in the drawing. It is interesting to notice that originally, A's three dimensions were polar contrast dimensions, but in order for this magnifi­ cation process to work, the dimensions have to be continuous.

Figure 12. Drawing focus in 358-362

Figure 13. Drawing focus in 363-365 In 363, we have the situation a few years ahead in time, after the develop­ ment above has taken place. The new spectrum in Figure 13 is immediately filled by the dimensional content from the bad part of the old spectrum.

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

231

In 366, when the new cutting line is drawn and probably focused upon, A quickly mentions both sides of the new spectrum (referring to the money di­ mension). This probably does not mean that the focus of attention moves to both poles of the spectrum. Rather, the cutting line metonymically refers to the two poles, which give meaning to the cutting line when mentioned, because it links the new cut back to what the authorities did to the original spectrum.

Figure 14. Probable drawing focus in 366

Figure 15. Drawing focus in 367-369 The long awaited finale in 367/368 takes us (by an and these) to the bad side of the new spectrum, which is marked very clearly in the drawing. We are suddenly placed in the present, and the previous discourse becomes an historical explanation of current Canadian road quality.

5. Focus Movements The material used here was recorded on audio tape for another purpose (a study of spontaneous descriptions of national identity, Holmqvist and Holsánová (1996)). We reconstructed the drawing process in Figures 1 to 15 by erasing

232

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

pieces of the final drawing (Figure 15 except for focus marking), according to what was said in the discourse. In doing so, we have paid attention to what is being focused on at each point in the discourse, and related that to elements in the drawing, as was indicated above. For the speaker, the purpose of focus movements is to guide the listener, to let the listener build her own internal image the way the speaker wants her to. What should count as a focus movement depends on what the focus movements are supposed to move over. Chafe (1994: 63-65) points out that every substan­ tive intonation unit verbalises an idea, and gives examples of types of ideas. The quoted offers in 342/343, the magnification in 358-361, and the contrast dimen­ sions are all ideas, to which, from which, along and between which our focus of attention can be made to move. When A fills up the original spectrum in 313-337, there are a number of focus movements in and between the three dimensions (money, quality and eth­ nicity) that are not marked, neither in the drawing nor by any markers in the discourse. The focus moves between 314 and 315, simply because A mentions one dimension (quality) after the other (money). In this sense, virtually every new substantive intonation unit means a focus movement in some domain. Some of these domains are not easy to integrate in a drawing. The causal relations in 317-319 cannot easily be mapped onto two static dimensions. The district in 340 has too many domains to be able to put it on paper: Which as­ pects of the district should be drawn? Speaker A chooses to draw the contrast dimensions and the development over time, and those two mappings occupy the two dimensions of the paper. Mapping further dimensions onto it would be messy. Most of the image remains internal and consequently most of the focus movements are internal.

6. Linguistic Markers for Transitions between Attention Focuses As we have seen in the analysis, transitions between the foci in discourse were often accompanied by hesitations, pauses, and linguistic markers. Focus move­ ments that are difficult for the listener to anticipate and to predict the direction of are probably the ones that are lexically marked in the discourse. The many then we have down here and similar markers mean: Now we are going to move the focus (regulation), and we are moving it to this place (direction/deixis), and we are going to stay in this neighbourhood for some time (planning/prediction). These markers give the speaker the time needed for careful description of the upcoming centres of interest, such as the poles of the spectrum (313, 321, 328, 330), the process of sending offers (338), and the second spectrum in analogy to the first one (363). The marked focus movements therefore also influence pronominal reference (which Redeker 1993a, b studied more closely). The pronoun referent is pri-

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

233

marily found within the focused parts of the image, thus establishing these cen­ tres of interest makes it simpler for both speaker and listener: There are fewer possible referents, and they can be referred to by short and efficient pronouns. (Grosz and Sidner 1986 implemented a stack of lisp-structures to simulate levels of discourse focus for the very reason of making nominal and pronominal reso­ lution more efficient). When there is no need to establish a new centre of interest, markers such as an and an then were used (347, 348, 355, 357, 358). The function is one of progression, in time or in causality, within the same centre of interest. When moving the focus to a new centre of interest, an (345), an now (338) and then (321, 363) were used. The marker then can also mark cases when the focus is moved back to a place already described, repeating the description, as in 330 (we will see the same in our second analysis, in Section 7.2, unit 6). A clearer marker of refocusing is so, as in 328. So prepares the listener for a general summary, a new centre of interest, in which some things may already have been described and others not. The place to which the focus moves is sometimes given by deixis (here in 357, down here in 350, these in 355, 367, all these in 339, this in 358), but more often simply by mentioning the new focus. In the literature, these linguistic markers appear under many names, de­ pending on the perspectives and goals of the respective researcher: Gülich (1970) speaks of Gliederungssignale, Quasthoff (1979) of Verknüpfungs- und Gliederungssignale, Schiffrin (1987) of discourse markers, Aijmer (1988) of discourse particles, Rudolph (1989) and Weydt (1989) of Partikeln, Stenström (1989) of discourse signals, Redeker (1990) of discourse markers and Redeker (1991) of discourse operators. There is only partial agreement about the classifi­ cation of these signals. They are often divided into different classes and attrib­ uted different functions by different authors. Discourse markers appear as conjunctions, adverbials, interjections, parti­ cles, final tags, or lexical clauses. Examples are the English well, but, cause, so, y'know, I mean, now, anyway, and so, okay, by the way. The transitions are not only marked in monologues, but also in dialogical situations as the following example from Schiffrin (1987: 199) shows: (1) Sally (2) Irene (3) Irene (4) Irene

You said your teachers were old fashioned. Did they ever hit kids, or Yeah. I had one teacher, her name was Frank, we used t' call her Frankenstein. So, yeh, she would hit kids with a ruler.

Sally puts a question to Irene, who starts her answer with a description of a person. This beginning could have developed into an entire story, but Irene in­ terrupts herself, using as a marker, and goes back to the question.

234

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

What functions do such markers fulfill? To us the most interesting definitions of discourse markers are: "a discourse marker is a linguistic expression that is used to signal the relation of the utterance to the immediate context," (Redeker 1990: 372), and "a discourse operator is a word or phrase ... that is uttered with the primary function of bringing to the listener's attention a particular kind of link­ age of the upcoming utterance with the immediate discourse context. " (Redeker 1991: 168). Schiffrin characterises discourse markers as "sequentially dependent elements which bracket units of talk." (Schiffrin 1987: 31). We want to propose a broader definition than the ones above. These mark­ ers should include not only the movements internal to discourse, as in the defi­ nitions above. We also want to include the external focus movements that take place in the speech situation. If, for instance, we make a drawing or point at a painting while speaking, deictic expressions such as here (excluded by Redeker 1991) are necessary to guide our focus movements over the drawing or painting. The deictic here signals that a new focus of attention is to be pointed out in the participant's immediate perceptual space. Stressed deictic referents like here in 350, 358 and 367 often correspond to stress in the drawing, see Figures 8 and 15 above. The similarity between discourse markers and deictic expressions is closer than it may seem: Neither can specify the location of the next focus by itself. They both need either the succeeding intonation unit or the physical pointing with a finger to locate the new focus. There are also combinations of discourse markers with deictic markers which have a regulative function. The phrases an now it's like this (338), an then they do like this (348) serve as a link to the fol­ lowing context and prepare the listener for a more complex explication that will follow. But primarily, both discourse markers and deictic expressions mark an imminent move to a new focus of attention, and therefore we have chosen to call them attention movers.

7. Second Analysis: Focus Movements and the Reconstruction of the Listener's Internal Image We will now discuss in detail how a valence analysis based on the works of Langacker (1987) and Holmqvist (1993), can help us understand how the lis­ tener's internal image is built up during the progression of discourse, even when there are no corresponding external images. The valence analysis is presented as a spatial layout, which has the same topology as the image, but in which we have not filled in any other image content from individual words than the rela­ tive locations of entities and relations. This second analysis makes use of an excerpt from the same discussion as above. The passage is only 11.8 seconds long, and there is no corresponding

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

23 5

drawing. Still, there are obvious focus movements, again across a polar con­ trast. 1(A) 2(A) 3(B) 4(A) 5(B) 6(A) 7(A) 8(A) 9(B)

...0.8 the americans have a small fall. ..0.i5 which is .0.i9 separate fro=m the real fall, ..o'35hm ...Q.33 the big fall is on the Canadian side. ..0.30 mhm. ...0.71 then the americans have a small fall and they make a great hullabaloo about their little fall. â'h ja mej [ja ja!] < A IS VENTING OR QUOTING > [hn hn]

To the listener, this excerpt appears to be a delimited passage, combined from several intonation units, and it may therefore constitute a super focus. We will reconstruct the listener's process of understanding language, using a valence analysis, in which we assume that the semantic pole of words is schematic and can be incorporated into the discourse image. We take valence relations to be all semantic relations between the different words that appear in discourse. Thus, the relation between small and fall is a valence relation, just as much as the relation between the americans and have. In intonation unit (1), the americans have a small fall, the obviously con­ nects to americans. We know this because we know the word order in English (originally Swedish). There is a corresponding connection in the semantic pole, between the schemata of the and of americans. In the terminology of cognitive linguistics, we say that the trajector (TR) part of the [THE] schema is placed in a valence relation with the schema [AMERICANS]. The TR is the most salient part in relational schemata (adjectives, preposi­ tions, cases, verbs, etc.). In verbs, the TR normally appears as the unelaborated agent, such as [HAVE].TR. Other parts of relational schemata are landmarks (LMs). LMs appear as unelaborated objects, patients, reference points etc., such as [HAVE].LM and [SMALL].REFP. The most common types of valence rela­ tions are found between schema parts such as TRs and LMs on the one hand, and an entire schema on the other hand. Such is the case with both the valence relations [AMERICANS]- [HAVE].TR and [SMALL].REFP-[NIAGARA FALL(S)]. 7.1. Intonation unit (1) In the intonation units preceding excerpt (l)-(9), the speaker had described the Canadian Niagara fall (in Swedish: niagarafallet, in singular) in general. The main topic had been how fast the erosion of the cliff moves the fall backwards. Just before the passage (l)-(9) appears, we therefore had a fairly detailed image

236

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

in mind of A's Niagara fall, combined with whatever pictures or films of it that we could remember. We then receive intonation units (l)-(9). Each of these units evokes a more or less drawable image. We place each image, one by one, upon our semi-active image of the Niagara fall(s). The super imposition of a new unit image onto the previous discourse image results in the next discourse image. The first valence relation in intonation unit (1) is between [THE].TR and [AMERICANS]. The essence of this valence relation is that the determined thing and the Americans are the same. Protoformalistically, we can say that in this valence relation, [THE].TR and the [AMERICANS] schemata have been made identical. In Figure 16, we represent this identity by placing [AMERICANS] and [THE].TR in the same entity box.

Figure 16. Entities and relations after intonation unit (1): the americans have a smallfall In Figures 16-20, boxes are entities and lines are relations. A name with a dot shows that it is the first occurrence of this entity or relation. The white spotlight marks the focused entities and relations. Bold schema names in the sans font mark schemata evoked by the last intonation unit. The next schema in intonation unit (1), [HAVE], is an ownership or spatial proximity relation between the [HAVE].TR (owner) and the [HAVE].LM (owned object). On the basis of word order, it is clear that [HAVE].TR is in a valence relation with [THE].TR and [AMERICANS]. Conversely, [HAVE].LM is in a valence relation with [A].TR, [SMALL].TR and [FALL]. With this, we have the left-hand side of Figure 16: The americans, the fall, and the [HAVE] relation that connects the two entities. We have taken the own­ ership relation to mean that the [AMERICANS] entity and the [FALL] entity should be close to one another in the image.

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

237

There is however more to intonation unit (1). The [A] and [SMALL] schemata not only express the indefiniteness and smallness of their respective TRs. They also contrast their TRs, the American fall, to a definite and well-known [A].CONTR, and to a not so small reference point [SMALL].REFP, both of which are identical to the Niagara fall(s). In Figure 16, the contrasted entities have been placed with a relatively longer distance between them. It should be obvious that it is the left-hand side of Figure 16 that is focused on. After hearing (1), we do not direct our attention to the Niagara fall(s) any­ more, but to the small American fall. The reasons for this are: First, the [AMERICANS], [FALL] and [HAVE] schemata all have a rich lexi­ cal content. This makes them the most important schemata of (1), and therefore draws focus to the parts of the image they are being built into. Second, the TRs of [HAVE], [A] and [SMALL] are focused on because they are the most salient parts of their respective schema. Third, the CONTR and REFP of [A] and [SMALL] are not focused on, because they are not salient parts of their sche­ mata. How do we know that [A] and [SMALL] contrast the American fall to the Niagara fall(s)? Simple word order tells us that their TRs are in valence relations with the American fall. Therefore the [A] and [SMALL] schemata are about falls. Therefore the CONTR and REFP parts must also be in valence relations with falls. In other words, CONTR and REFP expect there to be another fall more or less focused in the discourse. What other falls are there in the dis­ course? The Niagara fall(s). The semantic mechanism establishing this connection is what Holmqvist (1993) calls semantic expectations. The hypothesis is that whenever two entities (such as the CONTR and the Niagara falls) have coinciding semantic properties, they are set in a valence relation to one another (unless there are explicit identity differentiators, as in the other fait). 7.2. Intonation units (2)-(9) Intonation unit (2), which is separate from the real fall, starts off from the small American fall ([WHICH].TR), which was in focus after intonation unit (1). Then three relations, [IS SEPARATE FROM], [THE] and [REAL], contrast this small fall with the real fall. The fall in (2) is identified with the Niagara fall(s), because they are both real falls and separate from the American fall (again semantic expectations). The three relations therefore describe the contrast between the two falls, together with the contrasting relations from intonation unit (1).

238

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

Figure 17. Entities and relations during intonation unit (2): which is separate from the real fall Focus has now changed to the contrast relation and to the two fall entities. The Americans are disappearing into the periphery. In addition to showing us again the same focus mechanisms as in (1), intonation unit (2) is also interesting because of the [IS SEPARATE FROM]. Many schemata that correspond to verbs, especially so-called events, contain change over time in some other domain. This change often follows a path, as in the example She went up the stairs, walked along the corridor and entered room no 12. The focus of attention sim­ ply follows this path. [IS SEPARATE FROM] also has a path in it, and our focus follows that path. In the beginning of (2), the small American fall is focused on. Towards the end, focus is on the real Niagara fall(s). In intonation unit (4), the big fall is on the Canadian side, focus is com­ pletely on the Niagara fall(s).-The new [BIG] and [THE CANADIAN SIDE] schemata provide not only spatial information, but also more contrast. The stress on big is of course another focus marker. Not only does the stress say that the difference in size between the Niagara fall(s) ([BIG].TR) and the American fall ([SMALLJ.REFP) is considerable. The salient part [BIG].TR is also given extra focus. Having presented both the [SEPARATE] and the [CANADIAN] schemata, the speaker has conveyed most of his image to the listener. Figure 18 is not only a valence analysis of intonation units (l)-(4). It is the framework for a drawing of the mental image that (l)-(4) evokes: It includes the Niagara fall(s), on the Canadian side. This is the big, real fall. It is separate from the small fall, which

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

239

the Americans have. The Canadian-American border can be drawn straight through Figure 18.

Figure 18. Entities and relations after intonation unit (4): the big fall is on the Canadian side In order to draw this image, we must not only have images for the lexical schemata in Figure 18, but also a mechanism for conjoining or superimposing the separate images onto one another. Finally, we have the focus on the Cana­ dian side, which tells us that now the focus of attention should be directed to that part of the image. However, during intonation unit (4), the speaker notices that he has been lead away from the discourse path to his punch line. The focus is now on the Canadian side, but his punch line will be about the Americans. He therefore has to move the focus back to the American side of the border. In order to notify us listeners of his focus move, he starts intonation unit (6) with the attention mover then: then the americans have a small fall. The rest of intonation unit (6) is a repetition of intonation unit (1), with added stress on small. Repeating intonation unit (1) may seem superfluous, but the speaker must ensure that our focus is moved to the proper parts of the im­ age: The Americans and their little fall. He moves the focus by mentioning the

240

KENNETH HOLMQVIST AND JANA HOLSÁNOVÁ

americans and by stressing small, which contrasts against the big in intonation unit (4). Since this is a big move to make in only one intonation unit, the atten­ tion mover [THEN] is inserted to notify us.

Figure 19. Entities and relations after intonation unit (6): then the Americans have a small fall Once the focus has been moved over to the American side, the speaker can get to his punch line. Intonation unit (7) is and they make a great hullabaloo about their little fall. In his evaluation of the current image, A uses salient schemata like [A], [GREAT] and [HULLABALOO], [THEIR], [LITTLE] and [FALL]. Intonation unit (6) moved the listener's focus of attention, and prepared her for the punch line. Without this move, the punch line could not refer to the Americans with a simple pronoun, and the hullabaloo idea would have got less focus than intended. The most important goal of the punch line is to describe the American atti­ tude to their little fall as exaggerated ([HULLABALOO]), childish and silly ([THEIR LITTLE FALL]). It is unclear what [HULLABALOO] exactly in­ volves. We have taken the Swedish [HALLA] to mean that the Americans are making exaggerated sounds of praise of their fall, and this interpretation is used in Figure 20.

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

241

Intonation unit (8), à'hja mej ¡ja ja!] is very difficult to translate. It may be a quote, in which the speaker tries to give a condensed abridgement of how he thinks that the Americans sound when they make their hullabaloo in praise of their fall. But it may also be the speaker just expressing his own feelings when thinking about the Americans.

Figure 20. Entities and relations after intonation unit (7): and they make a great hullabaloo about their little fall This second analysis shows how we can reconstruct the internal images and the focus movements over them. The first step of the reconstruction process results in predrawings such as Figures 16-20, in which only the relative posi­ tions of entities have been used from the word content. The next step is to evoke the remaining image or image schema structure of singular words and construc­ tions. Each individual lexical image contributes to the discourse image, and is added to it. Such an addition of an image onto another is sketched in Holmqvist (1993) under the general term superimposition. However, before we print out the drawing of the image, we have to choose what domains are to be visible. For instance, the spatial, geographical domain is easy to include. The size and other intensity domains in Figure 20 are not all easy to map onto a 2D surface. The sound domain is even more complicated to map onto paper.

242

KENNETH HOLMQVIST AND JANA HOLSANOVA

The text also includes a perspective. We are together with the speaker on the Canadian side in the image. The Americans are the -characters in the us them contrast. One way of including that perspective in the image would be to impose a spatial perspective: Turn the spatial, geographical domain so as to make the Canadian side appear close and the American side further away.

Figure 21. A drawing produced as part of a listener's explanation of speaker A's allegedly twisted truths. Compare to Figures 18-20 Figure 21 was produced by a person who at a seminar had listened to the excerpt (l)-(9) and heard the above analysis. He protested, not against the analysis, but against what he thought were the twisted truths in (l)-(9). He ar­ gued that the [HAVE] relation probably meant "have visual access to", and that the Americans have the big fall on their side, but have bad visual access to it (only from a bridge to the right of the fall). The Canadians instead have excel­ lent visual access to the American fall, but only a small fall on their own side.

8. Discussion It is sometimes argued that we do not visualise, at least not generally, when we understand language. As long as the answer to this question depends on intro­ spective observation, the matter cannot be objectively settled. There is however ample indirect proof in favour of visualization of language: Speakers who spon­ taneously draw and listeners who can draw what they think a drawing speaker is drawing, even in nonspatial domains. Above all, the traces of the restless wan­ dering of the speaker's attention over a reconstructable internal image appear in all of language, even if the speaker denies conscious access to those images.

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

243

It is interesting in itself that these visual processes can be revealed in linguistic communication. For the computational linguist it might also be worthwhile to consider that the use of an image-oriented semantics for natural language proc­ essing promises to simplify traditional NLP problems considerably, at the ex­ pense of refocusing part of the research towards visual computation.

Appendix I Symbols used in our transcription of speech, adopted from Chafe (1994). skflt fra'n

primary accent (a pitch deviation accompanied by loudness or lengthening) secondary accent (a pitch deviation without loudness or lengthening), here slightly displaced for typographical reasons

. 0..31 ...125 = ,

a brief pause, shorter than 0.5 seconds a pause longer than 0.5 seconds lengthening of the preceding vowel or consonant a terminal contour which is not sentence-final (speaker proceeds) a sentence-final falling pitch exclamational terminal contour loudness, stress non-verbal action; the only use of capital letters in transcripts overlapping speech

! big < SHOWS > [ ]

Capital letters are never used for speech. Border pauses between intonation units are always written in the beginning of the second intonation unit.

Appendix II The Swedish original for analysis 1. 312(A) 313(A) 314(A) 315(A) 316(B) 317(A) 318(A) 319(A) 320(B) 321(A) 322(A) 323(A) 324(A) 325(A) 326(A)

här har vi hèla spéktrat. ... 140 härharvi ...0 79 manga pengar och myvcket gòd kvalité. . .0.35 m h m

•••0.55 do=m gör ett gótt jobb, men dom vet att det kostar lite mer agö'ra ett [bràjòbb]. [mhm] ...0.62 sa dà har vi här nere har vi ..Q.27 nissarna som ..·! 50 kommer frân italien â alla dom dar länderna dom gör av med snabbt pèngarna à dom ..0.36 bry'r sig inte.

244

KENNETH HOLMQVIST AND JANA HOLSANOVA

327(B)

..0.42 m h m

328(A) 329(A) 330(A) 331(B) 332(A) 333(B) 334(B) 335(A) 336(A) 337(B) 338(A) 339(A) 340(A) 341(A) 342(A) 343(A) 344(A) 345(A) 346(A) 347(A) 348(A) 349(B) 350(A) 351(B) 352(A) 353(A) 3 54(A) 355(A) 356(A) 357(A) 358(A) 359(A) 360(A) 361 (A) 362(B) 363(A) 364(A) 365(A) 366(A) 367(A) 368(A) 369(B)

sa vi har mèr eller mindre skandinaver och skottar här uppe sa har vi italienare och portugis[er] [hn] och annat.. 0 . 23 [si/] [haha]ha ha[haha] [tráms] < SKRATTANDE > ...Q.64 a annat slö 'dder i bòtten < SKRATTANDE > ..0.3ohnhnhn ...0.52 â nu e de sa att < ALLVARLIGARE > . .0.42 alla dom här kommer in med búd till kommuner, â â provinser â sâdana. nú nu kan vi bygga vä'g och allt sànt här. ..o.nhnhn . .0.35 â kommunen tittar pà . .0.14 pâ alla buden, â sa tittar dom i sin kássa, â sa gör dom sa hä 'r. ..o.i2Jaha . . 08 sa handlar dom hä 'r nere. ..·! οι mhm. ...0.62 à naturligtvis, när . .0.4i dom hä 'r har fâtt arbete tillräckligt manga gá ' nger, sá ... gàr dom här i konkúrs. ...0.9 och försvinner. ..-0.50 â sàkapas de áv här. ... 109 â sa sprids den hä ' ut. ...Q.97 med sàmma kvalité, de e bara de att de e ett vidare spèktru[m], [m]hm ...i 34 sa har vi dà en ny ' grupp här som nä 'sta generation vägbyggare. · 1.38 dy 'rast à bílligast. ...i.76âdeedomha'r som nu bygger vägarna i kánada. ... 261 mhm

The Swedish original for analysis 2. 1 (A) 2(A)

.. . 0 g amerikánerna har ett litet fall. . . 0 1 5 som e . .0.19 skìlt frâ=n det riktiga fallet,

3(B)

..0.35 h m

4(A)

...0. det stóra fallet e pà den kandensiska sidan.

FOCUS MOVEMENTS AND THE INTERNAL IMAGES OF SPOKEN DISCOURSE

5(B)

..0.30 m h m .

6(A) 7(A) 8(A) 9(B)

...ότι sen har amerikanerna ett litet fall â dom har ett himla hallâ om sitt lilla fall. â'hjamej [jaja!] [hn hn]

245

References Aijmer, Karin. 1988. "Now may we have a word on this.' The Use of 'now' as a dis­ course particle." In M. Kytö, O. Ihalainen, and M. Rissanen (eds), 15-33.Corpus Linguistics, Hard and Soft. Proceedings of the Eighth International Conference on English Language Research on Computerized Corpora. Chafe, Wallace L. 1980. "The Deployment of Consciousness in the Production of a Narrative." In W. L. Chafe (ed.) The Pear Stories: Cognitive, Cultural, and Linguistic Aspects of Narrative Production. Norwood, NJ: Ablex. Chafe, Wallace L. 1994. Discourse, Consciousness, and Time. Chicago: Chicago Uni­ versity Press. Grosz, Barbara and Candace Sidner. 1986. "Attention, Intention and the Structure of Discourse." In Computational Linguistics, 12. 3 1996, 175-204. Gülich, Elisabeth. 1970. Makrosyntax der Gliederungsignale im gesprochenen Fran­ zösisch. München: Wilhelm Fink. Holmqvist, Kenneth. 1993. Implementing Cognitive Semantics. (Lund University Cognitive Studies 17). Lund. Holmqvist, Kenneth and Jana Holsánová. 1996. "Jag längtar efter svensk kvalitet och svenskt tänkande". (Lund University Cognitive Studies 49). Lund. Langacker, Ronald. 1987. Foundations of Cognitive Grammar. Volume 1. Stanford: Stanford University Press. Quasthoff, Uta. 1979. "Verzögerungsphänomene, Verknüpfungs- und Gliederungssig­ nale in Alltagsargumentationen und Alltagserzählungen." In H. Weydt (ed.). Die Partikel der deutschen Sprache. Berlin/New York: Mouton de Gruyter, 39-57. Redeker, Gisela. 1990. "Ideational and pragmatic markers of discourse structure." Journal of Pragmatics 14: 367-381. Redeker, Gisela. 1991. "Linguistic markers of discourse structure." Review article. Linguistics 29: 1139-1172. Redeker, Gisela. 1993a. "Discourse Markers as Attentional Cues to Discourse Struc­ ture." Poster presented at NATO Advanced Research Workshop "Burning issues in Discourse", Maratea, Italy, April 1993. Redeker, Gisela. 1993b. "The Attentional Structure of Spoken Discourse." Presenta­ tion at the 3rd International Cognitive Linguistics Conference in Leuven, July 1993. Rudolph, Elisabeth. 1989. "Partikeln in der Textorganisation." In H. Weydt (ed.), Sprechen mit Partikeln. Berlin/New York: Mouton de Gruyter, 489-498. Schiffrin, Deborah. 1987. Discourse markers. Cambridge: Cambridge University Press.

246

KENNETH HOLMQVIST AND JANA HOLSANOVA

Stenström, Anna-Brita. 1989. "Discourse Signals: Towards a Model of Analyzis." In H. Weydt (ed.) Sprechen mit Partikeln. Berlin/New York: Mouton de Gruyter, 561-574. Tomlin, Russell, in press. "Mapping Conceptual Representations into Linguistic Rep­ resentations: The Role of Attention in Grammar." In J. Nuyts and E. Pederson, With Language in Mind. Cambridge: Cambridge University Press. Weydt, Harald (ed.). 1989. Sprechen mit Partikeln. Berlin/New York: Mouton de Gruyter.

Pauses, Cognitive Rhythms and Discourse Structure: An Empirical Study of Discourse Production Joost Schilperoord & Ted Sanders Utrecht University, The Netherlands

1. Introduction One of the main concerns of Cognitive Linguistics is to develop integrated theories of language structure and language processes. In this paper we focus on one particular mode of language use: written discourse production} The goal of this paper is to examine the cognitive processes that take place during discourse production, and the way these processes are related to the structure of the dis­ course. Such a goal requires us to employ research methods to describe both the process and the product of discourse production. To this end, we combine two methods: text analysis (see note 1) and pause time analysis. By using textanalytical methods, structural descriptions can be assigned to discourse, provid­ ing insights into the cognitive representations that discourse producers build or maintain during production. Production pauses, on the other hand, are observ­ able and measurable "cues" to the processes in production. By describing and analysing both the location and duration of pauses relative to discourse structural characteristics, insights can be gleaned into the nature of these processes and into the way in which discourse structure is involved in production. We believe the text-analytical approach to be highly relevant for the "Cognitive Linguistic enterprise", not only because it shows how results from neighboring disciplines (i.e. psycholinguistics) complement linguistic analyses (cf. Gibbs 1996), but also because it has the promise to overcome the traditional emphasis on processing at the sentence level, thereby taking the challenge of "Cognitive Linguistics on the discourse level" seriously. The research reported in this paper draws on the idea that the variable "discourse level" can be operationalized in terms of descriptions of the hierar­ chical structure of discourse. The validity of this idea, however, hinges crucially on the account provided of the hierarchical structure of discourse. If such an account were based on intuitive judgements of discourse structure, the results of pause analysis would not have much to say about underlying cognitive structure and processes. Therefore, we use a procedural method of text analysis, called PISA (Procedures for Incremental Structure Analysis) which seeks to circum­ vent problems related to intuitive accounts of discourse structure. Although PISA is not an algorithm that can be run automatically, it relies on wellarticulated sets of production rules that take linguistic features of the discourse

248

JOOST SCHILPEROORD AND TED SANDERS

as input and delivers rule-based descriptions of the hierarchical structure as out­ put (Sanders & Van Wijk 1996a). The organization of this paper is as follows: Section 2 presents a concise review of the few existing studies of on-line production of discourse. It turns out that in these studies the linguistic characteristics of the structure of discourse are almost absent. And for as far as descriptions of discourse structure are taken into account, they seem to be based merely on intuitive judgements. Section 3 discusses PISA as a procedural method of text-analysis which has the promise to overcome the problems noted in section 2. In section 4, PISA is applied to a specific corpus which contains routine business letters produced by employees of Dutch law firms. In addition, this section presents some of the first results of connecting pause patterns that were registered during the production of these letters to the PISA-derived descriptions of hierarchical discourse structures. Section 5 concludes the paper. 2. Cognitive Rhythms in Monologues Studies of pause time distribution in language production start from the assump­ tion that: (...) hesitations in spontaneous speech occur at points where decisions and choices are being made. On this basis, the patterning of hesitations should pro­ vide clues as to the size and nature of the encoding units which are operative (Boomer 1965: 148). This section briefly discusses some pausological studies on cognitive rhythms in the production of oral monologues. These studies illustrate that cognitively meaningful relations exist between the structure of monologues on the one hand, and the overall distribution of pause time on the other.2 Hender­ son, Skarbek & Goldman-Eisler (1966) selected passages from transcripts of recorded monologues, and plotted the successive pause and speech durations sequentially with speech time. One such plot is shown in Figure 1 (taken from Henderson et al. 1966: 209). In Figure 1, successive speech and pause periods are represented along the X-axis and Y-axis respectively. The sequential temporal pattern reveals alternat­ ing periods of different speech-to-pause ratios, indicating that pause time is not equally distributed over the total speech time. The pattern deviates from a hypo­ thetical straight line fitting the overall temporal sequence: periods with a rela­ tively high pause-to-speech ratio alternate with periods with a relatively low pause-to-speech ratio. Henderson et al. (1966: 210) propose a cognitive account for these cyclic patterns:

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

249

If planning occurs in spontaneous speech and this cognitive activity goes on in the hesitations [i.e. pauses, Schilperoord & Sanders] within the flow of sponta­ neous speech, then an alternating pattern of the kind observed might well repre­ sent the cognitive framework underlying the final speech output.

Figure 1. Temporal cycles in monologue production (0-20 Pause time in sec­ onds; 5-40 Speech time in seconds) If proven adequate, this account implies that production rate parameters, such as pause time distribution, are indeed responsive to the underlying cogni­ tive processes of retrieving information from memory and converting it into coherent speech. In this view, each combination of a shallow and a steep graph corresponds with a plan-execute sequence in speech production (cf. Schilperoord 1996: 4ff). For this reason, the pattern depicted in Figure 1 is often referred to as the cognitive rhythm of speech. Henderson et al. present evidence supporting this cognitive interpretation of time series data. For example, they show the cyclic patterns to be absent in production circumstances in which there is no need for planning, such as reading out loud. In addition, they demonstrate hesi­ tation phenomena, such as false starts, repetitions and filled pauses, and linguis­ tic processing phenomena, such as grammatical errors, to occur more frequently during predominantly hesitant periods then during fluent periods (Henderson et al. 1966; Goldman-Eisler 1967). Finally, the authors argue for the non-random nature of cyclic patterns by showing that the amount of pausing in a hesitant phase is a linear function of the amount of speech in a fluent phase. This mathematical dependency suggests a psychological relationship between hesitant and fluent phases. During a fluent phase, a long range plan is executed that has been formed during a hesitant phase. In addition to such temporal explanations, attempts have been made to ac­ count for speech cycles by connecting the temporal cycles with the semantic and structural characteristics of the associated parts of the discourse. Butterworth

250

JOOST SCHILPEROORD AND TED SANDERS

(1975) tried to find out whether cognitive rhythms in speech were somehow systematically related to the semantic structure of monologues (see also Hender­ son 1974).3 He argues that if the cognitive rhythms in speech were indeed sys­ tematically related to the semantic structure of monologues, then a substantial argument could be provided supporting the claim that psycholinguistic processes underlying speech are represented in temporal cycles. He therefore had subjects speak on social and political topics and recorded their monologues. By using Henderson et al.'s plotting procedure, he was able to produce the same alternat­ ing patterns of steep and shallow slopes which are shown in Figure 1. In addi­ tion, however, Butterworth asked independent jurors to assess the structure of the monologues by segmenting these into so-called "idea units". These idea units were assumed to represent informal intuitions about the semantic units structuring the monologues. An idea boundary was (somewhat arbitrarily) put at locations that were picked out by at least half of the jurors. The main result of this study was that these locations coincided significantly with the temporal cy­ cles identified earlier. That is, junctures between idea-units coincided with hesi­ tant periods during production, suggesting that successive idea units are planned during these periods, and are worded during the subsequent fluent periods (Butterworth 1975: 81, 83). This would indeed suggest a non-random relation­ ship between the location of long pauses and the semantic structure of the dis­ course. However promising these results may be, Butterworth's research design has its flaws.4 A crucial problem concerns his procedure for identifying the structure of the recorded discourses. Not only were jurors in the experiment obliged to respond to the rather impressionistic notion of "idea units" considered to be the basic structuring elements, they were also asked to decide for themselves what such a unit might be. As Butterworth points out himself, the large intra-subject variation that was found in dividing the discourses according to a juror's intui­ tive understanding of what an idea unit is, suggests large differences in the kinds of tasks they set for themselves. This leaves Butterworth's (1975: 85) results to be of a rather undecided nature. Acknowledging this flaw, he remarks:"It is likely that some way can be found of organizing (...) units into a hierarchy and that the intuitions underlying it can be captured by some semantic formalism." In the next section we introduce a text-analytical method that may approxi­ mate such a "formalism". 3. PISA and the Hierarchical Structure of Discourse So far, we have seen that research of cyclic pausing patterns in discourse pro­ duction has not paid sufficient attention to the linguistic characteristics of the structure of discourse. Butterworth's study is a positive exception to this rule, but it was shown that his way of incorporating linguistic discourse characteris-

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

251

tics is problematic. It follows, that if we want to account for pausing patterns in discourse production, the notion of discourse structure should be treated in a more principled and procedural manner. Our emphasis on discourse structure follows from the general conviction in both analytical and cognitive studies of discourse that structure and coherence are constituting principles of discourse. Without these a discourse would be but a random set of utterances (see, among many others, Van Dijk 1977; Hobbs 1990; Mann & Thompson 1988; Redeker 1990; Sanders, Spooren & Noordman 1993). Therefore, of all possible linguis­ tic characteristics of discourse, a valid description of its structure is most likely to reflect aspects of the underlying mental representation of discourse which the producer uses as a production plan (Sanders & Van Wijk 1996 a, b). Thus, the structural descriptions assigned to discourse are regarded as prints of underlying cognitive activities (cf. Van der Pool 1995: 103). Discourse structure has both hierarchical and relational aspects (Sanders 1992). The relational aspect concerns the meaning of connections. Segments can be connected as, e.g., Evidence, Result, Concession, or Specification. The hier­ archical aspect reflects the dominance of one segment over another (the connec­ tion is subordinative, coordinative, or superordinative), the distance between connected segments (adjacent or non-adjacent segments) and, in many cases, their informational relevance (from high/central to low/peripheral). Although both aspects of discourse structure are obviously related (relations that differ in meaning often differ in hierarchical properties as well), most accounts of coher­ ence and structure concentrate on one of them. Outstanding examples of analyti­ cal models that consider both aspects of structure are the Rhetorical Structure Theory (abbreviated RST, Mann & Thompson 1988) and the Linguistic Dis­ course Model (abbreviated LDM, Polanyi 1988). 3.1. PISA : A procedural account of hierarchical structure In this paper we focus on the hierarchical aspects of discourse structure. The notion of hierarchy accounts for the intuition that the ordering of information in a discourse usually varies from important to unimportant - with the important information situated highest. It also accounts for the intuition that language users know what a discourse is about at a certain moment. The discourse can handle a certain topic for some time, then digress somewhat, to return to the initial topic. These digressions from and returns to the main line in the discourse suggest a hierarchical structure in discourse (see Sanders 1992: 5). During the past few years an analytic procedure has been developed for the analysis of hierarchical structure, called PISA {Procedure for Incremental Structure Analysis, see espe­ cially Sanders & Van Wijk 1996a, and Van der Pool 1995). PISA combines intuitions about structural regularities of particular types of discourse with in­ sights from the text-analytical,5 cognitive linguistic and functional linguistic lit­ erature.

252

JOOST SCHELPEROORD AND TED SANDERS

The original PISA-design applies to expository texts, varying from simple de­ scriptions to technical explanations. In the following we will highlight two cru­ cial features of PISA. First, PISA aims at psychological plausibility, and so its mode of operation should reflect important features of cognitive processes, be it in reading or in producing discourse. To this end, two such features have been incorporated (see also note 4): (i) People do not process language without any guidance: discourse pro­ duction is guided by plans, and discourse understanding by expecta­ tions. This has been implemented in the form of discourse schemes ex­ pressing a prior knowledge of discourse structure (cf. Bereiter & Scardamalia 1987). In the case of explanatory discourse, the discourse scheme determining the global organization of this genre are: - an Action-line: how is the topic used, how does it work? and - a Property-line: what are the features, parts, properties of the topic? (ii) People tend to process language in an incremental way, that is with little preplanning (Levelt 1989). This has been implemented by having discourse segments analysed according to their linear surface order without a preceding global inspection of the discourse. Second, PISA aims at maximizing the inter-analystrellability and is there­ fore designed as an analytic procedure. The actions a PISA-analyst should un­ dertake when analyzing a text are exactly specified so as to avoid heavy reliance on world knowledge (which would, at all times, lead to a decrease in interanalystrel lability). These specifications are formatted as IF-THEN decision rules, making explicit the analytical decisions the analyst should take for a wide variety of linguistic cues contributing to discourse structure. For instance, when a connective such as or is found, the THEN-part of a decision-rule stipulates that the segment is to be coordinated to the preceding segment (for details, see Sanders & Van Wijk 1996a, Van der Pool 1995). 3.2. From clauses to tree structure PISA accepts discourse segments as its input. Segments are identified on the basis of syntactic form. Basically, each clause is considered a segment (See Sanders & Van Wijk 1996a for segmentation rules; cf. also Mann & Thompson 1988, Schilperoord & Verhagen 1996). The analysis itself starts with the initia­ tion of some relevant discourse scheme as discussed in the previous section. For instance, in the case of an explanation of a telephone, the general topic descrip­ tor is set to "Telephone" and two main lines are initiated: the Action-line (a

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

253

sequence of actions or events) and the Property-Ymt (a list of static descrip­ tions). These are the main discourse lines to which the incoming segments are to be attached. The exact nature of these lines is, of course, genre-specific. Once the knowledge of the genre and the topic is initialized, the actual analysis proceeds incrementally, that is the analyst considers the first segment, and attaches it to one of the main lines or to the topic, without a prior inspection of the entire discourse. Then he moves to the following segment, attaches it to the evolving structure, and so on for each following segment until the analysis is finished. Each time an incoming segment is integrated within the existing tree structure, the crucial question is: is it subordinated or is it coordinated to an existing segment or main line? 3.3. The cognitive interpretation of PISA-structures It is an attractive hypothesis to assume that the hierarchical tree structures ar­ rived at by means of PISA resemble relevant features of a writer's cognitive representation of a discourse. In fact, recent research suggests that structural analyses of text corpora enable us to identify different ways writers use to ar­ range information in an explanatory text. For example, Sanders & Van Wijk (1994) had a group of (young) writers produce explanatory texts on topics such as "Telephones". These writers turned out to organize the information either according to a temporal order of actions or events ("first you do this, then you do that..." and so on), or by focusing on the topic to be described and mention­ ing all kinds of topical properties ("It has an X, and it has a Y ...," and so on). Such strategies for organizing information can be deduced from the systematic differences that were exposed by a structural analysis of the explanations. Moreover, these structures seem to reflect the way in which the writer has or­ ganized the information during production. Temporally ordered explanations (the ones dominated by action-Ymts), for instance, can be produced by running through episodic memory; the actions or events are mentioned in a temporal succession and end with a closing of an episode. Explanations dominated by property-Ymes, on the other hand, are probably produced by searching semantic memory in an associative way. They list information that is related to the topic, and which is potentially relevant to explain. In Sanders and Van Wijk's study the structure-analytic approach to writing has been substantiated, pursuing the idea that structure analysis yields insight into the knowledge structures that direct the composing process and the deci­ sions underlying the selection and organization of the information in the text (see also Bereiter & Scardamalia 1983). It is only logical to proceed from this "cognitive text-analysis" to the next issue: how is this knowledge used and how are these decisions made during the actual course of composing? Descriptions of discourse structures enable us to deduce hypotheses regarding on-line discourse

254

JOOST SCHELPEROORD AND TED SANDERS

production, which, we believe, can be empirically investigated by focusing on pause-patterns in production.

4. Structure and Processing Characteristics in Discourse Production This section presents and discusses some preliminary results of connecting the overall distribution of pause time and the PISA-structures obtained for one par­ ticular type of written discourse. We used a corpus of short business letters that were produced by lawyers working in Dutch law firms (for a more elaborate description of this study, see Schilperoord 1996, especially chapter 4). The main cognitive process that can be found within this type of routine discourse production, is the process of retrieving information from memory (Bereiter & Scardamalia 1987: 185; Schilperoord 1996). Granted the assumption that pauses reflect cognitive activities, the main underlying process reflected by pauses must be the process of retrieving, that is, activating information in order to express that information verbally and connect it coherently to the growing discourse. Differences in pause times should then reflect differences in the effort needed to retrieve the appropriate information. Now, what kind of relations can we expect to find between the hierarchical structure of these letters and the dis­ tribution of pause time? Global effects on pause times can be predicted from different hierarchical positions of segments. In particular, we would expect pause times to increase at points where a major structural juncture can be found in the discourse. In the appropriate PISA-terms, pause times can be expected to increase at switches between main lines because at these points the writer's at­ tention must shift from one structural frame to another one. On the other hand, once such a frame is activated, we may expect information retrieval to be less costly, which will manifest itself by shorter pause durations. Hypotheses like these can be examined by regarding the variable "pause length" as the dependent variable, and the variable "pause location" as the ex­ planatory variable. Obviously, our approach implies that the latter variable is to be defined in terms of the hierarchical structure of the letters. To show how this was done, section 4.1. describes how the PISA-procedure is applied to the analysis of routine judicial letters, and what the main structural characteristics of these letters are. Section 4.2. deals with the issue of defining pause locations in terms of hierarchical structures and exemplifies how pause patterns relate to these structures. Finally, section 4.3. describes a quasi-experiment7 that brings together these matters by analysing differences in pause lengths in relation to structurally defined pause locations.

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

25 5

4.1. The hierarchical structure ofjudicial letters In section 3, it was pointed out that the hierarchical structure of discourse is determined by three factors: (1) The topic of the discourse; (2) The pre-analytical initialization of the global organization, or discourse scheme of the discourse type at hand. This global characterization is derived from observed regularities in the letters; (3) The assumption that each discourse segment may either dominate other segments, or stand at the same level with other segments. In the first case, one segment is superordinate or subordinate with respect to another segment. In the second case, two (or more) segments are coor­ dinated. The most central element is the topic of judicial letters, which is usually specified by the producer as "Case X," or "In the matter of X versus Y." With concern to the pre-analytical initialization, the global organization of judicial letters is determined by three principles (see Fig. 2), which will be briefly illus­ trated here.

Figure 2. The hierarchical structure ofjudicial letters Superstructures. The letters are organized according to a global superstruc­ ture defining the functional syntax of the genre (Van Dijk & Kintsch 1983). In routine letters, superstructural categories are conventional parts of letters such as "greeting," "orientation," "central part," "additional remarks" and "closing" (abbreviated G, O, CP, AR and  respectively). Main lines. The substance of information in judicial letters can be found in the central part. This part is organized according to three "main lines." First, lawyers often judge particular states of affairs, an action undertaken by the "other party" or some intended actions on the part of their clients. Segments

256

JOOST SCHILPEROORD AND TED SANDERS

expressing such judgements constitute the first type of main line in judicial let­ ters - the Judgement-line. Second, lawyers often make appeals to their oppo­ nents. They urge the other party to perform some action, or they just propose it. Segments expressing appeals constitute the second type of main line in judicial letters - the Appeal-line. Inspection of the corpus of letters reveals that the two main lines distinguished here are recurrent elements in judicial letters. They can both appear in the same letter, but at least one of them is always present. If both are present, the Appeal-line has primacy over the Judgement-line, for appeals are usually based on judgements. This is reflected in Figure 3, where the Ap­ peal-line is located above the Judgement-line. In consequence, in aligning the segments in central parts, appeals appear after judgements. Because of the central nature of these main structural lines, the remaining information in judicial letters can only be analyzed in close connection with judgements and appeals. This remaining information for the most part serves to justify the speech acts. Inspection of the corpus reveals this to be a stable struc­ tural configuration. With almost no exceptions, judgements and appeals are justified by presenting arguments, by giving references to judicial rules or by listing relevant facts and the like. Because of this stable character the remaining information (that is, segments not expressing judgements and/or appeals) is or­ ganized according to so-called Justification-lines that are attached to segments expressing speech acts. Main lines in judicial letters are speech-act centered rather than topiccentered (cf. Pander Maat 1994). Judging, appealing and justifying are all typi­ cal speech acts. These speech acts differ from each other with respect to their perlocutionary strength. Whereas addressees can simply take judgements and justifications for granted, an appeal more or less forces them to respond. In fact, even a non-response to appeals will be interpreted by the addresser as a particu­ lar response. Local elaborations. The information in central parts displays several local ways of structuring. Often, segments expand or elaborate segments expressing judgements, appeals or justifications. Such segments make up the local structure of the letter's central part, and they behave according to the usual rules for sub­ ordination and coordination (the third factor determining the hierarchical make­ up of letters). However, local level extensions need not be restricted to segments in central parts of letters. Segments attached to the Superstructural line can also be expanded or elaborated. 4.2. Pause patterns and hierarchical structure In order to give an idea of how pause time is distributed across the hierarchical structure of judicial letters, this section presents all particular pauses occurring between the individual segments within one dictated letter (for the full text of the letter, see the Appendix). In Figure 3 its hierarchical structure is repre-

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

257

sented, together with pause data. For the moment, it is assumed that the super­ structure line, the main lines and the local elaborations indicate three significant hierarchical levels of organization. The three hierarchical levels of interest are indicated with roman numerals I, II, and III. The sampled pause data are indi­ cated below the dotted line. Measures are given for each pair of adjacent seg­ ments. In Figure 3 the segments (both numbered segments and a, b, ... segments) are presented according to their linear order of production and alignment in the letter. Therefore, each pair of adjacent segments defines one particular segmenttransition in the production process. A transition between segments thus exists of the step from one segment to the next in line. Furthermore, it is possible to define each transition in terms of the hierarchical level involved in such a tran­ sition, simply by detecting the highest hierarchical level involved. Hence, tran­ sition 1 --> 2 is a linear transition which involves the level of "superstructure," whereas transition 6 — > 6a only involves a local structural level. Transition 8a — > 9 involves what is marked as a "switch" between main lines. Having fin­ ished a Justification-line, the producer of the letter proceeds by developing an Appeal-line. So it turns out that each linear transition can be represented in terms of the underlying hierarchical structure. Accordingly, each pause location can be described in terms of transition types, so we have pauses between "local segments," pauses between a main line switch, and so on. The relevant observation in Figure 3, is that pause times are not equally distributed across segments. Far from it, in fact. The overall pattern shows sev­ eral divergences from the average pause length. Figure 3 shows that the five longest pauses all occur between transitions of segments involving the upper hierarchical levels. One pause occurs at a transition between segments involving the Superstructural line (segments 1 and 2). Figure 3 also shows that the five shortest pauses all occur at transitions involving a local structural level. All other pauses occur at transitions between segments that involve main lines and switches between main lines. All in all, it seems that different transition types are associated with different pause times, and hence, with different degrees of required "cognitive effort" to arrive at the next segment. 4.3. An experiment Research question. There seems to be a systematic relation between pause times for transitions and the hierarchical level involved in transitions: High-level transitions take up more pause time than low-level transitions. The stability of this pattern was examined by analyzing pause patterns in a substantial number of letter-production processes. The main research question thus is: How is pause time distributed over different hierarchical positions?

258

JOOST SCHILPEROORD AND TED SANDERS

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

259

Materials. A sample of twelve letters was drawn from a corpus of about 200 letters.8 These letters were all dictated by professional lawyers, and served real life purposes such as informing clients about legal procedures to be undertaken, or communicating with colleague-lawyers about cases. Dictation is a very com­ mon way for (Dutch) lawyers to produce their written messages. Letters are dictated on a dictation machine and subsequently transcribed by a secretary. We took special care to ensure that all twelve dictations were indeed full-fledged letters, rather then a set of instructions addressed to the secretary to produce a letter. The sample letter in the Appendix is a characteristic example, and it is presented there precisely as it was dictated by one lawyer. Preparation. All records of dictation were transcribed verbatim, with pause occurrences indicated. A pause was defined as a silence in the speech stream, which was equal to, or longer then .3 seconds (see Dechert & Raupauch 1980). This is the generally accepted cut-off value of pauses reflecting cognitive activi­ ties (rather than breathing pauses, or brief interruptions resulting from the mus­ cular functioning of the vocal tract). The transcribed dictations were analysed according to the PISA procedure as outlined in the former sections.10 The goal of the experiment is to explore the distribution of pause time in terms of hierarchical structure locations. This implies that the nature of the transitions between each pair of adjacent segments has to be established. On account of the properties of the hierarchical structure of judicial letters, eight mutually exclusive types of transitions can be distinguished. The types are listed below: 1. 2. 3. 4. 5. 6.

The Superstructural line; The Judgement-line; The Appeal-line; A Justification-line attached to judgement segments; A Justification-lines attached to appeal segments; Switches between two main lines: judgement-appeal, judgement-justi­ fication, appeal-justification; 7. At least one segment expressing local elaborations attached to segments either belonging to the Superstructural line, the Judgement-line, the Ap­ peal-line, or the Justifications-lines; 8. Two local segments.

Each transition was subsequently scored in terms of these types. The ana­ lytical question amounts to an estimation of mean lengths for pauses occurring at different types of transitions. Hypothesis. Obviously, it would be rather pointless to test all eight transi­ tion types for their possible effects on pause time. For example, there is no rea­ son to assume that transitions at the level of Judgement-lines will take up more processing time than transitions at Appeal-lines or at Justification-lines. The

260

JOOST SCHILPEROORD AND TED SANDERS

same holds for transition types 7 and 8. On the other hand, significant differ­ ences in mean pause times for certain transitional types can be expected. First of all, we can expect a general effect of "global" versus "local" transition types, that is, between types 1 - 6 on the one hand, and types 7 - 8 on the other. Let X be the mean pause length of a particular location type, the this first approxima­ tion of the research hypothesis is (1): (1) Now, consider the Superstructural line. Arguably, this line represents the most conventionalized knowledge people have of formal letters (at least in Dutch). Moreover, apart from the central part of letters, segments attached to the Superstructure line are usually of a highly conventionalized nature. It is therefore reasonable to expect that pause times for superstructural transitions will be shorter than transitions involving other main lines. Since we do not as­ sume pause times to differ with respect to other types of main lines, the hy­ pothesis becomes (2). (2) It can furthermore be argued that transitions involving two main lines will be particularly more time-consuming than transitions involving the same main line. At these transitions the producer must switch between two globally deter­ mined kinds of segments - for example, from segments expressing judgements to segments expressing appeals, or from a Judgement-line to a Justification-lint. However, since the latter switch still involves the Judgement-line, it was decided to define switches of main lines as switches between Judgement-lines and Ap­ peal-lines. Notice that expectations with respect to pause times for this kind of transition do not necessarily contradict the fact that these lines express general knowledge of judicial letters on the part of the writer. Such knowledge may ex­ ist independently from actual discourse or actual usage, but actually switching between main lines means switching from one domain of content to another. On the other hand, retrieving information to be attached to the same main line does not involve such shifts. Therefore, the research hypothesis can be further speci­ fied as (3). (3) Finally, the previous considerations lead us to expect within-segment pauses to be shortest on average, since these pauses occur at levels beneath that of local elaborations. So, the ultimate research hypothesis that was tested is (4). (4)

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

261

Data model. In order to test the hypothesis derived in the former section, a sta­ tistical analysis was conducted aimed at estimating the mean pause times for the five selected transition types, and at testing the differences in pause length. A cell means model was therefore constructed with dummy variables assigned to each transition type (for details, see Schilperoord 1996). Let Y{ be the i-th pause in the corpus, then the model for each pause can be written as (5).

In (5) the dummy variables are SLM (= switch between main lines), ML (= transition on a main line), SS (= superstructural transition), Lo (= local transition), and Wi (= within segment transition). A dummy variable is turned "on" if the associated transition type is found (regardless of whether an actual pause occurs at that instance), and turned "off" otherwise. Hence, the model allows for an estimation of the average pause times for all transition types (the fixed parameters b ^ ) and the error-residues (the so-called random parameters, given between brackets). The latter parameters allow us to estimate the amount of pause time variance that is explained by the explanatory variable "location." Results and discussion. Table 1 shows the results of the analysis of mean pause times for the five structural transition types. Table 1. Estimates of mean pause times for five structural transition types. In order to be significant, estimates must be 1.96 times as large as as­ sociated standard errors (between brackets) transition types

estimates

switch between main lines attachments to main lines superstructural transitions local attachments within-segment locations

9.283 (1.909) 5.979 (.976) 4.085 (.529) 1.856 (Λ2Ί) .767 (.034)

All transition types show a significant effect on pause times (Chi-squares for simultaneous contrasts = 1492.06, df = 5, ρ < .0001). It can be concluded that pause time is not randomly distributed across transition types. Moreover, the direction of differences is as hypothesized. The longest pauses are found at switches between main lines, and the second longest pauses occur at transitions to main lines. The intermediate position is for pause times of pauses occurring at superstructural transitions. The shortest pauses are found at local transitions and transitions within segments. To sum up, the distribution of pause time reflects the gross hierarchical distinction between "global" and "local" transitions. In

262

JOOST SCHILPEROORD AND TED SANDERS

addition the more specific expectations are also borne out by the data. Finally, estimations of the random parameters showed that location types account for 27% of the pause time variance. At first sight, this may not seem very much, but one must always reckon with the fact that numerous other factors, not in­ cluded in the data model, may affect pause times. 5. Conclusion In this paper we have argued that research which combines two different empiri­ cal methods is likely to promote insight into the "black box" of a discourse pro­ ducer's cognitive representation; these two methods are analysis of discourse structure and analysis of the location and duration of pauses. Where the first analysis results in "static" structures which inform us about the product of writing or speaking, pause registration informs us about its dynamic process. We proposed to consider the cyclic patterns in the overall distribution of pause time the phenomena to be explained, and the hierarchical structure of the dis­ course as the main explanatory variable. As we argued in section 2, a prerequi­ site for such an approach is that we are able to assign structural descriptions to discourse in a reliable and intersubjective way. Such a way can be obtained if the analysis of discourse is based on its linguistic features, especially those fea­ tures that are relevant to discourse structure. The PISA-procedure, described in the third section, aims at doing just that. The data reported in section 4 show significant correlates between the hier­ archical structure of formal business letters and the real-time course of discourse production. Discourse producers are inclined to pause longer before segments located high in the structural hierarchy than before segments located low in the hierarchy. Although we have only just started to explore these relations in de­ tail, these preliminary results are at least suggestive. They suggest that the pro­ ducer's cognitive activities during writing are — to a certain extent - reflected in the hierarchical structure of the discourse produced. If it is correct to assume that differences in pause times reflect differences in cognitive effort needed to retrieve information from Long Term Memory, then we hypothesize that the hierarchical structure of discourse reflects the on-line level of accessibility of information. Suppose that (1) is the general format of each transition between two linearly arranged segments A and B'. (1)

A - > (pause) - > 

For reasons of convenience, let us assume that A and  correspond to two mental concepts A' and B'. The pause between A and  can be taken to reflect the process of retrieving B', and the length of that pause is, to some extent, de­ termined by the degree of the current, on-line accessibility of B'. The question

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

263

then is, what variables determine the on-line accessibility of B'? Arguably, the nature of the connection between A' and B' will be one such variable, while at the same time this connection is determined by the hierarchical structure of the discourse. After all, the structure of discourse accounts for our intuition that the segments in a discourse are connected to each other. The data reported, then suggest that if this connection involves the higher hierarchical levels, the acces­ sibility of B' will be relatively low as compared to circumstances in which this connection involves only the lower levels of the structural hierarchy. This line of reasoning may ultimately allow us to treat structural descriptions, such as "switch between main lines" or "local elaborations", in terms of on-line proc­ essing and the accessibility of information to be retrieved (cf. Schilperoord 1996, 186ff). This is a strong claim indeed, and one that needs careful exami­ nation. We intend to substantiate this claim in further work on discourse struc­ ture and discourse processing.

Appendix The sample letter discussed in section 4.2 1 2 2a 2 3 3a 3 4 5 6 6a 6b 7 8 8a 9 10 10a 11 12 13

Dear confrere ("my learned friend"), On behalf of Mr. NAME, living at ADDRESS, I hereby strongly object to the summary dismissal given to him by FIRM. First of all, this dismissal cannot be motivated considering your client's letter to my client, as you are well aware. And second, there is no motive at all. The fact that my client was in jail for a couple of days in connection with a case dating from long before he started working for your firm, is of no significance in this case. You have also informed me that my client has used abusive language, that is, that he started to taunt and rave after he was given his dismissal. He allegedly cursed your client. Again this is no reason for a dismissal, quite apart from the fact that he was given his dismissal beforehand. I hereby summon your client to withdraw the summary dismissal by return post. After careful consideration and with due reservation of all his rights, my client has decided that he will not resist a termination of his labor contract on an un­ employment-benefit neutral basis, if he were to receive to his benefit a supplementary amount of money of AMOUNT guilders. For your information, I should also like to add that my client foresees that it will not be simple to find another job in the short term. To this it can be added that his wife is severely disabled and [that] he has a family to take care of.

264 14 15 16

JOOST SCHOLPEROORD AND TED SANDERS

Moreover, his wife had to call in a doctor with some speed after the dismissal. My client has therefore suffered demonstrable damage due to your client's be­ havior. Please inform me at your earliest convenience,

Yours sincerely Note: all main clauses are numbered, subordinate clauses and interjections are marked alphabeti­ cally.

Acknowledgements We like to express our gratitude to Huub van den Bergh for his statistical advice, and to Bregje Holleman, Gisela Redeker, Carel van Wijk and two anonymous referees for their comments on an earlier version of this paper.

Notes 1. A note on terminology: In this paper, we will use the term discourse, rather than text, for two reasons. First, with respect to the issues addressed here, there are no principled differ­ ences between spoken and written (monological) discourse. Therefore, we prefer the more general term discourse. Second, the data gathered in the experiment concern dictated letters, therefore, the term discourse production is more appropriate. However, we also use the term text to refer to research that is restricted to written language. In addition, it is used to prevent misunderstandings; the field referred to as 'text analysis' is different from the one referred to as 'discourse analysis'. Therefore, the first term is used only to distinguish it explicitly from the field of discourse analysis. 2. The evidence presented in this section originates from studies of oral discourse production. Obviously, oral production differs from written production in many respects, but the ques­ tion is to what extent these differences need to bother us. In fact, we argue here that for the main cognitive processes studied here, i.e., those of retrieving information from Long Term Memory, spoken and written discourse are quite similar. 3. This was done in a response to Jaffe, Breskin & Gerstman (1972), who had claimed that the temporal cycles in Henderson et al. were but the artificial result of pause length variance dis­ tributed randomly over speech time (see also Henderson 1974). 4. The preliminary and exploratory nature of his study becomes especially apparent when he remarks: "Considering that cycle and Idea boundaries could occur between any two words in the text, the probability of this occurring was less than .01, calculated by means of a X2 test." (Butterworth 1975: 81). In other words, the conclusion that processing characteristics (i.e. cyclic patterns) and discourse characteristics (i.e. boundaries between ideaunits) are re­ lated, rests on the assumption that both cycles and idea boundaries may occur anywhere, which leaves his null hypothesis little chance to survive statistical testing. 5. The procedure is partly based on the text-analytic models mentioned earlier. At the same time, however, it tries to overcome a number of their shortcomings. Like LDM, the analysis proceeds incrementally and keeps track of the openness of attachment points. Yet, the appeal to world knowledge is restricted to dictionary-like knowledge. In line with RST, the need for

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

265

a global, i.e. top-down, guidance of the analysis is accepted, but the guiding features are as­ sumed to be part of a priori knowledge of discourse structure, instead of being based on an inspection of the entire discourse to be analyzed. 6. For other aspects of writing tasks that were examined using the text-analytical method, see: Sanders, Janssen, Van der Pool, Schilperoord & Van Wijk (1996), Sanders & Van Wijk (1996b), Schilperoord (1996), Van der Pool (1995). 7. "Quasi", since, obviously, it is impossible to manipulate the independent variable. 8. The experiment reported here is described in more detail in Schilperoord (1996, chapt. 4). 9. The fact that lawyers dictate written messages, rather then oral monologues, follows from several characteristics of their dictation. For example, they punctuate their letters, they insert corrections by deleting the original parts, and they provide lay-out instructions such as 'new line' and 'new paragraph'. For more details concerning the sampling and preparation of ma­ terials, and for additional justification favoring the idea that dictation is in fact written dis­ course production, the reader is referred to Schilperoord (1996, chapters 1 and 2). 10. The analyses were conducted together with Els van der Pool, Tilburg University (see Schilp­ eroord & Van der Pool, to appear).

References Bereiter, Carl and Marlene Scardamalia. 1983. "Levels of inquiry in writing research." In P. Mosenthal, L. Tamor and S. Walmsley (eds), Research on Writing; Prin­ ciples and Methods. New York: Longman, 3-25 (a shortened version is included in their 1987-compilation). Bereiter, Carl and Marlene Scardamalia. 1987. The Psychology of Written Composi­ tion. Hillsdale N. J.: Erlbaum. Boomer, Donald S. 1965. "Hesitation and grammatical encoding," Language and Speech 8: 148-158. Butterworth, Brian. 1975. "Hesitation and semantic planning in speech," Journal of Psycholinguistic Research 4: 75-87. Butterworth, Brian. 1980. "Evidence from pauses in speech." In Brian Butterworth (ed.), Language Production. Volume I. Speech and Talk. London: Academic Press. Dechert, Hans-Wilhelm and Manfred Raupauch (eds). 1980. Temporal Variables in Speech. Studies in Honour of Frieda Goldman-Eisler. The Hague: Mouton. Dijk, Teun A. van. 1977. Text and Context. Explorations in the Semantics and Prag­ matics of Discourse. New York: Longman. Dijk, Teun A. van and Walter Kintsch. 1983. Strategies of Discourse Comprehension. New York: Academic Press. Gibbs, Raymond W. 1996. "What's cognitive about cognitive linguistics?" In Eugene Casad (ed.), Cognitive Linguistics in the Redwoods. The Expansion of a New Paradigm in Linguistics. Berlin: Mouton de Gruyter. Goldman-Eisler, Frieda. 1967. "Sequential temporal patterns and cognitive processes in speech," Language and Speech 10: 122-132. Henderson, Allan I. 1974. "Time patterns in spontaneous speech - cognitive stride or random walk? A reply to Jaffe et al. 1972," Language and Speech 17: 119-125.

266

JOOST SCHILPEROORD AND TED SANDERS

Henderson, Allan I., Frieda Goldman-Eisler and A. Skarbek. 1966. "Sequential tempo­ ral patterns in spontaneous speech", Language and Speech 9: 207-216. Hobbs, Jerry R. 1990. Literature and Cognition. Menlo Park, CA: Centre for the Study of Language and Information, Stanford. Jaffe, Joseph, S. Breskin and L.J. Gerstman. 1972. "Random generation of apparent speech rhythms," Language and Speech 15: 68-71. Levelt, Willem J.M. 1989. Speaking: From Intention to Articulation. Cambridge MA: MIT Press. Mann, William C. and Sandra A. Thompson. 1988. "Rhetorical Structure Theory: Toward a functional theory of text organization," Text 8: 243-281. Pander Maat, Henk. 1994. Tekstanalyse. Een pragmatische benadering. (Text analysis. A pragmatic approach) Groningen: Martinus Nijhoff Uitgevers. Polanyi, Livia. 1988. "A formal model of the structure of discourse," Journal of Pragmatics 12: 601-638. Pool, Els van der. 1995. Writing as a Conceptual Process; A Text-Analytical Study of Developmental Aspects. Doctoral dissertation, Tilburg University. Redeker, Gisela. 1990. "Ideational and pragmatic markers of discourse structure," Journal of Pragmatics 14: 305-319. Sanders, Ted. 1992. Discourse Structure and Coherence. Doctoral dissertation, Tilburg University. Sanders, Ted, Daniël Janssen, Els van der Pool, Joost Schilperoord and Carel van Wijk. 1996. "Hierarchical text structure in writing products and writing proc­ esses." In Gert Rijlaarsdam, Huub van den Bergh & Michael Couzijn (eds), Theories, Models and Methodology in Writing Research. Amsterdam: Amster­ dam University Press, 473-492. Sanders, Ted, Wilbert Spooren, and Leo Noordman. 1993. "Coherence relations in acognitive theory of discourse representation," Cognitive Linguistics 4: 93-133. Sanders, Ted and Carel van Wijk. 1994. "Procedures voor Increméntele Structuurana­ lyse (PISA). Uitleg en toepassing op verklarende teksten" (Procedures for In­ cremental Structure Analysis (PISA). Explanation and application on explanatory texts). In Fons A. Maes, Peter van Hauwermeiren & Luuk van Waes (eds), Per­ spectieven in taalbeheersingsonderzoek. Dordrecht: ICG, 191-203. Sanders, Ted and Carel van Wijk. 1996a. "PISA - A procedure for analyzing the structure of explanatory texts," Text 16: 91-132. Sanders, Ted and Carel van Wijk. 1996b. "Text analysis as a research tool. How hier­ archical structure explains for Sentence Combining performances." In C M . Levy and S.A. Ransdell (eds), The Science of Writing. Hillsdale NJ: Erlbaum, 251-269. Schilperoord, Joost. 1996. It's about time. Temporal Aspects of Cognitive Processes in Text Production. Amsterdam: Rodopi. Schilperoord, Joost and Els van der Pool. To appear. "Schrijfpauzes en tekststructuur. Naar een cognitief model van realtime tekstproduktie" (Writing pauses and text structures. Towards a cognitive model of real-time text production.) To appear in Gramma/ITT.

PAUSES, COGNITIVE RHYTHMS AND DISOURSE STRUCTURE

267

Schilperoord, Joost and Arie Verhagen. 1996. Conceptual Dependency and the Clausal Structure of Discourse. Paper presented at the second CSDL-conference, Buffalo, USA, April 1996.

SUBJECT INDEX a contrastive processing 45 action scenario 207, 213 agreement 21, 27, 29, 49, 61, 64-66, 68, 79, 81, 140, 143, 146, 233 AIDS 176, 178-180 analogical reasoning 150, 151 an exceptional case processing 45 artificial dialogue 179 attention movers 228, 234 backwards processing 46 bias, 109, 116, 123, 124, 132, 137, 138, 139 caricature 21, 29, 30, 32-35 chimerical concept 163, 180 cognitive grammar 21, 22, 25, 37 text analysis 253 representation 86, 94, 96, 247, 253, 262 rhytmus (during production) 247, 248-250 coherence 65, 185, 249 commissives 101, 215, 218 conceptual blending 185, 186, 192, 195, 198, 202 mappings 185, 188, 189, 202 metaphor 115, 122, 124, 137, 144, 185 constraints on implicatures 46-48, 57 construction 3-8, 10, 13, 15-17, 28 39, 70, 71, 74, 78, 79, 82, 107, 157, 161, 168, 179, 187, 198, 209 construction-task dialogue 81 context selection 39, 41 contextual effect 40-43, 47, 49-53, 58 convention 25, 71, 83 creative problemsolving 179 creative thinking 149, 159, 179

cyberfuture metaphor 187, 190, 199, 200 cyberspace 185-187, 189, 190, 192197, 199-201, 203 denial of a contextual assumption 4042, 49-52 deontic modality 86, 97, 99, 101, 103, 106, 110 Depth of intention 61, 62, 64-66, 68, 69, 70, 82 determination of depth of intention 82, dictation 259, 265 directives 100, 101, 206-215 disagreement 49, 62, 64-66, 69 discourse analysis 264 production 247, 250, 251, 262, 264, 265 strategy 4, 5, 14, 16, structure 247, 248, 251, 252, 253, 262, 263, 265 distance 237, 251 economics 115-119, 121, 123-128 elaboration 76, 79, 80, 211, 256, 257, 259, 260, 263 epistemic modality 86, 97, 99, 103, 106, 146 evidential 16,96, 111 experience 22, 25-27, 33, 35, 83, 115, 116, 152, 188, 189,224,225 explicature 41, 42, 49-51 expressives 26 focus movements 223, 225, 231, 232, 234, 235, 241 focus of attention 223, 229, 231, 232, 234, 238, 239, 240 grammaticalization 4, 5, 13-16

270

SUBJECT INDEX

grounding 22, 23, 25-27 HEALTH 58, 116, 118, 123-127 hedged performatives 206, 213, 215, 216 hierarchical 247, 248, 250-257, 259, 261, 263 HIV 160, 161, 165, 174-176, 180 hypertext 160 icon 21, 22, 29-32, 35 iconographie reference 21, 26, 27, 35 ideology 22, 24, 25, 27, 28, 35, 115, 118 image schémas 115, 116, 185, 189, 241 indirect speech acts 206, 207, 209, 213,217 information highway metaphor 187192, 195 interjection 11-15, 17, 233, 264 internal images 223, 241 interpellation 22, 24, 25, 35 intonation units 224, 226, 235-238, 243 invariance hypothesis 116 involvement 196, 197, 203 judgement 224 Lebenswelt 152 linguistic relativity 115, 128 (literary) critical theory 22 meaning 4-6, 11-17, 27, 37, 40, 43-45, 49, 57, 58, 62, 66, 69, 79, 81, 85, 89, 90, 93, 94, 97, 98, 102, 109, 161, 165, 167, 180, 194, 217, 251 mental space theory 37 metaphor 13, 26, 115-118, 120, 122128, 131-133, 135-146, 149, 150, 151-163, 165-170, 172, 173, 176180

after-the-fact 169 before-the-fact 169 metaphorical reasoning 150, 151, 153, 154, 156-161, 170, 173, 179, 185 190, 191, 193 metonymie reference 205 metonymy 12-14, 205, 206, 214-216, 218 modality 3, 8-12, 16, 85, 86, 91, 96, 97,99, 100, 102, 103, 106, 110 marker 13, 14 modalization 10 molecular biology 177, 178 mutual understanding 61 natural inference schemata 219 non-monotonic reasoning 90, 93 optimal relevance 40 paraphrase 40, 75, 80 parallel processing 46, 48, 49 particle typology 37, 57 PATH 14, 115, 116, 118-123, 126, 127 pauses 224, 232, 243, 247, 247, 248, 254, 257, 259-262 perspective 4, 5, 16, 22, 23, 25, 85, 86, 88-91, 94-96, 98, 99, 101, 103105, 107-109, 143, 146, 149-152, 163, 166, 168, 179, 233, 242 perspectivization 86, 88, 90, 93, 94, 96, 97, 100, 102, 104, 106, 107, 109, 110 persuasion 125, 205 polarity 3 , 8 , 10, 11, 13-15 politics 21, 27, 29, 127 popular science texts 179 pragmatic operations 149, 150, 154, 155, 158-159, 179 precision 69 problemsolving 149-151, 179 procedural meaning 37, 40, 43, 45, 49,

INDEX

57 processing effort 40, 52, 53, 55, 58 radio broadcast 149, 169, 176 Relevance Theory 61 scientific journalism 168 selection 25-28, 33, 39, 41, 43, 153, 173, 254 semantic operations 149, 150, 156, 158-160, 179 shared knowledge 153, 154, 167, 177, 179 social cognition 169 social constructionist 25 software piracy 198-200, 201 spatialization 197 subjectification 4, 5, 16, 86, 91, 93, 94, 96, 97, 101-104, 106, 107, 110 specification 80, 86, 157, 251, 252 speech acts 27, 86, 206-213, 217, 256 spoken discourse 89, 102, 103, 223, 224 subjectivity 5, 85, 86, 88, 91, 93, 94,

271

96, 97, 99, 100, 102-104, 106, 107, 109, 110 superstructure 255, 257, 260 tag question 3-6, 8-10, 13-17 taxonomy of metonymies 207, 213 technology policy and metaphor 185, 203 text analysis 247, 264 transitions between attention focuses 232 truth 242 up-/downwards processing 43, 44, 54 verbal interaction 149-154, 156-160, 169, 179 viewing 62, 126, 133, 139, 142 virologist 160, 161, 180 virology 154, 176, 177 virus 160, 165-167, 174-178 visualization of language 242 WAR 21, 27, 116, 125-127, 179

List of Contributors

Frank Boers Free University of Brussels Belgium

Wolf-Andreas Liebert University of Trier Germany

Maria Josep Cuenca University of Valencia Spain

Klaus Panther University of Hamburg Germany

Murielle Demecheleer Free University of Brussels Belgium

Tim Rohrer University of Oregon U.S.A.

Michele Emanatian University of MassachusettsAmherst, U.S.A.

José Sanders Nijmegen University The Netherlands

Bruce Hawkins Illinois State University U.S.A.

Ted Sanders Utrecht University The Netherlands

Isao Higashimori Kobe College Japan

Joost Schilperoord Utrecht University The Netherlands

Richard Hirsch Linköping University Sweden

Wilbert Spooren Tilburg University The Netherlands

Kenneth Holmqvist Lund University Sweden

Linda Thornburg Eötvös Loránd University of Budapest, Hungary

Jana Holsánová Lund University Sweden