Approaches to Cognition through Text and Discourse 9783110892895, 9783110177916

Citation preview

Approaches to Cognition through Text and Discourse

W DE G

Trends in Linguistics Studies and Monographs 147

Editors

Walter Bisang Hans Henrich Hock Werner Winter (main editor for this volume)

Mouton de Gruyter Berlin · New York

Approaches to Cognition through Text and Discourse

edited by Tuija Virtanen

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

Printed on acid-free paper which falls within the guidelines of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Approaches to cognition through text and discourse / edited by Tuija Virtanen p. cm. - (Trends in linguistics. Studies and monographs ; 147) Includes bibliographical references and index. ISBN 3-11-017791-9 (alk. paper) 1. Discourse analysis — Psychological aspects. 2. Cognition. I. Virtanen, Tuija. II. Series. P302.8.A68 2004 401'.41-dc22 2004018522

ISBN 3-11-017791-9 Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at . © Copyright 2004 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Christopher Schneider, Berlin. Printed in Germany.

Acknowledgements

The present volume contributes to bridging the gap that exists between discourse linguistics and cognitive linguistics - fields which share an interest in issues of discourse and cognition but differ in the frameworks and perspectives adopted for study. The first step towards this goal was a one-day thematic session on discourse approaches to cognition which I organized at the 6th International Cognitive Linguistics Conference in Stockholm, Sweden, 10-16 July, 1999. Several of the studies included in the present volume are revised versions of papers originally presented, in person or on video, in that context and I wish to thank the participants in the very popular theme session for making it such an inspiring event. While the conference papers have been extensively revised and in some cases given a different focus, other studies not presented at the conference have since been added to extend the range of topics covered by the individual chapters of the volume. It is obvious that the number of topics that could be covered is very large; the present selection includes an overview of language, discourse and cognition, and eight studies which explore cognitive aspects of information structuring, coherence, foregrounding, knowledge structuring, negotiation for meaning, and interpretation of recontextualized material. I wish to thank the contributors both for their interesting contributions and their enthusiasm for this project. I am grateful to Professor Werner Winter, the series editor of Trends in Linguistics. Studies and Monographs (TiLSM), Dr Anke Beck, editor-in-chief at Mouton de Gruyter, and Birgit Sievert, the managing editor of TiLSM, for accepting the present volume for publication in the series. I also wish to thank the anonymous reviewer of the manuscript for insightful comments and suggestions. Special thanks are due to M.M.Jocelyne Fernandez-Vest for encouraging me to go ahead with the planned volume. I am also indebted to Nils Erik Enkvist and Jan-Ola Östman for illuminating discussions concerning discourse and cognition. Last but not least, my heartfelt thanks go to my husband, Fredrik Ulfhielm, whose unfailing support and interest constitute a sine qua non of my work. December 2003, Turku/Abo, Finland

Tuija Virtanen

Contents

Acknowledgements

v

Chapter 1 Text, discourse and cognition: An introduction Tuija Virtanen

1

Chapter 2 Language, discourse, and cognition: Retrospects and prospects Robert de Beaugrande

17

Chapter 3 On the discourse basis of person agreement Anna Siewierska

33

Chapter 4 The information structure of bilingual meaning: A constructivist approach to Californian Finnish conversation M M Jocelyne Fernandez- Vest

49

Chapter 5 Point of departure: Cognitive aspects of sentence-initial adverbials Tuija Virtanen

79

Chapter 6 What is foregrounded in narratives? Hypotheses for the cognitive basis of foregrounding Brila Wärvik

99

Chapter 7 From legal knowledge to legal discourse - and back again Lita Lundquist

123

Chapter 8 Conditionals: Your space or mine? Anne Marie Bulo\v-M0ller

149

viii

Table of Contents

Chapter 9 Communicative fragments and the interpretation of discourse Martina Björklund

177

Chapter 10 Drawing the line: A contested conceptual model in Danish 'child care talk' Peter Harder

187

Index

209

Chapter 1 Text, discourse and cognition: An introduction Tuija Virtanen

Linguists of different orientations, focusing on very different areas of the study of language, regularly come into contact with phenomena that can be related to cognitive aspects of language use. While much of the work done within the overlapping fields of text linguistics and discourse analysis touches upon phenomena which can be characterized as cognitive, much of the work within cognitive linguistics is, in its turn, concerned with areas that are also familiar to us from the study of text and discourse. The shared interest in discourse and cognition urgently calls for discussions between these and other groups of linguists, despite the present gap that exists between them caused by differences in the perspectives and frames of reference adopted for study. The aim of this volume is to contribute to bridging that gap, by offering a forum for discussion with a relatively broad scope to meet the current general interest in the area of discourse and cognition. The volume opens with an overview chapter on the relations between language, discourse and cognition, in retrospect and prospect. The rest of the volume consists of another eight chapters focusing on central issues and highlighting methodological differences motivated by the frameworks from which these issues are approached. While we have recently been able to witness a welcome increase in studies which adopt a cognitive perspective on discourse (cf. e.g. van Hoek et al. (eds.) 1999; Koenig (ed.) 1998; Lundquist and Jarvella (eds.) 2000; Sanders et al. (eds.) 2001), it is important to note that the individual chapters of the present volume clearly opt for a discourse approach to cognition. In this chapter the concern is first with basic notions in text and discourse linguistics - a term used to refer to the overlapping fields of text linguistics, discourse analysis/studies and conversation analysis. The main purpose of the discussion below of 'text', 'discourse', 'context', and variation across these, is to address issues which most clearly reflect the kinds of changes that have taken place in the study of text and discourse over the years. Unless clarified, these notions also constitute a threat to successful communication between discourse linguists and cognitive linguists. Section 2 is an introduction to the individual chapters of the volume.

2

Tuija Virtanen

Both 'text and discourse linguistics' - or 'discourse linguistics' for short and 'cognitive linguistics' are here used in a broad sense, as umbrella terms for a range of different frameworks. Text and discourse linguists thus focus on text and discourse in context, while the concern in cognitive linguistics is primarily with individual and/or distributed cognition. The chapters of this volume attempt to combine the two interests.

1.

Approaching cognition from the perspective of text and discourse

Let me start this section with a brief personal note. In the 1980's when I was working in the research group 'Style and Text as Structure and Process', directed by Professor Nils Erik Enkvist in Abo, Finland, the focus was on text and discourse as process in terms of various text (or discourse) strategies involving parameter weighting in relation to particular communicative goals. The texts analysed in terms of structure were, in the first place, considered products of processes constituting the actual purpose of the analyses; analysing structures was a method to get at the processes which were assumed to lie behind them and the motivations that had made particular text producers in particular contexts opt for one set of alternatives rather than another, to give textual parameters a particular weight and value in view of particular communicative goals. Such goaloriented weightings of decision parameters were investigated to understand the ways in which particular text strategies worked towards the elimination of the interlocutors' uncertainties through the exclusion of alternatives, and how they facilitated the production and/or interpretation of discourse. They were also investigated to find out what they told us about the construction of textuality through information structuring, coherence and text segmentation, what they told us about intertextuality, and about the 'textual fit' of sentences in written texts. Discussions thus also dealt with relations between sentence grammar and text where these could be seen to be pulling in different directions. Furthermore, studies focused on the relative salience of particular linguistic material in relation to its immediate context and the Figureness of a profile found in a text in relation to its Ground, the effects of experiential iconicity detected in texts, and the influence of perspectivization on the form of a text. Impromptu speech was examined to understand how hesitation phenomena disclosed aspects of on-line discourse processing, and the ways in which interlocutors negotiate meaning. Central questions also included what inferences need to be made for interlocutors to come to grips with what is left implicit in texts and discourses, what expectations people set up in text processing, and how interlocutors cope with

Text, discourse and cognition: An introduction

3

expectations not met in the subsequent text. The interplay of the textual factors affecting the form of the actual text, i.e. the product - their 'conflicts and conspiracies' in text processing, to adopt Enkvist's phrase - were ultimately connected with the basic ability of human beings to make goal-oriented, context-based decisions, and to the interpretability of texts and discourses. To cite Enkvist (1989: 166) on 'interpretability': Text comprehension and interpretability can thus be seen as a highly complex, incremental process involving the interplay of bottom-up and top-down processing, as well as zig-zagging between the text, the universe of discourse meaning the universe at large within which the text can be placed, and the specific world of text with its specific, usually highly constrained states of affairs.

The views that we held on text and discourse as structure and process were well in line with those prevailing at the time (cf. e.g. de Beaugrande 1980; de Beaugrande and Dressier 1981; Brown and Yule 1983; van Dijk and Kintsch 1983). In fact, as de Beaugrande notes in Chapter 2 of the present volume, "text linguistics has always had a resolutely cognitive orientation because the text must be described as both product and process." Several of the basic notions of text and discourse linguistics have since developed in ways which make them, at one and the same time, even more dynamic but also much more indeterminate, and not necessarily grounded in assumptions of rationality and intentionality in human communication. In light of the relations of discourse and cognition, it therefore seems in order to pay some attention to the primary notions of 'text' and 'discourse', their relation to 'context', and the kinds of variation that can be found across them. To start with text and discourse, it is obvious that for many linguists discourse consists of text and its situational context; for others texts are primarily written while discourse refers to spoken interaction; and for others still, only one of these concepts is necessary to cover the field of their study. All in all, however, both terms have acquired a more processual and interactional reading over time, and the question that cognitive linguists might wish to ask at this point is thus whether the two notions are needed and if so, what senses will best be ascribed to them if we try to keep both of them transparent for the benefit of all linguists. In what follows, the two terms will be used in the British sense of discourse including text and its situational and socio-cultural context. Discourse can be viewed as a cognitive process involving interaction, adaptation, and negotiation between interlocutors, but it can also be viewed as a social phenomenon in a very wide sense of the term. To fully understand what people do with language it is necessary to take into account both cognitive and social concerns. This raises the issue of the status and scope of the concept of

4

Tuija Virtanen

'context'. For the purposes of cognitive linguists, context is a mental phenomenon, included in the broad notion of 'conceptualization'. In text and discourse linguistics context has traditionally been regarded as having three dimensions: we speak of the textual or linguistic context (the 'co-text'), the situational context, and the socio-cultural context. The different dimensions of context thus have to be related to the notion of context as used in cognitive linguistics, since all of them can be considered mental representations of different kinds but activated in parallel. Furthermore, the development of text and discourse linguistics has witnessed a change in the relative status of the notions 'text' and 'context'. Whatever reading is given to text and discourse, it is important to note that from the beginning both have been viewed as closely tied to their situational and socio-cultural contexts. Recently, however, the focus of this relationship has shifted to the context itself. Hence, while the form that a particular text takes is undeniably affected by its context, it is also true that texts and discourses themselves mediate contexts and in so doing contribute to constructing, maintaining or altering those contexts. This bidirectional view of text and context has turned text producers and text receivers into interlocutors or participants in discourse, engaged in discourse practices, be they spoken, written, signed, or computer-mediated multimodal phenomena. The recognition of a two-way traffic between texts and contexts has, in turn, raised the issue of the inseparability of the two. And this view has important implications for the cognitive notion of context as a mental phenomenon. Much study in the area of text and discourse has traditionally been centred around individual cognition, the inferences that people are required to make to interpret discourse, and the assumptions they seem to be making about their interlocutors' consciousness and memory constraints, as manifested in discourse. Metadiscoursal exponents reveal awareness of context-based strategic planning, irrespective of whether such planning is considered rational or common-sensical and what interpretation is given to the idea of intentionality. Criticism has since been voiced of the idea of 'autonomous minds' (cf. e.g. Linell 1998), i.e. the assumption that people communicate as individuals with ready-made content in their minds. It is, however, worth noting that even in studies of discourse governed by the idea of individual cognition, the goal of the analyses and their focus has been the processes that take place when people communicate with one another in authentic situational and socio-cultural contexts. The recent shift of interest to 'distributed cognition' has brought with it a greater concern for indeterminacy in discourse. We can see that communicative goals can be of many kinds: interlocutors can act in an explicitly intentional fashion but do not do so in all contexts and for all purposes. Communicative goals can be negotiated and adapted to, not only in impromptu speech but also

Text, discourse and cognition: An introduction

5

in the discourse shaping the ideologies of a given society and culture at a given point in time. Further, interlocutors can be aware of and content with the fact that the inferences they draw are not necessarily shared by others and that interpretability can be a matter of several different or overlapping alternatives activated simultaneously. The motivation for selecting a particular interpretation in a particular context can be related to the construction of that context, in the sense of adaptation, negotiation, persuasion and other dimensions of the activity of co-construction of discourse by interlocutors. The focus on distributed cognition highlights the kind of intersubjectivity that emerges through interaction both locally in a particular situational context and on a macro-level, in a particular culture and across cultures. Still, even in a globalising world - whichever of the many senses we choose to give to that notion - it is important to remember that in the first instance, people do live locally. Interlocutors are regarded as being engaged in discourse practices of various kinds which manifest and constitute socio-cultural patterns of action and thought. And interlocutors can show metalinguistic awareness of such shared patterns, even though they do not necessarily conform to them in practice. This suggests that the role of variation across texts and discourses is another central area in the study of discourse and cognition. A burning issue here is how we can best capture the dynamism connected with these phenomena. Two dimensions of variation have traditionally been in the focus of text and discourse linguistics. To start with, text types - or discourse types - originate in the rhetorical tradition and have to a large extent been studied by those interested in building theoretical models of language to increase our understanding of language use. Genres, again, have been adopted into linguistics from the study of literature, predominantly by those whose primary aim has been to apply the notion to practical purposes such as the study of professional discourse or learner language of a particular kind. Both notions are essentially prototypical, and they should be given a dynamic interpretation and studied in terms of the dialectical relation in which they enter with contexts of situation and culture. Furthermore, both notions seem necessary since text/discourse types cut across genres in interesting ways. Yet, however dynamically modelled, no typologies are as such sufficient to explain the variation that emerges from the actual use of texts and discourses and the mechanisms that allow interlocutors to cope with and profit from such variation in their daily lives. Apart from obvious pedagogical purposes, text classifications are still justified precisely because they help us to understand linguistic categorization and the ways in which people make dynamic use of them in continuously adapting and contributing to variation across texts and discourses, making sense of their experience of the world, and recontextualizing linguistic material in ways that have a bearing on

6

Tuija Virtanen

their participation in speech communities and professional discourse communities. Such issues have recently been the focus of studies of 'intertextuality' and 'interdiscursivity', and of the interdisciplinary field of 'genre analysis'. Traditional typologists have argued that text/discourse types reflect the way in which we view reality; thus even such static types would seem to have an obvious cognitive basis. While prototypical types of text have been shown to differ across cultures and some of them are connected more readily than others with particular educational systems and the ideologies that these convey and help to construct, some rather high-level distinction between narrative and non-narrative can probably be given a primarily cognitive justification. Hence, the way in which we construct reality through a narrative flow of time involving human beings in a dynamic series of actions that have an outcome of some sort which is different from the situation at the beginning of the series, is very different from the way a location, a concrete object or an abstract notion is described or explained to other people or the way in which values and beliefs are constructed, mediated, and negotiated through discourse of the openly argumentative kind. If people use text types as prototypical categories against which they can match texts to detect similarities and differences and on which they can rely in recontextualizing and creating other texts in context, then the study of intertextuality in terms of text/discourse types should indeed also be very much part of the core of cognitive linguistics. Idealised types emerge and change through authentic texts, and idealised types stored in memory affect the form of authentic texts. Finally, what people actually do in their daily lives with and through narrative, and why, has proven to be fundamental for our understanding of the relations between discourse and cognition. In this light, genres seem much more context-oriented than text types. Genres emerge and are maintained or altered in particular socio-cultural contexts, helping to construct those contexts in the first place. Essentially this takes place through interdiscursivity, in the sense of genre-mixing and embedding of emergent and established genres in one another. Hence, to a greater extent than text types, genres appear to be in line with the socio-cultural concerns that are in focus in discourse linguistics, pragmatics and social sciences today. While it is possible to argue that anything in language use can be ascribed a cognitive function and it is indeed possible to study any linguistic phenomenon from a cognitive point of view, the very origin of genres seems to be the discourse community and socio-cultural context which they help to construct. Genres are thus essentially part of our distributed cognition, of our social memory, or the shared knowledge of a particular discourse community. Perhaps we can conclude by placing text/discourse types in a framework where the focus is on the first member of the text-context pair, whereas genres, as the dynamic constructs

Text, discourse and cognition: An introduction

7

that they are regarded as today, would rather tend to highlight the second member of the pair. Genres would thus seem to make 'context' more Figure-like than the actual texts representing or helping to construct or develop those genres. Taken together, the study of text/discourse types and genres is an area which will add to our understanding of the interplay of the cognitive and social concerns and constructs. Such an interplay is manifest in discourse, and it motivates people to use language in particular ways. Variation across texts and discourses thus demands some consideration in cognitive linguistics at least as concerns the mechanisms of emergence and use - or indeed, non-use - of text categorizations of various kinds, the distinction between text/discourse types and genres, and their relations to the notion of 'context' as a mental phenomenon. Among text types, an important and cognitively highly relevant area of study concerns the status of narrative in relation to other types. Genres have been in focus in the study of professional discourse and the interaction between professionals of various kinds and/or non-professionals in institutional discourse. At the same time, the study of genre dynamism in non-professional discourse seems to be an area where models of analysis are at present lacking. This is a task where a combination of the expertise of both discourse linguists and cognitive linguists might prove to be particularly advantageous. Furthermore, another area of growing relevance in today's rapidly changing world of communication is multimodality, in the sense of the construction and interpretation of multisemiotic meaning. Hence, the dynamism of multimodal variation emergent in the new media awaits thorough study, and this is yet another forum where cognitive linguists and discourse linguists can profitably meet. Text and discourse linguistics constitutes a very wide field of study in which proponents of different frameworks and methodologies meet to investigate a shared interest: text or discourse. Similarly, in cognitive linguistics the common denominator is what constitutes the object of study, i.e. cognition. As Geeraerts (1995: 114) rightly points out, what people with different orientations and methodologies refer to as cognitive linguistics is "not a single theory of language, but rather a cluster of broadly compatible approaches." The meeting point of discourse linguistics and cognitive linguistics has traditionally been the individual mind, engaged in interaction with other minds. Recent developments in the humanities and social sciences have, however, brought social cognition very much to the fore. Hence, to cite for instance Edwards (1997: 28), "cognition is [...] clearly a cultural and discursive matter." The studies included in this volume have the purpose of contributing to our understanding of the area in which the individual and distributed cognition meet, i.e. discourse. In doing so, they also forcefully testify to the benefits of communication across academic disciplines and invite further discussions in that spirit.

8

Tuija Virtanen

2.

The studies included in the present volume

The individual chapters of the present volume focus on both text and discourse, and their textual, situational and socio-cultural contexts. The data range from narrative to non-narrative, written to spoken, informative to literary, experimental to authentic, professional genres to impromptu speech, and from public to semi-private or private discourse. Several languages are discussed, including discourse originating in bilingual speech communities. We also find cross-linguistic data at a high level of abstraction where a certain extent of decontextualization seems mandatory, and we find data originating in automatic searches run on very large corpora of authentic texts and discourses, yet obviously recontextualized to appear in new formats on computer screens and printouts. The discussion proceeds from the construction of textuality, in terms of considerations of information structuring, coherence, and foregrounding, in (inter)textual and interdiscursive contexts, and across languages of very different kinds, to examinations of negotiation of meaning, distribution and recognition of knowledge, and the recontextualization and interpretation of discourse in textual, situational and socio-cultural contexts. Hence, the context taken into account expands as we proceed through the different chapters. The volume highlights linguistic elements of relatively small size within and across languages and cultures; it also highlights local and global discourse phenomena ranging across parts of texts or entire texts. Such discourse phenomena are signalled with the help of linguistic elements that serve important cognitive functions in the use of language. The cognitive issues brought to the fore by the individual studies range from accessibility and assumptions we make about our interlocutors' memory constraints to information structuring, the relative salience we give to elements of discourse, pro to typicality, figure-ground distinctions reflected in language use, knowledge structure, locative concerns, iconicity, recontextualization of communicative fragments originating in social memory, negotiation between people or groups of people of mental spaces and the establishment of shared conceptual models within society at large, pertaining to culture and ideology. The studies thus vary in their relative orientation towards individual and distributed cognition. What is common to the contributions to this volume is a concern for context. In light of cognitive linguistics, these studies open avenues to ways of conceptualizing this primary concept of text and discourse linguistics which is notorious in resisting definition but without which we can hardly hope to approach text and discourse at all. The focus of several studies is on individuals processing text and discourse in a given situational context, making sense of and mediating their experience of the world through discourse, constructing

Text, discourse and cognition: An introduction

9

conceptual models and activating mental spaces, recontextualizing elements of earlier texts and discourses in novel ways, and negotiating common ground through adaptation to and manipulation of what they assume to be their interlocutors' activated conceptual models and their possibly shared mental spaces. But individuals form part of and help to create socio-cultural contexts that are of prime relevance to a discourse approach to cognition. Hence, other chapters rather focus on what can be regarded as distributed cognition, the collective memory of a particular speech community, socio-cultural knowledge shared by most members of a speech community, or again, professional knowledge shared by members of discourse communities, which is, in turn, constitutive or regulative of such communities. Chapter 2, entitled "Language, discourse, and cognition: retrospects and prospects", is by Robert de Beaugrande. Examining discussions of language and cognition in the linguistic literature, the chapter opens with the claim that while the wheel has turned a long way, the relation between cognition and language has not, as yet, been adequately defined. A dialectical model of 'language' and 'discourse' is then presented such that language specifies the standing constraints of discourse while discourse manifests emergent constraints for language. The traditional division into 'theory' and 'practice' is thus fundamentally called into question as the dialectics between language and discourse turn the theory of language into 'theoretical practice', essentially 'practice-driven', and the practice of discourse into 'practical theory', essentially 'theory-driven'. A similar dialectical model is suggested to account for the relations between 'cognition' and 'language', such that cognition generates meanings while language determines meanings. The fundamental problem for linguists is of course the fact that "meanings will not hold still or remain constant while we 'analyse' them." This chapter provides us with a number of analogies which we can use in our attempts to understand cognition and its relation to language and discourse. Tacit assumptions and explicit statements originating in several schools of thought are addressed in light of current models of cognitive processing. The chapter closes with an agenda for some of the most urgent tasks for what the author calls 'cognitive text linguistics' - and which is what by any other name the contributors to this volume are promoting by simply participating in this endeavour. The variety of methods and frameworks represented in this volume, together with many others outside it, testify to the richness of work in this field but also to the fact that there is no one way of accomplishing what these and other linguists are engaged in doing. What is thus urgently needed is a forum for discussion between proponents of the different frameworks and practitioners of the

10

Tuija Virtanen

different methodologies. Rather than adding to the fragmentation of today's linguistic map, we need to create connections between text linguistics, discourse analysis, conversation analysis, pragmatics and related fields, to make explicit the ties between this area of study and cognition. And as in this volume, we need to build bridges between discourse linguistics and cognitive linguistics. One way of starting to do this, de Beaugrande suggests, is to study meanings in very large corpora of text and discourse. The study of collocations, he predicts, can lead us to a level between 'language' and 'text/discourse' with the missing links partly explicated, and to the content of meanings in terms of the multiple activations in networks with other meanings of the events thus presented to us. Chapters 3-6 deal with ways of structuring information, creating coherence, and signalling foregrounding. Here we are focusing on the cognitive aspects of the construction of textuality. In Chapter 3, "on the discourse basis of person agreement", Anna Siewierska examines person agreement marking cross-linguistically, addressing the issue of the discourse context for its development in light of two diachronic scenarios that have been proposed in the literature. In the first of these, person agreement markers originate in the third person while the second, based on accessibility theory, postulates first and second person pronouns as the source of grammaticalization of person agreement. Studies of discourse have often tacitly assumed that narrative constitutes a good exemplar of discourse, and the focus has usually been on third person narratives, which may, but need not, contribute to the discourse basis of the first scenario. The encoding of discourse referents in impromptu speech, however, would rather seem to point to the necessity of taking into account the fact that first and second person referents are inherently more accessible in memory than third person referents. While critically examining the claims concerning the discourse bases of the two diachronic scenarios, this chapter argues for an extension of the second model, based on the accessibility of referents in memory, such that (a) not only high accessibility but a wider range - or different levels - of accessibility be taken into account, and (b) third person pronouns, as well as both subject and object functions, be included in the scenario. In this way, Siewierska points out, it will be possible to fully assess the merits of the model in terms of cross-linguistic data. Chapter 4 presents a series of studies by MM Jocelyne Fernandez-Vest, of "the information structure of bilingual meaning", in which she adopts "a constructivist approach to California Finnish conversation". Information structuring is examined in impromptu speech emerging in both monolingual and bilingual contexts, using a tripartite model of Theme-Rheme-Mneme. The last of the three concepts refers to elements following the Rheme which carry a flat intonation, convey shared knowledge, and receive affective modulation. Analysis

Text, discourse and cognition: An introduction

11

of the dialectics of bilingual meaning as the outcome of a co-determination of discourse-pragmatic and morphosyntactic factors in American Finnish oral narratives reveals a tendency for the Rheme or the Mneme to be marked by codeswitching, from Finnish to English. In non-narrative, again, code-switching is mainly motivated by situational needs such as the presence of a monolingual interlocutor or the need to be exact about referents related to life in the USA. This latter need is also shown to occasionally invite metalinguistic comments. The second main motivation for code-switching in impromptu speech has to do with the interlocutors' memory processes. When another piece of discourse is recontextualized in the form of a memorized quotation or social situation with a high degree of affect which took place in English, these tend to trigger code-switching into the language of the memorized situation. Finally, the chapter deals with the social parameters of quantitative memory in Sami contexts, where men and women are shown to manifest different mechanisms of remembering dates, estimating distances, and so forth. Applied to American Finnish contexts, the analysis shows that the sex or gender differences here concern code-switching at temporal signposts of a story. All in all, the chapter reveals a high degree of creativity in the way bilingual interlocutors construct textuality through information structuring where code-switching serves an important function. Through the decision to focus on information structuring in bilingual discourse, this chapter presents a new approach to code-switching, linking it closely to cognitive concerns. In Chapter 5, entitled "Point of departure: cognitive aspects of sentence-initial adverbials", I revisit sentence-initial adverbials, exploring their discourse functions in written texts from a cognitive point of view. To start with information structuring, sentence-initial adverbials appear in the theme position, which constitutes the prominent starting point of the sentence: theme can be seen as Figure in relation to the Ground of the rheme. Elements that interlocutors choose to place in this position are thus informationally foregrounded at the stage at which they appear in the text. Subsequently, of course, they will be integrated into the Ground. In terms of persuasion, it is important to note that thematic adverbials can mediate information, beliefs and attitudes that the writer wishes the reader to interpret as given even when they cannot be automatically assumed to be so. Sentence-initial adverbials can, in fact, convey information of any kind. The extremes of brand-new information and textually given information are, however, less usual here than inferrable and unused information, which are found in the middle of the given-new scale. This variation also accounts for the efficiency with which sentence-initial adverbials contribute to the signalling of textual boundaries. After a discussion of thematic adverbials in light of coherence, text type, and iconicity of two different kinds, the chapter concludes by

12

Tuija Virtanen

claiming that the sentence-initial slot itself constitutes a rich source of discourse meanings precisely because of its cognitive relevance for our processing capacities and memory constraints. After three chapters on information structuring and coherence, we proceed to a discussion of (fore)grounding in Chapter 6, entitled "What is foregrounded in narratives? Hypotheses for the cognitive basis of foregrounding". In this chapter Brita Wärvik explores parallels between the textual foreground-background distinction in narrative and perceptual and cognitive principles of organization. The relative foregrounding and backgrounding of elements in narrative is here considered from three cognitive perspectives, i.e. those of the figure-ground distinction, prototype, and salience. The chapter also includes a discussion of relevant aspects of iconicity. One of the most urgent tasks for students of grounding, Wärvik argues, is to devise a model to come to grips with the relative weightings of the different grounding criteria so far established for narrative. Another obvious, but similarly in no way straightforward task is to systematically establish grounding criteria for non-narrative text. Scrutiny of the cognitive basis of the meticulous system of grounding criteria presented in this chapter reveals important links and networks of links between different approaches to foregrounding in the study of discourse. It also serves to relate principles of textual organization to corresponding phenomena in other fields of human activity. At this point a terminological note is in order. The use of the term 'grounding' in text and discourse linguistics and the study of literature, to indicate the relative degree of foregrounding vs backgrounding of elements in a text, does not correspond to the use of the term 'grounding' in cognitive linguistics (manifest in this volume in Chapter 10, where the focus is on the 'discourse grounding' of a conceptual model). Further, when the term 'foregrounding' appears in Chapter 8 (on the use of conditionals in interaction), it simply functions as a metaphor for dominances of scales connected with discourse-pragmatic processes in face-to-face argumentation. Thirdly, Chapter 5 touches on the status of sentence-initial elements as 'informationally foregrounded' at the stage where they appear in the text. This status is, however, only related to the Figureness of the theme in relation to the rheme. In contrast, (fore)grounding as dealt with in Chapter 6, which is entirely devoted to the phenomenon, pertains to the systematic study of the organization of textual material in narrative in terms of its cognitive basis. Chapters 7-10 discuss ways in which interlocutors distribute and recognize knowledge, negotiate meaning, recontextualize linguistic material, and interpret discourse. Hence, Chapter 7 is a study of the relations between legal knowledge and legal discourse, Chapter 8 is concerned with the furnishing of mental

Text, discourse and cognition: An introduction

13

spaces in business talk, Chapter 9 focuses on communicative fragments used for literary purposes, and Chapter 10 deals with a particular conceptual model mediated through discourse in a given socio-cultural context. Adopting experimental methods for a linguistic and cognitive analysis of legal texts, Lita Lundquist investigates expert and non-expert knowledge of two types of legal concepts: 'contract' and 'judgement'. In Chapter 7, entitled "From legal knowledge to legal discourse - and back again", we learn what 'knowledge' is, how specialised knowledge differs from that of laypeople, and more particularly, what the structure of legal knowledge can look like. Through an investigation of three types of 'qualia' - semantic notions used to analyse the data - the knowledge structures of expert and non-expert texts are shown to vary such that experts use a higher number of qualia than non-experts. It also turns out that an expert rater recognizes more qualia than a non-expert one. Furthermore, the type of qualia present in the 'contracts' and 'judgements' produced by experts and non-experts vary in interesting ways, which supports hypotheses concerning the existence of important differences in the knowledge structures available to these two groups of subjects. All through the chapter, Lundquist engages in a critical discussion of the methods used to obtain the results, thus stressing the fundamental problem of studying discourse and cognition through discourse and cognition themselves. This discussion is reminiscent of de Beaugrandes's worries in Chapter 2 concerning the fact that "we do not have any language-independent modality for accessing cognition beyond the limits of what we happen to perceive with our senses", which is also "partly pre-organized by language, so its potential for testing language remains limited in principle." Lundquist's discussion is highly instructive to anyone involved in the study of discourse and/or cognition. Chapter 8 is a study of conditionals in context, entitled "Conditionals: your space or mine?" Adopting a scalar view of the phenomenon, Anne Marie BiilowM0ller here shows that classifications based on decontextualized sentences, which cognitive linguists often rely on to study conditionals, do not hold in practice. This is so because authentic interaction manifests blends of such uses of conditionals and secondly, because the idea of mapping mental spaces together, to form a shared space is too simple for the purposes of actual analysis. This chapter shows that furnishing mental spaces, be they yours, mine, or something in-between, is what the interlocutors engaged in business talk do, and they do so for strategic purposes. The study accounts for the use of conditionals in authentic argumentation by postulating a number of fundamental scales, which can be foregrounded or backgrounded (cf. above) for different communicative purposes. The model thus also allows for interlocutors, given their stake, to mean more than one thing at a time, which is what regularly happens in au-

14

Tuija Virtanen

thentic discourse, where interlocutors are in the process of negotiating meaning, adapting to situations, and constructing contexts through discourse. It is obvious from this chapter that conditionals are used in argumentative discourse in ways which cannot be accounted for without full consideration of the relevant context and the communicative strategies of the interlocutors engaged in the discourse. Focusing on literary discourse, Martina Björklund sets out to explore "communicative fragments and the interpretation of discourse" in Chapter 9. Presenting an analysis of a Russian short story as a showcase, she emphasizes the crucial impact that such communicative fragments - prefabricated patterns, or simply, 'recycled' linguistic material - have on the interpretation of the text. It is evident that the recognition of recontextualized elements, present in discourse in new or modified combinations, and in new and always unique contexts, is a prerequisite if we are to fully appreciate a piece of literary discourse. This has important implications not least for translators of such texts. More generally, Björklund's study contributes to our understanding of the interpretation of discourse, literary and non-literary, by providing us with a tool for the analysis of its many voices, originating in the distributed cognition related to a given sociocultural context. The effect of communicative fragments on the interpretation of discourse constitutes a cognitive concern of a very central kind. While our metalinguistic awareness may allow us to reflect on the intertextual links activated in a given communication situation - in relation to literary or non-literary discourse - the relevant communicative fragments ultimately only pertain to that very context - textual, situational and socio-cultural - thus making us see the discourse we are engaged in as new, as something unique whose construction and interpretation we are participating in, here and now. This separation of our automatic use of language from our metalinguistic awareness may in part explain why so few linguistic models still reckon with a store of prefabricated linguistic expressions in the memory which are put into use in daily communication. As Miller (1984: 156) puts it in an entirely different context, "recurrence is an intersubjective phenomenon, a social occurrence"; "what recurs cannot be a material configuration of objects, events and people, nor can it be a subjective configuration, a 'perception', for these, too, are unique from moment to moment and person to person." Communicative fragments help people in the task of interpreting contexts and relating to them; yet, not enough is known today about the uses of recontextualized prefabricated patterns, lexico-grammatical and textual, which regularly appear in the web of discourses of many kinds. The final chapter of the volume, entitled "Drawing the line: a contested conceptual model in Danish 'child care talk'", is by Peter Harder, who investigates

Text, discourse and cognition: An introduction

15

the negotiation of common ground in a wide socio-cultural context, through an analysis of the conceptual models and norms which we construct in society. He considers the issue in terms of a case study of a particular conceptual model in a Danish context, that of 'drawing the line' in the interaction between adults and the children for whom they are responsible. The chapter analyses the on-going collective discourse through various dimensions of the metaphor activated here, showing that it is not enough to refer to the basic bodily grounding (cf. above) familiar to us from cognitive linguistics. What we also need to consider is the discourse grounding of the metaphorical mapping, i.e. the processes taking place in the interaction between interlocutors seeking to align the metaphor with their own experience and to analyse the mappings of their opponents in ways that result in a polarization of the public debate and hence a battle ground in which even a lack of an acceptable mapping for the current metaphor is one discourse strategy among others. We are here dealing with a mapping from source domain, to target domain the analysis of which can only be partial when conceptualization is seen as a product. Instead, Harder argues that we need to consider conceptualization as an on-going discourse process "where mental structures meet actual experience and there is a struggle to impose some conceptual order on it." It is evident from this analysis that the socio-cultural context in which the metaphor is activated is crucial for the understanding of conceptual models being constructed, altered or maintained through discourse. To sum up, the variety of frameworks represented in this volume highlights methodological issues concerning the study of, on the one hand, text and discourse, and on the other, cognition - and the tacit assumptions that come with the packages. While the methods adopted for study thus vary a great deal across the chapters, what is shared is an awareness of the fact that discourse and cognition can only be studied with the help of discourse and cognition. This unavoidable state of affairs has important implications for the discourse of discourse: scholarly discourse, too, serves to create contexts. At the same time, what emerges from the different chapters is the uniqueness of discourse as a site for the study of both individual and distributed cognition and their interplay in authentic situational and socio-cultural contexts. All the chapters also show that both discourse linguistics and cognitive linguistics profit from interdisciplinary discussions, which have greatly contributed to bringing these areas of study closer to one another in the recent past. It is thus hoped that in addition to the discoveries and insights of the discourse-based approaches to cognition included in this volume, the present selection of studies also serves to inspire linguists of various orientations to go on building bridges across academic boundaries.

16

Tuija Virtanen

References de Beaugrande, Robert 1980 Text, Discourse, and Process: Toward a Multidisciplinary Science of Texts. (Advances in Discourse Processes 4.) Norwood, N.J.: Ablex. de Beaugrande, Robert and Wolfgang Dressier 1981 Introduction to Text Linguistics. London/New York: Longman. Brown, Gillian and George Yule 1983 Discourse Analysis. Cambridge: Cambridge University Press. van Dijk, Teun A. and Walter Kintsch 1983 Strategies of Discourse Comprehension. New York: Academic Press. Edwards, Derek 1997 Discourse and Cognition. London/Thousand Oaks/New Delhi: Sage. Enkvist, Nils Erik 1989 Connexity, interpretability, universes of discourse, and text worlds. In: Sture Allen (ed.), Possible Worlds in Humanities, Arts and Sciences: Proceedings of Nobel Symposium 65, 162-186. (Research in Text Theory 14.) Berlin/New York: Walter de Gruyter. Geeraerts, Dirk 1995 Cognitive linguistics. In: Jef Verschueren, Jan-Ola Östman and Jan Blommaert (eds.), Handbook of Pragmatics: Manual, 111-116. Amsterdam / Philadelphia: John Benjamins, van Hoek, Karen, Andrej A. Kibrik and Leo Noordman (eds.) 1999 Discourse Studies in Cognitive Linguistics: Selected Papers from the 5th International Cognitive Linguistics Conference, Amsterdam, 1997, 91—110. Amsterdam / Philadephia: John Benjamins. Koenig, Jean-Pierre (ed.) 1998 Discourse and Cognition: Bridging the Gap. Stanford, California: Center for the Study of Language and Information. Linell, Per 1998 Approaching Dialogue: Talk, Interaction and Contexts in Dialogical Perspectives. (Impact: Studies in Language and Society 3.) Amsterdam / Philadelphia: John Benjamins. Lundquist, Lita and Robert J. Jarvella (eds.) 2000 Language, Text, and Knowledge: Mental Models of Expert Communication. (Text, Translation, Computational Processing 2.) Berlin / New York: Mouton de Gruyter. Miller, Carolyn R. 1984 Genre as social action. Quarterly Journal of Speech 70: 151-167. Sanders, Ted, Joost Schilperoord and Wilbert Spooren (eds.) 2001 Text Representation: Linguistic and Psycholinguistic Aspects. (Human Cognitive Processing 8.) Amsterdam / Philadelphia: John Benjamins.

Chapter 2 Language, discourse, and cognition: Retrospects and prospects1 Robert de Beaugrande

1.

Language and cognition as a (non)topic in linguistics

The status of cognition as a concept in the science of linguistics remains uncertain. I recently re-scanned my survey of major linguistic discourses (in Beaugrande 1991) and I could not find the term 'cognition' in the writings of such avowed mentalists as Saussure and Sapir, nor (less surprisingly) in the writings of such avowed behaviourists as Bloomfield and Firth. The first occurrence of the term 'cognition', or 'cognitive faculties' to be exact, was found in the writings of Noam Chomsky, which at the time signified a programmatic renewal of mentalism, as we shall see. The long absence of the term 'cognition' in linguistics seems interesting in view of the well-established term 'cognitive linguistics', and merits some scrutiny. For the early mentalists, the contents of cognition, variously called 'knowledge', 'ideas', 'thoughts' and so on, seemed under-determined. For Saussure (1966 [1916]: 111), 'thought' was Only a shapeless and indistinct mass', an 'indefinite plane of jumbled ideas'. For Hjelmslev (1969 [1943]: 52), 'purport' [Danish 'mening'] was 'an unanalysed entity', an 'amorphous thoughtmass'. These linguists surmised that language imposes determinacy: 'thought' 'becomes ordered in the process of its decomposition' as 'language takes shape' and enables 'clear-cut consistent distinctions between two ideas' (Saussure); 'linguistic form' 'lays arbitrary boundaries on a purport-continuum' that 'depends exclusively on this structure' (Hjelmslev). So mentalist linguistics might assume that the study of cognition would be left with nothing. For the behaviourists, in contrast, the contents of cognition seemed overdetermined, thanks to the notion that 'mental images, feelings, thoughts, concepts', and 'ideas' 'are merely popular names for various bodily movements' (Bloomfield 1933: 142). For Bloomfield, 'the meaning of a linguistic form' is not a 'mental event', but 'the situation in which the speaker utters it and the response which it calls forth in the hearer'; then he wondered if 'the situations which prompt people to utter speech include every object and happening in their universe' (1933: 142, 139f). If so, 'the study of speakers' situations and

18

Robert de Beaugrande

hearers' responses' might be 'equivalent to the sum total of all human knowledge' (1933: 74). And 'to give a scientifically accurate definition of meaning for every form in the language' would require 'scientifically accurate knowledge of everything in the speaker's world'. So behaviourist linguistics might assume that the study of cognition would be left with everything. At all events, these mentalists and behaviourists did leave cognition out of their accounts, and concentrated on the units of sound (phonemes) and the units of form (morphemes), which plainly constitute orderly systems of 'distinctive features'. And when linguistics and semantics eventually embarked on the study of meaning, they predictably reapplied a conception developed in phonology and morphology, namely the system of basic units. Parallel to the 'phonemes' and 'morphemes', these units were variously termed 'semes', 'sememes', 'semantemes', 'semantic features', or 'semantic primitives' (e.g. Hjelmslev 1957; Katz and Fodor 1963; Greimas 1966). However, 'phonemes' can be securely defined by articulation, e.g., a 'voiced dental stop'; and 'morphemes' can be defined, though less securely, by the processes of word-construction, e.g., for the Tenses or Aspects of Verbs. What could be the basis for defining the units in semantics? Several possibilities were aired, such as: (a) the 'linguistic image of properties, relations, and objects in the real world' (Albrecht 1967: 179), as in Pottier's (1963) taxonomy of chairs; (b) the distinctive elements arising from the 'apperceptive constitution' of 'human beings in regard to their environment' (Bierwisch 1966: 98), a wellknown example being colours (cf. Heider 1972); (c) as conceptual elements into which a 'reading' decomposes a 'sense' (Katz 1966); (d) as elements for constructing a semantic theory (Katz and Fodor 1963); (e) as constituents of a metalanguage for discussing meaning (Greimas 1966). Either the units are given in advance by the 'real world' or the 'human environment'; or they are generated by the processes of human comprehension; or again they are purely theoretical constructs of linguistics and semantics. In all these interpretations, the units are presumably cognitive rather than strictly linguistic; they cannot be simply a set of words used for writing definitions in the manner of a dictionary. Moreover, the catalogue of units would have to be universally applicable for all the meanings of a language, or even the meanings of all languages. And the units would constitute a system analogous to 'phonemes' and 'morphemes' only if a given unit is applied repeatedly and consistently to many meanings. In the most strictly orthodox view, the units would determine the 'sense properties' whose 'sum' is the 'sense of any expression'- 'its

Language, discourse, and cognition: Retrospects and prospects

19

indispensable hard core of meaning', 'deliberately excluding any influence of context or situation of utterance' (Hurford and Heasley 1983: 91). An immediate problem was how to label these cognitive units of meaning. If we use words, then the words can have meanings of their own beyond the basic units they are supposed to label. Let us consider a seeming straightforward example I recall hearing discussed at the time: constructing the meaning of'kill' from CAUSE + DIE (Chomsky 1971 [1968]: 188f). In authentic discourse, this construction works for some instances like (1), and breaks down for others, like (2). (1) 'It would always be a heavy thought to me, if I 'd caused his death, even in a just cause.' 'Yes', said Phineas, 'killing is an ugly operation...' (Uncle Tom 's Cabin} (2) Is'(I that chase theefrom the country and expose Those tender limbs of thine to the event Of the none-sparing war? [...] Whoever shoots at him, I set him there. Whoever charges on his forward breast I am the caitiff that do hold him to Ϊ. And though I kill him not. I am the cause His death was so effected (All's Well that Ends Welll\\,2, 105-108, 115-119) The problem here lies not in the correct choice of labels for 'semantic features' or 'sense properties', but in the orthodox aspiration to describe an 'indispensable hard core of meaning' free of all 'context'. Whilst these discussions were in progress, the so-called 'cognitive revolution' set about transforming the human sciences from the 1960s onwards. The most 'revolutionary' idea, in my view, was that cognition is a unifying central faculty encompassing a wide range of human capabilities and processes that had hitherto been studied in isolation and with narrow methods - intelligence, learning, memory, skills, problem-solving, and, ultimately, language and meaning (cf. overview in Newell 1990). The response of linguistics was a return not merely to the mentalism of its early years (e.g. Saussure, Hjelmslev) but to the idealist and rationalist philosophy of 17th and 18th centuries (e.g. Descartes, Leibnitz). Early idealism held the contents of cognition to be determinate, but proposed to explain this factor at a single stroke by invoking a 'universal, language-independent' 'system of possible concepts' (Chomsky 1965: 160). The order of meanings in language would be pre-determined by the construction of the human mind; each language

20

Robert de Beaugrande

makes its own selection out of the universal order. For philosophers like Descartes, the ultimate source and creator of this order would be the mind of God. In modern linguistics, God was dispensed with by linking cognition directly to biology (as if to outflank behaviourism): 'there is a highly determinate, very definite structure of concepts and of meaning that is intrinsic to our nature, and as we acquire language or other cognitive systems these things just kind of grow in our minds, the same way we grow arms and legs' (Chomsky 1991: 66). This renewed mentalism in effect set out to study cognition instead of language, aspiring to construct 'a theory of linguistic structure' without 'reference to particular languages' (Chomsky 1957: 11). Eventually, real language - now called 'externalised language' (or 'e-language' for short) - was declared to be a mere 'epiphenomenon' (Chomsky 1986: 25). In these 'radically different theories', 'there are no constructions; there are no rules' (Chomsky 1991: 81). This version of cognitive linguistics is concerned solely with 'the structure of mental representations' and expressly excludes 'the relationship between' 'mental representations' 'and things in the world' (Chomsky 1991: 93). The exclusion carries ironic (and probably unintentional) echoes of Bloomfield's much earlier exclusion, quoted above, upon confronting 'the sum total of all human knowledge'. The wheel seems to have turned a long way. Cognition has passed from being first an 'indistinct mass', then a stupendous catalogue of 'every object and happening in the universe', then a determinate system of 'hard-core meanings', and finally a 'universal system of possible concepts'. As far as I can tell, none of these conceptions adequately defines the relation between cognition and language; and I shall suggest an account that is markedly different in theory (what cognition and meaning are) and in practice (how we go about describing them through data from language and discourse).

2.

Language and discourse as theory and practice

Applying our terms broadly, we can define a language as a theory of human knowledge and experience (what speakers know how to talk about), and discourse as its practice (what speakers do talk about). Language and discourse would thus interact in a dialectical cycle, with each side informing and guiding the other, as modelled in Figure 1. Language specifies the standing constraints (e.g., what the words usually mean), whereas discourse manifests emergent constraints (e.g., what the words mean in this particular stretch of discourse). The practices are heavily 'theory-driven' in the sense that discourse compels its participants to 'theorise' about what words mean, what makes sense, what

Language, discourse, and cognition: Retrospects and prospects

21

specify standing constraints theory-driven LANGUAGE

DISCOURSE practice-driven

manifest emergent constraints A DIALECTICAL MODEL

Figure 1.

people intend, and so on. Yet the theory is also heavily 'practice-driven' in the sense that discourse continually tests and adjusts implicit theories about meanings, senses, or intentions. In this account, discourse is probably the most theoretical practice humans can perform, particularly in respect to meaning. In return, language is the most practical theory humans can devise, offering us the resources to shape and guide almost any of our practical activities. Yet the 'theoreticalness' of language is extremely well-hidden from most speakers who practice it. They would regard discourse as a 'practical matter'; they would be surprised if we told them they possess a 'theory of their language' that makes them 'theoreticians'. All human practices produce knowledge, which is stored as the contents of cognition. Language must have originated and evolved as a refined means for delimiting, organising and stabilising those contents into meanings, and for sharing the meanings among the community. The content of cognition and the content or meaning of language are thus phenomena of the same order, differing in degree more than in kind. By talking about what we know and sharing it with other people, we also build or change our knowledge in ways that could hardly be done without the use of language. Cognition and language would thus also interact in a dialectical cycle, with each side informing and guiding the other, as modelled in Figure 2. Most essentially, cognition generates meanings, whereas language determines meanings. But here also, the difference is only in degree, and these dialectical processes are inseparable. A helpful analogy here might be cognition as the landscape and language as the map (compare the 'cognitive maps' in Neisser 1976). We could live in the landscape and move about after a fashion with no map, but language makes our lives and our movements much more organised and productive. Similarly, each of us can have a 'mental map' of the landscape, but without the language-

22

Robert de Beaugrande generate meanings

COGNITION

LANGUAGE

determine meanings A DIALECTICAL MODEL

Figure 2.

map, we could not compare and improve the goodness of fit. The meanings of language are in turn not just names of points on the map, but directions about how to navigate with the aid of the map. And discourse is the activity of actually walking through the landscape by means of the map. How accurate the map might be in any literal cartographic sense could never be established, because both the landscape and the map are continually evolving. Indeed, just using the map can change the landscape or the map, or both. We can only try to notice how helpful the map proves to be in practice, and hope that we will get where we want to go and not be led too far astray. Nor again could we establish which language is a better map of human cognition; or whether cognition is the same for all humans whilst only our maps differ from one language to another. Such questions are unresolvable because we do not have any language-independent modality for accessing cognition beyond the limits of what we happen to perceive with our senses. And even what we see, hear, and so on is partly pre-organised by language, so its potential for testing language remains limited in principle. The analogy to the map is inappropriate in that a map is deliberately drawn by an explorer or cartographer, whereas language evolves both spontaneously and communally. Somehow, the contents of cognition get organised as meanings of language with remarkable precision and delicacy even though speakers are not following an explicit plan or design. The organisation is 'delivered for free', so to speak, and one might empathise with the 17th century idealists who imagined it all to be the work of God. Perhaps a different analogy than a map might be helpful here. The termite hill on the African veldt is an impressively complex structure built by an immense number of individual events and agents, each of these eminently simple in itself. A single termite certainly possesses no knowledge of the total design

Language, discourse, and cognition: Retrospects and prospects

23

of the termite hill; such knowledge could not be justly attributed even to the entire termite colony. Instead, the design is generated by a dynamic convergence among millions of movements of termites moving tiny grains of sand or earth. An intriguing fact about the architecture of termite hills was demonstrated in computer simulation by Peter Kugler when I was at the Institute for Medical Engineering of the University of California, Los Angeles. He was able to show that the main supports of the structure correspond to the points where the trajectories of the several termites most frequently intersect. The structure thus emerged from the crossing of pathways - like Order out of chaos' (Prigogine and Stengers 1984). In an analogous manner, the complex architecture of cognition could arise from the intersections among the trajectories in networks of knowledge representing the relations among classes of perceptions, sensations, and so on, each one of which is simple in itself. What something 'means' or 'represents' is a function of what it is connected to - the pathways leading toward or away from it in the network. When some item of knowledge gets put into practice, a point in the network, or more likely a cluster of points, gets 'activated'; and this activation automatically spreads out along the contiguous pathways without the intervention of executive control (cf. Collins and Loftus 1975). The relevant pathways retain and increase their activation, whereas the irrelevant pathways lose their activation in about half a second - a mode of processing cognitive psychology has been able to demonstrate under the label of the 'constructionintegration model' (Kintsch 1988, 1989). A strangely counter-intuitive consequence is that all meanings of a word are initially activated, and not just the one relevant to the context, as 'relevance theory' complacently asserts without regard for the findings of experimental psychology. The relevant one is the one whose activation level is sustained or raised. What is conventionally called a 'meaning', 'concept', 'thought', 'idea' and the like is not a standing unit like an entry in a dictionary or a message in a pigeonhole, but an activation pattern. Each instance of activation may strengthen its links, but may also adjust or alter them, depending on other areas of parallel activation, which together constitute the 'context'. In this account, the contents of cognition and the meaning of language are products of massive parallel distributed processing (Rumelhart, McClelland et al. 1986). They are not so much 'stored' as 'prefigured' in the cognitive architecture of human knowledge networks, which would be the psychological and neurological instantiation of language as 'theory'; their activations would be the instantiations of discourse as 'practice'. So language would have naturally emerged as a 'user-friendly front end' of cognition, which itself possesses Ian-

24

Robert de Beaugrande

guage-like qualities. Language is a product and process of 'higher-order consciousness' that reconciles the individual and the social factors of cognition (cf. Edelman 1992). Now, this account could explain the failure of projects in linguistics and semantics for describing meaning in basic units that constitute the 'hard core of meaning independent of context or situation'. Meanings are not units at all, but events in a dialectical process which always has a context as its cognitive architecture. If semanticists purport to be analysing as meaning out of context, they are in fact just replacing natural contexts with artificial ones, I cannot see why the results could claim to be generally valid or representative. I would go on to point out, as a logical corollary, that meanings will not hold still or remain constant while we 'analyse' them. On the contrary, analysis necessarily entails an increased or specialised activation, the meaning will keep on spreading and thus gaining in richness and complexity. We might draw here upon the tradition of 'gestalt psychology' (e.g. Koffka 1935) by modelling cognition as the ground, and a language as the figure. In a typical gestalt, the figure is generated by receiving the central focus of attention or consciousness, whereas the ground receives only peripheral attention. This model would be recursive in the sense that the meaning of a discourse is also the figure set off by focused attention against the ground the meaning in the language, which also receives peripheral attention. The participants in the discourse understand the current meanings with the aid of the usual meanings being, in the commonplace metaphor, in the 'back of the mind'. This model too indicates the impossibility of describing meaning by 'indispensable hard-core senses'; a meaning does not have such a core. Speakers know what words mean, and how those meanings can be shared, through the convergence of partially similar events. The meaning of a given word evolves in a dialectical cycle between what it might mean by itself and what other sorts of words it is likely to appear with in collocations, i.e., typical combinations of lexical choices (Sinclair 1991; Hunston and Francis 1999). In cognitive terms, the meaning takes on its characteristic content by virtue of multiple activations in networks with other meanings. A grand challenge for cognitive linguistics, I would suggest, is to analyse meanings by examining large sets of such events as provided by very large corpora of authentic text and discourse (Beaugrande 2000). Let us return to examples of 'kill' and 'die'. Using my general corpus of 15 million words of British and American writers on the WordPilot programme (Milton 1999), I found 442 occurrences of 'killed'. In the large majority, the thing 'killed' was indeed a person or an animal (especially a snake) being caused to die. But this was not an 'indispensable core'. One can histrionically speak of being 'killed'

Language, discourse, and cognition: Retrospects and prospects

25

by sadness (3) or discomfort (4). Or, one can use the term to mean 'put a sudden and complete end to' something that is neither a person nor an animal (5-6). (3) How should I ever have borne it! It would have killed me never to come to Hartfield any more! (Emma) (4) The heat was excessive; he had never suffered any thing like it—almost wished he had staid at home—nothing killed him like heat (Emma) (5) I'm in love with change and I've killed my conscience (This Side of Paradise) (6) you have killed my love. You used to stir my imagination. Now you don't even stir my curiosity. You simply produce no effect. (Dorian Gray) My corpus of Shakespeare plays returned some highly creative meanings, such as cease an emotion (7), cancel one vow with another (8), outshine a light (9), close one's eyes (10), or torment a person with misplaced kindnesses (11). This last has become a standing collocation in English. (7) How will she love when the rich golden shaft Hath kill 'd the flock of all affections else That live in her

(Twelfth Night)

(8) You do advance your cunning more and more. When truth kills truth, Ο devilish-holy fray! (Midsummer Night s Dream) (9) Arise, fair sun, and kill the envious moon, Who is already sick and pale with grief That thou her maid art far more fair than she. (Romeo and Juliet) (10) To bed, to bed! Sleep kill those pretty eyes, And give as soft attachment to thy senses As infants 'empty of all thought! (Troilus and Cressida) (11) This is a way to kill a wife with kindness, And thus ΙΊΙ curb her mad and headstrong humour (Taming of the Shrew)

26

Robert de Beaugrande

Among the 789 occurrences of 'died' in my general literature corpus were a fair number of things that could not be sensibly said to get 'killed', e.g. (1215). In the English of literature at least, certain kinds of things 'die' or 'die away' or 'die out' when their continuation would have been desirable, such as sounds, including speech (12), light (13), and warmth, including courage (13) and emotion (14). I found no data for the opposite expressions, such as 'silence', 'darkness', or 'coldness' being said to 'die'. These are already absences or terminations, and their continuation would be undesirable. The idea of a 'mountain dying' (15) stands out as the least predictable collocation; Muir wanted to portray how Mt. Hood 'varies greatly in impressiveness and apparent height at different times'. (12) He opened his lips to speak, but his accents died away ere they -were formed. (Wieland) (13) the illumination died away in a few seconds, and my heart died away within me as it went. (Arthur Gordon Pym) (14) His face was ruggedly formed, but it looked like ashes—like something from which all the warmth and light had died out. (My Antonio) (15) Next year you return to the same point of view, perhaps to find that the glory has departed, as if the mountain had died (Steep Trails) I was surprised to find from my Shakespeare corpus that only people were said to have 'died'. No other uses of the term occurred. It is customary in studies of language or literature to distinguish between literal meaning and metaphoric meaning. The implication is that a listener or reader first supplies the literal meaning and, noticing this does not fit, casts about for a suitable metaphoric meaning. But the 'spreading activation' or 'construction-integration' models of cognitive processing suggest that the suitable meaning could be produced without such a detour through the literal meanings. Certainly such could be expected if the meaning is familiar, e.g. 'sound died away' covering a range of other things that 'died away' in my corpus data, such as 'voice', 'whisper', 'footsteps', 'trampling', 'music', 'rhythm', and 'cadence'.

Language, discourse, and cognition: Retrospects and prospects

3.

27

Cognitive text linguistics?

Whereas 'cognitive linguistics' is a well-known field with an international association, a journal, and some 8211 attestations on the Internet (returned by Alta Vista in August 2000), 'cognitive text linguistics' is an unknown field and does not appear on the Internet even once. The absence is due more to academic convention than to substantive divergence. In my own work at least (e.g. Beaugrande 1980, 1997), text linguistics has always had a resolutely cognitive orientation because the text must be described as both product and process. Perhaps cognitive linguistics originally shared a border with psycholinguistics, which routinely chose the sentence as the main unit of language. But no comparable restrictions persist today. In the purview of text linguistics, the collocations we can identify in very large corpus data represent a special phenomenon: they are more specific than the language, yet more general than the text, like some category of 'missing links' in between (Beaugrande 2000). In terms of cognition, the account I have proposed would explain collocations as interactive priming loops among expressions whose probability of being chosen together is significantly higher than the normal base state prior to activation. The links among, say, 'sleep - close - eyes' would be closer to activation, whereas no such linkage would be primed among 'sleep - kill - eyes'. Looking at multiple occurrences of an expression in corpus data is a special means of distributing our attention between figure and ground. Instead of just determining the meaning of an expression in a single discourse, we are seeking to assess the similarities and differences among its meanings in multiple discourses. Some of those meanings will match our intuitions, e.g., how 'voices died away', whilst others may not match, e.g., how a 'mountain died'. Usually the match is a matter of degrees rather than a yes-or-no division. Still, our knowledge of collocations is part of the cognitive and linguistic ground for the meanings that figure in any single text. Quite plausibly, the greater challenge for a 'cognitive text linguistics' would be to account not just for the figures in the focus of processing, i.e., the expressed meanings of a text, but for the ground at the peripheries of processing, i.e., the (possibly) implied meanings. If activation works itself out through the cognitive architecture of knowledge and meaning without executive control, then determining its full breath or scope may prove difficult indeed. Consider this brief report about someone being 'killed' in a road accident, taken from the US news media (Cullingford 1978):

28

Robert de Beaugrande

(16) A New Jersey man was killed Friday evening when a car swerved off Route 69 and struck a tree. David Hall, 27, was pronounced dead at Milford Hospital. The driver, Frank Miller, was treated and released. No charges were filed, according to investigating officer Robert Onofrio. Our cognition specifies not merely how road accidents typically occur, but also what is typically reported about them. In this text type, we expect to learn the names and ages of the people involved and the time and place of the accident, even if such knowledge is totally irrelevant to our own personal situation. We do not expect to hear that the driver's nice new suit got all messed up; or that one passenger broke a fingernail, and another missed an appointment with her hairdresser - even if all this actually happened. Such knowledge entails the wrong focus and degree of detail for the text type. In exchange, we seem to know all manner of things implied but not stated. We know that the hapless Mr. Hall was 'killed' by the impact of the car against the tree, rather than being bumped off just then by the New Jersey Mafia. Miller lost control rather than deliberately heading for the tree with the plan of knocking it down and carting it home for firewood. The 'tree' was a large sturdy outdoor tree rather than a dwarf bonsai on a window sill or a Christmas tree in a shopping mall. What was left of Hall got taken to 'hospital' by an ambulance and not by canoe, camel, or skateboard. Miller was 'treated' by dressing his injuries and not by giving him a fancy-dress dinner. The 'charges' would be documents for criminal proceedings and not swift advances of mounted cavalry. These excluded alternatives may sound rather zany or scurrilous, but how and why we know they are excluded is by no means a trivial question. The psychological evidence I have cited for the 'construction-integration model' indicates that large amounts of irrelevant knowledge are indeed initially activated, such as these alternate meanings of'tree', 'treat', and 'charge'. Now contrast this passage from a news item from the Ipswich Journal (Jan. 12, 1878): (17) The solicitor was going to the Court, when he staggered as if in a fit, and fell against the wall. The watchman and a policeman, running to his assistance, took him into a room. Some brandy was administered to no effect, and Mr. Bond, the surgeon of Parliament Street, arriving, he pronounced him dead. From a purely linguistic standpoint, the Pronouns 'he' and 'him' could be ambiguous about who pronounced whom dead. But from a cognitive standpoint, no doubt can arise. The actions of 'staggering' and 'falling' (to say nothing of not wanting good brandy) are nicely apropos prior to becoming 'dead'; and

Language, discourse, and cognition: Retrospects and prospects

29

a 'surgeon' is just the person who has the authority to make such 'pronouncements'. Notice that we are not actually told that the solicitor has died. But just to introduce a modest margin of doubt would require some macabre ingenuity, e.g.: (17 a) A large draft of brandy was administered to the most amazing effect, for the solicitor became so intoxicated and confused that when Mr. Bond, the surgeon of Parliament Street, arrived, he pronounced him dead. The apparent ordinariness and simple-mindedness of texts like (16) and (17) must in reality be the effects of complex cognitive and linguistic processes we will eventually have to account for. At present, the most promising account might be to build the collocations emerging from very large corpus data into a 'spreading activation' or 'construction-integration' model. But thorny problems regarding the choice of data and the methods of interpreting them must first be resolved (cf. Virtanen 1990; Beaugrande 2000). In particular, we must show how cognition adjusts the strengths of collocations to fine-tuned distinctions among text types, registers, or styles, the most significant example in English being the works of Shakespeare. Whether we elect to call such explorations by the term of 'cognitive text linguistics' is a purely academic matter. What counts will be the vast dimensions and richness of the terrain, for which at the moment we are still struggling to attain some cognitive map.

Note This paper is a fully revised version of a video lecture recorded in Botswana and presented at the Conference of the International Association of Cognitive Linguistics in Stockholm in July 1999.1 am deeply indebted to Ibolya Maricic of Växjö University for the transcription.

References Albrecht, E. 1967 Sprache und Erkenntnis. Berlin: Deutscher Verlag der Wissenschaften. Beaugrande, Robert de 1980 Text, Discourse, and Process. Norwood, NJ: Ablex. 1991 Linguistic Theory: The Discourse of Fundamental Works. London: Longman.

30

Robert de Beaugrande 1997

New Foundations for a Science of Text and Discourse. Stamford, CT: Ablex. 2000 Text linguistics at the millennium: Corpus data and missing links. Text 20/2: 153-195. Bierwisch, Manfred 1966 Strukturalismus: Geschichte, Probleme, Methoden. Kursbuch 5, 77-152. Bloomfield, Leonard 1933 Language. Chicago: University of Chicago Press. Chomsky, Noam 1957 Syntactic Structures. The Hague: Mouton. 1965 Aspects of the Theory of Syntax. Cambridge: MIT Press. 1971 [original 1968] Deep structure, surface structure, and semantic interpretation. In: Danny Steinberg and Leon Jakobovits (eds.), Semantics: An Interdisciplinary Reader in Philosophy, Linguistics, and Psychology, 183-216. Cambridge: Cambridge University Press. 1986 Knowledge of Language. New York: Praeger. 1991 Language, politics, and composition (with Gary Olsen and Lester Faigley). In: G. Olsen and I. Gales (eds.) Interviews: Cross-Disciplinary Perspectives on Rhetoric and Literacy, 61-95. Carbondale: Southern Illinois University Press. Collins, Allan and Elizabeth Loftus 1975 A spreading-activation theory of semantic processing. Psychological Review 82: 407^28. Cullingford, Richard 1978 Script Application. New Haven: Yale University dissertation. Edelman, Gerald M. 1992 Bright Air, Brilliant Fire: On the Matter of the Mind. New York: Basic Books. Greimas, Algirdas Julius 1966 Semantique structurale: Recherches de methode. Paris: Larousse. Heider, Eleanor 1972 Universals in color naming and memory. Journal of Experimental Psychology 93: 10-20. Hjelmslev, Louis 1969 [original 1943] Prolegomena to a Theory of Language. Madison: University of Wisconsin Press. 1957 Semantique structurale. In Essais linguistiques, 96-112. Copenhagen: Sprog- og Kulturvorlag, 1970. Hunston Susan and Gill Francis 1999 Pattern Grammar. Amsterdam: Benjamins. Hurford, James R. and Brendan Heasley 1983 Semantics. Cambridge: Cambridge University Press. Katz, Jerrold 1966 The Philosophy of Language. New York: Harper & Row.

Language, discourse, and cognition: Retrospects and prospects

31

Katz, Jerrold and Jerry Fodor 1963 The structure of a semantic theory. Language 39: 170-210. Kintsch, Walter 1988 The role of knowledge in discourse comprehension: A 'construction-integration model'. Psychological Review 95/2: 163-82. 1989 The representation of knowledge and the use of knowledge in discourse comprehension. In: Rainer Dietrich and Carl Graumann (eds.), Language Processing in Social Context, 185-209. Amsterdam: North Holland. Koffka, Kurt 1935 Principles of Gestalt Psychology. New York: Harcourt, Brace & Co. Milton, John 1999 Lexical thickets and electronic gateways: making text accessible by novice writers. In: Christopher N. Candlin and Ken Hyland (eds.), Writing: Texts, Processes and Practices, 221-243. London: Longman. Neisser, Ulric 1976 Cognition and Reality. San Francisco: Freeman. Newell, Allan 1990 Unified Theories of Cognition. Cambridge, MA: Harvard University Press. Pottier, Bernard 1963 Recherches sur I'analyse semanlique en linguistique et en traduction mechanique. Nancy: University of Nancy Press. Prigogine, Ilya and Isabelle Stengers 1984 Order out of Chaos: Man s Dialogue with Nature. London: Heinemann. Rumelhart, David, James McClelland et al. 1986 Parallel Distributed Processing: Explorations in the Microstructures of Cognition (2 vols.). Cambridge, MA: MIT Press. Saussure, Ferdinand de 1966 [original 1916] Course in General Linguistics (transl. Wade Baskin). New York: McGraw-Hill. Sinclair, John McHardy 1991 Corpus, Concordance, Collocation. Oxford: Oxford University Press. Virtanen, Tuija 1990 Problems connected with the choice of data and the use of standardised multipurpose corpora in linguistics. In: Keith Battarbee and Risto Hiltunen (eds.), Alarums and Excursions, 163-173. Turku: University of Turku Press.

Chapter 3 On the discourse basis of person agreement Anna Siewierska

It is generally acknowledged (see e.g. Givon 1976; Moravcsik 1978; Lehmann 1982; Corbett 1991; Anderson 1992) that subject person agreement markers typically originate from independent person pronouns by phonological reduction, cliticization and affixation, as shown in (1). (1) indep Pro > unstressed Pro > clitic > affix It is also generally accepted that the above development is motivated by discourse factors. More controversial is the nature of the discourse factors involved. The hitherto most widely accepted explanation for why and how this process comes about is the so-called NP-detachment account primarily associated with the name of Givon (1976). An alternative explanation of the rise of subject person agreement markers has been recently elaborated by Ariel (2000) based on her (Ariel 1990) version of accessibility theory (AT). Ariel contends that while both the NP-detachment and AT lines of development from pronoun to agreement marker are plausible, her AT analysis carries the advantage of simultaneously accounting for the dominant form of person agreement paradigms, which she takes to be the overt realization of first and second person markers and zero realization of third person forms. Under the NP-detachment scenario, this person asymmetry must be attributed to subsequent loss or reanalysis of the third person forms.1 Under the AT analysis, on the other hand, third person forms simply tend not to arise at all. Since, accordingly to Ariel, lack of third person agreement markers is the cross-linguistic norm, she sees the AT line of development of subject agreement as the cross-linguistically favoured one. The present paper takes issue with Ariel's claim under her current assumptions with respect to the rise of third person agreement. It is argued that unless third person markers can be accommodated within the AT developmental scenario, AT cannot be seen as the major driving force behind the emergence of person agreement. The discussion begins with a brief overview of the main differences between the NP-detachment and AT accounts of the rise of person agreement. The cross-linguistic implications of one of these differences, the grammatical status of the person forms, are examined in section 2. The person asymmetries which lie at the heart of Ariel's AT account of the rise of agreement

34

Anna Siewierska

are investigated in section 3. Cross-linguistic data is presented which reveal that the person asymmetries in question constitute the minority rather than the majority pattern. Section 4 considers a way of accommodating the emergence of third person agreement markers with AT and provides additional reasons for doing so. Section 5 concludes the discussion.

1.

Two accounts of the rise of person agreement

According to the NP-detachment analysis, person agreement markers originate as anaphoric pronouns in topic-shifted, left- or right-detached constructions such as those in (2). (2) a. b.

Sally, she came early. She came early, Sally.

The claim is that as a result of over-use such topic-shifted constructions become reanalysed as neutral clauses and the anaphoric pronouns, formerly functioning as topic agreement markers, become subject agreement markers. This reanalysis is seen to be typically accompanied by the cliticization and subsequent affixation of the anaphoric pronoun to the verb, producing structures such as those in (3). (3)

Sally she-came early.

The pronominal forms commonly continue to perform a referential function which over time may be lost, resulting in forms that only redundantly express person and number and/or gender.2 Such forms may undergo phonological erosion and subsequently be lost altogether. Under the AT analysis the development from pronoun to subject agreement marker takes place not in topic-shifted constructions but rather in simple clauses with a pronominal subject such as those in (4). (4) a. b.

I arrived late. You won.

The bounding of such subject pronouns to the verb is seen to be a reflection of the attenuated encoding of highly accessible discourse referents. In AT the coding of discourse referents is seen to reflect speakers' assumptions as to the degree of accessibility of the referents in the memory store of the addressee. The more accessible the referent the less coding required. Accessibility is taken to be a function of several factors, the most relevant of which in the context of this discussion is entity saliency, where other things being equal

On the discourse basis of person agreement

35

mental entities to the left of > in (5) are seen to be more salient than those on the right of >. (5) a. b. c. d. e. f. g.

Speaker > addressee > non-participant (3rd person) High physical salience > low physical salience Topic > nontopic Grammatical subject > nonsubject Human > animate > inanimate Repeated reference > few previous references > first mention No intervening/competing referents > many intervening/competing referents

The higher the accessibility of the referent, the more attenuated its encoding, as shown in (6), taken from Ariel (2000: 205), where accessibility decreases from left to right. (6) The accessibility marking scale zero < reflexives < poor agreement markers < rich agreement markers < reduced/cliticized pronouns < unstressed pronouns < stressed pronouns < stressed pronoun plus gesture < proximal demonstrative (+NP) < distal demonstrative (+NP) < proximal demonstrative (+NP) + modifier < distal demonstrative (+NP) + modifier < first name < last name < short definite description < long definite description < full name < full name + modifier. As reflected in (6), agreement markers are considered to be high accessibility coding devices, second only to zero and reflexives. Thus, according to AT, they should be used as the sole means of referent encoding, only in the case of highly accessible referents. How highly accessible a referent needs to be in order to warrant encoding solely by an agreement marker remains an open question. It should be clear from the above that the NP-detachment and AT analyses of the rise of subject person agreement are in fact in two important ways complementary. First of all, they differ with respect to the grammatical status of the person markers that they consider. The NP-detachment analysis seeks to provide an account of the emergence of person markers co-occurring with corresponding nominal arguments, i.e. prototypical agreement markers. Ariel's AT analysis, by contrast, deals with person markers functioning as bound pronouns. And secondly, the two analyses make different predictions in regard to the realization of the markers of the three persons. We see that the NP-detachment analysis in (2) is illustrated on the basis of the third person, while the AT analysis in (4) on the basis of the first and second person. This is not co-incidental. Since left- and right-dislocated topics are typically nominal constituents, under

36

Anna Slewierska

the NP-detachment analysis subject person agreement is generally assumed to originate in the third person and then to spread to the other two persons. Ariel's AT analysis, on the other hand, involves primarily the first and second person. As captured in (5a), in AT the speaker and hearer are considered to be inherently more accessible than third parties. Consequently, first and second person pronouns are prime candidates for undergoing phonological reduction, cliticization and affixation. Third person pronouns by contrast are not as their referents are not consistently highly accessible enough to warrant encoding by high accessibility markers. Thus "third persons are predicted to not often be coded by bound pronominal forms" (Ariel 2000: 211). The above considerations suggest that the NP-detachment vs AT origin of agreement markers should in principle be discernible on the basis of synchronic distributional restrictions. This is essentially the position that Ariel adopts. Her contention is, however, that in most languages synchronic data clearly favour an AT origin of agreement markers. My cross-linguistic investigations of person agreement do not support this. Let us therefore turn to the cross-linguistic data.

2.

From pronoun to person agreement marker

The person markers found in languages may be classified into several types. In some languages person markers occur only in the absence of an independent subject argument; they are simply bound pronouns.3 Such markers may be treated as agreement markers only in that they agree in person/number (and/or gender) with a referent in the preceding or following discourse. But they are not clause internal agreement markers. A language which displays such person markers is Retuarä, a Tucanoan language of Colombia (7). Retuarä (Strom 1992: 218-219) (7) a.

b.

Toma-re hose-re Thomas-TERM JOSC-TERM 'Thomas saw Jose.'

ia-ko?o see-PAST

Sa-ki-ba?ako?o it-he-ate-PAST 'He ate it.'

In other languages the person markers occur both in the presence of the corresponding independent subject argument and in its absence, as in the following examples from Tauya.

On the discourse basis of person agreement

37

Tauya (MacDonald 1990: 118) (8) a.

b.

Fena?-ni fanu-/0 woman-ERG man-ABS 'The woman saw the men.'

nen-yau-a-?a 3pL-see-3so-iND

Nen-yau-a- ?a 3pL-see-3so-iND 'She/he saw them.'

The person markers are thus functionally ambiguous. In the presence of independent lexical or pronominal arguments they function as clause-internal agreement markers. In the absence of independent lexical arguments they function as bound pronouns. In yet other languages the person markers always co-occur with corresponding independent subject arguments; they no longer have a referential function and are fully grammaticalized agreement markers. Such markers are illustrated in (9) on the basis of Dutch. Dutch (9) a.

Piet zie-t Kees Piet see-2/3so Kees 'Piet sees Kees every day.'

elke every

dag. day

b.

*(Hij) zie-t Kees he see-2/3so Kees 'He sees Kees every day.'

elke every

dag. day

The three types of person markers may be seen as reflecting three stages in the grammaticalization of agreement as shown in (10).4 (10) bound anaphoric pronoun > referential (rich) agreement marker > nonreferential (weak) agreement marker As noted earlier, the AT analysis deals specifically with the emergence of bound pronouns, i.e. with stage one of the diachronic development in (10). The NPdetachment analysis, on the other hand, skips over stage one and begins with stage two.5 Ariel is happy to accept that stage one is not a necessary prerequisite to stage two (but see note 2). Nonetheless, she considers the dominant line of development of agreement to begin with bound pronouns, i.e. stage one. A possible way of testing whether this is indeed so, is to examine the distribution of languages with respect to the three stages of the grammaticalization of agreement. Given that grammaticalization is seen to be a continuous process ongoing in all languages at all times, we would expect to find synchronic evidence of all

38

Anna Siewierska

the above stages of the grammaticalization of agreement. And indeed we do. It is less clear what our expectations should be with respect to each of the stages of the grammaticalization of agreement. Should we expect all to be equally frequently attested or rather that some should be more common than others? Rather than speculate further on the issue, let us consider the cross-linguistic data. My investigation of person agreement is based on a genetically and areally stratified set of 332 languages. For the purpose of this discussion I will cite statistical data only on the nature of person agreement markers corresponding to the S, i.e. the subject of intransitive clauses.6 Such agreement is manifested in 259 languages (see the appendix). These 259 languages will be referred to from here on as the sample. In the vast majority of the sample languages (88%), the person markers are referential agreement markers, i.e. markers reflecting the middle stages of grammaticalization. Non-referential subject markers are very rare. They occur in only three of the languages in the sample, namely Dutch, Vanimo, and Anejom. And bound pronouns are also quite uncommon.7 Only 27 languages have such S forms. The relevant data are presented in Table 1. Table 1. Type of S person agreement (N=259) Bound pronouns Referential agr Non-referential agr

27

11%

229

88%

3

1%

How do these data bear on the two postulated origins of person agreement markers? A result that would clearly favour the AT over the NP-detachment diachronic scenario would be a high instance of languages with bound pronouns in complementary distribution with free arguments. This is clearly not the case; only 11% of the languages in the sample manifest such markers. The evident preponderance of person markers capable of co-occurring with corresponding free forms, on the other hand, could be seen as indicative of the dominance of the NP-detachment origin of person markers. If so, this would be highly damaging for Ariel. However, I am hesitant to interpret the data in this way as such person markers, though problematic for AT, constitute a possible later stage of development from bound pronouns. Ariel (2000: 207-208) suggests that the co-occurrence of bound pronouns with corresponding free arguments depends on a high degree of fusion between the person marker and the verb. The greater the fusion, the higher the degree of accessibility that the person marker encodes and thus the more restricted the contexts in which it alone can be used. She argues that such fusion discourages, though does not preclude, a referential interpretation of the person marker. Thus independent arguments start co-occurring alongside the person markers. This

On the discourse basis of person agreement

39

is seen to be particularly likely in situations where the degree of accessibility associated with the referent of the bound pronoun is not deemed high enough to merit reference by the bound pronoun only. While the above account of the emergence of actual agreement markers from bound pronouns sounds very plausible, it also suggests that only those agreement markers which exhibit a high degree of fusion with the verb potentially qualify as having an AT origin. A high degree of fusion with the verb is, however, also fully compatible with the NP-detachment origin of the markers. Thus a consideration of the degree of fusion with the verb of person markers cannot discriminate between the two diachronic scenarios, unless most referential agreement markers do not exhibit a high degree of fusion. Such a situation would provide additional support for the NP-detachment over the AT analysis. I have not been in a position to carry out a comprehensive analysis of the degree of fusion with the verb of the referential agreement markers in my sample.8 I do, however, have some data on the ordering of person agreement affixes relative to other verbal affixes which, arguably, may be viewed as suggestive of degree of fusion. Taking a very liberal and AT friendly view of "high fusion", I treated as highly fused with the verb any person marker occurring next to or close to the stem and preceded (in the case of prefixes) and followed (in the case of suffixes) by at least one other major inflection affix. Two, relatively simple, cases in point are illustrated below, both involving a third singular form followed (11) and preceded (12) by a tense marker, respectively. Amele (Roberts 1987: 163) (11) S Horn uga wali-ag

ho-i-a

Silom 3sc brother-3SG come-3sG-TODAY PAST 'Silom's brother came.' Tuscarora (Mithun 1976: 41) (12) Wi:rv:n wahra:-kv? William AORIST- 3soM-see- PUNCTUA! 'William saw the dog.'

tsir dog

In 37 languages in the sample the S person marker appears to be such an inner marker obligatorily and in a further 51 languages in at least some circumstances. Another potential manifestation of fusion is sensitivity of the person marker to tense/aspect and/or modality distinctions as in the Austronesian language of New Guinea, Labu, illustrated in (13).

40

Anna Siewierska

Labu (Siegel 1984: 103,106) (13) a. Ai yututu iya ko I IsGiPAST-fire dog with Ί hit the dog with the stone.' b. Ai ngwa nda- taka I PUT lsc:iRR-cut.with tool 'I'll break open the coconut.'

hu stone ni the

poso coconut

Various forms of sensitivity to the ΤΑΜ system are displayed in another 31 languages in the sample. Thus if we take the above two factors as indicators of high fusion, about 55% (127/229) of the languages with referential agreement markers in the sample may be seen as having markers attributable to either AT or NP-detachment. The agreement markers in the remaining 45% of the languages, in turn, are more compatible with a NP-detachment origin. In sum, a consideration of the cross-linguistic distribution of the three stages of grammaticalization of person markers does not evidently favour either of the two sources of person agreement. A small minority of languages display person markers which can be accounted for in terms of AT but not NP-detachment. These are the languages with bound pronouns in complementary distribution with free forms. A larger number of languages have markers which do not evidently manifest a high degree of fusion with the verb and thus are more amenable to the NP-detachment analysis of the emergence of agreement. The person markers in the majority of languages, however, appear to be in principle compatible with either diachronic pathway. Crucially, there is no indication that the AT origin is the universally favoured one. This conclusion will be further reinforced by the data on person asymmetries to be considered below.

3.

Person asymmetries in subject agreement paradigms

That the exponents of subject person agreement markers in the case of the third person, particularly in the third person singular are frequently zero is a commonly made observation. Benveniste (1971) has even suggested that overt third person subject agreement markers are quite exceptional. It is therefore not altogether suprising that Ariel assumes that this is indeed so, particularly in the absence of any large scale investigations which would indicate otherwise. Nonetheless, despite the many examples in the literature of languages which have zero markers in the third person, a systematic study of the formal realization of agreement markers in the 210 languages in my sample reveals that this is by no means the dominant pattern. The relevant data are presented in Table 2.

On the discourse basis of person agreement Table 2. Zero exponents of S person markers (N=259) All zero Sg zero Some zeroes 2% 1 0.4% 5 1st 3 1% 2nd 2 0.7% 3 1% 0 32 12% 3rd 28 11% 25 10%

41

No zeroes 250 97% 254 98% 174 67%

The "all zero" column depicts languages in which all the forms of subject agreement for the person in question (irrespective of number, gender or tense/aspect/ mood) are non-overt. The "sg zero" column presents languages in which only the sg form is zero while the non-singular forms have overt markers. And the "some zeroes" column groups the languages which have at least one zero allomorph in the singular or non-singular. Recall that according to AT, third person referents are not consistently highly accessible enough to warrant encoding by high accessibility markers. Ariel's prediction therefore is not that third person markers should have a zero exponent, but rather that they should not arise at all, or only rarely. Consequently, the figures in Table 2 directly relevant to Ariel's claim are those in the "all zero" column. These figures confirm, in no uncertain terms, that third person zeroes are overwhelmingly more frequent than first or second person ones. First person markers (for the S) are lacking in only three (Nez Perce, Trumai and Dutch) of the 259 languages and second person markers in only two (Nez Perce and Trumai).9 Third person markers, by contrast, are lacking in 28 languages. The languages in question are listed in the appendix. The data in Table 2 do not, however, confirm Ariel's contention in regard to the dominant form of subject agreement paradigms. Only 11 % of the languages in the sample lack third person agreement markers altogether. Overt realization of only first and second person markers is therefore clearly the minority rather than the majority pattern, contrary to what Ariel assumes. It could be argued that in some of the languages which have a zero exponent only for the third person singular, i.e. those in the "3sg zero" column, the third person non-singular marker is actually a number rather than a person marker. This is the case in the Uto-Aztecan language of El Salvador Pipil, as shown in (14). Pipil (Campbell 1985:54)

(14) Isg 2sg 3sg Ipl 2pl 3pl

niti0 //- t an~t 0- -t

42

Anna Siewierska

Such languages would then also qualify as lacking real third person agreement. However, even if we make allowances for this, the number of languages lacking third person agreement is not likely to rise dramatically. It is worth noting in this context that not only the number of languages which lack third person agreement is far lower than Ariel assumes but that the same applies to third person zeroes overall. Only 33% of the languages in the sample have any zero exponents for the third person. Therefore even if we were to treat all the languages in question as lacking third person agreement, they would still constitute the minority. In sum, Ariel is clearly incorrect in her claim that overt realization of the first and second person only is the "universally predominant inflectional pattern" (Ariel 2000: 213). The above findings do not invalidate Ariel's AT analysis of the rise of person agreement. They do, on the other hand, invalidate her claims as to the widespread AT origins of agreement. If most languages displaying subject person agreement have overt third person markers, like the data in Table 2 suggest, but such markers only rarely have an AT source, as argued by Ariel, it follows that AT cannot be the driving force behind the emergence of agreement in the majority of languages.

4.

Accessibility and third person markers

We have seen that though, contrary to the predictions of AT, third person agreement markers are highly frequent, they do nonetheless arise less often than first and second person ones. Whereas 99% of the languages in the sample have first and second person markers, the corresponding figure for third person markers is 87%. These findings could be reconciled with the AT account of the rise of person agreement if person markers are treated as markers of a wider range of accessibility, not necessarily only of high accessibility. In fact, I cannot see how such a move can be avoided if Ariel's AT developmental scenario is to be extended to other person forms, most notably object agreement markers. Since according to the accessibility hierarchies in (5), referents coded by objects are inherently less accessible than those coded by subjects, the bounding to the verb of pronominal objects must involve conditions of lower accessibility than those of subjects. Thus if first and second person object bound pronouns are lower accessibility markers than first and second person subject ones, a similar distinction should also be possible in the case of bound first and second person pronouns as opposed to third person pronouns. Ariel (p.c.) is not sure whether her AT analysis of the rise of person agreement also holds for object agreement. However, as the NP-detachment analysis also includes object agreement, the

On the discourse basis of person agreement

43

AT analysis could hardly be treated as a serious competitor if it too did not embrace object person forms. Significantly, the person asymmetries which appear to drive the AT account of the rise of person agreement in the case of subjects, also hold with respect to objects (see Siewierska and Bakker, to appear). The above notwithstanding, accommodating the rise of third person agreement within AT is not unproblematic in regard to the marking of the first and second person. A consideration of the accessibility hierarchies in (5) and of the accessibility marking scale in (6) presented in section 1 leads us to expect that the most inherently highly accessible referents, the speaker and hearer, should receive the most extreme form of attenuated encoding, i.e. they should be zero marked. Yet they are not. Though Ariel does not discuss this issue, the extreme rarity of zero first and second person agreement markers may be seen as a natural consequence of the absence of third person markers. If, however, the latter are "permitted" to develop, there is no immediate AT-based explanation for why the first or possibly second person should not be zero marked. Somewhat paradoxically, though Ariel does not consider the initial creation of third person agreement as being motivated by AT, she views the actual presence of third person agreement in a language as fully compatible with the theory. She argues that as AT is a theory about referent encoding, in assessing its applicability the total means used for encoding a given discourse referent needs to be taken into account. What is important for AT is that third person referents be consistently encoded by relatively lower accessibility markers than first and second person referents. And this is typically so irrespective of whether a language does or does not have third person agreement. Since third person referents are not as inherently accessible as first and second person ones, they tend to be encoded by full NPs much more frequently than by pronominal expressions let alone reduced pronominals. Whether or not such NPs are accompanied by third person agreement markers is immaterial to AT. A NP with or without an agreement marker is a less attenuated form of encoding than a first or second person bound pronoun or even a first or second person independent pronoun plus a corresponding agreement marker. In fact Ariel mentions only two types of encoding which would be incompatible with AT, both of which involve languages which allow zero anaphora for third person subjects. If in such a language first and second person subjects must be overt, or if only first and second person subjects manifest agreement, then they will be encoded by lower accessibility markers than third person referents. Such languages do not appear to be attested.

44

Anna Siewierska

5.

Concluding remarks

I have argued that though first and second person agreement markers are more common than third person ones, contrary to what Ariel assumes, overt realization of agreement markers for all three persons is the cross-linguistic norm. Therefore if AT is seen as providing an account of the emergence of first and second person agreement only (or primarily), then it cannot be considered as the dominant force driving the rise of person agreement on a cross-linguistic basis. Alternatively, if AT seeks to account for the universally dominant form of agreement markers, it must allow for the regular (as opposed to incidental) development of third person agreement. Since there is no reason to doubt that third person referents are indeed less inherently accessible than first and second person ones, if the development from independent pronoun to bound pronoun of all three persons is to be AT driven, then it must be assumed to be motivated by different levels of accessibility. There is, of course, the possibility of maintaining that whereas first and second person agreement markers are due to AT, third person markers arise via NP-detachment. One piece of evidence that argues against such a general compromise is that of the 27 languages in my sample that have bound pronouns as opposed to referential agreement markers for the S, all but four have overt third person forms. And recall that bound pronouns in complementary distribution with free forms is the one form of person agreement that falls outside the scope of the NP-detachment analysis. I see no principled reason why Ariel's AT account of the rise of person agreement should not be extended to third person forms or to other than subject agreement markers. In fact, I very much hope that it will, as only then will we be able to fully assess its cross-linguistic potential and its merits vis-ä-vis the NP-detachment analysis.

Appendix The agreement markers considered are those for the agentive argument of an intransitive verbal predicate. Languages which lack third person agreement Ainu, Basque, Dakota, Dogon, Guarani, Kashmiri, Katla, Koasati, Kutenai, Maricopa, Nandi, Navajo, Newari, Nootka, Northern Tepehuan, Tetelcingo Nahuatl, Nunggubuyu, Pipil, Plains Cree, Seri, Sierra Popoluca, Squamish, Tlingit, Warao, Waura, Wichita, Wintun, Yukulta

On the discourse basis of person agreement

45

Languages in which the third person singular agreement marker is zero while the third person non-singular is/are overt Armenian, Asmat, Atakapa, Atayal, Bygansi, Cahuila, Chepang, Comox, Cora, Crimean Tatar, Finnish, Gooniyandi, lea, Jacaltec, Kalkatungu, Kiowa, Kobon, Kunama, Mangarayi, Nambiquara, Ngalakan, Paamese, Paumari, Pitjantjatjara, Slave, Tonkawa, Turkish, Tzutujil, Lower Upquaa, Wambon, Wardaman, Welsh Languages in which at least one of the major allomorphs of the third person agreement marker is zero Akkadian, Amuesha, Cayuvava, Djingili, Doyayo, Egyptian Arabic, Harar Oromo, Hebrew, Kurdi, Lango, Latvian, Makushi, Marunguku, Otomi, Passamaquoddy Maliseet, Persian, Quechua, Summerian, Takelma, Tarascan, UrubuKaapor, Waorani, Warekena, Yapese, Yele Languages with overt third person agreement markers (by macro-area) AFRICA: Afro-Asiatic (Amharic, Beja, Bilin, Central-west Gurage Coptic, Dizi, Dongolese, Geez, Hausa, Iraqw Tamazight, Tigrinya) Khoisan (Sandawe) Niger-Kordofanian (Diola Fogny, Ewe, Fula, Koma, Kongo, Koromfe, Krongo, Luvale, Maba, Ndonga, Nkore Kiga, Swahili, Zande, Zulu) Nilo-Saharan (Bagirmi, Berta, Fur, Kanuri, Mesalit, Murle, Ngiti, Pari, Turkana) AUST-NG: Australian (Alawa, Malak-Malak, Maung, Ngankikur, Ngiyambaa, Nyulnyul, Tiwi, Ungarinjin, Wambaya) Indo-Pacific (Alamblak, Amele, Anem, Au, Bukiyip, Daga, Ekari, Gapun, Grand-Valley-Dani, Hua, Kapau, Kewa, Maisin, Marind, Maybrat, Nasioi, Salt-Yui, Selepet, Sentani, Tauya, Tehit, Una Vanimo, Wanuma, Waskia, Yava, Yimas) EURASIA: Altaic (Evenki, Ju Chen, GDgur, Khalka Mongolian) Northwest Caucasian (Abkhaz) Kartvelian (Georgian) Chukchi-Kamchatkan (Chukchi) Elamo-Dravidian (Kannada, Brahuri) Austric (Mundari), Indo-Hittite (Albanian, Dutch, Greek, Hindi, Hittite, Italian, Ossetic, Polish) Language Isolates (Burushaski, Hurrian, Ket) Uralic-Yukaghir (Hungarian, Yukaghir, Evenki) N-AMER: Amerind (Achumawi, Acoma, Awa Pit, Lealo Chinantec, Chalcalon Mixtec, Chocho, Comanche, CopalaZoque, Copala Trique, Hannis Coos, Karok, Mohawk, Mountain-Maidu, Nez-Perce, Oneida, Quileute, Salinan, SS-Miwok, Coast Tsimshian, Tunica, Upper-Chinook, Washo, Yuchi, Yurok, Zapotec) Eskimo-Aleut (Greenlandic Eskimo, Yupik Eskimo)

46

Anna Siewierska

S-AMER: Amerind (Apurina, Arawak, Ashaninca, Aymara, Bororo, Candoshi, Canela Kraho, Capanahua, Carib, Cavinena, Chacobo, Hishkaryana, Huave, Iquito, Mapuche, Mataco, Miskito, Nadeb, Piraha, Rama, Retuara, Southern-Barasano, Suena, Trumai, Tuyuca, Wari, Yagua, Zapotec) SEA and OC: Sino-Tibetan (Bawm, Lushai, Rawang) Austric (Anejom, Chamorro, Fijian, Indonesian, Kali-Kove, Kapampangan, Kilivila, Konjo, Larike Mono-Alu, Muna, Palauan, Sahu, Taba Temiar, Tinrin, Tolai, Tsou, Tukang Besi, West Makian, Uma).

Notes 1. The zero marking of third person agreement forms is typically seen to be economically motivated, i.e. it is considered to be a reflection of the tendency to shorten the linguistic expressions used most commonly. For some discussion see Haiman (1985) and Croft (1990). A somewhat different explanation has been offered by Koch (1994). 2. Ariel (2000) seems to assume that the NP-detachment analysis is incompatible with the continuing referential status of the person agreement markers. Givon (1990: 302) does not share this assumption. I will follow Givon. 3. In the following discussion the term bound pronoun will be used in the above sense, i.e. for forms in complementary distribution with free forms. Often the term is used more widely and encompasses also referential agreement markers. 4. This is, of course, a simplification. What we are dealing with is a grammaticalization cline. While many, if not most, languages can be shown to display markers corresponding to the three stages shown in (10), there are also languages with markers showing in between properties, e.g. complementarity with full NPs but not pronouns (e.g. Welsh) or vice versa (e.g. Palauan) or with proper names but not common ones (e.g. Kusiean). 5. Strictly speaking the actual bounding of the pronominal form to the verb (or some other constituent) is not a necessary aspect of the NP-detachment analysis. There are languages in which the person agreement marker is a free form, e.g. Woleaian and Khasi. 6. I decided to concentrate on agreement in intransitive rather than transitive clauses as this avoids the issue of which marker to take into account in languages with ergative or hierarchical agreement. In languages manifesting active agreement where the intransitive subject takes two types of markers, I took into consideration the agentive marker. 7. Bound pronouns (in the sense of the term used here) are considerably more common for objects than for subjects. Of the languages in the sample, 27 have bound pronouns for the subject as opposed to 43 for the object. Moreover, if a language has person

On the discourse basis of person agreement

47

agreement with both of the arguments of a transitive clause, if the subject marker is a bound pronoun, so is the object marker but not vice versa. See Siewierska (1999). 8. An obvious factor to consider is the degree of phonological similarity between the person marker and the independent pronoun, relative lack of similarity being suggestive of the time elapsed since the bounding of the pronoun to the verb. It must be remembered though, that the current independent pronouns need not be the source of the person agreement markers. In fact, the latter may give rise to the former. 9. Nez Perce has a very unusual person agreement system where the intransitive subject for the first and second person and also the transitive subject if the object is a speech act participant are zero marked. The third person has overt marking. The person marking is not sensitive to number. This is marked by a separate affix. See Rude (1985). In Trumani (Guirardello 1999: ch 3), there is a third person enclitic -n/-e (attached to the last element of the verb phrase, typically the verb) which occurs in complementary distribution with the absolutive (S/O) head noun. There are no corresponding clitics for the first and second person.

References Andersen, Stephen R. 1992 Amorphous Morphology. Cambridge: Cambridge University Press. Ariel, Mira 1990 Accessing NP Antecedents. London: Routledge. 2000 The development of person agreement markers: from pronouns to higher accessibility markers. In: M. Barlow and S. Kemmer (eds.), Usage-based Models of Language, 197-260. CSIL Publications. Benveniste, Emile 1971 Problems in General Linguistics. Translated by Mary Elizabeth Meek. Cora Gables, Fa: University of Miami Papers. Campbell, Lyle 1985 The Pipil Language of El Salvador. Berlin: de Gruyter. Corbett, Greville 1991 Gender. Cambridge: Cambridge University Press. Croft, William 1990 Typology and Universals. Cambridge: Cambridge University Press. Givon, Talmy 1976 Topic, pronoun and grammatical agreement. In: Charles Li (ed.), Subject and Topic, 151-188. New York: Academic Press. Givon, Talmy 1990 Syntax: A Functional-Typological Introduction. Volume 1. Amsterdam: John Benjamins. Guirardello, Raquel 1999 A reference grammar of Trumai. Ph.D dissertation. Rice University.

48

Anna Siewierska

Haiman, John 1985 Natural Syntax. Cambridge: Cambridge University Press. Koch, H. 1994 The creation of morphological zeroes. In: G. Booji and J. Van Marie (eds.), The Yearbook of Morphology, 31-71. Dordrecht: Kluwer. Lehmann, Christian 1982 Universal and typological aspects of agreement. In: Hans-Jakob Seiler and J. Stachowiak (eds.) Apprehension: Das sprachliche Erfassen von Gegenständen. Vol. 2, 201-267. Tübingen: Gunter Narr Verlag. Macdonald, Lorna 1990 A Grammar ofTauya. Berlin: Mouton de Gruyter. Mithun, Marianne 1976 A Grammar ofTuscarora. New York: Garland. Moravcsik, Edith A. 1978 Agreement. In: Joseph Greenberg (ed.), Universals of Human Language. Vol. 4, 331-374. Stanford: Stanford University Press. Roberts, J.R. 1987 Amele. London: Croom Helm. Rude, Noel 1985 Ergative, passive and antipassive in Nez Perce. In: M. Shibatani (ed.), Passive and Voice, 547-560. Amsterdam: John Benjamins. Siegel, Jeff 1984 Introduction to the Labu Language. Papers in New Guinea Linguistics 23: 83-157. Siewierska, Anna 1999 From anaphoric pronoun to grammatical agreement marker: why objects don't make it. Folia Linguistica. 33/2: 225-251. Siewierska, Anna and Dik Bakker Forthc. Person asymmetries and the grammaticalization of agreement. Strom, Clay 1992 Retuarä Syntax. (Studies in the Languages of Colombia 3). Dallas: Summer Institute of Linguistics and The University of Texas at Arlington.

Chapter 4 The information structure of bilingual meaning: A constructivist approach to Californian Finnish conversation1 MM Jocelyne Fernandez-Vest

The recent introduction in cognitive science of discourse approaches makes it possible to tackle from an interdisciplinary point of view the structuring and accessibility of information - particularly the discourse strategies of topicalization and focalization - the interactional dimension of mental spaces (cf. the importance of 'shared knowledge'), and more generally, the cognitive aspects of textual coherence and prominence. Moreover, the initial universalism of cognitive research has been counterbalanced by the emergence of typological concerns and a recognition of the diversity of languages in the debate on 'grammar and cognition' - see Hopper 1987, 1998; Langacker 2000, 2001; Talmy 2000a, 2002b; Thompson 2002; Thompson and Hopper, fc. In the French domain of linguistics - or 'sciences du langage' as it is nowadays called - this evolution, strongly supported by the CNRS (French National Center for Scientific Research), has gone through several stages during the last decade. These stages can be somewhat arbitrarily reduced to the following four (see Fernandez-Vest, 2003). (1) A collective volume, a thematic issue of the journal of Association pour la Recherche Cognitive (Fernandez-Vest (ed.) 1995); (2) A workshop organized at the X Vlth International Congress of Linguists, Paris 1997, see Fernandez-Vest (ed.) 1998; (3) A research network created in 1999 at CNRS 'Diversite et evolution des langues: enjeux cognitifs' (see Fuchs and Robert (eds.) 1997, 1999); (4) A research laboratory founded in 2000, OSTERLITS, UMR 7108, CNRS - University Paris ΙΠ-Sorbonne Nouvelle - Ecole Pratique des Hautes Etudes - Institut National des Langues et Civilisations Orientales (see Fernandez-Vest (dir.) 2004b). The research referred to below was originally initiated within the program Orality and cognition', part of another CNRS laboratory (LACITO), and it has been mainly conducted by the new team presented under (4), above.

50

MM Jocelyne Fernandez-Vest

1.

Information structure and language contacts

In this section the focus is first on information structure in monolingual discourse (1.1), and then on information structure and code-switching in bilingual meaning (1.2). 1.1. The thematic construction of meaning in monolingual discourse The thematic construction of meaning in monolingual discourse is here approached from the perspective of a Finno-Ugric interlanguage: Finnish spoken by bilingual Sami (1.1.1). After discussing some of the definitions and data needed to enlarge the perspective (1.1.2), I shall move on to a broader comparison of information structure in different languages and the basic Schemas necessary to study its strategy (1.1.3). 1.1.1. Sami bilingualism and interlanguage My first initiative as an ethnolinguist was triggered off by a certain conception, still dominant in the 70s, according to which the bilingualism of weakly educated minority people could only be evaluated in comparison with the two language systems in contact (see the debate about semilingualism in Hansegärd 1979; Fernandez-Vest 1989; Hagege 1996). A "shorter" Ph.D. dissertation on 'Finnish spoken by bilingual Sami in Utsjoki-Ohcejohka' (Fernandez-Vest 1982) took the pretext of a syntactic and 'enunciative' (i.e. pragmatic) description of 'Finnish influenced by Sami' for rejecting this type of method. Northern Sami, spoken in Finnish and Norwegian Lappland, is the majority language of the Samic branch of Finno-Ugric languages. As a reaction against the conventional measuring of multilingual situations, the study aimed to be a functional presentation of a variant of Finnish spoken by non-native speakers of Finnish. The introduction to the published volume parodied the normative studies as describing an instrument of communication so out of tune that nobody could believe it has any use for communication. And a final protest of Galilean tone rose: "But it does function though!" Beyond the morphological and syntactical inventory of this interlanguage, the most innovating part was the third one, which, leaving aside the comparative point of view (spoken Finnish was still very little described at the time), explained the functioning of this system at the message level. Based on the three main thematic devices (i.e. prosody, order of constituents, and the formal marking of the dichotomy Theme / Rheme), this discourse analysis skissed a perspective 'beyond the sentence'. In the conclusion - "Strategie discursive" -

The information structure of bilingual meaning

51

I considered the basic hypothesis verified: bilingual speech achieves some of the latent possibilities of the two language systems. This idea of a potential exploited in a creative way by the bilingual speaker needed to be investigated further and backed up by the study of thematic devices in monolingual corpora. That is what most of my subsequent research has been devoted to, in Sami and some other languages, particularly Finnish and French (Fernandez-Vest 1987b, 1994b). 7.7.2. Defin it ions and data In the approach adopted here, deeply tentative in the 60s, then strengthened by the study of various corpora, meaning, as linguistically analysable, results essentially from a co-determination by enunciative factors (with universal tendencies) and morphosyntactic factors (typological, products of linguistic systems). My analysis of enunciative constituents meets the principle of a triple organization of the utterance recognized by several language theoreticians (Pierce 1932; Danes 1974; Hagege 1978,1993,2000a; Guimier (ed.) 1999), but it states the hierarchical and chronological primacy of Level 1, the enunciative level, on Level 2, the morphosyntactic one (see also Chafe 1987, 1994). My choice of a textual and interactional definition of the Theme/Topic ("what is spoken about") and Rheme/Focus ("what is said about it"), implied the recognition of a third element, the Mneme, which is characterized by formal properties (a post-Focus marked by a flat intonation) and semantic ones (supposedly shared knowledge, affective modulation, etc.; Fernandez-Vest 1994a: 197-200). This third element, already spotted in a few non-Indo-European languages by several linguists (Kuno 1978 for Japanese, Erguvanli 1984 for Turkish), has been variably labelled in the literature as 'right detached element', 'de-focused NP', 'afterthought NP', 'post-predicate constituent'. Furthermore, the Mneme has later even proved to be similar to two notions elaborated within separate frameworks: the 'tail' of Functional Grammar (Dik 1978; Valdduvi 1990) and the 'antitopic' of Construction Grammar (launched by Chafe 1976 and developed by Lambrecht 1981, 1994). My personal conception, gradually shaped by the analysis of diverse corpora, meets the definition of Lambrecht's Antitopic in general terms. His definition relies on the Principle of the Separation of the Reference and the Role (Lambrecht 1994: 184-191, 202, 238, 252, 268). Yet my view differs from Lambrecht's definition in several details: the textual Mneme does not submit to the 'constraint of strict localization' (i.e. the Antitopic must immediately follow the clause containing the topic expression, Lambrecht 1994: 205), nor does it conform to the impossibility of a double occurrence of Antitopics (Lambrecht 1998: 42; Michaelis and Lambrecht 1996). To put it

52

MM Jocelyne Fernandez-Vest

briefly, the object is very similar, but the frame of analysis (utterance vs. text) and the levels of assignment (morphosyntactic vs. enunciative/pragmatic) are different - which has inevitable implications on the exploitation of the corpora (Fernandez-Vest 2004a). Oral data of different origins have been presented in my books on the Northern Sami discourse and Discourse Particles (DIPs, 'Particules Enonciatives'), see Fernandez-Vest (1987a, 1994a). The description of particle systems in several unrelated languages (mostly Indo-European and Finno-Ugric), a description achieved through the complementary itineraries of ethnolinguistics and the Enunciation theories, permits some typological and diachronical hypotheses: the particle modulation as opposed to intonation, the demotivation as a morphogenetic device (grammaticalization), see Fernandez-Vest (2000b). This method for a phenomenological analysis of impromptu speech, which is the basis of my thematic model with three constituents (Theme - Rheme - Mneme), has been recently applied to a long-distance field: Finnish spoken by bilingual Californians with a Finnish origin. 7.7.3. The construction of monolingual meaning The works mentioned above have led me to adopt a constructivist position, which can be briefly formulated as follows: through observations of thoughts in progress and their linguistic expression, the analysis of impromptu speech gives us access to cognitive processes that underlie the construction of meaning (Fernandez-Vest 1995a). One can choose four of the definitory dimensions of impromptu speech (see Enkvist (ed.) 1982; Enkvist (ed.) 1985; Fernandez-Vest 1994a: 138-141): (1) the duophoric ('dialogical') relation (2) the immanent constructivism, manifested by the importance of discourse articulators (e.g. the DIPs) and lexical inventories ('listings') (3) the prevalence of marked enunciative strategies (topicalizing devices) (4) the implicit reference to context. Each of these points, which were originally put forward by the pragmatic investigation of a Finno-Ugric oral language, Northern Sami (Fernandez-Vest 1987a: 301-603, 1994a: 55-65, 160-171), have also been illustrated by the analysis of French corpora (regional French spoken in the South-Western part of France). The analysis of the narrative parts of these corpora has confirmed the following two basic thematic strategies: - the binary strategy 1 (Theme - Rheme), with a topic currently marked by prejection, characteristic of oral discourse, and

The information structure of bilingual meaning

53

- the binary strategy 2 (Rheme - Mneme), which often makes it possible to complete an iconic (circular) cohesion of the text (Fernandez-Vest 1995a, 1998b,2000a). This thematic model has been applied to other types of corpora, namely 'simulated' oral texts like comics and cartoon versions of the Aventures de Tintin by Herge and their translations into Finnish (Fernandez-Vest 1995b), and even to argumentative strategies in film dialogues - like Woody Allen's Hannah and her sisters and its French version (Arleo and Fernandez-Vest 2004). The systematic differences of thematic strategies between French and Finnish can be illustrated by two excerpts of 'simulated speech', from the Tintin comics corpus and its Finnish translation (Tintin et les Picaros, 1974; Tinttija Picarot, 1976): (1) - C 'est bien possible, mais moi I 'air en conserve, je n 'aime pas 'it can be possible, but me canned air, I don't like that!' - Voi hyvinkin olla, mutta minua ei säilötty ilma miellytä! lit. '..., but me does not canned air please!'

!

-^ The binary strategy 1, with a theme marked by prejection in French, is rendered in Finnish by an integrated construction (and no left detachment), as the object, always marked by a specific case (here the partitive minua), can be fronted - followed in this utterance by the connegative (ei) + the subject (säilötty ilma) + the main verb (miellytä). (2) ~(^an 'arrive qu 'a moi, des choses pareilles ! 'It happens only to me, such kind of things!' - Tällaista ei voi kyllä sattua muille kuin minulle! lit. 'Such a thing can not really happen to others than me!' -> The binary strategy 2 in French, Rheme - Mneme, is rendered by a straight construction and no right detachment in Finnish.

1.2. Bilingual meaning and code-switching An interesting question is whether, considering the great differences of thematic strategies implied by the systems of the two languages already analysed (see examples (1) and (2) above), one can without difficulties apply the above mentioned model of discourse analysis to Code-switching (CoSwit), which is a natural device of bilingual speech (Auer 1984; Baetens-Beardsmore 1982; Paradis 1978; Romaine 1989).

54

MM Jocelyne Fernandez-Vest

Like the study of language contacts in general (Weinreich 1953), the study of CoSwit constitutes a challenge for linguistic reflexion. The mixed languages, often stigmatized, are little codified, or at least not explicitly standardized. One can nevertheless observe how the most genuine and naive speakers switch easily from one code to the other, and common tendencies can be easily spotted among speakers of the same group (Muysken 2000). A regular subject of terminological debate has been the distinction between 'loan' and 'CoSwit' (e.g. Saari 1989; Myers-Scotton 1992, 1993). One of the criteria generally used for 'loan' has been the phonological assimilation, which is for many reasons unsatisfactory. On the one hand, this choice seems too strongly deduced from the contacts between genetically related languages (often Indo-European). More appealing is the consideration put forward by linguists working on languages with a rich morphology (which is the case of Finno-Ugric languages) that a morphologically integrated item still remains a CoSwit unit. Consider (3) below (from Halmari 1997: 17): (3) - Kerran sä olit pannu sita mun lunchbox+iin. Once you had put it into my lunchbox.' As for the opposition 'matrix language' vs. 'embedded language', claiming that (1) the definition of the 'matrix language' has to rely on enunciative criteria more than syntactic ones; (2) CoSwit should be looked at as an interactional strategy comparable to the register changes of monolingual discourse (Halmari and Smith 1994) seems to me well founded. Just as the speakers of a monolingual population are not familiar with all the registers of their mother-tongue and can find it difficult to discuss certain subjects (in certain types of relations, etc.), in the same way, in a bilingual community, the elements of the communicative situation will determine which of the two available languages will constitute the 'frame' (a kind of unmarked choice) in which the elements of the other language can be inserted. On the other hand, research aiming at the definition of universal morphosyntactic constraints put on CoSwit (e.g. Poplack 1980; Poplack et al. 1987) is not directly convincing as long as the nature and function of oral syntax are not sufficiently described in the languages in question. The distinction between loan and CoSwit can be seen as resulting from a conventional and inadequate approach to discourse phenomena expressed by words formally belonging to two systems but interpretable in situational and interrelational terms (Stroud 1992). The definition of a gradation of roles (more or less symbolic, according to the degree of independence of the roles from a preexisting situation, e.g. diglossia) for CoSwit as 'indexical of social negotiations' (Myers-Scotton 1988) seems more realistic.

The information structure of bilingual meaning

55

As the conclusion of a relatively narrow corpus study based on different kinds of interaction within variably steady bilingual communities (conversations between (1) native and non-native speakers of Finnish in Finland, (2) bilingual Helsinki Finland-Swedes, (3) Finnish tourists and their Estonian hosts in Tallinn - a temporarily bilingual speech community, with two genetically related but sociopolitically different (Baltic-)Finnic languages) CoSwit can be characterized as both a device for expressing the dialogicality of language, and a strategy for activating several contexts and several genres within a single conversation. From this perspective, multilingual polyphony is thus recognized as a constitutive part of the communicative competence of multilingual speakers (Kalliokoski 1995). The Nordic countries offer an ideal field for research about language contacts, minority languages, and bilingualism. Concerning the future of indigenous minorities, the Finnish microcosm presents an original situation: coexistence on the same national territory of 'Swedes' protected by a strong bilingual legislation, and of Sami ('Lapps') with only a regional status. The disparity relies upon historical and sociocultural differences. The linguistic interest of the situation is increased by a rare confrontation of Indo-European and Finno-Ugric languages. Sweden offers a field unique in Europe of a balanced treatment of immigrated minorities: the interlanguage of migrants and the construction of a multilingual identity are served by diverse programs involving teachers and social workers as well as researchers (Fernandez-Vest 1989; Huss 1999; Wände 1984, 1996,2000). Many of the Nordic research projects, Scandinavian and specially Finnish, have made up the background of my present concerns - the explicitly multilingual studies as well as the innovating contributions of what is nowadays known as the Finnish School of Conversation Analysis (Hakulinen 1989). The objective of the methodology presented here, applied to an extra-European type of Finnish bilingualism and relying upon a less studied type of Finnish-based bilingual corpus (West American Finnish), is both less ambitious and more theoretically oriented than the majority of these highly appreciated studies. Codeswitching, limited in its scope to certain components of the utterance, will not be for instance, as is often the case, included in the problematics of second language acquisition and intercultural communication. It will serve here as a new approach to a theoretical domain, broadly defined as (not necessarily formal) cognitive semantics.

56

MM Jocelyne Fernandez-Vest

2.

American Finnish bilingualism

This section deals with American Finnish bilingualism on the basis of two kinds of oral data: life stories and spontaneous conversation. It is to be noted, however, that the distinction between 'narrative' and 'conversation' is an artefact: authentic impromptu speech corpora normally include both dialogue and monologue (monologue expanding from dialogue and being frequently punctuated with interactional signals of feedback) - which is the case of our Californian corpus, a field study conducted in Northern California (Bay Area and Sonoma County), during eight months in 1996-2001, thanks to several grants from the Centre National de la Recherche Scientifique (CNRS) and the Finnish Cultural Foundation (Suomen Kulttuurirahasto, Helsinki). Although inspired by other experiments and intuitions, the conclusions below are drawn concretely on the basis of this special corpus: 12 older speakers of Californian Finnish, average age 70, second generation of immigrants, both men and women - five of whom have lived the main part of their adult lives with "no Finnish" and regained their mother tongue through different activities (translation, teaching, theater, tourist guiding) after their parents' death. Before proceeding to the discussion of bilingual meaning in Californian Finnish narratives (2.2.) and conversations (2.3.), a short introduction to American Finnish is in order (2.1.).

2.1. American Finns The expression 'American Finns' refers to residents of Finnish origin in the USA and Canada. Haifa million Finns have emigrated over the Atlantic Ocean. The majority of them have settled in the USA, about 50 000 in Canada and 20 000 in Australia. The first departures took place 350 years ago, the most recent ones during the last decades. The most intensive period of emigration was at the end of the 19th century and the beginning of the 20th century, but many of the emigrants settled in their final country only in the 1950's and 1960's. The Finnish immigration is, on the whole, more recent in Canada than in the USA. Further, it is more recent in the Western part of North America than in the Eastern and Central regions (Kero 1986; Virtaranta et al. 1993: 12-34). The research on American Finns, which traditionally distinguishes three generations of migrants, has shown that the border between the USA and Canada does not correspond to a linguistic border (Virtaranta 1971: 81). The majority of immigrants, who arrived in North America at the shift of the century, settled in the region of the Great Lakes: in Minnesota, the higher part

The information structure of bilingual meaning

57

of Michigan, Wisconsin, Ohio, and Ontario. In the East of the United States, Finnish-speaking communities were still to be found a few decades ago in New York, Massachusetts, and Pennsylvania. Younger Finnish-speaking communities, heterogenous but active, were constituted progressively in the West: in the states of Washington, Oregon, and California. The linguistic situation of Finnish Americans, although abundantly described by historians, sociologists and novelists, has been relatively scantly studied by linguists. Until the eighties, the Finnish research on American Finns was dominated by the strong national tradition of dialectology. Field studies have been conducted in New York, in the states of Massachusetts, Illinois, Minnesota, some incursions in California: in Los Angeles, San Francisco and the Valley of San Joaquin. Recent approaches take into account the advances of modern linguistics: the semantic and pragmatic levels have been introduced, and American Finnish corpora are even studied within the frame of specific theories, see, for instance, Halmari 1997 for a clear choice of Government and Binding (GB). Several studies presented at universities in Finland (Helsinki, Jyväskylä, Abo Akademi) and Sweden (Umea, Uppsala) describe American Finnish from the points of view of - phonology and morphology (e.g. Martin 1988, 1989, 1990) - casual syntax (e.g. Jönsson-Korhola 1980, 1987, 1990) - syntactic-semantic and sociolinguistic variation (Karttunen 1977; SahlmanKarlsson 1976; Pietilä 1989). The reference book on these questions in Finnish is Amerikansuomi ('American Finnish', 1993), a book written by the Finnish Academician Pertti Virtaranta and three younger colleagues - Virtaranta being an experienced researcher on American Finnish (Virtaranta 1981, 1988, 1992). The urgency of collecting Finnish-English material appeared to me during an official visit as a visiting scholar in Northern California 1996, cooperating with George Lakoff's and Eve Sweetser's cognitivist team, at UC Berkeley. The main reasons were: (1) The large variety of motivations (to emigrate and then maintain or regain one's ethnic language) which can be found in the Finnish-speaking or bilingual (Finnish/English) corpus collected in Northern California from 1996 on - i.e. at the very last moment, considering the extreme state of decline of Finnish in the American West - sheds an interesting light on the relations between 'language and identity'.

58

MM Jocelyne Fernandez-Vest

(2) Like several ethnolinguists (cf. the great attention nowadays paid to 'endangered languages'; see Fernandez-Vest 2000c, 2002; Hagege 2000b), I believe that to observe how a language is declining can also reveal its stronger characteristics: Amerikansuomi on kuolemassa. Se on tärkeä tutkimuskohde siksi, että sen tutkimuksella on kiire, mutta myös siksi että kielentutkimuksen kannalta tällaisella kielelläon "tavallisilta"kieliltäpuuttuvia etuja... Kuolevan kielen tarkastelu voi parhaimmillaan tuottaa uutta tietoa kaikista kielen osa-alueista, ennen kaikkea siitä... mikä kielessämme on tinkimätöntä ja olennaista. (Martin 1993: 196-197). 'American Finnish is dying. It is an important object of study because time is short for its investigation, but also because such a language offers, from the point of view of research, several advantages generally lacking from "ordinary" languages... The observation of a dying language can at its best provide new knowledge about all part-domains of language, in the first place about... what is irreducible and essential in our language.'

2.2. Bilingual meaning in Californian Finnish narratives In this case study of bilingual code-switching in life stories told by 2nd generation Californian Finns, one of the main objectives has been to evaluate the role of CoSwit as a specific device used by speakers to construct their meaning. The transcription conventions used below are the following. All utterances begin with a capital letter and finish with a dot (terminal intonation). The notation (...) stands for a cut-off passage. The internal segmentation is marked by dashes. Short pauses are marked by two dots [..], longer ones by four dots [....]. The sign =» indicates a quick tempo which turns a sequence of short utterances, separated by a terminal intonation, into a semantic unit, [d] is a phoneme of hesitation ('filled pause'). A few simple criteria permit to detect some of the categories regularly affected by CoSwit, some of them already well inventoried in the specialized literature. To mention the lexical items, we have - place-names, e.g. (4) — Ja vanhempanne olivat tulleet Suomesta? 'And your parents had come from Finland?' Ne tulivat Kanadaan. (...) Ja he tulivat Kaliforniaan /kun Kaliforniassa oli.. kaikki tiet oli /KULTAA. You know oli niinpaljon rahaa täällä / Kaliforniassa /kun ne Ifarmit.... no ne tulivat sinne /suoraan Central Kaliforniaan._

The information structure of bilingual meaning

59

'They came to Canada. (...) And they came to California / when in California .. all roads were / GOLD. You know there was so much money here / in California / when the / farms .... well they came there / directly to Central California.' - socio-economic terms, e.g. (5) - Isä tahtoi että me menimmeyliopistoon (...)/ veljeni kävija /sitten tuli oikein SCIENTIST! (...). Ja minäkin kävinpari vuotta /yliopistoa / Junior College'ia 'Father wanted us to go to the university (...)/ my brother went and / then he became a real SCIENTIST (...). And me too I went a couple of years / to the university / to the Junior College' and among longer utterances, - set phrases which occur in a position of glosses (function as glosses), e.g. (6) [entäs Kalevala?] Todella upea! Oh it s wonderful! ['what about Kalevala?'] 'Really gorgious! Oh it's wonderful!' But these code-switches are not to be interpreted exclusively on a lexical level: they are components of a communicative strategy. It can be shown that the switched words are mostly used in a position of Rheme, (7) - Jo se oli nun kun sidebenefit 'Yes it was a kind of sidebenefit' often with an effect of argumentative cohesion, (8) - Han ei ollut professor / hän oli instructor. 'He was not a professor / he was an instructor.' Both the binary schemes, 1 and 2, can be set into action, combined with a CoSwit in each of the successive segments. This can give a special strength to the argumentative strategy chosen by the speaker, at least in a bilingual environment: (9a) Han 'he

aina always

(9b) mutta minä 'but I

sanoi / että said / that

minä I

ölen am

en ollut / was not /

nationalist/ a nationalist / Mn

nationalist a nationalist' Rh

60

MM Jocelyne Fernandez-Vest

(9c) mä oon enempi internationalist [they laugh] kun.. kun nationalist am more internationalist [they laugh] than.. than nationalist' Rh Mn This rhematic use of code-s witching in the part of discourse usually considered the most intentional strongly suggests a conscious strategy, which probably presupposes shared knowledge of a common bilingualism. My conclusion has thus been that producers of life stories which are situationally anchored in a matrix language (here Finnish within an inquiry on Finnish-speaking Californians) reserve the words directly imported from their major linguistic environment to the most information-loaded part of their utterances, i.e. the rhematic part. Far from revealing a failure of communication, this type of code-switching functions as a primary means for the speakers to construct their meaning. It is most probably a conscious strategy, albeit favored by the shared knowledge of a common bilingualism, which in its turn generates a perceptible complicity. Certain utterances could not succeed in the same way in a non-bilingual community, e.g. (Fernandez-Vest 1998a, 2000a): (10) sitten minä olin vapaa / completely VAPAA [everybody laughs], 'then I was free / completely FREE [everybody laughs].' Rh The climax, resulting in a general outburst of laughter, is here reached by the speaker through a typically bilingual construction. The final AdjP (=Rheme) has a double emphasis: a strong accent on the (repeated) Finnish form of the adjective 'free' (vapaa); and an intensifying adverb in its English form (completely), pointing to (a shared knowledge of) the American tendency - different from Finnish - to use superlative qualifiers (e.g. absolutely, totally, definitely). 2.3. Bilingual meaning in Californian Finnish conversations In this section the focus is on conversations, rather than life stories as in 2.2, above. Several sources (or 'brain starters') for code-switching can be detected. I shall mention two of the more recurrent categories of phenomena, i.e. situational needs (2.3.1) and memory processes (2.3.2). 2.3.1. Situational needs The switch can be induced by the presence of a monolingual addressee, a situation well known to conversationalists. The addressee is not automatically a human being, as in the following example.

The information structure of bilingual meaning

61

(11) - Ja mitä teidän miehenne teki? 'And what did your husband do?' - Hän soitüja tilasi... [speaks to the cat] "Nowyou (hinkyou can make it. OK. TRY again then. There she did it! You had to jump over there before you jumped down!" 'He phoned and ordered... [speaks to the cat] "Now you think (...) down!'" Even though convinced (which is revealed by the code-switched quotations) of the bilingual competence of their partners, speakers can be willing to use a word, or even a phrase or a whole sentence, from the second (majority) language in order to be more exact (see the examples (5) and (8) in section 2.2, above). In (12) below, the speaker first mentions the meeting place for Finns as suomalainen halli ('Finnish Hall') with a hesitating tone and then corrects it by hall. The following mention is in English, in spite of a coherent and syntactically correct Finnish discourse otherwise. The speaker thus manifests that halli is to be taken in the original meaning of the word hall, which is far more current than the Finnish loanword halli and has a different social connotation. At the end of his turn, as well as in the answer to the subsequent question, the English phrases (nickname, the quotation), directly borrowed from the Englishspeaking social context, testify to the knowledge of a shared bilingualism. The speaker's initiative is validated by the hearer's immediate reaction: they laugh, as expected by the speaker. (12) - Oliko pal/on suomalaisia / sünä työssä? 'Were there many Finns / at that working-place?' - Siellä oli PALJONkin. =» Tämä pemkka missä me asumme / nyt / oli paljon suomalaisia täällä oli. Meidän takapihalla / tuolla vähän /se.. se palo oli /suomalainen .. Halli Hall missä nepiti kokouksiaja /ja.... tanssejaja/näytöskappaleita. =» Se oli.... TYÖVÄENHall (-Jaa.). (...) Niit oli.... niit oli.... siellä oliyksi oli.... yksi suomalainen minun isän .... ikäinen kanss. =» Sen nimi oli Lauri Leckman /ja sen haukkuman nimi oli "SLOWMOTION" [they laugh]. 'There were MANY in fact. This corner where we live / now / there were many Finns here yes. In our backyard / there a little / this .. this burnt ground was / the Finnish .. Hall .... Hall where they held meetings and/ and .... dances and / theaterplays. =» It was .... THE WORKERS' Hall (- Oh yes.). (...) There were .... there were .... there was one was .... a Finn in the same age .... as my father =» His name was Lauri Leckman / and his nickname was "SLOW MOTION" [they laugh].' - Miksi?

62

M M Jocelyne Fernandez- Vest

'Why?' - Kun se hiljakseen vaan kävelija teki töitä oikein sillä että. =» Jase aina katoi meidänpäällejaja/aina vain ja sanoi: "American born Russian Finns!" [they laugh]. Se haukku meitäsittenjoo! 'Because he walked along slowly and did his work this way. =» And he always looked at us and and / all the time and said: "American born Russian Finns!" [they laugh]. He made fun of us finally yes!' The intention to be exact is particularly relevant if the circumstances related are by-gones, as in the following example, and it can even raise an explicit metalinguistic activity, as in (14), below. (13) - Meit oli.. San Franciscan satamassa /oli.... niinkö .... d d neljä tuhalta viisi sataa /satamatyöläisiä (...)ja sotan aikana d.... sotan aikana /mind olin holme vuotta "chief dispatcher ". Sepidettiin d.... meitän .... liiton puolestaja laivan omistajien puolesta / oli.. "Hiring Hall" / mistä miehet lähetettiin löihin /minä olin kaksi vuotta "chief dispatcher". Yksi vuosi "regular dispatcher" /siellä on ammatti dispatcher että/se oli kolme vuotta siinä meni (- Aah.). 'We were .. in the harbor of San Francisco / we were .... kind of ..dd four thousand five hundred / harbor workers (...) and during the war d .... during the war /1 was three years "chief dispatcher". We had .... on our .... Union's part and on the ship owner's part / there was .. "Hiring Hall" / from where the men were sent out to work /1 was two years "chief dispatcher". One year "regular dispatcher" / there is professional dispatcher so that / three years then went (- Oh.).' (14) - Ja sitten /minun isä oli satamatyöläinen /minä menin (...) hänen tilalle töihin sinne ja / (- satamaan?) satamatyöläinen ja "longshore man " mennä amerikankielellä sanottiinja (- Ja-ha.). 'And then / my father was a harborman /1 went (...) in his place to work there and / (- in the harbor?) harborman and to go "longshore man" one said in the American language yes (- Oh I see.).' 2.3.2. Memory processes Memory can be connected to past conversations in the second language which, although monolingual by themselves, when quoted introduce bilingual turns in the present exchange. The psycholinguistic status of these turns needs a thorough investigation as they are rarely limited to the quotations proper. On the contrary, they have a tendency to develop into whole stories: the quotation

The information structure of bilingual meaning

63

serves as a releasing device and the speaker is carried along irresistibly by the language of the memorized situation, as in (15) below. (15) - Ne ovat usein matkalla? 'They are often on travel?' - Ne ovat kyllä.... They are yes .... They are so social! She said that /Adidas /sent them to a soccer meeting in d.... Holland And then they went to .. They came over here to New York, d And they d stayed.... somebody else put them up there [laughs]. And then they went I and stayed/at someone 's home in .... in Long Beach. The switch is frequent in connection with a memorized social situation, and although little investigated, its probability seems to be higher still when there is a strong affective connotation (marked by paralinguistic devices, a shivering voice etc.), as in the following example. (16) - Olitko sinäkin löissä? 'Were you too working?' - Jaa. Minä kävin d / sitten kun lapset oli / isompia että you know etten minä tartte ola kotona (...). Ja sitten olinpois työstä KAKS KYMMENTÄ KUUS VUOTTA /ja sitten menin sinne samaanpaikkaan .. työhön (...). Mulla oli samat d d temp.... you know / upper / the people / were.. my bosses were all the same people that I had known before / except they were / higher up bosses (- Oh really!) So. It was like going back home. They all KNEW me. Oh yes. I was d I later when the children were / older so that you know I don't need to be at home (...). And then I was away from work TWENTY SIX YEARS / and then I went back to the same place .. work (...).! had the same dd .... youknow/(...)me.' The quantitative memory deserves a special chapter: it is a well-known fact that the mechanisms for counting usually get stabilized in one language, but as for the memorized distances and dates, things seem to be more complex. Due to its narrow connection with the contrastive construction of meaning, this topic will be treated in next section.

3.

Cognitive processes and bilingual meaning

This section sketches a few central characteristics of bilingual meaning, compared first with monolingual (3.1) meaning, especially their distinct relevance for a study of quantitative memory (3.1.1) and of Information Structure from

64

M M Jocelyne Fernandez- Vest

a textual perspective (3.1.2), and then with interpreted meaning, a valuable source for the investigation of multilingual communication (3.2). 3.1. Bilingual meaning vs. monolingual meaning Some outstanding features of the quantitative memory are here first exemplified on the basis of a monolingual (Sami) and a bilingual (Finnish Califomian) corpus (3.1.1). I shall then present some properties for the textual analysis of information structure, including prosody (3.1.2). 3.1.1. Social parameters for the quantitative memory On the basis of comparisons of several corpora, collected in diverse language communities, I should like to assert that the quantitative memory has a direct correlation with the speakers' social integration. This partly explains why a sexual difference can often be observed in the speakers' use of memorized calculations. Two examples of this will be given below, one 'exotic' and monolingual, the other bilingual. Sami memory of places and dates In a previous study of Northern Sami women's language and their representation of space and time, I made the following remarks. The identification of places and the archiving of traditions attached to toponymy generally rely on the memorial capacities of the informants recognized as the most active of the Arctic space - the men. Beyond the place-names and their etymology, Samispeaking men have proved to be competent in quantification: the estimation of distances and dates does not seem to be a major problem for them. They frequently and explicitly produce efforts of memory to dig out old dates or evaluate the length of an itinerary proposed by the inquirer; see Fernandez-Vest (1993), for instance, for the long negotiation about a walking distance, a prototype of everyday conversation between old Sami people. The competence of women in equivalent subjects is more insecure: their estimation of space is relatively unreliable. Due to different types of external activities, their efforts of quantification on the whole sound painful. The search for dates in the long term memory often fails after backing itself on intimate landmarks. The family circle, commonly resorted to for recollecting places, is also present in the struggle for assigning time in long term memory - children in particular, as in the following example.

The information structure of bilingual meaning

65

Sami (17) - Dat leaAIGA aiga. Dalle mun ledjen nuorra. =» Mun .... eai lean vel .. na gal dat lei manna. Jo guokte. =» Maireja.. Ravnna-guot Mairein. 'It is long long ago. At that time I was young. =» I.... I had not yet.. well yes I had a child. Two already. =» Maire and .. both Ravnna and Maire.' Another Sami woman, when inquired about some of the mountain-lakes with difficult access, remembers having been there, but after refreshing her memory through the state of growth of the children, she admits that she is not able to date the event with certainty. Sami (18) - Savzajavrres mun leanfitnan. =» Dallehan ledje visot dot ollesmanat .. mun leanfitnan gal dalle muhto mun in diede daid vuoaddloguid. have been at the lake of Sävzajävri. =» By then of course all children were grown up .. I have been there at that time but I don't know the year.' In cognitive terms, if one wants to establish a distinction between men and women as to their perception and their memory of space and time, one is tempted to characterize the representations of women, compared to those of men that are often 'type-representations', as Occurrence-representations'. As for the type of knowledge implied (see Kintsch 1970), what appears in the discourse of women about space is procedure knowledge; men, on the contrary, are more at ease in presenting declaration knowledge (for instance linked to specific practices of reindeer-care or fishing). The model of representation of space, as it is built up by the explicit memory (Paradis 1994) of these women informants (linguistically marked by a search for familiar landmarks, as we saw) is strictly issued from the personal experience and refers to their know-how more than to an abstract knowledge (Fernandez-Vest 1996, 1998b). Quantitative memory and Californian Finnish CoSwit An audacious return to the Californian Finnish situation brings out complementary aspects of a socially anchored sexual difference. The question is whether older women with a Finnish origin reveal their uncertainty in memorized quantification. My analysis shows that the situation here is not exactly the same: they do remember places and dates, but they have a marked tendency to quote them in the majority language, English - which is generally not the case for men. Consider, for instance, the following example.

66

MM. Jocelyne Fernandez-Vest

(19) - Milloin vanhempasi tulivat Amerikkaan? 'When did your parents come to America?' d Minu ISA.... lähti merelle /kun hän oli kuudentoista vuotias. (...) Mutta minäen tiedäkoska hän/juuri tuli/Amerikkaan. Hän olid.... nineteen nine (—Aah.) hän oli jo täällä. (...) Sitten minun veil syntyi siellä vuosi nineteen fifteen [coughs] anteeksi /ja / sitten ne lähti täällä San Franciscoon. Ja minä synnyin täällä sitten nineteen seventeen. (...) Minä menin / nun kaikki nuoret silloin meni / Veljesseuran Hallille (- Hm-m). Siellä oli kaikki suomalaisia. (...) Siellä minä tapasin Reinon. Ja 3.... me menimme naimisiin 8.... nineteen thirty-eight /ja me olemme ollut.... viisikymmentä.... pian ensi kuussa viisikymmentä seitsemän vuotta /naimisissa (- Ja-ah.). (...) Ja/sitten mä.... minäd/yhdyin /d Sisarseuraan/d vuosi nineteen forty. Minä sanon noita englanninkielellä kun se on sokelampi minulle. 'My FATHER .... went on sea / when he was sixteen years. (...) But I don't know when he / exactly came / to America. He was d .... nineteen nine (- Oh yes.) he was aready here. (...) Then my brother was born there in nineteen fifteen [coughs] excuse me / and / then they left here to San Francisco. And I was born here then nineteen seventeen. (...).! went / well all young people then went / to the Brotherhood Hall. (- Hm-m). There they were all Finnish. (...) There I met Reino. And d .... we got married d .... nineteen thirty-eight / and we have been .... fifty .... soon next month fifty-seven years / married. (- Oh yes.). (...) And / then I .... I d /joined / d the Sisterhood / d in nineteen forty. I say them in English as it is easier to me.' It is interesting to note that this female speaker usually counts in Finnish, despite the specific difficulties of the Finnish number morphology. In the first utterance, she fails partly, (over)declining the numeral determinant (kuuden/toista instead of kuusi/toista, as vuotias 'aged' does not require a genitive but a nominative). Later she succeeds in surmounting the obstacle of the double partitive construction in viisi/kymmentä (partitive, required by the compound 'five-ten') seitsemän vuotta (partitive, required after a numeral) 'fifty-seven years'. But the dates have been memorized, or at least they are delivered automatically, in their invariable English form. There must be some deeper explanation than the facility (sokelampi) which she finally refers to.

The information structure of bilingual meaning

67

3.1.2. A textual approach to Information Structure As long as the analysis is restricted to short utterances, it is not difficult to show that there is a preferential correspondence between certain information units and the use of the two competing languages in a bilingual situation. Recall the examples (5), (6), (7) and (8), and also (12), (13), (14) and (15), where the source of CoSwit even includes a quotation and/or a metalinguistic activity or a preferred language for quantitative memory. But Impromptu Speech involves long utterances as soon as one goes beyond conducted interviews. That is why linguists working on authentic corpora are tempted to adopt a textual approach, which I did from my first study on Samiinfluenced Finnish (cf. 1.1.1, above). This has some consequences for the Information Structure units, both as to their definitions and their perception in natural discourse (see also Carter-Thomas 1999, 2000; Niska 1999, 2000). The three main units that have been referred to here, i.e. Theme, Rheme, and Mneme, can thus be not only words or phrases but, as is frequently the case in discourse, whole clauses. This can be illustrated by a set of French examples. (20a) Independent clause: Sa mere \ s 'etait egaree dans laforet 'Her mother | had gone astray in the forest' Th Rh (20b) Subordinate clause: Apr es que sa mere s 'etait egaree dans laforet / lafille a ete placee a I Orphelinat 'After her mother had gone astray in the forest / the girl was put into an orphanage' (20b') Apres que sa mere ...foret | lafille ... a I Orphelinat | Th Rh (20b") [Quand est-ce que lafille a ete placee a I Orphelinat?} [When was the girl put into an orphanage?] Apres que sa mere ...foret \, lafille ...a I Orphelinat —» Rh Mn With different intonation patterns (f rising, J, falling, —» flat), the complex clause can achieve different strategies, i.e. the binary strategy 1 (20b') or the binary strategy 2 (20b"). The complex clause in (20b") serves as an answer, where the Mneme, as often, repeats some element of the question (here the whole clause).

68

M. M. Jocelyne Fernandez- Vest

Contrasted with Finnish, this example enlights a specificity of the Finnish language: the use of dense synthetic constructions, the so-called 'quasipropositions' (verbo-nominal equivalents to the Indo-European subordinate clauses in several Uralic languages), rare in impromptu speech but commonly used in written discourse: Finnish (21 a) Kun äiti oli eksynyt metsään | tyttöjoufui lastenkotiin \, 'After the mother had got lost in the forest, the girl was put into an orphanage' (21b) Aidin mother-of(gen.)

eksyttyä... after the "having-been got lost" (passive past participle partitive) 'After the mother's "got lost"

The use of a quasiproposition in (21b) has two main implications for the first and especially the third Information Structure unit: (1) As the topical sequence is fixed in the synthetic quasiproposition (an exception to the Finnish rule of '(syntactically) free word order'), the possibilities of marking the Topic are limited in such a normative style. (2) The Mneme-segment (possibly a whole clause) can be separated from the Rheme-segment (possibly a whole clause) by other additions, contrary to one of the conditions required for Antitopic ('strict localization', see 1.1.2, above). Consider example (22) below, which is an excerpt from (19), above. (22) d Minu ISA.... lähti merelle/kun hän oli kuudentoista vuotias. (...) Mutta minä en tiedä koska hän /juuri tuli /Amerikkaan. Mn 'My FATHER .... went on sea / when he was sixteen years. (...) But I don't know when he / exactly came / to America.' The recognition of the last information unit (Amerikkaan) as a Mneme explains that, although we remarked in 2.2. upon a frequent correlation between CoSwit and rhematic place-names, a CoSwit is practically excluded in the last position in (22). The final place-name is a mere repetition, with a flat intonation, of the question-topic. Another difference is the content itself. We admit that right-detachment (Mneme), unlike left-detachment (Topic) cannot indicate a new topic or a topic shift (Lambrecht 1994: 204); it can nevertheless modify, make more precise or reduce the content of the Topic, as in (23) below, which is an excerpt from (12), above.

The information structure of bilingual meaning

69

(23) Siellä oil PALJONkin. =» Tämäperukka missä me asumme / nyt / oli paljon suomalaisia täällä oli. 'There were MANY in fact. =»This block where we live / now / there were many Finns here yes.' The first two utterances of the answer, linked together by a quick tempo (=»), can be characterized as a 'multiple answer' (Fernandez-Vest 1987a: 443-460). Interesting is the speaker's slight shift of view-point. The answer stricto sensu (first utterance) is a mirror image of the question (same mode, same deictic pronoun) reinforced by the Discourse Particle -kin ('also', 'in fact', 'indeed'). But the second link of this chain answer brings forward an appropriation by the speaker of the facts referred to: the 'working-place' > 'this () where we live', backed by a partial repetition of the first two words, now moved into the speaker's personal deictic sphere (see Fernandez-Vest 2000b). On the basis of examples of this type, I should like to claim that a mere syntactic approach to Information Structure is inadequate for the analysis of Impromptu Speech, as it obscures, if not totally excludes, some of the fundamental processes of ordinary speech: in this case the circular cohesion present in (22) and (23) with some differences (simple repetition vs. personal appropriation). To the differences between Mneme and Antitopic already mentioned in 1.1.2. above, I should thus like to add a generally conceptual one: referents are not automatically activated vs. non-activated, since the use of the two binary strategies, and the possibility for them to combine, reflects a scalar rather than dichotomic conception of information (Fernandez-Vest 1994a: 84-85, 193-194). 3.2. Bilingual vs. interpreted meaning Another angle of approach to the bilingual construction of meaning is the research in progress about Interpretation. As some recent investigations have shown, Simultaneous Interpretation (SI) is distinct from other forms of speech behaviour in combining several features which preclude direct extrapolation from experiment and observation in other disciplines, but at the same time make SI data a variable source of inference in the study of language communication. Among the main characteristics (goal orientation, external sourcing, and so forth) of this still scantly investigated linguistic activity is the fact that input and output are in different languages: interpretation aims to maintain the message (both the prepositional content and intentionality) while changing the code (Setton 1999: 2, 7-1, 58-60).

70

M M. Jocelyne Fernandez- Vest

The two main directions chosen within a Franco-Nordic text-linguistic project on 'Interpretation and bilingual cognition' initiated in 2001 do have a strong connection with the questions raised above about CoSwit: (1) Can the difference of word order (cf. the synthetic constructions of Finnish, see (21) above) be correlated to the Effort model elaborated by several theoreticians of Simultaneous Interpretation (Gile 1995a, 1995b, 1997)? (2) Can the practice of interpreters regarding difficult items like numbers be assimilated to their reaction to 'innovations' (new terms, proper nouns)? (See Fernandez-Vest 200 Ib; Niska and Wände, in press.) The combined observation of these two socially and economically distant types of communication, Impromptu bilingual Speech and Simultaneous Interpretation, should be able to increase our knowledge of the information structure of meaning as conceived, perceived and/or translated by a bilingual brain.

Note 1. This chapter is a synthesis of four papers presented at recent conferences: "La construction thematique du sens, du discours unilingue au discours bilingue" (Fernandez-Vest 2000a), "Morphogenesis in discourse as a new approach to code-switching" (ICLC'99, Stockholm, 10-16 July 1999, thematic session 'Discourse Approaches to Cognition'), "Cognitive processes and the construction of bilingual meaning: the case of Finnish spoken in California" (FU9, Tartu, 7-13 August 2000; Fernandez-Vest 200la), "Cognitive processes in the construction of interpreted vs. bilingual meaning" (ICLC 2001, Santa Barbara, California, 22-27 July 2001; Fernandez-Vest 200Ib). I am grateful to several other participants for their remarks and comments, specially Ad Foolen (Nijmegen), Jyrki Kalliokoski (Helsinki), Maisa Martin (Jyväskylä), Lorenza Mondada (Bale, Lyon), Jan-Ola Östman (Helsinki), Len Talmy (Buffalo), Tuija Virtanen (Växjö), Erling Wände (Stockholm).

References Arleo, Andy and M.M.Jocelyne Fernandez-Vest 2004 De quelques particules anglaises et francaises dans le dialogue cinematographique: Hannah and her sisters I Hannah et ses sceurs de Woody Allen. In: M.M.J. Fernandez-Vest and S. Carter-Thomas (eds.), Structure informationnelle et Particules Enonciatives: Essai de typologie, 243-285. (Grammaire & Cognition 1-2.) Paris: L'Harmattan. Auer, Peter 1984 Bilingual Conversation. Amsterdam / Philadelphia: John Benjamins.

The information structure of bilingual meaning

71

Baetens-Beardsmore, Hugo 1982 Bilingualism: Basic Principles. Clevedon: Multilingual Matters. Carter-Thomas, Shirley 1999 La Strategie thematique: son importance dans Panalyse textuelle. In: Actes de la Seme Journee de la Formation Doctorale, 49-64. Departement de Linguistique Generale et Appliquee, Universite Rene-Descartes/Paris V. 2000 La coherence textuelle: Pour une nouvelle pedagogie de l 'ecrit. (Langue et Parole: Recherche en sciences du language.) Paris: L'Harmattan. Chafe, Wallace L. 1976 Givenness, contrastiveness, definiteness, subjects, topics and points of view. In: C.N. Li (ed), Subject and Topic, 25-55. New York: Academic Press. 1987 Cognitive constraints on information flowing. In: R. Tomlin (ed.), Coherence and Grounding in Discourse, 21-52. Amsterdam: John Benjamins. 1994 Discourse, Consciousness and Time. Chicago / London: The University of Chicago Press. Danes Frantisek 1974 Functional sentence perspective and the organisation of the text. In: F. Danes (ed.), Papers in Functional Sentence Perspective, 106-128. Prague: Academia. Dik, Simon C. 1978 Functional Grammar. (Linguistic series, 37). Amsterdam / New York / Oxford: North-Holland. Enkvist, Nils Erik (ed.) 1982 Impromptu Speech: A Symposium. (MSÄAF 78.) Abo: Abo Akademi. 1985 Coherence and Composition: A Symposium. (MSÄAF 101.) Abo: Abo Akademi. Erguvanli, Eser 1984 The Function of Word Order in Turkish Grammar. (University of California Publications in Linguistics 106.) Berkeley / Los Angeles: University of California Press. Fernandez-Vest, M.M.J. 1982 Le finnois parle par les Sames bilingues d'Utsjoki-Ohcejohka (Laponie finlandaise): Structures contrastives, syntaxiques, discursives. (L'Europe de Tradition Orale 1.) Paris: SELAF. 1987a La Finlande trilingue l: Le discours des Sames - Oralite, contrastes, enonciation. Paris: Didier Erudition. 1987b Analyse contrastive du discours: vers un modele enonciatif franco-finnois. In: O. Välikangas (ed.), Actes du Hie colloque franco-finlandais de linguistique contrastive, 200-217. Publications du Departement des Langues Romanes 7, Universite de Helsinki.

72

M.M. Jocelyne Fernandez-Vest 1989

1993

1994a 1994b

1995a

1995b

1996

1998a

1998b

2000a

2000b

Eire bilingue en Fenno-Scandie: un modele pour l'Europe? In: F. de Sivers (ed.), Questions d'identite, 77-112. (Sociolinguistique 4.) Paris: PEETERSSELAF. Cognition spatiale, oralite, passage ä l'ecriture: I'exemple du same (Laponie). Actes du Colloque Interdisciplinaire du Comite National de la Recherche Scientifique 'Images et langages: Multimodalite et Modelisation cognitive', Paris, C.N.R.S., 1-2 avril, 1993. 161-172. Les Particules Enonciatives dans la construction du discours. (Linguistique nouvelle.) Paris: Presses Universitaires de France. Particules Enonciatives et competence dialogique: pour construire l'interlangue. In: J. Schön (ed.), Actes du Colloque 'Le dialogue en question /Questioning dialogue', Toulouse-Lagrasse, 5-8 October, 1993. Cahiers du Centre Interdisciplinaire des Sciences du Langage 10: 409-420. Morphogenese orale du sens: de l'espace des langues aux objets de discours. In: M.M.J.Fernandez-Vest (ed.), Oralite: invariants enonciatifs et diversite des langues. Intellectica 1/20: 7-53. Tintin bilingue ou l'animation contrastive des categories enonciatives. In: O. Valikangas and M. Helkkula (eds.), Actes du Ve colloque franco-finlandais de linguistique contrastive, 35-52. Publications du Departement des Langues Romanes 9, Universite de Helsinki. Deixis and Cognition in the Construction of Sami and Balto-Finnic (SBF) Discourses. In: H. Leskinen, S. Maticsäk and T. Seilenthal (eds.), FU 8: Congressus Octavus Internationalis Fenno-Ugristarum, Jyväskylä, 10-15 August, 1995. Jyväskylä: Gummerus. Pars IV, Contactus linguistici et status hodiernus linguarum et cetera linguistic'. 23-27. Morphogenese orale du sens: de l'espace au theme. In: M.M.J. Fernandez (-Vest) (ed.), Table ronde/Workshop 'Cognition incarnee, cognition situee? Le temoignage des langues orales - Embodied cognition or situated cognition? The testimony of oral languages'. In: Actes du XVIe Congres International des Linguistes (CIL 16.). Elsevier. CD-ROM. Femmes de Laponie: migrante memoire, 1'arctique au quotidien. In: M.M.J. Fernandez (ed.), Parier femme en Europe - Lafemme, image et langage, de la tradition ä l Oral quotidien, 209-250. Paris: L'Harmattan. La construction thematique du sens, du discours unilingue au discours bilingue. In: L. Mondada and A.C. Berthoud (eds.), Modeies du discours en confrontation, 131-150. (Sciences de la communication 61.) Bern / Berlin / Bruxelles: Peter Lang. Deixis, interaction, grammaticalisation: le cas des particules enonciatives en same du Nord et en finnois. In: M.M.J. Fernandez-Vest (ed.), Grammaticalisation areale et semantique cognitive: les langues fenniques et sames - Areal Grammaticalization and Cognitive Semantics: the Finnic and Sami Languages. Actes du Colloque International du C.N.R.S. tenu

The information structure of bilingual meaning

73

les 9 et 10 avril 1999 en Sorbonne, Tallinn, Eesti keele sihtasutus, 65-80. (Oural-Ural 1.) 2000c Quel avenir pour une langue de 1'oralite: la langue same, memoire finnoougrienne de 1'Europe? In: Les langues en danger. Memoires de la Societe de Linguistique de Paris, Nouvelle serie, Tome VIII, 121-136. Paris: Peelers. 200 la Cognitive processes and the construction of bilingual meaning: the case of Finnish spoken in California. In: FU 9. Congressus Nonus International Fenno-Vgristarum, Tartu, 7-13 August, 2000, 237-246. Tartu: Trükk Oü Paar. 200 Ib Cognitive processes in the construction of interpreted vs. bilingual meaning. Paper presented at ICLC 2001, Santa Barbara, California, 22-27 July 2001. 2002 Typology, pragmatics, standardization: will Northern Sami survive as Finno-Ugric? In: R. Blokland & C. Hasselblatt (eds.), Finno-Ugrians and Indo-Europeans: Linguistic and literary contacts, Proceedings of the Symposium at the University of Groningen, November 22-24, 2001, 64-71. (Studia Fenno-Ugrica Groningana 2.) 2003 Semantique cognitive et diversite des langues: de quelques orientations en sciences du langage francophones. In: Actes des Journees franco-russes de Linguistique, Saint-Petersbourg 25 octobre - 4 novembre 2000. Societe d'Histoire et d'Epistemologie des Sciences du Language. 2004a Mnememe, Antitopic - le post-Rheme, de Penonce au texte. In: M.M.J. Fernandez-Vest and S. Carter-Thomas (eds.), Structure informationnelle et Particules Enonciatives: Essai de typologie. (Grammaire & Cognition 1-2.) Paris: L'Harmattan: 65-104. 2004b Information structure and typological change: Northern Sami challenged by Indo-European models. In: M.M.J. Fernandez-Vest (dir.), Les langues ouraliennes aujourd'hui: approche linguistique et cognitive - The Uralic Languages today: a cognitive and linguistic approach, Preface de Claude Hagege. (Bibl. de l'Ecole des Hautes Etudes ä la Sorbonne.) Paris: Ed. Honore Champion: 563-576. Fernandez-Vest, M.M.J. (ed.) 1995 Oralite: invariants enonciatifs et diversite des langues. Intellectica 1/20. 1998 Table ronde/Workshop 'Cognition incarnee, cognition situee? Le temoignage des langues orales - Embodied cognition or situated cognition? The testimony of oral languages'. Actes du XVIe Congres International des Linguistes (CIL 16.) Elsevier. CD-ROM. 2000 Grammaticalisation areale et semantique cognitive: les langues fenniques et sames — Areal Grammaticalization and Cognitive Semantics: the Finnic and Sami Languages. Actes du Colloque International du C.N.R.S. tenu les 9 et 10 avril 1999 en Sorbonne, Tallinn, Eesti keele sihtasutus. (OuralUral 1.)

74

M M Jocelyne Fernandez- Vest

Fuchs, Catherine and Stephane Robert (eds.) 1997 Diversite des langues et representations cognitives. Paris / Gap: Ophrys. 1999 Language Diversity and Cognitive Representations. (Human Cognitive Processings 3.) Amsterdam / Philadelphia: John Benjamins (English revised edition). Gile, Daniel 1995a Regards sur la recherche en interpretation de conference. Lille: Presses universitaires de Lille. Gile, Daniel (ed.) 1995b Interpreting Research. Special issue. Target 111. Gile, Daniel 1997 Conference interpreting as a cognitive management problem. In: Joseph E. Danks, Gregory M. Shreve, Stephen B. Fountain and Michael K. McBeath (eds), Cognitive Processes in Translation and Interpreting, 196-214. Thousand Oaks / London / New Delhi: Sage. Guimier, Claude (ed.) 1999 La thematisation dans les langues. Actes du Colloque de Caen, 9-11 octobre 1997. Bern / Berlin: Peter Lang. Hagege, Claude 1978 Du theme au theme en passant par le sujet. Pour une theorie cyclique. La linguistique 14/2: 3-38. 1993 The Language Builder: An Essay on the Human Signature in Linguistic Morphogenesis. (Amsterdam Studies in the Theory and History of Linguistic Science Ser. IV, Current Issues in Linguistic Theory 94.) Amsterdam / Philadelphia: John Benjamins. 1996 L 'enfant aux deux langues. Paris: Odile Jacob. 2000a Les langues ä mi-chemin du cognitif et du social: pressions des systemes et operations humaines (exemples fenniques). In: M.M.J. Fernandez-Vest (ed.), Grammaticalisation areale et semantique cognitive: les langues fenniques et sames - Areal Grammaticalization and Cognitive Semantics: the Finnic and Sami Languages. Actes du Colloque International du C.N.R.S. tenu les 9 et 10 avril 1999 en Sorbonne, Tallinn, Eesti keele sihtasutus, 205-210. (Oural-Urall.) 2000b Halte a la mart des langues. Paris: Odile Jacob. Hakulinen, Auli 1989 Keskustelun luonnehtimisesta konteksti- ja funktionaalisten tekijöiden nojalla. In: A. Hakulinen (ed.), Suomalaisen keskustelun keinoja /, 41-72. (Kieli 4.) Department of Finnish, University of Helsinki. Halmari, Helena 1997 Government and Codes-witching: Explaining American Finnish. (Studies in Bilingualism 12.) Amsterdam / Philadelphia: John Benjamins. Halmari, Helena and Wendy Smith 1994 Code-switching and register shift: Evidence from Finnish-English child bilingual conversation. Journal of Pragmatics 21/4: 427-445.

The information structure of bilingual meaning

75

Hansegärd, Nils 1979 Kaksikielisiä vai puolikielisiä. (Tietolipas 81.) Helsinki: Suomalaisen Kirjallisuuden Seura. Hopper, Paul J. 1987 Emergent grammar. Berkeley Linguistic Society 13: 139-157. 1998 Emergent grammar. In: Michael Tomasello (ed.), The New Psychology of Language: Cognitive and Functional Approaches to Language Structure. Mahwah, NJ: Lawrence Erlbaum. Huss, Leena 1999 Reversing Language Shift in the Far North: Linguistic Revilalization in Northern Scandinavia and Finland. (Studia Uralica Upsaliensia 31.) University of Uppsala. Jönsson-Korhola, Hannele 1980 On the syntax of the second generation American Finns. In: Congressus Quintus Internationalis Fenno-Ugristarum 3B, Summa dissertationum. Turku: Suomen kielen seura. 1987 Amerikansuomen kaasussyntaksin erityispiirteitä. Licentiate Thesis, Department of Finnish, Abo Akademi University. 1990 American Finnish syntax: Some observations on the use of cases in second generation speech. In: Congressus Septimus Internationalis Fenno-Ugristarum 3B, Sessiones Sectionum, Dissertationes Linguistica, 11-11. Debrecen. Kalliokoski, Jyrki 1995 Koodinvaihtojakeskustelun moniäänisyys. Vmttäjä 1: 2-24. Karttunen, Frances 1977 Finnish in America: A case study in monogenerational language change. In: Ben G. Blount and Mary Sanders (eds.), Socio-cultural Dimensions of Language Change, 173-185. New York: Academic Press. Kero, Reino 1986 Suomen siirtolaisuuden historia I. (Publications of the Department of History 10.) University of Turku. Kintsch, W. 1970 Models for free recall and recognition. In: D.A. Norman (ed.), Models of Human Memory. New York: Academic Press. Kuno, Susumo 1978 Japanese: A characteristic OV language. In: Winfred P. Lehmann (ed.), Syntactic Typology: Studies in the Phenomenology of Language, 58-138. Austin, TX: University of Texas Press. Lambrecht, Kiuid 1981 Topic, Anti-Topic, and Verb Agreement in Non Standard French. Amsterdam / Philadelphia: John Benjamins. 1994 Information Structure and Sentence Form: Topic, Focus and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Press.

76

MM Jocelyne Fernandez-Vest 1998

La relation formelle et fonctionnelle entre topiques et vocatifs. Cahiers d'etudes et de recherches francophones 1/1: 34—45. Langacker, Ronald 2000 A dynamic usage-based model. In: M. Barlow and S. Kemmer (eds.), Usage-Based Models of Language, 1-63. Stanford: CSLI Publications. 2001 Discourse in Cognitive Grammar. Cognitive Linguistics 12/2: 143-188. Martin, Maisa 1988 The role of code-switching among Canadian Finnish-English bilinguals. In: Michael G. Kami, Olavi Koivukangas and Edward W. Laine (eds.), Finns in North America, Proceedings of Finn Forum III (5-8.IX.1984), 366-377. (Migration Studies 9.) Turku: Institute of Migration. 1989 Amerikansuomen morfologiaa ja fonologiaa. Licentiate Thesis, Department of Finnish, University of Jyväskylä. 1990 On the morphological distribution of English loan verbs in American Finnish. In: Congressus Septimus Internationalis Fenno-Vgristarum 3B, Sessiones Sectionum, Dissertationes Linguistica, 68-71. Debrecen. 1993 Amerikansuomi kielentutkimuksen kentässä. In: Pertti Virtaranta, H. Jönsson-Korhola, M. Martin and M. Kainulainen (eds.) Amerikansuomi, 190197. (Tietolipas 125.) Helsinki: Suomalaisen Kirjallisuuden Seura. Michaelis, Laura and Knud Lambrecht 1996 Toward a Construction-based Theory of Language Function: The case of Nominal Extraposition. Language 72/2: 215-247. Muysken, Pieter 2000 Bilingual Speech: A Typology of Code-mixing. Cambridge: Cambridge University Press. Myers-Scotton, Carol 1988 Codeswitching as indexical of social negotiations. In: M. Heller (ed.), Codeswitching: Anthropological and Sociolinguistic Perspectives, 151186. Berlin: Mouton de Gruyter. 1992 Comparing Codeswitching and borrowing. Journal of Multilingualism and Multicultural Development 13/1-2: 19-39. 1993 Duelling languages: Grammatical Structure in Codeswitching. Oxford: Clarendon Press. Niska, Helge 1999 Textual processes and strategies in simultaneous interpreting. Licentiate Thesis, Department of Finnish, Stockholm University. 2000 Text linguistic models for the study of simultaneous interpreting. In: Ingo Warnke (ed.), Schnittstelle Text - Diskurs. Frankfurt: Peter Lang. Niska, Helge and Erling Wände In press Text linguistics, translation theory and simultanous interpreting: Models for the analysis of simultanously interpreted LSP-discourse. In: Brynja Svane (ed.), Translation and Intercultural Communication: Reports from the Research programme 'Translation and Interpreting - A meeting between cultures'. University of Stockholm.

The information structure of bilingual meaning

77

Paradis, Michel 1978 The stratification of bilingualism. In: M. Paradis (ed.), Aspects of Bilingualism, 166-175. Columbia: Hornbeam Press. 1994 Neurolinguistic aspects of implicit and explicit memory: Implications for bilingualism. In: N. Ellis (ed.), Implicit and Explicit Language Learning, 393^419. London: Academic Press. Pierce, Charles 1932 Collected Papers. Cambridge, MA: Harvard University Press. Pietilä, Päivi 1989 The English of Finnish Americans with reference to social and psychological background and with special reference to age. (Turun yliopiston julkaisuja B 188.) University of Turku. Poplack, Shana 1980 Sometimes I start a sentence in Spanish TERMING EN ESPAGNOL: Toward a typology of code-switching. Linguistics 18: 581-618. Poplack, Shana, Susan Wheeler and Anneli Westwood 1987 Distinguishing language contact phenomena: Evidence from Finnish-English bilingualism. In: Pirkko Lilius and Mirja Saari (eds.), The Nordic Languages and Modern Linguistics 6: Proceedings of the Sixth International Conference of Nordic and General Linguistics in Helsinki, 18-22 August, 1986, 21-56. Helsinki: Helsinki University Press. Romaine, Suzanne 1989 Bilingualism. (Language in Society.) Oxford: Basil Blackwell. Saari, Mirja 1989 Kodbyte och län. Folkmälsstudier 32: 197-209. Sahlman-Karlsson, Siiri 1976 Specimens of American Finnish: A field study of linguistic behaviour. Uppsala: Acta Universitatis Upsaliensis. Setton, Robin 1999 A Cognitive-Pragmatic Analysis of Simultaneous Interpretation. Amsterdam / Philadelphia: John Benjamins. Stroud, Christopher 1992 The problem of intention and meaning in code-switching. Text 12: 127155. Talmy, Len 2000a Grammar from the perspective of cognitive semantics. In: M.M.J. Fernandez-Vest (ed.), Grammaticalisation areale et semantique cognitive: les langues fenniques et sames - Areal Grammaticalization and Cognitive Semantics: the Finnic andSami Languages. Actes du Colloque International du C.N.R.S. tenu les 9 et 10 avril 1999 en Sorbonne, Tallinn, Eesti keele sihtasutus, 17-26. (Oural-Ural 1.) 2000b Toward a Cognitive Semantics. Vols. 1-2. Cambridge: Cambridge University Press.

78

M.M.Jocelyne Fernandez- Vest

Thompson, Sandra A. 2002 Object complements' and conversation: Towards a realistic account. Studies in Language 26/1: 125-163. Thompson, Sandra A. and Paul J. Hopper To appear Transitivity, clause structure, and argument structure: Evidence from conversation. In: Joan B. Bybee and Paul J. Hopper (eds.), Frequency and the Emergence of Linguistic Structure. Amsterdam / Philadelphia: Benjamins. Valdduvi, Enric 1990 The informational component. Ph.D. dissertation, University of Pennsylvania. Virtaranta, Pertti 1971 Finskan i Amerika. Spräk i Norden, 79-109. 1981 Finnish dialects in America: Some experiences and problems. In: Michal G. Kami (ed.), Finnish Diaspora II, 303-316. Toronto: The Multicultural Society of Ontario. 1988 Considerations on American Finnish, particularly its vocabulary. In: Michael G. Kami, Olavi Koivukangas and Edward W. Laine (eds.), Finns in North America: Proceedings of Finn Forum III (5-8 September, 1984), 355-366. (Migration Studies 9.) Turku: Institute of Migration. 1992 Amerikansuomen sanakirja. A Dictionary of American Finnish. Turku: Siirtolaisinstituutti. Virtaranta Pertti, Hannele Jönsson-Korhola, Maisa Martin and Maija Kainulainen 1993 Amerikansuomi. (Tietolipas 125.) Helsinki: Suomalaisen Kirjallisuuden Seura. Wände, Erling 1984 Two Finnish minorities in Sweden. Journal of Multilingualism and Multicultural Development 3-4: 225-241. 1996 Bilinguisme finnois-suedois dans la Vallee du Torneä: Aspects linguistiques et ecologiques. In: M.M.J. Fernandez-Vest and Raimo Raag (eds.), Contacts de langues et de cultures dans I'aire baltique - Contacts of Languages and Cultures in the Baltic Area: Melanges offerts a Fanny de Sivers, 335-348. (Uppsala Multiethnic Papers 39.) Uppsala: Acta Universitatis Upsaliensis. 2000 Spatial relations in Finnish and Swedish: Considerations from a cognitive point of view. In: M.M.J. Fernandez-Vest (ed.), Grammaticalisation areale et semantique cognitive: les langues fenniques et sames - Areal Grammaticalization and Cognitive Semantics: the Finnic and Sami Languages. Actes du Colloque International du C.N.R.S. tenu les 9 et 10 avril 1999 en Sorbonne, Tallinn, Eesti keele sihtasutus, 43-62. (Oural-Ural 1.) Weinreich, Uriel 1953 Languages in Contact: Findings and Problems. New York: Linguistic Circle of New York.

Chapter 5 Point of departure: Cognitive aspects of sentence-initial adverbials Tuija Virtanen

In his 1981 article Levelt discusses what he refers to as the 'speaker's linearization problem' - a highly relevant issue in both cognitive linguistics and discourse linguistics. One of the major problems in this field is the choice of a starting point. Linearization problems exist both at the level of the sentence and larger units of text, as well as at the level of the entire text or group of texts. As shown by Enkvist (1981), linearization at the sentence level is affected by the text in which the sentence appears. This chapter approaches cognitive aspects of linearization in discourse in the light of adverbial placement. More specifically, the discussion deals with cognitive motivations for sentence-initial placement of adverbials of various kinds, i.e. their placement before the subject of the clause or sentence. It will be shown that sentence-initial position is particularly interesting from a cognitive point of view. To start with, the focus is on the prototypical adverbials of time, place, and manner, which form part of the propositional meaning of the sentence. Of these, adverbials of time and place appear in the sentence-initial slot more frequently than do adverbials of manner, which can partly be accounted for on cognitive grounds. Secondly, attention is also to some extent paid to adverbials which are syntactically peripheral in the sentence in which they appear, i.e. connectors and adverbials indicating the text producer's beliefs and attitudes concerning the propositional content. Unlike sentence-initial adverbials of time, place and manner, these have become professionalized in the tasks that they serve in text organization and expression of affect, respectively. The degree of professionalization of the adverbial in performing given discourse functions influences its placement in the sentence, and this placement can be to an extent related to cognitive concerns. All realization types of adverbials - i.e. adverbs, prepositional phrases, noun phrases, and clauses - are included in the analysis. The data consist of written discourse, where the text producers have had time to give some attention to their discourse strategy - conscious or not - and its exponents in the text. The types of text include (i) narrative and (ii) description, which have been shown

80

Tuija Virtanen

to manifest chains of sentence-initial adverbials of time and place, (iii) instructions, where manner adverbials, too, are found introducing sentences, and finally (iv), non-narrative texts proper, in which the discourse strategy is commonly based on various thought processes concerning the discourse topic, such as the taxonomy of its parts or other relations between the abstractions at hand. While most authentic texts are multitype, manifesting embeddings of several independent types and various degrees of hybridization of different types, it is assumed that as language users we share a small number of prototypical types of text or discourse which can be characterized using text-internal and cognitive criteria. Hence we are continuously able to make flexible use of these prototypes, which affects their development through time (for further discussions of these and related issues, see Virtanen 1992a, 1992b). In what follows the concern will first be with the sentence-initial slot. After a brief discussion of why this position is interesting from a cognitive perspective (section 1), the focus will be on sentence-initial adverbials of time and place signalling text strategy in narrative, descriptive and instructive texts (section 2). From their characteristics we move on to iconic concerns related to these and other initially placed adverbials (section 3), and then proceed to briefly examine sentence-initial adverbials of time, place and manner conveying crucial information in instructive texts (4). Finally, other adverbials appearing in the initial slot, predominantly in non-narrative texts, will be dealt with from a cognitive perspective (sections 5 and 6).

1.

Sentence-initial position

Sentence-initial position is interesting cognitively for a number of reasons. To start with, the element placed here can be given the job of tying what is to come to what can be assumed to be present in the text world that readers are constructing on the basis of the text, its context, and their knowledge of the world. This suggests that the reader will expect given or inferrable information to appear in this slot (cf. Enkvist 1989; Prince 1981; Virtanen 1992a). Readeroriented writers will thus tend to fill this slot with material that, in Chafe's (1994) terminology, they assume to be active (or semi-active) in the reader's mind at that stage of text processing - or alternatively, they will signal deviation from the given-new order through linguistic devices of various kinds available to them in the context, such as marked focus in speech or the cleft construction in writing (cf. Enkvist 1984). Starting points are assumed to be light, small in size, and consist of given information. The reader's main inferencing effort is expected to take place later in the sentence, and that is where

Point of departure: Cognitive aspects of sentence-initial adverbials

81

new information to be integrated into the text world under construction therefore usually appears. Secondly, elements placed at the outset of a sentence also help readers anticipate what is to come as they pinpoint what the sentence is about and how it relates to the discourse topic. In conveying topical information, sentence-initial elements thus leave the end of the sentence free for commentized material, which is assumed to include weighty, new information. In addition to topical information proper, other elements can appear in initial position for similar reasons: in light of the flow of information running through the text, it can be expedient to position particular elements late in the sentence while sentence-initial material can be relied on to remain active in the working memory for some, albeit very short time. Furthermore, it is occasionally profitable to start with what is regarded as 'crucial information' (Enkvist 1989) in the context, irrespective of whether this information is treated as given or not. Examples of decisions to place crucial information in thematic position can be found in impromptu speech; in writing, instructive texts tend to make use of this strategy, as will be shown in section 4 below. Östman and Virtanen (1999) argue that in each of the three pairs of information-structural notions, i.e. 'theme' - 'rheme', 'topic' - 'comment', and 'given' - 'new', one of the two notions can be seen as Figure, against the Ground of the other. In the first pair, Figureness falls on 'theme' - a positional notion, which constitutes the prominent starting point of the sentence. In contrast, in the case of the other two pairs, it is the second notion that is Figure-like, i.e. 'comment', in the interactional sense of what is said about a topic, and 'new information' in the cognitive sense of the focus of the sentence, which is not assumed to be previously activated in the interlocutor's memory. Interlocutors are thus expected to be looking for the goal of the message in this second part. The sentence-initial position is of special interest because the element placed here is thematic and therefore informationally foregrounded at the stage at which it appears in the text. Subsequently, of course, it will merge with the Ground of the now Figure-like 'comment' and 'new information' - if, as in the default instance, none of these coincides with the 'theme'. The three pairs are distinguished on the basis of whether they rely on position (theme-rheme), aboutness (topic-comment), or focus in the sentence (given-new information). While none of the pairs is purely oriented towards the text-producer or the text-receiver, 'theme' can be argued to represent a more producer-oriented notion than the concepts constituting the other two pairs. In this light, there is reason to postulate that the sentence-initial slot itself has cognitive implications. In other words, what is chosen as the starting point

82

Tuija Virtanen

in a given context will have consequences for the processing of the rest of the sentence and for the text world that interlocutors are constructing and reshaping, their anticipations of what follows, and the order in which their uncertainties are eliminated.

2.

Sentence-initial adverbials as signals of text strategy

Sentence-initial adverbials of time and place in narrative and descriptive texts tend to form chains of text-strategic markers which have two basic functions in the discourse. They help create coherence and at the same time they signal text segmentation. Let us take a look at two simple examples, the first one a fable and the second an extract from a travel guide. (1) The Horse, the Ox, the Dog and the Man When Zeus made man, he only gave him a short life-span. But man, making use of his intelligence, made a house and lived in it when winter came on. Then, one day, it became fiercely cold, it poured with rain and the horse could no longer endure it. So he galloped up to the man s house and asked if he could take shelter with him. But the man said that he could only shelter there on one condition, and that was that the horse would give him a portion of the years of his life. The horse gave him some willingly. A short time later, the ox also appeared. He too could not bear the bad weather any more. The man said the same thing to him, that he wouldn Ί give him shelter unless the ox gave him a certain number of his own years. The ox gave him some and was allowed to go in. Finally, the dog, dying of cold, also appeared, and upon surrendering part of the time he had left to live, was given shelter. Thus it resulted that for that portion of time originally allotted them by Zeus, men are pure and good; when they reach the years gained from the horse, they are glorious and proud; when they reach the years of the ox, they are willing to accept discipline; but when they reach the dog years. they become grumbling and irritable. (Aesop, The Complete Fables, Fable 139, Penguin 1998.) As fables usually do, this fable has two parts, the narrative and the moral. In this instance the lesson to be learnt from the fable is presented so that it conforms to the chronology of the narrative. There are thus two chains of sentence-initial adverbials of time in the text, and each member of the chain signals the starting point of a new stage in the narrative, i.e. a textual boundary between units of

Point of departure: Cognitive aspects of sentence-initial adverbials

83

text of different sizes. Each adverbial creates coherence by participating in the chain of text-strategic markers, and at the same time, helps constitute the temporal setting for the textual unit that it introduces. The first adverbial activates the narrative schema, to the extent that it is not already activated by the context; temporal sequentiality is the basic characteristic of narrative and hence expected in the context. Since my concern here is with adverbials, I will not discuss the role or form of the references to participants in these and other sentences in relation to the signalling of coherence and text segmentation. An interesting discussion of two different models of participant reference in discourse is, however, found elsewhere in this volume (Anna Siewierska, in Chapter 3, above). The second example is an extract from a travel-guide, and it manifests a chain of adverbials of place in sentence-initial position. These adverbials function in the same way as the temporal adverbials in narrative: They create coherence and segment the text into units of various kinds. They are inherently temporal in the sense that the reader is instructed to follow a route, mentally and/or physically, by moving to a new location at each new sentence-initial locative. Both adverbials of time and place appear non-initially in the sentence, in texts of these kinds, when they are not needed in initial position to signal the activated temporal or locative text strategy. (For a detailed discussion of the temporal and locative text strategy, see Virtanen 1992a.) (2) Beyond the Taverns of the Fishmongers along the Decumanus Maximus, two insulae are set side by side; the Insula of Dionysos and Insula of the Eagle. Both of them built in 125-130 A.D., Hadrian's epoch, they were restored and partly modified around the middle of the HI century and this late remodelling considerably changed the face of the old insulae. Inside, several mosaics and frescoes show the considerable refinement of their decoration. In the room at the rear of the Insula of the Eagle is preserved the Mithraeum of the Seven Doors which boasts a fresco representing a lovely garden. [...] (Ostia: Guide to the Excavations, Storti 1985, p 39.) In what follows, the concern is with two aspects of strategy-marking adverbials: (i) their information status, which is important to consider in light of the default expectations that we have of thematic material, and (ii) the way in which these adverbials contribute to the signalling of a hierarchy of textual boundaries, which adds to our understanding of the cognitive role that they play in the discourse.

84

Tuija Virtanen

2.1 Information status of sentence-initial adverbials Sentence-initial adverbials of time and place are thematic but they only occasionally convey textually given information. Instead, they typically convey information that has been characterized as 'inferrable' by Prince (1981). While the kind of information that these strategy-marking adverbials represent can usually be situated in the middle part of a scale reaching from given to new information, it is important to note that they can have any information status on such a scale. For discussions, see Virtanen (1992a, 1992c). To illustrate the typical information status, let us take another look at (2), above. Here the reader is, for instance, assumed to be able to infer the fact that inside refers to the inside of the insulae which have been mentioned in the immediately preceding text, and which can therefore be assumed to be active in the reader's mind at that stage of text processing. At some point, the reader is similarly assumed to be able to infer that insulae are buildings of some kind which have rooms (cf. in the room at the rear of the Insula of the Eagle). Alternatively, to an informed audience this can be 'unused' information (Prince 1981) which they are assumed to identify while reading the text, and the information which they then need to infer is conveyed by the prepositional phrases starting with in and at the rear of.1 Similarly, the prepositional phrase beyond the Taverns of the Fishmongers along the Decumanus Maximus consists of inferrable information conveyed by the prepositions beyond and along anchored to textually, and possibly also situationally, given information conveyed by the two noun phrases of the 'absolute' kind (as contrasted to 'text-related' and 'speech-act' related references; cf Enkvist 1981). The information conveyed by the entire sentence-initial cluster of adverbials is accessible to the reader through the locative text strategy opted for in the text and the confirmation of anticipations concerning its continuation that are invited at this point. The reader, if a tourist on the spot, will be expecting new moves leading to new points of interest to be signalled in the text. As pointed out above, the locative text strategy is inherently temporal in the sense that the reader is assumed to follow a given route to visit the sight. New adverbials of place in sentence-initial position indicate new stages in the text, i.e. new stops on the reader's route. To return to (1), its first adverbial when Zeus made man conveys unused (new) information, which is assumed to be identifiable by the intended audience. Even when it is not, the reader will infer from the context that it should be and consequently adapt to the situation by either finding out (brand-new information) or inferring for instance that Zeus must be a God if s/he can be claimed to have made man (inferrable information encapsulated in the adverbial clause

Point of departure: Cognitive aspects of sentence-initial adverbials

85

and available from the activated fable frame). In all cases, the indication of a point of time by the subordinator -when serves to activate the temporal text strategy, which is expected in narrative. The three adverbials then, a short time later and finally are inferrable from the activated narrative schema and the previous temporal marker in the textstrategic chain. One day constitutes brand-new information. Yet, it is not too hard to process since it is anchored to the inferrable then. Finally, the series of sentence-initial adverbial clauses in the moral part of the fable (the starting point of which is signalled by the connector thus) show a high degree of lexical overlap with the preceding text and represent inferrable information. It is interesting to note that the starting point of the moral of the fable is indicated by a connector - one of those adverbials which have become 'professional' in their services to the text (see section 6, below). Thus here signals to the listener/reader the existence of a major textual boundary. The closure of the first temporal chain has been conveyed by the adverbial^na/(y. The fact that an adverbial of a different kind now appears in the sentence-initial position is a relatively sure sign of the beginning of something new in the text, the outset of another major unit of the text. It is indeed only after the transition indicated by thus it resulted that that a new chain of temporal adverbials is introduced, as an instruction to the audience to rewind the temporal sequence of the narrative, in order to match the chronology of its events with the succession of the points to be made in the moral. To sum up, strategy-marking adverbials can convey information of any kind. The extremes of brand-new information and textually given information are, however, less usual than inferrable, and to a smaller extent unused, information, which are found in the middle of the given-new scale. In addition to creating coherence through the chain that they form, these adverbials facilitate processing of the text by indicating textual boundaries, and, as will be seen next, the hierarchy of such boundaries. 2.2 Adverbials segmenting the text Both the information status and the number and size of text-strategically important adverbials in narrative and descriptive texts reflect the hierarchy of text structure, which the writer is obliged to express in a linear form. Hence, boundaries between major textual units tend to have more markers, larger markers and markers which convey newer information than those found at local boundaries in the text. This suggests that local boundaries need little marking because the reader is assumed to be able to keep the necessary information activated across

86

Tuija Virtanen

such boundaries. The continued activation status of referents, states and events, and relevant circumstances helps to maintain the unity of the textual unit within which these local boundaries appear. In contrast, major textual boundaries need more marking to allow the reader to temporarily discard aspects of text strategies not needed in the immediately following text and reactivate others that are useful at the outset of a new textual unit. Virtanen (1992a, 1992c) shows how sentence-initial adverbiale signalling text-strategy in narratives and descriptive texts participate in the marking of a textual boundary as minor or major. The text-internal differences that such adverbials manifest in terms of (i) their size and number, and (ii) the kind of information they convey, reflect the position of the boundaries in the overall hierarchy of the text: The size of a given boundary is iconically reflected in the size or number of the marker(s), and the processing load that its information status puts on the reader. Hence, in (2) above, inside serves to signal a minor boundary while in the room at the rear of the Insula of the Eagle marks a major boundary in the text. Narratives longer than our example above can make use of a full range of adverbial markers to signal the overall hierarchy of the text. In terms of this kind of 'structural iconicity', the number, size, and/or information status of strategy-marking adverbials reflect the size of the textual boundary that they mark, thus helping indicate the hierarchy of text structure. This finding is in line with Givon's (1984, 1985) work on iconicity, and the idea of an accessibility scale as proposed by Ariel (1990), both of which are discussed from another perspective by Anna Siewierska elsewhere in this volume.

3.

Relying on iconicity

Sentence-initial adverbials can be related to iconicity of different kinds. In addition to the iconic reflection of the hierarchy of text structure, they are obviously also related to iconicity in the very basic sense of linear ordering in the text. It is thus worth making a distinction between 'structural iconicity', as discussed in the previous section, and 'experiential iconicity' (for the latter notion, see Enkvist 1981). We know that the temporal order of the events and actions constituting a narrative is assumed to conform to their order in the story-time unless otherwise indicated. When a text is iconic with respect to the temporal order and adjacency of events and actions, no explicit markers such as then are necessary for the reader to understand the sequentiality of bounded events. We do, however, find then in narrative for other reasons, such as indicating a textual boundary or

Point of departure: Cognitive aspects of sentence-initial adverbials

87

contributing to the signalling of a peak profile in the narrative. Other adverbials of time signalling the temporal text strategy which convey a lexically weighty content and often indicate deviation from the iconic order or from temporal adjacency, are, of course, needed in such texts. The ordering of adverbials of place in descriptive texts, again, is inherently temporal. We have seen above that it conforms to the order in which the reader is assumed to move to see the sight, or more generally, the order which is assumed to be helpful for the reader conceptualizing the description (see Linde & Labov 1975). Temporality can therefore be left implicit and a locative strategy is usually chosen instead to move the reader along the given route in a particular order. A temporal marker, however, sometimes appears among the locative ones; this is what happens in the descriptions of apartments that Linde and Labov studied. Consider, in this light, extract (3) from a travel guide, in which the signal of a new stage in the text, very soon, is immediately followed by the locative subject the main street, the Grande Rue, serving to indicate the exact location of the next stop. Travel guides instructing tourists to follow a route outside urban areas, in fact, tend to conform to a series of temporal markers, interspersing the chain with indications of distance in space or spatial markers where appropriate; cf. (4) below, from the same text. (3) It is best to [...] enter the town through the Porte Pignerol. Very soon majnjir^eJ^jhe^r^nde^Ruej is reached. This is famous for its beautiful period houses and the stream that runs down the middle, which is known locally as the Grande Gargouille. To the right of the main street is the Church of Notre-Dame, also built by Vauban with an eye to defence. It has high walls, twin clock towers and hardly any windows. To the left of the old town is the Citadelle, which is approached through the Porte Dauphine after a climb up the rocky slope. [...] Above the Citadelle is a large statue, La France, and a nearby orientation map names the peaks of the surrounding Briangonnais region. (P. Scola, The Visitor's Guide to France: Alps & Jura. Ashbourne: Moorland 1990, pp. 240-1.) (4) Later there is a turn to the beautifully named village ofOz, which is developing as a ski resort and has linked its lift system in with that of L 'Alpe d'Huez. The road then crosses the river and climbs through the trees. After passing the village ofLe Rivier d'Allemont. the road turns and can be seen for some miles ahead as it contours along the mountainside in the rather desolate Defile de Maupas. While driving towards it, look behind for a dramatic view of the Grand Pic de Belledonne, rising above the valley. Four miles (6 '/2 km) later, the second artificial lake can be seen down

88

Tuija Virtanen

on the right, the Lac de Grand'Maison, presently the largest man-made lake in France, held back by the Barrage de Grand'Maison. Beyond the lake the Col du Glandon is soon reached, (p. 232) It is intriguing to examine the decisions that writers make concerning the placement of temporal and locative elements in such, often extremely condensed, texts. Some of these are needed at the outset of the textual unit which they introduce, to signal text strategy, while others can be placed at the end of the sentence as specifications; cf. e.g. the following specifications in (3): through the Porte Dauphine after a climb up the rocky slope; and in (4): rising above the valley; down on the right; soon. In such texts experiential iconicity can be detected in the relative ordering of the elements participating in the text-strategic chain. But is can also be found within the sentence so that the new location is indicated before information concerning the sight is given. In fact, experiential iconicity here enables the writer to leave out the copula between the adverbial signal and the noun phrase, as in the following sentence from an extremely condensed travel guide text: On the wall, Lord Roberts (1833-1914), bust by John Tweed. In contrast, specifications have a textually local and syntactically narrow scope and they can therefore easily break this ordering, as seen above; consider again the following sentence from (3): To the left of the old town is the Citadelle (iconic), which is approached through the Porte Dauphine after a climb up the rocky slope (non-iconic). Also, a narrative involving movement of characters may well blend temporal and locative signals in one and the same text-strategic chain, as in (5), below. (5) Once upon a time there were three billy-goats called Gruff. One fine day, the three billy-goats Gruff set off up the hillside. They were going to look for some sweet grass to eat so that they could grow fat. On the way up the hillside, the three billy-goats Gruff came to a river. [...] (The Three Billy-Goats Gruff, retold by V. Southgate. Loughborough: Ladybird Books 1968.) Furthermore, the ordering of a cluster of sentence-initial adverbials can be iconic. Consider the following two examples. The first one is entirely iconic in terms of temporal ordering ('dawn' - 'waking up' - 'running to window' - 'admiring the garden'). Here, the second sentence-initial adverbial conveys an event that takes place after the starting point of the time indicated by the first adverbial, and it is included in the time span of the first adverbial. In (7) readers are directed to a given sight through a cluster of sentence-initial markers whose iconic ordering helps them to zoom in on the sight.

Point of departure: Cognitive aspects of sentence-initial adverbials

89

(6) The next morning, as soon as they woke up, they ran to the window to admire their garden. (P. Rowand, The cats who stayed for dinner. In: B. Ireson (ed.) The Faber Book of Nursery Stories. London: Faber & Faber 1966, p 152.) (7) 2'/2τη. Α short distance to the 1. of the road, on lower ground towards the river, among farm buildings, is the church of the monastery of [...]. (I. Robertson, Blue Guide: Cyprus. London: Ernest Benn 1981, p. 100.) Iconicity is, however, only conformed to if it is not in conflict with other factors affecting the form of the text. The task of signalling a text strategy - temporal or locative - makes demands on the ordering of sentence-initial clusters so that adverbials indicating a text strategy usually appear before other sentence-initial adverbials. When signals of both local and global strategy are included in one and the same sentence-initial cluster, those indicating a local boundary in the text tend to precede the ones belonging to a global chain of text-strategic markers. This is so because local phenomena can be assumed to be activated in the reader's mind while the global strategy may need to be reactivated. But a temporally iconic placement of an adverbial at the outset of the sentence can also have other cognitive motivations, as will become apparent in the following section.

4.

Crucial information first

In (8) below we find a locative right at the beginning of the sentence, which conforms to the intended temporal iconicity of the instruction: 'first find a baking dish, then butter it, then arrange the ingredients in the dish in the very order in which they are given'. This kind of iconic ordering can be found in recipes, where such sentence-initial locatives appear sparingly and do not participate in a text-strategic chain. In (9), from the same text, we can, in fact, see that the text strategy is action-oriented, which is another inherently temporal way of organizing texts (place - add - pour). As experiential iconicity predetermines text organization in instructive texts, its temporal signals can be left implicit when they simply convey a temporally adjacent sequentiality of the series of actions that the reader is instructed to take. However, in (8) explicit temporal signals appear to emphasize the importance of ordering, as first, then. In (9), again, the clause-initial after 10 minutes conveys information that is urgent at that point; compare this ordering to the last sentence of the example, where the indication of time (after another 10 minutes) is not as urgent at all and will thus adopt the

90

Tuija Virtanen

task of specifying the time-point of the instructed action. (For discussions of recipes, see e.g. Enkvist 1981; Prince 1981.) (8) In a buttered baking dish arrange first a layer of potatoes, then a layer of onions and anchovies, and finish with a layer of potatoes. (B. Tims (ed.), Food in Vogue. London: Pyramid Books 1988, p. 70.) (9) Place in a preheated oven (Gas Regulo 6, 400 °F) and after 10 minutes add half the cream. Pour in the remaining cream after another 10 minutes. These examples bring us to the occasional need for interlocutors to adopt what Enkvist (1989) has called the strategy of placing 'crucial information first' in the sentence - because this information is urgent at the stage at which it appears in the text. Hence, the availability of the buttered dish in (8) and the fact that the first half of the cream needs to be added after 10 minutes in (9) both constitute crucial information in the context. At the same time, this ordering has the advantage of leaving the end of the sentence free for the weighty information concerning the layering of the ingredients in the dish, in (8), and the fact that only half of the cream is to be added at this stage, in (9). Crucial information need not constitute given information; irrespective of whether readers will find such information inferrable from the context, new but easily available in their relevant encyclopedic knowledge, or brand-new, the fact that it is placed first helps them follow the order of the steps they are instructed to take. The strategy of starting with crucial information also partly explains why manner adverbials, typically conveying new information, sometimes occur initially in instructive texts; other, concomitant motivations have again to do with what the writer wishes to position last in the sentence. Consider the following examples from instructive texts, where the crucial indication of the manner of the action is placed first: (10) comes from a recipe while (11) is a step in a yoga exercise. (10) Gradually add the sugar. (11) Gently, in slow motion, bend your elbows outward The examples in this section would also seem to illustrate Osgood's (1980) 'natural' and 'unnatural' salience (cf. also Chapter 6 in this volume, by Brita Wärvik). Hence, iconic linearization conforms to the 'natural' order of temporal sequentiality whereas in the last two examples the initial expression of manner is made 'unnaturally' salient by placing these adverbials first. Note that 'unnatural' salience is, in fact, no less natural in a given context; it can, in fact, be the most natural option available for the writer. Placing crucial information first can thus result in sentences which are 'textually unmarked', though syntacti-

Point of departure: Cognitive aspects of sentence-initial adverbials

91

cally marked (cf. Enkvist 1984). The examples discussed in this section are also reminiscent of the point Jespersen (1949: 54) makes: "what is at the moment uppermost in the speaker's mind tends to be first expressed"; this he calls the 'principle of actuality'.2

5. Creating common ground and persuading Starting points can manifest lexical overlap with the preceding text and they typically convey given or inferrable information. They help suggest common ground for what is to follow. They can indicate the setting, a source, or a restriction of some kind, for what follows, or a perspective that the reader is expected to adopt in processing the text. Such elements are not necessarily spatio-temporal though they can be abstractions from these. The fact that they are placed initially in the sentence invites the reader to interpret them as background for what follows. Consider the following examples from non-narrative texts. In (12a-c), all from a section giving practical information to the travel-guide reader who wishes to go cycle-touring, the adverbials in initial position set the background in terms of specifying the grounds or authority to make the subsequent recommendation (12a), giving the point of view from which to interpret the sentence (12b), and indicating the way in which the subsequent statement is intended to be understood (12c). (12a) After consulting with various experienced cyclists, and from my own travels. I recommend 80km (50 miles) a day as a good average for a cycling tour. (R. Neulands, Cycle Touring in France. Sparkford: Oxford Illustrated Press 1989, p. 32.) (12b) Traditionally, cycling clothing is brightly coloured to attract the attention of myopic motorists, close-fitting to cut down wind resistance, and partially wind-proof to keep the cold out. (p. 26) (12c) [...] and if the truth were told I don't think that clothing matters, at least until the weather turns chilly. (p. 27) In (13a-c), three sentences from a reply to a complaint received by the present author, we similarly find background information initially in the sentence. In this argumentative text, initial placement of adverbials has an inherently persuasive function. This is where we find elements to which the reader is assumed to assign background status and which s/he can be hoped to be taking for grant-

92

Tuija Virtcmen

ed, i.e. the fact that very careful thought has been given the matter in (13a), the information concerning the cooked wheat grain being crushed between a pair of heavy rollers - professionally given to the text-producing body but not to the uninformed reader, in (13b); and the authority of the text-producing company in(13c). (13a) Having given the matter very careful thought, we can think of only two possible sources of the stone that you found. (13b) However, as the cooked wheat grain is flaked by being crushed between a pair of heavy rollers, we can reasonably assume that a stone would also be crushed, the small particles falling clear during later sieving. (13 c) ... and we would like to take the opportunity of assuring you that as a major producer of breakfast cereals we do maintain very high standards of quality and are particularly conscious of our responsibilities... In (14a-b), from newspaper editorials, adverbials indicating attitudes and beliefs are placed initially, to instruct the reader of the way the statement is designed to be interpreted. Hence, in (14a) unfortunately, an affect-marker among others in the argumentative text, introduces the writer's main criticism. (The temporal adverbial at the outset of the editorial serves to set the scene, and the anaphoric on its merits at the beginning of the second sentence creates coherence, restricts the perspective on what follows to 'merits', and leaves the end of the sentence free for the writer's positive evaluation of the proposal, to be subsequently contrasted with criticism concerning the behaviour of its backers.) In (14b), the source of information is (vaguely) indicated sentence-initially (in survey after survey}, before the information itself is presented. Yet in the subsequent sentence conveys the writer's stance: the American public is ambivalent in their attitudes to election campaigns. (14a) On Election Day, New York City voters will be faced with a proposal to change the local term limits law by extending the maximum time an elected city official can hold one job from 8 years to 12. On its merits, the proposal is a good one. Unfortunately, its backers have been inept and evasive in pressing their cause. Their behavior has made an otherwise easy decision difficult and somewhat distasteful. (The New York Times, Editorial, Oct 21, 1996.) (14b) What the public is seeing here is no mystery but rather a mirror of its own ambivalence. In survey after survey, Americans say they are turned off when the attacks get too personal. Yet deep down they know that

Point of departure: Cognitive aspects of sentence-initial adverbials

93

American campaigns have always been in part an adversarial spectator sport. (The New York Times, Editorial, Oct 13, 1996.) The above examples serve to demonstrate the fact that the sentence-initial position is a handy slot for material that is designed to be interpreted as common ground, to be taken for granted - simply because it is placed where interlocutors by default would expect given information, where they are normally heading for guidance to orient themselves concerning text organization, and where they know they will find information that needs to be kept active while waiting for the weighty, new information to come. This is where we find adverbials indicating beliefs and attitudes, ranging from ways of speaking and estimated degrees of truth to a multitude of markers that indicate how the propositional content should be interpreted, what its implications are, what the source ofthat information is, and why it is a 'fact'. The sentence-initial position thus offers a high degree of potential for persuasive purposes. It serves as an important slot for various restrictions and markers of affect and point of view concerning the subsequent text (cf. also, in this light, the discussion of conditionals in Chapter 8 of this volume, by Anne-Marie Bulow-M011er). But because it is informationally a foregrounded position - consider the discussion of 'crucial information first' above - persuasive adverbials are also often placed right after the subject, i.e. initially enough to serve as markers of a suggested common ground but less foregrounded informationally, and hence less salient. (For a different, and detailed, discussion of the relative foregrounding and backgrounding of textual elements, see Wärvik, in this volume.) The effects of sentence-initial adverbials of the kinds illustrated in this section on the construction of contexts for, for instance, semi-public and public discourse could profitably be examined from the perspective of distributed cognition.

6.

'Professionalized' strategy markers

It has been pointed out above that the sentence-initial slot can be filled with items that have been 'professionalized' to indicate metatextual/metadiscursive information, such as adverbial connectors (e.g. to start with, furthermore, finally; thus; in contrast; however; yet). Such signals of text organization indicate the kinds of thought processes that are involved in the text. As they are aimed at contributing to the way readers interpret the text, they can also serve a persuasive function. They invite the reader to construe particular logico-semantic relations between units of the text and thus they possess an interpersonal func-

94

Tuija Virtanen

tion in the same way as metadiscourse does in general. It is no wonder that they appear - and are expected to appear - in written expository and argumentative texts. These are texts where planning is usually evident and where it is in the interest of writers to make the readership understand the way they are thinking and to see their point. Adverbial connectors are 'professionalized' in the task of indicating text strategy and they therefore need not necessarily appear at the very beginning of the sentence; indeed there are culture-specific stylistic reservations against too frequent placement of such items sentence-initially in informative texts. Writers can thus opt for a placement of some of these elements after the subject of the sentence; here they can still function in their text-organizing role while leaving the very beginning of the sentence to propositional elements that can only do so in that very position. Adverbials conveying beliefs, attitudes and the like (e.g. fortunately, correctly, obviously, surely, frankly) can also be argued to have been 'professionalized' in their task of indicating affect. The cognitive motivations for placing them first in the sentence are connected to their main function of constructing an affect-oriented point of view for the subsequent text, as described in section 5 above. But we can also note that adverbials which are part of the propositional structure of the sentence can become 'professionalized' in the discourse functions that they repeatedly serve. The narrative then indicating 'after that' is a good example of a 'professionalized' signal of the temporal text strategy - a job it can only do when placed at the beginning of the sentence, or right after the subject. Sentence-initial adverbials of time can also carry out intertextual and interdiscursive tasks as text-type or genre markers; consider, for instance, the cognitive expectations set by once upon the time at the outset of a text.

7.

In conclusion

The above analysis of sentence-initial placement of adverbials demonstrates that the sentence-initial position itself has a great deal of cognitive potential. Elements placed here serve to set the stage as important and salient points of departure, create coherence and segment the text at one and the same time, and indicate affect and determine point of view for what follows. The information status of sentence-initial adverbials can be argued to be partly related to the discourse work that such elements accomplish in the text. Some of these discourse functions rely on structural iconicity, others conform to experiential iconicity, and still others are non-iconic in nature.

Point of departure: Cognitive aspects of sentence-initial adverbials

95

In the default instance the informationally foregrounded, sentence-initial material is expected to remain activated in the reader's memory in wait for the main inferencing effort to take place at the end of the sentence, where the weighty information and its salient focus usually fall. Sentence-initial information is assumed to be active, semi-active, or at least to some extent accessible, throughout the processing of the textual unit that it introduces. When crucial information appears first, it is its very placement at the outset of the sentence that contributes to the signalling of this information as urgent and hence important in the context. The cognitive implications of placing adverbials initially can also account for the 'professionalizing' in given discourse functions, of particular categories of adverbials which appear in this slot repeatedly and with ease. Once 'professionalized', connectors and adverbials indicating manner of speaking, beliefs, attitudes and the like, can, for various reasons, even yield the very first position to other elements and still keep their discourse organizing or affect-marking status. Furthermore, sentence-initial adverbials can function as text-type or genre markers, thus activating intertextual and interdiscursive expectations. The study of sentence-initial adverbials in context also sheds light on the cognitive aspects of persuasion through discourse. For instance, readers interpreting initial material as conveying given or background-constituting information may take this information for granted while anticipating and focusing on what follows. Adverbial placement is also connected to the socio-cultural context in which particular linearization strategies are put to use as such strategies can also contribute to constructing the very contexts. In other words, sentenceinitial placement of adverbials can be related to both individual and distributed cognition. To round off, linearization is an area of study where the interests of cognitive linguists and discourse linguists meet. The sentence-initial slot is a rich source of discourse meanings, the study of which opens important avenues to cognition and contributes to our understanding of the cognitive aspects of central discourse phenomena.

Notes 1. An insula is an apartment in an apartment building (Ostia: Guide to the Excavations, Storti 1985: 60). 2. I am grateful to Marita Gustafsson for pointing this out to me.

96

Tuija Virtanen

References Ariel, Mira 1990 Accessing NP Antecedents. London: Routledge. Chafe, Wallace 1994 Discourse, Consciousness, and Time; The Flow and Displacement of Conscious Experience in Speaking and Writing. Chicago/London: The University of Chicago Press. Enkvist, Nils Erik 1981 Experiential iconicism in text strategy. Text 1/1:97-111. 1984 Contrastive linguistics and text linguistics. In: Jacek Fisiak (ed.) Contrastive Linguistics: Prospects and Problems, 45-67. Berlin: Mouton. 1989 Connexity, interpretability, universes of discourse, and text worlds. In: Sture Allen (ed.), Possible Worlds in Humanities, Arts and Sciences: Proceedings of Nobel Symposium 65, 162-186. (Research in Text Theory 14.) Berlin / New York: Walter de Gruyter. Givon, Talmy 1984 Syntax: A Functional-Typological Introduction. Volume 1. Amsterdam / Philadelphia: John Benjamins. 1985 Iconicity, isomorphism and non-arbitrary coding in syntax. In: John Haiman (ed.) Iconicity in Syntax, 187-219. (Typological Studies in Language 6.) Amsterdam / Philadelphia: John Benjamins. Jespersen, Otto 1949 A Modern English Grammar on Historical Principles. Part VII: Syntax. Copenhagen: J. J0rgensen. Levelt, Willem J.M. 1981 The speaker's linearization problem. The Psychological Mechanisms of Language, 91-101. London: The Royal Society and The British Academy. 91-101. (l st publ. in Philosophical Transactions of the Royal Society of London B:295, 305-315.) Linde, Charlotte and William Labov 1975 Spatial networks as a site for the study of language and thought. Language 51/4: 924-939. Osgood, Charles E. 1980 Lectures on Language Performance. (Springer Series in Language and Communication 7.) New York: Springer-Verlag. Östman, Jan-Ola and Tuija Virtanen 1999 Theme, Comment, and Newness as Figures in Information Structuring. In: Karen van Hoek, Andrej A. Kibrik and Leo Noordman (eds.), Discourse Studies in Cognitive Linguistics: Selected Papers from the 5th International Cognitive Linguistics Conference, Amsterdam, 1997, 91-110. Amsterdam / Philadephia: John Benjamins.

Point of departure: Cognitive aspects of sentence-initial adverbials

97

Prince, Ellen F. 1981 Toward a taxonomy of given-new information. In: Peter Cole (ed.), Radical Pragmatics, 223-255. New York: Academic Press. Virtanen, Tuija 1992a Discourse Functions of Adverbial Placement in English: Clause-Initial Adverbials of Time and Place in Narratives and Procedural Place Descriptions. Abo: Abo Akademi University Press. 1992b Issues of text typology: Narrative - a 'basic' type of text? Text 12/2: 293310. 1992c Given and new information in adverbials: Clause-initial adverbials of time and place. Journal of Pragmatics 17/2: 99-115.

Chapter 6 What is foregrounded in narratives? Hypotheses for the cognitive basis of foregrounding Brita Wärvik

The purpose of this chapter is to look for parallels between the textual foregroundbackground distinction and perceptual and cognitive principles of organization. I'll start with some general remarks about the foreground-background distinction or as it is also known, grounding. The discussion deals exclusively with narrative texts; grounding studies started with stories, typically folk tales and myths, and with very few exceptions they have stayed with narratives (Bäcklund 1988 on expository prose). In text and discourse linguistic studies of narratives, the term foreground has been used in two different, but related senses. On the one hand and originally, foreground stands for the temporally sequential story-line, the parts of the narrative that forward the story; background then stands for the non-temporal material, such as descriptions and comments, and also that temporally linked material which is not on the sequential story-line, that is flashbacks and foreshadowings. On the other hand and as a development of the first sense, foreground has been used to refer to the main story-line, the central events and actions that form the gist of the story, the red thread through the story; background to this type of foreground is then the supportive, downgraded or commenting material. Note that in this second sense, foreground is not bound to narratives: all kinds of text can be seen as organized around a backbone or main-line. Note also that in simple stories, in plain accounts of events, these two types of foreground coincide, because there what is central is what happens, but in more elaborated stories they tend to diverge. In addition to the linguistic concepts of foreground as sequential story-line and as main story-line, there is a third sense in which the term foreground has been used in the study of narrative structure, that is in literary studies, where foreground is used for what is striking, deviant or unexpected, particularly in literary or poetic texts; background in this framework is the norm or normal text against which something appears as foregrounded. Although the poetic foregrounding and the linguistic kinds of foregrounding may appear totally contradictory to each other, some models of grounding have tried to integrate all of them, their common denominator being salience or prominence.

100

1.

Brita Warvik

Foreground-background distinction in narratives

Let us start with a few general points about the foreground-background distinction. First of all, though most researchers talk about grounding distinctions in terms of a foreground and a background, these distinctions cannot be viewed as a simple dichotomy, but they ought to be seen as a scale with maximal foregrounding at one end and maximal backgrounding at the other. Secondly, grounding is best seen as a cluster concept, so that the foregroundedness vs. backgroundedness of a clause is dependent on several criteria: each of them affects the grounding degree of the clause, but none of them is alone decisive. Such grounding criteria have been collected most notably by Hopper and Thompson in their Transitivity Scale (1980; Hopper and Thompson (eds.) 1982) and Chvany in her Saliency Hierarchy (1985a, 1985b). Thirdly, apart from the complications arising from the nature of grounding as a continuum and as a cluster, models of grounding have to integrate contextuality, because the real grounding value of a clause is, like all textual phenomena, dependent on the context. In other words, it is only in relation to other clauses in a text that we can assign a specific degree of foregroundedness vs backgroundedness to a clause. Fourthly, grounding is a textual phenomenon which does not exist as such in the represented world (pace Tomlin 1983 and later). Thus, the events, actions and states that the textual elements depict and the participants that are involved in them are not foregrounded or backgrounded in themselves; their grounding value emerges only when a text producer presents them in a certain way in a certain context. The fifth point is that the foregroundedness vs backgroundedness of a clause depends on formal, linguistic criteria and on the content of the clause. The formal criteria, which we can call grounding markers, are of varying forms, ranging from subject and topic markers through aspects to pragmatic particles. The grounding value suggested by the formal markers usually contributes toward the same interpretation as the grounding value of the content criteria, but they do not obligatorily support each other. To solve such problems our model of grounding should involve information about the weightings and relative dominances of the different criteria so that we could measure the opposing criteria against each other; unfortunately we are not at that stage in grounding research yet. Where we are is probably best shown by a list of grounding criteria in narrative text; research has shown that the features in Figure 1 are involved in determining the foregroundedness vs backgroundedness of a clause. This list is my synthesis of the various suggestions for grounding criteria, and it subsumes

What is foregrounded in narratives?

FEATURES OF FOREGROUNDING

FEATURES OF BACKGROUNDING

NUMBER OF PARTICIPANTS two or more participants > one participant > impersonal

oi

υ

I OH


positioner/recipient > instrument > affected > locative/temporal/eventive > prop it > no subject TOPICALITY OF PARTICIPANTS given/old new current evoked>displaced evoked>inferrable>unused>brand new AFFECTEDNESS OF THE MOST AFFECTED PARΉCIPANT created/destroyed>moved/changed>touched/perceived/experienced>unaffected TENSE story-now, past other 'times'

..

3

ASPECT, SITUATION TYPE perfective imperfective simple, non-progressive progressive dynamic static telic atelic punctual durative unique multiplicable ACTUALITY: MODALITY, AFFIRMATION real, affirmed > 'less than real* > irreal, negated

i ω on D

,_j

WLITIONALITY purposeful.volitional > > accidental, non-controllable independent combined paratactic nonrestrictive finite

dependent embedded hypotactic restrictive nonfinite

TEMPORAL SEQUENTIALITY events forwarding the story-line events off the story-line

ί Figure 1. Grounding criteria for narrative text

101

102

Brita Warvik

the Transitivity parameters and the criteria of the Saliency hierarchy, some of them reformulated or reorganized (Wärvik 2002). The features are expressed here in a rather fuzzy form, because this is not a language-specific list. Work on grounding has been done on a variety of very different languages and these features can be coded in very different forms. If we think of today's Standard English, some of these features are explicitly coded or grammaticaticalized, and others are not. The list starts with participant criteria, features of the nominal elements of the clause. The first is the number of participants, which is the basic feature of transitivity in the traditional sense where transitive clauses are ones with subject and direct object; this criterion is also included in Hopper and Thompson's Transitivity scale (1980, (eds.) 1982) and Chvany's Saliency hierarchy (1985a, 1985b), which take into account all kinds of objects. In terms of this criterion, the higher the number of participants, the more foregrounded the clause is likely to be. Secondly, there are the individuation features, which apply to all participants. Thus highly individuated participants are more typically highly foregrounded than participants with low individuation scores. If a story starts John was drinking coffee, we are more likely to assume that the story is going to be about John than about coffee. Individuation arises out of a combination of different features, which form groups of interdependent features: the groupings vary from one language to another, but a common constellation is one where the features human, animate and concrete are connected; likewise countability and number, referentiality, specificity and definiteness and finally proper noun, pronoun and common noun (cf Chafe 1976; Chvany 1985a, 1985b; Givon 1984; Hopper and Thompson 1980; Lyons 1977; Yamamoto 1999). Thirdly, the agentivity of the subject affects the foregroundedness of the clause: the more agentive the subject the more foregrounded the clause is likely to be. Agentivity refers to the degree to which the subject is capable of acting as an agent. This criterion is problematic to model and what appears here is my suggestion adapting Quirk et al's (1985) hierarchy of types of subject roles in English, but other subjectization hierarchies, or hierarchies of accessibility to subjecthood could as well have been used here (eg Ertel 1977; Keenan 1976; Lyons 1977: SOOff; Sundman 1987; Tomlin 1983). Prototypical agents, that is, initiators or direct causes of the happening, are most foregroundable. So that if the story starting with John was drinking coffee, continues // was steaming, we tend to see that as more backgrounded than He was steaming, because of the different degrees of agentivity of the subjects. The fourth criterion is Topicality of participants, which refers to the information status, givenness, thematicity or continuity of the participants. The more

What is foregrounded in narratives?

103

topical or continuous a participant is, the more foregrounded it tends to be. So if the story about John and coffee really goes on about coffee, then coffee is foregrounded in that discourse. Simplifying a bit, we can say that major participants or main characters tend to be more foregrounded than minor characters and props (on the role of main characters in text processing see e.g. McGann and Schwartz 1988; Garrod and Sanford 1988). The parameter of Topicality is here modelled according to Prince's taxonomy of assumed familiarity, which is a scale of given and new information (Prince 1981; Brown and Yule 1983). The fifth and last of the participant criteria is the affectedness of the most affected participant, which in finite transitive active clauses is the direct object, but for instance in corresponding passive clauses is the subject. The scale for evaluating affectedness that we have in the list is from Chvany's Saliency hierarchy, which classifies participants fairly impressionistically in four classes (Chvany 1985a, 1985b). The idea is that the more effect an action or event has on a participant, the more foregrounded that action or event appears in the narrative (cf Transitivity as effectiveness of action in Hopper and Thompson 1980). The second group of features are the verbal criteria. First of them is tense, which has the in theory simple relation to foregrounding that situations, that is events, actions and states reported in the tense associated with the story-wow, which is often the past tense, are more foregrounded because they typically forward the story-line, and other tenses are backgrounded (Dry 1983; Hopper and Thompson 1980; Reinhart 1984). In practice the correlation is not that simple, the most notable exception being the historic present (Wolfson 1981). Secondly in the verbal group we have the features associated with aspect and situation type. These are the features that were the first to attract the attention of discourse-minded students of narrative, because in many languages, for instance in French and Russian, the behaviour of aspectual distinctions can sometimes seem haphazard, but there are regularities that can be explained by considering their role in marking grounding distinctions (Grimes 1975; Hopper 1979; Longacre 1981). Aspect and situation type come here together, because in some languages, such as English, they are very intricately intertwined and must be considered together. The features that are relevant include perfectivity, progressivity, dynamicity, telicity, punctuality and the uniqueness of the situation (eg Bartschat 1987; Chung and Timberlake 1985; Comrie 1976; Fleischman 1990; and articles in Hopper (ed.) 1982; Tedeschi and Zaenen (eds.) 1981; and Vet and Vetters (eds). 1994). The third feature concerns the actuality of the situation, the grammatical features of modality and polarity. The relation with foregrounding is here too simple in theory, because the standard correlation is that situations that are real

104

Brita Warvik

and affirmed are more likely to be foregrounded than situations that are uncertain, potential, irreal or negated (Chvany 1985a, 1985b; Hopper and Thompson 1980; Wallace 1984). So John finished his coffee is more foregroundable than John didn 't finish his coffee or John could have finished his coffee, but it is obvious that context is in these cases extremely important and even more so are our expectations about what usually happens (cf Chvany 1985a, 1985b). The fourth and last of the verbal criteria is volitionality; this feature is related to agentivity and in the Saliency hierarchy it is merged with it (Chvany 1985a, 1985b; Hopper and Thompson 1980). They are however independent, as agentivity refers to the capacity of the subject of acting as an agent and volitionality refers to the degree to which the situation requires volitional action by a participant; for instance the volitional action of drinking coffee matches the agentive subject John. In theory, only agentive participants can act volitionally, but they may just as well perform non-volitional actions (eg John burnt his tongue) and we need not look further than personifications for instances of nonagentive participants acting volitionally (eg The coffee helped him stay awake). In sum, the more volitional an action is, the more foregroundable it is. Finally of the grammatical criteria in the list is clause status, which subsumes features having to do with subordination and finiteness. In fact the relation of clause status to foregrounding is an ancient one, as in writing guides we can find instructions to put important things in main clauses and less important things in subordinate clauses, which tallies with the relationship between clause status and grounding as well. The main complication comes from the feature of clause status itself, as it has turned out to be more complex than a simple dichotomy. Here it is decomposed into five components: dependency, type of combination, taxis, restrictiveness and finiteness (cf Foley and van Valin 1984; Lehmann 1988). Last on the list is a feature that is on a different level than the others because here we need more context to decide how to classify a clause. The feature of temporal sequentiality is, however, the very first defining feature to be used in linguistic studies of narratives (Labov and Waletsky 1967; Reinhart 1984) and it still has a crucial role, though in a slightly modified form. Grounding distinctions have been relevant in all languages that have been the object of studies of narrative structure. The pervasiveness of the foregroundbackground distinctions has give rise to the question Why? In their article on Transitivity Hopper and Thompson say that Transitivity is important because it encodes grounding distinctions, but others have since asked the question: Why are grounding distinctions so important? Why is discourse organized in that way? There are three suggestions for correlates of grounding, which are not mutually exclusive.

What is foregrounded in narratives ?

2.

105

Correlate 1: Figure-ground distinction

Wallace (1982) and Reinhart (1984) have suggested that the foreground-background distinction is the textual correlate of the Figure-Ground distinction in visual perception. According to Gestalt psychology, the Figure-Ground distinction is a basic principle of perceptual organization. Figure 2 presents the main distinctive features of Figures vs Grounds. FIGURE thing-like, solid, discrete well-defined, tightly organized enclosed, surrounded, bounded small near above, in front lies upon, above, in front of the ground more in the centre of attention symmetric 'meaningful', familiar

GROUND unformed, diffuse, shapeless less definite, unstructured, loosely organized boundless, open large far below, behind continues unbroken under or behind the figure less in the centre of attention irregular 'meaningless', unfamiliar

Figure 2. Characteristics of Figure vs Ground

In terms of these features, we tend to see the small, closed, contoured shape as a Figure, as opposed to a larger, open, less definite shape or area, which we perceive as Ground. The element interpreted as Figure appears as lying on the Ground, which, consequently, appears to continue without interruption under or behind the Ground. The Figure also appears to be more in the centre of attention than the Ground. Such a characterization applies to the differentiation of a single Figure from a single uniform Ground; in general, however, the field of perception has a more involved structure with several potential Figures against various possible grounds. In such cases the Figures are organized into more complex Figures according to certain grouping principles, as listed in Figure 3. Let us look at some examples of how these principles work in visual perception; these are familiar from text books in psychology (eg Köhler 1964; Krech and Crutchfield 1982). The first set of images illustrates the principle of good continuation: the most natural interpretation of them is one that allows most continuation of lines and shapes and avoids breaking internal lines. So when we look at image A we tend to see two different and overlapping waves rather than a white version of image B; this interpretation allows us to see the two lines as continuous.

106

Brila Warvik

PROXIMITY

Elements that are spatially or temporally nearer each other tend to form a group.

SIMILARITY

Elements that in some way resemble each other are more likely to form a group than dissimilar elements.

GOOD FORM

Good continuation: the organization of elements that permits most continuation of a line, a contour, or a movement is better than one involving discontinuity or breaking of internal lines. Svmmetrv: symmetrical and balanced groupings are preferred to asymmetrical ones. Closure: the organization of elements in a closed or complete whole gives a better Figure than a grouping in an open or incomplete pattern. Common fate: groupings that combine elements moving or changing in a common direction will be better than groupings of elements moving or changing in separate directions in the field.

Figure 3. Gestalt principles for perceptual grouping

Similarly, in C we tend to see two vertical black lines which are crossed by a horizontal white line, rather than three white and two black squares on both sides of a white rectangle; this interpretation allows us to see the white line as continuous. B

Figure 4. Examples of the principle of good continuity

The images in Figure 5 show how size affects Figureness. Smaller areas are better Figures than larger ones. In the A image the black and white areas should be equally large, but in B the white areas are smaller. It is thus easier to see a white cross on a black circle in B rather than a black cross on a white circle, while A should be inherently ambiguous.

A

^—'

B

Figure 5. Examples of the criterion of size

What is foregrounded in narratives?

107

The last set of examples illustrate the criteria of proximity and closure. Proximity explains why A in Figure 6 tends to appear as three thin stripes plus one line at the right rather than three thick stripes and one line at the left. Closure is a strong criterion and outweighs proximity, as we can see in B, where we have square brackets instead of the lines of A. It is now easier to see three thick stripes and one extra bracket on the left. Another example of the strength of closure is provided by 4B: that we primarily see a group of more or less square black shapes can be explained by the closedness of the black areas.

A

B

Figure 6. Examples of the criteria of proximity and closure

As general characteristics of Figure and Ground, we may note that they are cluster concepts, that is each Figure and Ground must have some distinctive features of its group, but it need not have them all. As in all clusters, some of the criteria are stronger than others and thus prevail in cases of conflict. Moreover, Figure and Ground are relative concepts: they can be discerned only in relation to each other. Similarly, their characteristics are relative: Figures are, for instance, small only in relation to larger Grounds. Let us now turn back to language and to the linguistic parallels of the FigureGround distinction. Figure 7 shows another long list, including the suggestions of several researchers, who argue that the criteria for identifying Figures and Grounds in perception and in language are similar or parallel (Chvany 1985a, 1985b; Reinhart 1984; Talmy 1978b; Townsend and Bever 1977; Wallace 1982). First in the list are the nominal categories, which include features that appeared in the list of grounding criteria under the title individuation. These nominal features characterizing linguistic Figures are remarkably similar to certain properties of perceptual Figures: thing-like parallels human, animate and concrete, discrete parallels singular and count, well-defined parallels definite and referential (cf Östman and Virtanen 1999 for a different but compatible view). Further, the foregroundedness of human participants can be seen as analogue to the tendency of perceptual Figures to be in the centre of attention (cf below in section 4 on Speaker-motivation). Secondly we have the verbal categories. For them the parallels with perceptual Figures are less obvious and we can see them as metaphorically similar. Thus first, it is suggested that the verbal criterion of aspect is parallel to the criterion of closure for perceptual Figures; aspect is then considered prima-

108

Brita Wärvik

LINGUISTIC FIGURES LINGUISTIC GROUNDS nominal categories human nonhuman animate inanimate proper common singular plural concrete abstract definite indefinite referential nonreferential count mass verbal categories event state punctual durative perfective nonperfective completed incompleted 'temporally-contained' 'temporally-containing' present-immediate nonpresent-remote real, Occurring' irreal, 'non-occurrent' affirmative negative transitive intransitive actional verb stative verb deliberative action accidental action clause status main clause subordinate clause discourse categories sequential story-line off-story-line foreground background

Figure 7. Linguistic Figures vs linguistic Grounds

rily as the opposition between completed and incompleted events and actions. Completion, telicity and perfectivity correspond to the closedness and boundedness of perceptual Figures, while incompleteness, atelicity and imperfectivity correspond to the openness and indefinite boundaries of perceptual Grounds. Second, it is suggested that the punctuality of foregrounded events is parallel to the small size of perceptual Figures; punctual events cover less time, like smaller shapes cover less space, and in contrast, durative events cover more time like larger shapes cover more space. Thirdly, the correlation of tenses with grounding finds a parallel in the perceptual Figure-Ground distinction. The tenses referring to the story-wow, like the simple past and historic present in English, appear as Figures by virtue of referring to immediate, nearer events, whereas other tenses by virtue of referring to more remote times are more distant, like perceptual Grounds. In the fourth place, actually occurring, affirmed events resemble perceptual Figures in being more discrete and localized, whereas, irreal, nonoccurring or negated events resemble perceptual Grounds in being more diffuse, shapeless and unlocalized.

What is foregrounded in narratives?

109

Next on the list we have the status of the clause. Main clauses act as linguistic Figures and subordinate clause as linguistic Grounds. I must point out however, that though this characterization of main clauses as Figures and subordinate clauses as Grounds appears in many studies, no researchers have suggested exactly how clause status parallels the perceptual distinction of Figure and Ground (eg Talmy 1978a, 1978b; Townsend and Bever 1977). Finally, there are textual and discoursal parallels to the Figure-Ground distinction. For one the sequential story-line appears as Figure against the Ground of off-story-line material. We can draw a parallel between the temporal sequentiality of the story-line and the good continuity of perceptual Figures: the linguistic elements constituting a temporally sequential and hence continuous line form a linguistic Figure like elements organized along continuous contours and shapes form a perceptual Figure. I would also like to see a parallel in the principle of common fate. If, in general, the elements that move or change in the same direction appear as Figures in relation to elements moving or changing in other directions, the sequential story-line can be said to be a Figure because story-line elements represent a move or change forward on the story-line as opposed to off-story-line elements which by definition do not forward the story-line, but move or change in other directions, if they move or change at all. For the other, the foreground-background distinction parallels the FigureGround distinction. Foreground is Figure-like because foregrounded material is more in the centre of attention than backgrounded material. Foreground also shares the properties of good continuation and common fate with perceptual Figures. The principle of grouping by similarity supports these latter parallels. As Chvany argues, in the same way as a group of similar perceptual elements form a Figure distinguished from the Ground, foregrounded clauses "tend to cohere into a 'line' against a less organized background" (1985a: 10). This characterization tallies also with the common way of defining background negatively as the parts of discourse that are not foregrounded. To conclude on this correlate of grounding, I'd like to take up two points about the nature of this correlation. First then, though the perceptual FigureGround opposition and the textual foreground-background distinction are both fundamental principles of organization in their own spheres of human life, they are obviously not the only principles of organization guiding us in those spheres. And secondly, what kind of relationship is there between the perceptual Figure-Ground distinction and the textual grounding distinctions? Are they both manifestations of what Wallace (1982: 218) calls "a very broad sort of contrast which applies across traditionally separated areas of human cognition and human behavior", that is equal partners in organizational scheme. Or is the temporal organization of narrative texts, as Reinhart (1984: 805) describes it, "a

110

BritaWärvik

metaphorical extension of the spatial perceptual system". The latter suggestion tallies with the fact that terms of spatial relationships are often metaphorically used for expressing temporal relations. However, though this metaphorical extension applies nicely to the sequential story-line, it is perhaps not as convincing for the foreground-background distinction in its more general sense, for which the former alternative appears more suitable.

3.

Correlate 2: The EVENT prototype

Studying Hopper's and Thompson's Transitivity, Delancey (1987) argues that both the prominence of Transitivity in the morphosyntax of different languages and its function in encoding grounding distinctions are derivable from underlying cognitive schemata. He argues that the component features of Transitivity form a coherent cluster because they encode different aspects of a coherent semantic prototype, the EVENT. The foregrounding function of Transitivity arises from the cognitive salience of prototypical EVENTS (cf below in section 4 on Salience). Applying Rosch's principles of categorization, Delancey characterizes Transitivity as a natural category (Rosch 1978). Its naturalness is derivable from the fact that Transitivity encodes a CAUSE > EFFECT relationship, which is based on our perception of causes and their effects in the 'real world'. Delancey describes the EVENT schema underlying the prototypical Transitive clause as "a sequence of two events: a volitional act on the part of the agent, and a subsequent and consequent change on the part of the patient" (1987: 61). Such highly Transitive EVENTs are perceptually salient, and when sequences of events are encoded in language, often in the form of a narrative, their salience is reflected in the Transitivity of the clauses. Delancey's suggestion finds support in studies of language acquisition. Slobin (1982) argues that certain notions are salient to the child learning a language, and by virtue of their salience they have a special status in the early stages of language acquisition. One of these salient prototypes is the Transitive EVENT, as defined by the component features of Transitivity of Hopper and Thompson (1980). Such salient prototypes are early learnt and consistently used by children with different first languages (cf Berman and Slobin 1994). Further evidence for the relevance of foreground-background distinctions in language acquisition can be found in studies of second language acquisition. Thus, Bardovi-Harlig (1995) shows that foreground-background distinctions play a role in the acquisition of tense and aspect in a second language. Studying conversational stories by second-language speakers, Orletti (1995) found

What is foregrounded in narratives?

Ill

"that the subjects used the linguistic and communicative resources at their disposal to highlight [the] foreground/background contrast as much as possible" (1995: 192). Andersen and Shirai (1994) present evidence for the importance of grounding and other principles of discourse organization in tense and aspect use in first and second language acquisition.

4.

Correlate 3: Salience

The first correlate of grounding, the Figure-Ground distinction and the second correlate, the EVENT prototype are not competing motivations, but they are well compatible with each other, two perspectives on the same phenomenon. The third correlate of grounding is not incongruent with the others either. As a third parallel phenomenon I would like to suggest Salience as it is defined by Osgood (1980; Osgood and al. 1957; Wärvik 1990, 2002). The effects of this type of Salience have mostly been studied in features on the sentence level and below, mostly as to the ordering of words within constituents, of constituents within clauses and clauses within sentences. The points of interest to grounding studies are the Naturalness principle of ordering and the reasons motivating what Osgood calls 'unnatural' Salience, including the inherent Salience of elements. The Naturalness principle of ordering states that the 'natural' order of linguistic elements correlates with the order in which their referents are cognized in prelinguistic experience (Osgood 1980: HOff, 127ff). In clause-combining on the sentence level this principle is manifested in the tendency for clauses to reflect the order of the corresponding events/actions in the real world. Osgood presents evidence for a hierarchy of modes of conjoining according to the semantic and syntactic complexity of the relations (1980: 128ff). This hierarchy shows, among other things, that the 'easiest' types of conjoining are those where the clauses occur in 'natural' order and the signal for their relationship comes between them: for instance John waited until Mary came home is more 'natural' than Before Mary came home John waited. However, text-producers do not always follow 'natural' orders and we can meet quite 'unnaturally' ordered combinations of linguistic elements. Motivations for such 'unnatural' choices come from the different variables that add up to Salience; they are Vividness, Speaker-motivation, and Topicality. Osgood defines Vividness as arising out of the inherent Salience scores of the component elements. The inherent Salience of elements is characterized by the values attributed to them on the Evaluation, Potency and Activity scales. Evaluation is defined in terms of oppositions such as good vs bad, kind

112

BritaWarvik

vs cruel and honest vs dishonest. Potency covers oppositions such as strong vs weak, hard vs soft and heavy vs light, and Activity oppositions like fast vs slow and active vs passive. These clusterings of oppositions have turned out to recur in different languages and cultures and Osgood suggests that they reflect an affective meaning system common to all human beings. In these terms elements are defined as highly Salient in terms of Vividness, when they receive highly affective, that is strongly positive or negative values on the Evaluation, Potency and Activity scales, while the elements with neutral values are non-Salient. The second Salience variable, Speaker-motivation, refers to the tendency of human beings to consider as important everything that concerns, or is near or like ourselves. Thus the elements that are closer to us in space, time, nationality, race, sex, generation, profession and so on appear to us more Salient. Note that while Vividness is defined as a universal kind of Salience, inherent in the elements themselves, Speaker-motivation is extrinsic to the elements and depends on the speaker (or the producer or receiver of discourse). The third type of Salience, Topicality, is bound to the situation, so that elements that are nearer (temporally or spatially) or more familiar, more "recently cognized in perceptual or linguistic processing" (Osgood 1980: 192) tend to be interpreted as more Salient. Thus, Osgood proposes two concepts that influence the ordering of elements: Salience and the Naturalness principle. In general the three Salience variables are in accordance with the Naturalness principle, but this need not be the case, and some or all of the variables may override Naturalness. The links between the concepts introduced by Osgood and the features involved in the foreground-background distinction become more obvious when we expand the terms used by Osgood in his definitions to cover not only sentences, but also texts. Thus, the Naturalness principle of ordering tallies with the pervasiveness of what Enkvist (1981) has called experiential iconicism in textualization (cf Virtanen 1992). Hence it relates to temporal sequentiality, which is a defining feature of narratives and one of the basic characteristics of the organization of narratives around a sequential story-line. Drawing the correlation further, we can view the important structural role of the sequential story-line and its foregroundedness in simple stories as a consequence of its adherence to the Naturalness principle. In the same way as Naturalness is manifested in story-line sequentiality, the Salience variables are reflected as different aspects of foregrounding. While Salience overrides Naturalness, the other characteristics of foregrounding override temporal sequentiality in cases of conflict.

What is foregrounded in narratives?

113

The first Salience variable, Vividness, refers to the inherent Salience scores of elements and explains the salient positions given to elements with marked positive or negative scores. Drawing a parallel to grounding, we can evaluate clauses in terms of the inherent Salience of their components, for instance with the aid of the Transitivity Scale or another cluster of grounding criteria. In this sense, the Vividness of a clause can be related to the tendency of the most intense, active, dynamic, vivid actions and events to appear most foregrounded; or in more concrete terms, it can explain why punctual, transitive, and perfective actions are likely to appear highly foregrounded. The second variable, Speaker-motivation can explain why certain types of participants are more likely to occur in foregrounded clauses than others. Looking through the list of grounding criteria, we saw that the participants associated with foregrounding are typically anthropocentric; in other words the closer to ourselves the participants and the more consequential to us the event or action in the clause, the more foregrounded the clause is likely to be. Concepts close to Osgood's Speaker-motivation are the 'Me-First' principle by Cooper and Ross (1975), Ertel's Ego-closeness (1977), and Kuno's Speaker's empathy (1976) among others. The third Salience variable Topicality brings in the importance of the context. As already mentioned, this type of Salience is attributed to elements that are familiar to the interlocutors in being near in time or space in the situational context or recently mentioned in the textual context. In texts, high Salience in terms of Topicality is attributed to themes, topics, and given elements rather than rhemes, comments and new elements. In terms of grounding, this type of Salience is manifested primarily in the Topicality of participants, but choices in the individuation features are also related. To sum up on the correlation between Osgood's Salience and grounding, story-line sequentiality and grounding can profitably be seen as parallels to the concepts of Naturalness and Salience as defined by Osgood. The Naturalness principle, which motivates the ordering of elements on the sentence level and below, can be seen as part and parcel of the same force, experiential iconicism, that gives the temporally sequential story-line its structural prominence within the narrative text. Both Naturalness and sequentiality can be overridden by forces of Salience and foregrounding. Such 'unnatural' Salience is motivated by Vividness, Speaker-motivation and Topicality. Similarly, foregrounding, when distinguished from story-line sequentiality, arises out of high Transitivity or Saliency, human importance and continuity or Topicality of participants.

114

5.

BritaWarvik

Iconicity

Without contesting the basic arbitrariness of most linguistic signs, it has been argued that iconicity is an important force within the basically symbolic system of language. The role of iconicity in influencing the organization of linguistic expressions is discussed by Haiman (1980, 1983, 1985a, 1985b) and Givon (1984, 1985) and other contributors in Haiman (ed.) 1985. The origins of the ideas are traced back to Peirce (1932), de Saussure (eg Culler 1976), Jakobson ([1965] 1971, 1966), Greenberg (1966), and Benveniste (1966: 225ff, 1974: 43ff). To say that something in language is iconic may mean that a linguistic expression iconically reflects some aspect of the existing (extralinguistic) reality or that the particular expression iconically reflects our perception of some aspect ofthat reality.1 Iconicity can be imagic or diagrammatic (Haiman (1980) on the basis of Peirce's framework (1932)). Haiman defines an iconic image as "a single sign which resembles its referent with respect to some (not necessarily visual) characteristic" (1980: 515); examples of linguistic images are onomatopoeic words. While the role of imagic iconicity is marginal in languages, the diagrammatic kind of iconicity is claimed to have a near-universal status. An iconic diagram is defined as "a systematic arrangement of signs, none of which necessarily resembles its referent, but whose relationships to each other mirror the relationships of their referents" (Haiman 1980: 515). Haiman (1980) distinguishes between two types of diagrammatic iconicity. First, isomorphism refers to "a one-to-one correspondence between the signans and the signatum, whether this be a single word or a grammatical construction" (1980: 515). This type of iconicity is reflected in the universal aversion towards synonymy and homonymy. Secondly, motivation refers to the type of iconicity "in which a grammatical structure ... reflects its meaning directly" (1980: 516). Iconic motivation is found, for instance, in cases where formal complexity correlates with semantic complexity (eg comparison of adjectives, tense-aspect-mood forms) and in cases of iconicity of sequence. One type of iconic motivation is automorphism, which refers to a correlation between different parts of the same system; in other words, structural similarities in different parts of the system can be seen as reflecting their conceptual similarities (cf Haiman (ed.) 1985/Introduction). The iconicity of sequence, which exploits the obligatory linearity of the linguistic expression, is cited by Haiman as a conspicuous example of diagrammatic iconicity (1980, 1985a). The iconicity of sequence plays an important role in narratives, where, it can be argued, the elements that form the sequential story-line are in an iconic relationship to the corresponding situations in the 'real' or imagined world. The order of the story-line situations reflects ordo

What is foregrounded in narratives?

115

naturalis, experiential iconicism (Enkvist 1981), or the Naturalness Principle (Osgood 1980; cf section 4 above). The iconicity of the sequential story-line is essentially diagrammatic in kind, as the clauses forming the story-line are not (imagic) icons of the situations they depict, but the temporal relations between the story-line clauses in the narrative are diagrammatically iconic with the temporal relations between the situations depicted in them. While the diagrammatic iconicity of the temporally sequential story-line can be claimed to account for its structural importance in narratives, the pervasiveness of grounding distinctions can be argued to be a diagrammatically iconic reflection of another phenomenon in our perception of the events to be depicted in a narrative, that is, the Salience of the events. It can be hypothesized that the degrees of foregrounding iconically reflect the prominence relations between story-world elements. Such relations can be said to exist on the basis of some cultural or social convention or perceptual principle, or they may be relations that we wish to create in between those elements. Against this background, the correlations between grounding and the Salience variables can be reformulated in terms of iconicity.2 The Vividness associated with foregrounded elements can be considered as an iconic reflection of the inherent perceptual Salience of elements. In this way, the distinctions of grounding degrees between the clauses in the narrative can be viewed as correlating to the distinctions of Salience between the events depicted in the clauses. Speaker-motivation can be said to establish an iconic relation between the interlocutors (producer and receiver) and the participants in text. Introducing the concept of icons of grounding, Chvany (1985a, 1985b) argues that the foregroundedness of highly individuated participants arises from one kind of iconicity: such participants are icons of the speaker. To adopt the view-point of the receiver, we should perhaps add that the receiver is, obviously, guided by largely similar principles of anthropocentricity. Thus, it can be suggested that the anthropocentricity or Me-First principle associated with foregrounded elements is an iconic reflection of the human importance of the participants involved in the events depicted in those clauses. This iconic relation can be viewed as based on distance: how far (dissimilar) the producer/receiver is from the participant in terms of species, nationality, social status, age, ideology, and so on (cf the individuation criteria in Figure 1). Hence, the participants that are nearer to the interlocutors are more foregrounded (ie more likely to occur in foregrounded clauses) than participants that are more remote from them. The third variable of Salience, Topicality can be claimed to reflect iconically the contextual Salience of the elements, defined in terms of their familiarity, or the given/new distinction. This relation is a fairly concrete instance of the

116

BritaWärvik

iconicity of distance. According to Haiman, iconicity of distance can be argued for in cases where the "social distance between interlocutors corresponds to the length of the message" (1983: 783). In parallel terms we can consider the distance between the interlocutor (producer or receiver) and the participant in the text: when the participant is (supposed to be) known or given, it is expressed by a smaller amount of textual material than when it is (supposed to be) new. For instance, known participants tend to be referred to by pronouns, while new participants are more probably referred to by proper names or noun phrases (cf eg Givon 1983; Gundel et al. 1993; Figure 1 above). To sum up, there are two kinds of diagrammatically iconic relations between the organization of the elements in the narrative and the corresponding events in the 'real' or imagined world. On the one hand, the sequential story-line reflects an iconicity of temporal sequence. On the other hand, the narrative foreground reflects an iconicity of prominence relations (defined in terms of Salience) between the elements in the text and those in the depicted world. As has been pointed out, in simple stories foreground and sequential storyline are merged. There the different relations of iconicity are fairly straightforward, as it can be claimed that the sequentiality of the story-line elements in simple stories implies their grounding degree, by attributing higher prominence to sequential elements. When we consider more complex narratives, we have to take into account the fact that the text-producer is, at least in principle, free to ground the text by her/his own weighting of the grounding criteria. Thus, we have to take quite literally the less ambitious view of iconicity, according to which a linguistic expression is iconic when its form or organization reflects some aspect of our perception of the 'reality'. This kind of iconicity covers also the Salience relations that reflect the specific perceptions that the producer of the narrative text wants to communicate and that its receiver interprets as existing between the story-world elements. The relations between the producer's intentions and the receiver's interpretations are of course not without problems, but here I have (naively) assumed that the interlocutors do not wilfully try to avoid reaching a common view of grounding distinctions (and, consequently, their correlates) in a narrative text.

6.

Conclusion

What can the preceding discussion add to our understanding of story-line sequentiality and grounding? Viewing these two organizational principles of narrative text in terms of iconicity does not give us any methods or instruments for analysis. What this perspective may contribute is a (potential) motivation for

What is foregrounded in narratives?

117

the structurally important role of story-line sequentiality and grounding in the narratives of the languages investigated to date. In the present framework of grounding, the iconicity of the relations is one motivation; another motivation is their Figureness. The interest and value of both concepts lies in their applicability to different fields of human activity, as principles guiding perception and guiding linguistic as well as other forms of expression.

Notes 1. The relations between reality and perceptions obviously raise a number of philosophical problems, which are outside the scope of this chapter. For my present purposes it is enough to accept the latter, more prudent view of iconicity, drawing a parallel only between perception and expression. 2. This tallies with Haiman's (1985a) discussion of symmetry, where he models temporal sequentiality and importance as two different, though intersecting dimensions. Parallel pairs of features can also be seen in Givon's (1995) semantic and pragmatic principles of ordering.

References Andersen, Roger W. and Yasuhiro Shirai 1994 Discourse motivations for some cognitive acquisition principles. Studies in Second Language Acquisition 16: 133-156. Bäcklund, Ingegerd 1988 Grounds for prominence: On hierarchies and grounding in English expository text. Studio Neophilologica 60: 37-61. Bardovi-Harlig, Kathleen 1995 A narrative perspective on the development of the tense/aspect system in second language acquisition. Studies in Second Language Acquisition 17: 263-291. Bartschat, Brigitte 1987 Aspekt und 'grounding' in russischen Texten. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 40: 758-771. Benveniste, Emile 1966 Problemes de Linguistique Generale I. Paris: Gallimard. 1974 Problemes de Linguistique Generale II. Paris: Gallimard. Berman, Ruth and Dan I. Slobin 1994 Relating Events in Narrative: A Crosslinguistic Developmental Study. Hillsdale, NJ and Hove, UK: Lawrence Erlbaum. Brown, Gillian and George Yule 1983 Discourse Analysis. Cambridge etc.: Cambridge University Press.

118

BritaWarvik

Chafe, Wallace L 1976 Givenness, contrastiveness, definiteness, subjects and topics. In: Charles N. Li (ed.), Subject and Topic, 25-55. New York etc.: Academic Press. Chung, Sandra and Alan Timberlake 1985 Tense, aspect, and mood. In: Timothy Shopen (ed.), Language Typology and Syntactic Description, Vol. Ill: Grammatical Categories and the Lexicon, 202-258. Cambridge etc.: Cambridge University Press. Chvany, Catherine V. 1985a Foregrounding, 'Transitivity', Saliency (in sequential and non-sequential prose). Essays in Poetics 10/2: 1-27. 1985b Backgrounded perfectives and plot line imperfect!ves: Toward a theory of grounding in text. In: Michael S. Flier and Alan M. Timberlake (eds.). The Scope of Slavic Aspect, 247-273. (UCLA Slavic Studies 12.) Columbus, Ohio: Slavica. Comrie, Bernard 1976 Aspect: An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge etc.: Cambridge University Press. Cooper, William. E. and John. R. Ross 1975 World order. In: Robin E. Grossman, L. James San, and Timothy J. Vance (eds.), Papers from the Parasession on Functionalism, 63-111. Chicago: Chicago University Press. Culler, Jonathan 1976 Saussure (Fontana Modern Masters). Glasgow: Fontana Paperbacks. Delancey, Scott 1987 Transitivity in grammar and cognition. In: Russell S. Tomlin (ed.), Coherence and Grounding, 53-68. Amsterdam / Philadelphia: John Benjamins. Dry, Helen 1983 The movement of narrative time. Journal of Literary Semantics 12/2: 19-53. Enkvist, Nils Erik 1981 Experiential iconicism in text strategy. Text 1: 77-111. Ertel, Suitbert 1977 Where do the subjects of sentences come from? In: Sheldon Rosenberg (ed.), Sentence Production: Developments in Research and Theory, 141-167. Hillsdale, N.J.: Lawrence Erlbaum. Fleischman, Suzanne 1990 Tense and Narrativity: From Medieval Performance to Modern Fiction. London: Routledge. Foley, William A. and Robert D. van Valin J. 1984 Functional Syntax and Universal Grammar. (Cambridge Studies in Linguistics 38.) Cambridge etc.: Cambridge University Press. Garrod, Simon and Tony Sanford 1988 Thematic subjecthood and cognitive constraints on discourse structure. Journal of Pragmatics 12: 519-534.

What is foregrounded in narratives ?

119

Givon, Talmy 1983 Topic continuity in discourse: An introduction. In: Talmy Givon (ed.), Topic Continuity in Discourse: A Quantitative Cross-language Study, 1—41. (Typological Studies in Language 3.) Amsterdam and Philadelphia: John Benjamins, 1-41. 1984 Syntax: A Functional Typological Introduction. Volume 1. Amsterdam and Philadelphia: John Benjamins. 1984 Iconicity, isomorphism and non-arbitrary coding in syntax. In: John Haiman (ed.), 187-219. Iconicity in Syntax. (Typological Studies in Language 6.) Amsterdam and Philadelphia: John Benjamins. 1995 Isomorphism in the grammatical code: Cognitive and biological considerations. In: Raffaele Simone (ed.), Iconicity in Language, 47-76. (Current Issues in Linguistic Theory 110.) Amsterdam / Philadelphia: John Benjamins. Greenberg, Joseph H. 1966 Some universale of grammar with particular reference to the order of meaningful elements. In: Joseph H. Greenberg (ed.), Universals of Language, 73-113. Cambridge, Mass.: The M.I.T. Press (2nd ed.). Grimes, Joseph E 1975 The Thread of Discourse. (Janua Linguarum. Series Minor 207.) The Hague / Paris: Mouton. Gundel, Jeanette K., Nancy Hedberg and Ron Zacharski 1993 Cognitive status and the form of referring expressions in discourse. Language 69: 274-307. Haiman, John 1980 The iconicity of grammar: Isomorphism and motivation. Language 56: 513-540. 1983 Iconic and economic motivation. Language 59: 781-819. 1985a Symmetry. In: John Haiman (ed.), Iconicity in Syntax, 73-95. (Typological Studies in Language 6.) Amsterdam / Philadelphia: John Benjamins. 1985b Natural Syntax: Iconicity and Erosion. (Cambridge Studies in Linguistics 44.) Cambridge etc: Cambridge University Press. Haiman, John (ed.) 1985 Iconicity in Syntax. (Typological Studies in Language 6.) Amsterdam / Philadelphia: John Benjamins. Hopper, Paul J. 1979 Aspect and foregrounding in discourse. In: Talmy Givon (ed.), Discourse and Syntax, 213-241. (Syntax and Semantics 12.) New York etc.: Academic Press. Hopper, Paul J. (ed.) 1982 Tense-Aspect: Between Semantics and Pragmatics. (Typological Studies in Language 1.) Amsterdam / Philadelphia: John Benjamins. Hopper, Paul J. and Sandra A. Thompson 1980 Transitivity in grammar and discourse. Language 56: 251-299.

120

Brita Wärvik

Hopper, Paul J. and Sandra A. Thompson (eds.) 1982 Studiesin Transitivity. (Syntax and Semantics 15.)NewYorketc.:Academic Press. Jakobson, Roman 1965 Quest for the essence of language. Diogenes 51: 21-37. (Reprinted 1971 in Selected Writings, Vol. 2, 345-359. The Hague / Paris: Mouton.) 1966 Implications of language universals for linguistics. In: Joseph H. Greenberg (ed.), Universals of Language, 263-278. Cambridge, Mass.: The M.l.T. Press (2nd ed.). Keenan, Edward L. 1976 Towards a universal definition of 'subject'. In: Charles N. Li (ed.), Subject and Topic, 303-333. New York etc.: Academic Press. Köhler, Wolfgang 1964 Gestalt Psychology: An Introduction to New Concepts in Modern Psychology. New York: New American Library. Krech, David and Richard S. Crutchfield 1982 Elements of Psychology. New York: Alfred A. Knopf. (1st printing 1958.) Kuno, Susumo 1976 Subject, theme, and the speaker's empathy: A re-examination of relativization phenomena. In: Charles N. Li (ed.), Subject and Topic, 417-444. New York etc.: Academic Press. Labov, William and Joshua Waletzky 1967 Narrative Analysis: Oral versions of personal experiences. In: June Helm (ed.), Essays on Verbal and Visual Arts: Proceedings of the 1966 Annual Spring Meeting of the American Ethnological Society, 12-44. Seattle: University of Washington Press. Lehmann, Christian 1988 Towards a typology of clause linkage. In: John Haiman and Sandra A. Thompson (eds.), Clause Combining in Grammar and Discourse, 181-225. (Typological Studies in Language 18.) Amsterdam / Philadelphia: John Benjamins. Li, Charles N. (ed.) 1976 Subject and Topic. New York etc.: Academic Press. Longacre, Robert E. 1981 A spectrum and profile approach to discourse analysis. Text 1: 337-359. Lyons, John 1977 Semantics. Vols, 1-2. Cambridge etc.: Cambridge University Press. McGann, William and Arthur Schwartz 1988 Main character in children's narratives. Linguistics 26: 215-233. Orletti, Franca 1995 Figure and Ground in second language narratives: Traces of iconicity. In: Raffaele Simone (ed.), Iconicity in Language, 171-195. (Current Issues in Linguistic Theory 110.) Amsterdam / Philadelphia: John Benjamins.

What is foregrounded in narratives ?

121

Osgood, Charles E. 1980 Lectures on Language Performance. New York etc.: Springer-Verlag. Osgood, Charles N., George J. Suci and Percy H. Tannenbaum 1957 The Measurement of Meaning. Urbana, 111.: University of Illinois Press. Östman, Jan-Ola and Tuija Virtanen 1999 Theme, comment, and newness as Figures in Information Structuring. In: Karen van Hoek, Andrej A. Kibrik and Leo Noordman (eds.), Discourse Studies in Cognitive Linguistics, 91-110. (Current Issues in Linguistic Theory 176.) Amsterdam / Philadelphia: John Benjamins. Peirce, Charles S. 1932 The Collected Papers of Charles Sanders Peirce (Edited by Charles Hartshome and Paul Weiss), Vol. II, Elements of Logic. Cambridge, Mass.: Harvard University Press. Prince, Ellen F. 1981 Toward a taxonomy of given-new information. In: Peter Cole (ed.), Radical Pragmatics, 223-255. New York etc.: Academic Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London and New York: Longman. Reinhart, Tanya 1984 Principles of Gestalt perception in the temporal organization of narrative texts. Linguistics 22: 779-809. Rosch, Eleanor 1978 Principles of categorization. In: Eleanor Rosch and Barbara B. Lloyd (eds.), Cognition and Categorization, 27-48. New York: John Wiley. Simone, Raffaele (ed.) 1995 Iconicity in Language. (Current Issues in Linguistic Theory 110.) Amsterdam / Philadelphia: John Benjamins. Slobin, Dan 1982 The origins of grammatical encoding of events. In: Paul J. Hopper and Sandra A. Thompson (eds.), Studies in Transitivity, 409-422. (Syntax and Semantics 15.) New York etc.: Academic Press. Sundman, Marketta 1987 Subjektval och diätes i svenskan. Abo: Abo Academy Press. Talmy, Leonard 1978a Relations between subordination and coordination. In: Joseph H. Greenberg (ed.), Universals of Human Language. 4 Vols, 487-513. Stanford: Stanford University Press. 1978b Figure and Ground in complex sentences. In: Joseph H. Greenberg (ed.), Universals of Human Language. 4 Vols, 625-649. Stanford: Stanford University Press. Tedeschi, Philip and Anne Zaenen (eds.) 1981 Tense and Aspect. (Syntax and Semantics 14.) New York etc.: Academic Press.

122

Brita Warvik

Tomlin, Rüssel S. 1983 On the interaction of subject, thematic information, and agent in English. Journal of Pragmatics 7: 411-432. Townsend, David J. and Thomas G. Bever 1977 Main and Subordinate Clauses: A Study in Figure and Ground. Bloomington, Indiana: Indiana University Linguistics Club. Vet, Co and Carl Vetters (eds) 1994 Tense and Aspect in Discourse. (Trends in Linguistics, Studies and Monographs 75.) Berlin / New York: Mouton de Gruyter. Virtanen, Tuija 1992 Discourse Functions of Adverbial Placement in English: Clause-initial Adverbials of Time and Place in Narratives and Procedural Place Descriptions. Abo: Abo Akademi University Press. Wallace, Stephen 1982 Figure and Ground: The interrelationships of linguistic categories. In: Paul J. Hopper (ed.), Tense-Aspect: Between Semantics and Pragmatics, 201-223. Amsterdam / Philadelphia: John Benjamins. Wolfson, Nessa 1981 Tense variation in narrative. Language 57: 45-62. Wärvik, Brita 1990 Grounding in narratives: What is foregrounded? In: Werner Bahner, Joachim Schildt and Dieter Viehweger (eds.), Proceedings of the Fourteenth International Congress of Linguists, Berlin/GDR, August 10-August 15, 1987, 2253-2256. Berlin: Akademie-Verlag. 2002 Grounding in Narratives: A Survey of Models and Criteria. (English Department Publications, Abo Akademi University, 5.) Abo: Abo Akademi University Press. Yamamoto, Mutsumi 1999 Animacy and Reference: A Cognitive Approach to Corpus Linguistics. Amsterdam / Philadelphia: John Benjamins.

Chapter 7 From legal knowledge to legal discourse - and back again Lita Lundquist

If one takes cognition to mean 'concerned with knowledge and the structuring of knowledge' - and this is exactly what the Encyclopaedia Britannica does in defining cognition as "the mental activities involved in the acquisition, processing, organisation, and use of knowledge" - cognition is involved in several respects in the present chapter. This chapter reports a production experiment in which cognition was tested by asking two groups of subjects to write what they knew about two concepts belonging to the specialised domain of legal knowledge. As the aim was to investigate differences in knowledge, one group of subjects was selected among legal experts, the other among non-experts, one group thus knowing more than the other. The approach taken to the specific cognitive problem of comparing differences in knowledge was via discourse, namely via the production protocols obtained in the experiment. In order to judge differences in knowledge, the following questions were posed: in what respect do experts know more than non-experts, how do we get at these differences, and how do we characterise them? The production experiment was carried out with the aim of getting at what experts and non-experts know from what they write. So, in more precise terms, the questions asked are, how do we isolate, categorise and evaluate differences between experts' and non-experts' knowledge via the texts they write? The general methodological problem raised at this point is that of judging cognitive categories from the linguistic surface structure of texts, and in the present study the more specific question of getting at cognitive categories which are typical of legal reasoning. Finding a method to establish one-to-one relations between linguistic and cognitive structures presents an intriguing task, just as it seems difficult to stop the process of going from knowledge to language (the production phase), from language to knowledge (the reception and analysis phase), from knowledge to language (the description phase), and so forth. There simply seems to be no last turtle here. I have not solved the problem of using language and thought as "instruments for analyzing themselves" (Fauconnier 1997: 3), but I venture to propose a method which combines linguistic categories with cognitive structures typical

124

Lita Lundquist

of legal reasoning. The linguistic structures to be used here are the so-called qualia from the semantic theory of the generative lexicon, and the legal structures stem from the dynamic reasoning of the if-then type. This method makes it possible to segment the text surface of the protocols into cognitive units, which then form the basis of both qualitative and quantitative analyses.

1.

Knowing more and knowing less

It is commonly accepted and has been experimentally proved (e.g. Barsalou 1992) that knowledge is structured, and different models, such as frames (Fillmore 1985), mental models (Johnson-Laird 1983), and Idealised Cognitive Models (Lakoff 1990), have been proposed to describe these structures. Knowing more, then, about a specific domain means having a more elaborate structure of that domain, an elaboration which consists in knowing more concepts (nodes) and more relations (links) between the concepts. The different types of mental organisation of knowledge are most often described in very general terms, such as: "a frame includes a cooccurring set of abstract attributes that adopt different values across exemplars" (Barsalou 1992: 37); and the distinctions between expert and novice knowledge are described in equally general terms: "a novice's frame for horse might contain fewer attributes and values than an expert's frame" (Barsalou 1992: 64). As concerns specialised knowledge, however, it is fair to ask what is specific for frames and mental models used to organise and structure it (Lundquist and Jarvella 2000). 1.1. Specialised knowledge Specialised knowledge has been addressed, if only indirectly, in two semantic theories, which claim to be 'realistic' in that they take into account that one or more of the contextual factors, such as the speaker and his/her knowledge at a given time and place, may vary (cf. Lundquist 1989). Thus Kripke (1972) operates with a priori properties of a term, known to lay people as the result of an 'initial baptism', as opposed to a posteriori properties which are discovered by experts via scientific investigation and named at a later baptising ceremony (Kripke 1972: 328). Kripke's theory applies first and foremost to 'terms for natural kinds' stemming from natural sciences, i.e., terms such as gold, heat, and water, but his distinction between 'sufficient properties' in ordinary language use, and 'essential, necessary properties' in scientific language use, may prove to be useful also to the analysis of concepts belonging to the domain of legal knowledge, i.e., concepts from social sciences.

From legal knowledge to legal discourse - and back again

125

Whereas Kripke's semantic theory is 'diachronic' in that it opposes a priori and a posteriori knowledge, Putnam (1975) presents a synchronic version of a realistic semantic theory. Furthermore, Putnam's version is 'social' in that it operates with the concept of the 'Social Division of Linguistic Labor' between experts and non-experts. For non-experts, again in the domain of natural sciences, stereotypes, i.e., conventional ideas, are sufficient because '"hidden structures' become irrelevant, and superficial characteristics become the decisive ones" (Putnam 1975: 241). Experts, on the other hand, are experts exactly because they know the hidden criteria of a term, again of terms belonging to natural sciences, and they know these via "methods of identifying ... which have grown incredibly sophisticated" (Putnam 1975: 235). If we turn from natural sciences to social sciences, it has been shown by Noordman, Vonk and Simons (2000), through a series of knowledge eliciting experiments, that within the domain of economic knowledge, expert knowledge as compared to non-expert knowledge is characterised by the following features: - Experts make more associations from and between economic concepts presented. - Experts can explain more relations between associated concepts. - Experts more often explain relations in causality chains than in superordinate and subordinate concepts. - Experts show more similarity in their answers than non-experts. These findings confirm the hypothesis put forward by the authors that "experts make use of more complex knowledge schemata in which the knowledge is organised according to more general categories and principles of the knowledge domain" (Noordman et al. 2000: 238). These categories and principles stem from economic theories, which may differ, but which have in common the fact "that the economic reality is considered a system of related factors." (Noordman et al. 2000: 236; my emphasis). Thus, specialised knowledge domains are different from each other in that they are structured along the specific theory/theories holding for them, i.e., in a "coherent body" (Noordman et al. 2000) of concepts and relations between concepts. This goes especially for social science domains where things are not discovered by scientific investigation as in natural sciences, but rather named, agreed upon and defined as "counting as": a ten dollar bill counts as ten dollars by social convention, and a contract counts as a contract by legal definition (Searle 1995, 1998). In this chapter, I shall focus on the latter type of knowledge, i.e., on the structuring of legal knowledge. Let us assume, then, that specialised knowledge involves knowledge of precise distinctions and criteria, relations and causes. This point was already

126

Lita Lundquist

mentioned by Aristotle, to whom we shall return, in his Metaphysics: "Clearly then wisdom is about certain causes and principles ... he who is more exact and more capable of teaching the causes is wiser, in every branch of knowledge" (Ackrill 1987: 257). 1.2. Legal knowledge For legal knowledge, it has been pointed out that a specific type of conceptual relation prevails. Several legal philosophers and other legal scholars have suggested that legal knowledge is rule-based. Rather than being organised in a static, vertical structure with super-concepts and sub-concepts, it is organised along a dynamic horizontal structure of implication, in an if-then structure. Kjaer (2000) convincingly describes the function of the if-then structure in legal experts' interpretation of different types of legal discourse, and as her approach seems to be useful for the analysis of the present production protocols, her argumentation will be set out briefly. Referring to a number of legal experts (e.g., to the Danish legal philosopher Alf Ross), she shows that it is "legal rules (that) determine or prestructure the expert's conceptual knowledge of his subject" (Kjaer 2000: 146; my emphasis). This conceptual model of rule-based reasoning can be formulated in a 'connected if-then structure', in which two if-then structures are nested into one. An example is: If 'purchase'

then Ownership' If Ownership' then 'right to dispose' (Kjasr 2000: 134, her figure 5b, quoting Eckhoff 1969: 65) This three-step model corresponds to three steps in the basic legal reasoning pattern: (1) the legal condition, (2) the insertion of the legal connecting concept, and (3) the legal effect. The three steps are held together in a dynamic model, going from left to right. legal condition

connecting concept

I

legal effect

Figure 1. Conditional model of the basic legal reasoning type

The model is an abstract model, or, as stressed in Kjaer, a "hypothetical construct" which has been "introduced into the science of law by jurisprudence as a simple and concise way of describing legal rules" (Kjasr 2000: 133). This

From legal knowledge to legal discourse - and back again

127

abstract rule-model is conceived of as a mental model which is basic for and characteristic of legal knowledge: As a characteristic pattern of legal reasoning, the if-then structure can be regarded as a mental model of legal knowledge, i.e., as a cognitive model which structures legal thinking and is used by legal experts informing categories and in reasoning. (Kjaer 2000: 150; my emphasis)

For the purpose of getting at cognitive categories in the production protocols, I shall retain from Kjaer's argumentation the central idea here, namely that the frame or mental model which is basic for legal knowledge is organised in a dynamic, horizontal structure of implication, with a condition and an effect, or, in even more general terms, in a dynamic structure with a 'before' and an 'after': before

concept

I

after

Figure 2. Dynamic model of the basic legal reasoning type

7.2.7. Knowledge structure of the legal concepts 'contract'and 'judgement' Here, I will report the result of a first step in the experiment on the knowledge experts and non-experts have of the two legal concepts, 'contract' and 'judgement', in order to illustrate this dynamic model of legal reasoning in a before and an after. A legal expert, who is a lawyer and assistant professor at the Law Department at Copenhagen University, was asked to explain how he conceived the basic structure of the two concepts in question. The categories he mentioned and the relations he explained between these basic categories are reproduced in Figures 3 and 4 for 'contract' and 'judgement', respectively. parties writing

agreement

binds (only parties) can be executed

Figure 3. Contract. Expert's explanation of categories and temporal sequencing

court

decision/conclusion of a legal case (civil/criminal)

execution

Figure 4. Judgement. Expert's explanation of categories and temporal sequencing

128

Lita Lundquist

It appears from Figures 3 and 4 that the expert's intuitive explanation of the basic structure of the two concepts 'contract' and 'judgement' contains a dynamic structure parallel to the one proposed by Kjaer, and illustrated in Figures 1 and 2, above.

2.

From legal structure to linguistic structure, and vice versa, via qualia

Using the mental model of the legal connecting conditional structure, we are still facing the problem of getting from the linguistic surface structure of the production protocols to the cognitive categories of the three steps of conditionconcept-effect; or, vice versa, how to get from the legal cognitive categories to units in the surface structure of the texts. In order to combine cognitive and linguistic units, I have used the semantic notion of qualia from Pustejovsky's The Generative Lexicon (1995). Pustejovsky operates with four types of qualia, which have been inspired by Aristotle's four 'modes of explanation' (aitiae). Aristotle's ideas are interesting because they are directly related to cognition, in that he addresses understanding via these aitiae., which explain the '"how and why' of things coming into existence and passing out of it" (Aristotle 1970: 194bl6). Aristotle's emphasis on knowledge and understanding appears from the following quotation: "It is clear that we must obtain knowledge of the primary causes, because it is when we think that we understand its primary cause that we claim to know each particular thing. Now there are four recognised kinds of cause" (Aristotle 1970: 17). These four main classes of 'essential and causal determinants' (ibid. 133) are described in Physics: Well then, (1) the existence of material for the generating process to start from ... is one of the essential factors we are looking for ... (Material aitia). Then, naturally, (2) the thing in question cannot be there unless the material has actually received the form or characteristic of the type ... (Formal aitia). Then again, (3) there must be something to initiate the process of the change or its cessation... such as the act of a voluntary agent... or more generally the prime, conscious or unconscious, agent that produces the effect and starts the material on its way to the product... (Efficient aitia). And lastly, (4) there is the end or purpose, for the sake of which the process is initiated... (Final aitia). (Aristotle 1970: 194bl6)

Aristotle's four modes of explanation have been the inspiration for Pustejovsky's (1995: 76) semantic description of the lexicon in terms of qualia: "By looking at modes of explanation for a word, we permit a much richer description of meaning than either a simple decompositional view or a purely relational

From legal knowledge to legal discourse - and back again

129

approach to word meaning would allow." Pustejovsky's aim is to describe how lexicon is creative and explain how it contributes to the enrichment and extension of meanings in context. His focus is on words, especially nouns and verbs, but he emphasises that "qualia are in fact structures very much like phrase structural descriptions for syntactic analysis" (Pustejovsky 1995: 76). He also points to the fuzzy interface between lexical knowledge and common-sense knowledge: "Once we start enriching our lexicon with information that, to a linguist, appears better suited for a knowledge base, there may appear to be no systematic means to judge where to stop" (Pustejovsky 1995: 232-233). These quotations give us a hint that qualia analysis may also be useful in the analysis of sentences and texts (see also Lundquist 2000), and in comparing linguistic and cognitive structures, be they common-sense or specialised. Reformulating Aristotle's four modes of explanation, Pustejovsky (1995: 8586) suggests the following four components of Qualia: - constitutive: the relation between an object and its constituents, or proper parts (such as material, weight, parts) - formal: that which distinguishes the object within a larger domain (orientation, magnitude, shape, etc.) - telic: purpose and function of the object (e.g., purpose that an agent has in performing an act) - agentive: factors involved in the origin or 'bringing about' of an object (e.g. creator, artifact, natural kind, causal chain). Whereas the two first Qualia components, Constitutive and Formal (corresponding to Aristotle's 'material' and 'formal' cause) refer mostly to the static structuring of knowledge, the two latter components, Agentive and Telic (Aristotle's 'efficient' and 'final' cause), clearly involve a dynamic notion as the Agentive quale describes the "bringing about" of the concept, and the Telic quale describes the purpose and the function, i.e., the "coming after". I have used this lexical approach as a methodological framework and as a tool for analysing the linguistic surface of the production protocols. As Constitutive and Form are difficult to keep apart in a concrete analysis, I have combined them into one, called Form in the following. I thus use the three qualia Agentive, Form, and Telic as cognitive-linguistic categories to segment the linguistic surface structure and combine it with the conditional model which is typical of legal reasoning: Agentive/Condition

Form/Concept

Telic/Effect

Figure 5. Qualia combined with the three step legal reasoning model

130

Lita Lundquist

In the production experiment reported in the next section, each protocol, which was produced by either an expert or a non-expert on the two concepts 'contract' and 'judgement', was analysed into these three qualia structures by two different raters: an expert in law (the above-mentioned lawyer and assistant professor, referred to as NN below) and a non-expert in law (myself, LL in the following). Three main hypotheses were explored in the experiment: (1) There will be a difference between experts' and non-experts' texts. This difference will be visible in knowledge structures as analysed in terms of qualia. (2) Experts point to more qualia than non-experts. This goes both for producers where the experts know more and can therefore write texts with more refined categories, and for receivers, here the raters, where the expert will perceive more and more highly refined categories than the non-expert. (3) Experts see more Agentive and Telic qualia than non-experts because they know more about causes and consequences. Non-experts will focus more on Form.

3.

The experiment

3.1. Method In the experiment reported here two groups of subjects were asked to produce two out of four different types of texts, namely encyclopaedic entries in a short or a long form for the two concepts contract and judgement. One group of subjects was composed of non-experts found among first year language students (39 students who were paid for their participation) at the Language Faculty at the Copenhagen Business School. The second group was composed of experts (8 experts), legal scholars from the Law Department at Copenhagen University. The students were not complete novices in the field of law since, as part of their language studies in LSP, they had taken an introductory course in commercial law. They were thus not 'no-knowledge' people but 'low-knowledge people' (Noordman et al. 2000: 236), which is of course a prerequisite for comparing with experts at all: The comparison only makes sense if the experts and non-experts have some knowledge in common: if they can be located at the same dimension. The nonexperts should be low-knowledge people and not no-knowledge people. The low-knowledge people should have some basic knowledge in the domain. They

From legal knowledge to legal discourse - and back again

131

Table 1. Examples of texts Expert Texts

Student Texts

Contract Short

Written agreement between 2 or more parties.

A written agreement between two partners.

Contract Long

Consists generally of legally binding declarations of intent from 2 or more parties. A declaration of intent is usually a promise whereby a contracting party undertakes to perform some service, which may be conditional on the other contracting party performing a service in return. A contractual agreement may exist without the existence of any express declarations of intent, the agreement being legally binding as a result of the conduct of the contracting parties in respect of each other.

A written agreement between two parties. Specifies in detail the conditions under which the agreement has been entered into. Lays down rules to be applied in cases of disagreement between the parties. Security against any form of misunderstanding occurring. Any breach of contract is settled by a court of law.

Judgement Short

Settlement by a court of law of a The decision made by a court conflict between 2 or more parties. of law in a case of legal action against a person who has acted in conflict with the law or who has committed a crime.

Judgement Long

A judgement is a decision made by a court of law which determines how a legal conflict should be settled. The judgement establishes the legal position of the conflicting parties. A judgement may be enforced. For example, a judgement pertaining to payment of a sum of money may be enforced with the aid of the enforcement court, and a judgement concerning parental custody means that the party awarded custody may resort to the bailiff in order to gain physical custody of the child.

After his or her case has been conducted in court, a person who is charged with having done something illegal can by a judge be given a sentence like a fine, a fortnight in detention, imprisonment for an indefinite period, and so on. The judgement consists in the final decision of the court as to what measures are to be taken against the person. How severe the judgement is depends on the degree of the crime.

132

Lita Lundquist

should know the concepts and have some understanding of the domain. Otherwise, what would we be dealing with in investigating them? (Noordman et al. 2000:236)

The expert group, on the other hand, can be considered to have the highest level of legal expertise, holding qualifications for an academic career in law. The expert group included both Ph.D. students, lecturers and full professors (see alsoPalsbro2000:213). The two groups were tested separately, each group receiving a booklet containing several tasks. Among these tasks the subjects were asked to write a short (a long) encyclopaedic entry for the concept contract (judgement). The subjects each wrote one short and one long text, on either the one or the other concept, e.g., a short text on contract and a long text on judgement, or vice versa. The outcome of the experiment was 39 χ two protocols + 8 χ two protocols, all of which were subsequently typed. For the present study, a subgroup of 8 protocols of each type was selected at random from among the non-expert protocols in order to balance the representation between the two groups. The set submitted to analysis comprises 32 protocols coming from 8 experts and 14 nonexperts^ and representing the two concepts contract ana judgement and the two forms short and long. Table 1 shows an exemplar of each type of text, i.e., eight texts, translated from Danish into English. To get an idea of the corpus analysed, multiply by four. Appendix 1 contains the Danish versions of the same texts. 3.2. Analyses and results 3.2.1. Ratings by a legal expert and his confidence In a first analysis, all 32 texts (in their Danish versions) were submitted to the legal expert, who was asked (1) to judge for each text whether it had been written by an expert or a nonexpert (2) to evaluate how confident he was in his judgement The results in Table 1 show a high degree of correctness in judging plus a high degree of confidence.2 This result supports our first hypothesis that there is a difference between experts' and non-experts' texts. In order to describe these differences more closely, the expert was further asked to explain on what criteria he had based this judgement. The list follows in Table 3, to which the expert added the general remark that he had felt most doubts in texts with lapidary style, such as "(a contract) written agreement between two or more parties", i.e., in short texts. It is also here that most incorrect judgements were made.

From legal knowledge to legal discourse - and back again

133

Table 2. Proportion of texts produced by experts and non-experts which were correctly identified as such by a legal expert, and mean confidence of his correct identifications (5-point scale; 1 = unsure, 5 = very sure).

Proportion correct Confidence rating

Contract Expert short long .75 1.0 3.3

4.7

Nonexpert short long 1.0 .75 2.5

3.7

Judgement Expert short long .63 1.0 3.3

2.3

Nonexpert short long 1.0 1.0 4.7

4.5

Table 3. The expert's self-acknowledged criteria for classifying texts as experts' or nonexperts' texts Experts' texts terms distinctions details

Non-experts' texts imprecise and incorrect use of terms no distinctions focusing on wrong and irrelevant details and parts focusing on wrong and irrelevant relations wrong content

These remarks give us a first hint as to the tight connection between knowledge, terms, distinctions and relations. In this chapter we shall focus on distinctions and relations as expressed in and between text segments which will be categorised and characterised as qualia structures. 3.2.2. Qualia structure in production protocols The second analysis addresses the problem of correlating linguistic units and cognitive units in a more direct way. This was done via the segmentation of the texts into qualia, a method which makes it possible to make a qualitative analysis as well as a quantitative analysis of the differences between experts' and non-experts' knowledge as expressed in texts. The 32 texts were analysed by the legal expert and myself separately. To the legal expert I had explained the qualia model in general terms by demonstrating on (another) text string the segmentation into the three conceptual categories: Agentive = the before, the coming about, the cause and origin Form = the concept/situation itself as to its content, constituents and form Telic = the after, i.e., the function, the effects and consequences.

134

Lita Lundquist

The legal expert was first asked to divide the running text of the protocols, which were unmarked for author, into units which he found corresponded to units of legal knowledge involved in the two concepts 'contract' and 'judgement'. Secondly, he was to categorise these legal cognitive units as either Agentive, Form or Telic. He was not given any linguistic explanation or cues, his analysis of the text string of the protocols thus being performed top-down from his expert conceptualisation of the legal domain in question. Below is an example translated into English of a production protocol (expert on judgement, long) analysed in qualia by the expert rater and the non-expert rater, respectively. Expert rater s analysis of expert's text on judgement, long. A judgement F/ is a decision F/ made A / by a court of law A/ which determines how F/ a legal conflict A/ should be settled. A/ (The judgement)3 establishes A/ the legal position F/ of the conflicting parties. A/ A judgement F/ may be enforced. T/ For example, a judgement F/ pertaining to payment T/ of a sum of money A/ may be enforced T/ with the aid T/ of the enforcement court T/, and a judgement F/ concerning parental custody means that A/ the party A/ awarded T/ custody A/ may resort T/ to the bailiff T/ in order to gain physical custody A/ of the child A/ The expert rater arrives at 7 Form, 12 Agentive and 7 Telic qualia, a total of 26 qualia, whereas the distribution in the non-expert rater's analysis (see below) is 8 Form, 7 Agentive and 4 Telic qualia, a total of 19 qualia. Non-expert rater's analysis of expert's text on judgement, long. A judgement F/ is a decision F/ made A/ by a court of law which determines how A/ a legal conflict A/ should be settled F/. (The judgement) establishes F/ the legal position F/ of the conflicting parties. A/ A judgement F/ may be enforced T/. For example, a judgement pertaining to payment of a sum of money F/ may be enforced T/ with the aid of the enforcement court A/, and a judgement concerning parental custody F/ means T/ that the party awarded custody A/ may resort to the bailiff A/ in order to gain physical custody of the child. T/ The two analyses contain several illustrative examples of how the expert and non-expert raters differ most in the distinctions they make in more specialised descriptions, as in the following expression where the expert rater saw three qualia and the non-expert rater only one. Expert rater: a judgement F/ pertaining to payment T/ of a sum of money A/ Non-expert rater: a judgement pertaining to payment of a sum of money F/

From legal knowledge to legal discourse - and back again

135

Nevertheless, a fairly high overlap can be observed between the legal expert and the non-expert rater in their segmentation and their characterisation in terms of qualia. The difference consisted in the legal expert isolating a larger number of qualia overall (p < 0.001), as shown in Table 4 below. This difference is especially pronounced in the expert rater's perception of more Agentive (p = 0.002), but also more Telic qualia (p = 0.016), whereas the expert and non-expert raters arrived at the same number of Form qualia (p = 0.818). These findings support the hypotheses above that expert raters perceive more qualia than non-expert raters, and that experts perceive more Agentive and Telic qualia than non-experts, because they know more about the causes and consequences. Table 4. Frequency of agentive, form, and telic qualia made when the 32-text data base was coded by a legal expert and a text linguist. The last column indicates the significant probabilities testing hypotheses of equal Poisson intensities of numbers in the same row. Qualia Agentive Form Telic Sum

Legal expert 227 153 117 504

Text linguist 166 147 83 395

P 0.002 0.818 0.016 < 0.001

The difference between qualia produced in texts by experts and non-experts (students) appears from Tables 5 and 6. Here we see that in total, experts produce more qualia than non-experts, and a higher number of Agentive and Telic qualia than non-experts. Table 5. Number of qualia produced by students and experts in texts defining judgements (as perceived by expert rater). Judgements

Students

Experts

Agentive Form Telic

62 28 18

80 39 26

Table 6. Number of qualia produced by students and experts in texts defining contracts (as perceived by expert rater). Contracts

Students

Experts

Agentive Form Telic

29 30 24

56 54 49

136

Lita Lundquist

A statistical test was run to determine whether the data show any significant differences in the distribution of qualia between expert and non-expert producers, between expert and non-expert raters, between text types (short and long) and between concepts treated (judgement and contract). This is discussed below. 3.2.3. Total distribution of qualia The data have been submitted to an analysis of variance (ANOVA) applying the factors: type (long/short), producer (expert/non-expert), text (judgement/ contract) and rater (expert/non-expert). Besides all main effects, also all first and second order interactions were included in the analysis.4 The analysis was performed using the computer program GENSTAT. The statistical analysis was separated into two parts. First the total number of qualia produced by each producer was analysed. Thereafter, conditioning on the total number of qualia, i.e., the percentage of Agentive, Form and Telic qualia, respectively, was analysed. Nowhere did interaction terms show any significance. The results of the analyses can therefore be reported as main effects. The mean number of qualia is distributed over the parameters expert/nonexpert (student) producer, short/long, judgement/contract, as shown in Table 7. Table 7. Mean number of qualia in text type (long/short), expert/non-expert (student) producer, judgement/contract, and expert rater (NN) and non-expert rater (LL)

Judgement Contract Rater

Long Expert

Student

25.12 19.62 NN 15.47

16.50 16.81 LL 12.38

Short Expert

Student

8.38 14.0

5.56 5.38

There is a significant difference in the mean number of qualia in long and short texts (19.52 > 8.33, p = 0.018), which is of course not surprising since longer texts invite the producers to develop a topic and thus produce more linguistic material. There is also a significant difference in the mean number of qualia produced by experts and students (16.78 > 11.06, p = 0.020). This confirms our prediction that legal experts can draw on a mental model of the legal domain with more fine-grained distinctions than non-experts. The difference in the mean number of qualia perceived by expert and non-expert rater is also significant (15.47 > 12.38, p = 0.015). This result shows that legal experts, also when they

From legal knowledge to legal discourse - and back again

137

read expert texts about legal domain concepts, categorise in more qualia, i.e., perceive finer distinctions than non-experts. There is no significant difference in the mean number of qualia between the textualisation of the two concepts judgements and contracts (13.89-13.95, p = 0.378). These results support the second hypothesis, according to which experts, both when they produce and when they read texts about their specific knowledge domain, categorise in finer distinctions. Whether there is a systematic distribution between qualia produced by experts and non-experts respectively, will be commented on below. 3.2.4. Proportional distribution ofAgentive, Form, and Telic qualia The distribution in percentage ofAgentive, Form, and Telic qualia is shown in Table 8. Table 8. Distribution in % of the three qualia over text type (long/short), expert/nonexpert (student) producer, judgement/contract, and expert rater (NN) and non-expert rater (LL). Long Expert

Student

Short Expert

Student

Judgement

Agentive Form Telic

47.8 32.3 19.9 100

52.3 36.7 11.0 100

63.8 30.3 5.8 100

68.2 24.3 7.6 100

Contract

Agentive Form

28.9 42.6 28.6 100

33.2 36.5 30.3 100

42.9 33.0 24.1 100

47.4 37.4 15.2 100

Telic

Agentive Form Telic

NN

LL

49.7 32.0 18.3 100

46.4 36.3 17.3 100

The data will be commented upon quale by quale.

The Agentive quale There is a significantly higher percentage ofAgentive qualia in judgements than in contracts (0.580 > 0.381, p < 0.001). This fact can be interpreted in the fol-

138

Lita Lundquist

lowing way. When defining judgements, people, both experts and non-experts, focus on the factors that cause a judgement to come into play:5 Example of Agentive qualia (in bold) in judgement, short definition, expert rater: a decision of /a conflict/ between two or more parties/, /made/ by a /legal court/. In opposition to judgements, contracts contain significantly more Form and Telic qualia than judgements do (Form 0.374 > 0.309, p = 0.016; Telic 0.245 > 0.111, p 0.406, p 0.464, p = 0.214), in the proportion of Agentives they produce, or perceive. The Form quale As mentioned above, there is a significantly higher percentage of Form qualia in contracts than in judgements (0.374 > 0.309, p = 0.016). An explanation may be that contracts are easier to define with respect to their concrete form and to the parts they consist of, as compared with judgements, which have no such concrete physical manifestations. Example of Form qualia (in bold) in contract, short definition, expert producer, expert rater: (a contract) is a /written/ /agreement/ between two parties. There is a tendency for long texts to contain a higher percentage of Form qualia than short texts. This can be interpreted as the effects of the producers asked to write a long text specifying more details about the constituents of the concept. This goes especially for the expert producers of long definitions of contract. There is a tendency for the non-expert rater to perceive more Form qualia than the expert rater (0.363 > .0320, p = 0.056), which supports hypothesis three. There is no difference, however, between expert and non-expert producers in this respect, i.e., in their recourse to the Form quale. This runs counter to our prediction that non-expert producers would rely more on Form qualia than on Agentive and Telic qualia.

From legal knowledge to legal discourse — and back again

139

The Telic quale There is a significantly higher percentage of Telic qualia in contracts than in judgements (0.245 > 0.111, p < 0.001). Whereas for judgements we saw a preponderance of Agentive qualia, i.e., of the 'before' (namely what causes a judgement to come about), for contracts the 'after' plays a bigger role. This seems intuitively natural for an essential feature about a contract must be its purpose. Example of Telic qualia (in bold) in judgement, long definition, expert producer, expert rater: a promise whereby/ a contracting party/ undertakes/ to perform some service/, which may be conditional/ on the other contracting party/ performing a service in return/. There is a significantly higher proportion of Telic qualia in long texts than in short texts (0.224 > 0.132, p = 0.016), which indicates that when people are asked to write more about a concept, i.e., develop a concept, they take into consideration what the consequences are. There is a small difference, but not significant, between expert and non-expert producers' use of Telic qualia when writing about the two legal concepts 'judgement' and 'contract' (0.196 > 0.160), and between expert and non-expert rater's identification of Telic qualia in the written texts (0.183 > 0.173). 3.3. Conclusion about qualia statistics The findings above in 3.2.3 support our three predictions that in total experts, as compared to non-experts, conceptualise in more and finer categories, shown in the number and form of qualia, both when they produce texts treating their knowledge domain, and when they read such texts. As to the distribution of qualia within groups, the data in 3.2.4 did not, however, completely support our third hypothesis, predicting that experts, both when they produce and read texts about their knowledge domain, will perceive more Agentive and Telic qualia than non-experts. There was a small but not significant difference between expert and non-expert producers' use of Telic qualia, and between expert and non-expert rater's identification of Telic qualia, but no difference between the two groups' use and identification of Agentive qualia. This result will be commented on in 4.1, below. What these data did bring out, and quite unpredictably, was that there was a significant difference in the way the different types of qualia were distrib-

140

Lita Lundquist

uted over type (long/short), and over text (contract/judgement). In fact, the data showed that: (1) The Agentive quale has a higher representation in short texts compared with long texts. (2) The Agentive quale is relatively more frequent in texts treating judgements than in contracts, which, in turn, have a higher score of the Form and Telic qualia. In the vein of the dynamic models in Figures 1, 2 and 3 above, short texts and texts defining judgements thus show a preference for the 'before', the bringing about of the situation described, whereas long texts and texts defining contracts bring more Form and Telic qualia to the field, thus developing what the object or situation consists of and aims at. Before/Agentive

Concept/Form

Judgements + short texts

After/Telic

Contracts + long texts

Figure 6. Distribution of texts and types on the dynamic three-step model

4.

Discussion

Although there was no significant difference among experts and non-experts in the proportional number of different types of qualia they produced or perceived, it was shown above that in total, experts produced and perceived a significantly higher number of qualia in the production protocols. The differences were especially marked in experts seeing more Agentive and Telic qualia than non-experts (Tables 3 and 4), as predicted by hypothesis three above. The results also support two of the expert rater's self-acknowledged criteria for classifying texts as experts' or non-experts' texts (3.2.1 above), namely that experts make more distinctions and focus on more (relevant and correct) details than non-experts. 4.1. Knowledge, words and terms There are, however, other criteria which point to differences in knowledge, as also mentioned by the expert rater, namely that experts and non-experts use terms, but that the latter use these in an imprecise and incorrect way. I take this to parallel the observations made by Putnam and Kripke (see 1.1, above), that lay people only know a priori properties and stereotypes, i.e. conventional

From legal knowledge to legal discourse - and back again

141

Table 1.2. Examples of terms and words in expert and non-expert texts. '=' indicates some kind of parallelism between expert's and non-expert's knowledge, but expressed in terms by the expert and in words by the non-expert. '*' marks wrong content. Expert Texts terms

Student Texts words focusing on wrong and irrelevant details, relations and content

Contract Short

parties

partners

Contract Long

legally binding declarations of intent

specifies in detail the conditions under which the agreement has been entered into

promise

lays down rules to be applied in cases of disagreement between the parties

contracting party contractual agreement Judgement short

settlement by a court of law

=

the decision made by a court of law

of a conflict between 2 or more parties

=

in a case of legal action against a person who has acted in conflict with the law or who has committed a crime

Judgement long

legal conflict

a judgement is a decision made by a court of law

after his or her case has been conducted in court, a person who is charged with having done something illegal* ... =

can by a judge be given a sentence

which determines how = like a fine, a fortnight in detention, a legal conflict should be settled. imprisonment for an indefinite period, and so on how severe the judgement is depends on the degree of the crime* enforced enforcement court parental custody resort to the bailiff physical custody

142

Lita Lundquist

ideas, whereas experts know a posteriori properties and hidden, essential, necessary properties. Without entering into the ongoing discussion of the differences between terms and words (Lundquist, in preparation; Raccah 2000), I shall baptise this difference here as experts using and knowing terms as opposed to non-experts using and knowing words. This difference can be illustrated by some examples from the texts presented in Table 1, reproduced (as fragments only) in Table 1.2 above. Experts use terms which are specialised in that they relate to new mental models of the legal reasoning type, e.g., legally binding, promise, contracting party, legal conflict, enforcement (court). These terms open up to new frames, what is specific for specialised frames: "Within a frame, each attribute may be associated with its own frame of more specific attributes" (Barsalou 1992: 33). Non-experts use words and expressions which can be characterised as being often wrong and specifying details which are person-focused ("after his or her case... a person who is charged with having done something illegal...") and narrative. In fact, the frequency in Table 1.2 of action verbs such as enter into, lay down, act in a conflict, commit a crime, conduct a case, charge somebody with something, do something illegal, give a sentence, supports the claim that 'non-expert reasoning is more like story understanding" (Noordman et al. 2000, 237). This fact may explain why non-experts, contrary to the hypothesis, use proportionally as many Agentive and Telic qualia as experts do. The verbs mentioned above activate an event structure with inherent agents, i.e., Agentive, and potential Telic qualia (Lundquist 2000). Also Telic qualia are realised by the non-experts, in that they emphasise consequences, often 'dramatic' consequences for judgements, such as a fortnight in detention, imprisonment for an indefinite period; it is less so for contracts: disagreement between the parties. Because Agentive and Telic qualia are perceived both in terms and words, and in verbs and nouns, especially of course in deverbal nouns but also in nouns in general, the hypothesis concerning the proportional distribution of these two qualia over expert and non-expert texts and raters has to be supplemented and counterbalanced by an analysis of vocabulary; i.e. terms, words, and word classes that embody the respective qualia.

5.

Perspectives and conclusion

The proposed method of applying qualia to relate linguistic and cognitive structures - in both directions - has proved practicable and useful in the comparison of differences in knowledge between experts and non-experts, and specially

From legal knowledge to legal discourse - and back again

143

apt to seize cognitive categories which are also typical of legal reasoning. The method, however, was not sufficient. As pointed out above in the discussion, the linguistic-cognitive categories of qualia have to be refined by a more systematic analysis of the linguistic material which constitute the qualia. This will be the topic of a forthcoming paper, but even such an analysis does not solve the fundamental problem raised in the introduction. Indeed, it rather exemplifies it. The problem is that of disentangling language and thought. It seems impossible not to confound 'the stars and the telescope' as Fauconnier (1997: 3) puts it in the following comparison: In studying supernovas or neutrinos, the phenomena, the theories and our reflections on them are kept apart with relative ease. For language and thought this is not the case: We produce our account of the phenomena under study by using language and thought, that is, by relying on the very phenomenon we are studying.

Studying language and knowledge with language and knowledge, and discourse and cognition with discourse and cognition, we seem to be caught in an infinite circle, with "turtles all the way down".

144

Lita Lundquist

Appendix Table 1.3. Original texts in Danish Expert Texts

Student Texts

Contract Short

Skriftlig aftale mellem 2 eller flere parter.

En skriftlig aftale mellem to partnere.

Contract Long

Bestär i almindelighed af tilkendegivelser fra 2 eller flere parter i et retsforhold. Tilkendegivelsen er oftest et 10fte hvorved parten forpligter sig til en ydelse, evt. betinget af en modydelse fra medkontrahenten. Et kontraktsforhold kan ogsa bestä uden, at der foreligger udtrykkelige tilkendegivelser, idet retsforholdet stifles som f01ge af parternes indbyrdes adfsrd.

En skriftlig aftale mellem to parter. Udspecificerer noje de betingelser aftalen er indgäet under. Foreskriver regier i tilfaelde af uenighed mellem de to parter. En sikkerhed for, at der ikke opstär nogen form for misforstäelser. Ved misligeholdelse af kontrakten afg0res sagen ved en domstol.

Judgement Short

Er en afgerelse af en konflikt mellem 2 eller flere parter, truffet af en domstol.

Den afgerelse som domstolen giver i en retssag til en person, som har handlet ulovligt eller har begäet en forbrydelse.

Judgement Long

En dorn er en afg0relse truffet af en domstol, som bestemmer hvorledes en retslig konflikt skal afg0res. De stridende parters retsposition fastlaegges herved. En dom kan tvangsfuldbyrdes, säledes at f.eks. en dom som vedrarer betaling af et be!0b kan tvangsinddrives ved bistand fra fogedretten, og en dom om forajldremyndighed indebasrer at den part som tillsgges foraeldremyndigheden kan fa barnet udleveret ved fogeds bistand.

En person, der er sigtet for at have gjort noget ulovligt, kan i retten efter at hans sag har vaeret f0rt af en dommer id0mmes en straf, der kan vsere for eksempel en b0de, 14 dages haefte, fasngsling pä ubestemt tid og sä videre. Dommen bestär i rettens endelige afg0relse af, hvad der videre skal ske med personen. Hvor streng dommen bliver afhasnger af graden af forbrydelsen.

From legal knowledge to legal discourse - and back again

145

Notes 1. Selected at random among the total set of non-expert protocols, the texts turned out to stem from 14 different subjects, two of whom were represented by two texts each. 2. The chance likelihood of judging 12 of 14 expert-produced texts as coming from experts is less than 1 in 100; similarly for judging 14 of 16 non-expert texts as coming from non-experts. 3. This noun is added by the translator. The Danish version contained a passive construction + herved('hereby'), which received no qualia. 4. As some of the producers produced two texts (short/long judgement/contract), a factor representing each of the producers is also included. However, as the producers are to be considered as two random samples of producers (expert/non-expert), the factor representing the producers is made random. Thus, the applied model is a mixed ANOVA with four systematic factors and one random factor. 5. Intuitively, it is of course also interesting to specify for a judgement what its consequences are, but maybe so more in the description of a concrete, actual judgement.

Acknowledgement The present work was funded partly by the Copenhagen Business School (the project "Mental Model in LSP texts", see Lundquist and Jarvella (eds.) 2000), partly by the Danish Research Council for the Humanities (the project "Production and reception of texts for specialised purposes"). I thank my colleagues Jesper Lau, Gorm Gabrielsen, Anne-Marie Biilow-M011er and Martin Aitken for having helped with the legal, statistical, and linguistic analysis.

References Ackrill, J.L. (ed.) 1987 A New Aristotle Reader. Princeton, Princeton University Press. Aristotle 1970 The Physics. With an English translation by P. H. Wicksteed and P.M. Comford (eds.). London, William Heinemann. 1975 The Metaphysics. With an English translation by H. Tredennic. London, William Heinemann. Barsalou, Laurence W. 1992 Frames, concepts, and conceptual fields. In: A. Lehrer and E. Kittay (eds.), Frames, Fields, and Contrasts, 21-74. Hillsdale, NJ: Lawrence Erlbaum.

146

Lita Lundquist

Fauconnier, Gilles 1997 Mappings in Thought and Language. Cambridge: Cambridge University Press. Fillmore, Charles J. 1985 Frames and the semantics of understanding. Quaderni di semantica 6/2: 222-254. Johnson-Laird, Philip N. 1983 Menial models: Towards a Cognitive Science of Language, Inference, and Consciousness. Cambridge: Cambridge University Press. Kjasr, Anne Lise 2000 On the structure of legal knowledge: The importance of knowing legal rules for understanding legal texts. In: Lita Lundquist and Robert J. Jarvella (eds.), Language, Text, and Knowledge: Mental Models of Expert Communication, 127-162. Berlin / New York: Mouton de Gruyter. Kripke, S. 1972 Naming and necessity. In: G. Harman and D. Davidson (eds.), Semantics of Natural Language, 253-355. Dordrecht: D. Reidel. Lakoff, George 1990 Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Lundquist, Lita 1989 Coherence in scientific texts 2. In: Wolfgang Heydrich, Fritz Neubauer, Jänos S. Petöfi and Emel Sözer (eds.), Connexity and Coherence: An Analysis of Text and Discourse, 122-149. Berlin / New York: Mouton de Gruyter. 1998 Cadres et espaces mentaux dans la structuration cognitive des connaissances - et des discours - juridiques. In: Yves Gambier (ed.), Les discours professionnels enfrangais, 123-150. Berlin: Peter Lang. 2000 Knowledge, events and anaphors in texts for specific purposes. In: Lita Lundquist and Robert J. Jarvella (eds.), Language, Text, and Knowledge: Mental Models of Expert Communication, 97-126. Berlin / New York: Mouton de Gruyter. Lundquist, Lita and Robert J. Jarvella 2000 Introduction. In: Lita Lundquist and Robert J. Jarvella (eds.), Language, Text, and Knowledge: Mental Models of Expert Communication, 1-9. Berlin / New York: Mouton de Gruyter. Lundquist, Lita and Robert J. Jarvella (eds.) 2000 Language, Text, and Knowledge: Mental Models of Expert Communication. Berlin / New York: Mouton de Gruyter. Noordman, Leo, Wietske Vonk and Wim H.G. Simons 2000 Knowledge representation in the domain of economics. In: Lita Lundquist and Robert J. Jarvella (eds.), Language, Text, and Knowledge: Mental

From legal knowledge to legal discourse — and back again

147

Models of Expert Communication, 261-284. Berlin / New York: Mouton de Gruyter. Palsbro, Lene 2000 Argumentation and knowledge: An empirical study on inference-making in expert and novice reasoning. In: Lita Lundquist and Robert J. Jarvella (eds.), Language, Text, and Knowledge: Mental Models of Expert Communication, 207-234. Berlin / New York: Mouton de Gruyter. Pustejovsky, James 1995 The Generative Lexicon. Cambridge, Mass.: MIT Press. Putnam, Hilary 1975 The meaning of 'meaning'. In: Hilary Putnam (ed.), Mind, Language and Reality: Philosophical Papers. Volume 1, 215-21Ί. Cambridge: Cambridge University Press. Raccah, Pierre-Yves 2000 Lexical and dynamic topoi in semantic description: A theoretical and practical differentiation between words and terms. In: Lita Lundquist and Robert J. Jarvella (eds.), Language, Text, and Knowledge: Mental Models of Expert Communication, 11-30. Berlin / New York: Mouton de Gruyter. Searle, John R. 1995 The Construction of Social Reality. New York: Free Press. Searle, John R. 1998 Mind, Language, and Society. New York: Basic Books.

Chapter 8 Conditionals: Your space or mine? Anne Marie Bülow-Moller

Conditionals have received a good deal of attention in cognitive grammar as a classic case of alternative mental spaces (e.g. Dancygier 1998; Fauconnier 1996; Sweetser 1996a, 1996b; Werth 1997). In essence, conditionals represent a grammaticalised cognitive stance: a categorisation of a state of affairs as possible but not actual. The cognition involved in processing a conditional construction is, then, to map assertions onto sets of spaces, or models, or possible worlds, in order to establish a dependency relation between them. But since 1986 scholars have also noted the need for work based on discourse, i.e. text in actual use (Akatsuka 1986; Ford 1993, 1997; Ford & Thompson 1986; Ramsay 1987), since it can be shown that in context, conditionals serve particular strategic aims: they are used to set up several types of models that the speaker wishes to present as provisionally shared. In this chapter I shall attempt to show that the analysis of discourse functions bears directly on the mapping process: it is not just a matter of mapping two possible worlds together to form a shared space - the status of 'my' space, 'your' space, and Our' shared space is delicately calibrated by speakers to serve their interactional goals. While the textual function of conditionals is certainly to provide coherence and relevance, as is pointed out in the text-based work mentioned above, the interactional goal relates to the speaker's stake as well as to his or her face. This is most clearly seen in argumentative discourse, and the examples that I shall discuss in section 2 are consequently drawn from a corpus of negotiation talk. In setting up agreements speakers clearly have a self-interest or stake, and a great deal of facework is evident. There is also a very large number of conditional clauses, precisely because the inspection of shared models for agreements is the business of negotiation talk. The analysis of conditionals in talk shows a kind of scaling and blending that leads us to query the categories that the cognitive linguists often use for conditionals. In actual use, the mapping and the spaces can only be understood in context, not from the shape of the sentence; thus slight variations in pitch and loudness can make one speaker's backgrounded, epistemic conditional into another speaker's foregrounded prediction, or at least move it to a point be-

150

A nne Marie Bulow-M0ller

tween the clear prototypes. In section 3.1 below, some criss-crossed categories will be discussed on the basis of Sweetser's influential division into 'content', 'epistemic' and 'speech act' conditionals.

1.

The status and meaning of conditionals

Speakers use conditionals for a great variety of discourse purposes. Consider a typical conditional construction, where the if p then q format has a causal contingency relation (when/? is realized it causes q). The example is lifted from Taylor (1997): (1) A: Does your cat scratch? B: If someone teases her. Taylor's point is that conditionals are centrally concerned with polarity, i.e. the question whether the if-clause rums out to be positive or negative. If positive (you tease her) the consequent takes effect (she scratches). The interesting thing in a discourse perspective is that text (1) can be taken to mean either yes or no, depending on the context: (1') Ann, restraining small child eager to grab cat: Does your cat scratch? Bob: (Yes) if someone teases her! (1") Ann, nervous about her extremely expensive new tights: Does your cat scratch? Bob: (No. only) if someone teases her. The difference can be seen to reside in the amount of relevance attached to the condition: if the interlocutors can see that the possibility of teasing is relevant, the mental space of 'scratching cat' is foregrounded, and the utterance counts as affirmative. Since this effect is undesirable, also for the speaker, who has a stake here, the conditional counts as a warning. If the possibility is less likely, the conditional acquires the status of the small print on contracts that applies only in extraordinary circumstances: backgrounded but still there if such a circumstance arises. In other words, we are dealing not so much with yes/no polarity as with a contingency relation: the more likely, the more foregrounded. The contingency relation has several interactional consequences. One is that if Ann tries to shift the amorous cat by nudging it away, and the cat then scratches, Ann is on record as having teased the cat. This is the result of a combination of implicatures surrounding the use of conditionals: (a) 'conditional perfection', whereby if is heard as if and only if, (b) the status as 'provisionally shared' that attaches to conditionals, and

Conditionals: Your space or mine?

151

(c) the two-way epistemological reasoning that follows: if//means iff ana is accepted as shared, and the consequent proves true, the antecedent must be true as well. So Ann must have teased the cat. The set of inferences involved here makes the conditional construction particularly interesting for the kind of discourse that involves reasoning. In the example above, speaker B is entitled to his conclusion through his assumption that the provisionally shared if-clause turned out to be positive (someone teased the cat, so she scratched). Ordinary conditional perfection ensures that utterances like If you mow the lawn, I'll give you ten dollars mean if and only if, not otherwise (for discussion, see Auwera 1997; Horn 2000; Levinson 2000). Speaker A, however, has the uphill battle of querying something she must regard as only provisionally shared (the teasing is evidently a sufficient but not necessary condition). How far A is linguistically boxed in by the conditional is in fact the crux of the problem in the cognitive handling of conditionals. 1.1. Presupposition, givenness, and commitment Some of the puzzlement over the versatile conditional has concerned the amount of commitment that is required for the speaker to use it meaningfully. Firstly, the conditionals typically deal with something that has already been brought up in the text or interaction ("Conditionals are topics", Haiman 1978; Schiffrin 1992); they are normally preposed and hence, in terms of information structure, in 'theme' position, where they can be treated as given. A speaker that treats an if-clause as given is clearly committed to the content in some form or other. Secondly and somewhat paradoxically, conditionals do not commit the speaker to the truth of the propositions (Comrie 1986; Taylor 1997; Werth 1997); they are used to create cognitive distance, because they denote speculation (Comrie 1986) and implicitly acknowledge the alternative point of view (Dancygier & Sweetser 1996; Ford 1997). Finally, conditional clauses may adopt a set of modal values and thus allow for different levels of alignment with the propositions (Fillmore 1990) from open possibility to counterfactuals (if you do, if you did, if you had done). In fact, in the corpus to be discussed in section 2, the conditionals are often used alongside modalized expressions (we may do...if we do). This apparent contradiction will bear a closer analysis; it appears that what speakers do with conditionals in discourse is to claim relevance for their point along some sort of scale of commitment. In the following we shall look at two aspects of such commitment: first at the levels of cognitive commitment inher-

152

Anne Marie Bülow-M0ller

ent in presupposed and assumed material (1.1.1 - 1.1.2), and then at the levels of cognitive dissociation that Fillmore calls 'epistemic stance' (1.1.3). 7.7.7. Presupposed and assumed if-clauses If-clauses cannot be logically presupposed in the same way as temporal or causal clauses can. Cognitive commitment is strongest in the case of entailment and presupposition, where one cannot meaningfully deny the subclause. A speaker who says Because you teased her, she scratched is logically committed to the truth of both propositions (cf. the logical impossibility of *Since/ when you teased her, she scratched, and I don't know if you did tease her). That kind of commitment is lacking with conditionals: it will always be possible to say If you teased her, she scratched, but I don't know if you did. Similarly, the kind of embedding that otherwise produces presupposition transfer does not work: I am sorry that you teased her presupposes 'you teased her', but lam sorry if you teased her does not. Conditionals cannot shed their element of speculation. A little less cognitive commitment is required for the logical formula if p then q, as in If all of the arrows hit the target, some of them did. There is a substantial literature on truth-conditions in this sort of conditionals; for recent treatments that involve cognition, see Johnson-Laird & Byrne (2002) or Lycan (2001). Here it is acknowledged that purely logical relations must be supplemented with an understanding of the reasoning processes that they involve. Hence the arrow example only makes sense if the antecedent is taken as implicature (Taylor 1997) or as conversationally presupposed, so that it can be glossed as 'let's operate in a space where it is true that all the arrows hit the target, ok?' The givenness is, then, negotiated: alone among conjunctions, conditionals crucially retain an interactional function (i.e. the necessity that the hearer accepts the instruction to share a model). This type of assumed sharedness is often found in the epistemic use of the conditional, where a piece of evidence is to hand that allows the speaker to draw a conclusion (Sweetser 1990). This is the type If their lights are out, they are not at home. The argument is successful only if both speaker and hearer assume the truth of the antecedent. In discourse the type is strategically used to demonstrate the basis of one's reasoning: hearers who have access to the evidence (and do not query it) are bound by the reasoning to accept the argument. However, if we assume that the speaker claims relevance, it requires some sort of explanation that it is perfectly possible to use a conditional in situations where there is no doubt whatsoever:

(2) A: Ow! B: Oh dear. If she scratched you, she must have thought you attacked her.

Conditionals: Your space or mine?

153

Somehow it is not good enough to say that the antecedent represents something that is assumed to be shared, for it is after all just assumed. In terms of Grice's maxims, this is a flagrant breach. Presenting something known as merely supposed is a misrepresentation, in the same way that in normal conditions it would be distinctly weird for a speaker to say (cooperatively), If this is my wife, I'd like you to meet her. However, the epistemic downscaling is widely used, cf. also types like Oh, did she scratch you? She must have thought (etc). Dancygier & Sweetser call this sort of distancing 'evoking' a space rather than setting it up (2000: 116). I shall argue below that in discourse this ploy serves to distance the speaker a little from an unpleasant reality for reasons to do with face - in this case, not politeness or mitigation to deal with something awkward for hearers, but for speakers themselves because of their stake in the situation. 7.7.2. Marked theme The argument from information structure follows Halliday's (1994) use of 'theme' as the point of departure, that which the speaker selects for 'grounding' the utterance; typically this is anything that is placed before the main clause verb (in statements). The principle applies centrally to the clause, but also, if perhaps to a lesser extent, to the sentence. The overwhelming tendency in English is to fill the theme slot with material that is not-new to the hearer, such as a light pronominal subject, whereas heavy semantic material that requires processing tends to be found at the end. Adverbs or subclauses before the subject are known as 'marked theme'. Subclauses have their own focus structure and distribution of given-new material, but unless they are under contrast focus or other emphasis, they will tend to be backgrounded as the starting point for the newsworthy information in the main clause. This means that the speaker treats anything in the theme slot as not-new (Fowler 1985; Edwards 1997, 1999) or as 'naturalised' (Fairclough 1995), and that the material can be taken as recoverable from cultural knowledge, shared background or situational context. Whether or not this actually corresponds to the speaker's and hearer's shared knowledge is immaterial, because in the interaction it is assumed that the inferential structures are available to suggest the mental spaces to be shared, for instance, in the shape of situational frames or cultural scripts. For example, in talking to a man of marriageable age, it is quite possible to understand the utterance My wife adores cats even for a hearer who did not know that the speaker was married: adjustment is made retrospectively. The main treatment of this so-called 'pragmatic presupposition' concerns definite nouns (for overview, see Abbott 2000); but in the same way, prenominal

154

Anne Marie Bülow-M0ller

qualifications and the items in theme position are naturalised because they cannot be challenged in the way the rheme can: (3) A: That stupid cat scratched me! B: *No she isn 't! (but: No she didn't!) A: When the cat scratched me, I yelled B: *No she didn't! (but: No you didn't!) The most important function of the theme position, whether recoverable from context or script, is that the speaker treats the material as uncontroversial, cf. also Grice( 1981). When material is treated as given in the sense of recoverable, it can serve a bridge function between the not-new and the genuinely new, or rather: newsworthy. For example, if a tourist guide starts a new paragraph with In the same street,,., or If you turn to your left..., this is obviously both a continuation and a reorientation (Virtanen 1992). Givon calls such preposed adverbials 'coherence bridges' (1995: 372), and points out that the strategy serves as a theme-switcher to supplement referential continuity. Textually, this is an important function for conditional subclauses: to provide a piece of shared ground as a reminder or continuation marker, and to serve as a pivot for a new development (Ramsay 1987; Ford 1993). The pivot is the claim of relevance; in the sequence If you are so clever, why aren't you rich?, the cleverness is specifically picked out (a) as referring back - not necessarily to the word clever but to previous discourse that can be summarized as 'you (think you) are clever', and (b) as the relevance point for the question, which is rather different from the unaccompanied question Why aren 'tyou rich? When an if-clause takes up contextual material, it may be given only in that sense, and not in the sense of speaker's commitment. In the example above the speaker's stake suggests that he/she is in fact holding the commitment at arm's length. With examples like, (4) A: Does your cat scratch? B: If she scratches it is because someone teases her, the recoverable 'scratching' is a quotation and in discourse sometimes presented with audible scare quotes. The antecedent is accepted only as a possibility, but it serves as a relevance point for the newsworthy 'teasing'. However, it is also possible to introduce unmentioned, and arguably unrecoverable material:

Conditionals: Your space or mine?

155

(5) A: Does your cat scratch? B: If someone TEASES her she does There are two ways of regarding this phenomenon: one is to treat the preposed if-clause as a left-dislocation, on a par with Wilson his name is, where the recoverable part is so uninformative as to be left out (as in Taylor's original example, text (1)), and only added as an afterthought under tail intonation. The other solution, more generally applicable, since speakers do in fact use a great variety of intonation patterns in the consequent clauses depending on the information weight they assign to them, is to do as Halliday says and treat anything in theme position as not-new. It is equally arguable that the if-clause always represents material that is (treated as) recoverable, not necessarily from the context, but rather from mutual or cultural knowledge of shared models. To say If you tease her she does is, then, to draw on a frame in which it is common knowledge that cats scratch if they are teased. This function, sometimes referred to as generic or generalizing conditionals, is just as effective in reasoning as epistemics proper: in discourse, surface forms corresponding to Cats scratch if you tease them, She scratches if you tease her, She (would have) scratched if you teased her, and If she scratched, you teased her, can all show up as explanations, or accounts, in the same situation. If we accept the view that in talking about the given, we refer to what the speaker treats as sufficiently recoverable from the co-text, situation or script, then that takes care of the objections raised by Matthiessen and Thompson (1988: 313), who claim that the subclause cannot be said to be given, known or inferrable. Their example concerns a woman with a hereditary condition who is consulting a specialist about possible hand surgery in her fear to lose the use of the thumbs /// don't do something now. True, the material is not mentioned and not in the context, but it is certainly part of a standard script about seeking expert help in case of medical problems. It was shown above that speakers may unilaterally downgrade shared knowledge (where the speaker mops up the blood while saying If she scratched you...). The not-new theme position affords the speaker a complementary opportunity to unilaterally upgrade an item and treat it as part of a shared space: (6) Thank you for looking after Pussy, it is very kind of you. There are three tins ofcatfoodon the table, and if she scratches, there is sticking plaster in the drawer. As in the former case, upgrading serves a particular strategic function. This is not a warning (cf. She may scratch) - it is a blithely assumed item in the looking-after-Pussy script. The speaker has an obvious interest in downplaying any

156

Anne Marie Bülow-M0ller

unpleasantness in connection with the service rendered; the imputed normality here safeguards his or her social face. Finally, if we assume that conditionals function as starting points because they are recoverable, some explanation is warranted for the large number of elliptical conditionals that are found in discourse: (7) If you could just put a finger there while I cut the plaster - there, that's better. It has been noted that these ellipses typically function as requests, and that the conditional therefore mitigates in the same way as a modal verb (van Dijk 1979), so that the polite list of hypothetical distance would run, can you?, could you?, if you can?, if you could? But we still need to explain why the (recoverable) antecedent is found while the (newsworthy) consequent clause is suppressed. I shall argue that again this is a function of the speaker's stake and face: overwhelmingly these clauses occur where the consequent would run ...that would be nice/helpful/desirable for me. The suppression simply removes the beneficiary from view. 1.1.3. Epistemic stance Treating something as a shared space is different from accepting the truth (if past) or likelihood (if future) of the semantic content. This kind of distancing work can be seen in the grammaticalised scale of epistemic stance discussed by Fillmore (1990), where the range of modal verbs signals the speaker's alignment with the possible world of the if-clause: (8) If you If you If you If you

tease her, she will scratch (and you may or may not choose to tease) teased her, she would scratch (but you probably wouldn'() were to tease her, she would scratch (but I'm sure you wouldn't) had teased her, she would have scratched (but you didn 't).

At the top of the scale, the possibility is open (the speaker is uncommitted); as we move down, the speaker takes up a stance, in this case that it is unlikely that you will tease the cat, and finally that it is untrue that you have teased her. Counterfactuals have a literature of their own. While the cognitive treatments abound (the process of mapping two impossible possible worlds on each other) there has been less interest in the mapping process seen as interaction: What is the information value of asserting that something that did not happen would have had consequences that it did not have? Since the mapping of shared spaces is instantly dismissed as not valid, the Gricean maxim of quantity begins to loom: that of giving too much information (after all, a very great deal of things did not happen).

Conditionals: Your space or mine?

157

Once again, in discourse the counterfactual conditional serves a pragmatic purpose. The more the speaker foregrounds the provisionally shared space, the more the speech act counts as past prediction (and possibly still valid, so that the utterance If you had teased her, she -would have scratched you can count as a warning that any future teasing will have the same effect). But the more the speaker foregrounds the provisionally shared (cancellable) space, we must look to other kinds of dismissals for an explanation: (9) A: Oh no, look at that! B: What are you so upset about? If she had scratched you I could understand, but this is after all just a bit of cat sick. Here the speaker is engaged in argumentation (on the basis that 'cat sick is minor damage'), and uses the counterfactual and more damaging scenario to demonstrate a reasonable attitude, thus protecting his social face (cf. also Garcia 1997). It would appear, then, that in discourse the degree of sharedness and commitment is intimately bound up with the mapping of shared spaces. To stay with the metaphor of foreground and background, it is as if the speaker can choose to present a conditional as my space mapped onto yours or your space mapped onto mine. The choice will reflect a pragmatic difference in the interaction.

2.

Conditionals at work: Examples from business talk

In the following I shall demonstrate some effects of this discussion when applied to discourse data. My material is drawn from many hours of recordings of people trying to get some sort of agreement, either face to face or on the telephone. The participants have different occupations (an insurance claims manager, several claimants (here bankrupted factory owners), business managers, an architect, a sales representative, MBA students) but in the discourse examples they fill the roles of 'sellers' and more or less reluctant 'buyers' either of a commodity or of an idea (and often both). In all negotiation activity, credibility and clout are major concerns for the participants, who regularly shape their contributions as power moves to create an impression of being in charge. At the same time, a good atmosphere is helpful, and speakers regularly use hedges and similar Other'-based traits, collectively known as 'facework'. The notion of 'face', as borrowed from Goffman (1967) and developed for the purposes of politeness theory by Brown and Levinson (1987), and later by e.g. Ting-Toomey (1988) and Scollon and Scollon (1995), has overwhelm-

158

Anne Marie Bülow-M0ller

ingly been used to describe the speaker's options vis-ä-vis the hearer s face. In this sense, it is 'positive politeness' to make the hearer feel appreciated, and 'negative politeness' to make the hearer feel respected, through giving options and refraining from imposing. In the present study, however, I shall return to Goffman's original notion efface as something a speaker strives for on his or her own behalf. Speakers hope to be accorded their share of social recognition, i.e. to maintain their face, through displays of social attractiveness to others (being nice, attentive to needs, etc.) and through displays of competence (being correctly informed and logical, showing initiative and efficiency, etc.) Attempts to build social face and competence face may be in conflict: hard-hitting reasoning may be heard as competent, but not very nice by the hearer, while attentiveness to the hearer's needs may be nice but not very clever. It appears from the corpus that the majority of the conditionals represent bids to boost the speaker's face; and that strategically this happens on the sort of commitment scale that has been outlined and to which we shall return at the end, running from if-clauses that foreground the speaker's idea as shared ('my space') to types where the speaker treats the hearer's space as provisonal ('your space'). Again I would like to stress that the weight given to the conditional, and hence the commitment, is relative and sometimes dependent on a mere inflection in the voice; we are talking about nuances rather than categories, but they are noticeable also in the reaction they draw from the hearer.

2.1. My space Speakers who are talking about their own concerns can assume a space as a matter of simple implicature: (10) (Rockley Glass: the seller is outlining plans for future investment in their plant) ...an investment requisition that, if I give the go-ahead, will improve quality yet again. But for interaction, acceptance is a strategic goal. Conditionals are a staple in descriptions of negotiations; thus Kennedy (1998) affirms that the most useful two-letter word in the language of negotiators is i-f, as in Who pays if the lorry breaks down? or If you pay cash, I might find you a rebate. The advice is to foresee all possible options in a script, and to offer no concessions without getting something in return. These conditionals demonstrate initiative; they are power moves that boost the speaker's competence face in mapping speaker's

Conditionals: Your space or mine?

159

space onto hearer's by assuming that his own suggestion can serve as common ground. At the speaker's end of the spectrum we find three main types of discourse function: the contingent quid-pro-quo types, including offers and explorations; the rational reasoning types, including coherence-building and interpretation; and the directive types, including the grammaticalised meta-discourse functions. 2.7.7. Contingent types (11) (Rockley Glass: the buyer is worried that the seller's capacity has decreased and is reluctant to place an order for several years. The seller tries a power move, a straightforward condition:) Seller: Pretty bluntly, if you 're prepared to commit to us and the stronger that commitment is. the stronger we can commit towards yourselves, yes. This extract is positively flagged as blunt, and buttressed by the speaker's habit of adding a yes after a suggestion. The assent that it projects does not mean 'Will you commit to us - yes?' but rather 'Have you understood the relation I am suggesting - yes?'. In this sense even the most hardnosed quid-pro-quo type is central in the process of reasoning: the shared space is presented as desirable for the hearer, and the element of shared stake adds to the shared space. When the space is contested, the interlocutor may query either the acceptability of the space or its likelihood or relevance, as in the following two examples: (12) (Dental Clinic Equipment: the seller is pushing for a packet, while the buyer wants only separate parts. Seller outlines an offer:) Seller:... 50 if we can go for something like I said ahm . we could certainly help with the MARKeting of the units Buyer [shaking her head]: but. it would double our cost (In this simplified transcription, only extra stress and micropauses have been marked, with upper case and with single spaced dots, respectively.) The seller uses a single breath group, clearly backgrounding the if-clause as uncontroversial; but the buyer treats it as provisional and refuses to accept the space as shared. (13) (Bridge 1: the claimant wants the local council to admit liability for a bridge which has fallen into disrepair and which he has been banned from using for his factory's heavy traffic. The council's insurance company's claims manager refuses, and the case is now moving to court. The claimant explores options;)

160

Anne Marie B low-M0ller

Claimant: Well who is going to put it right if it falls down then? Manager: If anyone had to put it right it would be British Waterways Board, but I don't know what their position is in law, it may be that if it fell down they wouldn 't have to replace it, I just don't know The claimant, the 'seller' of the idea, offers his model as part of a shared script: it makes sense to ask about putting it right only inside the model where the bridge falls down. The claims manager queries the presupposition that someone must put it right, and treats this space as distant and hence irrelevant ('if it fell down'). All four speakers defend their competence face in a fashion that can be expected from their stake: before the agreement phase of a negotiation the partners use the interaction to dissociate themselves from each other's spaces. 2.1.2. Reason ing and accounts Because of the danger of dissociation, the speaker has an interest in creating coherence from his own space. The bridging function, which links back textually, is strategically important in keeping the speaker's model shared and on the table. It can be done both from the immediate context and from the preceding discourse: (14) (Rockley Glass: buyer is being invited to list the "items we need to talk about today":) Seller: .. .that s why I think we want ίο get all your concerns out on the table, to understand them. If we do that, we can address them This kind of explanatory note or account is typically found in connection with an interactionally sensitive move, like a refusal or some sort of imposition (Antaki 1994). It serves as mitigation for a direction ("we want your concerns out on the table") and simultaneously as reasoning: the direction makes sense. Cohesion created from longer stretches can be used to gather in results: (15) (Rockley Glass: attempt to close the opening round) Seller: Let me summarize all that. If we can give you assurances on capacity ahm - capacity to produce through the future, if we can get agreements on the technical back-up. (...) then can I take it that we can form some agreement this afternoon? Buyer: Definitely (...) IF all that can be resolved at this meeting, then sure, we can (...) make the agreement for the whole year. It is a position of considerable power to be the party that does the summing up of what may be a very protracted round. The seller is heard as offering his

Conditionals: Your space or mine?

161

interpretation as a shared space for both parties, but he is also heard as doing it on the textual basis of the shared context, thus sharing the assumptions behind his conclusions. The argumentative power of the conditional is substantial and often works like conditional perfection in reverse: the discourse value is not so much 'If and only if I can give you assurances...then we can form an agreement'; the interactional value is rather 'If we can give you those assurances [which we now assume that we can] ...then [I conclude] that we have fulfilled our side of the bargain and you will have no excuse for not forming an agreement'. This move shifts the conditional usage to a point very close to the epistemic type, cf. the discussion of reasoning in Dancygier & Sweetser 2000. In salesmen's manuals the ploy is referred to as the 'tackle objections and lock' move. But here, too, the battle of commitment can be seen: the buyer looks after his social face by aligning himself with the seller, i.e. agreeing to share the space, and then, a little later, he introduces a slight contrast stress on if, which foregrounds the provisionality once again, thereby implicitly referencing the seller's stake. This buyer is aware of the interest reflected in the summary. The same buyer combines boosts for social and competence face in his use of qualifying afterthoughts: (16) (Rockley Glass: the seller, Rockley, is a long-term supplier, but they had a problem last year when they delivered chipped glass on one occasion) Buyer: No, no, the quality has been FINE, if you take last year as an exception The speaker uses the tail intonation typical of uncontroversial backgrounding; thus he shows willingness to assume that it was an exception, and at the same time takes the concession back by making it speculative. 2.1.3. Directive conditionals and politeness In talks between professional people, imperatives are used exclusively for register-shifted man-to-man requests that do not really hurt, like Come on, give us something to shoot at. The real power moves are scantily hedged commands like the following: (17) (Rockwell Glass: the buyer side is represented by two people, of whom the finance manager plays the role of the tough cop, with very few but hard-hitting contributions;) Finance Manager: / think. if you 'd like to make us a better offer. a rather better offer on the price. ahm that might help us along

162

Anne Marie Bülow-M0ller

Secondly, directives are used as meta-discourse to direct the talk. Controlling the agenda is almost as important as controlling the summary. A large proportion of conditionals in the corpus are used to redirect the topic: (18) (Rockley Glass: chief buyer speaking) Buyer: No, it was just that we are obviously if you like, if we can come back to the pricing issue, because now is most probably the appropriate time, that we see for example another company, a major supplier, very good reputation [cont.] (19) (Dental Clinic Equipment) Seller: Perhaps [clears throat] it may be an idea to proceed into negotiation now, if we can proceed into that. erhm to coordinate a MARketing scheme [cont.] These two speakers do exactly the same thing: they explicitly introduce a new topic and ask permission to do so in the middle of their utterance, without slowing down or otherwise acknowledging the hearer's acceptance. In shape, the conditional is a hedge, in both cases flanked by other hedges (an account and epistemic modals, respectively); but in function it flags the power move it seeks acceptance for: the introduction of the speaker's space as the shared topic. Despite their different situational roles, these two speakers were dominant in their respective negotiations. The meta-discourse conditionals are often elliptical, as with text (7), suppressing a judgement like if we can proceed into that, that would be good/desirable (for me). Interestingly, the same mechanism that keeps the consequent clause suppressed can serve a different strategic goal. Here is an example where acceptance is definitely sought, because the speech act in question is an offer: (20) (Rockley: haggling round) Seller: Can I float something. If we were to come in . around. four point . two? The suppressed consequent clause would run ...would that be good enough for you? Where ellipsis dictated by social politeness would cut out good (for me), this time the speaker requires protection for his competence face (the seller would reference his embattled state too explicitly by completing the structure). The offer is much more in the 'shared space' than the quid-pro-quo of text (11); here, where he is on the defensive, it is in the speaker's interest to keep his own space floating as tentative, so that it may resemble the contingent type a little. If the offer is rejected, it was, after all, just made tentatively. Similarly, a negative evaluation may be kept back:

Conditionals: Your space or mine?

163

(21) (Rockley Glass) Seller: But what IS your concern? Buyer: Well, if we see a loss in productive capacity at the same time as we get these three other comparable offers . I mean- [outbreath and worried frown]. The buyer knows full well that capacity has been lost; he is therefore using the conditional epistemologically, as a basis for reasoning. It is a polite mitigating trait that he does not treat it as a fact, but just a conversationally given; furthermore he withholds the conclusion, which would damage the hearer's face. We are now near the centre of the cline, where speakers dissociate themselves a little from their own space to accommodate their hearer. 2.2. Halfway house: culturally shared space Further along the cline we find the kind of mapping where the space that is suggested as shared cannot be said to be Owned' by either party. As argumentation the shared space draws authority from the assumption that the assertion is common knowledge in the culture or script, or logically self-evident. In being Objective' in this way, the speaker plays down his own stake. First, a logical deduction: (22) (The Budget Cuts Case: the managers are trying to find substantial savings, preferably in someone else's department) A: You still haven't convinced me that sixteen people can't do the jobs of eighteen here B: Well you haven't convinced ME that they CAN. If a person's working eight hours a day at 100% productivity, how on earth can you expect them to carry another 20% load? Secondly, a general rule, here backed by experience: (23) (Budget Cuts) Obviously, one would prefer to do without security, but we know what happens . if there 's no security . s tuff will walk out of the place. The speaker is in charge of the security; in leaning on both a rule and genuine shared experience, he plays down his own interest. Thirdly, the analogy or example, chosen so that the hearer must do his own inferencing work in order to make the utterance relevant: (24) (Bridge 1: the claimant had a small claim partially paid out some years earlier)

164

Anne Marie B low-M0ller

Claimant: It stinks, Steve. If you owe 50 pounds and pay 10. it stinks ... /£ you go hungry and you want a loaf, and get offered a slice, you take it 'cause you 're hungry, but it stinks. This application is tendentious because it smacks of the speaker's stake: the judgement may be culturally shared in general, but the implication is that the unflattering analogy is relevant in this particular case. The availability of a general rule or commonplace lends both authority and pathos to the judgement and protects the speaker's social face (cf. the more directly unpleasant variety / think you stink, Steve -your paying me only a fraction stinks). 2.3. Your space A speaker who agrees completely with the hearer may supplement him or her with further examples, thus explicitly making the hearer's space shared. There are no examples at all in the corpus of such supportiveness; however, just as one can summarise in a way that reveals one's stake, so one can adopt the hearer's space in a way that catches the hearer on the wrong foot: (25) (Bridge 2: the claimant, A, is trying to talk the manager, B, out of going through with the court case) A: But he [the solicitor] only had half the facts, surely, truly B: (...) your statement of claim will have to be formally amended if you want ίο rely on additional facts. Then we ΊΙ serve an amended defence, if we need to, in response to those amendments in the statement of the claim. A: Yes, I mean, it 's no good amending things if you think you have a bad case, is it. The central characteristic of the conditionals grouped here as 'interlocutor's space' is that the speaker examines a space that the hearer assumes, or treats it as if the hearer assumes that it is shared. For the vast majority of the tokens in the corpus, the mapping procedure in this group is the opposite of what we saw above, where the speaker invites the hearer to share his space, which is treated as given; in the examples below he queries the status of something apparently given by foregrounding the speculative element. The most important discourse function of the conditionals in this group is cancellation of inference, primarily of expectations that the hearer draws from context, situation or script. A large proportion of 'your space' citations are followed by but, glossable as Ί am allowing your point for the sake of argument, but I am about to cancel it'.

Conditionals: Your space or mine?

165

If we regard the family likenesses of this function, it is apparent that also the concessive conditional is part of the same argumentative cline. The concessive function, typically marked with even if, is normally disregarded in discussions of conditionals because it specifically does not set up a contingent relationship; even if it rains they won't stop the match is apparently a completely different relation from the contingent ///'/ rains, they will stop it. In context it can be seen that these conditionals behave like other circumstances attached to negations: they address themselves to an assumption that should be cancelled; cf. the work on polyphony that projects a 'voice' that expects that they will stop the match (Ducrot 1980; see also Auwera 1983; Dancygier & Sweetser 1997). The counterfactual conditional (if they had stopped it) is another natural member of the family, not only because it cancels its own truth value, but because in discourse it is typically used to cancel a given proposition that, if true, would have been an argument against the speaker. In the following the group has been divided into sections that deal with dissociation from a given space, reasoning against a given space, and cancelling the space of an imagined Other'. 2.3.7. Dissociation In cases where the speaker has an interest in putting distance between himself and the assumption, the conditionals are often accompanied by other references to the hearer's space; this ties the hearer to the logic of the rebuttal of his model: (26) (Bridge 1: the claimant insists that the council has promised to repair the bridge. The manager foregrounds the speculative nature of the assertion that the interlocutor treats as given:) Manager: Yes, but the fact that Mr Jones may or may not have made an offer to the council wouldn 't place them under any obligation. They said that /£ and I refer precisely to the agreement which you have mentioned yourself that they would proceed with the work quote "Once started [3 syllables inaudible] strengthening or reconstructing the bridge with all expedition " - that 's all they agreed to do with [company], they said that IF they did do this job, because it was never positive and certain that they would do it, if they once STARted this job they would complete it with all speed. That 's the only obligation they committed themselves to. The dissociation continues with the use of concessives:

166

Anne Marie Bülow-M0ller

(27) (Bridge 2: different claimant, same problem) Claimant: But these houses would be stranded then, if the bridge falls down. Manager: If it did, even assuming, even if it WERE to fall down, it still wouldn 't be [name] Council s responsibility. The smooth parallel between if it did and even if it were to goes to show that all that is conceded with this use of the conditional is that the hearer, or some projection of the hearer, holds a given model to be possible. The use of epistemic distance cancels the hoped-for inference, glossable as 'Let's allow the unlikely situation that the bridge falls down: it does NOT lead to the conclusion you expect'. The concessive use is, then, a marked choice on the same commitment cline as the distantly allowed space. 2.3.2. Reasoning: creating coherence from the hearer 's space Texts (14) and (15) above showed how reasoning works when the speaker creates coherence from his (own) given model. But reasoning may of course also be a matter of shifting the interlocutors' view of their own model. (28) (Rockley Glass: in this round, the chief buyer has shown concern over "closure of one of the facilities ... and I would presume, a loss of productive capacity as well") Seller: Ok, do you see, Burt. in there a kind of quid pro quo? I mean if_ we shut capacity that 's because we say well, look, have we got an order for tomorrow to fill in that plant (Q.: Sure) yes - or the day after. I mean, the stronger you can say to us "we believe this is the sort of number we are going to want at some point in the future ", the more evenly paced can be our own investment decisions and capacity planning, which will then reflect through to you in terms of the overall package you get, in terms of price, quality and the rest of it-yes? The seller here uses the buyer's wording (shut capacity) instead of calling it what he normally does (putting one plant on stand-by). He offsets this choice by the conditional in order to make it less real (as in Let us say that we have "shut capacity", as you say); but by providing an account, he cancels the implication (that shutting capacity is damaging). The construction is curiously like the metalinguistic negation, of the type he didn 't lose a finger, he lost an arm, which also stops inference by supplying an amended, more real version or model. The construction simultaneously buys into the common-space regularity that makes this an eminently reasonable decision: just as cats scratch if they are teased, so plants should be closed if it makes economic sense. His account of a damaging

Conditionals: Your space or mine?

167

decision is thus lifted from a private action to a instance of rule-following. This saves competence face. Cohesion of a sort that we could call 'eristic quotation' occurs when the speaker cites a previous turn to mock it or take issue with it: (29) (Dental Clinic: the buyers are resisting the efforts of the sellers to make them buy the whole packet, not just the cheap dental chair) Buyer: We have discussed and decided that we are only interested [1.5 sec] in the CHAIRS. [several turns on the feasibility of splitting the product line] Seller: I'd like to mention-. As you know, we are negotiating with two Danish representatives, [missing section describing how the other company will incorporate the chair in their "total clinic"] Ok? Ok. Now, i£_ we 're talking just chairs with YOU, I mean, if we 're gonna talk on the basis of just CHAIRS? - of course we would go to the OTHer one. This is clearly a conditional that supplies coherence by reaching back over several turns, and announces the relevance point for the upcoming newsworthy suggestion. But it is not backgrounded: it is not only repeated, it is decked out with emphasis points that flag the weirdness of the scenario, from the speaker's point of view. It is a quotation that uses the buyer's space to reason from; by showing the undesirable consequences, the seller accomplishes a move that takes the buyer's space and rubs his nose in it. The conditional implicitly references the alternative, as noticed by Ford (1997); here that means that the buyer is invited to cancel his model and supply one where they can talk more than "just chairs". (30) (Budget Cuts) A: We have two storemen responsible for the goods inwards and [B. Yeah] and and we have the four under the control of the dispatcher. is there an opportunity to consolidate their functions there. they can't be working all the time? B: Well they ARE erhm and I'm a bit concerned that we 're getting bogged down . and I- yeah, I think if we have to go through job by job jus-just to find out what each person does, we 're gonna be here all afternoon The conditional in this example ostensibly refers back to A's solution to their problem. But in comparison with text (15), where the gist was provided on the background of a positive interest, this time the speaker is negative. He uses this stake-biassed piece of reasoning to show that an undesirable antecedent (which he imputes to A) will yield undesirable results. The further the speaker is from

168

Anne Marie Bülow-M0ller

A's original in his wording, the more we can look for a stake. In fact, this is the mechanism of the well-known eristic argument type known as 'arguing against a straw man', a statement composed for the occasion to look like the interlocutor's model, but exaggerated and hence an easy target. The function of the conditional is, as above, to invite a different model. 2.3.3. Cancelled space: Counterfactuals With counterfactuals, the speaker does the complete job: instead of cancelling something provided by the interlocutor, he supplies it himself on behalf of the Other'. In the simple form, counterfactuals have no particular argumentative value in themselves; they hold up an already cancelled model of the past, where things would have been done differently: (31) (Bridge 2) Manager: / wasn 't a party to the negotiations, but all I can say is I wouldn 't have paid him a penny had I been in control at that time However, even this usage is stake-bound: the speaker argues against the assumption that because a claim was paid out in the past, it should create a precedent; this is the inference that is scotched at speech act level. A more typical function for counterfactuals is the negative argument, which explicitly contradicts the hearer's assumption: (32) (Bridge 1: the claimant argues that the Council must be responsible for the bridge) Claimant: [the owners] wouldn't have planning permission if provisions hadn'/ been made. Because if they ain 't had planning permission, they couldn 't have built the factory, could they? But epistemic distance also serves a facework purpose in argumentation: (33) (Bridge 2) Claimant: / mean, why can't you settle this amicably Manager: Well we would if we thought there was a liability, Mr Jones, but we don't think there IS a liability Why mention something and cancel it? This use of the interlocutor's space demonstrates that the speaker has actually thought about the opponent's point of view; if he has entertained the idea he is entitled to cancel it when he can give good reason. This is one of the cases where the tentative conditional boosts both social and competence face. Naturally, this sympathy-booster, too, can be used strategically:

Conditionals: Your space or mine?

169

(34) (Bridge 1: the claimant argues that the insurance company can afford to pay out the claims) Claimant: They ΊΙ still be bankrupt after that and it 's only bloody money, Steve, if it "was life it would be different, but it s only money. The strategy is the same: suggest a bigger alternative which you are not asking for, and you appear more reasonable. The speaker here manages to suggest that it is reasonable of him to ask only for money from the insurance company.

3.

Consequences

We are now in a position to revert to the introduction and conclude that a view of conditionals that relies on the commitment scale has repercussions both for the established cognitive categories, which have been shown to contain blends and cross-pollinations, and for the pragmatic functions, which have been shown to modify the simple shared-space model. This gives rise to a tentative model of conditional functions plotted along the cline. 3.1. Blends Spoken language is not as orderly as written language, and not at all as orderly as constructed examples. For example, in text (23) there is a choice of two consequent clauses for the if-clause: we know what happens if we have no security stuff'walks out of the place; the intonation suggests that the speaker 'forgets' the completed structure and completes it again. Spoken language has a tendency to telescope something that written language would make more formally explicit. This means that it is hard to make meaningful distinctions where all the conditionals can be assigned to a particular class on the lines of Sweetser (1990). In her classification conditionals are subdivided into three groups: (a) Content: the classic contingent prediction; e.g if he loves her, he will type her thesis (b) Epistemological: reasoning from evidence; e.g. if he typed her thesis, he loves her (c) Speech act conditional: information offered contingent on a relevance condition; e.g. if you must know, she paid him to type it for her In the latter case, the consequent clause is true regardless of the antecedent.

170

Anne Marie B low-Moller

In practice, i.e. in discourse, it turns out that blends occur because of the epistemological status of the antecedent: if the speaker regards a model as accepted, the conditional will be classifiable as epistemological, but if it is assigned to interlocutor's space, it will be contingent. Since one cannot always check speakers' beliefs, we have no choice but to regard the use of conditionals as a mapping somewhere on a cline rather than a choice of a particular category. Consider the following example: (35) (Bridge 3: this claimant has rung up to complain that the insurance company's solicitor, mr X, has been unwilling to talk to the claimants' newly appointed solicitor, mr Y) Manager: I didn 't know you had a solicitor on the record (...) well, he certainly hasn 't written on the case Claimant: Well he has made many phone calls to mr [X] and he s contacted him and he can't see him for the next two weeks. Will you have a word with mr [X] to see if he can see mr [Y] as a matter of urgency (...) ask him if he ΊΙ ring mr [Y], will you please Manager: Well, ifmr [Y] is on the record I will, but I have no reason to believe that mr [Y] has confirmed to anyone in writing that he is on the record here The if-clause could equally reasonably be regarded as exponents for two different classes: Content: Epistemological:

If and only ifmr Y is on the record, I'll ask Ok, ifmr Y is on the record, I'll ask.

In practice, it is probably in-between: the manager has protected his social face by allowing the possibility that the solicitor is in fact on the case, although he does not believe it. But note the consequences: if the claimant hears an epistemological conditional, as is likely, given his stake, he hears a promise (the manager will ring mr Y); if the manager intended a content conditional, as is likely, given his stake, he has not promised anything yet. The borderline between contingent conditionals and speech act or meta-discourse conditionals is equally fuzzy. It has been argued above that conditionals function as hedges that mitigate directions, and that this use is almost codified, so that phrases like if you would like to sit here can do without the consequent (that would be helpful). There are also semi-independent forms like if you like and fully grammaticalised forms likeplease (ifyou please), both of which function as ordinary adverbials with no dependency relation implied. It may be this face-based function that rubs off on blends like the following example:

Conditionals: Your space or mine?

171

(36) (Budget Cuts): A: There s no point if it s just a loan B: It's definitely a loan. If you look at this [points in his papers] it is definitely a loan On the one hand, this is contingent: If you look at this, you will see that it is definitely a loan. On the other hand, the 'definiteness' is regarded as true in any case, and the if-clause concerns only the relevant circumstance for the information, rather like the canonical examples of speech act conditionals, If you need anything, my name is Susan or If you are thirsty, there is beer in the fridge. Again it makes sense to conclude that if the speaker regards the if-clause as fully given and backgrounded, it approaches the speech-act formula that does not actually ask for confirmation. A view of conditionals as a bundle of functions based on a speculative element and plotted along a cline of commitment has the advantage that it includes the embedding use of // In discourse there would seem to be fuzzy distinctions between subordinating conjunctions and embeddings following verbs of doubt: (37) Ask them I doubt What It seems weird There is no point

if it 's just a loan

In each case it is the epistemological status of the possibility that is in focus. The distance is very short to the otherwise separately treated concessive clauseinternal if, as in well-meaning if pointless. Supplying one's own concessions, attributable to a different 'voice', is a little less committed, a little more doubtful, than a fully-fledged although or despite. 3.2. The cline of commitment Finally, we can conclude that the question of the 'shared' status of the if-clauses cannot be solved just with reference to the given-new contract, not even in its augmented form, where we include material that is broadly recoverable rather than just contextually available. To describe their status in functional terms, we need to consider the element of commitment. Because a conditional construction involves two spaces that must be mapped together to create common ground, we have a more complicated structure than

172

Anne Marie Bülow-M0ller

the given-new contract inside the clause. It has been argued above that we are dealing with a cline rather than a polarity, and that the deciding factor is the foregrounding or backgrounding of commitment to the space represented in the given/recoverable theme. A little simplistically, it can be illustrated along a single axis: My space 1. My space and I expect you to share it.

Your space

2. My space and I expect you to query it.

3.

4.

5.

Our space, externally agreed.

Your space and I share it.

Your space and I query

it.

In the corpus, and, I believe, as a general tendency, the functions have core examples: Type (1) : backgrounded assumptions, as in proper contingent relations Type (2): accounts and reasoning Type (3) : rule-governed general conditionals Type (4): supportive examples Type (5): counterfactuals Finally, it seems that the functions that we have seen demonstrated are all argumentative, not because they were drawn from argumentative discourse, but because there is something inherently two-sided in conditionals, where the mapping function cannot take place without some sort of projected agreement on a shared space. Any parent knows that the contingent 'if you do (eat your spinach), I'll do y (give you strawberry icecream)' is an argument; information that 'the cat scratches if it is teased' is an argument for not teasing it; if 'the game is cancelled', some course of action is advocated; examples are regularly given to illustrate a point; rules and generalities are regularly cited as backing in argumentative chains; and reasoning from evidence is of course a central argumentative action. I would like to suggest that where causatives can be said to function as monological arguments, conditionals are dialogical, and because of their mapping structure, singularly sensitive to interactional use, and a great resource in furnishing spaces.

Conditionals: Your space or mine?

173

References Abbott, Barbara 2000 Presuppositions as nonassertions. Journal of Pragmatics 32: 1419-1437. Akatsuka, Noriko 1986 Conditionals are discourse-bound. In: Elizabeth C. Traugott, Alice ter Meulen, Judy S. Reilly and Charles A. Ferguson (eds), On Conditionals, 333-352. Cambridge: Cambridge University Press. Antaki, Charles 1994 Explaining and Arguing: The Social Organization of Accounts. London: Sage. Auwera, Johan van der 1983 Conditionals and antecedent possibilities. Journal of Pragmatics 7: 297309. 1997 Conditional perfection. In: Angeliki Athanasiadou and Rene Dirven (eds.), On Conditionals Again, 169-190. Amsterdam: John Benjamins. Brown, Penelope and Stephen C. Levinson 1987 Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press. Comrie, Bernard 1986 Conditionals: A typology. In: Elizabeth C. Traugott, Alice ter Meulen, Judy S. Reilly and Charles A. Ferguson (eds), On Conditionals, 77-99. Cambridge: Cambridge University Press. Dancygier, Barbara 1998 Conditionals and Prediction: Time, Knowledge and Causation in Conditional Constructions. Cambridge: Cambridge University Press. Dancygier, Barbara and Eve Sweetser 1996 Conditionals, distancing and alternative spaces. In: Adele E. Goldberg (ed.), Conceptual Structure, Discourse and Language, 83-98. Stanford, California: CSLI publications (Center for the Study of Language and Information). 1997 Then in conditional constructions. Cognitive Linguistics 8/2: 109-136. 2000 Constructions with if, since and because: Causality, epistemic stance, and clause order. In: Elizabeth Couper-Kuhlen and Bernd Kortmann (eds.), Cause - Condition - Concession - Contrast: Cognitive and Discourse Perspectives, 111-142. Berlin: Mouton de Gruyter. Dijk, Teun A. van 1979 Pragmatic connectives. Journal of Pragmatics 3: 447-456. Ducrot, Oswald 1980 Les echelles argumentatives. Paris: Minuit. Edwards, Derek 1997 Discourse and Cognition. London: Sage.

174

Anne Marie Bülow-M0ller

1999

Shared knowledge as performative and rhetorical category. In: Jef Verchueren (ed.), Pragmatics in 1998: Selected Papers from the 6th International Pragmatics Conference. Volume 2, 130-141. Antwerp: IPrA. Fairclough, Norman 1995 Critical Discourse Analysis. London: Longman. Fauconnier, Gilles 1996 Analogical counterfactuals. In: Gilles Fauconnier and Eve Sweetser (eds.), Spaces, Worlds and Grammar, 57-90. Chicago: University of Chicago Press. Fillmore, Charles J. 1986 Varieties of conditional sentences. Eastern States Conference on Linguistics 6: 163-182. 1990 Epistemic stance and grammatical form in English conditional sentences. Papers from the 26th Regional Meeting of the Chicago Linguistic Society, 137-162. University of Chicago. Ford, Cecilia 1993 Grammar in Interaction: Adverbial Clauses in American English Conversations. Cambridge: Cambridge University Press. 1997 Speaking conditionally: Some contexts for /^clauses in conversation. In: Angeliki Athanasiadou and Rene Dirven (eds.), On Conditionals Again, 387^413. Amsterdam: John Benjamins. Ford, Cecilia and Sandra Thompson 1986 Conditionals in discourse: A text-based study from English. In: Elizabeth C. Traugott, Alice ter Meulen, Judy S. Reilly and Charles A. Ferguson (eds.), On Conditionals, 353-372. Cambridge: Cambridge University Press. Fowler, Roger G. 1985 Power: Language as social practice. In: Teun A. van Dijk (ed.), Handbook of Discourse Analysis. Volume 4, 61-82. London: Academic Press. Garcia, Angela 1997 Interactional constraints on proposal generation. Discourse and Society 8/2: 219-247. Givon, Tom 1995 Functionalism and Grammar. Amsterdam: John Benjamins. Grice, H. Paul 1981 Presupposition and conversational implicature. In: P. Cole and J. Morgan (eds.), Speech Acts, 183-198. (Syntax and Semantics 3.) New York: Academic Press. Goffman, Ervin 1967 On face-work: An analysis of ritual elements in social interaction. Orig. 1955; reprinted in Interaction Ritual: Essays on Face-to-Face Behavior. Chicago: Aldine. Haiman, John 1978 Conditionals are topics. Language 54: 564-589.

Conditionals: Your space or mine?

175

Halliday, M.A.K. 1994 An Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Horn, Laurence R. 2000 From if to iff: Conditional perfection as pragmatic strengthening. Journal of Pragmatics 32: 289-326. Johnson-Laird, P. N. and Ruth M.J. Byrne 2002 Conditionals: A theory of meaning, pragmatics, and inference. Psychological Review 109/4: 646-678. Kennedy, Gavin 1998 The New Negotiating Edge: The Behavioral Approach for Results and Relationships. London: Brealey. Levinson, Stephen 2000 Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MASS: MIT Press. Lycan, William G. 2001 Real Conditionals. Oxford: Clarendon Press. Matthiessen, Christian and Sandra A. Thompson 1988 The structure of discourse and 'subordination'. In: John Haiman and Sandra Thompson (eds.), Clause Combining in Grammar and Discourse, 275329. Amsterdam: John Benjamins. Ramsay, Violeta 1987 The functional distribution of preposed and postposed "if and "when" clauses in written discourse. In: Russell S. Tomlin (ed.), Coherence and Grounding in Discourse, 383—408. Amsterdam: John Benjamins. Schiffrin, Deborah 1992 Conditionals as topics in discourse. Linguistics 30: 165-197. Scollon, Ron and Suzanne Wong Scollon 1995 Intercultural Communication: A Discourse Approach. Oxford: Blackwell. Sweetser, Eve E. 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. 1996a Reasoning, mappings, and meta-metaphorical conditionals. In: Masayoshi Shibatani and Sandra A. Thompson (eds.), Grammatical Constructions: Their Form and Meaning, 221-233. Oxford: Clarendon. 1996b Mental spaces and the grammar of conditional constructions. In: Gilles Fauconnier and Eve Sweetser (eds.), Spaces, Worlds, and Grammar, 318333. Chicago: University of Chicago Press. Taylor, John R. 1997 Conditionals and polarity. In: Angeliki Athanasiadou and Rene Dirven (eds.), On Conditionals Again, 289-306. Amsterdam: John Benjamins. Ting-Toomey, Stella 1988 Intercultural conflict styles: A face-negotiation theory. In: Young Yun Kim

176

Anne Marie Bülow-Moller

and William B. Gudykunst (eds.), Theories in Inlercultural Communication., 213-235. Newbury Park: Sage. Virtanen, Tuija 1992 Given and new information in adverbials: Clause-initial adverbials of time and place. Journal of Pragmatics 17: 99-115. Werth, Paul 1997 Conditionality as cognitive distance. In: Angeliki Athanasiadou and Rene Dirven (eds.), On Conditionals Again, 243-271. Amsterdam: John Benjamins.

Chapter 9 Communicative fragments and the interpretation of discourse Martina Björklund

In the last fifteen years or so linguists have become increasingly aware of and interested in the abundance of idioms and other more or less fixed expressions in natural language. Those who study corpus data find a multitude of collocations (e.g. Sinclair 1991; de Beaugrande 1999) and prefabs (e.g. Erman and Warren 2000). That prefabricated or memorized fixed linguistic expressions are not marginal phenomena has also been noticed by those working within the framework of Construction Grammar (see e.g. Fillmore, Kay and O'Connor 1988) and by Jackendoff (1995). In Langacker's Cognitive Grammar such linguistic expressions play a prominent part. As Langacker (1991: 15) sees it, a speaker's linguistic knowledge is a "structured inventory of conventional linguistic units", where 'unit' stands for "a thoroughly mastered structure, i.e. one that a speaker can activate as a preassembled whole without attending to the specifics of its internal composition." That a linguistic description of this type does indeed characterize the "abilities that constitute a speaker's grasp of linguistic convention" in a "cognitively realistic fashion" (Langacker 1991: 15) finds support in Tomasello's (1999: 138-140) observation that in building up their knowledge of a language small children have an item-specific way of using language. In other words, children do not generalize about constructional patterns across verbs, but use so-called 'verb island constructions', i.e. specific verbs where the slots for participant roles are marked only on an individual basis. Thus children seem to learn their language 'unit' by 'unit', to use Langacker's term. What all the approaches mentioned above have in common is that they view more of adult language as item-specific than has generally been usual. With regard to this line of thought Gasparov (1996) holds the most radical view of all. In fact, he outlines a model of language use which he calls lingvistika iazykovogo sushchestvovaniia ('linguistics of existence with and within language') in which the item-specific way of using language forms the basis of all use of language and whose basic unit is termed the 'communicative fragment'. The aim of this chapter is first to describe Gasparov's approach at some length and then to discuss the impact of 'communicative fragments' on the comprehension and interpretation of discourse through a case study. Since interpre-

178

Martina Björklund

tative processes come more sharply to the fore in literary texts, which generally assume a more conscious attention to their design, I base my discussion on the analysis of the short story "Strana" ('The Land') by Liudmila Petrushevskaia.

1.

Communicative fragments

Gasparov (1996: 42-51) points out that there are two types of linguistic activity that we as human beings are capable of: reflection on or analysis of language (from children forming new words by analogy to professional linguists refining their linguistic models) and the intuitive activity of everyday language use. It is the latter ability that interests Gasparov, who develops Bakhtin's (1986: 87; see also Voloshinov 1986: 67-70) claim that when speakers produce utterances, they normally take the words from utterances previously encountered in their linguistic experience, rather than from the language system. In other words, for the speaker, linguistic forms exist only in the context of specific utterances. Gasparov (1996: 104) thus sees a speaker's 'linguistic memory' as a vast conglomerate that accumulates and develops as long as the speaker lives. And he describes this conglomerate not as a system but as an " -dimensional kaleidoscope" (p. 144) formed by a huge supply of communicatively loaded linguistic elements of different sizes, from separate forms of words to fragments of utterances and even entire texts, carrying with them the whole communicative 'landscape' to which they belong, i.e. the contexts in which they typically occur or even the concrete texts in which they have been encountered. The basic unit of such a mnemonic command of language is the 'communicative fragment'. Communicative fragments are defined as discourse fragments of various lengths, stored in the speaker's memory as whole fixed elements (p. 117-118); i.e. they are not generated by the speaker according to grammatical rules (cf. Langacker's 'units'). Gasparov argues that it is with such preassembled elements that we basically operate when we produce and interpret utterances. He also shows (pp. 51-68) that in the long run, considering that linguistic expressions are used over and over again, it is more 'economical' for our memory to work with discrete whole units than to form them or construct them anew by means of rules every time (cf. Langacker's (1991: 15) view of the unit as a cognitive routine). Communicative fragments have the following characteristics (Gasparov 1996: 124-142). First, as is also clear from the definition, they are directly given in the speaker's memory, familiar and recognizable. Secondly, every communicative fragment is crystallized in our consciousness in the form of an integrated representation of a type of situation with all its intellectual and aesthetic con-

Communicative fragments and the interpretation of discourse

179

tent, associative valencies, thematic potential, emotional overtones, and characteristics of genre and style. Gasparov argues that such expressions as chital knigu ('read [imperfective past tense, masculine] a/the book') and chitalgazetu ('read [imperfective past tense, masculine] a/the newspaper'), well known and frequently used by any speaker of Russian, evoke different integrated images of situations. But each of them evokes a whole integrated situation, not a higher order complex formed of smaller units. Thirdly, communicative fragments are given, but in a dynamic way with vague and mobile outlines. This means that speakers of a language will not be able to say with certainty where one communicative fragment ends and another begins; or whether a certain expression constitutes one single fragment or a contamination or combination of several fragments. For instance, is the expression vsia nasha zhizn'est'ne chto inoe, kak ('all our life is nothing but') one fragment or is it a combination of vsia nasha zhizn' esl' and est' ne chto inoe, kak. According to Gasparov (p. 128) there is no final objective answer to such questions, nor can there be. Fourthly, rather than being definite quotations, communicative fragments are allusive. They remind us about something; they hint at something; they refer us to something, but they do it in a manifold vague and elusive way. Thus a communicative fragment does not have any correct, ideal, or initial form. Fifthly, communicative fragments are communicatively loaded with the styles and genres, the plots and thematic fields that they evoke in the consciousness of the speaker. Every communicative fragment that we encounter immediately plunges us into a whole world of potential story lines, modalities of genre, and so forth. Further, communicative fragments are characterized by the plasticity of the mental representations they evoke. Such a representation emerges at the intersection of many potential associative channels along which it can spread and modify itself. For instance, the fragment chital knigu evokes different situations and different contexts and genres depending on who does the reading. Moreover, communicative fragments exist in the form of integrated 'monads' of linguistic experience, forming a unique linguistic world. For instance, the expressions chital knigi ('read [imperfective past tense, masculine] books') and chital gazety ('read [imperfective past tense, masculine] newspapers/the newspapers') are not only plural modifications of the expressions chital knigu and chital gazetu discussed above, but they tend to evoke quite different text worlds. The fragment chital knigi does not evoke an actual act of reading so much as a type of activity or an interest that a person used to have, while the fragment chital gazety easily evokes both types of situation. Finally, in the conglomerate of linguistic experience, communicative fragments are connected not through paradigmatic similarity or contrast but through

180

Martina Björklund

superposition and contiguity. Thus they form not a system but a flexible everchanging continuum of elements of linguistic material that merge and fade into one another in many different directions and in many layers. Communicative fragments are therefore characterized by two types of opposed features: on the one hand, features that isolate each familiar expression as an individual momentarily recognizable linguistic Object', and, on the other hand, those that assimilate and fuse together various expressions in fields of more or less obvious analogies. Gasparov (p. 144) sees every act of language use as a compromise between these two opposing forces. It is the associative links between communicative fragments that make it possible to use them with endless variation and to create new combinations and modifications of familiar linguistic material. According to Gasparov (pp. 151-161) there are three types of basic technical device which enable such creative work with linguistic material. The simplest type of interaction between two or more communicative fragments is analogy. In this case, a certain fragment is understood, manipulated or created according to the model of another familiar fragment (or several other fragments). Analogy can, for instance, explain the mechanism of the use of different 'grammatical forms' of expressions known to speakers. According to Gasparov (p. 157) expressions such as posetite [nash] magazin ('visit [formal imperative] [our] shop'), poseshchenie magazina ('visiting the shop') will undoubtedly be included as memorized fixed expressions in the linguistic memory of many speakers of Russian. But an expression such as posetila [nash] magazin ('visited [feminine] [our] shop') would most probably not be prefabricated. It is, of course, possible to imagine situations in which it could be used, but these situations are not evoked with the same immediacy in the form of recognizable precedents as is the case in the former two expressions. Nevertheless, Russians have no difficulty either in forming or understanding the latter expression since there are communicative fragments with analogous forms and similar spheres of use: in this case, for instance, posetila vystavku ('visited [feminine] an/the exhibition'). Incidentally, Gasparov's intuition is here corroborated by the use of these communicative fragments on the Internet (www.rambler.ru 19 December, 2002). The fragment posetila nash magazin only appears on 3 websites, whereas posetite nash magazin appears on 139 sites and poseshchenie magazina on 418. Compare also posetila vystavku, which can be found on 80 websites. The second type of associative interaction is called contamination (p. 158). This device is more complex and brings about a more radical reshuffling of two or more communicative fragments so that the disintegrated components of all the fragments coexist in the consciousness of the speaker in the form of a kind of kaleidoscopic whole. To illustrate, Gasparov experiences the frag-

Communicative fragments and the interpretation of discourse

181

ment original 'nost' etoi mysli porazitel'na ('the originality of this thought is striking') as a contamination of the more fixed expressions: porazitel'no original'naia mysl' ('strikingly original thought'), mysl' porazhaet [svoei] original'nost'iu ('the thought strikes you with [its] originality'), and eta mysl' [vovse] ne original 'na ('this thought is not original [at all]'). The third device, amalgamation, is even more complex and manifold. In this case, some of the underlying fragments do not even leave any explicit trace in the resulting expression. Their contribution is only an allusive hint. Nevertheless, this hint is necessary for the identification and apprehension of the new whole. As an example, Gasparov (p. 160) analyses a line by the young Pushkin (about Karamzin the historian) "On tarn, v dymu stoletii" ('He is there, in the smoke of the centuries'). Gasparov explains the suggestive richness of the expression v dymu stoletii by the fact that it evokes a multitude of resonances in a speaker of Russian, but indirectly through allusive hints, rather than through direct analogies. There is, however, the familiar expression v dymu srazheniia ('in the smoke of the battle') which forms a kind of analogic background. Other fragments that play a part in the formation and appreciation of Pushkin's line include, for instance, [skrylsia] v tumannoi dali ('[hid himself] in the misty distance'); skvoz'dymku smutno prostupali... ('through the haze there vaguely appeared ...'); minuvshie stoletiia ('bygone centuries'); and many more. According to Gasparov, all three devices are probably at work in one way or other in the formation and interpretation of every concrete expression, but one of them may be more obvious than the others. In his book, Gasparov further describes different ways of putting together communicative fragments through what he calls seam/junction (shov) and how this putting together is steered and restricted by the 'communicative contour of the utterance'. A discussion of these is, however, beyond the scope of the present chapter.

2.

Communicative fragments in the short story "Strana"

Before going into the discussion of communicative fragments in the short story "Strana", I will briefly comment on the narrative form of this text. It is a very short story (only six short paragraphs) about the tragic life of a lonely mother who is an alcoholic and her little daughter. In terms of the narrative, this short story concentrates on the final situation - the outcome of the narrative sequence of events - described in the present tense. The present tense description is interspersed with fragments evoking the initial situation and the preceding sequence of events. (In short, a beautiful woman lived with a handsome man. They had a

182

Martina Björklund

daughter. The man met another woman and married her. The woman lost both her partner and her friends. She started drinking, and drank more and more, but she thought that she always took care of her daughter.) In an earlier article (Björklund 1998), on the basis of an analysis of the linearization and perspectivization of the text, I argue that the story is construed as the attempt of the main character to fathom her situation, looking at herself from the outside. Another possible reading would be that the main character and her life is described by a 5fozz-narrator, who is an average person, probably a woman, with certain intellectual aspirations amounting to a rather trivial literary taste. Let us look at the first paragraph of the text to see what discourse elements seem immediately recognizable communicativefragments- and what communicative landscapes they evoke. Kto skazhet, kak zhivet tikhaia, p 'iushchaia zhenshchina so svoim rebenkom, nikomu ne vidimaia v odnokomnatnoi kvartire. Kak ona kazhdyi vecher, kak by ni byla p 'ianoi, skladyvaet veshchichki svoei docheri dlia detskogo sada, chtoby utrom vse bylo pod rukoi. 'Who can say how a quiet, drinking woman lives with her child, seen by nobody in a one-room flat. How every evening, no matter how drunk she is, she packs her daughter's little things for the kindergarten so that everything is at hand in the morning.'

This initial paragraph introduces the main contexts and lines of associations to be further developed throughout the text. The text starts out with a fragment that, in this literary context, plunges the reader into the genre of unusual stories that are unknown or difficult to grasp: Kto skazhet, kak ('who can say how'). The same literary motif recurs in the fragment nikomu ne vidimaia ('seen by nobody'). The fragment p'iushchaia zhenshchina ('a drinking woman'), which refers to the main character of the story has a written flavour. It is a milder, more neutral, more literary and intellectual way of naming such a person than, for instance, alkogolichka ('alcoholic [feminine]'). The communicative fragment p 'iushchaia zhenshchina is, for example, widely used in various kinds of text where such women are discussed from a medical point of view. The above-mentioned literary and 'intellectual' expressions are combined with communicative fragments that evoke, on the one hand, everyday contexts of city life that could very well be spoken: [zhivet] v odnokomnatnoi kvartire ('[lives] in a one-room flat'); dlia detskogo sada ('for the kindergarten'), and, on the other hand, the motherhood and motherly care of the main character: zhenshchina so svoim rebenkom ('a woman with her child'); skladyvaet veshchichki svoei docheri ('packs her daughter's little things'); chtoby vse bylo pod

Communicative fragments and the interpretation of discourse

183

rukoi ('so that everything is at hand'). In the second fragment, the motherly care is reinforced by taking the child's perspective on 'the things' through the use of the diminutive form veshchichki. To the reader it is, however, a grim picture of motherly care, due to the preceding fragment kak by ni byla p 'ianoi ('no matter how drunk she is'), which could also be a combination of kak by ni byla and byla p'ianoi. Thus, the communicative fragments of the beginning of the short story evoke a tragic everyday situation, described partly with aspirations to objectivity, or at least without condemning or looking down on the woman alcoholic portrayed, and partly in the key of the literary convention of peculiar, often sad stories about the life of a tragic heroine. All the types of communicative context established through the communicative fragments that have been discussed so far are further developed throughout the text. In the following, I will, however, concentrate on the literary communicative fragments, since they stand out as the "evaluative meta-structure" (Polanyi 1982: 518) that build up the 'point' of the story. In the second paragraph, the appearance of the main character is described in the following communicative fragments: First, so sledami byloi krasoty na litse ('with traces of her former beauty in her face'), which has become a cliche and is normally used in fictional texts of minor literary merit, but it can also appear in everyday texts, such as newspapers and advertisements. Then, brovi dugami ('arched brows') and nos tonkii ('thin nose'), which belong to the standard literary description of a beautiful woman. If the text is read as an attempt by the main character to see herself from the outside, the passage under consideration reads as what the woman sees when looking at herself in the mirror and how she puts this image into words. The words that she has at hand come from rather trivial literature. In the same paragraph we also learn about the father of the child, and he is also described in standard literary fragments: iarkii blondin ('bright blond-haired man') and s iarko-krasnymi gubami ('with bright red lips'). In the third paragraph, we learn about the new wife of the blond-haired man. She is also described in words that evoke a literary context, words that imitate literary spoken language (Larisa Mokroborodova, personal communication): ne spuskaet nikomu nichego ('does not forgive anybody for anything'). In the fourth paragraph, i.e. halfway into the text, a long and complex full NP with a literary tone refers to the already well established main character: mat', u kotoroi doch Ot blondina ('the mother who has a daughter by the blond-haired man'), an expression which is not a standard communicative fragment, but rather an amalgamation of different fragments, bringing to mind unhappy unmarried mothers standing outside society (e.g. in Dostoyevsky's early sentimental novels).

184

Martina Björklund

In the last two paragraphs, we return to the frame of the unusual, unknown story. The fragment nikto na svete ne znaet, [kak oni zhivut] ('nobody in the world knows [how they live]') comes as an answer in the fifth paragraph to the question kto skazhet ('who can say') of the first paragraph. This fragment also evokes childrens' categorical way of thinking and expressing themselves through [nikto] na svete ('[nobody] in the world') (LM, p.c.). In the last paragraph, the fragment nikto ne znaet ('nobody knows') is repeated twice. What nobody knows is "what divine dreams" the mother and the daughter have and how they fall asleep as soon as they put their heads on the pillow "to return to the land" (chtoby vernut'sia v tu stranu) "which they will leave again early in the morning". This is the only time the word strana ('land') of the title appears in the body of the text. And it is now clear that the strana of the title is a truncated form of such standard communicative fragments as strana snov and strana snovidenii ('land of dreams') (LM, p.c.). The analysis of literary communicative fragments thus pinpoints the image of the woman, whether it is her own image of herself, or one that a skaz-narrator has of her. The story is built of communicative fragments evoking basically two different contexts, viz. first, the context of a contemporary, lonely mother, and her everyday life with her daughter in a city, and second, literary contexts, especially of sentimental stories with heroines, such as Karamzin's 'Poor Liza", the characters in Dostoyevsky's early sentimental novels, and later less well-known and more stereotyped sentimental literature. Thus, to a reader with a sufficiently wide spectrum of communicative fragments in her 'linguistic memory', the contexts evoked by the communicative fragments, in addition to the linearization and perspectivization of the text, will trigger the image of a woman whose life is framed in terms of the fates of the tragic heroines portrayed in sentimental literature. It then turns out that at least one 'point' of the story is that the main character practises escapism in every conceivable way: she drowns her sorrow and unhappiness in liquor (cf. Poor Liza, who drowned herself in a pond), she conceives of herself in terms of literary characters and their fates, and the world of dreams is where she would prefer to stay. Compare the last fragment of the text: nuzhno bylo by nikogda ne prosypat'sia ('they should never have woken up'), which, incidentally, echoes the last words of a famous Russian romance: O, esli b nikogda ia vnov'ne prosypalas'... (Oh, if I never woke up again') (LM, p.c.).

Communicative fragments and the interpretation of discourse

3.

185

Conclusion

If Gasparov is right, then from the point of view of language use as an intuitive activity ('the existence with and within language'), an essential part of the reception/interpretation of any discourse (whether artistic or not) is the recognition of communicative fragments together with their accompanying mental representations and the whole communicative landscape that they evoke: genres, styles, plots, thematic fields, even concrete texts and utterances. In other words, much more than the 'basic' referential information is received: the utterance is anchored to certain types of 'world views', situations, speech genres, registers, etc., which may produce 'readings between the lines', impressions of the speaker/text producer, and the like. This is normally done momentarily, without effort and reflection, especially in everyday spoken communication. But if an utterance or discourse includes too many unfamiliar communicative fragments that do not trigger the mental representations of relevant communicative landscapes, understanding is impaired in various ways and misunderstanding may arise. Mixing, combining, and fusing communicative fragments from everyday speech and contemporary life with fragments of more or less stereotyped literary expressions, Petrushevskaia has composed a strongly suggestive text of great literary merit. In my analysis of "Strana" I have traced the communicative fragments that evoke the landscape of sentimental literature and stand out against the background of more everyday fragments. As I have shown, these communicative fragments contribute to the impression that the narrator or the main character sees the situation at hand through the prism of rather trivial literature. This dimension of the text would be lost on a reader with no experience of a certain type of literary text. It is thus only through relevant previous linguistic experience, or in other words, through the possession of a vast enough supply of communicative fragments in her 'linguistic memory' that a reader can fully appreciate the text, whether perusing it in the mode of 'language use as an intuitive activity', without too much reflection, or in a more reflective mode of reception/interpretation, of which my analysis is an example. Gasparov's ideas about language viewed from the perspective of the intuitive activity of everyday language use form a fresh and challenging contribution to linguistic discussion. But, as mentioned in the introduction of the present chapter, many other linguists working within different linguistic frameworks have also begun to pay attention to the abundance of prefabricated or fixed expressions in actual discourse. It is my conviction that such studies of different aspects of various kinds of preassembled units are of prime importance, if we want to describe language in a "cognitively realistic fashion" (cf. Langacker 1991: 15).

186

Martina Bj rklund

References Bakhtin, Mikhail Mikhailovich 1986 Speech Genres and Other Late Essays. Austin, TX: University of Texas Press, de Beaugrande, Robert 1999 Discourse studies and ideology: On 'liberalism' and 'liberalisation' in three large corpora of English. Discourse Studies 1/3: 259-295. Bj rklund, Martina 1998 JlHHeiiHoe pasBHjue H nepcneKTHBa B HappaTHBHOM τεκστε. Studio Slavica Finlandensia 15:49—59. Erman, Britt and Beatrice Warren 2000 The idiom principle and the open choice principle. Text 20/1: 29-62. Fillmore, Charles J., Paul Kay and Mary-Catherine O'Connor 1988 Regularity and idiomaticity in grammatical constructions: The case of let alone. Language 64/3: 501-538. Gasparov, Boris Mikhailovich 1996 flsbiK, noMxmb, oopas: numeucmumf&biKoeozo cyufecmeoeaHua. MocKBa: HoBoe JiHTeparypnoe ooospenHe. Jackendoff, Ray 1995 The boundaries of the lexicon. In: M, Everaert, J-E. van der Linden, A. Schenk and R. Schreuder (eds.), Idioms: Structural and Psychological Perspectives, 133-165. Hillsdale, New Jersey: Lawrence Erlbaum. Langacker, Ronald W. 1991 Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin / New York: Mouton de Gruyter. Petrushevskaia, Liudmila 1995 Strana. In: Taina doma: Povesti i rasskazy, 54-55. Moscow: SP "Kvadrat". Polanyi, Livia 1982 Linguistic and social constraints on storytelling. Journal of Pragmatics 6: 509-524. Sinclair, John McH. 1991 Corpus, Concordance, Collocation. Oxford University Press. Tomasello, Michael 1999 The Cultural Origins of Human Cognition. Cambridge, Massachusetts: Harvard University Press. Voloshinov, Valentin Nikolaevich 1986 Marxism and the Philosophy of Language. Harvard University Press.

Chapter 10 Drawing the line: A contested conceptual model in Danish "child care talk' Peter Harder

The nature and role of conceptual models in human understanding is one of the core concerns of cognitive grammar. One aspect which has been the subject of intensive discussion is the metaphorical principle of using structure recruited from one domain, the 'source', as a way to understand another, 'target', domain. In their most basic form, such mappings have a natural directionality based on bodily grounding: the overall source of understanding is generally assumed to be primary bodily experience, which is then mapped on to less basic forms of experience. 'Force dynamics' as investigated by Talmy and others is one of the most powerful illustrations of how the basic concerns of cognitive grammar can be brought together in throwing light on grammatical constructions, metaphoricity and grounding. In principle, the metaphorical process does not depend on overt linguistic articulation of these mappings and the categories they reflect; they arise more or less automatically as part of the structure of human experience. However, even with the most basic types of metaphorical mapping, there is another grounding dimension that is sometimes overlooked in cognitive linguistics and which makes the natural directionality less straightforward than it may sometimes appear. This dimension has to do with the way patterns of understanding are shaped by the interactive experience of the subject, rather than simply by the fundamental nature of bodily experience, cf. Sinha (1999), Harder (1999). It is the role of this element which is especially salient in the area this chapter will be concerned with. In the case of the pervasive 'force' model and its mapping, e.g. into the force of an argument in the epistemic domain (cf. Sweetser 1990), this involves the landscape of forces in which the discourse is currently embedded - how much do arguments count, who determines what the force is, and what back-up may arguments have if they fail to do the job on their own? Unless you know that, you do not know precisely what is meant by 'the force of an argument' in actual discourse. Before exploring this subject further, I would like to stress that the emphasis on interactive shaping should not be seen as creating an alternative to bodily experience, which is fully as important as cognitive linguistics literature makes

188

Peter Harder

it out to be, and which retains its polemical force in relation to abstract Objective' reference as discussed in the logical tradition. The point is rather that bodily experience underdetermines the actual shape of processes of understanding. Life offers more than one way of grounding and aligning experiences, within the limitations that go with having, or rather being, a body of a certain kind. The physical wiring-up of the body and the nature of the experiences it generates are systematically shaped from before birth by the kinds of response we receive, and in social interaction we see the process at work. This is true of metaphorical mappings no less than of other forms of conceptual structurings of experience. The pragmatic career of a conceptual model may involve a struggle within the speech community to (dis)agree on the proper fixation of the actual mapping, influenced but not determined by the inferences licensed by the mapping and the credentials of the source domain. This chapter discusses a case in point, the grcense metaphor (henceforth GM) as applied to the interaction between adults and children for whom they are responsible. The Danish word can mean 'limit', 'boundary' or '(border)line', and the metaphor is used about situations where the issue comes up of whether the adult On duty' should restrain the child's (children's) activities or not. One reason for taking up the GM is that its popularity and role in child care talk has been described and discussed by a research group interested in its implications for the situation of children in institutions. A point of interest will be the role of mapping from source to target domain as a social process rather than a purely conceptual one. This will also illustrate the difference of perspective between conceptualization as product, i.e. as part of a fully mapped-out conceptual world, and conceptualization as an ongoing process where mental structures meet actual experience and there is a struggle to impose some conceptual order on it. Further, the data illustrate the struggle between various motivating factors in determining the use and interpretation of the GM, focusing on the position of the research group, which was itself part of the process in which interactive factors influenced the potential of the source domain. Among the important factors in the process is the role of particular 'language games' played by the participants when they invoke the model. In this case, the focus is on confrontational public debate. Finally, bodily grounding is cast in a role that differs from being simply the quasi-natural bedrock of understanding: the crucial 'source' boundaries of the GM can be stipulated to be the physical and mental boundaries of a person with whom one interacts (harmoniously). Taken like this, the presence of other people in the universe is seen as creating borderlines that should not lightly be transgressed by a (mature and competent) human agent. Thus conceived, the

Drawing the line: A contested conceptual model in Danish 'child care talk'

189

model extends basic human experience into a general pattern for the relationship between the child and the surrounding world - but this is not a guaranteed source of understanding the GM. First of all there is no guarantee that this is what the child's experience is like, secondly the metaphor itself does not automatically invoke this experience.

1.

The model, the research project and the historical context

Between 1993 and 1996, a group of psychologists and child care professionals in Denmark carried out a large-scale survey of the role of a conceptual model whose core term is Danish grcense. The model has played a considerable role in the last decades in Danish conversations about how to interact with children, and has been important especially for those who are in positions of responsibility, such as parents and teachers. It is difficult to translate adequately into English, since the keyword grcense (most frequently used in the plural grcenser) translates into border, (borderline, boundary or limit according to context. The etymon is the Polish wordgramca, i.e. 'edge' (which reached Denmark via German Grenze). At the root, there is thus a purely spatial meaning, designating the external boundary of a horizontal surface. In present-day Danish usage a social element has become salient in the understanding of the word, and the most immediate interpretation of the word would probably be the dividing-line between nations. Although purely physical applications can also be found, as in the tree-line and the snow-line which are rendered with grcense taking the place of line in the Danish compounds, you would not now speak of the grcense of a physical surface such as a field or a floor. The word thus typically designates a socially significant boundary between two areas, one of which is the 'home' area (whose 'edge' reflects the root sense), and in some cases the boundary of the home area is understood as the (natural) ending point of a trajectory starting from zero, in which case it translates into English limit(s) rather than line. The word enters into several collocations that are relevant for the case in question, the most important of which is the phrase scetle grcenser ('mark out boundaries/limits', 'draw the line'). Another context is the vioragrcensesegende ('boundary-seeking') used about children whose behaviour can be construed as designed to provoke adults to 'draw the line'. As applied to the target domain of responsibility for children, the model suggests that the freedom of scope for children can be understood as belonging in a landscape in which there are certain socially sanctioned boundaries, a home area and an outer territory that is out of bounds. The responsible adult should

190

Peter Harder

therefore mark out those boundaries clearly to the child whose behaviour shows ignorance of them or even desire to have them demonstrated. In applying the model, one is therefore immediately faced with the question of exactly where those boundaries are and how they should be marked out. What role should insistence on limits or boundaries or borderlines play, and what forms should it take? The research group did not take up the issue out of neutral scientific interest, but because the model worried them; they belonged to the progressive, antiauthoritarian movement in pedagogy and education, and wanted to understand what was happening primarily because they were sceptical about the implications of the model (for reasons similar to those invoked by Lakoff in his critique of the 'strict father model', cf. Lakoff 1996). The empirical material collected included systematic excerpts of publications, questionnaires sent out to key groups and random instances that came to the attention of the investigators. They did not look in vain for cases to justify their scepticism: the data include numerous examples of problematic consequences of the way this term is used (such as non-assignment of custody to a father who was judged not know how to draw the line/impose limits). Nor was their worry about the impact of the model quantitatively unjustified: in the case of the questionnaire investigation, 96 per cent of the respondents used the terminology associated with it. Neither the model itself nor the frontiers of the battle appeared on the scene out of nowhere. Ordinary everyday experience has always made it necessary for individuals to decide how much they are going to accept in the case of activities that impinge on them: 'you gotta draw the line somewhere' is a well-established commonsense maxim. And in relation to children, the discussion conjures up a well-trodden battle ground between a progressive (liberal) and authoritarian (conservative) orientation with respect to norms of behaviour and educational principles. Most recently, the 'permissive' climate associated with the sixties and the conflicts that have followed in its wake have drawn up the lines of confrontation that essentially reappear in the 'grsenser' argument in Denmark. To understand the specific relevance of the model, one must consider the historical development that led up to its success, including such things as increasing affluence and increasing parental absence from home, removing older 'natural' limits/boundaries around children. This development also led to anomie in conflict situations: while previously conditions in the home that constituted the child's natural world imposed quasi-natural limits on what children could do, in the modern situation where children's world is shaped by multiple forces from outside the home environment, it is no longer clear if, when or how adult caregivers are entitled to lay down the law. The most obvious facet of this historical change in Denmark is the radically increased institutionalization

Drawing the line: A contested conceptual model in Danish 'child care talk'

191

of children's lives: most Danish children spend the major part of the day away from home, since the typical pattern is for both parents to work. When traditional authoritarianism started to lose ground, most people would now agree that a number of desirable and healthy consequences ensued for children. They were no longer subjected to an arbitrary code of conduct that expected them to stay silent at meals, or be submissive towards adults in all cases, and so forth. On the other hand, as pointed out by Aries (as quoted in Sigsgaard and Vanning 1996), the process that undermined parental authority also meant a loss of identity for children, because they were no longer a natural part of a world in which adults lived and worked. Children's situation was thus both freer and more problematic. Both aspects play a role for how to understand the role of the 'grasnse'metaphor. A rough outline of mental history in Denmark would be to say that traditional forms of authority remained strong in the fifties and early sixties, but were challenged in the cultural climate associated with the youth or student revolt in 1968 - and lost the argument. For a while the direction of the attitudinal trend was to expose and dismantle traditional forms of authority wherever they could be found, and the accepted culture of institutions populated by children and teenagers changed radically as a result. One linguistic consequence was that the polite address form disappeared entirely from such institutions, where the norm until then had reflected the asymmetric 'power' semantics (cf Brown and Oilman 1960): the children said V and were given T in return. I was among the last years of school children to do this. I graduated from 'gymnasium' in 1967; my sister who graduated in 1971 already used T with all her teachers. In that climate it became abhorrent to parents and teachers to see themselves as exerting authority in restraint of the natural dispositions of children. The ideal teacher or parent would never need to use any form of power, but would always take the inclinations of children and students into account in such a way that a harmonious process of participatory decision-making would reign in home, kindergarten and classroom. This development ran its course in the generation after 1968; but then around 1990 the 68 generation began to be ripe for debunking. The spoilt children of the sixties were now in positions of wealth and power, and times were less optimistic; also, when the wall fell, the anti-Americanism of the 1968 generation and their glorification of the idea of revolutionary socialism seemed an additional reason to consign the sixties generation to the wastebasket of history. There was a sense that it was time to look critically at the values of the sixties generation, including permissiveness. In this cultural climate confrontations between adult plans and children who resist them began to appear in a new light. The GM in this sense was an idea whose time had come: once again it became possible to

192

Peter Harder

assert adult standards in opposition to children's wishes without feeling like a neanderthal. Therefore the model seemed attractive to a considerable number of parents and teachers in the eighties (including myself). After this rough outline of the context in which the metaphor operated and still operates, the point below will be to demonstrate how the understanding of the structure associated with the source domain is subjected to the forces of the discourse process that takes its point of departure in the context outlined.

2.

The battle of mappings and the discourse process

There are several interesting features in this situation. There is, first, the dual grounding: the source domain and its inherent structure on the one hand, the discourse grounding on the other. Then, within the area of discourse grounding, there is the confrontation between different bids for how to map it in context. The indeterminacy left open by this situation is another problematic feature: Can the GM be useful if it is surrounded by so much controversy? Note that this is not because the GM is especially vague. To take a favourite example, the 'love is a journey' metaphor does not make it clear how to interpret the goal element of the 'journey' (Lakoff and Johnson 1999 talk about the agreed purpose of the couple as constituting the goal, but the existence of agreed goals does not really make them the purpose of the love relationship itself, which might just as plausibly be seen as an end in itself, at least nowadays; the subject comes up in the Magic Flute in the duet between Pamina and Papageno; as seen by Pamina, the goal is reaching for the divine, but (already then) Papageno does not appear to be entirely convinced.) And finally there is the question of whether bodily grounding can return as a way to normatively steer the concept into a preferable mapping that will make it useful as a communicative and conceptual tool in getting a shared and healthy understanding of how to act responsibly in relation to children for whom one is responsible. In order to understand the role of the discourse processes, it is necessary to look at the inherent potential of the source domain. There can be little doubt that the GM by virtue of its source domain, whether physical or social, suggests constraint; a version of the Danish proverb about 'drawing the line' still uses the physical term 'edge' rather than the social term 'borderline'. The model indicates where something ends, and the metaphor could never be used to conjure up a vision of unrestricted options, whatever else one might disagree about. The question is, then - how does one apply this way of thinking to the target domain of the appropriate freedom of scope of children? Here the discourse pressures come in. In ordinary collaborative discourse, there would be a proc-

Drawing the line: A contested conceptual model in Danish 'child care talk'

193

ess of 'standard pragmatic enrichment' that would mediate between the utterance and the context of situation. Few people would disagree that any concrete context would open a range of plausible ways of interpreting where the line goes, with scope for disagreement between interlocutors, but subject to normal social pressure to mediate the differences. However, in spite of what Gricean maxims might suggest, there is no guarantee built into linguistic interaction that there are 'mutually accepted purposes' involved in a communicative exchange. Collaboration is only inherent in the whole idea of linguistic communication in so far as the recognition of communicative intentions is concerned, not when it comes to the realization of the communicative intentions. In some forms of communication, each party devotes all its energy to preventing the communicative intention of the other party from succeeding. This, for instance is true in the case of political arguments between opposed points of view. There is nothing to prevent spokesmen for opposing viewpoints from refusing mediation of differences, and from refusing collaboration when it comes to agreeing upon the most reasonable interpretation of what the other party is saying. If the contested issue is seen in terms of a binary choice between two opposite directions, attempts to mediate would be tantamount to giving up your own communicative purpose. This type of discourse grounding of conceptual models introduces an extra dimension in the understanding of utterances drawing upon the model, because the mapping process itself becomes a battle ground. The model, arising in the historical context described above, situated itself in a macro-context of educational discourse with political implications. In terms of the struggle to give children as individuals, pupils and students greater opportunities rather than constraints, as conceived by progressive educationalists such as those constituting the research group, a conceptual model licensing and encouraging the imposition of constraints upon children was first and foremost bad, whatever else it might be. At this point I am arguing that this is in a legitimate sense the only possible way of seeing the model, granted the discourse context in which spokesmen for progressive educationalists find themselves. Their cause can hardly be furthered by finding the best possible way of mapping a landscape of natural boundaries or edges onto the target domain of the scope of children. So they do not try to do this. If the only factor in understanding a metaphor were the inherent potential of the source domain for being mapped upon the target domain, such problems of understanding would be inexplicable.1 The main feature of the discourse context is its binary, polarized nature. The subtitle of all three books (Sigsgaard and Vanning 1996; Kragh-Muller 1997;

194

Peter Harder

Sigsgaard et al. 1998) is the unmitigated alternative gramer eller ej? ('boundaries or not?'). This clear-cut dichotomy of options sets the stage for the battle of mappings. A collaborative interpretation of the model (as used by an interlocutor) would aim to find a mapping that made sense in terms of her discourse purposes. The point would be to look for what the home area was and where Out-of-bounds' begins such that children were better off inside than outside the home area in terms of the current problem: should John junior be allowed to play with daddy's hand-gun, for instance, or where do we draw the line? However, if it is decided by the way the discourse battle lines have already been drawn that the model is bad, it follows that the mapping process should not even be allowed to occur. Yet in order to discuss the (de)merits of the model, it is necessary to argue from an assumed mapping, so there has to be a mapping to talk about. However, from the point of view of the research group, the less attractive a mapping, the better in terms of their discourse situation. Although the empirical material exemplifies a number of different ways of mapping the GM onto both children's and adult's lives, the possibility of understanding it as reflecting a harsh and authoritarian attitude is kept firmly in view: in the first of the three major publications of the project (Sigsgaard and Vanning 1996), grcensescetning ('drawing the line') is mentioned twice in chapter one (before the empirical material is discussed), in both cases in a compound NP, conjoined with 'physical punishment'. To understand the discourse context, this strategy must be confronted with the other side of the argument, those who support the model. They can be divided into two parties. The first group are the ideal opponents as seen from the point of view of the research group, namely those who are actually in favour of resurrecting a rigid and control-oriented way of dealing with children. They are those for whom the GM corresponds to the familiar but obsolete maxim, 'spare the rod and spoil the child'. When a doctor in the data invokes the GM in a situation where he is irritated that a child is struggling to avoid getting stabbed with the syringe, what he means is clearly: control this kid, so I can get my job done! It has nothing to do with what is good for the child, only what is good for the adult. The second group consist of those who believe that the model is useful because it highlights the fact that there are cases where you have to stop children doing what they are doing, not (only) for the adult's sake but also for the sake of the children themselves. And they create a more tricky discourse situation for the research group's purposes. In such a binary, confrontational discourse, where there is a public that is the secondary target for both parties, each party has a goal that involves the frustration of the other party's goal. With a mixed metaphor, the aim is to occupy

Drawing the line: A contested conceptual model in Danish 'child care talk'

195

the moral high ground while placing the other party in a check-mate position. Public argument may not quite be war, but it certainly is maneuvering around strategically in a mined territory. In terms of this picture, each discourse move is made based on considerations of how best to place opponent and self closer to check-mate vs. moral high ground, respectively. Each party has a mapping of the GM plus a model of what the alternative is, and each wants to establish their own model as the high ground and place the opponents in the slough of their dystopian alternative. A quick guide to the minefield might be the following. The mapping of the GM opponents: The metaphor is based on artificial boundaries like the iron curtain: marking out boundaries means 'authoritarianism and oppression of children's needs' (cf. 'trespassers will be prosecuted'). For GM opponents the metaphor licences adult oppression: the question is not how to restrict children's domain of activity but how to give them options enough, including the chance to explore life on their own terms. The mapping of the (non-authoritarian) GM supporters: The metaphor is based on the reality of natural boundaries as part of the world, including those between 'buffer zones' of parties involved in shared projects, and the boundaries between safe and wholesome vs. dangerous ground; marking out boundaries means 'helping children to stay in the right area'. For GM supporters the metaphor stresses adult responsibility: leaving the children to move off in any direction they might be heading in is a way of exposing the children to physical risks as well as to disharmonious relations with the world around them. The two check-mate positions: (1) For opponents: to be put in the position of allowing children to create chaos around them and get hurt in the process (2) For supporters: to be put in the position of claiming that prohibition is better for children than permission The two winning positions: (1) For opponents: to stand as the party who want to give children the scope they need (2) For supporters: to stand as the party who ensure that children develop a harmonious relationship with the world including its boundaries

196

Peter Harder

In the appendix, there is a transcription of a television interview with a member of the research group. It would have been better from a discourse point of view with a dialogue between an opponent and a supporter of the GM, but the interviewer assumes enough of the role of supporter to make the pattern recognizable. After the presentation of the interviewee, the interviewer gives a (possibly invented) case from his own experience, his child butting his head into daddy's behind during the cooking. Is that ok or not? The interviewee responds by re-conceptualizing the confrontation as being really about the child's natural desire to be with his (too busy) father - so that imposing the GM by stopping the child is really a case of depriving the child of love and care and imposing authoritarian constraint and punishment instead, cf. checkmate position (2). Towards the end, when asked about the alternative to the GM, he gives an anecdote of a child bringing his own money on a kindergarten hike, wanting to buy cakes for everybody, where the GM would mean suspicion and rejection of the proposal and the alternative is to let him buy the cake (which happened in the actual case, and everybody had a good time, cf winning position (1). Ignoring for the moment that the opposition offered by the interviewer is something less than vigorous, the first conclusion here is that the discourse process has produced an ending in which no attractive and useful mapping is discernible between source and target. Apart from indicating that there might be a damaging and repulsive mapping, the lack of a better one is also part of the discourse strategy. The GM, according to the position of the research group, not only licenses wrong inferences, it also does not make it clear precisely what inferences are intended; the word is vague and useless. The closest one comes to the positive potential is when it is said that some adviser who recommends the GM might mean that the parent should spend some more time with the child, but since it could equally well suggest to the parent that he should do 'something desperate', the word should never be used. Exit the GM, knocked out by discourse.

3.

'Just say no': can we avoid conceptualizing?

It is not a point here to discuss what the most reasonable attitude to the GM is (although it will be fairly obvious what the author's views are); when I now look at the other side of the argument, it should be seen as a further exploration of the field of forces in which the metaphor must be understood. What supporters of the GM focus on are cases where children's activities or desires are in conflict with concerns, attitudes and priorities of the world around them. A real-

Drawing the line: A contested conceptual model in Danish 'child care talk'

197

world example was a child who refused to eat whatever breakfast cereal was already available, with the result that the kind parents had to go and buy a new packet every morning. When they reached the forty-second brand they came to the family therapist and asked for advice. A natural question might be why they needed a therapist to tell them not to buy any more cereals. But the fact that they did reveals something about the field of forces around the role of 'caregiving adult' at this time in history, cf. above: the felt pressure to let children have their way is strong and is not counterbalanced in any culturally uncontroversial way by competing pressures. The research group naturally do not suggest that children should always have their way. What they say instead is something that is interesting from the point of view of cognitive linguistics and cognitive science generally. A recurring answer when this issue comes up is that one should avoid generalizing about such cases. In the interview, Kim Rasmussen parallels the GM with an abstract word like quality and says that such words are dangerous because they sound natural but do not translate in any precise way into real definable action. When Sigsgaard (1998) replies to an objection from the point of view of a psychologist who criticizes the alternative to a world in which the GM in some form applies, he replies that the research group did not want to think in terms of an alternative model at all. And this reflects the overall approach. In the concluding remarks of Sigsgaard and Vanning (1996: 172), the first of the three official publications of the project, the authors say The trivial, ordinary and everyday thing, to say 'no' or 'this far and no further', or 'don't' or 'get down from that tree', or Ί don't want to play any more', or 'don't touch me in that place' gets transformed into only one code, which may occur in a few subsidiary forms. The will, the tree, the football and the place where I don't want to be touched, concrete, personal and felt things are replaced by general, psychological and abstract categories.

From a cognitive linguistic point of view this statement is reminiscent of the critique of logical, extensional semantics referring to abstract categories rather than conceptual models grounded in bodily experience. But clearly even conceptual models grounded in experience can be applied in different ways to their target domain; the structure that is transferred needs to be abstracted out of the source domain and concretized into the target domain. The more 'primary' a metaphor is, cf. Grady (1997), the more the mapping is tied to recurrent experience; but unless there is an element of deconflation (cf. Johnson 1997) with the abstraction that involves, there is no mapping between fields at all, only a complex experiential whole.

198

Peter Harder

From that point of view, the position of the research group with respect to those cases that supporters of the GM would put forward can be paraphrased as follows: there are cases where it is right to say no to what children want and stop what they are doing. But any model that generalizes away from the concrete (the will, tree, football and place, cf. above), is said to involve a dangerous abstraction. This raises an important question of principle, not only for understanding based on metaphor, but for all forms of understanding. When Kim Rasmussen in the interview says that the word grcense lacks a core meaning, his rejection of it would affect a 'classical' as well as a metaphorical understanding. But can we understand a concrete everyday problem when we are in doubt without seeing the concrete case in the context of other cases? Is there a way to avoid the difficult question of how to apply properties gathered elsewhere to the case at hand (category membership in the classical case, source domain structure in the metaphorical case)? Although I suggest that the answer is no, the issue is obviously too fundamental to be satisfactorily covered on the level of principle within the scope of this chapter. There are two ways in which I believe the need to go beyond the concrete can be shown to be in fact inevitable, however. One is the difference between cases where the need to say no manifests itself obviously and naturally as part of the concrete event. In the ideal case, this is the best possible outcome; but if that was the way things worked, the GM would never have had its phenomenal success, because there would be no need for it. The success is only understandable in the context of a situation where concrete cases give rise to dilemmas, since solutions do not present themselves concretely in the situation. The parents with 41 cereals in the cupboard did as a matter of fact not simply say no, but needed advice. If they were told concretely to say no, they would still need advice the next time round. Secondly, since the topic in this context is the discourse process, there is the role of justification in communication. Habermas (1971) suggests that human communication is bound up with normative principles, one of which is that acts and assumptions can be challenged in terms of the need for justification. As parents will know, children do not have to read Habermas to be aware of the possibility of challenging adult decisions in terms of grounds for justification; and if other adults are present, the silent tribunal of peer evaluation is always in session. Once personal conviction and culturally uncontroversial grounds for saying no are lacking, adults subjectively and communicatively need to look beyond the concrete event for justification. The ban on abstraction, therefore, must be understood as a discourse move; it cannot be understood as a tenable universal principle of understanding or communication.

Drawing the line: A contested conceptual model in Danish 'child care talk'

4.

199

Bodily grounding as a normative criterion: the Other as boundary

The outcome of this battle of mappings is still uncertain. The GM remains popular in everyday parlance, but members of the research group express satisfaction that professionals have become more reluctant to invoke the GM in recommending how to deal with children. Assuming this is true, it rounds off the argument that I have presented to show the importance of discourse factors in metaphorical mapping: discourse pressure has reinforced the awareness of a dangerous mapping, in terms of which the GM would resurrect harsh authoritarianism, and it has impeded the use of alternative mappings as a source of understanding the target domain, and finally it has created a partial void of models, a void which is argued to be beneficial: thou shalt make no general image of when to say no to children! From a cognitive-linguistic point of view, the role of basic bodily experience can be discussed in a new light, suggested by the advice of Jesper Juul, a family therapist who uses the GM (albeit with some caution). On an instructional video he presents the problem (including the '42 cereals' case), warns against harshness and restrictions as an answer to problems, and then comes to the way he wants to advise parents. The core of his position may be summarized as follows: what children need and want are parents who react as real people, showing who they are and what they like and do not like, simply by being themselves in interaction with their children. The point is that by behaving like that parents and other responsible adults constitute boundaries that the children come up against in everyday life. Parents, by being there and reacting to children, impose constraints on children's plans and desires. This is part of their duty as responsible adults, and has perspectives for children's general upbringing in that the respect for boundaries as part of the human world will then be grounded in recurring experiences of how interaction with others comes with the constraints arising from the need to respect the reactions of others. If this were the way things inevitably happened, we would have something like a primary metaphor in Grady's sense: just as the mapping between warmth and affection works through repeated experience of bodily contact with loved ones, so does the mapping between boundaries and harmonious relations with the world arise from repeated experiences of how the joys of human relationships and the ability to accommodate oneself to them go together. This is perhaps why defenders of the metaphor (cf., e.g., the newspaper Information Sept 5/6,1998) emphasize the naturalness of boundaries in harmonious lives, and the hazards of blindness towards these boundaries.

200

Peter Harder

But as we all know that is not inevitably the way it works. In less ideal cases, adults may feel it is their duty to suppress their own 'adult' reactions, yielding habitually to pressure from children - this is the case that conservatives cite in attacking excessive permissiveness. In other cases, parents have too little time to be with their children to give them the association between harmonious relations and the ability to respect other people's 'boundaries' - this is where progressives rightly warn against prohibition as a substitute for responsiveness. The GM as a metaphor in the Danish speech community lives and works in an environment where all these situations play a role - in addition to the discourse battle that I have described.

5.

Conclusions

Metaphorical mappings are processes occurring in a field of forces that include discourse factors as well as factors due to the source domain and to individual experience. The pressure may act to prevent as well as to support particular mappings, and understanding depends on how these forces interact in a concrete situation. One discourse factor is the nature of confrontational as opposed to collaborative forms of communication. Polarization is a common element in conflictive discourse, reducing chances of collaborative sense-making: if one party is determined to stress the destructive potential of a conceptual model, the chances of finding a useful application of it are reduced. In this case, the conflictive discourse itself increases the chances that opponents end up being right: the GM is unhelpful. The example illustrates how bodily grounding may have a more complex role than as a quasi-natural part of the world: by offering one among several possibilities of understanding a conceptual metaphor, it may re-ground the model in a conceptual and interactive context that is arguably normatively fundamental. But there is no guarantee that understanding via this grounding will be available in all discourse contexts.

Drawing the line: A contested conceptual model in Danish 'child care talk'

201

Appendix Interview with mag.art. Kim Rasmussen, collaborator on 'Gra?nser eller ej, volume 3' (cf. Sigsgaard et al. 1998) DR2 1998. I: / lobet afden uge der er gäet har vi hört adskillige historier om at b0rn i Danmark vantrives. Hver 5. harjcevnligpsykisk betinget hovedpine, hver 4. mobbes jcevnligt alvorligt, og det er blevet vcerre de sidste 14 or. Hvis sky Id er del? 'In the course of the past week we have heard several stories about children in Denmark being treated badly and not feeling well. One in five has headaches caused by mental problems, one in four are suffering from persecution in peer groups, and things have been getting worse for the past fourteen years. Who is to blame?' - Voksensamfundet 'The adult world' I: Og hvem er det sä? 'And who might that be?' - Bade den mädefamilien har indrettet sig pa, og den made arbejdslivet fungerer pä, 0get konkurrence, den slags ting og sager. Sä det er noget med den generelle samfundsudvikling. 'Both the way family life is developing, the way working life is going, increasing competition, that kind ofthing. So it's to do with general trends in society' I: Men det er jo nogle personer ...Hvem er det som star bag det i forhold til b0rnenes vantrivsel? 'But we're talking about people..Who is behind it when it comes to children who are not feeling well?' - Altsä samfundets udvikling 0h det er jo bade nogle institutioner og det er nogle relationer, og den made som Institutionen har udviklet sig pä 0h viljeg i h0j gradfremhceve. Og hvis man sp0rger b0rnene selv fremhcever de jo ogsa tit at ..at deres forceldre er fortravlede, og de oplever at der er for lidt (id til dem ogfor lidt tid til det de gerne vil. 'well societal development..er..that is both institions and relationships, and the way institutions are developing ...er... that kind ofthing is what I'd strongly emphasize. And if you ask the children themselves they often say that their parents are too busy and they feel there's not enough time for them and for the sort ofthing they'd like to do'

202

Peter Harder

I: Dvs hvis man skal sige det kort, sä er det forceldrene ogpcedagogerne. 'That is, to put it briefly, it's parents and kindergarten staff' - 0h..de har et vcesentligt ansvar 'er.. .they have a major part of the responsibil-

ity' I: Men när jeg nu fa star derhjemme og laver mad om aftenen efter en lang arbejdsdag og min s0n pa ncesten 3 or kommer for femie gang og slar hovedet ind i bagdelenpä mig, ogjeg sä siger meget klart og meget tydeligt, Hold sä op!, jeg skal have lavet mad! Sä markererjeg meget klart overfor ham: her er min grcense gäet, den har du overträdt, lad vcere med det. Er detforkert? 'But when I for instance am at home, cooking after a long working day and my son of nearly three comes for the fifth time butting his head into my behind, and I say very pointedly: Stop it! I have to get the cooking done! ..then I mark out very clearly, here I draw the line, you have stepped over it, don't do that. Is that wrong?' - Det er jo en made at forstä grcense pa, og sporgsmälet er hvis du siger til et barn: her er min grcense, om dit barn overhovedet forstär hvad du siger, ikke? Jeg vil sige barnets reaktion kunne jo ogsä fortolkes pa en helt anden made. A lisa barnets reaktion kunne jo ogsä tolkes som at han har vceret...at han vil gerne vcere sammen med dig, og han har set for lidt til dig og han har ikkefäet sit behovfor samvcer med dig opfyldt og du kunne sä drage ham ind pä andre mäder, altsä prove at gore ham delagtig i madlavningen, prove at sige at tingene ikke skulle foregä sä strikt og sä tidsbundet. 'That's one of way of understanding [graense], and the question is if you tell a child, here's my [graense], if your child even understands what you're saying, isn't it? I'd say the child's reaction could also be understood quite differently. You know it could also be interpreted as saying that he'd been... he'd like to be with you, he'd not seen enough of you and not had his need to be together with you satisfied, and you might then try to make him part of what you're doing, help you cook, try and say that things didn't have to be so strict and time-bound.' I: Men der er jo nogle ting der skal gores i enfamilie engang imellem, og sä mä man vel sige til bornene, hold sä op, hvis vi skalfungere sammen, er der en grcense her, og den mä du ikke overskride? 'But there are some things that have to be done in a family from time to time, and then I suppose you have to tell the children, stop it, if we are going to be together there is a [graense] here, don't step over it?'

Drawing the line: A contested conceptual model in Danish 'child care talk'

203

- Skal det vcere so tidsbundet? Dei vedjeg ikke. Det synesjeg kan (p) vcere et godt sporgsmal. 'Does it have to be so time-bound? I don't know. I think that might be a good question to ask.' I: Du veden masse om b0rns trivselfordi du har vceretpä det her store forskningsprojekt, 'Grcenser eller ej', der har varet 7 är, og som blevfcerdigt her i starten afugen, da sidste del afdet udkom...hvadtcenker du när du horer forceldresom mig ogforst ogfremmest pcedagoger og de professionelle ifolkeskolen gär og taler om at born har behovfor greens er? 'You know a lot about children's conditions and well-being, because you've been part of this big research project, '[Graenser], yes or no', which has been under way for seven years, and which was completed last week, when the last publication appeared.. .what do you think when parents like me and professionals in schools and institutions are talking about children needing [graenser]?' - Altsä, jeg tanker ferst ogfremmest at det er et begreb som er blevet meget naturlig at bruge, uden at man egentlig har spurgt sig kritisk hvad det er der menes med det. Og det er ogsä det grundlceggende (?) spergsmal der ligger bag projektet hvor vi sä har spurgt os selv ogforceldre, Iceger og psykologer og andre som bruger begrebet grcense: Hvad mener I egentlig med det...när vi snakker umiddelbart er det tilsyneladende sädan at viforstär hvad det er der menes med det, men sä när man her om at fä nogle konkrete eksempler osv, sä siger vi at det kan betyde alt muligt - det kan betyde nej, det kan betyde gribe ind, det kan betyde at man skal tvinge b0rn, det kan betyde at man skal bestemme over born og en lang rcekke andre ting og sager. Og hvis professionelle rädgivere siger tilforceldre at de skal scette nogle grcenserfor deres born, sä kan det vcere at de tilsyneladende forstar hinanden i en sädan samtale, men at hvis rädgiveren egentlig mente at 0h forceldrene skulle vcere noget mere sammen med deres barn, og vcere noget mere intensive i deres samvcer med bornene, sä kan det vcere at forceldrene omvendt gär hjem og i vcerstefald mäske griber fat i dem eller g0r et eller andet mere desperat. 'Well, first of all I think it's a concept that has become very natural to use without anybody asking themselves critically what it means. And that's also the basic question that's behind our project, where we've asked ourselves, parents, doctors and psychologists and others who use the concept [graense]: what do you really mean by it? When we're just talking it seems to make sense to us, but when you ask for concrete examples etc, then we say that it can mean anything - it can mean 'no', it can mean intervening, it can mean forcing children, it can mean you should lay down the law and lots

204

Peter Harder

of other things. And if professional advisers tell parents that they should set [graenser] for their children, then they may appear to understand each other in such a conversation, but if the adviser really meant that.. .er.. .the parents should spend more time with their child and be more intensive in the way they spend time with them, then there's a risk that the parents, in contrast, go home and in the worst case perhaps seize them physically or do something more desperate.' I: Sä det er altsä simpelt hen et sporgsmäl om en sprogbrug, at man bruger et modeord 'groenser' uden helt at vide hvaddet er det drejer sig om? 'So it's simply a question of the language we use, that we use a fashionable word [graenser] without quite knowing what it's all about?' -Vi har i vores tid en tendens til at bruge nogle mere abstrakte ord, hvor 'grcenser'er etjeg kan ncevne. Kvalitetsdebatten er et andet, meget abstrakt ord, hvem kan umiddelbart vcere imod kvalitet, hvem kan umiddelbart vcere imod grcenser, det synes som noget meget naturligt noget. Men i det 0jeblik vi gär ind og sporger hvad der egentlig menes med det, sä er det et meget taget ord og et meget därligt ord. 'We have, these days, a tendency to use some more abstract words, where [graenser] is one I can mention. The debate on 'quality' is another very abstract word, who can be against 'quality' on the face of it, it looks as if it's something very natural. But the moment we go in and ask what's meant by it, it's very vague word and a very bad word.' I: Men i den forste afjeresforskningsrapporter, der ligger et par är tilbage, der siger ncesfen alle, 96% trorjeg det er aj'dem I sporger, at born er grcensesogende, born har brugfor grcenser. Det er jo de professionelle, dem der moder bornene i hverdagen i deres institutioner...er det ikke sädan noget man mä tage ad notam, sädan er det äbenbart? 'But in the first of your research reports, a couple of years back, almost all, I think it was 96%, of those people you ask say that children are [graense] seeking, children need [grasnser]. They are the professionals, those who have to do with children in their everyday life in their institutions...don't you just have to accept that's the way it is?' - Nej (p) det mäjeg klart a/vise (p) Ogjeg synes at hvis der skal vcere fornuft i det, sä mä man sporge sig selv: hvordan kan mange ellers begavede mennesker komme med et sä ubegavet svar som , som grcense. 0h det viser sig at nor vi sporger de professionelle, sä er det et begreb uden en egentlig betydningskerne.

Drawing the line:

contested conceptual model in Danish 'child care talk'

205

'No ... I must definitely reject that.... And I think if we are to act sensibly, you should ask, how can many otherwise intelligent people come up with such a stupid answer as, as [graense]. Er...it turns out when we ask the professionals, it is a concept without a real core meaning.' I: Skal man aldrig sige NEJ til et barn? 'Should one never say 'no' to a child?' - 0hm (p) jeg vil ikke, jeg vil ikke stä og sige at man aldrig skal sige nej, og at man ikke kan komme udfor at skulle for holde s ig til born pa en made hvor at man vil sige at nu g0r man 0h et eller andet som man — hvad skal vi sige, altsä at del har en karakter hvor man er indgribende pa en eller anden made. Selvfolgelig kan man komme udfor del. Men jeg synes det er meget vigtigt at tcenke pa, hvad er det der er borns motiver? Det eksempel du siger med din dreng, ikke? Hvis han kommer og slär dig bagi, sä i siedet for at rea-, at agere pa rygmarven, sä prove atforstä barnet, og sä synes jeg det meget vigtige, det er - born er forskellige fra voksne pa mange punkter, men vi har ogsä noget som - hvor vi er helt ens. Og lige sä vel som vi som voksne ikke bryder os om at blive ireltesat eller blive grebet ind overfor, eller blive tvunget til noget, sädan har bornene det ogsä, og de vil fole s ig krcenket, og det synes jeg er meget vcesentligt at vcere opmcerksom pä, sä derfor so den made vi omgäs born pä, der vil vcere en kerne i det hvor jeg vil sige, der kan vi godt prove at scette os noget mere ind i hvordan at bornene defoler. Hvis vi selv var i deres sted, ville vifole det pä den samme made. '-Erm...I won't, I won't stand up and say that one should never say 'no', and that you can't find youself in a situation where you have to deal with children in such a way that now you do ..er. .something that you.. .how shall I put it, I mean that it is in the nature of something where you intervene in some way. Of course you can come across that kind of situation. But I think it's very important to keep in mind, what are children's motives? If he comes and hits you in your behind, so instead of reac-, instead of acting in a kneejerk fashion, then try to understand the child, and then I think the crucial thing is - children are different from adults in many ways, but we also have things that - where we're the same. And just as adults don't like correction or intervention from others, or being forced to do something, children are exactly the same way, and they'll feel hurt, and I think that's very important to be aware of, so that's why the way we interact with children, there's a core where I'll say we might try to put ourselves more in their place, how they feel. If we were in their place, we'd feel the same way.' I: Den 3. ogsidste del afjeresforskningsprojekt, den header 'Andre mader 'som lige er udkommet. I den beskriver I andre mader at behandle born pä, forst og

206

Peter Harder

fremmest i insütutionerne. Kan du give nogle eksempler pä hvad det er for nogle ting der skal til for at ikke bornene oplever den her grcensescetning? 'The third and last part of your research project, it's called Other ways' which has just appeared. In it you describe other ways of dealing with children, especially in institutions. Can you give me a few examples of what is needed in order that children are not met with this boundary-marking?' - Altsä vi har talrige historier, hvisjeg skulle ncevne noget fra alle de institutioner vi har, vi har fra mange forskellige institutionstyper... mean we have lots of stories, if I were to mention from all those types of institutions we have, we have from many different types of institutions...' I: Kan du give sädan et par typiske eksempler? 'Can you give us a couple of typical cases?' - Jeg kanfortcelle nogle historier - en historic ville vcere eksempelvis et barn som kommer over i bornehaven en dag og har nogle penge med, og siger til pcedagogerne: Pr0v og se her, jeg har penge med, vi skal ud og k0be kager i dag. Den historie, den kunne vcere en anledning til at pcedagogerne tcenkte 'grcenser': Hvad er det her nufor noget, hvor har han pengene fra, han skal ikke komme her og bestemme osv, ikke? Det pcedagogen g0r i siedet for, det er at prove at huske tilbage, at engang var de pä tur og gikforbi en bager hvor pcedagogen sagde at det ser Icekkert ud det derinde. Og nu har drengen sä selv taget nogle penge med ogfaktisk en plan for noget de kan gore den pägceldende dag, noget der var meningsfuldt for barnet, ikke? De gik sä hen og k0bte kagerne og spiste - de k0bte et par stykker og spiste nogle i bornehavne og tog nogle med hjem til far og mor ogsä, og havde det smaddergodt. Det er jo en helt anden made at mode barnet pä, med tillid, i siedet for med kontrol og mistro. can tell some stories - one story would be for instance a child coming over to the kindergarten one day who's brought along some money and says to the kindergarten staff: look, I've got some money, let's buy cakes today. That story could be an occasion to think in terms of [granser]: what's all this, where did he get that money, he's not going to decide what we're doing, etc. But what they do instead is to think back, once they were on an outing and passed a baker's shop where he said, 'that looks good in there'. And now the boy has brought some money and has a plan for something they can do, something that makes sense to the boy, right? So they went along and bought the cakes and ate them, brought home a couple for mum and dad as well and had a wonderfuld time. That's quite a different way of responding to the boy, trusting him, rather than in terms of control and suspicion.'

Drawing the line: A contested conceptual model in Danish 'childcare talk'

207

I: Dei er altsä s impel t hen noget - del er en anden made at tcenkepa. Del er ikke et sporgsmäl om at revolutionere hele maden at vcere institution pä? 'So it's simply something...it's a different way of thinking. It's not a question of a revolution in the whole way of being an institution?' - Det er ikke noget med at revolutionere. Der er ikke noget med fundamentalisme i det her. 'There's nothing revolutionary about it. There is no sort of fundamentalism in this'

Note I should add that I think the research project is extremely valuable, and that the explicit statement of its values is a great merit, absolving them of all charges against presenting a biased account rather than the objective truth. There is also a great deal of interest in the publications regardless of one's position on the issue. But I think the value orientation leaves little scope for intermediate nuances.

References Brown, R. and A. Oilman 1960 The pronouns of power and solidarity. In: Thomas A. Sebeok (ed.), Style in Language, 253-276. Cambridge, MA: MIT Press. Grady, Joseph 1997 Foundations of Meaning: Primary Metaphors and Primary Scenes. Ph.D. dissertation, University of California, Berkeley. Habermas, Jürgen 1971 Vorbereitende Bemerkungen zu einer Theorie der kommunikativen Kompetenz. In: Jürgen Habermas and Niklas Luhmann (eds.) Theorie der Gesellschaft oder Sozialtechnologie, 101-141. Frankfurt am Main: Suhrkamp. Harder, Peter 1999 Partial autonomy: Ontology and methodology in Cognitive Linguistics. In: Theo Janssen and Gisela Redeker (eds.), Cognitive Linguistics: Foundations, Scope and Methodology, 195-222. (Cognitive Linguistics Research 15.) Berlin / New York: Mouton de Gruyter. Theo Janssen and Gisela Redeker (eds.) 1999 Cognitive Linguistics: Foundations, Scope and Methodology. (Cognitive Linguistics Research 15.) Berlin / New York: Mouton de Gruyter.

208

Peter Harder

Johnson, Christopher 1997 Metaphor vs. conflation in the acquisition of polysemy: The case of SEE. In: M.K.Hiraga, C.Sinha and S.Wilcox (eds.), Cultural, Typological and Psychological Issues in Cognitive Linguistics. (Current Issues in Linguistic Theory 152.). Amsterdam / Philadelphia: John Benjamins. Juul, Jesper No date At scette groenser, Scandinavian Kempler Institute (video). Kragh-Müller, Grethe 1997 B0rneliv og opdragelse: Grcenser eller ej? Volume 2. Copenhagen: Hans Reitzels Forlag. Lakoff, George 1996 Moral Politics: What Conservatives Know that Liberals Don't. Chicago: University of Chicago Press. Lakoff, George and Mark Johnson 1999 Philosophy in the Flesh: The Embodied Mind and its Challenge to Western Thought. New York: Basic Books. Sigsgaard, Erik 1998 De professionelle og grasnseterminologien. Dansk Pcedagogisk Tidsskrift 1:56-60. Sigsgaard, Erik and Ole Vanning 1996 Voksnes syn pa b0rn og opdragelse: Grcenser eller ej? Volume 1. Copenhagen: Hans Reitzels Forlag. Sigsgaard, Erik, Kim Rasmussen and S0ren Smidt 1998 Andre mader: Grcenser eller ej? Volume 3. Copenhagen: Hans Reitzels Forlag. Sinha, Chris 1999 Grounding, mapping, and acts of meaning. In: Theo Janssen and Gisela Redeker (eds.), Cognitive Linguistics: Foundations, Scope and Methodology, 223-225. (Cognitive Linguistics Research 15.) Berlin / New York: Mouton de Gruyter. Sweetser, Eve 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Talmy, Leonard 1988 Force dynamics in language and cognition. Cognitive Science 2: 49-100.

Index

aboutness 51,75 accessibility 8, 49, 84, 95, 102 accessibility theory 10, 33^45, 49, 86 activation 4-5, 9-10, 14-15, 23-24, 26-29, 55, 69, 80-81, 83-86, 89, 93, 95, 142, 177 adverbial placement 11, 79-95, 151, 153-154 affect 10, 11,51,63,79,92-95, 112 affectedness 101, 103 affix(ation) 33-34, 36, 39, 47n affix order 39 agentivity 22, 44, 46n, 101-102, 104, 110, 188 see also quale, agentive agreement marker 10, 33^5,46n subject 33-38, 40-44, 46-47n object 42^3, 46-47n amalgamation 181, 183 analogy 9, 107, 163-165, 178, 180-181 anticipation 81-82, 84, 95 see also expectation argumentation 6, 12-14, 53, 59, 91-92, 94, 149, 152, 157, 161, 163-165, 168, 172,187-188, 190-181, 193, 196, 199 aspect 18, 39^1, 100-101, 103, 107-108, 110-111, 114 background(ing) see grounding basic unit 18-19, 23-24, 27, 177-179, 183 behaviourism 17-18, 20 bilingualism 8, 10-11, 49-64, 67, 69-70 business talk 13, 149, 157 categorization 5-7, 27, 110, 123-125, 127-130, 133-134, 137, 139, 143, 149, 169,187, 197-198

child care talk 14, 187-188 clause status 67, 101, 104, 108-109 code-switching 11, 50, 53-55, 58-61, 65, 67-70 cognition 1-15,17-24, 27-29, 49, 70, 93, 95,109,123,128,143,149, 169 cognitive 1^, 6-15, 17-20, 23-24, 26-27, 49, 52, 55, 63, 65, 79-81, 83, 89,94-95,99, 123, 149, 151-153, 156, 159, 169, 177, 185 cognitive linguistics 1-4, 7-10, 12, 15, 17-18, 20, 24, 27, 49, 55, 79, 95, 124125, 128, 149, 156, 177-178, 187, 197 cognitive process(ing) 3, 8-9, 11-12, 18, 21, 23-24, 26-27, 29, 52, 63, 80, 82, 84-86, 91, 93, 95, 103, 112, 123, 149, 152-153, 156, 159, 187-188, 193-194, 200 parallel distributed processing 23 cognitive structure/structuring see structure, structuring coherence 2, 8, 10-12, 49, 61, 82-83, 85,92,94, 109-110, 125, 149, 154, 159-160, 166-167 cohesion 59, 69, 160, 167 comment 81, 113 see also focus, rheme commitment 151-152, 154, 156-159, 161, 166,169,171-173 common ground 9, 15, 54, 91, 93, 154, 159,171 communicative fragment 8, 13-14, 177-185 communicative goal/purpose 2, 4, 13, 81, 93, 128-129, 139, 149, 158, 162, 193-194 see also intention

210

Index

conceptual model 8-9, 12-14, 134, 139, 187-189,192-193, 196,200 see also mental model conceptualization 4, 8, 15, 23-24, 87, 113, 154-155, 164, 188, 196,200 conditional 12-14, 93, 149-172 conditional model 126-129 see also mental model conditional relation contingent 159-160, 162, 165, 169-172 directive 159, 161-163, 170 reasoning 159-161, 163, 165-167, 169,172 consciousness 4, 24, 178-180 construction-integration model 23, 26, 28-29 contamination 179-180 collocation 10, 24-27, 29, 177, 189 context 1-15, 19, 23-24, 38, 52, 55, 6061, 63, 67, 80-84, 90, 93, 95, 100, 104, 113, 115,124,129,149-150, 153-155, 160-161, 164-165, 171, 177-179, 182184,189-194,198,200 situational 2-6, 8, 10, 13-15, 24, 52, 54, 60, 63, 67, 80-81, 83, 90, 93, 95, 113,115,125, 129, 149-150, 153-155, 160-161, 164-165, 178-180, 182-185, 192-194, 198,200 socio-cultural 3-6, 8-9, 13-15, 61, 63, 93, 95, 153, 155, 163-165, 178-179, 182-184,189-192 textual (co-text) 2, 4, 8, 10, 14, 82-84, 90,95, 100, 104, 113, 115, 155, 160-161, 164, 177-179, 182-184, 189, 192-194,200 see also conceptualization, mental representation contract 13, 125, 127-128, 130-142, 144, 145 counterfactuals 151, 156-157, 165, 168-169, 172 cross-linguistic data 8, 10, 33-34, 36—44, 49, 51, 52, 54, 102-104, 110, 112, 114, 117

Danish 14-15, 17, 132, 144, 145n, 187-189 detachment 33^0, 42, 44, 46n, 51, 53, 68,155 left-detachment 34-35, 53, 68, 155 NP-detachment 33-^0, 42, 44, 46n right-detachment 34-35, 51, 53, 68 see also mneme determinacy 9, 17-21, 51, 89, 94, 100, 126, 188 dialectics, dialectical 5, 9, 11, 20-21, 24 discourse 1-15, 17, 19-24, 27, 33-34, 36, 43, 49-50, 53-54, 60-61, 65, 67-69, 79-83,93-95,99, 103-104, 109, 111112, 123, 126, 143, 149-157, 159-162, 164-165, 170-172, 177-178, 182, 185, 187-188, 190-196, 198-200 collaborative discourse 192-194, 200 confrontational discourse 15, 188, 190192, 194, 196,200 literary discourse 14, 177, 181-185 see also text discourse strategy see strategy discourse type see text type domain 15, 123-125, 129, 132, 134, 136137, 139, 187-189, 192-193, 197-200 source domain 15, 187-188, 192-193, 197-198,200 target domain 15, 187-189, 192-193, 197,199 endangered language 58 English 11, 25-26, 29, 57, 60-61, 65-66, 102-103, 108, 132, 134, 153, 189 event 10, 14, 17, 22, 24, 85-86, 88, 99101, 103, 108, 110-111, 113, 115-116, 142, 181 expectation 2-3, 28, 80-81, 83-85, 91, 93-95,99, 104, 164-165, 172 see also anticipation

Index experience 5, 8, 15, 20, 65, 101, 111, 178179,185, 187-190, 196-197, 199-200 see also iconicity, experiential expert 13, 123-142, 144, 145n see also non-expert

211

iconicity 2, 8, 11-12, 53, 80, 86-90, 94, 112-117 experiential iconicity 2, 86-90, 94, 112-113, 115 impromptu speech 2, 4, 8, 10-11, 52, 56, 67-70,81, 169, 182-183, 185 indeterminacy 3-4, 192 face 149, 153, 156-158, 160-164, 167inferencing 2, 4-5, 11, 80, 84, 95, 151, 168,170 153, 163-164, 166, 168, 188, 196 female speakers 11, 64-66 see also information, inferrable figure 2, 7-8, 11-12,24,27,81, 105-111, information 2, 8, 10-12, 60, 69, 80-81, 117 83-87,89-95, 101-103, 113, 115-116, see also ground 151-156, 161, 163-167, 169, 171-172, Finnish 10-11,45,49-53,55-58,60-61, 185 64-68,70 crucial 80-81, 89-91, 93, 95 focus 24, 27, 51, 80-81, 95, 153 given 11, 80-82, 84-85, 91-93, 95, 101see also comment, rheme 103,113,115-116, 151-155, 163-166, force dynamics 113-114, 180, 187-188, 171-172, 178-179 192,196-197,200 implicit 2, 52, 87, 89, 151, 161, 167 foreground(ing) inferrable 11, 80, 84-85, 90-91, 101, 155 see grounding new 11, 80-81, 84-86, 90, 93, 101-103, French 49, 51-53, 67, 103 113,115-116, 153-155, 172 unused 11,84-85, 101 genre 5-8, 55, 94-95, 179, 182,185 information status 60, 67-68, 83-86, 94, grammaticalization 10, 37-38,40,46n, 101-103, 155 52, 102, 149, 156, 159, 170 information structure, structuring 2, 8, ground 2,8,11-12, 24, 27, 81, 105-111 10-12, 49-50, 63-65, 67, 69-70, 80-81, see also figure 83-86,89-95, 151, 153 grounding 12,99-117, 187-188, 192, see also thematic strategy; theme, 197,199-200 rheme, mneme; topic, comment; focus background(ing) 12, 91, 93, 95, 99-112, individuation 101-102, 107, 113-115 115, 149-150, 153, 157, 159, 161, 167, instructive text 80-81, 83, 87-89 171-172,181,185 intention 3,4, 21, 60-62, 69, 116, 192 bodily grounding 12, 15, 187-188, 192, see also communicative goal/purpose 197, 199-200 interlanguage 50, 55 discourse grounding 12, 15, 192-193 interpretation 3-5, 7-8, 11-12, 14, 38, 91, dual grounding 192 92-93,95, 100, 105-106, 112, 116, 159, foreground(ing) 8, 10-13, 81, 93, 95, 161, 177-178, 181, 185, 193-194 99-117, 149-150, 157-159, 161, interpreting, simultaneous 64, 69 164-165, 172 interdiscursivity 6, 8, 94-95 grounding criteria 12, 100-104, 107, intertextuality 2, 6, 8, 94-95 113, 116 informational foregrounding 11-12, 81, judgement 13, 127-128, 130-142, 144, 95,153 145n

212

Index

knowledge 6, 8-9, 12-13, 17-18, 20-23, 27-28, 49, 51, 60-61, 65, 80, 90, 123130, 133-137, 139-143, 152-153, 155, 163,177 declaration knowledge 65 encyclopedic/world knowledge 80, 90, 185 legal knowledge 12-13, 123-127, 134137, 140, 142 procedure knowledge 55 shared knowledge 6, 9-10, 49, 51, 60-61, 152-153, 155, 163 socio-cultural knowledge 9, 153, 155, 163 specialised knowledge 9, 13, 123-125, 133-142 see also structure, knowledge language 1, 3, 5-10, 13-14, 17-24, 27, 55,57,107,110,114, 123,143,177178,185 linearization 77, 85-86, 90, 95, 114, 182, 184 mapping 13, 15, 149, 156-158, 163-164, 170-172, 187-188, 192-200 matrix language 54, 60 meaning 2, 7-10, 12, 14, 17-28,49-53, 56, 58, 60-61, 63-64, 69-70, 79, 95, 128-129, 150, 189, 198 memory 4, 6, 8, 10-12, 14, 19, 34, 60, 62-65, 67, 81, 95, 177-178,180, 184-185 quantitative memory 11, 63-65, 67 male speakers 11, 64-65 mental model 124, 126-129, 134, 136, 139, 142, 149, 155, 160 see also conceptual model, conditional model mental representation 4,7, 17, 20, 23, 35, 64-65,83,100,179, 185,188 mental space see space

mental structure see structure mentalism 17-20 metalanguage; metadiscourse, metatext 4-5, 11, 14, 18,62,67,93-95, 159, 162, 166, 170, 183 metaphor 12, 15, 24, 26, 107, 110, 157, 187-189, 191-200 methods 2, 13, 15, 19, 24, 29, 50, 52, 55, 116, 123-125, 127-130, 132-133, 140, 142-143 experimental methods 13, 56, 123-125, 127-130, 132-133, 140 corpus studies 8, 10, 24-27, 29, 51-53, 55-57, 64, 67, 149, 151, 158, 162, 164, 172, 177 typological, cross-language studies see cross-linguistic data mneme 10-11,51-53, 67-69 see also right-detachment narrative 6-8, 10-12, 52, 56, 58, 60, 62, 79-80, 82-83, 85-88, 94, 99-104, 107117, 142, 178, 181-185 norm 15, 33, 44, 99, 190-192, 197, 199 non-expert 13, 123-125, 127, 130-142, 144, 145n see also expert orality 49, 52 organization see structure, structuring perception 12-14, 22-23, 65, 70, 99, 101, 105-110,112, 114-117, 130, 135-140, 142 see also structuring, perceptual person agreement see agreement marker person asymmetries 33^4 person paradigms 10, 33-34, 40, 43 perspectivization, point of view 2, 91, 92-94, 115, 151, 167-168, 182-184, 194

Index persuasions, 11,91-93,95 polarity 150, 172 prefabricated pattern see communicative fragment presupposition 60, 151-153, 160 product vs process 2-4, 12, 15, 23-24, 27, 188 see also cognitive processing pronoun 10, 28, 33^0, 42-44,46-47n, 69,101,102,116,153 bound pronoun 35^0, 42-44, 46-47n prototype 5-6, 8, 12, 35, 64, 79, 80, 102, 110-111, 150 quale,qualia 13, 124, 128-130, 133-140, 142-143 agentive quale 129-130, 133-140, 142 see also agentivity form quale 129-130, 133-138, 140 telic quale 129-130, 133-140, 142 see also telicity quantitative memory see memory recognition 8, 12-14, 178, 180, 182, 185, 193 reference 10, 34-44, 46n, 52, 69, 83-84, 86, 101-102,107-108,111, 114, 116, 154, 161-162, 165, 167, 179, 182-183, 185,188 relevance 14,23, 149-152, 154, 159-160, 163-164, 167, 169, 171, 185 rheme 10-12, 50-53, 59-60, 67-68, 81, 113,154 see also comment, focus Russian 14, 103, 179-184 salience/saliency 2, 8, 12, 34—35, 90, 93-95,99-100, 102-104, 110-113, 115-116, 187, 189 Sami 11, 50-52, 55, 64-65, 67 sentence-initial position 11-13, 79-95, 151,153-154

213

scalar model 11-13, 35, 43, 69, 84-86, 100-103,111-113, 149, 151, 156, 158, 163, 165-166, 169-172 space mental space 8-9, 12-13, 49, 149-153, 155-172 shared space 149, 151-153, 155-159, 161-164, 166, 169, 171-172 provisionally shared space 149-151, 157-159, 161 stake 13, 149-150, 153-154, 156, 159161, 163-164, 167-168,170 stance 92, 149, 152, 156-157, 166, 168 epistemic stance 152, 156-157, 166, 168 story see narrative story-line 99, 101, 103, 108-110, 112117,179 strategy communicative strategy 13-14, 54, 59 text/discourse strategy 2, 15,49, 52-53, 55,59-60, 80-90, 92-95, 149, 152, 154-155, 158, 160, 162, 169, 194-196 locative 83-89 temporal 11, 82-89, 94, 109-110 see also temporal sequentiality see also thematic strategy structure 2-3, 8, 13, 15, 17, 20, 22-23, 89,93,99, 104-106, 109-117, 123-130, 134-138, 140, 142, 177, 187-188, 192, 197-198 see also structuring cognitive structure see mental structure knowledge structure 8, 13, 123-130, 138, 177 legal structure 123-124, 126-128, 134137, 140, 142 linguistic structure 20, 123-124, 128129, 142, 177 mental structure 15, 20, 99, 105-106, 110, 123-125, 127, 129, 142, 187-188, 192,197-198

214

Index

text structure 2-3, 8, 10-11, 83, 85-86, 89,93,99,104,109,111-117 see also information structure structuring 12-13, 21-22, 79, 89, 94-95, 99, 104-106, 109,111-112, 114, 116, 123-127, 129, 143, 177, 188 see also structure cognitive, conceptual structuring 12-13, 21-22,99, 105-106, 109, 177, 188 knowledge structuring 123-127, 129, 143 perceptual structuring 12-13, 99, 105106, 109 see also perception text structuring 12-13, 79, 89, 94-95, 99, 104, 109, 111-112, 114, 116 see also information structuring

text segmentation 2, 82-83, 94, 124, 133, 134 text(ual) unit 79, 82-83, 85-86, 88, 93, 95 textual boundary 11, 82-83, 85-86 text strategy see strategy text structure see structure, structuring text/discourse type 5-7, 11, 28-29, 79-80, 94-95, 130, 132, 136-137, 140, 149, 172 text world 3, 80-82, 100, 115-116, 149, 156, 179, 188 theme 10-12, 50-53, 67, 69, 81, 83-84, 102,113,151,153-156,172 marked theme 153-156 thematic position 11-12, 81, 151, 153-156 thematic strategy 50, 52-53, 67, 69, 81, 83, 89-90 third person zero 33, 35, 40-43, 45, 46n topic 34-35, 49, 51, 52, 68, 80-81, 100103, 111-113, 115, 151, 162 transitivity 38, 44, 46-47n, 100, 102-104, 108,110,113

telicity 101, 103, 108 see also quale, telic temporal adjacency 86-87, 89 temporal sequentiality 11, 83, 85-90, 99, 101,104,108-110,112-117,181 see also text/discourse strategy, temporal; story-line text 1-15, 24, 27-29, 51, 53, 63-64, 67, 79-95,99-102, 109-110, 112-116, 123-124, 128-142, 144, 145n, 149, 151, universal 18-20, 40, 42, 44, 49, 51, 54, 154, 160-161, 177-178, 181-185 112,114, 198 see also discourse