The Construal of Space in Language and Thought 9783110821611, 9783110152432

Citation preview

The Construal of Space in Language and Thought

w DE

G

Cognitive Linguistics Research 8

Editors Rene Dirven Ronald W. Langacker John R. Taylor

Mouton de Gruyter Berlin · New York

The Construal of Space in Language and Thought

Edited by

Martin Pütz Rene Dirven

1996 Mouton de Gruyter Berlin · New York

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

International L.A.U.D.-Symposium (19th : 1994 : Duisburg, Germany) The construal of space in language and thought / edited by Martin Pütz, Rene Dirven. p. cm. — (Cognitive linguistics research ; 8) Selected papers of the 19th International L.A.U.D. Symposium, held at Duisburg, Mar. 22-25, 1994. Includes bibliographical references and index. ISBN 3-11-015243-6 1. Space and time in language-Congresses. 2. Cognitive grammar-Congresses. I. Pütz, Martin, 1955- . II. Dirven, Rene. P37.5.S65I58 1996 415—dc20 96-34598 CIP

Die Deutsche Bibliothek — Cataloging-in-Publication

Data

The construal of space in language and thought / ed. by Martin Pütz ; Rene Dirven. — Berlin ; New York : Mouton de Gruyter, 1996 (Cognitive linguistics research ; 8) ISBN 3-11-015243-6 NE: Pütz, Martin [Hrsg.]; GT

© Copyright 1996 by Walter de Gruyter & Co., 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. Printing: Gerike GmbH, Berlin Binding: Mikolai GmbH, Berlin Printed in Germany

Acknowledgements

This collection of papers is the result of the 19th International L.A.U.D. (Linguistic Agency University of Duisburg) Symposium on "Language and Space" which was held at the Gerhard Mercator University of Duisburg, March 22-25, 1994. While compiling this volume we have incurred a number of debts that we wish to acknowledge. We are deeply indebted to two of the three editors of the book series "Cognitive Linguistics Research" Ronald W. Langacker and John R. Taylor, whose support was crucial to the emergence of this volume. Furthermore, our sincere thanks go out to the authors, who have responded with professionalism to all the requests that have been made of them. In this regard, the editors of this volume would also like to express a great debt of gratitude to the expertise of the many scholars who acted as our referees: Carlo Serra Borneto, Willem J. Botha, Eugene H. Casad, Kenneth W. Cook, Osten Dahl, Wolfgang Dressier, Gilles Fauconnier, Bruce Hawkins, Dirk Geeraerts, Louis Goossens, Gottfried Graustein, John Haiman, Richard Hudson, Yoshihiko Ikegami, Laura A. Janda, Ekkehard König, Zoltan Kövecses, Bernd Kortmann, Tomasz Krzeszowski, Barbara Kryk-Kastovksy, Ronald W. Langacker, Adrienne Lehrer, Stephen C. Levinson, Dragoslav Momirovic, Peter Mühlhäusler, Winfried Nöth, Günter Radden, Brygida Rudzka-Ostyn, Rainer Schulze, Michael Stubbs, Elzbieta Tabakowska, John R. Taylor, Jan Ulijn, Claude Vandeloise, Anna Wierzbicka, and Cornelia Zelinsky-Wibbelt. Beyond these, we would like to thank the student helpers at the University of Greifswald, namely Silvio Hermann and Jacqueline Färber for compiling the Subject Index and Hendrik Wieck for proof-reading major parts of the manuscript. Finally, we would like to express our thanks to Anastasia Mora'itis, who did a marvellous job in designing the layout of the book and in taking care of the laser print-out.

vi

Martin Pütz and Rene Dirven

It is hoped that this volume will contribute to stimulating and broadening the research on the conceptualization of space, its functioning in human thought, and consequently to the further expansion of the new paradigm of Cognitive Linguistics.

Martin Pütz

Rene Dirven

Contents

Introduction: Language and the cognitive construal of space Martin Pütz PART A: SPACE IN LANGUAGE

Section 1: Pointing, deixis, and distance The Japanese verbal suffixes as indicators of distance and proximity Haruko Minegishi Cook Demonstratives as locating expressions Walter De Mulder 'Here' and 'there' in Croatian: A case study of an urban standard variety Milena Zic Fuchs Prosodic and paralinguistic signals of distance Janina Ozga

Section 2: Conceptualizing space in prepositions and in morphology The German über Elena Bellavia The separability of German über -: A cognitive approach Robert B. Dewell Prepositional prototypes Sally Rice

viii

Contents

Space and movement in the English verb system Carlos Inchaurralde

167

The representation of space in English derivational morphology Dieter Kastovsky

197

PART B: SPACE AS A CULTURAL ARTEFACT

Section 3: Can language use cope with space? Spatial deixis in Afrikaans dictionaries Willem J. Botha

211

What good are locationals, anyway? Eugene H. Casad

239

Iconicity in verbal descriptions of space Karin Wenz

269

Section 4: Variability in the conceptualization of space The syntax and semantics of locativised nouns in Zulu John R. Taylor

287

Distinguishing the notion 'place' in an Oceanic language Deborah Hill

307

The linguistic, cognitive and cultural variables of the conceptualization of space Barbara Kryk-Kastovsfcy

329

Rethinking some universale of spatial language using controlled comparison Sabine Neumann and Thomas Widlok

345

Contents

ix

PART C: SPACE AS A BRIDGE TO OTHER CONCEPTUAL DOMAINS Section 5: From one meaning to another Polarity and metaphor in German Carlo Serra Borneto

373

Metaphors of 'total enclosure' grammaticizing into middle voice markers Susan Strauss

395

Section 6: From space to time, events, and beyond The story of -ing: A subjective perspective Marjolijn Verspoor

417

The temporal use of Hawaiian directional particles Kenneth William Cook

455

The spatial structuring of events: A study of Polish perfectivizing prefixes Ewa Dqbrowska

467

Temporal meanings of spatial prepositions in Polish: The case of przez and w Agata Kochanska

491

Viewpoint and subjectivity in English inversion Heidrun Dorgeloh

509

How do we mentally localize different types of spatial concepts? Cornelia Zelinsky-Wibbelt

527

χ

Contents

PART D: SPACE AS AN ORGANIZING PRINCIPLE OF THOUGHT Section 7: Discourse as space Space in dramatic discourse

553

Vimala Herman How space structures discourse

571

Lorenza Mondada The (meta-)textual space

599

Winfried Nöth

Section 8: Abstract worlds as space From one meaning to the next: The effects of polysemous relationships in lexical learning

613

Steven Frisson, Dominiek Sandra, Frank Brisard, and Hubert Cuyckens Metaphorical scenarios of science

649

Olaf Jäkel Language, space and theography: The case of height vs. depth Jean-Pierre van Noppen

679

List of contributors

691

Subject Index

697

Introduction: Language and the cognitive construal of space Martin Pütz

The symposium held in Duisburg in the spring of 1989 marked "the birth of cognitive linguistics as a broadly grounded, self-conscious intellectual movement" (Langacker 1990: ix). What Langacker was referring to here, was the 14th International LAUD Symposium {Linguistic Agency University of Duisburg) held in Duisburg, Germany, in 1989. This symposium was the First International Cognitive Linguistics Conference. It was during that conference when the International Cognitive Linguistics Association (ICLA) was founded, the journal Cognitive Linguistics launched, and a new series Cognitive Linguistics Research set up. The proceedings of the Duisburg conference were published in the new series by Geiger & Rudzka-Ostyn (1993). Five years later, from March 22-25, 1994, the newly named Gerhard Mercator University of Duisburg once again invited researchers from all over the world to present their views and insights on a conference theme entitled "Language and Space". On the 400th anniversary of Mercator's death, the University of Duisburg renamed itself as Gerhard Mercator University to honour the great 16th century Flemish cartographer and universal scholar who spent 50 years of his lifetime in Duisburg, measuring and describing the "space" of the whole world by putting together all his maps in his "Atlas" (posthumously published in 1606). The "Language and Space" Symposium in 1994 aspired to be a tribute to Mercator of the linguistic world. The celebration of the Mercator Year and the fifth anniversary of the International Cognitive Linguistics Association's public existence was more than just a happy coincidence. "Space" is also at the very heart of all conceptualization and consequently at the very heart of the new cognitive paradigm in linguistics that seeks to explore the fundamental, spatial basis of conceptualization in and through language. Originally, Langacker even intended to label his new grammar model as "space grammar" and only later, in line with other scholars, was the more abstract name "Cognitive Grammar" established. The 30 contributions to this volume are a selection of the papers presented at that 19th International L.A.U.D. Symposium. The papers, which have been arranged in four parts, reflect some of the major aspects of the interaction between language and space:

xii

Martin Pütz

Part A: "Space in language", i.e. the way it is reflected in its lexical and grammatical structures. Part B: "Space as a cultural artifact", i.e. the variability in the construal of the domain of space. Part C: "Space as a bridge to other conceptual domains", i.e. the metaphorical exploitation of space and its role as a dominant principle of thought. Part D: "Space as an organizing principle of thought", i.e. the fundamental, "spatial" way of conceptualization itself. These four main parts of the volume are further subdivided so that altogether the book contains eight sections. Central among all contributions is the notion of 'construal', which refers to the relationship between a speaker (the conceptualizer's choice) and a situation that s/he conceptualizes and portrays in one of many alternative ways. This 'construal' relationship implies an active role on the part of the language user in organizing and structuring, i.e. "construing" his or her world (Taylor 1995). The 30 papers of this volume are a comprehensive account of the choices of conceptual and linguistic alternatives available to the speaker (hearer) which determine the construal of space in language and thought.

PART A: SPACE IN LANGUAGE Section 1: Pointing, debds, and distance Since man is "in space" and since space is also "in man", the human conceptualizer must create a number of deictic pointers to trajectors in space which may be framed in collaboration with demonstratives (the prototypical device), but also in totally unexpected ways, e.g. as verb suffixes or prosodic markers. The speaker's choice of certain linguistic expressions may suggest a construal of a situation in terms of communicative needs and intentions. In this respect Haruko Minegishi Cook's contribution "The Japanese verbal suffixes as indicators of distance and proximity" focusses on the more active role of the language user in portraying a given situation. Cook analyzes the Japanese verbal suffix -masu as an expression of 'social deixis' or distance and demonstrates how distance is interpreted in various speech contexts. She emphasizes the importance of recognizing the distinction between the interpreted meanings of a deictic word and the encoded meaning of its lin-

Introduction

xiii

guistic form. The distinction is certainly necessary in order to account for the various meanings associated with the Japanese addressee-honorific suffix masu. A more prototypical device in creating deictic pointers is the use of "Demonstratives as locating expressions". In his paper, Walter De Mulder attempts to show that demonstratives are not pointers which indicate where in the context their referent is to be found. Rather, by analyzing French demonstrative noun phrases such as ce Ν 'this/that N', he shows that the only meaning they convey is that their referent must be identified with the help of a contextual element within the spoken discourse. Identifying the referent of these opaque deictic expressions then means that it will be necessary to combine their linguistic meaning with perceptual interpretative strategies. To strengthen his views, De Mulder exploits the figure-ground model as a fundamental feature of cognitive organization, i.e. the perception of a visual scene divided into the foregrounded figure and the fixed background. In the same vein, the importance of spatial deixis in language is taken up in Milena Zic Fuchs' paper entitled "'Here' and 'there' in Croatian: a case study of an urban standard variety". 2ic Fuchs focusses on issues of spatial deixis and in particular on a number of the Croatian equivalents of the English demonstrative adverbs here and there, such as the pronominal demonstrative adverbs ovdje, tu, and ondje 'here' which form a three-number paradigmatic set for designating location. One of Zic Fuchs' conclusions is that, historically speaking, this three-way spatial distinction based on participant roles (proximal to the Speaker/Hearer and non-proximal to either of them) was originally found in the Stovakian dialect, and then taken over into Standard Croatian. She leaves the question open for further diachronic research as to why systems of deictic expressions change within languages and dialects. In more marginal or peripheral ways, the construal of space encoded in "Prosodic and paralinguistic signals of distance" is the topic of Janina Ozga's contribution. Ozga analyzes the relation between prosody and distance from a cognitive point of view and examines the universal principles underlying the spatially constrained use of prosody and paralanguage. She shows that certain prosodic and paralinguistic features are non-trivially and non-randomly associated with linguistic forms involving the notion of space.

xiv

Martin Pütz

Section 2: Conceptualizing space in prepositions and in morphology Going beyond these first "localization" concepts from the speaker's viewpoint, the human conceptualizer has to structure the whole of his visual field along a number of vertical and horizontal axes so that subfields can be created into which motion can be projected. Here, it is especially prepositions and morphology which come into play. Accordingly, Elena Bellavia's paper "The German über" proposes a radial lexical network symbolizing the whole semantic area covered by the lexical unit über 'over'. Exploiting the trajector/landmark distinction, Bellavia gives a detailed account of the different uses of 'über' (+noun) and 'über' as a verb prefix and then attempts to compare and unify their meanings in a radial lexical network approach. She concludes that the prepositional phrase 'über+nouri represents a more general characterization of movement, while verbs prefixed by über are specifications for the purpose of construing particular aspects. It is a happy coincidence that Robert B. Dewell also focusses on the German preposition über with respect to how the speaker "construes" his conceptualization in a specific way. In contrast to Bellavia, however, he concentrates on "The separability of German über", discussing the question of when this prefix is separable and when inseparable. He first offers a sharp criticism of traditional treatments of the subject, based on the ground that data on prefixes are not described adequately and, moreover, that basic underlying patterns are not being investigated. Instead, Dewell argues for new directions of inquiry, i.e. the consideration of some basic cognitive issues such as the representation of path, the development of specialized semantic variants from spatial-path images, and the role of prefixes in conveying aspectual contours or transitivity. He convincingly argues that there are regular patterns and semantic constraints which underlie and explain the alternate use of prepositional phrases both as separable and as inseparable prefixes. In her paper "Prepositional prototypes", Sally Rice is more cautious in proposing an actual network for spatial prepositions, offering instead a "working 'regional' model". The model accounts for a prepositional network which may "ultimately reveal a structure with a temporal region and core sense, a spatial region and core sense, and a more diffuse but coreless abstract region surrounding them both". Rice presents an empirical analysis of the three most basic English spatial prepositions to show that there are prototypical semantic values for at, on, and in which are spatial in meaning. Furthermore, she suggests that in the subjective lexicons of speakers there may be multiple prototypes for a certain preposition, some of which cannot be spatially defined at all.

Introduction

xv

The semantics of English verbs with a focus on the tense modality system is the topic of Carlos Inchaurralde's paper "Space and movement in the English verb system". He argues that space has certain topological properties that can be expressed mathematically. Resorting to mathematical formulae, he adopts a geometrical perspective through which he demonstrates that the tense modality system and the lexical characterization of verbs can be analyzed through the concepts "space" and "movement". In line with the general theme of the section - the structuring of the visual field along a number of vertical and horizontal axes - Inchaurralde concludes that the same modal form, e.g. would, could be used for indicating remoteness either on the vertical axis of time or on the horizontal axis of hypotheticality or in a combination of the two. The last contribution in this section is Dieter Kastovsky's paper "The representation of space in English derivational morphology". It is devoted to the interplay between the functions of morphological and word-formation issues and the notion of "space" as a "pervasive lexical-semantic category". Kastovsky focusses on various types of English word-formation patterns such as compounding, prefixation, and suffixation with reference to their involvement in creating space-denoting lexical items: in other words, processes in which spatial/locative relations play a certain role. It becomes obvious that English word-formation serves to encode spatial relations for nominating as well as for pronominalisation purposes and that e.g. prefixation exclusively yields relational formations, i.e. "formations which encode a spatial relation but do not directly refer to a space manifestation".

PART B: SPACE AS A CULTURAL ARTIFACT Section 3: Can language use cope with space? Having all these devices for the conceptualization of space at hand, the question arises: how do they function in actual language use and, even more fundamentally, do they function in any satisfactory way? The question is taken up by Willem Botha in his paper entitled "Spatial deixis in Afrikaans dictionaries". Botha turns his attention to spatial deictic problems the lexicographer must take into account. First, s/he must portray the individual meanings of typical deictic words such as here, there, front, back. Second, s/he must project the meanings of certain localities which get their meanings from the locality of a conceptualizer in relation to certain arbitrary reference points, using typical deictic words which could not properly be understood outside a particular context. Botha addresses these

xvi

Martin Pütz

problems by analyzing the way in which some spatial deictic words are portrayed in an Afrikaans monolingual dictionary. Eugene Casad provides us with data from Cora, a Uto-Aztecan language of Northwest Mexico and poses the question "What good are locationals, anyway?" In other words, do they function in any satisfactory way? The author attempts to reply to the question by illustrating some of the functional usages of various grammatical elements such as adverbs, particles, demonstrative pronouns, etc. which indicate various kinds of spatial, temporal and logical concepts. His general conclusion is that in Cora, many locational and directional morphemes and lexical items relate to other domains of the grammar: they may distinguish one word from another, or they may help to distinguish different tense aspect forms of the same verb. In short, "without locationals, the Coras would have to talk about the world in an entirely different way", which certainly emphasizes the active role of the language user in organizing and structuring his or her world. Karin Wenz's paper "Iconicity in verbal descriptions of space" also begins with the assumption that the description of space presupposes previous cognition, namely internalization of the environment. She assumes that the verbalization of spatial experience yields insights into the processes of perceiving space. The iconic principles which govern the semiotic relationship between spatial cognition and the textual representation of space are empirically discussed and evaluated. The notion of 'iconicity' (Haiman 1985) as the conceived similarity between conceptual structure and linguistic form can therefore be seen as one of the salient aspects to the construal relationship discussed at the beginning of the Introduction.

Section 4: Variability in the conceptualization of space The general idea presented in the previous papers that the conceptualization of space presupposes the knowledge of man's broad or particular environment inevitably leads to the next step in the exploration of the conceptualization of space: to what extent is space a cultural artifact or a social construct created by the mind of man as a social being, dependent on his environment. Indeed, the conceptualization of space varies considerably among non-European cultures, especially African and Oceanic. An elucidating example revealing the variability in the conceptualization of space is John Taylor's paper "The syntax and semantics of locativized nouns in Zulu". Taylor provides a cognitive account of locativized nouns expressing spatial relations in Zulu, the largest language group of the Republic of South Africa. In particular, Taylor addresses the semantics of locativized nouns with reference to their expression in conjunction with

Introduction

xvii

motion verbs, e.g. the use of these nouns in copulative constructions to indicate the place of a trajector entity. One of Taylor's suggestions is that the function of the locative morphology in Zulu is to convert a "thing" concept into a "place" concept. Furthermore, he claims that locativized nouns do not contain any notion of "to" or "from". They do not denote a place construed as goal, source, or path, but as a pure "place". Yet another 'case study' emphasizing the cross-cultural variability in the conceptualization of space is illustrated by Deborah Hill in her paper "Distinguishing the notion 'place' in an Oceanic language". Hill analyzes data from the Oceanic language Longgu which is spoken on the island of Guadalcanal in the Solomon Islands. She argues that in that language there is a linguistic and cultural importance attached to the notion "place" or, in more specific terms, "home place", which is the linchpin for the whole system of spatial reference. Hill concludes that there are recurring patterns of conceptualization or 'themes' in language. Thus the distinction between what is "home" and what is "not home", reflected in the distinction between direct and indirect possession (what is part of me and what is not part of me), can be seen as different manifestations of the same theme. Variability in the conceptualization of space per se is also a prominent topic in Barbara Kryk-Kastovsky's paper "The linguistic, cognitive and cultural variables of the conceptualization of space". She convincingly argues that the variety of deictic systems operating in a multitude of languages requires a complex and multidisciplinary approach. From a synchronic perspective, deictic systems should not only be studied according to their use in standard varieties. In order to establish contrasts, dialectal features should also be part of the investigation. Furthermore, diachronicallly, deictic systems call for intralinguistic analyses so as to be able to detect the history of particular forms within the development of a single language. In their contribution "Rethinking some universale of spatial language using controlled comparison", Sabine Neumann and Thomas Widlok present field research data from two languages of Southern Africa. The paper explores the potential of regional comparison in the field of spatial conceptualization and in particular discusses the question of whether the division between Bantu-speaking agropastoralists and Khoisan-speaking huntergatherers corresponds with two distinctive ways of encoding and conceptualizing spatial relations.

xviii

Martin Pütz

PART C: SPACE AS A BRIDGE TO OTHER CONCEPTUAL DOMAINS Section 5: From one meaning to another Now, even if "space" never exists or existed independently 'out there', and is basically a product of man's mind and social context, the concept of space has always served as the 'primary' conceptualization onto which many other conceptual domains have been mapped. Thus, spatial concepts can be shown to become a principle for the organization of the conceptualization of ever more abstract domains such as that of "total enclosure", as framed by the 'middle voice', and also of concepts such as viewpoint or subjectivity. The construal of our world is not limited to the categories provided by our culture and fixed in our language. Human thought processes are largely metaphorical (Lakoff/Johnson 1987), which suggests that in creating new concepts from existing ones we are also able to extend our repertoire of linguistic expressions. The analysis of metaphorical concepts or nonliteral expressions is at the center of Carlo Serra Borneto's contribution entitled "Polarity and metaphor in German". Serra Borneto first examines the notion of "conceptual polarity", which reflects a general human tendency in thinking. Polarity is at work at all linguistic levels and refers to linguistic items which are conceptually in competition with one another such as the German verbs liegen 'to lie' and stehen 'to stand'. Serra Borneto attempts to show how the schema of polarity can be applied to a number of conceptual and orientational metaphors involving the use of the 'locational' verbs liegen and stehen. More abstract domains are discussed in Susan Strauss' contribution "Metaphors of 'total enclosure' grammaticizing into middle voice markers". Strauss illustrates the value of cross-linguistic research and draws attention to the fact that human languages exhibit strikingly similar and consistent grammatical patterns in "describing, reacting to, and expressing particular types of experiences and events". Strauss examines how and to what basic degrees auxiliary verbs in Japanese and Korean function as grammatical analogues to the reflexive marker in Romance or the middle voice marker in Spanish.

Introduction xix

Section 6: From space to time, events, and beyond One of the major domains that have been conceptualized in terms of space is time, which is even commonly referred to as "temporal space". But other linguistic and conceptual construals such as complementation, perfectivity, viewpoint and subjectivity are also relatable to notions of space. The syntax of English complementation is certainly an area which cannot be adequately accounted for without conceptual explanations. The topic of sentential complementation is taken up in Marjolijn Verspoor's paper "The story of -ing: a subjective perspective", whereby she investigates the general cognitive principles that motivate complement distribution. One of her main concerns springs from the observation that each type of complement may occur with typical action verbs and/or mental causation verbs and also with typical perception verbs or cognition verbs. Since those different verb classes comprise both spatial and non-spatial members, she sets out to ask the question whether there is one abstract schema that plays a role in complement selection with both groups of verbs. She eventually shows that the query can be responded to positively. Likewise, the notion of 'time' as conceptualized in terms of space is taken up by Kenneth W. Cook in his contribution entitled "The temporal use of Hawaiian directional particles". Cook deals with the extension of Hawaiian directional particles into the domain of time. The Hawaiian case seems to be peculiar in that the directionals are used to indicate degrees of temporal remoteness. Furthermore, Cook's paper motivates these degrees of temporal remoteness of Hawaiian directional particles in terms of human bodily experience. In particular, the direction indicated by aku ('away from the speaker') has no limits as one can traverse the earth/sea endlessly, whereas 'up' and 'down' are limited in traditional human experience: "one can only go up as far as one can jump or climb, and one is similarly limited in one's downward movement". Conceptual construals such as perfectivity are taken up by Ewa Dabrowska in her article entitled "The spatial structuring of events: a study of Polish perfectivizing prefixes". Using a cognitive grammar perspective, she gives a detailed analysis of five polysemous prefixes which are related to spatial prepositions and which are said to have clearly spatial meanings. Her main purpose is to show how these prefixes change the meaning of the verb to which they are attached and what determines which prefix is chosen in a given situation. Also based on the Polish system of time is Agata Kochaiiska's account "Temporal meanings of spatial prepositions in Polish: the case of przez and w", where the temporal senses of two Polish prepositions are discussed within the cognitive framework. This approach makes it possible to analyze

xx

Martin Pütz

temporal senses of prepositions as members of coherent categories centered around the prototypical spatial meanings. Kochanska also makes it clear that cognitive linguistics seems to account for the fact that the grouping of temporal and spatial senses of prepositions in one polysemous category is highly meaningful and natural for speakers of Polish and many other languages. In any construal of a situation, the notion of 'perspective' is central to the construal relationship. It may refer to the conceptualizer's viewpoint, viz. the "mental route" that a speaker takes in presenting a scene or event. The choice of a particular perspective or viewpoint in construing the world is discussed in Heidrun Dorgeloh's contribution "Viewpoint and subjectivity in English inversion". Dorgeloh sets out to investigate three inversion types - after deictic adverbs and following locative as well as non-locative constituents - which are related in a specific way. Inversions contain a reference point as vantage point from which a scene or event is seen, thereby expressing a particular viewpoint or perspective which the speaker assumes. Inversion can then be described as one device "whereby the immediate nature of an experience is reproduced in a subjective manner". The section concludes with a paper by Cornelia Zelinsky-Wibbelt in which she poses the question: "How do we mentally localize different types of spatial concepts"? Zelinsky-Wibbelt is mainly concerned with the behaviour of spatial predicates and with their metaphorical and metonymic derivations. She ends up with a model in which three different contextual functions operate on accordingly different lexical representations: contextual 'selection' operates on lexical representations which are equally valid, contextual 'configuration' operates on an abstract schema, and contextual 'shift' operates on the prototypical sense of the spatial predicate.

PART D: SPACE AS AN ORGANIZING PRINCIPLE OF THOUGHT Section 7: Discourse as space Space serves as an organizing principle of thought not only in the iconic reflection of events in sentence order, but also in the structure of discourse as a whole. "Discourse space" thus becomes a complex metaphor reflecting various principles of organization. In the same vein, Vimala Herman's paper "Space in dramatic discourse" looks at how space is created linguistically in fictional texts and how it can be put to use in interaction. Vimala's work also seeks to explore how the

Introduction

xxi

non-verbal organization of physical space via architecture, stage design, and the relationship of stage and audience have contributed to the ways in which space may be conceptualized within the constraints of dramatic use. 'Conversation' as a genre is yet another example of discourse dealt with in Lorenza Mondada's paper entitled "How space structures discourse". Mondada sets out to analyze spatial expressions used as markers to structure discourse, e.g. 'to come to our subject', in a particular text corpus. These texts are descriptions of space, exploiting the ambiguity of those markers' referential domain, which can refer both to territory or to textuality. Mondada makes it clear that the ambiguity is used by speakers to structure and motivate their discourse. "Discourse space", as a complex metaphor reflecting various principles of organization, is also dealt with by Winfried Nöth in his contribution "The (meta-)textual space". Nöth discusses two kinds of spatial metaphors of the metatext, i.e. "our way of referring to the text". These metatextual metaphors may be alive or transparent, at times opaque and, according to Nöth, apparent only to the etymologist. We can trace the geometry of these metatextual spatial concepts as it consists in its zero dimension of points, in its first dimension of lines, in its second dimension of surfaces or planes and in its third dimension of bodies. Nöth provides the reader with numerous other examples of metatextual metaphors, e.g. as they may be observed in the three dimensions of textual space reflecting the three main dimensions of human orientation: the horizontal (right/left), the vertical (above/ below), and the sagittal (front/back).

Section 8: Abstract worlds as space Space also becomes an organizing principle in the structuring of other important domains of experience such as language acquisition, the understanding of science and of specific subfields of it such as theography. The domain of language acquisition is focussed upon by Steven Frisson, Dominiek Sandra, Frank Brisard, and Hubert Cuyckens in their article "From one meaning to the next: the effects of polysemous relationships on lexical learning". The authors attempt to investigate the hypothesis that extension processes are an inalienable aspect of human categorization, one that is motivated in part by its warranting high efficiency in lexical learning. According to the authors, there seems to be clear evidence for the claim that the concept of semantic distance can serve as a key notion in the field of lexical acquisition. Thus, a series of experiments led to the assumption that notions of vagueness, homonymy, and polysemy obviously play an important role in determining the exact status of the results and

xxii

Martin Pütz

their implications. A more abstract domain - metaphorically conceptualized in terms of more concrete domains - is examined in Olaf Jäkel's paper entitled "Metaphorical scenarios of science". Within a cognitive theory of metaphor, Jäkel seeks to investigate metaphors as observed in a highly abstract domain such as the discourse of science. Jäkel is mainly interested in the whole conceptual domain of science as it is organized by a scenario which consists of several elements such as nature, a scientist, methods, theories, and scientific progress. Jäkel assumes that this generalized science-scenario "includes all the ingredients necessary for a fully-fledged conceptualization of science". Concluding this section is Jean-Pierre van Noppen's contribution entitled "Language, space and theography: the case of height vs. depth" which focusses on a religious domain of metaphorization processes. The author is concerned with descriptive theology ('theography') which, in many cases, deals with metaphorical modes of expression such as the spatial language of localization and orientation and which, according to van Noppen, provides important insights into the relationship between the human and the divine. In line with the cognitive commitment, van Noppen states that these spatial metaphors are based on the "universal human experience of interaction between the body and its environment". The symposium's original theme was "Space in language and language in space". Both aspects of this theme have been developed in an extremely rich and encompassing manner. Space has been shown to be conceptualized by almost all word categories and by the most unexpected grammatical devices; space has moreover been conceptualized with such a great diversity in various languages that the concept of space is by and large a cultural artifact, just like emotions are now claimed by social constructionalists to be cultural creations. The second half of the theme "Language in space" has been widened and deepened as 'language and thought in space', whereby the concepts of space available to a speech community enable the creation of further abstract conceptualizations and the organization of thought as such. Space is therefore at the very heart of thought.

References Geiger, Richard A. and Brygida Rudzka-Ostyn (eds) 1993 Topics in Cognitive Linguistics. Amsterdam/Philadelphia: Benjamins. Haiman, John (ed.) 1985 Iconicity in Syntax. Amsterdam/Philadelphia: Benjamins. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago: University of Chicago Press.

Introduction

xxiii

Langacker, Ronald 1990 Concept, Image, and Symbol. Berlin/New York: Mouton de Gruyter. Mercator, Gerardus 1606 Atlas sive cosmographicae meditationes de fabrica mundi et fabricati figura ... Amsterdam: Judocus Hondius. Taylor, John 1995 Introduction: On construing the world. In: Taylor, John and Robert E. MacLaury (eds). Language and the Cognitive Construal of the World. Berlin/New York: Mouton de Gruyter, 1-21.

Part A Space in language

Section 1 Pointing, deixis, and distance

The Japanese verbal suffixes as indicators of distance and proximity Haruko Minegishi Cook

1. Introduction Since language is used by human beings, it is situated in time and space. In this sense, we assume that all languages encode spatial notions. Languages extend the notion of space from the physical to the temporal, psychological and social domains. Honorifics are typically seen as a marking of social ranking and politeness but they can also be seen as markers of distance (cf. Brown and Levinson 1987). This paper takes the view that a Japanese honorific form is an indicator of distance and that the indication of politeness is one manifestation of distance. It analyzes the Japanese addressee honorific form masu and its nonhonorific counterpart, the plain form, as indicators of distance and proximity. 1 Although such a proposal has been previously made, the present study makes the following contribution: (i) Distance/proximity is not only interpersonal but also intrapersonal; (ii) The meanings of distance and proximity are encoded in the linguistic form and other meanings (including polite and non-polite) are implicatures in the speech context. This analysis can account for various meanings associated with the masu and plain forms which could not be explained in the conventional analysis. The data used in this study come from various genres of natural speech. In Japanese, sentences end in either the masu form or in the plain form, as shown in (la) and (lb). ga Mary to dekake-masu. (masu form, honorific) 2 S with go out 'John goes out with Mary.'

(la)

John

(lb)

John

ga Mary

to

S with 'John goes out with Mary.'

dekake-ru. (plain form non-honorific) go out

4

Haruko Minegishi Cook

The referential meanings of (la) and (lb) are identical but their social meanings differ. 3 Previously a number of scholars (Goldstein and Tamura (1975), Harada (1976), Ikuta and Ide (1983), Martin (1964) and Neustupny (1978) among others) analyzed the masu form as a polite speech level marker. Under this analysis, sentence (la) expresses politeness to the addressee, and sentence (lb) expresses intimacy or lack of politeness to the addressee. In much of the literature, it has been stated that the polite style is used when the addressee is a person who is socially superior or equal to the speaker (cf. Harada 1976) or is used to talk with outsiders (cf. Shibatani 1990). Mutual plain form exchanges indicate that there exists an informal, casual or inside relationship between the interlocutors. In this respect, the masu and plain forms respectively resemble the second personal pronouns V (as in French vous) and Τ (as in French tu) that occur in many European (and some other) languages. However, the difference is that the V and Τ pronouns are used only when the referent is the second person whereas in Japanese the masu and plain forms are used regardless of the referent of an utterance.4 In this way, they are far more pervasive. The masu form has been also analyzed as a marker of distance (Hinds 1976, 1978; Ikuta 1983; Jorden and Nöda 1987; Shibatani 1990). Based on his findings in natural interview data, Hinds (1976) claims that the masu form is chosen when there is perceived distance between the speaker and the addressee. Hinds (1976) also reports that the masu form appears when a high status person other than the addressee is present, and when the situation is formal. Both Ikuta (1983) and Shibatani (1990) propose that the masu form indicates psychological distance between interlocutors. Examining conversations with respect to the masu and plain forms, Ikuta (1983) also notices that the masu form tends to mark topic shift. Thus she proposes that the masu form is a marker of not only interpersonal but textual distance. The assumption in these proposals is that the plain form is a marker of interpersonal or textual proximity. Recently, Maynard (1991, 1993) has claimed that when speakers are more "aware" of the addressee as a separate entity, they are more likely to use the masu form and when they are less aware of the addressee, they are more likely to use the plain form. Maynard explains that in an intimate conversation the plain form is normally used because in such context, due to amae "psychological and emotional dependence", "the speaking self finds less need to address 'thou' as a completely separate and distinct entity" (1993: 178). Since separateness and oneness involve the notion of space, her proposal can be seen as one of distance and proximity. Although these proposals are insightful, these scholars have not clarified how their analysis relates to the conventional meaning of the masu form and its other social meanings.

Japanese verbal suffixes

5

In sum, there are at least three important questions yet to be answered: i) How is the meaning of distance/proximity related to the meaning of politeness/non-politeness? ii) How is the meaning of distance/proximity related to other social meanings? iii) Can the types of distance so far proposed (physical, psychological/interpersonal, and textual) adequately account for many uses of the masu and plain forms? In what follows, I will attempt to answer these three questions by showing that there is a difference between the encoded and situational meanings of these forms.

2. The encoded meaning The idea of encoded meaning vs. implicature is not new in the literature. Grice (1975) discusses the notions of natural and non-natural meanings.5 Natural meaning is directly linked to the linguistic form and non-natural meanings are implicatures that are derived from the literal meaning of an utterance used in a given context. For implicatures to arise, the literal meaning must be present. Similarly, in the case of deictics, for situational meanings to arise, the encoded meaning must be present. I use the terms encoded and interpreted meanings for Grice's natural and non-natural meanings, respectively. Both Hanks (1990, 1992) and Levinson (1979, 1983) argue for the importance of recognizing the distinction between the interpreted meanings of a deictic word (i.e. conveyed meanings, situational meanings, or implicatures) and the encoded meanings of the linguistic form. To understand and explain the nature of honorifics and the extent to which they are used in various social situations, it is important to clarify this distinction. More specifically, such a distinction can account for pragmatic processes in which various usages or situated meanings are established. The question is what counts as an encoded meaning? Although Hanks (1990) states that there is no clear test to distinguish the encoded meaning from situational ones, he mentions, "The division [of the two levels of meaning] is mainly a matter of the relative consistency or constancy of association between form and meaning, not of the kind of information" (1990: 53). In this paper I assume Hanks' claim concerning the difference between encoded and situational meanings and further propose that the encoded meaning of a social deictic expression is the one that is always present and that it is a part of the meaning(s) of an utterance in any instance of use. In this formulation, there may be more than one meaning assigned to a form simultaneously: the encoded meaning is one, and the others are situational meanings derived from the encoded meaning in a given context.

6

Haruko Minegishi

Cook

In some instances, the encoded meaning may be the only meaning in the context. In other instances, since situational meanings are interpretations, there may be more than one situational meaning assigned to a deictic feature. 2.1. The encoded meaning of the masu form In this section I propose that the encoded meaning of the masu form is not politeness but distance. Furthermore, my proposal is different from the previous analyses of the masu form as an indicator of distance in that it includes distance between the self and his/her social role (i.e. intrapersonal distance). It is difficult for the conventional analysis of the masu form (i.e. politeness) to explain the use of masu in some scicial contexts such as that of scolding a child. Consider example (2). Here the mother is angry at child Κ (a seven-year-old boy) because he is not sitting at the dinner table properly and because he is shaking the table. She scolds Κ by using desu, the copula form of masu, and -masen, the negative of masu. (2) Mother 1

—>Dooshite soo yatte gatan gatan why so do rattle rattle 'Why are (you) shaking (the table)?'

2

— > Tatehiza ikemasen. erect knee no good 'Don't draw up your knees.'

suru do

η desu ka! NOM COP INT

The use of masu in example (2) does not index politeness. It can be explained by the analysis that the masu form indexes distance between the speaker and the addressee. When the mother scolds the child, she creates psychological distance from the child. Since the encoded meaning must be present in all instances of the form, the meaning of politeness cannot be the encoded meaning of the masu form. It is known that the deictic meaning of space is often extended to the feeling level. For example, based upon a study on demonstratives, R. Lakoff (1974) claims that emotional and spatial distance and closeness are clearly related. Haiman (1983: 800) also notes, "physical distance is an obvious metaphor for social distance ...". Thus, it is not unreasonable to assume that the distance encoded by the masu form can be physical and/or

Japanese verbal suffixes

7

psychological (i.e. interpersonal). The claim that the encoded meaning of the masu form is distance can account for the fact that this form is generally used between speakers who are not in a close personal relationship and when the addressee is not present in the speech context (see Hinds 1976). Family members who usually speak in the plain form in face-to-face conversation often use the masu form when they are physically apart (e.g. in letters). 6 In my data, when the family members are making a cassette tape to send to one of the daughters, H, who lives in the U.S., the speakers generally shift to the masu form, which is illustrated in example (3). Normally AT, who is a sister of H, speaks to Η with the plain form if Η is in the immediate speech context. However, in (3) AT uses the masu form consistently with H. (3)

[AT is recording a message for her sister H]

1

— > Kurisumasu kaado ο todokimashita. Doomo arigatoo. Christmas card ο reach past much thank '(I) received (your) Christmas card. Thank you.' ->Eee kochira kara no Kurishumasu FI this side from LK Christmas todokimashita kal reach past INT 'Did (you) receive (our) Chirtmas card?'

kaado card

mo also

->Taakii wa umaku yakemashita kal turkey Τ well bake past INT 'Did (you) bake (your) turkey well?' Eee, hajimete no Kurisumasu yama de sugosu no FI first time LK Christmas mountain at spend NOM ->totemo ii desu ne. very good COP FP 'Well, (it)'s very nice to spend the first Christmas (after marriage) in the mountains.' • > Urayamashii

desu.

envious COP '(I)'m envious of (you).'

8

Haruko Minegishi

6

Cook

Atashi mo tabun Ichigatsu I too maybe January ikoo' nante kangaete, go what Q think boonasu mo deta koto

de

eee (.) hokuhoku

bonus

COP

Fl

— > shite do

also

paid

NOM

ni in

koo 'sukii this ski

ni to

delighted

imasu. ((laugh)) exist

'I'm also thinking of going skiing in January, and (I) got (my) bonus and well (.) (I)'m happy.' *

-mashita is the past tense of -masu.

It is true that letters and recorded messages are potentially available to a third party. In this sense, they may lose some of the intimacy associated with a face-to-face conversation between people in a close relationship, which might motivate speakers to use the masu form. Although this effect is partially responsible for the use of masu form in letters and recorded messages between intimate parties, the fact that even letters which include highly private matters can be written in the masu form suggests that physical distance is also responsible for the use of the masu form in these genres. The previous studies on the masu form which claim that it marks distance only consider the distance between topics in text and between the speaker and the addressee or the third party. However, as shown in example (4), this formulation is not adequate. In example (4), which comes from a dinner table conversation, the mother uses a masu form when she tells the children that there is more omelette. Note that child C does not use a masu form but uses a plain form hoshii 'want'. There was neither a higher status person nor an outsider in the speech situation.7 (4) Mother

Child C

Tamagoyaki hoshii hito omelette want person — >mada arimasu yo, okawari. still have FP seconds 'Anyone who wants omelette, there's still some more.' Kamaboko hoshii. fish cake want '(I) want fish cake.'

Japanese verbal suffixes

9

The mother's use of the masu form is difficult to explain within the proposal that it indexes distance between the speaker and the addressee or the third party. I propose that the masu form also indexes distance between the self and his/her social role that the self presents to others at the time of interaction (cf. Scollon and Scollon 1981). To speak like a 'mother', 'teacher', 'doctor', 'businessman' etc., the speaker creates distance from the innate mode of self and puts on a mask of a particular social role. Hinds (1976) also observes that professional opinions and evaluations can cause a shift from the plain to masu in adult-to-adult conversation. I claim that the masu can mark distance within one speaker. This analysis can account for example (4). One of the important responsibilities associated with the role of the Japanese mother is to provide food to the members of the family. The mother's use of the masu form here indexes distance between the speaker's self and her social role as a 'mother'. In sum, I propose that the encoded meaning of the masu form is distance, which includes distance between the innate mode of self and his/her social role (i.e. intrapersonal distance).

2.2. The encoded meaning of the plain form The previous studies on the masu form all suggest that the plain form marks intimacy or psychological proximity since a lack of interpersonal distance is intimacy. However, this formulation does not explain some uses. Consider example (5) which comes from a neighborhood quarrel. In this example, both speakers use the plain form. Here the landlord, Mr. Suzuki, is very angry because one of his tenants put out his trash without separating burnable and unburnable items.8 A neighbor is defending the tenant. Clearly, in (5) the speakers are opposed to each other and do not share a sense of intimacy or oneness. S=Suzuki, the landlord; Ν=neighbor (5)

S: Ne, chanto sooiu fuu ni hakkiri to wakatte iruba FI properly such way in clearly knowing exist if — >watashiwa okoranai yo. I Τ angry NEG FP 'You see, if you clearly separated it [the trash], I would not get angry.' N: Kare, he

shiranakatta η know NEG past NOM

da COP

kara because

10

Haruko Minegishi

Cook

— >ma shooganai. Shooganai tte iu ka. Fl help NEG help NEG Q say INT 'He did not know about it, so, it can't be helped. We may say it can't be helped.' — > S : (Soide) shiranakatta

tte

iu η ja

anta

komaruyo.

and know NEG past Q say NOM you trouble FP '(Then) if he says he did not know, that's a problem, you know.'

If the plain form indexes a lack of distancing in the speaker's own self (i.e. a lack of social role), the use of the plain form in (5) can be accounted for. In (5) both speakers, in particular Mr. Suzuki, are emotionally involved in the argument to the extent that they lose their composure. In such a state, it is difficult to act out their social roles. The speakers in the quarrel use the plain form to interact in the innate mode of self without marking their social roles. I propose that the encoded meaning of the plain form is a lack of distance or proximity between interlocutors as well as between the self and his/her social role. Underlying this hypothesis is Haiman's (1983) claim that there is an iconicity between linguistic expressions and nonlinguistic features of both cognitive and social contexts of speech. In this analysis, the more morphemes in a word or the more words in a sentence, the greater the conceptual or social distance becomes. Haiman states: The linguistic dimension is that of distance between linguistic expressions— which corresponds directly to, and in this sense is motivated by, a variety of conceptual dimensions (1983: 781). The verbosity or prolixity of formal registers may then be a verbal icon of an envelope around the speaker's actual message. The addressee is protected by this envelope from the speaker's ideas in the same way that he is protected by physical distance from other emanations of a personality (1983: 801).

Crosslinguistically, it seems true that so-called polite or formal expressions are more morphologically complex. For example, euphemism, technical terminology, and honorifics are typically longer or contain more morphemes. Now let us consider Table 1 which compares the morphology of the masu and plain forms of the verbal, adjectival and nominal predicates.

Japanese verbal suffixes

11

Table 1. Morphology of the masu and plain forms

verbal predicate: to look masu form

plain form

mi-mas-u stem-dis-prt

mi-ru stem-prt

mi-mas-ita stem-dis-past

mi-ta stem-past

adjectival predicate: to be large masu form

plain form

ooki-i des-u stem-prt COPdis-prt

ooki-i stem-prt

ooki-katta des-u stem-past COPdis-prt

ooki-katta stem-past

nominal predicate: to be a book masu form

plain form

hon des-u nom COPdis-prt

hon da nom COP-prt

hon des-ita nom COPdis-past

hon datta nom COP-past

dis=distance morpheme COP=copula prt=present tense past=past tense As Table 1 shows, the plain form consistently contains fewer morphemes. The plain forms of verbal and adjectival predicates consist only of the present tense markers -(r)u for verbs and -i for adjectives or of the past tense markers -ta for verbs and -katta for adjectives. The plain

12

Haruko Minegishi Cook

form of the nominal predicate consists of a noun plus the present tense copula da or its past tense form datta. In contrast, the counterparts of the masu form include distance morphemes {-mas for the verbal and des- for the adjectival and nominal predicates). Given the iconicity proposal (Haiman 1983), which predicts that the morphologically simpler forms are more plain, it is reasonable to propose that the encoded meaning of the plain form is lack of physical, psychological and interpersonal distancing between interlocutors and between one's own self and his/her social role. 9

3. Situational meanings

3.1. The conventional meaning of the masu form In my analysis of the masu form, the conventional meaning is one of the situational meanings. Due to a fit between the encoded meaning of the linguistic feature and native speakers' expectations of speech contexts (or written genres) in which the masu form is used and due to its frequent use in common speech contexts or written genres, native speakers are more aware of the conventional meaning than they are of the other situational meanings. Generally, native speakers feel that speech with the masu form is polite. The proposed encoded meaning of the masu form (i.e. distance) can account for the politeness which is conventionally associated with the masu form. As discussed by R. Lakoff (1973) and Brown and Levinson (1978), distance implicates the conventional meaning of politeness. R. Lakoff (1973) proposes that non-imposition, which presupposes the speaker's distance from the addressee or the content of speech, is a form of politeness. Related to what Lakoff proposes is the fact that in some Romance languages, a polite address and referent term for the second person takes the third person verb form, which, given that the third person is usually farther away from the speaker than the second person, can be said to indicate distance (e.g. Spanish usted, Portuguese voce, Italian Lei). Similarly, Brown and Levinson (1978) propose that 'negative politeness', which minimizes the threat to the addressee's 'negative face want' (i.e. desire to be unimpeded), is indicated by various linguistic and nonlinguistic devices which create social distance. Brown and Levinson also note that honorifics are cases of 'frozen outputs of politeness strategies' (p. 184). Taking all of these proposals into consideration, it makes sense to consider

Japanese verbal suffixes

13

that the conventional meaning of the masu form (politeness) derives from its encoded meaning of distance. According to Brown and Levinson (1987), politeness is an outcome of a strategy that minimizes face risk. However, politeness does not have to be strategic (see Ide 1989). Certain speech events call for a polite use of language. These include ceremonial, business, and public situations among others. In these social situations the participants generally use the masu form. 1 0 Example (6) comes from a Diet interpellation, in which the speakers mainly use the masu form. 11 All three speakers here use the masu form at the end of the clause. The masu form is an appropriate speech style in a public speech event such as the Diet interpellation where the participants are expected to show interpersonal distance as well as the social role of a member of the Diet. (6)

Κ = Konishi, opposition party member I = Inoki, opposition party member Ka = Kaifu, Prime Minister

1

— > Κ: Ee, Inoki giin ni tatchi itashimasu. Fl Inoki M.P. to touch do 'Well, (I)'11 give the floor to Inoki M.P.'

2

— >1: ( ) konn chotto kinchoo shite this a little tense do '( ) (I) am a little tense but.'

3 4

orimasu exist

ee, saru juugatsu no juuyokka ni Aizu Wakamatsu FI past October LK fourteenth on Aizu Wakamatsu ni okimashite, ee gookan ni osowareru to iu jiken in

Fl ruffian by attacked

Q

say incident

5

— > in aimashite, ookega ο shimashita ga with meet serious injury Ο did but 'Uhh, last October 14, in Aizu Wakamatsu, uh, I was attacked by a ruffian, and was seriously injured but'

6

—>ee taihen goshinpai

ο

okakeshimashita

ga

FI very worry Ο caused but 'uh, (I) caused you to worry about me greatly but'

7

ga but

maa

kyoo

kono

yoo ni genki

FI

today

this

way

health

ni

natte, become

14

8

9 10 11

Haruko Minegishi Cook

—>tanin ο shite mairimashita. leaving hospital Ο do existed 'well, today I was discharged from the hospital having recovered like this.' ee supootsu

wa kokkyoo

FI

sport

Τ

ο

chooetsu

shite

sekai jinrui

Ο

rise above

do

world

—>bunka

da

to

ya

hada no iro

national boundary and

skin

LK color

ga moteru kyooyuu no

human

S have common

LK

omoimasu.

culture COP Q think 'uh, (I) think that sports is a culture commonly shared by all human beings regardless of nationality and race.'

12

— >Ka: Saisho ni giin no first

13

14

itaitashii atama no shiroi

M.P. LK painful

head

LK white

hootai bandage

—>o mite kokoro kara omimai mooshiagemasu (.) Ο see heart from sympathy say 'First, seeing the white bandage on your head, I sincerely express my sympathy.' Geejutsu,

bunka no suijun

wa dore kurai ka to

art

culture LK standard

Τ

how much INT Q

15

— > osshaimasu keredomo say but '(You)'ve asked how high is the standard of art and culture (of Japan) but'

16

kuni ni yotte bunka, supootsu, sorezore country depending culture sports each — > rekishi ya dentoo ya tokui shumoku ga aru to

17

history and

tradition and

special item

s exist Q

omoimasu. think '(I) think that every country has a different history, tradition, and specialties in sports.' The masu form can be used strategically. In line 1 of example (7), which comes from the neighborhood quarrel, Mr. Kobayashi does not use the masu form in asking the addressee if he is the manager of the apartment. Once Mr. Suzuki has said that he is in fact the manager (line 2), Mr.

Japanese verbal suffixes

15

Kobayashi shifts his speech to the masu form (lines 6 and 8), even when he is quarreling with Mr. Suzuki. Mr. Kobayashi's use of the masu form can be interpreted as an indication of socially acceptable politeness on his part as a younger tenant toward the older manager of his apartment (i.e. indexing of interpersonal as well as intrapersonal distance). Such conduct is considered appropriate in Japanese society. (7) K=Kobayashi, the tenant; S=Suzuki, landlord 1

K: Otaku wa

koko

no

you Τ here LK 'Are you the manager here?'

kanrinin

na

no?

manager

COP

FP

yo.

Yatten no, do FP We are doing it.'

uchi we

wa. Τ

Kotchi wa mada zembu yarn this side Τ still all do 'We are still doing all of it but'

kedo but

sa

Hito

no uchi

made

sa, konna

oitette sa (

other

LK house

even

FP this

leave

S: Atarimae of course 'Of course.

FP

FP

FP

'You leave it [trash] even at another's house.'

[ K:

Dakara ( so 'So, ( )

~ > dashitaku

nai

put out NEG Ί wouldn't like

η

) dakara so so'

desu

kedo but to leave it but' COP

S: Un 'Uhuh'

8

K: Hikkoshite ashita wa inai kara move tomorrow Τ exist NEG so ~>shooganai ja nai desu ka. help NEG NEG COP INT 'Since I will move and won't be here tomorrow, isn't it the case that it can't be helped?'

16

Haruko Minegishi Cook

In sum, typically, the encoded meaning of the masu form, distance, is interpreted as politeness in social contexts in which politeness is socially expected.

3.2. The conventional meaning of the plain form Contrasted with the masu form, the plain form is conventionally perceived as an intimate speech form. This is accounted for in the following way. The encoded meaning of the plain form, i.e. a lack of distance between the speaker and addressee, can imply interpersonal closeness when used in conversation in which intimacy or closeness is expected among interlocutors. Example (8) illustrates the use of the plain form in a family conversation. Notice the lack of masu forms in (8). 12 Here the parents and two male children, Κ (7 years old) and Η (5 years old) are around the table after dinner. The mother addresses the younger child Η while the father speaks to the older child K. Except for in H's utterance, which ends with the gerund form -te, in all utterances the verbs are in the plain form. (8) M: Hiro warui kedo, ano Hiro bad but Fl 'Hiro, (I) bother (you) but, uh.' F: Ano, ojiichan ga sugoku kitte motten no shitterul Fl grandpa S a lot stamp have NOM know 'Well, do (you) know grandpa has a lot of stamps?'

M: ((to H))

Mame denki motte kite kurerul bean light bring give (me) 'Will (you) bring the small light?'

F: Suggoi takai kitte motteru yo, ippai. very expensive stamp have FP a lot '(He) has very expensive stamps, a lot of them.' K: Nani ( what 'What (

) )

motteru? have does he have?'

Japanese verbal suffixes

H: (

)

'(

)

17

motte kite. bring come bring and.'

In sum, the encoded meaning of the plain form can imply interpersonal closeness when used in conversation in which intimacy or closeness is expected. 3.3. Non-conventional situational meanings Besides the conventional meanings, there are numerous non-conventional situational meanings, of which native speakers are normally not consciously aware. Some situational meanings can be evoked by the contrast created by switching from one form to the other. In other cases, situational meanings can be derived from the overall pattern of the combination of the two forms and a given social situation. We should note, however, that since the social and linguistic contexts in which a particular form occurs are numerous, the following examples are by no means exhaustive and that in many instances multiple situational meanings are indexed simultaneously.

3.3.1. Situational meanings of the masu form

3.3.1.1. Social roles In family conversations, the norm is to use plain forms. However, as mentioned in section 2.1., occasionally the members switch to masu forms, which evoke certain social roles of speakers. These instances cannot be adequately described if the masu form is merely a marker of polite speech. Certain of the social roles defined by their rights, duties and responsibilities are associated with a mode of self which is distant from one's own self. Thus, the masu forms can index certain social roles. For example, the parents' use of the masu form can evoke their social responsibilities (e.g. to teach children what they need to know in order to become members of society). In (9) a mother and her child are talking about the time that the child should go to bed. The mother uses desu (the masu form of the copula) when she tells the child that he must go to bed at eight o'clock the day before he goes to school.

18

(9)

Haruko Minegishi Cook

Mother

Child Κ Kuji kana? nine wonder '(I) wonder whether (it's) nine.'

Kuji de ii no, neru nine COP good FP sleep 'Is nine o'clock OK, the bedtime?'

jikoku? time Un. 'Uhn.'

Natsuyasumi η nattara summer vacation become when 'When the summer vacation comes.'

ne. FP Un. 'Uhn.'

Sono kawashi, gakkoo iku mae instead school go before — >no hi wa hachiji desu LK day Τ eight o'clock COP 'Instead, before the days (you) go to school, it's eight o'clock.'

yo. FP

In (9), the masu form indexes the aspect of the mother's social role as the person who brings up the children. In this light, the use of masu forms in business or career contexts can be seen as an index of social role. It is possible to interpret the politicians' use of masu forms in the Diet interpellation in example (7) as an index of social role as well.

3.3.1.2. Negative affect Hinds (1976) mentions that when the speaker expresses displeasure at the topic of conversation s/he often uses masu forms. In my family conversation data, parents often use them when they scold a child. When closeness is the norm between the speaker and the addressee, masu forms can mark negative affect, for sudden distancing from the intimate relationship can create coldness. The conversation in (10) illustrates negative affect expressed by masu forms. In (10) the mother is angry at child Η because he is taking too long to finish his dinner and the mother has become impatient.

Japanese verbal suffixes

19

Here the mother's use of the masu form indicates negative affect. (10)

Mother

— >Mama saki nemasu kara Mom before sleep so 'I'm going to bed before (you).'

Child Η ne. FP Uun. 'Uhn.'

Hiroaki matte tara yo ga akechau. Hiroaki wait if night S lift 'If (I) waited for you, the day would break.' In sum, the sudden creation of psychological distance by masu forms in an intimate family conversation can index negative affect. These uses of masu forms cannot be explained coherently if masu forms only mark polite speech.

3.3.2. Situational meanings of the plain form The encoded meaning of the plain form can also index a variety of situational meanings contrasted with the meaning of the masu form and a given social situation. 3.3.2.1. Immediate and spontaneous reaction Occasional use of the plain form in the midst of a social context in which primarily the masu form is used indexes a value opposite of the one that is indexed by the masu form. Thus, when the masu form indexes a social role in a particular context, the occasional use of the plain form in that context indexes the lack of the social role. Without the facade of a social role, the speaker behaves spontaneously. Maynard (1993) also reports that the plain form is used to mark immediacy and vividness of a situation. In my TV interview data, in which the host interviews a chef in a yakitori restaurant in Tokyo, several times the interviewer switches from masu to plain forms. He does so consistently when he expresses himself spontaneously. This use of plain forms suggests that the interviewer momentarily steps out of his social role as a TV interviewer and talks spontaneously in the innate mode of self. Consider example (11). Here, the chef is demonstrating the technique of piercing a stick into a piece of chicken

20

Haruko Minegishi Cook

and green onions. Just as he has completed the demonstration, the interviewer describes the technique that he has just observed using the plain form. This is the interviewer's immediate and spontaneous reaction to the demonstration. (11)

C=Chef; I=Interviewer

1

C: Chotto mawasu little turn 'Turn it a little bit.' Kyokutan

η NOM

desu COP

ni mawasanakute mo,

ne FP

kagen

de

extremely turn NEG also degree COP 'Even though you don't turn it a lot, by adjustment.'

2

I:

Haa 'Yes'

C: warenai yoon break NEG 'it won't break, so' 4

I:

5

C: Mawasanaito

narimasu become

kara so

Haa 'Yes' hosoi

negi

slender onion

wa koo

kushi

ga futoi to

τ

stick

S

turn NEG

if

toku

ni warete shimaimasu

this

big

if

shi

particular in break end up and 'If you don't turn it, slender green onions, in particular when a stick is big, will break and'

6—>

I:

Sasu shunkan ni mawasu. pierce moment at turn 'The moment you pierce (the onion), you turn (it).'

The interviewer's use of a plain form in line 6 could be a professionally calculated technique so as to give the effect of fresh sensation. It is not possible to tell whether the interviewer genuinely stepped out of his professional role or whether he purposely did so for effect. This distinction is not an important one here. In either case, the effect of switching from masu to plain in professional talk can index an immediate and spontaneous reaction.

Japanese verbal suffixes

21

3.3.2.2. Spontaneous assertion of the speaker's thought In the data of the Diet interpellation, which is mainly carried on in masu forms, one of the major functions of plain forms in this speech event is to assert the speaker's position. In these instances, the utterance in plain forms is followed by a meta-sentence commenting on the assertion. The commenting sentence contains an anaphoric form, which refers to the speaker's belief or desire framed by the plain form. Since an anaphoric form refers to the sentence as if it were 'at a distance' from the speaker, it backgrounds the information referred to. This discourse structure of backgrounding information can be schematized in (12): (12)

Discourse structure of backgrounding information Utterance 1:

Speaker's belief or desire typically framed by a plain form 13 (backgrounded by Utterance 2)

Utterance 2:

Speaker's meta-comment on Utterance 1 ending in a masu form (Utterance 1 is referred to by an anaphoric device in Utterance 2)

The speaker asserts his belief or desire in Utterance 1 in the plain form, and an anaphoric device in Utterance 2 backgrounds this assertion. Utterance 2 is typically marked with the normative form for a given social context. In example (13) the utterances ending in plain forms are followed by anaphoric devices such as soo iu 'such'. Note that the strong assertion in Utterance 1 is followed by masu forms, which indicate politeness in the Diet interpellation. (13)

Finance Minister Hashimoto

Utterance 1 1 ...rironteki

na

theoretical

2 3

genshuu

no ooku

no bubun to iu

reduction

LK much

LK part

Q

mono

say thing

ga shoohisha

ga sore

dake

yasui

kakaku de mono

ο

S

S

that

only

cheap

price

Ο

consumer

at

thing

koonyuu

suru

koto

ga dekiru to iu jootai

ni

purchase

do

NOM

S

to

can

Q

say situation

22

Haruko Minegishi Cook

4

~>tenka ο itashite oru. (.) transfer Ο do exist '... a large part of the theoretical tax reduction is reflected in the situation in which consumers can purchase goods at a lower price.'

Utterance 2 5

Soo such

iu jittai situation

de gozaimasu. COP

'Such is the actual situation.' Here, the finance minister is defending a recent sales tax bill by stating the government position that introducing sales tax leads to overall tax reduction. This is the official belief of the finance minister. In Utterance 1 (lines 1-4) he asserts this position using the plain form itashite oru 'are doing1 in line 4 and in Utterance 2 (line 5) his opinion is backgrounded by the anaphoric device soo iu 'such'. Morphologically marking asserted information as the spontaneous expression of a speaker's belief or desire (with plain forms) and backgrounding this asserted information in discourse structure achieves two communicative goals: letting the addressee know the speaker's spontaneous assertion and redressing the face-threatening act (Brown and Levinson 1987) of the assertion by backgrounding it. Furthermore, the spontaneous expression of the speaker's voice in Utterance 1 is framed by a socially more appropriate masu form in Utterance 2. As long as the framing verb in Utterance 2 is socially appropriate, the form of the verbal in Utterance 1 has little effect on social appropriateness of the speech event. In this sense, Utterance 1 is similar to an embedded sentence. From the above discussion, we can conclude that for these politicians backgrounding information in discourse structure is a strategy to politely convey their views and opinions on record. Discourse structure of backpounding information is not only used in the Diet interpellation. Maynard (1991, 1993) found that in literary essays, written predominantly in masu forms, occasionally sentences end in a plain form, which is backgrounded in the following comment clause. She proposes that the use of plain forms can organize discourse structure in literary essays. Maynard's finding is not contradictory to the present finding. Indeed, these findings provide a good illustration of the very nature of deictic features, which function differently according to the specific aspects of the given context.

Japanese verbal suffixes

23

3.3.2.3. Higher social status A higher status person typically does not keep a distance when talking to a lower status person whereas a lower status person typically keeps a distance when talking to a higher status person (see Goffman 1967). Thus the nonreciprocal use of the plain form can index a socially higher status and that of the masu form socially lower status. Consider a part of example (7), which is repeated here as (14). In this example, Mr. Kobayashi, the tenant, shifts to masu forms once he realizes that Mr. Suzuki is the apartment manager. The latter, however, does not reciprocate and keeps using plain forms in line 7. By so doing, he implicates that his social status is higher than that of Mr. Kobayashi. (14)K=Kobayashi, the tenant; S=Suzuki, the landlord 1

K: Otaku wa koko no kanrinin you Τ here LK manager 'Are you the manager here?'

na COP

2

S: Atarimae yo. Yatten no, uchi of course FP do FP we O f course. We are doing it.'

wa. Τ

3

Kotchi

wa mada

zembu yarn

this side τ still all do 'We are still doing all of it but'

no? FP

kedo

sa

but

FP

((a f e w more lines by Suzuki))

4 Κ·.— >dashitaku nai

η

desu

kedo

put out NEG COP but Ί wouldn't like to leave it but'

5 S:

Un 'Uhuh'

6 K:

Hikkoshite ashita

wa

move

Τ

tomorrow

— > shooganai ja

nai

help NEG

NEG

inai exist

desu COP

kara NEG

so

ka. INT

'Since I will move and won't be here tomorrow, isn't it the case that it can't be helped?'

24

Haruko Minegishi Cook

7 S:

Dakarasoo iu toki wa chanto hanashite kurereba sa, So such time Τ properly speak give FP uchi mo chanto soo iu toki yarn kedo sa. we also proper such time give but FP 'So at such a time if (you) told (us), we'd certainly do that for you at such a time.'

In non-reciprocal uses of masu and plain forms, the one who uses plain forms, which do not index distance, implicates a higher status, and the one who uses masu forms, which index distance, implicates lower status.

4. Conclusion In sum, in this paper I have shown that the spatial notions of distance and proximity are the core of the social deictics of addressee honorifics and the non-honorific counterparts. Although the previous researchers have proposed that the addressee honorifics mark distance, they have not considered distance between the speaker's own self and his/her social role. I have demonstrated that such a distance/proximity must be part of the meaning of the masu and plain forms. I have also shown how these spatial notions evoke various situational meanings (i.e. implicatures) in different social contexts, of which at least one is the conventional meaning. I claim that these spatial notions are directly linked to the linguistic forms even when their implicatures change from context to context. The present analysis of these forms in terms of the encoded and situational meanings has at least two advantages over the previous analyses: first, it can systematically account for various situational meanings, including the conventional ones, some of which were not accounted for by the claim that the masu form is a polite speech marker and the plain form is an intimate speech marker; secondly, it provides a better understanding of pragmatic processes in which various meanings associated with the forms are indexed.

Notes * 1.

I would like to thank Patricia Clancy, Shoichi Iwasaki, Naomi McGloin and an anonymous reader for valuable comments on earlier versions of this paper. Aiiy shortcomings that remain are my responsibility. As I will mention in section 3.2., the suffix masu consists of the so-called honorific morpheme mas- (des- for the copula) and the present tense marker -u. The plain form is also referred to as the abrupt form,

Japanese verbal suffixes

2.

3.

4.

5. 6.

7.

8. 9.

10. 11.

dictionary form and da style. The following transcription conventions are used: (.) short pause ? rising pitch ((text)) information for which a symbol is not available ( ) incoherent string (word) conjectured string —> position of illustrative element '(word)' a word that does not occur in the Japanese but is necessary in the English translation. Also, for glossing the Japanese data, the following abbreviations are used: various forms of a copula COP filler FI final particle FP interrogative INT linking nominal LK NEG negative NOM nominal izer Ο direct object PAS passive quotative marker Q subject marker S Τ topic marker An anonymous reviewer has suggested that the use of the masu form correlates with addressee-oriented sentences and the plain form with contentoriented sentences. However, as a native speaker linguist, I do not find this suggestion plausible. Since the Japanese masu forms show politeness to the addressee who may not be mentioned in the sentence, Comrie (1976) classifies them as addressee honorifics. In contrast, Comrie refers to the T/V pronouns as referent honorifics, for they show respect to the referent which happens to be the addressee. Natural meaning is sometimes referred to as sentence or literal meaning and non-natural meaning is considered speaker meaning. Quasi-face-to-face conversations such as telephone conversations between speakers in close relationships are marked by a frequent use of the masu form. Since speakers are at a distance when they talk on the telephone, it makes sense in the present analysis that they would often use masu. The mother's use of the masu form in serving food and teaching social norms in family conversations occur frequently in my data. Clancy (1986) also reports similar examples. In Japan trash must be separated into burnable and nonburnable items, and the two types are collected on different days of the week. The plain form is also used in academic and scientific writings. This raises the question of why it is used in these written genres if it indexes a lack of distance. To answer this question, future research needs to examine the cultural meaning of writing in Japanese society. This does not mean that they use only masu forms. As discussed below, even in these situations, they once in a while switch to the plain forms. I will discuss the use of plain forms in the Diet interpellation in section 5.3.2.

25

26

Haruko Minegishi Cook

12.

As discussed in examples (2) and (4), this does not mean that in the family conversation masu forms do not occur. Utterance 1 may also be framed by the quotative to.

13.

References Brown, Penelope and Stephen Levinson 1978 Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press. Clancy, Patricia M. 1986 The acquisition of Japanese. In: Slobin, Dan (ed.), The Cross-Linguistic Study of Language Acquisition. Hillsdale, New Jersey: Erlbaum, 373524. Goffinan, Erving 1976 Interaction Ritual: Essays on Face-to-Face Behavior. New York: Pantheon Books. Goldstein, Bernice and Tamura, K. 1975 Japan and America. Tokyo: Charles E. Tuttle. Grice, Paul 1975 Logic and conversation. In: Cole, Peter and Jerry Morgan (eds). Syntax and Semantics 3: Speech Acts. New York: Academic Press, 41-59. Hai man, John 1983 Iconic and economic motivation. Language 59: 781-819. Hanks, William 1990 Referential Practice: Language and Lived Space among the Maya. Chicago: University of Chicago Press. 1992 The indexical ground of deictic reference. In: Durand, Alessandro and Charles Goodwin (eds). Rethinking Context. Cambridge: Cambridge University Press, 43-76. Harada, S. I. 1976 Honorifics. In: Shibatani, Masayoshi (ed.), Syntax and Semantics S.Japanese Generative Grammar. New York: Academic Press, 499-561. Hinds, John 1976 Aspects of Japanese Discourse Structure. Tokyo: Kaitakusha. 1978 Conversational structure: An investigation based on Japanese interview discourse. In: Hinds, John and Irwin Howard (eds). Problems in Japanese Syntax and Semantics. Tokyo: Kaitakusha, 79-121. Ide, Sachiko 1983 Japanese sociolinguistics: Politeness and women's language. Lingua 57: 357-385. 1989 Formal forms and discernment: Two neglected aspects of universals of linguistic politeness. Multilingua 8: 223-248. Ikuta, Shoko 1983 Speech level shift and conversational strategy in Japanese discourse. Language Sciences 5: 37-53. Ikuta, Shoko and Sachiko Ide 1983 Shakai gengogaku ni okeru danwa kenkyuu (Discouse study in sociolinguistics). Gengo (Language) Vol. 12 No. 12. Tokyo: Taishuukan, 77-84. Jorden, Eleanor and Mari Nöda 1987 Japanese: The Spoken Language, Part 1. New Haven: Yale University Press.

Japanese verbal suffixes

Lakoff, Robin 1973

27

The logic of politeness: Or minding your p's and q's. Proceedings of the Ninth Regional Meetings of the Chicago Linguistics Society, 292-305. 1974 Remarks on this and that. Proceedings of the Ninth Regional Meetings of the Chicago Linguistics Society, 435-456. Levinson, Stephen 1979 Pragmatics and social deixis. Proceedings of the Fifth Annual Meeting of the Berkeley Linguistic Society, 206-223. 1983 Pragmatics. Cambridge: Cambridge University Press. Martin, Samuel 1964 Speech levels in Japan and Korea. In: Dell Hymes (ed.). Language in Culture and Society. New York: Harper and Row, 407-415. Maynard, Senko 1991 Pragmatics of discourse modality: A case of da and desu/masu forms in Japanese. Journal of Pragmatics 15: 551-582. 1993 Discourse Modality: Subjectivity, Emotion and Voice in the Japanese Language. Amsterdam: Benjamins. Neustupny, Jiri V. 1978 Post-Structural Approaches to Language: Language Theory in Japanese Context. Tokyo: University of Tokyo Press. Scollon, Ron and Suzanne Β. K. Scollon 1981 Narrative, Literacy and Face in Interethnic Communication. Norwood, New Jersey: Ablex. Shibatani, Masayoshi 1990 The Languages of Japan. Cambridge: Cambridge University Press.

Demonstratives as locating expressions Walter De Mulder

0. Introduction Near the end of the comic Les Bijoux de la Castafiore by Remi (on top of p. 59), there is a series of drawings in which Tintin tries to indicate to his companions, captain Haddock and the two detectives Dupondt, the exact place where an emerald, which had been stolen from the famous opera singer Castafiore, has been hidden by a magpie. In order to do so, he points upwards and says: 'Regardez la-haut! ... C'est la que se trouve certainement la clef du mystere!' (Look up there! That's certainly where the key to the mistery is to be found!). It is interesting to note the reactions of both the two detectives Dupondt and captain Haddock: Dupont: La-haut? (Up there?) Dupond: Ou ςα, la-haut? (Where then, up there?) Haddock: Oui, ou Ιά, ςα-haut? (Yes, where there, that-up?) This passage shows that a pointing gesture as such is not sufficient to identify a referent or its location. It is only when Tintin, in the following drawing, adds a nominal content to his gesture, by saying: 'La-haut, dans cepeuplier ...' (Up there, in that poplar), that his interlocutors can identify the intended referent, as is confirmed by Haddock's answer at this point: 'Dans ce peuplier? ... Tout ce que je vois, c'est un nidi' (In that poplar? ... All I see is a nest!). The purpose of this article is, first, to confirm what this passage suggests: contrary to what is suggested by their very name, demonstratives are not pointers, they do not indicate where in the context their referent is to be found. This thesis has already been defended by Kleiber (1983a) with respect to the French demonstrative determiner ce and I will start by resuming his arguments. However, I will then show that his analysis of the French demonstrative noun phrase ce Ν ('this/that N') as ce + est + un Ν ('this/that + is + an N') is also problematic and propose another analysis of these French demonstrative noun phrases used to refer to objects present before the eyes: it will be shown that the only meaning they convey is that

30

Walter De Mulder

their referent is to be identified by means of an element of their context of utterance. They are thus opaque deictic words (Kleiber 1983a, 1986a) and to identify their referent, one has to combine their linguistic meaning with perceptual interpretive strategies (Searle 1983; Roberts 1993).

1. Demonstratives are not pointers To say that demonstratives are pointers is to say that demonstratives, when used to refer to an object right before the eyes, convey an indication as to where their referent is to be found. This idea is, of course, based on the simple observation that I can say: "This is the Wolfsburg", with some kind of ostensive gesture, or even, in a lot of cases, without such a pointing gesture, and that you then readily identify what I am referring to. There is, however, an apparent objection to this conclusion: if demonstratives were pointers, if they were to localize their referent by themselves, then the ostensive gesture would be entirely redundant and we should be able to leave it out without any problems. This is not the case, however, at least in French. 1 If I say: 'Regarde cette fenetre' (Look at that window), 1 have to point at one of the windows in this room, otherwise you would not be able to single it out (Kleiber 1984: 104-105). 2 This objection leads to a revision of the theory that demonstratives are pointers. In its new version, the theory accepts that demonstratives do not by themselves localize their referent and that the pointing gesture is indispensable; the revised theory then holds that the meaning of the demonstrative is the sense of the demonstration that accompanies it. This is an interesting idea, especially since it enables one to explain a demonstrative version of the traditional Fregean puzzle of identity. 3 Indeed, if the meaning of a demonstrative is the sense of the demonstration that accompanies it, a sentence like 'this is that' or 'this ship is this ship' (where is expresses identity) can be informative, since my first demonstration can be to the stem of the ship and my second to its stern. The sentence can then be informative if, for instance, the ship is the aircraft-carrier Enterprise and there is a building between the interlocutors and the Enterprise, such that we only see stem and stern and not what is in between them and connects them (Kaplan 1977). The revised theory thus presents some advantages. Nevertheless, such a Fregean theory of demonstratives also has its problems. It would make demonstratives infinitely ambiguous, since one would have to assign to them as many meanings as there are demonstrations. To a linguist, it seems more appealing to hold that the meaning of the demonstratives is always the same, and that it is only the demonstration that changes (Kleiber 1983a:

Demonstratives as locating expressions

31

106). This means, in fact, that the meaning of a demonstrative is not to be defined in the Fregean way, as the way it identifies its referent, if a demonstrative noun phrase indeed identifies its referent by the accompanying demonstration. If this were true, a Fregean would have to hold that the meaning changes constantly, whereas the linguistic meaning of the demonstrative is expected to stay the same in all its uses (Bürge 1979, 1990). There is another problem with the Fregean idea that the meaning of a demonstrative is the sense of the accompanying demonstration: it would make true propositions that are in fact false and vice versa. This is shown by the following thought-experiment devized by Kaplan (1977). Suppose I point to Bill Clinton and say: 'That man now lives in Washington D.C.'. What I say, the proposition this sentence expresses, is certainly true at this moment. Now suppose that I point to Frangois Mitterrand instead, and utter the same sentence: "That man now lives in Washington D.C.". What is said now, the proposition expressed, is false. However, suppose that Mitterrand is disguised as Bill Clinton and plays the saxophone. Would the proposition expressed by "That man now lives in Washington D.C." be true in such a case? Kaplan holds it would not, since, of course, Mitterrand lives in Paris, France, and not in Washington. The problem is that a Fregean theory could make this proposition true, since it would hold that the sentence expresses a proposition that can be circumscribed as 'That man - that is, the man I am now pointing at and that is identified by the sense of my pointing - now lives in Washington D.C.'. This proposition seems to be true if Mitterrand has successfully disguised himself as Bill Clinton and has fooled us into thinking that the speaker is in fact talking about Clinton, whereas in fact he is talking about Mitterrand, who does not live in Washington D.C. This means, according to Kaplan (1977: 512 and 516-517), that the Fregean approach makes true a proposition that is clearly false and, thus, that it has to be rejected. There is one last objection to the Fregean theory: if the meaning of the demonstrative were the sense of the associated demonstration, I should have the feeling that when a speaker does not point right, he makes a bad use of the demonstrative. In other terms, when I say: 'That table is white' and point in fact some five meters beside it, you should have the feeling that I don't know the meaning of the demonstrative. In fact, however, you will just think that I point wrong (Kleiber 1983a: 106). We must conclude, then, that we must reject the Fregean version of the theory that demonstratives are pointers that localize their referent. There is, however, still a way to defend, in a very weak sense, the idea that demonstratives are pointers. Indeed, one could claim that demonstratives do not really localize their referent, but that they convey a more vague meaning,

32

Walter De Mulder

something like 'the X which I am now pointing at', where the demonstration's role would be to act as a complement, to make this meaning more precise (Kleiber 1983a: 107). The demonstrative only signals to the hearer that the speaker is showing or localizing an object. However, this idea of 'showing' or 'localizing' a referent only applies naturally to the context considered here, that is, to uses of demonstratives to refer to objects present before the eyes, whereas demonstratives are used in a lot of other contexts: they refer to abstract objects, as in this love is killing me, to temporal objects as in this century is really decadent, or they can be used as anaphors, in which case they mean something like the X just mentioned, and so on. In most of these cases, it is difficult to understand exactly what it means to 'point' at the object identified by the demonstrative noun phrase. One could of course hold that these are typical cases where a spatial meaning is used to talk about more abstract domains and describe the metonymic and metaphorical processes underlying such transfers. Nevertheless, what one ends up with are cases where the demonstrative no longer literally indicates that the speaker points at a referent. The meaning of the demonstrative in those abstract and anaphoric uses is, rather, to indicate that the object referred to is to be found by making use of elements in the context of use of the demonstrative. But then, this meaning can also be attributed to the other deictic expressions such as /, here, now, and the like: they all indicate that the referent is to be found by using elements of the context of their use. Kleiber (1983a: 113) thus rightly concludes that the 'weak version' of the theory that demonstratives are pointers eliminates the difference between demonstratives and other indexical expressions. There is a difference, though, between demonstrative determiners such as the French ce and other indexicals. It can readily be observed that, contrary to the demonstrative determiner, je ('I'), maintenant ('now'), ici ('here'), and so on, clearly indicate what element of the context they refer to. Omitting further complications, it can be said that je refers to the speaker, maintenant to the moment of speaking, and ici to the place of speaking. The demonstrative determiners do not seem to convey such a clear rule, especially not in French, where, contrary to other languages, there is only one demonstrative determiner, ce, and not two or more (like this and that in English or even more elaborate systems in Eskimo etc.). 4 Although there are slight differences in the distinctions drawn by different authors, let us say, for the moment, that demonstratives are opaque deictic expressions, whereas the other deictic expressions are transparent, meaning that they convey a rule that uniquely gives their referent once the context of their use is known, whereas demonstratives do not convey such univocal information. A demonstrative, then, functions as a signal, a bell that goes and tells us that there is a referent to be identified by elements of its context of

Demonstratives as locating expressions

33

use. But the demonstrative by itself does not tell us where the referent is to be found, contrary to the other deictic expressions (Kleiber 1983a).

2. Demonstrations are not localizers Kleiber (1983a) has shown that demonstratives are not pointers. Nevertheless, Kleiber (1984) analyzes demonstrative noun phrases like ce Ν as Ce + est + un/du Ν and holds that the demonstrative determiner ce seizes the referent as not yet classified ('non-nomme') and that the Ν classifies this referent as (an) X that is of the class denoted by N. Does this mean, then, that the demonstrative, as it were, identifies the referent all by itself, by means of the accompanying pointing gesture or (in texts) 'by proximity' (Corblin 1987)? Kleiber (1984) defends his idea that the demonstrative adjective ce seizes the referent as 'not yet classified' by a comparison with the demonstrative pronoun ce which is opposed to the personal pronoun il in the following examples: (1) (2) (3) (4) (5) (6)

Paul, c 'est un acteur / Paul, c 'est un monstre. Paul, it is an actor / Paul, it is a monster. *Paul, il est un acteur / *Paul, il est un monstre. Paul, he is an actor / Paul, he is a monster. Paul, il est monstrueux / *Paul, c 'est monstrueux. Paul, he is monstrous / Paul, it is monstrous. Paul est acteur / *C'est acteur / II est acteur. Paul is actor / It is actor / He is actor. Ma voiture, c 'est une Peugeot. My car, it is a Peugeot. * Ma voiture, eile est une Peugeot. My car, she is a Peugeot.

What explains the distribution of the pronouns ce and il in these sentences, is the character of the predicate: if the predicate is classifying (X est unN), ce is used; if, on the contrary, the predicate is not classifying (X is Ν), il is used. Kleiber goes on to say that the purpose of a sentence like (1) is to classify an object, i.e. to say that its subject is apprehended as an object that is yet to be classified. Thus, the use of the pronoun ce in such sentences shows that ce presents the object it refers to as an object that is yet to be classified. This is even more clear in (7), where, according to Kleiber (1984: 70), ce cannot be used, precisely because in the first sentence, its referent has already been classified as a car:

34

(7)

Walter De Mulder

J'ai achete une voiture. Cette voiture/Elle est grise. * C'est gris. Ί bought a car This car/She is grey'. 'It is grey'.

But of course, this example also seems to show that ce used as an adjective cannot be compared to ce used as a pronoun. For whereas ce (the pronoun) is excluded, ce Ν (with the demonstrative determiner) is not. In fact, why is the use of the noun phrase possible in this case, if its structure is ce + est + un ΝΊ Doesn't this structure imply an elaborate but vacuous operation whereby the referent is first seized as not classified and is thus declassified, only to be re-classified afterwards in the same class as before? Kleiber's answer to this problem is that, in using a demonstrative noun phrase, the act of classifying is presupposed, whereas in a sentence like c'est gris it is asserted. He shows that the act of classification is presupposed by the traditional tests (8a-c): the negative and the interrogative versions of the second sentence of (7), (8a)

Cette voiture η 'est pas grise 'This car is not grey.'

(8b)

Est-ce que cette voiture est grise? 'Is this car grey?'

still presuppose that the object referred to by the demonstrative noun phrase is a car. Moreover, to refuse the classification, the interlocutor has to use a polemical negation: (8c)

Mais ce η 'est pas une voiture! 'But this is not a car!'

But if the classification is presupposed, how can Kleiber still hold that the determiner ce seizes its referent as not yet classified, just as the pronoun ce does? If it did, ce would seize a referent whose classification is already presupposed and then de-classify it, only to classify it again as it was classified before! In other terms, the problem with (7) still exists. One must conclude then, that the classification is indeed presupposed, but that this means that ce - the determiner, not the pronoun - grasps its referent as already classified. In fact, this is in conformity with the position defended in Kleiber (1990: 252), where it is shown that the descriptive content of the noun Ν in the demonstrative noun phrase has a role to play in the identi-

Demonstratives

as locating expressions

35

fication of the referent of this demonstrative noun phrase: it must be possible to consider the referent in some relevant way as an N. This approach enables us to solve the enigma of (7): (7)

J'ai achete une voiture. Cette voiture/Elle est grise. * C'est gris.

The demonstrative determiner ce now no longer has to 'seize the referent' of itself. Its only function is to be a signal, a bell sound, to quote Kleiber (1983a), that the referent is to be identified by elements of its context of use (a demonstration, or, in this case, an element present in the linguistic context). To decide which referent the speaker refers to, the hearer then uses, among other things, the classification already introduced by the speaker. As Kleiber (1990: 252) says, the context must allow the hearer to understand the classification of the referent as N. The advantage of this model is that it does not use an otiose mechanism of de- and re-classification, as Kleiber's (1984) approach still seems to do. Moreover, it allows us to explain all the facts explained by Kleiber, since it does not reject his explanation of ce used as a pronoun; it only rejects the idea of transferring the sense of 'not yet classified' from the use of ce as a pronoun to its use as a determiner. All this confirms what was already suggested by the passage from Tintin at the beginning of this article: it is only when the demonstrative and its associated demonstration are combined with the noun phrase that the referent can be identified. This is not a surprising conclusion, since it can already be found in Wittgenstein (1958: §28): 28. Now one can ostensively define a proper name, the name of a colour, the name of a material, a numeral, the name of a point of the compass and so on. The definition of the number two, 'that is called 'two' - pointing to two nuts - is perfectly exact. - But how can it be defined like that? The person one gives the definition to doesn't know what one wants to call 'two'; he will suppose that 'two' is the name given to this group of nuts! - He may suppose this; but perhaps he does not. He might make the opposite mistake; when I want to assign a name to this group of nuts, he might understand it as a numeral. And he might equally well take the name of a person, of which I give an ostensive definition, as that of a colour, of a race, or even of a point of the compass. That is to say: an ostensive definition can be variously interpreted in every case.

36

Walter De Mulder

29. Perhaps you say: two can only be ostensively defined in this way: 'This number is called 'two'. For the word 'number' here shows what place in language, in grammar, we assign to the word. But this means that the word 'number' must be explained before the ostensive definition can be understood. The word 'number' in the definition does indeed show this place; does show the post at which we station the word.

or in Lyons (1972: 65): Any theory of deixis must surely take account of the fact (much discussed in the philosophical treatments of ostensive definition) that the gesture of pointing of itself will never be able to make clear whether it is some entity, some property of an entity, or some location that the addressee's attention is being directed to.

3. Figure and ground For a demonstrative noun phrase used to refer to an object present before the eyes, the preceding paragraph suggests the following division of labour: (1)

The demonstrative gesture as such does not by itself identify the referent; it only serves to point to a region in space where the referent is to be located. The same function can be fulfilled by additional linguistic information, such as 'Look up there' in the example from Tintin at the beginning of this article.

(2)

The descriptive content of the noun phrase gives a classification of the referent making it possible to distinguish the referent from other objects in the region of space already singled out by the demonstration and accompanying linguistic information.

As has been pointed out by Roberts (1986: 1993a-d), this means that when we use demonstrative noun phrases to refer to objects present before the eyes, we apply the figure-ground model as used in perception to the domain of natural language communication. Indeed, what is proposed here is that the demonstration and, possibly, complementary linguistic information like regardez la-haut, identify a background, whereas the linguistic meaning of the noun gives the information necessary to distinguish a figure against this background. The combination of these elements then allows the hearer to localize the referent.

Demonstratives as locating expressions

37

This is, of course, but another way of saying that demonstratives are grounding predications (Langacker 1991, 1994). However, what interests us here, is to unravel the contribution made by the different components of the use of a demonstrative to the grounding process. Roberts' idea implies that reference by a demonstrative NP is fundamentally a perceptual process: what starts the demonstrative process is the very perception of the demonstrative, its sound, or its written form. It is this perceptual form that is spatio-temporally linked to other elements of the Zeigfeld, as Bühler (1934: II) called it: the spatio-temporal structure surrounding speaker and hearer with all the objects contained in it. As Bühler also held, it is the perception of the demonstrative expression that starts the process of demonstrative reference: the demonstrative 'rings a bell', as Kleiber says; it says that there is a referent to be identified, but does not say of itself where this referent is to be found. This is even more true if one accepts that demonstratives, contrary to other deictic expressions, are opaque in the sense mentioned above: they do not express a rule which leads to their referent. 5 The only rule demonstrative determiners convey is the general deictic rule that the referent is to be identified by means of elements of the context of utterance of the demonstrative determiner itself. This means that demonstratives are self-referring, meaning that, to identify the referent of a demonstrative noun phrase, it is necessary to use its very utterance (Searle 1983: 223; Benveniste 1966). As Searle stresses, this does not mean that they perform a speech act of referring to their own utterance; it only means that they 'show themselves', that, in order to find the referent, one must take into account the very act of uttering them, with its perceptual qualities.6 Now, if perceptual qualities start off the process of demonstrative reference, it should come as no surprise that alongside the phonetic qualities, other, e.g. visual qualities such as pointing gestures, may have a role to play. This tight connection between these different aspects of the context is not brought about by association by contiguity, however (Peirce 19321954: 2.306). It is the result of a process whereby the hearer, upon hearing the demonstrative, constructs a context, actively integrating incoming information from different sources into a unifying scheme (Tracy 1983: 101103). The only function of pointing gestures in this Zeigfeld is to attract attention to a particular region in it; to find the referent, one needs a figure, which is given by the descriptive content of the noun phrase. The perceptual nature of this process implies that the information given by the noun phrase does not function on the model of a definite description: the referent is not picked out because the descriptive content of the noun is true of the referent (Roberts 1993a-d). In Kripke's (1972) and Kaplan's (1977) terms, the descriptive content only serves 'to fix the referent'. It is sufficient for

38

Walter De Mulder

the descriptive content of the noun phrase to enable the hearer to identify the referent in the background, as the preposition and the noun do in Tintin's in that poplar in our very first example; it does not really have to be true of the referent. It only has to be effective in enabling the hearer to perceptually identify the referent. This is nicely illustrated by another passage of Les Bijoux de la Castafiore (p.23), where two journalists walk in the gardens of the castle where Castafiore stays, hoping to find more information about the wedding they think is going to take place between Castafiore and Captain Haddock. Suddenly, they see Professor Tournesol, who is working in the garden. However, they don't realize that it is Tournesol who is before them. The following conversation ensues: Journalist 1: Journalist 2: Journalist 1:

Journalist 2: (Journalist 1: Journalist 2: Journalist 1:

Journalist 2:

Oh! la un jardinier ... Viens, nous allons essayer de lui tirer les vers du nez ... 'D'ac!...' 'Mais ... Ce jardinier ... C'est le professeur Tournesol! ... Celui qui a ete sur la Lüne avec Tintin ...II doit etre dans le secret, lui... 'Certainement!' 'There, a gardener ... Come on, we will try to interrogate him ...' 'OK'

'But ... this gardener ... It's professor Tournesol! ... The one who's been to the moon with Tintin ... He must be in the secret...' 'Sure')

Professor Tournesol is only an occasional gardener, therefore it is not really correct to characterize him as 'a gardener of the castle'. Nevertheless, the qualification works in the context at hand and it is easy to see why: it enables journalist 2 to pick out the referent as a figure against a ground, since the professor looks like a gardener. The preceding example shows, in fact, that the noun phrase does not have to identify the referent as such, but the referent as perceived, as seen from a certain point of view. This is even more clear in the following use of a demonstrative noun phrase in a passage from Les Bijoux de la Castafiore (p. 50), where the two detectives Dupondt explain how, in their view, some gipsies have stolen the emerald of Castafiore: Non seulement, on retrouve chez eux une paire de ciseaux appartenant ä la bonne de madame Castafiore, mais, dans une roulotte ... on d6couvre un singe dress6! ... Or, finalement, le vol de l'imeraude n'a pu etre commis que

Demonstratives as locating expressions

39

par escalade; et encore, par un homme d'une agilit6 prodigieuse ... Cet homme, nous l'avons d6masqu6: c'est le singe! ... Bien sür, toute la bande nie farouchement!

Not only do we find at their place a pair of scissors belonging to the maid of madam Castafiore, but in a trailer, ... we find a dressed monkey! ... Now, in the end, the emerald could only have been stolen after a climb; and even then, by a man of an astonishing limberness ... This man, we have unmasked him: it is the monkey! ... Of course, the whole gang denies vehemently!

Admittedly, this is not a use of a demonstrative noun phrase to refer to an object before the eyes, but a textual use. But it again reveals that what guides the choice of the nominal component is not the referent as it is (a monkey), but the referent as it is known at that point in the discourse, or, rather, as the speaker thinks the hearer knows it and thus, as known in the common ground of speaker and hearer (Clark, Schrueder and Buttrick 1983), as given in the current discourse space (Langacker 1991: 97). As Roberts rightly notes, this has consequences for the thought experiment by which Kaplan meant to criticize a Fregean theory of demonstratives. Recall that, according to Kaplan, such a Fregean theory would identify what is said by 'This man now lives in Washington D.C.' as 'The man I now point at now lives in Washington D.C.'. However, when Mitterrand disguises as Clinton, this analysis makes the sentence true, whereas in fact it is false that Mitterrand now lives in Washington D.C., which is the proposition the sentence expresses when Mitterrand is disguised as Clinton. But is Kaplan right? Is 'This man now lives in Washington D.C.' false with Mitterrand disguised as Clinton? Only, in fact, when one already knows that the individual pointed at is Mitterrand, and not Clinton, and when one holds that it is this individual that is part of the proposition expressed. But this knowledge is not given by the use of the demonstrative noun phrase as such. If all that is used to find the referent of the demonstrative noun phrase is perceptual information, then this sentence can be true for the hearer. 7 Let us resume: demonstrative determiners of themselves do not identify or localize the referent; they only signal that the referent is to be identified by means of elements of the context. These elements are, first of all, the perceptual aspects of the use of the demonstrative (its physical sound and the accompanying gestures), and, second, the noun that follows the determiner. The determiner of itself only signals that there is a referent to be identified. The spatial pointing gesture tells us that the referent is to be identified in a scene that is perceptually given and that is, thus, structured

40

Walter De Mulder

along a figure/ground model. This perceptual scene, the Zeigfeld, is not objectively given. As was already pointed out above, it is already itself a product of mental constructive activity, where incoming information is integrated according to a unifying frame (Tracy 1983: 101-103). Such frames or schemes can be seen to underly our dispositions to act and to interpret actions in a particular way; they are implicitly present in our practices, largely unreflective and 'embodied' (Hanks 1990). 8 Such schemes are used at different levels: to interpret our initial example, with Tintin pointing and uttering 'Look up there!', at least two schemes must be used. The first allows the detectives Dupondt to interpret Tintin's bodily activity as a pointing gesture - this is why they look in the direction of the pointing finger and not to the finger itself. The second allows them to assign an interpretation to the indication given by up there. It can be identified as a viewer-centered frame of reference: Tintin points above with respect to his body-centered coordinate system. Because this frame of reference does not allow the two Dupondt to find the referent, an indication is given with a more 'objective' value: 'Up there, in that poplar'. The word poplar introduces an object-centered frame of reference, where up there can be interpreted with respect to the intrinsic axes of the object (Carlson-Radvansky and Irwin 1993: 224, Vandeloise 1986: 89-106). Not only does this show that we use schemes in interpreting pointing gestures and referential expressions, the passage also demonstrates that the choice of scheme is a fundamentally conversational enterprise: the speaker must use a frame of reference that he can assume to be shared with his hearer (Clark and Wilkes-Gibbs 1986). And Hanks (1990) has shown that the schemes themselves are already social products, the results of communicative activities. The schemes used to interpret demonstrative noun phrases must not be limited to the frames of reference or of bodily movement of the preceding paragraph. All kinds of general knowledge frames can be made accessible by the content of the noun phrase and the context and used by the interlocutors. This is nicely shown by an example of Kleiber (1987). Suppose two participants for the LAUD-conference at Duisburg are waiting for the train that will take them from Antwerp to Duisburg; however, the train is late. In such a situation, one can say to the other: 'Ce train a toujours du retard' (This train is always late). In this situation, there is no element in the context that can be classified as train. The interlocutors use elements that are present in the context of utterance (the platform and the hour of arrival of the train) to identify the final referent of this train via a piece of common knowledge (x, which is a train, normally arrives at platform Ρ at hour H). 9

Demonstratives as locating expressions

41

One could of course claim that in such cases the demonstrative noun phrase as such identifies an element of the context, the demonstration, which leads to the final referent via some kind of conversational implicatures. Such an analysis would, however, face the following problem: if this train finds its final referent by first identifying an element of its context of use, then why would the noun phrase provide us directly with a classification of the final referent? Thus, it seems safer to hold that we try to make relevant the classification proposed by the speaker and that, in doing this, we directly identify the referent, without first identifying the demonstratum.10 Consequently, these pieces of common knowledge have to be integrated in the analysis of demonstrative noun phrases. I take it, then, that the speaker will give information that allows the hearer to find the referent in a scene structured by using general knowledge frames. The gestures and some linguistic information, together with the schemes, give the ground, and the noun gives a figure that directs our attention to something that can be classified as Ν. I thus endorse Roberts' idea that in perceptual situations, demonstrative noun phrases identify their referent by using the figure-ground model. However, I do not see this figure-ground model as part of the semantics of the demonstrative noun phrase. The linguistic meaning of a demonstrative is neatly captured by describing them as opaque deictic expressions; the use of the figure-ground model is a consequence of the search for the referent they require. In short, to interpret demonstrative noun phrases used to refer to objects present before the eyes, we have to combine the semantic value of the demonstratives with the strategy we use in visual experience, an idea that has already been defended by Searle (1983: 226) and Husserl (1901: I,§26,89). Does this mean that we use different strategies to interpret demonstratives used in discourse? Not necessarily, since the figure-ground organization is a fundamental feature of our overall cognitive functioning (Langacker 1991: 120). To show that the figure-ground organization associated with demonstratives readily extends to their discourse uses, it is possible to take as a starting point the idea of Kirsner (1979: 358-359) that demonstrative determiners express deixis. This means that they signal that a search for the referent is called for. In this respect, demonstrative determiners must be distinguished from definite articles, which signal that, in the speaker's opinion, the hearer already has the necessary means to distinguish the referent. That the demonstratives have an imperative-like force (Kirsner and Van Heuven 1988: 236), urging the hearer to find the referent, is explained by the figure-ground model as it is interpreted by Roberts. For in his view, when using a demonstrative, the speaker asks the hearer not to match the descriptive content of the referential expression to a list of things, but to use the actions and descriptions accompanying the re-

42

Walter De Mulder

ferential expression as a guide to the referent (Roberts 1993d: 31). The difference between the demonstrative determiner and the definite article can now be reformulated as follows. The definite article signals that the referent can be found by matching its descriptive content with referents that are already given. This implies that the speaker assumes that the hearer already has access to a context where the referent is given. The demonstrative determiner, on the contrary, signals that the descriptive content cannot be matched with referents in a context already given to the hearer. It signals, consequently, that a search for the referent is called for, a search where the hearer has to identify a referent by using the elements of the context to set up a ground and by using the descriptive content of the noun as a figure. This can be illustrated by the following example of Kleiber (1986b: 175177). Whereas in French (9) is acceptable: (9)

Un avion s 'est ecrase hier. L'avion/Cet avion venait de Miami. 'Yesterday, an airplane crashed. The airplane/That airplane came from Miami.'

(10) is not: (10)

Un avion s 'est ecrase hier. *L 'avion relie habituellement Miami a New York. 'Yesterday, an airplane crashed. *The airplane usually connects Miami to New York.'

In this context, the use of a demonstrative, as in (11), is much more natural: (11)

Un avion s'est ecrase hier. Cet avion relie habituellement Miami ά New York. 'An airplane crashed yesterday. This airplane usually connects Miami to New York.'

Kleiber shows that the respective uses of the determiners can be explained by the differences in coherence between the first and the second sentences in these examples. Indeed, in (9), the second sentence is interpreted in the scene set up by the first one: the airplane clearly is the one of which it is true that it crashed yesterday. This is not possible in (10) and (11), however: the use of habituellement (usually) makes it impossible to interpret the second sentence in the scene set up by the first. In other words, the use of the definite article is impossible because the context is not

Demonstratives as locating expressions

43

given, but has to be set up. The demonstrative is more appropriate in this case, since it signals that a search for the referent is needed. And the final effect of the use of a demonstrative noun phrase in this context can be likened to that of the pointing finger: just as the pointing finger singles out a referent, pushing the indexical frame into the background, the use of a demonstrative noun phrase in discourse attracts attention to its referent and pushes the context that allows the identification of the referent into the background (Kleiber 1991: 86). 1 1 The figure-ground model thus allows us to describe the functioning of demonstrative noun phrases in discourse and to explain their use as a marker establishing discourse coherence. 12

4. Conclusions We can conclude that demonstratives are not pointers or localizers, since neither the demonstrative nor the demonstration identify the referent of themselves. Demonstrative noun phrases rather function in the following way: (1)

The French demonstrative determiner is an opaque deictic expression: it signals that the referent is to be identified by using elements of the context of its own utterance, without identifying those elements of itself. It thus signals that the referent is to be searched for and that its identification cannot be taken for granted, as is the case for the definite article.

(2)

When the demonstrative noun phrase is used to refer to an object present before the eyes, this means that the demonstrative is linked to elements of a situation that is modelled, as perceived situation, along a figure-gound model. The linguistic meaning of the demonstrative thus gets combined to the cognitive structure imposed on the perceptual scene.

(3)

In such a case, the demonstration or any other perceptual means (an ostensive sniffling, for instance), or even linguistic information ('Regardez la-haut', Look up there), interpreted in accordance with general knowledge schemes, identifies a region of space that functions as a background, in which the descriptive content of the noun phrase then singles out a figure.

(4)

Since the figure-ground organisation is a fundamental feature of cognitive organization, it is not surprising to find that it can also be

44

Walter De Mulder

used to explain the use of demonstrative noun phrases in discourse, where they have a particular cohesive force, to be distinguished from that associated with definite noun phrases.

Notes 1. 2.

3. 4.

5.

6. 7.

8. 9. 10.

Haviland (1992: 9) suggests that such a redundancy would exist in Guugu Yimidhirr. Let me point out here that one does not always need pointing gestures: what counts is that the referent gets attention. Thus, other devices than pointing, e.g. ostensive sniffling, can attract attention, or the referent can be salient by itself, as when I see one man running down the street and say to a companion: 'That man is in a hurry'. I am, of course, referring to Frege's (1892) famous identification of the morning and the evening star. There is, of course, in French the possibility of adding -ci and -la to mark an opposition that seems roughly parallel to that between this and that, but I am here interested in the most common demonstrative noun phrase in French, which is just ce followed by a noun (+ adjective, relative phrase, etc.). In fact, it is not clear whether there are transparent, non-opaque deictic expressions, if this means that the rule expressed by them is sufficient to find their referent. This is, e.g., shown by Nunberg (1992, 1993), using examples like: I am traditionally allowed to order whatever I like for my last meal, (uttered by a prisoner in death row) To understand the relevance of this sentence, one needs to interpret I as Ί, as a condemned prisoner', a qualification that is not given as such by the rule 7 refers to the utterer of this token of /'. Hence the original title of R6canati's (1979) introduction to pragmatics: La transparence et I'enonciation (Transparency and uttering). Kaplan (1977) holds that sentences containing demonstratives express singular propositions, that is, propositions that contain the individual itself which the demonstrative noun phrase refers to, or, in a slightly weaker version, propositions that one cannot understand if one does not know the individual referred to. This is why he interprets This man now lives in Washington D. C. as Mitterrand (the man himself!) now lives in Washington D. C.. But then, of course, his argument against the Fregean theory is a petitio principii (Kleiber 1983b). Moreover, as shown, e.g. in Nunberg (1992, 1993) and Roberts (1993a-c), it is not part of the semantics of demonstrative noun phrases in natural language that they always express singular propositions. See Hanks (1990) for further references. Nunberg (1978) calls such pieces of common knowledge 'referential functions'; Fauconnier (1984) speaks of 'connectors'. Nunberg (1992: 291-292) points out that there are comparable problems for a Gricean analysis of cases of deferred reference (Quine), as when I say: 'Murdoch bought that for 50 million dolllars' and point to a newspaper copy, whereas what I want to say is that he bought the company that publishes the newspaper. Nunberg rejects an analysis of such deferred references via conversational implicatures, where some kind of metonymy

Demonstratives as locating expressions

11.

12.

45

would link up the demonstrata to their final referents, because in the absence of demonstrata, there would be no (misplaced) literal meaning that could provoke the mechanism of conversational implicatures. However, recent proposals show that conventionalized conversational implicatures can make a contribution even before the level of literal meaning is reached. In the terms of Searle (1979, 1983): even literal meaning requires a background of common 'knowledge' and shared practices. However, for reasons adduced, e.g. in Bosch (1983: 197), Bach (1987), Nunberg (1992/1993), Tasmowski (1990), I reject the idea that demonstratives would express direct reference. Moreover, contrary to Kleiber, I prefer to think of reference in terms of success rather than of truth or satisfaction. Briefly put, reference is a speech act and is a question of speaker's intentions. For the different ways in which discourse uses of demonstratives can be connected to their spatial uses, also see Kirsner (1993).

References Bach, Kent 1987 Thought and Reference. Oxford: Clarendon Press. Benveniste, Emile 1958 De la subjectivit6 dans le langage. In: Benveniste, Emile. Problimes de linguistique generale 1. Paris: Gallimard: 1966, 258-266. Bosch, Peter 1983 'Vagueness' is context-dependence. A solution of the sorites paradox. In: Ballmer, Thomas and Manfred Pinkal (eds). Approaching Vagueness. Dordrecht: Elsevier, 189-210. Bühler, Karl 1934 Sprachtheorie. Die Darstellungsfunktion der Sprache. Jena: Fischer. Bürge, Tyler 1979 Sinning against Frege. The Philosophical Review 88: 398^32. 1990 Frege on sense and linguistic meaning. In: Bell, David and Neil Cooper (eds). The Analytic Tradition. Meaning, Thought and Knowledge. Oxford: Basil Blackwell, 30-60. Carlson-Radvansky, Laura and David Irwin 1993 Frames of reference in vision and language: Where is above? Cognition 46: 223-244. Clark, Herbert, Robert Schreuder and Samuel Buttrick 1983 Common ground and the understanding of demonstrative reference. In: Clark, Herbert (ed.). Arenas of Language Use. Chicago: University of Chicago Press: 78-99. Clark, Herbert and Deanna Wilkes-Gibbs 1986 Referring as a collaborative process. Cognition 22/1: 1-39. Corblin, Francis 1987 Indefini, defi.ni et demonstratif. Genfcve: Droz. Fauconnier, Gilles 1984 Espaces mentaux. Aspects de la construction du sens dans les languesturelles. Paris: Minuit.

46

Walter De Mulder

Frege, Gottlob 1892

Ueber Sinn und Bedeutung. In: Frege, Gottlob. Kleine Schriften. Hildesheim: Georg Olms 1967, 144-162.

Hanks, William Referential Practice. Language and Lived Space among the Maya. 1990 Chicago: Chicago University Press. Haviland, John Anchoring, Iconicity, and Orientation in Guugu Yimidhirr Pointing 1992 Gestures. Cognitive Anthropology Research Group at the Max Planck Institute for Psycholinguistics. Working Paper No. 8. Horn, Lawrence Toward a new taxonomy for pragmatic inference: Q-based and R-based 1984 implicature. In: Schiffrin, Deborah (ed.). Meaning, Form and Use in Context. Linguistic Applications. Georgetown: Georgetown University Press, 11-42. Husserl, Edmund Logische Untersuchungen IUI. Untersuchungen zur Phaenomenologie 1901 und Theorie der Erkenntnis. Den Haag: Martinus Nijhoff: 1984. (=Husserliana XIX/1). Kaplan, David 1977 Demonstratives. An essay on the semantics, logic, metaphysics, and epistemology of demonstratives and other indexicals. In: Almog, Joseph, Perry, John and Howard Wettstein (eds). Themes from Kaplan. Oxford: Oxford University Press, 481-614. Kirsner, Robert 1979 Deixis in discourse: An exploratory quantitative study of the modern Dutch demonstrative adjectives. In: Givön, Talmy (ed.). Discourse and Syntax. New York: Academic Press, 355-375. 1993 From meaning to message in two theories: Cognitive and Saussurean views of the Modern Dutch demonstratives. In: Geiger, Richard and Brygida Rudzka-Ostyn (eds). Conceptualizations and Mental Processing in Language. Berlin: New York, 81-114. Kirsner, Robert and Vincent van Heuven 1988 The significance of demonstrative position in Modern Dutch. Lingua 76: 209-248. Kleiber, Georges 1983a Les ddmonstratifs d6montrent-ils? Sur le sens des adjectifs et pronoms dSmonstratifs. Le frangais moderne 51/2: 99-117. 1983b A propos de l'analyse adjectif dimonstratif = article defini + iliment diictique ou sur I'irr6ductibilit6 des symboles indexicaux. Actes du VI lerne congris international de linguistique et de Philologie romanes (Aix-en-Provence, 29 aoüt - 3 septembre 1983). Vol. 4: 195-212. 1984 Sur la sdmantique des descriptions d6monstratives. Linguisticae Investigations VIII/1: 63-85. 1986a D6ictiques, embrayeurs, 'token-reflexives', symboles indexicaux, etc.: comment les ddfinir? L'information grammatical 30: 3-22. 1986b Adjectif ddmonstratif et article d6fini en anaphore fidfcle. In: David, Jean and Georges Kleiber (eds). Determinants: syntaxe et simantique. Paris: Klincksieck, 169-185. 1987 L'6nigme du Vintimille ou les ddtenninants 'ä quai'. Langue frangaise 75: 107-122.

Demonstratives as locating expressions

47

Kleiber, Georges 1990 Sur l'anaphore d6monstrative. In: Charolles, Michel (ed.)· Le discours. Repräsentation et interpretations. Presses Universitäres de Nancy, 243265. 1991 Du nom propre non modifii au nom propre modifi6: Le cas de la ditermination des noms propres par l'adjectif d6monstratif. Langue frangaise 92: 82-103. Kripke, Saul 1972 Naming and necessity. In: Davidson, Donald and Gilbert Harman (eds). Semantics for Natural Language. Dordrecht: Reidel, 253-355. Langacker, Ronald 1991 Foundations of Cognitive Grammar. Volume II. Descriptive Application. Stanford: Stanford University Press. 1994 Viewing in Cognition and Grammar. Talk at the IAUD-Conference on Language and Space, Duisburg, March 22-26, 1994. Levinson, Steven 1989 A review of relevance. Journal of Linguistics 27: 107-161. Lyons, lohn 1972 Deixis as the source of reference. In: Keenan. Edward (ed.). Formal Semantics of Natural Language. Cambridge: Cambridge University Press, 61-83. Nunberg, Geoffrey 1978 The Pragmatics of Reference. Bloomington: Indiana University Linguistics Club. 1992 Two kinds of indexicality. In: Barker, Chris and David Dowty (eds). Proceedings of the Second Conference on Semantics and Linguistic Theory, May 1-3, 1992. Ohio State University Working Papers in Linguistics No. 40: 283-301. 1993 Indexicality and deixis. Linguistics and Philosophy 16: 1-43. Peirce, Charles 1932-54 Collected Papers of Charles Sanders Peirce. Cambridge: Harvard U.P. Recanati, Frangois 1979 La transparence et l'enonciation. Pour introduire ä la pragmatique. Paris: Minuit. Remi, Georges (Herg6) 1963 Les aventures de Tintin. Les bijoux de la Castaflore. Tournai: Casterman. Roberts, Lawrence 1986 The Figure-Ground Model for the explanation of the determination of indexical reference. Synthese 68: 441-486. 1993a The foundations of Kaplan's Direct Reference Theory for demonstratives (Manuscript). To be published in: Philosophia 23. 1993b Pure Indexicals vs. True Demonstratives: A Difference in Compactness. Manuscript. 1993c Perry on indexical semantics and belief states. Communication and Cognition 26/1: 77-96. 1993d How Reference Works. Explanatory Models for Indexicals, Descriptions, and Opacity. Albany: State University of New York Press. Searle, John 1979 Literal meaning. In: Searle, John (ed.). Expression and Meaning. Cambridge: Cambridge University Press, 117-136.

48

Walter De Mulder

Searle, John 1983

Intentionality. An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Sperber, Dan and Deirdre Wilson 1986 Relevance: Communication and Cognition. Oxford: Blackwell. Tasmowski, Liliane 1990 Les dömonstratifs fran^ais et roumains dans le texte et dans la phrase. Langages 97: 82-99. Tracy, Rosemarie 1983 Cognitive processes and the acquisition of deixis. In: Rauh, Gisa (ed.). Essays on Deixis. Tübingen: Narr, 99-148. Vandeloise, Claude 1986 L'espaceenfrangais. Paris: Minuit. Wittgenstein, Ludwig 1958 Philosophical Investigations. Oxford: Blackwell ( 1 9 7 2 . Transl. by G E M. Anscombe).

'Here' and 'there' in Croatian: a case study of an urban standard variety V

Milena Zic Fuchs

1. Introduction The importance of deixis in language, and especially its inseverable ties with the situation of utterance, is recognized today by many linguists. During the last couple of decades deictic phenomena have been attracting more and more attention, which is not surprising since they can be seen as the meeting place of semantic, syntactic and pragmatic aspects of language, aspects through which the relationship between language and context is reflected in the structure of the languages themselves. Research in the domain of deixis can cover various aspects corresponding to the socio-spatio-temporal dimensions of the speech event, and the following kinds of deictic expressions are traditionally distinguished: personal deixis, such as Τ and 'you', spatial deixis, such as 'here' and 'there', temporal deixis, such as 'now', 'today', and 'yesterday'. Relatively more recent research also includes social and discourse deixis (see for example Fillmore 1975, Lyons 1977 and Levinson 1983). The main focus of this paper will be centered on spatial deixis, more specifically on some of the Croatian equivalents of the English demonstrative adverbs here and there, adverbs having primarily a locative function.

2. The Croatian language and the Zagreb standard Interest in the various possibilities of expressing spatial relations in Croatian was initially triggered by the analysis of a corpus of recordings of naturally occurring spoken language recorded in the city of Zagreb, the capital of Croatia, a city numbering approximately one million inhabitants.1 Traditionally, the Croatian language is classified into three major dialects - Stokavian, Cakavian and Kajkavian - the names of the dialects being based on the forms of the interrogative-relative what, that is, sto, ca

50

ν

Milena Zic Fuchs

and kaj respectively. Standard Croatian is based on the Stokavian dialect, but should not be fully equated with it since it has undergone its own specific development during the last couple of centuries. The city of Zagreb is located in the midst of a large Kajkavian dialect area. Thus, the linguistic situation in the city is characterized on the one hand by the Croatian Standard, which is the language of the press, mass media, schools, government institutions, etc., and on the other hand by a specific variety of urban Kajkavian predominantly used in everyday informal situations. It should be noted that the Zagreb Kajkavian koine is distinct from rural Kajkavian, and that through its close contact with the Stokavian standard it has developed specific features which can be recognized on all levels of linguistic analysis (Magner 1966, Sojat 1979). At the same time Zagreb Kajkavian has inflected Standard Croatian spoken in Zagreb, resulting in a special type of Stokavian which Sojat (1983) identifies as a special variety and names Zagreb Stokavian. Sojat (1983) also claims that Zagreb Stokavian, or Zagreb Standard Croatian, is gaining more and more ground in everyday (even informal) speech situations, while Zagreb Kajkavian can be primarily heard within the family circle. 2 Thus, the Zagreb Standard can be seen as a prestigious idiom, and should not be considered substandard, but rather as one of the varieties of Standard Croatian.

3. The data The above mentioned corpus of naturally occurring spoken language is in its greatest part a corpus of Zagreb Stokavian. It was not only the very high frequency of deictic expressions in the recorded material that arose our interest, but also the fact that some of the deictic expressions were not used according to the prescribed grammatical norm of Standard Croatian found in the grammars and dictionaries of the Croatian language.^, This prompted us to interview an initial group of 40 speakers of Zagreb Stokavian, as to when and how they use various expressions for spatial deixis, particularly those that are equivalents of here and there in English. They were also exposed to different elicitation contexts, concrete situations that prompted them to use various demonstrative adverbs as well as other deictic expressions. 3 Out of the forty informants interviewed, 22 were women ranging in age from 18 to 66; 18 were men ranging in age from 18 to 73. The level of education spanned from those with a high school education to university professors. As far as occupation was concerned, businessmen, doctors, housewives, students, shop assistants, etc., were interviewed. Special attention was paid to place of birth and to the number of years the

'Here' and 'there' in Croatian 51

informant had permanently lived in the city of Zagreb. This group of informants was comprised of those having been born in Zagreb and having lived their entire lives in the city, and those having lived in Zagreb from very early childhood. The point of selecting suchjnformants was to obtain a representative group of speakers of Zagreb Stokavian or the Zagreb Standard and not a mixture of city dwellers that may have come to live in the city from other dialectal areas, particularly not at some later stage in life. 4. The Croatian pronominal demonstrative adverbs ovdje, tu, ondje It is a recognized fact that standard grammars very often do not provide satisfactory descriptions of deictic systems, and the majority of Croatian grammars are no exception in this respect. The prominent Croatian linguist Pranjkovic (1992), in a recent paper, stresses the fact that one of the weak points of Croatian grammars is their treatment of adverbs and various deictic systems. He specifically refers to the paradigmatic sets ovdje - tu ondje and evo - eto - eno, which correspond to the English adverbs here and there. And it was precisely these expressions that speakers misused in the recorded material or, in other words, did not adhere to the prescribed standard norm. Fillmore (1982: 35) points out that the analysis of deixis may be centered on the concrete appearance of various deictic expressions in the sense that we specify the speaker, orientation of the speaker's body, etc. in specific acts of utterance, or that it may be directed towards sets of words or grammatical categories, viewed as parts of the linguistic system. Both these approaches may prove interesting in themselves, but when it appears that a deictic paradigmatic set may be undergoing changes in a specific speech community, then it is necessary to see how concrete usage of deictic terms is affecting the system. What is more, one is faced with questions as to why such changes are occurring, questions for which it is not always easy to provide answers. Seen from the viewpoint of the prescribed norm or correct usage in Standard Croatian, ovdje - tu - ondje form a three member paradigmatic set for designating spatial relations, more specifically, for designating location. Pranjkovic (1992) classifies them as pronominal adverbs because they reflect or are based on the three member pronoun set Ί ' , 'you' and 'him'. Thus, one of the older dictionaries of the Croatian language (Ivekovic-Broz 1901), under the entry for ovdje, clearly states that the correct usage should be

ν

52

Milena Tic Fuchs

ovdje tu ondje -

where / am where you are where he is.

In terms of spatial relations indicating location, this three member paradigmatic set can be expressed as

ovdje tu ondje -

'proximal to the Speaker' 'proximal to the Hearer' 'remote or non-proximal both to the Speaker or Hearer'.

In other words, deictic spatial terms in Croatian are organized with respect to contrasts between participant-roles, rather than simply by terms indicating distance from a fixed deictic center, namely the Speaker's location at the time of utterance. This is also apparent in another closely related paradigmatic set relevant for the present analysis, that is, the three member set ovamo, tamo, onamo,5 whose members indicate direction towards a goal in the following way:

ovamo tamo onamo -

'direction towards the Speaker' 'direction towards the Hearer' 'direction away from both Speaker and Hearer'.

Thus, if we compare Standard Croatian and English, we are confronted with two three-way contrasts in Croatian, one for location {ovdje, tu, ondje) and one for direction (ovamo, tamo, onamo), which correspond to the English here and there distinction. More precisely, ovdje and ovamo can be said to correspond to here, while tu, ondje and tamo, onamo can be seen as the equivalents of there.6 The obvious difference between the two languages lies in the distance contrasts they recognize as well as the Pivot around which the distance categories are anchored. Namely, English shows a two-way contrast anchored around the Speaker as Pivot, which Fillmore (1982: 48) symbolizes as here there

( + Proximal) - 'relatively close to the Speaker' (- Proximal) - 'relatively remote from the Speaker'.

'Here' and 'there' in Croatian

53

On the other hand, Standard Croatian has both Speaker and Hearer as Pivot as in the three-way contrast described above, while the distance contrasts can be presented as ovdje tu ondje

(Proximal) - 'relatively close to the Speaker' (Medial) - 'relatively close to the Hearer' (Distal) - 'remote or non-proximal both to Speaker or Hearer'.

5. The Zagreb Standard and ovdje, tu, ondje Differences such as these are interesting in themselves for cross-linguistic research. However, investigation of actual usage in specific speech communities, may indicate that the deictic systems as described above for Croatian are not stable as one might expect, and that certain changes may be occurring. Analysis of the corpus, as well as subsequent interviews and elicitations, indicated that speakers of Zagreb Stokavian Standard do not use the prescribed norm, and that differences in actual usage are apparent in the contemporary Zagreb Standard. One of the most striking features noticed in the speech of Zagreb citydwellers was that ovdje and tu were both used in the sense of proximal to the Speaker (where I am) and that the participant role distinction was not adhered to. Thus the difference between ovdje and tu was neither implemented in concrete situations nor was the difference felt by the majority of the informants interviewed. According to their statements, they perceive ovdje and tu as being synonymous, and what is more the tu form was strikingly more frequently used in contexts where the ovdje form should have been used. This was evident even in the simplest of examples from the set-up elicitation contexts, as for instance when a book was placed right next to the informant, and he or she was asked Where is the book? The most frequent answer obtained was Tu je. (There - proximal to the Hearer - it is) instead of Ovdje (Here -je.proximal to the Speaker - it is)

54

ν Milena Zic Fuchs

The same thing occurred when a child was left standing next to the informant, while the interviewer called from the next room Where is the child? Almost as a rule speakers of the Zagreb Standard answered Tu je, kod mene, u sobi. (There - proximal to the Hearer - next to me, in the room.) instead of Ovdje je, kod mene, u sobi. (Here - proximal to the Speaker - next to me, in the room.) Another illustrative example which shows the discrepancy in usage of these Croatian deictic expressions is a simulated telephone conversation (assumedly between speakers in two different cities) which went as follows: Interviewer:

Kakvo je vrijeme? (What's the weather like?)

Informant:

Tu je suncano, a tamo kod vas? (There - where the Hearer is - it is sunny, and there - direction towards the Hearer - how is it?)

Interviewer:

Ovdje pada kisa. (Here - where the Speaker is - it is raining.)

The informant's correct answer should have been: Ovdje je suncano, a tu kod vas? (Here - where the Speaker is - it is sunny, and there - where the Hearer is - how is it?) A similar shift in usage was noticed in utterances with motion verbs such as Come here ('Come to where the Speaker is'). According to the prescribed norm the correct Croatian equivalent would be Dodi ovamo, instead of which the most frequent construction was Dodi tu ('Come to where the Hearer is'), or in a few cases Dodi ovdje. Apart from not distinguishing between Ί ' and 'you', the forms used for indicating location appeared as a rule instead of the form ovamo, which indicates movement towards the speaker.

'Here' and 'there' in Croatian

55

6. Rural Stokavian and ovdje, tu, ondje After collecting and analyzing the data provided by the speakers of the Zagreb Standard, an additional group of informants was interviewed. These were speakers of the Stokavian Standard who had lived in the city of Zagreb for different lengths of time (depending on their age), but had been born and had spent their childhood and formative years in areas, where the rural Stokavian dialect is spoken. The group also included a smaller number of Stokavian speakers residing in rural regions. 7 Informants from this group did as a rule differentiate between the forms ovdje and tu, that is between 'proximal to the Speaker' and 'proximal to the Hearer' and what is more, expressed irritation by what they termed the incorrect usage of the above mentioned forms. This irritation was also evident when the Zagreb Standard speakers used ovdje and tu in their gestural and symbolic functions as defined by Levinson (1983: 65). Utterances indicating gestural usage such as Put in here (referring to a location close to the Speaker) revealed two possibilities with the Zagreb Standard informants. Namely, Stavi to ovdje and Stavi to tu both appeared as alternatives, and again the latter example with tu (which indicates proximity to the Hearer according to the prescribed norm) was more frequent. As before, Stokavian Standard speakers interpreted these utterances differently, that is they used ovdje and tu in order to differentiate between the spatial differences embodied in the participant role distinction. The same shift in usage was apparent in the symbolic function that ovdje and tu can have in sentences, such as in I've lived here all my life. Ovdje sam proveo cijeli svoj zivot. Tu sam proveo cijeli svoj zivot. (I've lived here all my life.) The urban speakers found both sentences acceptable, and what is more, most of them stated that they saw no difference between them, while especially the rural Stokavian speakers considered the second sentence either unacceptable or unclear. The second interesting point that the research on the urban Zagreb Standard provided was how the Croatian equivalents of there were used in the Zagreb speech community. The most frequent form that appeared in the recorded corpus was tamo, which according to the prescribed norm indicates 'direction towards the Hearer' and should not be used to indicate location. According to the data collected, on the other hand, tamo is most frequently used to designate

56

Mi lern Tic Fuchs

location in the sense of 'remote or nonproximal both to Speaker and Hearer', reserved traditionally for the deictic ondje. Thus, utterances such as On je tamo (He is there-direction towards the Hearer) appeared almost as a rule instead of On je ondje, which would be in accordance with the prescribed usage one finds in Croatian grammars. Ondje was used by the rural Stokavian speakers, while the Zagreb informants claimed in subsequent interviews that they used ondje very rarely, and what is more, that they used it to indicate location extremely remote from the speaker.

7. Changes in the Zagreb Standard The above analysis indicates that the described Standard Croatian paradigmatic sets of locational deictics are undergoing changes or, in fact, have changed in the Zagreb speech community, and that we are faced with a 'new' combination of paradigmatic deictic elements. The changed paradigmatic set for indicating location in the Zagreb Standard can be presented as ovdje

- here, in the sense of 'proximal to the Speaker'

tu tamo

- there, in the sense of 'relatively remote from the Speaker'

In fact, the most frequent combination would be tu and tamo functioning very much like the English pair here and there. Generally speaking, ovdje, tu and ondje were taken over initially from the Stokavian dialect and thus became part of Standard Croatian at the time of its formation, but due to various influences and lines of development evident in urban varieties, the differences among these deictic expressions have begun to fade. In other words ovdje and tu have become synonymous, that is, both express proximity to the Speaker and both can be seen as the equivalents of English here. The expression tamo, that indicates location in the Zagreb Standard, has in fact been shifted from the paradigmatic set indicating 'direction towards' and has acquired a new sense indicating location remote from the Speaker. The above changes in the paradigmatic set for indicating location in the Zagreb Standard are not only changes on the linguistic level, but are also conceptual changes in the sense that the whole system for designating spatial relations has obviously undergone modifications. The major conceptual

'Here' and 'there' in Croatian 57

difference is related to the Pivot, that is to the shift towards the Speaker, but is manifested even more strongly in major changes in how spatial distances are reflected and symbolized by language or - as in Talmy's view (1983, 1988) - how language structures space. The analysis of the Zagreb Standard shows that the changes described above have occurred on both the linguistic and conceptual levels as manifested through the collapse of the participant role distinction reflected in the synonymous usage of ovdje and tu, and the appearance of tamo as locational deictic. The result of these changes is that from a three-member paradigmatic set we now have a two-member one, and the difference lies not only in the set numbers, but more importantly in the way distance features are conceptualized. Namely, the Zagreb Standard does not reflect the Medial distance category prescribed by Croatian grammars, which is still felt and implemented by rural Stokavian speakers. One can conclude at this point that space conceptualization has changed on an urban-rural basis, a fact not too surprising since urban and rural communities can differ greatly in environmental as well as cultural features.

8. The Croatian sentential demonstratives evo, eto, eno Apart from the locative deictic adverbs discussed so far, here and there can also be expressed in Croatian by the three member paradigmatic set evo, eto, enoß As in the previous case, this set also reflects a three-way contrast based on the participant role distinction, i.e. evo [Proximal] eto [Medial] eno [Distal] -

'proximal to the Speaker' 'proximal to the Hearer' 'remote or non-proximal both to the Speaker and Hearer'

The data analyzed show that evo, eto, eno frequently appear as single word sentences - Evo\ Eto\ Eno\, which can be taken to mean 'Behold!' or 'Look at this!'. They also appear frequently with pronouns and nouns in sentences such as Evo ga! (Here he is') or Evo ti! ('Here you are'). This indicates that syntactically speaking we are dealing with sentential demonstratives having primarily a presenting or gestural function. When compared to the above discussed demonstrative adverbs, the first apparent difference, related to their specific sentential syntactic nature, manifests itself in the fact that they cannot be used symbolically. The sentence 'I've lived here all my life' cannot appear in Croatian as

58

ν

Milena Tic Fuchs

Evo sam proveo cijeli svoj zivot nor can 'John is here' be expressed as Iva je evo. The second difference lies in the fact that evo, eto, eno regularly appear at the beginning of utterances and do not possess the flexibility of ovdje, tu, ondje, which can appear in different positions within the sentence, a feature mainly based on the fact that Croatian has a very free word order. Grammars of Croatian have a hard time with evo, eto, eno. They can be found classified as adverbs, but this claim is then followed by statements claiming that they represent a special type of adverb since they do not correspond to the standard division of adverbs into those denoting place, manner, time and cause. Some grammars claim they are interjections, and some simply place them into the category of particles, thus avoiding more specific classification. It is basically their specific syntactic feature of being sentential demonstratives, as well as the inherent three-fold spatial distinction, that prevents easy classification into standard grammatical categories. The situation is complicated further by special usages of the deictic evo. For instance, one quite frequent usage of evo may appear confusing at first glance. We are referring to cases when evo is used in the sense LOOK - BEHOLD, but indicates something or somebody situated far away, that is distal. In such cases evo does not adhere to 'the proximity to the Speaker' feature. The explanation for this not infrequent usage lies in our opinion in the semantics of LOOK. LOOK may indicate or point to entities not just close to the Speaker, but also to entities that are more distant. When a speaker says "Evo ga!" {'Here he is' or 'Here he comes') and is referring to a person far off or just entering his visual field, then such utterances usually have an additional feature of 'surprise' or 'expectation'. A typical example would be when one is waiting for someone, and then spots him in the distance, or when one sees in the distance a person that one has not seen for a long time. Usages such as these can at least partially explain why evo, eto, eno are classified as interjections in some grammars. Various specific usages of evo, eto, eno in concrete speech situations can be described at great length, thus providing insights into the many nuances of variation. However, for the purpose of the present analysis it is interesting to see what is happening to the three-way distance contrast in the Zagreb Standard. Investigation of the actual usage of evo, eto, eno indicates that apparently the paradigmatic set is not stable in the sense that speakers of the Zagreb Standard readily distinguish between the three contrasts. To be more precise, evo meaning 'proximal to the Speaker' causes no

'Here' and 'there' in Croatian

59

problems, but it is the Medial eto and Distal eno that are problematic. The majority of speakers interviewed simply do not "feel" the difference between them, and the data collected indicate that eto and eno are both used very much like there in English, in the sense of 'remote from the Speaker'. It should also be stressed that eto is the form used most frequently of the two. Thus we have a very similar situation with these sentential demonstratives as was the case with the demonstrative adverbs analyzed in the previous sections. The paradigmatic set evo, eto, eno can be presented as follows for the Zagreb Standard: evo -

here, in the sense of 'proximal to the Speaker'

eto eno

there, in the sense of 'relatively remote from the Speaker'

The data collected on two very important and frequent paradigmatic sets used as spatial deictics (data collected from speakers of an influential urban standard variety) indicate that speakers do not use and do not differentiate the distance categories inherent in the prescribed grammatical norm. It should also be stressed that speakers of the other two dialects, Kajkavian and Cakavian, also have to learn the difference between, for instance, ovdje and tu in the process of acquiring the standard language because their dialectal idioms for spatial deixis do not contain distinctions as those described above. Although it is impossible and possibly not very wise to predict whether certain tendencies in a language will prevail, in our opinion, the additional evidence from the other dialects, as well as that from the Zagreb Standard, may bring about changes in the sense that the participant role distinction and the distance categories inherent to it may eventually fade and change the system itself. Such predictions can at this point only remain on the level of conjecture.

9. Concluding remarks Historically speaking, the three-way spatial distinction based on participant roles was taken over into Standard Croatian from the Stokavian dialect. The rural variant of Stokavian has still preserved the described distinctions, which is not surprising since rural dialects are more conservative to change and are not exposed to the varied influences and dynamics of change as is

60

ν

Milena Tic Fuchs

frequently the case with urban varieties. On the other hand, the analyzed Standard Croatian spoken in the city of Zagreb indicates that it is the Medial distance category or 'proximal to the Hearer' which is disappearing. This is evident in the paradigmatic sets analyzed. In the first case ovdje and tu have merged and simply mean 'proximal to the Speaker', while tamo means 'relatively remote from the Speaker'. The same is happening in the three member set of sentential demonstratives evo, eto, eno. Evo in the sense of 'proximal to the Speaker' is stable, while the Medial category eto has merged with eno and has acquired the new meaning 'relatively remote from the Speaker'. This implies that the participant role distinction has collapsed in the case of the Zagreb Standard, and that the Speaker has become the fixed deictic center from which then distance categories are determined. Finally, one is faced with questions to which probably only tentative answers can be provided. I am primarily referring to questions as to why such changes do occur. At best one can provide suggestions as to what causes variation in spatial deictic paradigmatic sets, particularly as to why the Medial contrast is being neutralized. Fillmore (1982) suggests that the Hearer-Pivot category is an evolutionary development out of the distance category Medial. This seems plausible since one can assume spatial categories to be more basic than participant role distinctions. However, from the point of view of distance conceptualization, the Medial contrast is particularly interesting. It can be said that spatial concepts are in general relative concepts such as 'proximal to the Speaker', for they are not determined by any fixed spatial coordinates. When we consider the Medial category, with Hearer as Pivot, then this spatial concept can be said to be even more relative in the sense that the Hearer may be very close to the Speaker, or very distant at the time of utterance. Thus the Medial category can be seen as a more flexible or loose spatial concept when defined by the relativeness of the Hearer's position. Keeping in mind the complexity of the prescribed Croatian Standard paradigmatic sets, one can also view the switch to the Speaker as the deictic centre as a form of simplification within the overall system, which, according to Lakoff (1987: 497), would be a more efficient cognitive system. The simplification in the case of Croatian has manifested itself primarily in an urban setting, and this raises questions as to how culture and environment (see Perkins 1992) influence the formation of spatial concepts, or more precisely, whether changes in the way of living and in the physical environment can influence changes, or provide the basis, for shifts in spatial concepts.

'Here' and 'there' in Croatian

61

Questions such as the above are naturally open to discussion and should be primarily seen as providing possible explanations as to why systems of deictic expressions change within languages.

Notes 1.

2.

3.

4.

5. 6. 7. 8.

These recordings were collected during the eighties and were used for various research projects at the Institute of Linguistics, University of Zagreb. See for example Biianic and Zic Fuchs (1981) and Zic Fuchs (1990). I agree with Sojat's (1983) claim that Zagreb Stokavian is becoming more widespread, however, certain elements of Zagreb Kaikavian are still present in everyday speech situations. See for example Zic Fuchs (1990) for the influence of Zagreb Kajkavian on the address system found in Zagreb. Neither interviews nor set-up elicitation contexts are ideal ways of collecting data, and this particularly applies to elicitation contexts since they are very difficult to organize properly. However, they were the only methods available for confirming some of the interesting points that came out of the analysis of the recorded material. The majority of Croatian dictionaries classify them as adverbs of place, while contemporary grammars emphasize their demonstrative, pronominal and spatial nature as in Babic et id. (1991). The fact that they resist easy classification into traditional grammatical categories is also stressed. There is also a third related paradigmatic set indicating Path - ovamo, tuda, onuda. However, this set is not relevant for the present analysis. Ovamo, tamo, onamo can be said to be similar to the obsolete directional forms hither and thither in English. The entire group numbered 12 informants. The data on evo, eto, eno is partially taken from an earlier paper. For a more detailed analysis of this paradigmatic set, see Zic Fuchs (19911992).

References Babic, Stjepan et al. 1991 Povijesni pregled, glasovi i oblici hrvatskoga knjizevnoga jezika - Nacrti za gramatiku. Zagreb: HAZU - Globus. Bicanic, Sonia W. and Milena Zic Fuchs 1981 Analysis of discourse at a place of work. Studio Romanica et Anglica Zagrabiensia, 26 (1-2): 263-278. Fillmore, Charles J. 1975 Santa Cruz Lectures on Deixis. Indiana University Linguistics Club Publication. 1982 Towards a descriptive framework for spatial deixis. In: Jarvella and W. Klein (eds). Speech, Place and Action. John Wiley and Sons Ltd.,31-59.

62

ν Milena Zic Fuchs

Ivekovic, Franjo and Broz, Ivan 1901 Rjecnik hrvatskoga jezika, Stamparija Karla Albrechta. Lakoff, George 1987 Women, Fire and Dangerous Things. What Categories Reveal about the Mind. Chicago: University of Chicago Press. Levinson, Stephen 1983 Pragmatics. Cambridge: Cambridge University Press. Lyons, John 1977 Semantics. Vol. 2. London: Cambridge University Press. Magner, Thomas F. 1966 A Zagreb Kajkavian Dialect. University Park: Pennsylvania State University Press. Perkins, Revere D. 1992 Deixis, Grammar and Culture. Amsterdam: Benjamins. Pranjkovic, Ivo 1992 Prilozi kao rijeci sviju vrsta. Suvremena lingvistika, No. 34, Hrvatsko filolosko drustvo, Zagreb. Sojat, Antun 1979 Ο zagrebackom kajkavskom govoru. Rasprave Zavoda za jezik, Vol. 4-5, Zagreb: Zavod za jezik Instituta za filologiju i folkloristiku. 1983 Zagrebacka Stokavstina. Rasprave Zavoda za jezik, Vol. 8-9, Zagreb: Zavod za jezik Instituta za filologiju i folkloristiku. Talmy, Leonard 1983 How language structures space. In: Pick, Η and L. Acredolo (eds). Spatial Orientation: Theory, Research and Application. New York: Plenum Press, 225-282. 1988 The relation of grammar to cognition. In: Rudzka-Ostyn, Brygida (ed.). Topics in Cognitive Linguistics. Amsterdam: Benjamins, 165-205. v Zic Fuchs, Milena 1990 Social deixis in contact and contrast. Languages in Contact, Proceedings of the 12th International Congress of Anthropological and Ethnological Sciences, Institute of Linguistics, Faculty of Philosophy, University of Zagreb, 274-279. 1991/ 1992 On contrasting expressions of spatial deixis in Croatian and English. Studia Romanica et Anglica Zagrabiensia, 36-37: 93-102.

Prosodic and paralinguistic signals of distance Janina Ozga

1. Introduction When collecting information on the prosodies of distance, I was not surprised to find that most of the relevant data came from studies concerned with man's social and proxemic behaviour. Sociolinguistics and ethnography of communication provide interesting material on prosodic correlates of social distance, defined in terms of status, familiarity, similarity, group identification, etc. (see e.g. Hymes 1964 and elsewhere, Coulthard 1977, Brown and Levinson 1987). Hall in his work on proxemics (1959, 1966) used changes in the volume of the voice as a source of information about distance zones within which human beings interact. It was not surprising, either, to find that most proxemic and ethnolinguistic studies focus on cultural variations in interpersonal space and emphasize the culture-specific nature of norms that govern these variations. Hall says at one point (1966: 139): Even though man may be physiologically and genetically one species, the proxemic patterns of the Americans and the Japanese often strike one as being as disparate as the territorial display patterns of the American grouse and the Australian bowerbirds ...

But is there really all that much culturally determined diversity in proxemic behaviour? I should like to argue that at least its vocal aspects display a certain invariance, connected with the nature of the vocal-auditory channel and the existence of a common experiential basis which influences people's conceptualization of space (see Sweetser 1990, Langacker 1991: 2-4). In this paper I shall examine the universal principles underlying spatially constrained use of prosody and paralanguage. I shall also attempt to account for some of the cultural differences that give rise to different prosodic markers of comparable relationships in physical and social space. Finally, I hope to demonstrate that certain prosodic and paralinguistic patterns are non-trivially and non-randomly associated with linguistic forms that refer to space. In other words, the prosody of talking in space shades into the

64

Janina Ozga

prosody of talking about space, through something like Sweetser's mapping of the physical domain onto the mental domain (1990: 23-48).

2. Talking in space In this section I shall take a brief look at the way speech carries through physical space and how people enlarge their auditory space. Then I shall analyze the factors involved in judgements that speakers make about distances separating them from their interlocutors and the way those judgements influence the choice of prosody.

2.1. Physical distance Speech is a temporal medium because it develops in time but it also has its spatial aspect, because it has to be carried through space from the speaker to the hearer. Speech can only perform its communicative function when it is heard, which is why we talk about being within/out of hearing or within/out of earshot. The physics of sound as well as physiological limitations imposed by the vocal apparatus and the hearing mechanism allow speech to be carried about 40 metres in a free field. Sounds inevitably fade away, to become inaudible at a distance beyond which we can cry ourselves hoarse without attracting the addressee's attention, to say nothing of being actually understood, as hailing distance is greater than hearing distance within which it is possible to make out what is being said. The maximum distance at which speech discrimination can occur is, however, not constant but depends on a variety of factors. There are, first of all, individual differences in auditory ability of hearers and the carrying power of the voice of speakers. Certain vocal sounds (e.g. open vowels) have greater inherent loudness while certain pitch patterns carry better than others (see call contours below). Hearing conditions (closed vs. open spaces) greatly influence the behaviour of the sound wave (with absorption, reflection, reverberation, delayed echoes, ambient noise, hot spots and dead spots being only a few of the terms used in acoustics). Auditory space may shrink because of interfering noise (especially in modern urban society), in spite of our ability to listen selectively ("cocktail-party phenomena", see Crystal 1987: 147). "I can't hear myself speak!" - this cry of despair plainly shows how auditory space may actually be reduced to almost nil. Aware of all these physical and environmental constraints on audibility people have tried, since time immemorial, to create optimal acoustic conditions for carrying out their

Prosodic and paralinguistic signals of distance

65

verbal transactions.

2.2. Enlarging auditory space A motif exploited in innumerable films is that of various forces and circumstances conspiring to render protagonists helpless because they cannot be heard (man overboard, earthquake victims, people separated by a milling crowd). In view of the fact that such situations do occur in real life, it is not surprising that people have always endeavoured to "secure uptake" (Austin's term, quoted after Gibbon 1976: 286). There are several ways this has been achieved: I.

Inventing and making the most of the written medium of language and other visual codes (Morse, smoke-; mirror-, flag-, and lightsignalling, bookmakers' and stockbrokers' signalling).

II.

Resorting to aural signalling systems with greater audibility (whistled speech, drum signalling).

III.

Utilizing the principles of architectural acoustics (that go back to the designs of ancient Greek and Roman amphitheatres) and environmental acoustics which deals with the problem of noise control.

IV.

Inventing devices that enhance audibility: from cupping hands round the mouth or the ear, through ear-trumpets and megaphones, to microphones, bugs, public address systems, broadcast systems and finally telecommunication satellites thanks to which the world has truly become a global village.

V.

For millennia now people with political ambitions, actors, public figures (from Demosthenes to Lady Thatcher) have been training their voices taking elocution lessons, to make their voices more resonant, their enunciation clear and pleasant to the ear. A good carrying voice must have been one of the requisite qualifications of the early leaders of men; only later hereditary rulers could afford to have thin reedy voices (often remarked upon in the memoirs of subjects astonished at the lack of this leadership quality in their sovereigns). There were always court officials (now spokesmen) who could speak in their name. They had heralds with Stentorian voices and now they have microphones. However, even today a good voice still ranks high on the list of desirable personality characteristics. I am discussing this seemingly

66

Janina Ozga

irrelevant problem here because I think the association of loudness and resonance with leadership, dominance, and elevation is firmly entrenched in collective human consciousness. One might ask whether the above can be reconciled with the paralanguage of the famous Wolof greetings (Coulthard 1977: 45-6), or a statement like the following: "[In Kirundi] upper-caste speakers seem never to raise their voices ..." (Crystal 1987: 38). Again the answer is: they can afford to, as their status is established and lower-caste persons are going to hang on their lips anyway. But the statuscreating association between loudness and dominance cannot be denied and has roots both in the physical elevation (leaders were wise to speak from higher levels with sound loss significantly reduced) and the atavistic cowering before thunder - the loudest natural sound and, by extension, any loud noise. This is why the powerful noise of unison speech or singing (war cries, battle chants, cheering, etc.) still have the function of undermining the morale of the enemy, opponent, or rival team. VI.

Unlike some of the preceding methods of enhancing audibility, the exploitation of the sonic potential of language is common to all language communities, regardless of how advanced they are technologically. A secluded village community may not need orators or beautifully designed auditoriums but its members need to raise their voices to secure uptake from distant interlocutors. Again loudness is the most common correlate of distance (FORTE) and is accompanied by decrease in tempo (LENTO). Vowels are lowered and prolonged to increase sonority (see Nathan 1989 and also Fonagy 1991 on the semantics of vocal gesturing). Nathan mentions an early experiment in which sonority is examined with respect to distance (1989: 62): In a dark street, the experimenter (O. Wolf) found the distances at which various sound differences were no longer audible. Needless to say, audible differences between voiceless consonants disappeared at a much closer distance than the audible differences between contrasting vowels, with /a/ versus higher vowels being the difference perceivable at the greatest distance.

An interesting correlation connected with this strategy of securing uptake can be observed in the use of Polish vocatives. There is a special vocative case (though increasingly speakers tend to use the nominative case instead). Those speakers who maintain the difference are likely to use the vocative at a closer distance, but the nominative at a greater distance, especially if the

Prosodic and paralinguistic

signals of distance

67

vocative ends in [u]: cf. Maigosiul (Maggie - Voc) and Matgosial (Maggie - Nom). Although in phonetic symbolism both [a] and [u] express distance while [i]-type vowels express nearness (Crystal 1987: 175), [u] is lower on the sonority scale (less prototypical), which accounts for the choices made by the speakers of Polish. Vowel epenthesis is common in calling (e.g. [pe:'li:z] for please·, Polish Fe-laszki, sze-maty for Flaszki, szmatyl used by people collecting bottles and old clothes). On the other hand, messages addressed to people at a distance tend to be brief, terse, even clipped, especially in set phrases, as in a Polish military greeting (1)

Czotem, panie prezydencie Hail, Mr.-Voc President-Voc 'Welcome, Mr. President.'

which becomes (2)

'czo - 'pa, pre - 'den -

where - marks "silent ictus". Compare: (3)

['jo: - ' P 9u]

which stands for Yorkshire Postl as heard from a seller in Leeds. Languages use vocalization as attention-getters at closer distances (Cruttenden 1986: 180) and interjections both at near and far distances (but interjections have other functions as well: on the prosodies and semantics of interjections in English, Russian and Lithuanian see an interesting study by Misevicene et al. 1987). As in this paper I am particularly interested in the prosodies of distance, I find by far the most interesting the use of vocal means to conquer distance which is associated with call contours. They are treated by some authors as stylized, stereotyped tunes (Ladd 1978, Cruttenden 1986: 125, Crystal 1987: 173) with the form described as a chanted or sing-song intonation pattern which steps down from one level pitch to another with an interval of approximately a minor third: 'Johnjiy (Couper-Kuhlen 1986: 169).

and the meaning as

68

Janina Ozga

What is signalled ... is the implication that the message is in some sense predictable, part of a stereotyped exchange or announcement. "Nothing you couldn't have anticipated" it says (Ladd 1978: 520).

However, as Gibbon (1976: 274-287) convincingly argues, the basic role of the call contour is the securing of uptake in circumstances where physical distance between interlocutors is of primary importance (Gibbon 1976: 280): (1) difficulty of communication is present at the physical level; (2) the difficulty is surmounted by increasing the carrying power of the utterance.

The final level tone "is admittedly an effective means to carry one's voice far or to ring it above other types of sound" (Abe 1962: 522, quoted after Gibbon 1976: 280). This call contour, which has its derived uses where the notion of distance is projected onto a more abstract plane, might be a good candidate for a distance-related intonation category, with the basic use as the prototypical member of the category (see Taylor 1989, Chapter 10). The foregoing discussion on the means of optimilizing audibility frequently stresses maximum value of loudness, sonority, etc. But the other extreme, i.e. silence, understood as a realization of the wish not to be heard/overheard, should also be taken into account, together with its social meanings: "staying silent is a sure sign of distance, alienation, even danger" (Crystal 1987: 10). Between the two extremes are the various voice levels and prosodies which signal speakers' assessment of distance, on the physical, social and emotional levels. As space is dynamic, such judgements are not stable, but constantly changing, depending on speakers' current location in physical and social space.

2.3. Judging distances Judgements of distance are grounded on people's sensory experience of physical space. This experience involves also some rudimentary understanding of acoustics, which allows people to adjust their voice levels to the physical and environmental conditions in which communication takes place. Sometimes such judgements are erroneous, as shown in the joke about the man with a banana in his ear. Voices may sound too loud if the acoustics of a hall are misjudged; amplifying equipment is frequently misused by people with insufficient understanding of its potential (the amplification of ranting political speeches or "fire and brimstone" sermons may lead listeners to

Prosodic and paralinguistic signals of distance

69

some sort of acoustic trauma). But readjustments are usually made instantaneously and semi-automatically through a process of accommodation. Judgements based solely on physical factors are made only if social distance/intimacy between interlocutors is firmly established, or in extremis: a student would yell at his professor to warn him of danger but not to tell him that he has forgotten his umbrella. If social distance rules are violated, significant - though often slight - adjustments in prosody, paralanguage, body language occur. A crowded bus is fertile ground for collecting data on such adjustments. A universal paralinguistic feature signalling a breach of social distance rules is the tension of vocal gesture (cf. Fonagy 1991). This can be observed e.g. in the use of whisper in socially different situations: (i)

no tension: whispered conversation between friends;

(ii)

tension (clarity of enunciation): answering a whispered query from a VIP sitting next to you;

(iii) tension (rigidity): whisper used by a woman trying to release herself from an unwelcome embrace without attracting attention (see also Brown and Levinson 1987: 231). I must admit to being on shaky ground when it comes to analyzing distance as an emotive variable. Fonagy, like Bolinger (1980 and elsewhere), appears to have no doubts that it is one; in his discussion of expression of emotions he argues that Intonation represents a decisive step in the strategy of encoding: glottal gesturing is projected into an ideal acoustic space. The events taking place at the glottal level are perceived as tonal movements and are described as such by musicologists ... and phoneticians ... A dynamic interpretation is implicitly contained in the German term Tonbewegung "vocal movement" and in the Hungarian hanglejtes "vocal dance". This enables pitch to depict real or imaginary spatial displacements [italics mine - J.o.], to conceive and to represent emotive states in a dynamic form. Anguish, for instance, is reflected by means of a strongly narrowed pitch-range, suggesting a similar contraction of the whole body of a person hunching up to pass unmarked, to escape notice (Fonagy 1991: 503-504).

Prosodic correlates of emotive distance (e.g. "cooling off" a relationship, sympathy, dislike of a sibling) surely exist but it is difficult to separate them from those that signal physical and social distance. If we try to

70

Janina Ozga

find out the spatial "meaning" of loudness, we can come up with a number of plausible versions. Thus, loudness may indicate the degree of the speaker's separation from the hearer. It may also indicate social separation (e.g. dominance). As an obvious exponent of anger, it may signal - in the case of non-directional "letting off steam" - that the speaker wants to release his tension and frustration and thus distance himself from the source of his anger. If anger is directed at another person, loudness may represent an instance of metaphorical mapping from the physical onto the mental domain (see Sweetser 1990); by indicating that although the object of the speaker's anger is close at hand, he cannot "get it through his thick skull" that he has done something wrong. A similar extension occurs in the case of some of the derived meanings of Gibbon's call contours, e.g. in talking to babes-in-arms, where the difficulty of communication does not occur at physical level but at a different level of comprehension (a communication barrier): "the question of understanding ... on the part of the listener does not arise" (Gibbon 1976: 280). I would also treat as metaphorical extension the use of loudness as an exponent (together with pitch and length) of emphasis, where the hearer is notified of the importance of certain parts of discourse by being, as it were, invited to "come closer to them". I am not prepared, at this point, to take a stand on which, if any, of the meanings of loudness belong to the prosody of "talking about space" but there are some prosodic patterns that are not associated with the communicative context of the speech act but directly and consistently with the actual linguistic forms used.

3. The prosody of talking about space When, on his election in October 1978, Pope John Paul II first addressed the bewildered and wary crowd gathered in St. Peter's Square (and millions of equally bewildered television viewers), he said that he came "from a distant country" (which later became the title of a film about him). His se e: (4)

pa

Ion ta: no

helped the audience to take a giant, though smooth and effortless, mental leap to the distant country and instantly created a bond of familiarity and intimacy that can still be re-lived by watching old newsreels reporting the scene.

Prosodic and paralinguistic signals of distance

71

The above is perhaps the most striking evidence for the existence of prosody of mental space, but in fact there are numerous examples that can be discussed under this heading. The most common is what Uldall calls "fairy-tale intonation" (Uldall 1939, quoted after Abe 1955), with a raised pitch occurring sentence medially. The pitch pattern together with other features (lento, glissando, drawl, rhythmic, etc. see Crystal and Davy 1969) signals the remoteness and unreality of the scenes evoked by language (There once was a king ...). Similar sets of features appear in utterances which represent "mental calculations", frequently accompanied by upward gaze, drawing pictures or characters in the air, and opening phrases such as Let me think ..., Let's see ... In such cases the speaker is focussing, as it were, on some distant point in mental space, where solutions to his problems might be found (Sweetser 1990: 33) on the relation between physical vision and mental vision. An even more complex system of prosodic markers of space operates in the theatre. To take one example: stage whisper performs several distancerelated functions simultaneously: it has to secure uptake from the audience, it has to suggest intimacy with the character to whom it is addressed and distance from the character from whom whispered information is withheld (on intonation in drama see Ozga 1984).

4. Conclusion No conclusions can be drawn from the foregoing discussion save one: the relation between prosody and distance deserves more attention, especially from cognitive linguists. The question that I should like to find an answer to, most urgently concerns the ratio of the constitutive to the concomitant forces which govern the prosodic shape of Edgar's speech to blinded Gloucester in Act IV, Scene VI of King Lear. Come on, sir; here's the place: stand still. How fearful And dizzy 'tis to cast one's eyes so low! The crowds and choughs that wind the midway air Show scarce so gross as beetles; half way down Hangs one that gathers sampire, dreadful trade! Methinks he seems no bigger than his head.

I hope to answer this question in another paper.

72

Janina Ozga

References Abe, Isamu 1955 1962

Intonational patterns of English and Japanese. Word 11.3: 386-98. Call contours. In: Proceedings of The Fourth International Congress of Phonetic Sciences. Helsinki, 519-23. Bolinger, Dwight 1980 Intonation and "nature". In: Foster, Mary Le Cron and Stanley H. Brandes (eds). Symbols as Sense. New York: Academic Press, 9-23. Brown, Pamela and Stephen Levinson 1987 Politeness. Cambridge: Cambridge University Press. Coulthard, Malcolm 1977 An Introduction to Discourse Analysis. London: Longman. Couper-Kuhlen, Elizabeth 1986 An Introduction to English Prosody. London: Edward Arnold. Cruttenden, Alan 1986 Intonation. Cambridge: Cambridge University Press. Crystal, David 1987 The Cambridge Encyclopedia of Language. Cambridge: University Press. Crystal, David and Derek Davy 1969 Investigating English Style. London: Longman. Fonagy, Ivan 1991 Paralinguistic universale and preconceptual thinking in language. In: Waugh, Linda R. and Stephen Rudy (eds). New Vistas in Grammar: Invariance and Variation. Amsterdam: Benjamins, 495-516. Gibbon, Dafydd 1976 Perspectives of Intonation Analysis. Frankfurt: Lang. Hall, Edward D. 1959 The Silent Language. New York: Doubleday. Hymes, Dell 1964 Introduction: Towards Ethnographies of Communication. In: Gumperz, John J. and Dell Hymes (eds). The Ethnography of Communication. American Anthropologist 66/6, 1-34. Ladd, D. Robert 1978 Stylized intonation. Language 54: 517-39. Langacker, Ronald 1991 Concept, Image and Symbol. Berlin: Mouton de Gruyter. Misevicene, I., L. Minaeva, J. Pikhchilingis 1987 Interjection. Vilnius: Ministry of Education Press. Nathan, Geoffrey 1989 Preliminaries to a theory of phonological substance: The substance of sonority. In: Corrigan, Roberta, Fred Eckman and Michael Noonan (eds). Linguistic Categorization (Current Issues in Linguistic Theory 61). Amsterdam: Benjamins, 55-67. Ozga, Janina 1984 Intonation in drama. In: Kaluza, Irena (ed.). Litterae et lingua: In honorem Premislavi Mroczkowski. Wroclaw: Ossolineum, 99-102. Sweetser, Eve 1990 From Etymology to Pragmatics. Cambridge: University Press. Taylor, John R. 1989 Linguistic Categorization. Oxford: Clarendon Press. Uldall, Elizabeth T. 1939 The Intonation of American English. Unpublished Thesis.

Section 2 Conceptualizing space in prepositions and in morphology

The German über Elena Bellavia

0. Introduction This paper is devoted to the German über, in spatial and non spatial expressions. The particle über is used as a preposition or as a verbal (or nominal) prefix. In this last function it can be separable (i.e. as a sort of stranded preposition) or inseparable (always occurring together with the verb stem). Über has some similarities to the English preposition over, which has been studied thoroughly by Brugman (1983) and Dewell (1993). Both scholars describe the meanings of this complex lexical unit according to the principles of cognitive linguistics. In this paper I will also refer to the same paradigm, first analyzing the different variants of über separately and then trying to unify their meanings in a lexical network. In the first part, the different uses of the prepositional phrase "Uber+ noun" will be investigated. The second part deals with über as a verb prefix. The first and the second section respectively examine the separable and the inseparable verb. The metaphorical expressions are of great importance in understanding the schemata expressed by each structure. I will also show how the schemata affect transitivity, since the inseparable verb is always transitive, while the separable verb is prototypically intransitive. In the third part, the three structures (prepositional phrase, separable and inseparable verb) will be compared in order to understand their functions in language. It will be shown how the prepositional phrase "über+ noun" represents a general characterization of movement, while verbs prefixed by über are specifications for the purpose of construal emphasizing particular aspects. In the final section, all the meanings of über will be represented in a network. All examples quoted in this paper are taken from Duden (1981) and Wahrig (1978).

74

Elena Bellavia

1. The preposition über

1.1. The basic meanings of 'über' The preposition über is a so-called two-way preposition, i.e. it can co-occur with both the Accusative (ACC) and the Dative (DAT). The use of these cases is governed by rather complex "rules" (cf. Smith 1987) which are partly referred to below. The first use I want to deal with refers to a scene in which the TR (=trajector) is located above the LM (=landmark[SMl])1 as in: (1)

Die Lampe hängt über demOAT Tisch. 'The lamp hangs above the table.'

(2)

Das Flugzeug fliegt über derDAj Stadt. 'The plane is flying over the town.' (inside the boundaries of the town)

In this kind of relationship the vertical superiority of the TR over the LM is profiled, i.e. the distance of TR's location from the origin of the vertical axis is greater than the LM distance (see Figure 1): the TR seems to include the LM along the vertical axis, whereas the LM seems to include the TR along the horizontal axis (cf. Langacker 1987: 223-5):

+Vert TR > ΤΜ

0

Figure 7. Die Lampe hängt über

demDAT

Tisch.

The German 'über'

75

In (1) the TR is motionless; whereas in (2) it is in movement; however, in both expressions the case required is DATIVE (DAT) since the relationship between TR and LM is unchanging (Smith 1987). In particular, in scene (2), the plane flies over the town but its flying is confined within the boundaries of the LM (the town):

Figure 2. Das Flugzeug fliegt über derDAT Stadt. In contrast, in a scene such as: (3)

Das Flugzeug fliegt über dieACC Stadt. 'The plane flies over the town.' (crossing the boundaries of the town)

TR

J LM Figure 3. Das Flugzeug fliegt über dieACC Stadt.

the TR crosses the boundaries of the LM area (the town) moving along the horizontal dimension and maintaining a linear, directional path. In this case the ACCUSATIVE (ACC) is required, since the relationship between TR and LM changes with respect to the locative configuration and is scanned in sequential fashion (Langacker 1991: 403). DAT signals motion (or static location) wholly within the preposition's search domain, whereas ACC indicates a trajectory that takes the search domain as its goal and penetrates its boundary.

76

Elena Bellavia

Note that in this kind of relationship (see Figure 4) the superiority of the TR over the LM along the horizontal dimension is profiled; the above mentioned vertical superiority, i.e. the vertical separation (see Figure 1) even though it is still present - seems to become less important:

+Veit

TR

'

y

-LM

tHotiz Figure 4. Das Flugzeug fliegt über dieACC Stadt.

Indeed, it can even be removed completely as in the following examples: (4)

Er geht über dieACC Straße. 'He crosses the road.'

where the TR and the LM are in contact, and the TR goes across the LM:

Figure 5. Er geht über dieACC Straße.

In other words: (a) (b)

if the TR as a mover remains within the LM area it maintains a vertical superiority over the LM (Figure 1); if the TR as a mover is in contact with the LM, the TR must necessarily cross the boundaries of the LM as in (4) {Er geht über

The German 'über'

77

dieACC Straße=He crosses the road); indeed, if both TR and LM are in contact and the TR moves remaining within the boundaries of the LM, auf must be used instead of über. (5) Er geht auf derDAT Straße. 'He walks along the road.' In (5) the TR moves inside the boundaries of the street; DATIVE is required even though the TR is in movement, since it does not cross the LM boundaries:

TR

LM

Figure 6. Er geht auf derDAT Straße.

Note that the path can also be a subjective path from the observer's viewpoint as in: (6)

Sie wohnen über derDAT Straße. 'They live across the road.'

The configuration in (6) - even though stable over time and therefore with DAT - shares some features with a process as in (4) {Er geht über dieACC Straße=He crosses the road), since the position of the TR is described with respect to the observer and it is seen as the result of mental scanning reaching a certain point across the road 2 :

B

LM

/

TR

T^

/

O" Figure 7. Sie wohnen über derDAT Straße.

78

Elena Bellavia

Up to now we have only discussed scenes in which, if the TR moves, the direction of the path is horizontal (across the schema). But if the TR goes over a vertically oriented LM (an obstacle) we may have two possibilities: (7)

Er klettert über dieACC Mauer. 'He climbs over the wall.'

(8)

Er springt über einACC Hindernis. 'He jumps over an obstacle.'

In (7) the TR and the LM are in contact, in (8) they are not; however, in both cases the schema is conceptually the same, because the parameter of contact is not distinctive. The schema at work is the ARC schema (the TR goes beyond the LM): TR

TR

LM

Er klettert über dieACC Mauer. Er springt über einACC Hindernis.

Figure 8. The TR goes beyond an obstacle.

A particular case of this schema is the excess liquid out of its container: (9)

Der Wein läuft über den Rand des Glases. 'The wine flows over the rim of the glass.'

where the shapeless TR is inside the LM-container. In order to go over the LM, the TR flows up, crosses the LM boundary (the rim) and turns upside down:

Λ ΤΪ V

The German 'über'

79

Consider the following situations, where the TR and the LM are in contact: (10)

Das Tuch ist über derOAT Nähmaschine. (motionless: DAT case) 'The cloth is over the sewing-machine.'

(11)

Die Frau breitet die Decke über denACC Tisch. (with motion: ACC case) 'The woman spreads the table-cloth over the table.'

The schema at work here is the absolute covering (Hawkins 1984), in which the TR obscures the entire LM-surface. In this case, the TR is on the LM: there is no separation between them along the vertical dimension and the TR acquires a dimensional superiority, i.e. it totally covers the LM surface:

Figure 10. Das Tuch ist über derDAT Nähmaschine.

To sum up the German über can describe various scenes: (a) The TR is above the LM: (al)

motionless:

Die Lampe hängt über demo AT Tisch. 'The lamp hangs above the table.'

static situation - contact unchanging relationship between TR and LM: DAT

superiority on the vertical axis: HEIGHT superiority

80

(a2)

Elena Bellavia

with motion: Das Flugzeug fliegt über derOAT Stadt. 'The plane is flying over the town.'

dynamic situation - contact unchanging relationship between TR and LM: DAT superiority on the vertical axis: HEIGHT superiority

(b) The TR crosses the LM-surface (path along the horizontal dimension): (bl)

without contact: Das Flugzeug fliegt über dieACC Stadt. 'The plane flies over the town.'

(b2)

dynamic situation - contact changing relationship between TR and LM: ACC relevant superiority on the horizontal axis: DISTANCE superiority from the reference point

with contact: Er geht über dieACC Strasse. 'He crosses the road.'

dynamic situation + contact changing relationship between TR and LM: ACC relevant superiority on the horizontal axis: DISTANCE superiority from the reference point

The German'über'

81

(c) TR goes beyond the LM-obstacle (path along the vertical dimension): (cl)

without contact: Er springt über ein ACCffindernis. 'He jumps over an obstacle.'

(c2)

dynamic situation - contact changing relationship between TR and LM: ACC relevant superiority on the horizontal axis: DISTANCE superiority from the reference point

with contact: Er klettert über dieACC Mauer. 'He climbs over the wall.'

dynamic situation + contact changing relationship between TR and LM: ACC relevant superiority on the horizontal axis: DISTANCE superiority from the reference point

(d) TR is across/bevond the LM: as the result of a subjective path: Sie wohnen über derOAT Straße. 'They live across the road.'

static situation contact: * irrelevant unchanging relationship between TR and LM: DAT

82

Elena Bellavia

superiority on the horizontal axis: DISTANCE superiority from the reference point

(e) TR covers the entire LM's surface:

(el)

motionless: Das Tuch ist über derDAT

Nähmaschine

'The cloth is over the sewing-machine.'

(e2)

static situation + contact unchanging relationship between TR and LM: DAT TR covers the LM surface: DIMENSIONAL superiority

with motion: Die Frau breitet die Decke über denACC Tisch. 'The woman spreads the table-cloth over the table.'

dynamic situation 4- contact changing relationship between TR and LM: ACC

TR covers the LM surface: DIMENSIONAL

superiority All these configurations can be subsumed under a single superschema: the ARC-schema. Through the ARC-schema we can more easily explain the different meanings of über: every scene can be thought of as the different activation of one point of the ARC (Figure 11; letters correspond to the subsections listed above).

The German 'über'

83

TR

ft

/

(a) 'above'

(c)'beyond' (dynamic) (+/-contact)

(b) 'across' (+/-contact)

TR

(d) 'beyond' (static)

(e) 'covering'

Figure 11. The ARC-schema as applied to "über"

In this respect the concept of German über is very similar to the concept of over, as interpreted by Dewell (1993): "the key to the analysis of 'over' is a curved arc-trajectory in the central schema, replacing the flat 'across' trajectory presumed by Brugman and Lakoff."

1.2. Extended meanings of über In this section I am going to illustrate how metaphors in extended meanings of über can be related to the schemata presented above.

84

Elena Bellavia

(a) Extended meanings related to the 'above' schema Being above means possessing a favourable position, from which one can control the situation (cf. Having Control or Force is Up/Being Subject to Control or Force is Down metaphor, Lakoff/Johnson 1980). Therefore the following metaphors are connected with the above schema: (12)

Er ist Herr über Leben und Tod. 'He is the master of life and death.'

(13)

Gewalt, Macht über jdn. haben. 'to have power, might over s.o.'

(14)

Das militärische Übergewicht über jdn. gewinnen. 'to gain military superiority over s.o.'

If the LM represents an imposed boundary, being above can be equated with crossing the norm and exceeding a conventional scale3. The excess can be numerical (cf. More is Up/Less is down metaphor, Lakoff/Johnson 1980) as in: (15)

Er hat über 5000 Mark gewonnen. 'He has won over 5,000 Marks.'

(16)

Das Thermometer steigt über Null. 'The thermometer rises above zero.'

LM

Figure 12. Das Thermometer steigt über Null.

The German 'über'

85

As for the contrast between auf/über (cf. also examples 4-5) compare now the following examples: (16)

Das Thermometer steigt über Null. 'The thermometer rises above zero.'

(17)

Das Thermometer steigt auf Null. 'The thermometer rises to zero.'

(18)

Das Thermometer steht auf drei Grad. 'The thermometer is at three degrees.'

Über describes the crossing of a certain point placed on the conventional scale (16), while auf can either express the reaching of that point - not going beyond it - (17), or a stable location on the scale (18). Über can also express an increase in quantity of the same object (More is Up): (19)

Das Kind stellte Fragen über Fragen. 'The child asked question after question.'

(20)

Er machte Schulden über Schulden. 'He piled up debt after debt.'

in which the quantity piles up and is always thought to be very large. Verticality can also symbolize preference (Brugman 1983), which evokes the image of a scale of value based on a subjective judgement of superiority (Johnson 1987): (21)

Musik geht ihm über alles. 'He likes music above all else.'

(b) Time metaphors in terms of the 'across' schema Unlike space, which is seen as three-dimensional and static, time is conceptualized as one-dimensional and dynamic, which implies a linear orientational direction (Ehrich 1989). Time has extension, and therefore it can be crossed:

86

Elena Bellavia

(22)

Er ist über 18 Jahre alt. 'He is over 18 years old.'

(23)

Er hat die Nacht über gearbeitet. 'He worked all night.'

In (22) the TR (Er) has crossed a certain age boundary. This limit placed on the time scale, which is conventionally horizontally oriented. (23) 'the man worked all night'; the whole period is focussed, and it crossed by the activity carried out by the agent. As in many other cases, here too the spatial schema can be applied temporal situations: (24)

Unser Spaziergang führt uns über die Altstadt hinaus. Our walk leads us beyond the old city.'

(25)

Sie ist über die besten Jahre hinaus. 'She is no longer in her prime.'

is In is to

In (24) (still only spatial) the TR reaches a certain place and goes over it; (25) (temporal) expresses the overcoming of a time quantity: a woman crossed a certain period of her life, her 'best' years, and now she is beyond it. As for the contrast between auf/über, compare the following examples: (26)

Er geht schon auf die Achtzig zu. 'He is approaching his 80th year.'

(22)

Er ist über 18 Jahre alt. 'He is over 18 years old.'

In (26) the TR is still entirely inside the LM (80 years), while in (22) the TR is already over the LM: the age boundary of 18 years. The preposition auf can also focus precisely on the end point without crossing the boundary: (27)

Er hat das Taxi auf fünf Uhr bestellt.4 'He booked a taxi for five o'clock.'

whereas über refers to a fuzzy region around the end point LM, and therefore can be seen as crossing it:

The Germern 'über'

(28)

Er kommt über Mittag nach Hause. 'He spends lunchtime at home.'

(c/d)

Bevond metaphors

87

The two basic schemata (c/d) differ in static/dynamic features only (see Figure 11). Beyond metaphorical expressions are dealt with in the same paragraph since they present both static and dynamic features. Consider the following examples: (29)

Wir sind mit den Schwierigkeiten noch nicht über denACC Berg. lit. Our difficulties are not yet over the mountain. ( = 'The difficulties are still not overcome.')

(30)

Der Kranke ist über den ACC Berg. lit. The patient is beyond the mountain. (='The patient has the worst behind him.')

(31)

Der Dieb ist über alleACC Berge. lit. The thief is over all hills. ( = 'The thief is beyond our reach.')

In (29-30) the positive solution of life difficulties or problems are described as the overcoming of a high mountain; in (31) the TR (the thief) is unreachable now. The fact that beyond metaphors share both features (static and dynamic) of the corresponding schemata is signalled at the sentence level by a static locative verb {sein 'to be') and the morphological feature of case (ACC, typical for motion). This combination (static verb + motion ACC) is not allowed in basic spatial sentences, but appears only in metaphors. (e)

The covering image

Coincidence of opinions is often interpreted as coincidence of surfaces. Thus from the covering schema: (11)

derives:

Die Frau breitet die Decke über den Tisch. 'The woman spreads the table-cloth over the table.'

88

Elena Bellavia

(32)

mit jdm. über etwas einig werden lit. to become only one thing with s.o. ( = 'to reach an agreement with s.o. on sth.')

in which two persons discussing something reach an agreement.

2. Über as a prefix

2.1. 'Über'as a separable prefix Über can also be constructed as a verb prefix; as such it can be either separable or not separable. The choice of one of the two structures is motivated and not arbitrary, since different mental schemata pertain to expressions with verbal prefixes. First of all I will deal with the separable verb as in: (33)

Die Truppen setzten zum anderen Ufer über. 'The troops crossed to the other side of the river.'

The separable verb (33) depicts a scene in which the TR moves from one side of the LM's area (the LM is the implicit river) to the other and crossing it:

Figure 13. Die Truppen setzten zum anderen Ufer über.

In this case the LM boundaries are the two banks: S is the source and G the goal of the TR path inside the LM area. The trajectory described by the separable verb seems to be iconically cut (as the morphological structure) at the crossing point with the LM edge. Note that the point reached by the TR is introduced by the prepositional phrase zum anderen Ufer (to the other side of the river).

The German 'über'

89

In this structure (separable verb) S and G can both be expressed as in: (34)

Die Gangster sind vom Nachbarhaus (aus) auf das Dach der Bank übergestiegen. 'The gangsters climbed over from the neighbouring house to the roof of the bank.'

or only the goal is expressed as in (33). The goal can also be implicit, as in the ARC-schema configuration (Figure 14): (35)

Der Funke springt über. lit. 'The spark leaps.' (it springs through the air from one point to another)

Figure 14. Der Funke springt über.

Let us consider now some metaphors expressed by the separable verb. From the scene depicted in (33) derive: (36)

Hunderte von Soldaten sind zum Feind, zu den Rebellen übergelaufen. 'Hundreds of soldiers went over to the enemy, to the rebels.'

(37)

Seine Begeisterung ist in Fanatismus übergegangen. 'His enthusiasm has changed into fanaticism.'

All metaphorical expressions describe a reversal of circumstances, i.e. the TR goes from the positive to the negative side: in (36) soldiers go to the opposite side: the enemy; in (37) enthusiasm changes into fanaticism. Let us concentrate on the ARC-trajectory (see Figure 14): if the TR jumps, in order to complete the entire ARC, there must not be any contact with other concrete objects; otherwise the TR stops without crossing the LM.

90

Elena Bellavia

Now, as mentioned before, the separable verb describes a scene in which the TR goes across the LM and keeps going. That is why we cannot use the separable verb when the TR jumps and the LM is a concrete object as in: (38)

*Er springt auf einen Tisch über. 'He jumps over onto a table.'

since, if the man reaches the table and touches it, the contact stops him so that he cannot continue. In contrast, in the metaphorical world this kind of configuration is allowed, because there are no material constraints which prevent the TR from going through the LM: (39)

Ihre Fröhlichkeit sprang auf alle über. 'Her happiness spilled over onto others.'

(40)

Er ging auf ein anderes Thema über. 'He moved on to another topic.'

2.2.

'Über' as inseparable prefix

Let us consider the inseparable verb : (41)

Er überbrückt den Fluß. 'He bridges the river.'

Here the LM is seen as a whole and the TR goes beyond it. In this kind of sentence the trajectory is continuous - as the morphological structure - and extends to the entire length of the LM: TR

Figure 15. Er überbrückt den Fluß.

Note that if the verb is inseparable, the LM is the direct object of the sentence: einenACC Fluß; while with the separable verb the LM is implicit

The German 'über'

91

in the configuration, and the crossed point, if mentioned, is introduced by the prepositional phrase (cf. 33). Now consider the following metaphorical situations: (42)

Er geht auf ein anderes Thema über. 'He moves on to another topic.'

(43)

Ich übergehe diesen Punkt zunächst und werde später darauf zurückkommen. Ί will skip this point and come back to it later.'

In (42), expressed by the separable verb, the TR moves on to another point crossing it: the speaker changes topic and discusses the next one. Whereas in (43), described by the inseparable verb, the TR moves over the entire LM without any contact with it: it goes beyond it. The trajectory is continuous, i.e. the speaker jumps over the point, avoiding it for the moment. The inseparable verb often describes a scene in which the TR jumps over an obstacle as in: (44)

Der Junge überspringt ein Hindernis. The boy jumps over an obstacle.'

TR

Figure 16. Der Junge «Zerspringt ein Hindernis.

From the concrete scene (44) the following metaphors derive: (45)

Er überging ein Kapitel. 'He missed out a chapter.'

(46)

Das übersteigt meine finanziellen Möglichkeiten, meine Kräfte. 'That is beyond my finances, my power.'

92

Elena Bellavia

Another concrete scene which is very productive is exemplified by the above mentioned (41) (Er überbrückt den Fluß)\ from (41) the following metaphors derive: (47) (48)

Gegensätze überbrücken 'to bridge contradictions' Überbrückungsgeld, Überbrückungshilfe 'Money, help to tide s.o. over'

At the basis of all these metaphorical expressions there is the ARCschema. Depending on the situations, it takes on different meanings: the missing out of a chapter, as in (45); the sense of general inadequacy as in (46); or, more positively, the overcoming of certain attitudinal problems as in (47). Compare now the following sentences: (49)

Der Lehrer übersah einen Fehler. 'The teacher missed a mistake.'

(50)

Ich kann von hier aus die ganze Straße übersehen. 'From here I can see the whole street.'

In (49) the teacher inadvertently missed a mistake. In (50) the verb is the same: the inseparable übersehen ('to oversee'). But here the TR looks at the LM (the street) and thanks to the favourable position from the top, he can get an overview of it. In spite of the different meanings I suggest that the verb übersehen assumes the ARC-schema (cf. Figure 11) in both sentences (49) and (50), with the following differences: in (49) the mistake (LM) is seen as an obstacle inadvertently overcome by the teacher (TR), and the meaning of the expression is to go beyond the LM:

TR

Figure 17. Der Lehrer übersah einen Fehler.

The German 'über'

93

Therefore, scene (49) has to be considered more dynamic than scene (50) in which the observer (TR) overviews the whole street (LM) from a certain point. Since the observer (TR) is conceived as staying on a dominant position with respect to the LM (the street), the schema implies the total control of the TR over the LM:

TR

LM ^

Figure 18. Ich kann von hier aus die ganze Strasse «Versehen.

Overviewing' something can also mean possessing complete knowledge in all its parts as in: (51)

Die Folgen von etwas übersehen 'to have an overview about the consequences of something'

The meaning of this sentence is: in seinem Zusammenhang erfassen, i.e. 'possessing a mental hold or understanding of the situation'; it is common knowledge in fact that one can better control something from the top (Control is up). To sum up, I suggest that the ARC-schema is conceptually central for the prefix über in both separable and inseparable verbs (and more generally for the preposition über)·, but if the TR displaces from one point to another, the speaker uses the separable verb as in (42) ('Er geht auf ein anderes Thema über' — 'He moves on to another topic'). While if the TR moves over the entire LM, the speaker uses the inseparable structure as in (45) (Er überging ein Kapitel. — He missed out a chapter.).

2.3. The excess of the liquid out of its container Consider the following example: (52)

Die Milch fließt über den Rand. 'The milk flows over the rim.'

(prepositional phrase)

94

Elena Bellavia

(52) is an instance of a configuration which can be called "the excess of the liquid out of its container". Comparing the verbs expressing this kind of scene, I noted that German selects the separable verb to express the process of overflowing, that is the crossing of the liquid over the container's rim as in: (53)

Das Benzin ist aus dem Tank übergelaufen. 'The petrol overflowed from the tank.'

The choice is consistent with the previous pattern, since this structure describes the displacement of the TR from one side of the LM to the opposite side crossing the LM border:

Figure 19. Das Benzin ist aus dem Tank übergelaufen.

On the other hand, the inseparable verb is selected to express the complete covering of the exterior LM as in: (54)

Tränen überströmten ihr Gesicht. 'Tears flooded down her face.'

Here too the choice is consistent since this structure describes a trajectory that extends along the entire LM surface embracing it. In other words, the selection depends on the different mental schema assumed by each verb. As for the process of overflowing, the speaker can focus either on crossing the border, or he can profile the covering of the external LM area; in the first case he chooses the separable verb as in: (53)

Das Benzin ist aus dem Tank übergelaufen. 'The petrol overflowed from the tank.'

The German 'über'

95

Λ

Λ

1R

LM

Figure 20. Das Benzin ist aus dem Tank übergelaufen.

In the second case he selects the inseparable verb as in: (54)

Tränen überströmten ihr Gesicht. 'Tears flooded down her face.'

1R V LM

Al Figure 21. Tränen überströmten ihr Gesicht.

Related with this schema there are several metaphorical extensions expressing FEELINGS. Distinctions are connected with the separable and inseparable form. Compare the following sentences: (55)

Das Wasser ist übergeschwappt. 'The water slopped over.'

Separable (concrete expression)

(56)

Er ist vor Zorn übergeschwappt. 'His anger burst out.'

Separable (metaphorical expression)

96

Elena Bellavia

(57)

Der Strom hat das Land überflutet. The river flooded the land.'

Inseparable (concrete expression)

(58)

Ein Gefühl der Scham überflutet sie. Ά feeling of shame overcomes her.'

Inseparable (metaphorical express.)

Both (56) and (58) describe feelings: in (56), expressed by the separable verb, it is a feeling of great anger. Feeling comes out of the BodyContainer metaphor and it seems that the body flows over (Lakoff 1987). The schema is the same as in the concrete expression (55), in which the water has boiled over. The angry man has a sort of active reaction, i.e. the feeling does not subdue him. In contrast, in (58), expressed by the inseparable verb, the woman is assailed and totally covered by the feeling of humiliation. The emotion has total control of her. The schema is the same as in the concrete expression (57) where the river has flooded the land. In other words, the schemata assumed by the prefix über in the feeling metaphors are: Separable Verb: Out of the Container Inseparable Verb: Absolute Covering and Total Control Consequently the separable verb is often used to decribe a great happiness or a great anger, i.e. emotions that come from our body and which we discharge outside: (59)

Er ist vor Wut übergekocht. 'He has boiled over with anger.'

(60)

Er sprudelte vor guter Laune über. 'He was bubbling with happiness.'

On the other hand, the inseparable verb is generally selected to describe the variedness of the emotions (anger, happiness, curiosity, fear, humiliation, nostalgia), and in general human conditions, felt as external forces by which we are assailed and totally controlled such as: (61)

Ein Gefühl der Scham, Angst überflutet sie. Ά feeling of shame, fear overcomes her.'

The German 'über'

(62)

Ein gewaltiger Schreck, Heimweh überfiel uns. Ά terrible shock, nostalgia overcame us.'

(63)

Angst, Neugier, Freude überwältigte ihn. 'Fear, curiosity, happiness overcame him.'

97

Note that in the inseparable structure the LM (that appears in the direct object position) takes over the role of patient in the sense of Dirven's (1992: 53) definition: Through the role of a patient (...) we can understand the 'holistic' character of the process, i.e. the referent must in some sense be subjected to some action, which presupposes that it is wholly involved.

He goes on to say that ... the congruence between subject and direct object follows from the notion of 'patient': in order to be affected at all, the referent of the direct object must be subjected to some action (...).

This can easily happen when a large quantity of some entity completely covers the LM as in the following covering relations, all expressed by the inseparable verb. From the image: (64)

eine Statue mit Gold überziehen 'to gild a statue'

the metaphorical expression derives: (65)

Die Abendsonne übergoldet die Dächer. 'The evening sun gilds the roofs.'

From the concrete scene (66): (66)

Der Himmel überzieht sich mit Wolken. 'The sky clouds over.'

derives: (67)

ein Land mit Krieg überziehen 'to turn a country into a battlefield'

98

Elena Bellavia

and from the concrete scene (68): (68)

Der Strom überflutete das Land. 'The river flooded the land.'

the following derives: (69)

Ausländische Waren überfluteten den Markt. 'Foreign goods have flooded the market.'

2.4. Communication metaphors The process of communication is conceptualized as the transmission of a given message via a certain medium (channel). The selected construction in German is always the inseparable verb: (70)

(Er funkte die Nordgruppe an) und übermittelte dorthin die Informationen. 'He radioed the North group and transmitted the information.'

(71)

Der Rundfunk überträgt eine Veranstaltung. 'The radio broadcasts an event.'

(72)

jdm. eine Nachricht übermitteln. 'to send somebody a message.'

(73)

jdm. ein Paket, eine Nachricht übersenden. 'to send somebody a package, a message.'

The activated schema is the ACROSS schema. The structure of the sentence is transitive: an agent transmits a given message to a receiver. Consider now the following example: (74)

Er bekam die Anschrift über einen Freund. 'He obtained the address through a friend.'

In this case the prepositional phrase (über einen Freund) introduces the medium, through which the transmitter and the receiver can communicate. In a sense the message goes beyond the medium represented by the friend in order to reach the receiver (therefore über).

The German 'über'

99

The inseparable verbs in (70-73) focus instead on another aspect of the communication metaphor: the action of crossing the transmitted object. The fact that the two structures (the prepositional phrase and the inseparable verb) focus on two different aspects of the same communication metaphor seems to be confirmed by the following expression: (75)

eine Anfrage über Femschreiber übermitteln 'to send an inquiry by telex'

in which both structures appear together in one sentence. The message is transmitted via the LM (Fernschreiber), i.e. the physical entity, the place in which the activity occurs, which is introduced by the prepositional phrase; while the inseparable verb describes the action of transmission.

3. The choice of the auxiliary verb: intransitivity of the separable über- and transitivity of the inseparable über

3.1. Basic uses There is a relationship between the separable/inseparable use of the prefix über5, the transitivity/intransitivity of the clause and the choice of the auxiliary verb in compound tenses (Present Perfect etc.). Synthetically this relationship shows the following characteristics: 3.1.1. Inseparable über The inseparable prefix über gives rise to transitive structures (i.e. a verb prefixed by über in an inseparable construction is always used in a transitive way): (76)

Das Pferd hat den Zaun übersprungen. 'The horse leapt over the fence.'

In (76) the TR moves beyond an obstacle. The LM is expressed as the direct object and the auxiliary verb required is HABEN. The same scene could be expressed in an intransitive clause as:

100

(77)

Elena Bellavia

Das Pferd ist über den Zaun gesprungen. 'The horse leapt over the fence.'

In both sentences (76; 77) the horse goes beyond the fence, but in (77) the LM (introduced by the prepositional phrase) is conceived as the purely spatial reference point for the TR to go beyond; whereas in (76) the transitive clause stresses the superiority of the TR over the LM (the direct object of the sentence) and the obstacle appears to be dominated by the mover. 3.1.2. Separable über The separable prefix über gives rise to intransitive6 and causative structures: (78)

Wir sind mit der Fähre übergesetzt. 'We crossed on the ferry.'

(Intransitive structure)

(78) describes the TR's movement across the LM (from one side of the to the opposite). The auxiliary verb required is SEIN. If the movement is caused by an agent, the clause has a transitive frame. The auxiliary verb required is HABEN: LM

(79)

Der Fährmann hat uns mit einem Kahn übergesetzt. (Caused movement) 'The ferryman took us across in a boat.'

4. Transformations "The motion of physical objects through space is fundamental to our experience"; that is why the "extension of expressions describing spatial motion to motion in abstract domains" is frequent (Langacker 1987:166). Along this line let us now consider: (80) Der Übergang der Truppen über den Fluß. 'The passage of the troops beyond the river.' In (80) the noun Übergang ('passage') - which derives from the separable verb übergehen - describes the process of transition, while the prepositional phrase über den Fluß ('over the river') introduces the LM in

The German 'über'

101

which the action occurs. From the concrete scene (80) the abstract transition from 'waking' to 'sleeping' can be derived: (81)

Der Übergang vom Wachen zum Schlafen. 'The transition from waking to sleeping.'

Every process of transition is felt as a confused, indistinct time, after which a lasting state will necessarily follow. This is reflected in several concepts in German. The Übergangszeit (interim) can be defined either as a period which must be crossed in order to go from one stage to another, or a transitional season such as autumn or spring: two seasons that are thought as 'transitional seasons' to reach the winter or summer. For the same reasons, the Übergangsgesellschaft - proposed by Marxist ideology - can be defined as a transitional stage between capitalism and communism that had to be crossed in order to achieve the revolutionary transformation of the social order and of the human consciousness, after which the more stable phase of the establishment of the new state would have followed. Note that the separable verb übergehen can also be used - as well as the above mentioned noun form Übergang - for describing the processes of metamorphosis. Moreover, the sense of entering into a new condition is generally introduced by the preposition which is representative for the metaphorical sense of transition (Dirven 1989: 564): (82)

Die Leiche war schon in Verwesung übergegangen. 'The corpse had entered a state of decay.'

(83)

Das Gold der untergehenden Sonne ging langsam in Rot über. 'The gold of the setting sun slowly changed to red.'

In (82) the dead body had changed from an intact state into one of decomposition; in (83) the colour of the setting sun slowly changed colour from gold to red. On the other hand, sometimes the inseparable verbs can also express processes of transition, as in: (84)

Er hat einen Text vom Englischen ins Deutsche übersetzt. 'He translated a text from English into German.'

and more importantly, the two verbal forms can be used alternatively as in: 7

102

Elena Bellavia

(85)

Der Tote wurde in die Heimat übeiführt (übergeßhrt). 'The dead man was returned home.'

(86)

Eine Flüssigkeit wird in den gasförmigen Zustand überfuhrt (übergeführt). Ά liquid is transformed into its gaseous form.'

In my hypothesis, the question of the alternative use of the two verbal structures for describing the same transitional metaphor can be solved with the help of the concept of transitivity. As already mentioned, the separable über- is prototypically used in intransitive sentences; as such it requires the auxiliary verb sein. In contrast, the inseparable über is always transitive and requires the auxiliary verb haben (cf. section 3.). Between these 'extreme' situations, there exist others placed along the continuum, in a 'vague' area in which the morphological and semantic features which influence the choice of the form are in competition. In the above mentioned sentences there is a passage to a different point (85; 86). Semantically, the situation should be described by the separable über- (see section 3.1.2.); nevertheless both verbs are allowed. Why? The base verb führen is transitive. It would therefore require the inseparable über, but semantically the situation would pertain to the separable über (displacement to another point): these two factors are in competition. One could object that the separable form can also be used to describe a caused movement (transitive structures like Der Fährmann hat uns mit einem Kahn übergesetzt=He took us across in a boat) and therefore only the separable form should be allowed. The reason why both separable and inseparable forms are accepted is that the situations are not prototypically transitive: in fact, in (85; 86) the agent is never mentioned, signalling situations conceived as being in between transitivity and intransitivity. Since transitivity and intransitivity are equally 'strong', definite preference is given to neither the separable nor the inseparable form.

5. Comparison of «Aer-structures From the above discussion it can be concluded that in German the choice of the structure is motivated as represented in Figure 22:

The German 'über'

103

über Prepositional Phrase Inseparable Structure

Separable Structure (b)

(b2)

(e2) (C2)

(cl

(al) Die Truppen setzten zum anderen Ufer über. 'The troops crossed to the other side of the river.'

(a) Er geht über die Brücke. 'He goes over the bridge.'

(bl)

(b) Das Flugzeug fliegt über (b2) Wir überflogen den die Stadt. Ozean. 'The plane flies over the town.' 'We flew over the ocean.'

(cl) Der Funke springt. 'The spark leaps.'

(c) Er springt über ein Hindernis. 'He jumps over an obstacle.'

(c2) Das Pferd hat den Zaun übersprungen. 'The horse leapt over the fence.'

(dl) Das Benzin ist aus Tank übergelaufen. 'The petrol overflowed from the tank.'

(d) Die Milch fließt über den Rand. 'The milk flows over the rim.'

(d2)

(e]) ****************

(e) Die Frau breitet die Decke über den Tisch. 'He spreads the tablecloth over the table.'

(e2) Der Strom hat das Land überflutet. 'The river flooded the land.'

Figure 22. Compared structures of "über"

(a2) Er überbrückt den Fluß. 'He bridges the river.'

************

104

Elena Bellavia

The prepositional phrase represents the more general characterization of the movement: (87)

Er geht über die Brücke. 'He goes over the bridge.'

The verbal prefix represents specifications of this schema with the function of focussing on different elements of the process. In particular: (a)

the separable verb profiles the displacement of the TR from one point to another, crossing it:

(33)

Die Truppen setzten zum anderen Ufer über. 'The troops crossed to the other side of the river.'

(b)

the inseparable verb profiles the entire LM as a direct object overcome by the TR:

(41)

Er überbrückt den Fluß. 'He bridges the river.'

Note that the 'more basic' prepositional phrase appears in all configurations (see Figure 22), while the two prefix patterns are not activated in some cases, for instance, in scene (b): (3)

Das Flugzeug fliegt über die Stadt. 'The plane flies over the town.'

(prepositional phrase)

(88)

Mr überflogen den Ozean. 'We flew over the ocean.'

(inseparable verb)

*(separable verb: non activated) and in the covering relation (e): (11)

Die Frau breitet die Decke über den Tisch, (prepositional phrase) 'The woman spreads the table-cloth over the table.'

(57)

Der Strom hat das Land überflutet. 'The river flooded the land.' *(separable verb: non activated.)

(inseparable verb)

The German 'über'

105

The inseparable verb is not activated in scene (d): (52)

Die Milch fließt über den Rand. 'The milk flows over the rim.'

(prepositional phrase)

(53)

Das Benzin ist aus dem Tank übergelaufen, (separable verb) 'The petrol overflowed from the tank.' *(inseparable verb: non activated)

6. Conclusion As a general conclusion, I would like to propose a network symbolizing the whole semantic area covered by the lexical unit über. All major schema (and examples) of über analyzed in this paper will be organized and connected together in this network. Starting from the representation of vertical superiority (schema I), configurations develop continuously towards horizontal superiority (V) and to the Excess and Covering schemas (VII and VIII). The central meaning is best represented by the ARC schema (VI).

TR

TR

i t Ϊ —» ι

' r-> ι

i'LM

— A y

/

LH

VII

VIII

Figure 23. Network of "über"

106

Elena Bellavia

(I)

Die Lampe hängt über dem Tisch. 'The lamp hangs above the table.' Das Flugzeug fliegt über der Stadt. "The plane is flying over the town.' (inside the boundaries of the town) Das Flugzeug fliegt über die Stadt. 'The plane flies over the town.' (crossing the boundaries of the town) Er geht über die Straße. 'He crosses the road.' Sie wohnen über der Straße. "The house is across the road.' Das Pferd springt über den Zaun. 'The horse leapt over the fence.' Die Milch fließt über den Rand. "The milk flows over the rim.' Die Frau breitet die Decke über den Tisch. 'The woman puts the table-cloth on the table.'

II)

(III)

(IV) (V) (VI) (VII) (VIII)

Notes 1. 2. 3.

4. 5. 6.

7.

The notions of trajector (TR) and landmark (LM) are used in the specific sense given them by Langacker (1987). This sentence can also have a meaning connected with the above schema, i.e. the house is conceived as located on a hill, whereas the road is seen at the bottom of it. For this meaning cf. Figure 1. "Scales (...) can be represented as vertically oriented paths with some 'normative character' attached (Johnson 1987: 123). Norms immediately evoke the concept of value, because if something is at or above the norm, it is considered positive, otherwise negative." (Serra Borneto: 1995) The use of the preposition auf with this meaning is not very common in contemporary German (the selected preposition is fur)·, however, it is still accepted in literary texts. I only refer here to verbs that describe the displacement of the TR in which über has the across/beyond meaning. The other meanings über can assume as prefix verb (super/upper) are not relevant for transitivity. The only factors that I consider relevant for Intransitive Clauses are: (a) lack of an object; (b) no transferring of the activity. The action seems to remain 'inside' the receiver, to affect only him. Examples (85-86) are taken from the DUDEN (1981) and have been checked with native speakers coming from different regions of Germany and Austria. The vast majority of them accept both versions, even if someone prefers the inseparable version.

The German 'über'

107

References Bouillon, Henri 1984 Zur deutschen Präposition 'auf. Tübingen: Günter Narr. Brugman, Claudia 1983 The Story of Over. Trier: LAUD. Dewell, Robert B. 1993 Over again: On the role of image-schemas in semantic analysis. Paper presented at the Third International Cognitive Linguistics Conference. Leuven, July 18-23, 1993. Dirven, Ren6 1989 A User's Grammar of English: Word, Sentence, Text, Interaction. Frankfurt: Lang. 1992 Metaphors in Thought and Communication. University of Duisburg. Duden 1981 Das große Wörterbuch der deutschen Sprache. Bibliographisches Institut Mannheim. Ehrich, Veronika 1989 Die temporale Festlegung lokaler Referenz. In: Habel, Christopher, Michael Herweg and Klaus Rehkämper (eds). Raumkonzepte in Verstehensprozessen: Interdisziplinäre Beiträge zu Sprache und Raum. Tübingen: Niemeyer. Hawkins, Bruce W. 1984 The Semantics of English Spatial Prepositions. San Diego: UCSD diss. Johnson, Mark 1987 The Body in the Mind. Chicago and London: University of Chicago Press. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago: University of Chicago Press. Lakoff, George 1987 Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. Chicago: University of Chicago Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar, vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. 1990 Concept, Image, and Symbol. The Cognitive Basis of Grammar. Berlin, New York: Mouton de Gruyter. 1991 Foundations of Cognitive Grammar, vol. 2: Descriptive Application. Stanford: Stanford University Press. Neumann, Dorothea 1987 Objects and Spaces. Tübingen: Narr. Serra Borneto, Carlo 1995 'Liegen' and 'stehen' in German. A study in horizontality and verticality. In: Casad, Eugene (ed.). Cognitive Linguistics in the Redwoods. Berlin/New York: Mouton de Gruyter, 459-505. Smith, Michael B. 1987 The Semantics of Dative and Accusative in German: An Investigation in Cognitive Grammar. San Diego: UCSD DISS. Wahrig, Gerhard 1978 Wörterbuch der deutschen Sprache. München: Deutscher Taschenbuch Verlag.

The separability of German über-: A cognitive approach Robert B. Dewell

1. Received accounts One of the most striking aspects of the German language is the alternation of plain verbs with compound verbs which have either separable prefixes (stressed particles such as auf- or ein-) or inseparable prefixes (e.g. be- or ver-). A particularly interesting subset of these compound verbs involves prefixes which are sometimes separable and sometimes inseparable. This set of variable prefixes consists essentially of durch (corresponding to the preposition meaning "through"), um (corresponding to the preposition for "around"), unter (corresponding to the preposition for "under"), and über (corresponding to the preposition meaning "over"). 1 The standard handbooks typically illustrate the phenomenon using starkly contrasting pairs of separable and inseparable verbs, such as these taken from the pedagogical grammar by Schulz and Griesbach (1960: B340): (1)

a. Er setzt die Leute mit seinem Boot über. He sets the people with his boat over 'He takes the people across in his boat.' b. Ich übersetzte den Brief ins Deutsche. I over-set the letter into German Ί translated the letter into German.'

(2)

a. Er kam in Frankfurt nicht unter. He came in Frankfurt not under 'He didn't find lodging in Frankfurt.' b. Bitte, unterlassen Sie diese Bemerkungen! Please under-let (IMPERATIVE) these remarks 'Please refrain from these remarks.'

110

Robert Β. Dewell

(3)

a. Das Auto fährt den Mann um. The car drives the man around (here = toppling over in the vertical plane) 'The car runs over the man [knocking him down].' b. Wir umfahren die Stadt. We around-drive the city 'We drive around [ = detour] the city.'

(4)

a. Die Mutter schneidet den Apfel durch. The mother cuts the apple through 'The mother slices the apple.' b. Das Schiff durchschneidet die Wellen. The ship through-cuts the waves 'The ship cuts through the waves.'

The received wisdom, presented with varying degrees of sophistication, is that such pairs exhibit these broad tendencies: the separable verbs tend to have "literal" spatial meanings, and the inseparable verbs tend to occur in transitive constructions with "figurative" meanings. Such an analysis does point out crude tendencies, but there are serious problems with it. To begin with, the standard treatments usually fail to deal very convincingly even with their own chosen examples. These SchulzGriesbach sentences, for example, include separable constructions which are transitive (la, 3a, 4a), separable constructions which are figurative (2a, arguably also 3a), and - of primary interest to us here - inseparable verbs which are spatial and literal (3b, arguably 4b). They would have us believe, in effect, that a separable construction like Wir fuhren den Baum um ("We knocked the tree down [by driving over it or into it]") is more literal and spatial than Wir umfuhren den Baum ("We drove around the tree") - even though the inseparable construction preserves the basic meaning of the spatial preposition (cf. Wir fuhren um den Baum) while the separable construction has no prepositional counterpart. Moreover, traditional accounts consistently neglect a significant portion of the data, namely the obvious regularity reflected in sets such as (5). A typically "figurative" inseparable variant such as (5d) can be systematically related to spatial constructions with a prepositional phrase (5a) or a separable prefix (5b), by way of an inseparable prefix (5c) which is essentially spatial and presumably "literal": (5)

a.

Sie springt über den Graben. 'She jumps over the ditch.'

The separability of German 'über-'

111

b.

Sie springt hinüber. She jumps DIRECTIONAL PARTICLE- over 2 'She jumps over.' c. Sie überspringt den Graben. She over-jumps the ditch 'She jumps (over) the ditch.' d. Sie überspringt die zweite Klasse. She over-jumps the second class 'She is skipping the second grade.'

Detailed studies such as those by Kühnhold (1973) and Mungan (1986) do acknowledge these regular alternations, but they emphasize the semantic consistency in all the variants without making a serious attempt to account for the regular differences among separable, inseparable, and prepositional constructions (Eroms 1982: 34). Eroms (1982) addresses the issues of primary interest to us here, but he only treats durch and um. As for über (or unter), he implies (p. 49) that it is more difficult to analyze adequately and that the difference between concrete and abstract variants plays a significant role. There are also some intriguing aspectual differences between separable and inseparable constructions, related to the obvious link between inseparable compounds and transitivity. Although such issues go to the heart of the function of these prefixes (see section 4 below), existing studies do not get beyond Eroms' (1982) observation that inseparable compounds tend to have intensified perfective (resultative) force. 3 Finally, the received treatments also ignore what might seem to be the most obvious observation of all about the core set of variable prefixes, at least to anyone not concerned with separating syntax from semantics, namely that they form a coherent semantic group. They correspond to the English expression "over, under, around and through" and form the essential German set of terms for route paths. They all describe a path characterized medially as passing through a defining point or points relative to a landmark (LM). Route paths can be contrasted with destination paths (cf. in/ein-, auf, an), which are characterized by a defining end-location relative to a LM, and with origin paths (cf. von/ab-, aus), characterized by a defining source location relative to a LM.4 In short, the traditional treatments do not describe all the data adequately, and they make little attempt to explore some intriguing underlying regularities. There is much more to be learned here about some very basic phenomena such as the representation of paths, the development of specialized semantic variants from spatial-path images, and the role of prefixes in conveying aspectual contours or transitivity. This paper will suggest some

112

Robert Β. Dewell

new directions of inquiry into these issues within a cognitive-linguistic framework. More specifically, it will concentrate on the prefix über-, focusing attention on the regular relationships reflected in the core spatial variants of examples like (5).

2. Separable iiber-

2.1. Gapped LMs and medial abstracting The analysis of separable über- is straightforward. It is essentially like the prepositional constructions with über, except that it has a gapped LM which is pragmatically recoverable from the given context (cf. Talmy 1993, who discusses gapping the whole medial section of a path, not just the LM). If Levin (1993: 43) can speak of "preposition drop" alternations in English, we might call the (b) variants of examples (6)-(8) "LM drops": (6)

a.

Sie klettert über den Zaun. 'She climbs over the fence.' b. Sie klettert hinüber. S h e climbs DIRECTIONAL-PARTICLE o v e r ' S h e climbs o v e r . '

(7)

Sie ist über den Fluß gefahren. She is over the river ridden 'She rode across the river.' b. Sie ist (auf die andere Seite) hinübergefahren. She is (to the other side) DIRECTIONAL-PARTICLE over ridden 'She rode across (to the other side).'

(8)

a. Er hat sie über den Fluß gefahren. He has her over the river driven 'He drove her over the river.' b. Er hat sie (auf die andere Seite) hinübergefahren. He has her (to the other side) DIRECTIONAL-PARTICLE over driven 'He drove her across.'

a.

A prepositional wfer-phrase - the (a) variants of (5)-(8) - describes a smooth, continuous path scanned sequentially from an unspecified source

The separability of German 'über-

113

location to an unspecified destination, related to the LM at each point along the way. A separable construction (the corresponding (b) variants) portrays essentially the same durative path, but with the LM gapped, so that the path is construed with a sort of medial abstracting - typically describing transition from one discrete location to another. Separable über- thus tends to occur in situations when the LM can be construed as a highly abstract boundary. 5

2.2. Overflow and covering The notion of "overflowing" is an obvious fit for separable über-. In examples such as (9b), the LM is naturally construed as an abstract, gapped boundary - the boundary of a container. Although an explicit LM can be provided, e.g. the edge in (9a), it is typically presumed vaguely as an inherent abstract part of the container, in this case its "active zone" (Langacker 1984). 6 (9)

a. Das Wasser fließt über den Rand. 'The water is flowing over the edge.' b. Das Wasser fließt über. The water is flowing over 'The water is overflowing.' b'. Die Wanne fließt über. 'The tub is overflowing'

Incidentally, the special nature of container LMs is reflected in variant constructions with the container as nominative subject, e.g. example (9b'). Compare the English "swarm" alternations (Levin 1993: 53-54) like The bees are swarming in the garden vs. The garden is swarming with bees. English also has "overflow" counterparts like The water is overflowing out of the tub vs. The tub is overflowing with water, or The wine is running over (out of my cup) vs. My cup runneth over. Unlike "overflowing", the "covering" sense associated with prepositional über (e.g. 10a, 11a) does not seem to fit separable über- very well. When a trajector (TR) such as the cover in (10) or the water in (11) spreads to cover a LM surface, there is a sustained interaction between the TR and the LM which would normally suggest a fairly salient LM rather than a gapped one. And in fact, this variant of separable über- is marginal, occurring only in highly restricted variants like (10b) or (lib) with a personal dative. In all cases, the precise LM is a gapped portion of the person's surface, an implicit active zone which the TR covers, so that the construction can best

114

Robert Β. Dewell

be related to prepositional constructions like (10a') or (11a') rather than (10a) or (11a). Again, separable über- occurs with a gapped LM. (10)

a. Sie hat eine Decke über ihn gelegt. 'She laid a blanket over him.' a'. Sie hat ihm eine Decke über die Schultern gelegt. She has him-DATlVE a cover over the shoulders laid 'She laid a blanket over his shoulders.' b. Sie hat ihm eine Decke übergelegt. She has him-DATlVE a cover over laid 'She laid a blanket over him.'

(11)

a.

Sie hat Wasser über ihn gegossen. 'She poured water over him.' a'. Sie hat ihm Wasser über den Kopf gegossen. She has him-DATlVE water over the head poured 'She poured water over his head.' b. Sie hat ihm Wasser übergegossen. She has him-DATlVE water over poured 'She poured water over him.'

2.3. General observations Since separable über-vcrbs essentially describe a durative spatial path, they are typically intransitive (with a sew-perfect). Transitive separable übercan occur, though, with causative root verbs as in (la), (8b), (10b) or (lib) that describe setting an accusative TR in motion. Like spatial path expressions generally, separable über- can be used metaphorically in nonspatial domains. Nonspatial dividing lines or "boundaries crossed" occur with changes of state (12), remainders or other extensions beyond a temporal boundary (13), changes of sides of an issue (14), changes of topic (15), and the spread or overflow of abstract "substances" such as emotions (16). (12)

Die Leiche war schon in Verwesung übergegangen. The corpse was already into decay over gone 'The corpse had already gone into decay.'

(13)

Sie haben einen Baum überbehalten? They have a tree over retained 'They left a tree standing.'

The separability of German 'über-'

(14)

Er ist ins feindliche Lager übergewechselt. He is into the enemy camp over changed 'He switched over to the enemy camp.'

(15)

Der Redner sprang auf ein anderes Thema über. The speaker jumped to another topic over 'The speaker shifted to a new topic.'

(16)

a. Die Epidemie hat auf andere Gebiete übergegriffen. The epidemic has to other areas over grasped 'The epidemic spread to other areas.' b. Ihre Fröhlichkeit sprang auf alle über. Her happiness jumped to all over 'Her happiness spread to everyone.'

115

It is fair to say, however, that separable «&er-verbs are used nonspatially only according to principles that apply to spatial path expressions generally (e.g. the metaphorical conception of changes of state as movement from one state-location to another). In other words, it is the «for-path structure which is being applied metaphorically to a nonspatial domain. The specific combination of über- with a particular root verb does not introduce new metaphorical possibilities which are not available to über-paths generally. Finally, comparing the constructions with prepositional über with the separable über-verbs - the (a) variants vs. the (b) variants in examples (5)(9) - reveals a consistent difference in cognitive salience. The prepositional (a) variants have a balanced focus, describing a continuous sequence of spatial relations between the TR and the LM. Separable über- also describes a sequential path, but it is a spliced path focusing on the TR's transition from one location to another, reducing the LM to a highly abstract boundary crossed.

3. Inseparable über-

3.1. Transitivity and salient LMs The basic nature of inseparable über- is considerably more difficult to describe, but a few observations are fairly clear. To begin with, it is well known that inseparable über-verbs are transitive.8 As in transitive constructions generally, the accusative object is construed as individuated and af-

116

Robert Β. Dewell

fected holistically (Hopper and Thompson 1980). Like the very similar beverb of (lid), an inseparable über-verb like (11c) implies that the accusative object is covered by the water completely - an implication not shared by (lib). (11)

a.

Sie hat Wasser über ihn gegossen. 'She poured water over him.' b. Sie hat ihm Wasser übergegossen. 'She poured water over him.' c. Sie hat ihn mit Wasser über gössen. She has him with water over-poured 'She doused him with water.' d. Sie hat ihn mit Wasser begossen. She has him with water BE-poured 'She doused him with water.'

Moreover, as (11c) and (5c) illustrate, it is the LM of über which becomes the accusative object in the typical transitive constructions. As a result, the LM in the (c) variants becomes especially salient, not only in comparison with the gapped LM of (b), but also in comparison with the oblique prepositional object of the (a) variants. There is a clear progression in LM salience from the separable (b) sentences with their gapped LM; to the oblique prepositional (a) sentences with their balanced interaction between TR, LM, root-verb activity and über-path; to the inseparable (c) sentences, which we can now consider in more detail.

3.2. Regular spatial alternations, pragmatic niches and specialized meanings A key observation is that each of the separable spatial types potentially has an inseparable, transitive counterpart; i.e., there are regular correspondences, in the spatial domain, between constructions with prepositional phrases, separable über-verbs, and inseparable über-verbs. We have already seen the full set of alternations in examples (5) and (11), and we can also add inseparable variants to (6)-(7). (5)

a. Sie springt über den Graben. 'She jumps over the ditch.' b. Sie springt hinüber. 'She jumps over.'

The separability of German 'über-'

c. d.

(6)

a. b. c.

(7)

a. b. c.

d.

117

Sie überspringt den Graben. 'She jumps (over) the ditch.' Sie überspringt die zweite Klasse. 'She is skipping the second grade.' Sie klettert über den Zaun. 'She climbs over the fence.' Sie klettert hinüber. 'She climbs over.' ?Sie überklettert den Zaun. She over-climbs the fence 'She climbs (over) the fence.' Sie ist über den Fluß gefahren. 'She rode across the river.' Sie ist (auf die andere Seite) (hinübergefahren. 'She rode across (to the other side).' *Sie hat den Fluß überfahren. She has the river over-ridden 'She crossed the river.' Sie hat die Kreuzung überfahren. She has the intersection over-ridden 'She drove past the intersection.' Sie hat ihn überfahren. She has him over-driven 'She ran over him.'

This regularity has not seemed as obvious as it would otherwise have been, because most purely spatial inseparable variants are not that useful pragmatically. There is normally no reason not to describe the spatial path using the basic prepositional construction of the (a) variants. The inseparable (c) variants only come into consideration when the speaker wishes to mark the event as transitive (and aspectually more punctual, with a pronounced focus on the LM). (5c) is thus restricted to specialized contexts in which the LM is an obstacle crossed in a single bound (cf. überwinden)·, and überspringen is normally associated with even more specialized uses such as skipping a portion of a book or a grade in school (5d). (6c) is comprehensible and sanctioned by the Duden Deutsches Universalwörterbuch, but it sounds very odd to native speakers, probably because it is odd to collapse its path into a single event without durative extension. (7c) is even worse, since it is definitely odd to construe its path as a single punctual event; even though (7c) does not exist, however, it is

118

Robert Β. Dewell

still comprehensible. (7d) and (7e), moreover, are fine, filling more specialized pragmatic niches consistent with the general nature of inseparable compounds. The fact is that there is definite motivation in the language system for such potential uses whenever the pragmatic need arises. Some inseparable verbs, especially those such as (17)-(19) which focus on the LM as an obstacle to be overcome, unquestionably have firmly entrenched spatial-path uses. (17)

Sie überschreitet die Schwelle. She over-steps the threshold 'She crosses the threshold.'

(18)

Wir überfliegen den Ozean. We over-fly the ocean 'We are flying over the ocean (crossing it by flying).'

(19)

Es ist nicht gestattet, die Absperrung am Ufer zu übersteigen. It is not permitted the barricade at the shore to over-climb 'Climbing (over) the barricade at the shore is not permitted.'

"Covering" variants would seem to be a more natural fit than normal paths with inseparable über-verbs, since there is clear interaction with the LM and thus "transitive" motion directed "toward" it. And in fact, "spreading to cover" variants like (11c), (20c) and (21c) are natural: (20)

a. Der Strom flutet über die Ufer. 'The stream is flooding over the banks.' b. Der Strom flutet über. 'The stream is flooding over.' c. Der Strom überflutet die Ufer. The stream over-floods the banks 'The stream is flooding its banks.'

(21)

a.

Sie streut Zucker über den Kuchen. 'She sprinkles sugar over the cake.' b. *Sie streut (dem Kuchen) Zucker über. She sprinkles (the cake-DATlVE) sugar over 'She sprinkles sugar over.' c. Sie überstreut den Kuchen mit Zucker. She over-sprinkles the cake with sugar 'She sprinkles the cake with sugar.'

The separability of German 'über-'

d.

e.

119

Sie überzuckert den Kuchen. She over-sugars the cake 'She sugars (ices) the cake.' Sie bestreut den Kuchen (mit Zucker). She BE-sprinkles the cake (with sugar) 'She sprinkles the cake (with sugar).'

Even in the "covering" sense, though, inseparable über- tends to have somewhat restricted usage, as reflected in the oddity of (10c). (10)

a.

Sie hat eine Decke über ihn gelegt. 'She laid a blanket over him.' b. Sie hat ihm eine Decke übergelegt. 'She laid a blanket over him.' c. *Sie hat ihn mit einer Decke überlegt. She has him with a cover over-laid 'She covered him with a blanket.' d. Sie hat ihn (mit einer Decke) bedeckt. She has him (with a cover) BE-covered 'She covered him (with a blanket).'

There are several reasons. For one, alternative expressions are available. An inseparable über-verb competes not only with prepositional phrases like (10a) or (21a), but also with be-verbs such as (lOd), (21e), or ( l i d ) , as well as with verbs that incorporate the TR as in (lOd) or (21d). For another, specialized uses of inseparable verbs in nonspatial domains, e.g. the entrenched use of inseparable überlegen to mean "consider, ponder", are often so well established that the association would disturb their use in sentences like (10c). Finally, the inseparable compounds work best when the root verb focuses on the covering process itself, the manner of interacting with the LM. Verbs such as legen or setzen focus more on the end location of the TR and thus are not as natural with inseparable über- as verbs such as streuen, fluten, ziehen, spannen, or decken are. 9 Even stronger pragmatic restrictions apply in "overflowing" situations like (9), where a construction with a separable über-verb is definitely preferred. If the containing LM or its surface is to be given marked prominence, that can be accomplished by promoting the LM to sentence subject (9b·). (9)

a.

Das Wasser fließt über den Rand. 'The water is flowing over the edge.'

120

Robert Β. Dewell

b. Das Wasser fließt über. 'The water is overflowing.' b'. Die Wanne fließt über. 'The tub is overflowing.' c. *Das Wasser überfließt die Wanne. 'The water is overflowing the tub.' In sum, the potential for spatial uses of inseparable über-verbs is there in the language, motivated by established uses and ready to be exploited in particular situations when the pragmatic need arises for a transitive construction with a salient LM. But the pragmatic need tends to arise mainly to fill particular, marked descriptive niches, since other means exist to express very similar general relations (e.g. prepositional phrases, be-verbs, container-LM subjects). Thus inseparable über-verbs have come to be associated, even in the spatial domain, primarily with highly specialized readings such as overcoming an obstacle (5c) or overwhelming (7e).

3.3. Nonspatial variants The specialized, marked, spatial meanings of inseparable über-ve,rbs extend nicely to fill particular niches in nonspatial domains as well, as illustrated in (22)-(26). The (a) examples are intended to give a very rough indication of the spatial source for the nonspatial extensions. Variants with accusative LMs include overcoming (22), overwhelming (23), bypassing or overlooking (24), and surpassing or outdoing (25), as well as "covering" extensions like competitive layering (25d) and holistic perception (26). (22)

a. Er hat die Hürden überlaufen. 'He jumped the hurdles.' b. Er hat seine Angst überwunden. 'He overcame his fear.'

(23)

a. Die Kompanie überrannte die feindlichen Stellungen. 'The company overran the enemy positions.' b. Angst überwältigte ihn. 'Fear overwhelmed him.'

(24)

a. Er Überschoß das Wild, (hunting jargon) 'He overshot the animal.' b. Er übersah den Fehler. 'He overlooked the mistake.'

The separability of German 'über-

121

c. Er überging unsere Einwände. 'He ignored our objections.' (25)

a.

Sie hat den Lastwagen überholt. 'She passed the truck.' b. Sie hat ihn im Sport übertroffen. 'She outdid him in sports.' c. Sie überlebte ihren Mann. 'She outlived her husband.' d. Der Chor übertönte die Solistin. 'The choir drowned out the soloist.'

(26)

a. Die Sonne überstrahlt das Tal. 'The sun shines over the valley.' b. Sie hat die Rechnung überprüft. 'She looked over the check.'

All of these verbs can be traced semantically to spatial constructions, but they are variants specific to the particular verbs as units, filling restricted gaps in the lexical system, not general path images applied to other domains. As the glosses suggest, English over- behaves similarly to überin specialized nonspatial extensions. (Note that über- often corresponds to English out- for competitive surpassing, e.g. (25).) There is also a very productive set of "excess" variants for "overdoing" virtually any activity. Typical examples such as (28) can be traced to the specialized spatial images of "overfilling" or "overloading" a LM (27), the transitive counterparts to "overflow" variants such as (9). 10 (27)

a. Der Saal war überfüllt. 'The room was overfilled (overcrowded).' b. Sie haben den LKW überladen. 'They overloaded the truck.'

(28)

a. Das Land ist übervölkert. 'The country is overpopulated.' b. Er hat seine Augen überanstrengt. 'He strained [overexerted] his eyes.' c. Man kann die Sparsamkeit übertreiben. 'One can exaggerate frugality.'

These "excess" variants, especially a sentence like (28c), might also be related to the linear-path image of extending over a normative bound on

122

Robert Β. Dewell

some presumed abstract scale (see the extensive treatment of these variants in Ritsch 1995). In that case, the image would be related directly to separable constructions with a gapped LM like Die Zweige des Baums hingen über ("The tree's branches hung over (a property boundary)") and the accusative object would be the TR of the spatial path rather than the LM. There is established precedent for this direct relationship between inseparable and separable über- in a fairly productive group of abstract "transfers", as illustrated in the examples of (29). These verbs are direct counterparts to the basic transitive separable verbs for causing a transition from one discrete location to another. They differ from the separable verbs mainly in that they are restricted to fairly precise extended domains, particularly transfer of possession (29a), transmitting information (29b), transfer of medium (29c, 29d), or persuasion to change sides of an issue (29e). (As the category designations indicate, English often uses trans- in these meanings.) (29)

a. Er übergab die Angelegenheit seinem Anwalt. 'He turned the matter over to his lawyer.' b. Das Konzert wird live im Fernsehen übertragen. 'The concert is being carried live on television.' c. Sie hat den Roman ins Deutsche übertragen. 'She translated the novel into German.' d. Diese Insekten übertragen die Krankheit. 'These insects carry the disease.' e. Er ließ sich nicht davon überzeugen. 'He couldn't (wouldn't let himself) be convinced of it.'

These are the instances which have received the primary focus in traditional accounts (example (1)), precisely because they can be directly contrasted with transitive separable verbs as more abstract "figurative" extensions from a basic spatial image. Essentially, they correspond to the transitive separable verbs in the same way that other inseparable constructions correspond to intransitive separable verbs. They are typical in most respects: they are transitive, and they tend to have marked, specialized uses tied to specific contexts. They are atypical only because their accusative object does not correspond to the LM of über.

4. The underlying constituent structures We can begin to account more precisely for the descriptive observations made thus far if we take the analysis to a more syntactic level and focus on

The separability of German'über-'

123

the sequence of processing the constituents. In order to highlight the distinctions between inseparable compounds and prepositional or separable constructions, we will again focus on examples such as (5) and (7). (5)

a.

Sie springt über den Graben. 'She jumps over the ditch.' b. Sie springt hinüber. 'She jumps over.' c. Sie überspringt den Graben. 'She jumps (over) the ditch.' d. Sie überspringt die zweite Klasse. 'She is skipping the second grade.'

(7)

a.

Sie ist über den Fluß gefahren. 'She rode across the river.' b. Sie ist (auf die andere Seite) (hinübergefahren. 'She rode across (to the other side).' c. *Sie hat den Fluß überfahren. 'She crossed the river.' d. Sie hat die Kreuzung überfahren. 'She drove past the intersection.' e. Sie hat ihn überfahren. 'She ran over him.'

4.1. Prepositional and separable constructions In the prepositional constructions of (5a) and (7a), the preposition über combines first with its LM (Graben, Fluß), producing a path image in which a TR moves through a sequence of spatial states relative to the LM, ultimately describing a characteristic shape. Über is the syntactic head (profile determinant). The prepositional phrase then elaborates the valence of the verb, sharing its TR (sie), and providing a more detailed description of the internal structure of its path, a sense of beginning, middle and end for the event. Most verbs which combine with an über-phrase (e.g. klettern, fahren) are what Langacker (1987: 87) calls "bounded episodes"; i.e., they are internally homogeneous and unstructured except for a constant recurring activity. They do presume temporal limits on the activity and the path, but they depend on other elements, such as an «Z>er-phrase, to provide a more detailed specification. (Springen is atypical in this regard, since it implies a starting propulsion, a flight and a landing, making the «Zw-path largely redundant with it except that the path is related to the LM.) As a result, the

124

Robert Β. Dewell

w£er-phrase provides the internal aspectual contour of the construal, while the main verb provides the overall temporal profile. The whole process is structured as a path, with successive stages scanned through time, and the end result is a new location for the TR.11 With separable über-, the processing is exactly the same except that the LM is implicit and must be inferred from context before the über-path is combined with the root verb. Because the LM is gapped, the resulting path image seems sketchy and disjointed compared with a full prepositional phrase, but nothing else changes. The path expression {über) and the root verb {springen, fahren) each retain their syntactic and intonational independence as profile determinants, and the resulting verb phrase is the straightforward result of combining the two components. Über provides the specific aspectual shape of the event as construed, i.e., as consisting of a medial peak and an implicit starting and end location for the TR with reference to the LM, and that path fits neatly into the root verb's temporal profile as a generic durative path.

4.2. Inseparable compounds

4.2.1. The problem of aspect The inseparable w^er-verbs, however, are fundamentally different, and more problematic. They do not simply describe a sequence of spatial relations persisting through time until an end-location is reached by the TR (a sequentially-scanned path characterized by a mn-perfect). They describe an act carried out by the TR with respect to the LM, as reflected in their perfect tense form with haben. Unlike (7a) and (7b), sentences such as (7d) and (7e) are not appropriate responses to the question 'Wohin ist sie gefahren?' ("Where did she ride/drive (to)?"); they answer a question like 'Was hat sie gemacht?' ("What did she do?"). They do not describe where the TR goes with respect to the LM; they describe what the TR does to the LM. The precise aspectual character of these verb phrases is difficult to specify, because there are seemingly contradictory construals involved. On the one hand, they suggest an almost punctual construal that is not associated with either component of the verb alone, a sense that a holistic event is occurring rather than an extended path, and that there is a focus on the interaction with the accusative object, a vague sense of transitive action "aimed" at the LM. On the other hand, they seem to be describing exactly the same extended trajectory that the prepositional and separable con-

The separability of German'über-'

125

structions do. Despite all the evident differences, there is a strong intuition that we are dealing with the "same" über (and the same springen and fahren) in all the expressions of (5) and (7), that there is a semantic constant running through all the variants. In short, all the constructions seem to describe essentially the same situation using the same basic elements in the same basic meanings, yet there are important aspectual differences setting the inseparable compounds apart. The rest of this paper will be concerned with tracing these differences in construal to the sequence in which the constituents are processed.

4.2.2. The sequence of processing In an inseparable construction such as (5c), über combines first with the root verb {springen), not with its LM. It is not a profile determinant at any point in the derivation, and its reduced status is reflected in its reduced stress. The resulting compound verb describes a composite path image with a temporal profile, an image which is essentially the same as the final image for (5a) or (5b) except for one crucial difference: the LM has not yet been inserted (or even gapped as an implicitly recoverable element in context). In separable or prepositional constructions, the processing of the verb phrase concludes by combining über with the root verb; in the case of inseparable über-, the verb phrase is not complete until an accusative object is supplied. In fact, über- is never directly combined with its own specific LM. The only LM explicitly represented is an immediate constituent of the verb phrase; it elaborates a slot within the valence structure of the compound verb, not of über alone. In contrast to the separable and prepositional constructions, inseparable compounds introduce a direct syntactic relation between the verb and the accusative object, a relation that underlies the basic "transitive" implications that the verb is "aimed" at the object.

4.3. Constituent-order profile In order to understand the somewhat perplexing situation described in section 4.2.1., it will help to distinguish different levels of sequential construal operating in parallel. Most obviously, there is a temporal profile inherent in the objective scene as the TR moves through conceived time, and the path usually has a more detailed aspectual contour specifying a sequence of stages (the beginning, medial or end points of the path). There is also, however, an inherent sequential profile contributed by the syntactic

126

Robert Β. Dewell

order of processing a sentence, the order in which we assemble constituents to form a whole composite image. This constituent-order profile also contributes to our construal of the objective scene, i.e. to the meaning of the sentence. 12

4.3.1. Normal transitive profiles When we process a transitive verb, we are aware that the verb phrase is incomplete as long as the accusative object remains unspecified. Filling that valence slot is the concluding moment of processing the verb phrase. In conceptual terms, we first set up the verb's path image and then we look to specify its LM slot as the culminating "target" of processing. With typical transitive verbs, such as schlagen or betreten, that constituent-order profile is in harmony with the actual temporal sequence being described. The accusative LM is not only specified as the salient concluding moment of processing, it is also the concluding point of the path that the verb describes. Our natural attention flow (see DeLancey 1981: 633-35) and our syntactic processing path both arrive together at the same destination. All paths converge at the same target. 13

4.3.2. Transitive route-paths and cognitive dissonance In the case of inseparable über-verbs, however, the aspectual profile of the compound verb does not merge harmoniously with the transitive constituent-order profile. The path described by über is normally a route path that leads "past" the LM in both the spatial and temporal senses of the word. When we finish tracing the TR's motion, the LM is history; it is intact and unaffected by the TR's passing. The transitive profile inherent in the constituent order of the inseparable-verb phrase, however, introduces the LM at that point as the culminating event of processing. Thus we get a kind of cognitive dissonance between the two sequential profiles, and tension arises between a durative path leading beyond the LM and a transitive act implying a punctual arrival at the LM. When we process the compound verb in a sentence like (5c), we conceive an über-path with an aspectual profile like that of separable über, so that our natural attention flow ends on the other side of a gapped LM. When we then insert the specified transitive object of the verb, we are forced to loop back conceptually from the end of the über-path. Put another way, we are forced to take the whole summarily-scanned über-path and relate it dynamically to a LM which does not receive any direct transfer of

The separability of German'über-'

127

energy from it; we collapse the path event into a conceptual unit, a punctual act applied to an accusative object. The culminating event of the constituent-order profile (the moment when the verb image unites with the accusative object) does not coincide with the culmination of the verb path as described. 14

4.4. Some effects of constituent processing on inseparable iiber-verbs It is this cognitive dissonance, then, which explains the conflicting intuitions about inseparable über-verbs discussed in section 4.2.1. Essentially, what is odd about the transitive iiber-verbs is that the culmination of the über-path does not coincide with any moment of interaction with the LM; the constituent-processing sequence does not fully accord with the conceived temporal sequence. The aspectual profile of the über-path and the transitive profile of constituent-order processing are delivering mixed messages, and the peculiarities of inseparable über-vtrbs can all be understood in this context.

4.4.1. Dissonance effects For one thing, we can account for the fact that inseparable constructions are understood primarily as acts (with a haben-perfect) rather than as paths. When the two profiles conflict, the one which predominates will be the last one imposed. The aspectual profile of the «for-path is processed first, as part of the compound verb, and then it is wrapped into a summarilyscanned gestalt and thus backgrounded. It is relatively less salient than the transitive effect of specifying the accusative object as the final event in the processing sequence. Moreover, the cognitive dissonance of transitive route paths helps explain why inseparable über-\trbs are used sparingly to describe concrete spatial paths such as those in (5c)-(7c), especially when the root verb describes an ongoing activity such as clambering or driving which is strongly associated with an extended temporal profile. The more vivid our image of the TR's route path is, the more aware we are of the clash between that path and the transitive profile. Since the transitive construction involves focusing on a concluding moment of interaction with the LM, the compound verbs become more natural when there is a real transfer of energy affecting the LM (7e), or when the LM is an obstacle to be overcome, as in (5c). An obstacle is a mental "target" which the TR focuses on, and so achieving separation from

128

Robert Β. Dewell

it becomes a decisive, culminating moment which the whole path has pointed toward. Similarly, a sentence like (7d) is typically construed to mean inadvertently missing a turn which the TR had "aimed at" (which was to have ended that section of the path), so that the TR's anticipation parallels our conceptual anticipation of "reaching" the LM. Conceptually, we conclude the processing by looping back from the end of the über-path to focus on a LM which has already been passed, just as the TR looks back to see a target which has been missed. In all cases, inseparable iiber-verbs focus on the LM, implying some kind of conclusive interaction with it. Since that interaction does not coincide neatly with the end location of the TR, the construction is marked and usually carries specialized connotations.

4.4.2. Conceptual separation of the transitive LM from the path-activity Since inseparable über is combined with a root verb without first specifying its LM, there are actually two conceptually distinct LMs involved in the processing - the maximally abstract LM of the w^er-path and the explicit accusative LM of the transitive compound verb. The transitive object typically denotes the same entity as the LM of the über-path, but constructions have developed where that is not the case. As we have seen, there are inseparable über-verbs such as (29) in which the accusative object is the TR, not the LM of the über-path. These constructions make sense under the analysis being proposed here. The root verb (e.g. geben, senden, liefern, bringen, tragen, setzen, nehmen) is itself transitive, describing a higher-order activity which operates on its accusative object; and that object in turn is the TR in a transfer path. 15 When über combines with these roots, it adds aspectual contour to the path just as it did in separable transfers, suggesting an abstract, gapped but implicit LM boundary crossed. But über's TR does not correspond to the TR of the root verb; it corresponds to the root verb's direct object, which has yet to be specified. As with the other inseparable compounds, when the über-verb is processed it contains an unspecified slot for a transitive object which must be inserted as the final piece of the constituency puzzle. But now the focus element in the constituent-order profile is the TR of über, not its LM. Otherwise the constructions of (29) are typical inseparable über-verbs. The embedded transfer path is processed first, and then our attention loops back to focus on the insertion of the path's TR. This constituent-order profile, as opposed to that of separable verbs like (12)-(16), focuses on the moment when the top-level TR (the agent) acts on the transitive object, not on the object's durative motion from one location to another. The construal

The separability of German'über-'

129

is particularly natural when the transfer is a well-defined specialized act construed as a single event applied to the transitive object.

4.4.3. Autonomy of the compound verb Another important trait of inseparable compounds, given their constituentorder profile, is their status as relatively autonomous units independent of the über-lM. The w£er-path and the root-verb activity are wrapped into a single concept, and the path is re-construed as a summarily-scanned event before being joined with the accusative object. The gestalt construal of the whole path in turn introduces an element of singularity. Just as the moment of inserting the transitive LM added a sense of punctuality to the construal, so also does the related holistic processing of the path. Furthermore, inseparable über-verbs are more abstract than separable über-verbs, since they involve both the true gapping of the iiber-lM and the summary scanning of the path. As a result, the verbs lend themselves to marked, specialized, often technical meanings, and there is some evidence that they are losing ground diachronically in comparison with separable verbs (Eroms 1982). Their unitary status also means that inseparable compounds are more like specific individual lexical units, as opposed to general path expressions. As such, they are freed from restrictions on path expressions generally. The inseparable verbs can and do often develop idiosyncratic variants which are peculiar to the individual lexical compound - hence their common association with "figurative" meanings not generally associated with spatial path expressions. Since separable verbs can also have figurative variants, however, it is more accurate to adopt Weinrich's (1993: 1067) term and say that the specialized inseparable variants are "demotiviert", not strongly suggested by general principles for applying spatial path expressions to nonspatial domains. They represent the odd combination of a route path with transitivity.

5. Conclusions There is obviously much more to the analysis of separable and inseparable über-verbs than has been accomplished here. An adequate analysis will need to be embedded in a more general framework which includes studies of the other variable prefixes and of prefixes generally, of aspectual contours and temporal profiles, of constituent-order processing, of transitivity, of salience, and of much more. The observations made here do suggest

130

Robert Β. Dewell

some tentative conclusions, however. There are deep regularities and semantic constants underlying the alternations among prepositional phrases, separable prefixes, and inseparable prefixes. Separable prefixes are essentially like prepositional phrases except that the LM is gapped, resulting in medial abstracting. Inseparable verbs also describe essentially the same path as constructions with a prepositional phrase, but there are important differences which can be traced to the constituent-order profile. As an inseparable prefix, über is not a profile determinant and its LM is left fully unspecified as it combines with the root verb. The empty slot in the valence o f the compound verb must be completed by the insertion of an explicit accusative LM (typically corresponding to the LM of über). Its appearance becomes the concluding moment of processing, lending the construction a sense of punctuality and a sense of aiming at the transitive object as target - both of which are at odds with the basic aspectual profile o f the path described by the compound verb. The various semantic interpretations and restrictions on the use of compound-verb constructions make sense within this framework.

Notes 1.

2.

3.

4. 5.

6.

There are two other, peripheral and unproductive members of the set of variable prefixes: voll- (corresponding to the adjective meaning "full") and wieder- (corresponding to a largely archaic preposition meaning "against", and to the modern particle meaning "again"). The directional particle hin in the (b) variant simply adds a deictic spatial element to the construction, bringing the destination into conceptual view. It (or its partner herüber, indicating direction toward the deictic reference point) tends to be required with fully spatial uses of the separable prefix when there is no other specification of the destination. Bellavia (this volume) provides a semantic analysis of über from a cognitive linguistic perspective, taking into account the regular alternations between inseparable, separable and prepositional constructions. She proposes a continuum of transitivity, with prototypical separable verbs at the intransitive pole and inseparable compounds with accusative TRs at the transitive pole. Route paths appear to have other interesting common properties which bear investigating, particularly with respect to image-schema transformations such as endpoint focus and the "covering" variants. As with spatial path constructions generally, the TR can also be an object that extends in the shape of the path (so our attention moves along the path in subjective motion as we scan the extending TR). When the LM is gapped in the separable constructions, there can also be a reflexive TR. (The common use of English over for reflexive-TR images like turn over or overturn, however, corresponds in German to um-, not über-; see example (3a).) Although the container is usually a three-dimensional volume, it can also be a two-dimensional enclosure. The image-schema transformation link-

The separability of German 'über-'

7. 8.

9.

10.

11.

12.

ing the two images is reflected in transitional instances like Das Feuer griff auf die umliegenden Gebäude über ("The fire spread to the surrounding buildings"), which is a path over a gapped LM bound, but also suggests spreading out of a containing region and "overflowing" into other areas. This usage is colloquial, not accepted by all speakers. Intransitive inseparable uber-\erbs are virtually non-existent. The only examples I have been able to find are either rare frozen prepositional phrases turned into verbs (übernachten, überwintern, überborden)·, or have clearly recoverable deleted accusative objects (Im Süden des Landes überwiegt das Laubholz. Es überwog die Meinung, daß .. Der Moderator hat um 3 Minuten überzogen.)·, or, in one truly odd case, a sentence which is apparently like the container-subject variants (9b'), but with a covered LM as subject: Die nasse Straße überfror ("The wet street froze over"). I am grateful to the editors of this volume for bringing this factor to my attention. It explains why a sentence like *1Sie überhängte den Käfig mit einem Tuch ("She covered the cage by hanging a cloth over it"), taken from the Duden Universalwörterbuch, is better than (10c) to the extent that it suggests a characteristic manner of handling the TR, but is still not as natural as überziehen, überspannen or überdecken. Another possible factor is the likelihood that the accusative object could be a TR for the über-path, i.e., the object of the root verb - cf. example (29). A few über-verbs with this kind of highly general semantic variant have a stressed prefix, namely those like überbelasten, übererfüllen, überdosieren whose root verbs begin with an unstressed syllable. The handbooks treat these verbs as having separable über-, maintaining the principle that separable prefixes are always stressed and inseparable prefixes are always unstressed. The problem is that these verbs are not separable! They tend to occur only as infinitives, past participles, or inflected verbs in subordinate clauses, i.e., in syntactic positions where the prefix is never separated in any event. And in the rare event that they are used in inflected positions requiring separation, they are left inseparable: Er überdosiert oft. Das Werk übererfüllt seine Norm. Du überbeanspruchst ihn. Er überbetont diese Mängel. (All examples are from the Duden Universalwörterbuch.) Rather than treat these verbs as defective separable verbs, it makes more sense to treat them as inseparable verbs with unusual stress patterns which cause uncertainty in use. Langacker (e.g. 1987: 74, 1991: 152) would say that prepositional über is scanned summarily, but this assessment seems problematic to me. Certainly the combination of verb + prepositional phrase has to be scanned sequentially, so why assume the path-prepositional phrase is first construed summarily (which entails sequential processing plus a more complex cumulative gestalt construal) and then converted back to a sequential construal? Why not assume it is simply sequential in the first place? All Langacker seems to gain is the conjecture that sequential scanning is an exclusive property of verbs, but perhaps the exclusive property of verbs lies in a temporal profile as distinct from sequential scanning. A prepositional w&er-phrase could be sequential, but the passing of conceived time is not profiled until it combines with a verb. Obviously, the processing sequence being discussed here is not necessarily identical to the actual surface ordering of the elements in the sentence. In sentences such as (5c) with the compound verb in inflected

131

132

13.

14.

15.

Robert Β. Dewell

position, the surface order reflects the constituent processing order fairly well. But in sentences such as (7) the accusative object occurs before the verb. Logically, however, it cannot be assembled into a composite image for the whole verb phrase until the verb has been processed as profile determinant. The actual order of appearance of the parts of a sentence is of course yet another level of sequential construal. In effect, the aspectual paths all reflect the natural image-schema transformation according to which the final state of a path is profiled - e.g. "gone" from "go", or the (subjective) endpoint focus variants of routepath prepositions in English as in She lives over the bridge. Transitive origin paths (e.g. with verlassen) also require a conceptual loop as the LM is inserted at the beginning of the path described by the verb; but at least the sense of punctual aspect associated with the transitive fits the moment of separation described by the origin path. Compare ent-verbs, which focus on the moment of separation from the LM in a generic origin path, with fte-verbs, which presume a generic destination path and focus on the moment of arrival at the LM. There is no corresponding prefix for generic route paths, which explains why über, um, etc. have inseparable uses while particles like auf-, ein- or aus- can only be separable. As for covering images, they have a similar but distinct kind of problem in transitive inseparable constructions. Covering is normally construed as a process with an extending t r , not as a momentary event with punctual aspect. There are also compounds with nontransitive root verbs like zeugen or reden, which are used with über- to mean "convince" by analogy with normal transitive transfer roots like tragen (compare the conduit metaphor).

References Bellavia, Elena The German über. This volume. DeLancey, Scott 1981 An interpretation of split ergativity and related patterns. Language 57: 626-657. Drosdowski, Günther 1983 Duden Deutsches Universalwörterbuch. Mannheim/ Wien/ Zürich: Bibliographisches Institut. Eroms, Hans-Werner 1982 Trennbarkeit und Nichttrennbarkeit bei den deutschen Partikelverben mit durch und um. In: Eichinger, Ludwig M. (ed.). Tendenzen verbaler Wortbildung in der deutschen Gegenwartssprache (Bayreuther Beiträge zur Sprachwissenschaft 4), Hamburg: Buske, 33-50. Hopper, Paul and Sandra Thompson 1980 Transitivity in grammar and discourse. Language 56: 251-299. Kühnhold, Ingeburg 1973 Präfixverben. In: Kühnhold, Ingeburg and Hans Wellmann (eds). Das Verb (Deutsche Wortbildung: Typen und Tendenzen in der Gegenwartssprache, Erster Hauptteil) (Sprache der Gegenwart 29). Düsseldorf: Schwann, 141-362.

The separability of German'über-'

133

Langacker, Ronald W. 1984 Active zones. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 10, 172-188. (Also in Langacker 1991, 189-201.) 1987 Nouns and verbs. Language 63: 53-94. (Also in Langacker 1991, 59100.) Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin/ 1991 New York: Mouton de Gruyter. Levin, Beth 1993 English Verb Classes and Alternations: A Preliminary Investigation. Chicago: UP. Mungan, Güler 1986 Die semantische Interaktion zwischen dem präfigierenden Verbzusatz und dem Simplex bei deutschen Partikel- und Präfixverben (Europäische Hochschulschriften, Reihe 1, Bd. 886). Frankfurt a.M.: Lang. Risch, Gabriela 1995 Verbpräfigierung des Deutschen: Skalierungsverben mit über- und unter-, PhD dissertation, Stuttgart. Schulz, Dora and Heinz Griesbach 1960 Grammatik der deutschen Sprache. München: Max Hueber. Talmy, Leonard 1993 The windowing of attention in language. Technical Report 93-5, Center for Cognitive Science, SUNY Buffalo. Also available from L.A.U.D Series A, paper no. 337. Weinrich, Harald 1993 Textgrammatik der deutschen Sprache. Mannheim, Leipzig, Wien, Zürich: Dudenverlag.

Prepositional prototypes Sally Rice

1. Prepositional polysemy and the network metaphor Prepositions and postpositions are traditionally described as forming closed, though extremely complex, lexical classes whose basic meaning is spatial. Correspondingly, it is generally assumed that their basic function is to mark location, direction, or some sort of configurational relationship between entities. However, the lexical items we call adpositions routinely mark temporal and abstract relationships as well. And, as is evident in the partial catalogue of grammatical functions and example sentences given in (1), the English prepositions are no exception [1]: (1)

a.

HEAD OF PREPOSITIONAL PHRASE

She hid his letters in a box. They met in 1989. He got the guarantee in writing. b.

[spatial relation] [temporal relation] [abstract relation]

HEAD OF ADVERBIAL PHRASE

This park is so pretty with the trees in bloom. C.

AD VERB/ADJUNCT

Despite the weather, they continued on. d.

COMPONENT OF PHRASAL VERB

He was yelled at by his supervisor. e.

COMPONENT OF VERB PARTICLE CONSTRUCTION

Can you pick his face out in the crowd? f.

VERBAL PREFIX

He outran the competition. g.

NOMINAL PREFIX

They've moved to the outskirts of the community.

136

Sally Rice

h.

PREDICATE

They outed him in the magazine and he sued. Needless to say, prepositions participate in a number of diverse morphosyntactic constructions. Because of their enormous semantic potential and wide syntactic distribution, these items make an ideal lexical class for postulating and testing theories of lexical representation and processing. They are often claimed to be polysemous, and linguists, Al researchers, and lexicographers have spent a considerable amount of energy arguing about how best to represent their multiple sense types as well as usage potential. Over the past two decades, a number of studies have emerged on the semantics of prepositions in a variety of European languages, many of which have been conducted within a cognitive linguistics paradigm (see Rice 1993 for references). In our desire to demonstrate that even the lowliest grammatical morpheme is meaningful, we cognitive linguists typically posit multiple senses for these items and treat them as complex lexical categories. The spatial metaphor or topological concept most often employed by cognitive linguists to describe the distinct though related senses of these lexemes is the lexical network (Brugman and Lakoff 1988) [2], Lexical networks are usually described as integrated structures containing multiple, linked nodes. These nodes are ambiguously taken to represent either separate senses or separate usage types of the lexeme in question (a stricter distinction between sense and usage type is no doubt warranted—a point which will be addressed in Section 2—but for now I will continue to vacillate between the two terms for ease of exposition). The nodes extend out from a central node whose value is commonly taken to be the prototype of the entire lexical category. An idealized version of such a network is given in Figure 1 with the prototype indicated by the solid node in the center and distinct senses represented by separate nodes radiating out from it.

Prepositional prototypes

137

Figure 1. An idealized lexical network

In addition, the nodes are understood to be situated conceptually at varying distances from one another and from the semantic center. Thus, rather ironically, a spatial metaphor, that of an interconnected network, is routinely employed for representing the meaning of what seem to be predominantly spatial lexical items. Unfortunately, cognitive linguists have left most aspects of network organization unspecified (see Sandra and Rice 1995 for a more comprehensive critique). Most researchers have been as vague about identifying the prototype or core node for a given lexical network as they have been about defining what a prototype is in the first place (although Lakoff 1986: 33-36, Langacker 1987: 371, and Taylor 1989 are rare exceptions). As yet, we do not know whether the prototype or, to be more neutral, the central reference point (henceforth, CenRP) is definable, let alone identifiable. (Determining the nature of the CenRP is the focus of Sections 3 and 4). Moreover, the vast majority of researchers have been equally remiss in their tendency to exclude non-spatial but still prepositional usages from their analyses of prepositions [i.e. usages like the two non-spatial ones given in (la)]. The failure to include both has led to lexical network models that have either ignored or exaggerated differences between similar spatial usages. In general, network models are especially dependent on the

138

Salty Rice

particular analytic skills or subjective aesthetic of the individual researcher. This dependence is only exacerbated by a too-narrow focus on a single usage domain (space). It is for this reason, and because I am expressly interested in the entire range of prepositional 'behavior', that I include spatial and non-spatial usages alike in my descriptive analyses and experimental studies as well as usages like those in (la-e) in my experimental stimuli. In the next section, I briefly discuss indeterminacies associated with lexical networks and characterize two different approaches to network models in the cognitive linguistics literature.

2. What do lexical network models of prepositional semantics represent? As increasingly popular and potentially revealing as lexical network models appear to be in cognitive linguistics research, they are not uncontroversial. Nor have they yet been demonstrated to be anything more than a convenient notational device for capturing semantic relationships between purportedly distinct senses of a lexical item. As argued in Sandra and Rice (1995), cognitive linguists routinely invoke a very powerful, but completely unexplored spatial metaphor — the lexical network — for characterizing lexical and conceptual organization, but they do so without paying heed to the psychological consequences of their claims. For example, do networks exhibit some sort of isomorphism with mental structure? Or should we presume no relationship between the proposed models and mental representation? Also, cognitive linguists rarely acknowledge the inherent subjectivity of the network models they propose. One is left with a confused impression of what network models are supposed to represent: (i) some sort of logical or historical arrangement of semantic senses associated with a lexical item in the language or (ii) a blueprint of the conceptual territory associated with that lexical item in the mind of a speaker. In short, do they represent something linguistic or something psychological? The situation is complicated by the fact that there are a range of network models being proposed in the cognitive linguistics literature, just as there are differences of opinion about what constitutes a distinct sense of a preposition. For Lakoff (1987), networks neatly exemplify the radial structure of conceptual categories. He believes that categories are organized around a single core member out from which all other members extend. I have given an idealized version of a Lakoff-type model in Figure 1. Figure 2 is the actual network he proposes for over (1987: 436).

Prepositional prototypes

UJ Ü. X

u.

u.

I

•

Ο X

t U υ . Χ. • >

υ ζ χ

•χ-

υ ζ

>

οί

UJ Oi

i s Ί «β- ^ » "δ * b * a, ~ * ». s> s .8 ® ss

LU U X

Η Ω

Jü δ

«< •ε. J J ° -g ^ g , 'a : JOm ^ 9 ^ ΪΪ -2i 'S 65 a Γβ. s §"5 §·5 I I J S J I 3 ω *gg^ 3® · "3 α χω * J: ft.45*' «*

8 J!

s -β s ••a. Ü 1 * » Φ ·„ « Ε- „ίΐδS £ * .2 ο ί «Φ -η« SHεi ϊ ο.ti TJ1„ °ζ οΐ Ζ ο-; Sfrα•υ α•ί-ο ζζ; Φ3u=25 ! « ΰ Λ 3 β > ®*» ον»S.S >ΐ ' ο§·οο; . S ο I«ε l i l a g Ε=3Λ c ' 5 8« . £ Ä : Έ ° ί ο 3l. Φ « >,Φ = 3 φ= ° ~ ί) V ι u ~ -β « .£ Φ ηΧ Φ •''-•JSj ΦS i l »Φ ?ι β .c β>χ φ ε Ε 7 £ χ ® Χ £ = Ο λ Λ iriinr-y £TOX ίί5ο£σίS; WW 3 £ Ρö w HcocoXH^CO^XOHC, ;• W Λ· C( Η, "TO e «9 0Q 0ΌÖ ν η η s -s; ι Ν —ι Μ ι es —' Μ • ) (Ο t—ισΝ γν c** LU £

ο

CA

ω

ο

c

•α

J=

-

-

- -

IV

η

-

to

α

υ

Figure 8. Hierarchical cluster analysis for mixed conditions of on (Exp. 3)

&

Prepositional prototypes

157

The tree's structure reflects the relative strengths of relations between sentences as they are perceived by the subjects. Pairs of sentences which are most similar are joined first (with lower branching). Higher branching indicates little perceived similarity. An inspection of Figure 8 reveals that subjects were quite able to sort sentences in a non-random way. Most interesting was the fact that for each of the 3 conditions of the 3 prepositions (that is, for all nine hierarchical cluster analyses), the major division was brought about by the presence or absence of a temporal sense. Moreover, within the non-temporal division, lower branchings indicating higher perceived similarity tended to occur for spatial usages. Abstract usages tend to cluster with spatial usages away from temporal ones, suggesting that semantic extension, metaphoric or otherwise, proceeds most productively from spatial senses. Temporal senses are relatively inert by contrast. Finally, a Principal Coordinates Analysis was performed on the data. This is a descriptive technique for representing multidimensional data in three or fewer dimensions. It can examine a large number of variables without having to make any a priori assumptions of what the variables are (forinstance, my SPATIAL, TEMPORAL, or ABSTRACT categories). This multivariate analysis, which takes the results of a similarity matrix as input, determines the number of variables which give the best account of the data. Essentially, it collapses 60 dimensions (representing the stimulus sentences) into 3. In the graph in Figure 9, which, again, represents the results from the MIXED condition for on, the analysis has condensed the information from the sorting task into a 3-D space, reflecting the distribution of spatial, abstract, and temporal usages.

158

Sally Rice

Figure 9. Principal coordinates analysis for

MIXED

condition of on (Exp. 3)

Prepositional prototypes

159

In Figure 9, the first or X axis (horizontal width) seems to correlate with a spatio-temporal dimension, with space and time usages strongly opposed. Sentences (64), On Tuesday evenings, I generally go to the movies, and (2), The book you want is on the top shelf, represent completely polarized values along this dimension. The second or Y axis (vertical height) seems to correlate with a concrete/non-concrete dimension which suggests that prototypes (be they spatial or temporal) are experientially grounded. Most of the sentences involving clear-cut spatio-temporal usages register high values along the Y axis, while nearly all of the sentences containing abstract usages register low values. It is difficult to determine what is being represented by the third dimension or Ζ axis (depth) at this time, the endpoints of which are exemplified by sentences (43), They spied on him through the two-way mirror, and (50), I'm sure he'll give you his opinion on almost any topic. In short, it is hard not to conclude that spatial and temporal usages are mutually exclusive, though similar in their opposition to abstract usages. This general pattern was noted for all nine sets of sorting data (3 prepositions χ 3 conditions).

4.4. General discussion Taken together, the findings suggest that there are indeed prototypical semantic values for the English prepositions, at, on, and in, which are definitely spatial in meaning. However, the results also indicate that temporal senses are just as salient, and moreover, seem to be equally concrete and completely independent semantically. At the least, I would suggest that in the case of at, on, and in, we are dealing with a minimum of two central reference points, a spatial one and a temporal one. Temporal and spatial usages seem to have little in common, even though time is commonly taken by cognitive and historical linguists to represent the most basic and completely straightforward metaphorical extension from space. Metaphorical mappings were perceived between the spatial and non-spatial domains, although these might be task-specific effects. For example, a sorting task might encourage subjects to keep spatial and non-spatial (including temporal) usages well apart, while a similarity task might encourage subjects to relate usages across different domains such as space and time. If spatial and temporal usages were maximally contrasted under a more controlled version of a similarity rating task (as in Experiment 1), then we might want to ask whether speakers still conceive of the TIME domain in terms of SPACE. These somewhat surprising results make sense if we consider that some spatial and temporal usages are determined by experientially-based, conceptual archetypes. However, none of the abstract usages are grounded in

160

Sally Rice

bodily experience, which may explain why they do not cluster but extend outward in a variety of dimensions from the more archetypal/prototypical usages—chiefly spatial ones. Although the results of none of these experiments give us specific information about which particular senses are distinguished, they give us some indication that different background or referential domains are relevant to sense and/or usage type. Now, whether we want to attribute these distinctions to the inherent, stored meanings of the preposition or to a computation based on contextual effects is another matter entirely (cf. Kuteva and Sinha 1994).

5. Conclusion

5.1. A working "regional" model The general conclusion I draw from these experiments is that, thus far, it is still too early to propose an actual network for these prepositions in any detail at all. We are not quite at the node level yet. However, there does seem to be evidence for generalized regions within these topologically conceived lexical spaces—a finding consonant with earlier experimental work done on lexical polysemy by Caramazza and Grober (1976). My working hypothesis is that a prepositional network may ultimately reveal a structure with a temporal region and core sense, a spatial region and core sense, and a more diffuse but coreless abstract region surrounding them both. I have attempted a crude illustration in Figure 10.

Prepositional prototypes

161

I doubt if we will ever find much clustering, relatively speaking, amongst abstract usages. It would certainly be difficult to identify an abstract CenRP. These, after all, are the usages of a preposition that are the most difficult to characterize semantically. Moreover, abstract usages are more easily substituted by other prepositions. Sharp category contrast may only obtain for core senses. There is a need to explore perceived similarity or differences between different prepositions. One question which I have not specifically addressed in this paper but which should worry all cognitive linguists attempting lexical semantics is how much representation is necessary? If lexical networks are models of lexical representation, are they storage models, retrieval models, or models of interpretation? Of equal importance is whether or not they can give an account of lexical acquisition (by both LI and L2 speakers) as well as lexical loss, as in cases of aphasia.

162

Sally Rice

5.2. Directions for future research To begin to answer this next generation of questions, we will need converging evidence from multiple sources. I suggest three strands of evidence below: (i)

More refined experimental methodologies Can on-line lexical monitoring (reaction time) tasks be brought to bear on questions of category structure? Will we find the task harder in the context of non-spatial usages while easier in the context of spatial or temporal usages, which are presumably experientially grounded? I would like to run a more controlled version of Experiment 1 in which subjects have to make similarity judgments between constantly changing target and probe sentences to reduce the possibility of lexical effects or habituation to the target. Gary Libben (personal communication) has suggested a task along the lines of Kellerman (1978) in which native speakers rate prepositional usages on the basis of perceived difficulty of transferability for second language learners. Concept formation tasks have also been suggested to me by a variety of researchers.

(ii)

Acquisition studies It seems very likely that by studying the acquisition of prepositional usages by very young children we can learn a lot about prototypical versus non-prototypical usages and when and how they come into existence in an individual's lexicon. Likewise, by looking at the ease or difficulty with which second language learners acquire prepositional usages could reveal information about the conceptual vs. linguistic basis underlying lexical organization. It is common knowledge that prepositions are notoriously difficult for non-native speakers to master. But do basic or extended usages cause the most interference?

(iii) Diachronic research Finally, we need to compare experimental results against diachronic evidence in order to determine whether the appearance of newer extended usages of these prepositions in the historical development of English mirrors the degree of extension from prototypical values as rated by subjects.

Prepositional prototypes

163

Clearly, this type of research is still in its infancy. There have been few studies in what is essentially experimental semantics. Nevertheless, there are ways in which we can begin to garner empirical answers to questions like those raised in this paper. The hardest task will be in convincing lexical researchers in cognitive linguistics to begin asking testable questions of the network models they invoke.

Acknowledgments This research was supported by research grant 410-930205 from the Social Sciences and Humanities Research Council of Canada (SSHRC). I am very grateful to Terry Baxter and Tom Welz for their assistance in designing the computer programs to run Experiments 1 and 3 as well as Keith Egger, Terry Nearey, and Gary Libben for their suggestions for experimental design and their willingness to perform statistical analyses on the data from Experiments 1 and 3.

Notes 1. 2. 3. 4.

Although it is fairly clear that all the uses of the lexemes in and out in (1) are not, strictly speaking, prepositional (see Rauh 1991). The term is intended to be distinguished from semantic network, which linguists, psychologists, and anthropologists have long used for representing interlexemic relationships within a semantic field. See Deane (1992) for a critique of the lexical network Lakoff posits for over and his attendant assumptions about lexical polysemy. Admittedly, prototype is a loaded term, which is why I have resorted to yet another, but hopefully neutral term, CenRP, for immediate expository purposes.

References Brugman, Claudia and George Lakoff 1988 Cognitive topology and lexical networks. In: Small, Steven, Garrison Cottrell and Michael Tanenhaus (eds). Lexical Ambiguity Resolution: Perspectives from Psycholinguistics, Neuropsychology, and Artificial Intelligence. San Mateo, CA: Morgan Kauffman Publishers, 477-508. Caramazza, Alfonso and Ellen Grober 1976 Polysemy and the structure of the subjective lexicon. In: Rameh, Clea (ed.). Semantics: Theory and Application [GURT on Languages and Linguistics 1976]. Washington, DC: Georgetown University Press, 181206.

164

Sally Rice

Choi, Soonja and Melissa Bowerman 1991 Learning to express motion events in English and Korean: The influence of language specific lexicalization patterns. In: Levin, Beth and Steve Pinker (eds). Lexical and Conceptual Semantics. Oxford: Blackwell, 83121. Colombo, Lucia and Giovanni Flores d'Arcais 1984 The meaning of Dutch prepositions: A psycholinguistic study of polysemy. Linguistics 22: 51-98. Craig, Colette (ed.) 1986 Noun Classes and Categorization. Amsterdam/Philadelphia: John Benjamins. Cruse, D. A. 1990 Prototype theory and lexical semantics. In: Tsohatsidis, Savas (ed.), 382402. Deane, Paul 1992 Polysemy as the consequence of internal conceptual complexity: The case of over. In: Bernstein, Michael (ed.). ESCOL '92. Ithaca: Cornell University Press, 32-43. Geeraerts, Dirk 1989 Prospects and problems of prototype theory. Linguistics 27: 587-612. 1992 Polysemy and prototypicality. Cognitive Linguistics 3: 219-231. Grieser, DiAnne and Patricia Kuhl 1989 Categorization of speech by infants: Support for speech-sound prototypes. Developmental Psychology 25: 577-588. Herskovits, Annette 1986 Language and Spatial Cognition. Cambridge: Cambridge University Press. Jackendoff, Ray 1990 Semantic Structures. Cambridge, ΜΑ: ΜΓΓ Press. Kellerman, Eric 1978 Giving learners a break: Native language intuition as a source of predictions about transferability. Working Papers in Bilingualism 15: 5992. Kuhl, Patricia 1992 Infants' perception and representation of speech: Development of a new theory. In: Ohala, John, Terrance Nearey and Bruce Derwing (eds). ICSLP '92 Proceedings. Edmonton, AB: University of Alberta, 449-456. Kuteva, Tania and Chris Sinha 1994 Spatial and non-spatial uses of prepositions: Conceptual integrity across semantic domains. In: Schwartz, Monika (ed.). Kognitive Semantikforschung. Tübingen: Gunter Narr Verlag. Lakoff, George 1986 Classifiers as a reflection of mind. In: Craig, Colette (ed.), 13-51. 1987 Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press. Langacker, Ronald 1987 Foundations of Cognitive Grammar, Vol. I, Theoretical Prerequisites. Stanford: Stanford University Press. 1991 Concept, Image, and Symbol: The Cognitive Basis of Grammar. Berlin: Mouton de Gruyter. 1993 Reference-point constructions. Cognitive Linguistics 4: 1-38.

Prepositional prototypes

165

Lehrer, Adrienne 1990 Prototype theory and its implications for lexical analysis. In: Tsohatsidis, Savas (ed.), 368-381. MacLaury, Robert 1991 Prototypes revisited. Annual Review of Anthropology 20: 55-74. Miller, George 1971 Empirical methods in the study of semantics. In: Steinberg, Danny and Leon Jakobovits (eds). Semantics: An Interdisciplinary Reader in Philosophy, Linguistics, and Psychology. Cambridge: Cambridge University Press, 569-585. Nosofsky, Robert 1988 Exemplar-based accounts of relations between classification, recognition, and typicality. Journal of Experimental Psychology—Learning, Memory, and Cognition 14: 700-708. Posner, Michael 1986 Empirical studies of prototypes. In: Craig, Colette (ed.), 53-61. Rauh, Gisa 1991 Prepositional forms in the lexicon: Problems and suggestions. In: Rauh, Gisa (ed.). Approaches to Prepositions. Tübingen: Gunter Narr Verlag, 169-223. Rice, Sally 1992 Polysemy and lexical representation: The case of three English prepositions. Proceedings of the Fourteenth Annual Conference of the Cognitive Science Society. Hillsdale, NJ: Lawrence Erlbaum, 89-94. 1993 Far afield in lexical fields: The English prepositions. In: Bernstein, Michael (ed.). ESCOL '92. Ithaca: Cornell University Press, 206-217. Rosch, Eleanor 1975 Cognitive representations of semantic categories. Experimental Psycholog: General, 104: 192-233. 1978 Principles of categorization. In: Rosch, Eleanor and Barbara Lloyd (eds). Cognition and Categorization. Hillsdale, NJ: Erlbaum Associates, 27-48. Sandra, Dominiek and Sally Rice 1995 Network analyses of prepositional meaning: Mirroring whose mind-the linguist's or the language user's? Cognitive Linguistics 6: 89-130. Taylor, John 1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon. Tsohatzidis, Savas (ed.) 1990 Meanings and Prototypes: Studies in Linguistic Categorization. London/ New York: Routledge. Vandeloise, Claude 1990 Representation, prototypes, and centrality. In: Tsohatsidis, Savas (ed.), 403-437. Wierzbicka, Anna 1990 Prototypes save: On the uses and abuses of the notion of "prototype" in linguistics and related fields. In: Tsohatsidis, Savas (ed.), 347-367.

Space and movement in the English verb system Carlos Inchaurralde

0. Introduction In the pages that follow we shall put forward two different suggestions of how spatial notions can be used for analyzing some issues concerning the semantics of verbs in English. More precisely, both the tense-modality system and the lexical characterization of verbs can be analyzed through the use of an extension of the primitive concepts 'space' (geometrically defined as the intersection of η dimensions; every «-tuple designating a location in space) and 'movement' (change of location in space along an additional dimension). A geometrical perspective, in which we take physical space as a starting point, will, therefore, be adopted.

1. Tense and modality in English as multidimensional space

1.1. Multidimensional

space-time

At the most intuitively clear level, which most readily represents our view of reality, space has three dimensions (let us denote it by S 3 ), and time is a fourth dimension along which spatial changes occur. To make things clearer, we can use projections. In a two-dimensional plane, a point is used to represent either a point or a line, whereas a line represents either a line or a plane (when we represent three-dimensional objects on a sheet of paper we are using this kind of projection) (see Fig. 1).

168

Carlos Inchaurralde

r

Projection of a 3D point or line

Projection of a 3D plane or line

Figure 1. 2 dimensions

In a similar way, we can think of projections of objects with four dimensions in a three-dimensional space. A point can then represent either a point or a line; a line can represent either a line or a plane; and a plane can represent either a plane or a volume (see fig. 2). This allows us to represent four-dimensional objects in much the same way as we do with three-dimensional ones in drawings.

Space and movement in the English verb system

169

Figure 2. 3 dimensions.

Coming now back to our basic space, we find that the projection of the movement of a point on a three-dimensional coordinate system is a line (see fig. 3). And this movement corresponds to different locations in time, which becomes then a fourth coordinate. The time axis is typically being expressed in verbs by means of different tenses. The internal structure of time is covered by aspect. 1

170

Carlos Inchaurralde

Figure 3. 3 dimensions (only space). (4th dimension - time- does not have an axis. It is only projected).

On a second level of analysis we have space with four dimensions (that is, space-time; we may call it S 4 ). Movement in this kind of space takes place along a fifth axis (hypothesis) which can be accounted for in terms of possible worlds, and which, if we assume a restricted perceived speed of change, only covers a stretch of space representable in a way similar to Minkowski's light cones2 (see Fig. 5). Again, we can make use of projections: The projection of the movement of a point in S 4 on a three-dimensional coordinate system is a plane (see Fig. 4). This kind of movement is covered by the realis-irrealis dimension, which is expressed in language by means of modality devices (e.g. modal verbs). 3

Space and movement in the English verb system

171

Figure 4. Projection in 3D of the temporal evolution of a point in 5D space (space + time + possibility) This hierarchy might be extended: S 5 has five dimensions (space-timehypothesis), and movement takes place along hypothetical hypothesis (let us denote it by HH); however, for practical reasons, we shall restrict ourselves to movement in S 4 .

1.2. The logical approach: possible times across possible worlds If modality adds a dimension to temporality, it should show in linguistic analyses borrowed from logic, which is in fact the case. Standard temporal logic (Prior 1967) was a prepositional language with operators F (Ff: 'It will be the case that f (at least once)') and Ρ (Pf: 'It was the case that f (at least once)'). With the use of these operators, several verbal tenses can be represented: PRESENT (p: 'Peter runs'), FUTURE (Fp: 'Peter will run'),

172

Carlos Inchaurralde

SIMPLE PAST (Pp: 'Peter ran'), PAST PERFECT (PPp: 'Peter had run'), FUTURE PERFECT (FPp: 'Peter will have run'), CONDITIONAL (PFp: 'Peter would run'). Apart from F and P, we also have operators G ('it will always be the case that') and Η ('it was always the case that'). This logic also makes use of a basic formal semantics with a model Μ = {Τ, < , V}, Τ being a set of time points t, < an order relation, and V a truth valuation which gives for the proposition ρ the set V(p) of times in which ρ is true. Van Benthem (1988: 7) gives the following recursive definition for a semantics of this kind:

Μ |= Φ[ί] ('Φ is true in Μ at t') Μ \= -K*[t] if not Μ |= Φ[ί] (and likewise for Λ /and, V/or, — > / i f ... then, etc.) Μ \= F$[t] if for some t' > t, Μ (= Φ[ί'] Μ P«[t] if for some t' < t, Μ (= *[t'] (and likewise for G, H) However, a temporal treatment may appear combined with modal elements, as van Benthem (1988: 32) himself shows. In this case, we have models with possible worlds across times (W χ T) where evaluation takes place in an index with two components: Μ (= $[w, t] ('Φ is true in Μ in the possible world w and the time t') Operators F and Ρ are defined in a way similar to their equivalents with only a temporal index: Μ (= ΡΦ|\ν, t] if for some t' > t, Μ (= Φ ^ , t'] Μ ΡΦ[λν, t] if for some t' < t, Μ (= Φ ^ , t'j What we have in this case is combinations of possible worlds in possible times. In this context, necessity can be interpreted in different ways (van Benthem 1988: 32): Μ Kl t] Μ μ Φ[νν', f ] Μ ^[w',t]

if for all w', t' for all w' R t w

In the second case we have a family of accessibility relations Rt for each temporal moment. This interpretation poses very interesting metaphysical implications. Van Benthem suggests that it expresses different parallel

Space and movement in the English verb system

173

developments in time, which can be represented by a tree with several branching paths of events:

t

(van Benthem 1988: 32)

1.3. The interaction time-modality Modality expressed logically in terms of possible worlds (certainty and probability) has been extended in order to interpret obligation/permission (deontic logic), and knowledge/belief (epistemic logic). All these categories may well fall under the label 'irrealis', which Givon (1993: 172 ff.) uses as denoting a characteristic feature of modality. We assume that modality has to do with the speaker's attitude towards a proposition; this attitude, as Givon states, concerns two types of judgement (Givön 1993: 169): 4 (a)

Epistemic judgements of truth, probability, certainty, belief or evidence.

(b)

Evaluative judgements of desirability, preference, intent, ability, obligation or manipulation.

All these can easily be explained by means of the possible worlds approach that has just been mentioned. 5 In addition, we also have high predictability of the following correlations between tense-aspect and epistemic modality (Givon 1993: 171), which is a proof of interaction between the time axis and the hypothesis axis:

174

(a) (b) (c) (d) (e)

Carlos Inchaurralde

Past Perfect Present Future Habitual

=

=

=

>

=

=

=

>

=

=

=

>

=

=

=

>

=

=

=

>

R-assertion (or presupposition) R-assertion (or presupposition) R-assertion iRR-assertion iRR-assertion

All this is congruent with the branching-path perspective mentioned above. The present is closed and comes from a single path of events that are already fixed in the past; but the future is open and there are parallel possible paths. Their number and 'width', so to speak, is limited only by the speed of change in a Minkowski-cone-like perspective.6 Givon (1993: 178, n. 10) gives a cognitive explanation for the overlap of the irrealis modality with the future. According to him, The reasons for this are most likely cognitive, having to do with differences in the strength and stability of the cognitive representation of memorized vs. imagined events/states.

Something could be added to this statement: our experience of reality tells us that the past is one and fixed, whereas future has different possible parallel paths. Of all these different alternative possibilities, only one will really take place, but we do not know which yet. This leads to a cognitively-felt tentative status for all the paths, a status which can be adequately termed as 'irrealis'. The most characteristic irrealis operators in the English language are modals, which may involve the two types of judgement mentioned above (epistemic and evaluative), although there seems to be a very direct relation and dependence between them. Epistemic judgement is implied whenever evaluation is involved, which means that a scale from negative to positive certainty with different probability values in between can be applied to all kinds of modality. Preference is equated with uncertainty (irrealis), "due to the fact that intention, ability, preference, permission and obligation are all future projecting" (Givon 1993: 172), and the future, as has already been mentioned, is typically an irrealis mode. English modals tend to code several irrealis sub-modalities. Thus, we have can for ability, probability, and permission; should and must for obligation and probability, etc. This can be considered to be a consequence of the above-mentioned relation across sub-modalities; although this relation is also to be understood as the result of a diachronical process of desemanticization which fixes different senses derived from the same original 'root' meaning (cf. Heine 1993; see also note 5).

Space and movement in the English verb system

175

1.4. A three-parameter system for modality We can integrate this modality dimension into a general scheme that is intended as an extension of Reichenbach's three-parameter temporal system. In his analysis (Reichenbach 1947), verbal tenses can be classified according to the following temporal points: time of event (E), time of reference (R), and time of speech (S). If we follow an order relation (denoted by < ; time overlap is denoted by = ) , we have the following combinations: Present Past Future Present Perfect Past Perfect Future Perfect

(R = Ε = S) (R = E) < S (R = S) < Ε Ε < (R = S) Ε < R < S S < Ε < R

The three parameters can now be used not only as points in temporal space, but also as points in hypothetical space. Thus, the expression of hypothesis can be compactly explained together with time. If we look at Fig. 5 again, we can see that in the PAST cone 7 there is an actual path (certainty), some other paths that could have been possible since they could have led to the PRESENT point {possibility), and an area outside the cone in which no path could have led to that point (impossibility). In the FUTURE cone the picture is pretty much the same, although, as we are now in the irrealis mode, there is no certainty path but only a highest possibility one.

176

Carlos Inchaurralde

Figure 5.

Space and movement in the English verb system

177

As it is clear from the cross-sections of the PAST and FUTURE cones, there are at least three degrees of 'reality' (certainty, possibility, and impossibility or uncertainty). Again, we can set three parameters: reality of event (Er), reality of reference (Rr), and reality of speech (Sr). The reality of event stands for the degree of reality that the main event has; the reality of reference may be the certain condition or assumption that is used for assigning realis or irrealis status to the event; finally, the reality of speech has to do with the utterance itself. If we now look at epistemic modality as it is expressed through modals, we can have the following possibilities, by using the three-parameter formulation for hypothetical space (All examples adapted from Swan 1980: 389 ff.): 1. Certainty

(Unmarked schema: [R^ = E r = S r ])

shall, shall not, will, will not, must, cannot, could not, would, would not e.g. (1) / shall see you tomorrow. (2) Things will be all right. (3) It won't rain this evening. (4) That can't be John - he's in Dublin. (5) I knew it couldn't be John. (6) This child would one day rule all England. (7) I told you you wouldn't be able to do it.

178

Carlos Inchaurralde

I Future + Certainty (Highest possibility): (1), (2) and (1)

Near future + Certainty (Highest possibility)

Present + Certainty (Highest possibility): (4)

I

Past + Certainty: (5)

Future of the past + Certainty: (6) ((7) the same, but uttered in the past: Certainty becomes Highest Possibility)

Space and movement in the English verb system

2. Probability

179

(Unmarked schema: [Rf = E r < S f ])

should, should not, ought to, ought not to, may (not) e.g. (8)

He should be here soon.

(9)

It shouldn 't be difficult to get there.

(10)

We may be buying a new house.

(11)

The water may not be warm enough to swim.

I Proximal future + Probability (High possibility): (8) and (10) (9) and (11) are better explained in terms of a structured world model (Goldsmith and Woisetschlaeger 1982; Langacker 1991: 264). 8

3. Weak probability

(Unmarked schema: [Ef < R r < S r ])

might, might not, could e.g. (12)

I might see you again - who knows.

180

Carlos Inchaurralde

(13)

Things might not be so bad as they seem.

(14)

We could all be millionaires one day.

Future + Low probability: (12), (13) and (14)

4. Theoretical or habitual possibility (Unmarked schema: [Rj. = E r < S r ]) can e.g. (15) (16)

How many people can get into a telephone box? Scotland can be very warm in September.

Both (15) and (16) can be well understood in terms of a structured world model (see above and note 8).

5. Conditional certainty or possibility (Unmarked schema: [Rr < E r = S r ]) would, would not, could, could not, might, might not

Space and movement in the English verb system

(17) (18) (19) (20) (21) (22)

181

If we had enough time, things would be easy. I wouldn't do this if I didn 't have to. If John came we could all go home. I couldn't do anything without your help. If you stopped criticizing I might get some work done. It mightn 't be a bad thing if we took a holiday next week.

Conditionals, whether clearly marked by a suitable subordinator or not, involve shifting the deictic center (cf. Langacker 1991: 266-269). In the three-parameter schema proposed here the reality of reference has a higher degree of uncertainty. If this condition were to become true, E r and S r should also become true (in a basic conditional; if we use modals like 'may', 'might', 'can', or 'could', we change their degrees of reality somewhat).

2. Abstract space as conceptual space

2.1. Abstract space The second topic that is going to be dealt with here concerns the use of non-physical space as a notion that helps in characterizing verbs and in classifying them. For a start, we need to look again at the close relation between space and time. Both have a similar 'cognitive' structure, which is shown by the fact that space concepts can substitute for time concepts, as is clearly seen in the large amount of spatial metaphors used for expressing time (cf. Lakoff 1991: 55 ff.). It is also evident in the fact that there are many notions that can be used for parametrization of both space and time: length (duration), path, interruption, location, start point, end point, limit, anteriority, posteriority, etc. Accordingly, kinetic and temporal situations may be represented in a similar way (this is possible as well for many other kinds of situational representation, which may become spatial by means of metaphorical extension). We can establish many correspondences between a passage-of-time situation (T) and a kinetic one (K): duration (T) « length of trajectory (K), factual time (vs. hypothetical) (T) actual path (K), interruption (T) = stop (K), temporal location (T) « spatial location (K), start (Τ) ~ source

182

Carlos lnchaurralde

(Κ), end (Τ) » goal (Κ), anteriority in time (order relation) (T) « spatial anteriority (K), posteriority in time (order relation) (T) = spatial posteriority (K), etc. The basic correspondences could also be found with respect to other kinds of situation, by extending spatial notions to non-spatial domains. This fact and the need for an explanation of how metaphorical extension takes place has been pointed out by Langacker (1987), as can be seen in the following quotation (1987: 168): Terms for motion and other spatial relationships are commonly extended to nonspatial domains. For example, the basic sense of go pertains to the movement of a physical object through space, (...), but this verb has many other conventional uses, (...). In analyzing such data, one's first inclination is to appeal to spatial metaphor, (...). However, recognizing that metaphor is a factor marks only the beginning of the descriptive task. Explicit analysis is required for both the nature of the metaphorical extension and the structures it relates.

Langacker mentions two different interpretations for this metaphorical mapping. The first one claims that no independent conception across domains is possible, for the target domain, as a matter of fact, deals with physical motion through space. This physical motion is, however, very alien to physical reality. The second one, which he clearly supports, claims that it is possible to construe the target domains and relationships independently and without reference to space. As evidence for independent conception, he mentions the existence of alternative expressions that do not rely on spatial metaphor: (a) (b) (c) (d)

It only takes five seconds to recite the alphabet. I read the book in just three hours. He can quickly change moods. This milk is about to become sour (from Langacker 1987: 169).

Figurative meanings that become established as a conventional value of an expression may lose their metaphorical character after being activated several times. As Langacker (1987: 169-170) says: This type of analysis presupposes a generalized notion of extensionality that is not specifically tied to our conception of physical space. It is a property of many domains, both basic and abstract, though the spatial domain stands out among them for its prototypicality and cognitive salience. By making this distinction (...), we can characterize "motion" in abstract terms applicable to any extensional domain.

Space and movement in the English verb system

183

This conception of 'motion' is the one that we are going to use in what follows: an abstract conception that is isomorphic with physical motion, as far as its internal structure and relations are concerned.

2.2. Eventive structures in descriptive texts In any kind of discourse, we find predications linked to entities (characters, actors, actants, or whatever we want to call them), which constitute events. 9 These events occur along time within certain settings in which several characters 'act'. From a generic point of view, event structure has been characterized, as referred to by Lakoff, as having a clear metaphorical relation with space, motion, and force in English. The general mapping he suggests (Lakoff 1991: 57) is the following: States are bounded regions of space. Changes are movements into or out of bounded regions. Processes are movements. Actions are self-propelled movements. Causes are forces. Purposes are destinations. Means are paths to destinations.

This mapping will need some minor modifications for its application to the analysis of specialized language, but basically it equates narrative structure with descriptive structure, since both can be described in terms of motion of entities through space(s). To illustrate the point, we can use here some examples taken from a sample article (Kranakis 1987) which was written using the conventions of mathematical language, of which the most important one is the coexistence of both a natural language (English) and a variety of mathematical symbolic language. In the text we are examining, the following verbs can be found: abbreviate (2), advise, apply, arise, assemble, associate, assume, be, build, call (3), choose, close, complete (4), consider (7), consist of (4), consult, contain, dedicate, define (16), delete, denote (4), derive (2), describe (2), desire, embed, enable, endow, exclude, exist (3), explain (2), express, extend (3), find, follow (8), force, give (6), give rise, guarantee, have (4), hold, identify (2), imply (3), include (2), indicate, interpret, involve, keep, know (2), leave (3), let (6), locate, make it possible (2), motivate, need (2), notice, obtain (3),

184

Carlos Inchaurralde

occur, omit (2), prove (8), quantify, range over, refer to, replace, require (5), satisfy (10), see (2), seem, show (11), state (2), study, summarize, suppose, take, treat (2), turn into, turn out, use (9), verify, want. These verbs can be considered to refer to predications that reflect basic scenes 10 , in which one of two alternatives is possible: they can reflect either states or changes of state. A mapping onto a space-and-motion domain is always possible; in this case, we can say that the former reflect space configurations, whereas the latter refer to spatial motion. For this mapping we need an abstract conception of space, similar to the one referred to by Langacker as "any extensional domain". Still, the concept SPACE can be relevant here, since there are some predicates which are typically spatial in their unmarked uses (such as move, come, go, group, etc.), whose meaning can be further extended, in order to deal with other domains. In this sense, we assume the spatial domain to be the basic one. Configurational predicates express basic spatial relationships between entities. In the text we have expressions that refer to spatial 'states'. However, these spatial relationships can be altered by means of motion. This motion can be undefined as to the number of spatial regions involved or it can clearly be motion from a bounded region to another. Characteristically, we have the following patterns: 11 Existence -

An entity enters space. An entity is in space.

Non-existence -

An entity leaves space. An entity is not in space

Kinetic

An entity moves from location or bounded region A to location or bounded region B.

Hiding (covering)

An entity moves from bounded region A to opaque bounded region B. An entity is in an opaque bounded region.

Uncovering -

An entity moves from opaque bounded region A to bounded region B. An entity is in a transparent bounded region.

Possession

An entity is part of a spatial configuration.

Space and movement in the English verb system

185

Grouping -

An entity moves to form part of a spatial configuration together with other entities.

Non-possession

An entity is not part of a spatial configuration.

Non-grouping -

An entity leaves a spatial configuration.

In all these situations, we are dealing with generic abstract space; and thus, the entities involved can be of any kind. However, there are some situations in which human entities (or other animated entities) are involved and which require a more specific type of abstract space: Sensory-perceptual

An animate entity perceives another entity through its senses. An entity is perceived as having certain sensory attributes.

Sensory-feeling -

An animate entity feels something.

Epistemicargumentative 12

A human entity has a mental experience (of knowledge, belief, etc.) and may or may not attempt to communicate it to another human entity.

In the latter cases, motion across bounded abstract spatial regions overlaps with motion across mind and body, as well as speaker and receiver. Apart from these basic patterns, which we could call basic or 'prototypical' situations, there are also some other notions that can be expressed by means of verbs such as possibility {can, could, may, might), necessity (need), and causation (which in Lakoff's mapping for event structures is equated with force). All these basic situations can be represented by referring to two kinds of motion (or lack of it): (a) motion through abstract generic space and (b) motion through interactantial space. Motion through abstract generic space means motion in the sense we have been considering so far: a conception close to Langacker's. Motion through interactantial space means interaction between different semantic roles as traditionally considered. A canonical example would be an object moving from an agent to a patient.

186

Carlos Inchaurralde

2.3. Abstract space and verbs in a sample text Now, if we look for predicates that articulate the mentioned situations in the text, we find the following: (a) Existence, non-existence, and transformation (both at the same time): be, exist, arise, give rise, turn out, locate, motivate, occur, make it possible, imply, delete, derive, replace.

Non-existenc

Non-existenc

Non-existen hiervan ['of [that^ here']] = > 'this one'

(19)

Laat my 'n boek kies. Ek hou däärvan. 'Let me choose a book. I like thät one.' [van [dit3 daar2]] — > däärvan ['of [that3 there 2 \ \ = > 'thät one'

In the development of an unaccentuated hiervan the demonstrative dit [idit3] is replaced by the referential dit [dit^]\ hier explicates the role of the speaker. In (18) and (19) the use of hiirvan

and däärvan

is usually accompanied

by an indexical gesture by the speaker. Example (19) also clearly illustrates the role of the speaker (conceptualizer). He chooses to reinforce the reference to a specific referent, thus moving within the blending of locality with entity stronger towards entity - therefore the selection of an emphasized pronoun anaphor. The speaker's role and the way he apprehends the situation will consequently determine the different underlying deictic meanings of hiervan and daarvan.

218

Willem J. Botha

This also confines the meanings of hier ('here') and daar ('there') in the following sections. (iv)

hier ('here') and daar ('there') in indefinite anaphoric use

In accordance with Ponelis (1979: 107) this category could be divided into four subcategories: (a)

existential use of hier ('here') and daar ('there')

Taking the differentiation between a strong and weak daar ('there') into account again, we can assume a similar development of the underlying deictic meanings of hier ('here') and daar ('there') in the following examples: (20)

Daar is olifante. which means: Olifante bestaan. ' There are elephants.' which means: 'Elephants exist.'

(21)

Daar is olifante (in Afrika). which means: Olifante bestaan erens (in Afrika). 'There are elephants (in Africa).' which means: 'Elephants exist somewhere (in Africa).' [daarx ] = > daar ['therej'] = > 'there'

The existential daar ('there') signifies a referent in the surrounding vicinity without focussing the attention on any specific region of the location, contrary to the existential hier ('here'), which signals deixis near2 (physical or conceptual) in relation to the speaker. Compare example (21) with example (22): (22)

Hier is olifante (in Afrika). which means: Olifante bestaan hier (in Afrika). 'Here are elephants (in Africa).' which means: 'Elephants exist here (in Africa).'

Spatial deixis in Afrikaans dictionaries

219

[daar{ hier] - > hier \'therex here'] = > 'here' In the process of amalgamation the strong hier ('here') absorbed the weak form daar ('there'). And compare (22) with (23): (23)

Daar is olifante hier (in Afrika). 'There are elephants here (in Africa).' [daar^] => daar ['therej'] => 'there' (b) hier ('here') and daar ('there') in impersonal (passive) constructions

In accordance with Ponelis (1978: 409) the assumption can be made that in these constructions hier ('here') or daar ('there') functions as a (formal) subject without any specific thematic meaning - although from a cognitive point of view it and there are considered as subjects denoting the setting (atmospheric or psychological); consequently the setting could be seen as rather unspecified. In contrast to the meaning of daar ('there') the meaning of hier ('here') signals deixis near, within the specific setting. Compare the following examples: (24)

HierIdaar word gedans. 'Here/there is danced.' (c) relativized hier ('here') and daar ('there') (cf. Ponelis 1979: 409)

Compare: (25)

Hierldaar bestaan die moontlikheid van 'n oorlog. which means: Die moontlikheid van 'n oorlog bestaan hierldaar. 'Here/there exists the possibility of a war.' which means: 'The possibility of a war exists here/there.' (d) indefinite hier ('here') and daar ('there')

220

Willem J. Botha

Compare: (26)

Hier/daar het 'n venster gebreek. which means: 'n Venster het hier/daar gebreek. Ά window broke (here/there).'

Some conclusive remarks on hier ('here') and daar ('there'): *hier ('here') signals deixis: near, *daar ('there') signals just deixis, but apart from the distinction between a strong and weak form, the examples clearly illustrate the conceptual relation between entity and locality3. Lyons (1977: 670) uses the term textual deixis to refer to the link which exists between the deictic and anaphoric functions of pronouns: The basically deictic component in an anaphoric expression directs the attention of the addressee to a certain part of the text or co-text and tells him, as it were, that he will find the referent there.

He also claims that in such cases the function of the specific pronoun would lie somewhere between the deictic and the anaphoric.

2.3. Time, space and demonstratives

2.3.1. Hier ('here') It is well-known that time is conceived of in terms of space. Therefore hier ('here') is very often used to refer to time. Compare the following example: (27)

Ek het hulle tot hiertoe (= nou) grootgemaak. Ί have raised them up to here (= now).'

Spatial deixis in Afrikaans dictionaries

221

2.3.2. Nou Similar to the distinction between two different "meanings" of hier ('here') and daar ('there') [hier ('here') and daar ('there') referring to certain locative points (dimension 0) and hier ('here') and daar ('there') referring to certain interiors {dimension 3], a distinction can be drawn between an instantaneous nou ('now') (referring to a specific point in time) and a comprehensive nou ('now'). Compare the following examples: (28)

Gee dit nou (onmiddellik/*tans) vir my! 'Give it now (immediately/^presently) to me!'

(29)

Dit reen nou (tans/*onmiddellik) baie. 'It is raining a lot now (presently/^immediately).'

Within the meaning of the comprehensive nou ('now') a further space related differentiation should be drawn. Compare the following examples: (30)

Hülle rekord inaggenome, moet die regering nou (= iewers in die onmiddellike toekoms) bedank. 'Considering their record, the government have to resign now (= from this point in time onwards).'

(31)

Ek het nou (= tot nou toe) genoeg gehad van die storie. 'Now (= up to now) I had enough of the story.'

In example (30) nou ('now') refers to a timespan stretching from coding time somewhere into the future. In example (31) nou ('now') refers to a timespan stretching from somewhere in the past up to coding time.

2.3.3. Hierdie ('this') and daardie ('that') Similar to the use of hierdie ('this') and daardie ('that') in spatial expressions to refer to localities in relation to the relative locality of the speaker, hierdie ('this') and daardie ('that') are also used to refer to points in time relative to coding time. We will not go into this matter now (cf. Botha 1991: 111-112). In this section the focus was directed at the deictic system underlying reference to spatial locations relative to the speaker's egocentric origin and coordinate axes. In section 3 we will deal with the intrinsic system.

222

Willem J. Botha

3. Intrinsic deictic relations Miller and Johnson-Laird (1976: 396) contrast the deictic system with the intrinsic system on the following basis: "(I)n the deictic system spatial terms are interpreted relative to intrinsic parts of ego, whereas in the intrinsic system they are interpreted relative to intrinsic parts of something else." With reference to spatial expressions such as in back of, in front of, and at the side of, which express a spatial relationship with respect to some point of reference, Tanz (1980: 6) distinguishes between their uses in a purely symbolic sense, or in an indexical sense·. In their symbolic sense the relationship is defined in terms of attributes inherent to the reference object, its permanent directional features. In the indexical sense of these expressions the spatial relationship is defined jointly by the position of the reference object and by a participant in the speech act.

Due to the fact that entities such as human beings, motor cars, houses, etc. are constantly spatially orientated with respect to inherent properties such as above, below, before, behind, right and left, Schwerdtfeger (1982: 57) considers such entities as intrinsic deictic (i.e. according to their symbolic sense), contrary to entities without such orientation. In this respect it should be mentioned that with certain objects which do not have their own intrinsic orientation (e.g. a ball, a rock, a tree, etc.), the coordinate system is transposed from one of the participants in the conversation - speaker or hearer. Therefore one can speak of "in front of the tree" (the tree is confronted with the speaker, and has a front-back orientation similar to a hearer in an orthodox conversation), "to the right of the tree" (the tree adopts the orientation of the speaker), etc. , These kinds of relations present the lexicögrapher with a special problem: he has to go onstage - due to the fact that conceptualization in this regard takes place from an intrinsic deictic point-of-orientation: thus construal can take place from a fixed point-of-orientation. The different pointsof-reference (i.e. landmarks) are therefore chosen in relation to the locality of the implied conceptualizer and in relation to a certain trajector. In Casad (ed. 1995) I have pointed out that neglect of the intrinsic deictic point-of-orientation constantly leads to an inadequate definition and such a definition does not reflect the correspondence between conceptual knowledge and linguistic structure and behaviour. Therefore I will not go into this matter here. Against the background of the preceding brief discussion of deictic spatial concepts, section 4 deals with an examination of a few lexicographical definitions taken from an Afrikaans monolingual dictionary.

Spatial deixis in Afrikaans dictionaries

223

4. Lexicographical portrayal of spatial deixis In the following section relevant lexicographical definitions from Verklärende Handwoordeboek van die Afrikaanse Taal are contested against the background of the previous discussion. Obviously the selected dictionary articles will only be representative of the category spatial deixis portrayed in this specific dictionary. Although the applicable dictionary articles are comprehensively quoted in Appendix A, the relevant lexicographical definitions are repeated on account of the analysis. The examination of the definitions will be in accordance with the order of the discussion in sections 2 and 3. 4.1. Deictic adverbs

4.1.1. Compare the following definitions under hier ('here'): (32)

(a) Op of in die plek waar die spreker hom bevind of wat hy aanwys. 'At or in the place where the speaker finds himself or which he points at.' (b) Op 'n tydstip of punt wat so pas bereik is 'At a point of time or point that has just been reached.' (c) Na die plek toe waar die spreker hom bevind 'To the place where the speaker finds himself.'

Apart from recognition of the deictic centre [waar die spreker hom bevind ('where the speaker finds himself')], the lexicographer also acknowledges that hier ('here') refers to a certain locative point (dimension 0) and to a certain interior (dimension 3) by using the two prepositions op ('at') and in ('in') in his definition - compare (32 a); in (32 b) he confirms the relation between space and time - therefore a quite useful definition.

4.1.2. Compare the following definitions under daar ('there'): (33)

(a) Op 'n plek, op 'n punt, tydstip weg van die spreker 'At a place, at a point, point of time away from the speaker.'

224

Willem J. Botha

(b) Om die bestaan van iets aan te dui 'To indicate the existence of something.' (c) Dadelik reg 'Immediately ready.' Recognition is explicitly given to the deictic centre [weg van die spreker ('away from the speaker')], to dimension 0 by applying the preposition op ('at'), to the existential meaning of daar ('there') - compare (33b); and to the relation between space and time - compare tydstip ('point of time') in (33a) and dadelik ('immediately') in (33 c); dimension 3 is only implicitly referred to by use of the noun phrase 'n plek ('a place'). In both instances [hier ('here') and daar ('there')] the entity-meaning remains unexplained.

4.2. Hier ('here') and daar ('there') in definite anaphoric use

4.2.1.

(34)

Compare the following definitions under hier'van, hier van ('of this'): (a) Van dit wat hier genoem of betrokke is of aangedui word O f that which was mentioned here, or which is involved, or which is indicated.' (b) Van verskillende dinge 'Of different things.' (c) Oor verskillende sake 'Of different aspects.'

Both its deictic use [reference to hier ('here')] and anaphoric use [reference to dit ('that')] is referred to, although very implicitly. According to this definition the dictionary user will not be able to recognize that (by using this word) the speaker (conceptualizer) has the opportunity to move within the blending of locality with entity towards emphasis of what ever he wants to stress. The definitions in (34b) and (34c) elucidate nothing about the deictic nature of hiervon ('of that'). Reference is only made to its anaphoric use.

Spatial deixis in Afrikaans dictionaries

225

No explicit explanation is given for the different uses of hier'van and hier van ('of this').

4.2.2. Compare the following definition under daar'van, daarvan', daar van ('of that'): (35)

Van die vooraf genoemde 'About the previously mentioned.'

In (35) the anaphoric use is identified; its deictic use is not mentioned; and no distinction is made between the different uses of daar'van, daarvan' and daar van ('of that'). Comparing the definitions in (32) and (33) with those in (34) and (35) the most obvious disparity lies in the fact that in the case of hier ('here') and daar ('there') the emphasis lies on its deictic nature (with negligence of its entity-meaning), while in the case of hiervon ('of this') and daarvan ('of that') the deictic meaning is to a certain degree overlooked.

4.3. Demonstratives

4.3.1. Compare the following definition under hierdie ('this'): (36)

Die ding(e) of persoon (persone) wat horn (hulle) naby die spreker bevind of net genoem is 'This/these thing(s) or person(s) which/who find themselves near the speaker, or which/who was/were just mentioned.'

Although recognition is given to the deictic centre [naby die spreker ('near the speaker')], the lexicographer was deceived by the entity-meaning of hierdie ('this') - in such a manner that the entire definition is based on this meaning: die dinge ... ('this/these things ...'), instead of: woord wat nabyheid aan die spreker aandui ('word denoting proximity to the speaker'); besides no reference is made to the relation between space and time.

226

Willem J. Botha

4.3.2. Compare the following definitions under daardie ('that'): (37)

(a) (byv.) Gese van 'n saak, persoon, ens., op 'n afstand van die spreker; teenoor hierdie 'Said of something, a person, etc. at a distance from the speaker; opposite of this.' (b) (seifst.) Gese van 'n saak, persoon, ens., reeds genoem 'Said of something, a person, etc. already mentioned.'

The definition in (37a) is analogous to the one in (36). Again the lexicographer was confused by the entity-meaning. The entire definition is based on reference to 'n saak, persoon ('something, a person'). Consequently its anaphoric use is explicitly referred to; although reference is also made to its deictic use, but in a rather secondary fashion. There is no reference to the relation between space and time. Furthermore, there is also no reference to the fact that conceptual proximity is denoted by daardie ('that'), exemplified by utterances where the speaker wishes to distance himself conceptually (mentally or emotionally) from something or somebody by using daardie ('that'), although the referred entity is nearby.

4.4. The deictic time word nou ('now') Compare the following definitions under nou ('now'): (38)

(a) Op hierdie oomblik 'At this moment' (b) Teenswoordig, in ons dae 'Presently, in our times' (c) Om 'n versoek te bekragtig 'To confirm a request'

Examples (38a) and (38b) reflect the lexicographer's recognition of the difference between an instantaneous nou ('now') and a comprehensive nou ('now'), although it is done implicitly - by using the prepositions op ('at') and in ('in').

Spatial deixis in Afrikaans dictionaries

4.5. Deictic words denoting orientational and dimensional

227

features

4.5.1. Voor ('front') and agter ('back') Compare the following definitions [voor ('front') used as a preposition]: (39)

(a) Aan die voorkant; teenoor agter 'Front side; contrary to at the back' (b) In teenwoordigheid van 'In the presence of (c) Met betrekking tot volg- of rangorde 'According to order/sequence' (d) Voorafgaande aan 'Former to'

Compare the following definitions [voor ('front') used as an adverb]: (40)

(a) Op 'n plek aan die voorkant 'At a place in front' (b) Om 'n volgorde aan te dui 'To indicate a sequence' (c) Om 'n gesindheid aan te dui 'To indicate an attitude' (d) Meer as ander 'More than other' (e) Vinnig 'Swift'

In examples (39) and (40) the inadequacies are very clear. The lexicographer in no way acknowledges a vantage point, and negation of a vantage point neutralizes orientational and dimensional features which are at issue. Compare example (39a) for instance. It exemplifies the fact that no recognition is explicitly given to the deictic centre as such; consequently both

228

Willem J. Botha

external and intrinsic deictic orientational features are nullified, and as a result voor is defined in relation to its antonym agter. In example (41a) a similar problem is encountered. Accordingly agter ('back') has to be defined in relation to its antonym voor ('front'), although implicit reference is made to orientational features: aan die rugkant ('at the back'). Compare the following definitions (where agter ('back') is used as a preposition and an adverb respectively): (41)

(a) Nie voor nie, aan die rugkant 'Not in front, at the back' (b) In die verlede 'In the past' (c) Sodat iemand of iets anders voor is 'In such a manner that somebody or something else is in front' (d) Aan die kant wot nie voor is nie 'At the side which is not in front'

In (41b) recognition is given to the relationship that exists between space and time.

4.5.2. Bo ('above') and onder ('under') Compare the following definitions where bo ('above') is used as a preposition and an adverb respectively: (42)

(a) Hoer as; bokant 'Higher than, above' (b) Hoer in rang, graad, aansien 'Higher in position, order, status' (c) Ouer, meer as Older, more than' (d) Liewer, beter as 'Rather, better than'

Spatial deixis in Afrikaans dictionaries

229

(e) Verderas 'Further than' (f) Sterker, luider as 'Stronger, louder than' (g) Hoer 'Higher' (h) In hoe of die hoogste posisie 'In a high or top position' In example (42a) hoer as ('higher than') and bokant ('above') require a spatial reference point in accordance with a vantage point, and no such reference point is mentioned. It also applies to the definitions in (42 b-g) where the comparative degree implies a reference point, which is absent: hoer ('higher'); ouer, meer as ('older, more than'); liewer, beter as ('rather, better than'); verder as ('further than'); sterker, luider as ('stronger, louder than'); hoer ('higher'). Similar to hoer as ('higher than'), bokant ('above'), etc. in examples (42 a-g), laer as ('lower than'), benede ('beneath'), etc., in the examples under (43), also require a spatial reference point in accordance with a vantage point. Again no reference point is mentioned. Compare the following defintions: (43)

(a) Laer as; benede 'Lower than, beneath' (b) Tussen, te midde van; in die kring van 'Amongst; in the company of (c) Naby, vlakby 'Near, nearby' (d) Gedurende, tydens 'During, while' (e) Minder as 'Less than' (f) Om ondergeskiktheid, aflianklikheid uit te druk 'To express subordination, inferiority'

230

Willem J. Botha

(g) In die beskerming van 'Under the protection of (h) Gedek, beplant met 'Covered with' (i) Laer, minder; aan die onderkant 'Lower, less; beneath' In spite of the inadequacies concerning spatial orientation (in the preceding definitions), reference is made to the link which exists between space and time (43 d). The spatial foundation in its metaphorical use is also acknowledged [cf. (43 f,g,h)].

4.5.3. Links ('left') and regs ('right') Compare the following definitions under links ('left'): (44)

(a) Aan, met die linkerhand 'On, with the left hand' (b) (fig.) Onhandig, lomp; agterstevoor 'Clumsy, lumpish upside down' (c) (politiek) Radikaal, in die rigting van die sosialisme, icommunisme 'Radical, in the direction of socialism, communism' (d) Bedrieglik, vals 'Deceptive, fraudulent, delusive'

By mentioning hand ('hand') in example (44 a), implicit reference is made to orientational features. Furthermore there is no reference to a spatial reference point in accordance with a vantage point. In example (45) implicit reference is also only made to orientational features by making use of the word hand ('hand'), but also with no verification of a spatial reference point in accordance with a vantage point. Compare the following definitions under regs ('right'):

Spatial deixis in Afrikaans dictionaries

(45)

231

(a) Wat die regterhand beter en met meer gemak kan gebruik as die linkerhand 'What the right hand can do better and with more ease' (b) Aan, na die regterkant 'On, to the right side' (c) Met die regterhand 'With the right hand' (d) (fig.) Konserwatief 'Conservative'

5. Conclusion The previously examined lexicographical definitions reveal one definite feature: deictic inconsistency - because the definitions as such do not reflect the correspondence between conceptual knowledge and linguistic structure and behaviour; it does not recognize the fact that deixis is "the domain par excellence where language and reality meet." Should the lexicographer systematically take this into account (also by way of lexicographic labels), he could succeed in formulating an adequate definition - like a Mercator projection succeeds in portraying a three-dimensional "world" on a "flat surface". A few suggestions could be made to propose a rectification action on the part of the lexicographer. In his minigrammar (guide to the use of the dictionary) the lexicographer should give guidance in this regard. It implies at least the following: The phenomenon deixis should be labelled within the lexicographical definition. Deixis as concept should be explained in the minigrammar, with special reference to the deictic centre: person deixis, time deixis and space deixis. In his explanation of space deixis it is imperative that reference should also be made to orientational and dimensional features in accordance with vantage point. By way of cross references the dictionary user could then be guided back to the minigrammar in those instances which require it. This mechanism will also enable the lexicographer to make optimal use of it in the creation of a lexicographical definition.

232

Willem J. Botha

Appendix A: Lexicographical definitions from Verklärende Handwoordeboek van die Afrikaanse Taal I.

hier bw. (1) Op of in die plek waar die spreker hom bevind of wat hy aanwys ['at or in the place where the speaker finds himself or which he points at']: Hier gaan ons eers 'n bietjie rusPlant die boom hier. Hier längs is dit maar droog. Tussen hier en die dorp is daar baie slaggate. Ek weet hy woon hier rond. Hoor hier wat se hy. (2) Op 'n tydstip of punt wat so pas bereik is ['at a point of time or point that has just been reached']: Hier moet ek nou die verhaal onderbreek. Is dit moontlik om hier raad te gee? Hy het eers hier teen die aand aangekom. (3) Na die plek toe waar die spreker hom bevind ['to the place where the speaker finds himself]: Kom bietjie hier, UITDR.: Hier en daar, op verspreide plekke. Hier en nou, op hierdie plek en die huidige tydstip.

II.

daar I. bw. (1) Op 'n plek, op 'n punt, tydstip weg van die spreker ['at a place, at a point, point of time away from the speaker']: Dit is daar, nie hier nie. Hy woon daar bo, onder. Ek het hom daar anderkant gesien. Kyk, daar kom Jan. Daar le die ding, dit is juis die moeilikheid. Daar storm die kinders uit. Daar gebeur dit toe, juis toe. (2) Om die bestaan van iets aan te dui ['To indicate the existence of something']: Daar word te min gedoen. Daar is mense wat dink ... Daar word vertel dat ... (3) Dadelik reg ['Immediately ready']: Toe ek haar roep, was sy daar. II. voegw. Aangesien, omdat, dewyl: Daar die kar gaan staan het, kon ons nie verder nie.

III.

hier'van, hier van bw.vnw. Van dit wat hier genoem of betrokke is of aangedui word ['Of that which was mentioned here, or which is involved,

or which is indicated']: Ek het net tien sent en hier van moet ek nog vir jou gee. Weet jy iets hiervon? Wat word nou hiervon as ons dit nie eet nie! UITDR: Hiervon en daarvan,

(a) van verskillende dinge ['of different

things'], bv. hy het hiervon en daarvan geeet\ (b) oor verskillende sake ['of different aspects'], bv. ons het hiervon en daarvan gepraat. IV.

daar'van, daarvan', daar van byw.vnw. Van die vooraf genoemde ['About the previously mentioned']: Gee my ook 'n deel daarvan. Niks daarvan nie, ek keur jou plan af. Wat daarvan!, wat wil jy daardeur se of wat wil jy gedoen he? Dit kom daarvan, dit is die gevolg (waarteen jy gewaarsku is).

V.

hier'die aanw. vnw. Di6 ding(e) of persoon (persone) wat hom (hulle) naby die spreker bevind of net genoem is - byvoeglik en selfstandig gebruik ['This/these thing(s) or person(s) which/who find themselves near the speaker, or which/who was/were just mentioned']: Hierdie boek(e) is myne. Ken jy hierdie kind(ers)! Vat jy daardie; ek sal hierdie neem. Dis 'n man honderd hierdie. OPM.: Die selfstandige gebruik van hierdie as onderwerp is meestal onder Engelse invloed.

VI.

daar'die aanw. vnw. (1) (byv.) Gese van 'n saak, persoon, ens., op 'n afstand van die spreker; teenoor hierdie ['Said of something, a person, etc. at a distance from the speaker; opposite of this']: Daardie man het nog nie betaal nie. Ek wil daardie hemp koop. Daardie datum pas my

Spatial deixis in Afrikaans dictionaries 233

nie. By daardie geleentheid, 'n genoemde geleentheid. (2) (seifst.) Gese van 'n saak, persoon, ens., reeds genoem ['Said of something, a person, etc. already mentioned']: Met hierdie en daardie nog gesels na die kerk. OPM.: Daardie as onderwerp is nie erkende Afr. nie, bv. Daardie is 'n mooi hoed. VII.

nou I. bw. (1) Op hierdie oomblik ['At this moment']: Doen dit nou. Nou en dan. Nou of nooit. Hy is nou net weg. Nou die dag, onlangs. Ek hoor dit nou eers, vir die eerste maal. So gou soos nou, baie gou. (2) Teenswoordig, in ons dae ['Presently, in our times']: Nou word dit ouderwets genoem. Tot nou toe was dit anders. Dit is nou die mode. (3) Om 'n versoek te bekragtig ['To confirm a request']: Hou tog nou op\ Wie sal nou so iets sei II. tw. Om aan te spoor, op te wek, ens. ['To encourage, to generate']: Nou wat dink jy daarvanl Nou toe nou\

VIII. voor I. voors. (1) Aan die voorkant; teenoor agter ['front side; contrary to at the back']: Ek staan voor die venster. Voor die huis staan 'n vlagpaal. (2) In teenwoordigheid van ['In the presence o f ] : Voor die landdros verskyn. Nie vloek voor die kinders nie. Voor getuies teken. (3) Met betrekking tot volg- of rangorde ['According to order/sequence']: Voor die titel van die boek staan die Franse titel. 'n Kolonel kom voor 'n kommandant. (4) Voorafgaande aan ['Former to']: Voor die oorlog, die runderpes. Voor agtuur. Kom asb. voor Vrydag, voor die einde van die maand. UITDR.: Dit staan voor die deur, kan elke oomblik gebeur. 'n Doel voor oe he, daarna strewe. Voor op die wa wees, voorbarig, indringerig, aanmatigend. Dit gaan voor die wind, voorspoedig. II. voegw. Voordat ['Before']: Voor hy dit gesi het, was ek al huis toe. Voor ek nog kon antwoord, het hy al weer begin. Ill s.nw. lets wat ten gunste is ['Something that is in favour']: Alles het 'n voor en 'n tee. Die voor en tee van 'n saak. UITDR.: Van voor af begin, van die begin af. IV. bw. (1) Op 'n plek aan die voorkant ['At a place in front']: 'n Huis met 'n tuin voor en agter. Die man loop altyd voor. (2) Om 'n volgorde aan te dui ['To indicate a sequence']: Voor en na, telkens. (3) Om 'n gesindheid aan te dui ['To indicate an attitude']: Ek is daar nie voor nie. (4) Meer as ander ['More than other']: Tweeputjies voor wees (gholf). Los voor wees. (5) Vinnig ['Swift']: Die klok loop voor. UITDR.: Dit is net taruiie (meneer, professor, doktor) voor en tannie (meneer, ens.) agter - gese van iemand wat oordrewe beleef is en dus titels, aanspreekvorme by herhaling gebruik. IX.

agter I. voors. (1) Nie voor nie, aan die rugkant ['Not in front, at the back']: Agter die kas, agter iemand, agter mekaar loop. (2) (van 'n tydsverloop) In die verlede ['In the past']: Dit Ιέ alles agter ons. Ons het die eksamen gelukkig nou agter die rug. OPM..· Agter die kinders kyk moet in goeie Afrikaans wees na die kinders kyk, die kinders versorg, oppas. II. bw. (1) Sodat iemand of iets anders voor is ['In such a manner that somebody or something else is in front']: Party seuns sal agter moet staan. (2) Aan die kant wat nie voor is nie ['At the side which is not in front']: Die huis moet agter reggemaak word. Iemand van agter pak. Gaan staan agter. Hierdie os trek agter. Agter gaan sit. III. b.nw. Laat; nie so ver as ander nie ['Late, not as far as others']: Die klok is agter. Agter wees met die rente, UITDR..· lets agter iemand aan doen, hom dit nadoen. Agter iemand aan wees, hom gedurig soek, agternasit; hom na-

234

Willem J. Botha

aap. Agter iets aan wees, bv. geld, eer, dit probeer verkry. Van agter beter lyk as van voor, lelik wees. Agter iets kom, die waarheid ontdek. Iemand liewer van agter as van voor sien, liewer sien weggaan as kom. Agter iemand om, agter iemand se rug, sodat hy niks daarvan weet nie. Daar sit, skuil, steek iets agter, daar is 'η verborge bedoeling; alles is nie pluis nie. Agter jou (die) tyd wees, ouderwets, nie op die hoogte van die tyd nie. Van agter, agterna, bv. van agter beskou, is dit maar goed dat ek nie gegaan het nie. Dit was Meneer voor en Meneer agter - gese van iemand wat oordrewe beleef is. Agter iets wees, daarna soek, dit probeer, kry. Agter die wiel sit, die voertuig bestuur. X.

bo I. voors. (1) Hoer as; bokant ['Higher than, above']: Bo die wasbak is 'η rakkie. (2) Hoer in rang, graad, aansien ['Higher in position, order, status']: Hy is bo my aangestel. (3) Ouer, meer as ['Older, more than']: Kinders bo 6 jaar. Twee grade bo vriespunt. (4) Liewer, beter as ['Rather, better than']: Eendrag bo alles. (5) Verder as ['Further than']: Bo my verstand. (6) Sterker, luider as ['Stronger, louder than']: Sy stem klink bo die lawaai uit. II. bw. (1) Hoer ['Higher']: Bo in die lug is dit koud. Hy woon bo, hoer as die onderste verdieping. (2) In hoe of die hoogste posisie ['In a high or top position']: Hy staan bo in sy klas. Hy sit bo aan tafel. Ook seifst.: Die seSn kom van Bo, van die hemel, van God. Kyk na bo. Van bo tot onder. UITDR.: Bo iemand se vuurmaakplek, bo sy begrip.

XI.

onder I. voors. (1) Laer as; benede ['Lower than, beneath']: Onder die stoep. Onder die bed inkruip. Onder 'n seil skuil. Onder die puin Ιέ. Onder die wol kruip, gaan slaap. Onder die hart dra, swanger wees van 'n kind. Onder seil gaan, afvaar. Onder 'n vlag vaar, die vlag voer. Onder die blote hemel. Daar is glad niks nuuts onder die son nie (Pred. 1: 9). Onder dak, in huis; binne. Die kat kruip onder die tafel deur. (2 a) Tussen, te midde van; in die kring van ['Amongst; in the company of]: Onder die menigte, skare was my kind. Onder vriende verkeer. Daar is baie kwaad onder hulle. Hulle twis baie onder mekaar. Onder andere gese as 'n mens een uit baie sake tot voorbeeld neem. Onder die mense kom, jou in die openbaar vertoon, met mense omgaan. Dit bly onder ons, bly geheim, niemand anders moet daar iets van weet nie. Onder ons [gesS], in ons eie kring; in vertroue. (b) Naby, vlak by ['Near, nearby']: Hy steel sommer hier onder my oe. Onder skot wees, binne trefafstand. (3) Gedurende, tydens ['During, while']: Onder die regering van Karel /. Onder die preek. Onder kerktyd, onder die ete. (4) Minder as ['Less than']: lets onder 'η rand koop. Onder die 10 jaar. (5) (fig.) Om ondergeskiktheid, afhanklikheid uit te druk ['To express subordination, inferiority']: Onder sy oom se voogdy. Onder iemand se invloed staan. Onder iemand dien. (6) In die beskerming van ['Under the protection of]: Onder vrygeleide. Hy is onder my toesig, sorg. (7) Gedek, beplant met ['Covered with']: 'n Groot land onder koring, katoen hi. II. bw. Laer, minder; aan die onderkant ['Lower, less; beneath']: Onder in die kan. Onder aan die bladsy. Onder woon, op die onderste verdieping. Onder längs, onder in, onder uit. Die son is onder. UITDR.: Ten onder bring, onderwerp, oorwin, bv. 'n volk ten onder bring. Onder woorde bring, deur woorde uitdruk.

Spatial deixis in Afrikaans dictionaries

XII.

links b.nw. en bw. (1) Aan, met die linkerhand ['On, with the left hand']: Hy woon links van my. Hou links! Links af, om. Sy skryf links. (2) (fig.) Onhandig, lomp; agterstevoor ['Clumsy, lumpish; upside down']: Linkse mattiere, 'η Soak links aanpak. (3) (politiek) Radikaal, in die rigting van die sosialisme, kommunisme ['Radical, in the direction of socialism, communism']: Die linkse partye. Links stem. Ver links wees. (4) Bedrieglik, vals ['Deceptive, fraudulent, delusive']: Linkse streke uithaal. UITDR: lemand (iets) links laat le, hom negeer, jou opsetlik nie met hom (daarmee) bemoei nie. Links van die regering wees, altyd teen die stroom op. Nie links wees nie, rats wees, iets dadelik aanpak.

XIII. regs b.nw. en bw. (1) Wat die regterhand beter en met meer gemak kan gebruik as die linkerhand ['What the right hand can do better and with more ease']: Die meeste mense is regs. (2) Aan, na die regterkant ['On, to the right side']: Loop meer regs; draai daar regs. Regs om! (3) Met die regterhand ['With the right hand']: Die oefeninge moet regs en links gedoen word. Hy skryf links maar eet regs. (4) (fig.) Konserwatief ['Conservative']: Die regse party. Polities taamlik regs wees. Regse neigings hi.

Notes 1. 2. 3.

4.

The English glosses are not meant to be grammatical sentences, only translations. Kirsner (1993: 97) differentiates between the three Dutch forms: deze ('hierdie'/'this'): signalling deixis: near; die ('daardie'/'that'): signalling just deixis; de ('die'/'the'): signalling differentiation required and made. In some sense Kirsner (1993: 94) also recognizes the difference between entity and locality when he points out that the "imperative force" of the demonstratives is derived and not basic, in the sense that it results from iconic factors, "namely as a consequence of the simultaneous signalling of both definiteness ... and locative information in a single unit, i.e. within a single word' (deze/'hierdie'/'this' and die/'daardie'/'that') in contrast with the collocations de/'die'/'the' ... hier/' hier'/'here' and del 'die '/'the' ... daar/'daar'/'there'" (cf. Kirsner 1993: 94). The Afrikaans single demonstrative units clearly illustrate this fact (cf. examples (17), (18) (19)). Consequently linguistic fusion correlates with conceptual fiision, while linguistic independence correlates with conceptual independence (cf. Kirsner 1993: 95). Although all the applicable dictionary articles are comprehensively quoted, only the relevant lexicographical definitions are translated into English. Citations are quoted for the sake of those who are familiar with Afrikaans, and to illustrate to what extent the lexicographer relies on it for the explanation of his definitions, contrary to the fact that he "must base his definitions on citations from the spoken and written language" to reflect "the dynamics of real language" (cf. Lombard 1991: 158).

236

Willem J. Botha

References Ballmer, Thomas T. and Waltraud Brennenstuhl (eds) 1981 Lexical analysis and language theory. In: Eikmeyer, Hans-Jürgen and Hannes Rieser (eds). Words, Worlds, and Contexts. New Approaches in Word Semantics. Berlin: Walter de Gruyter, 414-461. Bierwisch, Manfred 1967 Some semantic universals of German adjectives. Foundations of Language 3: 1-36. Botha, Willem J. 1980 Die grammatika van die eerste persoon. Doctoral dissertation, University of South Africa. 1991 Deiktiese tydsverwysing. Acta Academica 22 (3): 102-115. 1995 The cognitive frame of a set of cricket terms. In: Casad, Eugene H. (ed.), 237-247. Casad, Eugene H. (ed.) 1995 Cognitive Lingusitics in the Redwoods. Berlin: Mouton de Gruyter. Combrink, Johan G. H. 1978 'n Afrikaanse plek-bespreking. In: Snyman, Henning (ed.). Vit vier windstreke. Goodwood: Nasionale Opvoedkundige Uitgewery. Eikmeyer, Hans-Jürgen and Hannes Rieser 1981 Word semantics from different points of view. An introduction to the present volume. In: Eikmeyer, Hans-Jürgen and Hannes Rieser (eds). Words, Worlds, and Contexts. New Approaches in Word Semantics. Berlin: Walter de Gruyter, 1-18. Geiger, Richard A. and Brygida Rudzka-Ostyn (eds) 1993 Conceptualizations and Mental Processing in Language. Berlin: Mouton de Gruyter. Harteveldt, Piet (ed.) 1991 Lexikos. Afrilex Series 1: 1991, Stellenbosch: Büro van die WAT 158182. Keenan, Edward L. (ed.) 1975 Formal Semantics of Natural Language. London/New York: Cambridge University Press. Kirsner, Robert S. 1993 From meaning to message in two theories: Cognitive and Saussurean views of the Modern Dutch demonstratives. In: Geiger, Richard A. and Brygida Rudzka-Ostyn (eds). Conceptualizations and Mental Processing in Language. Berlin: Mouton de Gruyter, 81-114. Lombard, Frederick J. 1991 Die aard en aanbieding van die leksikografiese definisie. In: Harteveldt, Piet (ed.). Lexikos Afrilex Series 1. Stellenbosch: Büro van die WAT 158182. Lyons, John 1977 Semantics I and II. Cambridge: Cambridge University Press. 1975 Deixis as the source of reference. In: Keenan, Edward L. (ed.). Formal Semantics of Natural Language. London and New York: Cambridge University Press, 61-83. Miller, George A. and Philip N. Johnson-Laird 1976 Language and Perception. Cambridge: Cambridge University Press.

Spatial deixis in Afrikaans dictionaries

237

Odendal, Francois F. (ed.) 1994 Verklärende Handwoordeboek van die Afrikaanse Taal. Johannesburg: Perskor. Ponelis, Fritz A. 1979 Afrikaanse sintaksis. Pretoria: VanSchaik. Rommetveit, R. 1968 Words, Meanings and Messages. Oslo: Academic Press. Schwerdtfeger, Anna-Mart 1982 Deiksis: Orientasie en konfrontasie in die ruimte. Klasgids 17(2): 52-63. Smith, Raoul N. 1985 Conceptual primitives in the English lexicon. Papers in Linguistics 18(14): 99-137. Snyman, Henning 1979 Vit vier windstreke. Goodwood: Nasionale Opvoedkundige Uitgewery. Tanz, Christine 1980 Studies in the Acquisition of Deictic Terms. Cambridge: Cambridge Taylor, John R. University Press. 1994 'Subjective' and 'objective' readings of possessor nominals. Cognitive Linguistics 5-3: 201-242. Weissenborn, Jürgen and Wolfgang Klein (eds) 1982 'Here' and 'there'. Cross-linguistic studies on deixis and demonstration. Pragmatics and Beyond III 92/3: 1-11.

What good are locationals, anyway? Eugene H. Casad

1. Introduction In a nicely stated paper Friedrich (1970) explored three categories into which the Tarascan language of Western Mexico encodes notions of shape and location. They included numeral classifiers, classificatory verbs and body part suffixes (1970: 379). In the course of his discussion, he asked the question "What good are locationals, anyway?" His answer was that he did not know, but he did know that all languages have them (Ibid.: 403-4). One clear implication of all this, as he himself noted, is that the subject is worth a lot of attention. The question that Friedrich felt was even more important than what good locationals are was the relation of the locationals to other domains of the grammar (Ibid.: 404). In this paper, I consider one aspect of this question, drawing on data from Cora, a Uto-Aztecan language of Northwest Mexico. 1 The lexicon of this language is characterized by a wide variety of free adverbs, particles, demonstrative pronouns, definite articles and verbal prefixes that indicate various kinds of spatial, temporal and logical concepts and perform a variety of functions in the grammar. I will briefly illustrate some of the functional usages of these grammatical elements with the intent of answering Friedrich's question by saying, "Well, beyond the ordinary use of telling the hearer where something is located with respect to the speaker, hearer or some other entity, they are useful for all kinds of other things and here are some of their functions and interactions in various domains of Cora grammar. " At the lexical level, Cora locationals commonly distinguish one word from another, or help signal that one tense of a verb or another one is being invoked and are the basis for a myriad of extensions to non-spatial domains within morphology and syntax. On the discourse level, they are used to begin discourses, end discourses, help mark the sequencing of events within given scenes, signal the progression from one scene to another within the discourse and to bring participants onstage and take them offstage. The particular locationals that I am concerned with in this paper include (a) topographic adverbs, (b) a variety of locative particles and (c) verbal prefixes

240

Eugene Η. Casad

of location and direction, as well as combinations of these. The locative particles include topographically based sentence initial particles that are morphologically related to both the topographic adverbs and to some of the verbal prefixes. In addition, certain locative particles are used as procomplements and there are extensions of the locatives into the temporal domain. Finally, there are also usages of the locatives as coordinating subjunctions. The semantics of these grammaticalized variants are all based on spatial models.2 2. Where are you going? The Cora topographic adverbs illustrate vividly the experiential basis of much of semantic structure. Orientation to the topography is a salient part of daily life for the Coras who live in a rugged area of the Sierra Madre Occidental in Northwest Mexico. This is elaborately built into the grammar of Cora via a set of topographic adverbs (cf. Casad 1977; 1982).

2.1. The topographic adverbs The topographic adverbs are morphemically complex and are marked by a set of suffixes that are neatly grouped into two sets of three, each set being further subdivided into "up," "down" and "over" dimensions. The two sets themselves pertain to the domains of the river and the hill, as can be seen in Table 1. Table 1. The topographic adverb suffixes

up down over

river

hill

-ϋαρ™α -ϋβ -ta(va)n

-ti(vi) -Id -ka 'i(wap

Summarizing Table 1, we can gloss these adverbial suffixes as follows: tfapwa = 'upriver', -fie = 'downriver', -ta(va)n - 'across the river', tfifvi) = 'uphill', -ki = 'downhill' and -ka'i(wa) 'over the hill'. 4 A more fine grained categorization of these adverbs is achieved by taking into account the locative particles that the suffixes attach to. As I have

What good are locationals, anyway?

241

previously described these elsewhere, I will not go into a detailed description of them here, but just note that they are distinguished along three parameters. For distance relative to the speaker, they are marked by y'proximal', m- 'medial' and zero 'distal'. In terms of boundedness, specific areas are marked by u- 'inside' and open or non-specific areas are marked by a- 'outside' (cf. Casad and Langacker 1985; Langacker 1991). These distinctions are marked in the sets of what I call Basic Series Particles, given in Table 2. These two series occur only with the suffix -ta(va)n 'over the river'. Table 2. The basic locative particles

here there off there

Outside

Inside

ya ma a

yu mu u

As an additional layer of complexity, these particles may be marked for position relative to the slope of a hill: areas at the foot of the hill are marked by a long a- or u- vowel, as illustrated in Table 2; areas within the slope are marked by either ah- or uh-, whereas locations at the head of the slope are marked by an-. These morphological markers are organized into very regular sets. The 'foot of the slope' particles are summarized below in Table 3. Table 3. The foot of the slope particles

here there off there

Outside

Inside

yaa maa aa

yuu muu uu

The 'foot of the slope' particles occur only with the 'downriver' and 'downhill' suffixes, i.e. -tie and -kt, yielding forms like άαϋε and yiiukt. On the other hand, the particles that signal locations within the face of the slope or at the head of the slope, occur with the full set of topographic

242

Eugene Η. Casad

adverbial suffixes. These particles are shown in Table 4. Table 4. Slope related locative particles

here there off there

outside

inside

outside

yah

yuh muh uh

yan man an

mah ah

The combination of locative particle plus topographic suffix yields a set of sixty-six topographic adverbs. These are summarized in Casad (1977; 1988) and are discussed in detail in Casad (1982). Typical usages of them are seen in some of the examples that follow in this paper.

2.2. Cross-reference to verbal prefixes There are numerous linkages between the topographic adverbs and the locative prefixes on the verbs as well as with other grammatical elements. In this section I consider briefly the linkages between the topographic adverbs and the locative prefixes of the verb system. There are both canonical pairings and non-canonical ones.

2.2.1. Canonical pairings In the canonical usages of the pairings between topographic adverbs and the locative prefix sequences on the verbs in a single sentence, the path implied by the verb leads away from the Speaker's position and the topographic adverbs designate the distal location at the end of that path. The canonical pairings are illustrated in (la-g). (1)

(a)

vü-auh ha ha '-«-Λ-ηΥέ-η5 I - LOC there BASE

DlSTAL-inside-slope-pass-PKTC

by

dn-ti(vi) on: top-uphill 'I'm going off to the top of the hill.

What good are locationals, anyway?

(b)

ny-auh ha ha '-u-ka-nye-n I - LOC there DlSTAL-inside-down-pass-PRTC BASE by άα-Μ in: flat-downhill 'I'm going off out in the flat at the foot of the hill.'

(c)

ny-auh ha ha '-u-re 'e-nye-n I - LOC there DlSTAL-inside-corner-pass-PKTC BASE by ά-h-ka 'iwa out-slope-overhill 'I'm going off over the side of the hill.'

(d)

ny-auh ha ha '-u-ra-ΐΰί-η I - LOC there DISTAL-inside-downriver-pass-PKTC BASE by άα-tye in: flat-downriver 'I'm going off in the flat downriver.'

(e)

tä-auh ha ha '-u-ta-nye-n I - LOC there DlSTAL-inside-psss-PRTC BASE by ä-h-tYapwa out-slope-upriver 'I'm going off into the slope upriver.'

(f)

ny-auh ha ha '-u-ta-nye-n I - LOC there DlSTAL-inside-pass-PRTC BASE by άη-ία(να)η on: top-across: river 'I'm going off to the top of the bank across the river.'

243

244

Eugene Η. Casad

A perusal of sentences (la-f) shows that the locative prefix sequence hä'-uh 'DlSTAL-inside-slope' cross references to -ti(vi) 'uphill' adverbs and M'u-ka- DlSTAL-inside-down cross refences to -kir 'downhill' adverbs. The other pairings of locative prefixes with topographic adverbs are as follows: ha'-u-re'e- DlSTAL-inside-around corner with -ka'i(wa) adverbs, ha'-u-rawith DlSTAL-inside-downriver 'downriver' adverbs and ha'-u-ta'DlSTAL-inside-across/ straight with both ap^a 'upriver' and -ta(va)n 'on the other side of the river' (cf. Casad 1977).

2.2.2. Non-canonical pairings Non-canonical pairings of topographic adverbs and locative verbal prefixes do occur and reflect a number of kinds of departures from the canonical ones. The following examples of both canonical (2a) and non-canonical pairings (2b) are taken from a Cora text that I collected from a native speaker from Jesus Maria, Nayarit. For easy reference, I title this text 'The children of Tuaacamu'una.'6 Sentence (2a) reflects the canonical situation in which the topographic adverb meetye 'right off there in the flat down river' designates the end point of a straight path whose downriver trajectory is signalled by the use of the locative prefix -ra 'downriver' on the verb. On the other hand, the use of müufle 'right downstream there along the river bank' in (2b) is not what I would have expected to find used with the locative prefix ra'a- 'around the corner'. Instead, I would have expected the adverb mahca'i 'right there at the side of the hill.' The anomaly gets straightened out, however, when one realizes that the path the trajectors follow is conceived as starting out from a location behind a hill away from the speakers conceived viewing point and then continues towards the speaker's conceived location, makes the bend and heads straight down river to a final point directly away from that conceived vantagepoint. (2) (a)

Αά mu ce'eta ha'-u-rda-ri^e there they same DlSTAL-away-downriver pass: by m-ee-fie C O M P L E T I V E , which is also ablauted to -u in this context.

What good are locationals, anyway?

(21)

(a)

263

Aa pu'u nu'u hih-ta-'ih himi there: out SUBJ: PAUS QUOT NARR-CNJ-SEQ far

y-a '-u-ta-m w ah. it-distal-COMPL-PERF-put: flat 'And then, they say, she left it there just a little ways off.' (b)

Ahtä άα w-i-y-ö-M-tautya-'a. CNJ there: outside them-NARR-(?)-o«i-awa>'-greet-lMPERF

'And then she gave them a parting greetings from off yonder.' (c)

Ahtä hi-y-ä'-u-raa. CNJ NARR-(l)-DISTAL-EXT-go: PST

'And then she left.' The sequencing of events is a crucial factor in ensuring a coherent text. In this area of textual organization, we also see abundant use of locational and directional elements in Cora. For example, the locative particles ah 'off out in the slope' and aa 'out there in the flat' are both used in discourse to sequence events. Immediate succession of one event by another is indicated by ah, whereas immediate location for successive events is signalled by aa. As earlier examples also suggested, auu LOCATIVE BASE serves to highlight the particular location where subsequent events take place. The use of ah in (22) is typical. After Possum gets captured and led off to the house of the owner of the sugar cane, he is tied up behind the house and left there. The very next event, as signalled by ah, is that he overhears his intended fate, as stated in the first clause of (22). The main verb carries the locative prefix sequence a'-u-m'a- DISTAL-inside-corner, which indicates the obliqueness of the path between the speaker's position and that of Coyote. (22)

Ah ρύ nii'u \yk-ha '-u-rn 'a-nam w a then SUBJ QUOT

OlSTR-DlSTAL-inside-corner-hezr

ti nu'u ayän ti'i-sa t vastakira'i tictn: SUBR QUOT PROCOMPL DISTR-say ART old: man QCOMPL

264

Eugene Η. Casad

Tya-ta'ah yeewi r-a'-u-tye-yaaca t ya'usu. we-purpose QUOT him-DISTAL-COMPL-P£7?F-boil ART possum 'And then from back around behind the corner of the house, he heard the old man say: "Let's boil that possum."' Anaphora and elaboration also figure prominently in Cora discourse. Certain locationals often serve as antecedents to more elaborate locative expressions that occur later in a sentence. A typical configuration is for the antecedent to precede the verb and the elaboration to follow the verb. Such relations are clearly based on the speaker's ability to construct mental spaces and set up reference points within those mental spaces (cf. Langacker 1993; van Hoek 1994, to appear). A common antecedent consists of the topographic particle άα 'out in the flat' in construction with a schematic existential verb such as ηά'α 'BE,' in which άα=ηά'α is elaborated by a locative subordinate clause. The examples in (23) and (24) are both taken from the story of Possum and Coyote. In (23) άα=ηά'α is elaborated by a postpositional phrase huye hece 'road in.' A second use of άα=ηά'α is seen in the subordinate clause of (24). In this case it refers back to huye hetze. (23)

Ahtä άα—ηά 'a hi-y-a '-u-τέ 'e-nYe CNJ there: out=bt NARR-(l)-DlSTAL-EXT-corner-axnve huye hece, hi-r-aa-seij tt puä'a nu'u άα—ηά'α road in NARR-it-COMPL-see SUBR COND QUOT there:

out-be

hi-y-a-wtt-kä'a i sihka. NARR-(l)out-away-\a.y ART wax 'And he came around the corner to there in the road when he saw that right there was the wax figure all stretched out across the road.' Sentence (24) illustrates a variation on the pattern άα=Χ (where X equals a Stative predication). In this case, άα is in construction with the schematic locative verb Μ'a 'BE (LOC). The preverbal antecedent äa=hä 'a is elaborated by a pair of a post-verbal adverbials, himt 'far off' and waρά'αοα EXTENSIVE-flat area 'way out in the flat.' The complex antecedent itself may occur as a discontinuous construction, as in (24b). In this case, the sentence is adding new details to the description of the scene and the discontinuous άα ... ha 'a refers back to the specific location mentioned in (24a).

What good are locationals, anyway?

(24)

(a)

265

Meentih rrn'u äin i wäave'e άα=Ηά'α Meanwhile QUOT DEM ART coyote there: out=LOC vici' i-ki· hi 'e-vi-vi' i htmt wa-pa'atfa. grasshopper-PL LOC-RDP-grab far EXT-flat 'Meanwhile, that old Coyote was off there yonder catching grasshoppers way out there in the flat.'

(b)

Αά ρύ hä'a w&-hi'e-vi-vi'i ί vici'i-ki. there: out SUBJ LOC them-RDP-grab ART grasshopper-PL 'He was off yonder there catching grasshoppers.'

5. Implications To recapitulate, in Cora, a wide variety of locational and directional morphemes and lexical items relate to other domains of the grammar in numerous ways. Some of these ways are at the level of the individual lexical item. They may distinguish one word from another, as we saw in the case of the verbs meaning 'carry,' 'take,' 'give,' 'bring' and 'to hold in the hand.' In other instances the locationals help to distinguish different tenseaspect forms of the same verb. The locational elements are also the starting points for lexical and semantic extensions that seem almost limitless. On the phrase and sentence level, many of them serve anaphorically, whereas others serve to elaborate their antecedents. Some also serve as procomplements in quotative sentences or as conversational closers. On the discourse level, Cora locationals help to get things started and bring them to an end. They also help to bring new participants onto the scene and take them off again or move them to a new locational or temporal setting for a later scene. Locationals are also used for structuring the way that events are sequenced in discourse and also to handle high level anaphora. To close, what good are locationals? They are good for lots of things, including the enrichening of theories of grammar and typology. The general conclusion is that, without locationals, the Coras (and Huicholes and Tarascans) would have to talk about the world in an entirely different way. 1 2

Notes 1.

I would like to express my thanks to an anonymous referee who made a number of very useful comments that have improved this paper considerably.

266

2.

3. 4.

Eugene Η. Casad

A systematic account of both the grammatical ization processes that relate these classes of locationals to one another and the semantic motivations for them is worthy of an entire monograph. Some of these matters are discussed in some detail in Casad 1982. For expositional clarity of the examples, I highlight both the locative elements in a given expression and their English glosses in italics. The following abbreviations are used in this paper. APPLIC: ART CAUS CNJ: COLL: COND: DEM: DISTR: DISTAL: DUR: EXT: FUT: HAB: IMPERF: LOC: BASE: MED: PRIV: SG:

5.

6. 7.

8.

Applicative Article Causative Conjunction Collective Conditional Demonstrative Distributive Distal Durative Extensive Future Habitual Imperfective Locative Base Medial Privative Singular

NARR: NEG: PASS: PST: PAUS: PERF: PERIP: PL: PROCOMP: PRTC: QNT: QUOT: RDP: REFL: RZD: STAT: SUBJ: SUBR:

Narrative Negative Passive Past Pausal Perfective Peripatetic Plural Procomplement Participle Quantifier Quotative Reduplication Reflexive Realized Stative Subjunctive Subordinator

As a convention, I use upper case lettering for the glosses of morphemes whose meanings are rather abstract and I employ lower case for those morphemes whose meanings are more like typical lexical items. The participial suffix -n functions in a variety of ways in Cora. Often it marks potential aspect, as in these examples. In other cases, it marks a subordinate clause, as in example (7a), Section 3.2.3. That particular clause is also understood as being in the imperfective aspect whereas the clause in (7c) that is marked with -n, shows it being used in the sense of 'having done X,' i.e. a perfective meaning. Finally, the use of -n in example (10c) and (lOd) illustrates its use with the present imperfective aspect. This text was provided by Justo Flores Aguilar in 1974-5. The an- in these verbs may be cognate to Nähuatl on-'away', but there are as yet no clear independently attested a:o vowel correspondences between Cora and Nähuatl. So any identification of the two prefixes is not yet established. To substantiate this will require careful comparative work. (Wick R. Miller, personal communication, July, 1989). This text was provided by the late Fidel de Jesus at the Mitla, Oaxaca workshop center in February of 1972.

What good are locationals, anyway?

9.

10. 11. 12.

267

The Quotative particle nu'u, as used in this example can be glossed as 'X is called Y,' e.g 'he is called the 'Secretary.' In its more common usages, nu'u marks narrated events in discourse and can generally be glossed 'so they say.' In other contexts it can be used to indicate indirect discourse and can be glossed 'he says X.' Cora also has other Quotative particles that mark direct discourse. The morphemic shapes are yee, yeehui and hui. These usages of these particles are discussed in considerable detail in Casad 1992. This text was also provided by the late Fidel de Jesüs at about the same time as the text in Footnote 2. This text was provided by Juan Celestino Laureano while he was at the S.I.L. Center Mois6s Saenz in Ixmiquilpan, Hidalgo, April 1974. This particular answer was given by my colleague Joe Grimes (personal communication, mid-November 1993).

References Casad, Eugene H. 1977 Location and direction in Cora discourse. Anthropological Linguistics 19: 216-241. 1982 Cora Locationals and Structured Imagery. Ph.D. Dissertation, University of California, San Diego. 1988 Conventionalization of Cora Locationals. In: Rudzka-Ostyn, Brygida (ed.). Topics in Cognitive Grammar. Amsterdam/Philadelphia: Benjamins, 345-378. 1992 Cognition, history and Cora yee. Cognitive Linguistics 3, 2: 151-186. 1993 Locations, Paths and the Cora Verb. In: Geiger, Richard A. and Brygida Rudzka-Ostyn (eds). Conceptualizations in Natural Language Processing. Berlin/New York: Mouton de Gruyter, 593-645. de Jesiis Serrano, Fidel and Eugene H. Casad. 1990 Cuento de los dos hermanos. Tlalocan XI: 173-178. Friedrich, Paul 1970 Shape in grammar. Language 46: 379-407. Grimes, Joseph E. 1964 Huichol Syntax. London/The Hague: Mouton. 1968 The Thread of Discourse. The Hague: Mouton. Langacker, Ronald W. 1993 Reference point constructions. Cognitive Linguistics 4.1: 1-38. Talmy, Leonard. 1988 The relation of grammar to cognition. In: Rudzka-Ostyn, Brygida (ed.). Topics in Cognitive Linguistics: Amsterdam/Philadelphia: Benjamins, 165-205. van Hoek, Karen Ann 1995 A cognitive analysis of bound anaphora. In: Casad, Eugene H. (ed.). Cognitive Linguistics in the Redwoods. Berlin/New York: Mouton de Gruyter, 753-791.

Iconicity in verbal descriptions of space Karin Wenz

1. Introduction The way we describe the experience of our spatial environment reveals important insights into the relation between language and cognition. Since the description of space presupposes a previous cognition of the environment, it may be assumed that the verbalization of spatial experience yields also insights into the processes of perceiving space. The study of verbalizations of space is hence fundamental to both linguistics and cognitive science. The topics of the present paper are: which structures of our perceptions of space are mirrored in texts, and how can we describe the semiotic relation between spatial cognition and the textual representation of space? Are there iconic principles which govern the linearization process of spatial descriptions? The empirical basis of my investigation is a corpus of 91 text passages from a London guide book in which typical places and buildings are described {London: A Handy Guide and Atlas, Sevenoaks, Kent: Geographer's A-Z Map, 1991).

2. The semiotic basis of textual representation of space Verbal representations of space are related to the space they describe by a sign relation comprising three components. The first is the sign vehicle consisting of the sequence of words which describe the spatial environment. The second is the object of reference, the spatial environment as such, and the third is the mental representation of the spatial perception. Within this triad, space is never an unsemiotically given piece of reality. Space is always a semiotic phenomenon insofar as its structure depends on the process of human perception. Even as a referential object, space is not an independently given phenomenon of the "real" world itself but depends on the cognitive capacity and structure of the perceiving mind.

270

Karin Wenz

2.1. Symbolic, indexical, and iconic representation of space Following Peirce, the relation between the signs of spatial descriptions and their object of reference is symbolic, indexical, or iconic. According to this semiotic typology, symbols are conventional, arbitrary, and general signs. Lexemes such as room, hall, attic, or corner are symbols in this sense. There seems to be no natural relation between these words and their object of reference. An index is a sign which stands in a relation of contiguity or causality to a particular object in a specific situation. In language, such indexical signs occur in the interactional situations in which utterances are used. They include indicators of the identity of speech participants, their location and orientation in space, and the time of the utterance. Deictic spatial and temporal expressions in language and many nonverbal expressions are examples of indexical signs. The iconic sign is related to its referent by similarity or analogy. I have to restrict my discussion to this third type of sign. Following Enquist (1986) and Ehrich (1985), texts are structured by principles of discourse organization and not by rules. What is the nature of these principles? According to Ehrich, the principles of textual organization are more communicative, aesthetic, and psychological than grammatical. To quote from Levelt (1989: 138), the aim is to "arrange information for expression according to the natural ordering of its content." This most basic principle of discourse organization is a principle of iconicity. Among other things it privileges a natural ordering which the ancients have called ordo naturalis. The principle, however, cannot be followed in texts which describe simultaneity of events and in particular the simultaneous perception of spatial configurations. In the light of Charles Sanders Peirce's theory of iconicity, the verbal patterns of spatial descriptions evince imaginal, diagrammatic, and metaphoric iconicity. These categories of iconicity represent their object with an increasing degree of abstraction or decreasing degree of similarity.

2.2. Imaginal iconicity Images are sign vehicles which share mere qualities with their object of reference. Stylistic means of imaginal iconicity in texts are repetition, rhythm, duration, and tempo. The qualities which these devices represent are structures of time, sequence, or quantity. In literature, delaying and accelerating of tempo are techniques of reflecting the protagonists' temporal sensation of actions and events. The possibility of reproducing space by

Iconicity in verbal descriptions of space

271

means of imaginal iconicity is limited. An example is typographic iconicity as used in visual poetry. Another iconic potential of imaginal iconicity is used in certain onomatopoeic features of the vocabulary of space. Consider the semantic opposition between cranny, nook, chest on the one hand and place or hall on the other. In these examples, the small spaces are expressed by means of short forms with a relative high phonetic constriction.

2.3. Diagrammatic icons of space Diagrams, according to Peirce, are icons which represent the relation of the parts of one thing by analogous relations in the sign vehicle. There are structural correspondences between the sign vehicle and its referential object. Since only relations and structures are considered, diagrammatic icons evince a certain degree of arbitrariness. Because of the digital and linear character of language, diagrammatic icons are more frequent in texts than imaginal icons. Diagrammatic iconicity appears in texts when linear relations within the text stand for temporal, spatial, causal, or social relations in the described world. These extralinguistic relations, which structure our experience as complex principles of order, are mirrored in the text as icons. A verbal description of space is necessarily less iconic than a description of time because complex visual expressions have to be transformed by a process of linearization into a temporal succession first. This projection of a temporal structure onto an originally spatial one has the following effect: static objects and configurations in space are now represented in the temporal medium language and are thus transformed from static to dynamic. The asymmetry between objects and configurations in space on the one hand and the temporality of spatial representation on the other is evident. In the following I can only focus on diagrammatic iconicity, with special attention to the different features of static and dynamic representation. As far as my text corpus is concerned, most descriptions are static ones. Since the aim of the guide book is to describe typical places and buildings and not touristic routes, there are only few purely dynamic descriptions.

272

Karin Wenz

3. Dynamic diagrammatic icons of space and some principles of semiotic coding

3.1. Temporal ordo naturalis In dynamic descriptions of space, cognitive structures are iconically represented in the mode of an ordo naturalis which requires a moving observer. The temporal order of his or her perceptions of space is reflected in the linearity of discourse (Nöth 1990a: 202): Dynamic space is the sequence of places experienced in the progression from a point of departure to a goal. It is associated with movement and thus with time.

The loci in space are paced off like the path of a journey. The complex spatial arrangement is transformed by means of an imaginary tour. In descriptions of a city, this mode of description is usually the one of a guided tour of its sights and institutions. There are only four purely dynamic texts in my corpus, but five additional texts evince a change from a static to a dynamic description, and three other texts which begin with a dynamic approach to the described building via a path and later continue with a description in a static mode. My first example is the description of Admiralty Arch. (1) Admiralty [1] and Admiralty Arch [2]; Whitehall [3]; The Mall [4] See the long view through from Trafalgar Square [5], the red surfaced processioned way of The Mall [6] leading direct to Buckingham Palace [7], In this and the following analyses, the loci to which there is reference in the text are marked by italics, followed by numbers in square brackets in the order of their textual occurrence. The syntactic equivalent of a locus is usually either a noun phrase or an adverbial phrase. Its semantic equivalent is either the name or description of a place or the designation of an object to be found at that place. The criterion of assigning a new number to a given local segment in the text is its local separation and distinctness from the previous place. Global loci which designate the sum total of several individual loci (e.g., West End or National Portrait Gallery) are designated by the sum total of the individual loci described at these larger places (e.g., [1=2-11]). Whenever the same locus is designated several times, each new

Iconicity in verbal descriptions of space

273

occurrence of this locus is marked by a superscript (e.g., [31]). In example (1), the observer is invited to follow an imaginary tour beginning with a first point of view at Trafalgar Square [5], then proceeding to Admiralty Arch on The Mall [6] and finally arriving at Buckingham Palace [7], The linearization follows an ordering from near to far. The dynamic mode of this description is underlined by the use of the motion verb "leading to". My second example is the description of pictures at the National Portrait Gallery: (2)

National Portrait Gallery [1]; St. Martins Place [2] ... The portraits [3] are in chronological order; coins [4] and early copies [5] represent the pre-Tudor monarchs [6], leading up to the unrivalled collection of 16th century works [7] commissioned by Tudor [8] and Elizabethan sovereigns [9], their advisors [10] and courtiers [11]. When the 18th century is reached [12], the display divides into two parallel series [13]: one shows rulers [14], statesmen [15] and men of action [16]; the other great names in art [17] and science [18] ...

The description of the loci of the pictures in this gallery is one of chronological order. It corresponds to the spatial order of the pictures in this gallery. The viewers' stations in passing through the gallery is rendered in the order of the description. Beginning with the pre-Tudor monarchs [6] up to the 18th century [12] the verbal sequence follows a temporal principle which is at the same time a spatial order. The 18th century is then taken as a new starting point for two different routes, both ending in the present time. The above descriptions (1) and (2) are iconic, following the principle of ordo naturalis in an imaginary tour, but the second text contains in addition a historical, chronological mode of diagrammatic iconicity. Whereas the ancient rhetorical term ordo naturalis was meant to describe word order reflecting a natural order of the real world, we have to specify that not only order in nature but also social, cultural, and cognitive schemata can be discovered as semiotic determinants of natural linearization. Textual order reflects principles of order which stem from various other, nontextual semiotic systems. In text (2), the sequence of [14], [15], and [16] illustrates a different nonchronological principle of ordering, namely a social one. In the following, such diverse systems will be referred to as codes. Following Nöth (1993), I will discuss three other or more specific codes in addition to the general code of human cognition, the perceptual code. These more specific codes are the egocentric code, the social code, and the axiological code.

274

Karin Wenz

3.2. Egocentric coding As Piaget's research has shown, the child experiences himself or herself as the center of his or her world. However, the egocentric perspective of the world is not restricted to children. To a lesser degree, every perceptual event is egocentric since every percipient has a perspective of his or her own. This egocentric perspective is reflected in discourse deixis whenever the speaker's standpoint coincides with the origo of discourse. In the linearization strategy of an imaginary tour, there is egocentric orientation insofar as the speaker's imagination is taken as the starting point of the description and not as an independent frame of reference. The hearer is simply invited to adopt the speaker's perspective. Even in the case of intrinsic perspective, where the object of description seems to be the only origo of representation, there is some subjectivity of the observer because this kind of representation is nevertheless based on a linearization of space which is only partly derived from features of the object. An additional factor in the perspective is the typical orientation of the user to the object. Intrinsic features are furthermore interpretations which are often made in analogy to structures of our own body (see Levelt 1984, 1986; Klein 1990). Therefore, Klein describes the intrinsic perspective as one form of a "frozen deictic origo" and draws our attention to the egocentric aspect of the intrinsic perspective. Insofar as perspective descriptions are subjective and egocentric, they can also be interpreted as iconic in the sense of the egocentric code. Consider the following description of Downing Street·. (3)

Downing Street [1]; Lincoln's Inn [2]; Fields [3] Here [1'], behind the hubbub of sightseers [4], diplomatic cars [5] and police [6], are numbers 10 [7] and 11 Downing Street [8], the Official residences [9=10-11] of the Prime Minister [10] and Chancellor of the Exchequer [11] ...

The description begins with the explicitly named deictic word here [1']. It first indicates the imaginary deictic perspective of the author to the reader right in the center of touristic interest and surrounded by the loci [4] to [6]. The position [1'] is then identified with the two buildings Downing Street 10 and 11 [7-8]. This privileged perspective of the reader behind the loci [2], [3] and [4] is necessarily an imaginary one since the place of the real tourists is actually among the sightseers which have no access to the doors of Downing Street 10 and 11.

Iconicity in verbal descriptions of space

275

3.3. Social coding The social code reflects orders of a social hierarchy mapped onto space. Such hierarchies reappear in terms of higher and lower or central and peripheral loci. Textually, the hierarchically higher rank is mentioned first. In text (2), the sequences of sovereigns [4], their advisors [5] and courtiers [6] and rulers [9], statesmen [10] and men of action [11] reflect such hierarchical orders (see above). Another example is: (4)

Madame Tussauds [1], Marylebone Road [2] ... Both historical [3] and contemporary world leaders [4] are represented, royalty [5], politicians [6], sportsmen [7J, famous personalities from all walks of life [8J;

Besides the chronological order in the phrase historical [4] and contemporary [5] the word order reflects the hierarchy of social roles of British society [5-8]. The generalization famous personalities from all walks of life [8] is at the same time an addition and therefore the end of a social hierarchy. The natural diagrammatic representation of this social order is based on conventional rules and evidently not on natural cognition.

3.4. Axiological coding The axiological code associates cognitive and social phenomena with evaluations. Insofar it is closely related to the social code. Consider the following description of the Barbican Arts Centre: (5)

Barbican [1] The Barbican Arts Centre [2=5-18] is the London equivalent of the Lincoln Center, New York [3] or the Centre Pompidou, Paris [4]; its wide range of facilities including Barbican Hall [5], Barbican Theatre [6], The Pit studio theatre [7], Barbican Library [8] (the City Lending Library [9]), art gallery [10], sculpture court [11], cinemas [12], conference [13] and trade exhibition areas [14], rooftop conservatory [15], restaurants [16], bars [17] and car park [18].

The enumeration follows an ordering from the most highly to the least highly evaluated places of the Barbican Arts Centre [2], beginning with the whole, which is characterized in analogy to other well-known places [3-4] and then listing its various buildings in the order of their cultural

276

Karin Wenz

importance [5-18] and not in the topological order in which we would reach these places in the course of a tour. Not only the listing of the various buildings reflects an axiological coding but also the order of Lincoln Center, New York [3] or the Centre Pompidou, Paris [4] mirrors a cultural order of importance of these two art centers. Texts of the evaluative type of order are frequent in my corpus. There are 69 texts (of 91) with 132 enumerations which follow the sequence from the most important to the marginal. Only five examples are incompatible with this schema.

3.5. Further principles of enumeration In addition to the above mentioned principles of ordering, there are two further typical schemata of linearization of loci, namely alphabetical order and circular tour. Let me begin with an example of alphabetical order. One of 58 examples of this type of order is the description of Courtauld Institute Galleries: (6)

Courtauld Institute Galleries [1] This famous art collection [l'=2-6] ... includes: the Courtauld Collection of Impressionist [2] and Post-Impressionist painting [3], the Lord Fareham Collection (14th-18th century art) [4]; the Roger Fry Collection (Bloomsbury Group and Impressionists) [5]; the Witt Collection of Old Master Drawings [6].

Besides the chronological ordering from Impressionist [2] to Post-Impressionist painting [3], this text reflects further an alphabetical linearization principle. The sequence from the Courtauld Collection [2] to the Fareham Collection [4], the Fry Collection [5], and finally the Witt Collection [6] follows an alphabetical schema from C to Fa, Fr, to W. Most examples which follow the alphabetical coding are mere lists of sights, which have nothing in common with a spatial description. The alphabetical ordering is a general means to avoid the evaluation which the reader associates with the first position in a text. The second principle is one which I would like to call 'circular tour'. My example is from a description of Holborn-Strand: (7)

Holborn-Strand [1] ... There are many other interesting places [2-4-10] tucked away [3] awaiting discovery- Sir John Soane's Museum [4], Old Curiosity Shop [5], Staple Inn [6]; the very special atmosphere of

Iconicity in verbal descriptions of space

277

Fleet Street [7]; the old pub Cheshire Cheese [8] and Dr. Johnson's House [9]; and the famous Strand [10] ... The ordering of this last example is a peculiarity of my corpus which occurs only once and which is only noticeable in comparison to a London map (see Figure 1 below). The ordering follows a circle around Sir John Soane's Museum or a spiral, beginning with this museum [4] and then going on in a clockwise order from Old Curiosity Shop [5] to The Strand [10] removing more and more from the starting point. From Sir John Soane's Museum [4], the way leads to the nearby Old Curiosity Shop [5], further on to Staple Inn [6], down Fetter Lane to Fleet Street [7]. When Fleet Street is reached, we arrive at the old pub Cheshire Cheese [8], Dr. Johnson's House [9] lies in a side street within easy reach. Walking down Fleet Street, The Strand [10] is reached, since both streets join at the height of Waterloo Bridge. (For examples of a circle around a source in form of a boundary see 4.2.) Smlihlinld

Lenim Ll Weilhir CentHolborn. n tfolb°I™

Lincoln'!

CCVESTi I GMj7

1Π W W " Sl • It^gSzZr /-m»CKFRUUtS

Cnute Bint

THAMES

William Itf St_V

StTltJrtMnC / HAHIHGx jmdss;

JX ^Ontri'i ^««1-ulFII.NjMjm.t % y///A iSSSr „Qwai sivt --jiVv"r y-Tel« i^lC

Beyil

tai.r'VA^V \ 5

Figure 1. A partial map of Central London

Ground Slreel

278

Karin Wenz

4. Static descriptions of space Static space is perceived from a fixed point of view. Its verbal representation can result in diagrammatic iconicity when the textual order corresponds to the order of the perceptual sequence. Results of the cognitive psychology show that there are laws which govern the intensity of our perceptions. The more intense or salient perceptions are also the ones that we notice first. Another principle determining the sequence of perceptions is egocentric proximity. Static descriptions of space tend to result in verbal linearizations following these principles of perceptual salience and subjective proximity. This order of perception is reflected in discourse linearizations when the spatial configurations are described in sequences such as: larger size before smaller size, figure before ground, center before periphery, near before far, top before bottom, bright before dark, and dynamic before static. Verbal ordering according to these determinants of perception is part of the perceptual code.

4.1. Figure - Ground The simultaneity of our perceptions with its overlap of information requires a distinction as to the relevance of this information. This distinction differentiates between central information, the figure, and peripheral information, the ground. The figure as the central topic in the text yields more information than the ground. As Givon (1985: 206) puts it: "The more important an item is in the communication, the more distinct and independent coding expression it receives." The textual sequence reflects the priority of the figure over the ground. 127 passages in my corpus maintain the sequence figure before ground, as for example the sequence of loci in the following headlines: (1)

Admiralty [1] and Admiralty Arch [2]; Whitehall [3]; The Mall [4]

(2)

National Portrait Gallery [1]; St. Martins Place [2]

(3)

Downing Street [1]; Lincoln's Inn [2]; Fields [3]

(4)

Madame Tussauds [1]; Marylebone Road [2]

The sequence of the places enumerated in these headlines clearly reflects the sequence of figure (marked by bold type in the original text) before ground (in regular type) both from the point of view of the reader, to

Iconicity in verbal descriptions of space

279

whom the distinction appears in typography, and to the tourist, in whose perception the places in bold type are in the center of attention against the ones in normal type, which remain in the perceptual background. Another text which reflects this principle of linearization is the following description of Scotland Yard: (8)

Scotland Yard [1] A purpose built block [2] constructed with granite [3] excavated by Dartmoor convicts, in the Scottish baronial style [4], to designs by Norman Shaw, was completed in 1891, overlooking the then new Victoria Embankment [5],

Here, the figure is the purpose built block [1] of the Headquarter, with the description of its material [3] and style [4]. The ground of this visual center of attention is Victoria Embankment [5], which is mentioned last.

4.2. Center - Periphery Comparable to this description are texts which follow a sequence from a center to the periphery. I want to demonstrate one of 43 examples which describe the periphery in the form of a boundary around a central place. Consider the following description of St. Paul's Cathedral·. (9a)

St Paul's Cathedral [1]; St Paul's Churchyard [2] ... Christopher Wren's Cathedral [3], built 1675- 1710, captures the peculiarly English trait of compromise, combining Gothic precedent [4] and new ideals [5], while balancing large scale forms [6] with human scale detail [7]. The cavernous octagonal intersection [8] between nave [9], choir [10] and transepts [11] forms the internal focal point [12],

After a general description of the architectonic style [4-7], the octagonal intersection [8] is situated within nave [9], choir [10], and transepts [11] and described as the focal point [12] that means the center. The description goes on: (9b)

and here [12'], 100ft above the floor [13], is the famous Whispering Gallery [14], where whispers against the wall on one side [15] can be heard near the wall opposite [16]. You can ascend higher [17] still;

280

Karin Wenz

After a frame is set and the origo is defined, the perspective changes from intrinsic to deictic with the adverbial here [12']. Then the further description follows the strategy of an imaginary tour. The starting point is the famous Whispering Gallery [14] leading further on higher and higher [17]. The sequence from near to far is inherent in these descriptions from center to periphery. Compare [15-16] which also follows this sequence from near to far. A description of a boundary in the form of a map is in the following example of Green Park: (10)

Green Park [1] Laid out by Charles II in 1667 as an extension [1'] to St James's Park [2], Great wrought iron gates [3] stand at each end of Broad Walk [4]: Dominion Gate [5] facing The Mall [6] and gates from old Devonshire House [7] towards Piccadilly [8], Queen's Walk [9] on the eastern edge [10] is overlooked by the Ritz Hotel [11] and many former mansions [12].

The park is only described as an extension of St James Park [2], Its triangular form is marked off by Dominion Gate in the South East [5], Devonshire House in the South West [7], and Queen's Walk from South to North East [9]. With these boundaries the contours of the park are determined. A linearization schema from top to bottom occurs in twelve examples among them a description of Albert Memorial: (11)

Albert Memorial [1=3-7]; Kensington Gardens [2] London's most eccentric monument and the epitome of High Victorian taste. The 180 ft spire of ornate stonework [1'] is encrusted with mosaics [3], gilt [4] and pinnacles [5], surrounded by allegorical statues [6] and relief portraits offamous artists [7]...

Beginning with the center of this monument the 180 ft spire [1'] with its ornaments [3-5] the description leads down to the surrounding statues [6] and relief portraits [7], Besides the linearization from top [1'] to bottom [67], this description follows a sequence from the center, the spire [1'], to the periphery, the surrounding statues [6-7].

Iconicity in verbal descriptions of space

281

4.3. Optical salience Texts which reflect the perceptual salience of bright objects are relatively rare in my corpus. There are only two examples which describe bright before dark objects as in the passage from Royal Opera House: (12)

Royal Opera House [1]; Bow Street [2]; Covent Garden [3] ... The interior [4] is of lush decoration [5] in white [6], gold [7] and deep crimson [8],

The colors of the decoration are listed in an ordering from the bright white [6], over the shining gold [7] to the dark color deep crimson [8]. The order of perception from the most to the less salient is maintained in the linearization of this passage. This order does not reflect a linear process of perception but a simultaneous perception of these colors which are listed according to their perceptual salience. 4.4. Size This principle derives the order of loci according to the criterion of size: larger size precedes smaller size. The enumeration from big to small as in the following description of the Natural History Museum is only one of 22 examples which reflect this sequence: (13)

Natural History Museum [1] The traditional displays include a range of exhibits from huge Dinosaurs [2], Whales [3] and Elephants [4] to tiny birds [5] and insects [6]\

The description starts with the largest animal we know, the Dinosaur [2], and leads to the smallest, the insect [6]. The sequence from dynamic to static objects does not occur in my corpus, which is a mere description of buildings and places and not of events. Up to this point the cited texts reflect various iconic principles which are discussed above. But the question arises: Are there no counterexamples? 5. Counterexamples as conflicts of coding There are counterexamples which do not follow the principle of iconic coding, particularly in two groups of coding, which have to be discussed now.

282

Kann Wenz

The first concerns the linearization from center to periphery. There are 43 examples which follow this principle, but 23 counterexamples. In these examples the description starts at the periphery and from there goes on to the center. Let us have a look at a passage of the description of Bloomsbury: (14)

Bloomsbury [1] An area [Γ] famous for national institutions [2], educational establishments [3] and Georgian domestic architecture [4] set around peaceful squares [5],.. Bloomsbury [1''] was once a highly fashionable residential area, especially in the early 1900's with the intellectual and artistic Bloomsbury Group; the area is now visually dominated by the colossal University of London Senate House [6],

Bloomsbury [1,1',1"] as an area with a special cultural history is the periphery or background of the visually dominant University of London Senate House [6], The reversal of order in examples like this is based on an ordering which is prototypical for the succession from theme to rheme in texts (cf. Dressier 1989). The theme, in our case the area Bloomsbury, was introduced in the previous text [1, 1'] and is the known element [ 1 " ] in this description. It is closely related to the previous context. The new, but perceptually more salient University of London Senate House [6] is mentioned for the first time and functions as the rheme. The thematic ordering dominates the order of the perceptual code. Such reversal of order is typical of texts which describe not only a building or place but a whole area. In the beginning the area is described in general with some interesting places and then the center is introduced in a second step. The second group with apparent counterexamples are descriptions which follow the sequence from top to bottom. 30 examples maintain this perceptual ordering, but 15 examples contradict it. Four of these examples are descriptions of the interior of a church. See example (9a) and (9b), the description of St. Paul's Cathedral. All four examples follow the strategy of an imaginary tour starting at the entrance or at some central point in the interior and then ascending up to the belfry, a typical touristic experience. One other description with the ordering from bottom to top is important because it reflects a conflict between two possible iconic codings. Consider the example of Regent Street: (15)

Regent Street [1] ... one of London's most ethereal views [2J: down Lower Regent [3] from Piccadilly Circus [4] over the tree tops [5] to the towers [6] and pinnacles [7] of Westminster [8=6-7].

Iconicity in verbal descriptions of space

283

The view from Piccadilly Circus [4] to the towers of Westminster [6] is on the one hand a description from the bottom (Piccadilly Circus) to the top (the towers), but on the other hand an imaginary tour which starts at Piccadilly Circus [3-4], leads further on to the nearby trees [5], then to the far away towers [6], and finally to the pinnacles of Westminster [8], The conflict which arises between two schemata of perception, the subjective proximity and the direction our vision follows from top to bottom, is solved as described above, because the temporal ordering in the form of an imaginary tour comes to meet with the linear and therefore temporal mode of a verbal description. The following table shows the different principles which are inherent in the discussed texts. These texts reflect the principles of social, cultural, and cognitive ordering just like textual ordering according to the distribution of information. These principles coexist as examples 1, 2, 4, 5, 8, 9, and 12 show or come into conflict as examples 3, 10, 11, 14, and 15 demonstrate ( + indicates the occurrence of, - the contradiction to a semiotic principle). The multiplicity of semiotic coding is obvious in these examples (see Table 1).

•ο

r+ *? 5 fS

+ ¥

size

«Λ

+ 2

+ 9-11/12

Karin Wenz

theme rheme

284

+

+ ιΗ ?

ο

fS

optical salience

ί

+ 5

i + 2

chronological order

ΪΗ

»•H ® + ^ + ·? + 00

+ 2

ι

2-5,1"6

+ 9-11/12 rs

ι » Ο ΙΛ

+ 2+ 2

ο + Τ •t

T

r»

I m j. »00 η

Μ +

ί-Η

00

+ '

+yrs

+ 2

^

Γ-

00 + 4

i t s J 8 J

texts

«ο

+ 2+ 2

egocen- social tric coding coding

+

«Λ h ι + $ 00

+ 5-6, 7-8

near far

+ 2

axiologi alphabe circular figure -cat tour -tical ground coding order

+ 2

center periphe -rie

top bottom

+ 2

Μ + τ NO

+yts

+ 2

+ 2

Tt

ν©

00

Ο

Μ

Table 1. The occurence of semiotic principles

m

Iconicity in verbal descriptions of space

285

6. Concluding remarks Let us remember that the ancient idea of mimesis and mimetic imitation of world in language is the precursor of the theory of iconicity. In its naive sense, mimesis means a direct mapping of world onto language. However, in the era of cognitive science, mimesis is being considered as a process of world construction and not as one of world imitation (Gebauer and Wulf 1992). Language and linguistic description can be used as a tool to construct a world via categorization. At the same time language provides access to the principles which are the basis of this cognitive construction. As Neisser (1967: 279) puts it, "Stored information consists of traces of earlier constructive acts, organized in ways that correspond to the structure of those acts." By categorizing the structures of our simultaneous perceptions we are making "sense of the world" (Smith 1990: 33). The interpretation of different perceptual cues take place at a higher cognitive level, where they are mapped onto the cognitive representation of space. That means that there is no direct connection between perception and language as the ancient theory of mimesis implies. Perceptual features are reflected in language only via conceptualization. As Lakoff (1987) defines it: "SEEING TYPICALLY INVOLVES CATEGORIZING." Seeing includes always seeing-as, that is categorizing what is perceived. Perception has to do with the interaction between the perceiver and the perceived world. Perception depends on cognitive skills, experience, and emotional states of the perceiver. "Perceivers pick up only what they have schema for, and willy-nilly ignore the rest" (Neisser 1976: 80). Lakoff and Johnson (1980) describe metaphor as a fundamental principle of conceptual organization, which is based on facts of our physical and cultural experience, as the field of spatial metaphors show. The natural connection between body and space (movement in space, upright posture) is the basis for our cognition of space. The embodiment of concepts provides an iconic connection between the perceiver and the perceived world. Iconicity directs our gaze towards categorization which is a motivated act of construction. The mediation between perception and language is achieved by social, cultural and cognitive schemata which can be named as semiotic determinants of linearization.

References Dressier, Wolfgang U. 1989 Semiotische Parameter einer textlinguistischen Wien: Akademie der Wissenschaften.

Natürlichkeitstheorie.

286

Karin Wenz

Ehrich, Veronika 1985 Zur Linguistik und Psycholinguistik der sekundären Raumdeixis. In: Schweizer, Harro (ed.)· Sprache und Raum: Ein Arbeitsbuch für das Lehren von Forschung. Stuttgart: Metzler, 130-161. Enquist, Nils Erik 1986 Linearization, text type and parameter weighting. In: Mey, Jacob (ed.). Language and Discourse: Test and Protest: A Festschrift for Petr Sgall. Amsterdam: John Benjamins, 245-260. Gebauer, Gunter and Christoph Wulf 1992 Mimesis: Kultur - Kunst - Gesellschaft. Reinbek: Rowohlt. Givön, Talmy 1985 Iconicity, isomorphism, and non-arbitrary coding in syntax. In: Haiman, John (ed.). Iconicity in Syntax. Amsterdam: Benjamins, 187-219. Klein, Wolfgang 1990 Überall und nirgendwo: Subjektive und objektive Momente in der Raumreferenz. Zeitschrift für Literaturwissenschaft und Linguistik 78: 9-42. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago: University Press. Lakoff, George 1987 Women, Fire, and Dangerous Things. Chicago: University Press. Levelt, Willem J.M. 1984 Some perceptual limitations on talking about space. In: van Doom, Andrea J., Wim A. van de Grind and Jan J. Koenderink (eds). Limits of Perception: Essays in Honour of Marteen A. Bouman. Utrecht: VNU Science Press, 323-358. 1986 Zur sprachlichen Abbildung des Raumes: Deiktische und intrinsische Perspektive. In: Bosshardt, Hans-Georg (ed.). Perspektiven auf Sprache: Interdisziplinäre Beiträge zum Gedenken an Hans Hörmann. Berlin: de Gruyter, 187-211. 1989 Speaking: From Intention to Articulation. Cambridge, ΜΑ: ΜΓΓ Press. Neisser, Ulric 1967 Cognitive Psychology. New York: Meredith. 1976 Cognition and Reality: Principles and Implications of Cognitive Psychology. San Francisco: Freeman. Nöth, Winfried 1990 Handbook of Semiotics. Bloomington: Indiana University Press. 1990a The semiotic potential for iconicity in spoken and written language. Kodikas/Code 13.3/4: 191-209. 1993 Iconicity of symmetries and asymmetries in syntactic coordination. In: Küper, Christoph (ed.). Von der Sprache zur Literatur. Tübingen: Stauffenburg, 23-36. Smith, Edward E. 1990 Categorization. In: Osherson, Daniel N. and Edward E. Smith (eds). Thinking: An Invitation to Cognitive Science. Cambridge, ΜΑ: ΜΓΓ, 3353.

Section 4 Variability in the conceptualization of space

The syntax and semantics of locativised nouns in Zulu John R. Taylor

1. Introduction Zulu is the major language of the Republic of South Africa, being spoken as mother tongue by some 8 million persons, or about 20% of the population. One of the most important resources of the Zulu language - and of a number of other Bantu languages, such as Sotho, Tswana, and Swahili - for the expression of spatial relations are the locativised nouns. 1 Generally, locativised nouns are marked by initial e-/o- and by suffixed -(w)ini/-(w)eni (Doke 1981: 232-236). (1)

indlu 'house' isikolo 'school' izulu 'sky' ubonda 'wall'

> > > >

endlini 'house-LOC' esikoleni 'school-LOC' ezulwini 'sky-LOC' obondweni 'wall-LOC'

In addition to the locativised nouns, Zulu has a number of complex preposition-like phrases, used for the expression of specific spatial locations. 2 Some of these are illustrated in (2) and (3). (2)

phezulu kwaphansi kwaphambi kwaphakathi kwaphandle kwa-

(3)

a. b.

'on, on/at the top of 'under, on/at the bottom of 'before, in front of 'between' 'outside o f

Inkomishi iphezu kwetafula 'The cup is on the table.' Ikati liphansi kwetafula 'The cat is under the table.'

288

John R. Taylor

c.

Intombazane iphambi kosofa 'The girl is in front of the sofa.'

In this paper, I specifically address the semantics of locativised nouns, as exemplified in (1), paying particular attention to the uses of these expressions in conjunction with motion verbs.

2. The locative as a place function One use of locativised nouns is in the copulative construction, to indicate the place of a trajector entity. (4)

a. Abantu basendlini people they-house-LOC3 'The people are in the house.' b. Inhlanzi isemanzini fish it-water-LOC 'The fish is in the water.' c. Isigqoko sisekhanda hat it-head-LOC 'The hat is on (his) head.' d. Isitembu sisemvilophini stamp it-envelope-LOC 'The stamp is in/on the envelope.' e.

Umgexo usentanyeni necklace it-neck-LOC 'the necklace is around (her) neck.'

Especially in light of their English translations, these examples could suggest that the Zulu locative is inherently vague with respect to the configuration of trajector and landmark entities. Thus, in (a) and (b), the locative designates an 'in'-relation, in (c) an 'on'-relation, whilst (d) is ambiguous between the two readings. This is not to say that Zulu speakers do not have the means to express more precisely the configuration of trajector and landmark. Complex prepositional expressions, like those in (2), for example, are readily available for this purpose.

Locativised nouns in Zulu

289

Locativised nouns are also used as complements of a motion verb. The following examples are from Doke (1981: 232). (5)

a. Ngakhwela endlini I-climbed-up house-LOC Ί climbed onto the house.' b. Ν gaya endlini I-went-to house-LOC Ί went to the house.'

c.

Ngangena endlini I-entered house-LOC

Ί went into the house.' d. Ngaphuma endlini I-left house-LOC

Ί went out of the house.' e.

Ngasuka endlini I-went-away house-LOC

Ί went away from the house.' An inspection of these examples could further suggest that locativised nouns are not only vague with respect to different kinds of static place relations ('in', 'on'), they are also underdetermined with respect to the distinction between a static relation of place, and dynamic relations of goal ('to', 'into', 'onto', etc.) and source ('from', 'out o f , etc.). The precise interpretation of the locative would be determined by context, especially by the meaning of an associated verb. A number of commentators on the Zulu language have remarked on the inherent vagueness of locativised nouns. 4 Thus, Ziervogel (1991: 373) writes that the locative "simply indicates 'in' or 'at' or 'to' and so on, and cannot be pin-pointed to either rest or motion unless determined by context". Likewise, for Doke (1981: 232), "the specific meaning of the locative is determined by the verb with which it is used, or the context in which it is found". Doke also raised the question of the general meaning of the locative, "isolated from the verb which gives its specific meaning'. This, the 'real meaning of the locative", might be captured, he suggested, by the phrase 'with respect to'. Thus, Doke proposed, as a more accurate rendering of (5a), Ί climbed-up with respect to the house'.

290

John R. Taylor

This traditional account of locativised nouns is unsatisfactory, in two respects. Consider, first, the alleged vagueness of locativised nouns vis-avis static relations of place vs. dynamic relations of source and goal. The following sentences, although superficially parallel to those in (5), are, for many speakers, ungrammatical, and even for those speakers who do find the sentences acceptable, the meaning is not that indicated in the gloss. 5 (6)

a.

*Ngihamba emsebenzini I-walk work-LOC Ί walk to work.'

b.

*Ngindiza eThekwini I-fly Durban-LOC Ί fly to Durban.'

c.

*Ngishayela eyunivesithi I-drive University-LOC Ί drive to the University.'

In (6a), emsebenzini 'work-LOC' is not able to be interpreted as the goal relation 'to (my place of) work'. If anything, (6a) would mean Ί walk around at (my place of) work'. Likewise, to the extent that (6b) is acceptable, it would mean Ί fly (around) in Durban', locative eThekwini cannot mean 'to Durban', even less, 'from Durban'. These examples clearly show that the notions of goal and source do not inhere in the locativised nouns themselves. Rather, these notions must derive from the associated verbs. The examples in (6) do not have a goal reading precisely because the verbs hamba 'go, walk', ndiza 'fly', shayela 'drive' express only the fact (and the manner) of motion, they do not incorporate the notion of motion to or from a place (nor even, as I shall argue later, the notion of motion in a place). The examples in (5) can have the given interpretations since the verbs do incorporate the notions of goal and source. Ya in (5b) is best glossed as 'go to', phuma in (5d) as 'go out of', and so on. A second aspect of traditional accounts is that locativised nouns are said to be inherently vague with respect to the spatial configuration of trajector and landmark entities - that, in other words, the locative is schematic for the various kinds of more precisely defined relations expressible by complex phrases such as those in (2). Again, this claim needs qualification. It is certainly not the case, for example, that any spatial configuration is compatible with the locative morphology. The cat under the table, the people who live near the church, the children playing behind the wall - these are

Locativised nouns in Zulu

291

locative relations which definitely can not be expressed by locativised nouns! Rather, the preferred interpretations of locativised nouns construe the relation of trajector and landmark as one of enclosure (the people in the house), or, in case the enclosure relation is conceptually incoherent, a relation of contact with a visible surface (the fly on the wall), or, failing this, the trajector is located within an interactional region contiguous to the landmark ('the man working at the desk'). Consider the following contrasts (Suzman, to appear). (7)

a.

Inyoni isemuthini bird it-tree-LOC 'The bird is in the tree.'

b.

Inyoni iphezulu

emuthini

bird it-up tree-LOC

'The bird is in the top of the tree.' c.

Inyoni iphezulu komuthi bird it-above of-tree 'The bird is above the tree.'

The locativised noun emuthini 'tree-LOC' in (7a) strongly evokes the interior of the region defined by the branches of the tree. If the bird is located in the upper part of this region (if the bird, say, is perched on an upper branch of the tree), (7b) is used. Here, the locativised noun occurs in conjunction with the locative adverbial phezulu 'at the top, upwards'. If the bird is above (i.e. not in) the tree, the locativised noun is inappropriate, and the complex prepositional phrase phezulu kwa- is indicated. In view of these considerations, my suggestion is that the function of the locative morphology is to convert a 'thing'-concept into a 'place'-concept. In other words, locativised nouns denote a place, or region in space, identified with respect to a thing. Topological, interactional, and functional properties of a thing may be important in defining the region. The region defined by a three-dimensional container, such as a 'bag' or a 'bucket', will typically be the interior; the region defined by a porous medium, such as air or water, will be the space occupied by the medium; the region defined by a wall will be its visible vertical surface; the region defined by a desk will be its working area, and so on. To this extent, certainly, the meaning of a locative is non-specific. It is not the case, however, that a locativised noun denotes any location in the neighbourhood of a thing. A crucial aspect of this claim is that a locativised noun denotes a place per se, it does not construe the place as the goal or source of directed motion.

292

John R. Taylor

Some evidence supporting the analysis of the locative morphology as a place function, comes from those nouns which fail to form locatives by the pattern exemplified in (l). 6 We can identify two classes of exceptions to the pattern. Firstly, there are some nouns whose locatives are marked only by initial e-/o~, and which lack the -(w)ini/-(w)eni suffix. To some extent, we have to do here with an idiosyncratic property of specific lexical items. But while the set of nouns which take a defective locative morphology cannot perhaps be predicted by rule, the nouns do seem to share a common property. This is that they denote entities which are, in a sense, already locative in character. Compare indlu 'house' and ikhaya 'home', 'birthplace'. The former denotes a concrete three-dimensional entity, and it takes full locative morphology: endlini. The latter denotes more an environment, or ambiance, and it takes the defective locative morphology: ekhaya/ *ekhayeni. Other nouns which fail to take full locative morphology include names for parts of a hut (8), regions of the human body (9), cardinal points (10), and vaguely demarcated geographical areas (11). Morphologically, many names of towns (see Doke 1981: 235f.) are defective locatives (12). (8)

umnyango 'doorway' iziko 'fireplace, hearth' umsamo 'back of hut'

> emnyango/*emnyangeni > eziko > emsamo

(9)

ikhanda 'head' umhlane 'back'

> ekhanda > emhlane

(10)

impumalanga 'East' intshonalanga 'West'

> empumalanga > entshonalanga

(11)

ihlane 'wilderness' ulwandle 'ocean

> ehlane > olwandle

(12)

umDloti (name of river) iNanda (hill range) igolide 'gold'

> emDloti 'Verulam' (name of town) > eNanda 'Lindley' (name of town) > eGoli 'Johannesburg'

Secondly, nouns denoting humans (that is, in the main, nouns of class 1 and 2), do not form locatives at all by the pattern in (1). From umuntu 'person' we do not get *omuntu, or *omuntweni. The reason, clearly, is that humans, being mobile, do not define a region in space. Class 1 and 2 nouns take alternative morphemes, e.g. ku- and kwa-. Ku- and kwa-, however, differ significantly from the place function under discussion here. The former may have a directional component (kubaba 'to father'), whilst kwa-

Locativised nouns in Zulu

293

denotes the fixed property, or territory, of a person or persons (kwaZulu 'Zululand'). 7 3. Things and places 'Thing' and 'place' are two of the 'ontological kinds' recognized by Jackendoff (1983: 48ff.); other ontological kinds include 'direction' (or path), 'action', 'event', 'manner', and 'amount'. One piece of evidence cited by Jackendoff for the validity of these distinctions is the use of deictic pro-forms. (13)

a. b. c. d.

This is your coat Here is your coat He went thataway Can you do this?

(thing) (place) (direction) (action)

This in (a) directs the hearer's attention to a thing, here in (b) directs attention to a place, thataway in (c) directs attention to a direction, or path, and so on. Lyons (1977) also remarked on the conceptual distinction between things and places. Things (at least prototypical things - what Lyons calls 'first order entities') comprise 'more or less discrete physical objects', including human beings and animals (p. 442). Places, on the other hand, are regions in space, where entities are located. Things, as Lyons (p. 693) observes, are not places, and places are not things.8 However, as Lyons pointed out, places tend to be 'ontologically ambivalent' (p. 442); as a result, locative expessions have 'a certain syntactic and semantic ambivalence' (p. 477). It is not only that places can be hypostatised, and construed as things, the name of a thing may serve to identify the place where the thing is located. The examples in (8)-(12) already show that the conceptual distinction between a thing and a place may not always be clear-cut. A doorway (umnyango), for example, is not so much a threedimensional thing (or first order entity, in Lyons' terminology), it is more a bounded space, which marks the point of access to another bounded space. Especially in English, the distinction between things and places is apt to be grammatically blurred. Canonically, of course, things are designated by nouns (house, table, bed), places by prepositional phrases (in the house, on the table, under the bed). Yet nouns can often designate what conceptually is a place. Consider the following contrast (cf. Lyons 1977: 475).

294

(14)

John R. Taylor

a. London is huge b. London is cold

In (a), London refers to the city as a spatially extended thing; is huge predicates a property of this thing. Although syntactically exactly parallel to (14a), (14b) ascribes a property to a place, not to a thing. It is not London qua thing that is cold, it is the place where the thing is located that is cold. Consider also Lyons' comments on the following sentence. (15)

I'll meet you at the car

Lyons (1977: 693) claims that in this sentence, the car is being used 'indirectly', to identify a place: i.e. the space that is occupied by the car. He offers the paraphrase: Ί will meet you at the place where the car is'. 4. Places and motion verbs Jackendoff gives a strikingly similar analysis of nouns functioning as prepositional objects. For sentences of the form I go to the house, The boy runs into the school, The baby crawls out of the room, Jackendoff (1983: 162ff.) proposed the following schematic conceptual-semantic representation. (16)

[Event~([Thing-][Path-([piace~([Thing-])])])]

According to (16), the sentence I go to the house denotes an event, which involves motion of a thing on a path. The latter is a function of a place, which in turn is a function of a thing. The full representation is as follows. (17)

[EventGO([ThingI][PathTO([piaceAT([ThingHOUSE])])])]

Note in particular that according to this representation, to takes as its complement, not a thing, but a place; also, that in English, the placefunction receives no overt expression. In other words, the constituent the house designates not only a thing, but also a place construed as a function of a thing. Lyons (1977: 480) speculated that there might exist languages which syntactically and morphologically distinguish place-denoting expressions from entity-denoting expressions. Arguably, Zulu is just such a language at least, with respect to the linguistic encoding of motion events. For in

Locativised nouns in Zulu

295

Zulu, the place-function receives obligatory marking, namely by the locative morphology. A second difference between English and Zulu concerns the expression of 'path'. In English, 'path' is generally expressed by a prepositional phrase. There are no path-denoting prepositions in Zulu. 'Path', if it is to be expressed in Zulu, must be incorporated into a verb. From these considerations follows the generalization in (18). (18)

Path-incorporating verbs in Zulu take as their complement a locativised noun.

Of the motion verbs in Zulu, only some incorporate a path function. Some of these verbs are given in (19). ya za ngena fika phuma susa wela dlula

'go to (place)' 'come to (place)' 'go into (place)' 'arrive at (place)' 'go away from (place)' 'take away from (place)' 'cross (place)' 'pass by (place)'

These verbs take as their complement, not a bare noun, denoting a thing, but a locativised noun, denoting a place. One does not enter, or leave, a house-gtta-thing, but a house-^Mö-place. (20)

ngena endlini 'enter house-LOC'; *ngena indlu dlula ekhaya/*ikhaya 'pass-by home-LOC' phuma eyunivesithi/*iyunivesithi 'leave University-LOC' fika emzini/*umuzi 'reach village-LOC'

Path-incorporating verbs sometimes contrast with respect to whether they incorporate goal or source. Compare beka 'put on' and thatha 'take from'. (21)

a.

Ubeke isitsha etafuleni He-put-on dish table-LOC 'He put the dish on the table.'

b. Bathathe izihlalo endlini They-took-from chairs house-LOC

296

John R. Taylor

'They took the chairs from the house.' Also thela 'pour into' and chitha 'pour from'. (22)

a. Ngithele amanzi ebhodleleni I-poured-into water bottle-LOC Ί poured water into the bottle.' b. Ngichithe amanzi ebhodleleni I-poured-from water bottle-LOC Ί poured water from the bottle.'

Zulu path-incorporating verbs are typically non-specific with respect to manner-of-motion, cause-of-motion, medium-of-motion, etc. 9 On the other hand, verbs which specify manner-of-motion typically do not incorporate a path function. Some of these verbs are listed in (23). (23)

hamba ndiza shayela gijima bhukuda ginqika gxuma xhuga

'go, walk' 'fly' 'drive' 'run' 'swim' 'roll (intransitive) 'jump' 'limp'

A problem now arises. If gijima 'run', ndiza 'fly', bhukuda 'swim', etc. do not incorporate a path-function, how can one express, in Zulu, the idea of 'running to a place', 'flying from a place', 'swimming past a place'? There are two grammatical resources in Zulu for the expression of these notions. The first involves the use of a verb-chaining construction. Verbchaining constructions make it possible to combine a path-incorporating verb and a manner-of-motion verb. Thus, the notion of 'run to' is broken down into two components: 'run', and 'go to'. There are two verb-chaining constructions in Zulu. One of these - the so-called consecutive construction (Doke 1981: 787ff.) - makes the path-incorporating verb dependent on the manner-of-motion verb. One possible rendering of The boy ran to (the) school is as follows. (24)

Umfana wagijima waya esikoleni boy he-ran he-went-to school-LOC' 'The boy ran and went to the school.'

Locativised nouns in Zulu

297

The main verb (in the past indicative mood) designates the manner of motion, whilst the dependent, path-incorporating verb is in the past subjunctive mood. The following sentences illustrate the same pattern. (25)

Wagijima wazungeza enkundleni10 He-ran he-went-round playing-field-LOC 'He ran round the playing field.'

(26)

Ugxume waphuma emanzini He-jumped he-left water-LOC 'He jumped out of the water.'

(27)

Ibhola liginqike laflka obondweni The-ball it-rolled it-arrived-at wall-LOC 'The ball rolled right up to the wall.'

More than one path-incorporating verb may be conjoined. (28)

Umfana wagijima wasuka ekhaya waya esikoleni boy he-ran he-left home-LOC he-went-to school-LOC 'The boy ran from his home to the school.'

The other verb-chaining construction is the so-called participial construction (Doke 1981: 341f.). Here, the main verb is path-incorporating, whilst the manner-of-motion verb appears in the participial mood. (29)

Umfana waya esikoleni egijima boy he-went school-LOC he-running 'The boy went to school running.'

Other examples of the same pattern: (30

Umfana wagijima ezungeza enkundleni boy he-ran he-going-round playing-field-LOC 'The boy ran round the playing field.'

(31)

Ingane yaphuma endlini inyonyoba child he-came-out-of house-LOC he-tip-toeing 'The child tip-toed out of the house.'

298

John R. Taylor

(32)

Abantwana benyuka entabeni betotoba children they-ascended mountain-LOC they-walking-slowly 'The children walked slowly up the mountain.'

(33)

Abantwana waqonda esikoleni egijima children they-went-straight-to school-LOC they-running 'The children ran straight to school.'

The second device makes use of the applied verbal extension. (34)

Umfana ugijimele esikoleni boy he-ran-APPL school-LOC

The applied extension -el- causes the locativised noun to become a complement, or argument, of the verb. The running takes place with respect to the school. One interpretation of (34), therefore, is that the boy ran to the school. The sentence can, however, have another interpretation. This is that the boy's running takes place at the place of the school, i.e. the boy ran (around) inside the school. The use of the locative without the applied, as in (35): (35)

Umfana ugijme esikoleni boy he-ran school-LOC

is marginally acceptable for some speakers. The meaning is like that just mentioned, i.e. 'run (around) inside the school'. It is on this reading that the sentences in (6) are marginally acceptable. Thus (6a) could mean Ί walk (around) at work'. Note that the use of the applied without the locative, as in (36)

Umfana ugijimele isikolo boy he-ran-APPL school

has a different meaning yet again, namely that the boy ran for the school, e.g. the boy represented the school in a race. It is worth mentioning also that the applied may be used in combination with a verb-chaining construction. (37)

Sishayele senyukela entabeni we-drove we-ascended-APPL mountain-LOC 'We drove up (to the top of) the mountain.'

Locativised nouns in Zulu

299

The use of the applied in locative expressions raises a number of interesting questions, especially concerning the general meaning of the applied. I would like to briefly address some of these in the remaining part of this paper.

5. Locatives and the applied extension The applied extension is a classic valence-increasing device. An intransitive verb (e.g. sebenza 'work') becomes transitive, a transitive verb (e.g. bhala 'write') becomes di-transitive, through the promotion of a peripheral participant (often, but by no means always, a benefactee) to direct object. (38)

a.

Ngiyasebenza I work

b. Ngisebenzela ubaba I work-APPL father

c.

Ngisebenzela imali I work-APPL money

(39)

a.

b.

Umfana ubhala incwadi boy writes letter Umfana ubhalela ubaba incwadi boy writes-APPL father letter

Syntactically, the promoted argument is a direct object. (Thus, the promoted argument may be taken up by an object concord, and the promoted argument becomes the subject if the verb is passivised). Semantically, however, the promoted object may bear many different kinds of semantic relation to the main predication. (38b) could mean that I am employed by my father, or that I work in order to benefit my father (in which case my father need not know of my intention). (38c) could convey that obtaining money is the goal of my working, or, alternatively, that the need for money motivates my working. In view of these various possibilities, it is not surprising that many sentences with the applied are ambiguous. (39b) could mean that the boy wrote the letter on behalf of his father (perhaps because the father was illiterate, or because the father asked the boy to write, or simply because the boy wanted to do the father a favour, without the father necessarily knowing anything about it). Another reading is that the boy

300

John R. Taylor

wrote the letter to the father, i.e. the father is the intended recipient of the letter. These few examples show that in non-spatial contexts the applied can convey abstract notions of goal/purpose, as well as source/cause. When used in a spatial context, the applied can likewise express the spatial notions of goal, and even source. The expression of a locative goal by means of the applied has already been exemplified (34). Here are some more examples. (40)

(41)

Ibhola liginqikele obondweni ball it-rolIed-APPL wall-LOC 'The ball rolled to the wall.' Umfana ugxumele emanzini boy he-jumped-APPL water-LOC

'The boy jumped into the water.' (42)

Ugogo uxhugela esontweni grandmother she-limps-APPL church-LOC 'Grandmother is limping to church.'

With an appropriate verb, the applied can also convey the notion of source. (43)

Amabhasi asukela esontweni buses they-depart-APPL church-LOC 'The buses depart from the church.'

(44)

Badlela esitsheni sami they-eat-APPL plate-LOC my

'They eat from my plate.' In these examples, the locativised noun has been promoted to an argument of the verb. Verbs such as gijima 'run', gxuma 'jump' do not incorporate the notion of goal; likewise, suka 'leave' merely denotes directed motion, it does not incorporate the notion of source. It is therefore not, in general, possible to construe these verbs with a locativised noun (or at least, not with a locativised noun expressing goal or source). The locativised noun has to be 'bound' to the verbal predicate, and this is achieved by the valence-increasing device of the applied. In corroboration of this account, note that applied motion verbs, such as gijimela, gxumela, sukela, absolutely require some kind of complement, be it a simple noun, or a locativised noun. 11

Locativised nouns in Zulu

(45)

a.

b.

301

*Uyagxumela he-jumps-APPL Ugxumela amanzi he-jumps-APPL water

'He jumps for (e.g. to get) water.' c.

Ugxumela emanzini he-jumps-APPL water-LOC 'He jumps into the water.'

Finally, let us consider a further use of the applied in locative contexts. This is the use of the applied to express the place at which an action or event occurs. This use was already mentioned in connection with (34), repeated below as (46). There follow some further examples of this use. (46)

Umfana ugijimele esikoleni boy he-ran-APPL school-LOC

'The boy ran (around) in the school' or 'The boy ran to the school.' (47)

Imoto bayipendela egalaji car they-it-paint-APPL garage-LOC 'They're painting the car in the garage.'

(48)

Inja yami yafela ekhaya dog my it-died-APPL home-LOC 'My dog died at home.'

(49)

Ngibhalela ehhovisi I-write-APPL office-LOC

Ί write at the office' or Ί write to the office.' (50)

Ugogo usengela esibayeni grandmother she-milks-APPL cattle-kraal-LOC 'Grandmother is milking in the cattle-kraal.'

The expression of place does not always require the use of the applied, however. The applied is not needed with the verbs hlala 'sit, stay, live' or sebenza 'work'.

302

(51)

(52)

John R. Taylor

a.

Sihlezi emanzini We-sit water-LOC 'We are sitting in water.' (idiomatically: 'We are in trouble')

b.

Uhlala eGoli he-stays Johannesburg-LOC 'He lives in Johannesburg.'

a.

Ngisebenza eyunivesithi I-work university-LOC Ί work at the University.'

b.

Usebenza eGoli he-works Johannesburg-LOC 'He works in Johannesburg.'

Sometimes we get a contrast between the expression of place with the applied, and expression of place without the applied. Thus, funda 'study' would tend to co-occur with eyunivesithi 'at the University' and e-Amerika 'in America', whilst fundela 'study-APPL' would tend to co-occur with etafuleni 'at the desk', ekamelo lokudlela 'in the dining-room'. (53)

a.

Ufunda e-Amerika he-studies America-LOC 'He's studying in America.'

b.

Ufundela etafuleni he-studies-APPL desk-LOC 'He's studying at the desk.'

Zulu speakers often report that the meaning of the applied in examples such as (46)-(50) and (53b) is 'restricted location'. Taken literally, this explanation makes little sense. Either they're painting the car in the garage, or they're painting it elsewhere; either the dog dies in the house, or it dies outside the house! Especially problematic is the fact that the notion of 'restricted location' is not easily relatable to any of the many other uses of the applied, either in locative or in non-locative expressions. The examples in (51) provide a clue to the explanation. Let us assume that the verb hlala 'sit, live' already incorporates the notion of place. A more accurate gloss of hlala could be 'sit-in (place)', 'live-in (place)'. Consequently, hlala takes, as its (obligatory) complement, a place expression, i.e. a locativised noun. The predicates in (46)-(50), in contrast, are

Locativised nouns in Zulu

303

not inherently locational. Fa must be glossed, quite simply, as 'die', not as 'die-in (place)'. We have already argued that gijima 'run' does not incorporate any notion of goal (or source); we now have to say that the verb does not incorporate any notion of place, either; the verb merely denotes the act, or activity, of running, divorced from any notion of where, to where, or from where, the running takes place. In English, of course, practically any predication can be freely augmented by a statement of the place at which the predication holds. In Zulu, this seems not, in general, to be possible. The place function must be incorporated into the predication, and this is achieved by means of the valence-increasing device of the applied. Once again, the general function of the applied is to promote the circumstantial place expression to a complement of the verb. On this view, native speaker intuitions that the applied can convey 'restricted location' certainly make sense. The applied 'binds' the locative expression to the main predication, making it an obligatory complement of the verb. Place thereby becomes a participant, not just a setting. 12 To put it another way: the predication is now characterised with respect to its location; the event is of a kind such that it occurs at a certain place, and nowhere else.

6. Concluding remarks I have argued that a locativised noun in Zulu denotes a place, construed as a function of thing. Crucially, I claim that locativised nouns do not contain any notion of 'to' or 'from', i.e. they denote a 'pure' place concept, not a place construed as goal, source, or place on a path. The syntactic environments in which locativised nouns occur follow from some further assumptions about the conceptual-semantic structure of Zulu predicates. Only some motion verbs incorporate a path function, the verbs, that is, contain, in their semantic structure, the notion of motion to (place), from (place), past (place), etc. These verbs naturally take a locativised noun as their complement. Similarly, only some predicates incorporate a place function. Again, these naturally take a locative as their complement. Otherwise, spatial notions require the use of the applied extension. The function of the applied extension is, precisely, to augment the semantic structure of the verb, by incorporating an additional argument in the verb's semantics, be it 'goal', 'source', or simply 'location'.

304

John R. Taylor

Notes 1.

2.

3.

4.

5. 6. 7. 8.

9. 10.

11.

12.

The term 'locativised noun' is from Louwrens (1992). The term nicely captures the derived status of the items in question, while at the same time leaving open the question of their categorial status. Doke (1981: 232) refers to the forms in question as 'locative adverbs'. Diachronically, these expressions derive from nouns of the locative noun classes, fully productive in some Bantu languages, but no longer so in Zulu. Thus phezulu derives from pha-(prefix of noun class 16) + izulu 'sky' (Doke 1981: 62). Kwa- is a reflex of a class 17 concord. Arguably, the forms in question have undergone substantial reanalysis, being now construed as prepositional phrases, not as nouns. In order to avoid visual clutter, I have glossed Zulu expressions only roughly, giving the semantic value only of those morphemes directly relevant to the discussion. Subject concords are glossed by resumptive pronouns. Analogous remarks have been made on locativised nouns in related Southern Bantu languages. Cole (1955: 341) says of locativised nouns in Tswana that they indicate, 'generally and non-specifically, the place or locality in connection with which some action is carried out. Thus, the ... locative nokeng, from noka ('river'), may convey the idea of to, from, into, out of, in, on, by, or at the river, the specific significance being determined by the verb with which it is used.' The examples in (6) were supplied by Pumla Mdonstwa, a teacher of Zulu at the University of the Witwatersrand, Johannesburg. The examples are typical of the errors made by English-speaking learners of Zulu. These paragraphs develop some observations of Louwrens and Posthumus (personal communication). Recently borrowed nouns are also apt to take defective locative morphology, or to form their locatives with ku-: iyunivesithi > eyunivesithi 'university-LOC', ummese (from Afrikaans mes) > kumese 'knife-LOC'. Note, however, that for Langacker, a place, or region in space, does count as a kind of thing (Langacker 1993: 16). For Langacker, the 'ambivalence' of place-concepts would lie in the fact that a place can be construed, not only as a nominal entity, but also as a relational entity. For these notions, see Talmy (1975, 1984). For their application to Tswana, see Schaefer (1985). Many speakers report a contrast between a locativised and non-locativised complement of zungeza, i.e. zungeza enkundleni (where the complement is a locativised noun) and zungeza inkundla (where the complement is a bare, non-locativised noun). The former means 'go round the inside of the playing-field', the latter 'go round the outside of the playing-field'. The contrast is consistent with the account of locativised nouns pursued here. In zungeza enkundleni, the circular motion denoted by zungeza is in the place occupied by the field, whilst in zungeza inkundla it is the thing itself that is encircled. A question of some interest is the categorial status of locativised nouns, that is to say, the question whether locativised nouns are nominal or relational in character. A full discussion of this issue is beyond the scope of this paper. Nothing that I have to say in this paper hinges on the resolution of this question. For the notions of participant and setting, see Langacker (1991: 284).

Locativised nouns in Zulu

305

References Cole, Desmond T. 1955 An Introduction to Tswana Grammar. Cape Town: Longman. Doke, Clement M. 19816 Textbook of Zulu Grammar. Cape Town: Longman. Jackendoff, Ray 1983 Semantics and Cognition. Cambridge, Mass: ΜΓΓ Press. Langacker, Ronald W. 1991 Foundations of Cognitive Grammar. Vol 2. Stanford: Stanford University Press. 1993 Reference-point constructions. Cognitive Linguistics A·. 1-38. Louwrens, Louis J. 1992 The conceptualization of spatial relationships as expressed by locative structures in Northern Sotho. South African Journal of African Languages 12: 107-111. Lyons, John 1977 Semantics. 2 Vols. Cambridge: Cambridge University Press. Schaefer, Ronald 1985 Motion in Tswana and its characteristic lexicalisation. Studies in African Linguistics 16: 57-87. Suzman, Susan To appear. Spatial terms in Zulu. Talmy, Leonard 1975 Semantics and syntax of motion. In: Kimball, John P. (ed.). Syntax and Semantics 4. New York: Academic Press, 181-238. 1984 Lexicalization patterns: Semantic structure in lexical forms. In: Shopen, Timothy (ed.). Language Typology and Syntactic Description, Vol III. New York: Cambridge University Press, 57-149. Ziervogel, Dirk 1971 The Bantu locative. African Studies 30: 371-384.

Distinguishing the notion 'place' in an Oceanic language Deborah Hill

0. Introduction It has been widely noted by Oceanic linguists that many Oceanic languages have a small set of nouns, referring to a "spatial reference point in the lives of the speakers (e.g. home, village, the sea)" (Durie 1988: 11). There are a number of reasons why it is worth paying attention to something that speakers of Oceanic languages clearly attend to: the notion of PLACE as represented by these inherently locational nouns. The first reason is that this notion of PLACE is reflected in conventionalized landmark-based directional systems, which are found in some Oceanic languages. The second is that an analogy can be drawn between the way the grammar distinguishes place nouns from common nouns and the way the grammar distinguishes inalienable possessive constructions from alienable possessive constructions. This paper argues that the underlying semantics of both grammatical distinctions have much in common because they express a distinction between what constitutes the essential part of one's life and what does not. In tying these points together, it is further argued that a landmark-based directional system can mirror what is going on elsewhere in the language, just as it reflects the environmental and cultural situation in which such a directional system is used. Conventionalized landmark-based directional systems appear to be, if not rare, then relatively uncommon amongst the world's languages. When in his 1983 paper, Cecil Brown asked the question "Where do cardinal directions come from?", he concluded that there were four sources for cardinal directions, of which environmental landmarks formed the smallest group (12 languages out of his sample of 127) (Brown 1983).1 The directional system discussed in this paper is one based on the landmarks SEA and LAND. Landmark-based directional systems are comparable to cardinal direction systems in that the reference points are fixed. They differ from cardinal direction systems, however, in that, at least along one axis, there are identifiable endpoints. In the case to be discussed in this paper, these are

308

Deborah Hill

the sea and the inland boundary of the language region. As Levinson has pointed out: "Spatial conceptualization is ... an interesting domain through which to explore issues of cultural relativity, as it is clearly highly constrained by the nature of the physical world as well as the nature of human psycho-biology" (Levinson 1991: 3). A landmark-based directional system provides a good example of spatial language that is constrained by the nature of the surrounding physical environment. The data for this paper comes from the Oceanic language Longgu which is spoken on the island of Guadalcanal in the Solomon Islands (Hill 1992). Longgu is spoken by about 1500 speakers in a coastal region that consists of a string of villages surrounded by the sea on one side and bush area on the other three sides. There is also a small collection of villages several miles inland. Although I am drawing on data from one language, the basic point that PLACE nouns are distinguished from other nouns holds for many Oceanic languages and has for example been noted in grammars of Lenakel (Lynch) and Paamese (Crowley), which are spoken in Vanuatu, in Fijian (Shütz), and in grammars of Rennel and Bellona (Elbert), and Kwaio (Keesing), which are spoken in the Solomons. Referring to the Fijian nouns vale 'home', wai 'sea', vanua 'land', Shütz, for example, notes that "the words seem to have taken on the properties of such locative nouns as cake 'up', ra 'down', and ruku 'under'" (Shütz 1985: 328). The directional systems that one finds in Oceanic languages are, however, variable and the one I will describe here should not be construed as entirely typical of the region (although it is by no means unique). The paper is organized in the following way. The first section describes how place nouns are distinguished from common nouns in Longgu. This is followed by a discussion of the use of some place nouns to describe both location and orientation. In the third section the directional system is described and its connection to the existence of a set of place nouns is discussed. The fourth section deals with the connection between place nouns and possession in Longgu.

1. The notion 'place' in Longgu In a recent paper John Lucy has cautioned those working on spatial language against the danger of imposing an external framework on a language and argued that researchers "often fail to address the most basic issue, namely, how the class of spatial lexemes itself is to be distinguished from other classes in formal terms such that it constitutes a valid object of linguistic comparison" (Lucy 1994: 625). In other words, how do we compare what we perceive to be "spatial language" from one language to another,

Distinguishing

the notion 'place' in an Oceanic language

309

and how, for any given language, do we isolate "spatial lexemes" from other lexemes? Moreover, do the categories that we bring from our own language (for example, the notions of location and direction) exist in other languages? Place nouns in Longgu fall into a class of nouns that includes both directional terms and terms used to express spatial relationships. They belong to the core set of spatial lexemes in the language, as distinguished on language internal grounds. In Longgu it is possible to isolate place nouns from other nouns on the basis of how they combine with simple prepositions. As is typical of Oceanic languages, there are very few simple prepositions in Longgu. These are a general locative i 'at', an allative preposition vu 'to, towards', and a preposition mi 'until'. The preposition mi 'until' appears to be derived historically from the conjunction ma 'and' and the general locative i. In addition there is a nominal preposition tana 'be at' that consists of a bound morpheme ta- and a possessive suffix -na. This preposition is not a simple preposition. The basic disctinction between place nouns and common nouns is that place nouns function as the object of simple prepositions whereas common nouns do not. Table 1 shows simple prepositions and the nominal preposition. If a noun is not inherently locative (i.e. is not a place noun) then the prepositional object consists of the nominal preposition tana 'be at' and the object noun whereas if the prepositional object is inherently locative (i.e. is a place noun) then the prepositional phrase consists of the simple preposition and the object noun. Table 1. Simple and nominal prepositions in Longgu

Location

Extent

Direction

Inherently locative

i 'at'

mi 'until'

vu 'to, towards'

Not inherently locative

ta-(na) 'be at'

mi ta-(na) 'until at'

vu ta-na 'to at'

Examples of nouns functioning as the object of the allative preposition vu 'to, towards' are given in (l)-(4). From these examples it can be seen that komu 'village' (example (1)) and asi 'sea' (example (2)) function in the same way as local nouns (such as ubu 'inside', vavo 'on, above' and buri 'behind') (example (3)) and as directional terms such as vu'a 'down' (example (4)).

310

Deborah Hill

(1)

mwane e man 3SG 'the man

la go went

(2)

e la vu 3SG go to 'he/she went

(3)

ririu turn 'turn

(4)

m -e CONJ-3SG 'and then

vu buri. to back, to the back, la go she

vu komu to village to (his) village'

asi sea towards the sea' ririu vu na 'o. turn to front, turn to the front,

ririu vu gege turn to side turn to the side'

hou ta 'asi-a na vu vu 'a thither throw-3SG PERF to down threw it down (to the bottom)'

Compare example (1) with examples (5) and (6) in which the motion is towards a location that is not inherently locative. In (5) and (6) the object of the allative preposition is a nominal prepositional phrase headed by tana 'be at'. In (5) the goal is a person, Peter, and in (6) the goal is a village (or villages) but not the village belonging to the referent of the subject of the clause. In other words, it is not their 'home' village. (5)

la vu ta-na go to LOC-3SG 'go to Peter'

Peter NAME

(6)

m -ara CONJ-3PL 'and they

vu ta-na tabalu to LOC-3SG some to some (other)

la go went

komu-gi-na village-PL-DEIC villages'

Using the ability to function as the object of a simple preposition as the criterion, we find that lexemes such as malaba 'garden', komu 'village', luma 'house' and asi 'sea' fall together with lexemes that refer to the names of towns and villages, and with lexemes that refer to directions (e.g. langi 'up' and vu'a 'down', ala'a 'east' and toll 'west'), as well as with local nouns. Local nouns such as ubu 'in' and vavo 'on' function as the head of possessive constructions that express spatial relations such as 'on the table', and 'in the house'. Local nouns and possessive constructions are discussed in greater detail below. Place nouns in Longgu are regularly polysemous (cf. Apresjan 1974). As well as functioning like directional terms, they may also function in the

Distinguishing the notion 'place' in an Oceanic language

311

same way as non-spatial lexemes. For example, the noun mala'u'u 'hill' can be quantified (e.g. rua mala'u'ugi 'two hills'), or modified by a deictic (e.g. mala'u'u nihou 'the hill over there'), just as common nouns such as iola 'canoe' can be (e.g. rua iolagi 'two canoes', iola nihou 'the canoe over there'). One of the senses of malaba 'garden' and komu 'village' then is the entity garden, or the entity village and the noun can refer to any garden or village. The other sense is that of the home garden or village and the noun then refers to a particular location, just as the name of a town refers to a particular location. Longgu codifies a distinction between 'place' as entity and 'place' as spatial relation. This point requires further discussion. Lyons (1977: 693) has argued that places are not entities. Wilkins, however, has argued that the failure to distinguish the notion of place as entity from place as spatial relation has been a fundamental error within semantic theory (n.d: 1-2). (Wilkins n.d: 1): A fundamental error surrounding the concept of place within semantic theory has been the failure of semanticists to recognize that the English word 'place' has a meaning range which covers two distinct semantic notions. One notion is that of 'place as entity'and the other is that of 'place as one of the two arguments of a locational/spatial predication'. Places in the second sense are generated into existence whenever such spatial predications (as represented by adpositions, spatial cases, or other morphological forms) apply. Such places are created out of any type of real-world entity and they exist as places as long as the predication holds true for them. Places in the first sense are a sub-type of entity and exist, like other entities, regardless of such predications.

Like Mparntwe Arrernte (an Australian language) which Wilkins discusses, Longgu distinguishes these two notions of place that are collapsed in English. Examples (1) and (6) illustrated that Longgu distinguishes what is a 'home place' from what is not a 'home place' at the syntactic level. A distinction between what is 'home' and what is not 'home' is of course also found in English, (for example in the difference between saying 'I'm going home', but not 'I'm going house') but in English the distinction is simply made between two separate lexemes. In Longgu, however, it can be said that each of these inherently locational nouns refers to a place that is home and that home includes the entire village and surrounding area in which people work and live and not just the house in which they live. 2 In Longgu, and in other Oceanic languages that distinguish inherently locational nouns from common nouns, these terms refer to both the entity

312

Deborah Hill

mala'u'u 'hill' and the location that is the familiar 'home hill', or the 'village' and the location that is the 'home village'. Table 2 provides a list of place nouns with appropriate glosses for the two possible senses. It should be stressed that the distinction between the two senses of these nouns is codified in the language and is not just polysemy at the level of the lexical unit. The following discussion presents two ways in which this coding is done. Table 2. Regular polysemy of place nouns

malaba

'garden'

'the garden I/we work in'

masu 'u

'bush'

'the bush on the outskirts of our village'

komu

'village'

'my/our village'

luma

'house'

'my/our house'

mala 'u 'u

'hill'

'the hill behind our village'

asi

'sea'

'the sea/beach at our village'

rara

'shore'

'the shore at our beach'

wai

'river'

'the river in our village'

'seawards'

First, note that nouns that are directions or names of villages or towns cannot function as an object of the nominal preposition tana 'be at' and that common nouns cannot function as objects of simple prepositions. It is only those nouns that I am calling place nouns that can function as objects of either type of preposition - the simple preposition and the nominal preposition. To reiterate the point, the meaning of inherent location or 'home place' only applies when a noun functions as the object of a simple preposition. This is exemplified in the following ungrammatical example. In (7) the name of the town Honiara cannot function as object of the nominal preposition tana 'at'. In (8) the common noun iola 'canoe' can not function as object of the simple preposition i 'at' because it is not a place noun (i.e. iola 'canoe' is an example of a noun that can function as 'place as spatial relation' but not as 'place as entity').

Distinguishing the notion 'place' in an Oceanic language

(7)

*nu

ii 'o

ta-na

Honiara

1SG

live

L0C-3SG

PLACE

313

Ί live in Honiara' (8)

*pitu

i

iola

wait LOC canoe 'wait at the canoe'

Table 3 lists nouns that function as either place or common nouns. Note that most nouns function as either place nouns or common nouns and that relatively few function as both. In other words, most nouns can function as either 'places as entities' (place nouns) or 'places as spatial relation' (common nouns) and a few can function as both. Table 3. Nouns which function as common nouns and place nouns

Place nouns

Common nouns

luma

'home'

luma

'house'

komu

'home village'

komu

'village'

iola

'canoe'

Honiara vu 'a

'down'

one'one

'sand'

langi

'up'

'ai

'tree'

ala'a

'eastwards'

mwane

'man'

toli

'westwards'

asi

'sea, seawards'

longa

'inland, towards the bush'

The second way in which place nouns are codified in the grammar is that place nouns are not marked for number and typically do not form the head of a possessive construction. Compare examples (9) and (10). In (9) komu

314

Deborah Hill

'village' does not form the head of a possessive construction and it is understood that the village is the village belonging to the referent of the subject of the clause. It would be odd to say the equivalent of 'I'm going to my village'. Rather, it is the construction (simple preposition plus unmarked place noun) that conveys that the village is the one that I belong to. In (10) komu 'village' is head of a possessive construction. The village referred to is not the home village of the referents of the subject of the clause. It is important to note, however, that in (10) it is the presence of the nominal preposition tana 'be at' and not the possessive construction that signals that komu 'village' is functioning as a common noun. (9)

m-e CON-3SG

'and she (10)

la mai vu komu go hither to village come to (her/his) village (he/she came home)'

ara lae lae mi ta-na komu ngaia Titipohona-i-na 3PL go go until at-3SG village 3SG N A M E -SG-DEIC 'they go on and on until at Titipohona's village'

Having established in this section that Longgu explicitly codifies a distinction between place nouns and common nouns, the use of place nouns as terms to describe orientation (i.e. information about the way something is facing or the direction something is going) is discussed.

2. Location and orientation It has already been mentioned that Longgu is an example of an Oceanic language which has a small group of nouns that refer to culturally salient places. It has been argued here that these nouns are regularly polysemous and that one of the senses of these nouns is that of HOME PLACE. Having established that certain locations, perceived as important in the culture, are distinguished grammatically in the language, the next step is to ask how these place nouns become used to describe orientation. To answer this question let me turn briefly to some literature on grammatical ization. In two studies on grammaticalization (Svorou 1986 and Bowden 1992) environmental landmarks turn up as a possible source of locative terms (such as IN, O N , U N D E R ) . Bowden's work deals specifically with Oceanic languages. One of his findings was that the landmarks SEA and L A N D are the source of locatives. Longgu is not one of the languages in which this happens. However, his results are further evidence that place nouns generally, and landmark terms specifically, have a special role in

Distinguishing the notion 'place' in an Oceanic language

315

Oceanic languages. 3 Svorou (1986, 1994) also deals with the development of terms for spatial relations rather than absolute notions. She specifically draws a distinction between landmarks that develop into locatives (or what she calls spatial grams) and landmarks that develop into a directional system. However, her model of the development from noun to spatial gram has some relevance to the relationship between noun (or location) and direction. In her model of the grammatical chain from noun to grammatical affix there is an intermediate stage of adverbial use. Without wanting to make a commitment to the view that what is going on in the relation between location and direction is a grammatical process, Svorou's model is relevant, as place nouns in Longgu are also used adverbially. When a place noun functions as an adverb modifying a verb of motion, it acquires a strong sense of direction. If we look at the Longgu data we find that directional terms and place nouns have both a nominal and an adverbial function (although taking the data as a whole - i.e. including names of towns etc., they are more appropriately described as unmarked complements rather than adverbs, but for the moment they can be described as adverbs) (see example (4) for a nominal function of the directional word vu'a 'down'). Examples of place nouns functioning as adverbs or unmarked complements are found in examples (11)-(14). This seems to follow the pattern of use of locatives in languages with case systems claimed by Tiersma (1982: 843): In languages with case systems, nouns referring to places are locally unmarked in the locative (or another local) case ...

In other words, in a language like Longgu which uses prepositions rather than cases to mark locatives, it is precisely those words that are inherently locational that do not have to be marked with a locative preposition. Nouns such as luma 'house', komu 'village', and asi 'sea' function as unmarked complements because they are inherently locational, just as the names of towns (e.g. Honiara) and directions (e.g. vu'a 'down') are inherently locational. (11)

tabalu pitu vu'a some wait down 'some waited below'

(12)

ara ii'o Honiara 3PL live PLACE 'they live in Honiara'

316

Deborah Hill

(13)

ara lau masu'u 3PL go to bush 'they went to the bush'

(14)

e

la mai Baulo go hither PLACE 'he came here (from) Baulo'

3SG

Note the verb lau 'go to' in example (13). This verb appears to be a contraction of the verb la 'go, travel' and the allative preposition vu 'to, towards'. It is only used with place nouns and can never be followed by the nominal preposition tana 'be at'. This suggests that an unmarked complement must refer to an inherent location (i.e. it must refer to 'place as entity' and not to 'place as spatial relation'). It seems plausible to suggest that it is the fact that place nouns, like directional terms such as UP and DOWN, can function as unmarked complements that has provided the right linguistic circumstances for a landmark to be used as a direction. By no means all Oceanic languages use a landmark-based directional system, so I am not suggesting a causal link. Rather I am suggesting that the conditions are right for such a directional system to develop. Interestingly, if we look at another common directional system found in Oceanic languages (a system based on terms for UP and DOWN), we find that these terms fall into the same linguistically identifiable category as place nouns in Longgu. Kwaio, for example, a Solomon Islands' language closely related to Longgu, also distinguishes place nouns and has a directional system based on terms for UP and DOWN (Keesing 1985). It has been suggested in this section that there is a correlation between the coding of place nouns and the use of a conventionalized landmark-based directional system. In the following section I want to argue that such a landmark-based directional system can mirror this distinction between place nouns and common nouns. Moreover, language use suggests that the relationship between location and direction is not unidirectional. That is, while terms for directions may have developed from terms for locations, directions can become associated with particular places through conversational implicature.

3. The directional system Let me first say something more about the sort of directional system I am referring to. In a number of papers Levinson (e.g. Levinson 1991, 1992) has described the difference between an absolute and a relative mode of

Distinguishing the notion 'place' in an Oceanic language

317

spatial description. A relative system of spatial description is one that relies on "relational concepts like 'in front of', 'behind', 'to the left of', etc. to describe the relative dispositions of objects from a particular point of view" (Levinson 1992: 2) and is one that uses terms such as left and right. An absolute system, by contrast, relies on fixed angles such as North, South, East, and West for spatial description. Some absolute systems, found for example in Pacific Islands, use fixed angles that relate to specific landmarks, in particular to SEA and LAND. While the system described here can be called a directional system, strictly speaking it is an orientational system. It subsumes both direction (involving motion) and facing information (involving static location). The directional system in Longgu consists of four terms that cover two axes. The terms along one of the axes are terms referring to SEA and LAND. Directions along the other axis are referred to by terms that can be glossed roughly as EAST and WEST and, according to Longgu speakers, are derived from the rising and setting of the sun (although comparative linguistic evidence suggests that historically, they could have referred to Up and Down [cf. Keesing 1985]). The directional terms in Longgu are: asi longa ala'a toli

'sea' 'inland' 'east' 'west'

In contrast to what we may suppose about directional systems on the basis of the use of North, South, East, West in English, the two axes (asilonga 'sea-inland' and ala'a-toli 'east-west') are not equivalent. They each have a different status (see Ozanne-Rivierre 1992 for similar examples in New Caledonia) which, it can be argued, comes from the derivation of the terms. The SEA-LAND axis is restricted in its use to directions within the land boundary of the Longgu speaking area while the EAST-WEST axis is used to refer to directions and locations well beyond this area. That is, the directional terms referring to geographical reference points that are intimately connected with the people's lives (sea-inland) function differently from the reference points (sunrise(east)-sunset(west)) that are seen as beyond the scope of the region. Thus while the term asi 'sea' is used to refer to directions only to the point where the land meets the sea and the term longa 'inland' refers to directions only within the area of land that belongs to Longgu speaking people, toli 'west' can refer, for example, to Gizo in the Western Province of the Solomon Islands, and ala'a 'east' can refer to the island of Makira and beyond. This distinction between the two axes nicely reflects the distinction between place nouns and common nouns. In

318

Deborah Hill

both cases, the places or directions which are most intimately connected with the lives of Longgu-speaking people have a different status from the places or directions which are more remote. In the case of place nouns this is coded linguistically; in the case of the sea-land (asi-longa) axis it is distinguished by language use. In addition, the Longgu directional system can be used to refer to smallscale space as well as large-scale space. For example, where in English one might ask someone to move a little to the left or right, in Longgu one asks someone to move a little seaward, or to move a little west. Small objects on a table are also most typically described using the directional system. 4

3.1. Direction or local landmark? This brings us to the question of whether we are dealing with directions or local landmarks. Only one of the directional terms (asi 'sea' 5 ) refers to both a local landmark (the sea) as well as a direction. The term longa 'inland' is a direction and not a place. This seems clear as there is another term used to refer to untended areas of land (masu'u 'bush') and one to refer to gardens (malaba 'garden'). The existence of the inland villages in a region where most people live along the coast allows us to question the extent to which the directional term asi 'sea, seaward' is truly a directional term rather than a local landmark. The fact that people in the inland region of Nangali use the term asi 'sea, seaward' suggests that it is part of the system of directions and not simply a local landmark. If it were a landmark only (used by coastal dwellers), we would expect that in the inland region other local landmarks (e.g. the prominent river in the area) would be used instead of asi 'sea'. This is not the case. People in Nangali use asi 'seaward' as a directional term as much as people at the coast and it is clear that it is a full member of the four-term directional system. In the coastal villages the direction asi 'sea, seaward' invariably corresponds to the local landmark (the sea). Indeed when standing in a coastal village facing the sea, the direction asi 'sea' can only be used to refer to the place more or less in front of the speaker. Even though the length of the beach in front of the village may only be 50 metres long, each end of the beach can be referred to as ala'a 'east' or toli 'west' (assuming that the speaker is roughly in the centre of the village). The further from the beach one is, the wider the area that can be referred to as asi 'sea, seaward'. Syntactically it is possible to see that asi 'sea, seaward' is a direction because it functions as a locative complement and that it is also a location because it functions as the object of the nominal preposition tana 'be at'.

Distinguishing the notion 'place' in an Oceanic language

(15)

lae asi go sea/seawards 'go seawards'

(16)

la vu ta-na go towards LOC-3SG 'go to the sea'

319

asi sea

None of the other directional terms (longa 'inland', ala'a 'east' and toli 'west') can function as objects of the nominal preposition tana 'be at'. They can all function as locative complements (suggesting that they are directions but not locations). lae toli

'go west'

lae ala'a

'go east'

lae longa

'go inland'

3.2. Conversational implicature associated with directions Although longa 'inland', ala'a 'east' and toli 'west' are directions and not locations, we find that specific locations have become associated with the direction longa 'inland' through conventional use. This can best be explained by describing the sorts of answers, and the implications of the answers, associated with the question 'Where are you going?' (or rather its Longgu equivalent [la vu ievel\). At this point it is worth noting that 'Where are you going?' is the most common form of greeting in Longgu when encountering someone (as it is in many other languages).6 It is also important to remember the geographical environment and settlement pattern of the area. The villages are located along a narrow coastal strip stretching (very roughly) along the (north)west-(south)east axis (toli-ala'a). When travelling along the toli-ala'a axis (i.e. from one coastal village to another, or from one of the villages to Honiara), people responding to enquiries about where they are going usually use directional terms (e.g. ala'a 'east' and toli 'west') rather than mentioning the name of the village to which they are going. Even when going to Honiara (the capital and only large town on the island), speakers usually say they are going vu toli 'towards west' and do not mention the name Honiara. Similarly, people from the village of Bwabwasu heading to Marau (the eastern tip of Guadalcanal, where they go to market once a month) will say they are

320

Deborah Hill

going vu ala'a 'towards east' rather than say they Significantly, the directional terms do not refer only any place in these directions. By contrast, particular places are conversationally Longgu 'I'm going to the sea' or 'I'm going inland'. land villages, then when someone says: (17)

are going to Marau. to these places but to implied by saying in If we exclude the in-

na ho la vu asi 1SG IRR go to sea Ί am going to the sea/seawards.'

there is a conversational implicature that that person is going to the seashore. The fact that there is no conversational implicature of this kind for the inhabitants of the inland region is clear because they must specify that they are going towards and will be at the sea. (18)

na ho la vu 1SG IRR go to Ί am going to the sea.'

ta-na LOC-3SG

asi sea

If it were the case then, that the Longgu region consisted only of the coastal villages and did not include inland villages, to say that one was going toward the sea would always imply that they were going to a particular place. As it is, this conversational implicature holds for coastal dwellers but not for inland dwellers. The other direction along the same axis is longa 'inland'. In this case there is a conversational implicature that, when a speaker says: (19)

na ho la vu longa 1SG IRR go to inland Ί am going inland.'

then he or she is going to their garden (which is located inland of each village). The motivation for having one axis along which directional terms can be used to avoid being explicit, and another along which the use of directional terms imply a particular place, is clearly the land formation and settlement pattern of the region. In all villages people's lives revolve around going to the garden to collect food every day and, as the gardens are located at the inland boundary of the language area, it is to be expected that the places (i.e. the gardens) have become associated with the direction.

Distinguishing the notion 'place' in an Oceanic language

321

The directional system in Longgu can be seen as a reflection of the distinction between what is 'home' (the landmark-based axis) and what is 'not home' (the east-west axis). The landmark-based axis is used only within the bounds of the language area and the lives of the people are basically carried out between each point of the axis - in what, in its broad sense, is their home of village, garden and sea. The conversational implicature that associates the direction longa 'inland, bushwards' with a particular place 'the garden', suggests that location and direction are so inextricably intertwined that not only do locations form the basis of directions (as in the example of asi 'sea'), but the use of directions can become conventionalized in such a way that the direction is equated with a location. Having argued that the directional system in Longgu mirrors the distinction between places that are 'home' and places that are 'not home', let me now turn to another area of the language where a distinction between what is, in a sense, 'part of me', is distinguished from what is 'not part of me'.

4. Location and possession The link between location and possession has often been made, at least implicitly, in the many languages that express a location by using 'possessed' locative nouns. Languages in Africa, Mesoamerica, and Oceanic come to mind. The link has also been explicitly stated in literature on part-whole relationships (Iris, Litowitz and Evens (1988: 284): All part-whole schemata are derived originally from physical knowledge of the world. For this reason, discreteness, formedness, attachment, spatial inclusion, and questions of alienable vs. inalienable possession are all implicated.

In addition it has been argued that semantic differences between inalienable and alienable possession reflects cultural differences (Mühlhäusler and Harre 1990: 212): Whilst the distinction between alienable and inalienable is found in many languages, the membership of the noun classes that can be alienably and inalienably possessed respectively differs from language to language, sometimes even among closely related languages or subsystems. This means that it is largely a cultural matter, a reflection of certain social practices and not, as one might suspect, any biological necessity.

322

Deborah Hill

In this section I want to argue that the culture specific differences reflected in possessive constructions mirror, at least in some way, the differences between the two axes of the directional system and the differences between place nouns and common nouns. Before doing this however, a caveat is required. It seems to be generally accepted that the terms "alienable" and "inalienable" are terms that have conflated semantic and formal differences. One does not expect to find complete overlap between the semantics of the relationship between a noun referent and its possessor throughout a language. Indeed, in Longgu, certain kinship terms form the head of alienable possessive constructions rather than inalienable possessive constructions (e.g. tia ngaia "his/her mother" - an alienable possessive construction, but burunga-na "his/her spouse" - an inalienable possessive construction). The history of the study of possessive constructions in Oceanic languages has been one where the pendulum has swung between analysing possessive constructions as if they were noun class systems with no semantic basis to analysing them as almost entirely semantically based systems. More recently the accepted view is one which recognises that possessive systems in Oceanic languages involve both types of analysis. That is, there is a degree of arbitrariness (the noun class approach) as well as a semantic basis for at least some nouns (see Pawley and Sayaba 1990). In Longgu only a relatively small number of nouns can form the head of either an inalienable possessive construction (indicated by a possessive suffix directly attached to the head noun) or an alienable possessive construction (indicated by a postposed independent pronoun). For these nouns the semantic basis for the distinction between the two types of possessive construction is very clear. It is thus only this set of nouns which allows us to draw some comparisons between possessive constructions and the way in which locatives are distinguished in the language. The following Table illustrates the sorts of distinctions that are made. The notion of part-whole relation is at the centre of the inalienable-alienable distinction. Thus for nouns which can function as head of either an inalienable or alienable possessive construction, the meaning conveyed by the inalienable possessive construction is that there is an intimate, inseparable connection between the possessor and the possessum. While when the noun functions as head of the alienable possessive construction there is no suggestion of inseparable connection between the possessor and the possessum. This distinction is particularly clear with nouns which refer to body parts but for kinship terms (most of which must function as head of either an inalienable or alienable possessive construction but not both) it does not seem to hold.

Distinguishing

the notion 'place' in an Oceanic language

323

Table 4. Inalienable and alienable possessive constructions

INALIENABLE ALIENABLE

GLOSS

kinship terms

gale-na

gale ngaia

'his/her offspring'

body part terms

kakasa-na

kakasa ngaia

'his/her rib'

mimi-na

mimi ngaia

'his/her bladder'

u 'unu-na

u 'unu ngaia

'his/her story'

tala-na

tala ngaia

'his/her path'

zalu-na

zalu ngaia

'her/its egg'

totohale-na

totohale

'his/her picture'

luma-na

luma ngaia

'his/her house'

ivi-na

ivi ngaia

'his/her clothes'

other

ngaia

The term gale 'child, offspring' forms the head of an inalienable possessive construction when the possessor is human, and forms the head of an alienable possessive construction when referring to the offspring of an animal. Nouns referring to body parts form the head of an inalienable possessive construction when the body part is intact but the head of an alienable possessive construction when the body part is dismembered (e.g. the pig's leg after the pig has been butchered). The inalienable possessive constructions headed by u'unu 'story', tala 'path', zalu 'egg', and totohale 'picture' express that the possessor is part of the possessum. For example, u'unu-na expresses that the story is about him or her whereas u'unu ngaia refers to the story told by him or her. Similarly totohale-na 'his/her picture' refers to the picture of him or her, while totohale ngaia 'his/her picture' is the picture that he or she drew, or took, or owns. An inalienable possessive construction headed by tala 'path' expresses the notion 'his/her path that he has travelled, the way he/she went'. An alienable possessive construction headed by tala 'path' refers to the road or path that belongs to him (or more usually to a village). The inalienable possessive construction zalu-na

324

Deborah Hill

'her/its egg' refers to the egg that she/it (i.e. the chicken) laid, while the alienable possessive construction zalu ngaia refers to the egg that he or she owns or is going to eat. For the nouns luma 'house', iola 'canoe' and ivi 'clothes' the important feature is whether the item is occupied by the possessor (in the case of house and canoe) or worn by the possessor (in the case of clothes). So, for example, luma-na 'his/her house' is the house that he or she habitually sleeps in while luma ngaia 'his/her house' is the house that he or she owns but does not sleep in. The example that is interesting from the point of view of a discussion of PLACE nouns is the example of luma 'house'. It is striking that luma 'house/home', which can function as a place noun, is also one of a comparatively small set of nouns that can form the head of either an inalienable or alienable possessive construction. When it forms the head of an inalienable possessive construction it can only refer to 'home'. The fact that one lexeme turns up in two parts of the grammar making a similar distinction suggests that the grammatical constructions are suggesting something similar. The inalienable possessive construction is used to express relationships that are inherent, that are seen as the most essential part of the person. And the nouns that are called place nouns refer to locations that are essentially bound to the everyday life of a person. Thus in two parts of the grammar, the notion of what is most intimately connected to a person's life is distinguished from what is less central.

5. Conclusion This paper has focussed on a basic and non-controversial point about Oceanic languages. It is simply that when looking at a range of languages from the Oceanic region it has often been noted that a fairly small set of nouns, which can be described as inherently locational, are distinguished from those that are not inherently locational. The purpose of the paper has been to argue that this point suggests that there is a linguistic and cultural importance attached to the notion 'PLACE', or more specifically 'HOME PLACE'. This paper has attempted to link the fact that inherently locational (or place), nouns are distinguished in many Oceanic languages with two other characteristics of Oceanic languages. Firstly, that it is not uncommon to find that the directional system is based on landmarks such as sea and land. Secondly, that Oceanic languages typically distinguish two types of possessive construction. Although the possessive constructions are not fully based on semantic differences, some nouns may function as head of either type of possessive construction with a consequent semantic difference.

Distinguishing the notion 'place' in an Oceanic language

325

It has been suggested that the significance of HOME PLACE is reflected in these two areas of the language. Firstly, and most obviously, it is reflected in the conventionalized landmark-based directional system. The landmarkbased directional system used in Longgu mirrors the distinction between what is home, in a broad sense of the English term, and what is not home. Place nouns function as the basis for directional terms, and one axis of the directional system is used only within the bounds of what is home territory. For a language that uses both a conventionalized landmark-based directional system AND distinguishes a set of culturally salient place nouns such as home, village and the sea, the link between location and direction is an extremely obvious one. The data from Longgu suggest that notions of LOCATION and DIRECTION should not always be seen as discrete categories. Terms for locations are used as terms for direction (e.g. the sea and the direction seaward) and, by conversational implicature, directions can be used to refer to locations (the example of the direction longa 'inland' referring to the location 'garden' malaba). Secondly, the distinction between what is home place and what is not home place is reflected in the distinction between inalienable and alienable possession. The semantic basis underlying inalienable and alienable possessive constructions - a distinction between what is closely connected to a person and what is not - is reminiscent of the distinction between what is home and what is not home. In particular the noun luma 'house' can be found both as a place noun expressing 'home' and in an inalienable possessive construction expressing 'home'. In conclusion, this paper has attempted to show that there are recurring patterns or themes in language. The distinction between what is home and what is not home, what is part of me and what is not part of me can be seen as different manifestations of the same theme. In a book appropriately titled 'Space and Place' (Yi-fu Tuan 1977) the author rather poetically describes the difference between the two notions in the following way (Tuan 1977: 3): Place is security, space is freedom: we are attached to the one and long for the other.

While Tuan's description of 'space as freedom' may not be what linguists have in mind when they discuss language and space, the association of the notion 'place' with security comes remarkably close to the mark for the discussion of the notion PLACE in Oceanic languages.

326

Deborah Hill

Notes 1.

2.

3. 4.

5. 6.

A fourth and final strategy for innovating terms for cardinal directions involves reference to environment specific features such as mountains, cliffs, rocky places, rough country, and so on ... clearly, of the four lexical source areas employed in cardinal point term innovation, the environment-specific feature area is least frequently used on a worldwide basis (Brown 1983: 138). 'Home' in English can of course be used to refer to a city or a country as well as to the house in which one lives, but the point i | that these places themselves can not function like the word 'home' (e.g. 'I'm going town' is grammatically unacceptable and the word 'town' can never refer to one's home town by occurring in a certain grammatical construction as the word komu 'village' can in Longgu.). Not surprisingly, Bowden's study showed that landmarks as a source of locative terms was less common than body parts as a source. More details of the use of the directional system to refer to small-scale space, and discussion of the directional system in relation to other means of spatial description (i.e. terms for 'back', 'front' etc.) can be found in Hill (1993a). The noun asi also means 'salt, saltwater'. The verbal suffix 'a derives the verb asi'a 'salty'; reduplication and the verbal suffix 'a derive the Stative verb asiasi'a 'to be rough (of sea).' It is clear that there is a level at which responses to expressions of greetings are formulaic and routine but nevertheless, within this context, there does seem to be something interesting to say about the implications when speakers respond with directional terms.

References Apresjan, Ju. D 1974 Regular polysemy. Linguistics, Vol 12: 5-32. Bowden, John 1992 Behind the Preposition: Grammaticalization of Locatives in Oceanic Languages. Canberra: Pacific Linguistics B-107. Brown, Cecil H. 1983 Where do cardinal direction terms come from? Anthropological Linguistics25(2): 121-161. Crowley, Terry 1982 The Paamese Language of Vanuatu. Canberra: Pacific Linguistics B-87. Durie, Mark 1988 Verb serialization and 'verbal-prepositions' in Oceanic languages. Oceanic Linguistics 27(1-2): 1-23. Elbert, Samuel H. 1988 Echo of a Culture: A Grammar of Rennell and Bellona. Oceanic Linguistics Special Publication No. 22., Honolulu: University of Hawaii Press. Hill, Deborah 1992 Longgu Grammar. Unpublished PhD, Canberra, ANU.

Distinguishing the notion 'place' in an Oceanic language

Hill, Deborah 1993a

327

Finding your way in Longgu: Geographical reference in a Solomon Islands language. Working Paper No. 21, Cognitive Anthropology Research Group, Max Planck Insitute for Psycholinguistics, Nijmegen. Spatial configurations and evidential propositions. Working Paper No. 1993b 24, Cognitive Anthropology Research Group, Max Planck Institute for Psycholinguistics, Nijmegen. Iris, Madelyn Anne, Bonnie E. Litowitz and Evens, Martha Walton 1988 Problems of the part-whole relation. In: Evens, Martha Walton (ed.). Relational Models of the Lexicon. Cambridge: Cambridge University Press, 261-288. Jackendoff, Ray 1983 Semantics and Cognition. Cambridge, Massachusetts: MIT Press. 1991 Semantic Structures. Cambridge, Massachusetts: MIT Press. Keesing, Roger 1985 Kwaio Grammar. Canberra: Pacific Linguistics B-88. Levinson, Stephen, C. 1991 Relativity in spatial conception and description. Working paper No. 1, Cognitive Anthropology Research Group, Max Planck Institute for Psycholinguistics, Nijmegen. 1992 Primer for the field investigation of spatial description and conception. Pragmatics 2(i): 5-47. Lichtenberg Frantisek 1985 Possessive constructions in Oceanic languages and in Proto-Oceanic. In: Pawley, Andrew and Lois Carrington (eds). Austronesian Linguistics at the 15th Pacific Science Congress. Canberra: Pacific Linguistics C-88, 93-140. Lucy, John A. 1994 The role of semantic value in lexical comparison: Motion and position roots in Yucatec Maya. Linguistics 32: 623-656. Lynch, John 1978 A Grammar of LenakeI. Canberra: Pacific Linguistics BSS. Lyons, John 1977 Semantics. Volumes 1 and 2. Cambridge: Cambridge University Press. Mühlhäusler, Peter and Romano Harr6 1990 Pronouns and People: The Linguistic Construction of Social and Personal Identity. Oxford: Blackwell. Pawley, Andrew and Timici Sayaba 1990 Possessive marking in Wayan, a Western Fijian language: Noun class or relational system? In: Davidson, J.H.C.S. (ed.). Pacific Island Languages. Honolulu: University of Hawaii Press, 147-171. Shütz, Albert J. 1985 The Fijan Language. Honolulu: University of Hawaii Press. Svorou, Soteria 1986 On the evolutionary paths of locative expressions, BLS 12, Berkeley, California. Svorou, Soteria 1994 The Grammar of Space, Amsterdam: John Benjamins. Tiersma, Peter Meijes 1982 Local and general markedness. Language 58(4): 832-849. Tuan, Yi-Fu 1977 Space and Place. Minneapolis: University of Minnesota Press. Wilkins, David no date Place. MS.

The linguistic, cognitive and cultural variables of the conceptualization of space Barbara Kryk-Kastovsky

1. Introduction Space is a fundamental concept in human cognitive and linguistic systems, since every human being is necessarily aware, consciously or subconsciously, of his/her spatial location. Therefore, it is only natural for language users to approach the notion of space in an egocentric fashion. Consequently, the deictic paradigms operating in natural language do not reflect any abstract spatial relations but are conceptualized and codified in terms of proximity and distance from the speaker's zero point, the Origo. The centrality of spatial cognition has long been recognized by researchers. Bühler was the first to suggest the basic role of the local dimension in his theory of the Zeigefeld der Sprache (Bühler 1934). Lyons (1977) and Fillmore (1975) have shown that all three deictic dimensions can be characterized by the local expressions proximal/distal. From a broader perspective, Traugott (1978) has offered a universalist explanation of how spatial terms are used to express temporal relations, apparently in all languages. Thus, in the sequencing of events some languages use the front/back pair of spatial deictics and some employ, in addition, a vertical up/down distinction. Similarly, Rauh (1983: 12) has stated that "the same descriptive system underlies various dimensions, i.e. egocentric-localistic, and therefore deictic dimensions are determined analogously to the local deictic dimension". Levinson (1992: 5) puts this in a nutshell by emphasizing that "spatial thinking intrudes into thinking about almost all other domains: when an intellectual problem can be spatialised, it can be conceived clearly". It will be shown below how languages differ as to the ways they conceptualize and grammaticalize the notion of space. To account for these contrasts, I will make the following claims:

330

Barbara Kryk-Kastovsky

(a)

The differences among languages as to their representations of the concept of space must be studied not only in terms of the traditional horizontal dimension of the distance from the Origo (based on the data from Indo-European languages), but also in terms of other dimensions likely to be found in non-Indo-European languages.

(b)

The apparent "simplicity" of the Indo-European systems of spatial terms as opposed to the "complexity" of the non-Indo-European systems can be accounted for by a variety of factors: on the one hand, by extralinguistic, cultural and ecological ones (the latter including, of course, the cognitive and perceptual factors which determine different conceptualizations of space characteristic of different speech communities) and, on the other hand, by intralinguistic factors, such as dialectal variation or language change.

As a guiding principle of my investigations I will accept Levinson's statement concerning the cautious approach the language analyst should take (1992: 25): ... We can use the semantic distinctions made in the various languages of the world as heuristics for setting up underlying conceptual distinctions. But any one language is likely to draw upon a subset of these distinctions, and possibly use quite different underlying conceptual systems in different parts of its grammatical system.

2. Some cross-linguistic data Even a brief glance at the ways various languages conceptualize and codify spatial distinctions suggests how difficult they are to capture by a unified theory (although some such attempts have been made, cf. e.g. Schmid 1972, Rauh 1983). The problem lies not only in the abundance of messy data coming from a variety of Indo-European and non-Indo-European languages, but above all in the number of possible linguistic and extra-linguistic factors that might account for these data. At this point, an explanation is in order. The contrasts mentioned below concern the ways how different languages have divided space in terms of proximity and distance. A two-way contrast represents the most basic proximal-distal distinction (of a distance from the speaker's zero-point, the Origo), a three-way contrast includes the proximal-distal distinction complemented by a neutral/medial category in between the two, and the re-

Variables of the conceptualization of space

331

maining manifold contrasts stand for a variety of points indicating the distance of an object from the Origo.

2.1. A two-way contrast A two-way proximal/distal contrast is found in the following languages: English, cf. e.g. Fillmore (1975), Lyons (1977); Danish, cf. Faerch (1980); Dutch, cf. Kirsner (1979); and Polish, cf. Kryk (1987). The distinction can be accounted for within Rauh's threefold system of egocentric-localistic criteria. These are: a) "coding place", i.e. the speaker's location; b) "in connection with coding place", i.e. close to the speaker (the addressee's location); c) "not in connection with coding place", i.e. away from the speaker, cf. Rauh (1983: 18). Thus, I will argue here that the languages analyzed do not distinguish between Rauh's domains a) and b), the speaker's location being collapsed with whatever is near him/her, as opposed to whatever is far from him/her. It will be shown below that most Germanic, Romance and Slavic languages are characterized by a threefold distinction, roughly corresponding to Rauh's system. The problem still remains why it is that English, Danish, Dutch and Polish are "the odd men out" in their groups. My attempt at an explanation will start with a few assumptions. First, I will follow Traugott (1982: 266) in assuming that when a language loses certain grammatical distinctions, e.g. a third degree of deixis (as in the case of the English system: here-thereyonder > here-there), the same semantic possibilities are present whether grammaticalized markers for them exist or not. Secondly, as I have stated elsewhere (Kryk 1990: 52), the grammaticalization of the spatial contrasts as either two or three distances from the Origo is highly relative, which can be supported by numerous data. For example in some English dialects this can be juxtaposed with the speaker's location marked as this here, parallel to the Polish, slightly substandard ten tu: (1)

This here guy is nuts.

(2)

Ten tu facet jest stukniety.

Moreover, Standard Polish allows for a perfectly acceptable ten oto 'this here', comparable to the Czech ten to, cf. below, denoting the speaker's location shared with some entity. Likewise, some English dialects still employ the otherwise obsolete forms yon {yonder), which complement the adjectival paradigms of this/that and the adverbial paradigm of here/there, cf. Jespersen (1933: 156).

332

Barbara Kryk-Kastovsky

My claim is that dialectal variation is the clue here. As we have seen above, both in English and in Polish the choice between the dyad and triad of deictic terms hinges upon the line one draws between the standard variety of language and its dialects. To give another example which contributes to the obscurity of the picture even more, some data from the Oravian dialect of Polish make us realize that due to dialectal variation European languages have more to offer in terms of the fine spatial distinctions than has been claimed so far. 1 It has been convincingly shown by Sikora (1989) that the Oravian dialect spoken in the Malopolska region of Southern Poland distinguishes four spatial adverbs tu/xaf/xan/tam, as opposed to the Standard Polish proximal tu 'here' and distal tarn 'there'. Interestingly, the usage of the four Oravian forms is governed by a different organization of space. Three locations are distinguished: (a) the speaker's location; (b) the addressee's location; (c) the location referred to in the conversation (whose visibility to the speaker and/or the addressee constitutes an additional parameter). This results in the following distribution of locative adverbs (Sikora 1989: 72f.): if (a) = (b) and (c) is visible to the speaker, xan is used; if (c) is not visible, tam 'there' is used; if (a) = (b) and (c) is visible to both the speaker and the addressee, tu 'here' is used; if (a) Φ (b) and (c) is visible to the speaker, χάη is used, if (c) is not visible, tam 'there' is used; if (a) Φ (b) and (b) = (c), tam 'there' is used; if (a) Φ (b) and (c) = (a), xaf is used (1989: 72f). Apart from these locative four adverbs, separate paradigms exist for ablative, adlative and perlative relations, each of the individual 16 forms having at least one additional contextual variant. It goes without saying that the complex system of Oravian locative adverbs is hard to comprehend to a speaker of Standard Polish with its two deictic forms.

2.2. A three-way contrast A triple contrast of the speaker's distance with reference to the deictic center can be found in most European languages, among others Romance languages like Latin, French, Italian, Spanish and Portuguese, as well as in Greek and German, cf. Frei (1944), Rauh (1983: 15). The threefold distinction, labeled proximal/neutral/distal, also holds in most Slavic languages, e.g. Russian, Czech, cf. Hauenschild (1982), and Serbo-Croatian,

Variables of the conceptualization of space

333

cf. Zee (1985), as well as in some non-Indo-European languages, like Nunggubuyu (spoken in the Northern Territory of Australia), cf. Heath (1980), Tagalog (spoken in the Philippines), cf. Fillmore (1975), or Swahili, cf. Opalka (1982). The problem arising here is that there are several criteria leading to this threefold localistic differentiation. In most cases the location between the proximity and distance from the Origo is occupied by the position of the addressee (as in Italian), or one can make a choice between the position of the addressee and the location surrounding the coding place (as in Nunggubuyu), or the location is neutral (as in the Slavic languages). Already at first sight this division of space is far from uniform, so that one of the possible ways of labeling it would be a vague cover term like Rauh's criterion b), cf. 2.1. above. This solution does not, however, account for the multiplicity of factors responsible for this division of space in particular cases. Consider the following: (a) In many languages the tripartite division is far from stable and, like in the case of the twopartite division, is subject to dialectal variation. For instance, the following differences of opinion arise when we analyze the German triad dieser/der/jener, usually glossed as 'this/that/that'. According to Rauh (1983: 22), dieser represents person deictic expressions determined with respect to her domain (a + b), i.e. "coding place" + "in connection with the coding place", jener with respect to (c), i.e. "not in connection with the coding place", while der is not determined with respect to (a + b), i.e. medial. Ehlich (1982: 316) claims that both dieser and der are "near demonstrative pronouns", whereas jener, along with der da, are "far demonstrative pronouns". Ehrich (1982: 45) considers dieser as a form equivalent to this, i.e. proximal, and both der and jener as equivalent to that, i.e. distal. In my opinion, this discrepancy shows that the medial category der is pragmatically fuzzy, i.e. oscillating between proximity and distance. Finally, sufficient evidence points to the fact that the paradigms of the German deictics are often subject to dialectal variation, and differences occur even within one local dialect. For instance, the speakers of Ziiritütsch (a variety of the Swiss German dialect spoken in and around the town of Zürich) have completely ruled out hier in favor of da (Rüegg 1994). According to Brandmaier (p.c.), in certain varieties of the Bavarian dialect of German, only one form, i.e. da, is used to cover all the three spatial distinctions. (b) Analogous variation occurs in Romance languages. Thus AndersonKeenan (1985) suggest French as a possible example of a minimal one-way deictic system, since the French paradigm ce/cette/cet does not encode distal distinctions. However, the language system has employed a defense mechanism by adding the suffixal -ci and -Ιά (as in ce chat-cil-lä 'this cat

334

Barbara Kryk-Kastovsky

here/there'), whereby the speaker indicates if the object is in his/her proximity or further away. Italian, on the other hand, is undergoing a process similar to English in so far as its triple paradigm (questo/codesto/quello) is gradually losing one of its members (in Standard Italian codesto has the tendency of becoming obsolete). (c) Slavic languages are another case in point here. Although Polish is usually quoted as a language with a two-way deictic contrast both in the pronominal and adverbial paradigms, and Czech, Russian and SerboCroatian are said to have a three-way contrast, my contention is that Slavic languages oscillate between the two. Some evidence with regard to Polish has already been given above. Russian and Czech present another problem. It has been convincingly demonstrated by Hauenschild (1982) that the twofold proximal/distal distinction traditionally employed to describe the demonstrative pronouns in Russian and Czech should be extended to a threefold paradigm, arrived at by investigating the functioning of the demonstratives in what she calls pragmatic, semantic and syntactic deixis. The two basic schemes, along with their German and English equivalents, look as follows (Hauenschild 1982: 168 and 183, respectively):2 (3) R. C. G. Ε. (4) R. C. E. G.

proximal etot ten dieser this

distal tot onen jener that

proximal etot tento this dieser der hier

neutral tot ten that der der da

distal (tot) tarnten {that) der dort (der da)

(d) The non-Indo-European languages with threefold deictic distinctions pose even more problems, since they often employ criteria absent from the Indo-European systems. Although Nunggubuyu has a tripartite division of space into what Heath (1980) calls "Proximate", "Immediate" and "Distal", the category "Immediate" is subject to considerable variation (it means that the referent is in the immediate surroundings, and although it focuses on the position of the addressee, it is often used when the location is closer to the speaker than to the addressee), cf. Heath (1980: 15). As to Swahili, Opalka (1982) notes that, like in all Bantu languages, the local deictic system reveals something about the orientation behavior of the speakers

Variables of the conceptualization of space

335

which is alien to the speakers of languages like English or German. It is a combination of a pragmatic-semantic specification and a morpho-syntactic one, so that the threefold spatial orientation of Swahili speakers is based on three features: distance, dimension and indefiniteness. In local deixis, this results in the following meanings: (i)

(ii)

(iii)

distance is relevant to objects which are "immediate", i.e. near each other or directed to the speaker, i.e. close to the IndoEuropean understanding of "proximate"; dimension is relevant to objects which are "in" each other or are instances of speaker-"inclusion" - a concept completely alien to the European way of conceptualizing space; indefiniteness describes objects that are "somewhere" or "away from the speaker", i.e. it partly includes the Indo-European "distal" category.

Notice that the difference between the Indo-European and the Bantu way of dividing the space lies in the egocentricity of the process in the Indo-European languages, where the only relevant point of reference is the speaker at the Origo, since the addressee's location is by definition speaker-oriented and the remaining points in space are lumped together under a category determined by default, i.e. neither at the speaker's nor at the addressee's location. This is why the third person has often been called a "non-person" (cf. Benveniste 1971). In the case of the Bantu languages, the perspective seems to have changed considerably, since it is always the location of some objects with relation to each other or some remote location that has to be considered along with their relation to the speaker. Thus, egocentricity is no longer the only criterion determining the division of space; both the speaker and the third party can be instrumental in the process. Finally, Japanese is an excellent example of combining a few of the problems discussed here in connection with a three-way deictic contrast. For one thing, there is no agreement as to whether Japanese actually distinguishes between three or four distances from the Origo. Frei (1944) enumerates it among languages with three such distinctions, whereas Coulmas (1982) suggests four, claiming, moreover, that the members of the quadripartite demonstrative paradigm employ the notion of pivot, as proposed by Bühler (1934) and developed by Fillmore (1966, 1968). The four Japanese demonstratives can be represented as follows:

336

(5)

Barbara Kryk-Kastovsky

kore sore are dore

speaker pivot: proximal hearer pivot: proximal; speaker pivot: medial speaker pivot and hearer pivot: distal indefinite

The English equivalents of the four demonstrative pronouns are roughly: this, that, that, which, respectively, cf. Coulmas (1982: 211). Similarly, the local deictics reflect the distinction between parallel categories: the speaker's location, koko 'here', the addressee's location, soko 'there', a definite place distant from the speaker and the addressee, asoko 'yonder', and an indefinite place, doko 'where'. It is worth pointing out that the Japanese system is reminiscent of the Slavic system, with its medial category close to the addressee. However, as follows from Coulmas' account, it is in fact much more complex, since the complete quadruple paradigm includes, apart from demonstratives and local deictics discussed above, also adnominal, adjectival, adverbial and directional demonstratives.

2.3. A four-way distinction Four distances from the Origo have been attested in Tlingit (spoken in Northwest coast of Canada) and in the Papuan languages (Eipo, Yale and Angguruk spoken in the former Dutch West New Guinea). Frei (1944: 115) glosses the deictic domains in Tlingit as: objet rapproche, mains rapproche, eloigne, tres eloigne et generalment invisible, whereas Heeschen (1982) claims that the way the speakers of Eipo divide space is much more complex. Although their natural environment is meticulously labeled by means of numerous names for mountains or rivers, they do distinguish between four deictic points of reference relative to ego's position (Heeschen 1982: 84): (6)

fleiouor-

here up there down there across there

These bound morphemes combine most frequently with verbs, nouns, postpositions and predicative suffixes. Notice, however, that apart from the horizontal proximal-distal distinction on which Indo-European languages base their deictic systems, the Eipo employ the vertical up-down distinction. In addition, the members of the Eipo deictic paradigm can be intensified by prefixing d- (Heeschen 1982: 86):

Variables of the conceptualization

dadeidoudor-

of space

337

here (in a wider area around the speaker and hearer, here and there) very far up there very far down there very far across there

It follows from the data above that non-Indo-European languages tend to employ more complex systems of spatial distinctions than do the Indo-European languages, hence the number of possible contrasts increases as one leaves the European territory. The following data on more elaborate spatial contrasts confirm this observation.

2.4. Other multiple deictic contrasts For reasons of brevity, the remaining examples of multiple contrasts will be discussed under one heading and only selected languages will be given more attention. As can be expected, the data come from non-Indo-European languages. According to Frei (1944), five deictic distinctions operate in the Bantu language Ronga, which have been glossed by him as: 1. ci, 2. la, la plus loin (ou: la oil tu es, toi avec lequel je parle), 3a. Ιά-bas, loin, 3b. tout labas, tres loin, 3c. I'objet qui est ä I horizon. He identifies the following six local deictics for Bisaya, a language of the Southern Philippines: la. proche du parleur, lb. eloigne du parleur, 2a. proche de l'interlocuteur, 2b. eloigne de l'interlocuteur, 3a. igalement proche du parleur et de l'interlocuteur, 3b. igalement eloigne du parleur et de l'interlocuteur (1944: 115). Also Kikuyu (a Bantu language) employs six deictic distinctions, described by Denny (1978) as non-extended here, extended here, non-extended there-in-field, extended there-in-field, non-extended there-out-of-field, extended there-out-of-field. Note that Ronga and Bisaya divide the space more or less equally between the speaker and the addressee in terms of proximity and distance, which makes them similar to the Slavic and Romance languages. However, Ronga makes more distinctions for third parties, objects away from both the speaker and the addressee (cf. categories 3a, 3b, 3c: 'overthere, far'; 'far away'; 'the object on the horizon'). An eleven-way contrast can be found in Tolai (an Austronesian language spoken in Papua New Guinea), cf. Mosel (1982). It is worth noting that, like in many other analyses of non-Indo-European languages, the categories describing spatial relations are markedly different from what is employed in Indo-European languages. Thus the Tolai deictics are not only distinctive

338

Barbara Kryk-Kastovsky

for the proximal-distal contrast {here/there), but can also be marked for the following factors: the location of a place with relation to the speaker (e.g. up/down)·, whether a place is the center, the goal or a source of an action, and whether the place is known to the hearer, cf. Mosel (1982: 111). Finally, the most multifarious contrast which has, to the best of my knowledge, been attested so far, can be found in Eskimo with its 88 spatial distinctions. According to Denny (1978: 75), the Eskimo system involves a contrast of "here" with five locations of "there", all but one of which ("in there") are classified as "extended" or "non-extended" - these are expressed by 11 roots. Each root can take the prefix ta (C) for "other field", yielding 22 prefixed roots. All these can take four local case suffixes, yielding 88 words.

3. Towards an explanation As follows from the above, the non-Indo-European languages differ markedly from the Indo-European ones in that their systems of spatial deictics are much more complex, both quantitatively and qualitatively. It has already been conjectured in the Introduction that the former conceptualize space in terms of much finer distinctions than do the latter, hence the number of spatial terms increases as we leave the European territory. And this is what has been confirmed by my data. Moreover, the data have revealed a striking diversity of dimensions necessary for the analysis of the non-Indo-European languages, as compared to the simple proximal/distal distinction sufficient in the case of the Indo-European systems. Indeed, it has been shown that other dimensions, like height, extension, or containment would have to be taken into account, which is consonant with my claim a) in 1. above. My claim (b), i.e. that the apparent simplicity of Indo-European systems of spatial terms as opposed to the "complexity" of the non-European systems can be explained by a variety of extralinguistic and intralinguistic factors, can be supported by the arguments formulated here in terms of answers to the following detailed questions: (a)

Why are the simplest two-way/three-way spatial distinctions limited mostly to European languages? Thus

(b)

Is there any correlation between the complexity of the spatial systems encoded by particular languages and the Indo-European vs. non-IndoEuropean distinction?

Variables of the conceptualization of space

(c)

339

If the complexity of the spatial differentiation present in particular languages can be explained along the Indo-European vs. non-Indo-European dividing line, how can one account for the variation between a two-way and a three-way contrast spread randomly over members of different groups of European languages?

The most plausible answer to questions a) and b) can be provided along the cultural/ecological lines, i.e. Denny's (1978) Cultural Ecology Hypothesis (the complexity of a deictic system is inversely proportionate to the extent to which the environment in which a language is spoken is manmade), and Perkins' (1992) claim that grammatical variation is related to the complexity of the cultures where the language is spoken3. As Perkins has shown by providing statistical data from numerous languages, the less complex the culture, the more complex the deictic system of the language spoken in it. If we conflate the two ways of reasoning, the following conclusion results. In European and other highly developed societies numerous man-made points of orientation provide excellent means of dividing the space into distinct, well-defined locations, e.g. behind the second gas station. Moreover, not only can the points of orientation be described, but they often bear proper names, e.g. the ESSO gas station, the Empire State Building, etc. Consequently, for the members of the societies with an advanced man-made environment a two- or three-way system of spatial expressions is sufficient to denote the distance of a given object from the Origo. The situation is markedly different in societies where human interference with the environment is minimal. Natural environment tends to have fewer distinct points of orientation than the man-made one, and more spatial distinctions are needed for man to find his way in a bush, jungle or tundra. There one has to rely on trees, rivers, or hills, which normally have fewer distinctive features and are more difficult to refer to unambiguously than artefacts like buildings or highways. This does not mean that there are no natural landmarks, but they have to stand out from their environment, like, e.g., Baobab Hill in Kruger Park, which early travelers to and from the east coast of South Africa used as a navigational aid: a little hill crowned by a majestic baobab tree in an otherwise nondescript plain. However, underdeveloped areas normally have very few such outstanding points of orientation. Therefore, the majority of potential locational reference points will remain unnamed in view of their multitude (one cannot possibly distinguish between thousands of similar trees not to mention the division of vast open spaces). Hence, space is conceptualized differently by the members of culturally less complex societies which necessarily develop more complex deictic systems. This has been confirmed by experimental studies of the differences in spatial perception between children from urban

340

Barbara Kryk-Kastovsky

and rural communities, e.g. Southern Quebec children and Northern Quebec Inuit children, respectively, cf. Pallascio, et al. (1993). It can be concluded at this point that the complexity of the spatial distinctions attested in various languages is accountable for by cognitive, cultural, and ecological factors, the correlation being as follows: The more complex the culture (i.e. the more human interference there is into the ecological system of a given speech community), the less complex the cognitive representation of space, resulting in a simpler linguistic representation thereof. The question that has been left unanswered here concerns the dialectal variation present in many European languages, which allows for an abundance of forms often comparable to the complexity of the non-Indo-European systems, e.g. the Oravian dialect of Polish (cf. above). The seeming paradox can easily be solved within the present framework: the dialects of Polish, English and other languages discussed here as having more complex deictic systems than the standard varieties of those languages are spoken in rural, i.e. less developed areas of the given countries. As to the problem posed in (c) above, I agree with Rauh (1983: 18ff) who has distinguished three criteria which constitute the foundations of deictic determination. It follows from my data that most European languages have adopted the threefold division of the distance from the Origo, while a few others are characterized by the proximal/distal dichotomy. Since this twofold contrast holds in the case of three Germanic languages and one Slavic language, one might conjecture at this point that the languages analyzed simply do not distinguish between Rauh's domains (a) and (b) - the speaker's location being collapsed with whatever is near him/her, as opposed to whatever is far from him/her. This seems to be an ongoing development in various European languages, which conflate their three deictic distinctions into two, as is the case, e.g. in English, Polish, Italian, or German. What follows from the above is that the twofold vs. threefold contrasts in European languages are relative distinctions sensitive to idiolectal and dialectal variations which certainly have historical roots. It is my conjecture that the threefold spatial distinction was the historically dominating one, as evidence from English and Polish has shown. In the course of time, some of the distances from the coding place gradually became superfluous and got absorbed by the remaining ones, in accordance with the general tendency of languages towards structural simplification. This process still seems to be in progress in certain European languages, where one category is already obsolete in the standard variety, although it is still acceptable in some dialects. A possible historical explanation of deictic contrasts calls for much more detailed studies in the field.

Variables of the conceptualization of space

341

4. Conclusion It has been demonstrated here that the variety of deictic systems operating in different languages calls for a complex, multidisciplinary approach. Not only is the degree of the complexity of particular deictic systems a linguistic phenomenon, which must be explained both along the synchronic and the diachronic lines, but it also hinges upon a multitude of perceptual, cognitive, and cultural factors: (a)

Synchronically, the deictic system of a language should be studied not only within its standard variety, but it should also be compared to its dialectal variants to bring out the existing contrasts and find the possible (synchronic and/or diachronic) explanations.

(b)

Diachronically, the deictic system of a language calls for intralinguistic analyses, tracing the history of particular forms within the development of a single language, as well as interlinguistic studies revealing the historical facts within particular language groups, possible Sprachbund phenomena, etc.

(c)

The deictic system of a language is also shaped by its speakers, in particular their perceptual and cognitive understanding of the notion of space. Hence, these factors should be investigated in connection with the geographical and socio-cultural context in which the members of a particular speech community are located.

If all these factors are given due attention, one day the dividing line between the European and non-European languages, allegedly spoken by representatives of more complex vs. less complex cultures, might be abandoned in favor of an elaborate system of isoglosses reflecting the division of our common space, the World.

Notes 1.

2.

Cf. Denny's (1978) Cultural Ecology Hypothesis concerning the simplicity of (Indo)European deictic systems as opposed to the complexity of the deictic systems spoken by societies with minimally man-made environments. Similarly, Perkins (1992) has tried to prove that the complexity of the grammatical systems is inversely proportionate to the complexity of the culture in which the language is spoken. Note that Russian employs the same form tot for both the neutral and distal uses. According to Hauenschild (1982), the systematic position of Russian tot can be described as either neutral or distal. Thus, it will be

342

3.

Barbara Kryk-Kastovsky

neutral in its function in syntactic deixis, but it will have the meaning of distance when "the syntagmatic contrast to the proximal term of the opposition h o t is explicit or implicit" (1982: 183). The complexity of a culture is measured on a Cultural Complexity Scale based on such variables as e.g. subsistence economy, type and intensity of agriculture, settlement pattern, mean size of local communities, craft specialization, class stratification, inheritance of real and movable property, regional organization, etc. For details, cf. Perkins (1992: 140).

References Anderson, Stephen and Edward Keenan 1985 Deixis. In: Shopen, Timothy (ed.), 257-308. Bach, Emmon and Robert Harms (eds) 1968 Universals in Linguistic Theory. New York: Holt, Rinehart and Winston. Bechert, Johannes, Giuliano Bernini and Claude Buridant (eds). 1990 Toward a Typology of European Languages. Berlin: de Gruyter. Benveniste, Emile 1971 Problems in General Linguistics. Coral Gables, Fla.: University of Miami Press. Bühler, Karl 1934 Sprachtheorie. Jena: Fischer. Coulmas, Florian 1982 Some remarks on Japanese deictics. In: Rauh, Gisa (ed.), 209-221. Denny, J. Peter 1978 Locating universale in lexical systems for spatial deixis. CLS, Papers from the Parasession on the Lexicon: 71-84. Ehlich, Konrad 1982 Anaphora and deixis: Same, similar, or different? In: Jarvella, Robert and Wolfgang Klein (eds), 315-338. Ehrich, Veronika 1982 Da and the system of spatial deixis in German. In: Weissenborn, Jürgen and Wolfgang Klein (eds), 43-63. Faerch, Claus 1980 A contrastive description of deixis in Danish and English. In: Fisiak, Jacek (ed.), 365-376. Fillmore, Charles J. 1966 Deictic categories in the semantics of come. Foundations of Language 2: 219-227. 1968 The case for case. In: Bach, Emmon and Robert Harms (eds), 1-88. 1975 Santa Cruz lectures on deixis 1971. Indiana University Linguistics Club. Fisiak, Jacek (ed.) 1980 Theoretical Issues in Contrastive Linguistics. Amsterdam: Benjamins. Frei, Henri 1944 Systfemes de diictiques. Acta Linguistica 4: 111-129. Givön, Talmy (ed.) 1979 Syntax and Semantics 12. Discourse and Syntax. New York: Academic Press.

Variables of the conceptualization of space

343

Greenberg, Joseph (ed.) 1978 Universals of Human Language. Stanford, Ca.: Stanford University Press. Hauenschild, Christa 1982 Demonstrative pronouns in Russian and Czech - deixis and anaphora. In: Weissenborn, Jürgen and Wolfgang Klein (eds), 167-186. Heath, J. 1980 Nunggubuyu deixis, anaphora and culture. CLS, Papers from the Parasession on Pronouns and Anaphora·. 151-165. Heeschen, Volker 1982 Some systems of spatial deixis in Papuan languages. In: Weissenborn, Jürgen and Wolfgang Klein (eds), 81-109. Jarvella, Robert and Wolfgang Klein (eds) 1982 Speech, Place and Action. Studies on Deixis and Related Topics. Chichester, N.J.: Wiley and Sons. Jespersen, Otto 1933 Essentials of English Grammar. Alabama: University of Alabama Press. Kirsner, Robert 1979 Deixis in discourse: An explanatory quantitative study of the Modern Dutch demonstrative adjectives. In: Givön, Talmy (ed.), 355-375. Kryk, Barbara 1987 On Deixis in English and Polish: The Role of Demonstrative Pronouns. Frankfurt/Main: Lang. 1990 Deixis - a pragmatic universal? In: Bechert, Johannes, Giuliano Bernini, and Claude Buridant (eds), 49-61. Lehmann, Winfred and Yakov Malkiel (eds) 1982 Perspectives on Historical Linguistics. Amsterdam: Benjamins. Levinson, Steven C. 1992 Primer for the field investigation of spatial description and conception. Pragmatics 2: 5-47. Lyons, John 1977 Semantics. Vol. 2. Cambridge: Cambridge University Press. Mosel, Ulrike 1982 Local deixis in Tolai. In: Weissenborn, Jürgen and Wolfgang Klein (eds), 111-132. Opalka, Hubertus 1982 Representations of local ni-deixis in Swahili in relation to Biihler's Origo des Zeigefeldes'. In: Weissenborn, Jürgen and Wolfgang Klein (eds), 65-79. Pallascio, Richard, Richard Allaire, and Pierre Mongeau 1993 Spatial representation of geometric objects: A North-South comparison. Etudes /Inuit/Studies 17(2): 113-125. Perkins, Revere D. 1992 Deixis, Grammar, and Culture. Amsterdam: Benjamins. Rauh, Gisa 1983 Aspects of deixis. In: Rauh, Gisa (ed.), 9-60. Rauh, Gisa (ed.) 1983 Essays on Deixis. Tübingen: Narr. Rüegg, Kathrin 1994 The concept of space in different language systems. Unpublished paper. University of Zurich.

344

Barbara Kryk-Kastovsky

Scbmid, Wolfgang P. 1972 Die pragmatische Komponente in der Grammatik. Akademie der Wissenschaften und der Literatur 9: 1-20. Reprinted in Rauh, Gisa (ed.), 61-78. Shopen, Timothy (ed.) 1985 Language Typology and Syntactic Description. Vol. 3. Grammatical Categories and the Lexicon. Sydney: Cambridge University Press. Sikora, Kazimieiz 1989 Ο niektörych przyslöwkach przestrzennych w gwarze orawskiej (About certain spatial adverbs in the Oravian dialect), Polonica 14: 67-76. Traugott, Elizabeth C. 1978 On the expression of spatio-temporal relations in language. In: Greenberg, Joseph (ed.), 369-400. 1982 From propositional to textual and expressive meanings: Some semanticpragmatic aspects of grammaticalization. In: Lehmann, Winfred and Yakov Mal kiel (eds), 245-271. Weissenborn, Jürgen and Wolfgang Klein (eds) 1982 Here and There. Cross-Linguistic Studies on Deixis and Demonstration. Amsterdam: Benjamins. Zee, Draga 1985 From demonstrative to evidential: The case of Serbo-Croatian to. Unpublished paper. Stanford University.

Rethinking some universals of spatial language using controlled comparison Sabine Neumann and Thomas Widlok

1. Introduction Comparison is the major tool for any research into the universality and diversity of how space is conceptualized in language. To discover that there are diverse and universal elements has been achieved by comparisons as wide and broad as possible. However, to understand diversity and the limits to variability requires controlled comparisons. This chapter explores the potential of regional1 comparison in the field of spatial conceptualization. It shows some of the advantages, in terms of both theory and methodology, that regionally controlled comparisons have against the other, more common, modes of comparisons. Regional comparison is a complement to those comparisons that either aim at the greatest possible scope of languages and cultures compared or which select cases either arbitrarily or according to other cultural and linguistic factors. 2 The research techniques that we have applied were developed to investigate spatial language and non-verbal thinking about spatial arrangements. They have been designed and used as part of a large cross-cultural research project of the Cognitive Anthropology Research Group in Nijmegen. Our contribution attempts to resolve some of the problems raised by this exploratory comparison carried out by the Research Group as a whole by presenting and analyzing some of the data from two languages of the southern African region. The material discussed was collected by Neumann with the help of Kgalagadi (Bantu) speakers in Botswana and by Widlok with the assistance of Hai || om (Khoisan) speakers in Namibia. In our discussion of this data we will show that by establishing the limits of variability, regional comparison can not only contribute to questions of universality and diversity but also improve our knowledge of languages in the region under consideration. Not all regional comparisons are effectively controlled comparisons. Discussing the recent work of Fourie (1992) we want to argue that an effective control of regional comparison should not only be concerned about

346

Sabine Neumann and Thomas Widlok

the selection of languages but primarily about the selection of features to be investigated.

2. Exploring the problem of investigation Non-linguistic experiments run by the Group in Nijmegen suggest that there which Dutch subjects on the one hand from non-Indo-European languages on spatial arrangements.

members of the Cognitive Research are differences between the way in and subjects in a variety of samples the other hand cognitively deal with

Estimated absolute tendency (%)

Figure

1.

Figure 1 for instance shows the results of a task called 'animals in a row' run with Dutch speakers and with speakers of two southern African languages, namely Hai || om and Kgalagadi. In this task, subjects are asked to remember three animals placed in a row before them, they are then asked to turn around 180 degrees and asked to rebuild the sequence of the three animals as they remember them (see Figure 2).

Universals of spatial language

Table 1

347

Table 2

Figure 2.

Two dominant solutions can be distinguished. The first is a relative solution, selected by almost all Dutch subjects, that is the direction and order of the animals is remembered with regard to the left and right of the observer. Alternatively an absolute solution is possible. In that case, chosen by most Hai || om subjects, the animals are remembered as walking say to the east and not say to the left of the speaker. And then there are apparently cases without a dominant solution, in this case exemplified by the Kgalagadi speakers. 3 The intriguing question then is, what is responsible for the fact that people like the Dutch chose a different solution than say people like the Hai || om. There are a number of possible candidates for plausible explanations and the following list does not claim to be complete. There is above all a different natural environment as well as a different man-made environment. One could argue (as Denny 1978 and Heeschen 1982 do) that our urban and technical environment including road traffic and machines of all kinds pushes Europeans to think in terms of left and right whereas the rural environment of many non-Europeans supports orientation in cardinal terms. Then different expectations and strategies with regard to experiments of this sort may play an important role. After all, experiments and cognitive tasks are not context free but simply represent one very specific context that is well-entrenched in our culture. Therefore, connected to that, the question arises which cultural factors and social practices of how problems of this sort are usually solved are involved. And of course there is language as a factor to be considered, if we want to find out which of these differentiating features it is that correspond to the different solutions chosen

348

Sabine Neumann and Thomas Widlok

in the tasks. Preferably then our comparative work should be controlled with regard to the factors identified. One way to do this is to try and eliminate all but one factor on this list and see whether there is still a difference in the way tasks like those in the 'animals in a row' experiment are solved. Our regional comparison is an attempt in this direction.

3. Cardinal directions in the region Kgalagadi speakers and Hai || om speakers share many features with regard to their natural as well as man-made environment. Both groups have a background of the same modes of subsistence consisting of hunting and gathering, livestock herding and a semi-nomadic lifestyle. They also share a similar degree of interaction with Western style industrialized society, in terms of transport, consumer items and schooling. What we consider to be important is the fact that these speakers were equally unaccustomed to cognitive experiments. The way in which they were approached by the researchers to do these tasks was very similar. One very clear factor in which Kgalagadi and Hai || om differ is their mother tongue, an eastern Bantu language in the former case and a Khoisan language in the latter case. There is very little commonality in terms of grammar or lexicon and there is no direct contact between the two languages. The difference in the language spoken corresponds with different solutions selected in the 'animals in row' task. Comparing the Hai || om and the Kgalagadi results in this experiment, there appears to be a bias towards an absolute solution in the Hai || om sample. This tendency toward absolute solutions does not exist in the Kgalagadi sample which nevertheless is also different from the Dutch sample. But it is exactly because the two languages are so very different from each other that on the basis of the nonlinguistic evidence it was impossible to say more precisely where the crucial differences had to be located. Are they in the lexicon, in particular syntactic or semantic features, in pragmatics, or everywhere? There is only one recent attempt in which spatial expressions in Bantu and Khoisan languages are compared. Fourie (1992) suggested that there are considerable semantic differences in the lexicon of these two language groups. Looking at the terminology of cardinal directions he noted that cardinal terms in Ndonga, Kwanyama and Herero (the Bantu languages in his sample) have a different etymological derivation than those in Nama, a Khoisan language very closely related to Hai || om. Table 1 summarizes the opposition Fourie draws.

Universals of spatial language

349

Table 1. Regional comparison of the semantics of cardinal terms (based on Fourie 1992)

BANTU

KHOISAN

(Herero, Ndonga, Kwanyama)

(Nama)

EAST-WEST

EAST-WEST

winds rising and setting of sun (culturally marked in hut (culturally marked in settlement layout and ritual, 'sun worshippers') construction and funeral rituals, 'moon worshippers')

NORTH-SOUTH

NORTH-SOUTH

contingent, names of neighbouring peoples (sometimes dichotomized)

fixed proper terms (sometimes interchangable)

According to Fourie, the Bantu languages under investigation derive cardinal terms from the course of the sun for the east-west axis whereas the cardinal terms for the north-south axis are etymologically linked to the presence of contingent neighbouring peoples living in these directions. In contrast, in Khoisan languages like Nama the east-west terms are linked to specific winds and there are proper, if interchangeable, terms for northsouth. In each case, Fourie tries to establish the way in which the various directions are marked in cultural, especially religious representations. But there are more commonalities than Fourie's opposition suggests. In both cases a two-axes model is used with a salient east-west (marked and fixed directions) and a weak axis (contingent or interchangeable terms for north and south). And, as we will show below, the opposition breaks down when other languages are included in the regional comparison. Fourie uses his conclusions drawn from a comparison of cardinal directions in languages of the region to make a more general theoretical point about the spatial domain. In his discussion about the conceptualization of cardinal directions he comments on the theoretical positions of 'nominalism', 'rationalism', and 'conceptualism' in discussions about linguistic categorization as put forward by Dirven (1991), Taylor (1989) and others. Together with the nominalists (and some conceptualists) he argues against 'universal semantic structures' determining how the domain of space is

350

Sabine Neumann and Thomas Widlok

conceptualized. In contrast to the rationalists (and some conceptualists), he proposes on the basis of his data that conceptualization in this spatial domain is 'determined more' by culture and language convention (Fourie 1992: 154). We would like to take issue with some of the conclusions drawn from Fourie's regional comparison and at the same time develop the argument a little further. There are a number of points in Fourie's argument where assumptions are introduced that are in need of further discussion. Fourie amalgamates language and culture into one unified factor and he presents conventions and universals as constituting two mutually exclusive explanations (see Fourie 1992: 143). Before posing the question as to where universals may be found, it is a necessary preceding step to clarify whether structural similarities are found on the level of language, of culture or on the level of cognition. Universals may be found in all three fields and so are non-universal patterns which are, however, not necessarily conscious conventions. Analytically it seems advantageous to keep all three fields separate. Compared to the amount of linguistic research available, we know rather little about non-linguistic cultural practices, such as map making and orientation when travelling, which also involve concepts of cardinal directions. But there is reason to assume that there are culture-specific situations in which the variables are to be found on the level of social organization and not primarily on the level of language or individual cognition. Moving from these general issues to Fourie's regional comparison, we soon realized that Fourie's opposition does not hold as soon as we include Hai || om and Kgalagadi. Once Hai || om and Kgalagadi are included in the sample, the differences are not as clear as it seems. Furthermore despite etymological differences that may surface, for the Kgalagadi and the Hai || om case at least, both systems seem to be remarkably similar. The east-west axis is fixed according to the position of the sun. Kgalagadi shows a suggested Khoisan feature in making use of terms derived from specific winds. Hai || om shows the supposedly Bantu features of using terms for north and south that are contingent on neighbouring peoples and landscapes and of using east-west terms derived from the course of the sun (see Table 2).

Universals of spatial language

351

Table 2.

Kgalagadi

Hai || om

EAST-WEST

EAST-WEST

course of the sun

course of the sun

NORTH-SOUTH

NORTH-SOUTH

proper terms (winds)

contingent terms (neighbouring peoples and landscapes)

This result did not lead us to the conclusion that conventional language plays no role in space conceptualization nor that regional comparison is use-less. Our main point is that it is not the languages that were wrongly selected but that the wrong features of these languages have been compared. Instead of focusing on etymologies of cardinals and taking them to represent the synchronic structure of culture and language of this particular domain we suggest to look at conventional language usage in problem solving situations that involve cardinal directions and other forms of spatial language.

4. The method of investigation Language elicitation tasks provided an empirical method to effectively select linguistic features and to sufficiently delimit the domain for our controlled comparison. The elicitation task used as a tool in our research is the 'Men and Tree Photo-Photo Matching Game1 which was designed to systematically elicit spatial distinctions on the horizontal plane with a toy man and a toy tree. 4 The game is played by two players. They are seated side by side with a screen separating them. The screen between the players prevents them not only from seeing each others photos but should ideally also prevent communication by gestures. Each of the players has the same set of pictures. One of them, the Director, is choosing the pictures one by one and

352

Sabine Neumann and Thomas Widlok

describing them to the other player, the Matcher. On the basis of this description the Matcher has to select from his or her photos that picture which matches the director's picture. The Director and the Matcher can talk back and forth as much as they want. The Men and Tree Photo Game consists of four sets of pictures. Each set of photos consists of 'targets' (the contrasts we are interested in) and a few 'distractors' (other contrasts, e.g. full versus partial containment). Set 1 is a complete training set and does not contain any 'target' photos at all. Within a set the 'target' pictures vary from one another by one or two parameters in the arrangement of the objects on the photos. 'Target' pictures in set 2 - 4 vary within the parameters of directionality and location on the horizontal plane by 90 or 180 degrees. They each contain either the toy man and the toy tree or two toy men. The game proved to be useful in eliciting language about preselected contrasts. Responses are easy to evaluate (mismatched or correctly matched photos) and the cause of the mismatch is relatively easily determined. Furthermore the Men and Tree Photo Game was the first attempt of the Cognitive Anthropology Research Group to run a formally comparative task across all our field sites. Thus we have data not only for Hai || om and Kgalagadi but also data from the other languages mentioned plus Dutch and English for comparison. For our comparison we have analyzed the spatial propositions given for the target pictures of set 2 by each individual pair of Hai || om and Kgalagadi speakers. Each photo shows one man and one tree. They vary within the same parameters as the target pictures of the other sets. In the following we are outlining the systematic features of spatial description which are found in set 2 in Hai || om and Kgalagadi according to a method developed by David Wilkins. He suggested that "[...] Each target subset is viewed as its own discourse world [...]. [...] Each director-matcher pair can devise and negotiate its own repertoire of linguistic means for identifying distinctions [...]" within the constraints of the linguistic systems and the nature of the task (Wilkins 1993). Thereby we may decide what functional equivalence is independent of semantic relatedness. Two or more "[...] propositions are functionally equivalent if they distinguish the same subgroup of pictures from the rest", not only within a director-matcher pair playing a certain game but also across different pairs of players playing the same game, and even across different languages (Wilkins 1993). Consider the following examples for identifying functionally equivalent propositions within a game played by one Director and one Matcher. Looking at Table 11, it can be seen that 'man in east' and 'tree in west' are functionally equivalent. They distinguish the pictures 2.5., 2.6. and 2.8. from the other pictures. No other functional equivalent propositions were

Universals of spatial language

353

used. In Table 12 'man to my right' and 'man at my side' are functionally equivalent. These propositions distinguish pictures 2.5., 2.6. and 2.8. from pictures 2.3., 2.4. and 2.7. At the same time the propositions 'tree in front of man', 'man behind tree' and 'man facing tree' are functionally equally adequate to distinguish 2.3. and 2.5. from the rest of the target pictures. The following examples refer to functional equivalence across pairs of subjects and across languages. Pictures 2.5., 2.6. and 2.8. are distinguished from pictures 2.3., 2.4. and 2.7. by 'man to my side', and by 'man at my side' (in the Hai||om case), by 'man at right', 'tree at left' (in the Kgalagadi case), as well as by 'man in east [or west]', 'tree in west [or east]' (in both languages). Similarly picture 2.4. is distinguished from the rest of the pictures by the following propositions: 'tree behind man', 'tree at man's back' (in Hai || om), 'man turning back to tree' and 'man gives tree the back' (in Kgalagadi). Propositions such as 'in the west', 'facing east' etc. were classified as 'absolute' (ground centred), propositions such as 'to my right', 'facing left' etc. were classified as 'relative' (speaker centred), and propositions such as 'behind', 'man facing tree' and 'man behind tree' were classified as 'intrinsic' (object centred).

5. The results Following the method as outlined above we have grouped the Hai || om and Kgalagadi phrases used in the 'Man and Tree' task as either 'relative', 'absolute' or 'intrinsic' which appear to form three different contrasting types of spatial systems that are used not only in these two languages but more generally. The definition of these three systems as used here are as follows (see Annual Report 1993: 65-6): Relative systems 'utilize the position and orientation of the speaker or viewer of a scene, often to derive an imposed facet of the Ground or landmark object'. Absolute systems 'utilize fixed bearings, derived e.g. from meteorological, astronomical or landscape features'. Intrinsic systems 'utilize inherent properties of the landmark object to derive a projected region'. The question is which system takes priority. To illustrate this we summarize below the percentages of absolute, relative, intrinsic propositions

354

Sabine Neumann and Thomas Widlok

given in Hai || om and Kgalagadi. The actual number of propositions is given in brackets. 5 This summary shows a preference in the 'Men & Tree' game in Hai || om for an absolute and an intrinsic system, while the Kgalagadi data shows a clear preference for relative propositions (for more detailed results see Tables 3-18 in the appendix).

Hai || om

Kgalagadi

Absolute

(22) 47.8%

(15) 23.8%

Relative

(5) 10.9%

(27) 42.9%

Intrinsic

(19) 41.3%

(21) 33.3%

(46) 100 %

(63) 100 %

This and other language elicitation tasks which we ran with Kgalagadi and with Hai || om speakers show that terminology relating to all three kinds of solutions is being used. In terms of the lexicon both languages allow equally well a description of one object as being on the left side or the western side of another object. Neither of the two presents a closed system in the sense that it would only allow an absolute, relative or intrinsic way to talk about spatial arrangements. The main difference between the two languages as it appears from the 'Man and Tree' elicitation task is one of quantity and not of categorical difference in the sense of language form 'determining' in any direct way how people think about space and solve related problems. The languages differ to the degree in which they extend these three types of spatial descriptions into fields of the natural world. In other words, it is a different pattern of conversational extension that is characteristic for each of these languages. It is possible that the distinct pattern of thinking as it appears in our non-linguistic tasks is responsible for the characteristic conversational extension that we observed (that would be a 'rationalist' position). It is equally possible that a habitual or conventional way of talking about space leads speakers to their characteristic ways of memorizing and thinking about space non-linguistically (that would be a 'nominalist' position).

Universals of spatial language

355

However, our comparison suggests that the spatial configurations out there in the world, that is on the photo, did not determine whether people chose an absolute, a relative or an intrinsic way of description. Conversational agreement seems to have more power than rationalist thought or properties of the world out there. The extension of ways to think and talk about space in the various fields of the real world is a communicative process that relies on a community of speakers who share a language but who also share other modes of interaction. When looking at the results of our elicitation task more closely one realizes that absolute, intrinsic and relative descriptive phrases are fairly evenly distributed over the various photos that we used. By contrast they are not at all evenly distributed with regard to the pairs of subjects who cooperated in a session with the whole series of photographs. This suggests that it is not so much a particular photo, a particular spatial constellation that triggers off specific ways of talking or conceptualizing. Rather conceptualization seems to be guided very much by the communicative setting, the negotiated interaction between subjects as to how they are going to solve this task. The convention, however, is much more of a negotiated and situated kind than the notion of 'culture' as a unifying and comprehensive whole might suggest. Furthermore there are in these tasks as in any other interactional setting - social roles and relationships that influence the way in which communication proceeds. In the context of our elicitation tasks it was the Director who preselected a strategy which the Matcher followed if possible. Thus the nature of the task has its influence on the share the interlocutors have in establishing the conversational agreement. There are limitations to the variability observed and it is quite possible that there are universal structures not on the surface level of the lexicon and the etymology of cardinals but on the underlying practice level of negotiating strategic language usage in a given situation.

6. Conclusion The regionally controlled comparison which we have presented in this chapter shows that under similar situational and regional conditions speech communities differ in the degree to which they apply cardinal terms and other absolute expressions to describe an identical set of stimulus material displaying a number of spatial constellations. Relying on a set of language elicitation tasks we were able to demonstrate diversity not only between languages but also within the group of speakers of a language. The diversity within the two speech communities (Hai || om and Kgalagadi) is larger than suggested by earlier comparisons of spatial conceptualization in this region which were based on a comparison of the lexicon and etymologies.

356

Sabine Neumann and Thomas Widlok

Culture and language, conceived of as total systems, seem to be less a force in shaping variability than are conventional ways of social interaction and language use. These results suggest that with regard to questions of universality in spatial language, comparisons that are directed at language pragmatics are promising both in terms of method and in terms of a theory of spatial language that takes full account of the speakers involved.

Appendix: The results in detail The following Tables summarize the propositions made in the course of 'Man and Tree' elicitation tasks carried out by Hai || om and Kgalagadi speakers. The first set of Tables (Tables 4-9), referring to individual pictures, allows a comparison of the functional propositions made by Hai || om and Kgalagadi with regard to each 'Man and Tree' photograph (pictures 2.3. to 2.8.). The second set of Tables (Tables 11-18) are based on the same body of data but refer to individual events. They allow a comparison of the strategies employed by the players/speakers. Tables 3 and 10 summarize the two sets of Tables. The propositions have been grouped as 'absolute', 'relative' and 'intrinsic' according to the definition set out in the main text above. The distribution of propositions according to pictures (see Tables 4-9) is as follows: Table 3.

A: absolute

Hai || om

Kgalagadi

R: relative

(22A, 5R, 191)

(15A, 27R, 211)

Picture 2.3.

3A, OR, 51

3A, 4R, 31

Picture 2.4.

3A, OR, 41

3A, 3R, 51

Picture 2.5.

2A, 1R, 21

1A, 3R, 61

Picture 2.6.

5A, 1R, 21

2A, 7R, 41

Picture 2.7.

4A, OR, 41

3A, 5R, 21

Picture 2.8.

5A, 3R, 2A

3A, 5R, 1A

I: intrinsic

Universals of spatial language

357

The results in detail ('#' indicates a proposition made after being asked/prompted by the Matcher):

Table 4. Picture 2.3. Hai || om

Kgalagadi

ABSOLUTE

tree is E/W man is E/W man facing East tree stands in south man towards south

#

χ χ χ

χ χ

RELATIVE

man facing Right tree is R stick is to me

χ (2x) χ χ

INTRINSIC

man facing tree tree in front of man man behind tree man walks towards tree

χ (2x) χ χ χ

χ (3 χ)

Table 5. Picture 2.4.

Hai || om

Kgalagadi

ABSOLUTE

man facing W/E tree W/E tree stands in south man stands in north

# χ χ

x (2x) x

RELATIVE

you can see man's face man facing Left

x x (2x)

358

Sabine Neumann and Thomas Widlok

INTRINSIC

tree behind man tree at man's back man turning back to tree stick in front of body man and tree following each other man next to tree man facing tree

χ χ χ (3x) χ χ χ χ

Table 6. Picture 2.5.

Hai II om

Kgalagadi

ABSOLUTE

man facing E/W tree stands in Ν man towards Ν RELATIVE

man facing Left man at R tree at L man at my side INTRINSIC

man facing tree man face to face with tree tree in front of man man walks towards tree stick not between man and tree

x(2x) χ X

Table 7. Picture 2.6.

Hai || om

Kgalagadi

ABSOLUTE

tree is E/W tree stands in Ν man facing S/N man stands in S

χ χ χ χ

Universals of spatial language

man towards Ε man walks to the Ε

359

x(2x) χ

RELATIVE

man man man man man

has stick at (my) Right with back to us facing to you facing the front not facing to us

X X X x(2x) X X X

man facing where we are facing to man's face not seen INTRINSIC

tree at man's Left side tree at man's side man turning back to tree man facing tree man facing yonder (twds. nothing) man has Zero on his back

X X

X X X X

Table 8. Picture 2.7.

Hai || om

Kgalagadi

ABSOLUTE

man is EAV tree is E/W man (stands) in Ν tree stands in S man towards Ε

x (2x) x χ (2x) x χ

RELATIVE

man at Left man has stick at Left man facing back

x (3x) x x

INTRINSIC

tree at man's Left man next to tree

χ χ (2x)

x

360

Sabine Neumann and Thomas Widlok

man not facing tree stick not between man and tree

Table 9. Picture 2.8.

Hai II om

Kgalagadi

ABSOLUTE

man in E/W tree in E/W tree in Ν man facing N/S man towards W

x, # X X

RELATIVE

man to my Right man at Right tree at Left man facing us man facing to me man walks towards me

x(2x) x(2x)

INTRINSIC

man next to tree tree at man's side stick between man and tree

χ χ χ

The Tables that follow summarize the strategy of each pair of players for the whole set of pictures for the Hai || om sample (Tables 11 to 14) and for the Kgalagadi sample (Tables 15 to 18) respectively. Here a simple dot indicates whether the proposition in that row applied to the photo in question while a letter indicates that the proposition was actually made. In order to facilitate the mapping of the 'absolute', 'relative' and 'intrinsic' categories onto these Tables we have used the letters A, R, and I respectively. The results for the four Hai || om Director-Matcher pairs and the four Kgalagadi pairs can be summarized as follows:

Universals of spatial language

361

Table 10.

Hai || om total:

22 absolute

5 relative

19 intrinsic

Hai || om pair 1

2 A

2R

61

Hai || om pair 2

OA

2R

11 I

Hai || om pair 3

6A

1R

21

Hai || om pair 4

14 A

OR

01

Kgalagadi total:

15 absolute

27 relative

21 intrinsic

Kgalagadi 1

OA

10 R

31

Kgalagadi 2

1A

7R

5I

Kgalagadi 3

14 A

OR

61

Kgalagadi 4

OA

10 R

7 I

The results contrast with the fairly even distribution of 'absolute', 'relative' and 'intrinsic' propositions in terms of pictures discussed across pairs (see Table 3). The results in detail:

Table 11. Hai || om I 3

4

5

6

7

8

STANDING

man next to tree tree behind man man with back to us man in east tree in west

.1 .1

.1

.R .A .A

362

Sabine Neumann and Thomas Widlok

FACING

man facing tree man not facing tree man facing us

.1

.1

Position selected

2nd

1st

4th

3rd

5th

6th

3

4

5

6

7

8

.1

.1 .

.1 .1

.1 .R

Table 12. Hai || om II

STANDING

man next to tree tree at man's side tree in front of man man behind tree tree at man's back

.

.

.

.

.1 .1 .1

tree at man's left side man to my right man at my side

. .R

.1 .

.1

4th

5th

.R

FACING

man facing tree

.1

OTHERS

stick in front of body

Position selected

.1 1st

2nd

3rd

6th

Table 13. Hai || om III The Director puts himself into the position of the man, that is he discusses the position of the man in the first person singular.

Universals of spatial language

3

4

5

6

7

8

STANDING

man in the man in the tree stands tree stands

north west in west in east

.A A A A

DIRECTION

man man man man

towards the east walks to the east walks towards me walks towards tree

Position selected

.A .A R .1 4th

.1 5th

1st

3rd

2nd

6th

Table 14. Hai || om IV note that the matcher repeated information in each instance

STANDING

tree stands in south (lit.'river land') man stands in north (lit. land of soft sand) tree stands in north (lit. 'land of soft sand') man stands in south (lit. 'river land')

.A

.A

.A

.A

.A

.A

.A

DIRECTION

man towards south (lit. 'river land') man towards north (lit. 'land of soft sand') man towards east man towards west

.A .A

.A

364

Sabine Neumann and Thomas Widlok

Position selected

1st

2nd

3rd

4th

5th

6th

Table 15. Kgalagadi I Note that the players are seated opposite, the Director is facing N, the Matcher is facing S. 3

4

5

6

7

8

STANDING

man at left man at right tree at left

.R .R .R

FACING

man facing right man facing east man facing left

.R .R .R

man face-to-face with tree man turning back to tree

.R .1

.1 you can see man's face man's face not seen (you only see man's back)

I

.R

.R

OTHERS

man has stick at left Position selected

.R 3rd

5th

4th

6th

1st

2nd

Table 16. Kgalagadi II Again the players are seated opposite, the Director is facing North, the Matcher is facing South.

Universals of spatial language

3

4

5

6

7

365

8

STANDING

man at (my) left tree at man's left

.

1

.R ?

?

FACING

.A

man facing west man facing to you man facing the front (man has Zero on his back) man not facing to us man facing yonder man facing to me man facing tree man give tree back

.R .R .1 .R .R .R .1

.1 .1

OTHERS

.R

man has stick at right Position selected

6th

5th

2nd

4th

3rd

1st

Table 17. Kgalagadi III The players are seated opposite, the Director is facing South, the Matcher is facing North. 3

4

.A .A .A

.A

5

6

7

8

STANDING

man is east tree is west tree is right man is west tree is east tree in front of man

.A .A A . .1

.

.A

.A .A

FACING

man facing east man facing north man facing south

.A

.A .A .A

366

Sabine Neumann and Thomas Widlok

man facing tree man has given tree the back

.1 .1

OTHERS

stick between man and tree (stick not between man and tree)

.

.1

Position selected

5th

1st

.1

.

.1

2nd

6th

3rd

5th

Table 18. Kgalagadi IV The players are seated parallel. Director and Matcher are facing East.

STANDING

man at tree at left man at (tree's) left tree at man's left man and tree following

right

.R .R .

.R .R

. .

?

.1

.1 .1

?

FACING

man (and tree) facing left man (&tree) facing right .R man facing back man facing front man facing where we are facing to man facing tree .1 (face to face)

.R .R .R I

.1

.R I

2nd

4th

1st

OTHERS

stick is to me

.R

Position selected

5th

6th

JJniversals of spatial language

Notes *

1.

2.

3.

4.

Research with Kgalagadi-speakers was carried out by Neumann between 1992 and 1994. Neumann would like to thank the Office of the President of the Republic of Botswana for permission to conduct research in the Matsheng area and the community of Hukuntsi for their kind cooperation. Neumann's research was funded by the Cognitive Anthropology Research Group at the Max Planck Institute for Psycholinguistics in Nijmegen. Research with Hai || om-speakers was carried out by Widlok between 1990 and 1994. Field research in Namibia was made possible by a University of London Postgraduate Studentship, as well as financial support from the University of London Central Research Fund, the James A. Swan Fund of the Pitt Rivers Museum Oxford, and between 1992 and 1994 by the Cognitive Anthropology Research Group, Nijmegen. An earlier version of this paper was presented at die 19th International L . A . U . D . Symposium on Language and Space in Duisburg, 22-26 March 1994. We would like to thank David Wilkins, Gunter Senft and an anonymous reviewer for helpful comments in preparing this written version. We are also grateful to Gertie de Groen who did the computer graphics. Although linguists are more familiar with the notion of areal investigations, we use the term 'regional comparison', which is a wellestablished procedure in anthropology for controlled comparisons within a natural and socio-cultural region and which has inspired our contribution. An example for the wide-scope comparison is the analytic synopsis of work done by members of the Cognitive Anthropology Research Group on Left/Right concepts (Annual Report 1992: 93-100), although more recently other regional comparisons have also been carried out (Annual Report 1993: 87-9). Slobin (in press) and Pederson (1993) provide examples for controlled, but not primarily regionally controlled, comparisons while Levinson (1992) as well as Choi and Bowerman (1991) use comparisons based on an arbitrary selection of cases. Other tasks developed and run by the members of the Research Group are based on the same general idea but focus on other aspects of cognition, for instance inference instead of memory with regard to spatial constellations. For more details see Pederson (1992), the Working Paper series and the 'Manual for the Space Stimuli Kit 1.0' of the Cognitive Anthropology Research Group. For the purpose of this comparison we have not included results from these other tasks. It is only in a combination of tasks that more specific problems of each task can be discussed. One such problematic feature is the fact that in the Hai || om as well as in the Kgalagadi sample the 'animals in a row' task produced a large number of monodirectional responses in which subjects rebuilt the sequence in the same direction in all five trials. The game is part of the set of tools developed by the Cognitive Anthropology Research Group. It was designed by the Research Group, especially by Eric Pederson, Eve Danziger and Penelope Brown, and members of the group have elicited data with the help of this game from Longgu, Kilivila, Totonac, Mopan Maya, Yucatec, Tzeltal, languages of the Arandic family, Guugu Yimithirr, Tamil, Belhare and from the languages on which we will focus here: Hai || om and Kgalagadi.

367

368

5.

Sabine Neumann and Thomas Widlok

For our comparison we have counted tokens rather than quotes, that is to say if a proposition was uttered more than once in the course of one picture description it was only counted once. It should also be noted that Kgalagadi speakers were allowed to sit opposite each other, which they preferred to do. This introduces another factor that influences the results which should ideally be avoided. For the purposes of this particular comparison the difference in seating enforces our general point. Sitting opposite one another makes the usage of relative terms more difficult because rotation is required. One would therefore expect a tendency to use non-relative terms among the Kgalagadi speakers which is however not the case. Equally it could be argued that sitting next to one another supports relative descriptions. But the Hai || om speakers who were seated this way nevertheless used only very few relative propositions.

References Annual Report 1992

Annual Report of the Max Planck Institute for Psycholinguistics No. 13. Nijmegen. 1993 Annual Report of the Max Planck Institute for Psycholinguistics. No. 14. Nijmegen. Choi, S. and M. Bowerman 1991 Learning to express motion events in English and Korean: The influence of language-specific lexicalization patterns. Cognition 41: 83-121. Denny, P. 1978 Locating the universale in lexical systems for spatial deixis. Papers from the Parasession on the Lexicon, Chicago Linguistic Society, 14-15 April 1978. Chicago. Dirven, Ren6 1991 How languages structure thought: The case of prepositions. Paper held at the International Symposium on Language, Thought and Culture, 2-4 April 1991 at the University of the Witwatersrand. Fourie, David 1992 Cardinal directions in Ndonga, Kwanyama, Herero and Nama: A cognitive linguistic analysis. Afrikanistische Arbeitspapiere 28: 141-155. Heeschen, Volker 1982 Some systems of spatial deixis in Papuan languages. In: Weissenborn, Jürgen and Wolfgang Klein (eds). Here and There. Cross-linguistic Studies on Deixis and Demonstration. Amsterdam: Benjamins. Levinson, Stephen 1992 Language and cognition: The cognitive consequences of spatial description in Guugu Yimithirr. Cognitive Anthropology Research Group. Working paper No. 13. Nijmegen. Pederson, Eric 1993 Geographic and manipulable space in two Tamil linguistic systems. In: Frank, A. and I. Campari (eds). Spatial Information Theory. Berlin: Springer. Slobin, Dan (in press) Typology of rhetoric: Verbs of motion in English and Spanish. In: Shibatani, M. and S. A. Thompson (eds). Grammatical Constructions: Their Form and Meaning. Oxford: Oxford University Press.

Universals of spatial language

Taylor, John 1989

Linguistic Categorization: Clarendon.

Prototypes

in Linguistic

Theory.

369

Oxford:

Wilkins, David 1993 Proposal for outlining the systematic features of spatial description in the 'men and tree' data sets. Circular. Cognitive Anthropology Research Group. Nijmegen.

PartC Space as a bridge to other conceptual domains

Section 5 From one meaning to another

Polarity and metaphor in German Carlo Serra Borneto

1. The notion of polarity The notion of polarity in linguistics is often connected with the concept of (lexical) antonymy and sometimes, in an even more restricted sense, with words and phrases with respect to their negation, i.e. pairs as each/none, some/any, etc. Lyons (1968) notes that polarity has probably a more general origin which goes beyond the scope of linguistics (Lyons 1968: 469): The existence of a large number of antonyms and complementary terms in the vocabulary of natural languages would seem to be related to a general human tendency to 'polarise' experience and judgement - to 'think in opposites'.

Thus, antonymy mainly refers to oppositional lexical pairs such as long/short, good/bad, easy/difficult, etc., but polarity, as reflecting 'a general human tendency' in thinking, can also be applied to items which are not in such clear-cut opposition, but nevertheless are thought as being conceptually in competition, as for instance the German verbs liegen (approx. Eng.: 'to lie') and stehen (approx. 'to stand'), which do not represent an antonymous lexical pair in the narrow sense, although they apply to the same conceptual domain and mainly refer to diverging aspects of it (as will be shown in this paper). In an even more general sense it has been affirmed that polarity is at work at all linguistic levels, from phonology (distinctive features) to syntax (heads and modifiers), morphology and so on (Osgood and Richards 1973). There have been many attempts to define antonymy (and to a certain degree also polarity) in a propositional way (cf. Geckler 1980 for a survey). Cruse (1986: 205) proposes two different kinds of representation: a diagrammatic one and a schematic one, which incorporates the graded features of antonymy into a general schema. Since polarity basically represents a way of distributing conceptual information in a continuous array, the latter approach seems to be preferable. A schematic representation for polarity can be conceived by referring to a two or three dimensional space, but a

374

Carlo Serra Borneto

(vertical) axis is best suited, since "many properties can be conceptualized in terms of 'more' and 'less', thus creating a scale" (Cruse 1986: 204). Therefore, I propose to adopt a representation like Figure 1, which includes: (I)

An axis divided into two parts or fields by a protypically central discrimination point (zero point). In reality this point is not often identifiable, since, generally speaking, conceptual information appears to be symmetrically placed around a neutral region of the axis (Sapir 1944 refers to it as a 'zone of indifference') where neither of the polarity terms apply or both can apply (with some slight difference in meaning, as we will see later). The choice of the relevant terms in this area is likely to be subjected to imagery (intended as the "ability to construe a conceived situation in alternate ways - by means of alternate images - for purposes of thought or expression" [Langacker 1987: 110]).

(II)

Two poles at the extreme points of the axis, corresponding to the protoypical and most differentiated instances of the polarity and also to the most conventionalized ones:

UP

Λ

• • Dividing point

DOWN Figure 1. Schematic representation of polarity

Polarity and metaphor in German

375

There are also some other conditions to be satisfied in order to obtain a proper polar distribution of conceptual information: (a)

Conceptual information should be part of a common cognitive domain and share some feature or property: long and red or beef and car are not in polar opposition, whereas long and short, car and aeroplane, and even sometimes morphological features such as Accusative and Dative can be conceived of as being in polar opposition.

(b)

Conceptual information is distributed along the axis in a graded way: this is apparent for chains such as icy-cold-cool-lukewarm-warm-hot and the like, but applies also to 'pure' antonyms (not lexically gradable), as for instance long and short, whose range of reference is situationally and culturally conditioned (see also Cruse 1992: 297300).

(c)

Conceptual information is generally asymmetrically distributed, i.e. one part (or pole) of the axis is usually preferred. This shows up in the fact that one term in a pair is unmarked with respect to the other (for instance long, as opposed to short, can occur in so-called 'impartial' contexts, as in 'How long is this xx?' versus *'How short is this xx?'), or is more frequent, or reflects a more positive attitude etc. (cf. Lehrer and Lehrer 1982, Lehrer 1985). This feature is represented in Figure 1 by the dotted cone sorrounding the upper pole.

My aim in this paper is to show how this schema can be applied to some conceptual and orientational metaphors in German connected with the use of the so-called 'locational verbs' liegen and stehen.

2. The dimensional axes Lakoff and Johnson (1980) observed that vertical alignment is a common and pervasive way of organizing fundamental concepts in metaphorical terms. They consider UP-DOWN spatialization metaphors as the basic type of so-called orientational metaphors, which can be represented as in Figure 2 (Lakoff and Johnson 1980: 20, slightly modified):

376

Carlo Serra Borneto

Having control of force

UP

Experiential basis

Being subject to control of force

Figure 2.

DOWN

Schematic representation of orientational metaphors (adapted from Lakoff and Johnson)

Since vertically "enters our experience in many different ways" (Lakoff and Johnson 1980: 19), the experiential basis represents the link between the general schema of verticality and the instantiated metaphor. Thus, UP can be linked with widely different conceptual experiences, like MORE, HAPPY, HAVING CONTROL OF FORCE (as in the Figure), and DOWN with LESS, SAD, BEING SUBJECT TO CONTROL OF FORCE respectively, but they all share the same spatial schema as their organizing pattern. Johnson specifies further that the UP-DOWN axis is an instance of the schema of SCALE, which he defines as a modified PATH schema with a fixed directionality (upward), cumulative (piling up quantities) and normative (value and so on) character, thereby pointing out not only its quantitative but also its qualitative aspect (Johnson 1987: 121-124). Basic schemata (such as the schema of verticality) are linked with our bodily experience, but the different metaphorical instantiations related with them are culturally conditioned. We have in German a similar mapping of structured metaphors as in English for the UP-DOWN axis, although of course there are also some culturally based differences. In particular, German has two very productive (frequent in use) so-called 'locational verbs': stehen (roughly equivalent to to stand) and liegen (roughly=to lie) which denote the vertical and the horizontal axis respectively and can also be used for related metaphors. In their prototypical meanings, these two verbs refer to the shape or orientation of located objects, i.e. stehen applies to the lo-

Polarity and metaphor in German

377

cation of vertical objects, such as buildings, tables, vases and so on and liegen applies to horizontal objects, such as newspapers, records, and so on (for a general account of the semantics of liegen and stehen cf. Serra Borneto 1989 and 1995). If an object can be either vertically or horizontally oriented (such as people standing or lying, or a book on a shelf or on a table), the corresponding verb is used. The use of one of the two verbs in competition is almost obligatory in German when referring to the location of an object, i.e. you would hardly find the semantically neutral verb sein (= to be) in a locative sentence, where it is common in other languages, like English or Italian. In its prototypical uses the selection of liegen or stehen is determined by the shape or orientation of the located object, i.e. there is no overt reference to a vertical (or horizontal) axis along which different objects or experiences can be aligned. But, in fact, stehen and liegen can also refer to this kind of alignment and sometimes the two parameters (shape/orientation of the objects and vertical/horizontal alignment along a conceived axis) are in competition; in the latter case the more general parameter (the axis-parameter) predominates. Take for instance the following examples (for the sake of consistency, in this paper liegen and stehen will always be translated by the 'neutral' verb to be; the translation will therefore appear sometimes rather clumsy): (1)

Die Apfelsine liegt (*steht) auf dem Tisch. 'The orange is on the table.'

(2)

Der Ball liegt (*steht) auf dem Boden. 'The ball is on the floor.'

In (1) and (2) only liegen is acceptable. Oranges and balls are neither vertically nor horizontally salient; rather, their shape is symmetrical from the point of view of their extensions along these two dimensions. For this kind of object only liegen can be selected, since liegen applies not only to horizontal objects but represents also a sort of default case for dimensionally undetermined objects (cf. Serra Borneto 1995). The shape/orientation parameter is still relevant here, but it seems to be 'outplayed' in other cases, where 'symmetrical objects' trigger stehen: (3)

Die Sonne steht (*liegt) hoch am Himmel. 'The sun is high in the sky.'

(4)

Die Sonne steht (*liegt) schon sehr niedrig/tief. 'The sun is already very low/(deep).'

378

Carlo Serra Borneto

In (3) and (4) only stehen is acceptable. This could appear rather surprising, if we hold that 'symmetrical' objects can only trigger liegen. But even if the sun as such is certainly classifiable as 'symmetrical', the image at work in this case is completely different. Clearly, the sun is not seen as a normal object which is or can be located in one or another place, changing position or orientation, but is seen as strictly connected with its usual trajectory in the sky which is equated with an UP-DOWN axis. What stehen signals is that the sun is part of this vertical axis and, interestingly enough, its relative position on this axis (high or low) is irrelevant in this respect (so you do not find stehen for high, liegen for low position, as might be expected). From this point of view the shape/orientation parameter (which would require liegen) is overshadowed by the concurrent axis parameter (vertical axis = stehen). On the other hand, liegen conceptualizes the horizontal axis both in spatial and in temporal expressions (Examples are taken from a large corpus of newspaper sentences collected and kindly put at my disposal by the Institut für deutsche Sprache (IdS) in Mannheim): (5) Zwischen der Shaba-Hauptstadt Lumumbashi und Kinshasa liegen (*stehen) fast 2000 Kilometer. (IdS 25.06.86) 'There are almost 2000 Km between Kinshasa and the capital of Shaba Lumumbashi.' (6)

... liegt (*steht) dieser Tag vor dem 1. Januar, so muß das Arbeitsamt die Neufestsetzung bereits im Jahre 1986 vornehmen. (IdS 17.15.86) "... if this day is (=lies) before 1st January, the labour exchange has to carry out changes as early as 1986.'

In (5) no horizontally (or vertically) located object is implied in the scene; rather liegen refers to the conceptualization of a spatial line which connects the two towns. Similarly, (6) is an example of location on the temporal axis, which is generally considered a one-dimensional horizontal axis (cf. Clark/Clark 1978, Johnson 1987 among others). In this respect stehen and liegen conceptualize two different axes (vertical and horizontal) and their use is mutually exclusive (you cannot select liegen for the location of a vertically conceived axis and, conversely, it is not possible to select stehen for a horizontal axis). Since they also belong to the same cognitive domain and share common features (location, alignments), the notion of polarity seems to apply. Vertically and horizontally conceived axes can be understood as the two extremities of a polarity, instantiated by the lexical units liegen and stehen (note that you would not consider them antinomic in a strict sense), which could be represented as in

Polarity and metaphor in German

379

= Stehen

Figure 3. Polarity in liegen and stehen

It was already mentioned (with respect to Figure 1) that the dotted cone in this schema symbolizes an asymmetry in the distribution of the conceptual information. If applied to Figure 3, this would imply that stehen is 'more informative' than liegen, or, more technically, that it is unmarked (or less marked) with respect to liegen. Indeed, there is some evidence in favour of this hypothesis, especially if we invoke one of the most semantic and in some sense also 'cognitive' theories of markedness, the so-called 'Natural Morphology' (Mayerthaler 1981, Dressier 1990). Mayerthaler postulates an inverse relationship between markedness and naturalness: if χ is 'natural', χ is 'unmarked'; and if χ is 'marked', χ is 'unnatural' (Mayerthaler 1987: 27). But 'natural' is not a binary class ificatory predicate, rather it is a 'gradient predicate': "'p is natural' is simply a more convenient formulation of the explicit alternative 'p is more natural than q with respect to a fixed naturalness scale S'" (ibidem). Naturalness scales refer to human cognition and preferences within human cognition. For instance, MORE ANIMATE and LESS ANIMATE form a scale of semantic relation in which the first term (MORE ANIMATE) is more natural (and therefore less marked) than the second one. Mayerthaler lists a large number of such features (Mayerthaler 1981: 13-21), among others VERTICAL (seen as more natural) as opposed to HORIZONTAL (less natural and therefore cognitively

380

Carlo Serra Borneto

less preferred), CONCRETE (more natural) as opposed to ABSTRACT, BELONGING TO THE REAL WORLD as opposed to BELONGING TO OTHER WORLDS, POSITIVE as opposed to NEGATIVE, TOPIC as opposed to COMMENT and many others; all of them share the same character: the more 'natural' and 'cognitively preferred' term can be conceptualized in German by the verb stehen, whereas the less preferred is conceptualized by liegen (for a detailed discussion of this point cf. Serra Borneto 1989: 371-378). Therefore, at least from the point of view of Natural Morphology, it is possible to state that there is an asymmetry in the distribution of the conceptual information conveyed by the lexical pair stehenlliegen, in the sense that one term (stehen) is more natural and cognitively preferred than the other one (liegen), as symbolized in Figure 3. Of course, some additional characteristics are implied by the notion of polarity as it was defined above, such as the graded distribution of information, a central neutral area with respect to the specificity of the lexical units and the possibility of construing images in different ways. The rest of this paper is devoted to the illustration of these points with special reference to the metaphorical usages of stehen and liegen.

3. The poles of the axis There are many examples of metaphorical distribution on the UP-DOWN axis with stehen, which I have already illustrated elsewhere (Serra Borneto 1995). I will concentrate here on one particularly interesting case. Lakoff/ Johnson (1980: 15) consider control of force as an instantiation of an orientational metaphor, whose poles are: HAVING CONTROL OF FORCE IS UP; BEING SUBJECT TO CONTROL OF FORCE IS DOWN. In German there are numerous examples of this kind of metaphor too; for instance: (7)

Landeshauptmann Silvius Magnago (72), der seit 1961 an der Spitze der Partei steht (*liegt). (IdS 08.04.86) 'Governor S. M. (72), who led (lit.: is at the top of) the party since 1961.'

(8)

Die Konferenz steht (*liegt) unter der Schirmherrschaft des Bayrischen Staatsministers P.S. (IdS 15.05.86) 'The conference is under the auspices of the Bavarian state minister P.S..'

An der Spitze (=at the top) and unter (=under) respectively signal the metaphorical location of the controlling instance (UP) and the controlled

Polarity and metaphor in German

381

one (DOWN); this metaphor was symbolized by Lakoff/Johnson in Figure 2 . Again, stehen appears in both instantiations of the metaphor, irrespective of the trajector's position (high or low), as was the case for the sun in (3) and (4). This is a clue that also in this case stehen conceptualizes the whole metaphorical (vertical) axis as such and not the shape/orientation or even the position of the objects located on it. Since the relevant conceptualization refers to the axis and not to the shape/orientation or the position of the objects, it is possibile to conceive a sort of recursive application of the same metaphorical pattern. Take, for instance, the UP-DOWN metaphor as a whole (as represented in fig. 2 ) and consider it as a pole of a larger metaphor of CONTROLLING; then the metaphors presented SO far (HAVING CONTROL OF FORCE IS UP; BEING SUBJECT TO CONTROL OF FORCE IS DOWN) will be an instantiation of the upper pole of the more general (meta)metaphor of CONTROLLING, giving rise to the notion of EXERTING CONTROL IS UP (which includes metaphors like [ 7 ] and [8]). In this case new metaphors (see below), where the notion of force is missing, can be attributed to the lower pole (LACK OF CONTROL IS DOWN). Figure 4 symbolizes this situation:

DOWN

Figure 4. Schema of the (meta)metaphor of

CONTROLLING

382

Carlo Serra Borneto

Both metaphors (7) and (8), implying control, are (metaphorically) located on the UP-pole (irrespective of their internal distribution with the controller on the UP-pole and the control lee on the DOWN-pole of the small box), whereas metaphors implying lack of control are located on the DOWN-pole. Thus, the more general metaphor (metametaphor?) of CONTROLLING constructs a new UP-DOWN axis (big axis in the figure) which is entirely conceptualized by stehen, as in the case of the vertical trajectory of the sun in (3) and (4). German has several examples of this kind of metaphor: (9)

... in einer Erklärung der FDA heißt es, den Ärzten stehe (*liege) es frei, Alphainterferon Patienten zu geben, die an sogenannter Haarzell-Leukämie leiden. (IdS 06.06.86) '... a statement by the FDA says that doctors are free to give alphainterferon to patients suffering from so-called hairy-cell leukemia.'

(10) ... wenn ja, dann stehen (*liegen) ihnen dafür zwei Wege offen. Sie können auf Antrag versicherungspflichtig werden oder aber freiwillig beitreten. (IdS 19.04.86) '... if yes, then two paths are open to them; they can apply for compulsory insurance or join voluntarily.' In (9) the FDA organisation does not exert ultimate control over the doctors (who are free to prescribe the medicine or not, according to their conscience); nevertheless some sort of control is still operative, insofar as the FDA is allowing the doctors freedom; in (10) the LACK OF CONTROL metaphor is even more evident, since people have the total freedom to choose between two alternatives (two paths). In both examples the presence of stehen as a locative verb signals the vertical axis along which the LACK 1 OF CONTROL metaphors are aligned. This is confirmed by the fact that in (10) stehen co-occurs with Wege (=paths): in non-metaphorical perceptual scenes, in fact, Weg is not compatible with stehen because it is conceived as lying on the horizontal axis rather than on the vertical one, as happens in the conceptualization of physical space: (11)

Der Weg zum Berg liegt (*steht) hinter dem Wald. 'The path to the mountain is beyond the forest.'

The path in (10) is instead metaphorical (and therefore not subject to the constraints of a real perceptual scene), and since the whole scene can be considered an instantiation of the (meta)metaphor of CONTROLLING which implies a vertical axis, stehen is not only accepted but represents the only possible instantiation.

Polarity and metaphor in German

383

As for liegen, we already saw in (6) an example of metaphor concerning the dimension of time which is commonly intended as an extension of the linear horizontal axis. Another example of this kind is: (12)

...es kann allerdings noch einige Zeit dauern, bis die städtischen Autos bleifrei fahren, denn ihre Lebensdauer in den Diensten der Stadt liegt (*steht) bei sieben bis zehn Jahren. (IdS 15.10.87) '... in any case, it could take more time before city council cars will use unleaded petrol, because their life span is between seven and ten years.'

In this case there is a reference to a time span rather than to the location of a point on the temporal axis. Nevertheless, liegen is still used as a signal of the distribution of time conceived on a horizontal axis. Similarly, both geometrical and numerical entities are also located on the horizontal axis, as is widely recognized; consequently, they require liegen too: (13)

Der Punkt liegt (*steht) auf der Diagonale. 'The point is on the diagonal.'

(14)

Liegen (*stehen) die Geraden g, h, k auf einer Ebene ... 'If the lines g, h, k are on one plane

(15)

Die Zahl der Selbsthilfegruppen insgesamt liegt steht) zur Zeit bundesweit bei rund 20000. (IdS 08.01.88) 'The number of self-help groups is at the moment about 20,000 in the whole German Republic.'

(16)

... wie schon berichtet, liegen (*stehen) die Meßergebnisse in Mannheim vom 4. Mai bei 100 BQ per liter. (IdS 07.05.86) ' . . . a s already reported, the results of the measurements in Mannheim on May 4th are 100 Bq per litre.'

There is, of course, a difference between the geometrical domain and the domain of numbers: geometry is a matter of space and therefore subject to the constraints of spatial processing, even in the more abstract cases (as for instance point/line relationships). Numbers are non-spatial entities and therefore can only be metaphorically understood in spatial terms. Thus, a horizontal axis in geometry is basically concrete and has a reference in the real world, whereas a horizontal axis in the domain of numbers corresponds to a conceived distribution of numerical entities on a metaphorical axis. Since the metaphorical alignment is less concrete, so its orientation is less

384

Carlo Serra Borneto

compulsory, i.e. numerical entities can also appear aligned (or at least partially aligned) on another axis (more specifically on the vertical one) in particular circumstances. For instance in the conceptualization of scalar measurements connected with instruments usually seen as vertically oriented, as thermometers and similar. Consequently a conflict can arise in the selection of the corresponding lexical units: the two locational (or 'axes signalling') verbs stehen and liegen are sometimes used alternatively in competing metaphorical settings. In the next section I will try to show their distribution with respect to this issue.

4. The central area of the axis In the previous section, we have seen how stehen is selected for metaphorical processing of the UP-DOWN axis (with particular reference to the metaphor of CONTROLLING) and liegen for metaphorical processing of a linear horizontal axis (especially the time axis and the alignment of numerical quantities). In this section, we deal with more complex scenes which share features of both kinds of processing and are therefore less binding as far as the selection of the verb is concerned. Before going over to this kind of phenomenon, I would like to mention some more metaphors uniquely connected with stehen. They refer to a kind of cultural value or judgement about the situation depicted: (17)

... ist es denn faßbar, daß ihnen Waffenhandel und Kommerz höher stehen (*liegen) als das Leben ihrer Mitbürger? (IdS 02.01.86) '... is it then conceivable that for them arms dealing and trade are higher (=more important) than the lives of their fellow citizens?'

(18)

...um die Zahngesundheitserziehung stehe (*liege) es 'miserabel'. (IdS 07.02.86) '... dental health education is in a very bad way.'

(19)

Dieser Anzug steht (*liegt) dir gut. 'This suit suits you (lit.: is/stands well to you).'

Metaphors of this sort involve values (a moral assessment, an aesthetic appraisal, an evaluation of qualities, and so on) and, although connected with the general metaphor MORE is UP/LESS IS DOWN (cf. Lakoff and Johnson 1980: 22-24), they show a higher degree of metaphoricity than the basic instances. As Johnson (1987) notices, there are certain correlations of structures in our experience which give rise to metaphorical projections in

Polarity and metaphor in German

385

which quantitative and qualitative aspects are intertwined. MORE IS UP/LESS IS DOWN is obviously a metaphor of quantity but very often qualitative aspects are also strongly implied in it: "Having more or less of something may be either good or bad, desirable or undesirable" (Johnson 1987: 123). Johnson maintains that this kind of metaphor, implying both a quantitative and a qualitative aspect, is based on a more general schema: the SCALE schema. This schema shows a set of features which are worth mentioning in our context: (a) a SCALE schema has a fixed directionality (like a vertical MORE IS UP/LESS IS DOWN axis where the increase in quantity corresponds to an increase in height along the axis); (b) it has a cumulative character (which implies continuity); (c) it has a normative character (implying values, qualitative judgements and so on). These features are perfectly compatible with polaritiy, as defined above. Johnson (1987: 123) notices further: One other aspect of SCALES that is extremely important in our culture is the possibility of imposing numerical gradients on a scale. We can calibrate scales in terms of regular, discrete fixed units of measurement.

This point is extremely important for us, since we have seen that, on the contrary, numerical entities (and therefore also measures related with numbers) are prototypically conceived as lying on a horizontal axis (the 'natural' axis for numerical alignment) and not on a vertical. On the other hand, we have just observed that the MORE IS UP/LESS IS DOWN metaphor is conceived as an instance of verticality. Apparently there are two facets of the domain of quantity (more/less and number/measurement) which are conceived differently, the one belonging to the schema of verticality, the other to the schema of horizontality. This could give rise to conflicts, especially if, as in German, there are specialized linguistic units (like stehen and liegen) for each of the two dimensions. Johnson (1987) maintains that the SCALE schema has a 'fixed directionality' but he does not specify how this directionality is oriented, although he clearly seems to prefer the vertical alignment. Therefore, for a certain number of complex metaphors, where an entity is conceived as being located on a scale, we can postulate that both axes are somehow implied, suggesting that a complex, somehow 'mixed' schema could be at work here representable in Figure 5 (numerical entities are represented only on the horizontal, more 'natural' axis):

386

Carlo Serra Borneto up located entity L ^ /I—•I—I I I

numerals

down

Figure 5. Competing axes in complex metaphors

Of course, if both axes are implied in a conceived situation, the selection of the specialized verb will not always be straightforward: sometimes it will be fully sanctioned in favour of one verb (stehen versus liegen), sometimes subject to imagery construal (both stehen or liegen). Yet the selection is not arbitrary; rather it is a matter of motivation. In particular, if one of the axes is conceived as predominant, the corresponding verb will be chosen. In other words, if the vertical alignment is conceptually emphasized (or focussed), stehen will be selected; if the horizontal alignment is emphasized, liegen will be selected. In Figure 6 this alternative is represented: up

up

located entity

located entity

down 6a (more vertical)

down 6b (more horizontal)

Figure 6. Axis predominance in complex metaphors

Polarity and metaphor in German

387

Since the MORE IS UP/LESS IS DOWN metaphor is clearly vertically oriented, all instances of this metaphor will be coded by stehen. In particular, a quantity metaphor with a qualitative character will be included in this group, since the more/less polarity is strictly connected with qualitative judgement (cf. Johnson's quotation above and examples 17-19). This would correspond graphically to the representation in Figure 6a. On the other hand, if the quantitative metaphor shows distinct numerical character, liegen is expected. This corresponds to the representation in Figure 6b. The intermediate cases will depend on the emphasis attributed to the metaphorical scene. One could therefore formulate the following generalization: 'The more the speaker wants to stress the quantitative/numerical aspect of a metaphor, the more likely he will select liegen; the more he wants to stress the qualitative aspect of the metaphor, the more likely he will choose stehen'. Of course, this generalization has not the import of a formal rule but represents a motivational guideline. The rest of this section is devoted to the illustrations of this hypothesis. Numerical quantity is conceived as being aligned on the horizontal axis, thus 'triggering' liegen: (20)

... benutzten 1970 noch 21000 Berufstätige Busse und Bahnen, so liegt (*steht) ihre Zahl heute nur bei 10500. (IdS 15.10.87) '... if in 1970 21,000 working people still travelled by bus and train, today they number only 10,500.'

In this example the word 'number' (Zahl) is mentioned explicitly. Exact measures are also expressed through numerical entities: (21)

... die Werte liegen (*stehen) bei 250 Bequerel je Kilogramm für Jod 131 und 100 Bequerel je Kilogramm Gemüse für Caesium 137. '... as for iodine 131, figures are at 250 Bequerel per kilogram and for cesium 137 they are 100 Bequerel per kilogram of vegetables.'

So far, the vertical axis does not come into play explicitly. The following examples are different because quantity is implied not only in terms of numerical entities but also in terms of MORE or LESS. Since MORE and LESS are conceived on the vertical axis, liegen (still triggered by numbers and measures) co-occurs with lexical items which signal the implied presence of a vertical dimension (such as above, lower, below etc.):

388

(22)

Carlo Serra Borneto

... bei der Dresdner und der Commerzbank liegt (*steht) diese Zahl zwar niedriger, aber am internationalen Standard gemessen, zählen auch sie zur Spitze. (IdS 23.04.88) '... as for the Dresdner and Commerzbank this figure is lower, but both banks are still among the top institutions if compared with international standards.'

(23)

Beispiele für Fachausdrücke aus Physik und Technik sind Ultraschall: 'Schall, dessen Frequenz oberhalb der menschlichen Hörgrenze liegt (*steht).' (IdS 1/12/88) 'Examples of terminology in the fields of Physics and Technology are: ultra-sound = a sound whose frequency is above the audiofrequency range.'

(24)

... alle Werte liegen (*stehen) deutlich unter 0,4 Bequerel (Bq) pro Quadratzentimeter, dem Grenzwert für Gemüse. (IdS 07.05.86)

'... all figures are clearly below 0,4 Bequerel (Bq) per square centimetre, which is the accepted limit for vegetables.'

Examples of this kind are plentiful and all require the use of liegen (stehen is not allowed). This means that the horizontal alignment induced by the numerical quantity still predominates (the evoked scheme is in Figure 6b) and the degree of metaphoricity is still comparatively low. A different situation arises when the linear ordering of instances is not determined by absolute figures (numbers or measures as above) but by a relative position of an instance with respect to others (for instance: χ is in first or second place among other runners). In this case numerical alignment co-occurs with rank, which has strong qualitative implications, at least in our cultures (first is 'good', arriving in last position is 'bad') and therefore trigger metaphors with higher degree of metaphoricity. The two competing parameters are of comparable strength, so that both verbs are alternatively allowed:

(25)

Der Weltmeister liegt (*steht) in Führung.

(26)

...so liegt das Land - von der Kaufkraft her gesehen - denn auch an dritter Stelle unter den deutschen Flächenstaaten. (IdS 28.

'The world champion is leading.'

04.88) "... thus, this region is in third place among German federal states from the point of view of buying capacity.'

Polarity and metaphor in German

389

(27)

... und schon jetzt steht die Bundesrepublik mit dem Wiedergebrauch von verschmutztem Wasser ('Recycling') an erster Stelle in der Welt. (IdS Ol. 12.88) '... even now the Federal Republic is in first place in the world in water recycling.'

(28)

... in westlichen Ländern steht sie als Todesursache hinter Herzund Kreislauf-Erkrankungen, Krebs und Schlaganfall an vierter Stelle. (IdS 10.06.88) '... it is in fourth place among the causes of death in Western countries after heart attack, cancer and stroke.'

Native speakers, if asked about these examples, maintain that (25) is only possible with liegen, whereas the other examples are acceptable with both verbs. Often they do not 'feel' any major difference between the two variants, although some of them believe that the liegen constructions give you the impression that the described situation is provisional, more dynamic; stehen constructions seem to be more resultative, more fixed and static. For instance in (25) the first position of the world champion in the race is not to be considered definitive because he can be overtaken any time by other contestants, whereas the fourth place in (28) is seen as definitive as the result of a statistical study. This can be interpreted as follows: during a contest the positions of the participants are seen as horizontally aligned because the relevant factor is their linear sequence, any evaluation in terms of ranking is premature (liegen is preferred). On the other hand, final positions already imply an evaluation in terms of ranking or even rating and it is more 'value-loaded' (stehen is preferred). For the same reason liegen is associated with a dynamic interpretation of the scene, whereas stehen evokes a more resultative and definitive setting. There are two interesting examples which support this hypothesis. The first one concerns again the expression of position (an erster Stelle = in first place) where only stehen is allowed. For instance:. (29)

An erster Stelle steht (*liegt) für mich die Ehrlichkeit. 'For me honesty comes first (lit.: is in first place.).'

The moral relevance of this statement is evident (it is extremely 'valueloaded') 2 . Furthermore, the whole situation is conceived as static, insofar as the scale of value is not likely to undergo rapid changes as would be the case in some contexts. The other interesting case concerns the use of liegen in a context in which we have already met stehen. In order to facilitate the comparison, I

390

Carlo Serra Borneto

will repeat the relevant example with stehen (7) along with the alternative with liegen: (7)

Landeshauptmann Silvius Magnago (72), der seit 1961 an der Spitze der Partei steht (*liegt). (IdS 08.04.86) 'Governor S. M. (72), who led (lit.: = is at the top of) the party since 1961.'

(30)

... auf manchen Gebieten liegen (*stehen) die Forscher der MP wieder an der Weltspitze. (IdS 15.07.87) 'In some fields MP researchers again lead the world (lit.: are again at the top).'

In both examples the locational verb co-occurs with the prepositional expression an der Spitze (= at the top). I classified (7) as an instance of the HAVING CONTROL OF FORCE IS UP metaphor, which requires stehen. Although the same expression an der Spitze also appears in (29), the sentence is not an instance of the CONTROL metaphor; researchers are often seen as competing with each other in order to reach the top in their research field. This position is not stable but always endangered by the struggle for primacy which permeates the scientific community. The metaphor of competition connected with instability justifies the use of liegen in this case. There are many other domains in which the alternative stehen!liegen plays a certain role. My last examples will be about the financial domain. Prices, currencies and so on are obviously connected with numerical entities (liegen); nevertheless, one can find very similar sentences with both verbs: (31)

... so lag der D-Mark-Kurs gegenüber dem Franc bei 3,3850 Francs. (IdS 14.06.88) ' . . . so the DM exchange rate was 3 . 3 8 5 0 against the franc.'

(32)

Der Dollar stand bei 1,78. 'The dollar was at 1.78.'

(33)

Der Preis der jungen Aktien liegt weit über pari. (IdS 15.10.87) 'The price of the new shares is clearly above par.'

(34)

Die Aktie steht gut. 'The shares are in a positive trend.'

Polarity and metaphor in German

391

There is no relevant difference between (31) and (32); again (31) gives you a flavour of less stability, whereas (32) suggests a more rigid, 'fixed' situation (probably the state of the currency after fixing). (34), as compared with (33), is more 'value-loaded' (g«i=positive trend; with a higher degree of metaphoricity), whereas (33) is neutral and only implies location on an imaginary axis. In all these latter cases the influence of imagery is particularly significant.

5. Conclusions Lexical alternation reflects imagery ('the ability to construe a situation in alternate ways for purposes of thought or expression'), if it refers to the conceptualization of very similar scenes or even the same scene. Stehen and liegen are basically locational verbs: they indicate that something is located somewhere and often refer to similar scenes. Nevertheless, their specific function is to emphasize a particular aspect in the scene, since they refer to different images ('expressions are often functionally equivalent but nonetheless different in meaning by virtue of the contrasting image they convey' [Langacker 1987: 111]). Therefore, they share differentiation and commonalities. Difference and similarity are features we also attributed to the notion of polarity. Differentiation concentrates around the poles and marks the most distinct and fully conventionalized uses of each lexical unit (often the prototypical meanings at the same time). For stehen they coincide with pure vertical location, for liegen with pure horizontal alignment. In between there are many possibilities of conceiving a locative scene, especially if the location assumes metaphorical character. The choice of the verb depends on how the scene is conceptualized and on the features selected for explicit attention. The more a scene is conceptualized towards the differentiated (prototypical) senses of one of the competitive verbs, the more the use is likely to be sanctioned (conventionally fixed through language). In other words, there are cases in which an alternative is not allowed, where the use is fully conventionalized (you can only use one lexeme and not the competing one), and there are situations which allow an alternative construal. Since polarity is a graded phenomenon, it is not possible to fully predict these cases, but - relying on motivation - one can say that the more a situation can be interpreted as an instance of prototypical use (or meaning), the more the form is likely to be fully conventionalized·, and the more the situation is complex, not clear-cut, not prototypical, the more it will be possible for the speaker to choose different linguistic means in order to express some facets of this complexity instead of others. I think the notion of polarity can be of some interest in this respect, because it helps us to

392

Carlo Serra Borneto

find out the areas which are likely to be subject to alternate construal. Figure 7 depicts, for instance, the distribution of the locative verbs on a polarity axis:

Polarity axis

Äreet

Sieben

Area of Liegen

Figure 7. Stehen and liegen on a polarity axis

The conceptualization in terms of pure verticality is symbolically located in the upper part of the area of stehen-, here the use of stehen is fully conventionalized (for instance examples 7-10 in this paper). In the lower part of the same area, conceptualizations are located which still mainly emphasize the vertical axis but also involve some sense of horizontality (example 29). In the area of imagery both axes are equally represented: the choice of the verb depends entirely on the intention of the speaker to stress one aspect or the other (examples 25-28). The upper part of the area of liegen contains images with a preference for horizontal alignment although verticality is still present in the background (examples 22-24). The lower part of the same area symbolizes the conventionalized examples in which only horizontality is conceived (13-16). As I have already mentioned, the schema should be intended as a continuum with no specific boundaries be-

Polarity and metaphor in German

393

tween symbols and areas. A very close analysis would probably show more perspicuous differences and allow us to organize the relationships among the various meanings in the form of a network.

Notes *

1.

2.

This is a companion paper to Serra Borneto (1995). Most of the data are common to both papers but are presented and discussed under different perspectives. I wish to thank Rend Dirven and an anonymous referee for their precious suggestions. More precisely the two verbs in (9) and (10) are freistehen and offenstehen respectively. As in all verbs of this type, prefix and base (stehen) interact semantically: the prefixes reinforce the metaphorical sense of LACKING OF CONTROL, stressing the aspect of freedom and openess implied in it. Therefore, stehen cannot be considered an independent and pure locative verb in these examples. Interestingly enough the two prefixes (frei- and offen) can occur only in combination with stehen and not with liegen (*freiliegen doesn't exist and offenliegen has a very restricted use in the language of administration: Die Pläne werden ab Juni im Rathaus offenliegen = 'Starting from June, the plans will be at disposal in the town hall'). There are exceptions in this kind of metaphor. Ren6 Dirven (personal communication) mentions some expressions of free choice occurring with liegen, for instance: Es liegt in meiner Gewalt, das zu tun, or: Die Wahl liegt bei Dir. In these cases there is no lexical reinforcement of the verb semantics (for instance in the form of prefixes, as above) and the meaning remains purely locative. Therefore a competition between liegen and stehen may arise (for this kind of competition and a partial explanation of the above mentioned exceptions cf. Serra Borneto, 1995, § 5.6). Again this example shows a higher degree of metaphoricity than for instance (26). Ren6 Dirven (personal communication) points out a correlation between a low degree of metaphoricity (which is connected with liegen) and a higher degree of metaphoricity (connected with stehen). The reason for this distribution (which is consistent with most of the examples presented in this paper) is not clear, at least not to me. This is in any case a very promising subject for further research.

References Clark, Herbert and Eve Clark 1978 Universale, relativity and language processing. In: Greenberg, Joseph (ed.). Universals of Human Language, vol 1. Stanford: Stanford University Press, 225-277. Cruse, David A. 1986 Lexical Semantics. Cambridge: Cambridge University Press. 1992 Antonymy revisited: Some thoughts on the relationship between words and concepts. In: Lehrer, Adrienne and Eva Kittay (eds). Frames, Fields, and Contrasts: New Essays in Semantics and Lexical Organization. Hillsdale: Erlbaum, 289-306.

394

Carlo Serra Borneto

Dressler, Wolfgang 1990 The cognitive perspective of 'naturalist' linguistic models. Cognitive Linguistics 1-1: 75-98. Geckler, Horst 1980 Die Antonymie im Lexikon. In: Kastovsky, Dieter (ed.). Perspektiven der lexikalischen Semantik. Bonn: Grundmann, 42-69. Johnson, Mark 1987 The Body in the Mind. Chicago and London: The University of Chicago Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar. Vol 1: Theoretical Prerequisites. Stanford: Stanford University Press. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago and London: The University of Chicago Press. Lehrer, Adrienne and Keith Lehrer 1982 Antonymy. Linguistics and Philosophy 5: 483-501. Lehrer, Adrienne 1985 Markedness and antonymy. Journal of Linguistics 21: 397-429. Lyons, John 1968 Introduction to Theoretical Linguistics. London and New York: Cambridge University Press. Osgood, Charles and Meredith Richards 1973 From Yang and Yin to And and But. Language 49: 380-412. Mayerthaler, Willi 1981 Morphologische Natürlichkeit. Wiesbaden: Akademische Verlagsgesellschaft Athenaion. 1987 System-independent morphological naturalness. In: Dressier, Wolfgang (ed.). Leitmotifs in Natural Morphology. Amsterdam/Philadelphia: Benjamins, 25-58. Sapir, Edward 1944 Grading: A study in semantics. Philosophy of Science 11: 93-116. (also in: Mandelbaum, David [ed.]. Selected Writings of Edward Sapir in Language, Culture and Personality. Berkeley/Los Angeles 1951: 121-149). Serra Borneto, Carlo 1989 Per un'analisi cognitivista dei verbi stehen e liegen in tedesco. Studi Italiani di Linguistica Teorica eApplicata 18: 343-382. 1995 Liegen and stehen in German. A study in horizontality and verticality. In: Casad, Eugene (ed.). Cognitive Linguistics in the Redwoods. Berlin/New York: Mouton de Gruyter, 459-505.

Metaphors of 'total enclosure' grammaticizing into middle voice markers* Susan Strauss

1. Introduction This paper will examine a phenomenon in which lexical verbs in both Japanese (shimau) and Korean (petita) (as well as in other languages) have grammaticized into auxiliary verbs (-te shimau and -ale petita) whose function strikingly resembles that of middle constructions. Other than Strauss (1994a, 1994b), no researcher has, to my knowledge, previously linked Japanese -te shimau or Korean -ale petita to middle voice. It will be shown that this function as a middle has developed as a result of the metaphor of 'total enclosure' stemming from the basic meanings of the verbs as lexical verbs. A comparative and contrastive analysis will also be made between these auxiliaries in Japanese and Korean with the 'Romance' reflexive or middle voice marker in Spanish. Finally, it will be demonstrated that Japanese and Spanish share many more characteristics with respect to grammatical capabilities and pragmatic usages of their respective constructions than do the typologically related Japanese and Korean.

2. Japanese shimau and Korean petita as lexical verbs — 'total enclosure' As a lexical verb, shimau in Japanese means 'to put away' or 'to finish (something)', as in examples (1) and (2), taken from the (1990) Koojien dictionary:

(1)

shigoto

ο

shimau

work OBJ finish 'to finish or stop working'

finish

396

(2)

Susan Strauss

omoide ο mune ni shimau memory OBJ heart LOC put away 'to store a fond memory in one's heart'

put away

Ρelita, as a lexical verb in Korean means 'to spoil' or 'to throw away', as illustrated by examples (3) and (4), from Lee (1990): (3)

ku-hun sinpal-ul

peli-ess-ta

he-TOP shoes-OBJ

spoil-pst

spoil

'He spoiled the shoes' (4)

na-nun hen tti-lul

peli-ess-ta

I-TOP

throw-pst

old belt-OBJ

throw away

Ί threw away an old belt' Both shimau and pelita are at least 2-participant verbs, taking minimally an agent and patient or object (also allowing an optional locative). 1 Also, as can be seen in examples (2) and (4), both verbs imply some type of 'total enclosure' of the object in question, whether the locative is expressed or not, e.g., 'to put something away (in some special place)' or 'to throw something away (into the trash can)'. Previous research on shimau and pelita touch upon this notion only lightly. As a result, these previous accounts fall somewhat short of being able to provide a comprehensive explanation of the underlying semantic messages of these verbs as lexical verbs, and hence as auxiliaries as well. In the case of shimau, very little importance seems to have been placed on its underlying semantic properties as a lexical verb. For example, Ono (1992: 369) basically limits his discussion to the fact that the lexical verb shimau "implies that when the thing has been put away, it is difficult for one to have access to it", and Ono and Suzuki (1993) simply state that shimau influences the auxiliary construction -te shimau in that it focuses on the "endpoint of the activity". With regard to Korean, Lee (1990: 300; 1993) invokes the 'container' metaphor in his discussion of Korean auxiliary verbs in general, indicating that "bounded space is a container" and a certain object can be either outside or inside of that bounded space. According to Lee (1990; 1993) the lexical verb pelita, 'to throw away' designates an event which is defined on the basis of this inside/outside dichotomy; pelita "denotes that the subject causes the objects to go out of his sphere of possession" (1990: 302). Lee extends this concrete notion to include abstract situations, including states, such that the meaning 'to spoil' can be derived from the same dichotomy: the object is either "in (or out of) the state of usability". In the case of

Metaphors of 'total enclosure'

397

pelita, the object would be considered out of the state of usability. Example (5) represents Lee's (1993: 238) conceptual schema for pelita: (5)

a. before pelita

b. after pelita

According to this schema, χ represents the object to be discarded, and Y is the source location of x. In (5a), χ is located within the sphere of Y. In (5b), χ is outside of the sphere of Y. Section 3.2. below will show how Lee (1990; 1993) applies this notion to the auxiliary function of -ale pelita.

3. Previous accounts of the auxiliaries -te shimau and -a/e pelita:

3.1. -te shimau The auxiliary -te shimau has been accounted for in reference grammars and linguistic writings as a marker of terminative aspect (e.g., Soga 1983), in addition to a grammatical means of indicating irreversability, the lack of control over a particular situation, the automatic or spontaneous occurrence of an event or situation, the speaker's reaction to some unexpected happenstance (e.g., Alfonso 1966, Martin 1975, Teramura 1982, Soga 1983, Makino and Tsutsui 1986), as well as a means to convey the notions of completion and totality (Martin 1975, Teramura 1982, Soga 1983). From a discourse analytic perspective, -te shimau has more recently become the subject of a number of studies using actual discourse data, the focus of which has been primarily on grammaticization (Ono 1992, Ono and Suzuki 1993) and subjectivity (Yoshida 1994a, 1994b). These studies all consider -te shimau as a means of expressing the speaker's attitude, e.g., frustrative (Ono 1992), speaker's negative attitude (Ono and Suzuki 1993), and violated expectations (Yoshida 1994a, 1994b). Ono and Suzuki (1993) propose a grammaticization model which attempts to illustrate how four

398

Susan Strauss

pragmatic meanings of the auxiliary -te shimau have developed from the meaning of the lexical verb shimau ('to put away/finish'). The four pragmatic meanings are nearly identical to some of the meanings already proposed by the traditional linguists as noted above: "inability to undo", "automatic", "speaker's negative attitude", and "speaker's guiltily positive attitude". Moreover, according to Ono and Suzuki (1993), all meanings associated with the -te shimau auxiliary are predominantly negative. According to Yoshida (1994a, 1994b), all instances of -te shimau are motivated by some type of violated expectation. 2 Common to both studies is the representation that -te shimau is a strictly optional verbal morpheme and, as such, indicates some type of affective stance on the part of the speaker; hence, no usage of this auxiliary involves an objective or neutral description. Example (6), reproduced below, which appears as Ono and Suzuki's (1993) example (12), represents an instance of the "automatic" meaning of -te shimau: (6) harawa-nakute mo ii η daroo kedo haratte shimau to pay-not EMPH good NOM guess but pay SHIMAU PTL '(I) guess (I) don't have to pay, but (I find myself) paying' Examples (7) and (8) (from Yoshida 1994a — examples (4) and (11) respectively) illustrate what Yoshida refers to as "violated expectations" on behalf of the speaker. Example (7) is from a story recall task which recounts in Japanese (Clancy 1980) the plot of a six minute film called 'The Pear Film' (Chafe 1980). (7) sono jitensha ni notta otoko that bicycle LOC riding male aruite itchatte3. ... de walk-go: CHAU: TE and 'the boy on the bicycle walks away ...

no GEN sorede then and the

ko wa toko toko child TOP ONM hanashiga kawatte story NOM change scene changes'

For Yoshida (1994a), what constitutes a violated expectation here is the fact that the main character has walked off the screen, and "this most prominent character's departure is viewed as irreversible" (p. 247), with the underlying implication that the main character must somehow have been expected to remain on screen for the duration of the film. An alternative analysis to Yoshida's interpretation will be provided in Section 6.2.1., which discusses in more detail the co-occurrence of -te shimau with the verb iku "to go" as in this example. Example (8) is taken from a narrative of a personal wartime experience (Iwasaki 1988):

Metaphors of'total

(8)

enclosure'

honde

hito-

minna atsui

atsu

tte

and

FRG

all

hot

hot

QT

mizu

η

naka

e

tobikonda

shito

water

GEN

inside

to

jumped

into people

PST

PRT

wa TOP

wake

yakechatta burn: CHAU:

399

'those who screamed O h hot!' and jumped into the water got burned to death'. In this case, a "violated expectation" reading could be entirely plausible, since, as Yoshida notes, people who were trying to escape the heat of the fire jumped into the water, and rather than finding cooling relief, were instead burned to death. As will be seen, in contrast with Ono and Suzuki (1993), while some type of negative evaluation does tend to be associated with the -te shimau morpheme, it is not used exclusively in negative contexts; and in contrast with Yoshida (1994a, 1994b), the notion of speaker expectation can and most likely does, to a certain extent, play a significant role in some aspects surrounding the usage of this auxiliary; however, the claim that 'violated expectations' is one of its required motivating factors is an overstated one, and such a strong interpretation causes one to overlook some quite fascinating properties of this construction.

3.2. -ale pelita Lee (1990; 1993), like Ono and Suzuki (1993) and Yoshida (1994a, 1994b), is also concerned more with the expression of the speaker's evaluative judgment than with particular aspectual characteristics of this auxiliary, and like Ono and Suzuki (1993) focuses on the phenomenon of grammaticization from main verbs into auxiliaries. Lee's (1990; 1993) framework, however, is based upon a much stronger and more explicitly detailed relationship between the semantics of pelita and the meanings associated with its grammaticized auxiliary. Thus, in the case of auxiliary -ale pelita, the sense of "throwing something away" or "spoiling something", remains as a core semantic notion both for concrete as well as abstract events. In Lee (1993: 239 and 242), the meanings of -alepelita are explained in terms of "spoiling the speaker's expectation", thereby expressing disappointment or regret, and "removing psychological hindrance", expressing relief. Examples of each are shown in (9) and (10) respectively, from Lee (1993, examples (3b) and (9b)).

400

(9)

Susan Strauss

ku-nun ka-peli-ess-ta he-TOP go-peli-pst 'To my regret, he went away'.

In example (9), Lee (1993: 240) explains that the subject of the sentence (he) went away and the speaker is disappointed because of it. (10)

wuli - nun ku umsik - ul meke - peli - ess - ta we - TOP that food - OBJ eat - peli - pst 'We ate the food, to our relief'.

In (10), Lee (1993 : 242) indicates that by adding the -ale petita auxiliary, the speaker can imply that the food was somehow a burden, and that eating it has caused the speaker some relief. Lee makes no explicit mention as to the optionality of the -ale petita morpheme, although, as will be shown in Section 6.2.1., -ale petita does appear to be strictly optional, in contrast with its Japanese counterpart -te shimau. Further, by Lee's own admission, it would be difficult without additional context or an explicit evaluative expression to determine whether the speaker has intended an implication of disappointment or one of relief, and an interpretation of these utterances at the sentence level would be ambiguous.

4. -te shimau and -ale petita in the light of transitivity and aspect As noted in Section 3.1. and 3.2., -te shimau and -ale petita have been discussed in the more traditional literature in terms of their function of aspectual markers, indicating terminative aspect, telicity, and punctuality. In order to examine these notions, particularly that of punctuality, consider examples (11) and (12) below, contrasting a durative reading in the (a) sentences with a punctual one in the (b) counterparts. (11)

(a) warn ga moemashita (hitobanjuu) (Japanese) straw SUBJ burn-pst (one night through) 'The straw burned (all night)'. (b) warn ga moete shimaimashita (*hito ban juu) straw SUBJ burn-TE shimau-pst (*one night through) 'The straw caught fire (*all night)'.

Metaphors of'total enclosure'

(12)

401

(a) ku ciph-un tha-(a)ss-ta (pam-say-tolok) (Korean) that straw- SUBJ burn-pst (night-spend-to the degree) 'The straw burned (all night)'. (b) ku ciph-un tha-peli-ess-supnita (*pam-say-tolok) thatstraw-SUBJ burn-peli-pst *(night-spend-to the degree) 'The straw caught fire (*all night)'.

What is interesting is that (11) and (12) are translations from the Spanish examples used by Hopper and Thompson (1980: 266) 4 in their discussion of PUNCTUALITY and its relationship to the relative transitivity of an event, which have been reproduced below as (13a) and (13b): (13)

(a) La paja ardio (toda la noche) the straw burned (all the night) 'The straw burned (all night)'. (b) La paja SE ardio (*toda la noche) the straw RFLX burned (*all the night) 'The straw caught fire (*all night)'.

Note the striking relationship between the function of the reflexive marker se in (13b) to the -te shimau auxiliary in ( l i b ) and -alepelita in (12b). In all three cases, the reading of the sentence changes from a non-punctual reading of "burned" in the (a) sentences to "caught fire" (or "burned up") in the (b) sentences, and in all three cases, by virtue of the respective grammatical addition, a durative reading of "all night long" is ungrammatical, or unnatural at best. According to the Hopper and Thompson (1980) framework, the se-marked clause for the Spanish example represents a more highly transitive event than the non je-marked clause, and the remarked clause resists any type of Stative reading; an identical characterization of the (a) and (b) counterparts for the Japanese and Korean examples in (11) and (12) also holds. It is also interesting to note that what is salient in the earlier examples (6) - (10), in addition to the speaker's affective stance of "automatic", "violated expectations", "regret", "relief", etc., is that some type of change of state, involving punctuality as well as telicity, has also taken place. These findings are consistent with the hypothesis suggested by ArceArenales, Axelrod, and Fox (1994: 7), in their discussion of the middle phenomenon with respect to Spanish se and English gei-constructions.

402

Susan Strauss

(I)n general, middle diathesis^ marking will have associated aspectual functions, such that the middle diathesis marker is associated with punctual, change of state events.

5. -te shimau and -ale petita in the light of the 'Romance' reflexive se and the middle voice We have seen in the previous section that there exists at least a superficial relationship between the auxiliary constructions -te shimau and -ale pelita with the "Romance" reflexive marker se in Spanish. The remainder of this paper will be dedicated to solidifying this relationship and providing evidence as to some of the semantic, pragmatic, and grammatical functions of -te shimau and -ale pelita as middle voice markers.

5.1. The 'Romance' reflexive and the middle voice Before examining in more detail the parallel between -te shimau, -ale pelita, and the "Romance" reflexive se, a brief discussion of the "Romance" reflexive in general is in order. Examples (14) and (15) below illustrate respectively the basic distinction between the "true" reflexive (and reciprocal) and the "Romance" reflexive. Note that the "true" reflexive/reciprocal in (14a) and (14b) has an overt accusative object, such that the subject and object are co-referential: (14)

(a) (Yo) me golpee (I) SUB me (OBJ/RFLX) hit PST (1st sing pret) Ί hit myself.' 'true' reflexive

(14)

(b) (Ellos) se (They) SUB them (OBJ/RECIP) 'They see each other.'

ven6 see (3rd pi pres) 'true' reciprocal

In contrast, the 'Romance' reflexive has no overt accusative object, as in (15): (15)

Me

alegro tanto del clima de California gladen so much of the climate of California 'I'm so happy with the California climate'.

RFLX

Metaphors of 'total enclosure'

403

With "Romance" reflexives, then, according to Garcia (1975: 3): "instead of a direct object referring back to the subject, we have no direct object at all: the verb, correspondingly, is 'intransitive', rather than the 'transitive' of the regular reflexive." This account is also reminiscent of Lyons' (1968: 373-374) definition of middle voice, an intermediate position between active and passive voices. Lyons states: "The implications of the middle (when it is in opposition with the active) are that the 'action' or 'state' affects the subject of the verb or his interests." The relationship of the middle construction as discussed by Garcia (1975) and Lyons (1968), in the sense of affectedness of the subject with intransitive verbs can be seen in the Japanese and Korean examples (7), (8), and (9) and in the Spanish example in (15). The verbs in these examples are all intransitive.7 As discussed earlier, the middle may also occur in transitive sentences, and regarding such transitive events, Lyons considers that the "action" is being carried out by the subject for his own benefit or in his own interests. In this light, note examples (16) and (17) below, which are the Spanish and Japanese translations, respectively, for the Korean sentence in (10), reproduced here for convenience: (10)

wuli - un ku umsik - ul rneke - peli - ess - ta we - TOP that food - OBJ eat - peli - pst 'We ate the food, to our relief'.

(16)

Nos

comimos

(we) RFLX eat PST

la comida the f o o d

'We ate up the food, to our regret, relief, etc.' (17)

shokuji ο tabete shimaimashita. food OBJ eat-TE SHIMAU-PST 'We ate up the food, to our regret, relief, etc.'

These (three) utterances, i.e., (10), (16), and (17) are all possible without -ale pelita, se, and -te shimau; in which case the reading would be a simple statement that we ate the food, without the indication of "completely eating it up", and without the implication of an evaluative judgment. This type of expression is referred to as a "hypertransitive" by Arce-Arenales et al. (1994: 7) by virtue of the fact that the action is done "to completion." In an extremely detailed, precise, and global analysis of middle markers and their inherent relationships to reflexives in many languages of the world 8 , Kemmer (1988; 1993) introduces the notion of distinguishability between entities, such that a direct reflexive "involves a volitional entity

404

Susan Strauss

acting on itself as it would act on another separate entity" (1988: 97). This relationship is represented schematically in (18), from Kemmer (1993: 71): s

(18)

Q ' . A

As Kemmer (1993: 72) points out, the schema in (18) represents the fact that the event involves two distinguishable participant roles, i.e., the agent of the action and the experiencer of it, both of which are in a co-referential relationship. The dotted line represents the co-referentiality. 9 This schema could be applied to the Spanish examples in (14a) and (14b). In (14a), "I hit myself", "myself" is in an accusative relationship with the subject "I" and the transitive verb "hit". "I" and "myself" are co-referential and indeed distinguishable, in the same manner that the participants would be in a variation on this sentence, "I hit him." In contrast, in the case of middles, the entity in question is far less distinguishable, as indicated by Kemmer's (1993: 71) schema in (19):

(19)

Here, the schema also indicates a kind of co-referential ity between both the subject and the experiencer in terms of affected entities; however, the degree of distinguishability between the two is minimal at best. In (15), for example, while the form used is reflexive in nature, it is not at all a direct reflexive. In other words, the sentence does not translate as: "*I make myself happy". The reflexive does not indicate an accusative relationship between the subject and the verb. Rather, it translates as "I am so happy ...", or "I get so happy ...," indicating that the subject and the affected experiencer are one and the same. Maldonado (1993), drawing on Kemmer's (1988; 1993) notion of distinguishability among participants, focuses his study on Spanish se and its relationship to middles, explicating many of its occurrences on the basis of controlled vs. uncontrolled occurrences, naturally expected vs. unexpected situations, suddenness of events, and so forth. In fact, Maldonado's (1993) characterization of "Energetic and Dynamic" constructions, involving se in Modern Mexican Spanish, seems to apply nicely to certain uses of -te

Metaphors of'total

enclosure'

405

shimau and -alepelita as well, as will be shown in more detail in Section 7. 5.2. zte shimau and -ale pelita — middle environments, middle functions The auxiliaries -te shimau and -ale pelita are both hypothesized to have grammaticized from a metaphor of "total enclosure", and having viewed Kemmer's (1993: 71) schema for the middle in (19), it becomes clearer that it is precisely this type of conceptualization that could best account for the parallel phenomenon in Japanese and Korean, rather than the schema proposed by Lee (1990; 1993), in (5). First of all, Lee's schema can only apply to Korean, since the Japanese lexical verb shimau simply means 'to put away', not 'throw away' or 'get rid of'. Hence, the schema suggested in (5b), i.e., after pelita, would only apply to Korean. Evidence from other languages such as Malay and Uyghur, indicate that a similar grammaticization has taken place involving verbs which imply 'total enclosure'. For example, Wee (1993) has pointed out that the Malay prefix TER-, used to indicate that an action occurred accidentally or without the speaker's control, may have originated from a verb meaning 'to pass through'; and Taub (1993) demonstrates that the Uyghur auxiliary verb salmaq 'to put into' has grammaticized into an auxiliary meaning 'accidental'. 10 Further, another basic feature of both Japanese -te shimau and Korean -ale pelita, as well as Malay T E R - 1 1 , is that as they all signal the notion of 'completion' and 'totality,' although in the case of TER- the reading does appear to be predominantly 'accidental'. 12 This last fact, combined with the semantics of the lexical items from which these morphemes have grammaticized, represent the basic elements for a grammaticization into middle constructions, the schematic representation of which would coincide with that in (19); and what is particularly fascinating here is that these verbal morphemes, at least insofar as Japanese and Korean are concerned, have grammaticized into middle constructions, much in the same way as the direct reflexive has in many other languages of the world. Both -te shimau and -ale pelita occur in discourse in similar environments as middle constructions, i.e., occurring overwhelmingly with intransitive verbs, as well as in many of the same environments as designated by Kemmer (1988: 343-349, 1993) for the middle voice, notably in the emphatic domain, with verbs of translatable motion (motion that relates to a path, e.g., 'go', 'fly', 'fall'), verbs of emotion (e.g., 'be amazed,' 'be afraid', 'be delighted'), and with cognition verbs (e.g., 'forget', 'realize'). Also, as demonstrated in the data sets for the current study, when these constructions do involve lexically transitive verbs, they tend to be of the

406

Susan Strauss

type in which it is the subject of the verb who benefits, e.g., 'to take (away)', 'to eat up', 'to steal', and so forth. However, as will be demonstrated in the following sections, while Japanese and Korean are typologically related, they evidence quite distinct patterns in the distributions of the usage of these respective middle morphemes, in contrast with the non-typologically related Japanese and Spanish.

6. The data In an effort to conduct an in-depth study of the auxiliary verbs -te shimau and -ale pelita and their relationships to the 'Romance' reflexive, two basic data types were used: the children's story, 'Goldilocks and the Three Bears' and narratives about "The Pear Film".

6.1. Goldilocks and the Three Bears 'Goldilocks' was chosen because of the recurring themes in the story of "eating up porridge", "breaking the baby bear's chair", and "falling asleep", and the corresponding prediction that these actions might be marked by the same grammatical feature(s) under investigation in all three languages. The results of the comparison between the Japanese and the Spanish versions are shown below in (20a) and (20b). It was impossible to locate a Korean version of this story anywhere in Southern California or Korea 13 , and consequently (20c) is based on native speaker intuition on how the Korean text might read if it were available. (20)

(a) (2) (1) (1)

Spanish: (4 tokens)14 comerse 'to eat up' [G ate up baby bear's porridge] dormirse 'to fall asleep' [G, in baby bear's bed] romperse 'to break' [baby bear's chair broke when G sat in it]

(b) (2) (1) (1)

Japanese: (4 tokens) tabeteshimau 'to eat up' [G ate up baby bear's porridge] nemurikondeshimau 'to fall asleep' [G, in baby bear's bed] nuiteshimau 'to break the bottom out of' [re: baby bear's chair]

Metaphors of 'total enclosure'

407

(c) Korean: (4 tokens predicted): (2) cwuk-ul ta mek-e peli-ess-supnita porridge-OBJ all eat peli-pst '(She) ate up the baby bear's porridge'. (1) cam tul-e peli-ess-supnita sleep peli-pst '(She) fell asleep'. (1) uyca-ka pwuse-cye peli-ess-supnita chair-SUBJ break-passive peli-pst 'The chair broke'. What each of these 12 verb tokens share, at least minimally, is the following: (1) the notion of totality, (2) a punctual action as opposed to a nonpunctual state, (3) telicity, with an orientation toward the termination/end point of the event [totality in the sense of 'eating up', 'falling [fast] asleep', and 'breaking the chair'], and (4) a substantial clue that a primary grammatical means to express the difference between 'eat' vs. 'eat up' and 'sleep' vs. 'fall asleep' in Japanese, Korean, and Spanish is -te shimau, -ale pelita and the 'Romance' reflexive, respectively.

6.2. Oral narratives

6.2.1. Story Recall - 'Pear' narratives The 'Pear' data consist of 19 narratives collected in Japanese (Clancy 1980), nine in Spanish, and ten in Korean 15 . The narrative task for each subject, based on Chafe (1980), was to monologically recount the plot of a six-minute, dialogue-free film, to an interviewer who claimed not to know the story. The main characters in the film, except for the pear picker, are all children. The plot involves the theft of one basket of pears by a boy on a bicycle, the boy's getaway, a distraction, a minor bicycle accident, and three passers-by who help the boy. Similar to the Goldilocks data, the 'Pear' data are rich in environments for middle constructions. The number of 'Pear' narratives for each language and the corresponding number of tokens of each grammatical feature are shown in (21). Each narrative was produced by a different speaker.

408

Susan Strauss

(21)

'Pear' Data

tokens/narrative ^/narratives 9 19 10

Spanish Japanese Korean

#/tokens 92 106 2

10.2 5.6 0.2

The most immediately obvious difference in (21) is the exceptionally low number of tokens of Korean -ale pelita, in contrast with the much higher frequencies of occurrence in the Spanish and Japanese data. Further, upon deeper investigation, we note another similarity between the Japanese and Spanish data with respect to these morphemes and their co-occurrence with the verb 'to go' in each language. In fact, for the Japanese data, of the 106 total tokens of -te shimau, a full 35 % co-occur with some form of the verb iku 'to go', which in most, if not all cases, constitutes an obligatory use of the -te shimau auxiliary, as originally noted by Alfonso (1966). Here, the semantic contrast between iku vs. itte shimau (-te form of main verb iku + shimau aux.) or itchau (the reduced phonological variant in the Tokyo dialect) is basically analogous to the contrast of 'to go' vs.'to go away' or 'to leave'. Note how example (22), excerpted from narrative #1 actually becomes an ungrammatical Japanese utterance when the -te shimau auxiliary is deleted, as in (22'): (22)

((From narrative #1)) booshi ga tobasareta wake, koronda toki ni ne? a sono ... ... motte kichatta16 otoko no kono ... — > ...de sono otoko no ko ga itchau η da kedo and

that

boy

SUBJ

go-TESHIMAU

PRTcopPRT

'his hat was blown off when he fell, right?, [the hat] of the boy who brought [the pears here] ... and then that boy leaves' ... (22')

... *de sono otoko noko ga and that boy SUBJ 'and then that boy goes ...'

iku η da kedo go PRT cop PRT

A parallel semantic distinction also exists in Romance languages, as illustrated in (23) and (24) below with examples from Spanish and French. The meaning change from 'to go' vs. 'to go away' or 'to leave' is accomplished by adding a reflexive marker in each case:

Metaphors of'total enclosure'

409

(23)

Spanish: (a) 'to go' (b) 'to leave'

infinitive ir irse

3rd pi. sing. va a la escuela '(She) goes to school' seva/ *va '(She) is leaving'

(24)

French: (a)'to go' (b) 'to leave'

aller s'en aller

ilvaal'ecole ils'en va /* il va

'He goes to school' 'He is leaving'

Examples (25) and (25') from the Spanish data are a fine parallel to the Japanese (22) and (22') above: (25)

((Spanish narrative #3)) ... le ayudan a montar lasperas en la bicicleta y el se va. 'they help him put the pears on the bike and he leaves'

(25')

*le ayudan a montar las peras en la bicicleta y έΐ va

On the basis of these examples from Romance as well as the tokens from the data (both Japanese and Spanish), it is evident that with respect to the verb 'to go', some additional notion of destination or directionality is relevant with respect to both the GOAL as well as the SOURCE locations. Thus, the use of -te shimau for Japanese or the 'Romance' reflexive for Spanish and French, constitutes not only an aspectual difference, but is also significant from the point of view of spatio-temporal deixis. Note in the French example (24b), the existence of the particle en, literally meaning 'from there,' i.e., direction away from a source, which is obligatory with certain verbs in French. Kemmer (1988: 218) posits that "it may be that the original overtones of change of state, perhaps shared by se and en, came in time to be expressed by se alone, and that the directionality so transparently provided by en came to be associated with and expressible by se in these constructions". Note, however, in (26) that in Korean, the distinction between 'to go' vs. 'to go away' is not obligatorily expressed by the verb kata 'to go' + -ale pelita (aux). In contrast with Spanish and Japanese, it is perfectly grammatical in Korean to use the verb kata + past tense marker to express the fact that someone went away. (26)

ku-nun

ka-(a)ss-ta

he TOP go- PST

'He went (away).', 'He left'.

410

Susan Strauss

The addition of -ale pelita, does in accordance with Lee's (1990; 1993) discussions, charge the utterance with an evaluative judgment on the part of the speaker, as was illustrated earlier in (9). An example of the 'hypertransitive' verbs, as discussed in 5.1., can be seen in (27) and (28), excerpted from the Japanese and Spanish data respectively. No such counterpart exists in the Korean data. Recall that according to Arce-Arenales et al. (1994), these verbs indicate that the action is carried out through completion. Note how the intensifier zenbu in (27) "all" underscores the notion of totality, as pointed out by Teramura (1982), Makino and Tsutsui (1986), among others. Actually, (27) is marked three times for totality, in the expressions kago goto, zenbu, and the -te shimau auxiliary. (28) shows a striking similarity to the Japanese example with respect to the underscoring of totality by other linguistic means, with the Spanish intensifer toda 'whole' and the explicit mention of fullness through the adjective llena 'full'. (27)

(28)

kago goto zenbu totchatta basket whole all take-TE SHlMAU-pst '(He) took the whole basket of pears'.

wake PRT

se lleva toda la cesta llena de peras RFLX take entire the basket full of pears '(He) takes the entire basket full of pears'.

As noted earlier, the 'Pear' data is abundantly rich in verbal environments which tend to be associated with the middle voice according to Kemmer (1988; 1993). We have evidence of translational motion in the verbs 'to go', 'to go away', 'to fall', 'to fly', 'to roll', 'to drop', 'to fall down'; in the lexically transitive types 'to take,' 'to steal,' 'to eat'; and even in the cognition verb 'to forget'. Interestingly, we see a nice parallel in the distribution of Japanese -te shimau and the "Romance" reflexive throughout these event types. In fact, there were a total of 95 tokens of reflexive verbs in the Spanish 'Pear' data, of which 92 were of the 'Romance' reflexive type. Only three were of the 'true' reflexive or reciprocal type: 2 tokens of golpearse17 'to get hurt' (lit. 'to hit one's s e l f ) and 1 token of hablarse 'to speak to each other.' Of these 92 tokens, 43 (or 47%) are used in describing the identical episodes that the -te shimau construction described in the Japanese versions. These verbs include: irse 'to go away', caerse 'to fall', llevarse 'to take away', tirarse 'to knock over', tropezarse 'to bump into', and toparse 'to bump into'. 1 8

Metaphors of 'total enclosure'

411

Regarding the two tokens of Korean -ale petita (uttered by different speakers), both occur in the description of the same event sequence: the boy falls from the bike/all the stolen pears spill out. In contrast, 15 of the 19 Japanese narratives use -te shimau to depict this sequence of events, and 9 of the 9 Spanish narratives use the 'Romance' reflexive. Perhaps what is being marked here in all three languages is that this event is the non-volitional counterpart which thwarted the boy's exceedingly volitional action of having stolen the pears in the first place. This pair of corresponding volitional/non-volitional actions is the only such type in the story. A more detailed discussion of this event sequence will follow in the next section.

7. Maldonado's dynamic and energetic reflexives Some of the incidents marked with -te shimau, -ale petita, and se in both the 'Goldilocks' data as well as the "Pear" data, particularly with reference to their occurrences with intransitive verbs, appear to share many characteristics of "Dynamic" and "Energetic" se discussed by Maldonado (1988; 1993), since se, too, has long been recognized as a marker of accidentally, similar to -te shimau, -ale petita, and other verbal morphemes mentioned in this paper. Maldonado's (1988; 1993) discussion of this phenomenon calls to mind one aspect of Yoshida's (1994a, 1994b) 'violated expectation' interpretation for -te shimau and also provides much support for considering both -te shimau and -ale petita as middle markers. According to Maldonado (1993: 344-345), the expectations of the conceptualizer can play a crucial role in the use of middle constructions. For example, using the following pair of examples19, Maldonado illustrates how one situation might be conceived of as a natural occurrence and the other as an unexpected one. In the case of Spanish, the use of se in the (a) version, constituting the natural situation, would render the utterance ungrammatical, just as would its absence in the (b) version, constituting an unexpected occurrence. (29)

(a) En el otono, las hojas (*se) caen de los ärboles 'In autumn, the leaves fall from the trees.' [e.g., during normal course of season]

(b) En la primavera, las hojas se (*•) cayeron de los ärboles 'In spring, the leaves fell from the trees.' [leaves expected NOT to fall]

412

Susan Strauss

Interestingly, -te shimau and -ale pelita pattern in a similar manner, as noted in (30) and (31). In the case of Japanese and Korean, however, the presence or absence of the respective marker would not necessarily render the utterance ungrammatical, but rather only 'unnatural.' Also, according to some native speakers of Korean, (31b) sounds like a highly emotionally charged utterance. (30)

Japanese (a) aki ni wa autumn in TOP

konoha the leaves

ga SUBJ

ki kara tree from

ochiru fall-NON PAST

'In autumn, the leaves fall from the trees'. (b) haru ni wa spring in TOP

konoha ga ki kara ochite shimaimashita the leaves tree from fall-TE SHIMAU PST 'In spring, the leaves fell from the trees'.

(31)

Korean (a) kaul ey-nun namwu eyse iph-i ttele-cinta autumn in-TOP trees from leaves-SUBJ fall 'In autumn, leaves fall from the trees'. (b) pom ey-nun namwu eyse iph-i ttele-cye-peli-ess-ta spring in-TOP trees from leaves-SUBJ fall-PELITA 'In spring, the leaves fell from the trees'.

The fact that Spanish se would be grammatically obligatory in an instance such as (29b) and that the corresponding markers in Japanese and Korean vary quite markedly in terms of degree of evaluative judgment in (30b) and (31b) might well account for the type of distribution of forms discussed in the previous section with regard to the Pear data — in the event sequence where the boy falls from his bike and the pears consequently spill out: 2 tokens of -ale pelita from the entire Korean data set, 15 of 19 Japanese speakers marking the verbs with -te shimau, and 9 out of 9 Spanish speakers using se.

8. Conclusion In this paper I have attempted to elucidate how particular verbal elements indicating 'total enclosure' have come to grammaticize as auxiliary verbs

Metaphors of 'total enclosure'

413

whose functions seem to strikingly parallel many aspects of other middle constructions. In particular, we have examined how and to what basic degrees these verbal auxiliary elements in Japanese and Korean function as grammatical analogues to the reflexive marker in Romance. Some of the data have shown that from a particular grammatical standpoint Japanese -te shimau actually appears to be more closely related to the 'Romance' reflexive than it is to Korean -ale pelita, which may provide us with some clues that the Japanese and Korean forms have grammaticized somewhat differently from their original lexical meanings. Another factor for the differences between -te shimau and -ale pelita could be the fact that as lexical verbs shimau and pelita are closely related from a semantic point of view, however, pelita is much more negatively charged, since its meaning is 'throw away' (i.e. discard) as opposed to the more neutral 'put away', in Japanese. This study has also attempted to illustrate the value of cross-linguistic analysis in locating some pieces of grammatical puzzles that may otherwise have remained unnoticed and to underscore the fact that human languages exhibit strikingly similar and consistent grammatical patterns in describing, reacting to, and expressing particular types of experiences and events.

TRANSCRIPTION CONVENTIONS: OBJ LOC TOP EMPH PTL

accusative marker locative topic marker emphatic particle final particle

ONM GEN FRG QT PST

onomatopoetic genitive truncated word (fragment) quotative past tense

Notes *

1.

This study has been funded in part by UCLA's Center for Japanese Studies. The author is grateful to Noriko Akatsuka, Marianne Celce-Murcia, Patricia Clancy, Haruko Cook, Kenneth Cook, Osten Dahl, Robert Kirsner, Dohee Koo, Masayoshi Shibatani, Sung-Ock Sohn, Sandra Thompson, Elizabeth Traugott, Lionel Wee, and the anonymous reviewer for their suggestions and comments relating to the ideas expressed in this paper as well as an earlier version, delivered at the LAUD Language and Space Conference in Duisburg, Germany, March, 1994. Special thanks go to Yasuhiro Shirai and Foongha Yap. The author is also grateful to David Palmquist for his assistance with the graphics. Any errors or inconsistencies are, of course, my own. It is interesting to note that both verbs are highly transitive in the sense of Hopper and Thompson (1980: 252), ranking high on the transitivity scale

414

2.

3.

4. 5.

6. 7.

8. 9.

Susan Strauss

for all ten components-PARTlciPANTS: 2 or more, i.e., A and O, KINESIS: action, ASPECT: telic, PUNCTUALITY: punctual, VOLOTONALITY: volitional, AFFIRMATION: affirmative, MODE: realis, AGENCY: A = high in potency, AFFECTEDNESS OF Ο: Ο = totally affected, and INDIVIDUATION OF Ο: O = highly individuated. Ironically, these verbs, when used as auxiliaries, in some cases have the effect of decreasing the transitivity of the main verb with which they occur. It has been argued elsewhere (i.e., Strauss 1994a, 1994b) that Yoshida's interpretation of 'violated expectations' for every and all instance of -te shimau is without empirical foundation, and that numerous counter-examples abound in modern colloquial Japanese in which a 'violated expectation' reading would be implausible. The -te shimau auxiliary often occurs in its reduced phonological form chau in the Tokyo dialect. Glosses in this paper will include both the full form, -te shimau, as well as the reduced form chau, and the same basic semantic, pragmatic, and grammatical interpretations can be said to apply to both. These are Hopper and Thompson's (1980) examples (47a) and (47b), originally excerpted from Garcia (1975). Arce-Arenales et al. (1994: 1) use the term 'middle diathesis' (from Klaiman 1988) to refer to syntactically active subjects which are semantically affected by the action of the verb; it is used in their article in the same way as middle voice or middle construction is used in the present paper. This sentence can also be interpreted with the reflexive meaning of se, 'they see themselves' (e.g. in the mirror). In both cases, the co-reference between subject and object still holds. Notice how Ono and Suzuki's translation of Japanese example (6) uses a reflexive construction in English Ί find myself paying'. While the verb 'to pay' is clearly transitive, the use of -te shimau relegates the activity to beyond the control of the speaker, and it is the speaker who is affected or suffers as a result of his own lack of control over the situation. Not included in Kemmer's discussion of middles are Japanese and Korean, among others. cf. Kemmer's (1993: 50) schema for a prototypical 2-participant event:

o A

10. 11. 12. 13. 14.

+o Β

Neither Wee (1993) nor Taub (1993) have suggested that these verbal morphemes belong to the domain of middles. As pointed out by Strauss (1994a, 1994b) and Wee (1993), respectively. Foongha Yap (personal communication) has pointed out numerous examples which support Wee's (1993) findings for the 'accidental' and 'lack of control' readings for TER-, In spite of the tireless efforts by Kyu-hyun Kim and Kyung Hee Suh. It must be noted here that because the 'Romance' reflexive and -te shimau do not identically overlap in their usage, a far greater number of reflexive tokens appeared in the Spanish data — however, the four that are listed parallel almost identically the usages of -te shimau in Japanese, as shown.

Metaphors of 'total enclosure'

15. 16. 17. 18.

19. 20.

415

I am deeply grateful to Kyu-hyun Kim and Kyung-Hee Suh for sharing these data with me. This verb, kuru 'to come' is also marked with -te shimau (+pst); however, this is not semantically relevant to the immediate discussion of iku + -te shimau, since it does not constitute an obligatory use of -te shimau. This is actually a borderline case between the 'Romance' reflexive and the 'true' reflexive, since it is used here in the sense of 'to get hurt' rather than in the transitive sense of 'to hit one's self. Although there are striking similarities between the Spanish 'Romance' reflexive and -te shimau, many differences exist as well, and it is not the intention of the author to suggest a strict one-to-one correspondence between these constructions. Examples (29a) and (29b) are from Maldonado's (1993: 344-345) examples (42b) and (43). I thank Noriko Akatsuka for this observation.

References Abraham, Werner 1993 English, Dutch, and German Middles. Talk given at UCLA Department of Germanic Languages. May, 1993. Alfonso, Anthony 1966 Japanese Language Patterns. A Structural Approach. Tokyo: Sophia University L.L., Center of Applied Linguistics. Arce-Arenales, Manuel, Axelrod, Melissa, and Fox, Barbara 1994 Active voice and middle diathesis. A cross-linguistic perspective. In: Fox, Barbara and Paul Hopper (eds). Voice: Form and Function. Amsterdam/Philadelphia: Benjamins, 1-21. Benveniste, Emile 1971 Problems in General Linguistics. University of Miami Press. Chafe, Wallace (ed.) 1980 The Pear Stories: Cognitive, Cultural, and Linguistic Aspects of Narrative Production. Norwood, NJ. Ablex. Clancy, Patricia 1980 The Pear narrative data. In: Chafe, Wallace (ed.), 127-202. Garcia, Erica 1975 The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North-Holland Publishing Company. Hopper, Paul and Sandra Thompson 1980 Transitivity in Grammar and Discourse. Language. 56, 2: 251-298. Iwasaki, Shoichi 1988 A Study of Speaker's Perspective in Japanese Spoken Discourse. Ph.D. dissertation, UCLA. Ide, Sachiko, et al (eds.). 1984 Shufu no isshukan no danwa shiryoo (Data from a day in the life of a housewife). Nihon Jooshi Daigaku. Jacobsen, Wesley. M. 1992 The Transitive Structure of Events in Japanese. Tokyo: Kurosio Publishers. Kemmer, Suzanne 1988 The Middle Voice: A Typological and Diachronic Study. Unpublished doctoral dissertation. Stanford University.

416

Susan Strauss

Kemmer, Suzanne 1993 The Middle Voice. Amsterdam/Philadelphia: Benjamins. Klaiman, Μ. H. 1988 Affectedness and control: a typology of voice systems. In: Shibatani, Masayoshi (ed.). Passive and Voice. Amsterdam: Benjamins, 25-84. Lee, Keedong Auxiliary verbs in Korean and subjectification. Papers from the Seventh 1990 International Conference on Korean Linguistics. Eung-Jin Baek (ed.). University of Toronto Press. A Korean Grammar on Semantic-Pragmatic Principles. Hangkuk Moon 1993 Hwasa. Seoul. Lyons, John 1968 Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press. Makino, Seiichi and Michio Tsutsui 1986 A Dictionary of Basic Japanese Grammar. Tokyo: The Japan Times. Maldonado, Ricardo 1988 Energetic reflexives in Spanish. Berkeley Linguistics Society 14: 153-165. 1993 Middle Voice: The Case of Spanish se. Ph.D. dissertation. University of California, San Diego. Martin, Samuel 1975 A Reference Grammar of Japanese. New Haven. Yale University Press. Ono, Tsuyoshi and Suzuki, Ryoko 1993 The development of a marker of speaker's attitude: The pragmatic use of the Japanese grammaticized verb shimau in conversation. Berkeley Linguistics Society 18. Ono, Tsuyoshi. The grammaticization of the Japanese verbs oku and shimau. Cognitive 1992 Linguistics 3-4: 367-390. Soga, Matsuo 1983 Tense and Aspect in Modern Colloquial Japanese. Vancouver. University of British Columbia Press. Strauss, Susan A cross-linguistic analysis of Japanese, Korean, and Spanish: -te shimau, 1994a -alepetita, and the 'Romance' reflexive se. Japanese/Korean Linguistics. Vol. 4. CSLI: Stanford. Noriko Akatsuka (ed.), 257-273. Japanese -te shimau examined in the light of the 'Romance' reflexive 1994b system, UCSB Working Papers in Linguistics. Volume 4, 205-225. Taub, Sarah Uyghur verb compounds: Semantic shifts in auxiliary verbs, uc 1993 Berkeley, ms. Teramura, H. 1982 Nihongo no shintaxu to imi. Part II. Tokyo. Kurosio Publishers. Traugott, Elizabeth On the rise of epistemic meanings in English: An example of subjectifi1989 cation in semantic change. Language 65; 1: 31-55. Wee, Lionel 1993 The Semantics of 'Ter'. UC Berkeley. MS. Yoshida, Eri 1994a Expectation motivating the use of the auxiliary shimau in Japanese spoken narratives. UCSB Working Papers in Linguistics, 245-263. Speaker's subjectivity and the use of shimau. Japanese/Korean Lin1994b guistics. Volume 4. Noriko Akatsuka (ed.). CSLI: Stanford, 183-199.

Section 6 From space to time, events, and beyond

The story of -ing: A subjective perspective* Marjolijn Verspoor

1. Introduction For many years, sentential complementation in English and other languages has been the subject of much interest, study, and controversy in different theoretical frameworks. The number of insightful observations rooted in this tradition is so great that an in-depth review of the literature would be beyond the scope of any article. The point of this article, however, is not to criticize any of these observations, but to extract from them general cognitive principles that motivate complement distribution. Within the generative account especially the semantics of the main verb (e.g. Karttunen 1971, Kiparsky and Kiparsky 1970) has been regarded as the main factor in complement selection. Within a functional account the semantics of the complement structure and its compatibility with the semantics of the main verb (Givon 1980) have been mentioned to account for the distribution of the different types of complements underlined in (1). (1)

a. Peter found the chair comfortable, (predicative adjunct) b. Peter made Harry leave, (plain infinitive) c. Peter saw Harry leaving, (-ing) d. Peter ordered Harry to leave, (to infinitive) e. Peter ordered that Harry leave, (that clause)

Within a cognitive account, it is also accepted that the compatibility between the semantics of the main verb and the semantics of the complement clause is a determining factor. In addition, it is postulated that the use of a specific complement depends on the construal of the conceptualizer. The construal, in turn, is related to space: the perspective of the conceptual izer.

418

Marjolijn Verspoor

Furthermore, it has been shown that even though they have distinct senses, depending on the construal of the conceptual izer, alternate constructions after the same verb such as a that clause or an -ing after see are semantically related in that they may denote the same event, objectively speaking. (2)

a.

We saw that Duco left.

b. We saw Duco leaving. Also the same complement structure after different types of verbs is often related; for example, a that clause after verbs of perception or a that clause after verbs of communication both symbolize propositions (cf. Langacker 1991: 35, 447). (3)

a.

We saw that Duco left.

b.

We admit that Duco left.

In other words, it is generally accepted that on the one hand different complement structures often have related senses and on the other hand one particular complement structure may have distinct, but related senses. Despite this complexity of interrelated factors (same structure, different senses; different structures, related senses) one might wonder if it is possible to extract an underlying schema for each of the constructions in Table 1. One of the main issues that has not been resolved entirely is that each type of complement may occur on the one hand with typical action and/or mental causation verbs (wipe, make, avoid, and refuse) and on the other hand with typical perception or cognition verbs (find, see, and believe). Is there one abstract schema that plays a role in complement selection with both types of verbs?

The story of -ing: A subjective perspective

Table 1.

419

Constructions: Schematic sentences with different types of complements subject

verb

predicative adjunct

Harry Harry

wiped found

the board clean the board clean

subject

verb

plain infinitive

Harry Harry

made saw

Ron leave Ron leave

subject Harry Harry

verb avoided heard

ing leaving Ron leaving

subject

verb

to infinitive

Harry Harry

refused believed

to see Sally Sally to be at home

subject

verb

that clause

Harry Harry

said knew

that Sally was at home that Sally was at home

This article shows that both questions can be answered positively. I will follow especially Goldberg (1995) in claiming that these specific sentence patterns should be seen as 'constructions' with semantic import and I will follow Langacker (1991: 438-463) in claiming that the differences in semantic import between these structures is related to the notions of 'subjectivity' and 'objectivity.' In short, a predicative adjunct typically symbolizes a direct result of an action or an assessment, interpreted as being induced by direct experience. A plain infinitive symbolizes a bounded event that is caused or perceived directly. An -ing also symbolizes an event, but caused immediately or viewed from such a close range that the

420

MarjoUjn Verspoor

boundaries of the event are not within the conceptualizer's visual scope. A to infinitive denotes a less immediate event, and a that clause, a more objective view of a state or event. In addition, I will claim that the compatibility of these different complement types with particular main verbs is motivated by a causal image schema inherent in the state or process denoted by the verb, not only in the case of typical causal verbs such as make and refuse, but also in the case of verbs of perception and cognition such as see and believe. To support this rather controversial claim, I have organized this paper as follows: Section 2 outlines the problem in more detail, and to relate verbs denoting mental states and processes to causation, Section 3 provides a sketch of the Anglo folk model of the mind. Section 4 outlines the theoretical framework and working hypothesis for the analysis, and in Section 5, I will attempt to apply the causal schema to -ing structures (participles and gerunds) in particular.

2. The problem As the overview of complement-taking verbs in Table 2 (see appendix) illustrates, there seems to be no rhyme or reason to the distribution of complement structures: finite complements (that clauses) and non-finite complements (predicative adjuncts, -ing structures, plain infinitives, and to infinitives) may each be found after many different semantic types of verbs: those that express a pure cause (cause), an act {wipe), an intention (avoid), a prior intention or willingness (agree), perception or mental processing (see, understand), a belief or mode of reasoning (believe, reason), a feeling or emotion (prefer), or a mode of speaking (order). Also as (4)-(8) illustrate, verbs that are semantically similar do not always take the same type of complement. For example, why does avoid take an -ing and refuse a to infinitive? Why does see take a that clause, but catch does not? Why does see take an -ing, but make does not? Why does deny take an -ing, but say does not? A general principle should account not only for the cases in which particular complements may occur, but also for these apparent inconsistencies and other gaps in the chart of Table 2. (4)

a. / avoid swimming. b. *I avoid to swim.

(5)

a. I refuse to swim. b. */ refuse swimming.

The story of -ing: A subjective perspective

(6)

a. I saw Duco leaving. b. I saw Duco leave. c. I saw that Duco left.

(7)

a. I made Duco leave. b. *I made Duco leaving. c. */ made that Duco left.

(8)

a. I caught Duco leaving. b. *I caught Duco leave. c. */ caught (in the sense of see) that Duco left.

421

Since there are five general categories of constructions (see Table 1), each with a huge array of sub-categories, it would seem likely that speakers do not learn these category members one by one, but operate in terms of some basic, productive, and fairly simple general schema (unless some specialized principle takes precedence). Taking Lakoffs discussion of Dixon's analysis of the Dyirbal classifier as an example (Lakoff 1987) and applying it to other component structures with unit status (Langacker 1991b: 103), it would seem likely that each of the constructions can be characterized as a complex radial category with some members that are more central, which have common properties that play a role in characterizing the basic schemas. Other members, not necessarily sharing these common properties, would be linked by a chaining process, characterized by basic domains of experience and idealized models of the world. It would also be quite conceivable that exceptions are distributed according to the basic opposition model; for example, exceptions that would go otherwise in the to infinitive category might go to the -ing category and conversely (Lakoff 1987: Ch. 6). However, before looking at constructions separately, it is important to see whether these constructions have common properties that may characterize a basic schema. Since it is generally acknowledged that the particular complements occur most typically after verbs that express a variety of mental states or processes and actions and since it is assumed that a complement structure must be compatible with the semantics of a verb, it would be appropriate to determine whether there are general cognitive schemas related to mental states, processes and actions. To do so, I will examine the folk model of the mind in the next section.

422

Marjolijn Verspoor

3. Basic notions A folk model of the mind might give us insight into image schematic structures that play a role in language, especially as they may be projected on verbs denoting mental states and processes. By means of in-depth interviews, D'Andrade concludes that the Anglo folk model treats the mind both as a 'collection of internal states' (a container) and as 'a set of internal processes' (a processor). Inside this container there are various sub-parts. Each of these sub-parts can be conceived of alternately as a smaller container or processor. As containers, the parts can take in and store information, and as processors they process information by interacting with each other and triggering other states or processes. The container as a whole, or some of its sub-parts, interact with the outside world. Besides regarding the mind as a container and processor, the folk model of the mind treats the conscious mental states as having central causal powers (D'Andrade 1987: 117-18): Except in pathological cases, what one sees, hears, and/or senses is understood to be caused by various events and objects external to the mind. What one knows or believes is usually considered to be a creation from within, a result of the operation of the mind itself. What one feels emotionally is more problematic. Sometimes emotions are treated as something caused at least in the sense of being triggered—by external events. At other times, emotions are treated as internally generated by the person. Desires, like emotions, are also seen as both internally and externally caused. Intentions and resolutions, however, are treated as directly caused only from within.

As Figure 1 illustrates by means of arrows, each of the mental states and processes (perception, belief or knowledge, feelings or emotions, desires or wishes, and intentions, will or self-control), involves the concept of cause: the idea that certain (mental) events are thought to bring about (mental) events.

The story of-ing: A subjective perspective

423

Figure 1. A folk model of the mind

The shortest and most direct causal chains are possible in perception, action and emotion. An outside stimulus may directly cause perception. (I see John because John is there.) An inside or outside stimulus may directly cause an emotion. (I feel sad because I see him or because I think about him.) An (inside) intention may directly cause an external action. (I raise my hand because I want to get the teacher's attention.) Inherent to directly caused perception, emotion, and action is the notion of temporal contiguity: When I see a cat, that cat causes my perception at that moment. When I turn the wheel of my car in an emergency to avoid an accident, I intend to turn the wheel. However, the intention and action typically occur together, so the intention causes the action at the same moment as or immediately after I form the intention. When I think about a sad event and the thought makes me feel sad, I feel sad at the same moment as the thought occurs. According to the folk model, the other mental states may involve a variety of more complex causal chains. For example, a belief may be based on past perception and/or cognition: Sally believes Lisa is her friend because she went through the process of assessing her relation to Lisa and finally concluded she was a real friend. A desire may be based on an emotion and then trigger an intention which results in an action: Sally sees John and feels angry because he is late. She may then form a desire to ignore him. The desire in turn may trigger an intention, and that intention

424

Marjolijn Verspoor

may trigger an action at a later moment. At one point, Sally may form an intention in action and walk away without saying anything. The more complex causal chain involved in beliefs, desires, and prior intentions are inherently indirect, either because there is some intermediary force or some delayed result. For example, if I hear a car coming up the driveway (direct perception), it may cause me to believe that someone is coming to visit me. So, in this case the belief is caused by the perception (an intermittent force). In fact, a mental state in itself can affect perception or another mental state. For example, if you are expecting a visitor, your mental state of expectation makes you more likely to perceive the car coming up the driveway. Desires and intentions are usually formed before an event takes place, so even though they may directly cause an event, there is a distance in time between the moment of forming the intention or desire and the event, which in turn would be caused by an intention made at that moment. For example, if I plan to get some coffee during my break, I have a 'prior intention'. At the moment that the break starts and I actually get up, the getting up is an 'intention in action' resulting from my prior intention. Of course, as Dirven (1995) points out, the concept of cause is quite complex and fuzzy and is not necessarily related to objectively ascertainable facts, but emerges as a consequence of perspectivization: even though it is often difficult to tell what is the cause and what is the effect, two events are construed as being linked together in a cause-effect relation. In other words, the term refers to the human conceptualization of causation, some fundamental way of schematizing perceived reality to allow humans to make sense of their physical and social environment. However, in human conceptualization such complex, fuzzy concepts may be reduced to rather simple ones. The English use of complement structures seems to make a rather simple distinction in the conceptualization of these causal chains: those that are construed as very direct, those that are construed as more complex and therefore less direct, and those that are construed as not causally related. As Figure 2 illustrates, a stimulus can directly cause a mental state or process, and a mental state or process can directly cause some form of action. Indirectness is involved when there is some mediation (in this case cognition/belief) between the stimulus and the mental state or when there is some mediation (in this case prior intention or desire) between a mental state and an action.

The story of -ing: A subjective perspective

425

Stimulus — > (Cognition) — > Mental State/Process — > (Prior Intention) — > Action

Figure 2. A folk model of a mental causation chain

As Langacker (1991b: 254-55) points out, different degrees of causation are reflected in the way the links in an action chain can be construed: as a single complex event or as a sequence of separate, but causally related, events: (9)

a. A — > b. Β — > A — > c. C — > Β — > A — >

When a chain is construed as a single event, it is usually coded by a single verb stem, but if a situation is construed as an action chain chunked into separate events, it is usually coded by separate verbs: (10)

a. John moved the table. b. John made the table move. c. John wanted to move the table.

The first one is most appropriate for a situation of direct physical causation, the second one for causation involving some perceived mediacy (by pressing a button, by pulling on a rope, by catching it with his foot). The third one involves some premeditation, and the effect does not necessarily have to be immediate; for example, John may first have formed a desire to move the table and decided later to act upon this desire. Or, there can be a mental state directed towards some action, but the action fails to take place for some reason. These degrees of temporal and spatial contiguity are in turn symbolic for an even more profound concept: the degree of subject involvement: (a) would be more typical of an intended, but not premeditated situation, (b) would be more typical of an intended, but mediated situation, and (c) for a premeditated situation. One might conclude that the degree of subject involvement (as construed by the conceptual izer) is inversely related to the number of words the complement structure contains (cf. Givon 1980), symbolizing the degree of directness: direct or indirect.

426

Marjolijn Verspoor

In this case, the different degrees of directness can be related to Searle's distinction between 'intention in action' and 'prior intention' (1983: 84-7). Cases of prior intention are those in which the agent has the intention to perform an action prior to the performance of the action. For example, with a prior intention, the agent knows what he is going to do because he already has an intention to do it, but as Searle points out (Searle 1983: 84): ... not all intentions are like that: suppose you ask me, 'When you suddenly hit that man, did you first form the intention to hit him?' My answer might be, 'No I just hit him.' But even in such a case I hit him intentionally and my action was done with the intention of hitting him. I want to say about such a case that the intention was in the action but that there was no prior intention.

With a prior intention, the agent acts on his intention, carries out his intention, or tries to carry it out; on the other hand, with intentions in action, the intention in action is the intentional content of the action; the intention and the action are inseparable (Searle 1983: 84). It is commonly accepted that a form of 'prior intention' is denoted in English with a to infinitive (I wanted to move the table). I assume that the 'intention in action' is denoted in English with a single verb {He moved the table) because, as D'Andrade points out, "in the folk model actions do not occur without intentions", but "following the Gricean maxims, we do not say what is obvious, [so] normally we do not explain an action by saying it was intended" (1987: 120). However, even though it is not always necessary to express that an action was intended as in (10a), the notion that an action was intended may be verbalized. To point out that there was an intention and at the same time an action, English may use the plain infinitive (He made the table move). I will use the term 'intention and action' to distinguish such a situation from an 'intention in action.' The examples in (10) have to do with some form of action, which has to do with the last part of the causal chain given in Figure 2, from mental state or process to action: (10a) denotes only the final part of the causal chain, the action (move), and the intentional mental state is not made explicit; (10b) makes the intentional mental state explicit (make) that is responsible for the simultaneous event (the table move)·, (10c) denotes a prior intention (want) that is responsible for a later (potential) event (to move the table). The beginning of the causal chain (Stimulus > (Cognition) > mental state/process) can be denoted by English sentences in a similar manner. For example, perception, one of the mental states or processes, is not usually explicitly stated in a sentence reporting on a perceived event (see 11a), but the mental state/process of perception may be made explicit as in (lib).

The story of -ing: A subjective perspective

427

Also, a sentence may report on a directly perceived stimulus which, through a mode of cognition, has led to a state of belief as in (11c). (11)

a.

The table moved. (This sentence typically denotes that an event (the stimulus) was directly perceived by the speaker, but the mental state of perception is not explicitly stated.) b. / saw the table move. (This sentence also denotes that an event was directly perceived by the speaker, but in this sentence the mental state/process is made explicit by the verb see.) c. I believed the table to have moved. (This sentence may also denote some direct perception (e.g. seeing that the table is now in another place), but the event itself as expressed by the complement was not directly perceived but inferred. Some form of cognition has mediated between the direct perception and the inferred event.)

To conclude this section, in the folk model of the mind, what one does, sees, hears, and/or feels is understood to be caused directly. What one believes and wants or intends to do is understood to be caused indirectly. In order to relate this folk model of mental causation to an analysis of constructions containing subject, verb and complement structures, the framework and working hypothesis are laid out in the next section.

4. The framework and the working hypothesis My analysis makes use of concepts developed by the theory of Cognitive Grammar (CG) especially by Langacker (1987, 1991a and b). Assuming that readers are familiar with its general tenets, I will introduce only those aspects most relevant to my analysis: 'schematic units,' 'construal relation,' and 'subjectification.' According to Langacker (1991b: 103), the proper domain of linguistic description is conventional linguistic knowledge, and it is assumed that this knowledge takes the form of conventional units. A unit is defined as a structural complex that has achieved 'unit status.' With this he means that it has been mastered by a speaker to the point that he can manipulate it as a whole without attending specifically to its internal parts or their relation to one another. Among the units that embody linguistic knowledge there are 'content units' (corresponding to specific statements) and 'schemas' (corresponding to general statements). A schematic unit is a category, which represents the content shared by its various instantiations at a higher level of abstraction. Each content unit elaborates the schema in different

428

Marjolijn Verspoor

ways and specifies it in finer detail along one or more dimensions. I presume that Langacker's schematic unit at sentence level is analogous to Goldberg's (1995) 'constructions.' One and the same situation may be described in many different ways, making use of different 'conventional units,' depending on the 'construal relation.' As Langacker points out: "There are many different ways to construe a given body of content, and each construal represents a distinct meaning; this is my intent in saying that an expression imposes a particular image on the content it evokes" (1991b: ix). Alternate constructions impose contrasting images on the conceived situation. One construal relation pertinent to complementizers in particular is 'subjectification,' illustrated in the diagrams below (Langacker 1991b: 317):

(b)

(a)

Θ

Ο \

j

J

!

OS

i

__

PF OS

PF

Figure 3a. Objective viewing arrangement

Figure 3b. Subjective viewing arrangement

In Figure 3a the conceptualizer (C) has an off-stage view of the perceived object (P). In this optimal viewing arrangement, C has clear perceptual (or conceptual) access to P, with the specific focus of attention within OS (the onstage, 'objective' region) and PF (the full expanse of the viewer's perceptual field), but C is external to both. In this case, although C's construal is subjective, Ρ is viewed as 'objectively' as possible.

The story of -ing: A subjective perspective

429

Figure 3b illustrates the 'egocentric viewing arrangement', where C may not only be self-aware (hence included in PF), but can even go onstage, taking its place within an expanded egocentrically determined OS region. At the extreme, C can itself become the focus of viewing attention (C=P). The closer C gets to Ρ the more 'subjectively' Ρ is construed. In Figure 3 and the examples in (12), the conceptualizer (speaker) and the subject of the sentence are one and the same. To account for sentences with subjects other than the conceptual izer (e.g. He regrets being late) it is conceivable that a conceptual izer is off-stage, but reports on a conceptual izer who is on-stage. In other words, a speaker reports an event from another conceptual izer's vantage point. Figure 4, a composite of Figure 3a and b, illustrates how one speaker (S) has an 'objective' view of another conceptual izer's (C2) 'subjective' view.

S (CI)

Figure 4. A speaker's objective view of a conceptual izer's subjective view.

Langacker gives an example how P, a mental state, may be construed more subjectively or objectively:

430

Marjolijn Verspoor

Suppose I experience an emotion, such as fear, desire, or elation. If I merely undergo that experience nonreflectively, both the emotion and my own role in feeling it are subjectively construed. But to the extent that I reflect on the emotional experience—by analyzing it, by comparing it to other such experiences, or simply noting that I am undergoing it—the emotion and my role therein receive a more objective construal.

This alternate construal is reflected in the type of complementizer chosen after a verb of emotion. In the case of a finite clause, the situation is 'grounded' and therefore construed more 'objectively.' In the case of a non-finite clause, the situation is not grounded and therefore construed more 'subjectively' (Langacker (1991b: 324). (12)

a. I regret that I am late. (The grounded that clause imposes a more objective construal of the event.) b. I regret being late. (The atemporal non-finite structure imposes a more subjective construal of the event.)

As Langacker (1991: 35) points out, "The effect of reification (marked by that) is 'to step back' from the situation—including both the event and its relation to the ground—and construe it as an abstract object or proposition capable of being manipulated, evaluated, and commented on." To be able to analyze the 'objective' or 'subjective' construal situation in sentences with different complement constructions, we should distinguish at least the following component and composite structures: S

Speaker: The 'speaker of the sentence' is also a conceptualizer,

SlTl

Situation 1: Realized as a sentence consisting of the following components:

SIT2

PI

Participant 1: Realized as subject of the main clause. This participant is a conceptualizer (C) who may or may not be the same as (S).

VI

Verb 1: A verb of causation, mental state or process, or mode of communication. Situation 2: Realized as a sentence-like part consisting of the following sub-components:

The story of -ing: A subjective perspective

P2

Participant 2: [schematically] realized as subject of the complement structure.

V2

Verb 2: [schematically] realized: zero, non-finite or finite.

COM

431

Complement: A possible complement of V2 which may be a direct object, adverbial, etc.

I claim that a sentence containing a finite or non-finite complement clause occurs when a speaker has as object of perception a participant (Pi) and his/her construal relation (vi) towards a state of affairs or event. This state of affairs or event may be symbolized by a finite or non-finite clause depending on the degree of objectivity or subjectivity with which the speaker thinks Pi has construed it. The degree of objectivity or subjectivity in turn depnds on the perceived degree of causal directness between the construal relation and the event or state of affairs. The direction of the causal relation, which may go from Pi to SIT2 (where Pi triggers SIT2 by means of Vl) or from SIT2 to Pi (where SIT2 affects Pi) is not evident until Vl is lexicalized. For example, in (13a) the causation may be from Pi to SIT2 depending on whether Vl is make or see. (13)

a. I [Vl] John leave.

(The composite structure expresses that Pi and SIT2 are causally related; the direction is not specified)

b. I made John leave. (The causal direction goes from Pi to SIT2)

c. I saw John leave.

(The causal direction goes from SIT2 to PI)

It may seem odd to consider a direct object as causally related to the mental state or process of a subject, but, as Langacker (1991b: 221) points out, the subject of a sentence such as Several witnesses saw the accident is an experiencer. Unlike a prototypical agent, who induces contact by moving to the patient, an experiencer makes mental contact with the entities represented (Langacker 1991b: 221): A sentient creature is capable of generating an internal representation of the world around him and also of entities that have no actual physical existence. By generating [bold face mine] such a representation the experiencer makes mental contact with the entities represented. We tend in fact to concretize our conceptualization of this process by construing the entity in question as external to the experiencer, who is conceived as mentally moving into contact with it.

432

Marjolijn Verspoor

Langacker (1991b: 222) uses the term 'absolute' for the role of the second participant as it is non-energetic: it is neither an 'energy source' nor an 'energy sink.' However, the term 'generating' implies that the experiencer generates the representation from the object observed. Without the object (real or abstract) no representation could be made. In other words, even though there is no energy flow from the object per se, the object is a source and enables (a form of causation) the generation of the mental representation. One sees an object because it is there. Therefore, even if the object does not exude energy, it has a causal relation to the subject's perception. For example, the sentence I saw the accident happen could be paraphrased as follows: 'The accident happened. I know this because I saw it.' We may conclude that the seeing event would not have come about if it had not been for the accident happening. Of course, the subject's state of perception involves some energy directed at the object, and the object does not exude energy, but the perception of the object would not be possible without the object. This is in accordance with Croft (1991: 219), who observes: There are two processes involved in processing a mental state (and changing a mental state): the experiencer must direct his or her attention to the stimulus, and then the stimulus (or some property of it) causes the experiencer to be (or enter into) a certain mental state. Thus a mental state is actually a two-way causal relation and is better represented as follows:

Experiencer

Stimulus direct attention to >




Figure 6. poy

(48)

Chodzilismy po lesie. we walked around in forest 'We walked around in the forest.'

The 'distributive' sense is related to another sense of the preposition po, exemplified in (49) and represented graphically in Figure 7. 3 It normally requires a plural landmark; if the landmark is singular, as in (50), the action is understood to have been repeated a number of times, producing visibly plural results (e.g. scratch marks all over the table top). The sense of completion is due to the fact that the LM consists of a finite set of objects (or that the action has resulted in a specific number of marks, etc.). (49)

Porozsypywafi mqk% po catym stole. po-he spilled flour over whole table. 'He spilled flour all over the table.'

A study of Polish perfectivizing prefixes

(50)

481

Porysowai (caly) stoi. po-scratched (whole) table 'He scratched the (whole) table/He made scratch marks all over the table top.'

Figure 7. po2 Other uses of po- are more difficult to relate to spatial meanings. In perfective forms such as poinformowac, 'to inform', poplQtac, 'to tangle', podzielic, 'to divide', the prefix does not seem to add to the meaning of the verb, and is hence often described as 'purely aspectual'. This conclusion is again premature. The 'purely aspectual' po- occurs only with verbs which profile a change (a change of state, an addition to the knowledge of an addressee in a speech event, etc.); and a comparison with another apparently 'purely aspectual' prefix, prze-, which also combines with change of state verbs, will help to bring out its semantic characteristics. The difference is not immediately apparent because the prefixes tend to occur with different verbs. Po- is used when the verb designates a transition from a 'zero' or normal value to a marked value—for example, poinformowac, 'to inform', implies a transition from a state of not knowing to a state of knowing, poplqtac, 'to tangle', designates a transition from an untangled to a tangled state, etc. Prze-, on the other hand, is used when the verb designates a transition from one value to another, and the values have equal status: przebudowac ('across-build'), 'to do structural work (to a building)', przettumaczyc ('across-explain'), 'to translate (transform from language A to language B)', etc. Thus, pomalowac ('po-paint') and przemalowac ('across-paint') differ in meaning: pomalowac means 'to cover with paint' and przemalowac, 'to change the color of LM by painting it'.

482

Ewa Dgbrowska

5. ZaThe prefix za- is very productive and has many senses. For the purpose of this paper, I will only look at two of these, both closely related to the prototypical spatial sense of the preposition za, 'behind' (Figure 8).

*

T R

Figure 8. ζάλη the configuration represented above, the LM is between the TR and an observer. It also makes the trajector invisible and blocks access to it. It is the latter two properties which make possible the use of the prefix za- in the senses exemplified in (51-52) and (53-55). (51)

Zakrylismy /Zakleilis'my plam% papierem. behind-we covered/behind-we glued spot: ACC paper: INSTR 'We covered the dirty spot with a piece of paper/We glued a piece of paper over the dirty spot.'

(52)

Zapisal /Zabazgrat /Zarysowal caty arkusz. behind-he wrote /behind-he scribbled /behind-he drew whole sheet 'He wrote/scribbled/drew all over the sheet.'

(53)

Zatarasowat/Zablokowat/Zagrodzil nam droge. behind-he blocked we: DAT way: ACC 'He stood in our way/blocked the way for us.'

(54)

Zamrozilismy /Zakisilismy /Zawekowalismy fasolkq. behind-we froze /behind-we pickled /behind-we canned string beans 'We froze/pickled/canned the string beans.'

A study of Polish perfectivizing prefixes

(55)

483

Sekretarka zapisata /zaprotokotowata/ secretary behind-wrote /behind-set down in the minutes/ zanotowata wypowiedz dyrektora. behind-noted

statement: ACC director: GEN

'The secretary wrote down/set down in the minutes/jotted down the director's statement.' In (51), the trajector ('the dirty spot') is covered by the landmark ('the paper') and hence made invisible. (52) is an extension of this sense: although the trajector ('the sheet of paper') is not made invisible, it is partially covered with writing, scribbles, or drawings. In (53), the (unspecified) landmark blocks access to the trajector (wherever the referent of the dative NP is going to). (54) and (55) are metaphorical extensions of this 'blocking of access' sense, motivated by the fact that if one blocks X's access to Y, one protects Y from X, and, by implication, makes it available for future use. All of the verbs in (54) and (55) have this sense of preserving/securing, which is present to varying degrees in the meaning of the simplex verbs, and reinforced by the use of za-. In both cases the action is completed when the result suggested by the meaning of the prefix (i.e. blocking access to the trajector or covering the trajector) is reached.

6. Factors determining the choice of prefix In this section, I will look at three verbs, cerowac, 'to darn', pracowac, 'to work', and pisac, 'to write', all of which can take several perfectivizing prefixes. I will show that each prefix reinforces a different aspect of the meaning of the simplex verb, and thus modifies its meaning in different ways. I will also show that the idea of telicity is arrived at by integrating the semantic contributions of the prefix and the landmark object. (56)

(57)

a. Babcia cerowata skarpetkς. grandma darned sock 'Grandma was darning the sock.' b.

?Babcia cerowata dziur% w skarpetce. grandma darned hole in sock 'Grandma was darning the sock.'

a.

(?) Babcia zacerowata skarpetkζ. grandma behind-darned sock 'Grandma darned the sock.'

484

Ewa Dqbrowska

b. Babcia zacerowata dziurz w skarpetce. grandma behind-darned hole in sock 'Grandma darned the hole in the sock.' (58)

Babcia pocerowata skarpetki. grandma po-darned socks 'Grandma darned (all) the socks.'

The verb cerowac evokes a semantic structure which incorporates, among other things, a garment with a hole in it. Either of these can function as the landmark: one can 'darn a sock' or 'a hole in a sock' (56), though skarpetka, 'sock', is the preferred object. The most neutral perfective form of cerowac is zacerowac, another example of the 'covering' sense of za- discussed earlier (cf. example (51)). This use of za- requires a landmark which can be covered—in this case, 'the hole in the sock'; and hole in the sock is the preferred object when the verb is perfective (57b). (57a), where the verb is perfective and the whole patient, rather than the directly affected part, is the object, is an example of profile-active zone discrepancy; though it is a possible way of expressing the proposition, it is somewhat elliptical (unless the sentence is taken to mean 'she covered the entire sock with darning patches'). Finally, (58) evokes the 'distributive' sense of po-, which requires a plural landmark; the action is delimited because the set of objects designated by the landmark is finite. (A 'limited duration' interpretation of (58) is also possible, though less likely.) (59)

Kowalski przepracowat dwa lata. Kowalski across-worked two years 'Kowalski worked for two years.'

(60)

Kowalski dopracowat do emerytury. Kowalski to-worked to retirement 'Kowalski worked until his retirement/until he became eligible to retire.'

(61)

Kowalski odpracowal wolny dzien. Kowalski from-worked free day 'Kowalski worked an extra day to make up for a day off.'

(62)

Kowalski popracowat (jakis czas/dwie godziny). Kowalski po-worked (some time/two hours) 'Kowalski worked for a while/for two hours.'

A study of Polish perfectivizing prefixes

(63)

485

Kowalski wypracowat nowq, metodq badawczq. Kowalski out-worked new method research 'Kowalski developed a new research method.'

The above examples contain perfective forms of the verb pracowac, which is intransitive, and hence limits of the action cannot be defined with respect to the direct object. One way to delimit a process that does not involve a patient is to give it temporal boundaries, and perfectivization with prze-, do- and od- requires a temporal landmark. 5 Prze- requires a landmark that designates a stretch of time (in (59), it is the period during which Kowalski worked). Do- requires a landmark which designates a point in time or an event which metonymically stands for a point in time (in (60), the moment at which Kowalski became eligible for a retirement pension). Odpracowac (example (61)) is slightly more complex. It is a temporal variant of the 'return' sense of od- (cf. (41-42)): what is returned here is the equivalent of the time that Kowalski owes his employer (i.e., the same number of hours). Po- (in its 'limited duration' sense) can also be included in this group, in that its semantic representation makes reference to a stretch of time of indefinite duration; in the case of this prefix, however, the landmark is usually rather indefinite (e.g. jakis czas, 'for a while') and can be left implicit. The use of wy- in the last example is made possible by the metaphor CREATION IS TAKING OUT (as in to make X out of Y; see Lakoff and Johnson 1980 for a discussion). The landmark is an abstract object created in the course of the action designated by the verb; the process is completed when this object reaches its canonical form—i.e., when the result of Kowalski's efforts can properly be described as a new research method.

(64)

a. Magda napisata

namurze

'telewizja ktamie'.

Magda on-wrote on wall 'television lies' 'Magda wrote 'television tells lies' on the wall.'

b. Magda napisata list /wiersz. Magda on-wrote letter/poem 'Magda wrote a letter/a poem.' c.

Magda napisata dwie

strony.

Magda on-wrote two 'Magda wrote two

pages pages.'

486

Ewa Dqbrowska

d. Magda dopisata postscriptum do listu. Magda to-wrote postscript to letter 'Magda added a postscript to the letter.' e. Magda podpisata list. Magda under-wrote letter 'Magda signed the letter.' f.

Magda wpisata swoje nazwisko w odpowiednie miejsce formularzu. Magda in-wrote her name in right place form 'Magda wrote her name in the blank on the form.'

na on

g. Magda przepisata /odpisata zadanie domowe. Magda across-wrote /from-wrote homework 'Magda copied her homework (off somebody else).' (65)

(66)

Magda zapisata dwie strony. Magda behind-wrote two pages. 'Magda filled two pages with writing.'

a. Magda zapisata moj telefon. Magda behind-wrote my telephone (number) 'Magda wrote down my telephone number.' b. Magda wypisata dane ζ raportu. Magda out-wrote data from report 'Magda copied some figures from the report.'

(67)

Magda wypisata dtugopis. Magda out-wrote pen 'Magda wrote until the pen ran out.'

My last example is the verb pisac, 'to write', which combines with a number of prefixes (na-, do-, pod-, w-, prze-, od-, za-, and wy-, among others). To appreciate the semantic contribution of each, we must first consider the meaning of the unprefixed form. The verb 'write' invokes at least the following elements: a writer, a writing system, a message, a vehicle (i.e. a physical object, e.g. a piece of paper, on which marks are made to represent the message), a text (i.e., a series of marks on the vehicle repre-

A study of Polish perfectivizing prefixes

487

senting the message), and a writing implement (e.g. a pen). Any of the latter four can be the landmark that delimits the action. If the landmark is the text, as in example (64), the action will be completed once the entire text is recorded on the vehicle; if it is the vehicle, as in (65), the action counts as finished when the entire surface of the vehicle is covered; if it is the information, as in (66), the entire message has been committed to writing; and finally, if the landmark is the writing implement, the action is completed when the implement is completely used up (67). However, the prefix communicates more than just the type of landmark, and hence finer distinctions are possible. For example, five different prefixes can combine with a text LM, and they all highlight different aspects of the meaning of the verb. Να-, 'on' is the most neutral of the set, since it designates the relationship between the text and the vehicle (the writing is on the page, etc.), and is used to indicate that the entire text has been transcribed. Do-, pod-, and w- all have special connotations: do- conveys a sense of adding to the existing text (cf. (32-34)), podpisac ('under-write') has the specialized sense of 'signing' (which, of course, is motivated by the fact that one normally signs under the main text of a document), windicates that the text is placed within a larger text (w = 'in'). Perfectives with prze- and od- convey the idea of copying, or transfering (prze, 'across') text from {od-, 'from') one vehicle to another. When one wishes to convey the idea of recording the content of the text, za- is the most appropriate prefix (recall that za- has a 'preserving' sense which closely matches the idea of securing information onto a piece of paper). Wy('out'), which can also be used in this function, has the additional sense of 'taking information out' of a source (in (66b), out of the report). The za- in (65) is an instance of the 'covering' sense, which of course requires a concrete landmark with a surface that can be covered with writing. The contrast between (65) and (64c) is a good illustration of how the prefix can pick out different senses of the object NP. The noun strona, 'page', can refer either to the vehicle or, metonymically, to the (length of) the text: since za- requires a LM with a surface one can write on, it forces the 'cover with writing' interpretation (the preserving sense of za- is irrelevant here, because strona cannot refer to information); na-, on the other hand, requires a landmark which is a body of text, and therefore strona must be interpreted as referring to the length of the text. 6

7. Deadjectival and denominal verbs I will conclude with a brief discussion of the role that perfectivizing prefixes play in the derivation of verbs from nouns and adjectives.

488

Ewa Dqbrowska

It is one of the basic claims of cognitive linguistics that all grammatical concepts, including word classes, can be defined in semantic terms. If this is so, then any morphological device that changes word-class membership will have to add to the base the semantic characteristics required for membership in the new class. According to Langacker (1987), verbs are expressions which designate processes, or relations with a positive temporal profile. Prototypically, of course, they designate processes which involve a change of state. The perfectivizing prefixes discussed here all designate relationships that involve movement, and hence necessarily imply both change and a positive temporal profile. Thus, they give a noun or adjective what it needs to become a verb, so to speak. The following examples will illustrate the point. (68)

a. b. c.

(69)

a.

odgruzowac, 'to remove the debris from' (gruz, 'debris') odsniezyc, 'to remove snow from' (snieg, 'snow') odwodnic, 'to drain, dehydrate' (woda, 'water')

zadrzewic, zalesic, 'to afforest, plant with trees' (drzewo, 'tree'; las, 'forest') b. zaludnic, 'to populate' (ludny, 'populous') c. zablocic, 'to cover with mud' (bloto, 'mud')

The derivation of the verb also involves adding a verbal ending ( - V c or -owac), which also signals the fact that the predicate to which it is attached has a positive temporal profile. However, it could be argued that the prefix symbolizes this relationship more transparently, since it does not change according to person and number, and, moreover, it has clear semantic links with the basic spatial sense of the corresponding preposition. The obvious question which presents itself at this point is: is the prefix necessary for deriving a verbal meaning? The answer, presumably, is 'no', since there are many unprefixed denominal and deadjectival verbs in Polish. However, it might serve this function more often than it might seem at first glance. Historically, many of these unprefixed verbs may have started out as prefixed forms, with the corresponding imperfective being derived from the perfective by suffixation. Then, as the verbs became more entrenched, and the prefixes gradually bleached, simpler, prefix-less forms might have begun to function as imperfectives, and gradually phased out the more complex forms. From a synchronic perspective, it is impossible to tell which member of a perfective-imperfective pair such as odszumowacszumowac, 'to skim off the scum (e.g. when making jelly)', is primary, i.e. whether the simplex imperfective form was derived by backformation from the prefixed perfective, or whether the perfective form was derived by ad-

A study of Polish perfectivizing prefixes

489

ding od- to the simplex verb. The question, therefore, requires a diachronic study. However, the synchronic co-occurrence of both simplex and prefixed-suffixed imperfective counterparts of the same perfective form (cf. (70)) seems to support the hypothesis that the prefix-less imperfective is derived from the prefixed perfective. (70)

spetnic

(pf) - spetniac,

pelnic

(imp), 'fulfill'

zaparzyc

(pf) - zaparzac,

parzyc (imp), 'infuse, brew'

zarqczyc

(pf) - zarqczac,

reczyc

(imp),

'vouch'

Notes 1.

2.

3.

4.

5. 6.

The concept of radial category is introduced and discussed at length in Lakoff (1987). Janda (1986) provides a detailed analysis of four Russian analogues of the perfectivizing prefixes discussed here (za-, pere-, do-, and of-). Pasich (1993) applies the concept to the Polish prze- and Rudzka-Ostyn (1984) is a cognitive study of another Polish prefix not covered here (ivy-). That 'time' is conceptualized in spatial terms is evident in the systematic polysemy of spatial prepositions such as in, from, to, throughout and many other words designating spatial dimensions (e.g. long, short, near, distant), all of which have temporal senses. Moreover, verbs of motion are often used metaphorically do designate the passage of time (e.g. time flies, three years went by, the minutes dragged on, there will come a day when ...), and many other grammatical categories (e.g. cases) have temporal analogues of spatial uses. There are also many non-linguistic manifestations of the metaphor: time is represented as space in graphs, time lines, etc.; in a clock, the passage of time is symbolized by the movement of a hand around the dial. See Lakoff and Johnson (1980) for a detailed discussion. The prepositionpo- also has a temporal sense of 'after', which may also play a role in motivating the 'distributive' sense: the episodes which make up the action designated by the prefixed verbs follow each other. Thus the distributive sense of po- may be said to be multiply motivated. Porozsypywac is another example of the layering of prefixes mentioned in the introductory section (cf. example (1)). Sypac, 'to pour' is imperfective; the addition of the prefix roz-, 'out', makes it perfective, but it also changes the meaning (rozsypac means 'to spill'). Therefore, the imperfective is derived by adding the suffix -ywac. Finally, the addition of po- introduces the idea of covering the entire surface of die LM. Prze- and od- also make the verb transitive. The expression length of the text is another illustration of the fact that texts are conceptualized as paths.

490

Ewa Dqbrowska

References Comrie, Bernard 1976 Aspect. Cambridge: cup. Dahl, Osten 1985 Tense and Aspect Systems. Oxford: Blackwell. Forsyth, John 1970 A Grammar of Aspect: Usage and Meaning in the Russian Verb. Cambridge: CUP. Janda, Laura 1986 A Semantic Analysis of the Russian Verbal Prefixes za-, pere-, do-, and ot. München: Sagne. Hopper, Paul J. and Sandra A. Thompson 1980 Transitivity in grammar and discourse. Language 56: 251-99. Lakoff, George 1987 Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago: Chicago University Press. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago: Chicago University Press. Langacker, Ronald 1982 Remarks on English aspect. In: Hopper, Paul J. (ed.). Tense-Aspect: Between Semantics and Pragmatics. Amsterdam: Benjamins, 265-304. 1984 Active Zones. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 10: 172-88. 1987 Foundations of Cognitive Grammar. Stanford: Stanford University Press. 1990 Concept, Image and Symbol. The Cognitive Basis of Grammar. Berlin: Mouton de Gruyter. Miller, John 1972 Towards a generative semantic account of aspect in Russian. Journal of Linguistics 8: 217-236. Netteberg, Kristine 1953 Etudes sur le verbe polonais. Copenhage: Rosenkilde og Bagger. Pasich-Piasecka, Agnieszka 1993 Polysemy of the Polish verbal prefix prze-. In: Görska, Elzbieta (ed.). Images from the Cognitive Scene. Kraköw: Uniwersitas, 11-26. Rudzka-Ostyn, Brygida 1984 Cognitive grammar and word formation. The case of Dutch uit and Polish wy-. In: Nowakowski, Jan (ed.). Litterae at Lingua: In Honorem Premislavi Mroczkowski. Wroclaw: Polska Akademia Nauk, 227-239. / Smiech, Witold 1986 Derywacja preflksalna czasowniköw polskich. Wroclaw: Wydawnictwo Polskiej Akademii Nauk. Talmy, Leonard 1983 How language structures space. In: Pick, Herbert and Linda Acredolo, (eds). Spatial Orientation: Theory, Research and Application. New York: Plenum Press, 225-282.

Temporal meanings of spatial prepositions in Polish: The case of przez and w Agata Kochanska

It has often been noted that in the case of prepositions, one linguistic form typically expresses multiple spatial meanings, as well as a number of more abstract, non-spatial senses. Temporal meanings seem to be among the most frequent abstract uses of many prepositions which in their primary sense denote relations between entities in physical space. This poses serious problems to any semantic theory which adheres either to the Aristotelian model of categorization, or to the view that categories encoded in languages mirror the objective structure of the world. Adopting the Aristotelian model of categorization results in the claim that different senses expressed by one linguistic form constitute a polysemous category only if there exists a set of essential meaning properties common to all senses under consideration and, at the same time, unique to that group of senses. Adhering to objectivism in semantics, on the other hand, allows for considering a group of meanings as belonging to one category only if the relatedness of senses mirrors the relatedness of entities in reality. Therefore, those semantic theories which take the above-mentioned assumptions for granted are forced to treat cases of e.g. expressing spatial and temporal relations between entities by means of the same linguistic form as instances of accidental homonymy. Such an analysis, however, does not account for the fact that grouping spatial and temporal meanings under one name is absolutely pervasive, both within one language and cross-linguistically. Also, it does not seem to agree very well with native speakers' intuitions in its claim that languages are overloaded with massive and unmotivated homonymy. In the present paper, temporal senses of two Polish prepositions, przez 'across' and w 'in', are analyzed within the framework suggested by cognitive linguistics (cf. Lakoff 1987: 416-461, after Brugman 1981; Lakoff 1990; Langacker 1987a; Radden 1991; Taylor 1989). It will be claimed that temporal meanings of przez and w constitute extensions from the

492

Agata Kochanska

prototypical spatial senses of these prepositions, which are motivated by our metaphorical understanding of abstract concepts such as time, process, state, or change. It will be argued that, contrary to the objectivist claim, the grouping of spatial and temporal senses within one category, far from being accidental, is indeed highly motivated and meaningful for speakers of Polish. In its basic spatial sense, przez 'across' profiles the complex atemporal relation (cf. Langacker 1987a: 248ί)* between two entities in which the TR (the figure within the relational profile (cf. ibid.: 217)) follows a path across the bounded region of the LM (the entity which serves as a reference point for assessing the location of the TR (cf.ibid.)). The relationship between the TR and the LM is complex since it does not reduce to a single consistent configuration, but involves a series of component states. These states, however, are scanned in a summary fashion, i.e. they are all simultaneously available to the conceptualizer. Thus, they are not construed as being distributed through conceived time (cf. Langacker 1987a: 168) and, consequently, the relationship is atemporal (cf. ibid.: 248). The basic spatial sense of przez is illustrated by example (1). (1)

Piotr przeszedt przez plac i usiadt przy stoliku w kawiarni Peter />rze-Perf.-went across square and «-Perf.-sat at table in cafe 'Peter went across the square and sat at a table in the cafe.'

In (1), przez profiles a series of component configurations between the TR and the LM. This complex scene is construed atemporally. The positive temporal profile of the conceptualization denoted by the whole clause results from the fact that the preposition (or, more precisely, the prepositional phrase) is integrated with the verb, which is the profile determinant (cf. Langacker 1987a: 289) at the clausal level.2 It should be observed that the basic meaning of przez may evoke the con-ceptualization of either the part of the path which is actually contained within the bounded area of the LM or, alternatively, of the whole path, starting outside the area, crossing its interior and boundaries, and ending again outside of the area. W 'in', on the other hand, prototypically designates a simple configuration (cf. Langacker 1987a: 248f) in which the TR is wholly contained inside the LM. This basic meaning of w is illustrated by (2): (2)

pierscionek w pudetku ring in box 'the ring in the box'

The case of 'przez' and W

493

The main difference between the logic inherent in the image-schematic structures of the prototypical conceptualizations evoked by przez and irrespectively seems to be that in the case of przez the path of the TR's movement may freely extend outside the limits of the LM. In the case of w, on the other hand, the TR must necessarily be wholly contained inside the LM. In certain contexts w may be used to locate not an object, but a path of the object's movement inside a bounded region. In such a case this path, being the TR of w, cannot extend outside the limits of the LM. This contrast between the two prepositions is illustrated by the sentences in (3) below: (3)

a.

Czzsto jezdze przez las rowerem Often I-ride across forest bicycle Ί often ride a bicycle through the forest.'

b.

Czqsto jezdzq w lesie rowerem Often I-ride in forest bicycle Ί often ride a bicycle in the forest.'

It may be observed that (3a) may denote both the situation when the path of the TR's movement is wholly contained inside the LM and the situation of the TR's movement starting and finishing outside the region of the LM. In contrast, the use of (3b) is limited to those cases in which the TR moves only within the boundaries of the LM. Like a great number of other spatial prepositions, both przez and w may also profile relations in the domain of time. According to cognitive linguistics, the fact that relations in space and relations in time may be expressed by means of the same linguistic forms is by no means accidental. Instead, it is claimed that relating such objectively quite disparate conceptualizations within one category is well-grounded in our cognitive system. As pointed out by Lakoff (cf. 1990: 55), one way of understanding time is by conceptualizing spans of time in terms of bounded spatial regions. The flow of time is then construed in terms of the observer's movement in space.3 This metaphorical understanding of the more abstract domain of time in terms of the more concrete domain of physical space and movement in this space motivates the use of przez and w in cases like those illustrated under (4a) and (4b), respectively. (4)

a.

Piotr napisat ksiqzkq przez dwa lata Peter «ö-Perf.-wrote book across two years 'Peter wrote a book in two years.'

494

Agata Kocfianska

b.

Piotr napisat ksiQzk$ w dwa lata4 Peter «ö-Perf.-wrote book in two years 'Peter wrote a book in two years.'

In the examples above, the temporal extensions of the profiled processes are specified by a /jrzez-phrase in (4a) and a w-phrase in (4b). These uses of the prototypically spatial prepositions are motivated by the fact that spans of time are metaphorically understood in terms of bounded regions in physical space. Therefore, the conceptualization of the flow of time in general, and of the temporal extension of the process in particular, seem to involve the subjective motion on the part of the conceptual izer (cf. Langacker 1987a: 171ff.) relative to the time space. In the case of processes involving a change in the relation between the TR and the LM, it may be suggested that this abstract motion is, at least partially, objectified. This objectification may take place to the extent to which the change brought about by a process is understood metaphorically. As pointed out by Lakoff (cf. 1990: 57), processes involving a change from one state to another may be metaphorically understood as the TR's movements from one region to another. These metaphorical regions with respect to which the TR of the process moves, that is, the component states of this process, are distributed through the conceived time (cf. Langacker 1987a: 168). Thus, conceptualizing certain processes in terms of metaphorical movement from one state to another seems to involve also the conceptualization of the abstract motion in time on the part of the TR of a given process. It may be observed that the details of a spatial configuration denoted by the basic meaning of a preposition are also relevant to our understanding of its metaphorical senses. To illustrate this point, let us consider the conceptualizations evoked by (5a) and (5b) below. (5)

a.

Piotr pisal ksiazke przez dwa lata Piotr Imperf.-wrote book across two years 'Piotr was writing a book for two years.'

b.

Piotr pisal ksiqzkt w dwa lata Piotr Imperf.-wrote book in two years 'Piotr wrote a book/each of his books in two years.'

In both sentences, the verbs are used in their imperfective forms. In Langacker's terminology, these verbs profile processes which are imperfective, that is, unbounded in time within the immediate scope of predication (cf. 1987b: 80). 5 It should be noted, however, that their unboundedness seems to be construed in two alternative ways, depending

The case of 'przez' and 'w'

495

upon whether the path of the subjective motion of the conceptualizer relative to the temporal space is specified by przez, or whether it is located in a region of time by means of the preposition w. The alternative construals follow from different logics of configurations which are denoted by the basic spatial senses of these two prepositions and which are preserved in their metaphorical senses. In (5a), the path of the abstract motion involved in the conceptualization of the temporal extension of the process is denoted by przez. This path simply crosses the region of time. Since it can freely extend outside that region, the verb may be understood as designating a single temporally unbounded instance of the process, as in example (6) below: (6)

Piotr pisat ksiqzkq przez dwa lata ijeszcze jej nie skonczyt Peter Imperf.-wrote book across two years and yet it not hefinished 'Peter has been writing a book for two years and hasn't finished it yet.'

In sentence (5b), on the other hand, the use of w denotes the fact that the path of the abstract motion involved in the conceptualization of the temporal extension of the process is construed as being wholly contained within the region of time. Therefore, the imperfective verb form in this sentence cannot be construed as designating a single instance of the process unbounded in time. Instead, it is construed as denoting repeated instances of the temporally bounded activity. The ultimate temporal unboundedness of the process profiled by (5b) results, therefore, from this replication, and not from the temporal unboundedness of a single instance, as it was the case with the process profiled in (5a). 6 In other words, each instance of the activity denoted by the verb is bounded within the lower-level immediate scope of predication established by the expression w dwa lata 'in two years'. Replicating such individual, bounded instances results in the temporal unboundedness of the process and hence, its imperfectivity within the immediate scope of predication at the level of the clause. This claim is confirmed by the sentences under (7): (7)

a. Piotr pisat kazdq SWQ ksiqzkq. W dwa lata Peter Imperf.-wrote each his book in two years 'Peter wrote each of his books in two years.'

496

Agata Kochanska

b.

?Piotr pisal te ksiqzkq, w dwa lata ijeszcze jej nie skonczyl Peter Imperf.-wrote this book in two years and yet it not hefinished ?'Peter has been writing this book in two years and hasn't finished it yet.'

In (7a), the conceptualization of repeated instances of an activity evoked by the word kazjda 'each' is perfectly consistent with the temporally unbounded construal of the process profiled by the imperfective verb form on the one hand, and the concept of bounding evoked by w on the other. In contrast, sentence (7b) does not seem to be acceptable, since there is a conceptual clash between the idea of a single instance of the process which is unbounded in time within the immediate scope (such a conceptualization being evoked by the imperfectivity of the verb combined with the definiteness of the object of this verb) and the bounding which is inherent in the image-schematic structure of the configuration profiled by w. Generally, the combination of an imperfective verb form with a przezphrase designating the temporal extension of the process is perfectly acceptable, since the default-case construal of processual imperfectivity is consistent with the image-schematic structure of the configuration denoted by this preposition. Combining an imperfective verb form with a w-phrase may be more problematic. The probability of construing a given imperfective verb form as being unbounded through the replication of individual temporally bounded instances of the process, rather than through unboundedness of a single instance, may be influenced by a number of factors. The degree of acceptability accorded to the combination of an imperfective verb form with the preposition w locating the temporal extension of the process within a region of time seems to be correlated with the degree of probability for such a construal.7 To illustrate this point, let us consider sentences (8)-(ll): (8)

a. Przed zebraniem Piotr przez 10 minut porzgdkowat swoje notatki Before meeting Piotr across 10 minutes Imperf.-sorted-out his notes 'Before the meeting Piotr was sorting out his notes for 10 minutes.'

The case of 'przez' and V

497

b. Przed zebraniem Piotr (zawsze) w 10 minut porzqdkowal swoje notatki Before meeting Piotr (always) in 10 minutes Imperf.-sortedout his notes 'Before a meeting Piotr (always) sorted out his notes in 10 minutes.' c.

?Wczoraj tuz przed. zebraniem Piotr w 10 minut porzodkowat swoje notatki Yesterday just before meeting Piotr in 10 minutes Imperf. -sorted-out his notes ?'Yesterday just before the meeting Piotr was sorting out his notes in 10 minutes.'

In both (8b) and (8c) above the temporal extension of the process is located within the bounded region of time space profiled by the phrase w 10 minut 'in 10 minutes'. In sentence (8b), however, assuming at the clausal level a broad immediate scope of predication evoked by the word zawsze 'always' allows for the possibility of construing the imperfective verb form as profiling the process which is temporally unbounded through the replication of bounded instances (rather than through unboundedness of a single instance). Such an interpretation is perfectly consistent with both the concept of imperfectivity denoted by the verb form and the idea of bounding conveyed by the w-phrase. Therefore, (8b) is perfectly acceptable. In sentence (8c), on the other hand, the immediate scope of predication at the level of the clause is established by the phrase wczoraj tuz przed zebraniem 'yesterday just before the meeting'. This scope seems to be too limited for replicating the temporal extension of the perfective process within its boundaries. Also, the phrase under discussion is normally understood as denoting a specific region in time space and not just a type of region which may be instantiated at various points of this space (an example of the latter would be the expression tuz przed zebraniem 'just before a meeting'). This blocks the possibility of construing the scope of predication as being a collection of points scattered in the time space and thus, having a sufficient extension. Therefore, the improbability of the replication construal results in the clash between the imperfectivity designated by the verb form and the bounding imposed by the w-phrase. This conceptual clash renders (8c) unacceptable. Another factor influencing the probability of the replication construal of the processual imperfectivity is illustrated by the sentences under (9):

498

(9)

Agata Kochanska

a. Piotr pisat list przez godzin^ Peter Imperf.-wrote letter across hour 'Peter was writing a letter for an hour.' b. Piotr (zawsze) pisat list w godzinξ Peter (always) Imperf.-wrote letter in hour 'Peter (always) wrote a letter in an hour.' c.

?Piotr pisat ten list w godzinς Peter Imperf.-wrote this letter in hour ?'Peter was writing this letter in an hour.'

Sentence (9b) is again perfectly acceptable. Sentence (9c) may also be acceptable, but only under the reading in which the definite noun phrase ten list 'this letter' refers to different versions of the same letter, that is, to different products resulting from completing the activity of writing a letter more than one time. Such an interpretation is consistent with construing the imperfectivity of the process denoted by the verb pisac 'to write' as resulting from replicating instances of this process, which in themselves are bounded, and it is precisely this construal that is imposed by the imageschematic structure of the configuration profiled by w. If, however, the phrase ten list 'this letter' is interpreted as designating just one particular physical object - the product of completing the activity of writing - the replication construal is no longer possible and the sentence stops being acceptable (cf. also the discussion of (7) above). Two more examples of unacceptability resulting from the relative improbability of giving the replication interpretation to the imperfective verb form are illustrated under (10) and (11) below: (10)

a.

Ten cztowiek podwajat majotek flrmy przez pot wieku This man Imperf.-doubled assets of firm across half century 'This man was doubling the assets of the firm for half a century.'

b.

?Ten cztowiek podwajat majotek flrmy w pot wieku This man Imperf.-doubled assets of firm in half century ?'This man doubled the assets of the firm every half a century.'

c.

Ten cztowiek podwajat majqtek flrmy w 5 lat This man Imperf.-doubled assets of firm in 5 years 'This man doubled the assets of the firm every 5 years.'

The case of 'przez,' and V

499

d. Kolejni wtasciciele podwajali majetekfirmy w pol wieku Successive owners Imperf.-doubled assets of firm in half century 'The successive owners of the firm doubled the assets of the firm every half a century.' (11)

a. Ranny Robin Hood umierai przez trzy dni Wounded Robin Hood Imperf.-died across three days 'Robin Hood was dying for three days when he got wounded.' b.

?Ranny Robin Hood (zawsze) umieratw trzy dni Wounded Robin Hood (always) Imperf.-died in three days ?'Robin Hood (always) died in three days when he got wounded.'

c.

Ranny Robin Hood (zawsze) wracat do zdrowia w trzy dni Wounded Robin Hood (always) Imperf.-returned to health in three days 'Robin Hood (always) recovered in three days when he got wounded.'

The sentences under (10) seem to suggest that the length of the span of time specified by the w-phrase may also influence speakers' acceptability judgements. In (10b), the span is half a century. This region in time space constitutes the temporal extension of the process profiled by the imperfective verb form. Since this temporal extension is comparable to the length of one human life and the TR of the process is a specific human being, construing the imperfectivity of the process as unboundedness through replication of temporally bounded instances does not seem to be probable. A replication of the process one instance of which lasts half a century is not normally thought of as being possible within one human lifetime. On the other hand, construing the imperfective verb form as denoting one instance of a temporally unbounded process is inconsistent with the topological structure of the configuration profiled by w. Therefore, sentence (8b) is normally judged unacceptable. If, however, the replication interpretation is made possible (e.g. by shortening the period of time, as in (10c), or by stating that the replication does not take place within one human lifetime, as in (10d)), the sentence may again be considered acceptable. Finally, as illustrated by the sentences in (11), the kind of activity denoted by the verb may also block the replication interpretation of the imperfectivity of the verb and thus, result in the unacceptability of a sentence. It seems that some processes are seen as one-time activities

500

Agata Kochanska

which, once completed, cannot be replicated (these are Tal my's one-way non-resettable processes (cf. 1988: 185)). For instance, a particular person, having died once, cannot die again (unless, of course, one believes in reincarnation). Such an understanding of some activities excludes the possibility of interpreting them as replicable. Thus, sentences such as ( l i b ) , if interpreted as denoting an imperfective process whose temporal unboundedness results from the replication of a temporally bounded activity, are judged unacceptable, in contrast to sentences such as (11c), in which the kind of activity denoted by the verb allows for interpreting it as replicable. It may be observed that what has been discussed so far is the influence of the configuration profiled in the domain of time by przez or w upon the construal of the imperfective forms of verbs which prototypically denote perfective, internally heterogenous processes - that is, processes whose component TR-LM relations are construed as changing through time (cf. Langacker 1987b: 79). Let us now briefly examine the way these prepositions interact with imperfective forms of verbs prototypically denoting either imperfective processes or processes which are perfective, but internally homogenous, that is, ones whose component states are construed as being effectively identical and which are bounded only externally (cf. ibid.: 85ff; see also note 3 below). As an illustration of the former case, let us consider the examples in (12) below: (12)

a.

Przez wiele lat Piotr wiedzial, co miatam na mysli Across many years Piotr Imperf.-knew what I-Imperf.-had on thought 'For many years Piotr knew what I had in mind.'

b.

?Piotr wiedziat, co miatam na mysli w 5 minut Peter Imperf.-knew what I-Imperf.-had on thought in 5 minutes ?'Piotr knew what I had in mind in 5 minutes.'

As could be expected, the imperfectivity of a verb like wiedziec 'know' is perfectly consistent with the configuration in the domain of time profiled by przez. On the other hand, integrating the unboundedness of the process with bounding imposed by a w-phrase does not result in a coherent image. Therefore, (12b) is unacceptable, unless, of course, it is given the perfective construal 'Piotr (always) got to know what I had in mind in 5 minutes'. A more interesting situation arises in the case of combining przez or w in their temporal meanings with imperfective forms of verbs prototypically profiling internally homogenous, but externally bounded processes. As an

The case of 'przez' and W

501

illustration, let us consider the sentences in (13): (13)

a. Piotr spat przez 5 godzin Piotr Imperf.-slept across 5 hours 'Piotr slept for 5 hours.'

b.

?Piotr spat w 5 godzin Piotr Imperf.-slept in 5 hours 7'Piotr slept in 5 hours.'

The combination of the imperfective form of the prototypically perfective, but internally homogenous verb spat 'sleep' with przez is, again, unproblematic. However, combining this verb with a w-phrase does not seem acceptable, no matter whether at the clausal level the verb form is construed as unbounded through replicating individual instances of bounded activities or, alternatively, whether it is conceptualized as a single instance of an activity bounded in time. What is more, even certain perfective forms of such verbs are not acceptable with w-phrases, as illustrated by (14) below: (14)

a.

?Piotr pospat w godzin^ Piotr po-Perf.-slept in hour ?'Piotr had some sleep in an hour.'

b. Piotr wyspat sit w godzinξ Piotr wy-Perf. -slept himself in hour 'Piotr managed to have his sleep out in an hour.' In both (14a) and (14b) perfective forms of the verb spat 'sleep' are used in combination with the temporal meaning of w. However, only the combination in the latter sentence seems to be acceptable. My tentative explanation of this difference in behaviour is the suggestion that the crucial factor influencing the acceptability of such combinations is the kind of bounding responsible for the perfective construal of the verb. It may be observed that the prefix po- in (14a) does not introduce any internal heterogeneity into the conceptualization of the process. Instead, it only specifies explicitly its bounded-episode character. Therefore, the verb is construed as denoting a state which is stable and uniform through time and which is bounded only externally, through contrast with other states. On the other hand, the integration of the prefix wy- with the verb spat results in the conceptualization of the process leading from the state of being sleepy and

502

Agata Kochatiska

tired to the state of being fresh and full of vigour through a series of nonhomogenous states. It may also be noted that combining a verb with w in its temporal meaning requires that the temporal extension of the process should coincide with the span of time denoted by a w-phrase. The examples under (15) may serve as illustrations here: (15)

a. Mielismy godzin^ na przeczytanie tego tekstu. Piotr przeczytat go przez godzinq. W tym samym czasie zrobil tez pewne notatki We had hour on reading this text. Piotr read it across hour. In the same time he made also certain notes 'We had an hour for reading this text. Piotr read it for an hour. At the same time he also made certain notes.' b.

?Mielismy godzinς na przeczytanie tego tekstu. Piotr przeczytat go w godzinξ. W tym samym czasie zrobit tez pewne notatki We had hour on reading this text. Piotr read it in hour. In the same time he made also certain notes ?'We had an hour for reading this text. Piotr read it in an hour. At the same time he made also certain notes.'

Example (15b), although not obviously unacceptable, sounds at least odd. Its oddity seems to result from the fact that the use of the expression w godzinξ (or any other such w-phrase) brings about the expectation that the activity was completed within the specified span of time only with the maximal amount of effort, with no free time being left. No such expectation arises in the case of the use of a przez-phrase, as illustrated by (15a). The requirement of coincidence between the span of time denoted by a w-phrase and the temporal extension of a process, together with the differences in the nature of bounding in the case of internally heterogenous and internally homogenous perfective processes seem to account for the differing behaviour of the respective groups of verbs in combination with w-phrases. For such a combination to be acceptable, an instance of a process has to be bounded within the immediate scope of predication defined by a w-phrase. In the case of internally homogenous perfective processes the fact that the limits of the process have to coincide with the boundaries of the immediate scope precludes the possibility of bounding within that scope. 8 Therefore, as illustrated by (13b) and (14a) the combination of a w-phrase with a verb denoting an internally homogenous perfective process does not seem to be acceptable, no matter whether the

The case of 'przez' and V

503

perfectivity of the process is overtly specified by a verb form, or not. On the other hand, as shown by (14b) (cf. also (4b) above), such a combination is perfectly acceptable with verbs denoting perfective processes which are internally heterogenous, since in the case of such processes bounding is inherent to each process, and does not result only from the contrast with surroundings within the immediate scope of predication. Summing up the basic points made above, it seems that expressing spatial and temporal relations by means of the same linguistic form is both an extremely common phenomenon and a problematic case for objectivist semantic theories. The fact that relations in space and relations in time are objectively quite different does not permit such theories for grouping spatial and temporal concepts within one category, and thus forces them to classify any instance of denoting a spatial and a temporal relation by means of the same form as an instance of homonymy. In contrast, the cognitive framework makes it possible to treat such cases as instances of perfectly natural and meaningful categorizing. Spatial and temporal relations, although quite different in reality, are nevertheless closely related in our conceptual system. This conceptual relatedness does not follow from the objective state of affairs in the world, but from the way we experience the world and understand it metaphorically. Conceptualizing time in terms of physical space is well-grounded in our experience and allows for understanding a more abstract and less structured domain in terms of a domain which is more tangible and more concrete. As such, it is highly meaningful and natural for the users of a given language. In the present paper, the temporal meanings of two Polish prepositions, przez and w, have been analyzed as metaphorically motivated extensions from the prototypical spatial senses. One of the aims of the paper was to show that the topological structures of the spatial configurations profiled by the basic senses of the prepositions, as well as the logics inherent in them, are preserved under the metaphorical mapping onto the domain of time, as predicted by the Invariance Hypothesis (cf. Lakoff 1990). This would seem difficult to explain on the assumption that spatial and temporal meanings of the prepositions are mere homonyms. However, if we assume that the grouping of spatial and temporal meanings within one category is an instance of metaphorically motivated polysemy and, further, that metaphor is a tool for understanding more abstract areas of our experience in terms of the more concrete, this is no longer surprising. It may be noted that the Invariance Hypothesis, as it was formulated by Lakoff (1990), has left many problems open (for a presentation of the most important ones, cf. Brugman 1990). One of the questions arising in connection with this version of the Hypothesis is the problem of the principles motivating the choice of certain domains of experience rather

504

Agata Kochamka

than others as source domains for metaphorical understanding of more abstract concepts. One possible answer here is Langacker's (cf. 1993: 3) suggestion that human beings possess certain innate image-schematic abilities which allow them to experience and conceptualize both the concrete and the more abstract areas of experience in analogous ways. Metaphorical mappings from the concrete to the abstract domains are possible by virtue of the existence of such abilities, which are not extracted from the basic bodily experience, but rather "initially manifested" (ibid.) in it. In this view, space and time, for example, are pre-metaphorically conceptualized in an analogous way as extensional domains (cf. Langacker 1987a: 153) thanks to our innate mental ability "to scan mentally from one point to another in any kind of representational space" (Langacker 1993: 3). Therefore, the similarities in the image-schematic structures of these two domains are present in our direct experience and are not a result of the metaphorical mapping. Instead, they are a precondition for its occurrence. Of course, any direct understanding of an abstract domain such as time must be fairly schematic, and thus there is a need for a metaphor mapping a more detailed structure from a domain which is experientially more basic and in which image-schematic abilities are first manifested. If we postulate that human beings are equipped with certain innate image-schematic abilities, then expressing relations in time by means of spatial prepositions may be seen as motivated not only by the metaphorical linking of temporal senses to the spatial prototypes, but also by the existence of schemas which capture commonalities between different subcategories of senses and which result from an analogous image-schematic experience and conceptualization of different domains. Therefore, in this view it is no longer surprising that the details of spatial configurations profiled by the prototypical senses of przez and w are also relevant to the conceptualizations evoked by their respective temporal meanings. As the present paper aims to show, the image-schematic structures of the spatial configurations profiled by the basic senses of these prepositions seem to influence, among other things, the choice of construal of the imperfectivity of verbs combined with temporal przez- or w-phrases. Also, the very possibility of combining verbs with such phrases appears to be conditioned by a degree of compatibility between the image-schematic structure of the preposition and the construal of aspect evoked by the verb form. If we assume Langacker's suggestion concerning the existence of innate image-schematic abilities to be valid, such observations are not surprising, but rather to be expected, since they may be considered to be direct consequences of conceptualizing objectively quite disparate areas of experience through the use of the same mental mechanisms.

The case of 'przez' and W

Notes

1.

2.

This paper is a revised version of a fragment of my unpublished M.A. thesis. I would like to express my special gratitude to Dr. Elzbieta Görska, who was the supervisor of this thesis. I would like to thank her for all the help she has offered me so far, and specifically for her directing my attention to the fact that Polish homogenous perfective verbs whose forms are explicitly marked for their perfectivity behave in a way parallel to verb forms which are traditionally called 'imperfective', but which denote externally bounded perfective processes. I would also like to thank Professor Günter Radden for his comments upon my analysis of przez, especially for his remarks concerning the conceptualization of the flow of time in terms of the subjective movement on the part of the conceptualizer. Words of thanks are due to two anonymous reviewers of the earlier version of this paper for all their valuable remarks and suggestions, particularly those related to the interesting questions and problems arising in connection with Lakoffs Invariance Hypothesis. Thanks to them the problem of how our direct experience of certain abstract domains restricts the possibilities of metaphorical mappings has once again been brought to my attention. And last, but not least, I would like to express my gratitude to Prof. Emma Harris, who kindly agreed to check and correct my English. Of course, any flaws and errors in this paper are entirely my own. It should be noted that the terms 'atemporal' and 'temporal' are used here in a sense characterizing the mode of scanning adopted in conceptualizing a given relation. This sense should not be confused with the meaning 'related to time', which is widely used throughout this paper. I would like to thank one of the anonymous reviewers of the earlier version of this paper for pointing out to me this potential source of misunderstanding, as well as the fact that the atemporal construal of przez itself should be clearly distinguished from the positive temporal profile at the clausal level, at which the preposition is integrated with a verb. Combining the component structures into the composite clause is achieved in several steps. At the first level of constituency a correspondence is established between the schematic LM of przez 'across' and the profile of the specific nominal plac 'square'. The nominal elaborates the LM of the preposition. The composite structure profiles a complex atemporal relation since it is the preposition which is the profile determinant at this level. Next, the path designated by the prepositional phrase is put into correspondence with the more schematic path present in the profile of the verb. Finally, a correspondence is established between the ra of przejsc ' Perf. -prze-go'and the TR of przez. These schematic TRs are elaborated by the profile of the specific nominal Piotr. Since it is the verb which is the profile determinant at the clausal level, the clause has a positive temporal profile. It may be noted that this description of the composition of the clause is, for the sake of brevity, largely simplified, in that it ignores e.g. the semantic value of the verb's ending (3rd person, singular, masculine, past tense) and the way it integrates with the rest of the clause. Also, the verb is treated here as an unanalysable whole, although the presence of the prefix prze- definitely adds to the semantic value of the composite structure.

505

506

3.

4.

Agata

Kochanska

Another pervasive way of understanding time in terms of movement in physical space is what Radden (cf. 1991: 17) calls the MOVING TIME metaphor: the conceptualization of moments in time as physical objects moving with respect to the stative observer. This variant of the spatial metaphor of time is not directly relevant to the topic of the present paper and therefore it will not be discussed here. It may be observed that in sentences such as (4b) an NP after w 'in' is used in the accusative case. This might suggest that such a temporal meaning of w is more directly linked to the dynamic spatial sense of this preposition (the goal sense), than to the prototypical stative meaning, in which w combines with NPs in the locative case: pierscionek w pudetku ring in box-Loc.

PLACE

'the ring in the box' wtozyc pierscionek w pudetko put ring in box-Acc. 'to put the ring into the box'

5.

GOAL

It should be noted that the temporal w-phrases in which the preposition is combined with a noun in the Locative and the ones with nouns in the Accusative behave differently with respect to the construal of the aspect of verbs. However, any investigation of the semantic differences between these two constructions responsible for this differing behaviour is beyond the scope of the present paper and will be ignored in the following discussion, which deals only with the W-NP-Acc-phrases. It may be observed that what is traditionally referred to as the imperfective verb form in Polish does not always have to profile an imperfective process in Langacker's sense, as is the case in (5). To illustrate, let us consider sentence (i) below: (i) Wczoraj Piotr shichat muzyki i spacerowatz psem Yesterday Piotr Imperf.-listened music and Imperf.-walked with dog 'Yesterday Piotr was listening to the music and walking with his dog' 'Yesterday Piotr listened to the music and walked with his dog' The verb forms used in (i) are imperfective. However, the verbs stuchac (muzyki) 'listen (to the music)' and spacerowac (z psem) 'walk (with a dog)' prototypically denote processes which, although internally homogenous (not profiling any change through time), occur in bounded episodes (external bounding) and hence, are perfective (cf. Langacker 1987b: 85ff). Adopting a sufficiently broad immediate scope of predication may allow the conceptualizer to construe the whole episode of the process together with its boundaries as falling within that scope, thus rendering the process perfective. Since the immediate scope of predication in (i) may be construed as sufficiently broad (cf. the second English equivalent above), it is possible to conceptualize the processes denoted by verbs as perfective. This interpretation seems to be confirmed by the fact that the verbs in (i) do not profile processes characterized by ex-

The case of 'przez' and V

507

pansibility, in contrast to verbs prototypically denoting imperfective processes (like e.g. lubic 'like'): (ii) ?Wczoraj Piotr stuchat muzyki i nadal stucha Yesterday Piotr Imperf.-listened music and he still listens •'Yesterday Piotr listened to the music and he is still listening.' (iii) ?Wczoraj Piotr spacerowat ζ psem i nadal ζ nim spaceruje Yesterday Piotr walked with dog and still with him walks •'Yesterday Piotr walked with his dog and he is still walking with him.' (iv) Piotr lubif swa tesciow^ w dniu slubu i lubi nadal Piotr liked his mother-in-law in day of wedding and likes her still 'Piotr liked his mother-in-law on the day of his wedding and he still likes her.'

6.

7.

8.

It seems that in the examples above the English translations are more obviously unacceptable than their Polish counterparts, which just 'sound odd' if the normal course of events in the world is taken into account. At least two factors may be responsible for this difference: (a) the differing meanings of the past tense morphemes in English and in Polish respectively (the Polish morpheme may denote both the past and the present perfect sense when combined with an imperfective verb form); (b) the fact that Polish imperfective past-tense forms of verbs may convey a conceptualization in which either an 'internal perspective' is taken on the profiled process (cf. Langacker 1987b: 85) resulting in its imperfectivity, or, alternatively, a broad immediate scope of predication is adopted, which makes the process bounded within that scope. This is, of course, only a very tentative suggestion since I haven't investigated the problem in any detail. As pointed out by Langacker, there is a close parallel between the perfective/imperfective distinction for verbs and the count/mass distinction for nouns (cf. 1987b: 80; after Mourelatos 1981). It may be noted here that if the imperfectivity of the process construed as temporal unboundedness of a single instance parallels a mass noun, then imperfectivity through replication is similar to a plural noun, whose ultimate mass construal (that is, unboundedness in the primary domain) also results from replicating individual instances, and not from the unboundedness of a single instance. Alternatively, imperfective verb forms may be construed as profiling processes which are temporally bounded. Such a construal, although in conflict with the conventional meaning of the imperfective verb form in Polish, is nevertheless consistent with the topological structure of the configuration denoted by w, and may therefore result in an increasing degree of acceptability. In my view, the situation of combining an internally homogenous perfective verb with a w-phrase denoting the temporal expanse of a process is analogous to Langacker's example of perceiving a red spot (cf. 1987b: 59f): if, by any chance, my visual field coincides with the boundaries of an otherwise homogenous spot, this perceptual experience will evoke the conceptualization of an unbounded mass. Therefore, such a con-

508

Agata Kochanska

ceptualization would be coded linguistically as I see red, and not as I see a red spot. Similarly, the coincidence between the limits of an internally homogenous process and the boundaries of the immediate scope of predication makes the process bounded by the scope, and not within it.

References Brugman, Carol 1981 Story of Over. M.A. thesis, University of California at Berkeley. 1990 What is the Invariance Hypothesis? Cognitive Linguistics 1-2: 257-266. Lakoff, George 1987 Women, Fire and Dangerous Things. What Categories Reveal about the Mind. Chicago: The University of Chicago Press. 1990 The Invariance Hypothesis: Is abstract reason based on image schemas ? Cognitive Linguistics 1-1: 39-74. Langacker, Ronald W. 1987a Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. 1987b Nouns and verbs. Language 63: 53-94. 1993 The reference-point constructions. Cognitive Linguistics 4-1: 1-38. Mourelatos, A.P.D. 1981 Events, processes and states. In: Tedeschi, P.J. and A. Zaenen (eds). Syntax and Semantics. Vol. 14: Tense and Aspect. New York: Academic Press, 191-212. Radden, Günter 1991 The cognitive approach to natural language. Series A. Paper No.300. Duisburg: L.A.U.D. Talmy, Len 1988 The relation of grammar to cognition. In: Rudzka-Ostyn, Brygida (ed.). Topics in Cognitive Linguistics. Amsterdam: Benjamins, 165-205. Taylor, John R. 1989 Linguistic Categorization. Prototypes in Linguistic Theory. Oxford: Clarendon Press.

Viewpoint and subjectivity in English inversion Heidrun Dorgeloh

1. Introduction Within the English word order system, where the position of the noun phrase before the verb is interpreted as marking the grammatical subject, only a limited range of reordering options, and thereby sentence patterns with deviant order of major sentence constituents, are available to the choice of the speaker. One of these options is inversion, in which "the NP of which some location, status, or activity is being predicated (the logical subject) appears to the right of the verb; [and] some other constituent appears to the left of the verb, where the subject would canonically appear" (Birner 1994: 234). The range of possible VS-constructions covered by this definition and as treated in the analysis here is exemplified in (1) - (3), also illustrating that not included are superficially similar constructions such as subject-auxiliary-inversion or more specialized types, for instance quotation inversion (Green 1982: 124) or inversion with no preposed constituent other than the verb: (1)

The economy, Berlin, the Common Market - here are three issues whose gravity has during the past few days led to regretful sighings over the impracticability of a National Government, (LOB, press: editorial)

(2)

Once upon a time there was a lake in the mountains, and in that lived a huge crab. (W. H. D. Rouse, The Giant Crab, p. 1)

(3)

These are shocking figures, but even more shocking is the fact that at least half the people behind these crimes will go undetected. (LOB, press: editorial)

It has become customary to refer to the prototypical inversion type exemplified by (2) as "locative inversion" (cf. Bresnan 1994) or "focus

510

Heidrun Dorgeloh

inversion" (cf. Levine 1989). However, in that locative implies a basic operation of locating discourse entities mentally or conceptually — discourse entities that in this way carry the sentence focus (cf. Enkvist 1980) and potentially become the new focus of attention in discourse (cf. Vallduvi 1992) — one can in principle extend this analysis to inversion after deictic adverbs (as in (1)) and following nonlocative constituents (as in (3)). In fact, I shall explicitly argue here that all three inversion types — after deictic adverbs and following locative as well as nonlocative constituents -- are very much related in kind. I shall analyze these inversions as a device of expressing point of view in discourse (cf. Ehrlich 1990), in that they contain a reference point as vantage point from where a scene or event is seen. They thereby establish a particular perspective which the speaker assumes. This analysis will further be differentiated using Langacker's (1990) concept of subjectivity. It can furthermore be shown that in written discourse, or discourse under conditions of displacement (cf. Chafe 1992) in general, the reference point expressed in an inversion is most often one from within the universe of discourse, namely one of a discourse entity previously established — or of a "ground" previously "set" (cf. Drubig 1988). Since I consider as the basic function of the front-shifted constituent involved in an inversion to induce a change of ground, my analysis does not in principle contradict other claims made about the discourse function of inversion, in particular not an information-packaging analysis of it (cf. Birner 1992, 1994). However, the approach taken here adds a component of (non-propositional) meaning involved that is apt to account for what are so far apparent imponderables about the usage of inversion in naturally-occurring discourse: although previous studies of inversion have plausibly argued that it reflects a preference for "presenting information which is more familiar in the discourse before information which is less familiar" (Birner 1994: 255) or serves an overall "presentative function" (Rochemont 1986, Langacker 1993), most of them have in fact neglected the question why in many contexts — types or parts of a discourse — one basically finds no occurrences of inversion at all (e.g. in most news texts), while in others (e.g. guidebooks, children's tales) often occur virtual clusters. By analyzing inverted constructions as expressing a point of view in discourse, by contrast, it becomes possible to make some predictions under which discourse conditions inversions are more likely to be used. This analysis also sheds some new light on previous treatments of inversion as a "root transformation" (Emonds 1969) or a "main clause phenomenon" (Green 1976). Such a status has generally underlain claims for a putative incapacity of inversion to occur in embedded clauses, though, from the onset of this discussion, additional pragmatic or discourse factors always

Viewpoint and subjectivity

511

had to be brought into play (Emonds 1969, Hooper and Thompson 1973, Green 1976). It can be shown that considering the point of view involved in an inverted clause and, in particular, to claim that this viewpoint has to be one which belongs to the speaker rather than to the (main clause's) subject accounts for most examples of acceptable inversion embedding, and for many examples of more controversial acceptability, where the matter of embeddability appears to be highly context-sensitive. Finally, I shall discuss two example passages from newspaper reports which display specific characteristics that make them hospitable to occurrence of inversions within them. In particular, when dealing with events that are "displaced" from the immediate environment of speaker and hearer (as is the case in news reporting, from which I draw my examples), the effect of an inversion can be described as adding an effect of "displaced immediacy" (Chafe 1992) to the contents of a discourse. In that, its occurrence often correlates with other linguistic forms symptomatic of a certain resemblance to immediate experience which a text can maintain. It will also be shown that the kind of displaced immediacy signalled by inversion usage produces subjectivity, while, alternatively, a speaker could also refer to previous experience objectively.

2. Viewpoint and subjectivity in different inversion types

2.1. From a deictic to a lexical presentative mechanism As argued in Drubig (1988), there is a clear interrelationship between inversions after here (and there, if stressed), i.e. following a deictic adverb pointing to a shared perceptual field, and inversions following more complex (— lexical) fronted constituents. The former type he calls the "deictic type of SVI [= subject verb inversion]", while non-deictic constituents are followed by a "lexical type" of inversion (1988: 91): [They] can be said to be complementary in distribution and discourse function: deictic svi is used in a situation of optimal perceptual access ..., where pointing is possible, whereas lexical svi typically occurs in non-optimal situations characterized by the relative distance or displacement of one speech act participant... from the region of optimal perceptual access.

As can be assumed on the basis of Birner's (1992, 1994) findings about the relative discourse-familiarity of the constituents affected by an inversion

512

Heidrun Dorgeloh

on the one hand, and of the kind of verbs involved ("informationally light", in particular copular be or verbs of appearance and existence; cf. Birner 1992: 191ff.) on the other, both types of construction can indeed be said to be presentative in kind: They consist of a scene-setting process and then direct the subsequent focus of attention to a discourse entity or topic newly or reintroduced. But while, in the deictic type of inversion, reference is made to a shared perceptual field, in lexical inversion, the reference point of the presentative mechanism is explicit lexical material, i.e. the addressee is provided "with all the lexical information necessary to construct a mental model of the reference situation" (Drubig 1988: 91). This is thus a process which relies on conceptualization, as opposed to the perceptually accessible situation to which deictics refer. Drubig (1988: 92) concludes that ... the function of lexical svi appears to be the inverse of that of the deictic type: it is an instruction for the addressee not to look, but to mentally construe the content of the svi sentence as an appearance, which allows him to process the information as if he were witnessing it.

If one considers (4a) vs. (4b) and (5a) vs. (5b), it becomes evident that there is nonetheless a crucial difference between the two kinds of construction: (4a)

Here is the paper I promised you last week.

(4b)

The paper I promised you last week is here.

(5a)

In front of the house is a tree.

(5b)

The tree is in front of the house.

The versions containing a deictic adverb do not necessarily possess the same truth-conditions in inverted as opposed to canonical word order. If, for example, (4a) is written to somebody in an e-mail message, the paper is expected to follow afterwards, i.e. in the minimal shared perceptual domain of the e-mail context. (4b), by contrast, is true if the paper is located where the speaker is. Due to this difference in meaning, one can consider such deictic inversions as presentative constructions in which the speaker-based reference, which can also be of a textual, "endophoric" (cf. Halliday and Hasan 1976) kind, has become grammaticalized, hence the difference in meaning produced may have an effect on the truth-values of the two versions.

Viewpoint and subjectivity

513

The presentative mechanism starting out from a lexical constituent creates a similar endophoric link to a prior discourse model, even though the link need not be overtly evoked, but may probably only "cue" (cf. Birner 1994: 255) the hearer to evoke it from very sentence-initial placement. However, since the scene-setting process consists in fully specifying by lexical means the ground from where the presentative mechanism is to start, the predication in principle remains the same; thus, there is no real truthconditional difference between (5a) and (5b). Instead, the difference in meaning which arises can be described as one of viewpoint or perspective, and (5a) and (5b) make this difference particularly explicit: In (5b), the tree is located in the viewer's line of sight, but what remains unclear is whether the side where the tree stands is the front or the back side of the house. An inversion as in (5a), by contrast, does not create such an ambiguity, because the scene is viewed from the house as its reference point, the location of the tree thus defined in relation to it.

2.2. Viewpoint analysis of different lexical inversion types This mechanism of expressing a viewpoint or perspective can be applied to different types of English inversion (for instance, to (2) and (3) from above), and most clearly to directional constituents such as in (6a) and (7a). Both examples display an effect of movement of the discourse entity introduced in relation to the speaker, which is not present in canonical word order, for instance in (6b): (6a)

Into the room sneaked a boy and was immediately hit on the head by me.

(6b)

—A boy sneaked into the room and was immediately hit on the head by me. [ = Kuno 1987: (8.5b)]

(7a)

Toward me lurched a drunk. [ = Bresnan 1994: (16a)]

(7b)

* Toward me looked a drunk. [= Bresnan 1994: (16b)]

(6a) is a variant of (6b), which has been taken from Kuno's (1987) illustration of the empathy phenomenon, or — using his metaphor — of a "camera angle" (1987: 203ff.), from where an event is seen. 1 The examples show that inversion affects empathy or the "position of the camera"; the change in viewpoint could accordingly be described as a "camera movement". While (6b) not only violates Kuno's "Topic Empathy

514

Heidrun Dorgeloh

Hierarchy", but, more important, the "Speech Act Empathy Hierachy" (1987: 212) he puts forward, i.e. the strong tendency of the speaker to empathize with himself, the sentence becomes more acceptable with an inversion as in (6a); the requirements of speaker empathy are more likely to be met here, since, by viewing the event from the room as reference point, one could say that the speaker "has moved his camera closer". (7a) as opposed to (7b) have been discussed within the framework of Lexical Functional Grammar (Bresnan 1994), where so-called "split intransitivity" has been noted in that "locative inversion in English ... does not apply to ALL intransitive verbs" (Bresnan 1994: 78, her emphasis). Apparently, directional prepositional phrases require a dynamic verb, which can now be accounted for following the viewpoint analysis undertaken above: in (7a), there is movement of the discourse entity in relation to the speaker, and inversion thereby acceptable: lurching is an event which the speaker can perceive as a participant on the scene, while simple looking, as in (7b), seemingly is not. That an implicit reference point, which the speaker can assume as his vantage point or "position for the camera", is also involved in inversions following adjective phrases is illustrated by examples (8) - (10): (8)

More troublesome to German planners is the country's inflation rate, which last month crept above the 4% annual pace tolerated by the Bundesbank, prompting the sharp interest-rate hike. {Time, 20 Jan. 92, p. 9)

(9)

These are shocking figures, but even more shocking [to the commentator] is the fact that at least half the people behind these crimes will go undetected, (LOB, press: editorial)

(10)

Most charming [to the reviewer] was· his tongue-in-cheek "Unheimliche Geschichten" (1920). five ghost stories with a light touch, [ . . . ] . ( L O B , press: review)

Here, the fronted constituents followed by inversion express nonlocative properties assigned to the subject. However, they also involve a reference point to which the presentative mechanism can be "anchored" and which the speaker can assume as his or her vantage point. This reference point may be named explicitly, as in (8), but more often is it the case that, due to the special semantic class to which most fronted adjectives followed by an inversion belong — normally they bear a meaning of comment or evaluation —, these are implicitly anchored to an underlying experiencer (as in (9)), or gain this evaluative meaning by virtue of being fronted, as in (10).

Viewpoint and subjectivity

515

Many fronted adjectives followed by an inversion appear in comparative form and in fact "seem to be more acceptable than their positive counterparts" (Penhallurick 1984: 43). Ultimately, in this case, they can be said to have a twofold anchor. Comparatives always express a form of reference — likeness being a referential property (cf. Halliday-Hasan 1976), since "a thing cannot just be 'like1; it must be 'like something'" (1976: 78) — and as such this point of comparison is also available as reference point linking the inversion to prior discourse. This point of comparison is sometimes given explicitly, of which (11) is an example: (11)

More important than Japanese concessions, many business executives on the trip believed, was the event's symbolism - a point that even the disappointed Big Three chairmen underscored. {Newsweek, 20 Jan. 92, p. 24)

By contrast, if no basis of comparison is mentioned as an explicit reference point, an adjective phrase is minimally anchored to the speaker himor herself (as in (9) and (10)) — subjectively, as will be argued in section 2.4. —, which is due to the fact that the attribution of nonlocative properties such as those found with inversion always involves an inherently speakerbased process of comment or evaluation.

2.3. Speaker's viewpoint and embedding The demonstration of point of view involved in English inversion shall next be substantiated by observations about the (non-)embeddability of these constructions. As already noted, inversions of the kind discussed here cannot be said to be entirely excluded from subordinate clauses, though evidence from naturally occurring discourse speaks for a nonetheless strong preference to invert main clauses almost exclusively.2 However, consider examples (12) - (14): (12)

Van Dijk (1982), for example, argues that between the unity of the sentence and the unit of the text or conversation exists a further unit of analysis, which he calls the episode: [...] (Ehrlich 1990, p.27)

(13a)

John says that standing in the corner is a man with a camera. and I think he's right. [ = Green 1976: (34a)]

516

Heidrun Dorgeloh

(13b)

*John says that standing in the corner is a man with a camera. and he's wrong. [ = Green 1976: (34c)]

(14)

He has forgotten that over the fireplace hangs a picture of his late wife. [= Birner 1992: (8), p. 87]

Hooper and Thompson (1973) have claimed that the main function of root transformations is to produce emphasis and "emphasis would be unacceptable in sentences that are not asserted" (1973: 172f.); hence the fact that they are not applicable in presupposed sentences. However, at least (12) looks very unemphatic to me, nor does the assertion criterion explain the acceptability of (14). A related suggestion, put forward by Green (1976), has been speaker's agreement as exemplified in (13a) vs. (13b), an hypothesis which probably comes closer to the truth. However, agreement is a quite vague contextual condition, and is obviously rare to be found overtly stated; thus, it may be worth reformulating this condition in terms of a speaker's point of view contained in an embedded predication. Following point of view theory (Banfield 1982, Ehrlich 1990), sentences with embedded-//?«? clauses in principle have two possible readings: one in which the speaker's point of view is represented and one in which the subject's point of view is represented (cf. Reinhart 1983: 170). Which of these potentialities is realised is generally dependent on the context as well as on the meaning of the verb. What one can show is that inversion is always acceptable if the embedded clause reflects the speaker's point of view: for instance in (12), the speaker uses the quote to develop her own line of thought, and in fact probably uses the inversion to virtually signal that in the indirect quote are united her and the quoted source's point of view. In the same vein, (13a) is acceptable in contrast to (13b), because the context of (13a) makes it clear that the subject's point of view and the speaker's point of view are compatible. That the necessary condition is not one of asserted as opposed to presupposed material is shown by (14): here, the inversion occurs in the complement clause of a factive verb and is therefore presupposed rather than asserted. What seems to be important, by contrast, is that the context allows for the speaker to make the predication in the embedded clause, to assume open responsibility for it. That is, the predication is not projected to the subject's perspective only, but is presented as a fact which the speaker is overtly aware of, while the subject's perspective is blocked due to the semantics of forget. In other words, the speaker makes this predication from his or her own point of view. Further evidence for these claims is found in (15a) as opposed to (15b), as well as in (16a) as opposed to (16b): inversion with a first person factive

Viewpoint and subjectivity

517

is acceptable in contrast to a third person factive, and is acceptable rin causal adverbial clauses if these represent an external reason or are non-restrictive, but not in internal or restrictive causal adverbial clauses: (15a)

We can support the claim that standing in the corner was a black umbrella. [= Green 1976: (56b)]

(15b)

ΊΊ

(16a)

Something must have happened because in came my father.

(16b)

*/ had to hide the book because in came my father.

·Ηε supports the claim that standing in the corner was a black umbrella.

To conclude on inversion in embeddings, I want to explicitly emphasize that judgements about their acceptability are a very vulnerable matter. This is reflected by all treatments in the literature quoted and best illustrated by Hooper and Thompson's (1973: 479) claim that, in the end, "a few speakers allow RTs [root transformations] in all that-S complements". Taking into account that embedded-that clauses may have a "speakerorientation" and/or a "subject-orientation" (Ehrlich 1990: 12) is quite apt to account for these uncertainties: context permitting, it is in the end the speaker's choice — or the judger's imagination — that shifts an embedded predication into the perspective of the subject or, alternatively, reconciles it with the perspective of the speaker him- or herself. Embedding contexts where the perspective of the subject to the definite exclusion of the speaker's point of view is least plausible are therefore those that most easily embed an inversion; hence the fact that a relatively large proportion of counterexamples discussed in the literature contain a first person pronoun or an epistemic matrix clause.

2.4. Subjectivity in lexical inversion On the basis of these illustrations of inversion being a device by means of which the speaker adopts the viewpoint of the fronted constituent, or, more precisely, of the reference point or discourse entity contained in it, one can now further analyze the different scene-setting processes in terms of Langacker's (1990) distinction of subjective and objective mechanisms of construing discourse participants. Langacker himself (cf. 1990: 6) uses the term "perspective" to subsume "orientation" and "vantage point", which reflects a close affinity with the point-of-view metaphor used so far. His

518

Heidrun Dorgeloh

subjective-objective dichotomy in a quite technical sense, however, allows for the treatment of any viewing arrangement, or of any construal of discourse entities, in terms of viewpoint or perspective, so that it can here serve to compare the processes of discourse entity construal involved in inverted as opposed to canonical word order constructions. The basic idea in Langacker's approach is that discourse entities as elements of conception are accessed ("conceptualized") by the language user ("the conceptualizer"), who is the subject of conception. In the "optimal viewing arrangement" (cf. 1990: 7), this subject, i.e. the speaker or viewer, is only implicit, non-salient ("offstage"), hence subjectively construed. This situation Langacker compares to the wearing of glasses which "fade from my conscious awareness despite their role in determining the nature of the perceptual field" (1990: 7). By contrast, discourse entities which are placed "onstage" in this viewing arrangement independently of the ground of the language user are objectively construed. In this sense, the construal of discourse entities by means of lexical inversion is at first sight entirely objective. Other than with deictic inversion, the reader's focus of attention is directed to new topic entities by mere reference to the conceptual scene expressed via language, i.e. to the universe of discourse established. The speaker or viewer, by contrast, remains maximally subjective, with the result that "the inherent asymmetry between a perceiving individual and the entity perceived" (1990: 7) is maximized. Compare (2a) and (3a) from above to the variants given in (2c) and (3b), or examples (17) as opposed to (18). In the cases of inversion, the speaker remains subjective and "offstage" and uses the fronted constituent, i.e. a part of the conceptual scene, as reference point of the presentative mechanism; while in (2c), (3b) and (18), some part of the ground, e.g. the speaker and/or the hearer, is overtly put "onstage" ("profiled", cf. Langacker 1990: 10). In that case, the ground itself becomes salient, objectively construed: (2a)

In front of the house is a tree.

(2c)

In front of the house I seel you find/ we see ... a tree.

(3a)

These are shocking figures, but even more shocking is the fact that at least half the people behind these crimes will go undetected. (LOB, press: editorial)

(3b)

These are shocking figures, but even more shocking / consider/ you will find/... the fact that at least half the people behind these crimes will go undetected.

Viewpoint and subjectivity

519

(17)

Next come the German Expressionists and the paintings of Nolde and Munch, in particular, have been carefully selected to indicate the important role that this School played in the formation of 20th century art. (LOB, scientific writing)

(18)

And then we come to the British section which is very revealing (for the French public anyway) in that the accent is much more on arts and crafts than on painting and sculpture. (LOB, scientific writing).

Another more objective construal of the ground would be (3c), because explicit reference to the ground by use of a personal pronoun represents "a more detached outlook in which the speaker treats his own participation as being on a par with anybody else's" (1990: 20); cf. also the parallelism with (8) from above: (3c)

These are shocking figures, but even more shocking to me is the fact that at least half the people behind these crimes will go undetected.

(8)

More troublesome to German planners is the country's inflation rate, which last month crept above the 4Ψο annual pace tolerated by the Bundesbank, prompting the sharp interest-rate hike. (Time, 20 Jan. 92, p. 9)

Following Langacker, the increase in objectivity as displayed, for instance, in (3c) has iconic reasons: "the formal distinction between overt and covert reference to the ground [...] iconically reflects its being construed with a greater or lesser degree of objectivity" (1990: 20). Thus, by way of lexical inversion, the speaker can construe discourse entities without letting his or her own ground come "onstage" in the text. One can therefore consider them a device of subjectivity, in that they express a viewpoint from within the text and do not let the speaker come objectively into the open. By means of word order, they nonetheless establish empathy with the reference points given and from here focus a new discourse entity. Written texts, in particular, in which normally a conceptual rather than the actual scene of the speaker is the focus of attention, are prone to managing their own topic supply in this way, they thereby look entirely objective, but contain a perspective which is more subjective than the one expressed by canonical word order clauses.

520

Heidrun Dorgeloh

3. Lexical inversion and displaced immediacy

3.1. Immediacy, displacement and displaced immediacy Discourse dealing with matters which are not part of — are displaced from - the immediate environment of both speaker and hearer is characterized by conditions of displacement, and may be termed displaced speech. Here, the context of communication as "experience that is immediate" (Chafe 1992: 231) is not, or only marginally, the focus of attention; instead, the consciousness of language producers and receivers focusses on "experience that is displaced from the immediate environment in which the possessor of the mind is located" (1992: 231). It can be shown that the kind of subjective focus management that inversions perform is particularly prone to such discourse conditions. What they contribute to displaced speech is that they treat displaced matters without letting the discrepancy between them and the immediate consciousness of the speaker be openly felt. Displacement as a property of human language has generally been defined as the capacity of language - in Chafe's (1992) terms of the "consciousness" of language producers (and receivers) — to "alternate its focus between experience that is immediate and experience that is displaced" (1992: 231). The author of a discourse is first of all the proximal consciousness of speech and operates in the immediacy of the speech situation, while a person who has experienced certain events at another location or time, which can now be reproduced via language, has a displaced consciousness of his own. If an author not only focusses on displaced matters, which — encoded into language — normally obtain the nature of events, but also on another, displaced consciousness, the events are presented in a way similar to immediate experience, the effect of which is called displaced immediacy. Figure 1 illustrates that, while experience is normally made in the immediacy of a situation and events are reported under conditions of displacement, one may speak of displaced immediacy if experience-like properties are maintained under conditions of displacement: immediacy

displacement displaced__ ' immediacy

experience

events

Figure 1. Immediacy and displacement reproduced into language

Viewpoint and subjectivity

521

3.2. Immediacy and displacement in news reporting It is one of the characteristics of journalistic being non-fictional discourse that the proximal consciousness of speech, i.e. the journalist, is a real person. Nonetheless, the main objective of, for instance, news reporting is to inform about political events; thus, journalists mainly focus on displaced matters and rarely on the immediacy of their own situation. Prototypically, they then treat political facts as events, and not as displaced experience of their own; in fact, it rarely is their own experience which they write about, but rather about information mediated through a number of instances, such as press agencies, eyewitnesses, official institutions and other "informed sources". As a consequence, journalistic writing about displaced political events often involves quite an extreme deviation in terms of what Chafe (1992: 236ff.) discusses as "three properties of immediate experience". First, in terms of continuity, normal news is presented in "isolated chunks" and thereby often in a form diametrically opposed to the "continuous" nature of experience. Second, this also implies that — instead of a fine-grained solution — the richness of detail available in the immediacy of an event happening is strongly reduced and very selectively treated. And, finally, deictic elements "that serve to locate experience in space, time, and with respect to an ego" (1992: 239) are virtually absent. Facts are in this way presented as entirely displaced events; the speaker, however, where present at all, is not displaced in his consciousness but — due to the irrelevance of his own position as mere mediator of information — lets his language flow from a location in space and time recognizably different from the events being talked about. A study of the usage of inversion over various newspaper genres has shown (cf. Dorgeloh 1994) that in the default case of reporting on mere events, i.e. in texts which maintain no or little linguistic indication of immediate experience, one can normally not expect to encounter inversions either. Primarily, this goes for those texts which constitute real or so-called "hard" news. 3 In some other newspaper genres, by contrast, the proximal consciousness of the journalist may enter "a displaced mode, where it focusses on another, distal consciousness" (Chafe 1992: 235). This is particularly plausible in the case of articles written by correspondants, who have experienced the events they write about "on the spot" -- or pretend to have done so. This means that one finds the author not only focussing on the events, but also on the experience or quasi-experience itself, an effect which is achieved by adding to the bare information about political news some of the characteristics of immediate experience in terms of temporal continuity, richness of detail and the use of deictic expressions. The result

522

Heidrun Dorgeloh

is that an effect of displaced immediacy comes about, where "language seems to flow from the displaced consciousness of the person producing it" (1992: 234), i.e. the author not only focusses on displaced events, but also on another, distal consciousness. By assuming this position of another consciousness, the speaker's ground can be said to shift from a journalist's position as a reporter to a previous — or a pretended - position as an experiencer. Since the nature of an experience is created by reference to the ground (of the speaker, who is the previous or claimed experiencer), this ground can now be construed objectively or subjectively, following Langacker's (1990) distinction introduced above. As explicated there, lexical inversion always involves the speaker remaining entirely subjective, and topic construal being performed objectively; hence, it contributes to a subjective construal of the ground or reproduction of experience. Thus, while it can potentially occur in all kinds of a discourse concerned with displaced immediacy rather than with displaced events only — including contexts which choose objective reference to the ground —, the actual occurrence of inversion in fact contributes to the creation of displaced immediacy in a subjective manner. The two alternative mechanisms are illustrated in the following passages (19) and (20): (19)

WELCOME to Travnik - Have a Nice Stay, read the sign as our Warrior sped into the town. We travelled along the main street, past broken windows, buildings riddled with bullet-holes, destroyed by shells. People waved their appreciation. [...] We pass the most shelled building in the town - the hospital - and on to the frontline village ofTurbe -[...]. / am with the Colonel Bob Stewart, commanding officer [...]. It is one of the best organised, most professional British operations I have ever seen - and one we can be proud of. My journey from Split on Sunday to Sarajevo brought home to me the sheer scale of the operation [...]. (The Times, 15 Dec. 92, p. 1)

(19) displays spatio-temporal displacement only in that the proximal consciousness of the correspondant, objectively construed by means of first person pronouns, focusses on his or her own displaced experience. An effect of displaced immediacy nonetheless comes about due to the maintenance of relative continuity, a richness of details rather unusual for ordinary news reporting, and most explicitly in the switch to immediate deixis in form of the present tense in the middle of the report. These are thus symptoms of a context where inversions are more likely to be found, even

Viewpoint and subjectivity

523

though, given that there is always the objective alternative, there is no instance of an inversion triggered by a locative reference point in this passage. The effect of displaced immediacy gets lost once the speaker sets in to reflect upon this experience (My journey from ...), which flows from a consciousness recognizably different again from the location of the events reported. An example of displaced immediacy arising out of a subjectively construed ground is exemplified in (20): (20)

Yesterday they muzzled their guns for the first time, but the atmosphere was tense. One Serb sharpshooter, the barrel of his AK-47 automatic rifle poking out of a second-floor window, glared at UN troops as they scrambled into position on both sides of the bridge. Spotting him, a French soldier flopped down with his arms curled around a sniper rifle with long sights, and aimed at the building.

[...] Two bursts of gunfire rang out. Journalists and some UN officers ducked in alarm. French soldiers herded the journalists back to the boulevard and behind the wreckage of two rusted trucks dragged across the intersection to shield civilians. [...] On one side of the street leading to the bridge was a burnt-out restaurant. On the other stood the shell of a newspaper kiosk. (The Guardian, 11 Feb. 94, p. 6) Here, the speaker remains entirely "offstage" or subjective in that no overt reference is made to his or her own ground as source of the original experience. Nonetheless, this passage also produces an impression of immediacy, which is due to its narrative structure, to the amount of details given, which go beyond what one would expect of a bare news (second-floor window, arms curled around..., rusted trucks), and also underlined by the two instances of inversion. By means of these lexical inversions, the speaker assumes the viewpoint of the locations contained in them, which belong to the universe of discourse evoked by the text. But instead of appearing objectively "onstage" here, which could be done by first person pronouns (On one side of the street [...], we see [...]), the speaker chooses to remain "offstage" and only subjectively expresses his role as an experiencer. In addition to spatio-temporal displacement, such an attitude is considered ego displacement: the speaker focusses on "displaced experience imported from a different ego" (Chafe 1992: 247). And since "ego displacement is made possible by empathy" (1992: 247, his emphasis), it is in

524

Heidrun Dorgeloh

line with the analysis given above that these are contexts where one is most likely to encounter inversions in news reporting.

4. Conclusion Patterns of occurrence of inverted word order in naturally occurring discourse can indeed be accounted for by taking into consideration that it expresses a particular viewpoint or perspective which the speaker assumes. In that lexical inversion departs from a reference point which belongs to the universe of discourse established by a text, and not from the speaker's own ground, this kind of inversion has been shown to be a subjective presentative mechanism, to involve a subjective kind of construal of discourse entities. This analysis of inversion has enabled me to narrow down to a certain extent contexts or kinds of a discourse in which the construction can be said to be more likely to be used. For newspaper language in particular, it has been noted that, in contrast to hard news, where the speaker is normally at a recognizably different location from the events reported, typical contexts of usage for inversions are, for example, correspondants' reports or eyewitness reports. This is due to the fact that these are texts where the political events are presented as ~ in fact probably are - experience and the mode of displaced immediacy is chosen. Inversion is then one device whereby the immediate nature of an experience is reproduced in a subjective manner. Ultimately, as Chafe (1992.) has argued, the proximal and the displaced consciousness are not mutually exclusive, but rather "peripherally aware" (1992: 235) of each other, so that there is always potential alternation between the immediate and the displaced mode. This means that it is not that one of them is pre-existent in a certain genre or context of speech; proximal or displaced speech has to be established by linguistic devices at any given stage of a discourse. It is one of the discourse functions of inversion 4 to produce temporarily displaced immediacy, to signal that the speaker enters his or her universe of discourse by assuming a viewpoint of an entity from within it. Since the speaker in this way does not objectively come into the open, inversion may be considered a construction expressing subjectivity.

Viewpoint and subjectivity

525

Notes 1.

2. 3.

4.

I have taken over Kuno's (1987) assessment about the questionable acceptability of (6b). Many, though not all, native speakers I have consulted share my impression that the version given in (6a) is at least slightly more acceptable. For instance, 93% of Biraer's (1992) corpus of 1778 tokens of inversion occurred in a main clause. For instance, from a range of various newspapers and magazines, the sections titled "News in brief' (The Times, The Guardian), "The week" {The Guardian Weekly), "In brief" (The Daily Telegraph, The Economist), "The world this week" (The Economist) or "Periscope" (Newsweek), as well as pages titled "news" in general and containing a range of short articles (Daily Express) have not been found to show any usage of inversion whatsoever. For a more complete analysis of the discourse functions of Modern English inversion cf. Dorgeloh (1994).

References Banfield, Ann 1982 Birner, Betty J. 1992 1994

Unspeakable Sentences: Narration and Representation in the Language of Fiction. London: Routledge and Kegan Paul. The Discourse Function of Inversion in English. Ph.D. dissertation, Northwestern University. Information status and word order: An analysis of English inversion. Language 70.2: 233-259.

Bresnan, Joan B. 1994 Locative inversion and the architecture of Universal Grammar. Language 70.1: 72-137. Chafe, Wallace 1992 Immediacy and displacement in consciousness and language. In: Stein, Dieter (ed.). Cooperating with Written Texts. Berlin: Mouton de Gruyter, 231-255. Dorgeloh, Heidrun 1994 Word Order in Discourse: Form and Function of Inversion in Modern English. Ph.D. dissertation, University of Düsseldorf. Drubig, Hans Bernhard 1988 On the discourse function of subject verb inversion. In: Klegraf, Josef and Dietrich Nehls (eds). Essays on the English Language and Applied Linguistics. On the Occasion of Gerhard Nickel's 60th Birthday. Heidelberg: Julius Groos Verlag, 83-95. Ehrlich, Susan 1990 Point of View: A Linguistic Analysis of Literary Style. London/New York: Routledge. Emonds, Joseph 1969 A structure-preserving constraint on NP Movement transformations. Papers from the 5th Regional Meeting of the Chicago Linguistic Society, 60-65.

526

Heidrun Dorgeloh

Enkvist, Nils Erik 1980 Marked focus: Function and constraints. In: Greenbaum, Sidney et al. (eds). Studies in English Linguistics for Randolph Quirk. London: Longman, 134-52. Green, Georgia M. 1976 Main clause phenomena in subordinate clauses. Language 52.2: 382-97. 1982 Colloquial and literary uses of inversions. In: Tannen, Deborah (ed.). Spoken and Written Language: Exploring Oratity and Literacy. Norwood, NJ: Ablex, 119-154. Halliday, Michael A.K. and Ruqaiya Hasan 1976 Cohesion in English. London: Longman. Hooper, Joan B. and Sandra A. Thompson 1973 On the applicability of root transformations. Linguistic Inquiry IV/4: 465-97. Kuno, Susumu 1987 Functional Syntax: Anaphora, Discourse and Empathy. Chicago: Chicago University Press. Langacker, Ronald W. 1990 Subjectification. Cognitive Linguistics 1.1: 5-38. 1993 Reference-point constructions. Cognitive Linguistics 4.1: 1-38. Levine, Robert D. 1989 On focus inversion: Syntactic valence and the role of SUBCAT list. Linguistics 27.6: 1013-1055. Penhallurick, John 1984 Full-verb inversion in English. Australian Journal of Linguistics 4: 3356. Reinhart, Tanya 1983 Anaphora and Semantic Interpretation. London: Croom Helm. Rochemont, Michael S. 1986 Focus in Generative Grammar. Amsterdam: Benjamins. Vallduvi, Enric 1992 The Informational Component. New York: Garland.

Primary sources The Guardian, February 11, 1994. The Lancaster-Oslo/Bergen Corpus (LOB) Newsweek, January 20, 1992. Rouse, W.H.D. 1973. The Giant Crab. In: The Giant Crab and Other Tales from Old India. London: Minerva Press, 1-7. Time, January 20, 1992. The Times, December 15, 1992

How do we mentally localize different types of spatial concepts? Cornelia Zelinsky-Wibbelt

1. Introduction In this paper we will be concerned with the localization of spatial concepts and with their metaphorical and metonymic derivations. We will consider the localization of senses with respect to the fundamental question of how the meaning of utterances is established. Although each meaningful interpretation is unique within its current discourse model, we assume a relaxed distinction between the activity of processing the obtained information on the one hand and the storage of lexical knowledge and discourse knowledge on the other hand. The fluency of this distinction becomes obvious if we consider Lakoff and Johnson's by now well-known example (cf. Lakoff and Johnson 1980: 35). (1)

The ham sandwich is waiting for his check.

Speakers obviously have no problems to make this 'impossible' sentence possible if they have sufficient knowledge about the respective discourse domain, that is they must be capable of activating their restaurant scenario. In addition they must have available sufficient information about the respective discourse situation in order to apply their metonymy rule which relates the POSSESSOR to its POSSESSED. More exactly, they must have noticed that there is a customer in the restaurant who is eating a ham sandwich and who therefore will become the POSSESSOR of the POSSESSED if he has paid his bill. (1) was meant to illustrate that methodologically linguistic understanding works with both knowledge and information, however dependent on context the distinction between knowledge components and composed information may be. The crucial question for us is how lexical units and spatial predicates in particular become meaningful in an actual usage event; the answer to this question will result in a distinction between knowledge which is lex-

528

Cornelia Zelinsky-Wibbelt

icalized with a spatial predicate and information which is computed contextually. From our point of view this distinction can only be made with respect to different types of spatial predicates, by considering their syntactic and semantic context-dependence. DIMENSIONAL and SCALAR adjectives are context-dependent in a principled way, as the realization of these categories depends completely on the conventions embodied in the discourse situation and domain: (2) (3) (4) (5) (6)

a long nail a long channel This table is too big for our kitchen. This table is large enough for eight plates. This table is really great.

Whereas in (4) the normal THREE-DIMENSIONAL extension of table is not necessarily projected, it is projected to its TWO-DIMENSIONAL board in (5) and to its ABSTRACT idea in (6). And depending on the DIMENSIONALITY different adjectives are used in English to designate it accordingly. Whereas the DIMENSIONALITY projection is highly dependent on context and the SCALARITY of long may adopt any value on an infinite continuous scale, this is not so for a MOTION predicate like enter or leave. It is mostly the syntactically provided landmark and trajector that provides the CONCRETE vs. ABSTRACT reading: (7) (8)

John left the university at 4 pm yesterday. John left the university, in 1980.

In ( 7 ) the MOTION verb leave clearly predicates the CONCRETE PUNCreading, which is induced by the PUNCTUAL PP. In (8) the PP provides a PERIOD OF TIME and this induces the ABSTRACT reading on leave. Thus we have observed from the behaviour of spatial predicates that different domains such as SHAPE, ABSTRACTION, MOTION are lexically vague to different degrees. It is the intent of this paper to operationalize different types of lexical representation and the corresponding contextual functions by which they are processed. We proceed from the hypothesis that space is man's original and archetypical experience in terms of perception and imaging; accordingly, space is the major organizational principle in all cognitive domains in terms of the EXTERNAL BOUNDING and the INTERNAL CONFIGURATION of an entity. Although these two parameters of EXTERNAL BOUNDING and INTERNAL CONFIGURATION are elaborated in all cognitive domains, it is much harder to operationalize them for ABSTRACT domain predicates than for CONCRETE TUAL

Types of spatial concepts

529

domain predicates. In the ABSTRACT domain DIMENSIONALITY has become projected in various ways, depending on our socio-cultural and communicative-pragmatic conventions. Although this occurs on the model of spatial perception, the immense complexity of the social domain prevents a straightforward application of spatial principles. SPATIAL

2. Theoretical position Before studying the behaviour of spatial predicates in a systematic way, let us set out our general theoretical position. (a) From our systemic perspective1 we can accept the principle of compositionality only in its weak form: on the one hand it is indisputable that the meaning of a complex utterance is composed of the meanings of its components, but in addition it is always the localization of an utterance in the respective discourse situation which effects the uniqueness and the novelty of the meanings which the components contribute to the composite whole. Although it is a drastic example, the 'ham sandwich' metonymy illustrates this fact. (b) In this system's theoretical perspective the holistic view is already implied. Beyond the 'oversummativity' assumption we claim that the mental processing of linguistic meaning is integrated within our overall cognitive behaviour. (c) The structure of the lexicon is symbolic in a principled way; that is, every formal distinction is originally meaningful, although its relevance may fade with time. (d) To achieve linguistic flexibility as exemplified above by Lakoff and Johnson's ad-hoc metonymy it is by necessity that concepts are lexically vague in order to provide for the variations between the activities by which speakers fix the actual senses. Methodologically speaking, this means that we assume a distinction between the mental construals of concept, referent and sense (cf. Johnson-Laird 1983; Strohner 1990): Concepts are the lexically encoded types or their contextual extensions. Conceptual types provide abstract semantic representations which, being activated, enable speakers to understand the actually occurring tokens. Conceptual types provide their representations either directly or by analogy with some related conceptual type; the latter holds for ad-hoc extensions such as the ham sandwich use above. The referent is the perceived external information with which the instantiated type concept becomes associated. The achievement of this relation is in no way a trivial activity, as may be supposed from the

530

Cornelia Zelinsky-Wibbelt

type concept necessary for the ham sandwich use; successful reference depends on the relation to make sense. But where do speakers take the type concept from in cases like the use of ham sandwich? Sense only results from this association within the speaker's mental model of the discourse situation; that is, from relating the concept with all components of the speaker's knowledge that become relevant in the respective situation. If this test results in a coherent structure, the concept may be integrated within the speaker's mental model representation. By this localization, by its relations with the situationally relevant components of knowledge, a concept is really used and endowed with sense. It is within this integration that lexically vague concepts are adjusted to the specific requirements of each individual discourse situation by semantic extension, shift, etc. The HUMAN AGENT sense of ham sandwich may be inferred by our knowledge of the restaurant scenario and by applying the well conventionalized metonymic rule of POSSESSOR-POSSESSED, by which speakers are licensed to project the POSSESSED ham sandwich to the customer as the HUMAN POSSESSOR. (e) The economy of this flexible linguistic behaviour consists in the phenomenon that, by the plasticity of concepts and by the reduction of irrelevant information, speakers can subsume ever more referential tokens under the same conceptual type. Linguistic flexibility is creative in that novel semantic extensions are normally understood at once by analogy with conventionalized polysemy types. An example for which less context is needed is the PROCESS sense of university, which is not yet conventionalized for German Universität, whereas it is well conventionalized in English as well as for the semantically related noun Schule ('school') in German. By analogy with (9), (10) would be processed without difficulties: (9) (10)

Die Schule fängt am Montag wieder an. 'School starts on Monday.' ? Die Universität fängt am Montag wieder an. 'University starts on Monday.'

The phenomena of lexical economy, flexibility and creativity by analogy can only be explained by an adequate balance between lexical structure and lexical redundancy. Speakers structure their mental lexicon by categorizing the perceived information into prototypes and schemata. Both are economical abstractions

Types of spatial concepts

531

by which speakers control the enormous variation of information. In our model a prototype is a hypothesis about the central, dominant sense of a polysemy structure (cf. Kleiber 1990); a schema represents conceptual vagueness which relies on a principled context dependence. That is, we rely on Langacker's notion of a schema as being a superstructure of its elaborations within specific usage events (Langacker 1987: 68). 2 In our holistic representation we do not make any distinction between linguistic knowledge and world knowledge; instead we consider the typical behaviour of categories, such as nouns designating individuals and verbs predicating change, i.e. DYNAMIC situations. It is from this default behaviour that we derive the lexical representation of categories (cf. Croft 1991); this typical behaviour may of course be modified and if a modification occurs sufficiently often this will always result in a conceptual conventionalization. From this viewpoint the lexicon is 'open' so that we must consider the continuous process of lexicalization in our model. In this sense the typicality space provides neither necessary nor sufficient conditions for the use of word senses, but a central tendency in the sense that, based on the frequency of usage that speakers have experienced, certain senses are expected to be more probable and thus more typical than others. There is sufficient psycholinguistic evidence that so-called 'dominant' readings emerge as hypotheses from experience and are used in the absence of conflicting conditions. The latter may defeat the default assumptions in accordance with the actual experience right from the beginning and derive the situationally fitting sense. Thus even the most 'impossible' senses may be inferred if they are sufficiently supported by contextual information and background knowledge. The following notions of prototype semantics support our assumptions: 1.

Linguistic processing is interactive·, only by a recursive analysis working bottom-up and top-down as well as forwards and backwards will information and knowledge merge into one coherent unit;

2.

It is only by this intricate interaction based on lexical vagueness provided by the plasticity of concepts that the mental reorganization during information processing may be explained (Johnson-Laird 1983);

3.

Of course this reorganization cannot proceed in arbitrary ways, but it must be possible to integrate the reorganized concepts into the current discourse context. This depends on conventionalized cognitive constraints, which are very much culturally determined;

532

Cornelia

Zelinsky-Wibbelt

4.

The interactive processing of linguistic information works on the encyclopedia of linguistic knowledge within which the lexicon has the methodological function of a 'bridge-head'3; in this function the lexicon relates to all those encyclopedic and pragmatic-communicative components of our knowledge which are relevant for understanding even the most 'impossible' and most creative innovations.

5.

Finally, all these assumptions of linguistic processing require a broad notion of compositionality and context.

For our ultimate aim of localizing the sense of a lexical unit within the respective discourse model, the following central question arises from our prototype position: What is conceptually represented as the meaning of a lexical unit and which categories are contextually 'computed'?

3. Contextual processing We distinguish major contextual functions by which word senses are identified. In making these distinctions we proceed from Langacker's basic principle in which the constitution of reference resides. This principle consists in structuring the information conveyed by an utterance into focus, foreground and background along the following dimensions: 1. Along the trajector/landmark distinction, whereby different parts of speech are put into foreground and background (cf. Langacker 1987: 231ff.). (a) The trajector (TR) as foreground is typically provided by the subject; (b) The landmark (lm) as background is typically provided by the object. These functions may be assigned at all linguistic levels so that the same part of speech is associated with different TR/LM functions at different linguistic levels. This is to say that Langacker's terminology is by no means a simple translation of the traditional notions of subject and object. 2. Along the profile/base distinction, whereby the information structure of an utterance is organized (cf. Langacker 1987: 273): (a)

The profile as foreground is the communicatively relevant part of the information, towards which the focus of attention is directed;

Types of spatial concepts

(b)

533

The base is just that relevant background knowledge of our encyclopedia which is needed for the interpretation of the profile.

3.1. Contextual selection We assume the function of contextual selection for polysemy types which Dirven (1993) calls conjunctive metonymy, because they cover a socioculturally determined field and the senses are communicatively equally relevant. The following examples are meant to illustrate that, depending on which social group a speaker belongs to - whether he is a pupil, a teacher, a parent, a caretaker or a neutral person - different senses may assume different typicality values and accordingly may be preferred differently during the selection process. The equal validity of each sense originates from their establishing constituent parts of an overall discourse model (— PDM), which is mentally apparent, though differently weighted in each discourse situation in which one sense is applied accordingly. (11)

a.

You can enter the school through the back door. [CONCRETE (PDM)]

b.

CONTAINER

PART

OF

DISCOURSE

MODEL

The school is famous for its teachers. [ABSTRACT CONTENT = CONVENTIONALIZED INSTITUTION OF ANIMATE BEINGS AS PDM]

c.

In principle, school starts at 8.00. [CONVENTIONAL ACTIVITY OF COLLECTIVE INSTITUTION = PDM]

In (11a) the PHYSICAL MOTION of the HUMAN TR can only be conceived of as moving into the CONCRETE building. In (lib) the adjective famous functions as LM by providing an ABSTRACT EVALUATION in a QUALITY domain in which only the ABSTRACT INSTITUTION may be localized. In (11c) the LM designates a POINT IN TIME and from this point the ACTIVITY predicated by the verb starts; with this ACTIVITY sense of start only the ACTIVITY sense of school is compatible.

534

Cornelia Zelinsky-Wibbelt

Analogous examples are provided by the following sentences: (12) (13)

The Tokyo office called. Cleveland hasn 't reported.

In both examples the INSTITUTION sense of Tokyo and Cleveland can only be located in the ABSTRACT SOCIAL domain on which the VOLITIONAL predicates are dependent, whereas SPATIAL predicates such as be situated or extend would require the SPATIAL sense of Tokyo and Cleveland. These are examples of fairly productive metonymy patterns also elaborated by the following nouns: (14)

school, university, church, parliament

At present we can notice that there are many institutions with which the use of the ACTIVITY sense doesn't work, such as government. This is evidence for our central tenets: metonymy patterns are not elaborated with complete regularity partly because of the dynamic nature of the lexicon because some lexical units have not yet developed the whole pattern.

3.2. Contextual configuration Contextual configuration operates on a schematic disposition in the lexicon. As defined above, a schema is a generalization; more specifically, a lexical schema generalizes over several contextually induced specializations. Langacker gives the following definition of a schema (Langacker 1987: 68):

If an instance (I) is categorized by comparison with a schema (S), S as a generalization over different specializations must always be included in I: IcS DIMENSIONAL and SCALAR adjectives are a clear and typical instance of contextual configuration. They are prototypical because they are those relational predications which are conceptually dependent on their nominal TRs in a principled way - their semantic behaviour is 'relative', so that they are lexically underdetermined. The lexicon only provides the dimensional grid and scale on which the states predicated by the adjectives are projected and graded in accordance with conventional norms provided by the respective context. We want to distinguish our notion of a schema clearly from that of the Bierwisch and Lang paradigm (cf. Bierwisch 1989; Lang 1989): our

Types of spatial concepts

535

point is not the strictly sequential processing of meaning from the linguistic-semantic to the conceptual level; we only want to represent lexical vagueness based on the relative behaviour of lexical units. To summarize, our notion of a schema is defined as follows: (a)

An S represents the conceptual relativity of a lexical unit;

(b)

Ss represent the conceptual relativity of only those relational predications which are conceptually dependent on their trajectors in a principled way, as prototypically established by DIMENSIONAL and SCALAR adjectives;

(c)

Conceptual indeterminateness of these lexical units is elaborated by inheriting the communicatively relevant DIMENSIONALITY or SCALARITY from its trajector;

(d)

A proposition is the minimal context for the configuration of a schematic disposition (cf. Pustejovsky 1991).

3.2.1. The configuration of an adjective's meaning The configuration of an adjective's meaning proceeds in the following way: the noun's own DIMENSIONAL extension, which has become communicatively relevant, unifies with the DIMENSIONAL disposition of the adjective. Whether the speaker designates the actual SPATIAL EXTENSION or only a PART of the referent associated with the noun depends on the communicative relevance, as illustrated by the translationally relevant use of the German DIMENSIONAL adjective groß in the following examples: (15)

Der große Stuhl steht vor dem Klavier. —

>

The big chair is standing in front of the piano. [WHOLE AS FUNCTIONAL UNITY]

(16)

Diese Wohnung ist größer als die alte. This flat is larger than the old one. [PART AS FUNCTIONAL REFERENCE POINT]

536

Cornelia Zelinsky-Wibbelt

(17)

Er hat sich große Mühe bei der Fertigstellung gegeben. ... > He took great pains to finish it. [BOUNDING A PART OF THE MASS ON A SCALE]

In (15) the discourse context does not change the actual THREEGestalt denoted by Stuhl, and this unifies with the schematic semantic representation of groß. The sense which is thereby represented is normally translated into Standard British English big. In the discourse context of comparing flats it is normally the TWODIMENSIONAL extension of the floor which becomes the communicatively relevant part of flat. The question is whether the flat has enough square meters to accommodate one's furniture. A comparison in terms of cubic meters would be very unnatural. Therefore the entity designated by flat in (16) is projected from the actual THREE-DIMENSIONAL WHOLE to the TWODIMENSIONAL PART. This dimensional value is unified with the dimensional disposition of groß and the normal translation into British English is large. In (17) the ABSTRACT MASS noun Mühe induces the ABSTRACT sense of groß. The noun Mühe denotes an ABSTRACT MASS concept with relative intrinsicness and it functions as the TR in the object NP. And this TR may only be localized in an ABSTRACT LM provided by groß, which is then located in an ABSTRACT QUALITY domain. Contextual configuration applies in a similar fashion to the VERTICAL DIMENSION of the adjective's semantic representation, which is triggered in (18) and (19) by their nominal TRS; the same adjective long, however, inherits the HORIZONTAL DIMENSION of the noun queue. Moreover, the adjective long also inherits the SPATIAL EXTENSION in the respective dimension, which increases from (18) to (20). DIMENSIONAL

(18) (19) (20) (21) (22)

long grass long legs a long queue a fast typist a fast trip

The semantic disposition of the TEMPORAL adjective fast is HIGH which, again, is clearly relative if we compare the MOTION of ants and planes. In (21) the semantic representation of the adjective inherits the PROFESSIONAL VOLITION of the noun and predicates the general capability which is necessary for fulfilling this profession; in (22) fast inherits the TEMPORAL EXTENSION of trip so that HIGH TEMPO specializes into SHORT TEMPO,

TEMPORAL EXTENSION.

Types of spatial concepts

537

3.3. Contextual shift Contextual shift occurs if the context introduces a marked perspective on a dominant sense. We take this to be the most frequent and vital operation in the constitution of lexical meanings. There is psycholinguistic evidence that the dominant sense and the context are equally important and interact during lexical processing (Hirst 1988; Prather and Swinney 1988; Simpson 1981). The dominant sense applies by default if it is not in conflict with the available context. Speakers interpret the meaning of a lexical unit in an actual usage event by comparing it with its dominant sense. This comparison is achieved under the following condition: An instance (i) must approximate the prototype (P) or match it in the ideal case: Ρ > I

3.3.1. Against the arbitrary grinding of matter One of the most frequent shift types is obviously that between nominal and COUNT concepts. It is our intention to give evidence against the often made claim that this shift type applies arbitrarily to any noun, i.e. that any COUNT noun may be turned into a MASS noun and vice versa (cf. Bürge 1979; Laycock 1975; Allan 1980; Ware 1975; Pelletier 1979; ZelinskyWibbelt in print). The operation of shifting is a pragmatically and especially culturally constrained convention. That is, it must be possible to integrate the shifted meaning into its situational context, but at the same time this context must be culturally licensed. For instance we can shift from the COUNT SENSE OF ANIMATE beings to its EDIBLE MASS only if this feature is positively anchored within our nutritional standards. Thus in German the use of the noun Pferd ('horse') as a MASS noun is generally unlicenced, because it establishes a taboo to think of a horse in terms of its MEAT. MASS

3.3.2. Shifting nominal meanings by figure-ground reversal A very productive shift type is currently being discussed under the term of figure-ground reversal. Having originated in Gestalt psychology this term already indicates that there is a dominant figure and a background supporting the figure. The dominant figure is provided by the WHOLE Gestalt of the entity referred to. With the less typical usages only PARTS of the

538

Cornelia Zelinsky-Wibbelt

are mentally activated, so that we have a case of PART-WHOLE relationships with this shift type. Langacker (1993) subsumes the mental processing of these relations under the speaker's 'reference-point ability', which is indispensable for our necessarily flexible linguistic behaviour. With this ability speakers can functionally foreground the 'active zone' of a referent, without necessarily changing the referent. If applied sufficiently often, however, this foregrounding can always result in a conventional conceptualization of an autonomous function, which implies a referential change. In (23) the meaning of the ACTIVITIES predicated by the verbs suffices to project the contextually relevant PARTS from the WHOLE GESTALT: In (23a) the verb open induces the normal function of window, which can only be provided by the WHOLE GESTALT. In (23b) the verb see predicates a VISUAL PERCEPTION which together with the preposition through requires a VISIBLE LM; this in turn can only be provided by the transparent window pane. In (23c) the verb jump predicates a MOTION which requires a MOBILE AGENT. Together with its modifying SPATIAL prepositional phrase the verb requires the open state of the window so that in (23c) the frame as the material boundary becomes the relevant part. Although different parts of the whole entity are focussed in (23), it is not the referent but only its 'active zone' that changes, since in these discourse situations the PARTS of the 4 WHOLE could not fulfil their normal function if separated from each other. WHOLE

(23)

a. b. c.

Please open the window. I can see him through the window. The cat is jumping through the window.

The referent in (24) and (25) also remains constant, according to the criterion of functional consistency. The WHOLE is activated in (a), the CONTENT in (b), the CONTAINER in (c). More specifically, in (24b) the speaker performs a mental shift by focussing on the BOUNDED MASS of water contained in the concavity established by the boundaries of the lake. In (24c) the speaker performs a shift by focussing on the external boundaries establishing the CONTAINER of the water. Analogously, in (25b) the speaker performs a mental shift to the bulb as being the CONTENT of the lamp, whereas in (25c) he shifts from the whole gestalt of the lamp to its shade as establishing the CONTAINER of the bulb.

Types of spatial concepts

(24)

a. b. c.

The lake is not far from here. There are a lot of fishes in the lake. Yesterday John swam all around the lake.

(25)

a. b. c.

The lamp is hanging from the ceiling. This lamp is too bright. The design of this lamp doesn 't go with this room.

539

In contrast to the examples given in (23), (24) and (25) the referent changes in (26). The PARTS, the furniture activated in (26b) and the SPATIAL CONTAINER provided by the BOUNDING walls of the room in (26c), may potentially be located at different places. The furniture establishing the CONTENT of the kitchen may still be in a department-store when the walls of the kitchen establishing its CONTAINER are painted. The same holds true for (27), where the eiderdown referred to by bed in (27b) establishing the CONTENT of the bed and the frame referred to by bed in (27c) may also be separated from the WHOLE GEST ALT in the stores. (26)

a. b. c.

The kitchen is next to the living room. We need a new kitchen built in. But before that we have to paint the kitchen.

(27)

a. b. c.

I would prefer to sleep in the bed next to the window. My bed is not warm enough. The prices of the beds refer to the frame only.

So the criterion of spatial separation has given rise to the development of different referents to the WHOLE, to the CONTAINER and to the CONTENT. A final example of conceptual conventionalization is (28), where different PARTS at different places are referred to: the WHOLE GESTALT is profiled in (28a); the wine as the CONTENT which may have been decanted and thereby may have left its original place is profiled in (28b); finally the empty bottle as the CONTAINER which may already have been refunded is designated in (28c):

540

(28)

Cornelia Zelinsky-Wibbelt

a. b. c.

This shelf is full up with wine bottles. This bottle of wine is excellent. You have to pay a deposit on the bottle.

3.4. Conditions for shifting the meaning of MOTION verbs

Although verbs are relational predications and as such conceptually dependent on the entities they interconnect, they do not configurate their senses in such a principled way as we have assumed for DIMENSIONAL adjectives; that is, the dependence of verbs on TR and LM does not consist in relativity, but rather in typical and less typical uses (Lakoff 1987; Langacker 1987). We illustrate this with MOTION verbs, of which the original SPATIAL sense is often preserved as the dominant one. The following analysis of the verb leave has been inspired by Taylor (Taylor 1994), who in turn argues against Bierwisch and Schreuder (Bierwisch 1992). Our analysis of this MOTION verb is yet another piece of evidence for the claim that PHYSICAL space functions as an iconic model for TEMPORAL space. More precisely, it is the TEMPORAL STRUCTURE

which the speech act imposes on what is called 'discourse space' in cognitive linguistics (cf. Langacker 1991: 21 Iff.) that is construed on the model of PHYSICAL space. 5

In (29a) leave is used in the CONCRETE sense, predicating John's PHYSICAL MOTION from within the university building or campus as the CONTAINER to outside the CONCRETE CONTAINER (Taylor 1994; Zelinsky-

Wibbelt 1995). This sense is induced by the secondary LM of leave, the PP which predicates a POINT IN TIME, which on the time axis is located no more than some hours before the TIME OF DISCOURSE. In contrast to this leave predicates an ABSTRACT MOTION in (29b) to be located as a CONVENTION in a SOCIAL domain. John functions as an ABSTRACT MEMBER OF

A SOCIAL GROUP either consisting of students or of the university staff. Again this ABSTRACT function of the TR is induced by the secondary LM of leave·, the PP predicates a TIME SPAN functioning as the CONTAINER within which the ABSTRACT MOTION is to be mentally located. That is, the speaker as the point of origin being temporally located at the TIME OF DISCOURSE generates an ORIENTATION to a TIME SPAN located some years before the TIME OF DISCOURSE. This is signalled by the past tense of leave and indicated indirectly by the absolute distance between the TIME OF DISCOURSE and the TIME OF ORIENTATION predicated by the PP as being the time towards which the speech act is oriented. This PP also induces the ABSTRACT INSTITUTION sense of University. Depending on its ABSTRACT TR and LM

Types of spatial concepts

541

the sense of leave is also mentally shifted to its ABSTRACT SOCIAL MOTION. Finally, in (29c) the secondary LM provided by the PP is semantically vague enough for the sense of leave to remain ambiguous in respect of both readings. (29)

a. b. c.

John left the University at 4 pm yesterday. John left the university in 1980. John left the university a short time ago.

In the following we give the lexical representations of the semantic categories which distinguish the polysemous lexical units of (29), namely the verb leave and the nouns university and John:

leave 1

leave 2

TEMPORAL EXTENSION (BOUNDED(ACHIEVEMENT)) MOTION CONCRETE

TEMPORAL EXTENSION (BOUNDED(ACHIEVEMENT)) MOTION ABSTRACT(SOCIAL(CONVENTION))

university 1

university 2

CONCRETE COUNT INDIVIDUAL PLACE

ABSTRACT(SOCIAL(INSTITUTION)) COUNT

John 1

John 2

ANIMATE(HUMAN(MALE)) AND PHYSICAL COUNT INDIVIDUAL CONCRETE

ANIMATE(HUMAN(MALE)) AND VOLITIONAL COUNT

COLLECTION SOCIAL INSTITUTION

INDIVIDUAL ABSTRACT

(Social(Group Member)) Another condition for a shift from the typical to a less typical verbal sense is mentioned by Lakoff (1987), namely the negation operator. We illustrate its application with ACTIONS located in TEMPORAL SPACE. According to many monolingual dictionaries leave is part of a class of verbs which in their typical sense predicate an ACHIEVEMENT. In Vendler's

542

Cornelia Zelinsky-Wibbelt

terminology an ACHIEVEMENT implies that the ACTION has already reached its aim at the TIME OF ORIENTATION. This is not the case in (30b), where by way of negating the ACHIEVEMENT of leave the truth of the foregrounded conditions is denied (cf. Lakoff 1987: 133); instead the negative modal profiles the likelihood of the ACHIEVEMENT occurring in the future and thereby projects the speaker's attention along a path extending from the TIME OF DISCOURSE to the TIME OF ORIENTATION establishing the BOUNDING POINT of the mentally scanned TIME SPAN. In Vendler's terminology this mental scanning operation corresponds to a shift from the ACHIEVEMENT to the ACCOMPLISHMENT sense (cf. Vendler 1967).

(30)

a. b.

The plane is leaving. The plane will not leave before 2 ο 'clock.

According to monolingual dictionaries die also adheres to the class of verbs which typically predicate an ACHIEVEMENT, and it does so perfectly in depending on the PUNCTUAL LM in (31a). In processing (31b), however, speakers have to perform a mental shift to the ACCOMPLISHMENT which is required by the DURATIVE LM.

(31)

a. b.

He died immediately after the accident. He died over a long and painful period of time.

(32b) is also a perfectly correct answer to the question in (32a). Here the progressive form introduces the DURATIVE perspective on the ACCOMPLISHMENT of finish taking place at the TIME OF DISCOURSE. The auxiliary be retemporalizes the ACHIEVEMENT as being mentally scanned sequentially within the immediate scope of predication with the TEMPORAL scope including the expected result, the finished paper (cf. Langacker 1991: 424ff.). (32)

a. b.

What are you doing? I am finishing the paper.

3.5. Prepositions: between configurational and shifting predications Being the most polysemous lexical units, prepositions have been investigated in detail with regard to their lexicalization patterns (cf. Rauh 1991; Zelinsky-Wibbelt 1993a). Their intermediate position between con-

Types of spatial concepts

543

figuration and shifting is more extreme than with other relational predications as for instance with verbs. It may be recognized in a more straightforward way than with any other lexical category that prepositions, being originally spatial predications, embody a typical sense from which the other senses seem to derive (cf. Lakoff 1987). In example (33) the typical sense is clearly predicated by in in (33a), which relates to a LM elaborated by a CONTAINER ACTUALLY BOUNDED AT ALL FACES; this LM induces the archetypical INCLUSION relation on in. The other senses have been ordered with respect to their degree of BOUN-DEDNESS, their extension in THREE-DIMENSIONAL SPACE and with respect to their MOBILITY. The perfect INCLUSION is no longer present in (33b) to (33f.), yet it is discernible as occurring in a relaxed way. 6 In (33b) the LM is BOUNDED and THREE-DIMENSIONAL, but internally HOMOGENEOUS so that it achieves an INTEGRATION of an NONMOVABLE TR. I n ( 3 3 c ) t h e LM is FUZZILY BOUNDED a n d DISCONTINUOUS, thereby allowing SUPPORT to the TR. In (33d) a BOUNDED MOVABLE TWO-DIMENSIONAL HOMOGENEOUS SURFACE integrates a TWODIMENSIONAL NON-MOVABLE PRIVATIVE being. In (33e) a THREEDIMENSIONAL UN-BOUNDED MOBILE HOMOGENEOUS MASS integrates a THREE-DIMENSIONAL FUZZILY BOUNDED HOMO-GENEOUS MOBILE object. Finally, in (33f.) a MOBILE ONE-DIMENSIONAL UNBOUNDED ACTIVITY is contained within a MOBILE ONE-DIMENSIONAL UNBOUNDED ACTIVITY.

We assume a more abstract INCLUSION relation to have been abstracted from the typical sense to be underlying all senses of in. It is from this abstract prototype represented by the symbol 'C' that all senses configurate as CONTAINMENT, ACCOMMODATION and INTEGRATION as explained. For clarification we represent each configuration underneath the respective example in terms of a predicate-argument structure, where the value of the abstract INCLUSION relation 'C' is given on the left-hand side of the hierarchical relation represented by the sign ' = ', followed by the TR and LM which the INCLUSION interconnects.7

(33)

a.

biscuits in the box C = SPATIAL CONTAINMENT (TR = 3 - D , BOUNDED, REPLICATE, MOVABLE, L M = 3 - D , SHARPLY BOUNDED AT ALL FACES, INTERNAL CONFIGURATION = HOLLOW)

544

Cornelia Zelinsky-Wibbelt

b.

raisins in the cake C = SPATIAL INTEGRATION ( T R = 3 - D , BOUNDED, REPLICATE, L M = 3 - D , BOUNDED, HOMOGENEOUS, MOVABLE )

c.

a bird in the tree C = SPATIAL ACCOMODATION ( T R = 3 - D , BOUNDED, MOBILE, L M = 3 - D , FUZZILY BOUNDED, DISCONTINUOUS )

d.

a hole in the tissue C = SPATIAL INTEGRATION ( T R = 2 - D , BOUNDED, NONMOVABLE, PRIVATIVE, L M = 2 - D , BOUNDED, HOMOGENEOUS)

e.

smoke in the air C - SPATIAL INTEGRATION (TR = 3 - D , FUZZILY BOUNDED, HOMOGENEOUS, MOBILE, L M = 3 - D , UNBOUNDED, HOMOGENEOUS, MOBILE)

f.

living in peace C = SPATIO-TEMPORAL CONTAINMENT ( T R = 1 - D , ABSTRACT, MASS, MOBILE, LM = 1 - D , ABSTRACT, UNBOUNDED, MOBILE )

3.6. Shifting into ABSTRACT space A look at spatial metaphors provides evidence of what we have mentioned in the introduction: the semantic analysis of ABSTRACT domain predications is very difficult. Their EXTERNAL BOUNDING and their INTERNAL CONFIGURATION in terms of their GRANULARITY, DIMENSIONALITY and their PARTS is so deeply entwined in the complexity of our socio-cultural conventions that a rough and ready analysis as done by logicians would not work. The spatial senses of the predicates run, reach and forward in (34) to (36) are extended to their metaphorical senses by being projected onto ONEDIMENSIONAL TIME, where they preserve their EXTERNAL BOUNDARIES: run remains UNBOUNDED in (34b), reach remains BOUNDED within the immediate scope of predication in (35b), forward is oriented towards a boundary in (36b).

Types of spatial concepts

545

In (37b) it is both the EXTERNAL BOUNDING and the INTERNAL CONFIGURATION which are essential for the metaphorical shift into SEMIOTIC space. Both senses in (37) clearly invoke the concept of a BOUNDED COLLECTION comprising different INDIVIDUALS. In (38b) the conceptualizer abstracts away from the HORIZONTAL dimension along which a DISTANCE is predicated. The subject NP in (38b) induces the projection of far onto the MENTAL domain, where it predicates the NEGATIVE force of problem by evoking the image of a long and painful journey of the mind into distant unknown spaces until a solution to the problem is found. The preposition between in (39b) shifts from the structure in TWO-DIMENSIONAL or THREE-DIMENSIONAL SPACE to the Structure of ONE-DIMENSIONAL TIME which may also assume a second VERTICAL DIMENSION by analogy with the SPATIAL sense: in the same way as there may be different heights of houses, there may be ups and downs in jobs. In spite of the indefinite noun phrase in the metaphor in (40b), the metaphorical phrase a sea of troubles induces the image of a MASS which on the model of the infinity of the sea covering earth is conceptualized as burying the experiencer by flowing over his head in infinite quantities. Although from a logical point of view the adjective deep in (41b) predicates a ZERO-DIMENSIONAL STATE of existence as the RESULT of a MENTAL ACTIVITY, we do not conceptualize this STATE as ZERO-DIMENSION AL; by analogy with the CONCRETE sense deep predicates an intrusion well inside the boundaries of the CONTAINER. We conceptualize this intrusion into the MASS of thought by structuring it into a web of thought, which on the model of the THREE-DIMENSIONAL brain and within the idealized cognitive model of THINKING IS BUILDING becomes conceptualized as THREE-DIMENSIONAL. (34)

a. to run through the darkness [CONCRETE MOTION] —

>

b. to run through the examples [ABSTRACT MOTION]

(35)

a. to reach a destination [CONCRETE MOTION] — >

b. to reach an aim [ABSTRACT MOTION]

546

(36)

Cornelia Zelinsky-Wibbelt

a. Just lean/move forward a bit. [CONCRETE ORIENTATION] — >

b. I am looking forward to the meeting. [ABSTRACT ORIENTATION]

(37)

a. to send a Christmas package to Helen [CONCRETE COLLECTION] — >

b. to send the information package to Helen [ABSTRACT COLLECTION]

(38)

a.

Our house is not fax from the centre. [CONCRETE DISTANCE] — >

b.

This problem is far from being trivial. [ABSTRACT DISTANCE]

(39)

a.

There is an old house between the new ones. [CONCRETE CONTIGUITY, DISTANCE, INTERCONNECTION] — >

b. I am between jobs. [ABSTRACT CONTIGUITY, DISTANCE AND INTERCONNECTION]

(40)

a.

The sea covers three-quarters of the earth. [CONCRETE INFINITE QUANTITY] —

>

b. A sea of troubles came over him. [ABSTRACT INFINITE QUANTITY]

(41)

a. He is deep in the forest. [GREAT DISTANCE FROM EXTERNAL BOUNDARIES IN CONCRETE SPACE] — >

Types of spatial concepts

b.

547

He is deep in thought. [GREAT DISTANCE FROM EXTERNAL BOUNDARIES IN MENTAL SPACE]

4. Conclusion Proceeding from assumptions about flexibility, economy and creativity in the mental processing o f polysemous lexical units w e have presented a model in which three different contextual functions operate on accordingly different lexical representations: contextual selection operates on equally valid lexical representations, contextual configuration operates on an abstract schema, and contextual shift operates on a prototypical sense. Each contextual function achieves a coherent sense by relating a TR to a LM. By exhausting the principle of analogy, which organizes the whole lexicon, these contextual functions control regularity on the one hand, while enabling flexibility in terms o f creativity on the other hand. W e have concentrated on h o w the EXTERNAL BOUNDING and the INTERNAL CONFIGURATION of an entity are established. Only having touched on the shift into ABSTRACT space, w e have shown its complex grounding in social conventions.

Notes 1. 2.

3. 4. 5.

6.

In the sense of systems theory. This notion is opposed to Bartlett's concept of 'schema' (Bartlett 1932) and his followers (Rumelhart 1975; Rumelhart 1980), which comes closer to Rosch's original definition of a prototype as the concept with the most prominent features of all instances associated with a word (Rosch and Lloyd 1978). While both notions imply the dynamic specialization of schemata within actual discourse, the abstraction of schemata is a perspectivization for the proponents of schema theory, while it is a generalization by the ability of schematizing along a hierarchy of concepts within Langacker's cognitive grammar (Langacker 1987: 378f.). This is a metaphor created by Strohner (1990). With this assumption we are not in agreement with Lakoff (Lakoff 1987: 418), who assumes window to be polysemous. See however other linguists who have criticized this view (Taylor 1994). cf. e.g. Greenberg (1985) as one example of the endeavour to put Reichenbach's schema into a cognitive-linguistic framework. Following Greenberg, Langacker and other cognitive linguists who have worked on tense and aspect, we refrain from using the Reichenbach terminology (Reichenbach 1979). This phenomenon has also been discussed elsewhere in the literature; cf. for example Herskovits (1986); Hottenroth (1993); Cuyckens (1993).

548

7.

Cornelia Zelinsky-Wibbelt

cf. Zelinsky-Wibbelt (1993c) for a detailed analysis and generation of prepositional meaning. Being meant as a model for machine translation the analysis applies to the source language and the generation to the target language during the process of translation.

References Allan, Keith 1980 Nouns and countability. Language 56 (3): 541-567. Bartlett, Frederic C. 1932 Remembering: A Study in Experimental Psychology. Cambridge: Cambridge University Press. Bierwisch, Manfred 1989 The semantics of gradation. In: Bierwisch, Manfred and Ewald Lang (eds). Dimensional Adjectives. Grammatical Structure and Conceptual Interpretation. Berlin: Springer, 71-261. Bierwisch, Manfred and R. Schreuder. 1992 From concepts to lexical items. Cognition 42: 23-60. Bürge, Tyler 1979 A theory of aggregates, NOUS 11: 97-117. Croft, William 1991 Syntactic Categories and Grammatical Relations: The Cognitive Organization of Information. Chicago: Chicago University Press. Cuyckens, Hubert 1993 The Dutch spatial preposition 'in'. A cognitive-semantic analysis. In: Zelinsky-Wibbelt, Cornelia (ed.), 27-71. Dirven, Ren6 1993 Metonymy and metaphor: Different mental strategies of conceptualisation. Leuvense Bijdragen 82: 1-28. Greenberg, Joseph 1985 Some iconic relationships about place, time and discourse deixis. In: Haiman, John (ed.). Iconicity in Syntax. Amsterdam: Benjamins, 271287. Herskovits, Annette H. 1986 Language and Spatial Cognition. Cambridge: Cambridge University Press. Hirst, Graham 1988 Semantic interpretation and ambiguity. Artificial Intelligence 34: 131177. Hottenroth, Priska-Monika 1993 Prepositions and object concepts: A contribution to cognitive semantics. In: Zelinsky-Wibbelt, Cornelia (ed.), 179-219. Johnson-Laird, Philip N. 1983 Mental Models: Towards a Cognitive Science of Language, Inference and Consciousness. Cambridge: Cambridge University Press. Kleiber, George 1990 La simantique du prototype. Categories et sens lexical. Paris: Presses Universitäres de France. Lakoff, George 1987 Women, Fire and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press.

Types of spatial concepts

549

George Lakoff and Mark Johnson 1980 Metaphors We Live By. Chicago: University of Chicago Press. Lang, Ewald 1989 The semantics of dimensional designation of spatial objects. In: Bierwisch, Manfred and Ewald Lang (eds). Dimensional Adjectives. Grammatical Structure and Conceptual Interpretation. Berlin: Springer, 263-417. Langacker, Ronald 1987 Foundations of Cognitive Grammar. Theoretical Prerequisites. Vol. 1. Stanford: Stanford University Press. 1991 Foundations of Cognitive Grammar. Practical Applications. Vol. 2. Stanford: Stanford University Press. 1993 Reference-point constructions. Cognitive Linguistics 4(1): 1-38. Laycock, Henry 1975 Theories of matter. Synthese 31: 411-442. Prather, P. A. and David A. Swinney. 1988 Lexical processing and ambiguity resolution: An autonomous process in an interactive box. In: Small, Steven L., Garrison W. Cottrell and Michael K. Tannenhaus (eds). Lexical Ambiguity Resolution: Perspectives from Psycho linguistics, Neuropsychology, and Artificial Intelligence. San Mateo, Ca.: Kaufmann Publishers, 289-310. Pelletier, F. J. 1979 Mass Terms: Some Philosophical Problems. Dordrecht: Reidel. Pustejovsky, James 1991 The generative lexicon. Computational Linguistics 17(4): 409-441. Rauh Gisa (ed.) 1991 Approaches to Prepositions. Tübingen: Narr. Reichenbach, Hans 1947 Elements of Symbolic Logic. New York: Macmillan. Rosch, Eleanor and B.B. Lloyd. 1978 Cognition and Categorization. Hillsdale, N.J.: Erlbaum. Rumelhart, David E. 1975 Notes on a schema for stories. In: Bobrow, D.G. and A. M. Collins (eds). Representation and Understanding: Studies in Cognitive Science. New York: Academic Press, 211-236. 1980 Schemata: The building blocks of cognition. In: Spiro, Β., B.C. Bruce and W.F. Brewer (eds). Theoretical Issues in Reading Comprehension. Hillsdale, N.J.: Erlbaum, 33-58. Simpson, Gregory B. 1981 Meaning dominance and semantic context in the process of lexical ambiguity. Journal of Verbal Learning and Verbal Behavior 20: 120-136. Strohner, Hans 1990 Textverstehen. Kognitive und kommunikative Grundlagen der Sprachverarbeitung. Opladen: Westdeutscher Verlag. Taylor, John R. 1995 Approaches to word meaning: The network model (Langacker) and the two-level model (Bierwisch) in comparison. In: Vanparys, Johan and Rend Dirven (eds). Current Approaches to the Lexicon. Frankfurt: Lang, 3-26. Vendler, Zeno 1967 Linguistics and Philosophy. Ithaca, N.Y.: Cornell University Press. Ware, Robert 1975 Some bits and pieces. Synthese 31: 379-393.

550

Cornelia Zelinsky-Wibbelt

Wierzbicka, Anna 1985 Lexicography and Conceptual Analysis. Ann Arbor: Karoma Publishers. Zelinsky-Wibbelt, Cornelia (ed.) 1993a The Semantics of Prepositions: From Mental Processing to Natural Language Processing. Berlin/New York: Mouton de Gruyter. Zelinsky-Wibbelt, Cornelia 1993b Introduction. In: Zelinsky-Wibbelt, Cornelia (ed.), 1-24. 1993c Interpreting and translating prepositions: A cognitively based formalization. In: Zelinsky-Wibbelt, Cornelia (ed.), 351-390. 1995 Insights from mental categorization for a computational model of lexicalization. In: Dirven, Reni and Johan Vanparys (eds). Current Approaches to the Lexicon. Frankfurt: Lang, 27-61. in print Reference as a universal cognitive process: A contrastive study of article use. In: Geiger, Richard (ed.). Reference in Multidisciplinary Perspective: Philosophical Object, Cognitive Subject, Intersubjective Process. Hildesheim: Olms Verlag. Now available as: EUROTRA-D WORKING PAPER 2 1 .

Part D Space as an organizing principle of thought

Section 7 Discourse as space

Space in dramatic discourse Vimala Herman

1. Introduction Drama as a genre makes complex and constant demands on us to conceptualize space. In fact, the art of the theatre has often been regarded as the art of space. The bare boards of the stage are the site of transformations since the physical spaces of stage and auditorium have to be transmuted into various fictional spaces and places and the conversions are wholly mental, cognitive and imaginative. Even the most realistic setting can only be mimetic of real world space and not the thing itself. Drama is a hybrid form having a dual existence as text in literature and as performance in theatre, with the language of the text simultaneously addressing a context of performance. Both visual and verbal modes of space depiction are consequently involved. The verbal is encountered in two forms - as dialogue in the primary text, and as stage directions in the secondary text, especially in the spatial arrangements that such directions propose: the former is directed at the dramatic figures while the latter is aimed primarily at the director. Both discursive forms verbalize space although their functions vary. Offstage, invisible spaces, for example, are constructed via the characters' dialogue in order to be conceptualized and visualized by the audience or reader. The dramatist's characterization of on-stage space in the secondary text, as already conceptualized space, has to be re-conceptualized by the director or stage designers if it is to be presented visually on stage. The linguistic medium is involved in both these instances. The paper seeks to investigate the relevance of some of the recent work on language and space to drama, since the data offered by the field, though neglected, are rich. It also seeks to explore the particular ways by which the non-verbal organization of physical space via architecture, stage design, and the relationship of stage and audience they signify, and the like, have also contributed to how space may be conceptualized within the constraints of dramatic use.

554

Vimala Herman

2. Domain constraints on space in drama: General Space in drama has its own history, aesthetics and traditions of use. The architectural design of theatres, stage design, scenography, and different dramaturgies have all exerted their pressures to constrain how dramatists use space in plays in different periods. The scale of Greek theatres was immense, about four hundred feet across and estimated to seat about 17,000 in the theatre of Dionysus and over 50,000 in the theatre of Epheseus (Styan 1979: 111). The physical scale and setting itself bespoke its communal importance. The size of the theatre, more like a modern sports auditorium, called for stylized rather than realistic modes of performance - seen in the use of masks, symbolic costumes, declamation rather than conversation, and of choral commentary (Pfister 1988: 20). As a ritual space, the circular, spatial organization of the huge amphitheatre, in spectacular settings, enclosing the nested organization of the stage, the central altar, chorus, actors, orchestra and contest areas, has been seen, iconically, as "a version of the social system which alternated agony and solidarity; it was open to debate and interrogation, but closed as to who was or was not a member, a citizen" (Schechner 1988: 160). English mystery plays of medieval drama with their makeshift platform stages and surrounded by the street audience on all sides provided a more intimate relationship between stage and audience, but even if less stylized than the Greeks, offered little scope for the verbal rendering of space. It was only with the advent of Elizabethan drama with its richness of dialogue, the secular themes, and the flexibility of the stage that the verbal rendering of space was to come into its own. Plays were not performed in the street but in permanent theatres with rounded or square auditoriums of about 80 feet across, open to the skies and surrounded by tiers of galleries. But plays were also performed in great houses, guildhalls, taverns, etc. Four basic essentials - an enclosing auditorium, a projecting platform which was as deep as it was wide, two upstage entrances and a balcony - have been proposed as characteristic of this theatre (Styan 1987: 12). The performance was during the day, and the audience surrounded three sides of the stage, thus both the scale or size of the bounded enclosures, the architecture of stage and theatre, and the degrees of proximity of the players to the audience contributed to the kind of theatre and drama that this organization of space made possible. The English Restoration theatres were smaller, and the audience was separated from the stage by a proscenium arch. The auditorium could hold about five hundred as opposed to the Globe theatre which could cater for an audience of two thousand. The proscenium curtain came later, in the late eighteenth century. The 'picture-stage' of the nineteenth and twentieth

Space in dramatic discourse

555

century, with the darkened auditorium and the illuminated stage which created the illusion of a room with the missing fourth wall also enforced the power of the autonomy of the dramatic world (Pfister 1988: 21-22). The separation of the stage from the audience was quite radical, the organization of space with the barrier of the proscenium arch again, iconic of the convention of separation of the stage and audience worlds. Although practioners like Brecht and the post-modernists have attempted to challenge this convention, the convention has been naturalized enough to function as prototype of all theatres for ordinary theatre-goers. The uses of space in plays have also changed with changing theatrical space. The classical unities of time, place and action had given way, especially in Shakespeare, to looser, more varied and multiple locations in a play. Domestic and interior settings were particularly suitable for the picture-stage while the modern stage often uses indeterminate locations or spills over, via action, into audience space as well. At any rate, conventions like asides, direct addresses to the audience, whether confined to prologues and epilogues in earlier drama or mobilized as a consistent part of the dramatic action as in a contemporary play like Peter Handke's Offending the Audience, are also reflexive uses of the whole deictic, spatial scope of the speech context of the overall stage event, which the conventions of the picture-stage tended to obscure. Movement, action, address, etc. can thus extend the scope of fictional space beyond the given, physical boundaries of stage and audience as self-contained divisions. Spaces can be re-cognitivized as a consequence within the terms of a given context of activity. The physical restrictions of the stage itself necessitate further sub-divisions of space into on-stage and off-stage space, that which is visible and those which are not. The first kind as mimetic space is communicated visually, while diegetic space, the second type, is mediated and communicated via dialogue (Issacharoff 1989: 55-6). The two kinds may interweave when a dramatic figure verbalizes the visual via references to objects, decor, etc. Of course, other media like sound may also be used to signify spaces; in fact, radio drama relies on the acoustic signification of space in the absence of the visual channel (1989: 61-62). The dramatic space that is utilized in a play-script is a combination of the two other kinds and varies from play to play. Plays may use either closed or open structures of space (Pfister 1988: 252-6). In closed structures, the space-scope is extremely limited and can be concentrated on one locale and is used in naturalistic as much as absurdist plays and can deliver effects that range from intimacy and immediacy to claustrophobia, with characters imprisoned in a place in deterministic worlds that provide no options but stasis. Open structures, on the other hand, provide more expansive worlds since both time and space are extended in the changes of locale. Thus,

556

Vimala Herman

space may also be conceived of in static or dynamic fashion with greater movement among places able to trace larger tapestries of themes or concerns. The option chosen in a play is significant since the extension of choice in space presentation, or its refusal, are both consequential. Space on stage has also been semioticized along the dimensions of up/down, left/right, front/back, which are operative in linguistic spatial descriptions. The divisions, however, do not remain neutral, but accrue various semantic values and reveal various relations like symmetry, opposition, contrast, etc. Whether characters belong to good or evil camps can be indexed via their entrances from the right or left, in some conventions. In Greek theatre, entrance from stage right could denote that the character has come from far afield, whereas stage left was reserved for those from a nearby city (Aston and Savona 1991: 77-8). Political opposition can be made visible between different parties, the French and English camps in some of Shakespeare's plays, for instance, because of the positioning of their entrances. Where groups of characters stand on either side of the stage, the physical distance between them can mean antagonism or hostility, which could well be negated when mingling occurs. The up/down dimension can signify social hierarchies, the descending movement from up to down can enact a fall in fortunes, as in the deposition scene in Shakespeare's Richard II. Feminists have noted that centre stage was reserved for men in public contexts with women's marginalized, social status to be seen in their physical placings on stage, in these instances, and generally, in the kind of spaces to which they were relegated, usually domestic interiors. The nearer the actor to the audience, on the back/front reach, the more intimate the relation with it. Even on the Greek stage, the placing of the chorus between the actors and the audience brought them into closer relation to it. The chorus, at any rate, articulated communal concerns of the audience which their physical placing underscores (Styan 1979: 112). Such contrasts have been patterned and extended to structure dramatic space globally across whole plays. Where spaces and places are multiple, the patterns adopted in inter-relating them can be consequential. In Shakespeare's A Midsummer Night's Dream, the basic contrast between the worlds of the Court and the Forest, as normative and imaginative worlds, is further structured spatially, so that the world of the forest is in medial position with the world of the Court beginning and ending the play. Sanity and the status quo are thus safeguarded in the end. The topography of the romances and comedies is generally located in far-off places - Illyria, Belmont, Verona, Athens, etc. Satires were located in London (Pfister 1988: 259-60). Spaces like gardens, too, have special significance, either as 'liminal' spaces blurring boundaries between country and city, inside and outside, domestic and wild, or used for special significance or contrast given their traditional

Space in dramatic discourse

557

associations with the Garden of Eden and the like (Hilton 1987: 73). Exits and entrances, as departures and arrivals, to or from other places, off-stage, can enlarge the spatial scope of a play. Space can be conceptualized, schema or frame-wise, via activities within them, by contrast with other places, whether adjudged to be 'near' or 'far' geographically, and so on. The use of space on-stage is also conditioned by various proxemic relations as spatial relations among characters and between stage and audience. Hilton, following E.T. Hall, divides such relations into four categories: intimate, personal, social, and public (Hilton 1987: 22-24). The schema is ego-centric, as in speech, with concentric circles radiating away into different proxemic zones although boundaries among the zones are not wholly determinate, and cross-cultural factors could well influence the designation of the different zones themselves. The intimate zone covers a space of about eighteen inches from the body, within direct touch and skin contact, a contingency that is open to taboo. Usually, only one person, in any instance, is located within it and the presence of others could signal either marked aggression or trust. Spatial proximity of this kind can help one gauge the degree of intimacy between people. But access to this zone is non-reciprocal since "The lord may enter the serf's space, or box his ears. The reverse is not true" (Hilton 1987). The space in front of the body is secure, generating feelings of control; the space behind is one of vulnerability and anxiety. Thus, in common parlance, we may "protect each other's backs" or get "stabbed in the back"; at any rate, we are enjoined to "cover our backs" as a precaution. The personal zone covers a range from eighteen inches to about four feet. Trust is evident but not with the same intensity as in the previous case, and groups and others may inhabit it as a matter of course, as in conversation. Access to this still exclusive space is via introduction or prior familiarity. The social zone extends from four to twelve feet. At its outer perimeter many others can be included. The proportions of rooms in terraced houses in England are between eleven to fourteen feet. The sense of exclusivity is weaker, but still present. And, finally, the public zone covers an expanse of twelve to twenty-five feet or more. But none of these can be adhered to rigidly since knowledge and expectations of behaviour in context, or via convention, in different circumstances, can often tolerate breaches or result in re-definitions of them. In crowded rush-hour journeys on trains, or buses, people are squashed together without respect to proxemic rights, and in parties, people may invade each other's spaces with conventional impunity. In performance, a further paradox is evident. The audience in public space is actually sitting side-by-side in close proximity, while the performers occupy the other two zones. Both share public space. But the

558

Vimala Herman

crossing of zones by characters on-stage are cues for the audience that significant dramatic matters are imminent.

3. Language and space Issues of space in drama intersect with language in two major ways, as noted above: in the dialogue, and in the "stage directions" or didascalia to do with setting. Dialogue in drama is meant to be spoken on stage by actors in face-to-face mode, usually. The play-text inevitably addresses a context of performance which entails that actors do not merely read or recite the lines, but transform the written lines into forms of interaction among themselves. The lines, therefore, are transmuted into speech contexts and realized as speech events, as instances of "situated" discourse (Herman 1991). The co-ordinates of the speech event in Hymes' mnenomic of SPEAKING (Hymes 1972) include the spatio-temporal context. One of the obvious functions for space in drama is as setting, although as seen earlier, space is used in other ways as well. The spatio-temporal setting, however, must not be regarded merely as passive backdrop or a decorative addition. Language and spatial context are integrated and the influence is two-way (Gumperz 1982). The setting influences participant speech behaviour, as in doctor-patient transactions in a surgery. But relevant speech behaviour actually realizes the context itself via various "contextualizing cues". Linguistic means for representing space are multiple as is well known, and vary cross-culturally in different languages. Additionally, space is never described in the abstract, but usually in terms which are other than itself, as spatial relations among objects, or as vectors and projections from reference points like the body of the speaker in the deictic centre of speech. Linguistic resources include deictic items like 'here' and 'there', demonstrative adverbs like 'this/that' calculated as distal or proxemic to the speaker, although other reference points could function as deictic centre. Various prepositions and prepositional phrases, like 'at', 'in', 'on', 'from', 'to', and 'in front o f , 'at the back o f , 'to the left/right o f , ' n e x t to','at the side o f , etc., and verbs like 'come/go', 'bring/take', as well as compass points, or their correlates in various cultures are included (Levinson 1992). Of the prepositions, some like 'at' locate objects at a place without respect to size or dimensionality. Those like 'on' locate objects in relation to surfaces in two-dimensional mode, whereas words like 'within' locate objects in enclosures in three-dimensional space (Lyons 1977: 698). Positional locations are generally static. Dynamic relations among objects are evident in prepositions like 'into', 'from', 'out of', etc. with objects understood as travelling from a source of some kind to a goal. Shape, too, is relevant in

Space in dramatic discourse

559

phrases like 'in the corner' as opposed to 'at the corner' since notions of boundaries, edges, angularity, roundedness, etc. can also be drawn upon (Lyons 1977: 697). Locative expressions and spatial relations, however, are a vaster area than the above summary has encapsulated (Leech 1969; Bennett 1975; Herskovits 1986). Underlying conceptual parameters that can be inferred from linguistic spatial terms are multiple although it is generally agreed that there is no one-to-one correspondence between linguistic and conceptual configurations of space. The canonical face-to-face encounter functions as a kind of cognitive primal scene with the speaker as deictic centre and point of reference for calculations of the fundamental up/down, left/right, front/ back dimensions. Generalizing over possible patterns, Levinson (1992) offers the Figure/Ground nexus as foundational with objects being fixed in space relative to the Ground. But absolute systems are also in evidence. The Figure, as the object to be located, is fixed in connection with some stable reference point, as Ground, generally the speaker, as origo, but there are also hearer and third-person centred sub-systems to be found. The cross-cultural and cross-linguistic record reveals that non-human bodies may also be used as Ground or primary reference points. Given the relative stability of the up/down co-ordinate - on Earth at any rate - owing to gravity, the left/right of the body of the speaker is usually mapped on to objects with no intrinsic left or right, like cupboards or bookcases. The front/back is similarly calculated with respect to the speaker, the front of an object being that aspect of it which confronts the speaker, in analogy with the canonical face-to-face encounter (Clark 1973). Other fronts are conventionalized by use - like the front of buildings, for instance. These modes of calculation are culture-relevant. Where a long line as of people in single file, one behind the other, is substituted for the face-to-face scenario, the 'front' of a tree in English would actually be the 'back' if the tree were ahead of one, a contingency evident in Hausa (Hill 1982). Other cognitive models of space include maps and journeys: many descriptions of apartments are organized from the perspective of someone moving through the rooms (Linde and Labov 1975). The whole area of prepositions and locative expressions often involves geometric conceptualizations of points on lines, surfaces, planes, enclosures, etc. in locating objects (Lyons 1977; Leech 1969; Herskovits 1986).

4. Space in dramatic discourse Whatever the nature of the cognitive or representational structures that can be invoked, as mediators between 'reality' and language, to explain the way

560

Vimala Herman

in which we perceive, order, understand and talk about space, or objects, spatial relations, movement, etc. in space, abstract conceptualizations lose their 'ideal' status and are constrained in various ways when cast into actual contexts of use. In fact, some researchers have questioned the value of the staticity of idealized types. Pick and Rieser (1982) propose more attention to the process of mapping and the role of spatial inferences and spatial updating. Hart and Berzok (1982), in their research on how children represent routes in landscapes, suggest that they have choices in fact as to how to extract regularities from the environment in the form of physical, social or event maps, some information being more relevant to a "Where am I?" type question, others to a "What next" mode of motivation (1982: 164). Representations are not the same for all purposes, and different types of representation of the 'same' geographical space may well be called for according to purpose. Herskovits, too, argues for the supplementation of what she calls "geometric meaning" (1986: 18) by pragmatic factors like relevance, salience, tolerance, and typicality in order to account for the difference that pragmatic motivations and context-dependency makes on how locative expressions are understood. For instance, the notion of 'salience' is operative in the metonymic shifts that occur when parts of objects are foregrounded and made to deputize for wholes. In an utterance like, "There is a rabbit under the bush", the rabbit is understood not as under the roots of the bush but under that part of the bush that is above ground, as perceptually salient. The notion of 'tolerance' is evident in instances where a book is described as "on the table" even if a set of files interposes between the top of the table and the book referred to. 'Typicality' motivates the relations like those between bottle and cap as the "cap on the bottle" rather than the logical equivalent, "The bottle is under the cap", owing to conventional modes of use. 'Relevance' has to do with communicative goals and choice of means - in Herskovits prepositional choice - among other things, and includes the kind of tolerance evident in situations when "Move the cat off the mat" is still relevant even if only the paws and not the whole cat is on the mat (Herskovits 1986: 73-85). Spatial representations via language in drama are highly use-conditioned, but not necessarily at the expense of the abstract, although ideal types are accommodated to contextual requirements. Descriptive or ideational and communicative or inter-personal constraints are both involved. Moreover, the symmetrical, ninety degree intersections of the up/down, left/right dimensions need to be complicated to cater for object location at points arrayed along the potential 180 degree arc of the visual field. Moreover, where multiple objects and spatial relations configure the perceived scene, linguistic description has to project the principles that underlie their

Space in dramatic discourse

561

ordering so that a world of singulars does not result. Relative distances among objects and degrees of 'depth' in space may also have to be catered for. Localizing techniques in drama are word-painting or spoken space strategies. The dramatic figures' utterances can be used to transform space, to paint the scene verbally in place of physical decor or technological means. In Shakespeare's Hamlet, in the opening scene, the events of the night of the appearance of the Ghost end with the verbal transformation of the spatio-temporal setting to one of morning in Horatio's speech. But look, the morn, in russet mantle clad, Walks o'er the dew of yon high eastward hill (1.1.166-7).

The speaker is the origo, so that the location of the space of the morning light is dependent on the direction of the gesture or gaze. The initial 'look', which generates a Gricean implicature of "look now", as well as the later 'yon' draw the 'morn' into the deictic spatial context of speech. The speaker's body is the primary point of reference, the ground, the light located with respect to it, but located only directionally, on the horizontal. Once fixed directionally, a secondary strategy is used to anchor the light more precisely, compass-wise as in the east, but via another ground and figure, the dew on the hill which also utilizes the vertical top-down. The two independent projections, the horizontal and the vertical, intersect in projected far space to locate the required object. This abstract geometrical schema is pragmatically adapted by yet another locative since the 'russet' light itself is further placed as "o'er the dew" and metaphorized and dynamized by the image of 'walks'. The choice of the 'dew' as 'ground' in this instance is pertinent, since it cannot be conceptualized as a solid mass, since dew is more gaseous, so that the boundaries are indeterminate, as is its shape (Herskovits 1986: 60). The light is made kinetic in being reflected off the dew and given an unbounded, 'russet' spread. Thus, not just location, but appearance is stressed. The whole can be conceptualized as a triangle, in situ, in this context. The angle of vision is oblique, as is the base angle, since the speaker here is placed on the battlements of the castle and not on the ground, as in the traditional face-to-face scenario, and pragmatic adjustments have to be made accordingly. The two primary, radiating lines are slanted and meet the vertical of the mountain at base and top. The reversed triangle is necessary to enable the imagination to 'metricise' height, which is calculated on the vertical, and helped by locating the observer in medial height in relation to the light which is placed on a 'high' hill. The line of vision moves 'up' but also all the way 'down'. Moreover, pragmatic, real world knowledge

562

Vimala Herman

would also curve the line of light against the outline of the hill, whatever the scope of its indeterminate spread, which is how it would be visualized. The capacities of secondary locative strategies are particularly interesting for their compositional functions in configuring space. In the following, in Thorton Wilder's play Our Town, the Stage Manager's utterances have to paint a whole town verbally for the audience, and locate it in the theatre. Well, I'd better show you how our town lies. Up here - [That is: parallel with the back wall] is Main Street. Way back there is the railway station; tracks go that way. Polish Town's across the tracks, and some Canuck families. [Towards the left] Over there is the Congregational Church; across the street's the Presbyterian. Methodist and Unitarian are over there. Baptist is down in the holla' by the river. Catholic Church is over beyond the tracks. Here's the Town Hall and the Post Office combined; jail's in the basement. Bryan once made a speech from these very steps here. Along here's a row of stores. Hitching posts and horse blocks in front of them. First automobile's going to come along in about five years - belonged to Banker Cartwright, our richest citizen... lives in the big white house up on the hill. Here's the grocery store and here's Mr. Morgan's drug store. Most everybody in town manages to look into those two stores once a day. Public School's over yonder. High School's still farther over. ... (Wilder 1938, repr. 1982: 22)

One suspects that any number of maps can be drawn to configure the layout of the town since the body of the speaker and the index finger are the only guides for tracing the initial vectors. Much depends on where 'here' and 'there' or 'over there', 'way back there' are located, and proximity and area covered varies among 'up here', 'along here', 'these very steps here'. And the speaker can move on stage as well, thus trailing the egocentric co-ordinates with him. The whole 360 degree rotation could be used, if necessary, although it is equally possible that only the 180 degree, shared, face-to-face area could be mobilized. Actors speaking with their backs to the audience is not as proscribed in contemporary drama as it had been in older ones, so that the 'face-to-face' need not remain static as generally presupposed in linguistic deliberations.

Space in dramatic discourse

563

The index finger fixes primary points of reference - Main Street, the railway station, the tracks in relation to the speaker. But there are also independent points of reference - the hill and the river as given, topographical features of the landscape. The erstwhile 'figures' of street, tracks, and railway station become, in turn, 'grounds' for other objects to be located. Thus, Polish town's across the tracks, and deeper still, the Catholic church is "over beyond" the tracks, possibly 'in' Polish town as an area at the further edge of it, or even 'outside' it. Similarly, public schools "over yonder", and High School is "still farther over". The Presbyterian Church is located across the street. Whereas some relations among areas and objects are conceptualized as in long lines, as depth relations, taking in middle and deep distances, others are placed indeterminately in the direction of the finger or gaze as blocks in figure/ground relation - the Baptist church near the river, the big house on the hill, the jail in the basement of the Town Hall/post office, the hitching posts and horse blocks in front of the row of stores. Nor is this an "imaginary tour" since the linear order of the various clauses or mention of objects do not in fact reflect the order of objects as encountered on the path of a journey. There are too many switches between near and far in the perception of objects for such a model to be credible. Sequential route or map strategies are available but not consistently. But survey strategies (Hart and Berzok 1982: 153-4), especially of the loose topographical type with locations specified grossly or vaguely as 'beyond', 'near', etc., are most in evidence. The description of space is actually static (Wenz, this volume), the order of perception moving from near to far, from the centre, where the speaker is located, to the periphery of its visual range and back again. And this movement is repeated so that lines across different vectors radiate from the centre. But the scope of both 'here' and 'there' is highly variable, and calibrated differently across the proxemicdistal range. Lateralized space is often drawn in in the way objects are arranged in the 'here' or 'there'. The shape or area occupied by designated objects as inferred pragmatically also affects how we read the space configurated in the passage, the hitching posts, as opposed to Polish town, for instance. Other ways of depicting space in drama, rarely investigated in linguistic research, include the use of figure perspectives on some place, especially contradictory ones so that different aspects of the 'same' space, via different experiences of it, may be revealed. The same landscape may be exciting to some but sinister to others. Communicational goals may require the use of one strategy to describe space rather than another, and proximity itself between speaker and object could be used differently for different dramatic purposes. Space may need to be miniaturized or symbolized for

564

Vimala Herman

plot or thematic purposes, and verbal resources may be called upon to paint dark or dismal places where the usual co-ordinates are not even visible and other means have got to be found. Qualities of spaces, whether cold, scented or moonlit, may also be represented verbally (Herman: forthcoming). The secondary text, too, is a useful area of study for the depiction of space, especially in those plays where the dramatist turns stage director and gives explicit directions for scenery or setting in the text. And domain constraints, here, require the adaptation of some of the everyday language coordinates. Locations like stage Right, or stage Left, for instance, are calculated via the left/right of the body, but not the dramatist's in the study, while composing the text, but that of the actor on stage during performance. Thus, a deictic shift is called for on the part of the dramatist and imaginary identification with another absent context into which appropriate co-ordinates are inserted. Moreover, the up/down of the stage is actually on the horizontal and calculated as near or far from the audience. There is the vertical, too, but different 'meanings' accrue to these positions depending on which of them is used. So, too, for front/back which does not use the bodies of the performing principals as the only determinant. Thus, in the following, most of the vertical references are to be read on the horizontal. The Porter's one-room flat in a large Midland town ... The scene is a fairly large attic room, at the top of a large Victorian house. The ceiling slopes down quite sharply from L. to R. Down R. are two small, low windows. In front of these is a dark oak dressing table. Most of the furniture is simple, and rather old. Up R. is a double bed, running the length of most of the back wall, the rest of which is taken up with a shelf of books. Down R. below the bed is a heavy chest of drawers covered with books, neckties and odds and ends, including a large, tattered toy teddy bear and soft, woolly squirrel. Up L. is a door. Below this is a small wardrobe. Most of the wall L. is taken up with a high oblong window. This looks out on to the landing but light comes through it from a skylight beyond. Below the wardrobe is a gas stove, and, beside this, a wooden food cupboard, on which is a small portable radio. Down C. is a sturdy dining table with three chairs, and, below this, L. and R. are two deep, shabby arm leather armchairs ... (Osborne Look Back in Anger 1957: 9).

The structure of this description would correspond to a gaze tour around the walls of the room (Ullmer-Ehrich 1982). The features of this type include the predominant use of secondary deixis like right, left, front, back, although in the present case the presence of such terms in the text is also motivated by the conventions used for stage descriptions. The configuration

Space in dramatic discourse

565

of space is conditioned by the reference frame adopted with its co-ordinates of the reference place, as a designated locality, like the entrance to a room, which functions as origo for the location of other places or objects; and the reference orientation, which is the visual orientation assumed with respect to the phenomena. In the above data, the reference place is not the entrance, as in living space descriptions, but the audience/viewer, placed on the other side of the invisible fourth wall of the stage. The reference orientation, however, begins at the top of the slope of the ceiling and then traces a path along the walls from right to left, but comes to rest at stage centre, thus bowing to the theatrical imperative. The objects, the pieces of furniture referred to, follow the normal order of such descriptions: each object takes its turn to come into view. Linearization principles are of two kinds according to Ullmer-Ehrich: the first, or Round About principle describes the furniture as arranged along the adjacent walls of the room, in spatial connectedness, while the Parallel Line principle allows the objects to be arranged in 'lines' with 'jumps' between them, and without spatial connectivity attempted among the different lines of description. Also, sub-areas within the wider spatial arrangement can be linearized, either in Sequencing or Group mode; the first mentions one object and then links the next with respect to the first, the second uses one object as a pivot around which others are spatially clustered. Round About descriptions mobilize the sequencing mode, while the Parallel Line option uses the Group mode. In the passage above, the Round About principle is generally in evidence, and although there are sub-areas like the living area, the bedroom area, and the kitchen area, only the last of these could be said to use the grouping principle in the contiguity of food cupboard and radio, and to some extent the stove as well. All other areas are linearized sequentially. Generally speaking, the fit between the principles as used in the stage description and the real life data given by Ullmer-Ehrich is remarkably close, Osborne's preference for stage decor obviously being for as realistic a setting as possible. Space may also be configured with minimal reference to spatial relations among objects but focus instead in container-contained fashion on the objects themselves as typical of a place. And this is the strategy that Bernard Shaw seems to use in the following extract. Lists of objects are linearized, and lists are an interesting phenomena in depicting space, with many uses. Here, only a kind of existential principle is at work, so that if the order of the objects listed were to be changed it would not materially affect the spatial configuration of the park. Only the outer boundary is specified, and the extent of the area itself. Internal spatial relations among the various objects are not given. A secure, primary reference point, too, is hard to come by.

566

Vimala Herman

One may posit a viewer, as origo, but the view itself can equally well be conceptualized as from a height, an ariel view, and the viewer could equally well be located at a fixed point or in motion above the landscape as in a helicopter or plane, even if functionally the whole has to be presented as a painted backdrop to the action on stage. The order in which objects appear cannot be conceptualized as that which strikes the eye of the viewer either, for many of these would more plausibly be grouped together - the lake and the trees, for instance, or the "plenty of greensward", the different flowerbeds and the patterns they trace, and the fence itself or parts of it, in relation to the area it encloses or demarcates. ... Near the outer end of the Hackney Road is a park of 217 acres, fenced in, not by railings, but by a wooden paling, and containing plenty of greensward, trees, a lake for bathers, flower beds which are triumphs of the admired cockney art of carpet gardening, and a sandpit, originally imported from the seaside for the delight of children, but speedily deserted on its becoming a natural vermin preserve for all the petty fauna of Kingsland, Hackney and Hoxton. A bandstand, an unfurnished forum for religious, antireligious, and political orators, cricket pitches, a gymnasium, and an oldfashioned stone kiosk are among its attractions ... (Candida 1898, repr. 1984: 93).

And finally, strategies used by non-realistic, non-illusionist drama could also be mentioned for the unconventional modes tried. In Thornton Wilder's The Skin of Our Teeth, the spatial context of Atlantic City in which the Convention is held is delineated via a montage of celluloid images. The viewer's sense of space and place is evoked via images and not by direct contact with space itself. The images themselves are drawn from various aspects of life as lived and include man-made objects, not natural objects or buildings alone, and people, in order to delineate the place and context of the Convention. There is no call for spatial relations among them. Only a chronological succession of single or grouped images one at a time is cued by the description. The objects are denoted in plural form, thus different tokens of each type are called for. Nor is one mode of conceptualization sufficient. A Source-Goal mode can be read into the details of the timetables for trains from Pennsylvania to Atlantic City, but the depiction of the latter mobilizes the Container-contained mode. Objects like churches, especially if projected as one kind of church following another, rely on the fact of numbers and kinds to delineate the place, as do hotels, Bingo parlours, rug merchants and the like. The linear order presented is open to various forms of exploitation as to which order prevails in any instance. With respect to linearity, only the principle of succession remains, and even

Space in dramatic discourse

567

this can be fractured by juxtaposing unrelated objects; iconicity is irrelevant. The whole is left for the viewer to construct at will. Towards the end of the intermission, though with the houselights still up, lantern slide projections begin to appear on the curtain. Timetables for trains leaving Pennsylvania Station for Atlantic City. Advertisements of Atlantic City hotels, drugstores, churches, rug merchants, fortune tellers, Bingo parlours (The Skin of Our Teeth 1942, repr. 1982: 127).

5. Conclusion This study has attempted to investigate the relevance of contemporary deliberations on language and space for the study of drama, and vice versa. Plays offer particularly valuable data for exploring the scope of the conceptual scenarios that are proposed in the fields of linguistics and cognition for how we organize the abstract notion of space verbally. The complex pressures exerted on verbal and visual delineations of space in drama not only offer support for some of the frameworks examined, but often resist or undermine them as well. The latter eventuality functions as a valuable corrective to any hegemonic tendencies lurking in any of the explanations available. The data provided from both dialogue and the stage directions reveal no great poetic or literary skill since they have a more practical role in their relations to issues of space. A great deal of variety is available in how space is conceptualized in language. Especially in discourse which eschews one-liners for longer stretches of text, mix and match conceptualizations, ambiguities, primary 'figures' transformed to secondary 'grounds' in turn, pragmatic dependencies, and productive flouts of preferred, normative frameworks are evident, with no threat to understanding or sense, which reveals variety in conceptualizations especially when functional considerations are taken into account. Speakers may well have a repertoire of possibilities which are used selectively in specific instances to represent the same space, spatial relations, or place. The inclusion of different domains like drama could only further attention to such issues given that it is one which exploits normativities, but is also an area of creativity and experimentation.

568

Vimala Herman

References Aston, Elaine and George Savona 1991 Theatre as Sign-System: A Semiotics of Text and Performance. London: Routledge. Bennett, D.C. 1975 Spatial and Temporal Uses of English Prepositions. London: Longman. Clark, Herbert 1973 Space, time, semantics and the child. In: Moore, Τ. E. (ed.). Cognitive Development and the Development of Language. New York: Academic Press, 28-64. Gumperz, John 1982 Discourse Strategies. Cambridge: Cambridge University Press. Gumperz, John J. and Dell Hymes (eds) 1972 Directions in Sociolingusitics: The Ethnography of Communication. New York: Holt, Rinehart and Winston. Handke, Peter 1969 Offending the Audience. London: Methuen. repr. 1971 Hart, Roger and Maxine Berzok 1982 Children's strategies for mapping the geographic-scale environment. In: Potegal, Michael (ed.). Spatial Abilities: Development and Physiological Foundations. New York: Academic Press, 147-69. Herman, Vimala 1991 Dramatic discourse and the systematics of turn-taking. Semiotica, 83 1/2: 97-121. 1996 Dramatic Discourse. London: Routledge. Herskovits, Annette 1986 Language and Spatial Cognition: An Interdisciplinary Study of Prepositions in English. Cambridge: Cambridge University Press. Hill, Clifford 1982 Up/down, front/back, left/right. A contrastive study of Hausa and English. In: Weissenborn, Jürgen and Wolfgang Klein (eds). Here and There: Cross-linguistic Studies on Deixis and Demonstration. Amsterdam/Philadelphia: John Benjamins, 13-42. Hilton, Julian 1987 Performance. London: Macmillan. Hymes, Dell 1972 Models of the interaction of language and social life. In: Gumperz, John J. and Dell Hymes (eds). Directions in Sociolinguistics: New York: Holt, Rinehart and Winston, 35-71. Issacharoff, Michael 1989 Discourse as Performance. Stanford: Stanford University Press. Jarvella, R. J. and W. Klein (eds) 1982 Speech, Place and Action. London: John Wiley and Sons. Leech, Geoffrey N. 1969 Towards a Semantic Description of English. London: Longman. Levinson, Stephen C. 1992 Primer for the field investigation of spatial description and conception. Pragmatics 2: 1, 5-47. Linde, Charlotte and William Labov 1975 Spatial networks as a site for the study of language and thought. Language 51: 4, 924-939.

Space in dramatic discourse

569

Lyons, John 1977 Semantics. Vols. 1,2. Cambridge: Cambridge University Press. Osborne, John 1957 Look Back in Anger. London: Faber and Faber. Pfister, Manfred 1988 The Theory and Analysis of Drama. Cambridge: Cambridge University Press. Pick, Herbert L. and John J. Rieser 1982 Children's cognitive mapping. In: Potegal, Michael (ed.), 107-128. Potegal, Michael (ed.) 1982 Spatial Abilities: Development and Physiological Foundations. New York: Academic. Schechner, Richard 1988 Performance Theory. London: Routledge. Shakespeare, William 1994 Complete Works of William Shakespeare. London: Harper Collins. Shaw, George Bernard 1898 Plays Pleasant and Unpleasant. Hannondsworth, Middlesex: Penguin, repr. 1946 Styan, J. L. 1979 Drama, Stage and Audience. Cambridge: Cambridge University Press. 1987 Shakespeare's Stagecraft. Cambridge: Cambridge University Press. Ullmer-Ehrich, Veronika 1982 The structure of living space descriptions. In: Jarvella, R. J. and W. Klein (eds), 219-249. Weissenborn, Jürgen and Wolfgang Klein (eds) 1982 Here and There: Cross-linguistic Studies on Deixis and Demonstration. Amsterdam/Philadelphia: Benjamins. Wilder, Thornton 1962/ rep. 1982 Our Town. The Skin of Our Teeth. The Matchmaker. Hannondsworth, Middlesex: Penguin.

How space structures discourse Lorenza Mondada

0. Introduction Relations between space and language have been classically investigated by focusing on the linguistic descriptors of space and by exploring to what extent the grammatical possibilities of the linguistic system have a structuring effect on space. However, space is not a referent as many others: while it is structured by the language which expresses it, it plays also a structuring role, serving to order and to inform non-spatial entities. This latter aspect is illustrated by well-known spatial metaphors and, more specifically, by models which use space to visualize and organize objects of discourse and objects of knowledge. In this article, we shall deal first with textual deixis as one of the manifestations of this structuring power of space. We shall further widen the domain usually covered by this category, with examples selected from a corpus which is particular but which allows a better observability of these phenomena. This will lead us to identify spatial macrostructures organizing the discourse. We shall focus on the iconic discursive effects of such an organization, suggest other domains where they can be observed, and insist on the generality of the processes that make them possible.

1. Textual deixis A certain number of spatial terms can be used to plan a discourse; they belong to the category of textual deixis, discourse deixis or logodeixis (Fillmore 1975: 70): Discourse deixis has to do with the choice of lexical or grammatical elements which indicate or otherwise refer to some portion or aspects of the ongoing discourse.

572

Lorenza Mondada

1.1. Markers of textual deixis The markers occurring in this category are the following: demonstratives (e.g. 'this/that paragraph/chapter/paper'); locatives indicating a position (located as proximal or distant from the origo: e.g. 'here', 'there', 'elsewhere'; in addition, they indicate a relation of inclusion: e.g. 'in this section/chapter/part of the book'); locatives indicating a direction (this direction is either calculated from the origo: e.g. 'so far', 'to this point'; or situated on a vertical axis: e.g. 'below/above', 'infra/supra'); temporal prepositions (e.g. 'now', 'earlier/later', 'preceding/following', 'next/last') and verbal tenses (e.g. perfect, future, cf. Conte 1988: 14); specific terms for textual deixis, e.g. 'former/latter'. Textual deixis is based on markers taken from other types of deixis, mainly spatial and temporal deixis. More seldom it has its specific markers, which are missing in certain languages. This fact raises two questions: what is the primitive dimension of textual deixis (1.2.) and whether textual deixis is a type of deixis in its own or pertains to other types from which it borrows its markers (1.3.).

1.2. The primitive dimension of textual deixis Analyzing the possible markers of textual deixis, Rauh (1978) describes the opposition between spatial and temporal markers. On one side, she remarks that they can be based on the textual graphic dimension, and imply, in this case, a conception of the text as "... eine Kette lokalisierbarer Elemente, auf der, wie auf einer Geraden, einzelne Punkte lokalisiert werden können, von denen der Punkt, an dem sich der Äusserer gerade befindet, als die Hier-origo mit dem lokalen Adverb 'here' markiert werden kann" (1978: 97-8). On the other hand, markers can be based on the textual spoken dimension, which is then conceived as "... eine Realisation, die sich in einem Zeitraum erstreckt, ein Zeitkontinuum bildet, in dem jeder Zeitpunkt der Äusserung ein Punkt jetzt ist und zur Jetzt-Origo erhoben werden kann" (1978: 98). Thus the use of the resources of temporal deixis by textual

How space structures discourse

513

deixis refers to spoken language as the use of the resources of spatial deixis refers to writing processes. This issue concerns the categorization of the linguistic materializations, as every mode of categorization implies specific processes. Research by Ong (1982) or Goody (1977) on writing has explored the sources and implications of the opposition between spoken and written language, and their radical consequences for the way speakers conceive objects of discourse and of knowledge. Rauh opposes both aspects without concluding on the predominance of one or the other. Fillmore (1975), on the contrary - although he describes textual deixis in spatial terms, i.e. "Discourse deixis has to do with the choice of lexical or grammatical elements which indicate or otherwise refer to some portion or aspect of the ongoing discourse" (1975: 70) - takes a radical position, identifying the temporal dimension as primitive (1975: 70): Most commonly, the terms of discourse deixis are taken from systems of deictic and non-deictic time semantics, for the very good reason that any point in a discourse can be thought of as a point in time.

He doesn't consider, however, the fact that temporal relations are often spatially metaphorized and modelled, and that a point in time is a place 1 . The point is to consider which conception of the text is proposed by language itself and by the speakers who use the linguistic system. This conception is not disembodied, but intervenes actively in the organization of the discourse, and eventually in the rationality expressed and constructed by it. We can then draw a parallel - to be questioned and validated empirically on contextually occurring data - between the 'domestication of the mind' through writing and the fact that spatiality is a textual organizing principle, appearing for instance in, but not only, textual deixis.

1.3. The specificity of textual deixis Depending on the authors, the spatial and temporal forms used by textual deixis do or do not represent a type of deixis. Thus, for Rauh (1983: 48): ... the fourth type of use of deixis expressions is characteristically represented by a centre of orientation that differs from type 1, 2 and 3 ['demonstratio ad oculos', 'deixis am phantasma', 'konstruktive Phantasie'] in its lack of correspondence to the real or reconstructed external situation of the encoder but which corresponds rather to his momentary situation within

574

Lorenza Mondada

the course of a text, considered either temporal or local and with respect to which either temporal or local domains of the textual context are determined.

Rauh, like Fillmore, considers textual deixis as a type on its own, i.e. as spatial, temporal or personal deixis. On the contrary, Conte (1988: 15) argues that textual deixis borrows forms belonging to other types of deixis, namely spatial and temporal ones: consequently it doesn't have specific markers. Moreover, Conte refuses to deal with textual deixis on the same level as the other types, because other markers work on the proper linguistic level, while textual deictic markers work on the metatextual level. Thus we shall consider not only the formal markers but also the processes involved. Moreover, other questions remain open: are different modes mutually exclusive or can they be combined? Where to locate the specificity of each type, at the level of forms or of the underlying processes?

1.4. Linearization process If we consider the operations underlying textual markers, we can say not only that they deal with the text as a space (cf. Rauh 1978), but that they actualize activities in this space, namely paths. As Ehlich (1983) or Kurzon (1985) point out, textual deixis orients the reader's attention to a spatial portion of the text, to its transitions, or to its global organization. This orientation is not static, based only on locatives, but it is dynamic, for it involves instructions addressed to the hearer, concerning his motion in this space. We can thus link locatives indicating a direction in the text with corresponding motion verbs, as in the following examples, extracted from spoken (ex. 5) or written academic discourses in English, French and Italian: (1)

You may now proceed to their exemplification in the following pages (Rau83-97).

(2)

I now return to the problem of insertion sequences introduced in Part I (Sche72-106).

(3)

Let us come back to the question, mentioned in the introductory section, of non-'canonical' reference points (Fu54).

How space structures discourse

575

(4)

Nous reviendrons plus loin (cf. 3.1.1.) sur la question des restrictions de la selection lexicale (Lam48). 'We will come back later (cf. 3.1.1.) to the question of lexical selection restrictions.'

(5)

ςava / ... autre chose ... ace sujet/bien ... alors avangons ... bon je disais ... (Mey-11). 'okay ! ... something else ... on this topic/good ... so let's go further ... well I was saying

(6)

Se andiamo un momento alia storia dell'organizzazione scolastica nei paesi europei, ecco, vediamo che e stata una storia abbastanza faticosa (Ber240). 'If we go for a while to the history of school organisation in European countries, then we see that it was a pretty difficult story.'

(7)

Nella prima parte mi domando se e possibile trovare un quadro teorico in cui trattare in modo unitario tutti i tipi di ellissi [..]. Prima di giunpere a questa affermazione, chiarisco che cosa intendo per ellissi (Mar255). 'In the first part I wonder if it is possible to find a theoretical frame for the unitary analysis of all kinds of ellipsis [..]. Before we arrive at this statement, I have to explain what I mean by ellipsis.'

(8)

Lasciando al lettore I'esercizio su altri connettivi esempliflcati [...], mi fermerei piuttosto sul 'perche' (Ber249). 'Letting the reader exercise himself on the other connectors mentioned [...], I would rather stop at the connector "why".'

We have here a list of motion verbs (to proceed (1), to return (2), to come back (3, 4), to advance (5), to go (6), to arrive (7), to stop (8)) used metalinguistically: in fact their arguments are textual elements. They transpose directions into movements, considering them as instructions. This is the case with directions signifying a distance from the origo on a linear path. The vertical dimension, on the contrary, is not used (we didn't find any instance). Text is thus conceived as a linearized space, with places where to stop, to go or to come back and where the speaker runs through. These places are designed as metalinguistic entities (the question, the problem, the point) but can also be topics (e.g.: "We come back to the problem of the sign's arbitrariness/to the sign's arbitrariness").

576

Lorenza Mondada

1.5. Textual deixis and reading process The structuring role of path spatiality is a phenomenon which Vuillaume (1990) relates to the organization of what he calls "secondary fiction". This meta-textual level explicitly refers to the narrator's and the reader's activities, where the latter is guided by the former or where both are joined in a common activity. The secondary fiction is separated and doesn't interfere with the primary fiction, but gives the narrator and the reader a specific point of observation on the character's actions. Consequently, motion verbs organize the text's progression in relationship to spatially metaphorized time as with space itself. Vuillaume's hypothesis consists in articulating the two dimensions of fiction: the first is the narrative of events, prior to its uttering, the second is the presentification of the described universe (1990: 69-71). These two dimensions correspond to two levels of enunciative landmarks: the first is related to the presumed date of writing and the second is related to the event of reading (1990: 78). In the frame of the secondary fiction, the narrator's and reader's moves can be represented along a path which, according to Vuillaume, is not metaphorical but real in the context of this fiction, and which must be distinguished from other metaphorical uses of the path, namely temporal uses (as, for example, "... l'epoque oü nous sommes arrives") (1990: 38). The examples provided by Vuillaume 2 illustrate the autonomy of the secondary fiction, and make the mechanism of text organization explicit. Vuillaume refers to Fauconnier (1985), who distinguishes mental spaces at different levels, of text, of reading, of path and of narrated universe (1984: 178): ... linguistiquement ces niveaux de r6cit sont des espaces mentaux en correspondance; en particulier, ä l'espace temporel 'univers racontd' correspond un espace spatial, le 'parcours', de telle sorte qu'ä un moment ou une p6riode du premier correspond un point ou un intervalle du second. L'axe temporel de Γ univers racont6 devient ainsi un axe spatial sur lequel s'effectue le parcours; mais ä I'irr6versibilit6 des temps sur le premier axe ne correspond aucune irr6versibilit6 des chemins sur le second: le dfiplacement dans l'espace parcouru est remarquablement libre. ... linguistically these narrative levels are corresponding mental spaces; in particular, to the temporal space 'narrated universe' corresponds a spatial space, the 'journey', in such a way that to a moment or to a period in the first corresponds a point or an interval in the second. The temporal axis of the narrated universe thus becomes a spatial axis on which a path is drawn;

How space structures discourse

577

but no irreversibility of paths on the second axis corresponds to the irreversibility of time on the first: the movements through this space are remarkably free, (translation L. M.)

How to describe the possibilities and the constraints of this freedom still remains to be known. We shall try to attempt such a description drawn from a particular corpus. 2. Discursive effects of isomorphic relations between textual space and territorial space.

2.1. Textual and territorial spaces One of the arguments invoked by Rauh in defence of the specificity of textual deixis is "... its lack of correspondence to the real or reconstructed external situation of the encoder" (1983: 48): textual deixis refers to a kind of space which has nothing to do with physical space. If that holds for the majority of texts, we can however imagine discourses which, instead of maintaining the separation between these levels, confuse them wilfully. The convergence between different levels becomes thus a textual strategy. We shall illustrate this possibility with a first series of examples, which all deal with the motion verb 'to quit', 'to leave', 'quitter' in French. All are extracted from a homogeneous corpus of 18th-19th French travelogues. (9)

Je vous ai parle du porphyre vert; il ne faut pas le quitter sans vous dire qu 'il y a trois especes de porphyre: le rouge, le vert et le noir (Bro2: 94) [followed by a digression on porphyry], Ί spoke to you about green porphyry; one cannot leave it without saying that there are three kinds of phorphyry: the red, the green and the black one.'

(10) Je ne quitterai pas Livourne sans mentionner un avantage par lequel eile se recommande aux gastronomes (Audi: 26). Ί will not leave Livorno without mentionning its gastronomic interest. '

578

Lorenzo Mondada

(11) Avant de quitter la Toscane, je dois dire que j'y ai vu le paysan partout vetu de drap, bien logi et nulle part des sabots (Duc37-38). 'Before we leave Tuscany, I must say that there the peasant was well dressed and well housed everywhere and nowhere with clogs.' (12) En attendant que nous y entrions [ä Lanciano], parlons un peu des heroiques Samnites, car bientöt nous quitterons le sol ou leurs ossemens sont enfouis, et nous aurons I'occasion de parier d'autres peuples. (Aud2a: 207) 'Before we arrive [in Lanciano], let us talk a little about the heroic Sannits, because we will soon leave the soil where their bones are buried, and we will have the opportunity to talk about other nations.' (13) Nous n'avons pas voulu quitter Naples sans aller visiter le tombeau venere de sainte Philomene (Ass219). [begins a chapter entitled: 'Pelerinage au tombeau de sainte Philomene'] 'We didn't want to leave Napoli without visiting the venerated grave of Saint Philomena.' (14) Avant de quitter Verone, consacrons quelques instans a visiter ses palais; le plus beau est celui que fit construire Louis Canossa. (Aud4a: 52) [followed by the palace's description] 'Before we leave Verona, let us spare some time to visit its palaces; the most beautiful is the one constructed by Louis Canossa.' (15) Nous quittames ce lieu sans avoir pu retrouver le tombeau d'Adelaide, femme de Roger, et reine des Deux-Siciles. (Cusl: 368) [end of the chapter] 'We leave this place without having found the tomb of Adelaide, wife of Roger, and queen of the Two-Sicilies.' (16) Nous allons quitter la capitale de la Lombardie et nous dinger vers celle du Piemont. (Mon3: 5) [beginning of the book] 'We shall leave the capital of Lombardia and make our way to the capital of Piemonte.' Example (9) uses the verb 'quitter' (in its negative form) as a textual resource to maintain the reader's attention on the actual portion of the text, postponing the transition to other sections of the text and to other topics. Here the verb argument is the 'green porphyry', which is more a topic than an explicit reference to a discursive domain (as it would be the case with 'the question of green porphyry'). As in the previous example, examples

How space structures discourse

579

(10), (11), (12) associate a spatial motion with a speaking activity (verbs like 'to say', 'to mention', 'to speak'). However, these three examples introduce a particular dimension: the spatial verb refers to spatial arguments ('to quit Livorno/Toscana/the land of Samnites'), in a context where a traveller is visiting Italy. Although Livorno, Toscana, the land of Samnites are topical heads on which the speaker intends to stop to add new informations, they are also places where the traveller intends to stop before pursuing his path. Another possibility is proposed by the subsequent examples, (13) and (14), where any reference to discursive activities disappears, and where consequently Naples and Verona are just places considered as landmarks for movement through territorial space, co-occurring with other spatial verbs, such as 'to visit' 3 . However, these references serve as a topical support for the description of the place: the tomb and the palace are topics that will be developed by the subsequent text. There is thus a parallelism between the descriptive path and the path of the visit: motion verbs, even if not associated with metalinguistic verbs, intervene in the textual organization. This aspect is illustrated by the last examples, (15) and (16), where a spatial movement indicates the boundaries of a textual portion, which can be a chapter or the entire book. Here we have a correspondence between 'to quit a place' and 'to close a textual segment' 4/5 .

2.2. Deixis and traces of writing processes Having examined these examples, one could ask what kind of underlying logic motivates them. The travelogues we have studied (Mondada 1994) are characterized mainly by their focus on the origo, the 'here' and 'now' of the enunciation. In fact, they designate, sometimes very carefully, the place and the moment of the enunciation. This is motivated by a pragmatic principle controlling the production of text, which could be summarized by the following: "Speak of a place when you are in this place". This principle is actualized by the co-occurrence of, on one side, the identification of the place of enunciation ('here') and, on the other side, the mention of a writing act referring to this place. For example: (17) Paestum, ce 9 mai 1812, a sept heures du soir. J'ecris assis sur les marches du temple de Ceres, ä Paestum! (Cusl: 251). 'Paestum, the 9th of May 1812, at seven o'clock in the evening. I'm writing on the steps of Ceres temple, in Paestum!'

580

Lorenza Mondada

(18) Au sommet du Visuve. A la lueur d'une eruption, ä minuit. J'ai traci ces deux lignes sur le sommet du Vesuve, a la lueur d'une eruption (DuP2: 152). 'On the top of the Vesuvio. At the light of an eruption, at midnight. I have drawn these two lines on the top of the Vesuvio, at the light of an eruption.' The reference to the circumstances of enunciation is the starting point of a description which will take the entire chapter. That allows both the identification of a unique object of discourse and its definition as an episode, which operates at the level of the textual planning and of the orientation of the reader's attention. Thus the attention is focused on the place as a landmark for the speaker's activities and the narrated events, as a discursive object, and as a structuring form. These multiple relations can be observed, not only in reference to the act of writing, but in their use as a metalinguistic planning of the oncoming text, focusing on the objects it will develop. This is especially the case when we have co-occurrence of metalinguistic verbs and localizations: (19) Puisque je suis actuellement dans I'amphitheatre, j'ai envie de vous en parier tout de suite (Bröl: 121). 'As I am now in the amphitheater, I would like to talk about it immediately. ' (20) Et maintenant que nous sommes arrives dans ces nobles murs, la fete de l'Italie, au milieu de ces chefs-d 'oeuvre que le monde adore, dans cette ville sans rivale parmi les villes italiennes, maintenant que nous sommes enfin a Florence. parlons de Florence, et parlons-en tout a notre aise; parcourons lentement ce vaste musee, tout rempli de souvenirs et de merveilles (Ja 107). 'And now that we have arrived in these noble walls, in the fair of Italy, in the middle of these masterpieces admired by the whole world, in this town without rival among the Italian towns, now that we are finally in Florence, let us talk about Florence, and talk at ease; let us wander slowly in this huge museum, full of memories and marvels.' Example (19) states explicitly and thereby satisfies the principle of speaking of the place of enunciation - in that it identifies the described place and the place where the activity of description is happening. In example (20), the moves and the actual position of the traveller and the speaker in Florence co-occur with metalinguistic verbs: to arrive some-

How space structures discourse

581

where corresponds to begin to speak of this place, to speak of a place is to loaf through this place. The search of a link between the two pertains to different mechanisms of text production. Firstly, it motivates the introduction of objects in the discourse: the discursive organization is not arbitrary and there are discursive places which are more adequate than others to speak of a particular object, especially if this object is a territorial place. This allows then a naturalization of the textual organization, not as inscribed in the speaker's strategy or choice but as obeying to the constraints of the world described. Finally, this implies an 'effect of truth', the text becoming thus a credible and faithful description of things. This constraint is explicitly stated in instruction manuals for travellers: we can find, for instance, in the Essai pour dinger et etendre les recherches des voyageurs written by L. Berchtold (Paris, Chez Du Pont, 1797) a chapter, called "Observations qui doivent etre redigees par ecrit", which advises travellers who want to write their travelogues, suggesting to have always a pen ready at hand, and to write down immediatly everything that can be noticed, even if it seems unimportant at first glance (vol. 1: 47): 1. Un voyageur soigneux aura toujours dans sa poche du papier, des plumes et de l'encre, parce que les notes 6crites ä la mine de plomb s'effacent ais6ment, et qu'on se trouve ainsi ρπνέ souvent du fruit de ses remarques. 2. Les voyageurs doivent confier au papier tout ce qu'ils trouveront de remarquable, tout ce qu'ils entendront ou liront, et meme les sensations que les difförents objets produiront sur eux; il est ä propos de le faire sur le champ, si le tems, le lieu et les circonstances le permettent; on prendra une pr6caution semblable meme pour les observations qui paraissent peu importantes. 1. A careful traveller will always have some paper, pens and ink in his/her pocket, because notes taken with a pencil are easily erased and one often risks to loose the result of his observations. 2. Travellers have to write on the paper all they find remarkable, all they hear or read, even the sensations provoked by different objects; it is better to do it at once, if time, place and circumstances allow it; one will take this precaution even for observations that appear less important, (translation L. M.)

These converging motivations, characteristic of travelogues, are identical to the principles which found the authority and the credibility of anthropological texts as field reports. In fact, the ethnologist's writing constructs the field, while presenting the act of writing as dictated by it (Clifford 1983; Atkinson 1992). Geertz (1988: 4-5) suggests:

582

Lorenzo Mondada

The ability of anthropologists to get us to take what they say seriously has less to do with either a factual look or an air of conceptual elegance than it has with their capacity to convince us that what they say is a result of their having actually penetrated (or, if you prefer, been penetrated by) another form of life, of having, one way or another 'been there'.

More generally, they can be viewed as resources available to every descriptive text which looks for a validity coming from the relation between the world and the words used to describe it.

2.3. Traveller's and reader's journeys It is as if the traveller's path drew the map of the speaker's text 6 . In the same way, the hearer is projected into the text, thanks to a reading contract established by the guidebook, which anticipates the path of the readertraveller: the reader is directed through this space, in a way that identifies the paths, differentiated by Fauconnier, in the text's world and in the reader's world. (21) D'abord, que le lecteur nous suive. s'il le juge a propos, au palais ducal (Aud4a: 99). 'First, may the reader follow us, if he agrees, to the ducal palace.' (22) Ne pas comprendre les Abruzzes, la Pouille et les Calabres dans un panorama de l'Italie, est une lacune que laissent frequemment ά leurs voyages MM. les touristes, et autres amis de la belle Italie [...]. Cette lacune nous l'eviterons, et c'est pleins du desir de la combler que nous allons conduire nos lecteurs dans le comte de Molise d'abord, pour ensuite le [sic] guider au travers du sol pulverulent, mais iminemment fertile, de la Pouille, et des verdoyantes et pittoresques Calabres (Aud2a: 205). 'Not including the Abbruzzi, Puglia and Calabria in a panorama of Italy is a mistake frequently made by tourists in their journeys, as well as by other friends of the beautiful Italy [...]. We shall avoid it, and full of the desire to repair it we shall first lead our readers in the county of Molise, and then lead him [sic] through the dusty but extremely fertile soil of Puglia and through the green and picturesque Calabria.'

How space structures discourse

583

(23) Avant que de sortir de la place Saint-Marc, je veux vous mener a la bibliotheque (Bröl: 173). 'Before we go out of the Piazza San Marco, I want to lead you to the library.' (24) Je vais d'abord vous camper au beau milieu de la ville, place des Saints-Apötres (Bro2: 91). [starting point of the itinerary described in the chapter] Ί shall first put you right in the middle of the town, in the Piazza Santi Apostoli.' But this projection of the reader into space doesn't lead him systematically to the identification with the speaker and traveller: there are moves which are specific to travel; there are moves which are specifically textual moves. Travelogues can use the reader's projection into space as a resource, while they reaffirm the singularity of the author's experience. The two levels are thus differentiated: (25) Maintenant que j'en aifini avec mes souvenirs de Come, je pourrais, mon eher ami, vous ramener encore ά Milan. Mais ce serait par trop abuser de votre longanimiti; transportons-nous done par la pensee, et tout d'un trait, sur les bords du lac Majeur (Lance 305). 'Now that I have finished with my memories of Como, I could, my dear friend, take you back to Milano. But that would be to impose too much on your forbearance; let us transport by our thought, in one go, along the shore of Lago Maggiore.' (26) Pendant cette digression, mon eher Blancey, je vous ai amene jusqu 'ά la Torre di Mezza-Via, maison isolee ou est la poste, puis -jusqu'a I'endroit ou. Von commence ä monter la montagne (de Brasses 1: 303). 'During this digression, my dear Blanchey, I took you up to Torre di Mezza-Via, an isolated house where the post lies, and then to the place where the mountain slope begins.' The reader is confined to one spatio-temporal dimension: although being parallel to the dimension of travel, it is not identical with the latter and can develop autonomously. Thus the length and hardships of the journey are cancelled (25), or the journey continues without the reader during the digressions (26). Relations between territorial and meta-textual spaces are multiple and can alter, converge or diverge.

584

Lorenza Mondada

2.4. Moves on textual linearity We saw that textual deixis conceives the text as a linear space. The use of verbs like 'to come' and 'to go' corresponds to local directions as 'further', 'after', etc., and strengthens this linearity (2.4.1.). On the other hand, verbs like 'to come back' and their synonyms make this linearity reversible (2.4.2.).

2.4.1. Strengthening linearity Here are some examples: (27) Venons a iarsenal. II est si celebre, que je fus d'abord assez micontent de trouver les salles des armes mal rangies (Bröl: 177). 'Let us come to the arsenal. It is so famous, that I was first quite displeased to find the weapon showrooms so badly ordered.' (28) Allans au Vieux-Palais et passons devant le Marchi-Neuf, construit en halle ä colonnades de bon goüt, au-devant de laquelle est un sanglier de cuivre qui jette de Veau. C'est un jeune gentilhomme fort bientourne. Ce Vieux-Palais [...] (Bröl: 253). 'Let us go to the Old Palace and proceed to the New Market, a hall with pleasant colonnades. In front of it there is a copper boar which spits water. It is a pretty handsome gentleman. This Old Palace [...]' (29) Et maintenant remontons dans notre navire, tendons les voiles, et puisse le vent nous etre favorable, et nous accorder un soujfle ami pour sortir des parages de la Corse et de I'lie d'Elbe. Nous allons ά Genes (Aud4b: 117) [end of the chapter, transition to the subsequent chapter on Genova]. 'And now let us get aboard our ship, spread the sails, and may the wind be favourable and help us out of the waters of Corsica and Elba Islands. We go to Genova.' (30) De lä je viens a la Solfatara, autrefois la marmite de Vulcain (Bröl: 352). 'From there, I come to the Solfatara, formerly called Vulcan's pot.'

How space structures discourse

585

(31) Je vais tout de suite du college au jardin des plantes, quoique ce soit fort loin (Bröl: 135). Ί go immediately from the college to the botanical garden, even if it is very far.' In all these examples, movements of the speaker allow to pass7 from a place to another, thanks to deictic verbs such as 'to go' or 'to come'. This path in the territory corresponds to a path in the textual space, for instance from one topic to another. The two deictic verbs create a different perspective on successivity: the former identifies the landmark in the source of the motion, the latter identifies it in its target. The point of view considered by 'come', in contrast with 'go', anticipates the place where one is directed to. Textual progression is constructed step by step with 'to go'; on the contrary with 'to come', it is already configured in a way that considers the target place as previously defined, foreseen, established and known. The two verbs organize differently, not only the point of view, but also the textual planning.

2.4.2. Subverting linearity Instead of following linearity, one can reverse its orientation, using verbs such as 'to come back', 'to return', 'to pass again'. Expressions like 'revenons ä nos moutons' in French or 'torniamo al dunque' in Italian illustrate this point. The text is represented as a space organized around a given anchoring point: from this point, there are different ways of continuing the text. If one wants to follow more than one line, one always has to come back to the central point of departure. The most interesting case is given by occurrences of 'to come back'. The target, which is its landmark, becomes the central point of reference, organizing the textual configuration. (a) 'To come back' plays the role of an instruction to secure text organization. This is explicitly stated when the verb co-occurs with a metalinguistic element: (al) the verb can have an argument which is a textual category. It allows thus the transition from a textual space to another (from history to travelogue, from interior landscape to exterior landscape, from digression to narration): (32) Je reviens ä mon voyage. (Pertl: 395) [after a digression on the Roman government]. Ί come back to my travel.'

586

Lorenza Mondada

(33) Apres ces details sur notre interieur, je reviens ä la route. (Pertl: 540) [after the description of the travel companions in the coach] 'After these details on our interior, I come back to the road.' (34) Pardonnez-moi cette digression, un peu longue, je l'avoue, et revenons ä l'iglise Saint-Ambroise (Lan90). 'Forgive me this digression, a little too long, I acknowledge, and come back to the San Ambrogio church.' (a2) the verb can have a metalinguistic value and refer to the places explicitly identified as objects of discourse: (35) Je dis done, pour revenir au propos dont je m'etais ecarte, qu'on ne pourrait rendre un plus grand service ä notre art favori que de donner, par des mosaiques, un vif colons ä ces admirables ouvrages du Vatican. (Bro2: 164). Ί was so saying - to come back to the topic from which I took some distance - that we could not do our favourite art a greater favour than to give, with mosaics, a lively colour to these marvelous masterpieces of the Vatican.' (36) Pour en revenir a la moderne Tarente et a cette montagne de coquillages dont nous parlions tout-ä-1'heure, on nous conduisit, pour y arriver, par un chemin que I'on appelle I'ancienne Rue des Orfevres. (StN73) 'To come back to the modern Taranto and to the mountain of shellfishes that we talked about before, one took us in order to reach it, along a path called the old goldsmiths' street.' (b) But the verb 'to come back' occurs in other cases which are more ambiguous, where textual and territorial spaces, text and travel organization mingle: (37) Revenons a nos beaux monuments de la Piazza S. Marco. (Nisl57) [after a digression], 'Let us come back to our beautiful monuments of Piazza S. Marco.' (38) Revenons ά la cour du Belvedere. D'abord, e'est un piristyle orne dans lefond [...]. (Bro2: 167). 'Come back to the courtyard of the Belvedere. First of all, it's a peristyle ornate in the bottom [...]'

How space structures discourse

587

(39) Pour en revenir ä la ville de Turin, quoique regulierement bätie, eile est, je le ripete, d'un ensemble tres-agreable (Lan36). 'To come back to the town of Torino, I repeat that it is a very pleasant whole, even if it is regularly built.' (40) Pour en revenir aux palais, la forme architectonique qui domine dans cette riche collection d 'habitations, c'est Vogive. (Lan237). 'To come back to the palaces, the dominant architectural form in this rich collection of houses is the rib.' The verb has a more or less abstract value depending on the objects to which one comes back. The object can refer to plurality (37), to a single place (38), to a complex and articulated place (39) which becomes more a notion than a physical entity, or to a class of objects (40). The value of the verb depends on the co-text, which is either a digression or a spatial description. Other non-deictic verbs play the same role, as does the verb 'to abandon': (41) Mais je n'ai rien dit encore de Spoleto, ni de Foligno: force m'est d'abandonner le lac pour retourner sur mes pas. (Gas2: 254) 'But I still have not said anything of Spoleto, nor Foligno: I have to abandon the lake treading back in my own footsteps.' Path organisation is a preliminary for the structuring of the text: (42) Nous laissons a droite le temple de la Tosse. [...] Revenons au temple de la Toux, situe tout pres de nous dans le jardin des chanoines de la cathidrale. (Castellan 2: 148; 150) 'We leave on the right the temple della Tosse. [...] Let us come back to the temple della Tosse, situated very close to us, in a garden of the cathedral canons.' (43) Des I 'entrie, on voit des escaliers balustres de marbre blanc, et ces escaliers, de face et de cöte, faisant mille retours et causant par consequent beaucoup d'effets tourmentes aux yeux, mais la richesse du marbre blanc est seduisante. Cet escalier conduit a une terrasse terminee par une grote percie ä jour d'ou sort une grande figure d'Hercule. II y a beaucoup d'effet, mais rien ä servir de modele. Revenons au plein pied. A gauche, en entrant, quatre sallons peints dans toutes leurs voutes [...] Je reviens encore au coup d'oeil des escaliers qui se trouve en face de la grille du jardin; c'est le tourment

588

Lorenza Mondada

pricisiment pour les yeux et comme copie d'apres les mauvais dessins de la Joüe; ce sont des reves. (Bergeret 39). 'From the entry, one sees banisters of white marble, and these stairs, in front and on the side, make thousands of twists and turns, causing a lot of tortured effects to the eyes; but the richness of white marble is appealing. This stair leads to a terrace which ends with a cave in which a big figure of Hercules can be seen through a bored hole. There is a lot of effect, but nothing useful as a model for us. Come back to the ground floor. On the left, when one enters, four rooms entirely painted in their vaults. I come back again to the perspective of the stairs in front of the garden gate; that's precisely the torture for the eyes as if it were a copy of the bad drawings of la Joüe.' (44) II vaut mieux revenir sur mes pas que d'oublier plusieurs palais que nous avons vüs, environ deux heures apres etre sortis de Savonne, c'est-ä-dire une lieüe environ, dans le village d'Ardisol, bien pres des montagnes sans pouvoir s'etendre. [suit la description des palais] (Bergeret 39). 'It is better to tread back on my own footsteps than to forget some palaces that we saw, some two hours after leaving Savona, that is, at about one league, in the village of Ardisol, very close to the mountains and thus with no possibility of extension.' When the speaker deals with complex entities, he faces a problem of linearization (cf. Levelt 1982a; 1982b; cf. infra 3.3.); one way of solving this problem is to describe a path which links different parts of these entities together; the more complex the figure, the more the path implies loops, roundabouts and returns.

2.4.3. Digressions Markers dealing with disruptures of text linearity are signposts which apply the itinerary model and, at the same time, allow exceptions to this model. The latter case serves to introduce digressions which are specially related to roundabouts: (45) Mais avant de parier de la route de Verone, nous ferons une digression pour lespetites villes de Bassano et de Roveredo. [...] Bassano η 'est qu 'ä douze lieues de Roveredo et de Trente; il y α des voyageurs qui vont visiter ces deux villes avant que d'aller a Virone; et c'est id le lieu d'en donner une idie. [...]

How space structures discourse

589

Apres cette digression pour Bassano et Roveredo, nous allons reprendre la route ordinaire de Verone (Lal7: 151; 154; 160) 'But before we talk about the road of Verona, we shall do a digression for the small towns of Bassano and Roveredo. [...] Bassano is only twelve leagues away from Roveredo and Trento; some travellers go to visit these two towns before they go to Verona; so it is here that one should give an idea of them [...] After this digression about Bassano and Roveredo, we shall take the ordinary road to Verona again.' The digression itself obeys to a proximity principle, being anchored to a place which is ideally 'here' but can also be nearby. Digression is represented as a detour. The end of the digression corresponds to a return to the central path which is also the thread of discourse. The iconicity of the linearity of the path is a structuring principle which is so strong that any disrupture must be explicitly signalled and delimited by spatio-temporal deictics. We find an identical constraint in another kind of text, the ancient periplus, a sort of maritime guide, functioning as a linear map, indicating distances, landmarks and particularities of the coasts. The periplus respects the linearity of the path: any digression or discontinuity must be signalled. This is frequently the case in the description of archipelagos, islands and crossroads where choices have to be made about possible paths, where linearity is suspended and other figures are preferred. Jacob (1980: 116) quotes organizing sentences such as: (46) Mais je reviens au continent d'oü je m'itais eloigne vers la Corse. (Pseudo-Scylax Periplus) 'But I come back on the continent from where I went away towards Corsica.' This sentence illustrates, like the examples above, a textual move to recover a linear order.

2.5. Textual deixis and iconicity When the text uses a path model, it is ruled by an isomorphism between text structure and world structure. Although, according to Kurzon (1985), narratives don't need any textual deixis because they are organized by an iconic sequentiality, when the structuring power of the path model is missing, they do in fact need explicit

590

Lorenzo Mondada

signposts. Even if spatio-temporal planning markers express alternative possibilities of text structure (for example in the form of a network), they continue to secure and to strengthen the efficiency of the path model. Path is an iconic syntactic model: it is based on the correspondence between the sequentiality of sentence components and the sequentiality of described actions: (47) Arrives presque a la nuit chez le general Leclerc, absent; son fils malade. A l'auberge de la Poste, remplie. Attendorn ä la rue impatiemment. Logis ä I 'eveche; y courons. (Des 192) 'Arrived almost at night at general Leclerc, absent; his son ill. At the inn of the Post, full. We shall wait impatiently in the street. Quartered at the bishop's palace; we will run there.' cf: (48) The baby cried. The mommy picked it up. (Sacks 1972) (49) John got up and left. (VS ?John Ιφ and got up.) (Givon 1989: 104) Such examples show that a preferential isomorphic relation exists between the order of actions in time and the order of propositions in the sentence8. If the classical examples quoted in the literature concern temporal order (linked to causality), the examples above use spatio-temporal order as a structuring resource. The markers analyzed above strengthen or instaure this iconicity or offer other solutions, which are also iconic although pertaining to rival models. We can thus speak of the iconicity of textual deixis, which was noticed by Greenberg (1985: 272): Within the text there is metaphorical pointing back so that we are in some sense talking about physical space as an iconic model for discourse. Also past mention is regarded as in some sense distant so that we see already one of the important characteristics of this mapping.

In our corpus, it is the path which models both text and space, thus allowing it to rule the linear distribution of discourse objects along the pages as along the itinerary. The examples generally quoted in the literature, argue in favour of the thesis iconicity of grammar; our examples refer to the discursive views of the system and not to its grammatical properties. Givon is one of the rare authors who speaks of iconic pragmatic principles (Givon 1989: 99). Thus, the isomorphism between text and space is less related to the proper

How space structures discourse

591

hypothesis concerning the iconicity of grammar, and more linked to a particular textual strategy, which aims to motivate, objectify and naturalize processes of object discourse construction. This interpretation avoids the risk of proposing realistic formulations of the grammar iconicity (which view language as mapping the 'real' world). It also indicates that speakers tend to structure their discourse in a way that suggests an isomorphic relation with the structure of its objects. Isomorphism is less a linguistic explaining principle than a resource used by speakers in order to produce meaningful texts and to structure intelligible and persuasive objects of discourse. Travelogue is a type of text which uses this resource in that it chains discursive objects together in a particular way: its goal is to present itself as a map of the described path or territory. This corpus focuses on the persuasive and structuring power of isomorphism, and the textual effects of transparency it produces. To describe a town or a region following the logic of the itinerary is thus not to realistically submit the text to the world's constraints, but to use the link to the world or to its experience as a resource to structure the text.

3. The structuring power of space as a resource for organizing discourse Through textual deixis and its widening to other uses and textual effects, we described some aspects of the structuring power of space, a resource used to organize other entities than space, namely discourse. This possibility of organization is crystallized in the denomination of some textual entities: for example, if the verb 'to proceed' is used for textual planification, its nominalization, 'proceedings' refers, in particular, to the record of papers given in a congress. The term 'discourse' itself refers to the idea of path ('discursus'), as the term of 'period', signifies 'the path around'. Moreover, it is not an accident if the term 'itinerary' is used in two different senses: one refers to the path, the other refers metonymically to the report of this path. The same kind of process is observable in the term 'theatrum', which refers, until late Renaissance culture, not only to a playhouse, but to an exhibition room where objects of knowledge are spatially distributed and organized, as well as to a treatise covering a particular domain. In the same way, the term of 'panorama' refers to an architectural device appearing at the beginning of the 19th century, showing a painted global view of a city, a harbour or a battlefield, and also, later, to a text presenting a synthetic view of a given domain. Another example is the term of 'exposition', which refers to a place but also, as certified by the

592

Lorenza Mondada

Grand Dictionnaire Universel du xixe siecle, a part of a literary work in which the subject and the general circumstances of its action are presented (cf. Hamon 1989). All these terms share the same characteristics: referring both to spaces and textual devices, as both organize and distribute objects of knowledge or objects of discourse in space. They help us to better understand more general processes and their underlying spatial structuring power. As we saw, spatial designations provide many terms referring to discourse. Among the markers of textual organization9, expressions such as 'in the first place', 'in the second place' ... allow to draw a topic agenda conceived as a succession of places based on the spatial positioning of the arguments on the page. These markers pertain to the techniques of the 'dispositio' in classical rhetorics. The space mentioned above is not merely metaphoric. It materializes in the arts of memory techniques (Yates 1966), developed by the rhetorical tradition since Quintillian and Cicero, leading during the Renaissance to encyclopedical and mystical theories. This mnemotechnical device consists of the identification of places (loci) where to locate images or arguments composing a discourse (imagines agentes). Artificial memory is based on the organization of discursive objects into spatial ordered dispositions; once a complex familiar space, such as a palace or a church has been chosen, particular spots (such as windows, columns, chapels, etc.) are picked to locate certain images. While producing the discourse, it suffices to mentally revisit the different places to remember the corresponding arguments. As shown by the arts of memory, as well as experimentally by Klein (1982), Levelt (1982b), Linde and Labov (1975), a path is a descriptive strategy allowing to unroll a set of ordered objects exhaustively. More generally, to distribute objects in space, following an intelligible form, secures their comprehension, their analysis and even their manipulation. Natural sciences proceed efficiently on this principle through graphical selections, schematizations and mathematizations of their objects (Lynch 1985); in the same way, social sciences make symmetries and oppositions of phenomena intelligible by organizing them into taxonomic tables (Goody 1977); cognitive linguists working on space grammar analyze semantic and cognitive relations of entities by means of diagrammatic visualizations (Mondada 1994). Such processes focus our attention less on space referred to by a description, and more on space as an underlying logic of description. The focus of analysis shifts from a representation of space to a space of representation. The efficiency of this space of representation is expressed by its structuring power, capable to order the world while naturalizing this order by means of isomorphisms and effects of transparency. Spatialization tech-

How space structures discourse

593

niques allow to construct an intelligibility which seems to originate from things themselves.

Notes 1.

2.

3.

4.

It is interesting to remark that linguists themselves use spatio-temporal metaphors to conceptualize discourse. See for example the ambivalent definition of the linearity of the signifier by Saussure: "Le signifiant, 6tant de nature auditive, se d6roule dans le temps seul et a les caractöres qu'il emprunte au temps: (a) il repr6sente une itendue, et (b) cette Stendue est mesurable dans une seule dimension: c'est une ligne" (1972: 103). E.g.: Revenons ä Milady, qu'un regard jet6 sur les cötes de France ous a fait perdre de vue un instant. Nous la retrouvons dans la position d6sesp6r6e ού nous l'avons laiss6e, se creusant un abime de sombres r6flexions. (Dumas, quoted p. 62) ('Let us come back to Milady, that a glance on the shores of France has put out of our view for a while. We find her again in the desesperate position where we left her, digging a gulf of dark reflexions'). Que le lecteur nous permette de le ramener ä la place de Grfeve, que nous avons quittee hier avec Gringoire pour suivre la Esmeralda. (Hugo, quoted p. 63) ('May the reader allow us to bring him back to the Place de la Grtve, which we Ι φ yesterday with Gringoire to follow the Esmeralda'). See: Nous allons suivre la route de Milan ä Venise; nous ne visiterons, dans ce volume, que cette partie de l'ltalie, qui s'appelle par d6rision le royaume Lombardo-Vinitien (Nisi00) [beginning of the chapter entitled «De Milan ä Venise»]. ('We shall follow the road from Milano to Venice; we shall visit, in this volume, only the part of Italy called in mockery the Lombardo-V6netian kingdom'). Je te demande pardon, mon eher ami, d'avoir flani si longtemps dans la via Dora-Grossa, au lieu de te faire connaitre au plus t6t, comme c'6tait mon devoir, le Turin monumental. [...] J'arrive aux monuments, et je commence par le Palais Madame, attendu que de la rue que nous quittons il n'y a qu'un pas ä faire pour l'avoir devant soi. (Lan29-30) ( Ί beg your pardon, my dear friend, for having wandered so long in the via DoraGrossa, instead of letting you know earlier, as it was my duty, the monumental Torino. [...] I arrive to the monuments, and I begin with the Madama Palace, since one step is enough to be in front of it from the street we leave'). Je ne vous ferai pas promener de refuges en refuges; vous avez hate comme moi de descendre en Italie [...] J'abrägerai done ma description. (Nisl3) ( Ί shall not lead you from hut to hut; like me, you are in a hurry to come down to Italy [...] /shall thus cut short my description.'). In these examples, the visit ('to visit', 'to lounge', 'to walk') and textual activities are explicitly associated. We saw that textual deictic markers could precise a textual localization as a relation of inclusion (e.g. 'in this paragraph'). This localisation corresponds to verbs such as 'to enter/to exit'. Thus entering into a territory corresponds iconically to beginning a chapter; the same holds for exit and close a chapter. Here are some examples:

594

5.

6.

Lorenza Mondada

On entre en Lombardie ä Sesto-Calende sur le Tessin, ä huit lieues de Milan. (Val27) [beginning of a chapter on Lombardia] ('We enter Lombardia at Sesto-Calende on the river Ticino, at eight leagues from Milano'). Je ne peux plus ni admirer, ni regarder, ni meme voir. Ma sensibilit6 est 6puis6e; je sors. (DuP2: 4) [end of the chapter] ( Ί cannot admire or stare or even look any more. My sensitivity is exhausted; I get out.'). Uses of the verb 'sortir' ('to exit', 'to go out') show the link between scattering positions in the text, delimitating parts, organizing the text into episodes, centering it on topics. Thus we usually say 'sortir du sujet' or 'essere fuori tema' to signal a topical break. Cf.: [general considerations on the site of Herculanum] Mais on doit consid6rer que les homnies, devant presque toujours au hasard les plus singuliferes dtecouvertes, n'y ajoutent que peu de choses par leurs seules lumiferes [...] Tdmoin, sans sortir du sujet, les vitres qui sont au plus du quatri&ne sifccle, quoique le verre füt connu avant la fin de la ^publique. (Due 147).('But one has to consider that men, as they almost always depend on case for the most original discoveries, add only very little by their own imagination. A proof of that are, without leaving the subject, the window panes which are at best from the fourth century, even though glass was known before the end of the Republic.'). Rentrons dans la ville, c'6tait en sortir trop töt\ l'objet le plus visible est la tour degli Asinelli, droite et menue comme un cierge. (Bröl: 204) ('Let us come back in the city, it was too early to come out·, the most visible object is the Asinelli tower, straight and slim as a candle.'). Textual space, with its boundaries, its inner and outer regions, is also structured by the opposition between center and periphery, as shown in the following academic examples: The issues [..] that this study addresses are centered around the following questions (Svol24). La verifica poträ evidenziare problemi, che qui accantono (Ber249) ('The verification will highlight problems that, for the moment, I put aside.'). Here problems of textual deixis concern motion verbs and touch upon problems of thematic markers. In this sense, to close a part or to go away from a topic represents convergent movements, both oriented towards the exterior and to a new space. See the following examples, the first from a travelogue, the second from an academic discourse. L'amphithiätre est tellement ä part des autres 6difices de la ville, que nous avons dü le signaler d'abord au lecteur. (Aud4a: 50) ('The amphitheater is so apart from the others buildings of the town, that we must signal it at first to the reader.'). To do this we shall look at a piece of actual conversation. Before doing so, however, I want to turn abruptly away from these questions to an entirely different area (Sche72: 79). Cf.: Maintenant sur quel point de l'ltalie allais-je me diriger d'abord? Force m'6tait de renoncer ä la marche que suit religieusement chaque touriste dans son pfclerinage: Chamb6ry, Turin, Genes, etc. et la 16gende consacr6e. [...] Commencerai-je par Naples? [...] Suivons la mode, et commerifons par une de ces belles r6publiques italiennes du moyen-äge. Que le val d'Arno re?oive mon premier salut. (Audi,20) [beginning of the volume] ('Now towards which point of Italy would I move first? I must renounce the path religiously followed by every tourist in his pil-

How space structures discourse

7.

8.

9.

grimage: Chamb6ry, Torino, Genova, etc. and the consacrated legend. [...] Should I begin by Napoli? [...] Let us follow the trend, and let us begin with one of these beautiful republics of the Middle Ages. May the Val d'Arno receive my first greetings.'). See this series of examples, all containing the verb 'to pass', showing the same process, based on a parallel between a textual move and a motion in a territory: (a) Dans le septifcme [cabinet], qui est Γ arsenal, toutes sortes d'armures antiques, modernes et orientales, d'une richesse et d'un choix surprenants. Je passe 16gferement lä-dessus, pour ne rapporter que le gros mousquet dont le canon est tout d'or. (Bröl: 258) ('In the seventh [cabinet], which is the arsenal, all kind of antique, modern and oriental armours, of a surprising richness and choice. I pass quickly on this, and I only mention the big musket with an entirely golden barrel.'). (b) De V6rone, dont notre dessein n'6tait d'offrir que les traits principaux, nous passons ä Brescia, ville ä 16 Heues de Milan, dans une belle plaine sur la Garza et la Melle. (Mon2: 220) ('From Verona, where we planned to give only the main characteristics, we pass to Brescia, a town situated 16 leagues from Milano, in a beautiful plain on the rivers Garza and Melle.'). (c) Passons done, sans nous y arreter, devant Vicence, la ville natale de notre maitre Palladio, et contentons-nous de saluer de loin, puisque nous devons y revenir, Γ antique Patavium, la moderne Padoue, la patrie de Tite-Live et celle aussi du tyran Angelo Malipieri. (Lanl31) ('Lei us pass then, without stopping, by Vicenza, birthplace of our master Palladio, and be satisfied with greeting from a far - since we shall come back - the antique Patavium, the modern Padova, birthplace of Tito Livio and also of the tyrant Angelo Malipieri.'). (d) La conf6d6ration des Samnites existait dans le comt6 de Molise, dont hernia, dans laquelle nous entrons, 6tait la mdtropole, comme la Philadelphie l'est des Etats-Unis; mais passons, sans trop nous arreter, parmi ces mines encore fümantes, et, poursuivant notre chemin parmi les d6combres gisans ä chaque angle de ses rues et de ses places publiques, p6n6trons dans la premifere des Abruzzes, aussi riante qu'est triste le comt6 de Molise. (Aud2a: 205) ('The Confederation of the Sannites existed in the county of Molise. The town of Isernia, where we enter, was its metropolis, as Philadelphia is for the United States. But let us pass, without stopping too much, across these still smoking ruins and pursue our journey across the debris scattered at every corner of its streets and its public places, let us penetrate in the first part of the Abruzzi, as smiling as the county of Molise is sad.'). 'To pass' can refer to a move towards a target which will be a topic further developed (b); or to a transitional movement (a), (c), (d). This property was first noticed by Jakobson: «If the chain of verbs 'veni, vidi, vici' informs us about the order of Caesar's actions, it is first and foremost because the sequence of coordinated perfects is used to reproduce the successivity of the related events. The temporal order of enunciation processes tends to mirror the order of the enunciated processes.» (1965: 27). Greenberg echoes him and widens his remark: "... the order of elements in language parallels that in physical experience or the order of knowledge (1966: 103). Cf. Haiman, 1980, 1985. In the same way, Sweetser (1990: 46) remarks that some connectors refer to path, as 'anyway' or Italian 'tuttavia'.

595

596

Lorenza Mondada

References Atkinson, Paul 1990

The Ethnographic Imagination. London: Routledge.

Textual Constructions of

Reality.

Clifford, James 1988 The Predicament of Culture. Cambridge: Harvard University Press. Conte, Maria E. 1988 Condizioni di coerenza. Firenze: Nuova Italia. Ehlich, Konrad 1983 Deixis und anaphora. In: Rauh, Gisa (ed). Essays on Deixis. Tübingen: Narr, 79-99. Fauconnier, Gilles 1984 Espaces mentaux. Paris: Editions de Minuit. Fillmore, Charles J. 1975 Santa Cruz Lectures on Deixis. Bloomington: Indiana University. Mimeo. Geertz, Clifford 1988 Works and Lives. The Anthropologist as Author. Cambridge: Polity Press. Givön, Talmy 1989 Mind, Code and Context. Essays in Pragmatics. Hillsdale: Lawrence Erlbaum. Goody, Jack 1977 The Domestication of the Savage Mind. Cambridge: Cambridge University Press. Greenberg, Joseph H. 1966 Some universale of language with special reference to the order of meaningful constituents. In: Greenberg, Joseph H. (ed.). Universals of Language. Cambridge: ΜΓΓ Press, 73-113. 1985 Some iconic relationships among place, time, and discourse deixis. In: Haiman, John (ed). Iconicity in Syntax, Amsterdam: Benjamins, 271-288. Haiman, John 1980 The iconicity of grammar: Isomorphism and motivation. Language 56/3: 515-540. 1985 Natural Syntax: Iconicity and Erosion. Cambridge: Cambridge University Press. Hamon, Philippe 1989 Expositions. Paris: Corti. Jacob, Christian 1980 Ecritures du monde. In: Cartes et figures de la terre. Paris: Centre Pompidou, 104-119. Jakobson, Roman 1965 A la recherche de l'essence du langage. Diogdne 51: 22-38. Kurzon, Dennis 1985 Signposts for the reader: A corpus-based study of text deixis. Text 5/3: 187-200. Levelt, Willem J. M. 1982a Cognitive styles in the use of spatial directions terms. In: Klein, W. and R.J. Jarvella (eds). Speech, Place and Action. New York: Wiley and Son. 1982b Linearization in describing spatial networks. In: Peters, Szanley and Saarinen, Esa (eds). Processes, Beliefs, and Questions, Dordrecht: Reidel, 199-220.

How space structures discourse

597

Mondada, Lorenza 1994 Verbalisation de l'espace et fabrication du savoir: approche linguistique de la construction des objets de discours, PhD, University of Lausanne: Linguistic Institute. Ong, Walter 1982 Orality and Literacy. The Technologizing of the Word. New York: Routledge. Rauh, Gisa 1978 Linguistische Beschreibung deiktischer Komplexität in narrativen Texten. Tübingen: Narr. 1983 Aspects of deixis. In: Rauh, Gisa (ed.). Essays on Deixis, Tübingen: Narr, 9-61. Sacks, Harvey 1972a On the analyzability of stories by children. In: Gumperz, John J. and Dell Hymes (eds). Directions in Sociolinguistics: The Ethnography of Communication. New York: Holt, Rinehart and Winston, 325-345. Saussure, Ferdinand de 1972 Cours de linguistique generale. Paris: Payot. Sweetser, Eve 1990 From Etymology to Pragmatics. Cambridge: Cambridge University Press. Vuillaume, Marcel 1990 Grammaire temporelle des recits. Paris: Minuit. Yates, Frances 1966 The Art of Memory. London: Routledge and Kegan Paul.

The (meta-)textual space Winfried Nöth

1. The semiotic paradox of the text as a space Although sound travels in all directions, spoken language and speech perception are basically unidirectional in their phonetic form. Speech develops its message in a temporal sequence, not in a spatial configuration. In spite of this basic linearity, everyday language refers to the product of human speech, the text and its structures, more typically by means of spatial than in terms of linear concepts: There are points or places somewhere in a text. The text may contain several lines of reasoning. Within the framework of a message, we refer to ideas developed further above, below, in the middle, or even from front to back. The spatial metaphors of the metatext, our way of referring to the text, are often alive and transparent, as in the examples given above, but sometimes, they are opaque and apparent only to the etymologist, as in the metaphors of the text as a 'tissue' (L. textus) and its structure as a 'building' or 'construction' (L. structura). Both forms of metatextual metaphors are the topics of the following discussion. According to Leibniz's classical definition, time is the "order of succession", and space is the "order of coexistence" (Letter to des Bosses of 16-61712): Due to its basically unidirectional extension in time, spoken language hence does not permit simultaneity of structures to a large extent. It is true that the so-called suprasegmental phonetic features of pitch, stress or intonation do contribute so to speak a second dimension to the linear sequence of phonemes, but several spoken, and basically also written, words can usually not be perceived simultaneously. On these premises, the semiotic paradox of the spatial nature of the text begins with the metaphor of the linearity of speech. Saussure (1916: 49) called it the spoken chain, and although emphasizing the mere temporality of speech, he nevertheless ascribes to spoken language the first dimension of space, namely linearity (1916: 70): "The signifier, being auditory, is unfolded solely in time from which it gets the following characteristics: (a) it represents a span [etendue/ extension], and (b) the span is measurable in a single dimension, it is a line."

600

Winfried Nöth

One might presume that the spatial conceptualization of the text has its origins in the invention of writing and in the metaphorical transfer of characteristics of the writing space to speech. Consider, in this context, Derrida's (1967: 39) definition of writing as the "spacing of speech". In fact, 'linearity' seems to be a literal way of referring to the geometry of the line in alphabetical writing, if we disregard the fact that the individual letter is itself a two-dimensional configuration. On the inscribed page, the onedimensional lines develop into the two-dimensionality of the writing surface, which, in our roman writing system, extends horizontally from left to right and vertically from above to below. With the binding of the written or printed pages to a book, the text even acquires a three-dimensional form, a volume, which allows references to places at the front or at the back of the text to be understood in a literal sense. Nevertheless, the metaphor of the text as a space is by no means primarily derived from the material substance of writing. There are several reasons why this metaphor is relatively independent of our writing culture. First, many spatial metaphors of the text are clearly independent of the form of writing. Thus, nothing reminds us of writing when we have a standpoint, put an argument into the foreground or into relief, place something on the agenda, raise a topic, pick up an idea, lay down, reject, or even throw out an argument. Secondly, even in written texts, the usual metatextual references are not always spatial, but frequently also temporal (as, e.g., previously mentioned, earlier on in this section, or: we'll come to it later). Thirdly, even when the written metatext contains spatial concepts, its spatial references are often literally inappropriate. Thus, the above mentioned passages of a written text may well be found below, on the previous page, and the arguments discussed further below may well follow above on the next page. Finally, a further argument against the primarily scriptural motivation of the spatial metaphor of the text is the fact that even language referring to time and temporal order is highly spatial in its metaphors (cf., e.g., Wunderlich 1985). Thus, e.g., metaphorical references to the length or the brevity of a text can be taken as referring either to the time of discourse or to the one-dimensional space occupied by the linearity of writing.

2. Geometry of the textual space Let us take an inventory of the loci in the textual space. We find them in transparent and opaque metaphors whose literal meanings (a) designate space in abstract geometrical terms (e.g., as points or lines), (b) refer to the spatial structure of concrete objects (e.g., field or chain of ideas), (c) presuppose spatial conditions (e.g., proceed or in the following), or (d) are lo-

The (meta-)textual space

601

cal deictics (e.g., here, there, at this point). The geometry of these metatextual spatial concepts consists in its zero dimension of points, in its first dimension of lines, in its second dimension of surfaces or planes, and in its third dimension of bodies (including containers).

2.1. Points Designations for points in the metatextual space are the geometrical (or geographical) metaphors of the point, the position, the locus, and the deictics here and there. In deictic usage, the metatextual points actually remain only zero-dimensional when the metaphors are used autodeictically (cf. Harweg 1990) or self-referentially, as in here, at this point or in this place. With heterodeictic (cf. 1990) expressions, such as elsewhere or at that point, we move from zero- to one-dimensionality, since there is reference from the speaker's point to another point further away, which implies the geometry of a line connecting these two textual points. In nondeictic metaphors of the zero-dimensional textual space, points are the static loci of something standing, attached, or visually fixed, e.g., a standpoint, viewpoint, point of view, focal point, position, stance, or thesis ('something placed down'). Related metaphors are points referring to the crossing of lines, such as crossroads or juncture. In static verbal metaphors, points are made, stated, or ajfirmed. In spite of their geometrical zero-extension, points made in a discourse typically refer to something pertinent, essential, or even crucial. This discrepancy between the geometrical and metaphorical usage of the word leads to a catachresis in the metaphorical expression to underscore a point. The points discussed so far mark ideas or concepts developed within the text. However, the metatextual space does not only consist of references to ideas expressed monologically or even self-referentially within the text. Standpoints often also refer to one participant in a discussion in relation to the other participant. In this case, the speaker's positions, which are sometimes oppositions, characterize loci within a "dialogical space", whose points are the places where the dialogic agents stand. In contrast to this dialogical space created by intertextual references, we will refer to the metatextual space created by intratextual (monological) references as "conceptual space".

602

Winfried Nöth

2.2. Lines When points and positions are dynamically connected, they are the end points of metaphorical lines which constitute the one-dimensional textual space. Along such lines, we get to, arrive at, reach, pursue, follow, leave, bring up, raise or drop a point. The resulting line is a thread which we can lose or pick up, a chain of arguments, a way, a path, & flow of thought, a stream of ideas, or a guide line. Any kind of heterodeixis (e.g., there, further above, in the following) draws a metaphorical line between the reference point and the given point of discourse. Metatextual lines can be bound by a point of departure and end point, directed or telic (from a source to a goal), or atelic with a vague or open end, as in see above/ below. A further topical locus along the line is its middle. In addition to such conceptual lines within the text, there is a rich potential for metaphors in the dialogical space of the text. For examples see Reddy's (1979) study on the conduit metaphor of communication and Brünner (1987).

2.3. Surfaces Two-dimensional textual spaces begin wherever conceptual paths branch out, middle courses are steered, or parallels are drawn. For the designation of the two resulting dimensions of the textual space, the designation of surface is not in use, but a two-dimensional conceptual space may have an extension in terms of length and breadth. We refer to its lateral extension by on the one side and on the other side. Borrowing from the geometry of the circle, we refer to the center and the periphery of a conceptual space. More specific metaphors of the two-dimensional textual space are derived from the language of the land surveyor. Thus, concepts are defined, subjects are limited, fields are marked, borderlines are drawn or crossed, depths of arguments art fathomed, or common places are discovered. The two-dimensional space of pictures and writing is transferred to the textual space when we refer to the frame (work) of a discussion, draw, sketch out, illustrate ideas, give marginal comments or mention something in a footnote. The latter concept is twice metaphorical, containing a first transfer from the domain of the human body to the printing space and a second transfer from printing to conceptual space.

The (meta-)textual space

603

2.4. Bodies in textual space The most unspecific kind of physical entities occupying the metatextual space are stuff or matters. Although etymologically referring to unshaped substance (L. materia), we can have the matter in the hand or let the matter drop. Physical bodies of an as yet unspecified shape in the metatextual space are things, objects, or pieces of them. Things are both given and produced in discourse: we can find or hit upon something, but can also make a thing of an affair. Objects are bodies whose place is at the goal or aim of a journey in textual space. Objects are furthermore presupposed in discoursive actions such as to represent, to lay down, or to put (or throw) in (a remark). Shaped bodies in the metatextual space are solid, hollow, externally or internally structured. Large solid bodies tend to be considered as textual impediments like mental blocks or stumbling blocks (see 4.3.). For hollow and open textual bodies, which may be full or empty of content, see Reddy's (1979) study of the container metaphors in discourse. Bodies whose external structure provides the source domain of textual metaphors are the ones which describe the text anatomically as a biological organism. The text may be a body (copy), or have an organization, parts, sections, headings), a foot{note), and its propositions may entail other ones. Dominant among the internally structured textual bodies is the metaphor of the edifice of thought. It is a building which may be closed, locked, and inaccessible, or open, permitting views, vistas, having an access or a way out. By steps we reach higher levels of the conceptual edifice. It has a structure and levels of which we can draw the contour, floor plan, or blue print. The architectural metaphor of textual structure or construction refers to the domain of the visible and connotes transparency and clarity. Its opposite is expressed by the metaphor of the labyrinth of ideas and by a variety of metaphors referring to changes of the original order (e.g., displaced or twisted ideas) or to objects of an invisible or uncertain internal structure. In such cases, ideas are, e.g., mixed up, implicit (L. 'wrapped up'), involved (L. 'rolled up'), or entangled.

2.5. Continuity, coherence, and visibility The topography of the ideal textual space, which permits an optimal presentation or development of concepts, evinces continuity of lines, coherence, integrity, and completeness of surfaces and of bodies as well as coherence and stability of its structures. Deficiencies in these respects are

604

Winfried Nöth

linear discontinuities such as mental leaps or jumps, interruptions or a breaking in (on the conversation). Surface impairments are stains, (wist)cracks, or weak points which give in. Bodily impairments are gaps, lacunae, breakdowns, or collapses, which result in pieces or fragments of the conceptual body. The ideal textual space is furthermore one in which the observer has a free visual orientation. The conceptual loci have to be clear (misunderstandings must be cleared up), the points have to be discernible, lucid, and in perspective so that we can see, notice, or discern the points in the text from many aspects or points of view and obtain a good survey and many insights. At best, there is a multiple visibility of the textual objects when they are transparent or reflected. The unclear is cleared up, the obscure is made transparent, the invisible is put in the right light, illuminated, spotlighted, or elucidated.

3. The three-dimensional textual space The three dimensions of the textual space reflect the three main dimensions of human orientation, the horizontal or perhaps better lateral (right/left), the vertical (above/below), and the sagittal (front/back). All three dimensions occur in the description of the textual space, but they differ in their frequency of occurrence and the semantic values associated with them. Hence, textual space is anisotropic, i.e., it is a space whose various directions have different structural values.

3.1. Verticality Although we write from left to right in our roman writing system, the predominant dimension of reference in written texts seems to be the vertical one. There are passages above (up) or below (down, at the bottom), heads, headings, headlines, top news, superscriptions, subscriptions, underlinings, underscorings, bottom lines, or footnotes. The vertical dimension of the textual space is also predominant with those metaphors which are not derived from writing. Up in the textual space are the arguments of top importance. We refer to them above all, as being at stake, or, using metaphors etymologically referring to the human head, as being of chief or capital importance. Further down from these superordinate matters, and in a relation of dependence ('hanging down') from them, are the subordinate or subsumed arguments. If we overlook or pass over a mistake which we can neglect, we apparently leave it "below" us.

The (meta-)textual space

605

Along the vertical dimension, ideas can be played up or down, over- or underestimated. While good ideas may be praised to the skies, bad ones may lead to pitfalls or even to an abyss. Such associations of up and down with positive and negative connotations are valid only in the sphere of the visible verticality. In the sphere of the vertically invisible, that which extends vertically below a surface, the spatial axiology is reversed. The more important, the deep structure is further down, while the less essential surface structure is further up. Basic research goes to the roots of the problems, and deep, deeply rooted, or profound (L. 'deep') thoughts are certainly more important than superficial or shallow ones.

3.2. Sagittality The sagittal (front-back) dimension in conceptual space is prototypically represented by the opposition between the foregrourid, in which we place an important idea, vs. the background, to which we relegate the less important arguments. Important things stand out in relief, come to the fore, or are anticipated (L. 'taken beforehand'), less important things take place behind the scenes, are postponed (L. 'set behind'), or set back. In dialogical space, the sagittal dimension is the line between the speaker and the hearer. Along this line there are approaches and withdrawals (e.g., of proposals). In written texts, the sagittal dimension is only of marginal importance. In books and newspapers we find passages or pages at the front or at the back. This description gives the printed space an intrinsic perspective: the front remains front irrespective of the reader's changing orientation in the process of reading. The front and the back of the writing space is also implicitly referred to in the expressions of the reverse, or the obverse side of the coin or the other side of the story. These idiomatic phrases are coined from the reader's deictic perspective, since the front and the back of a coin or page depend on and change with what the observer sees at a given moment.

3.3. Laterality The right-left dimension in metatextual space is only weakly developed, mostly in references to the sides of the textual space, sometimes with reference to the speaker's hands. The sides can be opposed to each other or to the center of the textual space. When the sides are in opposition to each other, we have to take sides in dialogical space or to consider both sides in conceptual space. In spite of

606

Winfried Nöth

the opposition inherent in these expressions, the two sides are usually not specifically associated with the left and the right, but they have to be differentiated by indexical expressions, e.g., on the one hand/side vs. on the other hand/side. The sides are in opposition to the center when we refer to side issues or brush an argument aside. Here, the opposition between the two sides is semantically neutralized, since both sides are equally remote from the center. From the domain of written space, the metaphor of marginal comments seems to conform to this schema of laterality, but, to be precise, a written page has not only lateral margins, but also upper and lower margins. A genuine example of reference to laterality in the written textual space occurs in business correspondence, where we can find a right and a left undersigned party.

4. Dynamics of the metatextual space The metaphors of the geometry, structure, and stratification of the text convey a static picture of the conceptual space. However, in the course of its textual development, the conceptual and dialogical spaces do not remain static, but develop a dynamics which describes the changes and transformations of concepts and points of view.

4.1. Movement and flexibility Changes affecting the position and shape of the objects in space are a first source of metatextual dynamics. The conceptual objects are either moved around by a discoursive agent or they develop a dynamics of their own. The author may bring ideas to the fore or to the spotlight, defer, postpone, or set them aside. Ideas begin to move on their own when they emerge, come to the fore, stand aside, get lost, or vanish. In addition to this positional dynamics, the objects of textual space also have a dynamics which affects their own shape. Ideas are flexible because their importance can grow, increase, shrink (in our eyes), diminish, or disappear altogether. The text as a whole may be lengthened, broadened, deepened, extended, expanded, blown up, or shortened, reduced, or become superficial.

The (meta-)textual space

607

4.2. Movement in monological and dialogical space In the dialogical space (see 2.1.), the discoursive loci, standpoints, and perspectives are determined either with reference to the speaker only or relationally with reference to his or her hearer or to a conceptual object pursued in the speaker's environment. The discoursive agent moves through the dialogical space on his or her own on the ground, as a wanderer along ways, paths, trails, traces, tracks, at places, on or behind scenes, as a pilot rising to lofty thoughts in flights of fancy, or as a navigator or swimmer in deep water, where he or she might get drowned, emerge again, or steer a given course. By land, the traveler through textual space may proceed step by step, hurry up, go on, wait, rest, or stop. The agent's (A's) movements are described with respect to the interlocutor (B) when A directs B's attention to a point, guides B, approaches, follows, or attacks B. A's travels in metatextual space are seen in relation to concepts when A follows up ideas, turns to or away from considerations, tackles problems, reaches conclusions or comes back to an argument. In most of these situations, the concept is the goal of A's travels. This goal may be reached intentionally or found by chance. The traveler's movements may be free or forced. They are forced when A is attacked and therefore has to evade Β or when unwanted detours become necessary because of various hindrances in textual space.

4.3. Impediments and Boundaries Concepts and positions which impede, prevent, or stop progress in the course of the journey through textual space may have the form of lines, surfaces, or bodies. They are objects which we encounter as obstacles in our way, or they are mere optical obstructions of visibility, such as obscurities or smoke screens, which prevent insights into a matter, or covers that cover up an affair. Impediments associated with linear extent are experienced transversely as dividing lines between two areas, sagittally as distances between the discoursive agent and an interlocutor, or as the end point of lines. Dividing lines such as limits, limitations, borders, fences, or boundaries, prevent access from one conceptual area (to which we may have a claim) to another, but can also be overcome or bridged. However, since space is not experienced as limitless, there are also natural dividing lines beyond which we do not necessarily have to step (but may do so by curiosity), e.g., horizons, confines, or frontiers. End points of lines are equally ambivalent in their

608

Winfried Nöth

connotation. They may be goals which we want to reach, but as impediments they are dead ends which force us to stop. Distances between interlocutors are usually impediments to a mutual understanding, and so we want to reduce them, but they can also be intended when we want to distance ourselves from somebody. Examples of surfaces which impede the traveling in conceptual space are a rough terrain which is difficult to cross, a morass (e.g., of details) in which we seem to be bogged down, or a wild fire. Bodily impediments are objects such as stumbling blocks, obstacles which block our way, cliffs, mountains, piles (of difficulties), walls (e.g., those of silence or those to which we have our back), or doors (shut in someone's face). Impediments require a detour or have to be removed or cleared away if we want to reach our goal. Minor obstacles are removed laterally. They can be put, set, or brushed aside if they are not eliminated or left out. Major obstacles have to be circumvented, shipped, or got around. By metaphorical force or military action, argumentative obstacles are destroyed or dispelled. Most military metaphors describe dialogical space, where the interlocutors may appear as military opponents separated by such obstacles as lines of fire, trenches, or ambush traps, and move about in terms of attack or retreat (see also Lakoff and Johnson 1980: 4). The ideal of free and unhampered movement in conceptual and dialogical space is valid for rational or argumentative discourse. In the realm of fiction and poetry, however, the reader's mental movements may well be detained by force. The reader likes to be captivated, gripped, enthralled, or held spellbound by a novel. Such metaphors, which testify to the psychological proximity of eros and thanatos, have a parallel in the vocabulary of the emotions, where passions mean 'suffering' and 'fascination' at the same time.

5. Similarity and opposition in textual space Identity of, and similarity between, semantic domains is iconically mapped onto the metatextual space by means of two kinds of topographic relationships, homology and contiguity. Expressions, such as parallel thoughts, congruent ideas, or the development of ideas in concentric circles are examples of similarities represented by geometrical homology. Concepts near to, next to, in the neighborhood of, or in connection with each other exemplify the representation of similarity by means of geometrical contiguity. Semantic difference and opposition is expressed by a variety of spatial schemata (cf. Nöth 1996). The first of them is the schema of separation by boundaries or other impediments (see 4.3.), which may be unbridgeable or

The (meta-)textual space

609

insurmountable. A second schema is the one of distance (see 4.3.)· It represents concepts as extremely remote or diametrically opposed. The semantics of antonyms is localistically describable in terms of this model, since antonyms are characterized by a maximum distance along a scale, e.g. from hot to cold or small to big. Notice that the very concept of difference contains an etymological metaphor of spatial distancing (L. 'to carry apart'). A third schema is one whose etymological root is contained in the term 'opposition' itself, according to which opposites are 'set against each other'. It is the schema of confrontation (L. 'bringing face to face', similarly to affront) and conflict (L. 'striking together, fighting'), which shows opponents in a hostile face-to-face position. The schema of confrontation is not only restricted to the image of persons opposed in space, but applies also to objects. In this sense, the confrontational schema is apparent in the etymology and current meaning of against (OE 'in direct line/over against') and in the Latin prefixes contra ('against') and anti ('in front of/over against'). A fourth spatial model of opposition is the counterpart schema (cf. G. Gegenteil). It emphasizes the common semantic core between opposing concepts, which appear as the parts of a whole to which they can be united again. In semantics, opposites of this kind are called complementaries, a term which also contains a spatial etymology (from L. completus 'filled up'). In metatextual discourse, expressions such as on the one hand/side ... on the other hand/side exemplify the counterpart schema of opposition. A fifth schema of oppositional metaphors in discoursive space is the one of directional reversal. It is prototypically represented by expressions, such as the reverse (from L. 'to turn back'), the converse, or to turn something upside down (also topsy-turvy). However, the schema of directional reversal is not only restricted to the semantics of opposition. It is also present in textual references to returning to previously mentioned topics. These spatial schemata of opposition seem to confirm the hypothesis concerning the cognitive origin of linguistic opposites which Lyons (1977: 282) expressed as follows: It may well be that our understanding, not only of directional opposition, but of opposition in general, is based upon some kind of analogical extension of distinctions which we first learn to apply with respect to our own orientation and the location or locomotion of other objects in the external world.

610

Winfried Nöth

6. Cognitive motivation of metatextual spatial metaphors Cognitive semantics has emphasized the importance of spatial metaphors in all domains of everyday language. The origin of this spatial substratum of everyday semantics is in the biological importance of prelinguistic visual perception and orientation in our spatial environment (cf. Lakoff 1987: 269-292; Nöth 1994). Even the terminology of the linguists abounds with spatial metaphors so that linguistic studies of language referring to space run the risk of formulating tautologies. The meaning of abstract concepts expressed in the form of spatial metaphors is often defined by means of new metalinguistic spatial metaphors. Consider, e.g., the following definitions by Lakoff (1987: 283) (in which the metalinguistic spatial metaphors have been marked in italics): Radial structure in categories is understood in terms of CENTER-PERIPHERY schemas. Foreground-background structure is understood in terms of FRONTBACK schemas. Linear quantity scales are understood in terms of UP-DOWN schemas and LINEAR ORDER schemas.

In the course of our own inquiry into the metaphors of the textual space, we have come across various aspects of cognitive motivation. Among the principles of spatial cognition which are apparent in this metaphorical domain are the laws of visual perception in static space and of bodily orientation in dynamic space. In accordance with the process of focalization in visual perception, the center of textual attention is represented by (focal) points (which we may emphasize). Textual as well as perceptual attention is directed first at the center, the inside, the foreground, and the things in front of us. Things at the periphery, the sides, the outside, the background, and in the distance are topics of less interest. Also, the optical prerequisites of visual perception, such as light and acuity of vision, are a source of metatextual metaphors. The predominant metaphorical schema representing semantic development in discourse reflects the cognitive conditions of unhindered vision and movement in space. The system of coordinates in textual orientation is cognitively anthropocentric and differs from the laws of Euclidean geometry. Like the language of spatial orientation (cf. Clark and Clark 1978: 243), the metaphors of textual orientation are uneven in the values of their coordinates (anisotropic). The differences which we found in the textual frequencies and values of the dimensions of metatextual space reflect differences in our cognition of space. The fact that the lateral dimension is only weakly developed in textual space reflects the equal values of right and left in spatial orientation due to the lateral symmetry of our perceptual or-

The (meta-)textual space

611

gans of seeing and hearing. The predominance of verticality reflects the biological importance of verticality for orientation in upright gait. As to the axiology of textual verticality, Lakoffs (1987: 276) schema MORE IS UP, LESS IS DOWN is applicable, but only above ground. Under ground and under water, the values associated with up and down are reversed into their opposites (see 3.1.). Whereas above surface, the axiological scales of positivity go up, they go down below surface. This reversal even stresses the cognitive basis of vertical metaphorical axiology. In contrast to Euclidean geometry, where verticality extends homogeneously in one direction without a beginning or end, metaphorical verticality reflects an egocentric cognitive perspective which extends from a standpoint of the human being in the center of a vertical axis going bidirectionally up and down, but with an opposing axiology. However, motivations of the metaphors in metatextual space come not only from nature, i.e., the biology of cognition. Culture, too, provides spatial models of metatextual metaphors (cf. Nöth 1993). These are the ones derived from the spatial form of cultural artifacts such as houses or vehicles, and from the spatial form of sign systems such as painting and writing.

References Brünner, Gisela 1987 Metaphern für Sprache und Kommunikation im Alltag und Wissenschaft, Diskussion Deutsch 18: 100-119. Clark, Eve V. and Herbert H. Clark 1978 Universale, relativity, and language processing. In: Greenberg, Joseph H. (ed.). Universals of Human Language. Vol. 1. Stanford: University Press, 225-277. Derrida, Jacques 1967/ Of Grammatology. Baltimore: Johns Hopkins Univ. Press. 1976 Harweg, Roland 1990 Studien zur Deixis. Bochum: Brockmeyer. Lakoff, George 1987 Women, Fire, and Dangerous Things. Chicago: University Press. Lakoff, George and Mark Johnson 1980 Metaphors We Live By. Chicago: University Press. Lyons, John 1977 Semantics, 2 vols. Cambridge: University Press. Nöth, Winfried 1993 Iconicity of symmetries and asymmetries in syntactic coordination. In: Küper, Christoph (ed.). Von der Sprache zur Literatur: Motiviertheit im sprachlichen und im poetischen Kode. Tübingen: Stauffenburg, 23-36. 1994 Opposition at the roots of semiosis. In: Nöth, Winfried (ed.). Origins of Semiosis. Berlin: Mouton de Gruyter, 37-60. 1996 The semantic space of opposites: Cognitive and localist foundations. In: Yamanaka, Kei I. (ed.). A Festschrift for Yoshihiko Ikegami. Tokyo.

612

Winfried Nöth

Reddy, Michael J. 1979 The conduit metaphor: A case of frame conflict in our language about language. In: Ortony, Andrew (ed.)· Metaphor and Thought. Cambridge: University Press, 284-324. Saussure, Ferdinand de 1916/ Course in General Linguistics. New York: McGraw-Hill. 1959 Wunderlich, Dieter 1985 Raum, Zeit und das Lexikon. In: Schweizer, Harro (ed.). Sprache und Raum. Stuttgart: Metzler, 65-89.

Section 8 Abstract worlds as space

From one meaning to the next: The effects of polysemous relationships in lexical learning* Steven Frisson, Dominiek Sandra, Frank Brisard, and Hubert Cuyckens

1. Introduction The concept of polysemy has received considerable attention in cognitive linguistics (CL). Indeed, it is probably fair to say that linguists have never before been so preoccupied with this phenomenon. Furthermore, the degree of polysemy characterizing individual lexical items in CL analyses such as Brugman's (1981) pioneering work on the preposition over has not been witnessed in earlier linguistic work. The main purpose of CL analyses such as Brugman's is twofold: (i) make an inventory of all senses of a lexical item and (ii) determine the interrelatedness of the various senses, i.e. show how one sense is sanctioned by the existence of another. A major claim is that cognitive principles govern the process of meaning extension. In an attempt to model the product of their analyses, linguists have appealed to the concept of a network structure, whose architecture consists of two kinds of entities: nodes, which represent senses, and links or connections between nodes, which represent cognitive principles underlying semantic extension. The number of intervening links between any two nodes and the strength of these links are expected to account for the cognitive distance between two senses of a polysemoiis item. Evidently, linguists are free to choose their own object of study and representational device, and there are several reasons why the network structure used in CL may be expected to yield new insights. However, there is more to cognitive-linguistic networks than meets the eye. Although the purpose of network analyses in CL is essentially a linguistic one, i.e. the semantic analysis of a lexical item, the theoretical framework in which the enterprise is embedded is strongly committed to a specific view on the relationship between language and cognition. For one, the view that the primary purpose of semantic analysis is not the identification of a highly abstract meaning but the presentation of a mosaic of interrelated senses, differing in degrees of prototypicality, is

614

Steven Frisson et al.

the linguistic inheritance of work on human categorization in cognitive psychology (see, e.g., Lakoff 1987). Furthermore, the assumption that cognitive principles, such as metaphor and metonymy, determine the extension of meaning reflects the basic working hypothesis in CL, viz. that language is grounded in general cognition. Given this state of affairs, the polysemy issue inevitably raises questions of a psycholinguistic nature. Even though the cognitive linguist does not bear the burden of explicitly answering these questions, there is a responsibility for the paradigm as a whole to formulate explicit, falsifiable hypotheses regarding the psychological status of the postulated cognitive mechanisms. Many cognitive linguists have, at least in passing, commented on the psychological relevance of the principles involved, and some have even made explicit claims in this respect. For example, Langacker (1987: 56) proclaims that he takes seriously the goal of psychological reality in his own model of Cognitive Grammar: The description of a language is nevertheless a substantive hypothesis about its actual cognitive representation, and linguistic investigation is an empirical enterprise, its claims to be tested against the facts of cognitive structure.

However, once the focus is turned away from the strictly linguistic aspects of CL and cognitive-psychological implications are at issue, it becomes crucially important to spell out what the empirical predictions of the theory are. This linkage of linguistic concepts used in the network approach to psycholinguistic ones is a necessary first step in generating sensible research questions, as it seems hardly likely that there will be a complete convergence of the linguistic model and its concomitant psycholinguistic commitments. Indeed, when pursuing the goal of psychological reality "[t]he word 'goal' must be emphasized. It is not suggested that a strong claim of psychological reality can be made for any particular linguistic analysis as currently constituted" (Langacker 1987: 56). Given the distance between theory and empirical facts, many cognitive linguists retreat to the safe option of making no clear commitments of a psycholinguistic nature at all, or they make very vague and general claims. Thus, at the outset of a psycholinguistic approach to CL concepts, the empirically minded researcher trods on terra incognita, having recourse to a theory but not to its predictions. Nonetheless, the particular ways in which she/he translates the theoretical claims into psycholinguistic statements will determine which aspect of the theory is studied (i.e. is made subject to falsification) and what form the empirical research will take. In an attempt to get a handle on the psychologically relevant aspects of lexical networks in CL, Sandra and Rice (1995) recently proposed to make

Polysemy in lexical learning

615

a distinction between predictions at the level of psychological processes and predictions at the level of psychological structures (representations). In the context of lexical polysemy, a focus on the process side would amount to examining the categorization process (or set of processes) that enables the language user to extend a sense of a lexical item to new referential situations. On the other hand, a representation-based approach would be concerned with the mental representational structures that language users develop for dealing with polysemy. In the latter approach, questions can be raised at various levels of specificity. At one extreme, labeled the "isomorphic hypothesis" by Sandra and Rice, networks in CL can be considered attempts to model the contents and the structure of the language user's mental lexicon (what is in the network is also in the mind). At the other extreme, the relevant properties at the level of mental representations are restricted to the basic ingredients of the networks, e.g. the fact that discrete or local semantic representations can be suggested, the adoption of a level of very fine-grained distinctions, the use of schemas and superschemas in some models, etc. In the research reported below, we decided to direct our attention to the extension processes that are assumed in network analyses of polysemy, rather than focus on aspects of mental representation (see Sandra and Rice, 1995 for an experimental study bearing on the representational question). Our choice was inspired by the consideration that extension processes are at the very heart of the network model, making up the mechanism behind the growth of a lexical category. If no unambiguous claims are made within CL with regard to the status of mental representations, at least there exists no doubt in the community as to the cognitive relevance of extension processes. Independently motivated diachronic considerations, 1 combined with psychological research on the model of semantic networks (e.g. Woods 1975, Collins and Loftus 1975) and on prototype theory (Rosch 1978), have resulted in widely accepted and no longer controversial claims regarding the general functioning of extension nodes and their accompanying links in semantic networks. The insight emerging from this research is that semantic extension is obviously not the result of a random process. Two important questions should be raised in this respect: What is the nature of the processes involved? What function do they serve in language (what are they good for anyway)? We decided to tackle the latter question. The fact that meaning extension occurs so often seems to suggest that it serves an important purpose in language use. There must be a reason why language users prefer to make a "flexible" (i.e. extended) use of a lexical item rather than invent an entirely new one. We subscribe to the view that semantic extension is favored because it is easier, more natural, and

616

Steven Frisson et al.

cheaper in terms of the amount of required cognitive effort (cf. Taylor 1989). This should come as no surprise to the psychologist. It is a variation on the well-known psychological theme that it is easier to process novel information when this can be linked to familiar (old) knowledge. This hypothesis makes sense from the perspective of both language production and reception. Speakers will find it easier to produce extended meanings and listeners will have no special problems understanding and learning them. We have now reached the point where a specific empirical prediction can be formulated with respect to extension processes in cognitive-linguistic networks. Language users should find that learning an extended meaning is easier than learning a meaning that is unrelated to a familiar one. In other words, one psycholinguistic function of extension processes (and, hence, of polysemy in language) would be to facilitate the acquisition of lexical categories. We are by no means the first ones to make this claim (Deane 1988: 353): One of the oldest observations about polysemy, dating back to Br6al (1896), is that the primary sense makes secondary senses easier to learn, use, and remember. Even though the relations in lexical polysemy are fairly weak, they will clearly have this effect (our emphasis).

As far as the present framework of research is concerned, the important point is that the concept of semantic extension can now be situated within the cognitive system of the language user. This enables us to put this proposal to the empirical test. According to the above quotation, relatedness between two senses of a single word should automatically give rise to facilitated learning. We will investigate this hypothesis by randomly taking two senses of a word's array of meanings and testing the effect of one sense on the other. When choosing these senses, we will not set out from an existing or, for that matter, presumed network analysis of the lexical item in question (i.e. a preposition, cf. 2.1.1.). In other words, we will not presuppose any structural characteristics of the preposition as far as degree of relatedness and exact linkages between the various extensions are concerned. Rather, in case such effects of relatedness should show from the empirical results of the experiments, we will, a posteriori, try to interpret them in terms of the notion "cognitive distance" and, thus, offer testable grounds that should allow cognitive linguists to come up with psychologically valid descriptions of the lexeme at issue. We want to emphasize that we do not intend this research to have any implications vis-ä-vis the psychological status of semantic networks in CL

Polysemy in lexical learning

617

that go beyond those outlined above. In particular, the experiments should not be interpreted as attempts to prove that polysemous lexical items are mentally represented as networks. The results are entirely neutral with respect to this issue, the focus being on the process aspect rather than the representational one. Nor is the goal an assessment of the validity of a particular network in CL or even of a network type. The only purpose is to investigate the hypothesis that extension processes are an inalienable aspect of human categorization, one that is motivated in part by its warranting high efficiency in lexical learning.

2. General description of the experiments

2.1. Method The prediction that knowledge of one sense of a polysemous item facilitates the acquisition of other senses was tested in the present series of experiments with reference to the learning of prepositions in a foreign language. In order to chart this learning process, we opted for a task where Dutch-speaking subjects had to gradually construe a preposition's senses, while being steered by feedback from the computer. We picked the word class of prepositions because - as polysemous lexical items par excellence they have been widely studied in CL. Finally, we chose to run these experiments in a foreign language learning situation because this is an obvious context for learning new words.

2.1.1. Materials Subjects learned the meaning of the "Scottish" preposition yeath. Yeath is actually a non-existing form whose meaning was inspired by that of the English preposition beyond, and was made to roughly correspond to it. (Subjects were evidently not informed about this.) The reason for substituting yeath for beyond was to avoid word-form recognition by those subjects that had already come across beyond and that had possibly built up a meaning representation of this preposition.2 The preposition beyond was chosen to underlie the semantics of yeath for this set of concept-learning experiments because beyond does not have an exact equivalent in Dutch. This is important since it eliminates the possibility of the escape hatch that subjects would find a direct counterpart for yeath in Dutch. Moreover, it is

618

Steven Frisson et al.

not a high-frequency preposition in English foreign language instruction, which considerably diminishes the chances of subjects recognizing its meaning.3 Three senses of the polysemous preposition yeath were singled out for inclusion in the experiments. Their description was inspired by dictionary definitions and Boers' (1994) analysis of beyond. (i)

The spatial sense (SP): - TR is located at the end of a mental path from an observer's point of view, past or via LM and to an area at the other or further side of LM;

e.g. What lies beyond the mountains ? - TR traverses a physical path from an observer's point of view, past or via LM and to an area at the other or further side of LM; e.g. You

(ii)

can find the station if you walk beyond the statue and take the first street on your left. The "exceeding" sense (EX): TR moves outside the confines or boundaries of a non-spatial category

LM; e.g. After 25 years, George is still a secretary; he has never tried to move beyond his secretarial status. (iii) The "out of reach" sense (OR): TR is located outside the reach of a non-spatial category LM; e.g. The

story was clearly beyond the children's comprehension; / don't think they understood anything of it. These senses interrelate in the following ways: (i)

SP and EX: in the dynamic interpretation of sensel, TR moves into a different part of space than the one to which LM belongs. Through the metaphor "CATEGORY = BOUNDED SPACE", a TR can also be seen as moving outside the category mentioned in LM. (ii) SP and OR: since in the SP sense, TR and LM are located in different areas of space, they are experienced as distant from each other. SP, then, relates to OR by the metaphor "INACCESSIBILITY = DISTANCE". (iii) EX and OR: both senses interrelate indirectly through their relation with the spatial sense. There is, however, also a great deal of direct semantic overlap between EX and OR, since something that has moved outside a category might also become inaccessible; in other words, inaccessibility between TR and the category LM implies a TR having exceeded the confines of LM (but not vice versa).

Polysemy in lexical learning

619

2.1.2. Design Each experiment comprised two or three experimental conditions and one control condition (condition 1 in all experiments). In the experimental conditions, subjects were presented with two blocks of 40 English 4 sentences, each instantiating a different sense of yeath. For each of these blocks, subjects then had to construe the sense of yeath from its occurrence in these sentences. More specifically, they had to judge, by answering "yes" or "no" for each sentence, whether the meaning of the preposition was compatible with the meaning of the rest of the sentence. For the first few trials of the first block, subjects could only guess, of course, given that they started the experiment with no knowledge of the preposition's semantic content. However, by means of the computer feedback they received after each sentence, telling them whether their answer was correct or not, they were able to gradually construe sensel of the preposition, which, in turn, enabled them to carry out the rest of their experimental task in block 1 with increasing success. In block 2, subjects had to learn a different, but related sense of the preposition from a new set of sentences instantiating this second sense. The task was the same as for block 1. The prediction was that subjects' knowledge of sensel would facilitate their learning of sense2. 5 This prediction was tested with different pairs of the three yeathsenses. Between the two blocks, there was — depending on the experimental condition — either an explicit message that "another but related sense" would follow ("+signal") or a short pause with no message ("-signal"). In the control condition, subjects did not study yeath in the first block but another prepositional form, viz. the so-called Scottish preposition goaf. The meaning of goaf is based on that of English since — as in Dedre returned to swimming yesterday for the first time goaf her car accident — and is as such unrelated to that of yeath. Accordingly, between the blocks, subjects were instructed that a completely new preposition would follow. The rationale for including this block of goaf-sentences was to ensure that, at the time of learning sense2, all subjects would be equated on (i) the amount of experience with the learning task and (ii) level of fatigue. Across conditions, each block of stimuli/sentences consisted of 60% correct examples and 40% incorrect examples. Since we intended to trace the learning process, we wanted to make the discrimination task sufficiently difficult. To that effect, the LMs in negative sentences were potentially compatible with the meaning of the preposition, as construed by the subjects, but were not compatible with the broader sentential meaning. The 6/4 ratio was opted for because (i) it is easier to learn the meaning of a word from correct examples, and (ii) incorrect examples inform subjects about what falls outside the meaning potential of a word without, however,

620

Steven Frisson et al.

clearly delineating its meaning. In addition, the 6/4 ratio should not be expected to bias subjects to positive responses. Here are a few examples of correct and incorrect examples: (i)

SP:

+ -

(ii) EX:

+ -

(iii) OR:

+ -

The farm is not on this side of the field, it is yeath the field. John had a pain right in the middle yeath his shoulders. The police made the driver stop because he had gone yeath the speed limit. Our teacher follows the textbook very strictly; his way of teaching really goes yeath the textbook. These rockets are so old that they are yeath all use now. After a couple of minutes the pilot managed to have the plane yeath control.

Finally, in order to avoid long series of identical responses, blocks were constructed in such a way that no more than three correct or incorrect sentences would follow each other consecutively. Translations of English words/expressions that could cause comprehension difficulties for subjects were provided between brackets.

2.1.3. Procedure Before the experiment actually started, subjects were instructed - both orally and in writing — about the general relevant aspects of the experimental procedure, i.e. that they would have to learn the meaning of one (or, alternatively, two) Scottish prepositions by discriminating between correct and incorrect examples and on the basis of computer feedback. During the experiment, subjects were sitting in front of a computer screen. After getting through their preliminary instructions, they were presented with the first block of 40 sentences. For each sentence, the computer gave subjects two choices: "A: correct" and "B: incorrect". Depending on whether subjects thought that the sense of yeath (or goaf), insofar as they had construed it up to then, was compatible or incompatible with the meaning of the sentence, they were invited to strike the A- or B-key on their keyboard. After each answer, subjects were given feedback: "Your answer was correct" or "Your answer was wrong". The original sentence, the subjects' answer, and the computer feedback remained on the screen until the

Polysemy in lexical learning

621

subjects pressed the space bar, at which time a new sentence appeared on the screen.

2.1.4. Subjects Subjects were last-year secondary school students (Experiments 1 and 3) or first-year Business undergraduates (Experiment 2). They were all intermediary speakers of English (as a foreign language); their mother tongue was Dutch. Each subject participated in one experiment only.

2.2. Statistical analysis The unit of analysis in all experiments will be the number of correct responses in a series of 13 consecutive trials. The motivation behind this is to minimize the chances that high performance would readily result from random responding. In order to be able to trace the evolution of subjects' performance across a block of sentences, three measuring points will be used in the analyses: trial 13 (setting off the initial trial series), trial 27 (middle trial series), and trial 40 (final trial series). A first step in each analysis will be to demonstrate that subjects had the opportunity to rely on the sense that they learned in block 1, by showing that they indeed learned sense 1. To that effect, t-tests will be used to compare subjects' performance in the initial and final trial series of block 1. Most importantly, we are interested in the effects of sense 1 on the learning of sense2. There are two ways of assessing these. One is to compare experimental subjects (with knowledge of sense 1) and control subjects (lacking this knowledge) on their performance in the initial trial series of block 2. The hypothesis would predict that experimental subjects perform better than subjects in the control group. Notice that caution is needed with this between-subjects comparison, as it was impossible (for lack of repeated availability of most subjects) to match the groups on variables that might also affect their performance (like learning speed, intelligence, etc.). The second way in which the effect of sense 1 on sense2 can be demonstrated is by comparing the performance in the final trial series of block 1 and the performance in the initial series of block 2. If subjects maintain or improve their performance level in the transition from block 1 to block 2, this would certainly mean that sensel facilitated the acquisition of sense 2 (in the absence of such an effect, one would expect an initial phase of random responses). In the opposite case, where there is a drop in

622

Steven Frisson et al.

subjects' performance, it would be difficult to decide whether the similarity was smaller than the dissimilarity (thus yielding an overall decrease in the performance level), or whether there was no effect of sense 1 on sense2 at all. In order to find out whether the effect of sense 1 on sense2 differs across conditions, we applied analyses of variance (ANOVAs) to the data. A final analysis (using ANOVAs again) will be performed to see whether the evolution of subjects' performance within block 2 is similar across conditions or dependent on sense 1.

3. Experiments

3.1. Experiment 1 We designed Experiment 1 with two concerns in mind. One was to find out whether learning facilitation is the automatic result of linguistic relatedness, in which case it would obtain for any two related senses. To this end, we decided to use the OR-sense of yeath in block 2, in order to see whether different facilitation effects are obtained following prior acquisition of the SP-sense, as opposed to the EX-sense. The second concern was to create optimal conditions for learning facilitation to occur. Subjects learning related senses of yeath were explicitly attended to this relationship by a message in between the blocks (saying that "another but related sense of yeath" would follow).

3.1.1. Materials and design The design of Experiment 1 is presented in Figure 1. In all three conditions, the OR-sense of yeath figured in block 2. In block 1, the control group (condition 1) learned a completely unrelated preposition (the socalled Scottish preposition goaf, corresponding in meaning to English since), one experimental group (condition 2) learned the SP-sense, and another experimental group (condition 3) the EX-sense.

Polysemy in lexical learning

Block 1

Block 2

Condition 1

GOAF (= since)

YEATH (out of reach)

Condition 2

YEATH (spatial)

+ signal

Condition 3

YEATH (exceeding)

+signal

Figure 1.

623

YEATH (out of reach)

YEATH (out of reach)

Design of Experiment 1 The word form used is shown in capitals, the sense can be found between brackets.

3.1.2. Subjects Fifty-two last-year secondary school students volunteered to participate in the experiment. They were randomly assigned to the three conditions: 17 subjects in condition 1, 20 subjects in condition 2, and 15 subjects in condition 3. The small differences in the sample sizes were unavoidable (this does not affect the analyses), as the context of a school environment did not allow a more careful grouping of subjects.

3.1.3. Results Figure 2 graphically represents the mean percentage of correct responses for each condition at various points in the experiment. A first observation is that subjects in each condition were quite capable of inducing the meaning of a preposition from a block of sentences, as indicated by the significant improvement in their performance between the initial and final trial series in block 1 (all ps < .005). Importantly, subjects in conditions S P / O R 6 and EX/OR achieved a high level of performance at the end of block 1 (around 80 percent). Under these conditions, it should be possible to measure facilitation from sensel to sense2.

624

- Steven Frisson et al.

100

t - S S — Condition 1 (-/OR)

95

— X — Condition 2 (SP/OR)

90

— · — C o n d i t i o n 3 (EX/OR)

13

27

Block 2 Figure 2.

Result of Experiment 1 13=results after trials 1-13 2 7 = results after trials 15-27 40=results after trials 29-40

In the initial series of trials of block 2, subjects in condition EX/OR did significantly better than those in the control condition -/OR (p < .05). Numerically, performance in condition SP/OR was slightly worse than in the control condition, but this difference was nonsignificant (p > .10). Between the end of block 1 and the beginning of block 2 (only for the conditions with related senses), subjects' performance improved in condition EX/OR and dropped in condition SP/OR. These opposite movements indicate that the SP- and EX-senses affect the learning of the ORsense in different ways. This is statistically supported by the outcome of an ANOVA for a 2 (conditions: EX/OR - SP/OR) χ 2 (measuring points: final trial series of block 1 - initial trial series of block 2) design, in which the interaction between condition and measuring point was significant (F [1,33] = 7.38, ρ < .05). Separate t-tests on the data for each condition indicated a significant drop in performance for condition SP/OR (p < .05) and a nonsignificant rise for condition EX/OR (p > .10).

Polysemy in lexical learning

625

In order to assess the evolution of the results within block 2, an ANOVA for a 3 (conditions) χ 2 (measuring points: initial - final trial series of block 2) design was applied to the data. The interaction between condition and measuring point, reflecting the effect of condition on performance evolution, was marginally significant (F [2,49] = 2.62, ρ < .10). This was essentially due to the unexpected evolution in condition EX/OR, where performance was significantly worse in the final trial series than in the initial one (p < .05), in contrast to the results of the other two conditions. Not surprisingly, two separate ANOVAs, comparing the atypical condition EX/OR to each other condition, both yielded a significant interaction between condition and measuring point (EX/OR versus -/OR: F [1,33] = 4.39, ρ < .05; EX/OR versus SP/OR: F [1,30] = 4.33, ρ < .05).

3.1.4. Discussion The main question is whether subjects learning the OR-sense of yeath are affected by their knowledge of another sense of the same preposition. The data provide a very definite affirmative answer to this question. Subjects in conditions EX/OR and SP/OR reached a virtually identical level of performance at the end of block 1, but nevertheless obtained very different levels of performance in the initial trial series of block 2 (a statistically significant interaction). This would indicate that their knowledge of sense 1 directly affected the acquisition of sense2. This conclusion is further supported by the finding, in condition EX/OR, that performance in the initial trial series of block 2 was slightly better than in the final series of block 1 (although the effect was nonsignificant). Here we find an unambiguous demonstration that subjects attempted to treat sense2 as an extension of sense1. The differences in the relatedness effects for conditions EX/OR and SP/OR clearly show that the amount of learning facilitation is not constant for any pair of related senses. The strength of the relationship (or, to use another metaphor, the cognitive distance between the senses involved) probably determines the magnitude of these effects. This would mean that the subjects saw the EX- and OR-senses of yeath as being closer to each other than the SP- and OR-senses. Linguistically speaking, this is quite reasonable, considering the fact that the OR-sense entails the EX-sense (adding the notion of inaccessibility to it). In other words, the present set of data can be made sense of in terms of the cognitive-linguistic notion of "cognitive distance". Although the observation of the differences in the relatedness effects is not problematic, interpreting them is another matter. As mentioned above,

626

Steven Frisson et al.

we find a clear case of facilitation in condition EX/OR, which suggests a strategy of polysemy-based learning. However, the effect in condition SP/OR is more difficult to interpret. Although the drop in performance suggests that sense 1 and sense2 were perceived as relatively dissimilar, it is unclear how sense2 was acquired: as a (remote) extension of sensel (the case of a large degree of dissimilarity, causing a decrease in performance) or as an unrelated sense (homonymy-based learning)? The data do not offer conclusive evidence for either view. One might want to interpret the performance drop between blocks 1 and 2 as an argument in support of the homonymy view. However, this proves to be an invalid conclusion, as the benefit resulting from semantic overlap between the two senses need not be so substantial as to produce high performance on sense2 almost instantaneously. Another argument in favor of homonymy takes the nonsignificant difference between the control condition and condition SP/OR in the initial series of trials in block 2. This, too, is a weak argument. The fact that we did not find any significant differences between these two conditions does not necessarily imply that subjects made use of the same learning strategies (the experiment may lack power to detect the expected difference), and the between-subjects comparison is dubious in light of possible mismatches of the conditions on learning-related variables. The evidence in support of polysemy-based learning is not compelling either. However, it is possible to take this polysemy perspective to account for the drop in performance between block 1 and block 2 and for the strange drop in performance within block 2 in condition EX/OR. As this would lead us into a number of subtleties, we will present these arguments in the general discussion. Since we found no clear evidence of learning facilitation in condition SP/OR, one may wonder whether this could be related to the shift from a spatial sense to a non-spatial (abstract) sense. Such a change was not involved in condition EX/OR. In the following two experiments, we will focus on this spatial/non-spatial contrast and try to create conditions that will allow us to observe clearer evidence for learning facilitation here as well. We set out from the finding that a number of factors in Experiment 1 might have affected the perception and/or use of semantic information in learning: (i) Cognitive distance: the distance between the SP- and OR-senses may be too great for subjects to perceive the similarities of both senses immediately; (ii) Between-blocks instruction: encouraging subjects to look for commonalities might induce unnatural learning strategies, which would cause interference, rather than enhance a natural tendency (maximization); (iii) Order of presentation: the situation where subjects learn a non-spatial sense after having learned a spatial one may allow for less facilitation than

Polysemy in lexical learning

627

the reverse set-up. Experiment 2 investigated the first two variables, while Experiment 3 examined the third.

3.2. Experiment 2

3.2.1. Materials and design The design is presented in Figure 3. In line with the conclusions of the previous experiment, only the effect of the spatial sense of yeath on a nonspatial sense was studied. The non-spatial sense here was the EX-sense, which does not include the notion of "inaccessibility" that is part of the ORsense and which, hence, may be more closely related to the SP-sense. In all experimental conditions, the SP-sense appeared in block 1 and the EX-sense in block 2. In the control condition, the unrelated preposition goaf was again presented in block 1.

Block 1

Block 2

Cond. 1

GOAF (= since)

-

—

W

YEATH (exceeding)

Cond. 2

YEATH (spatial)

- signal — —

W

YEATH (exceeding)

Cond. 3

YEATH (spatial)

+signal

^

YEATH (exceeding)

Cond. 4

GOAF (= YEATH, spatial)

(form=signal^

YEATH (exceeding)

Figure 3. Design of Experiment 2

628

Steven Frisson et al.

The nature of the instruction between the two blocks varied across conditions. In the control condition, subjects were simply informed that a completely new preposition would follow. The important experimental contrast concerns the condition in which subjects were encouraged to relate the EX-sense to the SP-sense (as in Experiment 1, " + signal") and the one where no such suggestion was given ("- signal"). In order to evaluate the strength of the semantic relationship, an additional condition was included in which any suggestion of similarity between the two senses was avoided by using a different prepositional form for each of them {goaf and yeath). If facilitation is obtained under these circumstances, this would mean that subjects automatically notice the possibility of a semantic relationship, even if the different word forms may suggest the opposite.

3.2.2. Subjects Sixty-eight subjects participated in the experiment. They were recruited among last-year secondary school students. There were 17 subjects in condition 1 (control), 18 subjects in condition 2 ( " + signal"), 17 subjects in condition 3 ("- signal"), and 16 subjects in condition 4 (form difference).

3.2.3. Results Figure 4 graphically represents the evolution of subjects' performance as a function of condition and time. Subjects were able to learn the SP-sense of yeath in each experimental condition, as evidenced by performance scores in the final trial series of block 1, which are higher than in the initial one (all ps < .005). They reached high performance levels (80-90 percent).

Polysemy in lexical learning

629

—53—Condition 1 (-/EX) —Ο—Condition 2 (SP/EX), no signal

-195

-

- Condition 3 (SP/EX), +signal

i

90 »t

Ii

/A / v ' y/ X •ty

,

i - -Conditior 4 (GOAF/EX)

%

Φ J, s

r

%

·

.·'

\\

.1 I

J1

.-•y^/ji \ \γ^Γ Λ

»

/

s

»

\

V •/

t

/

1 13

27

40

Block 1

13

27

40

Block 2

Figure 4. Results of Experiment 2

In the initial trial series of block 2, no condition was found to differ from the control condition (ps > .10), with the exception of the " + signal" condition, where subjects were instructed to relate the EX- and SP-senses (p < .05). In all three experimental conditions, there was a significant drop in performance between the final trial series of block 1 and the initial trial series of block 2 (all ps < .05). This drop was non-distinct across conditions, as evidenced by the nonsignificant interaction between condition and measuring point (F [2,49] = .84, p> .10) in an ANOVA for a 3 (conditions) χ 2 (measuring points) design. This interaction remained nonsignificant in a separate ANOVA applied to the data of any two conditions (all ps > .10).

630

Steven Frisson et cd.

Also, the learning pattern in block 2 proved to be essentially the same across conditions, including the control condition. This is attested by the nonsignificant interaction between condition and measuring point (F [3,65] = .65, ρ > .10) in an ANOVA for a 4 (conditions) χ 2 (measuring points) design.

3.2.4. Discussion The picture emerging from the experimental results closely resembles the outcome for condition SP/OR in Experiment 1. The use of the EX-sense of yeath and the application of different ways to guide subjects' learning strategies in block 2 failed to yield clear signs of facilitation. Note that we do not claim that there is no facilitation, but rather that it is difficult to distinguish between polysemy-based learning and homonymy-based learning on the basis of the available data. There are some suggestions in the data that a homonymy-based account might be entertained. One is that the drop in performance between blocks 1 and 2 did not differ significantly between conditions "+signal" and "signal" (where identity of form would suggest polysemy-based learning) on the one hand, and the experimental condition GOAF/EX (where non-identity of form would suggest homonymy-based learning) on the other. In other words, subjects in the former two conditions, who had reason to believe that they could benefit from sense 1, did not differ from those in the latter condition, who were not led to believe that sense2 could be treated as an extension of sensel. The second point is that a failure to find any facilitation at all relative to the control condition in the initial trial series of block 2 can be taken as a probabilistic argument in favor of homonymybased learning. It seems indeed rather improbable that control subjects are always better learners than subjects in the experimental conditions, and that, accordingly, the facilitation effect that we are looking for is systematically masked. However, as mentioned in the discussion of Experiment 1, the possibility of polysemy-based learning should not be rejected out of hand. The theoretical relevance of the experimental data deserves an in-depth analysis, which will be made in the general discussion.

3.3. Experiment 3 In this experiment, we investigated whether clear facilitation would occur if the order of presentation for the spatial and non-spatial senses was reversed

Polysemy in lexical learning

631

with respect to the previous experiments. In this experiment, the OR-sense or the EX-sense preceded the SP-sense.

3.3.1. Materials and design Figure 5 represents the experimental design. In all conditions, the SP-sense had to be learned in block 2. Sense 1 was the OR-sense in one experimental condition, the EX-sense in the other experimental condition, and the unrelated preposition goaf in the control condition.

Condition 1

Block 1

Block 2

GOAF (= since)

(spatial)

YEATH

Condition 2

YEATH

+ signal

YEATH

Condition 3

YEATH (spatial)

(out of reach)

+signal

(exceeding)

YEATH (spatial)

Figure 5. Design of Experiment 3

3.3.2. Subjects 54 last-year secondary school students volunteered for this experiment. There were 19 subjects in condition -/SP, 16 subjects in condition OR/SP, and 19 subjects in condition EX/SP.

3.3.3. Results Figure 6 graphically represents the results of the experiment. In all conditions, subjects' performance was significantly higher in the final trial series of block 1 than in the initial series (all ps < .005), i.e. subjects were actually learning sensel.

632

Steven Frisson et al.

1 —53— Condition 1 (-/SP) —·—Condition 2 (OR/SP) - - X - -Condition 3 (EX/SP) Ν/1 1 «\

/ tt /

t Φ0 f

/ f/

«/ κ

/

\

\

ι

' < «

/

•

/

γ

. . " f /

) c~

1 13

27

40

Block 1 Figure

13

27

40

Block 2

6. Results of Experiment 3

In the initial trial series of block 2, no significant differences were found between the three conditions (all ps > .10). In the two experimental conditions, there was a significant drop in performance between block 1 and block 2 (ps < .005). This effect did not differ between conditions (F [1,31] = 1.59, ρ > .10). The evolution of subjects' performance in block 2 was virtually the same across the three conditions, as evidenced by the nonsignificant interaction between condition and measuring point in an ANOVA for a 3 (conditions) χ 2 (measuring points) design (F [2,49] = .01, ρ > .10).

3.3.4. Discussion The pattern of results is highly similar to that of Experiment 2 and of condition SP/OR in Experiment 1: performance in the experimental conditions significantly decreases when going from sensel to sense2, and, compared to the control condition, it does not reach a higher level in the initial trial series of block 2. Changing the order of the spatial and nonspatial senses apparently does not affect the learning process in block 2.

Polysemy in lexical learning

633

As in Experiments 1 and 2, the theoretical significance of the drop in performance in between blocks 1 and 2 is not obvious, which makes it all the more difficult to discriminate between polysemy-based and homonymybased learning. The only argument that one might make in relation to the issue of facilitation builds on the fact that, once more, the experimental conditions failed to do better than the control condition after thirteen trials into block 2. These between-subjects effects (or the lack of them) recur with a remarkable consistency across all three experiments. If unsuccessful matching of subjects in the different conditions (on variables like speed of learning, intelligence, ...) entirely accounts for this repeated null-effect (in other words, if a facilitation effect is cancelled each time by bad matching), this would mean that chance has played a bad trick on us on each occasion. In other words, even though the between-subjects effects might be unreliable in themselves, probabilistic considerations would suggest their plausibility in the context of an interpretation based on polysemy. In the following general discussion we will analyze the psychological mechanisms that might be involved in depth. As will become clear, the interpretation of the data may require more subtleties to be taken into account than the simple polysemy/homonymy-dichotomy would suggest.

4. General discussion One of the basic findings of the reported experiments seems to be that the links that make up semantic networks have different status. Within CL, various scholars have pointed out the possibility of varying distances between senses, e.g. Langacker (1987: 181): "The relation associating two nodes represents an established transition between them, and a network defines a set of actual and potential transition chains. Distances can thus be computed for various paths through the network." Langacker further claims that different kinds of distance relations will affect the linguistic system. 7 As can be concluded from our experiments, they also affect the processes involved in actual language learning. Before we proceed with the discussion, it seems advisable to explicate the explanatory status of linguistic concepts in psycholinguistic experiments like these. In particular, we make use of terms such as "cognitive distance" and "similarity" when trying to characterize the exact nature of the apparent relationships that hold between various senses of the artificial yeathcategory. It is important to see that such terminology is not used to bias the reader in any way when it comes to interpreting the results of the experiments. 8 Rather, we would like to point out that these are the linguistic criteria that, in our view, should be called upon to account for the

634

Steven Frisson et dl.

effects observed. What we are looking for, concretely, is an experimentally justified and ultimately linguistic interpretation of the apparent distance that is perceived between the SP-sense and the other senses of yeath on the one hand, and the obvious underlying similarity judgments on the part of the subjects for the EX-sense and the OR-sense on the other. We argue for the plausibility of such an approach on the basis of the striking match between the concept of linguistic similarity and the psychological clustering that subjects have made in their evaluation of the stimuli, as reflected in the transfer effect. The implications of these effects for the postulation of the much more complex notion of semantic network in the learning process are certainly less straightforward, and, accordingly, their translation into linguistic terminology will prove more hazardous. For this reason, any conjectures with respect to the potential construction of a network during these experiments will not be addressed in this study. Alternatively, starting from the psychological effects only, as pure experimental data, would not suffice in the present context. We have to keep in mind that notions such as transfer and inhibition effect are to be interpreted, initially at least, as purely quantitative terms, indicating a rise or drop in performance level only. But these effects need to be interpreted by using linguistic concepts that function within a well-defined theoretical framework, i.e. CL. Moreover, if linguistic importance can be attributed to the results of the conducted experiments, this will necessarily have to be based on more than just quantitative factors such as clustering or level of performance. In light of these observations, the following discussion will be an attempt to take into consideration possible linguistic justifications for the phenomena observed, without however taking for granted what appear to be plausible linguistic constructions on the basis of linguistic intuitions alone. As such, the linguistic description of the different senses involved cannot be viewed as the starting point of our analysis. Although we should be careful not to make any claims that would go beyond the analysis under consideration, in general we might derive from the results the proposal that the establishment of a link seems to become more difficult when a shift from one sense to a weakly related one is involved. We make this claim on the basis of reported facilitation/inhibition effects and other experimental data that will be introduced in the following sections. Thus, making extensions from one sense to another in actual language use appears not as straightforward as might be suggested on the basis of, for instance, diachronic evidence alone (cf. Introduction). Moreover, if they ever were transparent to actual language users, this no longer needs to be the case now, and linguistic descriptions of networks may, in fact, be of little relevance to the contemporary language user.

Polysemy in lexical learning

635

4.1. Degrees of similarity By themselves, the available results only allow for a definite analysis of the apparent connections between different senses in the context of language learning, where some links, i.e. those involving extensions between distant senses, are found to be conceptually harder to make than others. In this light, it comes as no surprise that different extensions are more easily connected when closely related senses are involved in the process (cf. the higher level of performance for the shift from EX to OR in Experiment l ) . 9 In these cases, the subjects do not seem to be significantly inhibited in their attempts to extend the conception of one meaning directly to a second, closely related meaning. For both these senses, the relevant semantic configuration is seen as largely overlapping: the EX-sense can be seen as part of, and entailed by, the OR-sense, which adds the meaning of inaccessibility to the abstract notion of excess (see section 3). Especially the transfer effect for condition EX/OR in the transition stage of Experiment 1 leads us to believe that subjects are at least capable of perceiving the relevant links relatively early on. This would corroborate the view that meaning extension is the natural consequence of general principles of human categorization, which does not seem to operate solely on the basis of necessary and sufficient conditions. Conversely, when extensions between weakly related senses are involved, we find the opposite learning pattern. In the present experiments, no unambiguous explanation can be provided for the lack of transfer effects in the distant types of shift. However, there are at least a few alternatives that might be considered. First, the initial drop in performance level when subjects learn the meaning of a second block could point to an interpretation of homonymy on the part of the subjects, in which case they do not perceive any relation between the two extensions. If this is true, then the drop might be attributed to the fact that subjects are not biased by the form similarity between the stimuli and start learning the second sense independently, which would result in them having to start from zero as it were, relying on pure chance. The discussion in section 3 has already entertained this hypothesis for the various experiments, relying mainly on probabilistic arguments regarding the masking of possible transfer effects. On the basis of this interpretation, it seems impossible to determine when subjects try to relate the second meaning to the first in the case of weak extension links, or whether they do so at all. However, the following account takes for granted an attempt to relate, at least initially, both senses, given the fact that they have the same word form. Moreover, general strategies in experimental research would advise the avoidance of the straightforward adoption of the null hypothesis if additional, be it tentative

636

Steven Frisson et al.

evidence, can be adduced to confirm the opposite hypothesis. In our case, control and experimental conditions are expected to yield significantly different results, if not in level of performance, then perhaps in other data that could indicate differing strategies used by the subjects to obtain similar results. In this respect, two factors seem to rule out the possibility of the subjects' immediate adoption of the homonymy-based strategy. First, subjects in the experimental conditions that display a drop in performance level in the first stage of block 2 were explicitly told to look for links between the new sense and the one previously learned. It seems highly unlikely that this signal was not at all taken into consideration by the subjects, especially considering the improvement in condition EX/OR (Experiment 1), which clearly suggests the possibility of subjects taking this option in our experimental set-up. And second, in most cases, the initial level of performance in block 2 proves higher than that of the corresponding stage in the first block. Although these effects are not significant, the tendency proves interesting if we consider that the same effects are witnessed when the order of presentation of the senses involved is reversed. (Thus, we can at least say that the results under consideration are not significantly influenced by the difficulty of the concepts to be learned.) However, at this point, we do not want to imply that the willingness to look for links is taken for granted, ruling out the relevance of homonymy on this basis alone. Therefore, we should look for additional evidence in favor of the adoption of a polysemy-based strategy by the experimental subjects. The least we can say is that initial inhibition effects might as plausibly point to the possibility that subjects, prompted by the form similarity between the two items, start looking for a specific link. In cases where this link proves too difficult to actually use, a drop in performance will be attested. The scenario we would like to propose at this point allows for the fact that subjects do perceive similarities and that they indeed try to establish links, in any of the experiments. The initial drop in the results could then be ascribed to simple performance errors of overgeneralization during the first stages of the learning process, after which the second meaning should be learned adequately. The typical U-shaped curve in the graphs suggests such an approach, which would imply a normal development in the acquisition of the two senses as related items of one category (three-stage learning model). More specifically, after having construed a schematic representation for the sense that was to be learned in block 1 (stage 1), subjects make use of this schema to sanction new instantiations presented in the following block (stage 2). As the fit will, in practice, not turn out to be very satisfying, resulting in worse performance and corresponding drops in the results, subjects are forced to readjust their

Polysemy in lexical learning

637

initial assessment and sometimes, depending on the actual discrepancies involved, change their strategy altogether. In this third stage, the level of performance will rise again. Evidence for such an interpretation is to be found in Figure 2, where, as noted earlier, we can witness a facilitating effect, instead of the usual inhibition, in the transition from block 1 to block 2 for the condition confronted with a shift from EX to OR (Condition 3), followed by a drop in performance level. Although the initial rise represents an atypical result in this series of experiments, it could give us a clue as to which considerations lie behind the subjects' learning strategies. In this particular case, we can recognize the effects of schematization, like in the other experiments. However, since the subjects are dealing with a shift that is easier to process, due to the close similarities of both senses (EX and OR), the schematic representation that is assumed to cover both individual senses each time the experiment proceeds to a following block will actually yield a considerable amount of success in the early stages. That is to say, as both senses do not differ greatly in their meanings, the chances of overgeneralizing (or, in other words, of construing a schema that sanctions too many instances, including those that would fall outside the scope of the second sense presented) become smaller. 10 Eventually, subjects are, however, bound to notice subtle meaning nuances in the second sense, as they gradually become more familiar with the range of semantic possibilities. Once this stage has been reached, they have two theoretical options: either they perceive the relation of extension between the first and the second sense and learn the second meaning on the basis of this perception, or they decide to discard the possibility of a semantic relation between sense 1 and sense2 and opt for the treatment of sense2 in terms of homonymy, i.e. as a new meaning to be learned in isolation. In both cases, a new stage has to be posited in the learning process, and the change of strategy will automatically be translated in initial drops of performance, followed by the usual acquisition of the concept to be learned. Strangely enough, the subjects in condition EX/OR reveal a pattern of learning that is not at all compatible with that of a scenario based on homonymy. This implies that the theoretical possibilities presented above are not directly applicable to the concrete experimental data we are examining here. If we go back to an analysis of the successful evaluations of right and wrong stimuli, we notice that the drop in performance between runs 1 and 2 (block 2 ) for condition EX/OR can mainly be attributed to a significant set-back in the evaluation of right sentences (p < .005), whereas the subjects' performance for wrong sentences rises considerably. (These results are diametrically opposed to those witnessed for condition SP/OR, the case of pure and immediate overgeneralization.) This would indicate

638

Steven Frisson et al.

that subjects, faced with the effects of schematization, suddenly become much more restrictive in their assessment of the semantic domain to be learned, and start to wrongly exclude many examples of the type they would have sanctioned just before (a case of "undergeneralization"). Given the stagnation in the learning curve, subjects will necessarily notice that some right stimuli are still evaluated wrongly by them on the basis of the constructed schema. As a result, the drop in performance level is assumed to occur in a stage where a new node is built. This new node will probably be linked to the schematic representation used in block 1, if only because of the fact that subjects do not seem to have to start all over again: their evaluation of negative examples remains accurate, which would indicate that learners attribute at least part of what cannot belong to the new sense on the basis of the negative examples in block 1. It is clear that such a strategy has little to do with homonymy, where we would expect learners to be less extreme in their exclusion of positive input, as is the case for the control condition in Experiment 1: compared to condition EX/OR, subjects in the control condition perform significantly better for positive sentences. Learning a new item that is seemingly unrelated to prior knowledge involves a different pattern altogether than the one witnessed in condition EX/OR, in that subjects faced with homonymy will initially operate with a central meaning representation that is not very abstract. Construing a schema that is relatively close to its instantiations in block 2 allows for an exclusion of negative samples, on the basis of which learners can gradually work on an increasingly accurate representation. The accompanying slight drop for positive sentences, however, does not resemble the highly significant set-back for condition EX/OR, which is, as suggested before, no doubt due to the misleading schema subjects are using in the beginning of block 2.

4.2. Strategies of learning In general, then, a pattern of changing strategies can be postulated for the reported experiments. In the case of distant senses, the postulated meaning differences are greater, and thus the chances of overgeneralizing augment considerably. If this is true, we might expect to see immediate drops in performance as soon as subjects are confronted with block 2, a finding which seems to be confirmed by the results of the following experiments 2 and 3. In this respect, we can posit the validity of a similarity index, which would allow us to measure the degrees of similarity between senses based on the following ratio: SIMILARITY/[SIMILARITY + DISSIMILARITY], (The denominator in this ratio includes both SIMILARITY and DISSIMILARITY,

Polysemy in lexical learning

639

because we are especially interested in the proportional import of SIMILARITY, i.e. in the relation of similarity features vis-a-vis the whole of the features involved in the semantic description.) Quantification of the (dis-)similarities between various senses obviously has to be based on a certain amount of semantic features that are shared, or not shared, by the items in question. 11 Without, however, going into the details of how to calculate such a ratio and which features would be included, it might help us in attributing the differences between various shifts to the effects of a similarity threshold, whereby the shift between EX and OR would surpass this threshold and thus result in initial transfer effects. In other words, if the index indicates a high degree of similarity, transfer effects are expected at least in the transition between blocks. In the opposite case, inhibition effects will surface, probably resulting in a more rapid adoption of an alternative interpretation on the part of the learners. What happens after the initial transfer effect witnessed in the case of high similarity, where the results show a typical drop in performance, becomes more plausible if we attribute the ensuing U-curve to a change in strategy, whereby subjects abandon the abstract schema they have construed in favor of a separate meaning representation linked to the first sense (as prompted by the intervening signal and as suggested by the results for positive and negative sentences). Thus, the learning process presents itself as consisting of a highly intricate set of strategies, and it would be misguided to try to isolate one and attach an exaggerated amount of importance to it. At this point, we should ask ourselves exactly why it is that such a complex pattern of learning strategies should have to be adopted even by the experimental groups, when we expect these language users to simply follow the invitation offered in the experiment to "look for a link". The answer can be found in the assumption that the mere presence of motivated links of extension, and especially the knowledge thereof, does not automatically enable learners to predict what the target of the learning process will be. If this is true, the knowledge of general principles of extension, whether induced or explicitly given, can, in certain cases, be helpful for the language user, but it does not excuse her/him from figuring out which specific extensions are at work in individual categories. The fluctuation in the performance of the relevant conditions in these experiments reflect this process of sanctioning, which cannot just be seen as involving a steady, undifferentiated improvement in performance. Concretely, this means that a hypothesis which seems plausible on linguistic grounds, namely that semantic overlap creates facilitation of acquisition, would wrongly predict such transfer effects in all cases where this overlap is noticed by language learners. 12 In fact, given the possibility that subjects may have perceived similarities even in cases of considerable

640

Steven Frisson et al.

semantic distance, the corresponding facilitation can be neutralized by the attempt to look for that part of the meaning that is not yet known to subjects. In this respect, two different strategies can be posited. Either subjects are immediately aware of some schematic meaning representation that they can deduce from their knowledge of the first sense, in which case all they have to do is find out the complementary (or missing) part of the meaning for the second item. This would imply the presence of transfer effects in each case (even with distant relations between senses), even though these effects can be neutralized and even overridden by the inhibition effects that go with the cognitive effort needed to establish the new information for the second sense. Or subjects start learning the second sense on the basis of the (non-schematic) meaning representation for sense 1, which they then reduce to a more abstract representation that would fit the semantic characteristics of sense2 better. The latter strategy seems more plausible, given the typical patterns noted above for positive and negative stimuli, and considering the fact that subjects have no way of knowing beforehand what level might be appropriate for construing schematic representations. Thus, the similarity index posited earlier would, at least partially, be determined by two semantic components: the amount of overlap and mismatch between the meanings of two related nodes. This observation makes it all the more reasonable to suppose that even cases of weak semantic links between senses would produce facilitation in the learning process, as suggested by Deane (see Introduction). The low degree of similarity in these instances could then account for the initial drop in performance witnessed in cases of distant shift, as the apparent mismatch between the different senses would be so great as to result in a lower level of performance. In other words, even if subjects perceive the overlap between two distant senses, the effort involved in establishing the missing parts of the new meaning will prevent the effects of this perception from ever showing up in actual learning curves, viz. as transfer effects. All we can say at this point is that the empirical facts that can be observed only indicate net results of the highly intertwined processes involved in learning related senses, and that it is impossible for us to tease apart the different components and make any definite claims with respect to their individual contributions. The idea that there may be different components involved in the construction of a new sense seems highly plausible but cannot be verified directly. It receives circumstantial support in the changing patterns of learning during the process of acquisition.

Polysemy in lexical learning

641

4.3. Vagueness or polysemy? In some way or other, the strategies we presented all imply a careful appreciation of what seems to become one of the major concerns in psycholinguistic research within the context of CL, viz. the determination of various levels at which cognitive mechanisms and natural language might interact (cf. Sandra and Rice 1995; Gibbs forthcoming). It seems as if future experimental research could shed more light on the distinctions between these levels, and eventually we could even hope for conclusive evidence in favor of, or, for that matter, against, well-defined interpretations of the status of extension as to their structural and/or processual characteristics in language production. For instance, the subtle difference between perceiving a link and using it in actual linguistic behavior definitely calls for experimental corroboration. Contrary to this, it is impossible for psycholinguistics to come up with definite analyses in cases where theoretical linguistics cannot provide clear definitions of the phenomena under investigation. More concretely, the problems concerning the issue of polysemy vs. vagueness could prevent us from choosing one or the other interpretation in ambiguous situations. For example, in the light of current linguistic debates, the question whether the apparent link between EX and OR in Experiment 1 can be attributed to vagueness (i.e. the case where both meanings are not dissimilar enough to produce different nodes) or to "real" polysemy is still open to discussion. If the link does not represent a case of vagueness, a position we have adopted in our preceding analyses, we might ascribe psychological reality to it and thus argue for the emergence of a small polysemous network under these conditions. Moreover, this might allow us to extrapolate a definite functional relevance of the attested link for the process of lexical acquisition, both in the domain of child language and of foreign language learning. As for the opposite case, Tuggy (1993) has drawn attention to the possibility that vagueness, as it stands, is nothing but a variation on the theme of polysemy, in that both phenomena entail the apriori postulation of lexical representations in terms of networks. The case of vagueness, then, differs from that of polysemy only in the degree of mental activation that can be attributed to the various nodes involved. Whereas polysemy implies an awareness of the instantiating nodes at the local level, vague lexical items display a high entrenchment of the more schematic part of the network. Thus, vagueness, we might say, is based on the same principles of representation as is polysemy, the main difference being level of awareness and level of analysis. In other words, the smaller extension links between nodes become, the more chances there are of the language user treating the nodes involved as one item, thereby taking recourse to a more schematic level.

642

Steven Frisson et al.

In the context of the present experiments, this approach seems very useful. It would indicate a perceived link in any of the cases, whether instances of a shift between closely related or between distant senses. The process of schematization/overgeneralization noted above is, in fact, to be understood in terms of the construction of a schematic representation, whereby the level of local representations is abandoned and one general schema is supposed to sanction both nodes, which are to be seen as vague instantiations of that same schema. It is important to point out, however, that this process is not at all irreversible for language learners. Thus, even if the meaning differences between EX and OR are not highly pronounced, subjects do seem to perceive them when prompted to and even change an initially successful strategy, possibly based on an interpretation of vagueness, in favor of one where both senses should be seen as represented separately, linked or not. The drop in performance for condition EX/OR in Experiment 1, after an initial rise, offers some evidence for this hypothesis: if subjects were to treat the second OR-sense as part of a vague schema shared with the first EX-sense, why would it be, then, that performance drops again in the course of block 2, where we would expect, on the basis of a continued vagueness-interpretation, the performance level to keep rising or at least stabilize? In other words, if OR and EX can be comprised under the same vague schema, this would entail only one item to be learned in the experiment. This hypothesis seems to be confirmed for the transition stage at least, where the performance level for condition EX/OR keeps rising, as if the same lexical item were to be considered in both blocks. But the notion of vagueness alone can ultimately not account for the variation in performance level, throughout block 2, within the same condition. Thus, we must conclude that vagueness, although undoubtedly part of the picture in these learning experiments, does not provide the whole story, and that subjects, apparently taking recourse to a strategy based on vagueness initially, can decide to follow different scenarios on the basis of criteria such as formal similarity and experimental design.

4.4.

Conclusions

There does not seem to exist an outspoken theoretical consensus within the cognitive community as to which independent factors are to determine whether semantic relations can be categorized as vague or polysemous (cf. Geeraerts 1993), and variation effects of the kind in Experiment 1 cannot be interpreted unambiguously until at least some relevant criteria have been identified in the course of empirical linguistic research. In general, however, the role of psycholinguistics, as it appears from the present series

Polysemy in lexical learning

643

of experiments, can be a stimulating one, if only in terms of the formulation of possible problematic areas in contemporary linguistic theory. Above all, it is important to assess the value of the findings we have reported here. We have found clear evidence, for instance, for the claim that the concept of semantic distance can serve as a key notion in the field of lexical acquisition, even though it remains unclear at this point how this factor specifically interacts with other relevant processes that are considered as contributing to the establishment of semantic extensions. These processes minimally involve mechanisms that were discussed in this article under the general term of similarity, which would cover metaphor and metonymy, but also various image-schematic transformations. In other words, the cognitive effort required in processing the links between different senses increases when weak extensions are involved, which is reflected by the fact that subjects tend to change strategies more easily when confronted with a shift from the spatial SP-sense to either one of the EX/OR-cluster. Ultimately, the observation of unambiguous, significant results in future experiments can help us resolve the question of whether this shift reflects the immediate adoption of a homonymy-based strategy or, on the contrary, important modifications in the type of semantic relation that is implicitly assumed by the subjects. Finally, notions of vagueness, homonymy, and polysemy, although unclear in the present state of investigation, obviously play an important role in determining the exact status of the results and their implications. Here, both linguists and psycholinguists have a responsibility to clarify these and other theoretical concepts, in order to proceed with more detailed analyses and perhaps even engage in a more content-based approach towards the investigation of particular mental representations. At any rate, both processual and structural questions are faced with the preliminary difficulties of establishing the various levels at which network models of lexical processing can or should be assessed. Indeed, one of the most prominent difficulties encountered in our analyses consists in the unavoidable indeterminacy as to the exact nature of what it is to perceive a link. The question whether knowledge of a link implies successful application of this knowledge is crucial in any attempt to solve the ambiguities between homonymy-based, polysemy-based, and vaguenessbased learning. If all of this entails a project far more ambitious than the series of experiments reported here, we take it as significant that some apparently highly relevant and problematic questions in network theory have surfaced in the present study. The analysis has allowed us to locate and formulate important trouble spots, issues that call for clarification if any decisive answers are to be expected in future research.

644

Steven Frisson et al.

Notes *

The research for this publication was carried out in the context of a research program supported by a Belgian government grant (IUAP-II, contract number 27). The first and the third author are research fellows for the Belgian National Science Foundation (NFWO). The third author also received a fellowship from the Belgian American Educational Foundation. We would like to thank Pol Cuvelier (UFSIA), Dirk Geeraerts (K.U.Leuven), Ron Langacker (UCSD), Sally Rice (University of Alberta); Ann Cooreman, Kerry Kilborn, and Anthony Sanford (University of Glasgow) for fruitful discussions on this topic. We are grateful to the anonymous reviewer as well for useful suggestions in the development of some arguments in the text. The first author also thanks his students and the secondary school teachers for their help in carrying out the experiments. Requests for reprints should be sent to Steven Frisson, Department of Linguistics, University of Antwerp (UFSIA), Prinsstraat 13 (D 134), B-2000 Antwerp, Belgium. E-Mail: FLW.FRISSON. [email protected]

1.

2. 3. 4. 5.

It is a simple fact of the diachronic evolution of word meanings that novel senses emerge along the way and that extended senses are transparently related to previously existing ones. See, for example, Eve Sweetser's various studies (e.g. 1987, 1990) on the diachronic development of different grammatical categories, where she has insightfully analyzed the numerous ways in which such networks might evolve. Sweetser's important claims with reference to polysemy state that this phenomenon is typically the intermediate step in semantic change, such that the same regularities observable in polysemy can be extended to an analysis of semantic change. The main question in this respect is whether it can be assumed that these historical mechanisms can be elicited from the knowledge base of the synchronic speech community or not. We thank the members of the "Glasgow Brain and Language Initiative" for their suggestion to replace beyond by the non-existing form yeath. We found that Dutch-speaking first-year undergraduates, who had already had four or five years of instruction in English as a foreign language, had considerable difficulties producing meaningful sentences with beyond. Indeed, as subjects were told, if the whole sentence were in Scottish, they would not be able to understand a single word of it. Note that there is an important difference between the experiment and real life learning situations. Whereas children or, in general, foreign language learners are rarely confronted with inadequate contextualizations, the subjects in the present experiments are prompted to make valid inferences concerning the semantic make-up of a lexical item on the basis of both correct and incorrect input. Especially the frequent presentations of the incorrect instances (40% of all stimuli in one block) enhances the artificial nature of the learning processes evoked by these experiments. We have to keep this complicating factor in mind when discussing the results later on. On the other hand, the feedback procedure, as performed by the computer, does not necessarily present the same danger of denaturalizing the typical learning environment: parents and other pedagogues are known to apply, among others of course, similar techniques (i.e. explicit evaluation plus possible correction) throughout the various learning stages of the subject.

Polysemy in lexical learning

6. 7.

8.

9.

10.

11.

12.

Conditions will be identified by giving the abbreviations for both senses, whereby block 1 is followed by block 2. The exact computation of the distance between nodes of a network, as put forward in Langacker's theoretical discussion, remains to be done in concrete linguistic studies. Psychologically speaking, such a computation might involve a host of relevant factors, some of them probably not even noticed at this point. Psycholinguistic on-line experiments like the present ones are equally limited in this respect, in that the selection of stimuli, before running the actual trials, happens on the basis of intuitions similar to those employed in linguistic studies. However, the analysis of subjects' learning patterns, particularly with regard to their implicit evaluation of the similarities involved, might shed more light on the types of semantic overlap that seem to be relevant to language users. In fact, a series of alternative linguistic notions could serve as valid candidates for the interpretation of the observed effects. For instance, the notion of distance could be related to that of domain, in that meaning extensions within one domain should be stronger than links that cross domain boundaries. In the context of the present experiments, this claim might be relevant insofar as SP would be characterized with reference to the basic domain of space only, whereas EX and OR would both refer to a host of abstract and/or experiental (i.e. non-spatial) domains that are seen as closely related. Similar results have been obtained in two other experiments, not reported here. In these experiments, subjects that were faced with the same type of shift (EX/OR) showed considerable improvement when they could learn the OR-sense relying on already acquired knowledge of the strongly related EX-sense. For the other conditions, the number of correct answers in the subjects' evaluation of right and wrong sentences respectively is extremely meaningful in this respect. In the hypothetical case of overgeneralization, a true schema is construed in the learning process to cover as many instances as possible of nodes 1 and 2 at the same time. Given the fact that such a schema will necessarily result in erroneous decisions in the context of the second sense alone, we can expect the performance level for right stimuli to be significantly better than that for wrong stimuli. If subjects overgeneralize, this means that the range of semantic possibilities will include correct instances of the sense at issue, but also that the schema is abstract enough to sanction a whole set of instantiations that fall outside its meaning. This is exactly the case for condition SP/OR in Experiment 1: compared to condition EX/OR, subjects in SP/OR perform significantly worse for negative sentences during the transition phase (F [1,33] = 5.63, ρ < .05). The gestalt-like (or image-schematic) qualities of semantic representations would, in principle, not allow us to consider the semantic features that make up the meaning of a lexical item as readily analytically separable. As a working hypothesis, however, the real question is how far the adoption and application of such a ratio would take us in adequately matching linguistic semantic descriptions with the kind of meaning structures obtained through empirical results from experimental testing. Again, transfer effect is used here as a purely quantitative term, indicating a rise in level of performance. Moreover, the facilitation that can be deduced from these effects implies a comparison between block 1 and block 2 only, with a different concept to be learned in both cases. As such, it does not tell us anything about the relative ease of learning a

645

646

Steven Frisson et al.

concept in one condition as compared to other conditions.

References Boers, Frank 1994

Motivating Meaning Extensions Beyond Physical Space. [Ph.D. dissertation, University of Antwerp] Brugman, Claudia 1981 The story of OVER, [MA Thesis, uc, Berkeley; reproduced by the IULC] Collins, Allan M. and Elizabeth F. Loftus 1975 A spreading-activation theory of semantic processing. Psychological Review 82: 407-28. Deane, Paul 1988 Polysemy and cognition. Lingua 75: 325-361. Geeraerts, Dirk 1993 Vagueness's puzzles, polysemy's vagaries. Cognitive Linguistics 4: 223272. Gibbs, Raymond W. 1995 What's cognitive about cognitive linguistics? In: Casad, Eugene H. (ed.). Cognitive Linguistics in the Redwoods. The Expansion of a New Paradigm in Linguistics. Berlin/New York: Mouton de Gruyter, 27-53. Lakoff, George 1987 Women, Fire, and Dangerous Things: What Categories Reveal About the Mind. Chicago: Chicago University Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. Stanford: Stanford University Press. 1988 A usage-based model. In: Rudzka-Ostyn, Brygida (ed.). Topics in Cognitive Linguistics. Amsterdam: Benjamins, 127-161. Rosch, Eleanor 1978 Principles of categorization. In: Rosch, Eleanor and Barbara B. Lloyd (eds). Cognition and Categorization. Hillsdale, N.J.: Lawrence Erlbaum, 27-48. Sandra, Dominiek and Sally Rice 1995 Network analyses of prepositional meaning: Mirroring whose mind - the linguist's or the language user's? Cognitive Linguistics 6: 89-130. Sweetser, Eve 1987 Metaphorical models of thought and speech: A comparison of historical directions and metaphorical mappings in the two domains. In: Aske, Jon, Natasha Beery, Laura Michaelis, and Hana Filip (eds). Proceedings of the Thirteenth Annual Meeting of the Berkeley Linguistics Society. Berkeley: CA: Berkeley Linguistics Society, 389-405. 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Taylor, John. R. 1989 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Oxford University Press. Tuggy, David 1993 Ambiguity, polysemy, and vagueness. Cognitive Linguistics 4: 273-290.

Polysemy in lexical learning

Woods, W.A. 1975

647

What's in a link: Foundations for semantic networks. In: Bobrow, Daniel G. and Allan Collins (eds). Representations and Understanding: Studies in Cognitive Science. New York: Academic Press, 35-84.

Metaphorical scenarios of science Olaf Jäkel

1. Investigating science: the general framework According to the Cognitive Theory of Metaphor^, abstract domains tend to be metaphorically conceptualized in terms of more concrete domains of experience. Probably one of the most abstract domains imaginable is that of science. Are there metaphors to be found in discourse on science? What are the source domains they draw on? And are there systematic patterns of metaphorization, or do we have to content ourselves with noting random use of metaphors?

1.1. Onomasiological cognitive metaphorology In asking questions like those above, I take an onomasiological approach to cognitive metaphorology.2 I am not so much interested in the lexeme, but rather in the whole conceptual domain of science. My study shows that it is useful to look at the structure of this conceptual domain as being organized by a scenario3; after all, science is but one special kind of human activity. The "science-scenario" consists among others of the following elements: -

scientist: agent, engaged in scientific study nature·, the scientist's object of study method: the scientist's activity, approach to the object theories·, result of scientific activity scientific progress: decision/choice between competing theories

This generalized science-scenario includes all the ingredients necessary for a fully-fledged conceptualization of science. The clues to target domain structure it provides may serve as a heuristic means: following up this basic inventory of aspects helps us to investigate the most important conceptual metaphors of science.

650

Olaf Jäkel

1.2. A diachronic and cross-linguistic corpus-study Science differs from other highly abstract domains like mental activity or economy* in that it hardly serves as a topic of everyday discourse, and where it does, that discourse lacks detail and differentiation. But while this is only one reason for a study of metaphors of science to take recourse to the writings of experts in the field, this corpus-linguistic approach has some more advantages to offer. 5 Among these, the possibility of providing some "hard" quantitative results is most welcome: drawing on a corpus of complete texts, we can not only list some conceptual metaphors that are used, but also make well-founded judgments concerning the question which metaphors are central to a text, and what is more, which metaphors can definitely be excluded from the writings of a certain author. However, my main interest is finding a substantial answer to this question: do different theories of science utilize different conceptual metaphors? My investigation is based on the seminal books by six important philosophers of science. If one were to give a concise outline of the history of the philosophy of science, these six scholars would be the prime candidates to acknowledge.6 Starting with Aristotle as the philosopher who first brought up the topic of empirical science, my investigation next includes the opposite poles of Rene Descartes and Francis Bacon, who stand for the beginning of modern philosophy in the early 17th century, one representing the rationalist/"continental", the other representing the empiricist/"British" branch. With his Kritik der reinen Vernunft (1781/87), Immanuel Kant undertakes a synthesis of the two competing camps. My study winds up with two influential contemporary philosophers: Karl R. Popper and Thomas S. Kuhn. In total, the corpus of philosophical writing analyzed (see Figure 1) consists of more than 4000 pages. This choice of authors adds two final characteristics to my project, making it a diachronic and cross-linguistic study of the metaphorical conceptualization of science. The languages involved are Greek (Aristotle), Latin (Bacon), French (Descartes), German (Kant, Popper), and English (Popper, Kuhn). With a truly "western" concept like science, a cross-linguistic approach seems not only justified but profitable. And as for the diachronic character of the investigation, it will be shown in what follows that this aspect in particular yields the most interesting results.

Metaphorical scenarios of 'science'

651

Aristotle Metaphysics (4th cent.B.C.) Bacon Novum Organum (1620)

Descartes Discours de la Methode (1637) Kant Kritik der reinen Vernunft (1781/87) Popper Logik der Forschung (1935) Conjectures and Refutations (1963) Objective Knowledge (1972)

Kuhn The Structure of Scientific Revolutions (1962)

Figure 1. Texts investigated In the following main section (2.) of this paper I will go through the metaphorical scenarios of the six scholars one by one. In each subsection, exemplary evidence is analyzed, followed by a detailed summary of the scenario's metaphorical mappings and a short discussion of the points of focus ("highlighting and hiding") of the model in question.7 The concluding section (3.) resumes a broader perspective, presenting a number of general conclusions and results of the complete investigation. Finally (in 3.2.1.), I will emphasize the diachronic perspective by pointing out systematic connections as well as crucial differences between the metaphorical scenarios of the approaches to science investigated.

2. Metaphorical scenarios of science One more preliminary remark concerns the concept of science as the object of this study: all of the authors investigated discuss scientific activity in general. In doing so, like most theorists of science, they draw on the natural sciences, and physics in particular, as their model. My general statements as to their historical place and impact on the theory of science merely reflect philosophical commonplaces, which will be found in any up-to-date introduction to the topic. 8

652

Olaf Jäkel

2.1. Aristotle: Science as Vision Aristotle (384-322 B.C.) is not only the first serious philosopher of science, but also one of the first theoreticians of language. It is from his writings on rhetoric that we know of his contempt for metaphor in scientific or philosophical discourse. Thus it is no wonder that in his own philosophical exposition of the topic of science in his Metaphysics he attempts to steer clear of metaphorical language. Yet despite of these conscious efforts, scattered over the passages on science we find a number of linguistic expressions that together provide strong evidence for one major conceptual metaphor: SCIENCE IS VISION. 9 The following two examples are linguistic variations on the crucial Aristotelian term theöriä, 'contemplation, vision of essences', which derives from Greek theöretn, 'be a spectator, look at': 1 0 (1)

(2)

on μεν obv ή φυσική θεωρητική nc εστι 'physics is a contemplating/observing science' 3

Εστίν επιστήμη τις η θεωρείτο ον η ον. 'There is a science that contemplates/observes all there is, insofar as it exists'

In Aristotle's model, what scientists do is observe, contemplate, look, and watch. The Greek verbs used to speak about scientific activity all come from the source domain of VISION. Besides theorem, they include episkopein, skopein, and synidein, as the following examples show:

(3)

οϋδεμία γάρ των άλλων έπισκοπεϊ καθόλου περί του οντος rj ον 'none of the other sciences looks in general at all there is as it is'

(4)

προς δ'άλήθειαν Ικ των υπαρχόντων δέΐσκοπείν 'But what is true must be observed/watched from what is really given'

(5)

ουδενί συνέβη την τοιαυτην συνιδεΐν αιτιαν 'nobody came to overlook/take into view this kind of principle'

Metaphorical scenarios of 'science'

653

These and similar passages contain the only instances of metaphor in the texts investigated that are worth mentioning. The metaphorical scenario of science which emerges is a relatively simple one. The scientist appears as a mere spectator, watching and viewing nature, or "all there is". This conceptualization of science does not include any active contribution on the part of the scientist. Other important aspects are also hidden: the status of scientific results, the role that alternative theories play, or ways in which a cooperation between scientists could be conceived of are completely out of the metaphorical focus. And indeed, these questions were of no concern to Aristotle, and came up much later in history. In its infancy, the philosophy of science started out with a seemingly "naive" construal of science as a mere extension of everyday perception. The metaphorical scenario of SCIENCE AS VISION is just in line with this construal.

2.2. Descartes: Science as a Journey Besides figuring as the founding father of philosophical rationalism, Rene Descartes (1596-1650) has to be acknowledged as a mathematician and natural scientist himself, working in such fields as physics and physiology. As his main contribution to the theory of science he proposes a unified scientific method in his celebrated Discours de la Methode. Descartes' general outline of scientific investigation in that book heavily draws on one conceptual metaphor, namely: SCIENCE IS A JOURNEY. Many linguistic expressions all over the Discours exploit this central metaphor and its specific mapping options. Only a few can be given as examples in the following. (6)

ce que nous conduisons nos pensies par diverses voies ..., et ceux qui ne marchent que fort lentement. peuvent avancer beaucoup davantage, s 'ils suivent toujours le droit chemin. que ne font ceux qui courent. et qui s 'en iloignent. 'we steer our thoughts on different paths, and those who walk only very slowly can advance much more when they always follow the straight path, than those who run and stray from it'

Here we have an instance of the PATH schema, which supplies the powerful basis of Descartes' metaphorical model of science. In this example, different paths correspond to different lines of thought or investigation that can be followed. In order to achieve optimum progress, the rationalist recommends a slow pace on the straight path rather than haste and variations of direction. The goal of this journey lies in a better understanding, which

654

Olaf Jäkel

can be seen in the next example: (7)

chercher la vraie methode pour parvenir ä la connaissance de toutes les choses dont mon esprit serait capable. 'to search for the right method to reach an understanding of all things of which my mind was capable'

"Pathfinder" Descartes introduces his unified method as the best path to reach the scientific goal. It is worth mentioning that this schematic mapping is already present in the etymology of the central Cartesian term: methode < Greek methodos, "pursuit of knowledge, mode of investigation", from metä, 'with, after' + hodos, 'way'. The following examples enlarge on the theme of progress: (8)

m'avancer. autant que ie pourrais, en la connaissance de la νerite, suivant la methode ... 'to go ahead, as far as I could, in the knowledge of truth, following that method'

(9)

j'ai forme une methode, ... moyen d'augmenter par degris ma connaissance, et de l 'elever peu a peu au plus haut point. ... Ί have developed a method, a means to extend my knowledge by degrees, and to raise it step by step to the highest point'

The idea underlying (8) is for any scientist to follow the method-path and go ahead as far as possible. But the schematic conceptualization of progress is not simply one of forward movement, but at the same time of an upward direction, as example (9) shows. The combination of the PATH schema and the VERTICALITY schema with its axiological GOOD IS UP orientation11 supply the source of this metaphorical model of scientific progress. The final example introduces a yet faint notion of scientific cooperation: (10)

afin que les derniers commenfant ou les precedents auraient acheve. et... nous allassions torn ensemble beaucoup plus loin que chacun en particulier ne saurait faire. 'so that the last began where the predecessors had ended, and all of us together would go much further than each could have done on his own'

If the idea of the scientific journey is to move forward and upward as far as possible, and presupposing that even a scientist is only a mortal being

Metaphorical scenarios of 'science'

655

with a severely limited lifespan, optimum progress in science must be regarded as a supraindividual enterprise. Within the Cartesian JOURNEY metaphor, this aspect appears in the kind of "relay" image suggested in example (10). The metaphorical scenario of SCIENCE AS A JOURNEY evoked in the Discours can be summarized as follows: the scientist is a traveller on a path of scientific investigation. Though different methods might be chosen as various routes to follow, the philosopher of science appears as the pathfinder who once and for all has found the best method. This recommended method is a straight path which leads upward towards knowledge as the journey's goal. Scientists make best progress towards this goal when they advance slowly but steadily. Like slow-motion relay runners, newcomers take up the scientific journey where their predecessors ended their stretch. Obviously, this metaphorical JOURNEY scenario focuses on methodical aspects of scientific activity. Early modern science is presented as a linear, goal-oriented process. Some real activity is required on the part of the scientist, who also cooperates with others, if only to the limited extent of forming a succession. One aspect that this model of science does not include is the role of alternative theories. These would have been easy to integrate into the scenario, figuring either as different routes to the same goal, or as various modes of travel. But this question had just not come up as a topic in the 17th century philosophy of science. Yet a second blindspot of the Cartesian scenario is much more revealing, as it mirrors the rationalist orientation, or preoccupation, of its author: the model allows no room for nature as the object of scientific inquiry. This aspect of science is systematically hidden by the very logic of the JOURNEY scenario, with its focus on a clear method and straight progress. The goal strived for lies in true knowledge, which is metaphorically located in some mystical hights - a rationalist echo of Plato's eternal idea space. Looking for a scenario of science which does not shun nature, we will have to consider an empiricist model.

2.3. Bacon: Science as the coercion of nature Francis Bacon (1561-1626) is one of the initiators and leading proponents of what became known as the British school of philosophical empiricism. The philosopher and politician, to whom the slogan "Knowledge is Power" is attributed, propagated his visionary project of employing science in the improvement of man's earthly estate. Bacon's trailblazing theoretical contribution to the beginnings of modern science, sketched in his main work,

656

Olaf Jäkel

the Novum Organum, lies in the outline of a new method of scientific inquiry: experimental research. This revolutionary change to the philosophy of science is also reflected in the central conceptual metaphor utilized by Bacon: SCIENCE IS THE COERCION OF NATURE. A few linguistic examples must suffice to explain some of the startling mapping details. All through his exposition, Bacon employs forceful, even violent images in describing the object of scientific inquiry: (11)

naturae constrictae et vexatae: nempe, cum per artem et ministerium humanum de statu suo detruditur. atque premitur et fingitur. 'nature bound and tormented; which by human art and activity is urged from its state, and is oppressed and manipulated'

These brutal images are reminiscent of Inquisition and torture. Nature appears as the victim of aggressive coercion by the scientist, being tied to the rack, tormented, and oppressed. Before we come back to this personification, we will provide two examples to introduce an important image schematic component of Bacon's science scenario: the CONTAINER schema. (12)

hujus ipsius veri mundi naturam introspicere et velut dissecare 'to look into the nature of this real world and, as it were, dissect it'

(13)

ut vero ad interiora et remotiora naturae penetretur 'to penetrate into the interior and the remoter parts of nature'

Here nature is metaphorically endowed with the in/out-orientation of a container. The scientist's aim is to look (12) or intrude (13) into this container. Its outer surface must put up some resistance, which is the reason why the scientist has to actually force his way into the container of nature (11-13), e.g. by means of dissecting it (12). But why should he want to get at the interior parts of nature in the first place? The following examples provide the answer, with example (14) enlarging on the theme of laborious toil on the part of the scientist. (14)

quanto magis laborandum est in hoc altero, quod ... ex naturae visceribus extrahiturl 'how much more do we have to toil with that which is to be drawn from the innards of nature!'

Metaphorical scenarios of 'science'

(15)

esse quae 'that have

657

adhuc in naturae sinu multa excellentis usus recondita. ... tarnen adhuc inventa non sunt there are many excellent things hidden in nature's womb that not been found out yet'

The aim of scientific inquiry is knowledge, conceptualized here as valuables hidden inside of nature. These have to be found out (15), or more drastically, drawn out (14). At this stage we come back to the general personification of nature which runs through Bacon's exposition of what science is all about. The application of the CONTAINER schema described above has to be situated within the conceptualization of nature as a person. It is from the innards (14) of this person that the scientist means to snatch what he seeks. Therefore, it is this person that has to be dissected (12) and, as a prerequisite, tied (11) to the operating table. One further specification of Bacon's personification of nature remains to be acknowledged. It is not only the Latin gender that characterizes Bacon's concept of nature as female. In example (15) the "innards" of nature are specified as "nature's womb". In this light, even the "penetration" of example (13) looks suspicious. And if we recall that the victims of religious Inquisition were mostly women, and that in Bacon's day and age all scientists were unquestionably males, we might detect some nasty connotations of rape suppressed in the conceptual underworld of the metaphorical model of SCIENCE AS COERCION. However this may be, SCIENCE AS COERCION is but a convenient shorthand to refer to Bacon's complex metaphorical scenario of science, which can be summarized as follows: the scientist is a brutal offender, inquisitor, torturer, intruder, maybe even rapist. He victimizes nature, which is personified as female. Inside her she hides secret knowledge, conceptualized metaphorically as valuables which the scientist wants to snatch from her. Thus she becomes a victim of aggressive coercion. The scientist's activity is a laborious process, highly armed for the forceful intrusion into nature's innards. The focus of this metaphorical COERCION scenario is on methodological aspects of scientific activity, and of experimental research in particular. In this model, the acquisition of scientific knowledge does not come easy or without effort on the part of the scientist. On the contrary, it presents a real, tough problem, requiring power and ingenuity. One aspect that is systematically hidden by this model is the role of theories in the scientific process. But this is just in line with the materialist and inductive approach as exemplified by the empiricist Bacon in his metaphorical scenario of SCIENCE AS THE COERCION OF NATURE.

658

Olaf Jäkel

2.4. Kant: Science as the building of an edifice Immanuel Kant (1724-1804) is generally regarded as the most important philosopher of the Enlightenment. His influential Critical Philosophy is an attempt not merely to combine, but truly synthesize the conflicting schools of rationalism and empiricism. One of the crucial questions tackled in his epistemological masterpiece, the Kritik der reinen Vernunft (1781/87), concerns the topic of science: "How is pure science of nature possible?" Kant's metaphorical scenario of science as reconstructed from the Critique comprises two different conceptual metaphors which can be formulated a s SCIENCE IS A JOURNEY OF DISCOVERY a n d SCIENCE IS THE BUILDING OF AN EDIFICE. How these two are interrelated to form one coherent scenario

of science will be shown in the following. As the first metaphor represents a prerequisite to the second, we will now treat them in their natural order. Having been trained in the rationalist school, Kant makes use of the JOURNEY metaphor which, as we have already seen, was employed first by Descartes. Yet, writing one and a half centuries after his French predecessor, he can already look back on the historical development of various sciences, seeing them on the move: (16)

Mit der Naturwissenschaft ging es weit langsamer zu [als mit der Mathematik], bis sie den Heeresweg der Wissenschaft traf. 'Natural science was very much slower [than was mathematics] in entering upon the martial road of science'

No longer is it only a lonesome scientist who travels his methodological path, but whole branches of science are on their journey. The scientific path they strive for has been in use for quite a while and by larger parties, as the term "martial road" reveals. If scientists thus travel in troop strength, Kant, the theorist of science, pictures himself as their pathfinder. In the following example he describes what he hopes to achieve with his critical epistemology: (17)

... Kultur der Vernunft durch den sicheren Gang einer Wissenschaft überhaupt, in Vergleichung mit dem grundlosen Tappen und leichtsinnigen Herumstreifen derselben ohne Kritik. 'For not only will reason be enabled to follow the secure path of science, instead of, as hitherto, groping at random and carelessly wandering about, without self-criticism'

Metaphorical scenarios of 'science'

659

Here Kant recommends his critical approach to reach and follow the secure path of science, while precritical science is presented as aimlessly wandering around. The next example enlarges on Kant's self-projection as a pathfinder in the service of science, adding aspects of exploration and cartography. There is a funny side to this in view of the fact that the philosopher spent his whole life in the East Prussian town of Königsberg, and never travelled. 12 (18)

Wir haben jetzt das Land des reinen Verstandes nicht allein durchreiset, und jeden Teil davon sorgfältig in Augenschein genommen, sondern es auch durchmessen, und jedem Dinge auf demselben seine Stelle bestimmt. Dieses Land aber ist eine Insel. ... [außer der] es sonst überall keinen Boden gibt, auf dem wir uns anbauen könnten. 'We have now not merely explored the territory of pure understanding, and carefully surveyed every part of it, but have also measured its extent, and assigned to everything in it its rightful place. This land is an island, ... [beyond which] there will be no other territory upon which we can settle'

It is the realm of possible knowledge that is presented here as a geographical territory. Kant claims to have explored this "terra incognita" in his epistemology, surveying its details and measuring its extent. He comes to the conclusion that the realm of possible knowledge is only a limited area: an island, so to speak, within whose boundaries any scientific enterprise will be confined. In Kant's Critique, these boundaries are those of sensual experience, delimiting the area of possible scientific knowledge to that which can be known empirically. 13 With the final sentence of example (18) we have reached the crucial link joining the two main conceptual metaphors in Kant's scenario of science. The aim of the epistemologist's pathfinding exploration was to provide some firm territory on which science could settle. Thus the JOURNEY OF DISCOVERY with all its metaphorical details of colonizing activity actually serves as a prerequisite for what is to follow: the BUILDING OF THE SCIENTIFIC EDIFICE.

We see an instance of the BUILDING metaphor in Kant's criticism of unscientific speculation: (19)

[Die Vernunftspekulation versucht] ihr Gebäude so früh, wie möglich, fertig zu machen, und hintennach allererst zu untersuchen, ob auch der Grund dazu gut gelegt sei.

660

Olaf Jäkel

'[Speculation tries] to complete its edifice as speedily as may be, and only afterwards to enquire whether the foundations are reliable' The convincing logic of the source domain brings home the point of this warning: it is extremely unwise to erect a building without taking care of solid foundations beforehand. Such a building is doomed to fall down sooner or later. What is presupposed in example (19), and supplying the conceptual basis of the remaining examples, is the conventional metaphor THEORIES ARE BUILDINGS. Kant uses it in describing his own role as the theorist of science: (20)

[Die Erkenntniskritik beschäftigt sich] mit einer nicht so glänzenden, aber doch auch nicht verdienstlosen Arbeit. nämlich·. den Boden zu jenen majestätischen ... Gebäuden eben und baufest zu machen. in welchem sich allerlei Maulwurfsgänge einer vergeblich, aber mit guter Zuversicht, auf Schätze grabenden Vernunft vorfinden, und die jenes Bauwerk unsicher machen. '[Critical epistemology occupies itself] with a less resplendent, but still meritorious job, namely, to level the ground, and to render it ready for building ... edifices of these majestic dimensions. For this ground has been honeycombed by mole passages which reason, in its confident but fruitless search for hidden treasures, has dug in all directions, and which threaten the security ofthat edifice'

Here the critical epistemologist's job is that of an engineer or construction worker, levelling the ground to make a proper building site. This task may seem humble, yet it is indispensable to the security of the theoretical building that is going to be erected. Within the general framework of the BUILDING metaphor, those dangerous "mole passages" may be taken to refer to any unsystematic research as part of a mere empiricist approach which is "blind" to rational procedure. 14 The importance of systematicity is emphasized in the final examples, which see the theorist as an architect drawing up the plan of the scientific edifice: (21)

Nicht technisch ..., sondern architektonisch ... kann dasjenige entspringen, was wir Wissenschaft nennen. 'Now that which we call science ... is not formed in technical fashion ..., but in architectonic fashion'

Metaphorical scenarios of 'science'

661

The scientific building-project will not prosper if construction workers or technicians merely assemble materials. It is basic source domain knowledge that any more complex building requires an architect to direct and organize what is to be executed by the construction workers. Kant's idea is that of one unified scientific system, designed by the critical epistemologist, to be completed by the empirical sciences: (22)

Es ist schlimm, daß nur allererst, nachdem wir lange Zeit ... viele ... Erkenntnisse. als Bauzeug gesammelt, ja gar lange Zeiten hindurch sie technisch zusammengesetzt haben, es uns dann allererst möglich ist, ... ein Ganzes nach den Zwecken der Vernunft architektonisch zu entwerfen eine Architektonik alles menschlichen Wissens ..., die jetziger Zeit, da schon so viel Stoff gesammelt ist. oder aus Ruinen eingefallener alter Gebäude genommen werden kann, nicht allein möglich, sondern nicht einmal mehr so gar schwer sein würde. 'It is unfortunate that only after we have spent much time collecting knowledge as building materials in somewhat random fashion ..., and after we have, indeed, over a long period assembled the materials in a merely technical manner, does it first become possible for us ... to draw up a whole architectonically in accordance with the ends of reason ...: an architectonic of all human knowledge, which, at the present time, in view of the great amount of material that has been collected, or which can be obtained from the ruins of ancient edifices, is not only possible, but would not indeed be difficult'

In this example, positive knowledge is conceptualized as building blocks and materials for the edifice planned. This scientific building-project as envisioned by the philosopher can even reutilize knowledge in the form of building materials from the ruins of ancient theories, integrating them into its own system-design. To summarize our findings, Kant's metaphorical scenario of science comprises two distinguishable stages. In the first stage of SCIENCE AS A JOURNEY OF DISCOVERY, the sciences are conceptualized as troops on the move. While precritical science was wandering around aimlessly, the critical philosopher of science has found the right path to be followed. The final goal of this expedition is the solid ground of possible, empirical knowledge. This territory has been explored by the pioneering epistemologist, whose cartography shows it to be an island of limited extension. The second stage of SCIENCE AS THE BUILDING OF AN EDIFICE reveals the final reason for the preceding expedition. Safe territory on solid ground

662

Olaf Jäkel

is needed for the settlement that science aims for. Scientific theory systems are buildings constructed out of the building materials of knowledge. While unscientific speculation is criticized as haphazard building without careful inspection of the development site, empirical science is described as building on safe territory. The theorist marks out this suitable territory, which he clears of natural flora and fauna, finally levelling it to make a proper building site. Furthermore, in his role as architect he organizes the construction work of ordinary scientists by designing the systematic plan of the complete scientific building. Several aspects are highlighted by this complex scenario of science. Kant views science as on a par with general understanding insofar as it is basically a constructive activity. This is mirrored in his distinguished use and elaboration of the BUILDING metaphor, which emphasizes the constructive aspects of scientific work. While its architectural component focuses on the importance of a theoretical system as the result of scientific activity, the preparatory exploration of the building site draws attention to the crucial epistemological prerequisites of such an undertaking. Moreover, the BUILDING metaphor allows to distinguish sharply between well-founded empirical science and unscientific speculation, which is one of Kant's major concerns. One target domain aspect that is clearly hidden by Kant's metaphorical scenario concerns nature in its yet unrecognized state. The role that natural phenomena play in the process of science is systematically left out of the picture - a blindspot probably due to Kant's rationalist legacy. Another feature absent from Kant's scenario concerns competition between different scientific theories. The philosopher of the enlightenment still believed in one unified scientific system as mirrored in his model of SCIENCE AS THE BUILDING OF AN EDIFICE.

To find this more modern aspect of theoretical competition reflected in metaphorical scenarios of science we will have to move on to the twentieth century and the scenarios developed by Popper and Kuhn.

2.5.

Popper: Science as an armed struggle for the survival of the fittest theory

Karl R. Popper (1902-1994) is certainly one of the most influential philosophers in our century. He sees his own theory of science, called Critical Rationalism, in a Kantian tradition. It is characterized by the strong weight it attaches to scientific criticism, stating that scientific theories are nothing but hypothetical conjectures to be critically tested by empirical counterevidence. It is because of this anti-inductivist "falsification" that Popper's ap-

Metaphorical scenarios of 'science'

663

proach is also termed "Falsificationism". Its first outline in the Logik der Forschung (1935) is developed further in Conjectures and Refutations (1963) and Objective Knowledge (1972). 15 With these later works the physicist and philosopher initiated a powerful interdisciplinary movement, joining philosophers and scientists under the label of "Evolutionary Epistemology". The linguistic metaphors in Popper's writings unequivocally suggest a scenario of SCIENCE AS AN ARMED STRUGGLE FOR THE SURVIVAL OF THE FITTEST THEORY. The conceptual basis of this complex model lies in the general personification of theories: (23)

... I consider tentative theories ... as the most important citizens of the world of ideas.

Throughout Popper's work, theories and hypotheses are described as living beings, persons, even citizens in a world of ideas, as in (23). On the basis of this personification Popper utilizes the whole of Darwin's theory of evolution by means of natural selection as the source domain for his metaphorical conceptualization of science, which is exemplified in the following: (24)

... that is the natural selection of hypotheses: our knowledge consists, at every moment, of those hypotheses which have shown their (comparative) fitness by surviving so far in their struggle for existence·, a competitive struggle which eliminates those hypotheses which are unfit.

As living beings, different hypothetical theories are engaged in a competitive struggle for survival. In this fierce process of natural selection, unfit theories are eliminated, while only the fittest theories sustain. That this Darwinian scenario is enriched with the concept of WAR can be seen in the following examples, which also introduce definite roles for scientists: (25)

... those who defend the thesis against the attacks of their opponents ...

(26)

Today the view of physical science ... has won the battle without another shot being fired.

As presented in (25), any given theory will be critically attacked by one group of scientists while being defended by others. Thus, scientists are characterized as warriors fighting in theoretical battles (26) with weapons of

664

Olaf Jäkel

criticism. Though conceptually close to it, this is not just the conventional ARGUMENT IS WAR metaphor. The scientific warriors only enter the stage set by the primary scenario of personified theories competing in their Darwinian STRUGGLE FOR SURVIVAL. This secondary role of scientific criticism is made explicit in the next two examples: (27)

... the struggle for existence is made harder by the conscious and systematic criticism of our theories.

(28)

... scientific criticism often makes our theories perish in our stead, eliminating our mistaken beliefs ...

The struggle for existence gains additional fierceness by the introduction of the weaponry of scientific criticism (27). It kills theories unfit for survival, thus artificially supporting, or reinforcing, the "natural selection". While weaker theories are eliminated (28), the final purpose of this belligerent activity is to further scientific progress by selecting those theories with stronger explanatory power. Before we summarize the main aspects of this metaphorical scenario of science, there are some points of interest concerning Popper's use of metaphors. Though he calls himself a "critical rationalist" and locates his own approach in a Kantian tradition, he dissociates himself from some core tenets of the theories of both Descartes and Kant. Interestingly enough, he chooses their central metaphors as targets of his critical attack: (29)

If anyone should think of scientific method as a way which leads to success in science, he will be disappointed. There is no royal road to success.

(30)

... my thesis that the empirical basis of our theories is far from firm·, that it should be compared to a swamp rather than to solid ground.

While in the first (29) of these examples Popper's criticism aims at Descartes' METHOD PATH, the second (30) is an assault on Kant's SOLID GROUND AS THE BASIS OF THE THEORETICAL EDIFICE. In his a r g u m e n t ,

Popper lets himself in for the metaphorical scenarios of his famous predecessors, while at the same time denying each of them a vital component crucial for the logic of the respective models: neither is there the "royal road" to success in science that the rationalist Descartes promised to have found in his unified method, nor does the swampy empirical basis supply

Metaphorical scenarios of 'science'

665

any unquestionable "firm ground" for the theoretical settlement, as envisaged by Kant. If the positive status that Popper ascribes to metaphor in theoretical argument needs further corroboration, this is made explicit in the following quotation. At the end of what has to be regarded a key passage on his evolutionary model of science, the core of which was presented above as example (24), the philosopher declares: "This statement of the situation is meant to describe how knowledge really grows. It is not meant metaphorically, though of course it makes use of metaphors." (Popper 1972: 261). In marked contrast to his predecessors, Popper indicates his awareness of the metaphoricity of his expressions. 16 He insists that these metaphors are anything but defective or "unscientific" ways of speaking. Not only are they meaningful, but they have real analytic value in describing a complex topic such as science. Popper's metaphorical scenario of SCIENCE AS AN ARMED STRUGGLE FOR THE SURVIVAL OF THE FITTEST THEORY can be summarized as follows: theories are personified as rivals competing in a fierce struggle for survival. Scientists appear as warriors, escalating the Darwinian struggle and its "natural selection" to warlike dimensions. They defend some theories against their opponents, while critically attacking rival theories in an attempt to kill them. Methodical criticism is featured as weaponry in this fight. The scientific progress that the warriors hope to enhance consists in the elimination of weaker theories and the selection of those with stronger explanatory power. This model clearly focuses on scientific progress and the competition between theories. In the twentieth century, this has become a major issue for the theory of science. Any further progress of science requires the choice of one of several rivalling theories. Popper's scenario expresses a good deal of optimism that this progress can be rational. "Naturally", theories will differ in fitness, and "naturally", the fittest will survive even the most critical assaults. Another aspect which is highlighted here is the importance of methodical criticism in the process of choosing between theories. This criticism cannot be too vigorous, as it will only reveal weak spots "naturally" displayed by unfit theories, while "naturally" stronger theories will survive all attacks. Once more, it is the role of nature as the scientists' object of study that is hidden by Popper's metaphorical scenario - just as in the preceding rationalist tradition of Descartes and Kant. As was discussed above, the aspects of a unified method and an empirical basis of science are not just left out of the metaphorical focus, but their existence is explicitly negated by Popper. This might be interpreted as a metatheoretical application of his own scenario: while critically attacking weak spots in both the Cartesian and the Kan-

666

Olaf Jäkel

tian model, the philosopher hopes to contribute to the defence and survival o f his o w n theory as depicted in the metaphorical scenario of SCIENCE AS AN ARMED STRUGGLE FOR THE SURVIVAL OF THE FITTEST THEORY.

2.6.

Kuhn: Science as a religious war over the acceptance of a certain kind of game

What kind of metaphorical scenario could emerge if a theorist of science, faced with the same problem-situation of competing theories, wanted to refute Popper's optimistic view and characterize scientific progress as rather irrational? He might choose the approach analyzed in this section. Thomas S. Kuhn (*1922), American physicist, philosopher and historian of science, presents his view of science in his most important work, The Structure of Scientific Revolutions (1962). He distinguishes two kinds of phases in scientific development. Periods of "normal science", during which research is guided by a shared "paradigm", or model theory, alternate with "revolutionary" changes of paradigm. This twofold, dialectical characterization of scientific development is mirrored in Kuhn's rich metaphorical scenario, which conceptualizes SCIENCE AS A RELIGIOUS WAR OVER THE ACCEPTANCE OF A CERTAIN KIND

OF GAME. The domineering role that theories, under the name of "paradigms", play in this model is once more based on personification: (31)

They had ... achieved a paradigm that proved able to guide the whole group's research.

(32)

... the paradigm forces scientists to investigate some part of nature in detail ...

(33)

Work under the paradigm can be conducted in no other way, and to desert the paradigm is to cease practising the science it defines.

A paradigm is here depicted as a sort of political leader, guiding (31) a whole group of scientists, even exerting force on them (32). The whole scientific community works under this leadership in humble submission (33), and abandoning this service is regarded as an act of desertion. That this leadership has massive religious overtones is displayed in the following examples: (34)

Though they may begin to lose faith ..., they do not renounce the paradigm that has led them into crisis.

Metaphorical scenarios of 'science'

(35)

... the scientist who embraces a new paradigm ...

(36)

... their members' conversion to the new paradigm.

667

Scientists are described as religious believers who will not easily renounce their faith in the charismatic leader they once chose to follow (34). What they will do, once they have lost faith completely, is embrace (i.e. "start believing in") a new religious leader (35), so that the process of theory choice as made by members of a scientific community can be called 'conversion' (36). But instead of being a peaceful act of voluntary conversion, the change of paradigm can take on revolutionary and warlike shapes, as the following examples show: (37)

... during scientific revolutions, the periods when paradigms are first under attack ...

(38)

Revolutions close with a total victory for one of the two opposing camps.

(39)

... the triumph of a new paradigm over the old one.

If larger parts of a scientific community thus abandon their paradigm, the old theory will come under attack (37) until being swept away by a fullscale revolution. As in religious wars, there can be no lasting stalemate or peaceful agreement between the opposing camps, but only total victory for one of them (38), and triumph of the new leadership over the old one that has been defeated (39). But after the triumphant victory there comes a lasting period of peace. Scientists settle down to their unspectacular everyday work under the victorious paradigm. The final examples show that, in marked contrast to the violence of scientific revolutions, this work of "normal science" now is a rather playful activity: (40)

... a game played under one set of rules ...

(41)

Normal science, the puzzle-solving activity we have just examined

(42)

... an attempt to force nature into the preformed and relatively inflexible box that the paradigm supplies.

668

Olaf Jäkel

Scientific work during these peaceful periods is conceptualized as a game (40) whose rules are determined by the paradigmatic leadership. In particular, this game is one of puzzle-solving (41). Yet the activity of the scientific puzzle-solver does not always display mere skill and ingenuity. On the contrary, like impatient children with pieces of a difficult jigsawpuzzle, they resort to sheere violence, attempting to "force nature into the preformed and relatively inflexible box that the paradigm supplies" (42). To summarize: Kuhn's metaphorical scenario draws a picture of SCIENCE AS A RELIGIOUS WAR OVER THE ACCEPTANCE OF A CERTAIN KIND OF GAME. Theories (or "paradigms") are personified as leaders dominating

the work of a whole scientific community. In times of peace, scientific work is a puzzle-solving game, in which scientists attempt to force nature into prefabricated boxes, following the rules laid down by their theoretical leadership. This leadership is of a religious kind, and scientists are believers in the charismatic leaders of their respective communities. Members of different scientific communities may try to convert each other to embrace their faith. But from time to time there will be a scientific revolution, a period of religious war over the acceptance of one kind of game instead of the other, resulting in a total victory for one community and its leadership over the other. This scenario focuses on the extremely important role of paradigmatic theories which govern scientific activity in periods of "normal science". In relation to the domineering theory, the scientist's role is reduced to that of a dogmatic believer, fanatical follower, and obedient serviceman. Scientists simply cannot do without some kind of theory guiding their research of nature and determining the kind of game to be played. One theory is abandoned only to be replaced by a successor, with this change displaying many symptoms of uprising, violence, and war typical of a political REVOLUTION. In this scenario, scientists only appear as members of different scientific communities, an aspect which is also highlighted by the RELIGION metaphors. The whole metaphorical model casts grave doubt on the rationality of scientific progress. If the choice between competing theories is a mere power struggle between religious leaders and their fanatical followers, it is a process beyond rational control, and any real progress or improvement of the scientific enterprise is in great danger. Worse than this, even "normal science", if sketched as the rather childlike activity of playing GAMES, is something not to be taken very seriously. All these features of blatant irrationalism, dogmatism, arbitrariness, and relativity are expressed in Kuhn's rich metaphorical scenario as formulated in our shorthand: SCIENCE IS A RELIGIOUS WAR OVER THE ACCEPTANCE OF A CERTAIN KIND OF GAME.

Metaphorical scenarios of 'science'

669

3. Some general conclusions and results This final section summarizes the results of my investigation, moving from the apparently trivial (3.1.) to the more interesting patterns of metaphorical systematicity (3.2.), to conclude with a closer inspection of the diachronic perspective (3.2.1.).

3.1. Trivial matters? The first result seems hardly exciting: discourse on science is not without metaphors. The examples of the linguistic data provided above make this plain to see. Even though philosophers of science are very much concerned with their use of language, and some, like Kant in particular, are renowned not as florid writers but for a most prosaic style bordering on tediousness, large parts of the texts investigated abound with metaphors. Though this result may seem like stating the obvious, it clearly corroborates Lakoff and Johnson's observation that abstract domains like science tend to be metaphorically conceptualized in terms of more concrete experiences.17 As a second result it can be stated that there are various metaphorical conceptualizations of science. It is interesting to note that these results of our diachronic investigation are borne out by synchronic research: WolfAndreas Liebert (1995; i.p.) reports that contemporary scientists' discourse on their own research activity is highly metaphorical. And what is more, the metaphorical models used bear a striking resemblance to those described above. 18 Thus taken together, synchronic and diachronic data seem to confirm the view that science is conceptualized metaphorically in terms of a relatively small number of source domains; a set which, having been accumulated in the course of western history, is still in use today.

3.2. Systematic patterns Beyond those general findings, our linguistic investigation of six important philosophers of science reveals some systematic patterns in their use of metaphors. Disregarding for a moment all historical aspects and treating those theories as just six alternative views of science, we can make the following assessment: different theories of science typically utilize different metaphor izations. Although in most cases there are also divergent linguistic metaphors scattered across the texts, each writer clearly uses some prevailing system of conceptual metaphors that suggest themselves to account for the bulk of

670

Olaf Jäkel

linguistic instances. Figure 2 provides an overview of those central structural metaphors which characterize each author's metaphorical scenario of science.

Aristotle SCIENCE IS VISION

Descartes

Bacon

SCIENCE IS A JOURNEY

SCIENCE IS THE COERCION OF NATURE

Kant SCIENCE IS THE BUILDING OF AN EDIFICE

Popper

Kuhn

SCIENCE IS AN

SCIENCE IS A

ARMED STRUGGLE FOR THE SURVIVAL OF THE FITTEST THEORY

RELIGIOUS WAR OVER THE ACCEPTANCE OF A CERTAIN KIND OF GAME

Figure 2. Different metaphorizations used in different theories of science

Some more details of the metaphorical mappings are given in Table 1 summarizing the six metaphorical scenarios. For each author, I go through the general target domain aspects, listing the corresponding set of "donor" concepts from the source domain. Taken together, these mappings constitute the author's metaphorical science-scenario.19 This way, also some "gaps" in the scenarios are pointed out (indicated by O), which represent target domain aspects systematically hidden as a result of the focus established by each metaphorical model. Having thus outlined the metaphorical scenarios of science characteristic of the six authors, one restriction needs to be made: there are also largely unspecific metaphors. Instances of the general BUILDING metaphor, for example, can be found in almost all of the texts, with the exception of Aristotle's writings. Though we have seen the special use and elaboration of this metaphor typical of Kant's theory of science.

Metaphorical scenarios of 'science'

671

Table 1. Overview of the metaphorical scenarios of science investigated (O indicates a target domain aspect systematically hidden by the metaphorical scenario)

method

theories

scientific progress

0

0

SCIENCE

scientist

nature

Aristotle

VISION

passive spectator, observer

objects of 0 observation

Descartes

JOURNEY

traveller

0

straight path, 0 slow but steady movement

forward and upward movement

Bacon

COERCION offender, OF NATURE inquisitor, torturer, intruder

female, biding valuables; victim of coercion

coercion and dissection of nature

0

snatching more valuables from nature's innards

Kant

BUILDING pioneer, AN EDIFICE architect, construction worker

0

inspection of the development site, architectural design of the building

buildings

completion of the edifice

Popper

STRUGGLE FOR SURVIVAL

warrior

0

weaponry

rivals competing in the struggle for survival

elimination of weaker theories, evolutionary selection of the fittest

Kuhn

RELIGIOUS WAR& GAME

fanatical toy items to rules of the charismatic revolutionary believer, be forced puzzlereligious change of serviceman, into boxes solving game leaders leadership, puzzle-solver victory in a religious war

672

Olaf Jäkel

More systematicity is revealed once we introduce the diachronic perspective.

3.2.1. Conclusions drawn from the diachronic perspective The results of our investigation clearly refute one possible hypothesis: the history of the theory of science is not marked by a development from a fancy or flowery discourse on science rich in metaphors to a more literal and less metaphorical discourse. The most sparing use of metaphors is noted with Aristotle. There is no noticeable difference in the frequency of linguistic metaphors between the other five authors. But maybe the most interesting question to ask from a diachronic perspective is the following: is there a correlation between the historical development of the theory of science, and the metaphorical scenarios used by the authors? As the philosophies of science investigated are not just six alternative views of science, but form a succession of seminal approaches to the topic, the scope of our study allows us to give a cautious answer. Before we end with a somewhat general observation, it may be illuminating to draw some diachronic comparisons between historical predecessors and their modern successors (cf. the overview in section 3.2., and especially Table 1). Aristotle holds a comparatively naive view of scientific understanding: the world and all its phenomena are just there, waiting to be observed and described by the scientist. Accordingly, of all the scholars investigated, he uses the most passive metaphors of science, drawing on the source domain of VISION. All the later scenarios construe scientific work as a much more active process, with Bacon's empiricist COERCION model topping the others in matters of aggressive action. Descartes, as Bacon's rationalist counterpart, uses the much more harmless and civilized activity of TRAVELLING as his source domain. His JOURNEY scenario reflects his rationalist focus on the linearity and goal orientation of scientific activity. Kant views science as on a par with general understanding insofar as it is basically a constructive activity. This is mirrored in his distinguished use and elaboration of the BUILDING metaphor, which highlights the constructive aspects of scientific work. This conceptual metaphor of science would never have occurred to Aristotle with his passive construal of science. As mentioned above, Kant's philosophical origins in the rationalist school founded by Descartes show up in his use of the Cartesian PATH metaphor, which he does not simply adopt, but which he elaborates and integrates in his sophisticated EXPLORATION scenario.

Metaphorical scenarios of 'science'

673

The rationalist background may also account for the fact that nature as such has no major role to play in either Descartes', Kant's, or Popper's metaphorical scenarios of science. Compare this to the empiricist Bacon, the philosophical advocate of experimental research: his personification of nature herself and the ensuing COERCION scenario would be completely alien to Descartes and Kant as well as to Popper. If we compare Kant to his modern successors Popper and Kuhn, the main difference lies in his unshakable vision of one unified scientific system. Two centuries later, the existence of competing theories has become a commonplace within the philosophy of science. Thus, scientific theories themselves are personified in both Popper's and Kuhn's metaphorical scenarios, and competition between theories is conceptualized in terms of WAR. As the two approaches differ in their attitude towards the rationality of scientific progress, the rather pessimistic Kuhn draws a RELIGIOUS WAR BETWEEN FANATICS scenario, while the more optimistic Popper utilizes the whole evolutionary STRUGGLE FOR THE SURVIVAL OF THE FITTEST as source domain in his conceptualization of science. Endowed with the unbroken spirit of Enlightenment of his day and age, Kant would not have dreamt of personifying theories, or of scientific progress as any kind of war. So far we have just compared the six metaphorical scenarios of science to each other. It was pointed out how some of the systematic connections as well as some of the most crucial differences between the theoretical approaches to science are mirrored in the main conceptual metaphors used. We will end these conclusions with a cautious generalization: in the course of the historical development of the theory of science, we witness the gradual unfolding of the science-scenario. With each theorist, there is a shift of focus. The generalized science-scenario that was introduced in section 1.1. can be seen as a result of the historical development that is presented in Figure 3. If we go through the metaphorical scenarios for a last time (compare the overview in section 3.2.), we find that Aristotle's simple VISION scenario focuses on the object of observation. As simply looking at natural phenomena presents no real problem, this metaphorical model hides all methodical aspects of scientific activity.

674

Olaf Jäkel

Aristotle focus on object

Bacon

Descartes focus on method (process)

rational procedure

empirical experiment

Kant focus on theory (and its prerequisites) (product)

Popper

Kuhn focus on scientific progress (meta-process)

ideal/rational

historical/sociological

Figure 3. Unfolding of the science-scenario and shift of focus The problems of method become a central issue to the 17th century theorists. Both Descartes' and Bacon's scenarios highlight procedural aspects of scientific activity. While the Cartesian JOURNEY metaphor concentrates on a linear, goal-oriented rational procedure, Bacon's COERCION model focuses on the powerful intrusion upon nature by means of experimental research. Kant's 18th century scenario brings about yet another shift of focus towards the theory, or the theoretical system, as the product of scientific activity. Thus his ARCHITECTURAL model highlights the constructive aspects as well as the epistemological prerequisites of a well-founded system of scientific knowledge.

Metaphorical scenarios of 'science'

675

Finally, in our century, the philosophers' interest turns to the issue of scientific progress. Thus, Popper and Kuhn present alternative WAR models to conceptualize the choice between competing theories, which can be regarded as a sort of meta-process on top of the scientists' activity of research. Popper's EVOLUTIONARY STRUGGLE scenario expresses hope for ideal or at least rational progress, while the RELIGIOUS WAR scenario behind Kuhn's historical and sociological approach to science highlights the irrational and dogmatic aspects of scientific progress. These results of our investigation, yielded by the diachronic approach, seem to be the most illuminating ones. The diachronic perspective allowed us to establish a complex correlation between the historical development of the theory of science and the metaphorical scenarios used by the most influential theorists in this field. Once more, Cognitive Linguistics proves useful not only for an understanding of the nature of language, but also for an understanding of other human cognitive achievements, in this case for a better understanding of the essence of different philosophical approaches to the nature of science.

Notes

1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

I would like to thank Günter Radden, whose constructive criticism of an earlier draft helped shape this paper. Thanks are also due to Eier Steffens for his kind assistance with Greek, as well as to Regina Jäkel for some helpful counselling on French. Any remaining errors are mine, not theirs. Among the most important contributions to the Cognitive Theory of Metaphor are Lakoff and Johnson (1980), Lakoff (1987, 1990, and 1993), Lakoff and Turner (1989), Sweetser (1990), and Johnson (1987, 1992, and 1993). A more comprehensive account of what I call "onomasiological cognitive metaphorology (ONCOME)" will be provided in Jäkel (forthcoming). The term scenario is used to refer to those idealized cognitive models that display ordered sequences of events, e.g. human activities like 'eating at a restaurant'. See Lakoff (1987: 285f.). Cf. Jäkel (1993a and 1995). The textlinguistic approach will be presented and discussed in detail in Jäkel (forthcoming). See, for example, Detel (1991), or Seiffert and Radnitzky (eds.) (1992). The complete investigation will be presented in a case study on "science" as part of my projected dissertation (Jäkel, forthcoming). Cf. e.g. Detel (1991), or Urmson and R6e (eds) (1992). Following the convention, conceptual metaphors are printed in CAPITALS. In my translations of the Non-English examples in this paper I have tried to reproduce the original metaphors as faithftilly as possible. Cf. Lakoff and Johnson (1980: 16-18). In my opinion, too much has been made of this discrepancy by authors who approach Kant's philosophy in a "psychoanalytic" manner. To name but one example, Böhme and Böhme (1985: 427-95) use these passages to

676

13.

14.

15.

16. 17.

18.

Olaf Jäkel

reproach the philosopher with compensatory boasting to make up for his suppressed narcissism. In the context of examples like (18) it seems much more apt to keep in mind that Kant was a keen and well-informed observer of the geographical journeys of discovery and exploration taking place in his time, like e.g. James Cook's voyages exploring the "terra australis". In a number of examples, which cannot be quoted due to lack of space, Kant metaphorizes unscientific speculation as vain attempts at transgressing those boundaries and leaving the firm ground of the island, either by boat or by flying. It is interesting to note that not only unsystematic research, but everything bad and detestable in science, like errors, unjustified assumptions, ideology, and dogmatism, is conceptualized by Kant in terms of naturalorganic entities. As wild plants growing rampant or hideous animals' nests, these are obstacles in the way of the scientific expedition and a hindrance to the building of the scientific edifice. Thus, the theoretical pathfinder has to do some real pioneering work, clearing the path as well as the building site of these obstacles. In this paper I abstain from quoting any examples from the German Logik der Forschung, although they clearly corroborate the conceptual metaphors presented here on the basis of linguistic evidence taken from Popper's English works. We find no such indication in the works of Descartes, Bacon, and Kant. Aristotle's contempt for metaphor in theoretical discourse was mentioned above. Cf. Lakoff and Johnson (1980: 105 and 109), Lakoff and Turner (1989: 132), Sweetser (1990: 30), and Lakoff (1993: 245). For a more comprehensive treatment of the unidirectionality hypothesis see Jäkel (1993b) and (forthcoming). The main source domains Liebert (i.p.) reports on are VISION/LIGHT, PATH/JOURNEY,

19.

BUILDING, MOSAIC/PUZZLE,

and

COMMUNICATION/ON/

INTERROGATION. While the latter may be seen as a moderated version of Bacon's COERCION, WAR metaphors are probably missing from the corpus of discussions between a group of virologists simply because competition and choice between different theories was not among the issues discussed. The head of the Table contains the main target domain aspects, which can be linked to their respective "donor" concepts from the source domain, thus spelling out the metaphorical cross-domain correspondences. I do not distinguish between "ontological" and "epistemic" correspondences (cf. Lakoff 1987: 387), as I find this distinction blurred and not very revealing.

Appendix: Sources The source of each example presented in this paper is provided by giving the page number of the respective works in the editions indicated. In the case of Aristotle and Kant, though, following the general convention, references are made to the pages of the original editions.

Metaphorical scenarios of 'science'

677

Aristoteles, Metaphysik. Hamburg (1989): Meiner (I) 1026a (2) 1003a (3) 1003a (5) 984b Aristoteles, Zweite Analytiken. Amsterdam (1987): Rodopi (4) 81b Descartes, Ren6 (1637) Discours de la Methode. Hamburg (1969): Meiner (6) 2f (7) 28 (8) 44 (9) 4 (10) 102 Bacon, Francis (1620) Novum Organum. Hamburg (1990): Meiner (II) 54 (12) 52 (13) 88 (14)44 (15)230 Kant, Immanuel (1781/87) Kritik der reinen Vernunft. Hamburg (1986): Meiner (16) Β XII (17) Β XXXf (18) Β 294f (19) Β 9 (20) Β 375f (21) Β 861 (22) Β 862f Popper, Karl R. (1972) Objective Knowledge. Oxford (1989): Clarendon Press (23) 300 (24) 261 (27) 261 (28) 261 (29) 265 Popper, Karl R. (1963) Conjectures and Rotations. Routledge (25) 315 (26) 99 (30) 387

London/New York (1989):

Kuhn, Thomas S. (1962) The Structure of Scientific Revolutions. Chicago/London (1987): University of Chicago Press (31) 22 (32) 24 (33) 34 (34) 77 (35) 122 (36) 19 (37) 48 (38) 166 (39) 147 (40)52 (41) 52 (42) 24

References Böhme, Hartmut and Böhme, Gernot 1985 Das Andere der Vernunft: Zur Entwicklung von Rationalitätsstrukturen am Beispiel Kants. Frankfurt a.M.: Suhrkamp. Detel, Wolfgang 1991 Wissenschaft. In: Ekkehard Martens and Herbert Schnädelbach (eds). Philosophie: Ein Grundkurs. Reinbek b. Hamburg: Rowohlt, 172-216. Jäkel, Olaf 1993a 'Economic growth' versus 'pushing up the GNP': Metaphors of quantity from the economic domain. L.A.U.D. 1993, (C:24) University of Duisburg. 1993b Is metaphor really a one-way street? One of the basic tenets of the cognitive theory of metaphor put to the test. C.L.E.A.R./ Cognitive Linguistics: Explorations, Applications, Research. No 1. University of Hamburg.

678

Olaf Jäkel

Jäkel, Olaf 1995

forthc. Johnson, Mark 1987 1992 1993

The metaphorical concept of mind: MENTAL ACTIVITY IS MANIPULATION. In: Taylor, John R., Robert E. MacLaury (eds) Language and the Cognitive Construal of the World. Berlin/New York: Mouton de Gruyter, 197-229. [Prepublished by L.A.U.D. 1993, (A: 333) University of Duisburg], Konzeptuelle Metaphern und die schematische Fundierung komplexer kognitiver Modelle. Ph.D. dissertation, University of Hamburg. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago/London: University of Chicago Press. Philosophical implications of cognitive semantics. Cognitive Linguistics 3-4: 345-66. Moral Imagination: Implications of Cognitive Science for Ethics. Chicago/London: University of Chicago Press.

Lakoff, George 1987 Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. Chicago/London: University of Chicago Press. 1990 The invariance hypothesis: Is abstract reason based on image-schemas? Cognitive Linguistics 1-1: 39-74. 1993 The contemporary theory of metaphor. In: Ortony, Andrew (ed.). Metaphor and Thought (2nd edition). Cambridge: Cambridge University Press, 202-251. Lakoff, George and Johnson, Mark 1980 Metaphors We Live By. Chicago/London: University of Chicago Press. Lakoff, George and Turner, Mark 1989 More than Cool Reason: A Field Guide to Poetic Metaphor. Chicago/London: University of Chicago Press. Liebert, Wolf-Andreas 1995 Metaphernbereiche der virologischen Aidsforschung. Lexicology 1: in press Zentrale Metaphernmodelle für den Forschungsprozeß. In: Wiegand, H.E. (ed.). Nachdenken über Wörterbücher II. Tübingen: Niemeyer. Seiffert, Helmut and Radnitzky, Gerard (eds) 1992 Handlexikon zur Wissenschaftstheorie. München: Ehrenwirth. Sweetser, Eve Ε. 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Urmson, James O. and R6e, Jonathan (eds) 1992 The Concise Encyclopedia of Western Philosophy and Philosophers. London/New York: Routledge.

Language, space and theography: The case of height vs. depth Jean-Pierre van Noppen

'Thought is a bird of space, that in a cage of words may indeed unfold its wings, but cannot fly'. K. Gibran: The Prophet.

Descriptive theology ('theography') frequently resorts to metaphorical modes of expression. Among these metaphors, the spatial language of localization and orientation plays an important role to delineate tentative insights into the relationship between the human and the divine. These spatial metaphors are presumably based on the universal human experience of interaction between the body and its environment. It is dangerous, however, to postulate universal agreement on meanings associated with spatial dimensions and directions, especially in the diachronic and diacultural situation of the Scriptures. Biblical and doctrinal theography offer two different views of space (an 'experiential' and a 'rational' one) which are not necessarily incompatible, but which reflect two different perspectives with different corollaries. Measurement of metaphorical meanings associated with different theographic utterances shows that certain spatial dimensions (here: 'height') may have lost some of their popular appeal and suggestive power (at least to a hypothetical 'secularized' audience), but the substitution of alternative spatial imagery (here: 'depth') does not allow to retrieve or replace allegedly 'lost' dimensions of meaning. 'Where is God?': in human theography, the issue appears as one of fundamental importance. In biblical contexts, the query is an expression of doubt or a challenge thrown at the believer by the heathen (Job XXXV: 10, Psalm XLll: 3 and 10, LXIX: 10, CXV: 2). Although the question is, in one sense, a category-mistake, it highlights the apparent need for a representation which renders the divine as real and manifest in the human situation, i.e. as known and met at certain places.

680

Jean-Pierre van Noppen

Of course, the 'places' assigned to the divine may tell us more about human feelings and experience than they do about God, for here as elsewhere, spatial models hold a pre-eminent place among the categories used to define and refer to the human existential situation. The language of space offers a concrete metaphorical anchorage for the expression of mental experiences and intellectual conceptions, and thus enables humans to 'situate' themselves with regard to reality by means of those relations and coordinates with which they normally orient and locate themselves in the sphere of physical experience. The loci and dimensions are invested with mental values, and the respective 'positions' of God and Humans come to define the relationship postulated between them. The theolinguist can, then, carry out a topo-analysis inasmuch as the human attempt at structuring his religious experience is reflected in spatial language, and notably in the lexical manifestations of the intersection between the cognitive metaphor MIND is BODY o r MENTAL is PHYSICAL a n d t h e b a s i c

'anthropomorphic'

projection GOD is SPATIAL.

This topo-analysis may opt for different approaches: thus one might decide to investigate the metaphorical meaning of particular loci, dimensions, orientations or polarisations, or choose to address the cognitive import of, say, static vs. dynamic representations, 1 or proceed in the opposite direction and seek to assess how particular categories of experience and thought are mediated by varying descriptions — not unlike the way in which Michael Reddy (in a paper that was to inspire Lakoff and Johnson) showed how the 'conduit' metaphor for language fostered a conception of linguistic interaction altogether different from that projected by the 'tool-maker paradigm' (Reddy 1979) —, i.e. how two different metaphors may each highlight or downplay very different aspects of the same referent (although this is a rather obvious observation which can be boiled down to an interactive view of metaphor). It may cause some surprise to observe that one and the same category for example, the presumably pre-axiomatic, universal spatial metaphors of verticality - may give rise to different systems of implication according to the underlying conception of space, and that thus, metaphoric representation may be partial in two senses of the word, i.e. not only selective but also biased. But the surprise is considerably abated when one comes to think of the fact that, especially in religious language, our spatial metaphors (as well as a number of others — think of the present-day feminist response to the Bible's androcentric representations) have come a long distance through space, time and culture, and may, like so many other linguistic resources, have taken on different values conditioned by changes in our world-view; but that (not least due to the conservative nature of much religious expression) antiquated conceptions may have survived alongside

Language, space and theography

681

present-day developments (e.g. in cosmology), just as archaic lexical and syntactic phenomena (the 'thees' and 'thous') co-exist in the religious register with modern linguistic forms. Hence, it may be short-sighted to postulate an unchanging value for a given category, even if at the outset, a set of interrelated and interdependent metaphors seem to add up to a coherent system rooted in a familiar cognitive process like, say, the assimilation between mental and physical activity based on the universal human experience of corporeal existence or - returning to the issue which concerns us here - on a hypothetical common experience of space underlying our linguistic habits. 2 Anthropologists readily oppose a 'secular' and a 'religious' conception of space. The modern, presumably scientific conception views the universe as an undifferentiated Euclidian medium, isotropic, impersonal and totally homogeneous:3 a space which may still be 'explored' in quantitative terms, i.e. beyond the bounds of what can be reached through the means of present-day technology, but which no longer offers any 'mystery' to be 'discovered'; whereas in the primitive, pre-scientific conception, the cosmos is a discontinuous espace vecu, a 'quality space' whose elements, loci, dimensions, and directions are fraught with affective charges (in general discourse) or spiritual values (in religion). The 'centre' of the cosmos may vary from one culture to another; and places and orientations may thus be felt to stand for different significates. It is important, in this perspective, to distinguish between the intellectual notion and the affective experience of space. Even in the language of everyday secular existence, our metaphors seem to be based on an underlying oriented, discontinuous view of space4: thus we speak of important things as central, and of secondary concerns as peripheral; of a significant experience as profound, and of one that leaves little impression as superficial; of ambition as climbing the social ladder, and of failure as a downfall; we speak of getting nowhere, of toeing the party line or following a line of thought, of straying off the point or reaching a dead end — as if mental activity were physical movement in a space where certain directions (upwards, inward, forward) are endowed with desirable values. Similarly, in a religious world-view, we find traces of a heterogeneous structure: in the Bible, and more particularly in the Old Testament (Chopineau 1 9 6 6 ) , the world is 'vectorialized' around centres of divine self-revelation: space is the locus of theophanies. On earth, the loci of theophany are qualitatively different from the rest of the human environment, and require a particuliar form of behaviour: humans must take off their shoes, keep away, or build an altar (Genesis XII: 7 , xxvill: 1 6 - 1 9 , Exodus in: 5). There are even 'degrees' in the sacredness of space: in the known world, the Holy Land is the sacred abode of God's chosen people;

682

Jean-Pierre van Noppen

within the Holy Land, Jerusalem is the Holy City; within Jerusalem, the Temple is more sacred than the other places, and within the Temple, the Holy of Holies. However, even in the Old Testament there appears a line of reasoning closer to the 'modern scientific' view, and which counterbalances the 'religious' experience of space described by e.g. Mircea Eliade. The Temple in Jerusalem is the visible sign of divine presence (IKings Vlll: 10), but at the same time it raises the issue of the divine abode: "Will God indeed dwell on earth? behold, the heaven and heaven of heavens cannot contain thee; how much less this house that I have builded ? " (IKings Vlll: 27, Jeremiah VII: 3-7). There is a clear tension between the metaphor which postulates divine omnipresence and the metaphors which localise God 'in' a particular spot (van Noppen 1988a): the representation of the divine at a particular locus of the universe is here a metaphor for the election of the Chosen People, and the point of contact between God and the world mediates God's presence to the rest of space, i.e. to all of humanity. But this presence is not to be taken for granted: it becomes a reality for Man only in and through the activity of worship; and thus, God's presence comes to be conditioned by human behaviour. The New Testament strengthens this view: the real temple of God is Man (ICor. ill: 16), and as a consequence, no place on earth can claim exclusive rights to divine presence. If the human heart is the only real temple, then no point in space can be called sacred. Every place in the universe becomes a potential centre, and the cosmos becomes homogeneous in all its dimensions. These two conceptions in turn interact with the 'finite container' view of space into which God 'enters' through incarnation without leaving his specific mode of existence — a projection which, despite appearances, does not postulate any separation between heaven and earth, but a union achieved in Christ (Torrance 1969). In traditional Christian theography, however, the space in which the divine is represented is clearly oriented or vectorialized, notably in its vertical dimension. The dimension 'up', which gives rise to images of height and heaven, ascension and descent, derives its pre-eminent value from the fundamental human experience of the body's erect position, from the difficulty experienced in elevating the body from the earth's horizontal surface, from the awe inspired by observation of the celestial vault, or even, according to some anthropologists, from the conception of the universe as a celestial tent or canopy. These metaphors of verticality constitute powerful and practically irreplaceable means to signify what is sublime, wonderful, worthy of admiration, and hence what is moral, spiritual or transcendent; thus they provide an appropriate frame for human aspirations towards selftranscendence. It is not surprising, then, that throughout a vast number of

Language, space and theography

683

cultures and religions, the loci situated at a distance upwards from the surface of the earth should be regarded as the abode of supreme and divine beings. In the Christian scheme of thought, the metaphorical location of God in the heavenly regions is congruent with more general categories of religious expression inasmuch as it bears similar connotations of divine glory, majesty, holiness, otherness and transcendence; but it differs from other religions in that it posits no equation between God and heaven and is not dependent on any deification of the sky (Smart 1969: 228); nor is the dwelling of God in heaven regarded as a natural presence: it remains a metaphor. Heaven may be used as an image of transcendence, but the heavens themselves are transient: they "roll up like a scroll" and "will pass away" (Isa. xxxiv: 4, Psalm CII: 26). As pointed out already, God's presence is not limited or guaranteed by particular loci — no matter how essential these may be to the communication of his nature and attributes. But even as 'mere' metaphors based on the analogy between positional and essential superiority, the categories of height and heaven may entail different theographies according to the frame of presuppositions within which they are interpreted. When human discourse seeks to 'chart' an extra-linguistic referent, i.e. seeks to provide it with a mental and/or linguistic representation, it employs a 'projection', not unlike the way in which cartographers resort to Mercator's or the orthogonal projection; and just as different projections result in different maps, the perception (and a fortiori the description) of an outside 'reality' may be warped by different systems of representation. An interesting case in point is the spatial category of verticality. The metaphorical language which localizes the divine in the celestial regions has been indicted for introducing a 'supernatural' projection — notably by Bishop Robinson in his theological bestseller Honest to God. The problem here is not so much that the spatial propositions are taken literally (as by Yuri Gagarin, who returned from his cosmic foray saying that he had seen no God in heaven), but that even as a metaphor, the category distorts the reality it is there to communicate. For a number of theologians5, representation of the divine in heaven is an 'extrapolation' which relegates God to a marginal position 'out there', and projects a separation between two worlds, between 'heaven' and 'earth'. This criticism, however, seems to wrest the metaphor of verticality out of its original cultural and historical context to invest it with structural associations altogether foreign to the Bible's theological schema: it is correct to say that biblical theology expresses the ontological difference between God and Man in terms of distance, and transcendence in terms of height; but the biblical corpus does not allow one to develop a two-world cosmology.6

684

Jean-Pierre van Noppen

It is in situations like these that the linguist feels impelled to come to the aid of the theologian, and to contribute his insight to diagnose the breach in communication, administer the proper antidote and avoid any undesirable secondary effects. Here as elsewhere (van Noppen 1983, 1988b), a delineation in terms of the misunderstanding or misapplication of metaphor logic may throw an altogether useful light on the types of misunderstanding involved. In this particular case, the confusion between a form of transcendence which maintains unity between God and Man on the one hand and a supranaturalist representation which suggests a break between them on the other, apparently results from two different readings of the vehicle of verticality, which may be used as a metaphor to both representations of the divine: while as pointed out above, the biblical schema seeks to differentiate Man and God but posits a fundamental unity between them, realized in Jesus Christ, the 'supranaturalist' reading of the same metaphor focuses on the concept of distance (which is only one relevant feature in the total semantic structure of the metaphor) and interprets it as an image of separation, i.e. as the projection of a relationship of indifference. This reading, however, simultaneously commits three interpretive fallacies: the mimetic fallacy, which views height in geometrical terms of distance rather than in spiritual terms of difference; the reductionist fallacy, which entirely bases its interpretation of the divine on this static, geometrical schema, and ignores other metaphors which speak of proximity and unity; and finally, the fallacy of context negation, which interprets a culturally conditioned metaphor in terms of presumed present-day associations and ignores its role and function within the biblical framework which gave rise to it. Spatial metaphors may be based on presumably universal human experiences; but this does not entail that one can postulate unchanging, self-evident meanings for them, elucidate them in terms of present-day commonlanguage associations, or treat them as 'open-ended' images which one is free to interpret ad libitum. There is a fundamental, but uncomfortable form of intellectual honesty which demands that one should not project onto a text meanings and intentions which were not in the author's mind or message in the first place. But retracing an author's original intention may require long and laborious hermeneutic detours, and one may without difficulty imagine a situation in which metaphors coined at a precise point of space and time require so much study or qualification that eventually they lose their meaning or their appeal, and thus give rise to the very interpretation errors we have just denounced. While an isolated individual's error will restrict its effects only to its author, errors committed - and widely published - by someone enjoying public status and authority, like Bishop

Language, space and theography

685

Robinson, may have widespread and disastrous results. Professor Robinson considered himself as the epitomy of a hypothetical 'Modern Man', whom he invested with his own confusions. On behalf of this abstraction, Robinson advocated recourse to the category of 1 depth' as a possible corrective to the traditional, presumably obsolete, imagery of height. 7 The transition from one complex projection, 'up and out' to another, 'down and in' is a beautiful case of theolo-fiction; but within the perspective of this volume on language and space, it may be worthwhile to try and investigate the impact of a shift in spatial language which is bound to radically upset the system of metaphorical, geometrical, physical and even elemental (van Noppen 1993) associations: the heights of heaven are replaced by profound abysses, and the spiritual connotations of sky symbolism by the tangible intimacy of the earth's depths. We may attempt first to assess the potential value of the alternative imagery, and subsequently try to evaluate its reception by the public. The imagery of depth is ambivalent in more than one respect: in the structural mode of meaning, it partakes of the dimension 'below' which contrasts with the heavens 'above' and which is thus associated with the earthly sphere of human existence; but when distance is measured downwards from the earth's surface, depth may come to refer to the underworld. On the other hand, it derives a positive value from architectural associations, in which a firm substratum constitutes a reliable support, and from organic connotations of the earth as deep source of nourishment. In the textural associations of downward orientation, we note a similar contrast: on the one hand, the qualitative connotations of height may be inverted, and depth may be fraught with negative value judgments; on the other hand, depth and ground imagery seeks truth value beneath the level of superficial, trivial and often deceptive appearances, and charges the downward and inward direction with positive associations of intensity and ultimacy beyond the grasp of humans, of interiority and immanence. In Bishop Robinson's 'pop' theology, depth holds a pre-eminent place as an alternative to the allegedly alienating imagery of height and heaven. Robinson borrows the metaphor from Paul Tillich, whose own use of Grund can be traced back, in turn, to Schelling, Heidegger and Boehme, but Robinson fails to give the terms the full scale of theographic resonance which they enjoy for these German authors (Tillich 1951-63, Dunphy 1977), and the alternative projection, for all its wealth of associations, misfires: not only do depth and ground take on impersonal and static overtones, but in addition, they lead to an unintended immanentist or even selfdeifying reading: 'If what I address [in prayer] is the depth of myself, says one reader, 'then I am talking to myself. (...) If Jones starts worshipping the God within Jones, he cannot but end up worshipping — Jones.'

686

Jean-Pierre van Noppen

In van Noppen (1979), I developed an ad hoc socio-semantic survey attempting to indicate (among others) whether, and to what extent, Robinson was right in assuming that 'height' metaphors had lost their appeal as a theographic category, and whether the meanings presumably lost could be usefully retrieved by the advocated recourse to 'depth' imagery. With this aim in mind, I submitted to a varied public a series of statements representing different realisations of the metaphors involved, and measured (in what could be called a form of Metapher-rezeptions-forschung) the deviance which these statements caused with regard to the global image of God which the informants had previously charted on a scale of theological meanings. 8 While a detailed description and breakdown of the responses to all the statements would take us well beyond what can be dealt with within the bounds of a short paper, the indications provided by the informants can be conveniently summarised to highlight the 'cash-value' which some realizations of spatial language may enjoy in a theological context. (1)

There is no being out there at all. The skies are empty: the first statement, borrowed from Robinson himself, was subjected to the informants' judgment to measure the impact of a form of metaphorical iconoclasm which might strip the image of the divine of its localization in the celestial regions. Even though at the literal level, the representation given here might be held to convey the rather obvious point that God is not 'a' perceptible being physically located in empirical space, it appears to have a certain shock-value as a metaphor: it seems to deprive the divine of its existence, of its personality and of all the positive evaluations (concern, presence, relevance, reality) which might be associated with it. The pattern of semantic association appears to be almost the symmetrical opposite of the image of God originally charted. Informants comment that the perspective of an empty sky is frightening, and that predicating nonspatiality of God amounts to implying His non-existence.

(2)

God the Son came down to earth. From 'out there' there entered into the human scene one who was not 'of it' and yet who lived completely within it. (3) In prayer, they lift their hearts to their ascended and triumphant Lord: these statements represent the traditional imagery in all its polyvalence, inasmuch as they combine the positive connotations of verticality with the category of personality, which projects the divine as accessible. The results charted by the respondents stay very close to the original conception of God, although with a slight loss of immanence and a marked gain in 'reality'. This might suggest that the 'mythological' representations

Language, space and theography

687

are not so much of an obstacle to correct theological understanding as Robinson suggested, and that christological images of 'incarnation' aptly convey the proximity, presence and concern of God which he believed were jeopardized by this very imagery. (4)

God is the supreme being, the Grand Architect, who exists somewhere out beyond the world. (5) God is very much a separate being, standing over against the world: here, the respondents are confronted with a Robinsonian caricature of transcendence: the locus of the divine is not invested with associations of height and glory, but fraught with connotations of separation, distance and unconcern. The public responds as expected: transcendence is stressed to the point where other aspects (proximity, personality, reality) of divinity are sacrificed. One might then say that in a way, Robinson was right in suggesting that too 'supranaturalist' a projection might prove harmful to an image of God which underlies religious practice and worship. It should be pointed out, however, that this representation is the result of Robinsonian, not Biblical, theography; and that the advocated 'displacement' of God from the heights to the depths, therefore, finds little justification. But as a matter of academic interest, we may pursue the matter a little further and measure the possible impact of 'depth' imagery on the popular image of God.

(6)

God is not 'out there', an Other beyond the skies, but the Ground of our very being; (7) We must give the name 'God' to the ground of our being, the deepest reality within us: statement 6 marks the transition from height to depth imagery, and, according to Robinson, from Bultmann to Tillich. The projection triggers off an increase on the axes of 'positive' qualities (presence, concern, reality and attractiveness) but on the other hand, a marked trend towards similarity and interiority, resulting in an 'immanentist' reading. Statement 7 consolidates this tendency, and causes a loss in transcendence to the point where the reading becomes pantheistic, i.e. negates the difference between God and creation, and even assimilates the two.

Basing ourselves on these indications, we may conclude, or rather confirm, that spatial metaphors profoundly affect our perception of the reality represented; that these metaphors partake of universal categories of perception and imagery, but that one cannot, indeed must not posit a universal interpretation of their value; that height and depth metaphors have altogether different systems of potential associations, and that while both

688

Jean-Pierre van Noppen

may contribute precious insights to theographic description, one cannot be substituted for another. Another precious indication is that the biblical use of spatial language, albeit 'mythological', conveys the appropriate theological insights within a cultural framework where knowledge of the imagery is fostered and maintained, 9 but that interpretation of the same metaphors in a different, present-day perspective may lose the meaning originally intended. In the hypothetical case where it would prove necessary to substitute one form of imagery for another, it would not be wise to recast everything in terms of a radically different dimension, lest one should introduce altogether foreign and irrelevant connotations. A more welcome solution might lie in recourse to a multi-model discourse, in which different categories of imagery complement but do not exclude each other, even if they contradict each other at the signifiant level (e.g. interiority and exteriority, transcendence and immanence, height and depth, localization and omnipresence, masculinity and femininity, but also traditional and modern imagery) (Crystal 1988). In the case of theography, the signifie is so vast and multifarious that it will not be satisfactorily encompassed by any single category of imagery.

Notes 1. 2. 3.

4. 5. 6.

7.

van Noppen (1980: 74-252) proposes a glossary of spatial terms in theographic use (5,795 tokens/426 types). The various syntactic patterns in which the metaphor is cast are described on pp. 306-24. It should be remembered, however, that Whorf (1953) pointed out how the Hopi thought world has no imaginary space. It is to be noted that the more modern Einsteinian conception of spacetime apparently has some difficulty finding its way into popular imagination, save perhaps in the form of the 'paradox of the twins', which is naively pictured as a form of time-travel. MatorS (1976) has performed a topo-analysis of spatial and geometrical metaphors in a corpus of present-day literary and journalistic texts (see also Hall 1966 and Lakoff and Johnson 1980). see esp. Robinson (1963: 14-16, 27, 41); for Robinson's sources (Bultmann, Feuerbach) and other theologians adhering to this view, cf. van Noppen (1980: 28-33, 345). Chopineau (1966: 362) points out that despite its dual lexical structure, the noun phrase "heaven and earth" is not to be understood as the expression of the existence of two entities, but as a phrasal term for 'the whole universe'; cf. also Barr (1961: 206-261; 1962: 160ff.). It must be pointed out here that the 'depth' imagery advocated by Robinson (1963), albeit his main alternative, is not the only possible relocation of the divine: there is also a 'horizontalist', 'immanentist' view which locates the divine 'among' humans or 'in' the process of history and human endeavour. These projections, when adopted at the exclusion

Language, space and theography

8. 9.

689

of others, raise problems of their own (e.g. a risk of pantheism); but this issue, studied in van Noppen (1980) will not be dealt with in the present paper. The full procedure is described in van Noppen (1979: 301-318). The method for measurement of metaphorical meaning by means of semantic association has been considerably improved by Steen (1992). Outside such a religious framework, meanings may evolve towards a more 'secular' reading, cf. Ader (1975).

References Ader, Armin 1975

Zum Wortschatz des Credo bei Jugendlichen. Versuch empirischer Verfassung. Linguistica Biblica 35: 80-83.

Barr, James 1961 The Semantics of Biblical Language. London, Oxford U.P. 1962 Biblical Words for Time. London, SCM Press. Chopineau, Jacques 1966 Space in the Bible. Student World LDC, 4: 358-66. Crystal, David 1988 Das Generieren theologischer Sprache. In: van Noppen, Jean-Pierre (ed.), 297-314. Dunphy, John 1977 Paul Tillich et le symbolisme religieux. Paris: Delarge. Hall, Edward T. 1966 The Hidden Dimension. New York: Doubleday. Johnson, Mark L. 1987 The Body in the Mind: The Bodily Basis of Meaning, Imagination and Reason. Chicago: Chicago University Press. Lakoff, George and Mark L. Johnson 1980 Metaphors We Live By. Chicago: Chicago University Press. Mator6, Georges 1976 L'Espace Humain. Paris: Nizet. Reddy, Michael 1979 The conduit metaphor: A case of frame conflict in our language about language. In: Ortony, Andrew (ed.). Metaphor and Thought. Cambridge: Cambridge University Press, 284-324. Robinson, John A.T. 1963 Honest to God. London: SCM Press. Smart, Nirian 1969 The concept of heaven. In: Vesey, G.N.Α.: Talk of God. London: MacMillan, 165ff. Steen, Gerard 1992 Metaphor in Literary Reception. Doctoral Dissertation. Vrije Universiteit Amsterdam. Tillich, Paul 1951-63 Systematic Theology. Chicago: Chicago University Press. Torrance, Thomas F. 1969 Space, Time, and Incarnation. London: SCM Press, van Noppen, Jean-Pierre. 1979 A method for the evaluation of recipient response. Technical Papers for the Bible Translator, 30, 3: 301-318.

690

Jean-Pierre van Noppen

1980 1983

Spatial Oleography. Ann Arbor, UMI. A metaphorological delineation of theographic expression, interpretation and errors of interpretation. Communication and Cognition XVI, 4, 439464. 1988 (ed.) Erinnern, um Neues zu sagen. Die Bedeutung der Metapher für die religiöse Sprache. Frankfurt/Main: Athenäum. In als theographische Metapher. In: van Noppen, Jean-Pierre (ed.), 2001988a 217. Metapher und Religion. In: van Noppen, Jean-Pierre (ed.), 7-51. 1988b Espace, ciel et terre: Les m6taphores qui forment et d6forment. Paper 1993 read at the conference 'La form delle immagini: per una metaforologia politico III: I quattro elementi e le loro metafore'. Verbania (Italy), October 1993. Whorf, Benjamin Lee 1953 Linguistic Factors in the terminology of Hopi Architecture. International Journal of American Linguistics 19, 141-145.

List of contributors

Elena Bellavia Via di Donna Olimpia 134 00152 Rome, Italy Willem J. Botha Randse Afrikaanse Universiteit Departement Afrikaans Johannesburg, South Africa Frank Brisard University of Antwerp (UIA) Department of Linguistics Antwerp, Belgium Eugene Η. Casad Summer Institute of Linguistics Tucson, Arizona, U.S.A. Haruko Minegishi Cook University of Hawai'i at Manoa Department of East Asian Languages and Literatures Honolulu, Hawai'i, U.S.A. Kenneth W. Cook Hawaii Pacific University English Foundations Program Honolulu, Hawai'i, U.S.A. Hubert Cuyckens University of Antwerp (UFSIA) English Language Group Faculty of Applied Economics Antwerp, Belgium

692

List of contributors

Ewa Dqbrowska University of Sussex Department of English Brighton, U.K. Robert B. Dewell Loyola University Modern Foreign Languages and Literatures New Orleans, Louisiana, U.S.A. Heidrun Dorgeloh University of Düsseldorf Department of English Düsseldorf, Germany Steven Frisson Belgian National Science Foundation/ University of Glasgow Department of Psychology Glasgow, U.K. Vimala Herman The University of Nottingham Department of English Studies Nottingham, U.K. Deborah Hill Cognitive Anthropology Research Group Max Planck Institute for Psycholinguistics Nijmegen, The Netherlands Carlos Inchaurralde Universidad de Zaragoza Facultad de Geologicas Zaragoza, Spain Olaf Jäkel University of Hamburg Graduiertenkolleg Hamburg, Germany

List of contributors

Dieter Kastovsky University of Vienna Institut für Anglistik und Amerikanistik Vienna, Austria Agata Kochanska University of Warsaw Warsaw, Poland Barbara Kryk-Kastovsky Perlhofgasse 17 Giesshuebl, Austria Lorenza Mondada Universite de Lausanne Institut de Linguistique et des Sciences du Langage Lausanne, Switzerland Walter De Mulder UIA-ROM

Wilrijk, Belgium Sabine Neumann Cognitive Anthropology Research Group Max Planck Institute for Psychol inguistics Nijmegen, The Netherlands Winfried Nöth University of Kassel Department of English Kassel, Germany Jean-Pierre van Noppen Universite Libre de Bruxelles Department of English Brussels, Belgium Janina Ozga Jagiellonian University Institute of English Cracow, Poland

694

List of contributors

Sally Rice University of Alberta Department of Linguistics Edmonton, Canada

Dominiek Sandra University of Antwerp (UFSIA) Department of Linguistics Antwerp, Belgium

Carlo Serra Borneto Universita degli studi di Roma Dipartimento di scienze del linguaggio Rome, Italy

Susan Strauss University of California at Los Angeles Department of Applied Linguistics Los Angeles, U.S.A.

John Taylor University of Otago School of Languages Linguistics Section Dunedin, New Zealand

Marjolijn Verspoor University of Groningen Department of English Language and Literature Groningen, The Netherlands

Karin Wenz University of Kassel Department of English Kassel, Germany

Thomas Widlok Cognitive Anthropology Research Group Max Planck Institute for Psycholinguistics Nijmegen, The Netherlands

List of contributors

Cornelia Zelinsky-Wibbelt Universität Hannover Englisches Seminar Hannover, Germany Milena iic Fuchs University of Zagreb English Department Zagreb, Croatia

Subject Index

-a/epelita, 395, 397, 399-403, 405-413 acoustics, 64, 65, 68 addressee honorifics, 3, 24 alienable possession, 321, 325 anaphora, 264, 265 antecedent, 264, 265 antonymy, 373 auditory space, 64, 65 axiological code, 273, 275, 276 axiology, 605, 611 axis horizontal, 74, 80, 81, 159, 192, 376-378, 382-385, 387, 460 vertical, 74, 79, 80, 159, 191, 374, 378, 381, 382, 387, 392, 460, 572, 611 backgrounding, 21, 22 Bantu, 287, 334, 335, 337, 345, 348350 bodily experience, 160, 376, 460, 461, 504 bounded episode, 123, 449, 506 bounding, 496, 500-503, 536, 539 external, 506, 528, 538, 544, 545, 547 call contours, 64, 67, 68, 70 cardinal directions, 307, 348-351 categorizing, 530 categorization, 141-144, 349, 614, 615, 617, 635 certainty, 173-175, 177, 178, 180 classificatory verbs, 239, 246, 252, 253

coding time, 221 place, 331, 333, 340 cognition, 211, 269, 273, 275, 285, 329, 350, 379, 405, 410, 418, 420, 423-427, 449, 567, 610, 611, 613 cognitive dissonance, 126, 127 distance, 613, 616, 625, 626, 633 linguistics, 71, 73, 136-138, 143, 163, 488, 493, 540, 592, 613, 675 semantics, 610 tasks, 347 comparison, 33, 102, 116, 129, 146, 147, 277, 308, 322, 345, 348-352, 354-356, 389, 406, 481, 515, 534, 536, 537, 621, 626, 672 conceived location, 244 concept, 60, 83, 101, 102, 129, 136, 143, 163, 167, 181, 184, 197, 198, 212, 213, 231, 239, 285, 291, 303, 311, 317, 329, 330, 335, 350, 373, 422, 424, 425, 427, 441, 451, 488, 492, 496, 497, 503, 504, 510, 529, 530, 536, 545, 599-603, 606610, 613, 616, 634, 636, 637, 643, 651, 652, 657, 663, 671, 684 conceptual space, 181, 601, 602, 605, 606, 608 conceptualizer, 212, 217, 222, 224, 411, 417, 418, 420, 425, 428-130, 433-135, 438, 440, 444, 445, 492, 494, 495, 518, 545 conceptualization, 57, 60, 63, 143, 198, 222, 285, 308, 329, 330, 349-351 355, 378, 381, 382, 384, 391, 392, 405, 424, 431, 442, 492-496, 501, 504, 512, 538, 559, 560, 566, 567, 600, 649, 651,

698

Subject Index

653, 654, 657, 663, 669, 673 conflicts of coding, 281 constituent-order profile, 125-130 construal, 73, 124-126, 128, 129, 222, 246, 386, 391, 392, 417, 418, 427 428, 430, 431, 437-439, 445, 447, 471, 495-498, 500, 501, 504, 518, 519, 522, 524, 653, 673 conventions, 183, 350, 528, 529, 544, 547, 555, 556, 564 conversational, 40, 41, 252, 253, 265, 316, 319, 320, 321, 325, 354, 355 corpus-(study), 650 Croatian, 49-60, 332, 334 cross-linguistic, 53, 330, 413, 491, 559, 650, 651 culture, 14, 60, 314, 322, 339-343, 345, 347, 350, 351, 355, 385, 388, 558, 559, 591, 600, 611, 680, 681, 683 degree of metaphoricity, 384, 388, 391 deictic(s), adverb, 57, 213, 215, 223, 510-512 pronoun, 213, 215 spatial, 52, 130, 212, 222, 270, 561 demonstration, 20, 30-33, 35, 36, 43, 515, 625, demonstrative(s), 6, 29-44, 57-60, 213-215, 217, 220, 225, 239, 266, 333-336, 572 demonstrative adverbs, 49-51, 57, 59 depth, 561, 563, 602, 633, 679, 685688, 666, 670, 672-674 description of space, 269, 271, 272, 278, 326, 563 diachronic, 162, 341, 489, 615, 634, 650, 651, 669, 672, 675, 679 dialectal variation, 330, 332, 333, 340, 341 dialogical space, 601, 602, 605-608 dialogue, 553-555, 558, 567

didascalia, 558 digression, 577, 583, 585-589 dimensional adjectives, 528, 534, 535, 540, 548,549 directional systems, 307, 308, 315-318, 321, 322, 325 directionality, 352, 376, 385, 409 discourse, 21, 22, 39, 41, 43, 44, 49, 70, 183, 191, 239, 259, 260, 262-265, 270, 272, 274, 278, 352, 397, 405, 510, 511, 513, 515, 517, 518, 520-524, 527-533, 536, 538, 540, 542, 558, 567, 571, 573, 574, 577, 580, 581, 586, 589-592, 600603, 608-610, 649, 650, 652, 669, 672, 681, 683, 688 discourse deixis, 571, 573 entity, 510, 512-514, 517-519, 524 organization, 270 displacement, 69, 94, 102, 104, 510, 511, 520-523, 687 displaced immediacy, 511, 520, 522-524 distance emotive distance, 69 physical distance, 6, 8, 10, 64, 68 social distance, 6, 10, 12, 63, 69 do-, 253, 468, 474-476, 485-487 domains, 3, 32, 100, 114-116, 119-122, 129, 142, 159, 160, 182, 184, 198, 239, 240, 259, 265, 329, 331, 336, 340, 390, 421, 503, 504, 526, 567, 571, 574, 608, 610, 649, 650, 669 hill, 240, 339 river, 240, 336, 339 drama, 71, 553-555, 558, 560-562, 566, 567 egocentric, 429, 562 code, 273, 274 embedding, 186, 511, 515, 517 emotional(ly), 68, 412, 430, 434, 439, 446448, 461 enclosure, 291, 395, 396, 405, 412, 554,

Subject Index

558, 559 encoded meaning, 5-7, 9, 10, 12, 13, 16, 17, 19 English prepositions spatial usages of, 137, 138, 151, 153 non-spatial usages of, 138 entity, 4, 36, 58, 97, 99, 128, 140, 183-191, 197, 200, 214, 215, 217, 220, 224-226, 239, 246, 247, 251, 262, 288, 292-294, 311-313, 316, 331, 385, 403, 404, 431, 450, 468, 492, 510, 512-514, 517-519, 524, 528, 536-538, 540, 547, 587 epistemic modality, 173, 177 event, 13, 21, 22, 49, 69, 98, 117, 123, 124, 126, 127, 129, 173-175, 177, 183, 185, 239, 245, 246, 256, 262, 263, 265, 270, 274, 281, 293, 294, 301, 303, 329, 356, 396, 397, 399, 401, 404, 407, 410-414, 418420, 422, 427, 429^36, 438-451, 463, 467, 470, 479, 481, 485, 510, 511, 513-515, 520-524, 527, 529, 531, 537, 555, 558, 560, 561, 567, 576, 580 external deictic, 212 Fallacies (in Metaphor Interpretation), 684 field of vision, 458 figure-ground, 36, 41, 43, 278, 537 future, 162, 171, 172, 174, 175, 177180, 221, 266, 442, 446-448, 457463, 470, 483, 542, 572, 641, 643 gapped landmark, 112-114, 116, 122, 124-126, 129, 130 genre, 3, 8, 12, 521, 524, 553 geometry, 383, 600-602, 606, 610, 611 German, 69, 73, 79, 83, 94, 98, 101, 102,

699

102, 108, 109, 111, 122, 332-335, 340, 342, 373, 375-377, 380, 382, 383, 385, 388, 514, 519, 530, 535, 337, 651, 685 goal 22, 52, 75, 88, 89, 182, 252, 272, 289. 291, 295, 299, 300, 303, 310, 338, 474476, 558, 560, 563, 566, 591, 602, 603, 607, 608, 614, 617, 654, 655, 662, 672, 674 grammaticization, 397, 399, 403 grounding predication, 37 Hai//om, 345-348, 350-363 Hawaiian, 455, 456, 458, 459, 461, 463 height, 79, 80, 159, 186, 187, 277, 338, 385, 561, 566, 679, 682-688 hearer, 32, 35-42, 52-57, 60, 64, 70, 213, 222, 239, 251, 256, 259, 262, 274, 293, 336-338, 449, 511, 513, 518, 520, 559, 574, 582, 605, 607 higher social status, 23 homonymy, 141, 491, 503, 626, 630, 633, 635-638, 643 honorifics, 3, 5, 10, 12, 24 hypothesis, 10, 102, 170, 171, 173, 175, 339, 341, 379, 387, 389, 401, 420, 427, 489, 503, 516, 528, 531, 576, 591, 609, 614-617, 621, 635, 636, 639, 642, 672 iconicity, 10, 12, 269-271, 273, 278, 285, 567, 589-591 image-schematic structure, 493, 496, 498, 504 abilities, 504 imagery, 374, 386, 391, 392, 679, 685-688 imaginary tour, 272-274, 280, 282, 283, 563 imperfective, 246, 252, 437^39, 444, 449, 470, 488, 489, 494-501 inalienable possession, 321, 325 Indo-European languages, 330, 333, 338 indexical, 32, 43, 211, 217, 222, 270, 606

700

Subject Index

internal configuration, 528, 544, 545, 547 interpersonal distance, 3, 9, 12, 13 intrapersonal distance, 6, 9, 15 intrinsic deictic, 212, 222, 228 intonation, 67-69, 71, 599 Invariance Hypothesis, 503 inversion, 509-524 isomorphism, 138, 191, 589, 592 Japanese, 3, 4, 9, 15, 63, 204, 335, 336, 395, 398, 400, 401, 403, 405413 Japanese verbal suffix, 3 Khoisan, 348-350 Kgalagadi, 345-348, 350-361, 664-366 Korean, 140, 395, 396, 401, 403, 405413 landmark -based directional system, 307, 308, 316, 325, 339 (language) elicitation tasks, 351, 354356 language usage, 351, 355, 356 learning experiments, 617, 642 lexical categorization, 136, 142, 146 networks, 73, 136-138, 141-143, 161, 614 representation, 136, 161, 528, 531, 541, 547, 641 semantics, 141, 142, 161 lexicon, 142, 162, 239, 251, 348, 354, 355, 529-532, 534, 547, 615 lexicographic, 211, 212, 231 linearity, 272, 566, 584, 585, 588, 589, 599, 600, 672 linearity of speech, 599 lines, 15, 22, 23, 56, 114, 142, 152, 162, 205, 339, 341, 383,

461, 558, 559, 561, 563, 565, 680, 599604, 607, 608, 654 local dimension, 329 locality, 212-214, 216, 217, 220-222, 224, 565 localization, 527, 529, 530, 580, 679, 686, 688 localizing, 32, 532, 561 locating expressions, 29 locational, 56, 57, 140, 239, 256, 259, 260, 262, 264, 265, 315, 324, 325 goal, 252 nouns, 307, 311 locative(s), 87, 140, 213, 215, 216, 223, 240, 260, 263, 264, 288-293, 295, 298301, 303, 308-310, 314, 315, 318, 319, 321, 322, 377, 382, 391-393, 396, 413, 455, 462, 510, 523, 559, 561, 562, 572, 574 adverbs, 214, 215, 332 particles, 202, 239, 240, 249, 251, 253, 254, 258-260, 263 prefixes, 242, 244, 249, 251, 253, 256258, 262, 263 Longgu, 308-311, 314-322, 325 loudness, 64, 66, 68, 70 macrostructure, 571 manner of motion, 296, 297 Maori, 462, 463 markedness, 379 masu form, 3, 4, 6-9, 11, 19, 21-24 meaning extension, 613, 615, 635 medial abstracting, 112, 113 mental space(s), 71, 143, 264, 439, 440, 547, 547, 576 lexicon, 530, 615 metalinguistic verbs, 579, 580 meta, 674, 675 metaphor(s), 6, 7, 83, 85, 87, 89, 91, 92, 96, 98, 99, 102, 135-138,

Subject Index

181, 182, 191, 285, 373, 375, 376, 380-390, 395, 396, 405, 470, 473, 485, 503, 504, 513, 517, 544, 545, 547, 571, 599-604, 606, 608-611, 614, 618, 625, 643, 649, 650, 652, 653, 655, 656, 658, 660, 662, 664, 665, 668, 670, 672-674, 679, 680688 metatext(ual), 574, 583, 599-603, 605611

metonymy, 527, 529, 533, 534, 614, 643 ad-hoc, 529 conjunctive, 533 part-whole relation, 323, 536 middle voice, 395, 402, 403, 405, 410 mimesis, 285 modals, 174, 177, 181 modality, 167, 170, 171, 173-175, 177, 191 motion abstract, 494, 495, 540, 545, 546 subjective, 494, 495 verbs, 54, 262, 273, 288, 289, 294297, 300, 303, 528, 540, 574-577 579 movement, 40, 54, 69, 73, 75, 77, 100, 102, 104, 115, 167, 169-171, 183, 191, 204, 272, 285, 460^62, 469, 471, 473, 474, 477, 479, 488, 493, 494, 513, 514, 555, 556, 560, 563, 575, 577, 579, 585, 606-608, 610, 624, 654, 663, 681 narrative discourse, 262 non-Indo-European languages, 330, 333, 334, 337, 338, 346 Oceanic, 307-309, 314-316, 321, 322, 324-326 od-, 468, 477-479, 485-489 opposition, 13, 160, 271, 348-350, 373, 375, 403, 421, 448, 556, 572,

701

573, 592, 601, 605, 606, 608, 609 ordo naturalis, 270, 272, 273 orientation, 51, 213, 222, 230, 240, 270, 274, 308, 314, 334, 335, 339, 347, 350, 353, 376, 378, 380, 381, 383, 407, 442, 517, 565, 573, 574, 580, 585, 604, 605, 609, 611, 654-656, 672, 679-681, 685 orientational, 85, 213, 227, 228, 230, 231, 317, 375, 376, 380 Origo, 197, 274, 280, 329-331, 333, 335, 336, 339, 340, 559, 561, 565, 566, 572, 575, 579 paralanguage, 63, 66, 69 past, 7-13, 126, 174, 175, 177, 178, 221, 228, 245, 257, 303, 409, 412, 413, 423, 439, 446, 450, 456, 458-463, 509, 522, 540, 590, 618 path model, 589, 590 perception, 36, 37, 120, 189, 211, 339, 418, 420, 422-424, 426, 427, 431-434, 436, 439, 440, 443-445, 449-451, 458, 528, 529, 563, 599, 610, 626, 637, 640, 653, 683, 687 perceptual code, 273, 278, 282 salience, 278, 281 perfective, 111, 246, 248, 252, 254, 256, 258, 444, 467, 470, 471, 481, 484, 485, 488, 498, 497, 500-503 philosophy (of science), 650, 653, 655, 656, 672, 673 place nouns, 307-310, 312-318, 322, 324, 325 plain form, 3-5, 7-12, 16, 17, 19-24 po-, 468, 479, 481, 484, 485, 488, 501 point of view, 60, 183, 202, 206, 211, 214, 219, 273, 278, 317, 324, 377, 378, 380, 388, 409, 413, 473, 510, 511, 515-517, 528, 545, 558, 601, 618 reference-, 100, 137, 144, 159, 229, 230,

702

Subject Index

264, 307, 317, 492, 510, 512-515, 517-519, 523, 524, 539, 554, 559, 565, 574, 602 pointing gesture, 29, 30, 33, 37, 39, 40 polarity, 373, 374, 378, 380, 387, 391, 392 Polish, 66, 67, 331, 332, 344, 340, 467, 473, 488, 491, 492, 503, 562, 563 politeness, 3-6, 12, 13, 15, 16, 21 Polynesian, 462 polysemy, 135, 142, 160, 312, 503, 530, 531, 533, 613, 614-618, 626, 630, 633, 641, 643, polysemy vs. homonymy-based learning, 626, 630, 633, 643 possession, 122, 184, 185, 188, 191, 308, 321, 325, 396 possessives, 307, 309, 310, 313, 314, 322-325 possibility, 115, 171, 174, 175, 177180, 185, 187, 191, 219, 270, 299, 380, 385, 439, 497, 500, 502, 504, 577, 579, 588, 591, 617, 628, 630, 633, 336, 337, 639, 641, 650 postposition, 135, 255, 336 pragmatic, 5, 24, 49, 116, 118-120, 197, 201, 205, 334, 335, 395, 398, 402, 510, 529, 560, 561, 567, 579, 590 prefix, 73, 88, 90, 93, 96, 99, 100, 104, 109-112, 129, 130, 135, 203, 204, 239, 240,242, 244-249, 251, 253, 255-258, 262, 263, 338, 405, 467-471, 473-475, 477-483, 487, 488, 501, 609 preposition, 38, 73-75, 86, 93, 101, 109, 110, 112, 123, 130, 135-137, 140, 142, 143, 145-148, 150-153, 155, 159-163, 197, 200, 203, 216, 223, 224, 226-228, 295, 309, 310, 312, 315, 316, 318, 319 468, 469, 474, 477, 480, 482, 488, 491-496, 500, 503, 504

538, 542, 543, 545, 558, 559, 616, 617, 620, 622, 625, 627, prepositional networks, 160 present, 3-7, 9, 11, 12, 22, 171, 178, 331, 334, 339, 340, 392, 483, 519, 522 probability, 173, 174, 179, 180, process

572, 613, 628, 631 174, 175, 450, 470, 496-498

homogeneous perfective, 502 imperfective, 449, 500 perfective, 497, 503 procomplement, 240, 259, 260, 265 prosody, 63, 64, 69-71 prototypes, 135-137, 141, 142, 144-146, 148, 151-153, 159, 504, 530, 615 proxemics, 63 proximal/distal, 242-245, 249, 525-254, 256-258, 260-264, 266, 329, 330, 331, 332, 334, 336, 338, 340, 520-522, 524 proximity, 3-5, 9, 10, 24, 33, 55, 56, 58, 225, 226, 278, 283, 329, 330, 333, 334, 337, 457, 554, 557, 589, 608, 684, 687 prze-, 468-472, 474, 481, 485-487, 492 quality space, 148, 681 radial category, 421, 468 Rarotongan, 462, 463 reference, 37, 40, 80-82, 100, 124, 137, 144, 159, 175, 177, 181, 182, 212, 217, 222, 224-226, 228-231, 242, 244, 260, 269, 270, 272, 274, 307, 317, 332, 335, 336, 339, 375, 377, 380, 383, 384, 392, 397, 411, 463, 468, 485, 492, 510, 512515, 517-519, 522-524, 530, 532, 555, 559, 561, 563-565, 578, 580, 585, 600602, 604-607, 609, 617 referent, 4, 12, 29-43, 97, 217, 218, 220, 270, 310, 314, 322, 334, 483, 529, 530, 535, 538, 539, 571, 680, 683 remoteness, 71, 191, 455, 461, 463 route path, 111, 126, 127, 129

Subject Index

Samoan, 462 scale, 67, 84-86, 122, 147, 174, 279, 318, 374, 379, 385, 389, 522, 528, 534, 554, 609-611, 685, 686 scenario metaphorical, 649, 651-653, 655, 657, 658, 661, 662, 664-666, 668, 675 scene, 39-43, 70, 71, 74, 75, 79, 82, 88-94, 97-99, 101, 104, 105, 125, 126, 184, 239, 245, 258-262, 264, 265, 353, 378, 382, 384, 387, 389, 391, 398, 492, 510, 512-514, 517-519, 556, 559-561, 564, 605, 606, 686 schema lexical, 534 scheme, 37, 40, 41, 43, 175, 334, 388, 683 science, 592, 649-675 semantic extension, 613, 615, 616, 643 features, 102, 205, 348, 639, overlap, 618, 626, 639 structure, 205, 240, 303, 349, 484, 684 survey, 686 semantics of prepositions, 136 semiotic, 269, 270, 272, 273, 283, 285, 599 sense dominant, 531, 537 sentential demonstratives, 57-60 silence, 68, 608 similarity and dissimilarity, 211, 270, 622, 626, 638, 639, 640, 642, 643 situational meaning, 5, 6, 12, 17, 19, 24 sonority, 66-68 source, 37, 63, 70, 88, 111, 112, 120, 141, 146, 162, 181, 211, 214, 252, 254, 277, 289, 290, 295, 300, 303, 307, 314, 338, 397, 409, 432, 449, 451, 477, 478, 504, 516, 521, 523,

703

558, 566, 573, 585, 602, 603, 606, 610, 649, 652, 654, 660, 661, 663, 669, 671673, 680, 685 Southern Africa, 345, 346 space representation of, 197, 269, 270, 285, 340, 592 Spanish, 12, 332, 395, 401-404, 406-412 spatial arrangements, 272, 553, 565 cognition, 261, 329, 610 deixis, 49, 50, 59, 211, 223, 573 deictics, 59, 60, 329, 338 description, 269, 270, 276, 317, 556, 587 distinctions, 58, 59, 330, 332, 333, 337340 expression, 73, 197, 198, 200, 221, 222, 339 relations, 49, 51, 52, 56, 115, 135, 199, 200, 204-206, 271, 287, 310-313, 315, 316, 329, 337, 557-560, 565-567 speaker's location, 52, 331, 332, 336, 340 stage, 71, 101, 124, 125, 245, 260, 262, 315, 474, 524, 553-558, 562, 564, 567, 635-638, 642, 661, 662, 664 subjectivity, 274, 397, 419, 431, 509-511, 517, 519, 524 Tahitian, 462, 463 -.teshimau, 395-413 temporal, 3, 32, 37, 49, 64, 86, 114, 123127, 129, 135, 146-148, 150-152, 154, 157, 159, 160, 162, 171, 172, 175, 181, 203, 239, 240, 245, 246, 258-260, 265, 270-273, 283, 329, 378, 383, 423, 425, 451, 455, 456, 458, 459, 461-463, 469471, 474, 478, 485, 488, 491, 492, 494497, 499-504, 521-523, 558, 561, 572574, 576, 583, 589, 590, 599, 600 tense-aspect, 173, 245, 246, 265 terminative aspect, 397, 409 text, 8, 33, 106, 183, 184, 186, 199, 220, 244, 248, 249, 259, 260, 262, 263, 296-

704

Subject Index

271, 273, 275, 276, 278, 279, 281, 282, 283, 356, 406, 486, 487, 511, 515, 519, 558, 565, 572-574, 576, 578, 581, 582, 584-586, 588-591, 599, 600, 603, 604, 606, 650, 651, 653, 669, 670, 672 textual deixis, 220, 571-574, 576, 577, 584, 589-591 space, 577, 583, 585, 599-608, 610, 611 theatre, 71, 199, 275, 553-556, 562 theatrical space, 555 Theography, 679, 682, 683, 687, 688 topic pivot, 262 topo-analysis, 680 total enclosure, 395, 396, 405, 413 trajector, 74, 113, 114, 198, 199, 201, 202, 204, 205, 222, 244, 288, 290, 291, 378, 381, 382, 460, 468, 469, 473, 474, 477, 480, 482, 483, 528, 532, 535 transitive, 73, 98-100, 102, 110, 1 ΜΙ 17, 120-122, 124-130, 299, 401, 403405, 410, 470, 479

transitivity, 73, 99, 102, 111, 115, 129, 130, 400, 401, 470 transparency, 591, 592, 603 über, 73-106, 109-130, 390 uncertainty, 131, 142, 174, 177, 181, 517 unidirectionality hypothesis, 676 urban-rural distinction, 57 vagueness, 144, 289, 290, 531, 535, 641-643 vantage point, 227, 229-231, 244, 329, 336, 510, 514, 517 variable prefix, 109, 111, 129 verticality, 85, 376, 385, 392, 604, 605, 611, 654, 680, 682, 683, 684, 686 viewpoint, 51, 77, 510, 511, 513-515, 517519, 523, 524, 531, 601 whisper, 69, 71, 447 word order, 58, 273, 275, 509, 512, 513, 518, 519, 524 za-, 482-489 Zeigfeld, 37, 40