Niemeier-putz - Theory & Language-acquisition.pdf

Applied Cognitive Linguistics I: Theory and Language Acquisition

W G DE

Cognitive Linguistics Research 19.1

Editors René Dirven Ronald W. Langacker John R. Taylor

Mouton de Gruyter Berlin · New York

Applied Cognitive Linguistics I: Theory and Language Acquisition

Edited by Martin Pütz Susanne Niemeier René Dirven

Mouton de Gruyter Berlin · New York 2001

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

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

Library of Congress Cataloging-in-Publication

Data

Applied cognitive linguistics / edited by Martin Pütz, Susanne Niemeier, René Dirven. p. cm. - (Cognitive linguistics research ; 19) Contents: 1. Theory and language acquisition - 2. Language pedagogy. ISBN 3110172216 (v. 1 : alk. paper) - ISBN 3110172224 (v. 2 : alk. paper) 1. Cognitive grammar. 2. Language acquisition. 3. Language and languages-Study and teaching. I. Pütz, Martin, 1955- II. Niemeier, Susanne, 1960- III. Dirven, René. IV. Series. P165 .A66 2001 418-dc21 2001044895

Die Deutsche Bibliothek — Cataloging-in-Publication

Data

Applied cognitive linguistics / ed. by Martin Pütz .... - Berlin ; New York : Mouton de Gruyter (Cognitive linguistics research ; 19) 1. Theory and language acquisition. - 2001 ISBN 3-11-017221-6

© Copyright 2001 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Printing: WB-Druck, Rieden/Allgäu Binding: Lüderitz & Bauer, Berlin Printed in Germany

Acknowledgements Most of the present contributions were originally presented at the 29th International LAUD Symposium held in Landau, Germany, from March 25-28, 2000. All were selected for inclusion only after a lengthy process of refereeing and, in some cases, extensive revising. Their work resulted in the publication of two volumes: Applied Cognitive Linguistics. Vol. I. Theory and Language Acquisition Applied Cognitive Linguistics. Vol. II. Language Pedagogy While compiling the two volumes we have incurred a number of debts that we wish to acknowledge. We are deeply indebted to two of the editors of the book series Cognitive Linguistics Research: Ronald W. Langacker and John R. Taylor, whose support was crucial to the emergence of the two volumes. We would like to thank the organizing staff of the symposium, in particular Susanne Heid, Alexander Kraft, and Caria M. Sandy as well as Heike Ramsauer for proofreading major parts of the manuscript. Thanks are also due to Anke Beck and Birgit Sievert (Mouton de Gruyter) for their kind assistance and cooperation with this venture. 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, we would also like to express a great debt of gratitude to the expertise of the many scholars who acted as our referees: Angeliki Athanasiadou, Frank Boers, Willis Edmondson, Carlos Inchaurralde, Dirk Geeraerts, Stefan Gries, Peter Grundy, Juliane House, Bernd Kortmann, Penny Lee, Lienhard Legenhausen, Bert Peeters, Mechthild Reh, Sally Rice, Francisco Ruiz de Mendoza, Doris Schönefeld, Rainer Schulze, Elzbieta Tabakowska, Jef Verschueren, Marjolijn Verspoor, Helmut Vollmer, Michael Wendt, Karin Wenz.

vi

Acknowledgements

Above all, we want to thank Birgit Smieja, who did a marvelous job in designing the layout of the book and in taking care of the laser printout.

The Editors Duisburg, Bremen, and Landau

July 2001

List of Contributors Jenny Cook-Gumperz University of California, Santa Barbara, U.S. A. Vyvyan Evans Georgetown University, Washington, D.C., U.S.A. CliffGoddard University of New England, Annidale, Australia Paul J. Hopper Carnegie Mellon University, Pittsburgh, U.S.A. Istvan Kecskes State University of New York at Albany, U.S.A. Amy Kyratzis University of California, Santa Barbara, U.S.A. Sydney M. Lamb Houston, Tx., U.S.A. Rice University, Ronald W. of Langacker University California, San Diego, U.S.A. Katharina J. Rohlfing University of Bielefeld, Germany Sang Hwan Seong University of Bonn, Germany Andrea Tyler Georgetown University, Washington, D.C., U.S.A.

Contents Acknowledgements

ν

List of Contributors

vii

Introduction Martin Pütz, René Dirven and Susanne Niemeier

xiii

Section 1: Cognitive approaches to the English tense system Cognitive linguistics, language pedagogy, and the English present tense Ronald W. Langacker Pretend play: trial ground for the simple present Jenny Cook-Gumperz and Amy Kyratzis The relation between experience, conceptual structure and meaning: non-temporal uses of tense and language teaching Andrea Tyler and Vyvyan Evans

3

41

63

Section 2: Facets of prototypes in grammatical constructions Grammatical constructions and their discourse origins: prototype or family resemblance? Paul J. Hopper

109

Transitivity parameter and prominence typology: a cross-linguistic study Sang Hwan Seong

131

χ

Contents

Section 3: Neurocognitive and cognitive issues of language acquisition in general Learning syntax - a neurocognitive approach Sydney M. Lamb Conceptual primes in early language development Cliff Goddard No preposition required. The role of prepositions for the understanding of spatial relations in language acquisition Katharina J. Rohlfing The 'Graded Salience Hypothesis' in second language acquisition Istvan Kecskes Subject Index

Contents of volume II Acknowledgements

ν

List of Contributors

vii

Introduction Martin Pütz, Susanne Niemeier and René Dirven

xiii

Section 1: Bottom-up approaches: Phrasal verbs and phraseological expressions English phrasal verbs: theory and didactic application René Dirven

3

Teaching English phrasal verbs: a cognitive approach Andrzej Kurtyka

29

A usage-based approach to modeling and teaching the phrasal lexicon Kurt Queller

55

Section 2: Top-down approaches: Metaphor and idiom study A cognitive linguistic view of learning idioms in an FLT context Zoltán Kövecses

87

On the systematic contrastive analysis of conceptual metaphors: case studies and proposed methodology 117 Antonio Barcelona

xii

Contents

Section 3: Systematical order instead of chaos in morphology and lexis A conceptual analysis of English -er nomináis Klaus-Uwe Panther and Linda Thornburg Basicness and conceptual hierarchies in foreign language learning: a corpus-based study Friedrich Ungerer

149

201

Section 4: Cultural models in education The African cultural model of community in English language instruction in Cameroon: the need for more systematicity Hans-Georg Wolf and Augustin Simo Bobda

225

Subject Index

261

Introduction Martin Pütz, René Dirven and Susanne Niemeier

The main title of the two volumes, i.e. Applied Cognitive Linguistics, not only focuses on the theory of cognitive linguistics as it can be applied to the teaching and learning of foreign languages; it also encompasses psycholinguistic models and theories with a focus on first language acquisition. Thus, Volume I is concerned with the interaction between language, cognition and acquisition in general (first and second language acquisition), and Volume II addresses a series of cognitive principles of linguistic, i.e. conceptual organization while acquiring, learning and teaching second or foreign languages. The two volumes address a number of important topics in the theory, acquisition and pedagogy of languages seen from the perspective of cognitive linguistics. We cannot discuss here in detail the main principles, aims and findings of the cognitive enterprise, but would nevertheless like to refer to the following assumptions and implications of the discipline of cognitive linguistics as set out by RudzkaOstyn (1993: 1): 1. As one domain of human cognition, language is intimately linked with other cognitive domains and as such mirrors the interplay of psychological, cultural, social, ecological, and other factors. 2. Linguistic structure depends on (and itself influences) conceptualization, the latter being conditioned by our experience of ourselves, the external world and our relation to that world. 3. Grammar is motivated by semantic considerations. 4. Language units are subject to categorization which commonly gives rise to prototype-based networks; much of it critically involves metaphor and metonymy.

xiv

Martin Pütz, René Dirven and Susanne Niemeier

5. Given the interaction among language subcomponents as well as the interaction between language and other domains of cognition, the various autonomy theses and dichotomies proposed in the linguistic literature have to be abandoned; a strict separation of syntax, morphology and lexicon is untenable; furthermore, it is impossible to separate linguistic knowledge from extra-linguistic knowledge. As a dynamic usage-based model (Langacker 2000: 9Iff.), cognitive linguistics is predestined to have an impact on applied research in such areas as language acquisition and language pedagogy, and, we may add, on the various facets of ideology and society as such. Although previous studies have already dealt with child language acquisition from a cognitive linguistics perspective (cf. Slobin 1985, Sinha et al. 1994, Niemeier and Achard 2000) as well as with the pedagogical implications of cognitive linguistics (Dirven 1989, Dirven and Taylor 1994, Taylor 1987, 1993), the Landau Symposium was the first to concentrate on "Applied Cognitive Linguistics" at a more principled and encompassing level. Even if cognitive linguistics is a rapidly expanding linguistic paradigm, the impact of this new and revolutionary linguistic theory on various branches of applied research and on their pedagogical implications is only now beginning to be felt. The two volumes provide a systematic attempt to carve out pathways from the links between language and cognition to the fields of language acquisition and language pedagogy, and to deal with them in one coherent framework called Applied Cognitive Linguistics. Nevertheless, the editors are open to critical, alternative proposals and invite critical reflections from authors not agreeing with the mainstream CL approach. The papers in the present volume have been arranged according to three major strings, namely (i) cognitive approaches to the English tense system (Langacker, Cook-Gumperz and Kyratzis, Tyler and Evans); (ii) facets of prototypes in grammatical constructions (Hopper, Seong); and (iii) neurocognitive and cognitive issues of language acquisition in general (Lamb, Goddard, Rohlfing, Kecskes).

Introduction

xv

The first three contributions deal with an account of the use of the English tense system seen from various perspectives. In his paper "Cognitive linguistics, language pedagogy, and the English present tense" Ronald W. Langacker examines the English present tense, as well as related phenomena like the progressive and the perfect/imperfect contrast, from the perspective of his well-known theory of cognitive grammar. Although Langacker does not specifically focus on issues of second language acquisition as such, he refers to several pedagogical implications of cognitive linguistic theory and possible directions for further research. Some of these directions follow from the usage-based nature of cognitive grammar, in which linguistic units are seen as being abstracted from usage events. Given the fact that regular constructions of full generality constitute only a small proportion of conventional patterns, Langacker assumes that complete mastery of linguistic rules does not assure any degree of actual fluency in a language. Furthermore, Langacker underlines the importance of the communicative, social and cultural context of any speech encounter and refers to the notion of 'construal', i.e. "our ability to conceive and portray the same situation in alternate ways". In conclusion, Langacker criticizes the traditional way of looking at tense mainly on the grounds that these approaches have been objectivist in nature. They lack an awareness of 'construal' and viewing arrangements and the subjective basis of factors like homogeneity and bounding. These are the two major viewing arrangements that we impose on all phenomena. In the domain of things we can impose a homogeneous view and conceptualize phenomena as homogeneous, unbounded substances, or else as bounded objects with an internal heterogeneous structure. This distinction corresponds in the language to that between mass nouns and count nouns. Similarly in the domain of processes, we impose a homogeneous arrangement and conceptualize homogeneous states or activities, or else we impose a bounded view and conceive of processes as bounded events. This distinction corresponds in the verb system to imperfective tense forms (progressive or habitual form) or to perfective, i.e. bounded tense forms. This major insight may stimulate many language pedagogues and psycholinguists to rethink their entire language pedagogy or acquisition

xvi

Martin Pütz, René Dirven and Susanne Niemeier

views. Although Langacker does not draw specific conclusions about teaching, he argues that a clear understanding of its conceptual import should at least allow pedagogical issues to be formulated in the proper terms. Some of the essential features of Langacker's analysis of the English present tense, e.g. the notions of 'construal' and 'viewing arrangement' are taken up by Jenny Cook-Gumperz and Amy Kyratzis in their article entitled "Pretend play: trial ground for the simple present", in which they illustrate Langacker's theoretical distinctions Their research pertains to the relationship between the conceptualizes and the situation being viewed, and how it enables participants and readers/hearers to shift perspectives on the story events. According to the authors, in storytelling a viewing arrangement is sometimes imposed as a self-conscious yet suspenseful perspective. Thus young children moving around with their toy cars or planes and verbalizing the events in which they are involved use the simple present to express the quick succession of events which they are bringing about themselves. A major argument of the paper is that English speaking children's early understanding of the tense system is a key prerequisite for literacy. In order to attain this goal, children's engagement in pretend play as well as in story telling and reading experiences with adults seems to be indispensable. The authors cite various other examples from children's books to illustrate the kinds of narrative experiences young children participate in before they learn to read for themselves. Their data are based on three studies, each focusing on the speech of little children (aged 3-7) who are engaged in collaborative character play with peers. In the center of the discussion is the use of the present tense for temporally-sequenced action and for habitual state, as well as present progressive for ongoing action in in-role play with 3 and 4 year olds. The authors arrive at the conclusion that through tense/aspect contrasts, children's stories are able to provide more nuanced views of the characters and their situation, thereby shifting perspectives and viewpoints on events. Uses of tense are also discussed by Andrea Tyler and Vyvyan Evans in their paper "The relation between experience, conceptual

Introduction

xvii

structure and meaning: Non temporal uses of tense and language teaching". While Cook-Gumperz and Kyratzis put much emphasis on the use of tense to shape different types of processes in terms of viewing arrangements, Tyler and Evans concentrate on the temporal function of tense, especially on the so-called exceptional, nontemporal uses of the present tense, which do not signal timereference but serve various other conceptual functions such as the expression of intimacy, salience, actuality, and attenuation. Tyler and Evans reject the traditional view that such non-temporal usages are arbitrary exceptions; rather they consider them as highly motivated uses, deriving directly from how we conceptualize time. The authors do not see these uses as metaphorical extensions of the temporal functions of tense, either. Instead they share a view of 'experiential correlations' that is also proposed by Joe Grady and Pinker. While perceiving processes in time, we undergo many other experiences, too. One set of these are spatial experiences. We associate these with each other and conceive of them as being 'close' or 'distal'. This higher principle can be invoked for both the temporal and nontemporal uses of tenses. What is close is that which is present here or can be intimate to us, or in the center of attention and actuality. Here present tenses come in. What is distal is non-present or irrealis, or to be softened or attenuated. The authors argue that the so-called 'exceptional' non-temporal uses of English tense is an area of grammar often either ignored altogether or treated as arbitrary by pedagogical grammars and language teachers. The main thesis in the paper is, as just shown, that the non-temporal uses of tense are related to its timereference function in a fully motivated way. Central to the discussion is the process of meaning extension or pragmatic strengthening, by which a conventional non-temporal meaning can become associated with a particular tense morpheme. Their rather elaborate theoretical and practical underpinnings of their view illustrate convincingly that by assuming the principles and methodology of cognitive linguistics it is possible to relate the non-temporal senses associated with tense and the time-reference function to each other in a plausible way. They thus help to solve the difficulty of language teachers trying to present insightfully the non-temporal uses associated with tense, i.e.

xviii

Martin Pütz, René Dirven and Susanne Niemeier

to explain to the learner why the foreign language should be as it is and to bring to the learner's consciousness the conceptualizations conventionally associated with the structures of the foreign language. The notion of 'exceptional uses' of grammar is also reflected in Paul J. Hopper's paper "Grammatical constructions and their discourse origins: Prototype or family resemblance?" Whereas Tyler and Evans emphasize the motivating force pertaining to so-called 'exceptional', peripheral uses of the present tense and drift away from a view of a prototypical core and metaphorical extensions, Hopper also takes up the implied notion of prototypicality but seriously calls into doubt its validity as suggested by many, though not all, proponents of cognitive linguistics, and also by many other scientists.1 The author concentrates on one complex construction, the English pseudocleft sentence, and discusses many examples taken from natural discourse in which such a pseudocleft sentence appears in a fragmentary 'incomplete' form. The author's view is that pseudocleft sentences do not primarily serve to highlight any single identifiable sentence constituent, but that this highlighting function is a mere side-effect. The real function of pseudocleft sentences is rather to delay an assertion for a number of pragmatic reasons. One of those may be the need to create more processing time. This also explains why so many pseudoclefts are broken off in mid-plan: if the processing problem has been solved, the construction can just be discarded. Thus the author shows that there is a need to revise the prototype model in grammatical construction and instead suggests adopting the family resemblance model. In such a view many different uses and meanings can be vaguely related to one another without any of them being an extension from a core use. 'Canonical' constructions should be seen as "highly stylized cultural artifacts ... put together and grammaticalized through stylistic and normative conventions" and not as prototypes or as the source of 'deviant' fragmentary instantiations in discourse. Hopper thus also indirectly criticizes mainstream cognitive linguistics in the sense that 'schemata' based on prototypical samples of language (i.e. planned discourse) often do not reflect the observable realities of language use in natural discourse. This observation, Hopper concludes, has important consequences for applied linguistics and

Introduction

χίχ

foreign language teaching. In general, teachers should pay attention to the role of "incomplete" non-prototypical utterances as the "building blocks of fluent and appropriate oral discourse". An interesting view of prototype theory is also proposed by Sang Hwan Seong, who analyses the relationships of the "Transitivity parameter and prominency typology" in the light of recent developments in cognitive grammar. Seong critically evaluates previous approaches dealing with particular linguistic schemata and prototypes organized in and across languages such as English, German and Korean. It is traditionally argued that active clauses high in transitivity typically undergo passivization in which agentive subjects, totally affected objects and action predicates are correlated. Seong claims that, on the contrary, the transitivity and prototype parameter should be constrained such that the languages with high semantic transparency and a material case-marking system such as Korean proportionally correlate with less grammaticization, e.g. they do not require overt subjects. Seong also suggests that the grammatical properties of a given language are closely linked to the presence/absence of the built-in subjective reference point. The effects of this theoretical position are also reflected in L2 data based on an analysis of the writing samples of German and Korean students. It was found, for example, that German learners of Korean tend to produce, on the basis of their German templates, sentences in Korean with overt subjects almost without any exception early on and then gradually learn to drop these subjects when they reach a higher level of proficiency. It was found, furthermore, that typical Korean phenomena such as topic particles and double nominative constructions are particularly difficult features for Germans to acquire, even in the later stage of learning. It is, in the editors' view, not exaggerated to conclude that prototypicality is a highly language-specific phenomenon, which therefore does not show up on a universal basis. A neurocognitive perspective of learning syntax is advocated by Sydney M. Lamb in his article "Learning syntax - a neurocognitive approach". Lamb asserts that the two approaches to language, i.e. the neurocognitive one and the analytical one, the latter including for him cognitive linguistics, have different concerns from each other. The

XX

Martin Pütz, René Dirven and Susanne Niemeier

distinction is based on the admittedly disputable claim that analytical linguistics is concerned only with the analysis of linguistic data without any reference to a cognitive and neurological basis, while neurocognitive linguistics takes as the object of its cognitive study the neurocognitive system of the individual.2 More particularly, Lamb criticizes the traditional view that the child or adult learning a second language must somehow acquire syntactic rules which the speaker has internalized. Contrary to this assumption, Lamb proposes an alternative view which holds that syntactic information is somehow attached to individual lexical items. According to such a view, the acquisition of syntactic knowledge is "part and parcel of the acquisition of lexical knowledge and therefore occurs little by little as individual lexemes are learned". Learning consists mainly of building and strengthening connections between synapses in the brain and adjusting thresholds of nodes. In the case of the child learning lexical items, it is commonly the case that conceptual connections are already present before they get connected to their linguistic expressions. Lamb illustrates this point with words like crawl and eat. For example, by the time a child learns these words, it has long since been crawling without a crawler - and it is aware that eating has two participants, since without eater and food there can be no eating. So the child must learn a phonological expression plus the relative ordering of the already known participants. Whereas Lamb's paper is an attempt to put forward a neurocognitive approach to learning syntax which holds for both first and second language acquisition, Cliff Goddard takes a stand on first language acquisition with a clear focus on children's semantic and conceptual development. The author is known to be one of the main proponents of the natural semantic metalanguage (NSM) approach originated by Anna Wierzbicka. To compare meanings expressed in different languages and different cultures, one needs to establish a semantic metalanguage. This metalanguage is the set of semantic 'atoms' or primitives which cannot be further reduced or made simpler. These primitives are also claimed to suffice to define or explain all other terms in all possible languages. Since the semantic primes are valid in all languages, they are by definition universal. These univer-

Introduction

xxi

sal semantic primitives are carved out of natural language and also have an inherent 'conceptual syntax'. Goddard's paper "Conceptual Primes in Early Language Development" tackles the problem that young children in their acquisition process do not start off with the seventy odd semantic primitives, but with more complex concepts. The author therefore explores various hypotheses about the nature and identities of the innate concepts which may underpin language acquisition. Since many of the child' s early words are more complex and cannot be explained in terms of conceptual primes, a possible explanation might be that these words have simpler meanings than the corresponding words in an adult's vocabulary. Goddard concludes from this that the child must have a certain 'conceptual vocabulary' of prime concepts prior to their emergence in language proper. In the remaining part of the paper Goddard engages in a detailed semantic analysis of the child's production vocabulary (i.e. the non-prime words and utterances) in order to investigate the conceptual primes which may be 'latent' in a child's early lexicon. An interesting observation refers to the fact that the production vocabulary of primes lags several months behind the conceptual vocabulary. As Goddard's study is based on the language development of one single child, i.e. Goddard's son Peter, it would be interesting to see to what extent Peter's acquisition story is representative for other children's production vocabulary and their acquisition of primes in the conceptual vocabulary. Whereas Goddard in general investigates the development of the linguistic and conceptual vocabulary of child language, Katharina J. Rohlfing's paper "No preposition required. The role of prepositions for the understanding of spatial relations in language acquisition" is a report on experiments focusing on the role of one single syntactic category, i.e. prepositions, for the understanding of spatial relations also in first language acquisition. Rohlfing presents the results of an empirical study of Polish-speaking children at the age of 2. The hypothesis of her study was that, in prototype situations, children understand instructions even without prepositions, which however they need in order to understand the instructions in abstract situations. The experiments show that infants rely not only on the appropriate loca-

xxii Martin Pütz, René Dirven and Susanne Niemeier

tive prepositions, but also on non-linguistic strategies, such as an object's relational character, its physical properties and the typicality of a situation. Comparing her work with Goddard's, we can come to an interesting hypothesis: the child may have a number of conceptual categories for which (s)he does not yet have linguistic expressions (Rohlfing) or preliminary labels from adult language (Goddard). The prepositions themselves are acquired much later than their supposed understanding indicates. Before the full acquisition of prepositions, infants are able to master the basic relations "in" and "on" with the help of non-linguistic strategies, and a linguistic instruction cannot be understood context-independently at this age. Rohlfing concludes that conceptual knowledge should be explored more task- and context-dependently, "because the non-linguistic strategies are a part of the infants' understanding and they establish the schematic meaning of spatial relations". Whereas Rohlfing's psycholinguistic study examines the role of prepositions for understanding instructions with spatial relations in first language acquisition, Istvan Kecskes considers both first and second language acquisition, although with a clear focus on foreign language learning. In his paper "The 'Graded Salience Hypothesis' in second language acquisition" the author discusses the validity of the 'Graded Salience Hypothesis' (GSH), which was developed to explain LI processing. The GSH claims that figurative and literal language use are governed by a general principle of salience in LI : Salient meanings (conventional, frequent, familiar) are processed first, and parallel processing is induced when more than one meaning is salient. Based on an empirical cross-sectional study, the author shows that its application in second language acquisition can also be very useful in the sense that salience is language- and culture-specific and that it derives from certain knowledge structures that are essential parts of the conceptual base of native speakers. When acquiring a non-primary language, students have to learn not only the forms of that particular language, but also the conceptual structures associated with those forms. Salience is a very important part of the sociocultural heritage of native speakers, and it is something that reveals a unique feature of the human mind.

Introduction xxiii

Notes 1. Problems of the prototype approach are also pointed out in CL, e.g. by Geeraerts (1988). On the other hand, Hopper and Thompson (1985) themselves invoke prototype theory for typology. Without calling himself a cognitive linguist, Kastovsky (1988) opposes prototype semantics as a superior alternative to structural semantics. One example out of the hundreds of non-cognitive linguists working with prototype ideas is Shaver et al. (1987). 2. Peeters (2001) devotes a whole paper to the - in his view - misnomer 'cognitive linguistics' for the CL paradigm, which he believes should call itself 'Cognitive Linguistics' with upper case, leaving the lower case appellation 'cognitive linguistics' to its inventor, i.e. Sydney Lamb.

References Dirven, René 1989

Cognitive linguistics and pedagogic grammar. In: Gerhard Leitner and Gottfried Graustein (eds.), Linguistic Theorizing and Grammar Writing, 56-75. Tübingen: Niemeyer. Dirven, René and John R. Taylor 1994 English modality: A cognitive-didactic approach. In: Keith CarIon, Kristin Davidse and Brygida Rudzka-Ostyn (eds.), Perspectives on English. Studies in Honour of Professor Emma Vorlat, 542-556. Leuven: Peeters. Geeraerts, Dirk 1988 Prototypicality as a prototypical notion. Communication & Cognition 21: 343-355. Hopper, Paul J. and Sandra A. Thompson 1985 The uses of the notion 'prototype' in typological studies. In: H.-J. Seiler and G. Brettschneider (eds.), Language Invariants and Mental Operations, 238-244. Tübingen: Gunter Narr. Kastovsky, Dieter 1988 Structural semantics or prototype semantics? The evidence of word-formation. In: Werner Hüllen and Rainer Schulze (eds.), Understanding the Lexicon. Meaning, Sense and World Knowledge in Lexical Semantics, 190-203. Tübingen: Niemeier. Langacker, Ronald W. 2000 Grammar and Conceptualization. Berlin/New York: Mouton de Gruyter.

xxiv Martin Pütz, René Dirven and Susanne Niemeier Niemeier, Susanne and Michel Achard (eds.) 2000 Cognitive Linguistics: Special issue on language acquisition. Cognitive Linguistics 11-1/2, 1-151. Berlin/New York: Mouton de Gruyter. Peeters, Bert 2001 Does Cognitive Linguistics live up to its name? In: René Dirven, Bruce Hawkins and Esra Sandikcioglu (eds.), Language and Ideology. Volume 1 : Theoretical Cognitive Approaches. (Current Issues in Linguistic Theory 204), 83-106. Amsterdam: John Benjamins. Rudzka-Ostyn, Brygida 1993 Introduction. In: Richard A. Geiger and Brygida Rudzka-Ostyn (eds.), Conceptualizations and Mental Processing in Language. A Selection of Papers from the First International Cognitive Linguistics Conference in Duisburg, 1989, 1-20. Berlin: Mouton de Gruyter. Shaver, P., J. Schwartz, D. Kirson and C. O'Connor 1987 Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology 52: 1061-1086. Sinha, Chris, Lis A. Thorseng et al. 1994 Comparative spatial semantics and language acquisition: Evidence from Danish, English and Japanese. Journal of Semantics 11:253-287. Slobin, Dan I. (ed.) 1985 The Crosslinguistic Study of Language Acquisition. Vol. 2: Theoretical Issues. Hillsdale, N.J.: Lawrence Erlbaum. Taylor, John R. 1987 Metaphors of communication and the nature of listening and reading comprehension interface: Journal of Applied Linguistics 1: 119-134 1993 Some pedagogical implications of cognitive linguistics. In: Richard A. Geiger and Brygida Rudzka-Ostyn (eds.), Conceptualizations and Mental Processing in Language. A Selection of Papers from the First International Cognitive Linguistics Conference in Duisburg, 1989,201-223. Berlin: Mouton de Gruyter.

Section 1 Cognitive approaches to the English tense system

Cognitive linguistics, language pedagogy, and the English present tense Ronald W. Langacker

1. Introduction Although I cannot claim any experience or substantial knowledge in the area of language pedagogy, I do recognize how essential it is. There seems little doubt that, in the broad field of language and linguistics, the applied, pedagogical side outstrips the theoretical side in terms of intrinsic importance, numbers of practitioners, and the securing of institutional tolerance. The applied folk are, as it were, carrying linguistic theoreticians on their back. Not only that, but the burden tends to be heavy, onerous, and quite unappreciative. Perhaps cognitive linguistics will prove to be lighter, less onerous, and more appreciative than certain previous theoretical burdens. I hope it will even prove useful. Let me try to contribute by sketching some pedagogical implications of cognitive linguistic theory, and some possible directions of research. I see the effectiveness of pedagogical applications as an important empirical test for linguistic theories. My suspicion is that, in the long run, cognitive grammar will not fare badly in this regard.

2. Implications of a usage-based model Cognitive grammar is a usage-based model of linguistic structure, in which linguistic units are seen as being abstracted from usage events by the reinforcement of recurring commonalties (Langacker 1988, 2000). Such units run the gamut from the fully specific to the maximally schematic, with specific structures and local regularities being

4

Ronald W. Langacker

at least as important as high-level generalizations. Moreover, special cases of general patterns can themselves have the status of linguistic units, provided that they are learned by speakers and conventional in a speech community. Since linguistic units arise by abstraction from usage events, a usage-based approach is necessarily a construction-based approach. That is, constructions are basic rather than epiphenomenal, and rules are nothing more than schematic constructions (cf. Goldberg 1995). A construction is merely an assembly of symbolic structures (formmeaning pairings). Networks of constructions are deemed sufficient for a full description of lexicon and grammar, which form a continuum. The usage-based and construction-based nature of cognitive grammar (and other approaches to cognitive linguistics) has certain implications for language pedagogy. One of them is the limited importance of fully general rules. Since regular constructions of full generality constitute only a small proportion of conventional patterns, complete mastery of linguistic rules (as normally conceived) does not assure any degree of actual fluency in a language. I can personally attest that, by thoroughly learning all the rules and vocabulary found in traditional textbooks and taught in traditional language classes, one does not come even close to being fluent in a language. To achieve fluency, one has to learn in addition a vast store of fixed expressions and normal ways of phrasing things in particular circumstances, out of all the ways the "rules" in principle permit. Only by controlling this immense inventory of conventional expressions and conventional modes of expression is it possible for speakers to put together a continuous flow of complex expressions in real time (Langacker 1987a: 35-36). If everything had to be computed from scratch, from general grammatical rules and traditionally recognized vocabulary, the exigencies of rapid, fluent speech would overwhelm our processing capacities. This is the state to which I was brought by traditional language training. I have often said - and I think fairly accurately - that lexicon and grammar as traditionally conceived and taught constitute only around 1% of the linguistic knowledge required for fluent speech, yet this

Cognitive linguistics, language pedagogy, and the English present tense

5

1% attracts around 99% of the effort and attention of linguists. The usage-based and constructional perspectives encourage a reallocation of effort that should bring these proportions more into balance. If it is accurate, my assessment of the discrepancy between what linguists focus on and what speakers need to know will not come as any great surprise to those engaged in language pedagogy. It should be evident to anyone confronted with real-life problems of language teaching. Nor will I suggest any specific pedagogical techniques or strategies these are better left to the experts. But perhaps this new theoretical perspective can support such initiatives by elevating the perceived status of conventional expressions and facilitating their description. Another facet of a usage-based model concerns the role of the communicative, social, and cultural context. In particular, cognitive grammar is contextually grounded because all linguistic units are abstracted from usage events, comprising the full contextual understandings of socially engaged interlocutors with specific communicative objectives in connected discourse. Any recurring facets of such events can be incorporated in the conventional meanings of linguistic units. It is only via progressive abstraction or decontextualization that linguistic units sometimes approximate the situation of having context-independent values. Yet I think they never fully achieve it. I believe that every expression, fixed or novel, is inevitably interpreted with respect to some presupposed context, if only a vanilla context derived from default assumptions. Hence the usage-based perspective provides a theoretical underpinning for what we all know in practical terms, namely the essential role of context and culture in language understanding and language learning. Let me illustrate with a banal example from my own experience. The event took place on a train in France in 1963. I was trying to hang up my jacket, but was having trouble due to the absence of the loop of fabric normally provided for that purpose (this notion appears to be a lexical gap in both English and French). One of the two French women in the compartment summarized the situation with the sentence in (1), which I remember precisely to this very day: (1)

Il n'y a pas de petit cordon.

6

Ronald W. Langacker

Without my telling you the context, you could probably not determine the specific import of (1) or of the phrase petit cordon. Conversely, prior to the utterance - having been trained in French in a classroom setting by traditional methods - I had no clue whatever about how to describe the situation or the loop of fabric. Yet this one utterance, immediately interpretable in context, was sufficient for me to learn a natural way of expressing this situation in French. (Admittedly, I have never since found myself in a situation where this knowledge was useful, but if it ever happens again I am ready.) Once more, I will not presume to make specific pedagogical recommendations. Let me simply state the obvious, namely that these considerations argue for pedagogical approaches which emphasize the interactive exposure to large quantities of natural speech in context.

3. Construal According to cognitive grammar (Langacker 1987a, 1990, 1991, 1999a), lexicon and grammar form a continuum, all elements of which are symbolic in nature. It follows that all grammatical elements are meaningful. For everyone but certain linguistic theorists, this is certainly a more attractive vision than the standard view of grammar as pointless drudgery arbitrarily imposed, hence very hard to learn. If this vision is correct - and by now I consider it essentially proven - it offers numerous pedagogical opportunities. The key to recognizing the meaningfulness of grammar lies in adopting a conceptualist semantics that properly accommodates construal, i.e. our ability to conceive and portray the same situation in alternate ways (Langacker 1993). As an inherent aspect of its conceptual-semantic value, every lexical and grammatical element incorporates a particular way of construing conceptual content - either its own content or that evoked by other elements. It is by no means an easy matter to correctly and convincingly describe this essential aspect of linguistic meaning, especially given the broad range of units that need to be taught. In principle, however, an accurate appreciation

Cognitive linguistics, language pedagogy, and the English present tense

7

of the construal a unit imposes on a situation should allow one to devise a more effective means of teaching it. Lexical and grammatical elements are conceptual tools, and in learning to use a tool it is helpful to know what it does. Linguistic structures are conceptual tools for imposing particular ways of viewing a situation. Hence their meanings are not directly discernible or objectively present in the situations described - rather they inhere in the cognitive process of apprehending those situations and construing them for expressive purposes. Failure to appreciate this fundamental point is the major reason why so many linguistic theorists do not accept the meaningfulness of basic grammatical constructs. As a case in point, they continue to insist, quite erroneously, that basic grammatical classes (noun, verb, adjective, etc.) are not semantically definable. It is argued, for example, that since a word like escape designates an event (and precisely the same event) whether it functions as a verb or a noun, it must have the same meaning as a member of either class, so verbs and nouns cannot be distinguished semantically. The nominalization of a verb, deriving nouns like performance or explosion, is therefore seen as having no effect on its meaning. Thus statements like (2) still reflect standard linguistic dogma, and (3) represents the only kind of argument advanced to refute the naive view that the parts of speech do have semantic import. (2)

"[N]o constant semantic effect is associated with the functioning of a morpheme as a noun, as a verb, or as any other part of speech." (Langacker 1968: 83)

(3)

"[L]et's ask whether each part of speech really denotes a consistent kind of meaning ... Now it is true that any word that names an object will be a noun. But on the other hand, not every noun names an object. 'Earthquake' names, if anything, an action, as does 'concert'; 'redness' and 'size' name properties; 'place' and 'location' pretty obviously name locations. In fact, for just about any kind of entity we can think of, there exist nouns that name that kind of entity. So the grammatical no-

8

Ronald W. Langacker

tion of noun can't be given a definition in terms of what kind of entity it names ... A particular kind of entity need not correspond to a single part of speech either ... We conclude that parts of speech ... are not definable in terms of meaning." (Jackendoff 1994: 68-69) What is missing from both traditional definitions and the standard refutation is any realization that meaning is not objectively given but reflects our apprehension of situations. Any degree of semantic sensitivity (which I clearly lacked in 1968) should tell us that a change in grammatical class involves a reconceptualization, an alternate construal of the same content. Rather than being semantically vacuous, it results in a subtly different meaning, in accordance with the abstract semantic values of the classes. Going from a verb like perform to a noun like performance involves a conceptual reification wherein a process is reconstrued as a kind of thing. Such reification might be characterized in terms of a process being construed metaphorically as a physical object (in the case of count nouns) or a physical substance (for mass nouns). But whatever the details, the standard argument against a notional definition of grammatical classes hinges on the tacit assumption that any characterization based on construal (rather than the objective situation per se) is not even worth contemplating. Numerous facets of construal are usefully thought of as being analogous to visual phenomena. This leads Talmy (1996) to speak of ception as the general process subsuming both perception and conception. In the same spirit, I use the term viewing for both vision in particular and conceptualization more generally, to the extent that they seem parallel (Langacker 1995). An example is our capacity to construe a situation either schematically or at progressively greater levels of specificity. Conceptually, this dimension of construal is reflected in the meanings of a series of expressions like (4): (4)

thing > animal > dog > retriever > golden retriever

Cognitive linguistics, language pedagogy, and the English present tense

9

This increasing semantic specificity seems quite analogous to the visual experience of seeing an object in progressively finer-grained detail the closer we come to it. In any case, we are clearly able to portray a situation at any desired level of specificity or schematicity, and lexical items provide ranges of options for this purpose. Some analogies are diagrammed in Figure 1. In vision, we can distinguish between the maximal field of view (everything visible to any degree at a given moment), within that the general locus of attention (or "onstage region"), and within that the specific focus of attention. The corresponding conceptual entities, adopted as semantic constructs in cognitive grammar, are the maximal scope (i.e. the full array of conceptual content an expression evokes), the immediate scope (the portion directly relevant for a particular purpose), and the profile (the entity the expression is construed as designating, or referring to). Maximal field of view-

Maximal scope (MS)

Focus of attention

Profile

Locus of attention ("onstage region")

Immediate scope (IS)

Figure 1. Viewing analogies

An expression's grammatical class is determined by the nature of its profile (not its overall content). Basic grammatical classes have conceptual characterizations. They invoke particular conceptual archetypes for their prototypical values (e.g. 'physical object' in the case of nouns), and particular cognitive abilities for their schematic descriptions (e.g. conceptual reification). Class membership is therefore neither arbitrary nor objectively determined. Rather, it inheres in a particular way of viewing a situation and construing it for expressive purposes. At the schematic level, a noun profiles a thing, abstractly defined as any product of conceptual reification. A verb profiles a process, defined abstractly as a relationship scanned sequentially in its evolution through time. Various other classes (e.g. adjective, adverb, pre-

10

Ronald W. Langacker

position, participle) profile relationships that are non-processual (thus viewed holistically, rather than scanned sequentially, even when they develop through time). Consider a term like aunt. It is clearly relational, so why does it function as a noun (and not, say, a preposition)? The reason, as shown in Figure 2, is that it does not profile this relationship, but rather the person who bears it to the reference individual (ego). Crucial though it is to the meaning of aunt, the relationship is not its referent but is merely evoked to characterize its referent. (Observe that there is no basis in this particular example for distinguishing between maximal and immediate scope. This is analogous to a word consisting of just a single morpheme, so that word and morpheme coincide.)

ego MS/IS aunt

Figure 2. A relational noun

Once we decide to use it, a form like aunt gives us no options: it is always a noun by virtue of imposing the scope and profiling shown in Figure 2. Despite polysemy and a certain flexibility of interpretation, lexical items are fixed expressions, so basically they embody packaged conceptualizations and their conventional symbolization. Ironically, it is grammar - so often thought of as an oppressor imposing arbitrary limits on our expressive freedom - that gives us some relief from these lexical strictures. Many grammatical elements have the specific function of adjusting the construals that would otherwise be inherited from lexical items. For instance, the verb perform profiles a perfective process, i.e. a process construed as being bounded. In Figure 3 a, the circle represents the focal participant; the vertical line and rectangle stand for the activity this participant is engaged in; and the horizontal line (with endpoints) indicates its progression through time (sequentially

Cognitive linguistics, language pedagogy, and the English present tense

11

viewed). Starting from this basic conception, the derivational morphemes -er and -ance do not add any new content. Rather, their semantic value resides in the construal they impose on that content. Thus -er shifts the profile from the temporally evolving relationship to its focal participant, as seen in Figure 3b. Performer is therefore a noun, since it profiles a thing rather than a process, although the original process provides its essential conceptual content. Performance is also a noun, but the thing it profiles is abstract, created by conceptual reification. The ellipse in Figure 3c stands for this abstract entity, a thing consisting of one instance of the process perform. (a)

perform

(b)

performer

(c)

performance

Figure 3. Profiling and grammatical class

Another example is the English progressive construction with be... -ing. I claim that the progressive is only applicable to processes construed as perfective, hence temporally bounded. It is an imperfectivizing construction, so it does not apply to processes that are already imperfective (not bounded), where its effect would be vacuous (Langacker 1987b). Occurrence in the progressive is of course a standard diagnostic for perfective (or "active") verbs, and non-occurrence for imperfectives (or "statives"). The other standard diagnostic, with the opposite distribution, is occurrence in the "true" present tense (with an actual, "right now" interpretation). Here are some typical examples: (5)

a. * He builds a house. [perfective] b. He knows the truth. [imperfective]

a'. He is building a house, b'. *He is knowing the truth,

12

Ronald W. Langacker

Intuitively, we can describe the progressive in terms of "zooming in" and taking an "internal view" of a bounded event. Technically, I describe it as imposing an immediate temporal scope that excludes the endpoints of the perfective process it applies to. This is shown in Figure 4b, where - since only aspect is relevant here - 1 have reduced the depiction of a process to a line representing its temporal extension. Thus, while the maximal scope is a span of time containing the full, bounded process, the immediate scope subtends only an arbitrary portion of its internal development. Only that portion is profiled, since - as a matter of definition - the profile is the focal point within the immediate scope (see Figure 1). The overall progressive expression is imperfective, because grammatical class is determined by the profile and the profiled process is not bounded. (Also, as with any imperfective, the profiled process is construed as being effectively homogeneous.) Perfective

Progressive

(b)

Imperfective MS

IS

t Figure 4. Basic aspectual classes

Because it only applies to perfectives, the progressive construction though itself imperfective - signals that the original process is construed as being bounded. The subtle contrast between a basic imperfective and one derived by using the progressive is seen in Figures (4b-c). Both profile a process which is unbounded within the immediate temporal scope (and construed as being effectively homogeneous). The difference is that a progressive expression creates this imperfective process by selectively attending to the interior of an overall occurrence recognized as being bounded. Thus, in (6a), both the simple present and the progressive indicate a current residence in Chicago, but the latter portrays this as part of an overall residential episode of limited duration. Likewise, with the past tense (6b) merely describes the situation as stable and unbounded within the time span

Cognitive linguistics, language pedagogy, and the English present tense

13

in question {in that era), whereas the progressive construes it in relation to a longer period within which the rule of dinosaurs is seen as being bounded. And in (6c), the progressive indicates that the statue's location in the plaza is only temporary. (6)

a. I {live/am living} in Chicago. b. In that era, dinosaurs {ruled/were ruling} the earth. c. A statue of Martin Pütz {stands/is standing} in the plaza.

It should go without saying that whether a process counts as being bounded is subject to construal. I say this anyway because I have often had linguists - even cognitive linguists - object to my characterization of perfectives by citing examples where bounding is not apparent just by examining the objective circumstances. One such example is (7). Since running around a pole is something that can go on indefinitely, with no intrinsic endpoint, how can it be said that the process is bounded? (7)

He is running around the pole.

The mistake here lies in assuming that the bounding of a process has to be objectively given, with an inherent endpoint observable in the situation itself. Ultimately, what counts for linguistic purposes is whether a process is conceptualized as some kind of bounded episode, irrespective of whether a natural endpoint is discernible. There is in general a strong tendency to conceptualize force-dynamic occurrences (those requiring the expenditure of energy - cf. Talmy 1988) as being bounded in duration, even when the process is internally homogeneous and nothing appears to be going on. Thus (8a) describes a stable situation, but since the stability results from a balance of opposing forces, the basic process (i.e. the dam contains the surging floodwaters) is construed as an episode of bounded duration, hence the progressive is possible. By contrast, in (8b) the basic process (i.e. the barrel contains water) is non-force-dynamic. It merely describes a spatial configuration, which as such can maintain itself

14

Ronald W. Langacker

indefinitely. It thus receives an imperfective construal, and consequently does not allow the progressive. (8)

a. The dam is containing the surging floodwaters. b. The barrel contains water.

English has many perfective predicates (e.g. sleep, dream, run, walk, sit, stand, lie, perspire, talk, chat, meditate, wear a tie) in which an internally homogeneous activity is nonetheless construed as occurring in bounded episodes. It is only their grammatical behavior - resisting the simple present, occurrence in the progressive - which alerts us to their perfectivity. This does not however imply that the distinction is "purely grammatical", with no conceptual basis. The grammatical classification hinges on a conceptual factor which is no less real for being subject to construal. Once the specific, subtly contrasting construals imposed by lexical and grammatical elements are elucidated, innumerable structural details that otherwise seem quite arbitrary turn out instead to have an intuitively graspable conceptual basis. The potential thus exists for devising effective ways of teaching them. How much of this should be explicitly taught to students? Should we burden the language learner with technical constructs like profile and immediate scope? I suspect not. It is not even evident that the classroom teacher should be responsible for such technical details, which might better be brought in at the level of overall planning and design of teaching materials. I will leave this matter for those competent to assess it. At the same time, these notions might find a natural place in a language arts curriculum. Generative grammarians have often proposed that linguistics ought to be more visible in school curricula at all levels. In particular, they put it forth as a way of teaching science: formulating hypotheses, testing them against the empirical evidence, developing skills of argumentation, etc. While this may have some merit, the very different vision of language embodied in cognitive linguistics suggests another option: the concepts and descriptions of cognitive linguistics might instead be used for inculcating an appreciation of language as a means of evoking and symbolizing alterna-

Cognitive linguistics, language pedagogy, and the English present tense

15

tive conceptualizations of experience. If language arts training of this sort were routinely available, it might then be possible to draw upon such notions, even explicitly, in the language classroom.

4. Conceptual substrate Traditional semantics focuses on the meanings and combinatory properties of overt elements. Cognitive linguistics has clearly demonstrated the limitations of this approach, showing that linguistic meanings rest on a vast and multifaceted conceptual substrate. The conceptions explicitly encoded by formal elements are merely the "visible" portions of far more extensive conceptual structures that support them and provide their coherence. Often left implicit - yet critical for determining and interpreting what does appear overtly are factors such as the presupposed viewing arrangement, the nature and force of the speaker-hearer interaction, and how expressions relate to the current discourse state. Linguistic understanding further relies on elaborate processes of meaning construction involving metaphor, metonymy, mental spaces, blending, idealized cognitive models, and the evocation of myriad "fictive" entities. Long ago, my own language training emphasized the translation or transformation of individual sentences totally isolated from any context. I thus find it interesting to contemplate how many levels of conceptual organization support the interpretation of even a simple expression. The expression itself - overt linguistic elements and the notions they directly encode - is of course merely the tip of the iceberg. The expression per se is part of a usage event, i.e. an actual instance of language use, comprising the interlocutors' full contextual understanding of the expression, including their apprehension of its interactive force. The usage event is usually part of a longer discourse, and is one facet of the interlocutors' overall social interaction. The interaction takes place in a particular situational context, which in turn is embedded in a culture, which develops as a way of coping with the world.

16

Ronald W. Langacker

We rely on information from any or all of these strata even for the interpretation of seemingly straightforward expressions. Let us once again consider example (1 ), Il n'y a pas de petit cordon. This made perfect sense given the situational context of wanting to hang up a jacket, but without this contextual support it seems rather pointless and hard to interpret. Apprehension of the situational context was itself dependent on the knowledge of certain cultural practices, notably that of hanging up jackets and the usual provision of a loop of fabric for this purpose. Moreover, certain basic properties of the world we inhabit - such as temperature, gravity, and force dynamics - are reflected in these cultural practices and implicitly invoked any time we think of them. A pivotal aspect of the conceptual substrate, quite clearly, is the ground, comprising the speaker, the hearer, their interaction, and the immediate circumstances. It is quite common - arguably even canonical - for the ground to remain offstage and not be mentioned. In the unmarked situation, the ground functions as the tacit location from which a scene is viewed and an expression's meaning is apprehended, as opposed to being onstage as the explicit focus of attention. For instance, a tense marker locates a profiled process with respect to the time of speaking (one facet of the ground), invoking it as a temporal reference point, but does not directly mention it. Likewise, a determiner - via its specification of (in)definiteness - invokes the speaker and hearer as the individuals seeking to identify the nominal referent, but leaves them offstage and unprofiled. The ground, then, is the locus of conception. We do of course have ways of putting facets of the ground onstage and referring to them specifically, e.g. with forms like I, you, here, and now. Even so, the ground's occasional status as focused target of conception coexists with (and is subsidiary to) its more fundamental role as the tacit locus of viewing. Inherent in every usage event is a presupposed viewing arrangement, pertaining to the relationship between the conceptualizes and the situation being viewed. The default arrangement finds the speaker and hearer together in a fixed location, from which they report on actual occurrences in the world around them. There are however numerous kinds of departures from this canonical circumstance. The

Cognitive linguistics, language pedagogy, and the English present tense

17

departures help make it evident that the default arrangement, so easily taken for granted, is nonetheless an essential part of the conceptual substrate supporting the interpretation of expressions. Whether canonical or special, the viewing arrangement has a shaping influence on the conception entertained and consequently on the linguistic structure used to code it. For expressive purposes, we sometimes invoke a viewing arrangement other than the actual one. On the most likely interpretation of (9), for instance, the speaker is actually in motion and the telephone poles are static (cf. Talmy 1996). Yet the situation is presented as if it instantiated the default-case viewing arrangement, with a static viewer observing the motion of other entities. Granted this fictive viewing arrangement, the description is accurate: it represents what the speaker actually sees, interpreted with respect to an arrangement that, while canonical, is non-actual. A correct apprehension of the speaker's intent requires that the expression's overt content be properly related to both the fictive viewing arrangement (which determines its form) and the actual one (where the viewer's motion generates the visual experience coded by that form). (9)

The telephone poles are rushing past at ninety miles per hour.

Of course, language is not just used for describing what happens. Most of us are unlike philosophers (or the idealized speakers sometimes imagined by philosophers) who spend their lives in detached contemplation, producing only objectively verifiable assertions purporting to truthfully describe the world. Beyond assertion and description, we use language for multifarious actions and purposes: for asking questions, giving orders, making promises, performing official acts; for attracting and directing attention; for eliciting approval, agreement, acknowledgment, permission, and cooperation; for the primal expression of pleasure, pain, fear, and disgust; and so on. The apprehension of these actions and purposes provides a conceptual substrate for the attachment of elements appearing overtly in expressions. Whether or not they refer to the substrate specifically, these

18

Ronald W. Langacker

elements - via their selection and their conformation - give witness to its presence and its character. For instance, modifiers sometimes pertain to the substrate rather than the overt content. The since-clause in (10a) does not give a reason for the bird being the kind of bird it is, but instead justifies the speaker's action of requesting this information (Sweetser 1990). Likewise, the honesty referred to in (10b) is that of the speaker in tendering the advice, not that of the addressee in following it. (10) a. Since you're an ornithologist, what kind of bird is that? b. In all honesty, you should give up painting and get a job. Part of the substrate resides in apprehending an expression's illocutionary force, i.e. the type of speech act the interlocutors are engaged in (Austin 1962). It is possible for the execution of that act to be put onstage as the focused target of description. The result is a performative sentence, as in (11a). More commonly, though, it is left implicit or signaled in a more peripheral manner. In the case of imperatives, where the envisaged action is to be carried out by the addressee, the agent can either be indicated in the usual way by the subject pronoun you, as in (lib), or else left implicit, as in (11c). Since the hearer's agentive role is an inherent specification of the speech act itself, it can be omitted from the expression's overt content unless explicit reference is felt necessary for emphasis. In particular cases, the speaker's role in the envisaged action can also be left implicit, as in (1 Id). To the extent that the speech event instantiates a standard type of interaction, the content overtly expressed may be limited to those facets of the situation not subsumed by the substrate. (11) a. b. c. d.

I order you to stay away from me. You stay away from me! Stay away from me! Stay away!

Apprehension of the speaker-hearer interaction (including illocutionary force) is always part of the conceptual substrate, even for canoni-

Cognitive linguistics, language pedagogy, and the English present tense

19

cal assertions. It is inherent in the construal of every expression as one facet of the viewing arrangement it presupposes. Expressions that are non-descriptive or have non-assertive force constitute one large class of departures from the standard viewing arrangement. There are however many other kinds of departure from it, even with expressions whose function is basically descriptive. Even when using language to report on what happens in the world around us, there are many ways in which we commonly deviate from the canonical situation of directly describing actual individuals and actual occurrences involving them. Some - like metaphor, metonymy, and implicature are extremely prevalent if not utterly pervasive. In the case of metaphor, as in (12a), we are not simply viewing and describing an entity in its own terms, but instead create a blend by selectively projecting properties from the source domain onto the target domain; although the target domain is the actual focus of our interest, it is the blend that is directly encoded linguistically. In metonymy, the entity we mention directly functions as a conceptual reference point, providing mental access to an associated entity that we are actually referring to, as in (12b). And of course, implicature allows us to convey the crucial information without having to express it directly. Thus (12c) can serve as an indirect means of telling the guests that it is time to leave the dinner table. (12) a. The brain tumor robbed him of the chance novel. b. The pen is mightier than the sword. c. The living room is much more comfortable,

to finish his [metaphor] [metonymy] [implicature]

I have already cited (9) as a case of invoking a fictive entity, namely the viewing arrangement itself. Essentially any facet of the overall conception an expression evokes can be fictive (or virtual) rather than actual (Langacker 1999b). For instance, a rhetorical question like (13a) can be analyzed in terms of a fictive speech act - the speaker only pretends to ask a question, it being understood that the actual interactive force is very different.

20

Ronald W. Langacker

(13) a. Would my client - a faithful public servant, a devoted family man, and a deacon in his church - even consider taking a bribe? Of course not! [fictive speech act] b. The trees got shorter as we approached the summit. [fictive change] In (13b), both the subject and the profiled event are fictive in nature. The trees does not refer to any actual set of objects, but is rather a role description, designating a feature of the landscape observable at any altitude. Nor does any tree or set of trees actually change in length. What the sentence describes is a virtual change generated by viewing the trees instantiating the role at different altitudes as if they were a single, changing entity. Finally, let me mention the frequent but seldom noted phenomenon of resorting to type specifications as a way of describing a set of actual occurrences that are alike in some respect. Imagine a series of actual events in each of which a single stranger - different each time - reaches over a fence and picks a single apple - also different each time. If there are three such events, the entire sequence can be summarized by sentence (14a). It is not essential for our purposes that the sentence is subject to alternate interpretations. What is essential is the possibility of using (14b) for exactly the same event sequence. There are three different strangers, and three different apples, yet these participants are referred to in the singular: a stranger, an apple. (14) a. Three times, strangers reached over the fence and picked apples. b. Three times, a stranger reached over the fence and picked an apple. I suggest that (14b) describes the events at the level of their common type characterization. The three actual events each instantiate the event type a stranger reach over the fence and pick an apple. At the type level, representing what the three events have in common, there is only one stranger and one apple. These are instances of the stranger and apple categories, but they are not actual instances or

Cognitive linguistics, language pedagogy, and the English present tense

21

specific individuals. Rather, they are virtual (or arbitrary) instances conjured up just to characterize the type of event in question. In short, even though (14a) and (14b) are used to describe precisely the same event sequence, they view it in different ways for linguistic purposes. The latter involves the conceptual operation of extracting a common event type, whose characterization invokes fictive entities distinct from any actual individual. This special viewing arrangement focuses on the type as the level that is directly coded linguistically, even though the description pertains to actual events. The adverb three times can be seen as an instruction for mapping between these two levels. The moral of all this, not at all surprising from the standpoint of cognitive linguistics, is that the relation between linguistic expressions and the world is non-transparent, being mediated by elaborate mental constructions even when we seem to be merely reporting on what happens (cf. Fauconnier 1997). For an accurate characterization of linguistic elements, we need a detailed understanding of the many subtleties of viewing, as well as the conceptual substrate that supports and shapes expressions. Though I will not suggest any specific pedagogical implications, I have to believe that some appreciation of these factors would be beneficial to both the language learner and the language teacher. In what follows, I will try to show their relevance to a proper analysis of a phenomenon noted for the pedagogical problems it poses, namely the English present tense.

5. The English present tense The one thing that is generally agreed upon concerning the English present tense is that it is really not a present tense, i.e. its value cannot be that of indicating that the process in question occurs at the time of speaking. The arguments seem straightforward. On the one hand, events that do occur at the time of speaking generally cannot be expressed in the present tense. This is the case with perfectives, as we saw earlier. As descriptions of actual, bounded events occurring

22

Ronald W. Langacker

at the time of speaking, sentences like (15a) are consistently infelicitous: (15) a. *Bill {sleeps/paints a fence/changes a tire/learns a poem} right now. b. Bill {is sleeping/is painting a fence/is changing a tire/is learning a poem} right now. To say these things, we must instead use the progressive, as in (15b). On the other hand, many uses of the present tense do not refer to the time of speaking. Standard uses of the "present" pertain to the future, to the past, to "timeless" situations, or even to "eternal truths": (16) a. Your driver's license expires on your next birthday. b. I'm driving home last night and I hear a siren. I pull over and stop. This cop comes up and starts writing me a ticket. c. Hamlet moves to center stage. He pulls out his dagger. He examines it. d. Pi is irrational. Despite these commonplace observations, I have long argued (e.g. 1987b) that the English present tense does in fact locate the designated process at the time of speaking (coincident with the ground). More precisely, the present tense indicates that a full instantiation of the profiled process occurs and precisely coincides with the time of speaking. The proposed account will serve as a case study illustrating many of the points made earlier. I do not know how the present tense should be taught, but an understanding of how it really works must surely be relevant to the problem. I analyze a tense marker as imposing an immediate temporal scope for the focused viewing of the process it grounds. For English, there are just two, as shown in Figure 5, where a box with squiggly lines represents the speech event. The past tense morpheme imposes an immediate scope located prior to the speech event, while the present tense morpheme (at least in English) imposes one that coincides with it.

Cognitive linguistics, language pedagogy, and the English present tense (a)

Past Perfective

IS

Present Perfective

MS

ΚΛΛΛΛΛ

t

(c)

23

Past Imperfective

IS

t

(d)

Present Imperfective

MS

wvWw ^

IS

MS

w w v

Figure 5. Tense and aspect

Observe that in the past tense there is no inherent limit on the length of the immediate scope, so a perfective process of any length can always be made to fit inside it. By contrast, in the present tense the immediate temporal scope must be the same in duration as the speech event. If an imperfective process endures for a span of time that includes the immediate scope (in either the past or the present), only that portion of it subtended by the immediate scope is profiled. Recall that an expression's profile is necessarily confined to its immediate scope (the general locus of attention). Moreover, since an imperfective process is internally homogeneous and not characterized in terms of bounding, any subpart singled out for profiling will itself constitute a valid instance of the process type in question. (Imperfectives are quite analogous to mass nouns in this respect.) Let me start by pointing out how much this analysis accounts for, straightforwardly and even rather elegantly. First, it accounts for imperfectives being able to occur in the present tense, as sketched in Figure 5d. Since any part of an imperfective process itself counts as a full instantiation of the process type, this will also be true for the portion that coincides with the time of speaking. Observe that the analysis does not imply that the stable situation described is valid only for the brief duration of the immediate scope. For instance, an utterance of (5b), He knows the truth, does not entail that his knowl-

24

Ronald W. Langacker

edge of the truth is limited to the time of the utterance. What is being claimed, instead, is that the speech event defines a "window" for focused viewing. In using a present tense imperfective, the speaker is taking a temporally coincident sample of the overall situation and observing that - for the portion sampled - the situation is stable and unbounded. It is possible, without contradiction, for the same overall situation to be sampled at different times, as in (17): (17) He knew the truth then, and he still knows it now. The same overall situation endures without interruption for a span of time that includes both immediate scopes, those imposed by the past tense in the first conjunct and by the present tense in the second conjunct. There is one overall situation, but from it each clause selects a different sample for focused viewing, resulting in two distinct profiled processes, each characterized by local stability and the absence of bounding. It is in the nature of imperfectives that a local sample is sufficient to reveal a stable situation of indefinite duration. Because progressives are derived imperfectives, the analysis predicts their occurrence in both the past and present tense, as in (18). This is diagrammed in Figure 6. Observe that two immediate scopes are indicated. The progressive itself imposes an immediate temporal scope, labeled ISi, whose boundaries are internal to those of the perfective process it applies to (see Figure 4b). The past or present tense marker then applies to the imperfective process profiled by the progressive. As shown in Figure 5, it imposes its own immediate temporal scope, given as IS2, which either precedes or coincides with the time of speaking. The profile of the complex expression is confined to the sample of the imperfectivized process that falls within the scope imposed by the tense. Recall that the ground is the locus of conception, and a tense marker specifies the locus of attention (or focused viewing) with respect to it. (18) She {was/is} {working/writing a letter/wearing a sweater}.

Cognitive linguistics, language pedagogy, and the English present tense

(a)

Past Progressive

(b)

25

Present Progressive MS IS, IS2

l

vvvvv

—

^

-

Figure 6. Tense and the progressive

The analysis also accounts for the usual infelicity of present tense perfectives, as in (15a). The difficulty is not however a matter of conceptual incoherence. Indeed, the conceptual configuration depicted in Figure 5b is perfectly coherent and non-anomalous. There is nothing inherently contradictory about a bounded event temporally coinciding with the speech event. And indeed, true present tense perfectives are sometimes permissible, as we will see. Instead, the problem with present tense perfectives is that certain factors make the configuration in Figure 5b hard to achieve in practice. A perfective process is bounded, so a full instantiation of such a process includes its boundaries. Thus, if a perfective process is to coincide with the time of speaking, its beginning point has to coincide with the initiation of the speech event, and its endpoint with its termination. This poses both a durational problem and an epistemic one. The durational problem is that there is no inherent connection between the length of the event described and the length of the speech event describing it. It takes longer to paint a fence, for example, than it does to utter the clause He paints a fence. The epistemic problem resides in having to observe an event in order to identify it as a prerequisite to describing it. By the time an event is observed and identified, it is already too late to initiate a speech event that precisely coincides with it. These problems do not arise with imperfectives, given their mass-like character and the property that any portion of the overall process counts as a full instantiation of the process type. Hence an imperfective has no specific duration, and a portion

26

Ronald W. Langacker

which follows a period of observation and identification can still count as a valid instance. If a perfective event has a long enough duration, therefore, and extends through the time of speaking, we can describe its occurrence "right now" by means of a present tense progressive. This was diagrammed in Figure 6b. The progressive derives an imperfective process delimited by ISi, and the present tense selects for focused viewing the portion of it that coincides with the time of speaking (IS2). Of course, this solution is not automatically available if the perfective event is punctual, effectively consisting of just an onset and an offset, with no interior phase of any significant duration. A sentence like (19a) is consequently infelicitous if intended as the description of something actually occurring right now (not as merely indicating that the popping is imminent). The reason is that the event is too short for imposition of an immediate scope that excludes its endpoints - there is no interior phase of sufficient duration for focused viewing. (19) a. * A balloon is popping (at this very moment), b. *She is blinking, [single blink] Likewise, (19b) is infelicitous if construed as referring to a single blink in progress. Of course, it can always be construed as repetitive, and since repetitives are perfective, they allow the progressive and require it in the present tense. We can also rescue (19b) by imagining a special viewing arrangement. It would be felicitous, for instance, if we were watching a slow motion film in which a single blink were viewed as occurring over a span of several seconds. Such examples show that well-formedness judgments depend on default assumptions about the world as well as a presupposed viewing arrangement. Also accounted for by the proposed analysis is a striking systematic exception to the usual non-occurrence of present tense perfectives, namely performatives, as in (20). Performatives clearly profile bounded events and not only tolerate but actually require the present tense. The reason is that a performative represents a special viewing arrangement in which the process put onstage and profiled is the speech event itself. Since the profiled process and the speech event

Cognitive linguistics, language pedagogy, and the English present tense

27

are the same, they have to be temporally coincident, as shown in Figure 7. Because of this property, the durational problem does not arise with performatives. Nor does the epistemic problem. The latter arises when it is necessary to observe an event and identify it prior to initiating its description. But with performatives, the speaker is responsible for carrying out the profiled event and acts with prior intent. As the intentional agent of the process, the speaker has no need to observe it in order to know its identity. (20) a. I promise to cooperate. b. I beg you to give me another chance. c. I hereby sentence you to 30 years in prison. Performative MS

is]

, ΙΛΛΛΛΙ ^ Figure 7. Performatives

Performatives indicate that the durational and epistemic problems are not problems with present tense perfectives per se. Rather, they stem from particular circumstances of viewing, namely the default viewing arrangement. In the default arrangement, the event to be described is independent of the speech event and beyond the control of the speaker, who merely observes the occurrence and then reports it. In this case the speaker can hardly begin its description coincident with its initiation, nor is its duration likely to match the time needed to utter a finite clause. But performatives, being intentional actions which implement the very events described, avoid these problems by their intrinsic nature. Are there other kinds of viewing arrangements which, by their nature, avoid the durational and epistemic problems? We simply need to imagine a situation where the speaker controls both the occurrence and the duration of the event described and can therefore make the description coincide with the occurrence. In one situation I

28

Ronald W. Langacker

can imagine, a child is playing with toy cars and a play village and accompanies each action she takes with a descriptive sentence. The successive utterances in (21) then coincide with the successive acts of pushing a toy car from one place to another. This use of the present tense for perfective events seems perfectly natural and unproblematic. (21) Now I drive to work. Now I go to the store. Now I drive home. More generally, the present tense is naturally used for the narration of demonstrations. Imagine an origami class, where each clause in (22a) accompanies the action it describes: (22) a. I pick up a sheet of paper and I fold it in two. I fold it again. Now I take the scissors, and I make an incision from one corner to the center, b. I put a tablespoon of butter in the pan. It melts quickly. Now I put the fillet in. I cook it at a low temperature for five minutes. Or a cooking program on television, as in (22b). Note that the subject does not have to be the speaker (e.g. It melts again). It is only necessary that the speaker have sufficient control over events to avoid the durational and epistemic problems. The last sentence in (22b) is included to raise certain issues concerning duration and coincidence. Obviously, the chef does not take five minutes to utter this sentence. Nor is the butter likely to melt in exactly the span of time required to say It melts quickly. There are two approaches we might consider for dealing with this type of problem. I suspect that both of them are valid and play some role in overlapping ranges of data. One approach is to recognize a certain amount of tolerance in applying the notion "precise coincidence" to actual circumstances. Except perhaps with performatives, exact temporal coincidence - in the strictest sense - is not a realistic expectation (for perfectives). It has to be understood as coincidence apparent on a certain time scale, or some approximation close enough for the

Cognitive linguistics, language pedagogy, and the English present tense

29

purpose at hand. This of course is characteristic of language use in general. A person who says that All politicians are dishonest is not considered to be lying if he believes that somewhere in the world there might be one who is honest. Tolerance in what counts as temporal coincidence clearly figures in the play-by-play mode of speech used by sportscasters. In a way, their role exemplifies the default viewing arrangement, since they occupy a fixed position and do in fact report on actual occurrences. How, then, do they overcome the durational and epistemic problems in their frequent use of present tense perfectives, as in (23)? (23) Stockton dribbles along the baseline. He passes out to Malone. Carl makes a nice grab. He puts up a three-point shot. He scores! Note that the events described in this way have approximately the right duration for temporally coincident description. In the context of a sporting event, they are also quite stereotyped, so the announcer has a good idea of what is likely to transpire at any instant. One is therefore able to shadow the events fairly closely, sometimes even to anticipate and describe them simultaneously with their occurrence. The goal at least is to come as close as possible to coincident description, and the conventions of play-by-play reporting rest on either the fiction that this is feasible or else the tolerance of a certain time-lag. Undoubtedly we have to recognize flexibility and degrees of approximation in what counts as "precise coincidence". These are, after all, matters of construal rather than objective scientific measurement. Yet this hardly seems adequate for examples like the final sentence in (22b). Moreover, I can easily imagine an alternative mode of narration for a demonstration, where - instead of coinciding with it - each statement precedes the action it describes, thereby telling the listener what to watch for. At least for cases like these, we appear to need some other approach. The approach I suggest is to posit a distinct viewing arrangement, one that does not specifically involve the simultaneous narration of actions. Although, in practice, the events in question are correlated with actions, they are conceptualized more

30

Ronald W. Langacker

abstractly as entries in a list, collectively constituting a kind of script or scenario that is being followed. In this respect they would be roughly analogous to stage directions, as in (16c). My proposal for such cases is that the present tense verbs are not in fact being used for the direct description of actual events - instead they indicate the reading off of entries on some kind of list or scenario. In other words, what is being coded linguistically is not the actual occurrence of events, but rather their virtual occurrence as part of a non-canonical viewing arrangement. The viewing arrangement is such that the virtual occurrence does coincide with the time of speaking. If this proposal should seem far-fetched, it is only because we are so accustomed to thinking in terms of the default viewing arrangement and regarding linguistic expressions as direct descriptions of the world. I have tried to show, on independent grounds, that there are many other kinds of viewing arrangements, and that we often resort to the direct linguistic coding of virtual entities even when our real concern is with actual ones. Recall (14b), Three times, a stranger reached over the fence and picked an apple. To describe an actual situation involving three strangers, three apples, and three events of picking, we resort to an expression that directly codes a virtual event involving fictive instances of the stranger and apple categories. We "conjure up" these fictive entities as a way of capturing the abstract commonality of the actual events. Once we recognize the true linguistic prevalence of virtual entities and non-canonical viewing arrangements, the proposal to posit a virtual event occurrence is not at all far-fetched, but rather the sort of thing we ought to anticipate. Let us begin with the scheduled future use of the present tense, as in (24a). My proposal is that such expressions relate only indirectly to the actual event in question. What a sentence like this directly describes is not the actual event per se, but rather a representation of that event on some kind of virtual schedule, some kind of plan or projection concerning the anticipated occurrence and timing of events in the future. Several considerations support the notion that something like a schedule is involved. For one thing, the scheduled future strongly favors a time expression, as seen by the infelicity of (24b). Moreover, it does not work well for events that cannot be scheduled

Cognitive linguistics, language pedagogy, and the English present tense

31

or anticipated. Thus (24c) is awkward unless it is uttered by God, or perhaps by a scientist with supreme confidence in a method of quake prediction. Finally, there is sometimes an actual schedule that is being read, as in (24d). (24) a. b. c. d.

Our new furniture comes tomorrow. ??Our new furniture comes. ??An earthquake strikes next week. There it is on the monitor - our bus leaves at noon.

A virtual schedule pertains to the future, but its own status and location are another matter. When a plan is in effect, the schedule itself is stable and mentally accessible through a span of time that includes the present. The schedule consists of virtual events, which are representations of anticipated actual events. Moreover, the time interval through which each virtual event is conceived as unfolding is identified with a particular time in the future, as shown by the dotted correspondence lines in Figure 8. However, the events constituting the schedule are only virtual. /

Eventj

Event2 I

/

I

~

Actual / hvents I t

Event3 / 1 1 / Virtual 1 1 ^ /Schedule

/

Figure 8. The scheduled future

Metaphorically, we can think of a virtual schedule as a "document" available to be "read" at any time. In producing a sentence like (24a), the speaker is essentially reading off one of its entries. Reading an entry amounts to the virtual occurrence of the event it comprises, and

32

Ronald W. Langacker

since that event is profiled by the sentence produced, a (virtual) occurrence of the profiled process precisely coincides with the time of speaking. Use of the present tense thus conforms to the proposed characterization, taking into account the special viewing arrangement in which the speaker is "reading" aloud from a virtual schedule. In that context, where all the events are virtual, they occur in the sense of being read, and the reading is necessarily coincident with the speech event. I take this as being typical of the so-called "non-present" uses of the present tense in English. Though details vary, a number of them are plausibly described metaphorically as the reading of a virtual document; the differences reside in the kind of document envisaged. In the case of (22b), the document would be an imagined script of how the cooking demonstration is supposed to proceed, step by step. In the case of stage directions, as in (16c) [Hamlet moves to center stage. He pulls out his dagger. He examines it.], the script may well be physically embodied. But even conceived as a virtual document, it comprises a series of inscribed events available to be read at any time. Reading them, and thereby apprehending the successive event descriptions, induces their virtual occurrence in the form of mentally constructing the prescribed event sequence. What about the historical present, as in (16b) [I'm driving home last night and I hear a siren. I pull over and stop. This cop comes up and starts writing me a ticket.]? Here the virtual document consists of a series of recalled events that the speaker can mentally "replay" at leisure, at the pace required for linguistic encoding. Another kind of recall figures in a photo caption, as in (25). Here the statement is physically instantiated and read quite literally. The photo captures one moment of the event described by the caption, and helps to evoke its virtual occurrence in the guise of apprehending the statement's import. (25) Nixon says farewell from the steps of his helicopter, [photo caption] More generally, the key to understanding "non-present" uses of the present tense lies in recognizing the special viewing arrangements

Cognitive linguistics, language pedagogy, and the English present tense

33

they presuppose. They all diverge from the default arrangement by invoking some kind of mental construction - such as a schedule, script, or mental replay - consisting of event representations. Even when these correspond in some fashion to actual events, the represented events are the ones directly coded linguistically and profiled by the present tense verb. What counts as the occurrence of such a process is therefore not an actual occurrence, but rather a virtual one consisting in its apprehension in the manner indicated by the special viewing arrangement (e.g. reading the virtual schedule, or running the mental replay). Indeed, the profiling of actual events in accordance with the default viewing arrangement can be seen as a special case of this more general scheme, the case of identity between the mental construction of represented events and the observation of actual ones. One broad class of mental constructions comprises generalizations extracted to represent the world's basic structure, as opposed to specific, contingent occurrences that arise within this stable framework (Goldsmith and Woisetschlaeger 1982; Langacker 1996, 1997). Expressions describing the world's basic structure - which I refer to as structural statements - include generics, habituais, and "timeless truths". Even when the verbs employed are perfective, they do not refer to specific, actual events. For example, (26a) does not designate any actual instance of a kitten being born. The profiled event is a virtual instance of the process type in question, conjured up to express a generalization about one facet of the world's basic nature. The events coded by structural statements belong to mental constructions purporting to represent the world's basic structure rather than any actual occurrences. Each of these event representations corresponds to an open-ended set of actual instantiations, distributed throughout the time span during which the generalization holds. Tense marking on a structural statement specifies the time at which the event representation can be consulted as a way of apprehending this facet of the world's structure. Present tense indicates its viewing, and thus its virtual occurrence, coincident with the time of speaking.

34

Ronald W. Langacker

(26) a. A kitten is born with blue eyes. [generic] b. My cousin goes to a singles bar on Friday night, [habitual] c. Water decomposes into hydrogen and oxygen. ["timeless truth"] Another special use of the present tense in English involves causal relationships between events and situations. This curious construction, illustrated by (27a), is often heard from sports announcers. I suspect it is rather common in casual speech. Despite the present tense, the sentence does not describe anything happening right now, nor is it a direct description of actuality. It is, moreover, ambiguous. Sentence (27a) may be construed as comparable to (27b), pertaining to a facet of future potentiality. Alternatively, it has an interpretation similar to (27c), pertaining to something that failed to materialize in the past. But unlike (27b-c), which make these interpretations explicit, in (27a) they are left implicit: how the events described are connected to actuality is not directly specified by any overt element. (27) a. He makes the freethrow and the game is tied. b. If he makes the freethrow, the game will be tied. c. If he had made the freethrow, the game would have been tied. An expression like (27a) profiles a sequence of two virtual processes, the first representing an event, and the second a situation resulting from that event. This mental construction manifests a special viewing arrangement allowing the simplified presentation of a contingency inherent in the past or future evolution of reality. Let's face it, (27b) and especially (27c) are quite complex both conceptually and grammatically, involving modals, perfect aspect, mental spaces, and shifts of viewpoint within those spaces. The viewing arrangement of (27a) does away with all this at the level of explicit linguistic coding. It simply abstracts the causally related processes and presents them as virtual occurrences available for direct viewing at the moment of speech. The complexity of their epistemic status is still all there, but

Cognitive linguistics, language pedagogy, and the English present tense

35

is incorporated in the presupposed viewing arrangement instead of being overtly expressed. The overall situation is sketched in Figure 9. The two virtual processes are connected by a double arrow to represent the causal relationship between them. It is only this virtual process sequence that is put onstage and profiled by the successive coordinate clauses in a sentence like (27a). Each clause is in the present tense, hence their profiles coincide with the respective speech events associated with the two finite clauses. In producing and understanding the expression, the speaker and hearer view and apprehend the virtual process sequence, and their apprehension of it constitutes its virtual occurrence coincident with the time(s) of speaking.

/

/ Α

/

1 Processai—^ Process2| Γ-

/ ^ /Virtual

,F ,Λ w ΛΛΑ, w w

Figure 9. A virtual process sequence

The presupposed viewing arrangement incorporates a conception of how the profiled process sequence relates to actuality. There are two options in regard to time. As indicated by the dotted correspondence lines, the time span during which the virtual process sequence is envisaged as occurring is identified with a temporal interval either prior to the speech events or subsequent to them. Under either option, the virtual processes are taken as representing a valid contingency, such that the actual occurrence of Processi results in the actual occurrence of Process2. However, the viewing arrangement also specifies that Processi is not (or not yet) actual. When applied to the past, where the course of reality has already been determined, the non-actuality of Processi entails that it did not in fact occur, and thus implicates that

36

Ronald W. Langacker

the situation coded by Process2 does not obtain. When applied to the future, where the course of reality has not yet been determined so that Processi is necessarily non-actual, presenting it for consideration has the effect of suggesting that its actual occurrence (and the subsequent occurrence of Process2) may be quite imminent. Finally, I should mention the use of present tense in certain types of subordinate clauses, as in (28): (28) {If / when / until / before / after / while} you make a decision, you should consider all your options. Despite its present tense form, in each case the predicate {make a decision) refers to a process envisaged as occurring in the future. Here I basically follow the analysis proposed by Fauconnier (1997) in terms of mental spaces. The subordinators introducing these clauses are space builders: //establishes a hypothetical space, and the others set up spaces defined by their temporal location. They further shift the viewpoint to the space they establish. In other words, they incorporate the instruction to adopt a special viewing arrangement in which the clausal content is apprehended from a temporal vantage point other than the actual time of speaking. Its apprehension does of course occur at the time of speaking, but when this is Actively identified with the time span internal to the mental space, a process conceived as occurring in this space is fictively viewed as coincident with the speech event.

6. Conclusion I have sketched an analysis of the English present tense that is quite non-standard and undoubtedly controversial. It is still preliminary (even after all these years), and certainly incomplete. For instance, following Brisard (1999), I have not sufficiently emphasized the epistemic immediacy that constitutes the flip side of temporal coincidence. Still, I have little personal doubt that the account is basically valid, or at least suggestive of what a valid account might look like.

Cognitive linguistics, language pedagogy, and the English present tense

37

For sake of discussion, let us suppose that you agree. What then follows? What conclusions can we draw for language pedagogy? I am of course eschewing any specific pedagogical proposals. I will however suggest that the traditional way of looking at tense, even in linguistics, engenders confusion by obscuring its basic nature. Standard discussions are objectivist in spirit. They ignore construal and the subjective basis of factors like homogeneity and bounding. They have no conception of the myriad viewing arrangements that mediate between objective circumstances and the formulation of linguistic expressions. Thus they attempt to account for tense directly in terms of the temporal relation between the actual time of speaking and the full duration of an envisaged actual occurrence. They do this even when - according to the analysis presented here - the process being viewed and temporally located is only a portion of the actual occurrence (notably with imperfectives), or else a virtual process connected to it in a manner specified by the viewing arrangement. It is no wonder, then, that a cogent description remains elusive, and that the present tense is claimed to be anything but a present tense. The consideration of pedagogical issues can only be aided by an accurate understanding of what is being taught. In the case of language, unfortunately, traditional and modern understandings are usually far from adequate, even for things as fundamental as the present tense. It is premature to suggest that cognitive linguistics is coming to the rescue. I do however see it as a positive development, providing new and revealing perspectives on specific problems as well as our overall conception of language and how it relates to culture, cognition, and social interaction. In short, I think we are starting to get a real grip on how things work. If so, it should eventually give rise to successful pedagogical applications, which will lend it empirical support.

38

Ronald W. Langacker

References Austin, J. L. 1962 Brisard, Frank 1999

How to Do Things with Words. Cambridge, MA: Harvard University Press.

A critique of localism in and about tense theory. Ph.D. dissertation, University of Antwerp. Fauconnier, Gilles 1997 Mappings in Thought and Language. Cambridge: Cambridge University Press. Goldberg, Adele E. 1995 Constructions: A Construction Grammar Approach to Argument Structure. Chicago and London: University of Chicago Press. Goldsmith, John and Erich Woisetschlaeger 1982 The logic of the English progressive. Linguistic Inquiry 13: 7989. Jackendoff, Ray 1994 Patterns in the Mind: Language and Human Nature. New York: Basic Books. Langacker, Ronald W. 1968 Language and its Structure: Some Fundamental Linguistic Concepts. New York: Harcourt, Brace & World. 1987a Foundations of Cognitive Grammar, Vol.1, Theoretical Prerequisites. Stanford: Stanford University Press. 1987b Nouns and verbs. Language 63: 53-94. 1988 A usage-based model. In: Brygida Rudzka-Ostyn (ed.), Topics in Cognitive Linguistics, 127-161. (Current Issues in Linguistic Theory 50.) Amsterdam/Philadelphia: John Benjamins. 1990 Concept, Image, and Symbol: The Cognitive Basis of Grammar. (Cognitive Linguistics Research 1.) Berlin/New York: Mouton de Gruyter. 1991 Foundations of Cognitive Grammar, Vol.2: Descriptive Application. Stanford: Stanford University Press. 1993 Universals of construal. Proceedings of the Annual Meeting of the Berkeley Linguistics Society 19: 447-463. 1995 Viewing in cognition and grammar. In: Philip W. Davis (ed.), Alternative Linguistics: Descriptive and Theoretical Modes, 153212. (Current Issues in Linguistic Theory 102.) Amsterdam/ Philadelphia: John Benjamins.

Cognitive linguistics, language pedagogy, and the English present tense 1996

1997

1999a 1999b 2000

39

A constraint on progressive generics. In: Adele E. Goldberg (ed.), Conceptual Structure, Discourse and Language, 289-302. Stanford: CSLI Publications. Generics and habituais. In: Angeliki Athanasiadou and René Dirven (eds.), On Conditionals Again, 191-222. (Current Issues in Linguistic Theory 143.) Amsterdam/Philadelphia: John Benjamins. Grammar and Conceptualization. (Cognitive Linguistics Research 14.) Berlin/New York: Mouton de Gruyter. Virtual reality. Studies in the Linguistic Sciences 29(2): 77-103. A dynamic usage-based model. In: Michael Barlow and Suzanne Kemmer (eds.), Usage Based Models of Language, 1-63. Stanford: CSLI Publications.

Sweetser, Eve E. 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. (Cambridge Studies in Linguistics 54.) Cambridge: Cambridge University Press. Talmy, Leonard 1988 Force dynamics in language and cognition. Cognitive Science 12: 49-100. 1996 Fictive motion in language and "ception". In: Paul Bloom, Mary A. Peterson, Lynn Nadel and Merrill F. Garrett (eds.), Language and Space, 211-276. Cambridge, MA/London: MIT Press/Bradford.

Pretend play: trial ground for the simple present Jenny Cook-Gumperz and Amy Kyratzis1

1. Introduction

Here is Edward Bear, coming downstairs now, bump, bump, bump, on the back of his head, behind Christopher Robin. It is, as far as he knows, the only way of coming downstairs, but sometimes he feels that there really is another way, if only he could stop bumping for a moment and think of it. And then he feels that perhaps there isn't. Anyhow, here he is at the bottom of the stairs, and ready to be introduced to you. Winnie-thePooh [Milne 926: 3].

And so begins one of the most self consciously literary, yet most frequently reissued children's books of the last 70 years. In this the original beginning to the story, since adapted by Disney and others into a universal tale of a "cute little bear", the literary devices of tense/aspect provide a specific set of what Langacker (this volume) calls the viewing arrangements that enable participants and readers/hearers to shift perspectives on the story events. The following text adds a further dimension to the tellers', participants', and reader/hearers' perspectives, when A. A. Milne inteqects his own character/voice: "When I first heard his name, I said, just as you are going to say, 'But I thought he was a boy?'" The author here enters the story briefly as a protagonist, that is an authorial "I" telling how the tale came to be told. After this interlude the story proper begins in the traditional story telling past tense, the third person preterite (Benveniste 1971). "Once upon a time, a very long time ago now, about last Friday, Winnie-the-Pooh lived in a forest all by himself under the name of Sanders." We cite these, and

42

Jenny Cook-Gumperz and Amy Kyratzis

later examples from the Dr. Seuss books, to illustrate the kinds of narrative experiences children participate in before they learn to read for themselves. In attending to the often told/read tales, they are exposed to linguistic constructions that can have complex semantic implications such as signaling shifts in perspective. Through tense/ aspect contrasts, children's stories are able to provide succinct information about the characters, the tale's context and the relationship of the teller to the protagonists that make simply told tales dependent on complex conceptualizations for their interpretation. As Langacker argues: "Inherent in every usage event is a presupposed viewing arrangement, pertaining to the relationship between the conceptualizers and the situation being viewed" (16). The progressive, as described by Langacker, for example, is distinct, not for its focus on the present time but for the viewing arrangement it imposes: allowing the "zooming in" and "taking an internal view" of a bounded event (12). In story telling, and in literary contexts, such a viewing arrangement is not appropriate; the events must be told as though the narrator can view their endpoint. In narrativizing discourse, then, to quote White's seminal paper on Narrativity: "we can say, with Benveniste, 'Truly there is no longer a 'narrator'. The events are chronologically recorded as they appear on the horizon of the story. Here no one speaks. The events seem to tell themselves" (White, quoting Benveniste 1981). Here the events are told as though they are unfolding at the moment of reading, and the narrator takes a distal, uninvolved viewpoint; s/he is not in the event. Yet sometimes, the narrator seeks to manipulate viewpoint to involve the reader more, by imposing a viewing arrangement that combines the affordances of the narrator's distal viewpoint which allows story events to be viewed unfolding, with a closer-seeming viewing arrangement that creates suspense. The simple present is a literary form which does just this. It imposes just such a self-conscious yet suspenseful perspective. The distal aspect of its viewpoint comes from reporting perfective events as though the teller could view their endpoint. As Langacker claims this imposes a seemingly "impossible" viewing arrangement for the conceptualizer unless she/he reconceptualizes the event as virtual rather than actual (19). Langacker

Pretend play

43

goes on to point out that the simple present is difficult to use in describing perfective events, occurring at the moment of speaking because of what he termed the durational problem. As he puts it "because a perfective process is bounded, a full instantiation of it includes its boundaries, so if a perfective process is to coincide with the time of speaking, its beginning point has to coincide with the initiation of the event of speaking, and its endpoint with its termination" (25). In other words, the speaker has to coordinate the event time with the beginning of speech and wait for the termination of the reported events to coincide with the time of speaking. The main use of the present would be in event casts, which describe actions taking place in the present time, such as sports commentaries, cooking shows, direction giving, and even more rarely, advanced driving tests. In event casts, such a viewpoint is possible because events are happening quickly and each event is completed before the other begins. Since such events rarely occur in everyday discourse, we tend to assume a virtual, that is a hypothetical event is at issue when we hear/see a simple present, as in the above beginning of the story. It is more usual in descriptions of present actions to use progressive constructions rather than the simple present. However with habituais, where there is no endpoint, the present is also used. When progressives ("imperfectives") are used, the speaker is taking a temporally coincident sample of the overall situation and observing that, for the portion sampled, the situation is stable and unbounded. For statements such as "I am driving to work", because a termination viewpoint is not necessitated, and similarly for generic statements or statements of habitual events (e.g. "I drive to work"), the endpoint does not have to be encompassed in the time of speaking. Therefore, to summarize Langacker's argument, there are two issues at stake in the choice of tense-aspect. One is the focus on the event time as simultaneous with or prior to the speech event, that is the choice as to whether one takes an involved or distal viewpoint; and the other is the viewing arrangement or the choice of perspective, that is whether the endpoint is revealed at the time of the telling. If we accept that the simple present is unusual in ordinary, everyday discourse, occurring only in special contexts, then children rarely

44

Jenny Cook-Gumperz and Amy Kyratzis

get practice with it. This argument is further substantiated in looking at child language studies. Already in Roger Brown's early "A First Language" (1973), he showed that the progressive present occurs much earlier than the simple present. Brown found a range of acquisition ages of 1;10-2;10, for the progressive to between 2;4 and 3;8 for the simple present in the three children he studied. Additionally, there is one area of children's early experience where the simple present is used: in picture-book storytelling between adults and children. This pre-literacy experience bears out Langacker's argument about the perspectival shift between continuative (or imperfective) action that you are a part of and the perfective aspect. In acquiring literacy, children must learn to negotiate this shift and adopt a perspectival role very different from the demands of everyday spoken discourse. They need to be able to take a viewpoint that looks in on events as potentially complete and finite, as well as viewing ongoing events from a distanced or continuative perspective. Children gain this experience in preliteracy through examining the pictures in a storybook as the story read to them unfolds. The events taking place in the pictures of a book are not realized until the reader/hearer looks in on them from their appropriate perspective which is further reinforced by the story text. As we saw in the first examples, the perspectival shift heightens involvement, since the present aligns the viewpoint of the reader/ hearer with that of the narrator, while simultaneously allowing the "viewing" of the unfolding of events (and their endpoints) that may in actuality take/have taken a long time to play out. From a writer's viewpoint, because viewing arrangements are virtual, using the simple present makes an increased involvement possible and so is a desirable literary move to take. The excerpt below from Dr. Seuss' "One Fish, Two Fish, Red Fish, Blue Fish", conveys this manipulation of viewpoint, although to fully appreciate what is happening both pictures and text need to be read together, as is common in children's books.

Pretend play

45

Here are some who like to run. they run for fun in the hot, hot sun. Oh me! Oh my! Oh me! Oh my! What a lot of funny things go by. We see them come. We see them go. Some are fast. And some are slow. Some are high. And some are low. Not one of them is like another. Don't ask us why. Go ask your mother. Simple present descriptions such as "we see them go" by implication put the reader/hearer into the frame of the action, and this involvement is quite literally represented by pictures showing a boy and girl watching the events as they are unfolding in the virtual picture space, as if the "real" little boy or girl hearer/listener were inside this space. However, the present tense conceptually also allows the events to appear limited with potential endpoints. This viewpoint is reinforced figuratively by the illustrations on each page which actually appear to delimit the action, because they end with the page. These contrasts allow the other constraint of "objectivity" in narrative, as suggested by narrative theorists such as Benveniste (1971) and White (White 1980), also to be realized. Children must begin to understand such complex and seemingly, "impossible" viewing arrangements in order to participate in early

46

Jenny Cook-Gumperz and Amy Kyratzis

book-reading routines and understand literacy events. In Heath's (1986) description of a two-year-old Roadville child's participation in a bookreading routine, the child was able to utilize the simple present "Puppy sees the ant"; "ant bite puppy" in responding to her aunt's questions ("what does the puppy do?") (Heath 1986: 107). Later, children will need to accomplish even greater milestones in using tense and aspect shifts for literary effect; they will have to learn to utilize the progressive not only in the sense of its affording of a "zooming in" effect on a bounded event, but for backgrounding in narrative. As noted by Hopper (1979), the contrast between ing-ed is not a sentence-level, semantic phenomenon but an essentially discourse-level one between events and states presented as backgrounded and events presented as individualized and sequenced. Our argument in this paper is that English speaking children's early understanding of the tense system is a key prerequisite for literacy, and that a major source of this preparation for literacy comes from children's engagement in pretend play with peers as well as with story telling and reading experiences with adults. A viewing arrangement for use of the simple present which avoids Langacker's durational problem is planning during pretend play, particularly planning talk for the manipulation of character figures. Planning action sequences for third-party characters in fantasy-play allows the speaker to achieve a distal viewpoint on the activity. Because the action sequence that is planned is hypothetical, or irrealis, a full instantiation of it and its boundaries is not necessitated. Since the actions themselves are imaginary, the boundaries of the actions can be hypothesized. In this way, planning action sequences for character figures in manipulative pretend play can be a context which scaffolds and provides a training ground for the use of the simple present. Our argument is grounded in a perspective that views grammar as discourse-embedded. More specifically, this view (Budwig 1990, 1995, Ervin-Tripp 1993, Gerhardt and Savasir 1986, Slobin 1982, 1985) maintains that children construct and appropriate grammatical contrasts to index particular clusters of semantic, pragmatic, and discursive notions. As stated by Ervin-Tripp (1987, 1993), "my advice to students of syntax is to get a video camera and look at what is go-

Pretend play

47

ing on" in the discourse. In other words, patternings in grammar are appropriated by children to construct or index patternings in discursive phenomena. For example, Budwig (1995) argued that some children use the contrast between "I-my" self-reference forms in first person to index a cluster of semantic/pragmatic features denoting the contrast between low and high agency/intentionality. Along similar lines, Gerhardt and Savasir (1986) argued that preschool aged children utilize the form contrast between simple present and progressive to index the contrast between normativity and negotiation. In this paper, we combine the arguments by Langacker (this volume) and Gerhardt and Savasir (1986) and argue that children use the simple present, not to index normativity alone, but to signal normative, temporally-sequenced action from a privileged viewpoint which allows the endpoints of perfective events not to have to occur during the time frame of the linguistic description of the events. This privileged viewpoint is the distal perspective afforded by the genre of hypothetical planning talk, particularly as it occurs during collaborative character play with peers.

2. Examples The data: excerpts in this paper come from three studies: one a study of two 3.6 year old girls who play regularly in each others' home (Cook-Gumperz 1985, 1992). The second set come from a study of character play with four- and seven-year-old dyads (Kyratzis 1992). The third set of data is taken from spontaneous play situations videotaped in a nursery school in a combined three- and four-year-old classroom (Kyratzis and Guo 2001, Kyratzis forthcoming).

2.1. Present tense for temporally-sequenced action: 3-year olds Two three and half year old girls, Lucie and Susie are playing a game of mummies and babies, indoors in small play space, surrounded by a collection of dolls and stuffed animals. The girls make up a loose

48

Jenny Cook-Gumperz and Amy Kyratzis

narrative plot for a game of Mummies and Babies where mummies feed, give drinks to their babies, scold them, put them to bed, go into the garden with their babies, go on a walk to the park etc. The game is fast-paced in that the girls are continuously talking or vocalizing sounds, such as drinking or soothing noises. Through their talk they create three distinct voices, a narrative planning voice in which they negotiate and jointly construct the plot, a voice of mothers to each other, or a voice of mothers speaking to their babies, each marked by a distinct tone of voice. For example, when the mothers address their babies their voice has short clipped phrasing with many rising/falling and lengthened pitch contours especially at the end of phrases. The narrative voice has an evenly paced tempo, close to their ordinary voices. The mother to mother voice is louder, with higher but more even pitch register. The narrative voice is shown with underline, the mummies to babies in italics, and the mummies to mummies in bold typeface. Ordinary voices are shown as unmarked text in the same type face as the rest of the paper. The first example shows the girls using the present tense as part of a narrative describing the play action that they are about to perform. Example

1.1

Susie:

We've got to go up here and cause it's a sunny day we've got to go outside and sit down in our garfden] Lucie: Oh... oh yes that., that can be there..that can be there., yes Susie: And we sit down and have a glass of orange iuicev. (whispered mummy to baby) there sit down. (play drinking and pouring sounds) Shh..shh..now cambe... it's

some more., maccamba

of.,

gumpy-shek..

it's

Susie: The babies don't like it Lucie: No the babies don't like macacamba [Two turns later] Susie: And he sits.... And the babies sit on our laps with us

mac-

Pretend play

49

In this example the two little girls used sounds, (Shh..Shh..) and made up words (maccacamba, gumpy shekt) to set up a rhythmic/metric beat for the game. Tense contrasts, along with voice-tone differences, mark the different perspectives represented and enacted by the "game voices". The narrative plot is predominantly represented in the simple present tense conveying a distal perspective as if the interactants are looking in on the stage where they are about to perform their actions. The initial statement by Susie relies on the model phrase "got to", but as the aspectual sense of these phrases has something of a habitual character to them, they could be interpreted as 'when it's sunny we always go and sit in the garden'. In Susie's second and third utterance the simple present describes the actions as they are about to happen, yet with the sense that these activities once performed can be visualized as computable actions. Their potential completeness contributes to their immediacy. In the next example the girls get into a dispute about the consequences of the action's narrative which is resolved in the last line by Susie's addition to the plot that signals her acceptance of the story line. Example 1.2

Lucie: Susie: Lucie: Susie: Lucie: Susie: Lucie: Susie: Lucie: Susie:

Anyway.... and you say.. "Sandra., have you got pins"., and I'll sav 'Ves" Have you got some pins? Yes I want to hold her.. I want., it's not fair She's having your pin She's mine ..Now you sav "Sandra., have you got pins., and I'll sav "yes" ..NO No you..And and you sav "Sandra..have you got pins to stick in your baby and and you and I say "yes" And um., then you come and give me two pins.

50

Jenny Cook-Gumperz and Amy Kyratzis

Susie's acceptance of Lucie's idea for the plot development is recognized by her conjoined contribution "and then you come and give me two pins". Initially the use of the discourse marker "anyway" is part of the narrative plot outlining speech, which along with use of direct speech markers 'you say' and 'I'll say' give the narrative voice something of the quality of a story in a children's book. The acting out of the plot takes place in a "mummy to mummy" voice with raised pitch, enacting the dialogue described by the narrative. These dialogue stage directions and enactments are illustrative of Langacker's durational problem. The talk describing the narrative plot can project a completable event or set of actions because it is hypothesizing the whole short sequence of dialogue, action and exchange from a distal perspective. These directions are then carried out but only in so far as they voice the suggested dialogue. If the girls' dialogue itself became the source for further actions arising either from mummy to mummy or mummy to baby talk, it is likely that they would use a tense shift to a present progressive to signal a new focus on continuing action, so 'zooming in' on the action as it takes place over a longer time span. This use of the present progressive in 'mummy-to-mummy talk' can be seen in the next Example 2.1. However while the simple present tense implies a temporal/durational aspect, as Langacker explains such a durational issue can be used to implicate habitual usage. Young children rarely use this form. In the first example (1.1) we have commented on the use of the habitual in a modal construction "we've got to go and sit down in our garden cause it's sunny" meaning 'when it's sunny we always go out in the garden'. The following example has something of a similar habitual generic quality to it, implying that a journey to Belgium is always similar to one being described. Example 1.3 Diane: How do you get to Belgium Susie? Susie: First you go on a boat and then you get to Belgium and go on a car the...then ..you get to Belgium Lucie: and then you go on a train and...and sea and ships

Pretend play

51

Susie:

ahh...and then you go in your car and you.your daddy put.gives.takes you to Belgium with your Daddy and Mummy and then you get to Belgium Diane: oh umm Susie: and it's a long way and it's inside Dominique's house and you'll see Dominique won't you?..

2.2.

The present progressive for ongoing action in in-role play with 3 year olds.

Another set of examples where the two girls alternate their "motherto- mother" talk with mother-to-baby talk. Apart from the distinctive voice/intonation patterning the former is more likely to be characterized by use of the progressive tense usually the present progressive as the girls (as mummies) report their activities to each other, and the mother to baby talk contains the greatest range of tense shifting and syntactic complexities, using conditionals, imperatives with justifications, indirect imperatives e.g. "I don't want you to shout at Sandra" reprimands in the form of questions e.g. "why did you spit water at Sally Manga?" Example 2.1 Lucie: Your baby is crying... Samolina...Santolina...allright...I'll give you to your mummy...I'll give you to your mummy now...Y our baby wants you S: [inaudible] [out of range] L: allright..here..she comes ..And then we are going to walk around..should we go for a little walk?...s7za// we go for a little walk...with Samolina..And try not to hurt her Samolina...Samolina look..don't hit her ...oh let's go for a walk you two...let's go for a walk with Sandra when she finishes There's Sandra...Sandra ..we are going for a little walk Susie: I know it.. where's my cup

52

Jenny Cook-Gumperz and Amy Kyratzis

Lucie: Susie: Lucie:

allright baby...have a drink on that one now...on mummy's knee...you know where it is... and you 7/ be happily ever after I hold my baby 'cause she was crying for me N0..N0 you hold my baby and I hold your baby

In this example Lucie is organizing the activity of her two 'pretend' babies and calling out to the other "mummy" (Susie), who is temporally given the name Sandra by Lucie at this stage in the game, that she is taking her babies for a walk. This statement serves as a directive to tell Susie about the new direction in the new plot. As Lucie's talk goes seamlessly back and forth between her babies and the other mummy, tense choices is one of the main indicators of her shifting game roles, the progressive describes actions being described by the mummies to each other such as "your baby is crying" or "we are going for a walk", the simple present either depicts mummies' talk to babies, particularly as directives or imperatives "let's go for a walk" "Samolina look., don't hit her".

2.3. Present tense for temporally-sequenced action: 4-year olds We now shift to examples of speech from character play involving four-year-olds. The narrative excerpts are from two four-year-old girls, Abby and Fran. With the circus props, these girls are emplotting a narrative about a Mom and twin children putting on an iceskating show. This excerpt is rich in tense shifting. Background, setting-like information uses the progressive, both past ("pretend I was moving this up and down" and "pretend somebody was ice-skating on the rink") and present ("and you're sitting where the man- lady zooms by") and copula verbs, both past ("there was a mom"; "pretend they were boyfriend and girlfriend, okay?") and present ("this is you, the teen-ager"; "now somebody's on the ice skate rink"; "but *he is the audience"; "he's my boyfriend"; "that's my mom"). In contrast, temporally-sequenced action is presented in past subjunctive ("I didn't let the seals fall off' "you got off the rink to watch for awhile"; "I got off') and present tense ("she slows down when she

Pretend play

53

gets to you"; "they say who's that"; "I say mama"; "you take a good look and say"; "I say that's my mom"; "and he says that's my mom"). In the narrative voice, when the girls say "I", they are referring not to themselves {in realis) but to the character of the Playmobile figure they have designated as acting out their role. In this voice, the girls seem to be reading off a script representation that is already constructed. As Langacker (this volume) states, "Tense marking on a structural statement specifies the time at which the (already constructed) event representation can be consulted as a way of apprehending the world's structure" (Langacker this volume). In contrast, when the girls are talking as themselves, in stage-manager voice using the first person, they include modal marking on the verb (e.g. "I'm supposed to act myself'; "pretend I could sit the seals on"), just as they also use commands and questions in this voice ("do it like this", "make it like this so the seals can sit"; "wait, I'll be right back"). The excerpts below exemplify how this dyad shifts tense and aspect to mark different phases of the play, using statives in the set-up, beginning phase, then shifting to progressive and simple present to mark background and temporally sequenced action, respectively. [up to now, they've been doing an animal tricks show. Now they shift to ice skating show scenario] Example 3.1 Fran: and pretend somebody was ice-skating on the rink, there was a mom. wait, pretend you got off the rink to watch for a while. [F grabs A's figure and places it in front A] Abby: okay, I got off it. [picks up the figure] Fran: now, now somebody's on the iceskate rink, 'cuz when she...[grabs A's figure again] I'll set her up to be standing. Abby: okay, but *he is the audience, [grabs a male figure and places it in front of F] wait. I'll be right back. [A goes to F's

54

Jenny Cook-Gumperz and Amy Kyratzis

side] pretend = t h e y were boyfriend and girlfriend, okay? [A returns to her seat] In these lines, the girls are setting up the scenario, and they use a lot of copulas "somebody is on the ice-skating rink" and statives "there was a mom", "they were boyfriend and girlfriend", including one in present tense "but *he is the audience". Statives are appropriate for the setting-up phase of the pretend. However, after this is accomplished, they shift to progressive and simple present. Example 3.2

Fran:

Abby: Fran: Abby: Fran: Abby: Fran: Abby: Fran:

and you're sitting where the man - where the lady zooms by, but she slows down when she gets to you. okay? [F runs her figure along the "ice" until it reaches F's side of the circus ring] they say 'who's that'? and I say- and I say- and I say [gets up and walks towards F] 'mama!' 'oh, he's my boyfriend', we're talking. no, no, you ask 'who's that?', and then you take a good look and say 'that's your mom!' um,'who's that'? and I say, 'that's my mom'. and he, and he says, 'that's my mom'. yeah, they both say that because they're twins.

In the first line, "and you're sitting where the man - where the lady zooms by, but she slows down when she gets to you", F presents an event sequence (lady zooms by, slows down), and uses the simple present to denote the temporally sequenced action. Simultaneously, she presents the background event (the audience sitting), using the progressive, since this is background material and its endpoint is not a focus of the story. Lower, three turns later, A uses progressive aspect to again denote background material - the explanation of her character's speech ("we're talking"). Throughout the rest of the excerpt, through a long chain of story action, the two girls use simple present to denote the temporally sequenced acts ("you ask", "and

Pretend play

55

then you take a good look" "and say", "and I say" and "he says". In the last line, they use third person in what Langacker called the stative sense, because the endpoint of the act "say" is not a focus and the information is merely explanatory, hence background. In example 3.2, as in the Dr. Seuss excerpt above, the simple present is used when there is an "observer" to the event (in this case, the Mom). Other times in their narrative, when there isn't such involved viewpoint, the girls use the preterite to tell and plan their story. Example 3.3 Fran: hello sweetie! Pretend you said, "hey that's my mama". Abby: 'hey, that's mama/and then she...no, pretend that, I'm then, the um, the trumpeters, well, like, Fran: ==the seals. Abby: the seals come on Fran: and I do the, um, the Abby: and they go 'shooo--(x) Fran: put that one on this! Abby: no, I'm gonna do it my way. Example 3.3 shows A and F advancing their collaborative story together, using simple present ("seals come on", "I do", "they go 'shoo'"), though they then break off into a stage manager argument (Wolf and Hicks 1989). They shift to modals ("gonna") and commands to do their arguing to one another in the stage manager voice. In contrast, in the voice that went with simple present, the "I's" and "you's" referred to characters, not to the girls themselves. The girls shift tense and aspect to index shifts in story levels (backgroundforeground) and story phases (stage-manager/director) in ways that are very impressive for this age. In sum, of the 10 four-year-old dyads in this study with character figures, only two used simple present in the sophisticated way shown here, for "consulting" or reading off temporally sequenced action from a script in the present time, and differentiating this planning phase both from the stage-manager voice and from background de-

56

Jenny Cook-Gumperz and Amy Kyratzis

tail. Abby and Fran displayed quite sophisticated tense-aspect shifting for children in the preschool period.

2.4. Present tense for habitual state, progressive for ongoing action in role-play with four-year-olds In Kyratzis' other data base with four-year-olds, which was role-play in the classroom, children seldom used character figures. They acted in roles themselves. Consequently, use of the simple present did not occur other than for habitual state (e.g. "I'm a Chinese sister, and I look pretty"; "I come from Korea".) For temporally sequenced action in their dramatic narratives, children used progressive tense (e.g. "I'm making rice"). We argue that this is because they are taking an involved, in contrast to a distal, viewpoint on the action, which they did in the ice-skating show excerpt featured above. In role-play, they are themselves engaged in the action. In contrast, in the character play excerpted above, they were manipulating character figures and talking about and reflecting on the action. In the excerpt below, three four-year-old girls (Viv, Aly, and Joan) are enacting a scenario where they are pretending to be Chinese sisters and make rice for their brothers. Example 4 Aly I'm making soup// (kalatoya)// (kalatoya)// (kulakepa)// Viv (malakapu)// pretend I'm making food// =(xx) pretend= I'm making Chinese stuff// Aly =I'm making food for my brother=// I'm making food for my brother// Viv pretend I'm making (rice) Aly == I'm making Chinese ri- soup for my brother// Viv yeah// and I'm making soup// In this scenario, the children are conducting the action themselves, as when the mummies in the Linde/Sophie example were describing their own actions while doing them. They use progressive to do this

Pretend play

57

("I'm making Chinese stuff"; "I'm making food for my brother"; "I'm making soup"), since their perspective on the action is quite involved. In addition to describing temporally sequenced action, these children use simple present to talk about habitual or generic action. For example, they describe habitual states ("I look pretty", "we always need it", "pretend I come from- from America"; "I come from China"). Example 5

Aly Viv Aly

Joan Aly Viv Aly Viv

Aly Viv

pretend I'm Chine::se// I'm a Chinese sister/ and I look pretty// pretend this (xx)// pretend (xx) are called (rices)// and those are called chaniza// because they're- they're special kind of tortilla// and we always need itII we make (lots of fresh rice) 'cause ("because") we're Chinese people// pretend/ I'm a Chinese// pretend/ this is cabbage// this is cabbage right? yeah// pretend I'm- pretend I'm American// pretend I come from- from America// = { I come from China [f]}// yeah// 'cause ("because") you're a China person// pretend I come from- from- from Korea 'cause ("because") I'm a Korean person =I'm a= =that's= where my mom comes from/ Korea//

These uses of habitual simple present, particularly the link to "always", tie in with the previous examples from the three year old girls. While the four year old girls in this example have lexicalized the habitual quality of their action with "always", it is clear that both the three and four-year-old girls understand the habitual perspective taken on the events described. As our six examples above show, between the ages of early three and late four, children are learning a great deal about temporality as

58

Jenny Cook-Gumperz and Amy Kyratzis

marked by tense. They show understanding of what we've referred to as the durational problem, and perhaps more importantly, the need to shift perspective on events for different discourse purposes. The ability to shift among the simple present for temporally sequenced action and for giving a generic description and the progressive for achieving a zooming-in perspective is quite advanced. As Langacker (this volume) states, "tense marking on a structural statement specifies the time at which the (already constructed) event representation can be consulted as a way of apprehending the world's structure".

3. Conclusion Children gain a special experience from the textual quality of stories as they are read aloud. An aspect of this textual quality is shifting in perspectival view, which is constituted by tense and aspect shifting. These shifts, in turn, are reinforced by the visual experience (i.e. looking at story pictures) as part of the textual quality in children's books. Perspectival shifts and visual experience taken together provide a textual experience beyond understanding of narrative plot. Through tense/aspect contrasts, children's stories are able to provide more nuanced views of the characters and their situation. These make the oral reading of simply told tales dependent on complex conceptualizations for their interpretation (as afforded by the syntax), and render the story reading interesting to the adult reader as well as to the child listener. From these oral readings, the child gleans a sense of the rich textual experience that stories provide. Although preschoolers are already fairly adept at perspectival shifting in their pretend play, as the data shown here suggests, their ability to recognize and make use of tense shifts is one of the things that changes most between three and four years. The three year olds shown here sometimes seemed to have an unstable use of tense, shown here in the interjection of future modals within present tense (contractions of the future auxiliary, as in "I'll", as though it were part of the pronoun). Moreover, the four year olds had a quite sophisticated shifting between progressive and simple present to mark

Pretend play

59

the contrast between foreground and background events. More practice with these oral narratives, and greater exposure to bookreading, no doubt help negotiate this developmental transition. Another preparation from literacy gleaned from character play is the use of simple present in the habitual sense. This may be an important precursor for literacy events, such as giving definitions and explanations (Kurland and Snow 1990, Aukrust and Snow 1998). In these genres of talk, habituais are common. For example, explanations featured in Aukrust and Snow's (1990) American data contained simple present habituais (e.g. "I like pancakes; They make me strong" (1998: 235) and "like when you talk, it goes on" (1998: 240). Definitions taken from Kurland and Snow (1997) use habituais as well (knife defined as "thing that cuts things" (1997: 609) and umbrella as "something that you hold over your head in case it rains" (1997: 610). The ability to give definitions is an important type of decontextualized language skill and is "related to being part of an academic culture" (Kurland and Snow 1997: 603), hence is an important aspect of emergent literacy. Children who have practice in using habituais in pretend play may have a leg up and a natural segue into literacy-related talk genres such as definitions and explanations, which utilize similar tense-aspect markers, thereby taking similar perspectives or viewpoints on events. However, it is from oral experiences outside of books as well, the character play with peers and parents, that children gain the tools to accomplish tense and aspect shifting, and thereby perspective. In character play, children rework the scripts of everyday experience. Onto these scripts, they import the language of perspective shifting, importing habituais, progressives, and simple presents from their preliterate experience of storybooks to rework and reflect upon their scripts of everyday experience.

Note 1. Names are listed in alphabetical order.

60

Jenny Cook-Gumperz and Amy Kyratzis

References Aukrust, Vibeke Grover and Catherine E. Snow 1998 Narratives and explanations during mealtime conversations in Norway and the U.S. Language in Society, 27: 221-246. Benveniste, Emile 1971 Problems in General Linguistics. Coral Gables, Florida. Brown, Roger 1973 A First Language: The Early Stages. Boston: Harvard University Press. Budwig, Nancy 1995 A Developmental-Functionalist Approach to Child Language. Mahwah, NJ: Lawrence Erlbaum Associates. Cook-Gumperz, Jenny 1985 Keeping it together: text and context in children's language. In: James Alatis and Deborah Tannen (eds.), Languages and Linguistics: The Interdependence of Theory, Data, and Application. Washington, D.C.: Georgetown University Press. 1992 Gendered contexts. In: Peter Auer and Aldo Diluzio (eds.), The Contetxualization of Language. Amsterdam: John Bejamins Ervin-Tripp, Susan M. 1987 Speech acts and syntactic development: Linked or independent? Keynote address to the Boston Child Language Conference, October 1987. 1993 Constructing syntax from discourse. In: Eve V. Clark (ed.), Proceedings of the twenty-fifth annual Child Language Research Forum, 333-341. Stanford, CA: Center for the Study of Language and Information. Geisel, Theodor Seuss 1960 One Fish, Two Fish, Red Fish, Blue Fish. New York: Random House. Gerhardt, Julie and Iskander Savasir 1986 The use of the simple present in the speech of two 3-year-olds: Normativity not subjectivity. Language in Society 15: 501-536. Heath, Shirley Brice 1986 What no bedtime story means: Narrative skills at home and school. In: Bambi B. Schieffelin and Elinor Ochs (eds.), Language Socialization Across Cultures. New York: Cambridge University Press.

Pretend play Hopper, Paul 1979

61

Aspect and foregrounding in discourse. In: Talmy Givón (ed.), Discourse and Syntax. Syntax and Semantics, vol. 12, 213-241. New York: Academic Press. Kurland, Brenda F. and Catherine E. Snow 1997 Longitudinal measurement of growth in definitional skill. Journal of Child Language 24: 603-625. Kyratzis, Amy 1992 Gender differences in the use of persuasive justifications in children's pretend play. In: Kira Hall, Mary Buchholtz and Birch Moonwomon (eds.), Locating power : Proceedings of the Second Berkeley Women and Language Conference 2, 326-337. Berkeley: Berkeley Women and Language Group. Forthcoming Language socialization across time and context: Emotion talk in preschool same-sex friendship groups. To appear in: Nancy Budwig (ed.), Language Socialization during the Preschool Years [Special issue]. Early Education and Development. Kyratzis, Amy and Jiansheng Guo 2001 Preschool girls' and boys' verbal conflict strategies in the U.S. and China: Cross-Cultural and contextual considerations. In: Amy Kyratzis (ed.), Gender Construction in Children's Interactions: A Cultural Perspective [Special Issue]. Research on Language and Social Interaction. Langacker, Ronald W. This volume Cognitive linguistics, language pedagogy, and the English present tense. Milne, Alan Alexander 1926 Winnie-the-Pooh. New York: E. P. Dutton Slobin, Dan I. 1982 Universal and particular in the study of child language. In: E. Wanner and L.R. Gleitman (eds.), Language Acquisition: The State of the Art, 128-170. Cambridge: Cambridge University Press. 1986 Crosslinguistic evidence for the language-making capacity. In: Dan I. Slobin (ed.), The Cross-Linguistic Study of Language Acquisition, Vol. 2: Theoretical Issues, 1157-1256. Hillsdale, NJ: Lawrence Erlbaum Associates. White, Hayden 1980 The value of narrativity in the representation of reality. In: W. J. T. Mitchell (ed.), On narrative, 1-23. Chicago: The University of Chicago Press.

62

Jenny Cook-Gumperz and Amy Kyratzis

Wolf, Dennie and Deborah Hicks 1989 The voices within narratives: The development of intertextuality in young children's stories. Discourse Processes 12(3): 329-351.

The relation between experience, conceptual structure and meaning: non-temporal uses of tense and language teaching Andrea Tyler and Vyvyan Evans

1. Introduction

1.1. The problem It is widely accepted that the primary function of English tense is to signal time-reference (Binnick 1991, Comrie 1985, Jespersen 1924; Langacker 1991, this volume, Quirk et al. 1973, Reichenbach 1947). While this paper is concerned with both tense and language teaching, we will not be primarily addressing the time-reference function associated with tense, nor will we be offering an approach for teaching tense. Nonetheless, the problem we address is directly motivated by an issue faced by language teachers. While tense morphology canonically signals time-reference, it is often associated with other nontemporal functions (Celce-Murcia and Larsen-Freeman 1998, Comrie 1985, Cutrer 1994, Fleischman 1989, Langacker 1991, Quirk et al. 1973, Riddle 1986, Swan and Smith 1987, Ter Meulen 1995, Westney 1994). There are four kinds of non-temporal uses of tense that we will consider. The first function relates to a designation of intimacy between the speaker/s and others. The second relates to what we will term salience (commonly referred to as foregrounding and backgrounding in the discourse literature). The third concerns what we will term actuality, in which tense is used to signal the extent to which the experiencer (or speaker) believes the event described corresponds to the actual world-state and conditions holding (this has

64

Andrea Tyler and Vyvyan Evans

been variously termed epistemic stance, cf. Fillmore 1990, or as a distinction between realis and irrealis). The fourth function concerns what we will term attenuation, in which certain speech acts are "softened" or mitigated in terms of their threat to face (cf. Brown and Levinson 1987). This function is commonly referred to in terms of linguistic politeness phenomena. These functions will be described in greater detail in Section 2. In previous approaches to tense it has been common to distinguish the time-reference function from such non-temporal functions on the basis of "meaning" versus "use" (e.g. Comrie 1985). That is, while the literal meaning of tense relates to time-reference, non-temporal uses are simply uses derived from contextual interpretation. The argument is that in a sentence such as: I just wanted to ask you if you could lend me a pound (ibid.: 19), the addressee can determine from context that the use of the past tense does not relate to a past desire, but rather to a current situation. This common approach has tended to reinforce the view that non-temporal uses should be treated as exceptions. The difficulty for language teachers, and one we have faced ourselves in classroom settings, is how to insightfully present the nontemporal uses associated with tense. The approach offered by received wisdom, as reflected in course books and pedagogical grammars, is to treat them as exceptions, or worse to ignore them altogether. For instance, Westney (1994) has observed that in pedagogical grammars: "[T]ime reference is treated as dominant and other uses are simply appended" (ibid.: 79). Riddle (1986) notes that most pedagogical texts ignore the uses of tense to signal intimacy, salience, and attenuation. If any of them are addressed, the general position is that these uses are arbitrary; presentation of non-temporal usages are often scattered throughout a grammar with no attempt to tie the non-temporal use back to the basic temporal sense. Such analyses have resulted in second language learners being instructed to simply learn formulaic phrases to express polite requests, indirect commands, conditionality, etc. with little or no explanation for why the tense marking in the phrases they are asked to memorize does not correspond to temporal uses of tense. The ultimate result, as Riddle

The relation between experience, conceptual structure and meaning

65

(1986) documents, is that second language teachers are at a loss for a satisfactory explanation of the phenomena and even relatively advanced second-language learners often experience difficulty acquiring these non-temporal uses of English tense.1 Our central thesis in this paper is that the non-temporal uses of tense are related to its time-reference function in a motivated way. Rather than treating tense as being analyzable into a literal timereference meaning and an unrelated range of non-literal exceptional uses, we suggest that two important aspects of the cognitive linguistic perspective allow us to view the non-temporal meanings as being related to the temporal meaning in a systematic way. First, cognitive linguists treat grammatical markers, such as tense morphemes, as being meaning-bearing elements in the same way, in principle, as full lexical items. This entails that tense markers can be treated as formmeaning pairings. The second is that symbolic elements (i.e. formmeaning pairings) are subject to usage-based meaning extensions. As such, through use, additional meanings can become associated with a particular form, resulting in the lexical form becoming related to a semantic network of distinct, although ultimately related, senses. We follow Elizabeth Traugott (e.g. 1989) in referring to this process of meaning extension as pragmatic strengthening. In other words, we argue that the non-temporal meanings associated with tense are conventional meanings or senses associated with a particular lexical form, in the same way as the time-reference meaning. While the time-reference sense might be the diachronically primary meaning element, our approach suggests that distinguishing meanings based on their "literalness" or whether or not a particular meaning counts as an exception may be misplaced. Moreover, we will argue in detail that due to the way in which we actually experience the notions of intimacy, salience, actuality, and politeness, namely in terms of proximal-distal spatial relations, and the fact that time-reference is experienced in terms of analogous spatial relations, in certain situations tense morphemes which canonically signal timereference can implicate a non-temporal relation. Through usagebased conventionalization, i.e. pragmatic strengthening, a conven-

66

Andrea Tyler and Vyvyan Evans

tional non-temporal meaning can become associated with a particular tense morpheme. The main purpose of the present paper, as we noted above, is not to offer a new way of teaching tense, whether we are concerned with temporal or indeed non-temporal meanings. Rather, the aim of presenting the rather elaborate study outlined below, is to illustrate that by assuming the perspective and methodology of cognitive linguistics it is possible to relate the non-temporal senses associated with tense with the time-reference meaning in a plausible way. This approach, we suggest, should facilitate language teaching by providing a systematic model of the links among the semantics of tense markers, which is to say between temporal and non-temporal meanings. Langacker (this volume) suggests that cognitive linguistics may "prove to be lighter, less onerous, and more appreciative than certain previous theoretical burdens. I hope it will even prove useful". We offer the present analysis in the same spirit, and suggest that by relating the various meanings associated with tense morphology, language teachers have a model by which to understand and so relate the diverse functions performed by tense markers. It is our hope that the proposed theoretical perspective may assist in the language classroom.

1.2. Previous analyses and overview of the proposed solution

Our goal is to present a motivated approach to the multiple uses of English tense. We argue that there is an experientially-based commonality reflected at the conceptual level that holds between the concepts of time-reference (externalized linguistically via tense), and intimacy, salience, attenuation, and actuality; this commonality motivates the use of the tense morphemes to implicate the multiple, attested meanings associated with tense. As such, the tense morphemes, which primarily encode time-reference, have come to be associated with non-temporal meanings. In our account, we depart in certain ways from previous explanations of tense within cognitive linguistics (e.g. Dirven and Radden 2000, Langacker 1991). We note two heretofore unaddressed prob-

The relation between experience, conceptual structure and meaning

67

lems with the analyses of the usage of English tense morphology. The first involves a contradiction between the representation of the basic sense of the past tense morpheme and several of its non-temporal uses. Langacker (1991), whose work forms the basis of many subsequent analyses, represents past time as being equated with "nonimmediate reality" (ibid.: 242), and present time with "immediate reality" (ibid.: 242). The situations or states of affairs being referenced by the English tense morphemes are represented as known reality, i.e. they "are accepted by a conceptualizer as being real" (Langacker 1991: 242). Reality status is argued to be signaled linguistically by the presence or absence of modal marking. Past time, which largely equates with non-immediate reality, is signaled by the past tense morpheme; present time, which largely equates with immediate reality, is signaled by present tense morphology. In contrast, future and modal forms are represented as signaling irrealis. However, English speakers regularly use the past tense morpheme to signal relatively less commitment to the reality of an event or state of affairs; past tense is also used in certain politeness phenomena, which are clearly not interpreted as states of affairs accepted as real. These uses are at odds with the "known reality" representation. The second problem concerns the explanatory power of certain claims about metaphorical distance and tense morphology. Recognizing that English tense morphemes are not solely interpreted in terms of temporal relations, Langacker (1991) argues that there is an epistemic opposition between "immediate and non-immediate reality" (ibid.: 245-246) which is marked by the absence or presence of past tense morphology;2 the present and past tenses contrast with "immediate and non-immediate irreality" (ibid.: 245-246) which are marked by various modal forms. On this basis, he labels the past tense morpheme as a "distal marker". Significantly, his discussion in relation to the past tense morpheme centers on temporal distance. Expanding on Langacker's analysis, Dirven and Radden (2000: Chapter 9) note that a sentence such as: I wanted to ask you a favor "illustrates a metaphorical shift of the past tense...The use of past tense in [I wanted to ask you a favor] achieves an effect of politeness; it distances the situation in time and, as a result makes the request

68

Andrea Tyler and Vyvyan Evans

seem less face-threatening." What is left unspecified in such an account is why placing a speech event in the past, i.e. in non-immediate reality, where it would normally be assigned the interpretation of known (or established) reality, should be understood as less facethreatening. For instance, the normal interpretation of: I wanted to ask you a favor still involves the understanding that the speaker is about to ask a favor. Why should establishing that the speaker had a favor to ask before the speech event, a state which continues at the time of the speech event, result in less face-threat? Simply saying this is an example of metaphoric shift and "temporal distance" (which itself is a metaphor that brings together two distinct domains - the temporal and spatial) does not actually offer an account of the phenomena. Crucially missing from this metaphorical analysis is an explanation of the relationship between 'known reality', time, and cognitive distance. Intuitively it seems that tense-related politeness phenomena do involve a notion of distance, but the distance seems to be interpreted in a non-temporal way. The nagging question remains as to why a morpheme which primarily signals non-immediate reality or, metaphorically speaking, "temporal distance", should come to be associated with "non-temporal distance". The analysis we put forward in this paper seeks to offer an explanation for these apparent dichotomies. In particular, we expand on two fundamental insights initially discussed in Grady (1997, see also Evans 2000). First, many linguistic phenomena which have previously been treated homogeneously as metaphor are more accurately understood as arising from different sources; the most important for our purposes involves experiential correlation, i.e. independently motivated and recurring correspondences in experience. Second, previous accounts of conceptual structure have tended to represent concepts pertaining to internal states as abstract and indirect while representing concepts pertaining to experiences with the external world as concrete and direct. Grady (1997) persuasively argues that this representation is questionable and that a more appropriate distinction can be made in terms of image and response concepts. Image concepts pertain to sensorimotor

The relation between experience, conceptual structure and meaning

69

information, derived from external experience. In contrast, response concepts pertain to information arising from internal states. Specifically, we suggest that the concepts of time-reference, intimacy, salience, actuality, and attenuation all represent response concepts that are structured or elaborated in terms of the same experientially-based image content. In certain contexts, tense (which denotes time-reference) may implicate one or another of these concepts. Through continued use, these implicatures can come to be conventionally associated with the tense morphemes. This process of meaning extension we term, following Traugott (e.g. 1989) and Hopper and Traugott (1993), pragmatic strengthening. In essence then, we will argue that morphologically bound English tense forms (past and present tense markers) are meaningful elements which typically prompt for time-reference meaning. However, these forms have been extended to convey non-temporal meanings in systematic and motivated ways. In view of the foregoing it is worth briefly contrasting the nature of the present analysis of tense with that presented by Langacker (this volume). In his paper, Langacker argues that canonically "the present tense indicates that a full instantiation of the profiled process occurs and precisely coincides with the time of speaking...[while]...[t]he past tense morpheme imposes an immediate scope located prior to the speech event" (ibid.: 22). Put another way, for Langacker, coding time (the time of speaking), and reference time exactly co-occur in canonical uses of the present tense. One of the issues that his paper addresses is the way in which present tense is employed in situations in which the present tense, for instance, does not refer to a situation in which coding time and reference time coincide. By way of example, a speaker who lives in London but works in New York might say: "I work in New York", even though it is the weekend, and coding time coincides with not working and being in London as opposed to New York. Langacker is at pains to address such 'non-present' uses of the present tense in terms of the possibility of multiple construals or viewing arrangements that change the interpretation of the particular linguistic element under examination. In some instances "what is being coded linguistically is not the actual occurrence of

70

Andrea Tyler and Vyvyan Evans

events, but their virtual occurrence as part of a non-canonical viewing arrangement" (ibid.: 30). While we in large measure subscribe to Langacker's stance, it is worth pointing out that in the present paper we are not dealing with 'non-present' uses of the present tense, in the sense of Langacker. Such 'non-present' uses are still, after all, temporal in nature. Rather, we seek to examine why tense morphology can be employed to code meaning elements which are non-temporal in nature. After all, politeness, for example, is not in any obvious way a temporal notion. Yet, in English, and indeed, in a range of other languages, notions such as politeness can be designated by utilizing tense morphology (cf. Fleischman 1989). Our approach, as intimated above, and as will become evident, is to claim that there are distinct meaning elements or senses, relating to concepts such as politeness, which are conventionally associated with the past and present tense morphemes. While construal is an important part of understanding non-canonical temporal uses of tense, some non-temporal meanings associated with tense are, we argue, due to the polysemy exhibited by tense morphology. This parallels the polysemy exhibited by free morphemes such as the much studied English preposition over, for instance (cf. Tyler and Evans forthcoming a, forthcoming b). The remainder of the paper is organized in the follow manner. In Section 2, we briefly present the linguistic phenomena under consideration. In section 3, we lay out the theoretical foundations for our analysis. First, we examine the evidence in favor of adopting the distinctions of image concepts versus response concepts rather than concrete concepts versus abstract concepts (contra Lakoff and Johnson 1980, 1999). We further argue that all the meaning elements associated with English tense that we discuss are best understood as response concepts. Next, we turn to Grady's insights regarding experiential correlation. We argue that experiential correlation provides a powerful mechanism for explaining how distinct events come to be associated at the conceptual level. In Section 4, we examine the nontense linguistic patterns associated with the domains of temporal reference, intimacy, salience, actuality, and attenuation and show that all are elaborated through distal-proximal image content. In Section 5, we argue that the fact that all these concepts are elaborated by the

The relation between experience, conceptual structure and meaning

71

same image content links them at the conceptual level. This conceptual linking, then, licenses the use of tense morphology to signal these various uses. Finally, in Section 6, we sketch some of the implications of the analysis for language pedagogy.

2. The phenomena In order to give a sense of the various meanings of tense to be investigated, we present some examples below. We also emphasize that we will only be considering the synthetic tense forms in English, namely the present and past tense bound morphemes. The present tense morpheme we will represent by [zero/s] or [0/s], to capture the fact that this tense morpheme constitutes the default verb marking. The past tense morpheme we will represent by [Id], which represents a generalization over regular and irregular past tense verb marking in English. These two tense morphemes can be employed to signal the following meanings: Time-reference (1) a. I work in advertizing (present time-reference) b. Yesterday I went to the cinema (past time-reference) Intimacy (2) a. A: Jane just bought a Volvo. B: Maureen has one. A: John, you've got to quit talking about Maureen as if you're still going together. You broke up three months ago. (Riddle 1986) The context in which the utterance occurred is that Speaker B, John, and Maureen were previously in an intimate relationship but the relationship ended and the two have not seen each other for some time. Speaker A interprets Speaker B's use of the present tense as an unjustified claim to intimacy.

72

(2)

Andrea Tyler and Vyvyan Evans

b. My daughter's father was Brazilian. He stays in contact with Suzanna, but I haven't seen him in years. (Tyler, personal data)

The past tense appears to be employed by the speaker to signal emotional distance from Suzanna's father. Given the context, we cannot interpret this use to mean that the father is dead or no longer Brazilian. Nor can we assign the interpretation that the man from Brazil is no longer in the role of father. Salience (3) a. In November 1859, Charles Darwin's The Origin of Species, one of the greatest and most controversial works in the literature of science, was published in London. The central idea in this book is the principle of natural selection. In the sixth edition.. .Darwin wrote: "This principle of preservation of the survival of the fittest, I have called Natural Selection." (Eigen and Winkler 1983: 53) Tense is employed to signal the relative status of the information or salience, i.e. past tense signals background and supporting status and present tense signals foreground status. (3)

b. Bateson introduced the notion of frame in 1955 to explain how individuals exchange signals that allow them to agree upon the level of abstraction at which any message is intended. Even animals can be seen to use frames to interpret each other's behavior, by signaling, for example, "This is play." Bateson (1972) insists that "frame" is a psychological concept, but to characterize it he uses "the physical analogy of the picture frame and the more abstract...analogy of the mathematical set", (p. 186) (Tannen 1993: 18)

Again, past tense is used to establish the background information and present tense is used to foreground the information the author deems most salient.

The relation between experience, conceptual structure and meaning

73

Actuality (4) a. I wish I knew what he'll say next. (Westney 1994) b. I wish the students liked phonetics. (Fleischman 1989) Past tense is used in conjunction with specific lexical items, such as wish, to signal a state of affairs, which the speaker believes to differ substantially from actuality. c. Suppose your house burned down. Do you have enough insurance? (Frank 1983) Past tense is used to signal that a situation is hypothetical, i.e. a situation which given the current world-state does not hold, and hence is imaginary or contingent in some way. d. If he studied harder now, he would get better grades. (Frank 1983) Past tense in the if clause (the protasis) signals a negative stance visà-vis the conditional situation and reality, i.e. the situation described is non-actual. The sentence roughly paraphrases as "He is not studying hard at the moment. If he changes the present circumstances, and he studies harder, it would be possible for him to get better grades." Attenuation (5) a. I was thinking about asking you to dinner. (Fleischman 1989) b. I was hoping we could get together next week. (Fleischman 1989) Past tense is used to attenuate invitations, and hence to decrease the threat to face for the hearer (e.g. the imposition of an invitation which must be responded to immediately), and to the speaker (e.g. possible immediate rejection of the invitation), (cf. Brown and Levinson 1987), and is thus conventionally interpreted as polite.

74

Andrea Tyler and Vyvyan Evans

c. Receptionist (answering telephone): Good afternoon, Dr. Keller's office. Caller: Yes, I wanted to ask you a question. (Davies personal communication) Past tense is used to attenuate requests. d. Did you want to take a look at this? (Fasold personal communication) e. I thought you might like to try this (Westney 1994) Past tense can be used to attenuate suggestions. f. It's high time we left (Westney 1994) Past tense attenuates commands and reprimands, rendering them conventionally more polite. In the foregoing examples we have seen that tense is crucially employed in English to signal a number of distinct and fundamentally non-temporal meanings. In (2), tense is used to signal intimacy. In (3) tense signals the relative salience or status of the information being conveyed. In (4), tense is employed to signal the stance towards the actuality of a particular scenario, i.e. speaker's degree of commitment to the reality of the scenario, or alternatively how likely it is to come about. Finally, in (5), tense can have an attenuating function in requests, commands and invitations, mitigating the amount of imposition on the addressee or mitigating potential threats to our public persona or face. In the remainder of this paper, we explore the hypothesis that the various meanings associated with tense are highly motivated, being related in a systematic, principled way. Thus, we attempt to move to a deeper level of explanation of the phenomena than afforded by previous metaphor-based accounts of tense.

The relation between experience, conceptual structure and meaning

75

3. The nature of conceptual structure

3.1. Image concepts versus response concepts One of the most important contributions of cognitive linguistics to our understanding of language has been the insight that much of everyday language involves discussing/understanding one concept in terms of another. Within cognitive linguistics it has traditionally been assumed that there is an a priori distinction to be made between abstract concepts, which are not directly perceived and with which we have less experience, e.g. time, and emotion concepts such as anger, etc., and concrete concepts, which are directly perceived and with which we have more extensive experience, e.g. motion, heat, etc. (Lakoff and Johnson 1980, 1999). It has been further argued that it is the distinction between direct and indirect perceptual experience that is responsible for the organization at the conceptual level of abstract concepts in terms of concrete concepts. Hence, an utterance such as Time flies is held to demonstrate that English speakers conceptualize and accordingly lexicalize the abstract, less directly experienced notion of time in terms of the more directly experienced notion of a physical entity in motion. However, it is not at all clear that so-called abstract concepts are less directly or less fully experienced than so-called concrete concepts. For instance, Ortony (1988) observes that putative abstract emotion concepts such as love are experienced by children much earlier and more extensively than putative concrete concepts such as journeys, which Lakoff and Johnson (1980, 1999; Lakoff 1987, 1993) and others have claimed serve to structure the abstract concepts. More recently, Grady (1997) has pointed out that while socalled abstract concepts may not result from external sensorimotor input (i.e. perceptual experience from the external world), this does not entail that they do not derive from equally direct experience. Grady's argument is that so-called abstract concepts pertain to internal, subjective states, but it does not follow that such internal states are not directly experienced. As Grady observes: "[T]he awareness

76

Andrea Tyler and Vyvyan Evans

that we are conscious (as opposed to unconscious) is perhaps the quintessential subjective experience, and yet we do not consider this awareness to be merely an intellectual construct or abstraction - it is a real and direct experience." (Grady in preparation: Chapter 5, 22). In a detailed study of the conceptual system for time, Evans (2000) argues that the concept 'time', which has previously been assumed to be a parade example of an 'abstract' concept, in the sense of an intellectual construct, can be traced to physiological mechanisms and processes. This work offers support for Grady's suggestion that 'abstract' concepts derive from frequently experienced internal states which humans are aware of at the level of phenomenological experience. As such, these experiences would seem to be just as basic as those arising from external sensorimotor input. Thus, there appears to be little justification for distinguishing many concepts based on their level of abstractness (Evans 2000, Grady 1997, in preparation). In addition, it is also worth reminding ourselves that even sensorimotor experience of the world 'out there' is mediated by our particular physiology and neurological architecture. As the developmental psychologist Jean Mandler (1992) points out, information arising from observation and experience of the real world does not directly enter into consciousness in an unmediated form. Even the most basic concepts are represented in the human conceptual system as redescriptions and only then are accessible to conceptual structure. That is, sensorimotor experience itself cannot be assumed to be 'direct'. Accordingly, following Grady, we suggest that the bifurcation in the nature of concepts has little to do with whether a concept is abstract or concrete, but rather is determined by what kind of information the concept represents a redescription of. In essence, a more accurate distinction may be that while image concepts represent redescriptions of 'external' i.e. sensorimotor experience, response concepts represent redescriptions of internal states (cf. Evans 2000). Perceptual information derived from external sensorimotor information and redescribed into a format accessible to the conceptual system we term image content. This represents substrate available at the conceptual level which serves to derive image concepts, such as

The relation between experience, conceptual structure and meaning

77

motion, heat, etc. Perceptual information derived from internal states (also redescribed into a format accessible to the conceptual system) gives rise to what we term, following Grady (1997), response content. Response content gives rise to response concepts. Response concepts derive from internal body states. In some cases, these internal states themselves may ultimately be responses to external information, hence the term response. For instance, love refers to an internal state, yet at the same time constitutes an emotional response to another being in the external world. What is clear from the foregoing is that while response concepts lack the 'objectivity' of image concepts, such as those pertaining to physical proximity or physical distance (deriving from sensorimotor information), they are no less basic (Grady 1997). As concepts such as time-reference (in the sense of temporal deixis rather than clocktime), intimacy, salience, actuality and attenuation are internal in nature and hence subjective, rather than being external and intersubjective, it follows that they constitute response concepts. The next question to consider is why it should be that response concepts tend to be elaborated in terms of image content. One possible explanation comes from the variability across individuals and the non-verifiability inherent in internal states.3 Perhaps because we do not have access to each other's minds, when we attempt to externalize information about our internal states through language, humans have resorted to talking about such internal states in terms of experiences which are verifiable, inter-subjective and so consistent across individuals. The issue may not be that the internal experience is less direct or less fully experienced than the sensorimotor experience for the experiencer/speaker but that the speaker's internal experience is less direct for the listener. A second explanation for why response concepts tend to be elaborated in terms of image concepts, may be because the parts of the brain which process response information may be less well connected to the conceptual system (the repository of concepts which language externalizes) than the parts of the brain which process image information (Evans 2000, Grady 1997, cf. Jackendoff 1987, 1992). In evolutionary terms it makes sense that the conceptual system should

78

Andrea Tyler and Vyvyan Evans

have better access (in terms of better neurological connections) to sensorimotor information than to response information. In terms of ecological viability, it is more pressing that external information be more readily available to the conceptual system (and hence consciousness) than response information (cf. Edelman's 1992 suggestions regarding the evolution of consciousness). Since response content may be only partially accessible to the conceptual system, response concepts may opportunistically appropriate image content in order to develop structure accessible (i.e. "visible") to the conceptual system, and hence available for encoding by language (Evans 2000, Grady 1997). On this view, it is due to the elaboration of a response concept in terms of image content that a concept such as time-reference (externalized by the tense morphemes [0/s] and [Id]) is elaborated in terms of locational content pertaining to proximal-distal relations, rather than in its own terms (whatever such terms might be). However, this does not deny that at least some of the content which elaborates a response concept is response content, as attested by the fact that we intuitively "know" what it is that tense symbolizes, without being able to adequately express this without the assistance of image content such as time-lines etc. (cf. Comrie's 1985 explanation of tense for instance, which appeals to notions such as a time-line in order to explain time-reference phenomena). Hence, the hypothesis that this response content is only partially accessible to the conceptual system explains 1) why response concepts such as time-reference are difficult to define, and 2) why response concepts tend to be expressed linguistically in terms of image concepts such as 'distance'. We now turn to a consideration of the mechanism which serves to elaborate response concepts in terms of image content.

3.2. Experiential

correlation

One of the remarkable insights to have emerged from cognitive linguistics has been the realization that conceptual structure is largely organized in terms of substrate deriving from external sensorimotor

The relation between experience, conceptual structure and meaning

79

experience, which we are identifying as image content. However, it remains to be explained what mechanism motivates response concepts to appropriate certain kinds of image content and not others. A promising candidate mechanism is experiential correlation, which has been studied in detail by Grady (1997, 1999a, 1999b, in preparation). Grady notes that a consequence of the nature of the interaction between humans and their environment is that certain kinds of experiences are frequently correlated. For instance, a common, recurring experience in the world is the correlation between the vertical elevation of a physical entity and an increase in the quantity of the entity, which is to say when there is an increase in vertical elevation, there is frequently a correlative increase in the original amount of the entity. By way of illustration, if there are two boxes stacked one on top of the other and a delivery person adds another two boxes to the stack, the height of the stack increases. Hence, height (vertical elevation) and number of boxes (quantity) are correlated in our experience. Similarly, if there is a certain amount of liquid in a container, and more liquid is added, the level of the liquid rises. So, humans frequently experience greater quantity in terms of an increase in vertical elevation. Grady argues that tight and recurring correlations in experience motivate associations at the conceptual level. Hence, as quantity and vertical elevation are correlated in experience, they come to be linked at the conceptual level, as attested by the example in (6): (6)

Prices have gone up recently.

This sentence has a conventional interpretation in which prices have increased. Yet, this reading is achieved by utilizing the linguistic prompt gone up. The point is that language systematically utilizes expressions, which conventionally denote vertical elevation to provide an interpretation of greater quantity. We suggest that correlations in experience serve to relate certain kinds of image content with specific response concepts. Kurath (1921, cited in Sweetser 1990) studied Indo-European words for emotions and noted that they are often derived from the physical ac-

80

Andrea Tyler and Vyvyan Evans

tions or sensations accompanying relevant emotions. Kurath attributed this historical development "to the inseparability of physical sensation from emotional reaction, or of emotional state from concomitant physical changes" (Sweetser 1990: 29). Putting this in present terms, the response concept 'anger' is elaborated in terms of image content pertaining to heat, precisely because anger and a physical sense of feeling hot (caused by increased adrenaline and increased heart rate) form a tight correlation in experience.4

4. Concepts elaborated by proximal-distal content During the course of this paper, we will argue that time-reference, the primary meaning associated with tense forms, comes to implicate non-temporal meanings, which in turn through entrenchment (a process we term pragmatic strengthening) comes to be conventionally associated with the tense morphemes. However, in order to able to offer a motivated account for the association of non-temporal meanings with the two tense morphemes, we need to establish why it should be that time-reference should implicate intimacy, salience, actuality and attenuation in the first place. We will argue that all five of these response concepts are elaborated in terms of locational content pertaining to the relative physical proximity of the experiencer because we conceptualize each of the concepts of time-reference, intimacy, salience, actuality and attenuation in terms of proximaldistal relations with respect to the experiencer. Hence, these distinct concepts, while not literally being spatial concepts are all elaborated in terms of spatial deixis, which, as we will demonstrate, is motivated in each instance by a distinct, tight correlation in experience. It is by virtue of being elaborated in terms of similar image content, we will suggest, that the tense morphemes, which denote time-reference, can come to implicate and ultimately denote non-temporal meanings. In the remainder of this section, we examine the evidence for experiential correlations which give rise to each of these distinct concepts being associated with the proximal-distal dimension. We will

The relation between experience, conceptual structure and meaning

81

also illustrate how these experiential correlations are reflected in nontemporal language patterns.

4.1. Time-reference As noted by a number of scholars (e.g. Comrie 1985, Rauh 1983, Traugott 1975, 1978) tense is a deictic phenomenon which signals time reference with respect to a reference point or deictic center. The reference point typically assumed is coding time, or time of speaking. Coding time is a temporal concept, which in principle is distinct from spatial image-based content. However, it quickly becomes clear, at least for English speakers, that it is virtually impossible to conceptualize temporal deixis without appealing to spatial image content. This is implicit even in formal definitions of tense. For instance, Comrie (1985) characterizes tense as "the grammaticalisation of location in time" (ibid.: 1). Given that location is a spatial notion and hence pertains to image content, by virtue of offering a definition of timereference in such terms, it is evident how deep-seated the conceptualization of time-reference is in terms of the locational information. Grady (1997) has noted that in experiential terms there is a tight correlation between the temporal concept of 'now' and the particular physical location, which is proximal to the human experiencer, i.e. 'here'. In other words, we cannot help but experience the present moment in terms of our immediate physical surroundings and our sensory perceptions of them. This tight correlation in experience between the present moment and the particular location we happen to occupy, motivates the elaboration of the concept identified by the lexeme now in terms of our experience of our physical location and the vicinity proximal to us. Extending this insight, we note that 'not now' is similarly experienced (through memory or imagination) as a location other than the immediate 'here'. In other words, 'not now' is experienced as 'not here.' A second experiential source linking time and space involves the human experience of getting from point A to point B. Traversing a certain distance inevitably correlates with the elapse of a certain

82

Andrea Tyler and Vyvyan Evans

amount of time. Thus, elements of the spatial domain, such as movement from one location to another and distance, have become strongly associated with the elapse of time. This is reflected in many aspects of language. For instance, it is not uncommon to hear exchanges along the lines of the following: (7)

A. How far is the restaurant from here? B. Oh, about a five minute walk.

Given that tense morphemes typically obtain their time reference with respect to the temporal 'now' (the coding time), which is elaborated at the conceptual level in terms of locational image content, and the experiential correlation between traversing a particular distance and the elapse of time which co-occurs with the traversing, it is hardly surprising that we should find that time-reference is conceptualized and lexicalized in terms of physical proximal-distal relations with respect to the experiencer. It is this elaboration, we suggest, which may be largely responsible for our ability to conceptualize temporal events in terms of physical location and distance, as attested by expressions such as: Christmas is getting close; The present moment has arrived, etc. (cf. Evans 2000, Grady 1997, Moore 2000). Moreover, expressions such as the near past, the distant past, etc., which employ the spatial language near and distant in order to signal time-reference with respect to coding time, illustrate that timereference is elaborated in terms of physical proximal-distal relations vis-à-vis the experiencer. In many ways, this account of the cognitive relationship between time and space may seem very like that for tense articulated by Dirven and Radden (2000), and Langacker (1991), and indeed the more general account of the concept of time offered by Lakoff and Johnson (1999). We readily acknowledge that the present account builds on those earlier ones. However, we believe the explanatory value provided by the notion of experiential correlation adds substantially to our understanding of how humans come to understand one concept from a distinct domain of experience, in terms of another, and begins to provide the illusive explanation for the apparent contradiction

The relation between experience, conceptual structure and meaning

83

between the primary sense associated with the tense morphemes and their non-temporal uses.

4.2. Intimacy Due to the nature of human interaction, there is a tight and recurring correlation in experience between intimacy and physical proximaldistal relations. In physical terms, two people cannot be intimate, e.g. touch, kiss, have face-to-face conversations, etc., unless they are physically proximal. In terms of familial relations, which are typically considered to be more intimate than the relationship between acquaintances, family members tend to spend a greater proportion of their time in physical proximity with each other than with acquaintances or casual friends. Consequently, there is a tight correlation between intimacy and physical proximity. We suggest that due to this recurring experiential correlation between intimacy and physical proximity, intimacy (which is a response concept), is elaborated at the conceptual level in terms of physical proximal-distal relations. This conceptual elaboration is attested by expressions such as the following: (8)

Peggy and I have been close for many years, but lately she has been acting a little distant.

In this sentence, the notion of physical proximity, as denoted by the word close, has a conventional reading of intimacy and that of physical distance, as denoted by the word distant, has a reading of lack of intimacy. In fact, the reading is so highly conventionalized that on first inspection we might fail to notice that the literal meanings of close and distant are not of intimacy and non-intimacy but rather of physical proximity and distance. The point is then, that the nature of experience, and particularly recurring correlations, gives rise to response concepts such as intimacy becoming elaborated at the conceptual level in terms of proximal-distal content. Hence a close rela-

84

Andrea Tyler and Vyvyan Evans

tionship is one which involves intimacy, while a distant relationship is lacking in intimacy.

4.3. Salience Salience is another concept elaborated in terms of image content pertaining to the physical experience of being proximal or distal. Due to the nature of our sense organs, particularly our eyes, that which is physically closer to us is more salient, that which is at a distance less salient. That which is closer tends to be that which is in foveal vision and more clearly observable, while that which is physically distant tends to be in peripheral vision and less clearly observable. That which is closer appears to be relatively larger, that which is distant appears to be relatively smaller. Entities which are located physically closer to humans, or events which take place physically closer are more likely to demand immediate attention than those which are physically distant. Thus, there is a tight correlation between the salience of an entity and how close it is to the experiencer. By way of example, in a situation in which a human can see two tigers, the tiger which is ten feet away is likely to be more salient than the one a quarter of a mile away. This tight experiential correlation between salience and physical proximity in human experience is illustrated by the following: (9)

We have to keep focused on the pressing issues of the day, those which are close at hand, not some distant threat or peripheral controversy.

In this sentence, the lexical items focused, pressing and close at hand, which represent image content pertaining to physical proximity and distance, are used to indicate which issues the speaker holds to be most salient; the items distant and peripheral are used to indicate those which the speaker holds less salient or important.

The relation between experience, conceptual structure and meaning

85

4.4. Actuality The next concept we examine is that which we are terming actuality. By actuality we mean that which a person believes to be objectively true and reliably known, particularly about the current world-state. Generally our cognitive commitment to (and hence belief in) the actuality of something which can be verified perceptually is much stronger than to something which we cannot verify. Given the physiological constraints on human sense-perception, that is, given that the ability of our sense organs allows us to see, hear, smell, etc. most acutely that which is within an area which is physically proximal to us, our beliefs about what currently holds in the world, correlate tightly with what we have personally experienced, by virtue of our sense-perceptory apparatus. Given that our sense-perceptory apparatus gathers information about our physical proximity, our actuality correlates with that which is physically proximal. This correlation is attested by expressions such as: (10) a. I saw it with my own eyes. b. I know someone was in the house. I heard the sound of footsteps with my own ears. Moreover, it follows that while we can be sure of that which is verifiable by our own senses, we are less sure of that which is not available to our own senses. Thus, we are cognitively committed to what is proximal and physically verifiable and we conceptualize these entities and events as constituting our actuality; we are much less committed to the actuality of that which is distant and not physically verifiable. In view of the foregoing we suggest that actuality is elaborated at the conceptual level in terms of content pertaining to that which is physically proximal to the experiencer.

86

Andrea Tyler and Vyvyan Evans

4.5. Attenuation The relation which concerns us here is that between attenuation of invitations, requests, suggestions, etc., on the one hand, and proximal-distal relations between the speaker and the addressee on the other. Attenuation represents a conventionalized linguistic means of mitigating the imposition placed on the addressee by a request, question, etc., and a means of mitigating the potential threat to face on the speaker, should the hearer fail or refuse to comply (cf. Brown and Levinson 1987). Like the other concepts considered in this paper, attenuation can be elaborated at the conceptual level in terms of physical proximity-distance. Talmy (1988) noted that real world force dynamics, in terms of barriers and forces needed to overcome those barriers, provides a powerful model for our understanding of much of the rest of our experience, including social interactions. Behind our understanding that certain linguistic acts (such as requests, commands, etc.) place impositions on the addressee, is a complex set of notions involving authority, status, desire to keep social relations in balance, and desire to be seen as cooperative, which are themselves largely seen in terms of force dynamics. Sweetser (1990) argues that language allows us to affect one another's actions without having to rely on physical force. "[Language] is ... our major means of intellectual and emotional influence on each other. As linguistically capable human beings, we have no need to constantly resort to physical pushes and pulls to influence other speakers of our language; we can do so in a far more sophisticated and effective manner via the vocal organs and the auditory sense-channel" (ibid.: 41). She adds, "In the real world, we don't usually use force unless we need to overcome reluctance on the part of the person we are forcing. ... In the real world, force is usually resented by the victim because freedom is valued" (ibid.: 61). In sum, Sweetser argues that consideration of socio-physical force lies behind many speech acts and much of politeness phenomena. In terms of politeness specifically, Sweetser notes that it is less face-threatening to enable your interlocutor to be cooperative rather than to evoke your restrictive powers of authority. Drawing on Prin-

The relation between experience, conceptual structure and meaning

87

ciple 3 from R. Lakoff s (1973) discussion of politeness: "Make the interlocutor feel good. Be friendly" - Sweetser notes that "[this] goal [is] best achieved by minimal exercise of overt authority." (ibid.: 153) One important way in which speakers give the appearance of lessening their overt power and simultaneously giving their addressee the option to be cooperative (R. Lakoff s 2nd principle "Give options"), is to linguistically implicate that the speaker is not in the position to force compliance. English has developed a number of conventional ways of enacting such mitigation. We believe that one of the most pervasive has to do with implicating that the speaker is physically distant from the addressee. In humans' everyday experience, there is a tight correlation between physical proximity and one's ability to affect an entity. For instance, if the experiencer is proximal to an object he or she can pick the object up, scrutinize it and manipulate it. If the experiencer is located away from the object he or she is unable to affect the object to the same degree. This experience is reflected in linguistic examples such as the following: (11) She kept a tight grip on the budget. In this sentence, the degree of control over and hence ability to affect the budget is articulated by the phrase tight grip, which literally denotes very close physical contact. Examples such as these are licensed by virtue of the experiential correlation between the ability to affect something and physical proximity. Experience also tells us that physical distance results in a lessening of the ability to affect a particular entity. This is reflected in sayings such as: (12) When the cat's away the mice will play. Invitations, requests, etc. create a situation in which the addressee's actions are potentially affected by the speaker. Attenuation represents an attempt on the part of the speaker to imply a lessening of his/her ability to affect the addressee. Given that being located away from an object correlates with the experiencer being unable to affect the ob-

88

Andrea Tyler and Vyvyan Evans

ject, we suggest that attenuation is elaborated in terms of locational content in which the speaker is physically distal with respect to the entity being affected, i.e. the addressee.

5. Accounting for non-temporal uses of tense To this point, we have argued that the concepts time-reference, intimacy, salience, actuality and attenuation represent response concepts, in part elaborated in terms of response content. As response content derives from internal states, it is less accessible to the conceptual system than externally derived image-based content. Hence, response concepts appropriate image based-content in order to elaborate themselves. Following Grady (1997) we suggest they do so by exploiting naturally occurring and ubiquitous correlations in experience. Each of the response concepts dealt with correlates with proximal-distal content with respect to the experiencer. Hence, each of these concepts shares similar image content, as depicted in Figure 1. Response Content

Response Concepts Legend:

Image Content

Image Concepts

1: time-reference; 2: intimacy; 3: salience; 4: actuality; 5: attenuation; 6: proximity; 7: distance

Figure 1. The elaboration of response concepts in terms of image content

The unshaded spheres on the left correspond to the response concepts: time-reference, intimacy, salience, actuality and attenuation

The relation between experience, conceptual structure and meaning

89

respectively. The two shaded spheres on the right correspond to proximity and distance, which represent redescriptions of sensorimotor information (image content) pertaining to physical proximaldistal relations. The response content is shaded in grey, indicating it is only partially accessible to the conceptual system, as further indicated by the broken arrows between the response content and the response concepts. Accordingly, image content (shaded in black) which is wholly accessible to the conceptual system is appropriated so as to make the response concepts accessible to the conceptual system. As noted earlier, in this paper we are dealing with the two bound tense morphemes in English namely [0/s] and [Id], which correspond to present and past tense respectively. Such morphemes constitute form-meaning pairings. Given the experiential correlation between the temporal reference point, i.e. the experiencer's awareness of now, and the location occupied by and proximal to the experiencer, time-reference is elaborated in terms of proximal-distal content, such that the present tense lexicalized by [0/s] is elaborated in terms of locational content proximal to the experience, and the past tense, lexicalized by [Id] is elaborated in terms of locational content distal with respect to the experiencer. We have also argued that other concepts such as intimacy, salience, actuality, and attenuation are elaborated in terms of similar locational content. Accordingly, we hypothesize that due to the parallel conceptual elaborations, in certain contexts the use of tense to signal time-reference can implicate a distinct concept, such as intimacy for instance. Through continued use, we suggest that tense morphology has become associated with the additional meaning, a process we term pragmatic strengthening, such that a new meaning component becomes associated with the tense morphemes.

5.1. Tense and intimacy

Now we return to the examples of non-temporal uses of tense in (2) through (5) with which we began our discussion. Turning first to intimacy, we have argued that intimacy is elaborated at the concep-

90

Andrea Tyler and Vyvyan Evans

tual level in terms of locational content proximal to the experiencer. This is due, we suggested, to the tight correlation in experience between intimacy and physical proximity. Now let us examine the example in (2) reproduced below: (2)

a. A: Jane just bought a Volvo. B: Maureen has one. A: John, you've got to quit talking about Maureen as if you're still going together. You broke up three months ago. (Riddle 1986)

In this example the use of the present tense form has by B, is interpreted by A as a claim of (unwarranted) intimacy. John does not have direct knowledge as to whether Maureen still owns a Volvo. He is speaking as if they are in an on-going relationship which would give him that knowledge. As such, in examples such as this the use of tense provides an intimacy reading. The point is that the tight correlation between intimacy and physical proximity is a corollary of the correlation between time-reference and physical proximity. Being temporally "located" provides immediate and verifiable experience. As such, knowing that Maureen has a Volvo in the present represents a stronger claim to intimacy than knowing that at some point in the past she had one. Due to an intimacy reading having become conventionally associated with tense markers, we suggest that A is able to interpret the tense usage not in terms of time-reference, but rather in terms of intimacy, due to tense being conventionally employed in certain situations to signal relative intimacy. A second example illustrates the use of past tense to signal lack of emotional intimacy: (2)

b. My daughter's father was Brazilian. He stays in contact with Suzanna, but I haven't seen him in years. (Tyler, personal data)

The relation between experience, conceptual structure and meaning

91

In this case, the speaker presents information about her daughter's father in the past tense. Out of context, this sentence is ambiguous and could suggest that the father is no longer alive. However, the additional information, that the father and the daughter continue to see each other, rules out the interpretation that the father is dead. Given our knowledge of the world, we can also rule out the interpretation that he is no longer Brazilian, as one's country of origin typically does not change. Finally, we can reject the interpretation that the man from Brazil is no longer Suzanna's father since biological fatherhood is unchangeable and he seems to continue, at least in some aspects, in the social role of father. We conclude that the speaker used the past tense to signal her own psychological/emotional attitude of non-intimacy towards her daughter's father. We hypothesize that intimacy has become associated with tense for the following reason. In certain situations, such as when talking about human relationships, as in (2a-b), relative intimacy is implicated. As time-reference and intimacy share similar image content, tense can become reanalyzed as the linguistic component in the sentence signaling the intimacy relation. Through continued usage of tense in situations in which intimacy is implicated tense develops intimacy as an additional meaning component, which can become instantiated in semantic memory along with the time-reference function. This process of conventionalization we term pragmatic strengthening.

5.2. Tense and salience We find an analogous recruitment of English tense to indicate salience (i.e. whether information is being foregrounded or backgrounded). We illustrated this phenomenon with the example in (3a-c) reproduced below: (3)

a. In November 1859, Charles Darwin's The Origin of Species, one of the greatest and most controversial works in the literature of science, was published in London.

92

Andrea Tyler and Vyvyan Evans

b. The central idea in this book is the principle of natural selection c. In the sixth edition... Darwin wrote: "This principle of preservation of the survival of the fittest, I have called Natural Selection." (Eigen and Winkler 1983: 53) The main point of these sentences is not the precise date of publication of this book, but rather the central topic of the book. The information presented in the first sentence establishes the frame for the focal information which occurs in the second sentence. The information in (3 c) provides supporting evidence for the key point, and as such is less prominent in terms of information status. Discourse analysts have often referred to the relative status of information such as that in sentence (3 a) and (3 c) as background and information such as that in sentence (3b) as foreground. In this example, tense is employed to signal the relative status of the information, i.e. past tense signals background status and present tense signals foreground status. As before, we suggest that tense comes to signal salience for the following reason. In certain contexts, when, for instance evaluating the relative importance of information, some pieces of information are implicitly more or less important than others. As time-reference correlates with physical proximity, so too degree of salience correlates with that which, as noted earlier, tends to be in foveal vision and thus physically proximal. Accordingly, as time-reference shares similar image content with salience, tense can in some contexts implicate relative salience. We suggest that through continued use of tense in contexts in which salience is implicated, namely pragmatic strengthening, tense morphology has developed a conventionalized meaning component, in which past tense denotes relatively less salience while present tense denotes greater salience. This use of past and present tense in order to signal relative salience relates to Langacker's (this volume) arguments concerning noncanonical uses of the present tense. He discusses vivid narrative, play-by-play sports casting, historical present, and other discourse uses of present tense which clearly do not conform to the typical representation of present tense usage. It is not entirely clear how his ex-

The relation between experience, conceptual structure and meaning

93

planation of non-canonical uses of the present tense as coding special viewing arrangements involving "virtual" events and texts might apply to the particular uses of present tense morphology we are considering. However, the use of the present tense to make a particular event more 'real' would seem to be related to the use of the present tense to denote greater salience and hence importance in terms of information structure.

5.3. Tense and actuality As we have previously observed, tense can be employed to signal actuality (the degree to which the experiencer believes the event described matches the current or actual world state). Let us reconsider example (4c), reproduced below: (4)

c. Suppose your house burned down. Do you have enough insurance?

Normal interpretation of this sentence is that the speaker is hypothesizing about an event which the speaker does not believe to have taken place, i.e. the speaker is not referring to an actual event in which the addressee's house burned down at some time before the moment of speaking. The fact that the speaker does not believe that the house concerned has actually burned down is signaled by the lexeme suppose.5 In addition, in this example past tense seems to underscore that the speaker does not believe that your house burned down references a situation holding in the world. That is, the use of past tense cannot be interpreted as referring to a past event, but rather signals a lack of commitment to the actuality of the situation. While English speakers regularly use the past tense in hypothetical constructions, they also have the option of using the present tense. So, Suppose your house burns down is also perfectly acceptable. Moreover, the choice of present tense would not entail that the speaker believes that the house is in fact burning down at the moment of speaking. Rather the choice between past and present tense seems

94

Andrea Tyler and Vyvyan Evans

to signal a somewhat subtle shift in the speaker's commitment to the probability that the hypothetical event will happen. In cases such as these, the choice of the present tense seems to signal a stronger commitment on the speaker's part to the possibility of the hypothetical state of affairs occurring. In terms of example (4c), without further context or access to the speaker, it is difficult to say with confidence why the speaker chose the past tense rather than the present. Given the language, this sounds like a sales pitch for insurance. It may be that the speaker was concerned about sounding overly aggressive and, not wanting to have his or her arguments dismissed out of hand as mere 'scare tactics', chose the past tense in order to soften the message. The point is that in these constructions, neither past nor present tense is interpreted in its canonical temporal meaning as coding information about events, which the conceptualizer believes to be real. Again, we suggest that as time-reference (signaled by tense) and actuality are both elaborated in terms of similar image content, i.e. the present correlates with the physically proximal, as does the notion of actuality, in situations in which a stronger or weaker commitment to the actuality of the event described is implicated, tense is reanalyzed as a lexical form contributing to this implicature. Through continued use of tense in such contexts, this meaning comes to be conventionally associated with the tense morphemes, such that the present tense form [0/s] signals a stronger commitment to the potential actuality of the situation, while the past tense form [Id] signals a weaker commitment.

5.4. Tense and attenuation We now turn to the final non-temporal meaning of tense, which we will address in this paper. Consider the following example, which is adapted from (5 c) above. (13) I wanted to ask you a question.

The relation between experience, conceptual structure and meaning

95

In sentences such as (13) we conventionally understand that the use of past tense does not place the desire to ask the question in the past, but rather that it attenuates and so makes such requests less facethreatening and hence more polite. We noted earlier that attenuation - which constitutes a linguistic lessening of the speaker's assertion of authority and an offering of options to one's addressee to cooperatively be affected by the speaker's request, hence lessening the face-threat involved - correlates with not being physically proximal to the addressee. That is, attenuation can be elaborated at the conceptual level in terms of proximal-distal locational content, the same image content used to elaborate time-reference. Put another way, while being in past timereference correlates with not being physically proximal with the current location, so too being polite correlates with not being physically proximal. After all, very close proximity can be perceived as being overly assertive and/or aggressive. This may be related to physical proximity being a pre-requisite for physical control. As the image content which serves to elaborate the concepts time-reference and politeness is closely related, we suggest situations in which politeness (and hence attenuation) are implicated, tense has come to be reanalyzed as the lexical means of signaling attenuation.

5.5. Further issues In view of the foregoing it is now apposite to ask why it should be tense, rather than some other linguistic form, which has become associated in English with the non-temporal meanings described in the foregoing. Indeed, tense appears to have developed some of the functions described above in a cross-linguistically robust way (cf. Fleischman 1989). This situation is even more intriguing as there is evidence that in some languages some of the meanings described in the foregoing are lexicalized not by tense but by other closed-class elements. For instance, in Japanese the demonstrative are 'that', which locates an object distant from both the speaker and listener, can serve to express attenuation:

96

Andrea Tyler and Vyvyan Evans

(14) A (male): doo shita no? how did interrogative marker 'What's wrong?' Β (female): kon 'na koto iu no are this such thing say complement that nan-desu kedo anata kimochi waruin-desu. be but you creepy be 'It might be that to say this to you, but you're creepy.' [Mitsuyo Sato] In this attested example, a female Japanese speaker employs are in order to attenuate a face-threatening remark. What is interesting is that the form used is a demonstrative, which locates an object distant from both the speaker and hearer. Given that Japanese employs a form with a primary meaning of a distal location in order to lexicalize attenuation, this pattern supports our suggestion that attenuation is elaborated at the conceptual level in terms of locational content distal to the speaker and hearer. Moreover, this pattern coheres with our explanation for the situation in English. We argued that as past timereference is elaborated in terms of locational content which is distant from the experiencer, the parallel elaboration between attenuation and past time-reference at the conceptual level means that timereference can come to develop an attenuation meaning. In English (and many other languages) the appropriation of tense as the relevant marker may simply be opportunistic based on frequency of appearance, i.e. due to usage. As English tense is so frequent, being marked on most verb forms, it is natural, given the parallel elaboration in terms of image content between time-reference and intimacy, salience, actuality and attenuation, that tense should be re-analyzed as the form which marks these meanings, rather than another form being developed for this purpose. However, the frequency of tense does not preclude another lexical item being developed to express these meanings, as we have just seen to be the case in Japanese.

The relation between experience, conceptual structure and meaning

97

6. Consequences for language teaching A number of researchers have noted the difficulty second language learners regularly encounter with non-temporal uses of English tense. For instance, Riddle (1986) documented that advanced learners of English often experience difficulty appropriately interpreting and producing non-temporal uses of tense in the areas of intimacy and salience. While advanced learners tend to learn to mitigate requests and suggestions through the use of the so-called past tense modals could and would, they experience a good deal of confusion over the type of attenuated invitations, requests, and suggestions exemplified in (5) above (Celce-Murcia and Larsen-Freeman 1998). Lock (1996) notes that uses of tense to indicate actuality presents problems for learners. We believe that at least part of this difficulty stems from the way English tense is represented in contemporary reference grammars and English as a Second Language (ESL) texts. As Riddle (1986) pointed out, ESL texts tend to present past tense solely as meaning 'completed before the time of speaking'; 'exceptional' uses, especially in the areas of intimacy, saliency, and attenuation, are generally ignored all together. For instance, while The Collins Cobuild English Grammar lists "vivid narrative" and "firm plans for the future" as the "other uses of tenses" (ibid.: 257), this information is placed in a usage note, several pages removed from the discussion of the primary uses of past and present tense. Moreover, no attempt is made to explain the connection between the temporal and extended uses. Their sections on politeness make no mention of the use of past tense to convey attenuation. Lock (1996) presents uses of tense in a variety of extended discourse contexts, but does not mention tense to convey foreground and background (i.e. salience); neither does he mention use of past tense to convey attenuation. When non-temporal uses are presented, for instance in hypothetical constructions, they tend to be discussed in terms of arbitrary constructions to be memorized. For example, from The Collins Cobuild English Grammar, "When you are talking about an unlikely situation, you use simple past tense in the conditional clause." (ibid.: 350).

98

Andrea Tyler and Vyvyan Evans

No explanation of why past tense is used in this situation is given. Similarly, Lock (1996) offers the following rule for unlikely conditions: "The general rule is that a condition is marked as unreal by the tense of the finite verb group being one step back in the past relative to its tense in the expression of a real condition..." (ibid.: 199). No further explanation follows. Finally, Westney (1994) reported that English teachers and texts often have difficulty in clearly articulating generalizations which cover many everyday uses of English tense. He notes that uses involving actuality and attenuated suggestions are simply unexplained exceptions which render rule-making very difficult. In sum, the typical ESL student appears to be presented with a partial list of uses of English tense, rather than a unified model, along with the advice that memorization of the arbitrary patterns is the best policy as there is no systematicity in the various non-temporal uses. We believe that insights from cognitive linguistics have real merit in offering more systematic, motivated accounts of how English works. However, we also believe that simply stating that tense can have additional usages due to "metaphoric extension" is in itself not very helpful. As we noted in Section 1.2., simply labeling something as metaphoric does not necessarily provide a revealing explanation, unless we actually explain how the information from the two domains come to be associated. We have attempted to show how the "exceptional" meanings associated with tense are grounded in experience, by virtue of experiential correlations, and derived as semantic elements associated with lexical forms, through language use, i.e. pragmatic strengthening. The nature of meaning extension is itself a function of treating grammatical elements as meaning-bearing units. This way of viewing language, we suggest, will be particularly helpful for teachers and language learners alike. Notwithstanding the utility and, we suggest, the plausibility of the present analysis, it is obvious that the details and concomitant complexity of the discussion presented here would be largely inappropriate for language learners or even many language teachers. Clearly further research is required in order to develop materials, based upon the foregoing proposals, which teachers could present in

The relation between experience, conceptual structure and meaning

99

a classroom setting. Such must be the aim of cognitively oriented applied linguists.

7. Conclusion In this paper we have argued that the primary meaning of tense is temporal reference, which is elaborated at the conceptual level in terms of spatial proximal-distal relations. The elaboration of this particular image content is motivated, we suggested, by experiential correlation. Due to the nature of tight correlations in experience, other response concepts such as intimacy, salience, actuality and attenuation are also elaborated in terms of similar image content. In certain contexts where tense is employed and these meanings are implicated, tense markers can come to be reanalyzed as the markers of these non-temporal meanings. This is the result, we have argued, of the parallel elaboration of time-reference on the one hand and intimacy, salience, actuality and attenuation on the other, in terms of similar image content. Through continued use, these meanings can come to be conventionally associated with the tense markers [0] and [Id], a process we have termed pragmatic strengthening. In terms of language teaching, this account has great utility as it provides a unified account of tense phenomena. Hence, it would be more teachable and coherent than accounts that assume that nontemporal meanings are arbitrarily related to the temporal reference meaning of tense.

Notes 1. Riddle's point is dramatically underscored by the presentation of tense in Collins Cobuild English Grammar (1990), which specifically purports to be a functional grammar aimed at second language learners with the intent of "concentrating on the real patterns of use in today's English." (back cover). No mention is made of the use of tense in relation to politeness in invitations, requests, or suggestions (ibid.: 204-206; 228-232). The exception to this silence is in a section 'Expressions used instead of Modals' where we find the state-

100 Andrea Tyler and Vyvyan Evans ment, "You can use 'want' instead of 'would like' to give an instruction or make a request... 'Wanted' is also sometimes used. It is more polite than 'want.' (ibid.: 240) No further explanation is provided. Similarly, in the section on expressing importance, there is no discussion of tense as a signal of the relative importance of information within the discourse (ibid.: 236-237; 257); the only remotely possible mention is in a usage note associated with the present tense under the rubric of 'Vivid narrative' (ibid.: 257). This deals only with personal narratives coded exclusively with the present tense; present tense is represented as a device to increase audience involvement. The only example involving tense given in relation to actuality is the use of the present tense in the following, "Suppose we don't say a word and somebody else finds out about it." Although native speakers also use the past tense in such structures, this use is not mentioned This grammar also clearly exemplifies the tendency to scatter non-temporal uses of tense, essentially representing them as arbitrary, and offering no attempt at a unified, systematic account that would tie them to the central temporal sense. 2. Specifically, Langacker argues that immediate reality is marked by the absence of a modal and the absence of the distal morpheme; non-immediate reality is marked by the absence of a modal and the presence of the distal morpheme. The immediate/non-immediate contrast is argued to be "a proximal/distal contrast in the epistemic sphere." (ibid.: 245) "Immediate reality coincides temporally with the time of speaking, so to the extent that the notion of time is specifically invoked, present time is conceived as one facet of immediate reality... In precisely analogous fashion, the predication of non-immediate reality is equivalent to one of past time...These notions are basically epistemic, i.e. they do not refer to time, yet they have an obvious interpretation with reference to the time-line model: since reality subsumes the past and present (but not the future), and immediate reality constitutes the present, the temporal projection of nonimmediate reality can only be the past. Presumably, then, the distal morpheme has a prototypical value that invokes the time-line model and is reasonably considered a past-tense predication. That, however, is only one manifestation of its basic epistemic import." (ibid.: 246). 3.Sweetser (1990) notes that experiences which have been labeled concrete tend to be those which allow continuity across individuals. For instance, "vision is...identical for different people - that is to say, two people who stand in the same place are generally understood to see the same thing ... Identity across people is a highly objective characteristic..." (ibid.: 39) In contrast, "abstract" concepts tend to be identified with internal states. Since human beings cannot communicate through mental telepathy, one human being cannot directly observe or know another person's mental or emotional state; we can only guess. Internal states simply do not allow for verifiable identity across people in the same way that entities and events in the external world do. Moreover, internal

The relation between experience, conceptual structure and meaning

101

states and subjective reactions are proverbially variable across individuals. We hypothesize that in order to communicate more effectively with their interlocutors about internal states, speakers have come to use concepts from the more observable, objectively agreed upon (intersubjective) world to discuss the unobservable (subjective) (cf. Grady 1997). 4. It is interesting to note that evidence is beginning to emerge that correlation may be a fundamental operation at all levels of cognitive processing. For instance, at the neurological level, integration or "binding" of perceptual information which is spatially distributed in the brain in order to form a coherent percept, may result from the correlated firing of the relevant neurons (Crick 1994, Crick and Koch 1990, 1998, Pöppel 1994, Stryker 1991). That is, the particular neurons associated with the sensory qualities constituting the perception of an object fire in correlated fashion. This synchronous firing serves to integrate the various spatially-distributed sensory qualities into a coherent percept, without requiring that the information be transmitted to and hence integrated at a single site in the brain. 5. In Fauconnier's (1994, 1997) terms, suppose is a space builder, which signals a hypothetical space.

References Binnick, Robert 1991 Time and the Verb. New York: Oxford University Press. Brown, Penelope and Stephen Levinson 1987 Some Universals in Language Use. New York: Cambridge University Press. Celce-Murcia, Marianne and Diane Larsen-Freeman 1998 The Grammar Book. Rawley, MA: Newbury House. Collins Cobuild English Grammar 1994 London and Glasgow: Collins Publishers. Comrie, Bernard 1985 Tense. Cambridge: Cambridge University Press. Crick, Francis 1994 The Astonishing Hypothesis. New York: Simon Schuster. Crick, Francis and Christof Koch 1990 Towards a neurobiological theory of consciousness. Seminars in the Neurosciences 2:263-275. 1998 Consciousness and neuroscience. Cerebral Cortex 8: 97-107. Cutrer, Michele 1994 Time and Tense in Narrative and in Everyday Language. USCD Dept. of Cognitive Science: Technical Report 9501.

102 Andrea Tyler and Vyvyan Evans Dirven, René and Günter Radden 2000 A cognitive linguistics grammar of English. Manuscript. Edelman, Gerald 1992 Bright Air, Brilliant Fire. New York: Basic books. Eigen, M. and R. Winkler 1983 Laws of the Game: How the Principles of Nature Govern Chance. New York: Alfred A. Krapf. Evans, Vyvyan 2000 The Structure of Time: Language, meaning and temporal cognition. Doctoral thesis. Linguistics dept., Georgetown University. Fauconnier, Gilles 1994 Mental spaces. Cambridge: Cambridge University Press. 1997 Mappings in Thought and Language. Cambridge: Cambridge University Press. Fillmore, Charles 1990 Epistemic stance and grammatical form in English conditional sentences. Papers from the Twenty-Sixth Regional Meeting of the Chicago Linguistics Society, 137-162. Chicago: University of Chicago. Fleischman, Suzanne 1989 Temporal distance: A basic linguistic metaphor. Studies in Language 13(1): 1-50. Frank, Marcella 1983 Modern English: A Practical Reference Guide. Englewood Cliffs, NJ: Prentice-Hall. Grady, Joseph 1997 Foundations of meaning: Primary metaphors and primary scenes. UC Berkeley: Doctoral Thesis 1999a A typology of motivation for conceptual metaphor: Correlation versus resemblance. In: G. Steen and R. Gibbs (eds.), Metaphor in Cognitive Linguistics. Philadelphia: John Benjamins. 1999b Crosslinguistic regularities in metaphorical extension. Paper presented at the Annual Meeting of the Linguistic Society of America, Los Angeles, January, 1999. In prep. Foundations of meaning. (Manuscript) Krasnow Institute for Advanced Study, George Mason University. Hopper, Paul and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Jackendoff, Ray 1987 Consciousness and the Computational Mind. Cambridge, MA.: MIT Press. 1992 Languages of the Mind. Cambridge, MA.: MIT Press.

The relation between experience, conceptual structure and meaning Jespersen, Otto 1924 Lakoff, George 1987

103

The Philosophy of Grammar. London: Allen and Unwin.

Women, Fire and Dangerous Things. Chicago: Chicago University Press. 1993 The contemporary theory of metaphor. In: Andrew Ortony (ed.), Metaphor and Thought, 202-251. Cambridge: Cambridge University Press. Lakoff, George and Mark Johnson 1980 Metaphors We Live by. Chicago: Chicago University Press. 1999 Philosophy in the Flesh. New York: Basic Books. Lakoff, Robin 1973 The logic of politeness; or minding your p's and q's. CLS 9, 292305. Langacker, Ronald 1987 Foundations of Cognitive Grammar, vol. 1. Stanford: Stanford University Press. 1991 Foundations of Cognitive Grammar, vol. 2. Stanford: Stanford University Press This volume Cognitive linguistics, language pedagogy and the English present tense. Lock, Graham 1996 Functional English Grammar: An Introduction For Second Language Learners. Cambridge: Cambridge University Press. Mandler, Jean 1992 How to build a baby: II. Conceptual primitives. Psychological Review 99(4): 587-604. Moore, Kevin Ezra 2000 Spatial experience and temporal metaphors in Wolof. Ph.D thesis. Linguistics dept., University of California at Berkeley. Ortony, Andrew 1988 Are emotion metaphors conceptual or lexical? Cognition and Emotion 2: 95-103. Pöpel, Ernst 1994 Temporal mechanisms in perception. In: O. Sporns and G. Tononi (eds.), International Review of Neurobiology 37, 185-202. San Diego, CA.: Academic Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik 1973 A Grammar of Contemporary English. New York: Harcourt Brace Jovanovich Inc.

104 Andrea Tyler and Vyvyan Evans Rauh, Gisa 1983

Tenses as deictic categories: An analysis of English and German tenses. In: G. Rauh (ed.), Essays on Deixis, 229-275. Tübingen: Gunter Narr. Reichenbach, Hans 1947 Elements of Symbolic Logic. New York: The Free Press. Riddle, Elizabeth 1986 The meaning and discourse function of the Past Tense in English. TESOL Quarterly 20: 267-286. Stryker, Michael 1991 Seeing the whole picture. Current Biology 1 (4): 252-253. Swan, Michael and Bernard Smith 1987 Learner English. Cambridge: Cambridge University Press. Sweetser, Eve 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Talmy, Leonard 1988 Force dynamics in language and cognition. Cognitive Science 2: 49-100. Tannen, Deborah 1993 What's in a frame? Surface evidence for underlying expectation. In: Deborah Tannen (ed.), Framing in Discourse, 14-56. Oxford: Oxford University Press. Ter Meulen, Alice 1995 Representing Time in Natural Language: The Dynamic Interpretation of Tense and Aspect. Cambridge, MA: MIT Press. Traugott, Elizabeth Closs 1975 Spatial expressions of tense and temporal sequencing: A contribution to the study of semantic fields. Semiotica 15(3): 207-230. 1978 On the expression of spatio-temporal relations in language. In: J. Greenberg (ed.), Universals of Human Language. Vol. 3: Word Structure, 369-400. Stanford: Stanford University Press. 1989 On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65: 31-55. Tyler, Andrea and Vyvyan Evans Forthc. a Reconsidering prepositional polysemy networks. Language. Forthc. b Spatial Scenes. New York and Cambridge: Cambridge University Press.

The relation between experience, conceptual structure and meaning 105 Westney, Paul 1994

Rules and pedagogical grammar. In: T. Odlin (ed.), Perspectives on Pedagogical Grammar, 72-96. Cambridge: Cambridge University Press.

Section 2 Facets of prototypes in grammatical constructions

Grammatical constructions and their discourse origins: prototype or family resemblance? Paul J. Hopper

"Language is a form of organized stutter." - Marshall McLuhan

1. Introduction1 Future historians of linguistics will, one might speculate, find the most significant turning point of the 1990's to be not the development of more subtle theories of grammatical structure, whether based on autonomous linguistic structure or on cognitive schémas, but the revolution in the nature of linguistic data brought about by massive storage and high-speed search software. In this respect, linguistics will be seen to have moved closer to other scientific fields in allowing the nature of its enterprise to be radically affected by changing technologies. The availability of large corpora as a source of data coincides with the rejection by many linguists of the exclusive use of "sentences" devised from intuition (introspection) and with a widespread movement to find motivations for grammatical constructions in cognitive dispositions and usage. In the present paper I will present a study of an English construction, and will suggest how observations about such constructions based on corpus data might lead us to different conclusions about their nature from those based on introspective data. For an example of this from syntax, we will consider the case of pseudocleft sentences. The orthodox construction of pseudocleft as laid out in the standard reference grammar of English (Quirk, Greenbaum, Leech, and Svartvik 1985: 1387-9) involves "an SVC [= Subject Verb Complement] sentence with a w/z-nominal clause as

110 Paul J. Hopper

subject or complement". This work illustrates some of the types using the following sentences: What you need most is a good rest What he's done is (to) spoil the whole thing What John did to his suit was (to) ruin it What I'm going to do to him is (to) teach him a lesson What I'm doing is teaching him a lesson Some of the uncertainty, the tension between what people say and write and what is "canonical", comes to the surface when the whclause is in the perfect and the resumptive verb is the -en participle rather than the base form or infinitive, as in ?What he has done is spoilt the whole thing. Revealingly, the authors suggest that this usage is a reduced form of two discrete utterances: "of doubtful acceptability...may be an ellipted form of: What he's done is this: he's spoilt the whole thing". In the by now considerable literature on pseudoclefts, two themes stand out. The first is that the function of the pseudocleft sentence is to background the material in the wh- clause so as to throw certain constituents of the predicate into contrastive relief. Thus in What my car needs is a new battery, only the direct object a new battery is new; the subject my car and the verb needs are both old or "given". The constituent-focus theory runs through both sentence-level and discourse studies of pseudoclefts. This distribution is claimed, for example by Prince (1978), to correspond to contexts in which (in this hypothetical instance) my car and needs are in some way recoverable ("given", "presupposed" in her terminology) from the previous discourse, either directly or inferentially. Similarly, in What John did to his new suit was ruin it, the agent and the grammatical object are already known and only the verb is asserted. Such treatments motivate pseudoclefts by reference to the ambient and preceding context, rather than by reference to upcoming segments. They derive from a linguistic theory that holds that linguistic forms code current meanings. I will here argue that in spoken discourse pseudoclefts are an-

Grammatical

constructions

and their discourse origins

111

ticipatory, and serve among other things to alert listeners to an imminent significant segment or to permit speakers a brief period of down time to prepare the formulation of an utterance. The second theme that permeates treatments of clefting is that "incomplete" pseudoclefts are exceptional and to be explained as reduced and deviant versions of the full construction. Thus Collins (1994) gives the following example of an incomplete pseudocleft from the London-Lund corpus: (1)

now what I'm fascinated is to know that COMMUNISTS (coughs) and FASCISTS are such bad SHOTS that when they have a battle between THEMSELVES ALL their MISSILES hit the UNFORTUNATE Mr. BROOKE, who as I understand is NEITHER (cited by Collins 1994:45. Some transcriptional detail has been omitted; caps as in original, italics are supplied.)

Fragments such as what I'm fascinated is are said (Collins 1994: 44) to be "system-deviant". System-deviance, in Collins' view, arises from processing factors: "There are, broadly speaking, two categories of incomplete pseudo-clefts and clefts: those whose completion is prevented by a variety of contextual or processing factors, and elliptical constructions in which omitted material is textually recoverable." In either case, the assumption is that the forms are exceptions to the normal realization of the pseudocleft. Speakers lose their drift or do not bother to state material that is obvious from the context. The language system (the speaker's competence) provides for canonical pseudoclefts, but speakers mutilate them in attempting to produce (i.e. perform) them. In this paper I will suggest that when natural spoken discourse is considered, it appears that pseudoclefts do not function primarily, or perhaps at all, to highlight any single identifiable sentence constituent such as a verb or a noun phrase. If such highlighting occurs, it is a by-product - an epiphenomenon - of a more rhetorical use of the pseudocleft: pseudoclefts function in natural discourse to delay an assertion for any of a number of pragmatic reasons. Furthermore, the "system-deviant" fragments are just as capable of carrying out this

112 Paul J. Hopper

function as are the more canonical pseudoclefts. Such functional equivalence suggests that we should hesitate to privilege "complete" constructions over "incomplete" or "fragmentary" ones, and should reject such notions as deviance and other terms implying a deficiency.

2. Pseudocleft fragments In what follows, many examples are considered in which a pseudocleft appears in a fragmentary, "incomplete" form. Of course, as we have seen, attributing incompleteness to these forms presupposes a certain view of the relationship between the ideal forms of grammar and the reality of spoken discourse such that spoken discourse is a degenerate or deviant partner to a grammatically perfect ideal construction. In order to avoid this presupposition, I will adopt the more neutral term piece. A pseudocleft piece is minimally a segment of discourse that begins with what and continues with a verb. The verb's subject is either the what or some other noun phrase. After these, there may be other elements (objects and complements). Often, but again not always, the pseudocleft piece terminates with the copula is/was. Pseudocleft pieces as described here are not a unitary schema, but link up in family-resemblance fashion with functionally similar pieces such as the thing, the one, the only, and several others. Following the pseudocleft piece there is usually, though not always, a continuation that may consist of a full clause: Λ

(2)

Anna:

But have you discussed the fact that you don't believe [...] that this person can actually help you with this person Mike: Yeah I have Anna. And what they're s what my psychologist has told me is that they will make arrangements for me to see a psychiatrist if they f th if I still feel this

or a complete predicate:

Grammatical constructions and their discourse origins

(3)

113

Yeah. But at the moment I I think it sh should be enhanced because what it actually does is looks through the whole of the die dictionary.

But while the continuation of a pseudocleft piece is frequently a clause or a predicate in a conventional sense, it should not be imagined that we can set up a pseudocleft "construction" such that the pseudocleft piece and the continuation together form a unit such as a "sentence". Quite often the pseudocleft piece is not followed by any continuation at all: (4)

the kind of newspapers that they want erm are are are really in a sense not er any longer er those down-market tabloids. And so it's it's b it's a struggle for in what I suppose er [= a?] population of about ten million I'm not quite certain myself exactly what it is but it's a huge population the two er main daily news er popular daily newspapers can barely sell eight hundred.

And at the other extreme, the continuation may be pursued over a gap of several subsequent clauses, as in this example: (5)

I mean maybe this is because of the school. I only have them on a Monday and then I have them next week Monday. Now what happens is we tried in the past different groups. You know different forms. So you ask somebody from each of these about three or four different forms Collect the homework for me on Tuesday...

The event presaged by what happens is is not in the continuation, nor even in the following clause. In fact it goes considerably beyond the scope of the given segment, during which time the interlocutor must patiently wait and listen. Similarly: (6)

...wood stain almost black. It's really beautiful and it's got all the cornices and the ceiling rose and high skirtings all original. Anyway then you go in on this it's like a parquet floor and then

114 Paul J. Hopper

what they've done is the cellar it's got a cellar and you go down the steps to the cellar but there's like a proper two proper rooms so on your left you've got a sort of cellar with a quarry tile. The follow-up to what they 've done is a general domain of architecture or interior decorating rather than a grammatical complement of the verb. The pseudocleft here marks an up-coming new segment introduced by the theme noun cellar. Many other examples of this kind of thing could be cited. Observations such as these should caution us against any explanation of pseudoclefts that involves some structural arrangement between the pseudocleft piece and the continuation, such as "focusing" on a constituent in the continuation, at least in less formal registers. There may be such a grammatical relationship in some cases, but there may not be. If there is a grammatical relationship, it is more likely to be a coincidental one, or else an interpolation from a more formal register. For whatever its role in formal and written registers and in the decontextualized sentences used by some linguists and philosophers, direct transcriptions of recorded speech point to pragmatic or rhetorical motivations rather than structural or strictly semantic ones as the functional basis of the English pseudocleft. While it is possible to identify several such functions, they all derive from a single fact: The pseudocleft works to delay the delivery of a significant segment of talk. It accomplishes this by adumbrating (foreshadowing) the continuation in general terms without giving away the main point: (7)

House to house collections are be going to begin this month in Hallgreen. What we 're going to do is we're going to target one area per month.

Here, what we 're going to do is points to an activity as the theme of the up-coming continuation, with we as the subject and are going to as the tense-aspect. Nothing else is supplied. This delaying function permits other useful strategies to be deployed. In addition to the gen-

Grammatical constructions and their discourse origins 115

eral "delaying" function, we may discern from the corpus at least the following five: i. To alert the listener that an upcoming utterance is noteworthy ii. To make an attitudinal comment on an upcoming utterance iii.To state a general theme for the upcoming utterance iv. To buy time while alternative wordings are considered v. To hold the floor pending the upcoming utterance Usually it is possible to see more than one of these at work, so that any single example might be used to illustrate several or all of them. The headings of the following examples are not therefore intended to suggest that the example is illustrative of only one of them. i. To alert the listener that an upcoming utterance is noteworthy A pseudocleft almost always suggests that an upcoming utterance or segment of an utterance is important, and that the listener should pay attention: (8)

...it need to kill something like a thousand times its own size. There's η there don't seem to be a r- real need. And in defence I mean what snakes or what animals try <pause> like what most animals try to do is if they tha have got a poisonous property is another animal attacks them they give them er a dose of venom which will not kill them it will just deter them next time.

The important point that the speaker wishes to focus attention on is the fact that most poisonous animals deliver less venom than would be needed to kill their victim. The build-up to this point includes also the conditional clause if they have got a poisonous property, revealed by the inclusion of a second occurrence of is to be part of the pseudocleft piece itself, and the further conditional another animal attacks them. The listeners are being prepared for the focal segment not only by the pseudocleft piece but by its extension with a first conditional

116 Paul J. Hopper

and by a second supplementary conditional. By this time they will have their pencils ready. ii. To make an attitudinal comment about an up-coming utterance A considerable number of instances of pseudoclefts include an expression of the speaker's attitude toward the up-coming statement. Typical examples include what worries me, what makes me smile, what amazes us, what saddens me greatly, what makes them exciting: (9)

Joe:

Er I mean if if people want to plant bombs there's always ways of doing it whatever the security. Rachel: Yeah. Joe: I mean what is worrying is that it's happening er people could get hurt and we don't know who's doing it really it might be the IRA it might not.

Interjecting a personal stance toward the upcoming segment naturally increases its prominence and provokes the listener's curiosity. While (as in this example) there may not be a first-person pronoun in the pseudocleft piece, attitudinal verbs like worry, amaze, etc. add a subjective element to the upcoming assertions, implicitly introducing a first-person reference where there otherwise was none.3 In this same vein, a number of examples show an interposed I suppose within the pseudocleft piece that again serves to subjectivize the discourse: (10) Robin: .. .you feel <pause> I mean <pause> the campaign has been going on in Scotland for longer than in in this country do you think it's erm a a victory for what I suppose you ΊI see it as a victory for the people over government I suppose would you? Walter: Yeah I do. I mean actually I mean it didn't work in Scotland from the start and that was like (11) to to clean up Mexico City air. And obviously this then became a pretty major objective for himself in terms of his own ambi-

Grammatical constructions and their discourse origins

117

tions and his own presidential er ambitions for er what I suppose must be the ninety-four ninety-four the next election ninety-four. Erm anyway the erm an emergency program was started er late I think in nineteen-eighty-nine. Erm I'm certain two or three At times, in fact, what I suppose functions as a virtually empty formula that delays the continuation while adding nothing in the way of independent meaning: (12) What what I wondering about er dr dr dr drifting away really from looking at individual ...claims and individual <pause> erm personal problems to what I suppose were policy issues... and er to do with the area of the er the type of service. . .that's offered to claimants The subjective nature of the pseudoclefi in spoken registers distinguishes it from both written registers and from the construction with which it is often paired, the cleft sentence (see Collins 1994: 103). iii. To state a general theme for the up-coming utterance A striking fact about pseudoclefts in natural discourse is that the verbs in the pseudoclefi are drawn from a very restricted lexical set. Among the most frequent are do, happen, need, make, use, say (usually in some form such as what I'm saying), tell, and a few others. These verbs have in common that they are semantically simple, basic, and general. They serve to adumbrate the continuation by supplying a general domain within which the more specific verb of the continuation will fit. Actions are adumbrated with do: (13) Derek: Thing is though you've got to preserve a a certain sort of scruffiness t to add to your programmer authenticity. Deirdre: No j just get an anorak. Jerry: Well what I thought I might do is erm Derek:

118 PaulJ. Hopper Jerry:

when I go into the office I'll nut the first person I see

(14) I mean I think Peggy and Sheila will be interested to hear what the software actually does here because this is relevant obviously to them. Erm (...) I mean basically what it does is it divides the the word are defined into various groups. Events are adumbrated with happen: (15) Frank: Lisa:

... she can ring Charles and speak to him. Yes. Erm <pause> try yes try and get her to ring you know <pause> t over the next day or so. Well I suppose what'll happen is she'll ring and you'll answer probably and it'll be good if I answer it won't I ooh-er

Happen here, unlike do in the earlier example, covers several single actions (ring, answer, answer). Its scope therefore extends over more than one clause, a fact that makes a functional explanation in terms of a single constituent in a single clause out of the question. Modalities of various kinds are expressed by want, need, and raodals: (16) Yeah but you see y y <sighs> <pause> you've got as much of a problem in expecting one of them to be [handy] when something happens as you have with a police officer. What we need <pause> actually is more money spent on the police isn't it? (17) John:

Imogene:

...what kind you know? I've got a few brochures today er d is there any kind of do you could you tell me what it must need to have you know like er in the way of disc drives and things? Well erm what you need is a erm there's a small disc drive you don't need a five and a half inch disc drive.

Grammatical constructions and their discourse origins

119

(18) If we're saving then what we want to know is how much more could we spe buy how much more goods and services could we acquire at the end of the year er by saving now (19) 'Cos like I wanted really what I wanted was I'd never seen anyone all day so what I suppose I wanted was someone to talk to (20) This is a kind of a special approach because we are operating from a standpoint that we don't have the necessary science in order to - for us to be exact in terms of our adjudication of these cases. So what we have to do is to make some conclusions based on what we refer to as the doctrine of reasonable doubt. (CSPAE) (21) Actually, what you might do is leave this part and say, these were the parameters originally to the committee. (CSPAE) The deployment of modality markers here is further significant because of the subjective dimension that such forms always contribute to the discourse. Almost always the subject of the modal verb is a speech act participant (first or second person pronoun). More generally, the speaker's strategy for choosing verbs in the pseudocleft piece is to anticipate some lexical material from the upcoming discourse such that an indefinite amount of projected utterance is preframed as "action", "event", "need", "want", "possibility", etc. This pre-framing is of course the speaker's choice, and is not determined by what is to come. It is therefore hardly surprising if, as in this next example (cited earlier), the pre-frame that is proclaimed in the pseudocleft does not appear, in retrospect, to match the actually implemented subsequent discourse: (22) and then what they 've done is the cellar it's got a cellar and you go down the steps to the cellar but there's like a proper two proper rooms so on your left you've got a sort of cellar with a quarry tile

120 PaulJ. Hopper

The speaker has here pre-framed the continuation as a set of actions of the builders/decorators rather than as a description of the interior; but her enthusiasm for the design overrides her admiration for the designers, and the continuation moves from being a narrative ("what they've done") to being a portrayal of the interior ("what it is now like"). In anticipating part of the lexical content of the up-coming utterance^), the speaker is furthermore distributing information over a wider stretch of discourse. This dispersal of information can be exploited rhetorically to introduce new information that would otherwise be more condensed in a more measured way, so as to be distributed over a longer time period and thus to be more easily digested by listeners. This feature no doubt explains the high frequency of pseudoclefts in academic discourse and other formal registers where complex issues are being articulated. /v. To buy time while alternative wordings are considered There is often reason to believe that the use of a pseudoclefit is related less to the focus demands of the discourse than to the need for time while formulating an utterance. It thus functions as a pleonastic segment that fills what would otherwise be heard as a hesitation (Ong 1982: 40-41). The formulas of epic verse work in a similar way. Homer's rhododaktylos ëôs "rosy-fingered dawn" comes as a readymade piece of two and a half feet of verse that can be "plugged in" whenever the sense and the prosody permit it. The singer, while producing it, is not thinking about the current line but planning the next one under the pressure of real-time performance (see Lord 1960: 22 and 30-67). The analogy is especially striking when we consider fixed and semantically empty pseudocleft pieces such as what I suppose is, what I suggest is, what happens is: (23) a rather modern play and it seems to avoid the kind of language <pause> traditionally expected from drama pointed wellformed eloquent witty erm and it's what I suppose is one of the reasons why we see this play as a sort of er modern or experi-

Grammatical constructions and their discourse origins

121

mental piece of drama 'cos it seems to deny itself a lot of the erm traditional resources of drama (24) Alan:

also that other facets of the curriculum had to be er brought on so people took on technology which will be coming on board after that Harriet: Mm. Alan: er history and geography. And what's happened is is as they've come on board and that we feel that <snif£> we have got to grips with them as well as we can at this moment in time we've abandoned them and then gone on to

The speaker "Alan"4 has a faltering delivery and is clearly fishing for words. The two hesitation markers in a short stretch, and even the sniff, and the resort to ready-made expressions all betray this. What's happened is is not the only cliché in Alan's utterances (facets of the curriculum, coming on board [twice], got to grips with, as well as we can, at this moment in time). To the reader of the transcript he is practically unintelligible, but of course the listener, who had access to the entire conversation to that point, may have had no such difficulty. The frequent association of pseudocleft pieces with disfluencies such as hesitation markers and clichés points to a role in processing spoken discourse on line. Pseudocleft pieces seem to occur with some regularity after a pause and also at points in the conversation where one speaker has not taken an expected turn and the other interlocutor has to step into the breach: (25) Fred: ...engineering at Aston and so we've always had a fair amount of autonomy Tim: Yeah. Fred: and so I don't think it's made a great deal of difference. Tim: Yes. <pause> Fred: But I think what has made the difference is er civil engineering degrees are now awarded by about forty institu-

122 Paul J. Hopper

tions and some of the polys take students of very low calibre indeed and so er in a way we could do Er I mean if if people want to plant bombs there's always ways of doing it whatever the security. Rachel: Yeah. Joe: I mean what is worrying is that it's happening er people could get hurt and we don't know who's doing it really it might be the IRA it might not.

(26) Joe:

The often noticed repetition of is in examples such as the next one, which is representative of a considerable number of such examples, suggests that the speaker has processed the pseudocleft as an unanalyzed unit to the extent that she is unaware of having said is, and feels bound to produce a further is to mark the boundary of the pseudocleft: (27) Adrian: I think that the difference is probably in interpretation. Whereas I would use the word power I can understand why you'd gib at the use of that word. Hannah: But surely surely what what this suggests is is that you have you have to have class erm community and culturally-specific notions of what masculinity Adrian: Mm. Hannah: and femininity are Hannah's struggle to produce these more complex ideas is seen not only in the repetitions {surely, what, you have) but in Adrian's background mm, probably to be interpreted as encouraging the speaker to continue in spite of the difficulty of the ideas. v. To hold the floor pending the upcoming utterance Several decades of work5 have confirmed the centrality of turn-taking in the organization of conversation, and a growing body of work suggests that grammar is heavily invested in strategies for delivering well-formed turns ("turn-construction units", see Schegloff 1996,

Grammatical constructions and their discourse origins

123

"transition-relevant places", see Ford and Thompson 1996). An important motive for deploying a pseudocleft in conversational discourse is that of shielding an utterance from possible interruption by signaling that the speaker's turn, contrary to expectations deriving from form, length and content, is not yet over. The speaker who is embarked on a discourse longer than a normal "turn" must always risk losing the floor through interruption by a listener. This risk - the loss of a rhetorical investment - is amplified if the speaker delays, for whatever reason, the delivery of an effective or otherwise prominent statement. Yet such delay is often very desirable. The pseudocleft piece is ideally structured to deal with this situation, for it keeps the airwaves moving (where silence would invite an interruption), holds the listener's attention by releasing small amounts of anticipatory material (what we would like to know is, what is especially worrying is, what these facts suggest is, etc.), and keeps the floor by avoiding the intonational fall that would mark the end of a turn. Haj Ross, in a recent paper (Ross forthcoming), has also drawn attention to the indeterminate nature of this construction, going so far as to label it a "constructoid". In an informal communication on the same subject, he remarks: "It is like this: there is a construction - NP be NP - and pseudos are attracted to it (maybe 'strange attractors' is not a bad metaphor) - but they don't get all the way there." Ross' observation seems to rest on the idea of a prototype: there is an ideal pseudocleft that is based on a simple copular sentence (as in What we use is a sonic aid [example from Cobuild]), and pseudoclefts approach it to a greater or to a lesser degree. This thought, however, leaves out the entire question of why pseudoclefts exist in the first place and what the functions of the various "imperfect" examples might be that are found in live natural discourse. It seems to me that by inverting this picture we obtain a clearer idea of both the construction and its discourse functions. In the case of the pseudocleft, the function appears to be not so much the focusing function that is generally attributed to the pseudocleft. To be sure, focus is part of the story, for the pseudocleft serves to alert the hearer to the imminence of a significant segment of talk, and no doubt the constituent focus-

124 Paul J. Hopper

ing function that has been noted in written or edited spoken texts derives from its less specific discourse focusing function. But more important, as has been argued here, seems to be the delaying function, a rhetorical effect of impressing the listener with the "social" significance of something about to be said, and making the listener aware that what follows is part of a considered argument worthy of attention and not a casual comment. From this delaying function, in one way or another, derive the various other functions that make the pseudocleft such a widely used and prominent construction in English. I would surmise that this general picture is true not only of pseudoclefts but of all of the more complex grammatical constructions. Grammatical constructions are normativized rationalizations of families of smaller and more fragmented quasi-lexical parts. They result from the grammaticalization of these fragments. A further observation is that the discourse motivations for the use of pseudocleft fragments are essentially temporal rather than structural. These motivations reside not in the static distribution of presuppositions across a completed construction, but in the timing and delivery of utterances: anticipating upcoming material, delaying an utterance for assessed rhetorical and cognitive effectiveness, finding strategies for warding off interruption. The presuppositional structure that is such a striking feature of written pseudoclefts is, one would surmise, a by-product of the transfer of the pseudocleft to a monologic atemporal medium and its expansion there as an extended grammatical construction.

3. Conclusions A common assumption of cognitive linguistics is that speakers frame utterances in terms of templates known as "schémas" that are based on prototypical instantiations of constructions.6 However, the leading researchers in cognitive linguistics have not usually seen any necessity to calibrate schémas beside actual utterances, since in principle utterances are themselves seen as mere reflections of immanent cog-

Grammatical constructions and their discourse origins

125

nitive structures.7 Yet isolated samples of language often bear only a partial similarity to the forms found in transcribed texts taken from natural discourse (Hopper 1996, Shirai 1990). Joan Bybee in a number of recent publications (Bybee 1998, forthcoming, see also Bybee and Hopper forthcoming) has shown the importance of considering structure in language as a function of the discourse frequency of forms, and stresses the need to study forms through their uses as exemplified in transcribed texts (usage-based phonology and grammar). The study of complex constructions like the pseudocleft from the perspective of their distribution in discourse points to a need to revise the prototype picture of constructions or schémas so as to bring grammar in line with what Mühlhäusler (1983, see Davis 1992) has called "the observable realities of language use." I suggest that the demands of consistency, the requirement that grammatical constructions be measured against a single ideal prototype, lead us to privilege in linguistic analysis sentence types that have achieved this consistency only in the context of reflective, planned discourse, and that in natural spoken discourse the visible parts of these canonical construction types appear as fragments. But these fragments should not be seen as incomplete partíais. Instead, the reverse is the case: the fuller "canonical" constructions should rather be seen as highly stylized cultural artifacts, amalgamations of fragments put together and grammaticalized through stylistic and normative conventions. The proper model for understanding this distribution, I would suggest, is not the central vs. marginal instance model (prototype model), but the family resemblance model, in which the fragments that end up as canonical instances are interrelated through partial formal-functional similarities. The grammaticization of apparent prototype instances such as: (28) What we need <pause> actually is more money spent on the police isn't it? should thus be seen as a normativized derivative of several fragmented possibilities, one in which written conventions and conventionalized and reflective ways of presenting complex information

126 Paul J. Hopper

have no doubt contributed to a complex "schema" that in formal and written registers controls the assembly of the formula. I would further suggest that these are not isolated examples, but are in fact typical of recognized canonical grammatical constructions in general. By the time a grammatical construction comes to the attention of linguists it is already highly normativized. Worse still, Westernized normative standards may be smuggled into descriptions of unwritten languages when linguists base their elicitations on English equivalents. To view the canonical constructions as prototypes and as the source of "deviant" fragmentary instantiations in discourse is to put the cart before the horse. It is sometimes said (e.g. Chomsky 1972, 1965) that children are exposed to "degenerate" data that cannot account for the fiali forms of language. Chomsky in fact explicitly denies that the fragments that occur in discourse could form the basis of a coherent systematic grammar: The native speaker has acquired a grammar on the basis of very restricted and degenerate evidence. (Chomsky 1972:23) Many children acquire first and second languages quite successfully even though no special care is taken to teach it to them and no special attention is given to their progress. It also seems apparent that much of the actual speech observed consists of fragments and deviant expressions of a variety of sorts. (Chomsky 1965:201).

Corpus studies suggest instead that these "degenerate" data are the true substance of natural spoken language, and that what our descriptive and prescriptive grammars give us are normativized assemblies of these fragments that tend to impress themselves on us as mental prototypes because of their greater social prestige - their associations with schooling, with literacy, and with complex discourse characterized by long periods and uninterrupted turns. This observation has significant consequences for both linguistic theory and applied linguistics. One moral to be drawn is that foreign language teachers and applied linguists in general should attend to the role of shorter, "incomplete" utterances, even ones that violate rules of canonical

Grammatical constructions and their discourse origins

127

grammar, as the building blocks of fluent and appropriate oral discourse.

Notes 1.1 am grateful to Rosalind Moon and Jeremy Clear for their help in introducing me to the Collins-Birmingham University International Linguistic Database (Cobuild, ) during a visiting fellowship at the Center for Advanced Research in English at Birmingham University in the spring of 1999; to Malcolm Coulthard for securing this appointment for me and for his hospitality; to the Faculty Development Fund at Carnegie Mellon University for financial support; and to David Kaufer, Head of the English Department at Carnegie Mellon, for arranging academic leave for me during the period of the fellowship. 2. Citations are from two sources. Those taken from the UK Spoken subcorpus of Cobuild are unattributed. A small number of additional examples have been taken from the Corpus of Spoken American Professional English, by Michael Barlow, available from Athelstan . These are tagged with the initials CSPAE. As is customary in corpus studies, citations are lifted from the original corpus without modifications of spelling or punctuation. 3.For a recent account of the importance of subjectivity in language, see Scheibman 2000. 4. Speakers in Cobuild are identified only as "male voice" or "female voice". Fictitious names have been supplied in the present article. 5. For a summary and important analysis of the mutual relevance of turn-taking to cognition, see the Introduction to Ochs, Schegloff, and Thompson (eds.), 1996. 6.Constructional schémas are roughly equivalent to prototypes. See Taylor 1998: 177-179 and footnote 1 (p. 199) for discussion. 7. On the Cognitive Linguistics view of schémas as abstractions that are immanent to linguistic expressions, cf. Langacker (1998): "It is reasonably supposed that schémas are immanent in their instantiating expressions, and emerge as cognitive entities by reinforcement of the structural properties they share at a certain level of abstraction" (page 13; italics as in the original). In Langacker's view, grammar resides in "schematized representations of sound-meaning pairings, abstracted from (and immanent in) the specific symbolic configurations observable in complex expressions" (page 2).

128 Paul J. Hopper

References Bybee, Joan L. 1998

The emergent lexicon. Papers of the 34th Annual Meeting. Chicago Linguistic Society. Forthcoming Phonology and Language Use. Cambridge: Cambridge University Press. Bybee, Joan L. and Paul Hopper Forthcoming Introduction. In: Joan L. Bybee and Paul Hopper (eds.), Frequency and the Emergence of Linguistic Structure. Amsterdam: John Benjamins. Chomsky, Noam 1965 Aspects of the Theory of Syntax. Cambridge: The MIT Press. 1972 Language and Mind. New York: Harcourt Brace Jovanovich. Collins, Peter C. 1994 Cleft and Pseudocleft Constructions in English. London: Routledge. Davis, Hayley 1992 Drawing the morphological line. In: George Wolf (ed.), New Departures in Linguistics, 90-115. New York: Garland. Ford, Cecilia E., and Sandra A. Thompson 1996 Interactional units in conversation: syntactic, intonational, and pragmatic resources for the management of turns. In: E. Ochs, E. A. Schegloff, and S. A. Thompson (eds.), 134-85. Hopper, Paul J. 1996 When grammar and discourse clash. Essays on Language Function and Language Type dedicated to T. Givon, 231-246. Edited by Joan Bybee, J. Haiman, and S. A. Thompson. Amsterdam: John Benjamins. Langacker, Ronald 1998 Conceptualization, symbolization, and grammar. In: M. Tomasello (ed.), 1-40. Lord, Albert Β. 1960 The Singer of Tales. (Harvard Studies in Comparative Literature, 24) Cambridge: Harvard University Press. Mühlhäusler, Peter 1983 Stinkiepoos, cuddles, and related matters. Australian Journal of Linguistics 3: 75-91. Ochs, Elinor, Emanuel A. Schegloff, and Sandra A. Thompson (eds.) 1996 Interaction and Grammar. (Studies in Interactional Sociolinguistics, 13) Cambridge: Cambridge University Press.

Grammatical constructions and their discourse origins 129 Ong, Walter 1982 Prince, Ellen 1978

Orality and Literacy: The Technologizing of the Word. London: Methuen & Co. Ltd.

A comparison of WH- and IT clefts in discourse. Language 54: 883-906. Quirk, Randolph, S. Greenbaum, G. Leech, and J. Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Ross, Haj To appear The frozenness of pseudoclefts: towards an inequality-based syntax. Papers of the Annual Meeting of the Chicago Linguistics Society. Schegloff, Emanuel A. 1996 Turn organization: one intersection of grammar and interaction. In: E. Ochs, E. A. Schegloff, and S. A. Thompson (eds.), 52-133. Scheibman, Joanne 2000 Structural Patterns of Subjectivity in American English Conversation. Dissertation (Ph.D.): Department of Linguistics, University of New Mexico. Shirai, Yasuhiro 1990 Putting PUT to use: Prototype and metaphorical extension. Issues in Applied Linguistics 1: 78-97. Taylor, John R. 1998 Syntactic constructions as prototype categories. In: M. Tomasello (ed.), 177-202. Tomasello, Michael (ed.) 1998 The New Psychology of Language. Mahwah, NJ: Lawrence Erlbaum.

Transitivity parameter and prominence typology: a cross-linguistic study* Sang Hwan Seong

This paper investigates the relationship of the transitivity parameter and prominence typology in light of recent developments in cognitive grammar. After critically evaluating previous approaches (Hopper and Thompson 1980, Shibatani 1985, Talmy 1988 and Voorst 1996) dealing with particular linguistic schémas and prototypes organized in and across languages, we will provide a parameterized cognitive functional perspective to account for language learnability and linguistic variation in English, German, and Korean. This research is based on the assumption that the grammatical structure of native language guides or influences the output of adult second language learners (L2). The grammatical structure of each individual language is shaped by the performance of the language speaker. Thus, this line of research is different from formalist acquisition theory which suggests that the way in which argument structure acquisition takes place can be reduced to merely linking a universal conceptual structure to syntactic components (cf. Goodluck 1991). By incorporating diachronic perspectives as well, the present topic will contribute to the study of cognitive principles of linguistic variations and organizations deriving from English, German and Korean. The present author advances a new cognitively motivated claim that the transitivity parameter should be constrained such that pragmatically determined word order languages with high semantic transparency proportionally correlate with less syntacticization.

132 Sang Hwan Seong

1. Prototype theory and transitivity Cognitive grammar (CG) views a linguistic construction as the pairing of a specification of form with a specification of meaning (Taylor 1995: 198). Langacker (1988: 18) also maintains that "grammar is inherently symbolic, and therefore distinct from neither lexicon nor semantics." It is therefore assumed that in linguistic constructions (or grammatical constructions) the syntactic component cannot be set apart from the semantic representation. Grammar has no independent existence apart from semantic and phonological structure (Langacker 1988: 5). In this line of argument, the interpretations of prototypical transitive constructions are subject to the more general theory of cognitive grammar. According to Hopper and Thompson (1980: 251), transitivity is traditionally interpreted "as a global property of an entire clause such that an activity is 'carried-over' or 'transferred' from an agent to a patient." Under this assumption, it is observed that the kinds of constructions that deviate from prototypical transitive constructions are also treated as marginal examples instantiated from the construction's prototype specification. Thus, we find the following unagent-like entities as grammatical subjects in English. (1)

a. This tent sleeps six. b. The room seats 500. c. The fifth day saw our departure.

The fact that these unagent-like noun phrases can appear as syntactic subjects in English does not pose serious problems in cognitive grammar since they are considered mere deviations from the central semantic specification of prototypical transitivity (Taylor 1995: 214).1 Compared to English sentences like Mary killed a spider or Tom hit a ball, the semantic relation between subject and each event involved in (1) is less intense. The subjects in (1) do not execute the processes denoted by the verbs. On the basis of Dutch, French and English data, Voorst (1996) demonstrates that there is cross-linguistic variation in the typical transitive constructions even with verbs such as break and buy as far as the level of intensity is concerned. Thus,

Transitivity parameter and prominence typology

133

transitive constructions in Dutch show a high level of intensity between the subject and the event denoted by the verb and between the event and the object, whereas in English both relations may be indirect or less intense (The rock broke the windshield / D *Het steentje heeft de voorruit gebroken). However, Voorst does not provide a functional account as to why this kind of cross-linguistic regularity exists. More recently, Talmy (1988) uses "energy transfer" as the metaphor for the transitive process. In Talmy's terms, energy transfer does not take place in (la-lc), because these subjects actually do nothing to execute the unfolding of the events denoted by the verbs. The same kind of idea is also very much visible in the semantic analysis of transitive constructions by Voorst (1996). Under this context it seems natural that active clauses high in transitivity or energy flow typically undergo passivization in which agentive subjects, totally affected objects and action predicates are correlated (Hopper and Thompson 1980: 292-294). (2)

a. *Six are slept by this tent. b. *500 people are seated by this room. c. G. ?? Unsere Ankunft wurde von dem fünften Tag gesehen. 'Our arrival was seen by the fifth day.' d. *Our departure was seen by the first day.

Thus, the passive constructions in (2) are all judged ungrammatical because the active counterparts of these sentences all exhibit low transitivity. The subjects of the active clauses are not agents. The arguments occupying the subject and object positions in the active clauses in (1) are not well differentiated semantically. Prototypical transitive constructions are cases in which the arguments filling the subject and object slots are diametrically opposed to each other involving "agent" and "patient" in the semantic make-up (e.g. kill and hit in contrast with resemble). Since the cases involved in (2) deviate significantly from the prototypical transitive constructions, the reason that the constructions in question in (1) cannot be passivized readily obtains. The following German examples can also be predicted not to undergo wm/ew-passivization due to the inherently non-actional

134 Sang Hwan Seong

lexical properties of the verbs involved. Note that the corresponding distinction in English is lost to a large extent. (3)

a. *Das Buch wurde nicht gekannt. (Helbig/Buscha 1994: 171) 'The book was not known.' b. *Der Pinto wurde von ihm besessen. 'The Pinto was owned by him.' c. *80 Pfennig werden von dem Heft gekostet. (Helbig/Buscha 1994:171) '80 pennies are cost by the notebook.'

Even in cases where the nature of the verb is prototypically actional, passives often fail to apply if the subject is not an agent. (4)

a. Die Mutter schneidet das Brot. 'The mother cuts the bread.' b. Das Brot wird von der Mutter geschnitten, (passive counterpart of 4a) 'The bread is cut by the mother.' c. Das Messer schneidet das Brot. 'The knife cuts the bread.' d. *Das Brot wird von dem Messer geschnitten, (passive counterpart of 4c) 'The bread is cut by the knife.' (Helbig/Buscha 1994: 165)

The subject in German sentence (4c) clearly denotes an instrumental case which plays an essential part in the cutting process. However, it does not control the energy flow per se involved in the event. Shibatani (1985: 831) makes a valid observation when he argues that "passives center around agents, and their fundamental function has to do with the defocusing of agents". If there is no agent to defocus in the active clause, passivization is usually impossible. The prototype approach significantly undermines the attempt to characterize the structure and/or function of the passive as a counterpart of

Transitivity parameter and prominence typology

135

the active voice (see also Shannon 1987). However, Shibatani does not explain why languages exist in which prototypical transitive constructions are not passivizable, when he states (1985: 831) that "even in transitive sentences, passives often fail to apply if the subject is not an agent." That is, he does not account for why not all prototypical transitive clauses with agentive subjects find passive counterparts in a language such as Korean. Our approach is based upon the functional motivation that the richer a morphological case-marking system a language has, the more degenerate the system of passivization becomes.

2. Transitivity and passives If the assumption that clauses high in transitivity are strongly correlated with prototypical passivization is right, we should find positive evidence in a language like Korean where the grammatical relations of transitive constructions are typically coded by nominative and accusative markers as illustrated in (5). (5)

Kyengchal-i totwuk-ul cap-ass-ta policeman-NOM thief-ACC arrest-past-Decl 'The policeman arrested the thief.'

This construction can be passivized as follows (cf. O'Grady 1991: 47): (6)

Totwuk-i kyengchal-eykey cap-hi-ess-ta thief-NOM policeman-DAT arrest-pass-past-Decl 'The thief was arrested by the policeman.'

In the passive construction (6) we notice that the accusative argument totwuk-ul in (5) is changed into the nominative argument in (6), whereas the nominative argument "policeman" is construed with an oblique case marker in (6). Usually the oblique marked noun phrase in (6) can be dropped, as is also the case with English and German

13 6 Sang Hwan Seong

because the passive construction involves "agent defocusing". We also note that in the construction (6) a passive suffix hi is inserted between the verbal stem cap and the rest of the verbal compound. To account for the structural properties of the Korean passive type in (6), O'Grady (1991: 47-52) takes a Categorial Grammar approach based on the formal system of morphological marking conventions in Korean. Sohn (1994: 301) also claims that there are three types of passives in Korean: (a) suffixal, (b) lexical, and (c) phrasal. Suffixal passives are the most basic and typical, while the other two types are a kind of pseudo passive. The choice among the three passives is dependent on what types of predicates are involved. In this paper, however, the present author does not treat lexical passives and phrasal passives as Korean passive constructions because they do not employ the grammatical passivization device (i.e. they are not morphologically derived passives). In lexical passives, despite the passive meanings they trigger, the forms of the stems are entirely different from active verb stems (e.g. macìa 'be hit' in contrast with chita 'hit'). Korean lexical passives are also composed of constructions in which the inchoative verb stem toy 'become' is attached to SinoKorean verbal nouns (e.g. wiim-ha 'entrust' vs. wiim-toy 'be entrusted'). Phrasal passives are made up of a verb stem and the inchoative verb cita 'get to be, become' (e.g. cwu 'give' vs. cwu-e cita 'be given'). The evidence for not treating lexical and phrasal passives as Korean passives comes from the grammaticization of Germanic passives. Valentin (1999: 145) argues that old Germanic passive constructions with werdhan (meaning 'happen', 'begin to exist', or 'enter into being') are perfective constructions, and that the ones with bim/was (first person singular present/preterite indicative form 'to be') are non-perfective. Valentin demonstrates that old Germanic passive constructions with werdhan do not coincide with the periphrases of New High German or Dutch, where the aspectual oppositions are exactly contrary to the old ones.2 The following New High German examples deserve our attention:

Transitivity parameter and prominence typology

(7)

137

a. Das Haus wird verkauft. 'The house becomes sold' b. Das Haus ist verkauft worden, (preterite version of 7a) 'The house is sold become' (i.e. the house has been sold) c. Das Haus ist verkauft. 'The house is sold.'

Unlike English, there are two different passive strategies available in Modern German: namely werden-passive and sein (statal)-passive respectively. In werden-passives (7a-b) expressing an action or occurrence it goes without saying that the change of perspective from the active counterpart takes place without affecting the lexical semantics of the verb verkaufen and that the original meaning of the verb werden (to become) is lost. The statai passive in (7c) which used to be a usage more prevailing in earlier periods should be interpreted as a derived construction of (7b), since the construction (7c) denotes a state that has resulted from previous action (see also Drosdowski 1995: 180). The Modern German statai passive with sein is a perfective construction. According to Valentin (1999), however, it is difficult to posit a passive as a grammatical device in the old Germanic dialects since wirdhu (first person singular present indicative form of werdhan) and bim/was exhibit the same "full" meaning; they are not auxiliaries. The present author assumes that this is also the case with the putative Korean passive constructions. We note that productive constructions such as the Korean lexical and phrasal passives with the inchoative verb stems toy and ci are fully compositional in their morphological make-up. The original meanings of these inchoative verb stems are retained. Thus the validity of the dominant view to classify the aforementioned constructions under the rubric of Korean passivization based on the translation equivalents (cf. O'Grady 1991: 50) does not simply obtain. With respect to prototype theory, the suffixal passives in Korean also pose problems. Despite the nominative and accusative marked transitive construction in (5), it cannot be maintained that all Korean prototypical active transitive constructions find corresponding passive constructions. For many passive sentences there are no active counterparts either. It is also the case that

138 Sang Hwan Seong many passives and their corresponding active constructions are not semantically equivalent. Nam and Koh (1993: 297) state that "Korean passives are not always derived from transitive counterparts and that there are more transitive verbs that cannot trigger derived suffixal passives than that can". (8)

a. Saram-dul-i mul-lul mani massi-nta. man-pl.-NOM water-ACC a lot drink-pres. 'People drink milk a lot' b. *Mul-i marti massi-i-nta. water-NOM a lot drink-pass-pres. 'Milk is drunk a lot [by people]'

In (8) we find a prototypical transitive clause with the verb drink in which no passive counterpart is available. Thus, as far as the Korean passive is concerned, a morphology-based structural description of the rule does not simply obtain. What interests us in the description of Korean grammar is not the transitivity phenomenon per se, but the animacy constraint imposed upon the voice system. Let us observe the examples in (9) and (10). (9)

a. ?mos-i os-ul ccic-ess-ta nail-NOM clothes-ACC tear-past-Decl Lit. Ά nail tore the clothes.' b. os-i mos-ey ccic-ki-ess-ta clothes-NOM nail-by tear-pass-past-decl 'The clothes were torn by a nail' (Sohn 1994: 307)

(10) a. *sikan-i na-lul ccoch-ko isseyo time-NOM I-ACC chase-ing is 'Time is chasing me' (Time is regarded as personified) b. nan-nun sikan-e ccoch-ki-ko isseyo I-TOP time-by chase-A/'-ing am Ί am being chased by time' (i.e. Things are hectic for me) (Klaiman 1988: 56-57; hi = passive suffix)

Transitivity parameter and prominence typology

139

The two transitive constructions in (9) and (10) tell us that inanimate subjects are avoided in controlling the state of affairs in Korean discourse, regardless of the argument's thematic status. This animacy constraint in the Korean diathesis might explain why the middle voice coupled with reflexivization is not available in the Korean and Japanese systems.3 (11) a. K. */ chaek-un casin-ul cai pania this book-TOP itself-ACC well sell b. i chaek-un cal pal-i-nta this book-TOP well sell-pass.-Decl. 'This book sells well.' c. G. Dieses Buch verkauft sich gut. (passive reading) this book sells (reflexive) well d. D. Deze stoel zit heel aangenaam. (Abraham 1995: 20) this chair sits very pleasantly 'This chair sits very pleasantly' One might argue that this kind of reasoning is not quite valid, given the fact that compared to English alone, Dutch and German transitive constructions show a relatively transparent semantic encoding system of subject selection analogous to Korean, while German is nonetheless identified as a language with reflexive construction. This seemingly plausible counter-argument is untenable, since we subscribe to the view that the reflexivization as an important feature of subjectprominence applies even to the German transitive construction in (11) with passive reading. That is, in German, even inanimate patient subjects trigger reflexivization in the middle reflexive construction. The corresponding Korean construction (lib) employs a pseudopassive with a passive morpheme. Reflexivization in Korean is restricted only to animate entities (cf. also Seong 1999b). When it comes to passivization with respect to transitivity, there is also a misunderstanding over the identification of the transitive object. Hopper and Thompson (1980: 259) suggest that "the special markings on definite objects, found in many languages, are better interpreted functionally as signals of the high transitivity of the

140 Sang Hwan Seong

clause as a whole." They also argue that "the arguments known to grammar as INDIRECT OBJECTS should in fact be transitive objects rather than what might be called 'accusative' objects, since they tend to be definite and animate". And this observation further supports the findings of Givón (1979: 54) that, in one pair of English texts, out of 115 indirect objects, 112 (or 97%) are definite and overwhelmingly animate. This is argued to be also true of some Bantu languages where topic-related syntactic processes including passivization are correlated with the fact that the 'dative' argument takes precedence over the patient noun phrase. Despite the crosslinguistic insights by Hopper and Thompson, the present author has some reservations based upon German and Korean data. Let us look at the following dative and accusative noun phrase serialization pattern of typical ditransitive constructions in German and Korean. (12) a. Hans hat dem Anwalt das Geld gegeben. (German) 'Hans has to the lawyer the money given' b. Hans-nun pyunhosa-eykey ton-ul TOP lawyer-DAT/BEN money-ACC cwu-ess-ta. (Korean) give-pret.-Decl. 'Hans gave the lawyer the money.' In both constructions in (12a) and (12b) we note that the dative noun phrase precedes the accusative patient noun phrase as the unmarked order.4 In German, this is only possible if the two combinations are in free variation. This tells us that the more referential animate object appears before the inanimate object, as far as the precedence rule is concerned. However, this cannot be taken to mean that the dative animate object is indeed the true transitive object of the verb, since the dative marked argument cannot be promoted to the nominative case in German or Korean passives (e.g. *Der Anwalt wurde das Geld gegeben; Das Geld wurde dem Anwalt gegeben). Thus, the identification of true transitive objects cannot be established unitarily by means of syntactic tests such as passivization (cf. *Er wurde ein Preis verliehen 'He was awarded a prize.'). This is in turn in conflict

Transitivity parameter and prominence typology

141

with another syntactic strategy called 'English dative movement' in which beneficiary/recipient argument precedes patient argument in the ditransitive construction. In English, 'dative movement' is also taken to support the preference selection of the dative recipient/beneficiary argument for identifying the true transitive object in English (cf. Hopper and Thompson 1980: 260). (13) a. Clara wrote a letter to Santa Claus. b. Clara wrote Santa Claus a letter. The two sentences in (13) have the same meaning (D. Shaffer, p.c.). The only difference may be a slight difference in focus. In the first, the focus is on letter: I wrote a letter (not something else) to Santa Claus. In the second, the focus is on Santa Claus: I wrote Santa Claus (not someone else) a letter. Since in (13b) the human recipient argument appears in the "object position" adjacent to the verb, this dative object would be interpreted as the true object according to the logic of Hopper and Thompson. Since the comparable nominal as an inherent case is not subject to promotion to the nominative argument in German and Korean passivization, the attempt to establish the true object in terms of animacy in ditransitive constructions does not succeed cross-linguistically. Thus, definiteness and animacy criteria associated with a high degree of transitivity cannot be a true diagnostic for selecting a true transitive object, at least with respect to passivization in German and Korean (cf. G. *Er wurde geholfen 'He was helped.'). In view of the criticism of the traditional transitivity theory, the transitivity parameter should be constrained such that languages with high semantic transparency and an overt case-marking system proportionally correlate with less syntacticization. The traditional analyses of passives centered around the formal aspects of the constructions per se or the lexical properties of the verbs involved (Haider 1985 for German and O'Grady 1991, Klaiman 1988 for Korean). Contrary to the previous studies, the present author attempts to account for the motivation behind the passive system by setting the discourse category "topic" apart from the grammatical category "subject".

142 Sang Hwan Seong 3. Subject and topic revisited Korean is well known for its subject and topic-prominence (Li and Thompson 1976). For the discussion of "subject and topic" the present author adopts Chafe's (1976: 50) functional definition of "topic" as something which "limit[s] the application of the main predication to a certain restricted domain," or as something which "sets a spatial, temporal, or individual framework within which the main predication holds." In previous research (cf. Hawkins 1986, Müller-Gotama 1994), it was pointed out that a grammatically prescribed fixed word order language like English shows more ambiguous surface syntax with respect to the pragmatic functions compared to German and the Slavic languages. In these approaches, however, the traditional notions of subject and object still play a central role in interpreting the core cases of grammatical organization. The present author maintains that for a more adequate description of the typological regularities, we need to set the notion of topic apart from the category of subject (Seong 1999a, 1999c). Gundel (1988) in her elegant study also illustrates what kinds of strategies are available to mark "topic-comment structures" across languages, following the tradition of Li and Thompson (1976). However, she does not show how the relative degree of topic-comment structure can vary among the languages investigated. In this section, we want to further investigate how the word orders of English, German, and Korean are pragmatically regulated in controlled dialogue situations. Let us compare the following question and answer pairs. (14) English (Chafe 1976: 48) Q: What happened to the lamp? Al : The dog knocked it over. A2: ?It was knocked over by the dog. (15) German Q: Was ist mit der Lampe passiert? A1 : Der Hund hat sie umgeworfen. Lit. 'The dog has it knocked over.'

Transitivity parameter and prominence typology

143

A2. Die hat der Hund umgeworfen. it-ACC. has the dog-NOM knocked over A3 : ?Sie wurde von dem Hund umgeworfen, (passive) A4: ?? Sie hat der Hund umgeworfen. (16) Korean Q: chundung-i etteke doinke-ya? lamp-NOM how became-Q [definite] [wh-rheme:focus] Lit. 'What has become of the lamp?' Al: (guguss-un) gae-ka nume-tturyu-ss-tta. [it -TOP] dog-NOM fall-cause-pret-Decl. [TOPIC] [FOCUS] [VERB] A2: ??guguss-un gae-eyeuyhae nume-ci-ess-ta. [it - TOP] dog-by fall-inchoative-pret-Decl Lit. 'It was knocked over by the dog.' In the sentences in (14-16) the present author provided a controlled dialogue situation whereby various answers (A) to the question (Q) 'what happened to the lamp?' can be formulated for the three languages in the order of preference. As we know from the functionalist research tradition associated with Talmy Givón, Charles Li, Sandra Thompson and others (cf. Traugott and Heine 1991), the fixing of discourse strategies influences the grammatical packaging of the discourse and evolution of the syntactic and morphological structure. We also hold the view that the grammatical structure of each given language is constantly shaped by the performance of the language speaker. Hence, the differences of discourse strategies identified in the dialogue exchanges (14-16) provide important implications for the grammaticization processes in each language investigated. In the English Q & A pair (14), we observe that while it is the topic, the dog is chosen here as subject. The preferred response (Al) treats the dog as the subject, although the disfavored response (A2) which promotes the topic it is entirely grammatical. In the German Q & A pair (15) as well, we find that der Hund is selected as the subject over the topical pronouns die or sie. Between English and German,

144 Sang Hwan Seong

we also note another difference - that passivization as a topicalization strategy can be a more feasible alternative answer in English than in German. In the Korean Q & A pair, the preferred unmarked answer (16A1) to the question (16Q) shows that the topical pronominal constituent with nun appearing in the sentence initial position precedes the pre-verbally focused rhematic element dog. This topical item can be freely dropped in the most natural discourse setting. It seems in Korean that even though the constituent with subject marker ka appears as the sentential subject in (16A1), its position is not syntactico-semantically but pragmatically determined (see also Shannon 1999 for Early New High German [ENHG] data on this issue).5 This topic-focus constraint also applies to the precedence rule ordering of dative, accusative and oblique arguments in Korean as the present author crucially demonstrated in his previous article on word order in German and Korean (cf. Seong 1999a). Based on the different discourse strategies attested in (14-16), the present author demonstrated that the linear arrangement of Korean as a topic-prominent [TP] language follows the scale of "communicative dynamism" (cf. Firbas 1992) far more closely than it does in a subject-prominent [SP] language. This finding suggests that the passivization strategy is highly restricted in TP languages, since they exhibit a "built-in topic-comment structure". In these languages the use of syntactic promotion to achieve topicalization of a non-subject participant is not necessary; such a strategy can be attained directly. In English, however, we do not find this strategy. Van Oosten (1984) already observes that agency or, more generally, "primaryhood" often wins out over topichood in English subject-selection in basic, active sentences. Even though German also seems to conform to this principle in basic sentences, the passivization in which topichood prevails over agency in English subject-selection does not strictly apply to German, due to the existence of verb-second rule (cf. examples in [15]). In English, the passive construction is employed in cases where a non-agentive topic beats the agent or the primaryhood for subject-selection, and English also has other various syntactic devices such as zY-cleft (or extraposition), there-existential, and towg/?-construction (e.g. This

Transitivity parameter and prominence typology

145

book is tough to read) to satisfy the secondary need to identify the topic of the sentence. In this respect, the topic or pragmatic pivot prevails over agency in English subject-selection, as claimed by Foley and Van Valin (1984: 115). Korean, on the other hand, strictly conforms to the topic- and focus (rheme) positional constraint in which two separate categories of topic and subject are available. Thus, the conflict over subject selection in the sense of van Oosten (1984) vanishes entirely as far as Korean is concerned.6 The following examples from Korean and ENHG further support our argument for the existence of the topic as an independent category. (17) Kokiri-nun ko-ga kil-ta. Elephant-TOP nose-NOM long-Decl. Lit. 'As for an elephant, it has a long trunk.' (i.e. An elephant has a long trunk.) (18) so sah er, das die Messer, die er geschliffen het, der Rück was als die Schneid... 'Then he saw that the knives that he had sharpened, the back was like the blade...' (taken from Shannon 1999) In the literature, this construction is known as a "double subject construction" in which the left-most constituent in (17) is not a subcategorized argument of the predicate kilta 'to be long'. In ENHG example (18) die Messer 'the knives' is also taken to be a free topic, which is not a part of the verbal valency. Such a construction is not available in Modern English or in New High German (NHG), since the grammaticization of a topic into the category of subject took place in the diachronic development of Germanic (Burridge 1993). Thus, it is clear that the older stage of modern Germanic languages conforms more faithfully to the pragmatically determined word order. Judging from these data, it is fairly obvious that the discourse frame setting strategy found in TP languages is completely different from that of SP languages such as English.

146 Sang Hwan Seong The historical rise of the opaque relationship between "subject category" and "topic" in English is also confirmed by the development of a sister Germanic language such as Scandinavian. Faarlund (1992) argues that the "subject" category in Scandinavian has changed from a relatively "non-prototypical" to a "prototypical" one as Scandinavian languages have evolved from non-configurationality to configurationality. He assumes that the nominative case in Old Scandinavian is associated with subjecthood, and that discourse- or referencerelated strategies involving theme-rheme organization are entirely independent of case marking. Specifically, the nominative-marked nomináis in modern Scandinavian languages have gained many reference-related (i.e. topicality in Chafe's term; cf. also Foley and Van Valin 1984) "subject" qualities that Old Scandinavian nominatives lacked. For example, while in Modern Scandinavian "the subject is always definite in some (specifiable) sense", Old Scandinavian nominatives are often indefinite (Faarlund 1992: 161). Often this means that "expletives" are used in the modern translation of Old Scandinavian indefinite nominative subjects. Thus, in Old Scandinavian, topicality and primaryhood (cf. Van Oosten 1984) are coded by completely different formal means. The basic sentence structure of the SP and TP languages can be schematically explained in the following diagram (cf. double subject constructions in [17] and [18]).

Topic Figure 1.

Comment

Diagram of topic-comment connection

In this system the category of topic need not be defined in terms of verbal valency of the lexical predicate. The vertical line in the dia-

Transitivity parameter and prominence typology

147

gram shows the functional boundary of topic and comment. The direction of the arrow in this diagram also captures the possibility that the category of topic functions as the core argument of the predicate as in SP languages. Schematic characterization of the English system would be the pattern in which the two categories "subject" and "topic" have converged with each other. Gapping between unlike constituents with differing grammatical relations also belongs to the general features of TP languages. Again, we find a striking typological similarity between Korean and early Germanic. (19) a. pasta-nun nae-ka imi mok-ess-jiman ajikdo TOP I-NOM already eat-pret-but still 0 nama-issta. (Korean) (subj) left-is 'Even though I already ate of the pasta, there is still (some) left' b. Die gefielen dem Herren so wol und [er] sprach zu DAT [NOM] ihm ... (ENHG) 'They pleased the Lord very well and [he] spoke to him ...' (Shannon 1999) c. Hi haelde mi (object) ende [ _ ] ginc met hem in die taverne (Middle Dutch) 'He fetched me and [I] went with him into the tavern' (Burridge 1993: 146) The Korean construction (19a) suggests that the initial constituent the pasta can be construed as both the object in the first clause and the omitted subject in the second clause. In ENHG example (19b) as well, the "dative-marked noun phrase" dem Herren 'the Lord' in the first clause functions as the controller of the missing "nominative nominal" er 'he' in the second clause. The highlighted constituent in Middle Dutch example (19c) represents the syntactic item omitted in the conjoined clause. This suggests that in Korean, ENHG and Middle Dutch the zero noun phrase-anaphor is not syntactically restricted

148 Sang Hwan Seong

and that the structural properties accrued to the category of "subject" in extreme subject-prominent languages like English are considerably limited. This means that we should consider two functional roles "subject" and "topic" in interpreting some core cases of the grammatical organization of the subject and topic-prominent languages. If this is the case, then we are justified in arguing that the structural topic properties mostly associated with the English subject category, i.e. grammatical relation changing rules such as passivization and raising, are highly restricted or not available at all in highly TP languages. Thus, our argument is based upon a discourse-oriented functional motivation different from the generalization of Plank (1983) in which the morphology based transparent coding of grammatical relations is directly linked to the restricted application of the syntactic rules in German as opposed to Modern English (e.g. raising constructions and extractions).

4. Parameterized prototype theory As we have seen in examples (9-11), the semantic restriction that the inanimate entities are low in controlling the state of affairs of Korean transitive constructions predicts the absence of the middle construction in the Korean voice system. This conceptual difference could be broadly interpreted so that the selectional restriction for the semantic content of the subject category in the transitive verbal valency becomes stronger when we move from grammatically determined word order languages like English to pragmatically determined word order languages like German and Korean (This tent sleeps four / G *Das Zelt schläft vier, The key opened the doorI G.*Der Schlüssel öffnete die Tür). This principle also holds for the relationships of verb and object in German and Korean. The case we have in mind is that in English, nomináis referring to very unpatient-like semantic roles appear as direct objects in transitive constructions (Taylor 1995: 212). (20) a. We laid a carpet in the room, b. We carpeted the room.

Transitivity parameter and prominence typology

149

c. He loaded hay onto the truck. d. He loaded the truck with hay. In (20b) we note that the patient participant denoting a locative is incorporated into the verb. This is, however, not a syntactic rule, but a lexical derivation, i.e. conversion (explanation attributed to R. Dirven). In (20d), the truck is construed as the transitive object of the same locative verb in (20c). In morphologically transparent systems like German and Korean, these opaque constructions are not likely to occur or less developed. (21) a. Wir haben einen Teppich in das Zimmer gelegt. we have a carpet in the room laid b. Wir haben das Zimmer mit einem Teppich ausgelegt. we have the room with a carpet laid out c. Er hat das Heu auf den LKW geladen. d. Er hat den LKW mit dem Heu beladen. e. *Er hat den LKW mit dem Heu geladen. f. Hans hat die Wand neu tapeziert. Hans has the wall new papered 'Hans repapered the wall.' (an opaque instance similar to English) In (21b) and (2Id) the German morpho-syntax of the corresponding clauses shows much clearer lexical choices than in English. The Korean equivalents in (22a) and (22b) reveal a similar transparent paradigm. (22) a. ku-nun cwimcha-ey kuncho-lul sil-ess-ta. he-TOP truck-LOC hay-ACC load-pret.-Decl. 'He loaded hay unto the truck.' b. ku-nun kuncho-ro cwimcha-lul chai-u-ess-ta. he-TOP hay-INSTR truck-ACC fill-caus.-pret.-Decl. 'He loaded the truck with hay.'

150 Sang Hwan Seong The fact that English grammar entails these opaque grammatical relations tells us that the English transitive construction has undergone enormous extension due to case syncretism in its history. Languages differ in the way features of actions are encoded by lexical items. However, it is no coincidence that German and Korean show a more transparent overall encoding system for the relationship of verb and object. Split intransitive constructions in German and Korean also fit well into our system of prototype parameters. The experiencer argument of a mental state is readily encoded as a transitive subject in English (e.g. The king likes pears). In German and Korean, however, the existence of the dative case is held to be responsible for encoding most typically the more topical or semantically salient "experiencer argument" of a mental state in a valency network of lexical predicate. (23) a. Ich höre, dass Studenten eine I hear that students-DAT a Bahnkarte fehlt. discount train card-NOM lacks Lit. Ί hear that students lack a discount train card.' b. *Ich höre, dass eine Bahnkarte I hear that a discount train card-NOM Studenten fehlt. students-Dat lacks Lit. Ί hear that students lack a discount train card.' c. * Thomas glaubt, dass ein Fehler der Thomas believes that a mistake-NOM the Mannschaft unterlaufen ist. team-DAT happened is. d. Na-eykey/nun ton-i mocara-nta. (Korean) I-DAT/TOP money-NOM lack-Decl. Ί lack money.' The split intransitive constructions (or inversion constructions) denoting a mental state with dative experiencer + nominative theme arguments in the German 'middle field' (Ger. Mittelfeld) and Korean

Transitivity parameter and prominence typology

151

show that the grammatical encodings specifiable in overtly casemarked systems are neutralized in English transitive constructions. This is due to the fact that unlike Modern German, English has become relatively opaque in the declension of case marking and the comparatively tight association of agent with nominative subject encoding. This diachronic change resulted in the integration into the subject category of formerly subjectless impersonal verbs with experiencer arguments (e.g. OEngl. Him/hine hyngrede 'He is hungry.' from Plank [1983: 11]). Given the lack of isomorphism between role and overt coding in a configurational language like Modern English, Noonan (1977: 377) argues that this system requires the identification of a level of grammar not relevant to direct role marking languages, i.e. a level of grammatical relations (e.g. subject and object). This is, then, the sense in which Noonan uses the term "subject": "the highest ranking syntactic slot in an indirect role marking system." In our prominence typology, the distinctive statuses of the two types of role-marking systems can be identified with English (indirect role marking system) and Korean (direct role marking system) respectively. In the English system, then, reliance on traditional grammatical relations to describe the morphological and syntactic process is most clearly indispensable. In contrast to English, the participant encoding strategy of pragmatically determined word order languages such as German and Korean provides a parameterized perspective with respect to the prototype view of constructions. Thus, the Korean system might be classified as a highly transparent system in which some form of overt coding (e.g. case marking) relates directly to semantic roles, without significant neutralization of role distinctions. Accordingly, we can take the German system to be a relatively transparent system in terms of how the form of overt coding is mapped onto the role system. By analogy we can also claim that Old Scandinavian had a highly transparent verbal encoding system for transitivity, since this language had two distinctive formal means of marking nominative argument and topicality, as we confirmed in the preceding section. It is in this sense that Noonan (1977) contrasts the transparent direct-role marking system with an English type "indirect role

152 Sang Hwan Seong

marking systems," in which syntactic slots and argument roles are arranged hierarchically à la Fillmore (1968: 33). In a grammatical theory dealing with typologically distinctive languages, we need to construct a syntactically, semantically and pragmatically balanced meta-linguistic apparatus according to which the grammatical properties of each language can be measured and tested. The following figure schematically demonstrates the relative degree of transitivity in the verbal encoding system identifiable among languages investigated via the subject and topic prominence parameter.

< more grammaticizing subject prominence subject = grammaticized topic

•

less grammaticizing topic prominence subject * topic

This figure indicates that the transitivity and prototype parameter is constrained such that languages with high topic-prominence strongly correlate with less grammaticization. Thus, this figure can be understood as a typological frame of reference, a scalar model to explain each grammaticization process of a given language with respect to subject and topic parameter. At the extreme left side of our system, we find an English type language whose properties such as allegedly looser selectional restrictions for verbs and greater freedom in the

Transitivity parameter and prominence typology

153

application of raising and passivization may be non-coincidentally related to the greater "opacity" specific to the system. We argue that these properties of a broader structural description of the rule are to be closely linked to the presence of the structural subject position in English. Since this configurational system requires grammatical relations to be indispensable to identifying arguments of the clause, the relationship between overt case-marking and the assignment of semantic roles is expected to be the least isomorphic in English as an SVO language (e.g. This tent sleeps three. / This bed was slept in. / G. *Dieses Bett wurde geschlafen in.). At the opposite end of our system we find a Korean type language in which pragmatically determined word order with a transparency principle motivates a more direct reflection of semantic roles in the morphology. Thus, the English subject is analyzed as a "grammaticized topic", whereas Korean as a typical SOV language retains an independent "topic category" in addition to the notion of subject. Prima facie evidence for the topic prominence parameter is presented via a historical account involving Germanic languages. It is clear that there is a typological distinction between construction-based languages and discourse-based ones. Recall that in chapter 1, we observed the generalization of Voorst (1996) that when it comes to the grammatical relations of transitive constructions in Dutch, French and English, Dutch shows the highest level of intensity (i.e. strict subcategorization) whereas English shows the lowest level of intensity (i.e. loose subcategorization). While Voorst does not provide any explanation as to why such crosslinguistic variation exists, it is obvious that our typology proposed here accounts for the cross-linguistic regularities at issue. In this section we demonstrated crucially the relative degree of subject and topic prominence that obtains among Korean, German and English, and that this regularity provides a better prediction of the grammaticization processes involved. In the next section we advance the view that this principle is directly reflected in the second language acquisition process for Korean.

154 Sang Hwan Seortg

5. Prominence typology and language acquisition There is substantial evidence manifested in L2 data that, due to the effects of prior knowledge such as in the form of transfer or interference phenomena, native speakers of SP languages experience systematic difficulties when learning TP languages such as Korean. The present account of the second language acquisition process for Korean is based on the part of the research database collected at the Korean program at the University of Bonn. The initial focus of this study is centered around the TP features of the target language, such as topic constructions with topic marker, zero topic (subject and object) constructions, and double nominative constructions. The participants were level 1 and level 2 German students with German as their native language. They were asked to describe the participant's family and what they did last summer. This task evaluated the participant's ability to acquire the target features in written production (free composition). To provide a baseline for each analysis, three native speakers of Korean also participated as the third group. Thus, the writing samples are taken from 3 groups at different levels of proficiency in Korean. See Table 1. Table 1.

Frequency distribution of overt subjects in the written production of three groups Korean Proficiency Levels (number of participants) Total subjects/ sentences

1(3) 16 / 20

2(3) 18 / 36

3(2) 19 / 44

Careful analysis of the writing samples demonstrates that German students at the first (beginners') proficiency level tend to write Korean sentences with overt subjects almost without any exception. Even in the four cases where empty subjects are found, it is the case that these empty subjects refer back to overt subjects in the immediately preceding sentences. Our data show that the tendency to drop overt subjects increases dramatically when it comes to proficiency level 2 consisting of second year students. Participants at level 2 dropped only the second sentential subject, which is discourse-bound

Transitivity parameter and prominence typology

155

to the first subject. On the other hand, level 2 students seem to overgeneralize this principle so that we often find the following ungrammatical sentence (24b) in which a subject noun with topic marker should be mentioned sentence-initially for the utterance to be grammatical. In the sample text we find the first sentence (24a) with three subjects (parents, older sister and older brother) which seem to function ambiguously as the discourse subjects of the sentence in (24b). (24) a. Pumonim-i keysiko, enni-wa parents-NOM be-honorif. older sister-with oppa-ka issupnita older brother-NOM be. Ί have parents, an older sister and an older brother.' b. *samsipo-nyun- june kyulhon-hae-ss-upnita 35 years ago marry-do-pret-honorif. deci 'They married 35 years ago.' This pattern suggests that German learners of Korean may first overproduce subjects (cf. level 1) and then gradually learn to drop them as their proficiency improves. However, it is clear that one cannot freely drop a subject in a context in which a bare noun with topic marker pumonim-un 'parents-topic' as the subject should be specified as the old information as evidenced in (24b). Among the different types of TP features of Korean, topic constructions with the topic marker (n)un seem to be relatively harder to acquire. Traditionally, it has been observed that the nominative marker is usually construed with a neutral description, whereas the topic marker triggers both a thematic and a contrastive reading (Seong 1999a: 362). The two separate readings become stronger when the topic particle is used twice in parallel constructions for conveying the information of equal status. The following example (25a) directly follows the sentence (24b) in the same sample text taken from one of level 2 participants. (25) a. *emeni-ka jubu-ipnita. Apuji-ka kepsuso-eyse mother-NOM housewife-be father-NOM waterworks-at

156 Sang Hwan Seong

sikmulhaJg'a-ipnita botanist-be 'My mother is a housewife. My father is a botanist at the waterworks.' b. emeni-nun jubu-i-mye apuji-nun mother-TOP housewife-be-conjunction father-TOP kepsuso-eyse sikmulhakja-ipnita waterworks-at botanist-be 'My mother is a housewife and my father is a botanist at the waterworks.' While writing about her family, the German female student produced (25a) instead of (25b). Since the parents of this student are previously identified information in the preceding context, the two nouns 'mother' and 'father' in (25a) should be marked with topic marker nun respectively. The expected grammatically correct version of (25a) would be (25b) in which the first NP marked with nun triggers a thematic reading, whereas the second NP with nun is associated with contrastive reading 'on the other hand'. Thus, it is evident that German students tend to overproduce nominative subject markers and that the acquisition of Korean by German students is guided by typological differences. We notice that the typological transfer of subject prominent features takes place continuously in the acquisition of Korean and that the pragmatically sensitive grammatical structures of Korean are cognitively hard to acquire for German native speakers. As regards the double nominative constructions as in (17) and (18), not a single instance is attested in the current database for level 1 and level 2 participants. It seems safe at this point to argue that compared to the other TP features such as null subject and the use of topic marker, the acquisition of double subject construction takes place later, although this construction is accessible to the learner as an early input. In this section, we have observed that when German students learn a TP language such as Korean, they rely on known linguistic knowledge before identifying the TP features of Korean. For a SP language

Transitivity parameter and prominence typology

157

such as English the primary requirement is that a subject should be specified, and that it should be realized before the verb. In addition, an object must also be mentioned in the case of transitive constructions (cf. SVO structure). For a TP language such as Korean or Japanese, it is required that the topic should be mentioned first in the sentence; the need to name a subject is a secondary matter: the subject is not marked by position and there may be no subject at all. As we have seen in (16) in section 3, the topical item or the shared information is usually deleted in the most natural discourse setting. In this section, we presented L2 evidence that the grammatical structure of native language controls or influences the output of adult second language learners considerably. It is clear that successful learning rests on the identification of the structural properties of the target language. The analysis of the data in this study also reveals that there are developmental stages of Korean acquisition by German native speakers. We have found that German learners of Korean overproduce subjects early on and then gradually learn to drop them as their proficiency improves, and that topic particles and double nominative constructions are particularly difficult features for Germans to acquire even in the later stage of learning. Thus, this process of learning Korean by Germans suggests that learners adopt the L2 acquisitional strategy of going for the typological indices that simplify to a maximum the procedure for learning a target-language grammar (cf. also Jin 1994 for Chinese).

6. Conclusion In this paper we have developed the idea that the transitivity parameter should be constrained such that pragmatically determined word order languages with high semantic transparency proportionally correlate with less grammaticization. The effects of this theoretical position are also reflected in the L2 data, such that the kinds of L2 competence that need to be acquired going from languages at each end of the typological continuum proposed in section 4 are either predominantly constructional or predominantly discoursal. It is evi-

158 Sang Hwan Seong

dent that SP languages are construction-based whereas TP languages are discourse-based. We have also seen that linguistic expressions that signify the same objective circumstances are construed differently from language to language. As Langacker put it (1988: 11), the ways in which functionally equivalent linguistic expressions are presented can be language-specific and in fact may vary from language to language. But it is not the case that the symbolic elements available to speakers are completely language specific and differ unpredictably. The idea which the present author wishes to promote is that the ways in which languages encode things are to a considerable extent parameterized in a regular fashion when comparable thoughts are expressed. In this respect we can state that the cognitive grammar view of languages is not completely relativistic.

Notes * I thank René Dirven and the other two anonymous reviewers for their valuable comments and suggestions on this paper. I am also indebted to John Attfield and Noah Isenberg for carefully reading the manuscript. All responsibility for errors remains of course my own. 1.As an anonymous reviewer points out, example la (intransitive verb) is of a different class from lb and lc (transitive verbs). However, this difference is not so relevant to this discussion because we are dealing with non-agentive subjects in English. 2. Some examples from the old Germanic dialects are in order (Valentin 1999): (1) jah warP in jainaim dagam... Gothic, Mk 1, 9 ('and it happened in those days') (2) uuerda thin uuilleo obar thesa uuerold alia. Old Saxon, Heliand 1604 ('your will be done all over the world') (3) a. that thi kind giboran...scoldi uuerdan. Old Saxon, Heliand 123 ('that the child should be born') b. nu ist Krist giboran. OS. Heliand 399 ('Christ is now born', 'Christ has now been born') (4) thaer hali-ern wearth tha geopenod and tha lac waeron in gebrohte. Old English ('the sacred house was opened and the offerings were brought in')

Transitivity parameter and prominence typology 159 3. Kuno (1973) illustrates the Japanese equivalent of the similar reflexive construction in (11). However, he does not explain why such a grammatical difference exists between English and Japanese. 4. There is evidence that the serialization of dative argument + accusative argument in ditransitive constructions should count as basic and unmarked order in German. Look at the following observation of Lenerz (1977: 44) which involves the focus pattern of dative and accusative objects: a. Ich habe das Geld dem KASSIERER gegeben. I have the money the cashier given b. *Ich habe das GELD dem Kassierer gegeben. c. Ich habe dem KASSIERER das Geld gegeben. d. Ich habe dem Kassierer das GELD gegeben. When the accusative patient precedes the dative recipient, the accusative argument has to be unfocused or unrhematised for the sentence to be grammatical. However, no such restriction applies to the reverse order "dative + accusative" serialization. Thus, the latter order should count as "unmarked" in the ditransitive constructions. This is also confirmed by Eisenberg (1994: 422). 5. Interestingly enough, we find this regularity in the diachronic development of German and Dutch. Shannon (1999) demonstrates, based on the following instances that the ordering of pronominal object and nominal subject in the middle field of Middle Dutch (MD) was an unmarked case and that Modern Dutch has largely given up this structure in favor of the opposite ordering. MD D

Doe sette hem sijn vader op sijn peert achter hem. Toen zette zijn vader hem op zijn paard, achter hem. 'Then his father set him on his horse, in back of him.'

Shannon also adds that German has preserved this pragmatically determined word order considerably, but has seen a substantial decrease. 6. An identical grammatical phenomenon is confirmed in Japanese (Y. Kasai [p.c.]). For details, see Seong (2000).

References Abraham, Werner 1995 Diathesis: The middle, particularly in West-Germanic. In: Werner Abraham, Talmy Givón, and Sandra Thompson (eds.), Discourse Grammar and Typology, Papers in Honor of John W. M. Verhaar, 3-47. Amsterdam: John Benjamins.

160 Sang Hwan Seong Burridge, Kate 1993 Syntactic Change in Germanic. Amsterdam: John Benjamins. Chafe, Wallace 1976 Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In: Charles Li (ed.), Subject and Topic, 25-55. New York: Academic Press. Drosdowski, Günther (ed.) 1995 Duden. Grammatik der deutschen Gegenwartssprache. Mannheim: Bibliographisches Institut. Eisenberg, Peter 1994 Grundriss der deutschen Grammatik. Stuttgart: J. B. Metzler. Faarlund, Jan T. 1992 The subject as a thematic category in the history of Scandinavian. Folia Linguistica. XXVI. 1(2): 151-68. Fillmore, Charles 1968 The case for case. In: Emmon Bach and Robert T. Harms (eds.), Universals in Linguistic Theory, 1-88. New York: Holt, Rinehart & Winston. Firbas, Jan 1992 Functional Sentence Perspective in Written and Spoken Communication. Cambridge: Cambridge University Press. Foley, William A. and Robert D. Van Valin 1984 Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Givón, Talmy 1979 On Understanding Grammar. New York: Academic Press. Goodluck, Helen 1991 Language Acquisition. Oxford and Cambridge: Blackwell. Gundel, Jeanette 1988 Universals of topic-comment structure. In: Michael Hammond, Edith Moravcsik, and Jessica Wirth (eds.), Studies in Syntactic Typology, 209-242. Amsterdam: John Benjamins. Haider, Hubert 1985 The case of German. In: Jindrich Toman (ed.), Studies in German Grammar, 65-101. Dordrecht: Foris. Hawkins, John A. 1986 A Comparative Typology of English and German". Unifying the Contrasts. Austin: University of Texas Press. Helbig, Gerhard and Joachim Buscha 1994 Deutsche Grammatik: Ein Handbuch för den Ausländerunterricht. Leipzig: Langenscheidt.

Transitivity parameter and prominence typology 161 Hopper, Paul and Sandra Thompson 1980 Transitivity in grammar and discourse. Language 56: 251-99. Jin, Hong G. 1994 Topic-prominence and subject-prominence in L2 acquisition: evidence of English-to-Chinese typological transfer. Language Learning 44(1): 101-22. Klaiman, Miriam H. 1988 Affectedness and control: a typology of voice systems. In: Masayoshi Shibatani (ed.), Passive and Voice, 25-83. Amsterdam: John Benjamins. Kuno, Susumu 1973 The Structure of the Japanese Language. Cambridge: The MIT Press. Langacker, Ronald 1988 An overview of cognitive grammar In: Brygida Rudzka-Ostyn (ed.), Topics in Cognitive Linguistics, 3-48. Amsterdam: Benjamins. Lenerz, Jürgen 1977 Zur Abfolge nominaler Satzglieder im Deutschen. Tübingen: Narr. Li, Charles and Sandra A. Thompson 1976 Subject and topic: A new typology of language. In: Charles N. Li (ed.), Subject and Topic, 457-89. New York: Academic Press. Müller-Gotama, F. 1994 Grammatical Relations: A Cross-Linguistic Perspective on their Syntax and Semantics. Berlin: Mouton de Gruyter. Nam, Ki-Sim and Koh, Young-Kun 1993 Standard Korean Grammar. Seoul: Top Publishing (in Korean). Noonan, Michael 1977 On subjects and topics. Proceedings of the Berkeley Linguistic Society 3: 372-85. O'Grady, William 1991 Categories and Case. Amsterdam: John Benjamins. Plank, Frans 1983 Transparent versus functional encoding of grammatical relations: a parameter for syntactic change and typology. In: Linguistische Berichte. 86: 1-13. Seong, Sang H. 1999a On word order in German and Korean. In: Gerald F. Carr, Wayne Harbert, and Lihua Zhang (eds.): Interdigitations·. Essays for Irmengard Rauch, 355-366. New York: Peter Lang.

162 Sang Hwan Seong 1999b

Semantic transparency and its implications: with special reference to German and Korean as SOV type. In: Irmengard Rauch and Gerald F. Carr (eds.), New Insights in Germanic Linguistics I, 233-260. New York: Peter Lang. 1999c Are Behaghel's Laws heterogeneous ones? To appear In: Sprachwissenschaft auf dem Weg ins nächste Jahrtausend: the Proceedings of the 34th Colloquium of Linguistics at the University of Mainz, Germersheim. 2000 Wie hätte Behaghel Koreanisch und Japanisch analysiert? Paper presented at the Ost-West Kolloquium für Sprachwissenschaft (3. 17-18,2000) at the Humboldt University in Berlin. (MS) Shannon, Thomas F. 1987 On some recent claims of relational grammar. In: Aske, Jon, Natasha Beery, Laura Michaelis and Hana Filip (eds.), Proceedings of the Berkeley Linguistic Society 13: 247-262. 1999 German word order change at the Millennium. To appear in: Sprachwissenschaft auf dem Weg ins nächste Jahrtausend: the Proceedings of the 34th Colloquium of Linguistics at the University of Mainz, Germersheim. Shibatani, Masayoshi 1985 Passives and related constructions: A prototype approach. Language 61: 821-848. Sohn, Ho-Min 1994 Korean. London: Routledge. Talmy, Leonard 1988 Force dynamics in language and cognition, Cognitive Science 12: 49-100. Taylor, John 1995 Linguistic Categorization: Prototypes in Linguistic Theory. Oxford: Clarendon Press. Traugott, Elisabeth C. and Bernd Heine (eds.) 1991 Approaches to Grammaticalization. Amsterdam: John Benjamins. Valentin, Paul 1999 Wirdhu and the Germanic passive. In: Gerald F. Carr, Wayne Harbert, and Lihua Zhang (eds.), Interdigitations: Essays for Irmengard Rauch, 141-46. New York: Peter Lang. Van Oosten, Jeanne H. 1984 The nature of subjects, topics and agents: A cognitive explanation. Unpublished Ph.D. Dissertation. University of California, Berkeley.

Transitivity parameter and prominence typology

163

Voorst, Jan G. van 1996 Some systematic differences between the Dutch, French and English transitive construction, Language Science, Vol. 18, Nos 1-2: 227-245.

Section 3 Neurocognitive and cognitive issues of language acquisition in general

Learning syntax - a neurocognitive approach Sydney M. Lamb

Not only every language, but every lexeme of a language, is an entire world in itself. Igor Mel'chuk

Like other papers in this collection, this one is concerned with learning, in particular, with the learning of syntax. It addresses the question of how children, or adults learning a second language, learn to handle what appear to be the syntactic categories needed for using a language. And in order to talk about this question there is a very important prerequisite: We need to understand just what it is that is being learned. According to a traditional view of the learning of syntax, the child, or the adult second-language learner, must somehow acquire syntactic rules. In one manner of speaking it is assumed that a speaker has "internalized" such rules. The assumption is that since syntactic rules are useful in descriptions of sentences, they must be present within the system that produces them. This mode of thinking comes from an unstated assumption, that patterns observable in linguistic data represent knowledge in the minds of those who produce such data. Is this assumption supported by any evidence? I have a hard time finding any basis for it. It is somewhat like supposing that since we can devise equations for describing the movements of planets around the sun, those planets must have internalized such equations. If we can find other sources for the patterns found, there is no reason to adopt this assumption (Lamb 1999: 227-247).

168 Sydney M. Lamb

Of course, there has to be some internal system that accounts for what people are able to say and to understand. But it need not be assumed to have the form of rules used in classifying utterances, nor is there any a priori reason for assuming that it contains the categories and other devices that may be employed by those attempting to describe them. And such attempts are particularly suspect as formulated by those who attempt to describe them in the most economical possible way, hence with the broadest possible categories and what have erroneously been called "linguistically significant generalizations". The discrepancies between the results of such taxonomic thinking and a more realistic view of the cognitive reality may be seen at all levels of linguistic structure, but they are nowhere more evident than in the area of syntax. An alternative is to treat most syntactic information, or even all of it, as attached to individual lexical items. In that case, the acquisition of syntactic knowledge is part and parcel of the acquisition of lexical knowledge and therefore occurs little by little as individual lexemes are learned. This view has become increasingly attractive in recent years. Yet there is also a lot of evidence for the existence in our cognitive systems of some kind of constructions, for example the argument-structure constructions described by Goldberg (1995). Such constructions, to be considered below, evidently make use of syntactic categories. And so we have a problem. Tomasello and Brooks (1999), who accept the cognitive existence of constructions of this kind (cf. Tomasello 1998) identify areas that require further investigation for developing a viable theory of syntactic learning, stating that "the various psychological processes involved in early syntactic development [...] need to be identified and characterised" (1999: 185). Of three such processes they identify, I would like to focus on this essential one: "[...] children's early skills to categorise not only isolated bits of language into item-based categories, but also their skills at categorising larger linguistic units into the various syntactic schémas and constructions that underlie much the productivity of human language" (1999: 185).

Learning syntax - a neurocognitive approach

169

To put it briefly: If there are syntactic constructions in a neurocognitive system, it is necessary to consider how they might be acquired by the child. Categories as such are not presented to the child for learning, only actual expressions.

1. What is it that has to be learned? Of course, before we can consider learning we need to understand what it is that is being learned. We need to consider the question of just what kind of information is involved. We commonly think of a category as a combination of objects - in this case linguistic objects. Two considerations make a difference here: First, what kind of objects? Second, what kind of combination? We can bypass such questions only at the danger of adopting unwarranted assumptions. One approach, too simple, would have it that the objects are morphemes. That is clearly to be rejected, as syntactic categories often have sequences as members. Moreover, the proper basic unit for syntax is not the morpheme but the lexeme. The term lexeme was coined by Benjamin Lee Whorf over half a century ago, but has yet to become widely used, despite what seems to me its obvious usefulness, indeed its indispensability. Part of the reason is that linguists have been so influenced by analytical rather than cognitive concerns. In the analytical tradition one attempts to analyze as much as possible, hence down to minimal units; for grammar, down to the morphemes. By contrast, the cognitive view of what a person learns, taking into account the workings of the brain, recognizes that people quite commonly learn larger combinations as units, and not just for language. For example, the lexeme activity is surely learned and used as a unit by English speakers despite the fact that it can be analyzed into three morphemes. This principle applies much more broadly as well. We learn whole phrases and clauses as units (hence, phrasal and clausal lexemes), like it doesn't take a brain surgeon to realize that ... and tell it like it is. Another common misconception is that words are the units with which syntax is concerned. But a lexeme can consist of multiple

170 Sydney M. Lamb

words or of just part of a word, for example, the plural ending -s or the past tense ending of verbs; they occur freely even with newly coined nouns and verbs. The other question posed above is of greater interest: what kind of combination is a category? Is it a set of lexemes and combinations of lexemes? For example, do we need to know a category Noun Phrase as a list whose members are all the noun phrases of English? Clearly not, for there are many possible noun phrases that, while perfectly acceptable once received in a suitable context, have never yet been uttered. We might say, adopting a commonly held point of view, that a syntactic category is a set consisting partly of individual forms (lexemes) and partly of combinations generated by rules. Such rules, representing constructions, of course use categories themselves, and their categories are similarly defined. According to this view, the knowledge one needs to learn consists of the rules for combinations plus a list of the individual members of the categories (like the nouns, verbs, etc.) not specified by rules. But I reject such a view, not only as too simple minded but, more important, as cognitively implausible. In this paper I develop an alternative view. Still under the heading of considering what it is that must be learned in order to have a command of the syntax of a language, let us next be explicit that syntax is concerned with combinations of lexemes and indeed with a hierarchy of combinations - phrases, clauses, sentences, etc. So we need to take a look at the kinds of combinations a person must be able to command, starting with the simplest, a "combination" of one. And of course we take a neurocognitive point of view. First, then, we have the lexeme. But I would like to suggest that we should recognize many more units as lexemes than are usually considered. The cognitive orientation forces us to accept that people learn as units any combination that has occurred with sufficient frequency or to which sufficient attention has been given, as a consequence of the brain's natural tendency to "absorb" repeatedly occurring phenomena. If a locution attracts enough attention, it will be learned on the basis of very few occurrences. For example, at the time of the Persian Gulf War, a new lexeme was introduced into

Learning syntax - a neurocognitive approach

171

English: the mother of all battles. It arose from a statement of Saddam Hussein, or rather from the English translation of his statement, which in Arabic was less colorful. It only took that one statement, heard perhaps once or a very few times by Americans in newscasts over the next few days, for that lexeme to be learned and, for a while, widely used. This one also illustrates that some lexemes have a rather transitory existence, as its use gradually declined over the next several months to the point at which ten years later one encounters it only rarely if at all. So we have simple lexemes and complex lexemes. The latter are not to be conflated with idioms, by the way. Idioms are those complex lexemes whose meaning is not clear from the meanings of their constituents. That is, a lexeme can be transparent or opaque, and these are not two values but the endpoints of a continuous scale. Some lexemes are quite transparent, like plastic bag, others are quite opaque, like red herring. A lexeme like blackboard is relatively transparent, but a blackboard doesn't have to be black. Another continuous scale relating to lexemes is that of entrenchment. A lexeme becomes more entrenched with more use: The neurocognitive pathways which support it become stronger the more they are traveled (Lamb 1999: 164-166). This is the first of several observations in this paper which support the hypothesis that a neurocognitive system has the form of a network. Transparent lexemes can be interpreted in two ways: either via the constituents or via the whole. This is no problem for a network approach. In fact, the fact that both kinds of interpretation do occur, presumably in parallel, constitutes further evidence in favor of the network model (cf. Lamb 1999: 167). It is surprising how much ordinary English text is made up of complex lexemes. This observation is important for the study of the cognitive operations relating to syntax since combinations which are "stored" in memory as units don't have to be constructed for their production or understanding. Consider the following illustrative examples of lexemes in English:

172 Sydney M. Lamb

Relatively short: horse sense no laughing matter not written in stone as clear as mud

strictly speaking a people person the bottom line a New York minute

Relatively longer: round up the usual suspects if it ain't broken don't fix it you can bet your bottom dollar between a rock and a hard place

painfully obvious a no-brainer a dumb question right then and there

it ain't over till it's over you know what I mean the truth of the matter is been there, done that

But of course there is a great deal that cannot be accounted for so simply. At the next level of complexity, I would like to suggest, we have the complex lexeme with a variable constituent. Any complex lexeme can be said to be composed of constituents; for example, red herring has red and herring as its constituents. For very complex lexemes, like a stitch in time saves nine, we can also distinguish immediate constituents from ultimate constituents. The variable constituent can be illustrated by what happened to the lexeme the mother of all battles soon after it was introduced into English. It developed into what can be called a mutable lexeme (Lamb 1999: 263-266), as the constituent battle mutated into others, like meteor. A story in the New York Times describes a spectacular meteor that was seen in the sky in the Eastern United States at around the time of the Persian Gulf War. An airline pilot who witnessed it was quoted as calling it the mother of all meteors. Moving on to the next level of complexity, we have what could be called the lexeme with more than one variable constituent. As examples, we have you don't have to be a <X> to that X: brain surgeon, rocket scientist Y: understand, appreciate, see it comes as no [great] surprise that

Learning syntax - a neurocognitive approach

173

The constituent enclosed in square brackets is an optional constituent, a type of variable in that it is a constituent varying with zero.

2. Constructions Pushing this notion further, we arrive at the limiting case, in which all constituents are variable and in which all have a wide range of values. This is what is more commonly called the construction, in which the variable constituents are called syntactic categories. We have as examples the argument structure constructions treated by Adele Goldberg (1998). Following Goldberg, we may identify these constructions: intransitive motion ditransitive caused motion resultative conative

The fly buzzed into the room he faxed Bill a letter she pushed the pencil off the table they wiped the counter clean she kicked at Henry

According to Goldberg, "Constructions which correspond to basic sentence types encode as their central senses event types that are basic to human experience." (1995: 39). I differ with Goldberg's treatment in considering all of these to represent verb phrase constructions rather than sentence types. This is a minor difference and does not require us to change Goldberg's names for them. The treatment as verb phrases is needed to account for their occurrence in the infinitive form, for example, (ditransitive) to fax Bill a letter (that is, with no subject expressed), as in The boss asked her to fax Bill a letter, or as participles, as in (intransitive motion) Sauntering into the room, she cast a glance my way. None of them have anything special in the relationship of the subject to the verb phrase. Instead, I believe we need to reckon with another basic construction with very broad scope, the Actor-Action construction (so-called by Bloomfield 1933: 172).

174 Sydney M. Lamb

Additionally, we need to consider, along with the verb phrase constructions listed above, the more basic one, transitive

she smacked Helen (cf. Goldberg 1995: 117-119)

In this illustrative example, I include the subject she not because it is a part of the construction but just to keep the example parallel to those listed above. As with them, the addition of the subject comes about in keeping with the operation of the actor-action construction. Related to the transitive construction, in fact dependent on it, is the passive

Helen was smacked [by her]

I take it as reasonable to assert that all combinations of lexemes, that is all of syntax, can be accounted for in terms of these types discussed above: complex lexemes, complex lexemes with variable constituents, and constructions (which could be called complex lexemes with multiple variable constituents, except that the term construction is more established and more convenient).

3. Syntactic categories and their members Now we are ready for our basic question, that concerning the nature of categories. We have to ask just what information must a child (or second-language learner) acquire in order to handle the syntactic categories of the language. Does such knowledge consist, for example, of a listing of the membership of the simple categories (like noun, verb) together with a set of rules for generating the more complex ones. If not, what? It is important to recognize that the notion of category comes from analytical linguistics, an essentially noncognitive endeavor. As with other tools of analytical linguistics, we are not obliged to suppose that they are internalized in the mental systems of speakers.

Learning syntax - a neurocognitive approach 175

In keeping with what has been observed so far, we could rephrase the question in terms of variable constituents rather than categories: How does a language learner learn how to use variable constituents? But for now, let us continue with the more traditional notion of syntactic categories. The first observation, an important one, is that as syntactic categories have been traditionally employed, they don't work. They are just rough approximations - not wholly useless, just not cognitively plausible. Approaching them with prototypicality notions helps, but ultimately, the closer we look at any syntactic category, the more it breaks down, until we get down to the individual lexeme. Consider, for example, the category of prepositions. If they do constitute a syntactic category, they behave alike syntactically. If they do not, how can the category specify their combinations? So let us take a look: We have in love but not *on love, yet we have on vacation but not *in vacation. The following are examples of the use of

basic prepositions with various objects for describing states that one can find oneself in or in which one can do things. We have: Preposition *at, *by, *at, *by, ?at, by, ?at, *by, *at, *by, *at, *by, *at, *by, *at, *by, at, *by, *at, *by, *at, *by, *by, at, ?at, *by, *at, *by, at, *by, *at, by, *at, *by,

?in, ?in, ?in, in, in, in, in, in, *in, in, in, in, ?in, *in, ?in, in, *in,

*out of, *out of, *out of, out of, out of, out of, ?out of, ?out of, *out of, out of, out of, out of, *out of, *out of, *out of, *out of, *out of,

?under, with *under, with *under, *with *under, ?with *under, ?with ?under, ?with *under, ?with *under, with *under, *with *under, with *under, *with *under, ?with *under, with under, *with *under, *with *under, *with *under, *with

Object assurance candor chance danger desperation doubt fear insecurity leisure love pain play pleasure pressure rest thought vacation

176 Sydney M. Lamb

Notice that the last noun listed, vacation, does not occur with any of the prepositions considered here, but does occur with on, and that on does not occur with any of the other nouns in the list. By the way, the judgements presented are mine and they may differ in some details from those of other native speakers. For a neurocognitive approach, that is no problem, since the object of investigation in neurocognitive linguistics is the neurocognitive system of the individual speaker, not some disembodied "language"; and we recognize that the system of every speaker differs from that of every other. Given findings like those listed above, what cognitive sense can we make of the notion that there is a construction utilizing the categories Preposition and Noun Phrase? Syntactic categories are based upon an old tradition, that of the "parts of speech", which goes back to the ancient Greeks. According to the doctrine of parts of speech, every word must belong to one or more of these categories - and there is a catch-all category "adverb" for the difficult cases. Now, what about tantamount! According to the ordinary dictionary, which treats the parts-of-speech myth as reality, it is classed as an adjective, as is the equally unique akimbo. Also of clearly unique distribution, but classed as a preposition, is between. Let's take a brief look at verbs. From the point of view of their syntactic distribution there are clearly many different kinds. We might be tempted to suppose that if we subcategorize to a sufficient degree we will get down to subcategories whose members behave alike. We might, for example, consider just the subcategory of verbs of perception: see, hear, listen, smell, etc. But only a little observation makes it clear that even this tiny subcategory doesn't help us to define what can and can't occur syntactically. Apart from the fact that we see visible objects but hear audible things, we have the different aspectual possibilities: You see and hear punctually and completively, but you listen [to] and look at duratively. We don't have to look very far to see that each of these verbs, and indeed every verb of any other subcategory, has its own distribution. And the same can be observed about members of any of the other parts of speech.

Learning syntax - a neurocognitive approach

177

And so we conclude that every lexeme has its own syntax. Now that is a conclusion which puts the validity of the concept of syntactic category into considerable doubt. How can there be syntactic categories, as constituting part of the information used by speakers of a language, if they don't work? And if they are useless, how can we explain why they come up so regularly in discussions of syntax? The answer to this question is that such categories result from applying the taxonomic techniques of analytical linguistics. Analytical linguistics is concerned with analyzing and categorizing and describing patterns found in things that people say. It is natural in such a pursuit to classify things that appear to share properties. In such classification it is easy to overlook that the things categorized together do not really behave alike. In any case, there is no justification for assuming that concepts of analytical linguistics can be taken over directly into an understanding of the cognitive basis of language. An alternative to the taxonomic approach is that most syntactic information, or even all of it, is attached to individual lexical items. In that case, most of the syntactic generalizations that can be formulated in rules are epiphenomenal, and the actual internal information that gives rise to them is widely distributed, among thousands of separate items. According to such a view the acquisition of syntactic knowledge is the acquisition of lexical knowledge.

4. Learning syntax as (mostly) learning lexicon If it is the case that every lexeme has its own syntax, then it follows that the only way to learn syntax is to learn lexicon. If this is so, then what seemed to be a process of learning syntax is really just the learning of vocabulary, a process that occurs one lexeme at a time. This view makes sense not only because every lexeme has its own syntax, but also because, as a consequence, you can't know how to use a lexeme without knowing how it connects with other lexemes. This conclusion is strongly supported by findings of Elizabeth Bates et al. (In press), who have examined the correlation between development of grammatical complexity and vocabulary size in chil-

178 Sydney M. Lamb

dren. They have found these two aspects of language development are very strongly correlated. As they write, ... the relationship between grammar and vocabulary development has turned out to be one of the strongest associations that has ever been observed in any aspect of human development (with the possible exception of height and weight!). ... the relationship holds at every point from 50-600 words (covering the period from 16-30 months of age). One certainly might have expected a relationship at the lower end of the distribution, simply because one cannot combine words until there is something to combine. We might also have expected some kind of "trigger" or "threshold" relationship between vocabulary and grammar, e.g. a critical number of words that need to be acquired for grammar to get off the ground. What we find instead is a continuous and accelerating function that holds at every point across this period of development.... The powerful relationship holds for very late talkers, and very early talkers as well.

Yet to conclude that learning syntax is just part of learning vocabulary leaves us with a lot to explain: What about the broad similarities in ordering: Subject before predicate (in the unmarked clause), preposition before noun phrase, etc.? And what about the general constructions like the argument structure constructions described by Goldberg? They surely use categories. And what about the ability, even of young children, to produce and to understand new combinations?

5. Participant roles Let's consider the case of a simple verb like eat, accepting the observation that its syntactic distribution is unique. That being the case, we are tempted to conclude that the operative knowledge used in producing a sentence like Mommy's eating an apple, with its ordering of the two participants in relation to the process - the agent before and the patient after the verb - is in large part information connected directly with the lexeme eat. Yet it seems also to be the case that the actor-action construction and the do-smthg-to-patient construction (i.e. the transitive construction) are also involved, even if their in-

Learning syntax - a neurocognitive approach

179

volvement can be seen as peripheral in relation to that of <EATER> EAT . Ordinary analytical linguists will balk at the suggestion of such a proposal, since the two methods of handling the combination appear to be in competition with each other. Why have two methods where one will suffice? But the cognitive considerations are essential here, as elsewhere. We have evidence that both kinds of knowledge may be presumed to be present in the cognitive system. For example, without the actor-action construction and the transitive construction, how can we account for the passive construction? How two seemingly competing methods of producing the combination can coexist is no problem for a network approach. The network allows multiple structures to operate in parallel (cf. Lamb 1999: 233-236). The presence of redundancy is likewise no problem in a cognitively realistic approach, as we need to recognize that our brains do learn many kinds of things redundantly. As a transitive verb, eat co-occurs with an agent and a patient, and so it entails two categories: that of its possible agents and that of its possible patients. We have two questions to consider: First, what does the information specifying the membership of each of these categories consist of, how is it organized, how represented in the mind of the child? Second, how is this information learned? How does the child (or the grown-up learning a second language) learn the ordering of these major constituents of the clause, often seemingly after having heard just a few examples? What, then, is the status of <EATER>? IS it a category? If it is, it is a semantic category. And as such it would seem to consist of all the possible agents of EAT. But the more relevant question is: What must a child learn, what must a speaker of English know, in order to use EAT productively? First, we have to understand that it is not possible to understand what eating is apart from eaters and eatees. Knowledge of a range of possible participants is part and parcel of the knowledge of the process - not something separate, as the lexemes Mommy and apple are separate from eat in the linguistic expression of the process MOMMY EAT APPLE. You can't have eating without <EATER> and <EATEE> - it is simply impossible. Therefore, the participant information must be attached to individual processes. The same observa-

180 Sydney M. Lamb

tions can be made for perception: Your visual system apprehends a person walking; the walking simply cannot be perceived without perceiving the person (or other legged creature) performing the process. Now we can explain why it is that a child doesn't need more than a very few examples like this to learn how to express instances of eating together with its participant categories. The child already knows what eating is through his/her own experience of eating and his/her observation of others. That is, she already has a subnetwork of semantic information, with connections in the motor and kinesthetic areas of her cortex, representing her own experience, together with visual and perhaps also auditory connections, based on observation of others. It is just a question of connecting this semantic subnetwork to a means of linguistic expression. As with lexemes generally, a child (or grown-up - in fact anyone except an analytical linguist) focuses on the meaning that a lexemic node is connected to more than on the lexeme itself. And you simply can't know the meaning of Leat without knowing that the process requires eater and food; and so the participant categories are simply EATER and EATEE. We don't need to ask that these categories be more clearly defined, nor that they be defined in some general way that could apply for the whole speech community. They will be defined for each child according to that child's semological system, according to that child's knowledge (already present) of what eating is, as it exists at the time of learning of the word eat. Later, as the child learns more, it may change its understanding of the categories, as an automatic consequence of changing her understanding of what constitutes eating. So what the child actually needs to learn is how to form linguistic expression for the already existing semantic information. The expression includes not only the phonological realization of eat but the sequencing information: The realization of the <EATER> comes first, then eat, then the realization of the <EATEE> (unless some other construction like passive intervenes). It only takes one example, if clearly understood, perhaps then supported by one or two additional examples for confirmation and reinforcement, for the child to have control of the lexemic realization of the process.

Learning syntax - a neurocognitive approach

181

In other words, since there are two participants in the (already known) process, the child has to learn not only a phonological expression for eat but also the ordering of the participants in relation to it. Until this ordering information has been learned the child knows from experience and observation that eating involves two participants but doesn't know their relative order in the linguistic expression of the process. As soon as she hears (and pays attention to) an example, like Mommy's eating an apple she knows, since she recognizes MOMMY as eater and APPLE as eatee. That is, she now knows that <EATER> precedes eat and <EATEE> follows. So now, without any additional information that is not already in her network, the child is able to use the word eat more or less like older members of the speech community. In the future use of this verb, a child (or adult) will produce and understand it in some ways that depart very little from those encountered earliest and most often, and in others that are more different perhaps some day for a car "eating" gasoline and even for acid "eating" pipe. It depends entirely on whether the process under consideration is construed as an instance of eating. And so <EATER> is not a category whose membership has to be learned. Rather, the value of <EATER> depends entirely upon the knowledge of what constitutes eating. For the producer of an occurrence of <EATER> EAT , it is entirely a question of whether or not the action under consideration can be construed as eating. If so, then it has an agent, as it must, since eating cannot occur without an eater. This point is at once very simple and altogether essential for an understanding of the knowledge underlying syntactic performance. Its simplicity makes its import easy to overlook, so I shall emphasize it by repeating, in different words: If the process under consideration is construed as eating, then whoever/whatever is doing it is the <EATER>, and by virtue of just that fact becomes at that moment a member of the "category" <EATER>. The consequence of this simple observation is that the child does not have to learn the membership of the category as such. Rather, the learning needed is just the learning of what eating is. And that knowledge is already present as conceptual and perceptual and motor network structures - even before the

182 Sydney M. Lamb

child learns the lexeme Leat. For any normal child has such structures, and therefore a concept of what eating is, from early on in its life. To be sure, the knowledge of what eating is may well undergo further development as the child matures and learns more about the culture, the mythology, stories, and so forth. Eventually a person may come to accept, or to reject, such notions as an automobile eating gasoline or a furnace eating coal. He may well come to appreciate the riddle What has three legs and eats marbles? (Answer: A threelegged marble-eater). Whether accepted or rejected, it depends not upon the "category" <EATER> as such but on EAT. If the process can be called eating, then whatever is performing it is the <EATER>. NO separate knowledge of that "category" is needed. The foregoing observations definitely simplify the task of the cognitive syntactician, for they eliminate the whole problem of determining what knowledge must be learned to learn such „categories". For the answer is that no additional knowledge is needed beyond knowledge of the process itself. To be sure, there is still plenty to investigate: What form does the knowledge of eating have in a person's brain, and how is that knowledge acquired? But note well that the problem of answering those questions was a problem already present for cognitive neuroscience anyway. What I am claiming is that no further knowledge beyond that is needed for syntactic purposes. In the preceding two paragraphs I have started to put category in quotation marks, because it is apparent that we are no longer talking about what the term has commonly meant in discourse about syntactic categories. The difference is clearly seen in the context of the question of what information in the cognitive system gives rise to the appearance of categories in analytical linguistics. More commonly that information would be seen as, in one way or another, specifying the membership of the category. That notion of category, as involving one-and-many, however useful it may be in analytical linguistics, is now seen to represent an illusion from the neurocognitive point of view. For this reason I prefer the term variable, free from such connotations, and so I shall use it from now on.

Learning syntax - a neurocognitive approach

183

6. Syntactic presupposition and variable constituents And so, as has been observed many times in the past, we have verbs presupposing participants while nouns do not. And we have not just theoretical reasons based on analysis of linguistic data for such an assertion. It is surely related to the fact that Broca's aphasies, whose area of damage is in the frontal lobe, typically have trouble not only with phonological production and with grammar, but also with verbs, much more so than with nouns. And they also have trouble with prepositions and with "function words" generally. All lexemes other than nouns evidently presuppose some other constituent or constituents with which they normally co-occur, just as eat presupposes <EATER> and <EATEE>. We have for example clause introducers like clearly, which presuppose a following clause or . Like it are other assertion introducers, including those described above as lexemes with variable constituents. We now see that what was written above in citing them was incomplete and that they should be written as follows, to include the presupposed constituent (with Lclearly also shown): you don't have to be a <X> to that X: brain surgeon, rocket scientist Y: understand, appreciate, see it comes as no [great] surprise that clearly, In the first of these examples we now see that there are three variable constituents, not just <X> and <7>. We also observe that a variable can range over a small number of values, like <X> and of this example, as well as over a large number of values, like , <EATER>, and <EATEE>. Similarly, the "mother of all ..." construction, identified above as a mutable lexeme, is seen, if we look at its syntactic properties, to be a nominal modifier; that is, it presupposes a noun or noun phrase, a

184 Sydney M. Lamb

realization of a member of the conceptual category notation like the above, we can write it as follows: the mother of all

.

Using

-S

Likewise, the passive construction may be written b e - e d [ b y ].

As the use of small caps is intended to indicate, the pertinent information for these variables is evidently at the semantic-conceptual level rather the lexemic. It's not categories like noun, verb, etc., but semantic information. On the other hand, the variables <X> and above range over lexemes, not semantic units (except insofar as lexemes are connected to semantic units). We can use the "mother of all..." lexeme to consider the process by which a lexeme becomes mutable. This change occurred in adult systems but is doubtless like that which occurs much more often in children and which accounts for much of their syntactic development. For the process to occur, the lexeme the mother of all battles has first to be segmented - in particular, battle has to be mentally segmented from the rest - and then other things are seen to be substitutable for battle. The segmentation of battle from the rest at the level of expression has a corresponding segmentation at the semantic level, so that the mother of all is seen to have some meaning of its own, perhaps paraphrasable as 'spectacularly large'. Such segmentation is automatic, since battle is recognized along with the whole lexeme, as are its other constituents. That is, the lexeme as a whole and also its constituents are activated in parallel in the neurocognitive network. When battle is segmented, the remainder of whatever meaning is connected to the mother of all battles is assigned to the remainder of the expression, i.e. the mother of all ...-s. The conditions for the occurrence of this process, for example, for the substitution of meteor for battle, are presumably that the situation involving the meteor was sufficiently similar to that involving battle to call up that (heretofore) fixed lexeme. This makes it easy to substitute meteor and other nouns

Learning syntax - a neurocognitive approach

185

for battle to describe situations in which something (anything) is spectacularly large or important. And when this happens, the mother of all has become another noun modifier. Note that the similarity involved here is semantic. And the range of values of the variable is exactly those things which can be construed as spectacularly large. It is an ad hoc variable, specific to this mutable lexeme. Garry Trudeau (the author of Doonesbury), more creative in his use of English than most, took it one step further in a column in the New York Times, March 21, 1991, in writing about the appearance of a can of Diet Pepsi on the table in front of General Norman Schwartzkopf, commander of the victorious allied forces in the Persian Gulf war, in the official pool photo of the cease-fire ceremony, beamed around the world to thousands of newspapers. He wrote, "It was the mother, father and next of kin of all product placements - too insanely great to be true."

7. Constructions as lexemes with variable constituents Now, finally, we have to consider the more general constructions, like Goldberg's argument constructions and the actor-action construction. These seem to be quite different from <EATER> EAT , since they are quite general in their coverage while the latter is tied to a specific lexeme. First, we can ask why they need to be recognized at all, if each of the constituents in any instance has its own syntactic information in accordance with the principle that every lexeme has its own syntax. The answer comes from the consideration of cases like Goldberg's example Frank sneezed the tissue off the table. The ability of sneeze to occur in such an expression is surely not a syntactic property we want to attribute to sneeze, normally an intransitive verb (Goldberg 1995: 152). Rather, its occurrence here is accounted for by the caused-motion construction itself, which evidently accommodates any verb whose performance can be construed under any circumstances as causing motion. Similarly anomalous from the point of view of its own syntactic properties is the use of conserve in But can

186 Sydney M. Lamb

Californians conserve their way out of the energy crisis?, heard on NBC Nightly News, January 26, 2001. Now conserve is a transitive verb, but it takes as patient something conserved, and in this example, way is not to be construed as such a patient. Rather, this usage is accounted for by the way-construction, seen also in such examples as he dug his way out of the prison and she fought her way into the room (Goldberg 1995:16 ) The caused motion construction (omitting the subject, which I do not consider part of it, see above) can be written <MOVE> , where stands for "directional", covering such possibilities as away, into the room, off the table, out of here, etc. Here, <MOVE> represents any action that can cause motion, and so it permits any verb that can be so construed, even sneeze, to impart motion to a suitable patient - that is, anything construable as a suitable patient. Notice that we are talking semantics and not about syntax as traditionally treated, which won't work. As soon as a speaker can construe any action as an instance of caused motion, the values of the variables are those which fit this construal of the situation. It's the same principle as that discussed above in connection with <EATER> <EAT> <EATEE>.

The question we must now ask is: How can such a construction, with all of its constituents variable, get learned, since the input to any child consists of specific utterances, not generalizations like this construction? A reasonable answer seems to be that the process is like that sketched above for the mutable lexeme the mother of all -s. The child may be presumed to start with a specific instance, say push it away or push it o f f , acquired first as a lexeme, and then to make it mutable by substituting other constituents for one of these three. Later, he substitutes also for another constituent, and then for the third. At this point all three constituents will have become variable. Notice that, as before, none of these variable constituents needs to be seen as a category whose membership has to be learned or stored. How could we ever devise a category definition that could accommodate in advance all of the objects that can participate in this construction?

Learning syntax - a neurocognitive approach

187

And so the difference between the construction and the lexeme with variable constituents is not that great. In fact a construction might be considered a lexeme with more than one variable constituent. Some constructions also include one or more fixed constituents, others do not. Among those which do are the way-construction (they made their way to the door) and the passive construction, which includes the verb be asa fixed constituent. And as we have seen, there are also variables which have a very small number of values, which can be enumerated, for example, it doesn't take a X to Y that , in which both X and 7 have just a few possible values.

8. Learning syntax in a second language The above rough descriptions of what I take to be reasonable scenarios for learning syntax are in the context of the developing syntax in one's native language. What about second-language learning? It happens that this same technique, as developed by Leonard Bloomfield and his colleagues, was being used in some of the second language teaching at Yale University during the time when I was an undergraduate there. It was used in my classes in German and Russian. The method was very simple: The course consisted of a series of lessons, each of which began with a number of sentences to be memorized. They were whole sentences, not just words or phrases. Each of them had a function that was both known and useful to the student. Since they were memorized (and the students were given a weekly test in which they had to reproduce them verbatim), they became lexemes within the student's cognitive system. Then each lesson guides the student in making substitutions for one or more of the constituents of these sentence-length lexemes. I still remember some of the early Russian sentences, over fifty years later: U mjinja yestj karandash Ί have a pencil'. U minja njet karandasha. Ί don't have a pencil', U mjinja estj pjiro Ί have a pen', U mjinja njet pjira Ί don't have a pen' (cf. Cornyn 1950). Since we were adult college students, the lessons also included a little grammar section in which were imparted some of the grammatical principles involved in the sentences we

188 Sydney M. Lamb

were memorizing and learning how to develop into mutable lexemes. But the grammar presentation was secondary. Another feature of this method was that we were not introduced to the writing system of either language during the first portion of the course, not until several weeks into the term, and we had to learn to pronounce with a high degree of accuracy. Native speaker coaches would make us repeat endlessly until we got the right pronunciation. Our learning thus had to utilize the auditory and motor cortices, for phonological recognition and production respectively. Had we been introduced early to writing, we would have used the college student's natural reliance upon written forms and would thus have been attempting to learn the forms using the visual cortex rather than the auditory and motor cortices. This method of language instruction seems to me to be based upon the same principles used by children in their native language development, and I don't believe that any better method of second language teaching has ever been devised.

9. Concluding observations Let me sum up by mentioning some of the principles that have guided this study and some of the conclusions reached. First, it is apparent that neurocognitive and analytical approaches to language have different concerns from each other. What I call analytical linguistics is concerned with analyzing linguistic data, utterances, sentences, and the like, and with finding patterns in such data, often guided by theoretical concerns that have little cognitive basis and usually no neurological basis at all. In neurocognitive linguistics, by contrast, while such data is still examined, the object of study is the neurocognitive system of the individual. This difference of focus has a number of consequences, not least of which is the recognition and acceptance that the system of every individual is different from that of every other. More important for the concerns of this paper is that by taking the natural operation of the human brain into consideration we recognize that linguistic information, like other kinds of informa-

Learning syntax - a neurocognitive approach

189

tion, is often represented redundantly in the neurocognitive system. The tendency of analytical linguists to seek out the most economical possible means of handling a given body of data is seen to be lacking any neurocognitive motivation. Second, I have argued that syntactic categories of the kind usually recognized are artifacts of analytical linguistics, based on approximations and motivated by the desire for economical description. They appear not to have any direct cognitive basis. The same can be said for some of the forms in which constructions based on such categories have often been conceived.. The human cognitive system represents information as connectivity in a network. It operates by means of widely distributed representations and parallel processing. As a consequence, linguistic forms can be recognized or produced by means of different structures operating in parallel. So for example, both a general construction and the information pertaining to lexemes may operate jointly to produce a given sentence or phrase. In contrast to the usual approach in analytical linguistics, we need not suppose that people use just a single unified strategy for producing and understanding combinations of lexemes. The brain thrives on redundancy and on multiplicity of strategies. I have also argued that every lexeme has its own syntax, and that learning syntax is mainly a matter of learning lexemes. This is a oneby-one process. And even the general constructions can be seen as lexical items, to be learned one at a time. There is a lexical scale from the fixed lexeme at one end, to the lexeme with a variable constituent of limited range, to the lexeme with variable constituent with broad range, to that with multiple variable constituents, and at the right end of the scale, the construction, with all constituents variable. Moreover, it is plausible that such constructions are acquired by a process that begins as the learning of a fixed lexeme, which then becomes mutable as segmentation occurs, leading to the possibility of substitution, whereupon a constituent becomes variable. And such segmentation is automatic, as soon as such a constituent is recognized as a unit. It is recognized as a unit while the lexeme as a whole

190 Sydney M. Lamb

is also recognized, in accordance with the parallel processing of activity in the network. Also, the knowledge needed to manage variable constituents like the participant roles of processes (e.g. <EATEE>) does not include any information other than that pertaining to the process itself (e.g. <EAT>).

Finally, I suggest that the best method of second language instruction may be that which comes as close as possible to utilizing these same techniques, presumably those used by children in native language development.

References Bates, Elizabeth, Donna Thai, Barbara Finlay and Barbara Clancy In press Early language development and its neural correlates. In: I. Rapin and S. Segalowitz (eds.), Handbook of Neuropsychology, Vol. 7: Child Neurology [2nd ed.]. Amsterdam: Elsevier. Bloomfield, Leonard 1933 Language. New York: Henry Holt Cornyn, William S. 1950 Beginning Russian. New Haven: Yale University Press. Goldberg, Adele 1995 Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Hockett, Charles F. 1997 Approaches to syntax. Lingua 100: 151-170. Lamb, Sydney M. 1999 Pathways of the Brain: The Neurocognitive Basis of Language. Amsterdam/Philadelphia: John Benjamins. 2000 Neurocognitive structure in the interplay of language and thought. In: Martin Pütz and Marjolijn Verspoor (eds.), Explorations in Linguistic Relativity, 173-196. Amsterdam/Philadelphia: John Benjamins. Mel'chuk, Igor 1988 Dependency Syntax. Albany: SUNY Press. 1998 The meaning-text approach to the study of natural language and linguistic functional models. LACUS Forum XXIV: 5-19.

Learning syntax - a neurocognitive approach 191 Tomasello, Michael 1998 The return of constructions. Journal of Child Language 25: 431442. Tomasello, Michael and Patricia J. Brooks 1999 Early syntactic development: A construction grammar approach. In: Martyn Barrett (ed.), The Development of Language. Hove, UK: Psychology Press.

Conceptual primes in early language development Cliff Goddard

1. Introduction In his inspirational book Acts of Meaning, the esteemed psychologist Jerome Bruner (1990: 72) argues that: "how we 'enter language' must rest upon a selective set of prelinguistic 'readiness for meaning'. That is to say, there are certain classes of meaning to which human beings are innately tuned and for which they actively search." Bruner here articulates a nativist approach to language acquisition, but a conceptual/semantic nativism rather than the syntactic nativism of Chomskyan linguistics.1 Bruner continues: "Prior to language, these exist in primitive form as protolinguistic representations of the world whose full realization depends upon the cultural tool of language". The present study explores certain hypotheses about the nature and identities of the innate concepts which may underpin language acquisition. These hypotheses have arisen from one of the most promising and productive approaches to cognitive semantics - the natural semantic metalanguage (NSM) approach originated by Anna Wierzbicka (1972, 1980, 1996, cf. Goddard 1998a, Goddard and Wierzbicka eds. 1994, forthcoming). Though the NSM approach has been responsible for literally hundreds of descriptive studies in lexical and grammatical semantics and pragmatics across a wide range of languages,2 it has not been applied very extensively to language acquisition. The only previous studies, Wierzbicka (1995) and Tien (1999), are unpublished. I hope to show, however, that the NSM approach generates interesting research hypotheses on language acquisition and allows for increased precision and testability in the notoriously diffi-

194 Cliff Goddard

cult area of child language semantics. In particular, it enables one to propose concrete and constrained semantic analyses of early "child meanings", proposals of a kind which are surprisingly sparse in the otherwise abundant literature on early lexical development (e.g. Bates et al. 1988, Bloom 1973, Bowerman 1978, Braine 1976, Clark 1993, Dromi 1987, Gopnik and Meltzoff 1997, Halliday 1975). The general outline of the NSM theory is well-known (but see Goddard 1998b for a discussion of popular misconceptions), so I will not go into it in great detail here. The essential claims are: (i) that at the heart of any language there is a specifiable set of simple, indefinable concepts, i.e. conceptual primes, also known as semantic primes, (ii) that these conceptual primes can be found embodied as the meanings of ordinary lexical items (words, bound morphemes, or phrasemes) in all languages (Goddard and Wierzbicka 1994), (iii) that conceptual primes have an inherent "conceptual syntax", in the sense that they can be combined in certain specifiable ways which make sense; such combinations, furthermore, can be expected to have equivalents in all languages, (iv) that conceptual primes and their combinatorial syntax3 can be thought of a comprising a "minilanguage" in terms of which all other meanings (whether expressed by words, grammatical constructions or prosodies) can be explicated. Most NSM researchers do not regard the natural semantic metalanguage merely as a useful analytical tool but rather as a conceptual and psychological reality. Conceptual primes are supposed to be, literally, elements of comprehension, and semantic explications framed in primes are supposed to be, literally, models of the conceptual meanings encoded in and expressed by linguistic forms. The current inventory of primes can be listed (using English exponents) as in Table 1. There are, of course, many aspects of the metalanguage and its realization in the English language which cannot be displayed in a summary presentation. For example, many of the words listed are polysemous, but in each case only one, specifiable sense is being proposed as a conceptual prime. The organization of Table 1, i.e. the grouping and categories, is intended to reflect some of the functional and combinatorial relationships among primes.

Conceptual primes in early language development

195

Table 1. Proposed semantic primes (after Goddard and Wierzbicka in press)

Substantives and substantive relations: Determiners and quantifiers:

I, YOU, SOMEONE/PERSON, SOMETHING/ THING, PEOPLE, BODY, PART OF, KIND OF THIS, THE SAME, OTHER, ONE, TWO, ALL, MUCH/MANY, SOME

Descriptors and evaluators: Intensifier: Mental predicates: Speech: Actions and events: Existence and possession: Life: Augmenter: Logical concepts: Time:

GOOD, BAD, BIG, SMALL VERY WANT, FEEL, THINK, KNOW, SEE, HEAR SAY, WORD, TRUE DO, HAPPEN, MOVE, TOUCH THERE IS, HAVE LIVE, DIE MORE NOT, MAYBE, IF, CAN, BECAUSE WHEN/TIME, NOW, BEFORE, AFTER, A SHORT TIME, A LONG TIME, FOR SOME TIME

Space:

WHERE/PLACE, HERE, BELOW, ABOVE,

Similarity:

LIKE

INSIDE, SIDE, FAR, NEAR

The claim of NSM researchers is that the primes indicated in Table 1 are necessaiy and sufficient to explicate all the meaning content expressible through the words, grammar and intonation patterns of the adult language. To what extent the same can be claimed for the language of young children is an open question, and one of the main concerns of this paper. A point I would like to emphasize, however, is that to the extent that the proposed primes (or a subset of them) turn out to be necessaiy and sufficient for modeling early child meanings, they will escape a potent criticism which has been leveled at other approaches to the semantics of child language. This criticism was first voiced by Susan Carey (1982). She pointed out that the majority of semantic descriptors used in child language research are highly abstract and theory-laden. Terms such as 'recurrence', 'agentive', 'dative', 'factitive', 'instrumental', 'regulatory', 'implication',

196

CliffGoddard

'dimension', 'polarity', and so on, come to mind. According to Carey, such "elegant and abstract" terms represent "a sophisticated schematization of knowledge by linguists". As such, they may be useful for certain purposes, but they are not plausible candidates for what Carey calls "developmental primitives", in the sense of "innate ... or at least very early-acquired concepts, out of which all other concepts are built" (p. 351). This criticism cannot be leveled at the NSM system, or at least, not with anything like the same force, because the NSM primes are not "abstract terms from theories which the child has not yet encountered". On the contrary, they are plain words and expressions of ordinary language, which the child is hearing every day, and which, in many cases, exist in the child's own active vocabulary. If we wish to fashion a system of semantic-conceptual representation for the child, then the NSM system is, at the least, a very good place to start.

1.1. Diary study of Pete Many of the examples discussed in this paper are drawn from diary notes of the language development of my son Pete (a pseudonym), who was born in August 1996. The notes were made by myself and by my wife, Mee Wun Lee, commencing (in earnest) from the time just before the child's second birthday. Our procedure was not as rigorous or systematic as some other diary studies, but it did yield a continuous sampling of the child's utterances over an 18-month period. I was the main record-taker. I tried to make a point, whenever I was with the child, of having some notepaper and a pencil with me, and I simply jotted down any utterance which seemed either typical of the way the child was speaking at that time, or interesting in the sense of showing "emergent" meanings or structures. Often some notes on the context or apparent intention of the child were also necessary. At different times, observations were made all over the house - at the breakfast table, in the bath, in the living room, in Pete's bedroom, and in the backyard, as well as in the car, at friends' houses, on shopping trips, at playgrounds, and so on. This is, admittedly, a rather

Conceptual primes in early language development

197

haphazard and intuitive technique but it is not different from the procedure followed by many linguists in taking field notes of an indigenous language in a naturalistic setting. Diary studies are most useful when they selectively focus on specific issues of interest to the investigators (Mervis et al. 1992). My observations were guided firstly by an interest in semantic primes, and secondly by an interest in general syntactic development. The fact that one is taking a day-by-day record tends to make one sensitive to new developments and thus more likely to record them; but on the other hand there were many hours each day when the child was not being observed, so the diary records cannot be taken to indicate the earliest occurrences of any word or structure. In most cases they probably pick up features which have been present for an unknown previous time. (We did not record utterances which were simply repetitions, in whole or part, of something which had just been said by an adult.) My wife, who is a native speaker of Cantonese, often spoke to the child in Cantonese when they were alone together; and especially in the early days I also used to address the child using my own rudimentary command of this language. Not surprisingly, a good proportion of Pete's earliest words were Cantonese, but for the purposes of this paper I have usually given English equivalents. Unlike several children who feature prominently in the child language literature, Pete is not an "early talker" but his general language development seems fairly typical of a child growing up in a middleclass Western household. From an examination of the corpus I have divided his early language development into the following rough stages. Prelinguistic stage Stage I (from about 14 months): single words Stage //(about 21 to early-26): two word combinations Stage III (early-26 to mid-29): some multi-word sentences, usually limited to a simple clause with an adjunct; also, only one utterance at a time. Stage IV (mid-29 to early-32): two or more related sentences in a row; begins to have conversational exchanges.

198

CliffGoddard

Stage V (early-32 to 36 months): onset of complex sentences (wantcomplements, when-clauses, clausal linking with and), early morphology (-ing, -s, -er) and grammatical verbs (e.g. inchoative ¿^-construction, have to, make-causative). Stage VI (36 to 41 months): past tense -ed, if-clauses, clitic copula -s. The present paper focuses primarily on the very early stages, i.e. the Prelinguistic stage, Stage I (one-word stage) and Stage II (two-word stage).

1.2. Semantic primes in the production vocabulary vs. the "conceptual vocabulary" It is useful at the onset to summarize the emergence, in Pete's production vocabulary, of lexical exponents of the proposed conceptual primes. This is done in Table 2, in which each column represents one of the Stages just described. Within the columns, the order does not represent chronological sequence but rather the groupings used earlier in Table 1. A couple of items (PART and MOVE) appear earlier than I would have expected from the research literature (cf. also Tien 1999: 115-141), but otherwise there is little in this sequence which will surprise child language researchers. The onset times given in Table 2 assume that certain primes are first expounded by words which are different to those of the adult language (cf. Wierzbicka 1995). For example, SMALL first appears as little, A SHORT TIME as a little while, BEFORE as first, AFTER as later, THE SAME as too, and WORD as called. It is not possible here to discuss and justify these decisions adequately, but in any case the adult exponents of these primes all appear by Pete's third birthday. The main general observation is simply that most of the proposed primes are clearly present by the child's third birthday; the remaining ones follow within a further six months or so (Goddard Forthc. a). Of course, one could certainly not expect all children to follow Pete's sequence, given the sizable individual differences in early lexical development even among English-speaking children (cf. Bates et al.

Conceptual primes in early language development

199

1988). Aside from personality differences, factors influencing the production sequence are individual preferences in "lexical selection" on phonological grounds (especially in the early stages) and the nature of the language input, e.g. Schwartz and Leonard (1982), Ninio (1992). Looking cross-culturally, differences in interactional style with infants is another potent factor (cf. Schieffelin and Ochs 1986). Table 2.

Stage I

Acquisition sequence of lexical exponents of primes in Pete's production vocabulary, up to 41 months Stage II

Stage III

I PEOPLE

THING/WHAT

PART

THE SAME(TOO)

Stage V

Stage VI

YOU

PERSON THIS

Stage IV

BODY KIND

OTHER

MUCH/MANY

ONE

TWO

ALL

SOME

GOOD BAD BIG SMALL(LITTLE) VERY(SO) WANT SEE

MOVE DO

HAPPEN

NOT(NO)

KNOW

SAY

HEAR

TOUCH HAVE

MORE

WORD(CALLED)

THINK THERE IS

CAN

FEEL DIE

MAYBE

IF

BECAUSE(COS) NOW BEFORE(FIRST)

AFTER(LATER)

A SHORT TIME

A LONG TIME

(LITTLE WHILE) WHERE/PLACE HERE

NEAR(NEXT TO) SIDE

INSIDE(LN)

FAR

ABOVE(UP) BELOW(DOWN) LIKE

I hasten to note that the presence of a lexical exponent does not mean that a prime is "fully acquired", in the adult sense. The child may have active command over only a small part of the prime's syntactic possibilities (as they exist in the adult language), with the result that

200 Cliff Goddard

its range of use may be highly restricted in comparison with adult usage. For example, though the word DO appears early in the third year it is a long time before the child is able to use it with the full range of complement and valency options, e.g. DO SOMETHING ΤΟ SOMETHING, DO SOMETHING WITH SOMETHING, DO SOMETHING TO SOMETHING WITH SOMETHING. Nevertheless, the presence of the lexical item, used with apparently the same meaning as its adult counterpart, is concrete linguistic evidence that the corresponding concept has some foothold, so to speak, in the child's mind. With this as background, the key issue to be addressed in this study can now be approached. With a small number of notable exceptions, most of the child's early words are not exponents of conceptual primes - for example, Stage I words such as mama, papa, bath, bird, duck, nana (banana), ball, and oh-oh, Stage II words such as mouth, hand, wheel, door, o f f , broke, made, noise, and many others. Dozens of early words appear whose semantic structure cannot be explicated in terms of the child's inventory of lexicalized primes. What can the NSM theory say about this situation? One response would be to observe that most of the child's early words must have simpler meanings than the corresponding words in an adult's vocabulary. This may be true enough, but it still seems impossible that dozens of Stage I words (for example) could be explicated in terms of a small handful of conceptual primes. The only possibility is that the child is making use of a larger set of primes, including some which are present conceptually but which still lack lexical exponents.4 As Bruner (1990) put it, there must be meanings which are present "in protolinguistic form" prior to their emergence in language proper. From a theoretical point of view, this deduction is hardly surprising. It is well-known that children's comprehension runs well in advance of their production vocabulary. Synthesizing results from various studies, Ingram (1989: 140-143) suggests that, as a norm, about 100 words are understood (in some fashion) even before the first recognizable words are produced. In Pete's case, a parental "selfinterview" identified about 60 words which we believed the child could understand (in some fashion) at 15 months. These included the following (or their Cantonese equivalents): dog, cannot, drink, eat,

Conceptual primes in early language development

201

water, go out, hurry up, milk, hand, foot, fly, ball, pick up, teddy, wet, hot, don't want, see, come, nappy, light, this, and many others. From the NSM point of view, if something is understood (in a linguistic sense) then it is necessarily understood in terms of some conceptual primes. In short, the child must have a certain "conceptual vocabulary" of prime concepts even before the onset of intelligible words. Theoretically, this is all well and good, but from a methodological point of view it is problematical. How can we identify conceptual primes which may be active in the child's mind, in the absence of tangible surface exponents? As far as I can see, there are two possible sources of linguistic evidence: (i) semantic analysis of the child's production vocabulary, and (ii) semantic analysis of the child's comprehension capabilities. The second option presents even more methodological difficulties than the first, and I have nothing to say about it in this paper. What I will try to do is to undertake semantic analysis of words in the child's non-prime production vocabulary, for if we can determine the meaning of these words this would furnish direct evidence of the conceptual vocabulary of the child at that age. In this I am taking a lead from Tien (1999), who, as far as I know, was the first to argue that conceptual primes may be "latent" in a child's early lexicon, in the sense of being hidden or implicit in the meanings of other, non-prime words. The task may seem like a daunting one, but it is not altogether different from that which faces a field linguist who undertakes semantic analysis of an unknown adult language from an unfamiliar culture. In either case we have to begin with close naturalistic observation of usage: documenting the range of contexts in which a certain expression is used - and not used - and comparing the usage of alternative expressions which can be found in different contexts. Then we experiment to discover the most economical semantic explications which match the attested range of usage. Obviously one must guard against the assumption that the child's meaning for a particular form corresponds to the adult meaning, i.e. against "adultocentrism", the child language analogue of ethnocentrism. In this paper I deal only with a fairly small number of early child words and utterances, and the fine details of the analyses in many

202

CliffGoddard

cases remain open to question. Even so, the exercise supports some highly specific proposals about which primes emerge first at a conceptual level and about the time lag between conceptual acquisition and lexicalization.

2. Semantic primes in early "conceptual vocabulary"

2.1. Explicating "proto-linguistic " meanings By general agreement, a child's first recognizable words are not his or her first attempts at intentional communication. The first words come after a lengthy period of prelinguistic (or proto-linguistic) expression by means of (consistent) signals composed of sounds, prosodies, gestures and gazing. For example, according to Halliday (1975), his son Nigel was producing meaningful proto-linguistic signals (sound-meaning pairings) from as early as 9 months, with this system of signals expanding in size and function for about six months prior to the child's first truly language-like utterances. At 13 months, for example, Nigel used a form ?nä Pnä ?nä ?nä to indicate something like Ί want that', and another form yi yi yi yi yi (high level tone) to respond when asked if he wanted a particular object to be given to him, meaning something like 'Yes. I want that' (Halliday 1975: 24). Halliday emphasizes that expressions like these are not "word-like" in the adult sense, but are direct form-meaning pairings; and also that they are for the most part not derived from forms of the adult language. Nevertheless they are intentional and systematic, they represent a child's long experience with his or her own system of expression, and they are fundamental in readying the child for language development in the adult sense. Similar conclusions have been reached by Carter (1978, 1979), Dore (1975), Dore et al. (1976), and Scollon (1976,1979). Regrettably I did not attempt to document and analyze Pete's proto-linguistic system. Even so, I think it will be useful to articulate some hypotheses about the kinds of meaning he was expressing in

Conceptual primes in early language development

203

this period. For this purpose, I have studied Halliday's account of Nigel and attempted to interpret some of what he says into NSM terms - with apologies to Halliday. The following ideas are ordered roughly according to Nigel's sequence of development and my confidence in them (the two orders happening to coincide). First, the primes WANT, SEE, DO, and THIS are involved in very early messages like the following, which Halliday assigns to the instrumental, regulatory, and interpersonal functions, respectively: WANT THIS DO THIS SEE THIS

Later, WANT and SEE form the basis of a "proto-mood" distinction in Nigel's speech, indicated by the contrast between high tone (i WANT) and falling tone (i SEE), respectively. This appears to correspond to the proposal by Bates et al. (1979: 115) that early child speech shows a distinction between proto-imperatives and proto-declaratives. Proto-imperatives - "the child's intentional use of the listener as an agent or tool in achieving some end" - are framed or introduced by the component ι WANT - . Proto-declaratives - "a preverbal effort to direct the adult's attention to some event or object in the world" - are framed or introduced by the component ι SEE - . In adult speech, of course, the declarative mood is associated, prototypically, with a more complex illocutionary component based not on SEE, but on KNOW, namely I WANT YOU TO KNOW SOMETHING. (Some researchers apply the proto-imperative vs. proto-declarative distinction also to the use of pointing by infant children, cf. Reddy 1999: 43.) Still at an early stage, the elements NOW and HERE are implied in some of Nigel's prelinguistic signals, in combinations such as the following: DO THIS NOW

differs from DO THIS by "an additional feature of urgency", Halliday's gloss 'Do that right now'

204

CliffGoddard

SOMEONE HERE NOW

a component of one of Nigel's "initiating expressions ... used typically when another person comes newly to his attention"

Before long, it seems that like: DO THIS MORE SEE MORE

MORE

enters the picture, in combinations

- Halliday's gloss 'do that (again)' - Halliday's gloss 'look, that's interesting'

Interestingly, the prelinguistic Nigel clearly seems to express the message DON'Τ-WANT. On an adult semantic analysis, this involves negation (NOT) in combination with WANT, but there is no direct evidence in Nigel's early vocalizations for negation as a separable component. Possibly DON'T-WANT is a single unanalyzable meaning for him at this stage. During the proto-linguistic stage Nigel also appears clearly to express some simple messages about his "feelings". From fairly early on, there are expressions glossed by Halliday as 'that's nice' and 'nice to see you' (pleasure). Towards the end of the period, Halliday lists two expressions which seem to express negative feelings; he glosses them as 'let's be sad' (shared regret) and 'I'm fed up' (complaint). In adult language, one would assign components such as I FEEL SOMETHING GOOD and Ι FEEL SOMETHING BAD but these seem too complex for the young Nigel, and there is no evidence for the presence of FEEL, GOOD, or BAD as separate items. I therefore suggest that in the early stages of development, there are two global and undifferentiated meanings which we can represent as: FEEL-GOOD and FEELBAD.

In summary, therefore, a consideration of proto-linguistic messages suggests that the following set of meanings may be present, in some incipient fashion, even prior to the first ordinary word: WANT, SEE, DO, THIS, SOMEONE, NOW, HERE, MORE, DON'T-WANT, FEELGOOD a n d FEEL-BAD.5

Conceptual primes in early language development

205

2.2. Explicating early Stage I words I will focus on the following small set of (non-prime) words from early in Pete's Stage I. There are indications (Dromi 1987, 1999) that the early and late periods of the one-word phrase may be qualitatively rather different, in terms of the child's word-learning strategies. mama, papa bath! nati 'hot/cold' (from Cantonese 'hot') oh-oh! bird! (Cantonese) duck! (Cantonese) broom-broom! 'car' What can we infer about the meanings of these words, when Pete was 18-20 months of age? Obviously we cannot attribute to the child anything like the semantic complexity of the comparable words in the adult lexicon. Nevertheless it is necessary to attribute some meanings to them, presumably the simplest conceivable meanings which are compatible with their range of use in his speech. The words mama and papa surely involve - minimally - the element SOMEONE(PERSON), presumably in combination with THIS. What else? One possibility would be to interpret mama and papa as, so to speak, proto-names; i.e. to attribute a semantic structure along the following lines (the use of inverted commas around mama in the explication is intended to refer to the sound of the word only). mama = this person this person is called "mama" (i.e. word for this person is "mama") papa = this person this person is called "papa" (i.e. word for this person is "papa")

206

CliffGoddard

According to these explications, the child's early words mama (and papa) have no descriptive content. They simply embody a recognition of a particular individual who is associated with a particular "label". Accepting this, however, means accepting that even very young children have some rudimentary notion of WORD. Some child language researchers would no doubt find this hard to swallow, for children at this stage; but others would not. Halliday (1975), for example, interprets the transition from the prelinguistic to the truly linguistic in terms precisely of the emergence of the "word" as a unit in the system: "The adult language is not a two-level system but a three-level system; it is composed not merely of meanings and sounds, but has another level of coding in between, one which, using folk-linguistic terminology, we may refer to as the level of wording' (p. 50, cf. also p. 34). A great deal has been written about the early onset of the so-called "naming insight", which some researchers see as explaining the vocabulary spurt which many children experience later in the one-word stage (cf. McShane 1979, Goldfield and Reznick 1990, Harris et al. 1988, Harris et al. 1995). I am inclined to agree with Harris et al. (1995), however, that the vocabulary spurt "might better be thought of as marking a transition from the view that some things have names to a view that everything has a name" [my emphasis]. It seems entirely plausible that Pete's earliest awareness of "names" or "labels", i.e. in NSM terms WORDS, would be in relation to those special "someones", his mama and papa.6 Various researchers have noted that very young children often use certain words in ways which defy the adult expectation of a distinction between "nouns" (words for things) and "verbs" (words for actions and events). Rather such words seem to stand for, or even to be part of, an "activity script" featuring the child him or herself in some daily routine such as going to bed, changing nappy, or having a bath. Early in Stage I, the word bath was like this for Pete. He would utter it upon seeing the bathtub filling up. The following explication presents bath! more as an exclamation than a noun or a verb; and like exclamations in adult speech (in this respect) it relies heavily on de-

Conceptual primes in early language development

ictic elements like THIS, HERE, and tional (WHERE) relationship.

NOW.

207

Also implied is the loca-

bath! = I see something something is happening here I (can) be in water (this 'stuff) now An important question which I will leave untouched is the status in the explication of the word water. I am prepared to take for granted that, experientially, water is directly recognizable to the child; but I am at a loss to know whether, and to what extent, the child has an articulated concept of water, deserving of its own "mini-explication". Pete's word nat derives from the Cantonese word for 'hot'. He often heard us say it to warn him about hot foods and drink, and also about the heater in our living room, but for a while Pete used his word nat not only about hot things, but also about very cold things mainly about ice cubes and stuff taken directly from the freezer. An explication along the following lines seems to be called for. Notice the apparent necessity for the term TOUCH - a recent, and as yet largely untested, candidate for the prime inventory. nat! ('hot/cold') = I don't want to touch this when I touch this, I feel something bad (or: I feel-bad) In his one-word period, Pete uttered the word oh-oh! when he saw that 'something bad' had just happened to something, typically that something had fallen over, been broken, dropped or spilled. Needless to say, this usage was heavily modeled for him by his parents, but Pete's oh-oh! embodied a simpler semantic structure than adult usage.7 oh-oh! = something bad happened to something now

208

CliffGoddard

Coming now to bird, duck, and car, we see what first appear to be clear examples of "nominal" terms, in the sense of words which are clearly and only used about things. In their very earliest uses, however, such words are used essentially to make an observation about immediate experience: the illocutionary frame of the one-word utterance is Ί see (Consistent with this, Halliday [1975] placed Nigel's early one-word utterances of this type under the "Personal" (expressive) function.) As parents too, we tend to use words like these to young children precisely to point out things: 'Look baby (a) bird!'. Clearly these words can have only a very simple structure compared to the enormous complexity of the adult words (cf. Wierzbicka 1985, 1996). Significantly, all three words refer to things that move. For bird and duck, I suggest a further salient feature is, so to speak, the locus of movement. Birds move up in the air; ducks move in the water. As for broom-broom 'car', I suggest its salient feature is that there is a person inside it. bird! = I see something it can move up-high this kind of thing is called "bird" (i.e. word for this thing is "bird") duck! = I see something it can move in water this kind of thing is called "duck" (i.e. word for this thing is "duck") broom-broom! 'car' = I see something it can move someone can be inside it this kind of thing is called "broom-broom" (i.e. word for this thing is "broom-broom")

Conceptual primes in early language development

209

Notice that these explications require the element KIND OF, even though the word kind does not "surface" till Stage V. Otherwise, they would be depicting the words bird, duck, and broom-broom as proper nouns, rather than as designations for recognizable classes of things. This is consistent with mounting psycholinguistic evidence for genuine categorization, i.e. categorization by kinds rather than simply by perceptual similarity, in very young children (Gopnik and Meltzoff 1997: Chapter 6, cf. Markman 1989, Keil 1989, Mervis 1987). Even from this small sample, then, we can see the implied presence of a dozen-and-a-half primes which do not surface as words in their own right for some months to come: I, SOMEONE(PERSON), SOMETHING(THING), KIND, THIS, HERE, NOW, SEE, WANT, DON'TWANT, BAD, HAPPEN, MOVE, TOUCH, CAN, ABOVE(UP), INSIDE, and

this list we can add DO, MORE, and FEEL-GOOD, which were indicated already in the proto-linguistic period (even though they have not turned up in the handful of Stage I words we have just looked at). At the one-word stage, it seems, Pete has a conceptual vocabulary of at least 20 semantic primes, about one-third of the eventual adult inventory. WORD(CALLED). TO

2.3. Explicating Stage II words In the three-month period between about 26.07 and 29.15 Pete learnt to say a large number of new words, and increasingly to produce twoword combinations (though he also continued to use a lot of singleword utterances). As set out in Table 2, by the end of Stage II the child had about 21 exponents of primes in his production vocabulary. This set is not sufficient, however, to plausibly explicate the wowprime vocabulary of Stage II. Rather, semantic analysis of his nonprime vocabulary suggests that Pete's conceptual vocabulary at this period already included most of the primes posited to appear in Stage III, plus several others which would not appear till subsequent stages. First, it is worth pointing out a few fairly simple "semantic molecules" (cf. Wierzbicka 1995):

210 Cliff Goddard

dark (here) = a person can't see here (this is) stuck = this thing can't move here (it 's) all gone = it was here before, it isn't here anymore A noticeable fact about Stage II is the appearance of sets of related nominal words (cf. Clark 1993, 1995), such as clothing words, bodypart terms, words for vehicles and for animals, and words for things to eat and drink. Clothing words were among Pete's first nouns, the first of them (pants, shirt) appearing before any of the body-part terms. Actually the words in question were Cantonese "baby talk" words. Fu-fu means roughly the same as 'pants', and saam takes in shirts and t-shirts. The distinction is between the top part of the body (saam) and the bottom part of the body (fu-fu). As far as I can see, to get plausible "child-level" explications for these and other clothing words we need not only to refer to certain parts of the body, but also to the idea that people 'do something with' these items and end up, as a result, with parts of their body inside them. For example: fu-fu (pants) = something a person does something with it afterwards the bottom part of a person is inside it saam (top) = something a person does something with it afterwards the top part of a person is inside it The terms 'top part' and 'bottom part' are obviously based on the primes ABOVE(UP) and BELOW(DOWN). The simplest thing would be to regard 'top' and 'bottom' as simply "adjectival" variants of ABOVE (UP) and BELOW(DOWN), respectively. The need for the prime PART is surely clear from the proliferation of body-part terms, such as mouth, eyes, nose, ear, head, and foot. However, I do not think we have to posit the prime BODY at this stage; it seems enough to explicate hands, for example, as 'parts of a

Conceptual primes in early language development 211

person' rather than as 'parts of a person's body'. Notice that many of the body-part terms {eyes, ears, hands, feet) are explicated here "in the dual", so to speak. This reflects the view, which can be traced back to von Humboldt (cf. Plank 1989), that the gestalt of a "pair" is impressed even upon an infant child from the experience of having, and seeing, dual body-parts. In the case of the eyes, in particular, their "two-ness" literally stares one in the face. eyes two things they are part of a person because of these things, a person can see

mouth = part of a person things can 'go' inside a person there a person can do things (to things) with this part

hands = two things they are parts of a person a person can do many things with them

head = one thing it is part of a person it is above the other parts

feet/legs (kiok) = two things they are parts of a person they are long they are parts of the bottom of a person a person can do some things with them

ears = two things they are parts of a person one is on one side of the head one is on the other side of the head

Another couple of Pete's early nominal words are wheel and door. The former is really a favorite word for Pete at age two, and perhaps would qualify as the prototypical part-term in relation to physical objects. An interest point about both words (especially clear in the case of door) is that they seem to require the semantic component SIDE. 8

wheel = part of something it is round (i.e. when you see it, it is the same on all sides) it moves when it moves, the other thing moves

door = something it is in a place someone can be on one side of it after this, it moves after this someone can be on the other side of it

212

CliffGoddard

Like many other children, in his two-word stage Pete began to use a largish number of animal names, such as horsie, dog, pig, cow, seal, cat, and monkey. To analyze these meanings in detail would be a fascinating project, which I believe would show that, even at this early stage, such words can involve semantic components describing the animal's size, some salient body-part features, characteristic sound, reference to habitat and (at least in some cases) reference to its typical food. This kind of study is beyond the scope of the present paper, however, and for present purposes I mention these words only to make the point that they further attest to the conceptual presence of the notion of KINDS. I will move instead to some explications for a set of "verbal" words. As with the nomináis, this listing is not exhaustive but it is a broad enough sample to indicate the range of semantic components which are needed. Perhaps the expression come here! is not entitled to be termed fully "verbal", since it is, at this stage, essentially an imperative formula; and the same applies to help me! Even simple structures such as the following make it plain that the prime YOU is called for. Indeed, one could claim that any example of genuinely "addresseedirected" speech implies YOU - for YOU is what an "addressee" is.9 come here! = I want you to do something you are not here I want you to be here

help me! = I want you to do something I want to do something (to this) I can't do it

The expressions fall down and bump both seem to imply the element HAPPEN. At age two, Pete used fall down not only about himself (after he fell), but also about something he threw or dropped. Bump not only seems to imply HAPPEN, but also TOUCH (without TOUCH, it is hard to see how the "contact" aspect of the event could be captured). The explications also seem more plausible with an explicit causal component BECAUSE OF THIS (as shown).

Conceptual primes in early language development

213

fall down = something happened to this thing/person because of this, it is down now bump (head) = something happened to me part of me (my head) touched something now because of this, I feel something bad now In connection with TOUCH, it is interesting to look at the word off one of the most common "verbs" in Pete's Stage II speech. He would say things like off shoe (after taking his shoes off), o f f f u or off socks (to describe what was happening, or to ask for it to be done). At this time, we used to have him sleep in a "baby sleeping-bag", rather than covering him with blankets in the cold Annidale winter. In the mornings, he would say off bag. He would also use the word off about his nappy, or (in the two-word mode) say off nappy. Similar verb-like uses of ochave been reported by Tomasello (1987), among others. The following explication for off has a kind of causative structure. off ('take off/remove/get out') = this thing was touching me somewhere before after this someone did something because of this, it's not touching me like this any more The word broke is an interesting one. It was first used in situations where one part of something, e.g. a toy, fell off, or when a Lego construction came apart. On one occasion he used it after the cap had fallen off a marker pen. The explication not only employs HAPPEN but also the element ONE, which is not strongly in evidence elsewhere at Stage II. broke = something happened to it because of this, it is not one thing any more

214

CliffGoddard

Give and make are an interesting pair. They both seem to demand an explicit "before-and-after" scenario: the situation as it was before is changed as a result of someone doing something. In the case of give, the change concerns possession (i.e. someone having something); in the case of make the change concerns the existence of something.10 (I) gave it (to someone) = this person didn't have it before after this I did something with it because of this, this someone else has it now I make house = there wasn't a house here before after this I did something with some things because of this, there is a house here now Verbs of physical manipulation, such as put, seem to call for a similar overall structure. The verb eat involves "doing" and a resultative component (including the component INSIDE). (someone) put (it) here = it wasn't here before after this someone did something with it because of this, it is here now (I) ate it = something was in my mouth I did something to it with my mouth because of this, it is inside me now Finally in this quick sweep through some of Pete's Stage II words, some evidence for the conceptual presence of HEAR (cf. Tien 1999: 104-111). First, Pete used the word noise in month 25, for example, to remark on the noise of water draining away in the bath. Similarly, he used the words cry and sing when not much older than two years, e.g. to comment on another child crying. The meanings of these

Conceptual primes in early language development

215

words surely call for HEAR. Second, this was a time when Pete was already quite interested in the characteristic sounds made by animals and birds: "quack-quack" for ducks, "tweet-tweet" for birds, "moomoo" for cow, "oink-oink" for pigs, and so on. He could produce any of these sounds at around two years of age. Surely this implies something like (for example): "quack-quack" = ducks do something when they do it, a person can hear something like this: "quackquack" There are also two primes, specifically, KNOW and SAY, whose presence is implied by functional (illocutionary) facts. They do not appear in Pete's production lexicon for three or four months, but their conceptual presence is implied by the fact that the child begins, in Stage II, to ask simple information-seeking questions (both polar and wh-), as shown below, and also to respond appropriately to questions asked by adults. The illocutionary intention of questions, conveyed by intonation, involves the components: Ι WANT TO KNOW SOMETHING, I WANT YOU ΤΟ SAY SOMETHING. (Interestingly, all the recorded examples of Pete's early questions concern location. I am not sure what to make of this: perhaps that is just the main kind of thing he wanted to know about.) 25.26 26.12 26.17 26.18 28.06

Moon there? Roo there? (re. a photo, i.e. is that the moon there? is that a kangaroo there?) (CG told Pete they were going to the coast tonight, to the sea) Pete: Seal there? CG: No, no seals. Pete: Boat there? Mama slippers where? CG: Pete, we're going to Jaew tse-tse's place for dinner. Pete: Baby there? Where plane?

In summary, from the sample of explications offered here one can deduce that the following set of elements are present as conceptual

216

CliffGoddard

primes at Stage II, despite the fact that they have not yet surfaced as individual words. They are presented in three groups: those which are destined to appear as words in the next stage (Stage III), i.e. within a couple of months; and those which do not appear until subsequent stages. To appear in Stage III:

PEOPLE, PART, OTHER, TWO, HAPPEN, TOUCH, HAVE, CAN, NOW, AFTER(LATER), LIKE

To appear in Stage IV: To appear in Stage V:

YOU, SAY, WORD(CALLED), THERE IS, SIDE ONE, KNOW, HEAR, BECAUSE

In the case of the Stage III group, what we are seeing is virtually the entire list of primes which appear at Stage III (with two exceptions: NEAR(NEXT TO) and FAR). This hardly seems like a surprising result any more. We saw the same pattern in relation to Stage I; and, as mentioned earlier, it is a pattern which is consistent with the fact that comprehension competence runs several months ahead of production competence. The conceptual presence of primes which only surface lexically four, five, or six months afterwards, however, does seem to call for some special explanation (see below).

3. Discussion The findings of this study of Pete's early semantic and conceptual development can be summarized as follows. All but a handful of the proposed NSM semantic primes have lexical exponents by 41 months, i.e. by age three-and-a-half (cf. Goddard Forthc. a). Semantic analysis of Pete's «on-prime words and utterances shows, however, that many conceptual primes are present well prior to acquiring surface lexical exponents. The general pattern is that the production vocabulary of primes lags several months behind the conceptual vocabulary. About 10 primes seem to be present in the conceptual vocabulary even in the proto-linguistic period, prior to the first recognizable word as such. They are: SOMEONE, THIS, WANT, DON'T-WANT, SEE, DO, MORE, NOW, HERE, and possibly a pair of "proto-primes"

Conceptual primes in early language development

217

and FEEL-BAD. By the end of the one-word stage, Pete's conceptual vocabulary had doubled to 20 or so primes, about onethird of the adult inventory, even though at this time only a small handful of primes had lexical exponents of their own. By the end of the two-word stage, Pete's conceptual vocabulary had again doubled, to account for over two-thirds of the adult inventory. Perhaps the most important question raised by this study is to what extent Pete's acquisition sequence is typical - in terms of production vocabulary, in terms of conceptual vocabulary, and in terms of the relationship between them. As mentioned earlier, one might expect there to be significant differences in the production sequence across children and across languages, which it would be interesting to map out and try to explain. 1 In relation to the acquisition sequence of primes in the conceptual vocabulary, I would like to advance the hypothesis that this is much more stable and consistent across children and across languages. These issues, it seems to me, have the potential to sustain a wide-ranging research program. A second research area concerns a range of specific questions about the developmental sequence of conceptual primes. For example: What explains which primes appear early and which appear late? Why don't more primes appear earlier in production, during the oneword and two-word periods? Why do some primes take longer than others to "bridge the gap" between their first appearance in the conceptual vocabulary and their eventual appearance in production? What explains the late lexicalization of certain primes? And so on. Developmental, functional and syntactic factors may all play a role here. For example, MORE, NO, and THIS may occur so early because they can be used as single words and still convey a functionally useful message, KNOW, HEAR, and BECAUSE may appear only relatively late in production because to manipulate these terms in production requires command over complement and adjunct structures, which may present too much of a processing problem for the child. A third area for further research concerns syntactic aspects of language development. Conceptual primes have valency, complementation and combinatorial properties (cf. Goddard and Wierzbicka in press) which, it can be argued, are the basis of syntactic structure at FEEL-GOOD

218 Cliff Goddard

large in the adult language. As mentioned earlier, however, the child's initial acquisition of a lexical prime does not normally constitute "full" acquisition because initially the child may command only a small part of the prime's syntactic possibilities. Thus, one could study the syntactic development of primes, both in production and in the conceptual vocabulary, and the relationship between this and general syntactic development. It should be clear that paraphrase in terms of the NSM semantic primes can provide a practical, yet highly constrained, method for formulating testable hypotheses about early child meanings, i.e. a serviceable method for actually "doing" child semantics. Furthermore, to echo the concluding words of Wierzbicka (1995), the approach generates new and interesting research questions about language acquisition. In the Introduction I mentioned that there have been surprisingly few studies which propose a substantial body of semantic analyses of "child meanings". One of the notable exceptions to this generalization is Michael Tomasello's (1992) book First Verbs. In his opening chapters, Tomasello (1992: 33) makes the point repeatedly that any such project requires "some form of cognitive-developmental theory to help us reconstruct the child's point of view from our point of view as we examine the contexts in which she used a particular word". His own study began with the assumption that the child "lived in a sensory-motor world of objects and their spatial, temporal, and causal relations. Also, she may be assumed to know a good deal about other persons and how they work ..." (p. 36). As a mode of representation he adopted image-schematic diagrams, somewhat in the style of Langacker (1987), and explicitly linked his empirical study with one of the main currents in contemporary cognitive linguistics.12 Needless to say, the present study can also be seen as an exercise in cognitive linguistics. There may be questions raised, however, about whether the NSM framework does allow us to "reconstruct the child's point of view", in the sense of safeguarding us against adultocentrism. Doesn't the framework start, in effect, from adult language primes and impose them on children? How do we know that young children don't have

Conceptual primes in early language development

219

their own conceptual primes which are different from, or even incommensurable with, the adult system? I will conclude by saying something about these theoretical questions. One cannot a priori rule out the possibility that the young child's semantic system is incommensurable with the adult system. On the other hand, practical experience shows that there can be a high degree of mutual understanding between young children and adults. A radical incommensurability thesis in relation to young children would also face the problem of accounting for the developmental continuity of child and adult understanding. In my view, the issue must remain open for the time being. One thing is certain, however. We will never get anywhere at resolving the issue unless we are willing to knuckle down and attempt serious semantic analyses of early child language. If we try, and fail, using a representational system which assumes some limited conceptual continuity with the adult system, then this will be evidence for incommensurability - and vice versa if we succeed. Above all, we have to try; and I submit that the NSM system offers a highly facilitative framework for doing just that. On the second point, it is not really correct to say that the NSM system uses adult primes to explicate child language speech (thus imposing an adult point of view). From the present study it would seem that the meanings of a child's early words can be adequately paraphrased in terms of the language of the same child - as it will be a few months down the track. That is, even when the explications cannot be framed entirely within the child's own production vocabulary (which may be the case until the child is, say, four years of age), they can at least be framed in terms which will soon be part of that child's production vocabulary. Another way of putting it is that although the language of toddlers and young children is not yet "metasemantically adequate" (as, apparently, all adult languages are) it does not stay that way for long. There is every indication that the entire NSM metalexicon exists in the production competence of a four-year old child. If so, the continuity/incommensurability issue should not be pitched in terms of child vs. adult, but in terms of twoyear old vs. four-year old. In this respect, the NSM primes retain a significant advantage over other representational systems such as

220

CliffGoddard

complex, technical descriptors ('recurrence', 'instrumental', 'joint attention', and the like) and complex, technical diagrams, which truly are beyond the grasp of children at any age.

Acknowledgements Mee Wun Lee has made a substantive contribution to this paper, especially in relation to the semantic analyses of Pete's early words. Vicki Knox also made a number of helpful suggestions. I am grateful to Anna Wierzbicka, René Dirven and Nick Enfield, who read and made valuable comments on earlier drafts, to participants in the LAUD Symposium held at Landau, Germany, in March 2000, and to an anonymous reviewer.

Notes 1.A similar position was long held by Dan Slobin (1985), Melissa Bowerman (1985), and others, but in recent years they have begun to repudiate their earlier emphasis on an innate prelinguistic conceptual basis for language acquisition, in favor of an emphasis on the characteristics of the adult "input" language, especially its language-specific aspects (cf. Bowerman 1996, Slobin 1997). In my view there is no necessary conflict between these two positions, but it is not possible to pursue the matter here. 2. The NSM bibliography is extensive and cannot be reviewed here for reasons of space. Aside from works cited elsewhere in the chapter, representative works include: Ameka (1990), Chappell (1986), Goddard (1996, 1997), Harkins (1990,1996), Hasada (1994), Peeters (1993), Travis (1998), Wilkins (1986). 3. It is true that in the NSM model it can make sense to speak of "universal syntax" or "universal grammar" (cf. Goddard and Wierzbicka forthcoming). For some readers such expressions may carry a connotation of autonomous syntax, so it is perhaps worth stressing that what is intended is literally a "conceptual syntax". What we are trying to say is that certain combinations of primes necessarily make sense and should be expressible in all languages, e.g. 'something', 'do', and 'good' can combine to form 'do something good'; 'say', 'something' and 'someone' can combine to form 'say something to someone'. Importantly, the potential for these combinations is inherent in and springs from the mean-

Conceptual primes in early language development

221

ings themselves. Thus the combinatorial syntax of primes is not autonomous (from meaning) - quite the opposite in fact. 4. Why, it might be asked, could the young child not be working in terms of "conceptual gestalts", in the sense of meaning complexes which are grasped as whole configurations without any apprehension of their individual conceptual constituents? In principle I have nothing against this suggestion, especially in relation to very young children (see my comments at the end of Section 2.1). However, the following analysis indicates that the same conceptual components (e.g. WANT, SEE, DO, THIS, SOMEONE, SOMETHING) apparently recur in m a n y a n d

varied early child words; and furthermore, that such concepts generally acquire surface lexical exponents within a few subsequent months. To my mind, these facts are most easily explained on the assumption that the concepts in question exist in a discrete fashion, so to speak, in the young child's mind. 5. If children's proto-linguistic signals can be explicated, and turn out to express combinations of a certain small set of conceptual primes, perhaps the same approach can be used about intentional communication by non-human primates (cf. Cheney and Seyfarth 1990, Tomasello and Call 1997). For example, the three alarm calls used by vervet monkeys to indicate the presence of three different predators (eagles, snakes, leopards) could perhaps be explicated as messages that 'something bad can happen, because something is moving up-above (for eagles)/down-below (for snakes)/near here (for leopards)'. See Jones (1999) for a discussion of conceptual primes in an evolutionary, phylogenetic perspective. 6. For some early talkers it might be that mama functions in its veiy earliest uses merely as a "calling device". In Pete's case, however, he used the word to indicate recognition of his mother, as well as to call her; and as the discussion in §2.1 indicated, there are reasons to believe that the "proto-declarative" function (with the illocutionary frame I SEE) is already present in the prelinguistic stage. With this in mind, the only alternative explication for mama (and papa) would seem to be as shown below. This explication does without WORD, but at the cost of attributing to the child something like a self-conscious generalization, which seems less than plausible. mama = this person when I see/want this person, I say "mama" 7. Adult Oh-oh! can be used in a broader range of situations; for example, I could say Oh-oh! upon reaching into my pocket for my keys and realizing that they are not there. Aside from conveying the idea that something bad and unforeseen is imminent, there is also a component of "minimisation" (Goddard 1998a: 190).

222

CliffGoddard

8. The prime SIDE is intended to be "spatial-relational" in character. That is, it describes the positional relationship between two entities: 'X is on this (one, two, all, etc.) side(s) of Y \ In some languages (including English) the same word can also function as a nominal, but this nominal usage is semantically complex (roughly, 'one side of X = one part of X, this part is on one side of all the other parts of X') and non-universal. The proposed explication for wheel, therefore, is not intended to imply that a wheel has sides. 9. If a young child does not comprehend Τ and 'you' that child would be quite at a loss to understand a lot of conversation between adults. It seems likely that children do understand the shifting referents in overheard speech long before they are able to carry out the same manipulations in their own speech. OshimaTakane (1999) makes an argument that observing (and understanding) adults using pronouns with one another is highly facilitative of the child developing his or her own command over shifters. 10. Re. HAVE, it is interesting to note the importance of "offer" gestures, and of "giving" and "taking" generally, in toddler play, cf. Bronson (1981). It is also worth noting that "giving-and-taking" a small object is a routine which many adults play with a young, preverbal child. 11. Esther Dromi's (1987) daughter Keren was different in many ways to Pete. Not only was she acquiring Hebrew rather than English, she was an early and prolific talker. Her first word came at 10 months and by the onset of the two-word stage, late in month 17, she had acquired an impressive 337 different one-word utterances. Dromi (1987: 171-179) lists Keren's entire one-word lexicon and it is interesting to extract from this list the words which appear to be exponents of semantic primes: (in order of appearance) HERE, WHAT, THIS, MORE, UP, WHERE, NO, WANT, BIG, VERY, SMALL, GOOD, MOVE, SEE, YOU. Admittedly, some of these putative identifications may be mistaken, mainly because the English "glosses" relate to the adult meanings of the words, rather than necessarily reflecting the child's own usage. Nevertheless, there is a very substantial overlap between this list and Pete's early vocabulary of primes. 12. The diagrams in Tomasello (1992) are sometimes so complex that they cannot be understood without verbal explication, which makes it debatable whether they really succeed in detaching the representation from adult language (for discussion, see Goddard Forthc. b).

References Ameka, Felix 1990

The grammatical packaging of experiencers in Ewe: A study in the semantics of syntax. Australian Journal of Linguistics 10(2): 139-181.

Conceptual primes in early language development

223

Bates, Elizabeth, Inge Bretherton and Lynn Snyder 1988 From First Words to Grammar. Individual Differences and Dissociable Mechanisms. Cambridge: Cambridge University Press. Bates, Elizabeth, Luigia Camaioni and Virginia Volterra 1979 The acquisition of performatives prior to speech. In: Elinor Ochs and Bambi B. Schieffelin (eds.), Developmental Pragmatics, 111-130. New York: Academic Press. Bloom, Lois 1973 One Word at a Time. The Hague: Mouton. 1991 Language Development from Two to Three. Cambridge: Cambridge University Press. Braine, Martin D. S. 1976 Children's first word combinations. Monographs of the Society for Reseach in Child Development 44(1): 1-97. Bowerman, Melissa 1978 The acquisition of word meaning: An investigation into some current conflicts. In: N. Waterman and C. Snow (eds.), The Development of Communication, 263-287. New York: John Wiley. 1996 The origins of children's spatial semantic categories: Cognitive vs. linguistic determinants. In: John J. Gumperz and Stephen C. Levinson (eds.), Rethinking Linguistic Relativity, 145-176. Cambridge: Cambridge University Press. Branson, Wanda C. 1981 Toddlers' Behaviors with Agemates: Issues of Interaction, Cognition, and Affect. Norwood, NJ: Ablex. Bruner, Jerome 1990 Acts of Meaning. Cambridge, MA: Harvard University Press. Carey, Susan 1982 Semantic development: The state of the art. In: E. Wanner and L. R. Gleitman (eds.), Language Acquisition: The state of the art, 347-379. Cambridge: Cambridge University Press. Carter, Anne Lindsay 1978 The development of systematic vocalizations prior to words: A case study. In: N. Waterson and C. Snow (eds.), The Development of Communication, 127-138. London: John Wiley. 1979 The disappearance schema: Case study of a second-year communicative behavior. In: Elinor Ochs and Bambi B. Schieffelin (eds.), Developmental Pragmatics, 131-158. New York: Academic Press. Chappell, Hilary 1986 The passive of bodily effect in Chinese. Studies in Language 10(2): 271-296.

224 Cliff Goddard Cheney, Dorothy L. and Robert M. Seyfarth How Monkeys See the World: Inside the Mind of Another Spe1990 cies. Chicago: University of Chicago Press. Clark, Eve V. 1993 The Lexicon in Acquisition. Cambridge: Cambridge University Press. 1995 Later lexical development and word formation. In: Paul Fletcher and Brian MacWhinney (eds.), The Handbook of Child Language, 393-412. Oxford: Basil Blackwell. Dore, John Holophrases, speech acts and language universals. Journal of 1975 Child Language 2:21-40. Dore, John, Margery B. Franklin, Robert T. Miller and Andrya L. H. Ramer Transitional phenomena in early language acquisition. Journal of 1976 Child Language 3: 13-28. Dromi, Esther 1987 Early Lexical Development. Cambridge: Cambridge University Press. 1999 Early lexical development. In: Martyn Barrett (ed.), The Development of Language, 99-132. Hove, UK: Psychology Press. Goddard, Cliff The "social emotions" of Malay (Bahasa Melayu). Ethos 24(3): 1996 426-464. 1997 Contrastive semantics and cultural psychology: 'Surprise' in Malay and English. Culture & Psychology 3(2): 153-181. 1998a Semantic Analysis: A Practical Introduction. Oxford: Oxford University Press. 1998b Bad arguments against semantic primitives. Theoretical Linguistics 24(2/3): 129-156. Forthc. a Semantic and conceptual primes in early language development: A diary study. Forthc. b Verbal explication and the place of NSM semantics in cognitive linguistics. In: June Luchjenbroers (ed.), Cognitive Linguistics: Investigations across Languages, Fields, and Philosophical Boundaries. Amsterdam: John Benjamins. Goddard, Cliff and Anna Wierzbicka (eds.) 1994 Semantic and Lexical Universals - Theory and Empirical Findings. Amsterdam: John Benjamins. Forthcoming Meaning and Universal Grammar - Theory and Empirical Findings. Amsterdam: John Benjamins.

Conceptual primes in early language development 225 Goldfield, Beverly A. and J. Steven Reznick 1990 Early lexical acquisition: Rate, content, and the vocabulary spurt. Journal of Child Language 17: 171-183. Gopnik, Alison and Andrew N. Meltzoff 1997 Words, Thoughts and Theories. Cambridge, M.A.: MIT Press. Halliday, M.A.K. 1975 Learning How to Mean - Explorations in the Development of Language. London: Edward Arnold. Harkins, Jean 1996 Linguistic and cultural differences in concepts of shame. In: D. Parker, R. Dalziell and I. Wright (eds.), Shame and the Modern Self, 84-96. Melbourne: Australian Scholarly Publishing. Harris, Margaret, Martyn Barrett, David Jones, and Susan Brookes 1988 Linguistic input and early word meaning. Journal of Child Language 15: 77-94. Harris, Margaret, Caroline Yeeles, Joan Chasin, and Yvonne Oakley 1995 Symmetries and asymmetries in early lexical comprehension and production. Journal of Child Language 22: 1-18. Hasada, Rie 1996 Some aspects of Japanese cultural ethos embedded in nonverbal communicative behaviour. In: Fernando Poyatos (ed.), Nonverbal Communication in Translation, 83-103. Amsterdam: John Benjamins. Ingram, David 1989 First Language Acquisition. Cambridge: Cambridge University Press. Jones, Doug 1999 Evolutionary psychology. Annual Review of Anthropology 28: 553-575. Keil, Frank C. 1989 Concepts, Kinds, and Cognitive Development. Cambridge, MA: MIT Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar, Volume 1. Stanford, CA: Stanford University Press. Markman, Ellen M. 1989 Categorization and Naming in Children: Problems of Induction. Cambridge, MA: MIT Press. McShane, John 1979 The development of naming. Linguistics 17: 879-905.

226

CliffGoddard

Mervis, Carolyn B. 1987 Child-basic object categories and early lexical development. In: U. Niesser (ed.), Concepts and Conceptual Development: Ecological and intellectual factors in categorization, 201-233. Cambridge: Cambridge University Press. Mervis, Carolyn B., Cynthia A. Mervis, Kathy E. Johnson and Jacquelyn Bertrand 1992 Studying early lexical development: The value of the systematic diary method. In: Carolyn Rovee-Collier and Lewis P. Lipstitt (eds.), Advances in Infancy Research, Vol. 7, 292-378. Norwood, NJ: Ablex. Ninio, Anat 1992 The relation of children's single word utterances to single word utterances in the input. Journal of Child Language 19: 87-110. Oshima-Takane, Yuriko 1999 The learning of first and second person pronouns in English. In: R. Jackendoff, P. Bloom and K. Wynn (eds.), Language, Logic and Concepts, 373-410. Cambridge, M.A.: MIT Press. Peeters, Bert 1993 Commencer et se mettre à: une description axiologico-conceptuelle. Langue française 98: 24-47. Plank, Frans 1989 On Humboldt on the dual. In: R. Corrigan, F. Eckman and M. Noonan (eds.), Linguistic Categorization, 293-333. Amsterdam: John Benjamins. Reddy, Vasudevi 1999 Prelinguistic communication. In: Martyn Barrett (ed.), The Development of Language, 25-50. Hove, UK: Psychology Press. Schieffelin, Bambi B. and Elinor Ochs (eds.) 1986 Language Socialization across Cultures. Cambridge: Cambridge University Press. Schwartz, R. and L. Leonard 1982 Do children pick and choose: an examination of phonological selection and avoidance in early lexical acquisition. Journal of Child Language 9: 319-36. Scollon, Ronald 1976 Conversations with a One-Year-Old: A Case Study of the Developmental Foundation of Syntax. Honolulu: University Press of Hawaii. 1979 A real early stage: An unzippered condensation of a dissertation on child language. In: Elinor Ochs and Bambi B. Schieffelin (eds.), Developmental Pragmatics, 215-228. New York: Academic Press.

Conceptual primes in early language development Slobin, Dan I. 1985

1997

Tien, Adrian 1999

227

Crosslinguistic evidence for the language-making capacity. In: Dan I. Slobin (ed.), The Crosslinguistic Study of Language Acquisition, Vol. 2: Theoretical issues, 1157-1256. Hillsdale, New Jersey: Lawrence Erlbaum. The origins of grammaticizable notions: Beyond the individual mind. In: Dan I. Slobin (ed.), The Crosslinguistic Study of Language Acquisition, Vol. 5: Expanding the Contexts, 265-323. Mahwah, New Jersey: Lawrence Erlbaum.

Early lexical exponents and 'related' lexical items as manifestations of conceptual/semantic primitives in child language. MA Thesis. Australian National University. Tomasello, Michael 1987 Learning to use prepositions: A case study. Journal of Child Language 14: 79-98. 1992 First Verbs: A Case Study of Early Grammatical Development. Cambridge: Cambridge University Press. Tomasello, Michael and Josep Call 1997 Primate Cognition. Oxford: Oxford University Press. Travis, Catherine 1998 Omoiyari as a core Japanese value: Japanese-style empathy? In: Angeliki Athanasiadou and Elzbieta Tabakowska (eds.), Speaking of Emotions: Conceptualization and Expression, 55-82. Berlin: Mouton de Gruyter. Wierzbicka, Anna 1972 Semantic Primitives. Translated by A. Wierzbicka and J. Besemeres. Frankfurt/M.: Athenäum Verlag. 1980 Lingua Mentalis: The Semantics of Natural Language. Sydney: Academic Press. 1985 Lexicography and Conceptual Analysis. Ann Arbor: Karoma. 1995 Universal semantic primitives as a tool for the study of language acquisition. Unpublished MS. 1996 Semantics, Primes and Universals. Oxford: Oxford University Press. Wilkins, David 1986 Particles/clitics for criticism and complaint in Mparntwe Arrernte (Aranda). Journal of Pragmatics 10(5): 575-596.

No preposition required. The role of prepositions for the understanding of spatial relations in language acquisition Katharina J. Rohlfing

1. Introduction In cognitive semantics locative prepositions play an important role for the processing of spatial relations. By using a locative preposition, the speaker's understanding of a spatial relation is marked. One of the most striking issues for cognitive science is whether there are universal relations that are first acquired and then marked via a preposition associated to them. Some studies (Clark 1973, Grieve, Hoogenraad and Murray 1977, Sinha et al. 1999) from developmental psychology indicate that this may be the case. In English and German, an order of acquisition has been observed, whereby the relations IN and ON are acquired earlier than e.g. UNDER. It is probable that these relations IN and ON are universal and therefore acquired first. On the other hand Bowerman's (1996a) results indicate a high degree of language specificity for spatial relations acquisition. Choi and Bowerman (1991) conducted a study, in which they collected and compared data of children learning English and Korean. This study is especially important, because the languages differ significantly in how they classify spatial configurations: the English IN-relation is matched by two different Korean relations, depending on whether the relation is a tight fit (like a video cassette in its box) or whether the relation is a loose fit (like an apple in a bowl). This different spatial configuration originates from a different categorization: "English makes a distinction between actions resulting in containment (put in) versus support or surface attachment (put on), while Korean makes a cross-cutting distinction between tight-fit relations (kkita) versus

230 Katharina J. Rohlfing

loose-fit or other contact relations (various verbs). In particular, the Korean verb kkita refers to actions resulting in a tight fit relation regardless of containment or support" (Choi et al. 1999: 241). The results from the study (Choi and Bowerman 1991) showed that infants distinguish very early between different forms in Korean for the one English IN-relation and acquire them without giving preference to any universal form. The acquisition of prepositions is, according to Bowerman (1996a), language-specific and leaves little opportunity to bear out universal relations. These results suggest that children do not necessarily develop preverbal concepts, but can construct concepts on the basis of the linguistic input (Bowerman 1996b). A recent study by Choi, McDonough, Bowerman and Mandler (1999), however, shows that children's sensitivity to languagespecific spatial categories begins at the age of 18 to 23 months. But children acquire an extensive knowledge about space, objects and events already before the age of 18 months. Another possibility therefore is that "children do map spatial words directly to pre-established spatial concepts - it is just that this set of notions is larger and more varied than has typically been supposed; for example, in addition to 'containment' and 'support', children might also have a prelinguistic notion of 'tight fit'" (Choi et al. 1999: 264). It is therefore important to elaborate which role the language-specific spatial categories exactly play in the child's primary process of learning about space. The psycholinguistic study I have undertaken examines the role of prepositions for understanding instructions with spatial relations in language acquisition. It examines not only the extent to which prepositions are important for understanding spatial relations, but also whether there are other, non-linguistic strategies for understanding that make universale at a non-linguistic level conceivable after all. Thus, the aim of the study was to examine whether the infants explicitly respond to prepositions and receive specific information through them. As I will show, the non-linguistic strategies contain culturespecific aspects and the input of language could therefore be found there already. However, even if we assume that these strategies are like an equation for a behavior with one culture-specific variable,

No preposition required

231

they can nevertheless let us get an insight into the learning processes: not the language-specific content of the spatial knowledge is more focused, but the (probably) universal processes which include some cultural-dependent aspects.

2. Non-linguistic strategies My study touches on a discussion about non-linguistic strategies that are said to be responsible for understanding locative instructions. According to the discussion, infants' understanding is based on nonlinguistic strategies and incomplete semantic knowledge, although the incompleteness of semantic knowledge concedes that children have and use semantic knowledge at this age. It is, moreover, thought that non-linguistic strategies make infants much more sensitive to the context than to the meaning of linguistic utterances. It is not my intention to deny that non-linguistic strategies are culture-dependent, rather, my main point is that these strategies are grounded in the context, i.e. in the environment of the child. The main task for the children in the study was to perform a relation between two objects, i.e. to set the trajector to the landmark. Note that already at the beginning of the relational performance, cultural aspects are crucial: the canonical orientation (Sinha 1982), i.e. in which position the object is linked to an activity or how an object is held to achieve something, plays an important role for the child handling the objects. The canonical orientation of an object is established by canonical rules1. The canonical rules, however, are provided by human social interest and value (Sinha 1983: 276). Performing the task, the child determines moreover which object is trajector and which landmark, decides about the proper relation between these two objects, about the appropriate motion of the trajector to the landmark, and about the direction or the kind of the motion. The result of my study suggests that non-linguistic strategies help children in their decision. Eve Clark (1973) mentions non-linguistic strategies in her studies with on average two-year-old infants. She claims that infants' be-

232 Katharina J. Rohlfìng

havior is based on non-linguistic strategies and incomplete semantic knowledge. Concerning non-linguistic strategies, infants may be guided by the features of objects. The physical properties of an object provoke a certain type of behavior and are the reason why the accompanying instructions are not taken into account. E.g. if an object includes a surface, the infants' behavior is influenced insofar as they will always put something on this surface, without following the semantic meaning of the instruction correctly. Clark (1973: 168) observed the same type of influenced behavior when the landmarkobject1 (LM-object) is a container. In conclusion, Clark claimed two rules of non-linguistic knowledge based on physical properties of objects: Rule 1: If the LM-object is a container, the TR-object (trajectorobject2) is inside it Rule 2: If the LM-object has a horizontal surface, the TR-object is on it. Note, however, that there is already a culture-dependent aspect in this physical-property strategy. When infants see a new object (e.g. a bridge), they associate it with their knowledge of the orientational and functional properties of an object they already know well (e.g. a table). In my study, only one child had some problems in placing the bridge in an upright orientation and placed it upside-down, as a crib. Thomas Thiel (1985) criticized Clark's non-linguistic strategies. In his study, he focussed the "relational character of objects [relationaler Charakter eines Objektes]" (Thiel 1985: 192), i.e. the infants link a certain activity to a type of object in its canonical orientation (Sinha 1982). Thus, the relational character is culture-dependent and subjective because of the child's previous experiences, i.e. which activity the child has repeatedly observed with a certain object. The relational character consists of two parameters: active - passive and one or more spatial relations. If a child observes that objects are put on a plate, the plate will have a passive relational character involving ONrelations (as a result, the objects are put on it), because the plate will repeatedly serve as a landmark-object supporting others. The objects

No preposition required

233

that are put on the plate (food or cups) obtain an active relational character. Chris Sinha (1982, et al. 1999) determined the relational character of objects on the basis of cultural differences. He focuses on the canonical orientation of objects as artefacts in a given culture. If, for example, an object is used as a container, the canonical orientation will be as follows: the object is supported by a surface, its top is open and the infants are inclined to put things in this object. Therefore it is very unusual for infants to use this object upside-down as a supporting object and to put things on the bottom of it. The canonical rules help children to handle the objects, i.e. they know in which orientation to use the objects to achieve something (e.g. how to hold a pen to write). Observing this, children develop an extensive knowledge about objects and the activities associated with them and try to preserve it. For the task in my study it means that as soon as the child recognizes an object (the trajector and/or the landmark), corresponding background knowledge (Sinha 1983) is activated. This background knowledge about one object (e.g. the landmark) can dominate over the relational behavior in the sense that the object serves as a target and motivates the motion needed in order to reach this target in a relation, while it does not matter what the trajector is and if it suits this relation (e.g. the TR-object is too big for the LM-object). To summarize, the non-linguistic strategies discussed here have the claim in common that they more strongly determine infants' behavior than their lexical knowledge3 does. To examine how strong the influence of language is, the following study was designed.

3. Experiment The study focuses on the claim of non-linguistic strategies and is based on two types of instructions, which have been designed to examine whether the infants explicitly react to prepositions and receive specific information through them: one type consists in a syntactically correct request relating two objects to each other in view of a

234 Katharina J. Rohlfmg

certain spatial relation; the other also consists in a request, but in this case it is syntactically incorrect: the preposition is omitted. The choice of preposition was guided by the order of prepositional acquisition in German (e.g. Thiel 1985) and in English (e.g. Clark 1973, Sinha et al. 1999). According to the order, the preposition IN is acquired first, the preposition ON / AUF second. This can also be observed in Polish for the prepositions DO (IN) and NA (ON) (Rohlfing 1998: pilot study). In Polish, however, there are two prepositions equivalent to English ON: a dynamic and a static one. Both are verb-derived but the latter depends on the motion of the verb and describes the spatial movement, whereby the former stands for a fixed location or a state. The dynamic or static character of a preposition determines the inflection of the cases (see below); in the case of the IN-preposition there are also two different lexemes: DO for a dynamic IN-preposition and NA for a static one. In this study only the dynamic form NA has been used, because the static form involves more syntactic complexity in the instructions4. In the following it will be marked with "->" to distinguish it from the static form. The other motivation for the choice of this preposition is the feature of inflection in Polish. Spatial relations are created with the aid of a preposition and, depending on the object's gender, with inflection of the cases. An instruction which contains this relation is: (1)

Daj konika NA—> most Give horse-AKK ON bridge 'Put the horse on the bridge'

(2)

Daj piticç NA-> kulç Give Ball-AKK ON Sphere-AKK] 'Put the ball on the sphere'

A syntactically incomplete instruction is:

No preposition required

(3)

235

a. Daj konika _ most Give horse-AKK_ bridge] 'Put the horse the bridge'

One can argue, referring to Slobin's (1973: 191) organizations' principles, that the infants know which preposition is omitted because of the inflection. One needs to keep in mind, however, that syntactically it is also possible to complete the incorrect instruction with other prepositions: b. Daj Give c. Daj Give d. Daj Give e. Daj Give f. *Daj Give

konika horse-AKK konika horse-AKK konika horse-AKK konika horse-AKK konika horse-AKK

POD-+ most UNDER bridge NAD-> most OVER bridge ZA—> most BEHIND bridge PRZED-> most IN FRONT OF bridge W—> most INTO bridge

Therefore, there must be more than a syntactical indicator leading infants to a correct understanding. According to these multiple possible solutions to complete an instruction syntactically, it is especially interesting to observe infants' behavior when they react to the instruction without a preposition.

3.1. Stimuli In two different situations, infants were instructed to relate two objects in an ON-relation. In a well-known situation, objects and toys were used that the infants knew from their everyday situations. The trials in the well-known situation consisted of three sets:

236 Katharina J. Rohlfing

Set 1: Pot (Trajector-object) and Table (Landmark-object); Set 2: Cup (TR) and Plate (LM); Set 3: LEGO®-Horse (TR) and Bridge (LM). In the second situation in which infants were tested the infants were confronted with an abstract wooden construction. The construction can be said to be abstract or 'not yet specified' because the child has never seen it before. Furthermore, it was designed not to provoke a certain relation with its physical properties and to give as little context as possible coming from the relational character of the objects used. Especially for the choice of relational objects in this wooden construction, it was important for them not to be associated with a relation, as e.g. an object with a surface associates the relation ON. Thus, two balls were used: the first was installed in a wooden construction named HiK5 and covered with Velcro®; the second was mobile and could be related (and attached) to the first one. Balls suggest few spatial relations, because they have neither a horizontal surface nor are they containers. They suggest only the well-known activity of "rolling".

Figure 1. The ^ - S i t u a t i o n

3.2. Design and procedure In two situations (an abstract ///^-situation and a well-known situation), instructions were given without and with a preposition:

No preposition required 237 Table 1. Design of the Experiment SITUATION well-known abstract HiK without preposition (INS„p)

AKA,

AKA 3

with preposition (INSNA->)

AKA 2

AKA 4

INSTRUCTION AKA: Number of correct performances according to the instruction

The order of the set (HiK, then sets 1, 2, 3 from the well-known situation) corresponded to the progressive interest of the children. For many infants the //zX-situation was too abstract and it was doubtful whether they would have wanted to continue to play with it, after they had seen the sets 1-3. Consequently, every child was seated in front of the HiK first. To attract her/his attention, questions like "What is it?" were asked. Then the names for the objects (sphere and ball) were introduced. Finally, instructions were given that were neutral in their formulation. They always began with Daj... 'Give...'. This formulation is semantically correct in Polish. Other verbs like Postaw... 'Put...' imply a state in which one object is supported by another (e.g. lies on a table).

3.3. Predictions These two situations correspond to two hypotheses. For the wellknown situation the first hypothesis can be formulated as follows: if the infants master both tasks (instruction with and without preposition) with the same success, it means that in a well-known situation infants understand instructions without prepositions as well as with prepositions (AKAi = AKA2), because the relational character of objects is crucial here. The focus will be on the child's first response to an instruction without preposition: if she or he carried out the expected relation, it would not be due to the preposition. For the abstract situation the second hypothesis is formed: if the preposition supplies important information, then the infants will pay

238 Katharina J. Rohlfmg

more attention to it in an abstract situation. This means that the instructions with prepositions will lead to better performance than instructions without prepositions (AKA3 < AKA4). In this case, the focus will be on the child's second response, i.e. to the instruction with preposition NA-» - can she or he specify the location according to this preposition? Finally, one can expect that the relational character of objects is a component of infants' understanding insofar as their behavior is guided by the relational character of objects, and it is easier for them to perform instructions that refer to well-known or typical situations. This means that toy-objects such as a table or a bridge imply the relation ON and, even when performing incomplete instructions, infants will choose this relation (in spite of the fact that linguistically there are other options to complete this instruction).

3.4. Participants

Twenty-four Polish-speaking infants (thirteen females and eleven males) with an average age of 23 months participated in this study. All the children were native speakers of Polish. The youngest participant was 20, the oldest 26 months old. With the aid of two pediatricians and one pharmacist the parents were invited to participate and contacted later. The experiment was carried out in three different places: in two doctors' surgeries and a parish room. Each subject was interviewed separately. The parents were asked to make no comments during the experimental trials, and to encourage the infants whenever they turned to their parents for help.

3.5. Scoring

The following table shows the infants' performance before scoring, i.e. how many infants performed the NA-» relation as suggested by the objects regarding different sets. The line 'other relations' in the table refers to how many infants performed different relations than

No preposition required

239

expected (e.g. POD [UNDER] instead of NA->). The line 'used different toy' shows how many children performed the NA-» relation with different toys as provided by the set (e.g. they put the horse ON the table instead of ON the bridge) — this performance was excluded from the final scoring. The term 'anticipation' refers to a situation in which a child performed the NA-» relation before the instruction could be expressed (as soon as both objects were presented). This reaction is also omitted for the final scoring. Table 2.

Infants' performance in a well-known situation (before scoring) set 1 pot/table

set 2 cup/plate

set 3 horse/bridge

performed NA-> other relations

14 0

13 1

14 0

used different invalid performed NA-> Anticipation

6 4 13 2

3 7 16 2

other relations

0

3 7 9 5 1+26

used different invalid

3 6

4

3 3

INSwp

INSTRUCTION

INSwp

INSna_»

3

0

Referring to the ///^-situation, the table shows how many infants performed a contact between the ball and the sphere due to the instruction without a preposition. In the case of an instruction with a preposition, infants were expected to specify the location of the ball on the sphere. If they could not specify the location due to the instruction, they mostly repeated the already performed relation - how many infants behaved this way can be viewed under 'other relation'. The line 'another action' refers to responses in which infants did not make a contact between the ball and the sphere but carried out other actions like throwing the ball away or trying to take the sphere down.

240 Katharina J. Rohlfmg Table 3. Infants' performance in an abstract ///^-situation (before scoring) performed contact

14

other relations

5

performed N A - >

4~

another action another action

4 11

INS

"τ INSTRUCTION

INSNA->

The final scoring (presented in Table 4 and 5) concentrates on a comparison between the different tasks' forms (with and without a preposition). Every infant's performance in the well-known situation containing an instruction with a preposition, e.g. in set 1, was compared to her/his performance resulting from an instruction without a preposition from another set, e.g. set 2. Altogether six comparisons were added up to establish the total result, but the performance in the well-known situation was not compared to the //zX-situation. The infants' performance due to instructions was coded as a solved or non-solved task: if a child performed a relation NA-> between the LM and TR-object in a well-known situation due to an instruction without preposition (INSwp) as well as due to an instruction with preposition (INSNA->) from a different set, this task was coded as a solved task. If a child performed another relation than NA-» between the LM and TR-object in a well-known situation, this response was coded as a non-solved task. In the case when the child performed the relation with a different object as provided by the set or her/his relational reaction came before the instruction could be uttered, this response was excluded from scoring. In an abstract situation, in the case of an instruction without any preposition a task was coded as solved if a child produced a contact between the ball and the sphere. In the case of an instruction with a preposition, a task was only coded as solved if a child located the ball in the correct NA-» position on the sphere (merely to produce a contact was not sufficient). Any other reaction, e.g. to put the ball on the bottom of the //zX-construction or on its vertical wall, was coded as non-solved.

No preposition required 241

4. Results The first result is my personal observation and concerns the infants' responses to the instructions without preposition. They were not surprised when the incomplete instruction was uttered. Rather they seemed to react as if it were a complete instruction and did not show any uncertainties in their behavior. The results of the experiment in the first situation are: Table 4. Results in the well-known situation

INS na ^ (AKA2)

solved non-solved

INSwp (AKA,) solved non-solved 48 1 1 0

According to the McNemar-test for dependent samples the tasks are equally difficult when the number of persons who solved the first task (INSwp) and did not solve the second task (INSna-*) is equal to the number of persons who did not solve the first but solved the second task. In the table above, the number is 1 in both cases. The results indicate that the tasks were equally difficult, and the hypothesis can be confirmed: AKAi = AKA2, i.e. that in a well-known situation, infants understand instructions without prepositions as well as instructions with prepositions. Note, that in the case of an instruction without a preposition, a certain relation (in this case: NA-») emerges especially from the objects. For the abstract ///.^-situation, the results are as shown in Table 5. Table 5. Results in the abstract ///^-situation

INSwp (AKA3) INSna_>

(AKA4)

solved non-solved

solved 2 10

non-solved 3 4

242 Katharina J. Rohlfmg

Because the sum of the interesting fields for the McNemar-test for dependent samples is 13 and therefore smaller than 60, the Binomialtest was applied to this data. Assuming a significance level of 0.05 the results of the Binomial-test indicate that the tasks were equally difficult. Moreover, the results suggest a tendency (-3 = 0.035) against the second hypothesis made for the abstract situation (if the prepositions supply important information, then the infants will pay more attention to them and the task INSNA-> will be easier for them to perform). In the abstract situation, it seemed to be easier for the infants to react to an instruction without preposition because this type of instruction implies only a contact between the sphere and the ball. In contrast, the type of instruction implying a concrete location of the ball on the sphere was difficult for the infants to carry out. Thus, the second hypothesis that infants need prepositions to understand locative instructions in an abstract situation cannot be borne out. On the contrary, the results tend to mean just the opposite: in the HiKsituation the location and the appropriate preposition of the ball on the sphere do not have a meaning; only the activity of attaching matters to the infants.

5. Discussion The results from the well-known situation show that the ΝΑ-» preposition does not have a prior role in understanding locative instructions. To solve the tasks, the relational character of objects used in the trials was crucial. As a consequence of the instructions without prepositions, infants carried out the NA—>· (ON)-relation as implied by the LM-object. Only in two cases during the cup / plate-set did the infants show other reactions. Their 'mistakes' were interesting in their own way. Marcel (23 months) received the instruction to put "the cup plate" (saucer) and he put it the other way around, i.e. the plate (saucer) ON the cup instead. I wondered about his reaction: according to Thiel's relational character, should not the plate suggest a passive ON-relation? After consultation with his mother it turned out that she always makes tea that way: she puts the tea in a cup and

No preposition required

243

then covers it up with a saucer to stew. Thus, while solving the task, this behavior imitated a situation that was well-known. In this example, the non-linguistic strategy was established by the typicality of the situation, depending on certain experiences of a child. According to my observations during the study, the typicality of a situation consists of a canonical spatial relation implying a canonical orientation of objects as noticed by Sinha (1982: 140). The child perceives the situation task-dependently, and the emerging relation is linked to a certain activity and a canonical relation (in Marcel's case: plate ON cup) with a canonical orientation (plate inverted, cup upright) of objects that depends on that situation. From the task-oriented situation follows the relational character of objects, which, as described by Thiel (1985), is dominated by the aspect of canonical orientation of an object. So, if the child observes a specific kind of tea-making and gets used to it (typicality of a situation), the child will associate the plate with the activity of covering, and the relational character of the plate is in this case: active, ON. The results presented here confirm Clark's partial semantics hypothesis to the extent that, at this stage, the infants' behavior is based on the combination of partial semantic knowledge and non-linguistic strategy. However, more than just the two rules formulated by Clark (1973) that are based on physical properties of objects are involved in the non-linguistic strategies; the typicality of a situation also belongs to them, which is an extension of the relational character of objects formulated by Thiel (1985) and includes cultural aspects as the canonical orientation of objects described by Sinha (1982). The tendency to refute the second hypothesis on the basis of the results from the abstract situation suggests that in a new and abstract situation the prepositions in an instruction do not supply any additional information. Moreover, in this procedure, a task in which children were expected to respond to a preposition directly and to specify the location accordingly, seemed to be more difficult. Nevertheless, a longer familiarization period could help to find out more about infants' behavior in the ///X-situation. For the time being, the results suggest that infants employ non-linguistic strategies to understand locative instructions, and a linguistic instruction cannot be under-

244 Katharina J. Rohlfmg

stood context-independently at this age. Infants seem to rely on their experience of what an object does - they cannot perform only on the basis of a linguistic instruction or its lexical meaning. The results from the well-known situation show that infants understand instructions containing the spatial relation NA-> (ON) very well. However, their understanding of instructions does not correlate with the lexical knowledge about this spatial relation. Many factors are responsible for infants' understanding of instructions in such a situation: non-linguistic strategies according to Clark (1973), the relational character of objects (Thiel 1985) including their canonical aspects (Sinha 1982) and the typicality of a situation. There seems to be a variety of strategies that establish the process of acquiring the lexical meaning of the spatial relation. Considering the observations presented here, some consequences can be drawn for studying conceptual and lexical knowledge: conceptual knowledge should be explored more task- and context-dependently, because the nonlinguistic strategies are a part of the infants' understanding and they establish the lexical meaning of spatial relations. Nonetheless some spatial relations like NA are easier to understand with the aid of nonlinguistic strategies than others (e.g. POD [UNDER]). Why? It seems plausible to speculate that the NA-relation can be mapped more directly to pre-established spatial concepts configured with the aid of non-linguistic strategies. In this sense, the NA-relation is supposedly more basic than universal, because it is based on non-linguistic strategies. Yet it is also culture-dependent, because non-linguistic strategies include culture-dependent aspects like canonicity, which corresponds to cultural values and experiences and therefore cannot be viewed as universal.

Acknowledgements The research reported in this paper took place in the Graduate Program "Task-Oriented Communication" (GK 256) at the University of Bielefeld, Germany and was supported by the Deutsche Forschungsgemeinschaft.

No preposition required 245

I am grateful to Dan I. Slobin for the idea of instructions without prepositions - and to Chris Sinha and René Dirven for their comments on this paper. I would like to thank Joanna Kosiedowska for her help in organizing this study in Poland, all the children and parents who participated, and Marc Rohlfing for support and videotaping.

Notes 1. Canonical rules, in general, are what govern relations between endostructure and exostructure in non-natural kinds, and their motivation is to be sought in the way in which human interest and value is both shaped by, and shapes, a natural environment. "Cups, for example, do not occur naturally, but must be produced; [...] Basically, a cup, to be a drinking vessel, must be a container; and containers must possess a certain (endo)structure. In this sense we may refer to the cup as a canonical container." (Sinha 1983: 276) 2. The terminology is adapted from Langacker ( 1991: 6, 1998: 5) 3. In the analysis of my study, the term lexical knowledge is applied taskdependently, concerning the main question of how much information a child receives through a certain preposition. Thus, by lexical knowledge I refer to the child's cognitive ability to abstract the particular meaning of an ON-relation and transfer it to every conceivable situation even if she or he has not seen this situation in their every day life. I distinguish lexical knowledge from conceptual knowledge, because I am convinced that conceptual knowledge develops much earlier than lexical knowledge can be determined. 4. An instruction including a static preposition could be constructed as e.g. Zrób tak, zeby konik bylna moscie 'Do this way that the horse is on [NA] the bridgeLOK]' - which is more complicated than the instruction Daj konika na most 'Put the horse on [NA-»] the bridge'. 5. The name HiK stands for "Heidelberger interessante Konstruktion", and refers to the fact that the idea for this construction was born in Heidelberg. 6. In this case only one child performed the converse relation (i.e. the plate on the cup), the other two put the cup on the table (on which the objects were presented, not on the toy-table).

246 Katharina J. Rohlfing

References Bowerman, Melissa 1996a The origins of children's spatial semantic categories: cognitive versus linguistic determinants. In: John J. Gumperz, and Stephen C. Levinson (eds.), Rethinking Linguistic Relativity, 145-176. Cambridge: Cambridge University Press. 1996b Learning how to structure space for language: A crosslinguistic perspective. In: P. Bloom, M. Peterson, L. Nadel and M. Garrett, (eds.), Language and Space, 385-436. Cambridge, Mass.: MIT Press. Choi, Soonja and Melissa Bowerman 1991 Learning to express motion events in English and Korean: The influence of language-specific lexicalization patterns. Cognition 41: 83-121. Choi, Soonja, Laraine McDonough, Melissa Bowerman and Jean M. Mandler 1999 Early sensitivity to language-specific spatial categories in English and Korean. Cognitive Development 14: 241-268. Clark, Eve V. 1973 Non-linguistic strategies and the acquisition of word meanings. Cognition3: 161-182. Grieve, Robert, Robert Hoogenraad and Diarmid Murray 1977 On the young child's use of lexis and syntax in understanding locative instructions. Cognition 5:235-250. Langacker, Ronald W. 1991 Concept, Image, and Symbol. The Cognitive Basis of Grammar. Berlin and New York: de Gruyter. 1998 Topic, subject, and possessor. Linguistic Notes from La Jolla 19: 1-28. Sinha, Chris 1982 Representational development and the structure of action. In: G. Butterworth and P. Light (eds.), Social Cognition: Studies of the Development of Understanding, 137-162. Brighton: Harvester. 1983 Background knowledge, presupposition and canonicality. In: T. Seiler and W. Wannenmacher (eds.), Concept Development and the Development of Word Meaning, 269-296. Berlin: Springer. Sinha, Chris, Lis Thorseng, Mariko Hayashi and Kim Plunkett 1999 Spatial language acquisition in Danish, English and Japanese. In: P. Broeder and J. Murre (eds.), Language and Thought in Development. Cross Linguistic Studies, 95-125. Tübingen: Gunter Narr.

No preposition required Slobin, Dan I. 1973

Thiel, Thomas 1985

247

Cognitive prerequisites for the development of grammar. In: Charles A. Ferguson and Dan I. Slobin (eds.), Studies of Child Language Development Crosslinguistic. Evidence for the Language-Making Capacity, 175-276. New York: Holt, Rinehart and Winston, Inc.

Räumliches Denken und Verständnis von Lokativen beim Spracherwerb [Spatial thinking and understanding of locatives in language acquisition]. In: Harro Schweizer (ed.), Sprache und Raum [Language and Space], 184-208. Stuttgart: Metzler. Wilcox, Stephen and David S. Palermo 1974 "In", "on", and "under" revisited. Cognition 3: 245-254.

The 'Graded Salience Hypothesis' in second language acquisition Istvan Kecskes

1. Objectives This paper discusses the validity of the 'graded salience hypothesis' (GSH) in second language acquisition based on a cross-sectional study conducted with 30 native speakers of English (NSs) and 51 non-native speakers of English (NNSs) from 12 countries who studied English as a foreign language in their home country for at least four years and have spent from six months to two years in the target language environment.

2. The Graded Salience Hypothesis The GHS claims that in LI processing figurative and literal language use is governed by a general principle of salience. Salient meanings (e.g. conventional, frequent, familiar, predictable, enhanced by prior context) are processed first, and parallel processing is induced when more than one meaning is salient (e.g. Gibbs 1980, 1984, Blasko and Connine 1993, Giora 1997, forthcoming). The hypothesis requires that the standard pragmatic model be revised: Instead of postulating the priority of literal meaning, the priority of salient (e.g. conventional, familiar, frequent) meaning should be postulated (Giora 1997, forthcoming). Consequently, it is not the figurative versus literal split, but the salient versus non-salient continuum that really counts when processing the meaning of words or utterances in the LI. Salient interpretation has unconditional priority over less salient interpretation. The most salient meaning of a word or utterance is

250 Istvan Kecskes

always activated, and is always activated initially. The more salient the meaning is, the more difficult it is to reject as the intended meaning. In the case of conventional metaphors or expressions, whose figurative and literal meanings are as salient, both are processed initially. Three types of processing can be distinguished (Giora 1997): i. Direct processing applies when highly salient meanings are intended. For example, the salient figurative meaning of highly conventionalized idioms or implicatures. ii. Parallel processing occurs when alternative meanings are equally salient (conventional metaphors whose metaphoric and literal meanings are equally salient). ili..Sequential processing applies when language is used innovatively, as in the case of literal uses of highly conventional idioms. The more salient meaning is processed initially, before the intended meaning is derived.

3. The Study 3.1. Selection of pragmatic units A study was conducted to investigate the use of situation-bound utterances (Kecskes 1999,2000a, 2000b) and implicatures (Grice 1968, Bouton 1994) by Non-Native Speakers of English (NNSs). The selection of these pragmatic units over other lexical phrases and expressions needs explanation: i. Implicatures and situation-bound utterances (SBUs) are especially suitable for checking the validity of GSH in second language acquisition because their interpretation is dependent on the context of a particular language use. However, they differ in the obligatoriness of occurrence. In institutionalized contexts the choice between actions that are theoretically available to the individual is much more restricted than in informal contexts. Reactions to

The 'GradedSalience Hypothesis'

251

situations are designed beforehand, and in many cases it is the SBUs that serve as a means of providing the required next move in the course of action (cf. Coulmas 1979). Some SBUs are very specifically bound to one type of situation while others are polyfunctional in usage. Implicatures differ from SBUs: they rarely occur in institutionalized contexts. If they do, there is a specific reason for that. Implicatures are usually not ready-made, prefabricated expressions but generated ad hoc for a particular situation. Some conventional implicatures, however, may be "routinized" and behave like SBUs. But as we will see later, even if they are "routinized" it is a type of conversation rather than a situational frame that they are tied to. ii. The proper application of SBUs and conversational implicatures in speech is one of the best signs of idiomatic language use and conceptual fluency (Kecskes 2000a, 2000b). NNSs' language production is usually not idiomatic enough for several reasons: (1) In principle every communicative function can be fulfilled in any language, but this does not imply that this is done in a like manner in different languages. (2) Types of social situations are not crossculturally invariant. (3) Without significant immersion in the target language culture NNSs do not have access to "conventionalized conceptualizations" of the L2 so they rely on the conceptual base of their LI. In the cognitive approach semantic representations which constitute the semantic pole of a linguistic sign are equated with "conventionalized conceptualizations" (Langacker 1988). Taylor argued that if the meanings of linguistic forms are equated with conceptualizations, and these conceptualizations are conventionalized in a language, then the conceptualizations are made available to speakers of a language by the language system that they have learned (Taylor 1993: 212). When acquiring a nonprimary language, students have to learn not only the forms of that particular language, but also the conceptual structures associated with those forms. It is not only grammatical competence but also conceptual fluency that governs the proper use of language (Danesi 1992, Kecskes 1999, 2000b, Kecskes and Papp 2000b).

252 Istvan Kecskes

3.1.1. Implicatures Mey (1993: 99) argued that an implicature is something that is implied in conversation, that is, something which is left implicit in actual language use. Grice (1968, 1975) made a distinction between conversational implicatures and conventional implicatures. Conversational implicatures are created ad hoc in the course of conversation and have to be calculated by the interlocutor. They are generated by a particular utterance for a particular receiver in a particular situation, so what is conversationally implicated is somewhat indeterminate and cancelable. The conventional implicature of an expression, on the other hand, is quite specific, and usually means the same for members of a speech community so its content is hardly cancelable. Conventional implicatures may become routine formulas and are used almost automatically. These expressions are routinely interpreted in a derived (i.e. non-literal) sense. For instance, "Please help yourself' is often used by Americans at the table to urge their guests to start to eat or take some more food. The original cognitive mechanism responsible for the situational meaning of the expression is no longer maintained consciously, and no inferential reasoning is necessary to find out that the speaker asks you to "help yourself' not because he does not want to. The linguistic form has acquired a pragmatically motivated sense which became conventionalized.

3.1.2. Situation-Bound Utterances (SBUs) SBUs are highly conventionalized, prefabricated pragmatic units whose occurrence is tied to standardized communicative situations (Kecskes 1999, 2000a). If, according to their obligatoriness and predictability in social situations, formulaic expressions are placed on a continuum where obligatoriness increases to the right, situationbound utterances will take the rightmost place because their use is highly predetermined by the situation. The acquisition of these units in an L2 requires the knowledge of the socio-cultural background of the target language because SBUs are functional units whose mean-

The 'GradedSalience Hypothesis'

253

ing can be explained only as functions of habitual usage (Kecskes 2000a). The pragmatic functions are usually not encoded in these linguistic units, therefore SBUs often receive their "charge" from the situation they are used in. It is generally this situational charge that distinguishes SBUs from their freely generated counterparts. Many of them can have both their figurative and the literal meaning as salient. However, the most salient meaning is usually the figurative one.

3.2. Data collection The participants were given a test which consisted of two parts. The first part focused on implicatures while the second part tested the use of SBUs.

3.2.1. Implicatures The first task presented the students with ten dialogs that included implicatures with their immediate context whose meaning the students were expected to explain. 1/1. Mr. R: Mr. M: Mr. R: Mr. M:

Two teachers are talking about a student's paper: Have you finished with Mark's term paper yet? Yeah, I read it last night. What did you think of it? Well, I thought it was well typed.

1/2. Amy: Don't you think Jim drinks a bit too much? Billy: Is the Pope Catholic? 1/3. Pam: Betty:

Do you need sugar in your coffee? About as much as I needed that last piece of cheese cake.

VA. Wife: Have you seen my lipstick? Husband: Your daughter was in your room this morning.

254 Istvan Kecskes

1/5. Ann: Bill:

How do you like your teacher? Well, he dresses nicely.

1/6. Alan: Do you think Mr. Herbert will give us a quiz today? Bob: Does the sun rise in the East? 1/7. Mary: Do you know how my car got scratched? Steve: I have a meeting to go to and I'm late. 1/8. Adam: How is Bill doing at his new job in the bank? Jim: Oh, quite well, I think. He hasn 't been thrown in prison yet. 1/9. Tim: Bill:

Where's my ham sandwich? Well, the dog looks happy.

1/10. Mark: Does Joe have a girlfriend these days? Tom: He's been making a lot of visits to Billings lately.

3.2.2. Situation-Bound Utterances In the second part students were expected to demonstrate their competence in the use of situation-bound utterances (SBUs) in three different tests: Test 1 In Test 1 students were asked to describe the meaning of situationbound utterances without context: S/l. S/2. S/3. S/4. S/5. S/6.

Take it easy. Piece of cake. Get out of here. Why don't you sit down? I'll talk to you later. Welcome aboard.

The 'GradedSalience Hypothesis'

255

S/7. Stick around. S/8. Tim is going out with that blonde. S/9. Please hear me out. S/10.1 am sorry, I didn't get your name. S/11. Hang on for a second. S/12. You are welcome. S/13. Help yourself. S/14. Give me a break. S/15. Give me a hand. Test 2 In Test 2 students were given dialogs with SBUs most of which have both a literal and a figurative meaning as salient. The participants were expected to recognize the most salient meaning in the given situation. Some expressions from the previous list (S/2, S/3, S/5, S/6) were repeated here to find out how student responses change if the expressions are used in a situational context. S/16. - Bill, I don't think I can agree with you. - OK, shoot. S/17. - Frank, I think you really deserved that prize. - Get out of here. S/l 8. - Hurry up. We will miss the train. - Come on, we have plenty of time. S/l 9. - Jim, do you think you can repair this coffee machine? - Piece of cake. S/20. - John, please clean up this mess from the desk. - Oh, give me a break, will you? S/21. - David, let me introduce you to our new secretary, Sally. - Nice to meet you, Sally. Welcome aboard.

256 Istvan Kecskes S/22. - 1 must be going now. See you soon. - OK, I'll talk to you later. S/23. - 1 think John was really rude to you yesterday. - You tell me about it. S/24. - Thank you very much for your kindness. - You bet. Test 3 The last part of the test focused on SBUs in which the figurative meaning is usually the most salient meaning. Some of these expressions (S/25, S/29) are rarely used in their literal meaning, others (S/26, S/27), however, quite frequently keep their original meaning. In the dialogs below these formulae were given in their literal meaning, and participants were expected to consider the literal meaning of the expressions as intended meaning in the given situation. S/25. - Bill, you must show me how to use this word processor. - Sorry, I am busy. Help yourself on that. S/26. - Jim, can I spend the night in your apartment? It's, kind of, too late to go home now. - You know jOH are always welcome. S/27. - Mary, why don't you sit down? There is a chair over there. - OK, OK, I prefer to stand. S/28. - Andy, can you tell me why you left me in the store without a word? - Not now. Don't you see I am busy. I'll talk to you later. S/29. - What do you want me to do? - Stick around here until I get back.

The 'Graded Salience Hypothesis'

257

S/30. - I feel like going to town. - I'm not going out with you now. I have a lot to do. S/31. - Tom, get me out of this hole, will you? - OK, give me a hand.

4. Hypothesis It was hypothesized that salience is not only a linguistic but also a socio-cultural phenomenon which is based on particular knowledge structures that are language- and culture-specific and depends on conceptual fluency in the target language. Not having those knowledge structures in their Common Underlying Conceptual Base1 (Kecskes 1998; Kecskes and Papp 2000a) NNSs have difficulty identifying salient meaning of an expression in the L2, and usually consider the compositional and literal meaning of the utterance as salient. The study attempted to answer the following questions: Can NNSs apply the principle of salience in their L2? What are the characteristics of conceptual fluency in the L2? In what cases do NNSs rely on their LI salience? How language-specific is the principle of salience?

5. Data analysis 5.1. Implicatures Data demonstrated that in most cases students had more difficulty interpreting conventional implicatures (1/1; 1/2:1/5; 1/6) than conversational implicatures (1/4; 1/8; 1/9; 1/10). This difficulty derives from the fact that conventional implicatures usually require more specific background knowledge and are more typical for a particular culture than conversational implicatures which are usually based on common sense.

258 Istvan Kecskes Table 1. Implicatures IMPLICATURE

CORRECT

MISINTER.

NO IDEA

Well, I thought it was well typed. Is the Pope Catholic? About as much as I needed that last piece of cheesecake. Your daughter was in your room this morning. Well, he dresses nicely.

35%

61%

4%

33% 29%

14%

53% 21%

Does the sun rise in the East? I have a meeting to go to and I'm late. Oh, quite well, I think. He hasn't been thrown in prison yet. Well, the dog looks happy. He's been making a lot of visits to Billings.

little 25%

much 26% 4%

10% 10%

88%

good teacher 24% 22% I did it I didn't 14% 12% 12%

96% 80%

4% 18%

86% 41%

61% 63%

welldressed 25%

17% 11% 0% 0% 2%

There is a special group of conventional implicatures that are especially problematic for NNSs: relevance implicatures. The uniqueness of this group is in that it is usually not the expression itself that is conventionalized but the function and/or the structure. When, for instance, there is a "yes/no" question, or a statement and the obvious answer from the perspective of the interlocutor is "yes", a relevance implicature may be used in a question format as in 1/1 : Amy: Billy:

Don't you think Jim drinks a bit too much? Is the Pope Catholic?

One and the same function ("Isn't this obvious for you?") can be lexicalized in a lot of different ways. This kind of implicature is not necessarily culture-specific but the way it is lexicalized can be, and this may result in an absolute misinterpretation of the implicature in L2 processing. The pragmatic unit "Is the Pope Catholic?" is a good

The 'GradedSalience Hypothesis'

259

example of how a conventional implicature becomes a fixed expression. Some of the explanations from respondents of different LI background and culture demonstrated very clearly that students relied on their LI socio-cultural background when processing relevance and misinterpreted the relevance implicature when its content was too specific in the American-English culture was too specific. For instance: 1/4. Russian student (2 years in US, female, age 23): "Catholics drink too much." 1/5. Spanish student (6 months in US, female, age 27): "Billy thinks that Catholic people drink too much." 1/6. Spanish student (8 months in US, male, age 22): "Billy confused Catholic with alcoholic." When the relevance implicature was expressed by an utterance whose processing required less specific knowledge the results showed that the connection was easier for NNSs to make. (See numbers in Table 1.) For instance in 1/2: Alan: Bob:

Do you think Mr. Herbert will give us a quiz today? Does the sun rise in the East?

Another interesting example for the relevance implicature is 1/1, and 1/5. What is conventionalized here is not the expression but the function: "if you do not want to say anything negative about something/someone, talk about things that are positive about that person or thing". The obvious irrelevance of response is expected to indicate negative opinion to the interlocutor as in 1/1 and 1/5: 1/1. Mr. R: Mr. M: Mr. R: Mr. M:

Have you finished with Mark's term paper yet? Yeah, I read it last night. What did you think of it? Well, I thought it was well typed.

260 Istvan Kecskes

1/5. Ann: How do you like your teacher? Bill: Well, he dresses nicely. The numbers in Table 1. (Correct: 1/1= 35%; 1/5.= 41%) demonstrate that students relying on the propositional meaning of the expressions had difficulty interpreting the intended meaning in these situations. In 1/5 many students noticed only the positive propositional meaning of the expression and concluded either that Bill had positive opinion about the way the teacher dresses (25%), or that Bill thought that the teacher was a good one (24%). Conversational implicatures were easier for NNSs to process because the salient meaning of the utterance is the propositional meaning that did not require any culture-specific interpretation. For instance: 1/8. Adam: How is Bill doing at his new job in the bank? Jim: Oh, quite well, I think. He hasn't been thrown in prison yet. 1/9. Tim: Bill:

Where's my ham sandwich? Well, the dog looks happy.

However, when the conversational implicature contained some culture-specific reference such as, for instance, a connection between cheesecake and proper diet, the full understanding of the propositional meaning of the utterance did not necessarily result in correct interpretation of the expression. (See numbers in Table 1.) For instance in 1/3: Pam: Betty:

Do you need sugar in your coffee? About as much as I needed that last piece of cheesecake.

The 'GradedSalience Hypothesis'

261

5.2. Situation-Bound Utterances Testi In the first test SBUs were listed without context. According to the GSH for information to be salient, i.e. to be foremost on one's mind, it needs to be stored or coded in the mental lexicon (Giora 1997, forthcoming). Stored information enjoys cognitive priority over unstored information such as new information or information inferable from context. Consequently, it is the most salient sense of a word or expression that is directly computable from the mental lexicon before any extra inferences based on contextual information are put to play . This approach claims that context has a limited role in the activation of salient meanings because salience is a matter of convention, frequency, and familiarity. Research (Rayner, Pacht and Duffy 1994, Giora 1997) suggested that even when prior context is heavily biased in favor of the less salient (e.g. less frequent, familiar and conventional) meaning of an ambiguous expression, salient meaning is activated first. So salient meanings cannot be bypassed in LI processing. Familiar metaphors and frequent SBUs often have both a figurative and a literal meaning as salient. They activate both their literal and figurative meanings simultaneously if both meanings are equally salient. However, one of the meanings is canceled based on contextual clues. In the first test the expressions could have more than one salient meaning. However, in most cases the figurative meaning was expected to be considered salient. Both the NS and NNS responses support GSH. The majority of respondents found the most salient meaning without context. The numbers demonstrate how important frequency of encounters with the expression is for both NSs and NNSs. Even the NSs made their decisions based on frequency when two salient meanings were possible. For instance: "Take it easy", "I'll talk to you later". Where parallel processing would be needed because two meanings are equally salient (for instance: "Get out of here", "Give me a break"), NSs are divided in their decisions since there are no contextual cues to rely on. NNSs, however, prefer literal meaning to figurative meaning, which makes sense, and supports our hypothesis about NNS language proc-

262 Istvan Kecskes

essing which usually prioritize literal meaning. The expression "Welcome aboard" is a good example to demonstrate how much NNSs rely on their experience, and how important familiarity with the expression is, when there is no context. A significant number of NNSs (22%) identified "Welcome aboard" with the expression they hear from the captain on an airplane. Table 2. SBUs without context SBUs

FIGURA- LITERAL OR F2 TIVE

Take it easy Piece of cake Get out of here Why don't you sit down. I'll talk to you later Welcome aboard

NNSs

NSs FIGURATIVE

relax relax bye 17% 75% 83% 90% 100% 0% kidding kidding leave 53% 8% 47% sit down let's talk 92% 83% 17% else bye 30% 75% 70% soc. accept. soc. accept. ship 61% 10% 90% 100% 0% 73% dating derog. 80% 20% 69% listen don't interr. 80% 20% 80% 4% 76% 96%

LITERAL OR F2

MISINTERPR

bye 18% 6% leave 82%

10%

0%

8%

18% airp. 22% 25%

7%

Stick around Tim is going out 6% with that blonde Please hear me 0% out I am sorry, I 10% didn't get your name 100% 0% Hang on for a 88% 4% second You are welcome no problem guest no problem guest 83% 17% 75% 6% Help yourself eat do yourself 87% 13% 41% 29% Give me a break disagree stop kidding rest leave me 47% 53% 31% 27% Give me a hand 98% 2% 96% 2%

7% 4%

17% 2% 25% 20% 14%

8%

19% 30% 42% 2%

The 'GradedSalience Hypothesis' 263

In some cases there are clear differences between NSs and NNSs. The expression "Give me a break" was interpreted by NSs as "I disagree with you" (47%), or "Stop kidding" (53%). The NNSs, however, thought that the intended meaning was either "Let me rest" (31%), or "Leave me alone" (27%). This also demonstrates that NNSs are usually more comfortable with a meaning that is close to the compositional meaning of the expression. The fact that NSs and NNSs do not share a common socio-cultural background also led to different interpretations of one and the same expression. Quite a number of the NSs (20%) said that the expression "Tim is going out with that blonde" was derogative. None of the NNSs thought that way. Test 2 In the second test context was biased for the figurative meaning of SBUs no matter whether the figurative meaning or the literal meaning of the expression was salient such as, for instance, in S/16 ("OK, shoot"), S/17 ("Get out of here"), S/18 ("Come on"). Table 3. SBUs withfigurativemeaning SBUs OK, shoot Get out of here Come on Piece of cake Oh, give me a break, will you? Welcome aboard I'll talk to you later Tell me about it You bet

FIGURATIVE 55% shut up 39% 76% 100% don't bother 61% 70% 84%

LITERAL OR F2 0% leave me 24% 10% 0% want rest 18% 5% 14%

agree 49% 92%

talk about it 22% 0%

MISINTERPRETATION 45% 47% 14% 0% 21% 25% 2%

29% 8%

264 Istvan Kecskes

There is absolutely no difference in how NSs processed the expressions. When parallel processing applied because both the literal and figurative meanings were equally salient, context played a decisive role. However, this was not necessarily the case for NNSs. The table below shows that their interpretation was based on frequency ("Piece of cake", which is very frequently used in everyday conversation was processed 100% correctly), and contextual cues not always helped them to find the correct salient meaning if it was not the literal meaning. For instance: S/16 ("OK, shoot"), and S/17 ("Get out of here"). It is interesting to compare the responses in two situations in Test 1 and Test 2. Context changed the interpretation of these two expressions significantly. "Get out of here" Test 1. S/3. "Kidding": 8% "Leave": 82% Misinterpr.: 10% Test 2. S/17. "Shut up": 39% "Leave": 24% Misinterpr.: 27% "Give me a break" Test 1. S/14. "Rest": 31% "Leave me": 27% Misinterpr.: 42% Test 2. S/20. "Don't bother me": 61% "Want rest": 18% Misinterpr.: 21% Test 3 In the third test SBUs were used in their less salient meaning which was the original literal meaning. Usually it is very difficult to find situations where an SBU is used in its original literal meaning. This is what Gibbs (1980) called "literal uses of highly conventional expressions." Responses of NSs demonstrated that the less salient meaning was processed sequentially with no problem if the context was clear. In a couple of cases, however, where the situation was open-ended, and the context was unclear sequential processing did not always occur in the responses of participants because some NSs and many NNSs processed the most salient meaning directly. From this respect the S/31 "Give me a hand" situation deserves attention.

The 'GradedSalience Hypothesis'

265

S/31 - Tom, get me out of this hole, will you? - OK, give me a hand. The most salient meaning of "Give me a hand" is "Help me", that is to say, help me, the speaker. 24% of the NS respondents and 45% NNSs interpreted "Give me a hand" in that way: "Help me so that I can help you". Most of the NSs (76%), however, felt that the situational meaning was "Extend a hand (and I will pull you out of the hole)", which is not the most salient meaning of the expression. In the case of the expression "I'll talk to you later" both the literal and figurative meaning can be salient. A part of the respondents interpreted the SBU literally: "When I have time I will explain to you why I left you in the store" while the other part understood it as saying "Don't bother me now". Table 4. SBUs with literal meaning

SBUs

NSs Literal Figurative

Literal

NNSs Figurative

100% 100% Help yourself You are always wel- 100% 96% come Why don't you sit 100% 100% down? I'll talk to you later explain don't bother explain don't bother 57% 57% 43% 35% Stick around 100% 100% busy I am not going out 100% 84% 10% with you now extend a help OK, give me a hand extend a help hand hand 41% 76% 45% 24%

Misinter. -

4% -

8%

6%

14%

266 Istvan Kecskes 6. Conclusions 6.1. Salience is culture-specific The study confirmed the validity of the hypothesis which claimed that salience is language- and culture-specific and derives from certain knowledge structures that are essential parts of the conceptual base of native speakers. The more frequent, familiar, conventional and prototypical a certain piece of information in a particular linguistic community is the more salient it is in that community. Salience is a very important part of the socio-cultural heritage of NSs. It is something that reveals a unique feature of the mind: Giora (forthcoming) argues that rather than accessing the context-compatible information first, our mind opts for the most accessible information. When it fails to integrate with the context, we access the next available interpretation. It is this most accessible information that is culture-specific and makes language processing difficult for NNSs because they do not have direct access to the most accessible information since it is not stored or coded in their LI-dominated mental lexicon. What is stored and encoded there is what is salient in the native tongue and culture of language learners, and that usually does not work in the target language.

6.2. Low level of conceptual fluency in L2 NNSs can hardly apply the principle of salience in the target language. The lack or low level of conceptual fluency in the L2 forces NNS to rely on linguistic signs rather than conceptualizations while processing L2. The study demonstrated this very well. NNSs usually had no difficulty identifying the compositional meaning of SBUs and implicatures. Problems occurred when literal meaning was not the most salient meaning. Then NNSs mapped target language expressions on LI conceptualizations, which often resulted in misinterpretation of expressions. Using the principle of salience, LI speakers processed figurative meanings directly without falling back on literal

The 'GradedSalience Hypothesis'

267

meanings. Adult L2 speakers, however, could hardly ignore the literal meaning.

6.3. The priority of contextual cues in L2 The GSH claims that context affects comprehension after highly salient information has been accessed (Giora 1997, forthcoming). There is empirical research that supports the hypothesis that available information is accessed initially in LI processing, regardless of contextual fit or speaker's intent (Keysar 1998; Keysar, Barr, Balin and Paek 1998). This is not so in L2 processing where contextual cues seem to have priority over salience for reasons discussed above. It is usually the linguistic context that NNSs rely on, and this is a direct consequence of the compositional interpretation of words and expressions in the target language.

6.4. Application of GSH in second language acquisition The GSH was developed to explain LI processing. However, its application in second language acquisition can also be very useful since it may reveal and explain several important features of L2 language use and processing, which can contribute to better understanding of what goes on in the multilingual mind.

Note 1. Bi- and multilingual development results in the emergence of a Common Underlying Conceptual Base that is responsible for the operation of each language channel (Kecskes 1998, Kecskes and Papp 2000a).

268 Istvan Kecskes

References Blasko, Dawn G. and Cynthia Connine 1993 Effects of familiarity and aptness on metaphor processing. Journal of Experimental Psychology: Learning, Memory and Cognition. 19 (2): 295-308. Bouton, Larry 1994 Can NNS skill in interpreting implicatures in American English be improved through explicit instruction? - A Pilot Study. In: Lany Bouton and J. Kachru (eds.), Pragmatics and Language Learning. Volume 5, 88-10. Urbana-Champaign: University of Illinois Press. Coulmas, Florian 1979 On the sociolinguistic relevance of routine formulae. Journal of Pragmatics 3: 239-266. Danesi, Marcel 1992 Metaphorical competence in second language acquisition and second language teaching: The neglected dimension. In: James E. Alatis (ed.), Georgetown University Round Table on Languages and Linguistics, 489-500. Washington, DC: Georgetown University Press. Gibbs, Ray W. Jr. 1980 Spilling the bean on understanding: Comprehension and appreciation. Journal of Experimental Psychology : Learning, Memory and Cognition 9 (3): 667-675. 1984 Literal meaning and psychological theory. Cognitive Science 8: 575-604. Giora, Rachel 1997 Understanding figurative and literal meaning: The graded salience hypothesis. Cognitive Linguistics 8 (3): 183-206. Forthcoming On Our Mind: Salience, Context, and Figurative Language. New York: Oxford University Press. Grice, H. Paul 1968 Studies in the Way of Words. Cambridge, M.A.: Harvard University Press. 1975 Logic and conversation. In: P. Cole and J. L. Morgan (eds.), Syntax and Semantics. Vol. 3: Speech Acts, 41-58. New York: Academic Press. Kecskes, Istvan 1998 The state of LI knowledge in foreign language learners. WORD, Journal of the International Linguistics Association, Volume 49, No. 3: 321-340.

The 'GradedSalience Hypothesis' 269 1999

Situation-Bound Utterances from an interlanguage perspective. In: Jef Verschueren (ed.), Pragmatics in 1998: Selected Papers from the 6th International Pragmatics Conference, Volume 2, 299-310. Antwerp: International Pragmatics Association. 2000a A cognitive-pragmatic approach to situation-bound utterances. Journal of Pragmatics 32: 605-625. 2000b Conceptual fluency and situation-bound utterances. In: Mia Victori (ed.), Autonomy in L2 Language Learning. Links & Letters. 2000. Issue 7: 143-158. Barcelona: UAB. Kecskes, Istvan and Tunde Papp 2000a Foreign Language and Mother Tongue. Hillsdale, NJ: Lawrence Erlbaum. 2000b Metaphorical competence in trilingual language production. In: J. Cenoz and U. Jessner (eds.), Acquisition of English as a Third Language, 99-120. Clevedon: Multilingual Matters. Keysar, Boaz 1998 Language users as problem solvers. Just what ambiguity problem do they solve? In: Susan R. Fussell and Roger Kreuz (eds.), Social and Cognitive Approaches to Interpersonal Communication, 175-200. Mahwah, NJ: Lawrence Erlbaum. Keysar, Boaz, Dale Barr, Jennifer A. Balin and Timothy S. Paek 1998 Definite reference and mutual knowledge. Process models of common ground in comprehension. Journal of Memory and Language. 39: 1-20. Langacker, Ronald 1988 A usage-based model. In: Brygida Rudzka-Ostyn (ed.), Topics in Cognitive Linguistics, 127-161. Amsterdam: John Benjamins. Mey, Jacob 1993 Pragmatics. Oxford: Blackwell. Rayner, Keith, Jeremy M. Pacht and Susan Dufly 1994 Effect of prior encounter and global discourse bias on the processing of lexically ambiguous words: Evidence from eye fixation. Journal of Memory and Language. 33: 527-544. Taylor, John R. 1993 Some pedagogical implications of cognitive linguistics. In: Richard A. Geiger and Brygida Rudzka-Ostyn (eds.), Conceptualizations and Mental Processing in Language, 200-223. Berlin/New York: Mouton de Gruyter.

Subject Index acquisition 44, 131, 153f, 156f, 168, 177, 193-222, 229f, 234, 249f, 267 Actor-Action construction 173f, 178f, 185 actuality 34f, 44,63,65f, 69f, 73f, 77, 80, 85, 88f, 93 f, 96-100 adultocentrism 201,218 analytical linguistics 174, 177, 182, 188f argument structure 131, 173, 178 attenuation 64, 66, 69, 70, 73, 77, 80, 86-89, 94-97, 99 background knowledge 233,357 canonical orientation 231-233,243 canonical rules 231, 233,245 categorization 209,229 child language 44,193-222 cognitive grammar 3 - 6 , 9 , 1 3 I f , 158 Common Underlying Conceptual Base 257,267 conceptual -fluency 251,257,266 - prime 193-222 - substrate 15-18,21 - vocabulary 201, 202, 209, 216218 conceptualization 8, 10, 15, 42, 58, 81,251,266 construal 6-8, lOf, 13f, 19, 29, 37, 69f, 186 construction 4f, l l f , 15, 21, 33f, 42f, 50, 93f, 97, 109, 111-113, 117, 122-127, 132f, 135-141, 144159, 168-170, 173f, 176, 178180, 183-189, 194, 198,213, 236, 240, 245 mental-21,33f corpus 109, 111, 115,126f, 197

decontextualized language skill 59 delaying function 114, 124 direct processing 250 direct role marking system 151 discourse 5, 15, 42-44, 46f, 50, 58, 63, 92, 97, 109-112, 116f, 119f, 123-127, 139, 141, 143-146, 148, 153-155, 157f, 182 - function 123f discourse- or reference-related strategies 146 distal viewpoint 42f, 46 Early New High German (ENHG) 144f, 147 English tense 63, 65-67, 69f, 91, 9 6 98 experiential correlation 68, 70, 7884, 87, 89, 98f explication 194, 201, 205-207, 209f, 212f, 215,219,221f family resemblance 109,112,125 fictive 15, 17, 19-21, 30, 36 fragments 11 If, 124-126 grammaticalization 124 ground 5, 16, 22, 24, 30, 41, 46, 98, 178, 231 habitual(s) 33f, 43,49f, 56f, 59,253 //¿^-situation 236f, 239-243 idiom 171,250f image concepts 68,70, 75-78, 88 immediate scope 9f, 12, 14, 22-24, 26, 69 imperfective 1 If, 14,23-26,37,43f implicature(s) 19, 69, 94, 250-253, 257-260,266 indirect role marking system 151 information 16, 18f, 42, 52, 55, 69, 72, 74, 76-78, 81, 89, 91-94, 97f, lOOf, 120, 125, 155-157, 168f,

272 Subject index 174, 177-182, 184f, 188-190, 215, 230, 233, 237, 242f, 245, 261, 266f - n e w 120,261 - o l d 155 intimacy 63-66, 69-71, 74, 77, 80, 83f, 88-91,96f, 99 Korean 57, 131, 135-145, 147-150, 153-157,229f language - acquisition 153f, 193, 218, 220, 229f, 249f, 267 - learnability 131 language teaching 5, 63, 66, 97, 99, 187f lexeme 81, 93, 167-172, 174f, 177180, 182-189 mutable - 172,183,185f, 188 lexical knowledge 168, 177, 233, 244f linguistic variation 13If, 153 literacy 44,46, 59,126 - events 46, 59 literal language 249 locative preposition 229 meaning extension 65, 69, 98 mental construction 21, 33f metaphor 15, 19, 68, 74, 123, 133, 250, 261 multilingual development 257 mutable lexeme 172,183,185f, 188 naming insight 206 narrative voice 48, 50, 53 narrativity 42 natural semantic metalanguage (NSM) 193f, 196, 200f, 203, 206, 216,218-220 network 4, 65, 150, 171, 179-181, 184, 189f neurocognitive 167, 169-171, 176, 182, 184,188-190

non-îinguistic strategies 230-233, 243f Old Scandinavian 146,151 one-word stage 198,206,209-217 parallel processing 189f, 249f, 261, 264 participant role 178,190 passives 134-138,140f perfective(s) 10-14, 21, 23-29, 33, 42-44,47, 136f performative 18,26-28 perspectival shift 44,58 pragmatic strengthening 65, 69, 80, 89, 9If, 98f pragmatically determined word order 131, 142, 145, 148,151, 153, 157, 159 present tense 11, 21-30, 32-37, 45, 47-50, 52-54, 56, 58, 67, 69-72, 89f, 92-94, 97,100 presupposed constituent 183 presupposition 112, 124,183 preverbal concepts) 230 primaryhood 144, 146 primates 221 prime(s) semantic - 194f, 197f, 202, 209, 216,218,222 conceptual - 193-222 principle of salience 249,257,266 processing direct - 250 parallel - 189f, 249f, 261,264 production vocabulary 198-201, 209, 216f, 219 profile 9-12, 14, 16, 20, 22-24, 26f, 32-35 progressive 5, 11-14, 22, 24-26, 4244,46f, 50-54, 56, 58f - present 44, 50f, 53f, 56, 58f proto-linguistic stage 204-206

Subject Index ITS prototype 109, 123, 125-127, 129, 13 If, 134, 137, 148, 150-152 proximal-distal spatial relations 65, 78, 80, 82f, 86, 88f, 95, 99 pseudocleft 109-117,119f, 122-125 -piece 112-116,119-121,123f relational character of objects 232f, 236-238,242-244 response concepts 68-70, 75-80, 83, 88f, 99 role marking system -direct 151 -indirect 151 salience 63-66, 69f, 72, 74, 77, 80, 84, 88f, 91-93, 96f, 99, 249, 257, 261,266f principle o f - 249,257,266 salient meaning 249f, 253, 255-257, 260f, 264-266 second language acquisition 153f, 249f, 267 segmentation 184,189 semantic - prime(s) 194f, 197f, 202, 209, 216,218,222 -transparency 131,141,157 sequential processing 250,264 situation-bound utterances (SBU) 250-256,261-266 socio-cultural background 252, 259, 263 spatial - configuration 13,229 - relation 65, 229f, 232, 234, 236, 243f

speech event 18, 22-27, 32, 36, 43, 68f spoken discourse 44, 110-112, 121, 125 subject and topic prominence 139, 142,152f syntactic category 167-70, 173-177, 182, 189 syntax 46, 58, 109, 142, 149, 167171, 174, 177f, 185-187, 189, 194, 220f temporality 57 tense 3, 1 If, 16, 21-30, 32-37, 4143,45-59, 63-74, 80-83, 88-100, 114, 170, 198 English - 63, 65-457, 69f, 91, 9698 present - 11, 21-30, 32-37, 45, 47-50, 52-54, 56, 58, 67, 69-72, 89f, 92-94, 97, 100 time-reference 63-66, 69, 71, 77f, 80-82, 88-92, 94-96, 99 transitivity parameter 131, 141, 152, 157 transitivity parameter and prominence typology 131,152 two-word stage 198,212,217,222 typicality of the situation 243f understanding of instructions 244 universal relation 229f usage-based model 3, 5 variable constituent 172-175, 183, 185-187, 189f viewing arrangement 15-17, 19, 21, 26f, 29f, 32-37,41-46, 69f, 93

Cognitive Linguistics Research Edited by René Dirven, Ronald W. Langacker and John R. Taylor Mouton de Gruyter · Berlin · New York

This series offers a forum for the presentation of research within the perspective of "cognitive linguistics". This rubric subsumes a variety of concerns and broadly compatible theoretical approaches that have a common basic outlook: that language is an integral facet of cognition which reflects the interaction of social, cultural, psychological, communicative and functional considerations, and which can only be understood in the context of a realistic view of acquisition, cognitive development and mental processing. Cognitive linguistics thus eschews the imposition of artificial boundaries, both internal and external. Internally, it seeks a unified account of language structure that avoids such problematic dichotomies as lexicon vs. grammar, morphology vs. syntax, semantics vs. pragmatics, and synchrony vs. diachrony. Externally, it seeks insofar as possible to explicate language structure in terms of the other facets of cognition on which it draws, as well as the communicative function it serves. Linguistic analysis can therefore profit from the insights of neighboring and overlapping disciplines such as sociology, cultural anthropology, neuroscience, philosophy, psychology, and cognitive science. 1 Ronald W. Langacker, Concept, Image, and Symbol. The Cognitive Basis of Grammar. 1990. 2 Paul D. Deane, Grammar in Mind and Brain. Explorations in Cognitive Syntax. 1992. 3 Conceptualizations and Mental Processing in Language. Edited by Richard A. Geiger and Brygida Rudzka-Ostyn. 1993. 4 Laura A. Janda, A Geography of Case Semantics. The Czech Dative and the Russian Instrumental. 1993. 5 Dirk Geeraerts, Stefan Grondelaers and Peter Bakema, The Structure of Lexical Variation. Meaning, Naming, and Context. 1994. 6 Cognitive Linguistics in the Redwoods. The Expansion of a New Paradigm in Linguistics. Edited by Eugene H. Casad. 1996. 7 John Newman, Give. A Cognitive Linguistic Study. 1996.

Cognitive Linguistics Research Edited by René Dirven, Ronald W. Langacker and John R. Taylor Mouton de Gruyter · Berlin · New York

8 The Construal of Space in Language and Thought. Edited by Martin Pütz and René Dirven. 1996. 9 Ewa D^browska, Cognitive Semantics and the Polish Dative. 1997. 10 Speaking of Emotions: Conceptualisation and Expression. Edited by Angeliki Athanasiadou and Elzbieta Tabakowska. 1998. 11 Michel Achard, Representation of Cognitive Structures. 1998. 12 Issues in Cognitive Linguistics. 1993 Proceedings of the International Cognitive Linguistics Conference. Edited by Leon de Stadler and Christoph Eyrich. 1999. 13 Historical Semantics and Cognition. Edited by Andreas Blank and Peter Koch. 1999. 14 Ronald W. Langacker, Grammar and Conceptualization. 1999. 15 Cognitive Linguistics: Foundations, Scope, and Methodology. Edited by Theo Janssen and Gisela Redeker. 1999. 16 A Cognitive Approach to the Verb. Morphological and Constructional Perspectives. Edited by Hanne Gram Simonsen and Rolf Theil Endresen. 2001. 17 Emotions in Crosslinguistic Perspective. Edited by Jean Harkins and Anna Wierzbicka. 2001. 18 Cognitive Linguistics and Non-Indo-European Languages. Edited by Eugene Casad and Gary B. Palmer. Forthcoming.