Contrasting Languages: The Scope of Contrastive Linguistics 9783110860146, 9783110121339

Citation preview

Contrasting Languages

Trends in Linguistics Studies and Monographs 51

Editor

Werner Winter

Mouton de Gruyter Berlin · New York

Contrasting Languages The Scope of Contrastive Linguistics

by Tomasz P. Krzeszowski

Mouton de Gruyter Berlin · New York

1990

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

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

Library of Congress Cataloging in Publication

Data

Contrasting languages the scope of contrastive linguistics / edited by Tomasz P. Krzeszowski. p. cm. (Trends in linguistics. Studies and monographs 51) Includes bibliographical references and index. ISBN 0-89925-590-6 (cloth acid-free paper) 1. Contrastive linguistics. I. Krzeszowski, Tomasz P. II. Series. P134.C58 1990 410 —dc20 90-45502 CIP

Deutsche Bibliothek Cataloging in Publication

Data

Krzeszowski, Tomasz P.: Contrasting languages the scope of contrastive linguistics / by Tomasz Krzeszowski. Berlin New York Mouton de Gruyter, 1991 (Trends in linguistics Studies and monographs ; 51) ISBN 3-11-012133-6 NE: Trends in linguistics / Studies and monographs

© Copyright 1990 by Walter de Gruyter & Co. D-1000 Berlin 30 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. Typesetting: Arthur Collignon GmbH, Berlin — Printing: Gerike GmbH, Berlin — Binding: Lüderitz & Bauer, Berlin — Printed in Germany

Acknowledgements

I am indebted to all my friends and colleagues who took pains to read the draft of this book in its entirety or in parts and suggested improvements. The complete list would have to include many students of mine, the first patient experiencers of the ideas contained in this volume. Facing the impossible task of mentioning all the numerous persons whose comments influenced my thinking, let me single out the following scholars: Prof. Dr. Edmund Gussmann, Prof. Dr. Ruta Nagucka, Doc. Dr. Elzbieta Muskat-Tabakowska, Dr. Tadeusz Danilewicz, Doc. Dr. Barbara Kryk, Dr. Wojciech Kubinski, Dr. Krzysztof Kwasniewicz, Ms. Janina Ozga, and Dr. Ewa Willim. The book owes its existence to the tradition of contrastive studies in Poland, inspired, promoted, and supported by their spiritus movens Prof. Dr. Jacek Fisiak, to whom I hereby express my special thanks. I am also very grateful to Marc Weinstein, who carefully read my manuscript and suggested numerous stylistic improvements. I also owe a debt of gratitude to Ms. Olga Sokolowska for her assistance in compiling the indexes. If the book suffers from inaccuracies, misinterpretations, and other flaws, it is partly because its author did not pay sufficient heed to the illuminating comments of its critics. Assuming total responsibility, the author apologizes to both: the critics, for not listening carefully enough, and to the readers, for obliging them to endure the remaining inadequacies. Gdahsk, March 1988

Tomasz P. Krzeszowski

Contents

Introduction

1

Chapter I What is contrastive linguistics?

9

Chapter II Tertium comparationis

15

Chapter III Towards a classification of contrastive studies

23

Chapter IV Three steps in "classical" contrastive studies

35

Chapter V Contrastive studies at various levels of linguistic analysis

47

Chapter VI Linguistic models and contrastive studies

107

Chapter VII Towards a theory of semanto-syntactic equivalence Chapter

147

VIII

Contrastive Generative Grammar 169 Quantitative Error Chapteranalysis, XIX contrastive interlanguage, studies and Contrastive Generative Grammar 203 189

VIII

Contents

Chapter XI Cognitive linguistics and contrastive studies

213

Chapter XII "Theoretical" and "applied" contrastive studies

233

Notes

245

References

251

Index of subjects

267

Index of authors

283

Introduction

Contrastive studies do not enjoy much respect among linguists. Although many scholars of repute occasionally or systematically do practise what, loosely speaking, falls within the domain of contrastive studies by evoking cross-language evidence to support some theoretical claims (for example, R. Lakoff 1972, Comrie 1976, or Siewierska 1984), they only do so compelled by general linguistic motivation rather than by the requirements imposed by the rules of an altogether different game called contrastive linguistics. Those who, by their own admission, undertake contrastive studies only involve themselves in that part which we shall presently refer to as contrastive analysis proper, paying insufficient heed to matters of principle, which motivate the analyses and provide them with methodological tools. In any case, the emphasis falls on actual practice and applications to the detriment of the theory and methodology of contrastive studies. Whatever issues arise in connection with these latter two aspects of contrastive studies, they are treated only marginally, as it were, in passing, and without sufficient attention paid to matters of finer detail. Consequently, the number of works explicitly and exclusively devoted to the theory and methodology of contrastive studies is negligible (but see Di Pietro 1971 and James 1980). For some years, I have been engaged in providing contrastive studies with a more rigorous format. Several other linguists have advanced illuminating comments and criticisms, notably Bouton (1976), Chesterman (1980), James (1980), and, most extensively, Van Buren (1976). Thus, although only embryonically, something that might be called a methodology of contrastive studies has begun to emerge. The present book attempts to bring under one cover some discussions and controversies connected with contrastive studies, to suggest possible answers to some critical comments, and to provide a synthetic outlook on the state of the art of contrastive studies with some modest suggestions of improvement. Contrastive studies have a very long history. As early as ca. 1000 A. D. Aelfric wrote his Grammatica, a grammar of Latin and English, based

2

Introduction

on the implicit assumption that the knowledge of grammar of one language may facilitate the learning of another language. Among later grammarians, John Hewes, in the 17th century, was the first to explicitly express the view that the knowledge of the native grammar can not only facilitate learning a foreign language but also interfere with it. In his A perfect survey of the English tongue taken according to the use and analogie of the Latine, published in 1624, he devoted a long introductory section to presenting fundamentals of English in order to provide the learner with a "right knowledge or censure of their owne Mother tongue, in regard it holdeth a great difference in it selfe from the dialect of the Latine" Having provided some intricate contrastive analyses of Latin and English, Hewes provides numerous translational exercises to counteract what appear to be the effects of negative transfer, focusing his attention on those phenomena which are different in Latin and English. Many other grammarians, like Howel(l) (1662), Coles (1675), and Lewis (1670?) applied the idea of facilitation (positive transfer in modern terms) by adjusting their grammars of English or of Latin to the needs of speakers of various native languages. It is very interesting to note that those early contrastive studies were motivated in almost the same way as modern contrastive studies in the United States were motivated. As early as 1670, Mark Lewis wrote the following words: The most facil (sic!) way of introducing any in a Tongue unknown is to show what Grammar it hath beyond, or short of his Mother tongue; following that Maxime, to proceed a noto ad ignotum, making what we know, a step to what we are to lean (sic!). One wonders whether nearly three centuries later Charles C. Fries was aware of these words, when he wrote the following: The most efficient materials are those that are based upon a scientific description of the language to be learned, carefully compared with a parallel description of the native language of the learner (Fries 1945: 9)· So, although the word "contrast" with reference to different phenomena across languages had not appeared until the end of the 18th century, when James Pickbourne (1789: 18) first used it ("I thought it would be useful to contrast [italics supplied] the English verb with the verb in other languages"), comparisons of languages for pedagogical purposes probably go to the very beginnings of foreign language teach-

Introduction

3

ing, while systematic written records of such procedures go back to at least the 15th century (cf. Krzeszowski 1985, 1986 a, and in preparation; Meech 1935). The early contrastive analysts did not concern themselves with methodological problems, although they did work out a method of comparison known as the "sign theory", the first method in contrastive studies (Krzeszowski 1985). For many years contrastive studies were practised and applied in the classroom in a more or less intuitive way, toutes proportions gardees, like folk medicine, without much theory and without much explanation. But modern linguistic theories that began to flourish in the 20th century could not fail to affect the state of affairs in contrastive studies. Interest in methodology and theory of contrastive studies began to grow. Linguistic explorations into the nature of language, its complex, multilayer, and hierarchical structure, its systematic but changing nature, its function in communication, and its relation to the human mind became a subject of very close scrutiny of modern linguistics. This is not to say that many of these problems never arose before. For centuries, people wrote grammars (including contrastive grammars) and for centuries they were interested in how languages reflect human thoughts. But modern linguistic theories have given new dimensions to old problems and have created new problems (cf. Fisiak 1975 b; Fisiak — Lipinska-Grzegorek — Zabrocki 1978). Obvious things ceased to be obvious, and completely new approaches to language were proposed (e. g., generative grammar). Modern linguistic theories have made contrastive analysts sensitive to methodological and theoretical problems in their own field. Contrastive studies began to aspire to the status of a rigorous scientific discipline. What was once a relatively simple, intuition-based procedure, began to assume the format of an algorithm, which culminated in the attempt to construct Contrastive Generative Grammar (see Chapter VIII). The rigorous, mathematical approach to contrastive studies has revealed a number of thorny problems which at best make such efforts extremely difficult and at worst make them hopeless. Ironically, the attempt to give contrastive studies a rigorous, mathematical format could turn out to be self-defeating: it could reveal the impossibility of conducting formalized contrastive studies (cf. Van Buren 1976: 315). The theoretical problems which bedevil contrastive studies can be summarized as three paradoxes: the grammatical paradox, the semantic paradox, and the pedagogical paradox.

4

Introduction

The grammatical paradox Any grammatical contrastive studies performed without reference (at least implicit) to meaning are doomed to failure. Yet, witness another paradox, the revival of contrastive studies in the 20th century was promoted by rather extreme versions of American structuralism, in which the semantic aspect of language was removed from the mainstream of linguistic analysis. What makes a purely structural approach to contrastive studies theoretically impossible is the fact that any comparison presupposes similarity as tertium comparationis (TC) against which differences can be stated. Therefore, languages are structurally comparable in the extent to which they are structurally similar. It must be remembered that early structuralists were not yet familiar with the notion of underlying structure nor with the notion of semantic representation from which, according to later theories, surface structures in various languages are derived. These concepts were acknowledged and appreciated in contrastive studies as soon as they had made their way in the linguistic world. Originally, however, contrastive studies concerned only those structures which later became known as surface structures. If one considers a taxonomy of surface structures in various languages, it is possible to envisage a sort of a cline of similarities for every pair of languages within the taxonomy. Theoretically, at one extreme, there will be a pair of languages so radically different that no common set of grammatical categories can be established for them, and at the other extreme, there will be two languages identical in all respects. In-between, there will be a spectrum of pairs of languages with varying degrees of similarities. In such a hypothetical cline, the languages situated at the two extremes would be completely incomparable in purely structural terms. The two languages that would have nothing in common could not be compared since there would be nothing that could serve as tertium comparationis. The two structurally identical languages at the other extreme, being identical, could not be compared either since there would be nothing different to consider. Graphically the cline can be represented as in Fig. 1, where L t and L 2 have no features in common while L n and L n + , are identical in all respects. In fact, Fig. 1 represents a fragment of a more complex situation since, in principle, the number of totally different languages and totally identical languages may be more than two, and the respective numbers of such

Introduction

5

Figure 1. Languages in a cline ranging from those with nothing in c o m m o n to those that are identical.

languages cannot be determined a priori. Therefore, more accurately, our cline should be represented by the following formulas: Let U stand for the set of all possible human languages U = [L„ L 2

Ln]

and let σ stand for the similarity ratio, i.e., σ = s/d where s stands for the number of similar elements and where d stands for the number of different elements, 1 and where both s and d > 0 , in order for the L's to be comparable. Let us assume, further, that reliable descriptions of all the languages constituting U are available, and that these descriptions are all made within the same theoretical framework or a set of mutually compatible frameworks such that equivalence can be established for these frame-

6

Introduction

works. Provided these conditions are met, all languages in U can be arranged pairwise on the scale of σ from η to m, where η > 0 and m < oo. The grammatical paradox consists in the resulting situation, viz. that languages are comparable only to the extent to which they are similar; thus, the lower the σ for a given pair of languages, the less comparable they are. It follows that the most voluminous contrastive analyses should be performed on those languages for which σ would be the highest, while contrastive analysis becomes increasingly impossible with the diminishing σ. In either direction, purely structural contrastive studies reach an impossible impasse, which cannot be overcome, even in principle. The additional paradox, as we pointed out earlier, is that contrastive studies in the 20th century began to flourish within the structural framework. The paradox underlying such studies led to the situation in which, thriving as they were, they were at the same time completely atheoretical.

The semantic paradox Even before the difficulties connected with comparability within the structural framework had been formulated, practising contrastive analysts abandoned the strictly structural positions in favour of employing semantic criteria in establishing comparability. At first, semantics was employed implicitly as translational equivalence, the term borrowed from the theory of translation as formulated by Catford (1965: 27 — 34) and used in the context of contrastive studies by Halliday et al. (1964: 111 — 134). Translation equivalence was employed as the main criterion in deciding what constructions and sentences in various languages are comparable. The following principle was formulated: If the items are not at least sometimes equivalent in translation, they are not worth comparing (Halliday et al. 1964: 115). This clearly implied the presence of semantics on the contrastive scene, since translational equivalence presupposes at least some degree of semantic resemblance: two sentences are mutually translatable if they share at least some semantic properties. Later, a more radical hypothesis was formulated. It claimed that equivalent sentences across languages have identical semantic representations. This hypothesis was connected with the so-called universal base hypothesis (Bach 1968: 91; Fillmore 1968: 51 ff.), which claimed that all

Introduction

7

sentences in all languages are derived from a universal semantic base, for which various metalanguages of representations were proposed, such as a modified predicate calculus (Fillmore 1968), a system of roles (Fillmore 1968), or a system of labelled graphs (Krzeszowski 1974). All these proposals had one thing in common: they assumed the existence of some universal, underlying semantic representation, free of language-specific syntactic categories, from which all sentences in all languages are derived through language-specific categorical and syntactic rules. These rules can be compared and contrasted in various languages, which provides a new dimension for contrastive studies (see also Lipinska 1975: 50 ff.). The hypothesis concerning the identity of semantic representations of equivalent sentences leads to the semantic paradox, which is based on the fact that what is identical is not subject to comparison, and what is different is not comparable. Since equivalent sentences across languages have identical semantic representations, it follows that differences occur at less abstract levels, due to the operation of different, language-specific grammatical rules. Therefore, languages differ at more superficial levels, while in their deeper structure they are increasingly similar, and at the level of semantic representation they are presumably universally identical. Thus, the most interesting linguistic insights and generalizations provide the least promise for contrastive studies since with the increasing universality of the grammar (whether expressed in terms of the universal base hypothesis or in quite different terms, such as more recent versions of the standard theory, as expounded by Chomsky 1975, 1982, 1984), there is less and less for contrastive studies to deal with. Here is how Preston formulates this paradox: That the drive for universality should deny comparative detail between even related languages should seem paradoxical is understandable at a superficial level, but if we recall Chomsky's assertion that universal grammar will eventually leave only idiosyncratic odds and ends and irregularity behind in particular grammars, the conclusion that better grammars provide less and less detail for contrastive analysis is selfevident (Preston 1975: 69). Therefore, what is left for comparison are essentially incomparable idiosyncracies, such as suppletion, pronominal irregularities, morphological curiosities, and specific selection features, as the only items of any interest in the grammar of a particular language. According to Preston, even if contrastivists do trouble themselves with all these idiosyncratic details across languages, they will "make no real contribution to linguistics" (Preston 1975: 65).

8

Introduction

The pedagogical paradox Accepting the inevitable conclusion that what is universal must be familiar to all foreign-language learners from their native languages and constitutes the main focus of interest of theoretical linguists, we must also face the conclusion that what remains to be learned are "non-comparable bits of peculiar and idiosyncratic information generally ignored in contrastive studies" (Preston 1975: 65). So the pedagogical paradox is that what theoretical linguists consider as trivial and uninteresting may be of utmost pedagogical importance. Language teachers all over the world are particularly sensitive to errors in the realm of morphology, transgressions against rare, exceptional forms or unique pronunciations. All these phenomena, mainly connected with surface structures, traditionally constitute the learner's hell and, if excessively focused on, occasion many a failure in learning foreign languages in classroom situations. Languages abounding in such phenomena have the reputation of being "difficult" in contrast to those languages which are relatively free of such surface phenomena as inflections. Neither the semantic nor the pedagogical paradox undermine the validity of contrastive studies since there is no way of telling what is universal and what is idiosyncratic without conducting thorough contrastive studies of as many languages as possible. One cannot accept the view that a thorough investigation of [!] large number of languages is really unnecessary for the universal features necessary to explanatory adequacy are natural by-products of close, descriptively adequate work on the structure of a particular language (Preston 1975: 67). T. Zabrocki presents some convincing evidence demonstrating that certain theoretical linguistic hypotheses cannot be verified without reference to cross-language data. In fact, Zabrocki is of the opinion that in order to test any claim assigning a universal value to particular syntactic rules or categories one should ideally test these claims "on a number of, possibly all, languages, within general theoretical linguistic studies" (Zabrocki 1976: 101). The monograph presented here attempts to resolve these paradoxes by suggesting such methods in contrastive studies which would be theoretically plausible and practically useful. In this way problems and controversies beclouding contrastive studies may lose some of their edge, which will give contrastive studies a chance to survive as an art, if not as an algorithm.

Chapter I

What is contrastive linguistics?

When two or more languages are compared, it is possible to focus either on similarities or on differences. 1 When a learner learns a new language, he usually focuses attention on differences and remains largely unaware of similarities. If he discovers some similarities, he is amused and surprised since he ordinarily does not expect to find them. Grammarians, on the other hand, quite early became interested in discovering what various languages have in common, in the belief that making such similarities explicit for the learner may facilitate the process of foreign language learning. Early contrastive studies were motivated precisely by this assumption (Krzeszowski 1985: 485). There are several approaches to linguistic comparisons. The 19th century witnessed the development of historical linguistics and the related field of comparative historical studies, which aimed at finding the common genetic background for whole groups of languages. Another field of linguistic comparisons emerged when languages were grouped on the basis of various characteristics which they share. The area of linguistics concerned with such comparisons is called typological linguistics. Languages grouped together in the same typological group need not be genetically (historically) related. For example, English and Chinese, which are not genetically related, share a large number of grammatical properties, such as relatively fixed and grammatically constrained word order, paucity of inflections, and prominence of function words. These shared features place the two languages quite close in the typological groupings, in spite of the genetic distance separating them. Contrastive linguistics is connected with yet another kind of comparison: noting and describing similarities and differences in languages rather than grouping them genetically or typologically. Ultimately, of course, all kinds of comparisons may yield results which are relevant to linguistic theory in general, as in the search for linguistic universals. Therefore, differences between typological and contrastive linguistics are largely a matter of focus: typological linguistics focuses on clusters of languages

10

Chapter I

united by some common feature or features, while contrastive linguistics focuses on pairs of languages and explores similarities as well as differences between them. Contrastive linguistics, like descriptive and historical linguistics, is dependent on theoretical linguistics since no exact and reliable exploration of facts can be conducted without a theoretical background, providing concepts, hypotheses, and theories which enable the investigator to describe the relevant facts and to account for them in terms of significant generalizations. But contrastive linguistics is also dependent on descriptive linguistics since no comparison of languages is possible without their prior description. In brief, then, contrastive linguistics is an area of linguistics in which a linguistic theory is applied to a comparative description of two or more languages, which need not be genetically or typologically related. The success of these comparisons is strictly dependent on the theory applied. As will be seen later, in extreme cases, the linguistic framework itself may preclude comparison. Therefore, contrastive linguistics imposes certain demands on the form and nature of the linguistic theory which is to be "applied" in such comparisons. In many less extreme situations the results of comparisons are strictly dependent on the theoretical framework adopted in the comparisons (see Chapter VI). Originally, all contrastive studies were pedagogically motivated and oriented. In recent years, however, distinctions have been drawn between "theoretical" and "applied" contrastive studies (see Fisiak 1981: 2 — 9). According to Fisiak Theoretical CS give an exhaustive account of the differences and similarities between two or more languages, provide an adequate model for their comparison, determine how and which elements are comparable, thus defining such notions as congruence, equivalence, correspondence, etc. Applied CS are part of applied linguistics. Drawing on the findings of theoretical contrastive studies they provide a framework for the comparison of languages, selecting whatever information is necessary for a specific purpose, e.g. teaching, bilingual analysis, translating, etc. (Fisiak 1981: 9). In Chapter XII we shall investigate to what extent this is a valid distinction. Assuming, however, that "applied contrastive studies" are sufficiently distinct from "theoretical contrastive studies", the former, as part of applied linguistics, especially when related to teaching, must necessarily depend not only on theoretical, descriptive, and comparative

What is contrastive

linguistics?

11

linguistics but also on other disciplines relevant to teaching; among them are psycholinguistics, sociolinguistics, didactics, psychology of learning and teaching, and possibly other areas which may be important in ways difficult to evaluate at the present moment. Finally, some comments are needed about terminology. Although the word "contrastive" is used most frequently with reference to crosslanguage comparisons of the sort described above, various authors have been trying to replace it with other terms, such as "cross-linguistic studies", "confrontative studies", and some even more esoteric terms, for example, "diaglossic grammar" (Dingwall 1964 a), which enjoyed but a brief existence. The word "contrastive" is likely to outlive all the competing terms since it appears in titles of monographs and collections of papers on the subject (cf. James 1980; Fisiak 1980, 1981, 1984). Terminological complications manifest themselves in a rich array of nouns which have been attested in collocation with the adjective "contrastive" In the current literature we find such terms as "contrastive linguistics", "contrastive studies", "contrastive analysis", and "contrastive grammar" Although consistency is certainly wanting, there is an observable tendency to select a particular collocation to refer to particular domains of cross-language comparisons. And so the term "contrastive studies" appears to be the least marked, as it fits all contexts in which other collocations with "contrastive" are also appropriate. The term "contrastive linguistics" is also often used with reference to the whole field of cross-language comparisons, with a slight tendency to focus on those instances when theory or methodology of comparisons come into play. The collocation "contrastive analysis" is often used interchangeably with the above two terms, but there is a tendency to restrict its scope of reference to comparison proper. In that restricted sense "contrastive analysis" would refer to the third of the three steps in classical contrastive studies, viz. description, juxtaposition, comparison proper, respectively (see Chapter IV). Finally, the collocation "contrastive grammar" is often used to refer to the product of contrastive studies, as a bilingual grammar highlighting differences across languages. In this sense, "contrastive grammars" constitute an outcome of "contrastive linguistics" Whether these observations are accurate or not, it would be desirable to aspire towards some consistency in the use of these terms along the lines suggested in the above generalizations.

12

Chapter I

Unfortunately, these observations do not exhaust all the contexts in which the adjective "contrastive" appears. Many other collocations can be encountered in an astonishingly varied assortment. Each such collocation consists of the adjective "contrastive" followed by all manner of nouns; so we get "contrastive pragmalinguistics" "contrastive sociolinguistics", "applied contrastive studies", "contrastive discourse analysis" "contrastive pragmatics", "contrastive syntax", "pragmatic contrastive analysis" "contrastive generative grammar", "contrastive phonostylistics" "contrastive language studies", "contrastive lexicon", "contrastive considerations", "transformational contrastive studies", "theoretical contrastive studies", "classical contrastive studies", "deeper and deeper contrastive analysis", "contrastive description", and quite a few more. The existing confusion is typical of a field where insufficient attention has been paid to various matters of principle. Preoccupied with analytic details, investigators often lose sight of general distinctions and aims of their research, and they do not question certain fundamental assumptions, which are often taken for granted. In contrastive studies, the situation is further aggravated by the deeply-nourished conviction of many practitioners in the field that theoretical problems of relevance arise and can be solved only in the domain of pure and descriptive linguistics. In the view of these linguists, contrastive studies are merely a set of procedures involving mechanical application of various findings in theoretical and descriptive linguistics. Implicit in this position is the belief that contrastive studies do not require any special theoretical framework to be fully effective and to bring forth whatever results are expected of such analyses. There is a handful of linguists concerned with contrastive analyses, who occasionally feel guilty of the "sin of omission", and they stop to ponder over certain vagaries of comparative procedures, trying to see them in more general terms than those available from actual practice (e.g., Van Buren 1974; Schwarze 1978; Kühlwein 1983). These few investigators are aware that contrastive linguistics requires its own theoretical framework in addition to specific linguistic models employed in the description of the compared languages. One of the purposes of this book is to remedy this situation by highlighting and discussing some crucial issues which bedevil contrastive linguistics. We are going to address notorious problems such as tertium comparationis, equivalence, the relation between linguistic theory and

What is contrastive

linguistics?

13

contrastive linguistics as well as between contrastive linguistics and foreign language teaching. With respect to the existing terminological jungle, referred to earlier on, we are going to suggest a taxonomy of contrastive studies, which will provide motivation for some terms and render other terms superfluous.

Chapter II

Tertium comparationis1

To compare them would be tantamount to putting ten-ton lorries and banana skins in the same class on the grounds that neither ought to be left on footpaths! Carl James

One of the reasons why contrastive studies continue to perform the role of the Cindarella of linguistics is the fact that its most fundamental concept, tertium comparationis, remains as hazy as ever. The existing contrastive analyses involve various platforms of interlinguistic reference, determined by specific linguistic models which they employ and specific levels of analysis which they embrace. Thus different tertia comparationis are used for comparisons in lexicology, in phonology, and in syntax. In few of these studies is explicit mention of any tertium comparationis made or any justification for a specific choice presented. 2 All comparisons involve the basic assumption that the objects to be compared share something in common, against which differences can be stated. This common platform of reference is called tertium comparationis. Moreover, any two or more objects can be compared with respect to various features and, as a result, the compared objects may turn out to be similar in some respects but different in others. Thus, a square and a rectangle are similar in that both consist of four sides at right angles. But they are also different, since in a square, but not in a rectangle, the four sides are of equal length. 3 If we compare squares and rectangles with respect to the angles, we ascertain that the two types of figures are identical. If, on the other hand, we consider the length of their sides, we find them to be different. Depending on the platform of reference (or tertium comparationis) which we adopt, the same objects turn out to be either similar or different. In cross-language comparisons, the choice of tertium comparationis will also constitute the determining factor in establishing similarities and differences between the phenomena compared (cf. Lipinska 1975: 48; Fisiak et al. 1978: 15). Since language is a complex hierarchical structure, operating at various levels of organization, and since it manifests itself

16

Chapter

II

as texts produced by its users, every aspect of language at every level of organization, as well as every text and its constituents, can undergo comparison with equivalent elements in another language. Therefore, various kinds of contrastive studies can be distinguished, depending on the tertium comparationis adopted and the kind of equivalence involved. Before we attempt to classify contrastive studies, let us take a closer look at various possible tertia comparationis. Theoretical discussions tend to be limited to only two types: formal correspondence and semantic equivalence (e.g., Lado 1957: 52 — 53; Spalatin 1969; Ivir 1969, 1970). Even a cursory glance at the wealth of the existing contrastive studies suffices to notice that these two types of tertium comparationis are not the only ones that are used in practice. Formal correspondence and semantic equivalence can serve as tertia comparationis for certain types of contrastive studies, such as syntactic and lexical. Other types of contrastive studies, for example phonological, pragmatic, or quantitative, must be based on other tertia comparationis. Moreover, neither contrastive studies based on formal correspondence nor those based on semantic equivalence are free from difficulties. For example, it has been pointed out that formal likeness alone cannot serve as a tertium comparationis without support from semantic equivalence (Liston 1970: 44; Lipinska-Grzegorek 1977: 1 — 10). At best a comparison based on formal criteria alone is incomplete, at worst it cannot be performed at all, and in many cases it is misleading (see also Spalatin 1969: 3 1 - 3 4 ) . If, for example, one compares Polish and English personal pronouns, a formal analysis will ascertain the equivalence between the English you and the Polish ty/wy and will be accurate as far as it goes. But such an analysis is incomplete as it leaves out such forms as Pan/Pani and other possible equivalents of you. These equivalents can only be established if other than formal criteria are employed (see Chapter IV). English articles cannot be compared to anything in those languages in which there are no articles, if only formal criteria are considered. Finally, in the case of such phenomena as the present perfect tense in English and passe compos0 in French, a formal analysis is misleading since the formal similarity is not matched, at least in this case, by semantic similarity, which creates a kind of situation which often causes considerable learning problems (see Politzer 1968). Therefore, it is generally recognized that a contrastive analysis based on purely formal criteria falls short of both theoretical and practical expectations. We shall return to this problem in Chapter VI.

Tertium comparationis

17

Somewhat less obviously, a contrastive analysis based on semantic similarity alone can also be inadequate and misleading. In the contrastive practice, semantic equivalence is often erroneously identified with translation equivalence: To establish that these [systems of deictics] are comparable, we first need to show their contextual equivalence; this can be done most simply by reference to translation (Halliday et al. 1964: 115). Chapter VII will discuss differences between semantic equivalence and translation. It will be shown that translation equivalents are often semantically non-equivalent. At this point, it must only be noted that semantic equivalence must be constrained formally, while translation equivalence may, but does not have to, be thus constrained. When one translates, one departs from semantic equivalence due to three types of reasons: (1) errors in translation; (2) formal properties of respective languages; and (3) what is loosely called "stylistic" reasons. These three types of reasons lead to situations in which actual translation practice, with the exception of that concerned with legal texts, seldom involves semantic equivalents in the sense defined below in Chapter VII. This means that only some translations can be used as data for systematic contrastive studies (cf. Ivir 1969), while translation as a method of contrasting must be regarded with caution: Translation must be viewed amorphously as the rendition of a text from one language to another. This is translation from the standpoint of la parole: the text, the act of speech or writing, is the thing. Or it may be viewed as a systematic comparison of two languages: this is translation from the standpoint of la langue (Bolinger 1966: 130). In fact, the use of translation in systematic contrastive studies is highly limited: Translation equivalence serves merely to help us isolate items of structure with shared meanings in the two languages (Ivir 1970: 15). Even if we do distinguish translation equivalence from semantic equivalence and base contrastive studies on the latter, we still face problems. As has been stated earlier, semantic equivalence involves "formal" constraints. Thus, semantic equivalence is inherently connected with at least some degree of formal correspondence. But the meaning of both concepts is richer than is commonly recognized in contrastive studies. For instance, "formal" can be extended to cover the entire plane of expression (cf.

18

Chapter II

Hjelmslev 1961: 59), whereas in most American studies the word "formal" is restricted to word order, function words, inflections, affixation, and suprasegmentale. In a broader perspective, "formal" would also embrace such aspects of expression as aliteration, rhymes, and rhythm. Many of these "formal" properties would find their place in the study of functionally (pragmatically) equivalent texts (see Chapter XI). The notion "semantic" is also often extended to cover matters of pragmatics, especially by those authors who identify semantic equivalence with translation equivalence: Our experience is that languages can be effectively contrasted only on semantic basis, specifically, on the basis of translation equivalence (Spalatin 1969: 34). In reality many authors have shown that semantic equivalence is not a necessary prerequisite of a good translation (cf. Rülker 1973: 29 — 35; Krzeszowski 1974: 13, 1981 a; Kopczynski 1980: 4 1 - 4 2 ) . What is expected of a good translation is pragmatic or functional equivalence (see Chapter V). It cannot be denied that pragmatic equivalence can serve as tertium comparationis for contrastive analyses of such matters as the structure of discourse, stylistic properties, and quantitative aspects of texts. But syntactic contrastive studies, the primary concern of earlier contrastive studies, must be conducted within the limits of the semantic component of the language, or more specifically that part of the semantic component which can be systematically and predictably correlated with the grammatical structure of sentences. This restricted sense of "semantic" still embraces some aspects of meaning which are traditionally relegated to "pragmatics" or "interpersonal function" of sentences (Halliday 1970: 143). According to Halliday, the systems of mood and modality are precisely those systems which relate sentences to their interpersonal functions. It seems obvious that the notion "sentence semantics" should cover those elements of "pragmatics" which can be correlated with the structure of sentences, even if consistency in this area is definitely out of the question; declarative, interrogative, and imperative sentences do not necessarily perform the functions of statements, questions, and commands, respectively. In so far as some correlation between form and function does exist, those "functional" aspects of sentence structure constitute the border area between pragmatics and semantics and should be included in any semanto-syntactic contrastive studies (for details see Krzeszowski 1974). Therefore, James (1980) suggests that for the purposes of contrastive analyses translation equivalents should be limited to those which are

Tertium comparationis

19

both semantically and pragmatically equivalent. However, this proposal also raises doubts. Presumably, what James means by "semantic" refers to Halliday's "ideational" function of sentences, while Halliday's "interpersonal" and possibly "textual" functions fall under "pragmatic" Under James' proposal many sentences across languages would exhibit both "ideational" (semantic) and "interpersonal" (pragmatic) equivalence; yet one would hardly wish to use them as data for syntactic contrastive studies. For example: (1) (2)

Did he kill that dog? Zabil tego psa? (literally: 'killed-he that dog?')

are equivalent, both ideationally (agent, transitive verb, patient) and interpersonally (general question); yet, they falsely suggest a relationship between syntactic types represented by (1) and (2) in English and Polish. A systematic syntactic equivalence would have to be ascertained between (1) and (3) rather than between (1) and (2): (3)

Czy on zabil tego psa? (literally: 'whether he killed that dog?')

since (3) typically represents interrogative sentences in Polish, just as (1) is a typical interrogative sentence in English. In an earlier work (Krzeszowski 1981 b: 123), I suggested that syntactic contrastive studies should be performed on data restricted in the following way: a contrastive grammar will take as its primary data (to be assigned the status of semanto-syntactic equivalence) the closest approximations to grammatical word-for-word translations and their synonymous paraphrases, if such forms exist. Such a constraining of primary data as the basis for syntactic contrastive studies bypasses the inherent difficulties of the proposals suggesting the use of unrestricted semantic equivalence as the basis for comparison. Accepting any translation as a possible basis for syntactic contrastive studies leads to two mutually exclusive and undesirable consequences. Either (1) no comparative generalizations become possible, as the number of well-formed translations of a particular sentence into another language cannot be predicted a priori; or (2) purely arbitrary decisions concerning formal correspondences in unconstrained translations must be made. Any non-arbitrary decision involves circularity: the investigator has to assume formal correspondences on the basis of syntactic and/or morphological features which the compared texts

20

Chapter

II

share. This circularity is even reflected in the use of the word "comparable" in certain contrastive grammars. For example, Stockwell et al. thus write about determiners in English and Spanish: Both English and Spanish have two sets of determiners, commonly referred to as definite and indefinite articles. In many respects they are comparable [emphasis is my own]; in others they are different (Stockwell et al. 1965: 65). The circularity consists in the following: we compare in order to see what is similar and what is different in the compared materials; we can only compare items which are in some respect similar, but we cannot use similarity as an independent criterion in deciding how to match items for comparison since similarity (or difference) is to result from the comparison and not to motivate it. To avoid this undesirable circularity, in deciding about formal correspondences, one needs a common tertium comparationis outside the formal properties. The underlying meaning of the closest approximations to well-formed word-for-word translations provides such a tertium comparationis. Sentences and constructions sharing identical semantic representations at the level of sentence semantics (but necessarily exhibiting certain idiosyncratic differences at the level of word-semantics) are semantosyntactically equivalent and constitute a constrained set of data for syntactic contrastive studies. The approach through constrained translations does not require the initial recognition of shared syntactic categories as tertium comparationis for syntactic contrastive studies. Such a recognition would illegitimately anticipate the results of contrastive studies. A detailed proposal along these lines will be presented in Chapter VIII. (See also Krzeszowski 1974 and 1979). Summarizing, let us say that formal properties alone do not provide an adequate tertium comparationis for syntactic contrastive studies, while a semantic tertium comparationis must be constrained through restricting the scope of translation equivalents as primary linguistic data for syntactic contrastive studies. Such constrained but rigorous contrastive studies have a very limited pedagogical relevance. Any extension of the scope of contrastive studies to make them pedagogically more useful increases the likelihood of their becoming less rigorous and hence less respectable as a "scientific" procedure. One has to look for ways of extending the scope of contrastive studies without losing any of the rigour characterizing syntactic contrastive studies. Formal and semantic tertia comparationis, discussed so far,

Tertium comparationis

21

will not suffice as bases for extended contrastive studies. For example, phonetic and phonological contrastive studies cannot rely on semantic equivalence as tertium comparationis. The crucial notion in identifying various kinds of tertia comparationis and determining their character is the concept of equivalence or the relation which provides justifications for why things are chosen for comparison, keeping in mind that only equivalent items across languages are comparable. The various principles motivating equivalence and, eo ipso, contrastive studies will provide grounds for dividing tertia comparationis and, consequently, contrastive studies into various categories, each being connected with a specific kind of equivalence which motivates the comparisons (see Chapter III). In other words, equivalence is the principle whereby tertium comparationis is established inasmuch as only such elements are equivalent for which some tertium comparationis can be found, and the extent to which a tertium comparationis can be found for a particular pair of items across languages determines the extent to which these elements are equivalent. Thus, equivalence and tertium comparationis are two sides of the same coin.

Chapter III

Towards a classification of contrastive studies1

Contrastive studies can be divided and subdivided according to various criterial principles. Fisiak's division into "theoretical" and "applied" contrastive studies is based partly on the aims of contrastive studies and partly on their methodology. 2 According to Fisiak, theoretical contrastive studies are performed for their own sake, while applied contrastive studies are performed for the purpose of some application. But Fisiak reinforces this division by claiming that applied contrastive studies are directional while theoretical contrastive studies are not. This roughly means that theoretical contrastive studies do not investigate how a given category present in language A is represented in language Β they look for the realization of a universal category X in both A and Β (Fisiak et al. 1978: 10). Another kind of taxonomy can be based on specific linguistic models applied in the description of languages involved in contrastive studies. Since contrastive studies can be conducted in a variety of models, we can speak of structural, transformational, stratificational, or systemic contrastive studies. Yet not all models are equally suitable as frameworks for contrastive studies. In extreme cases, a particular model may turn out to be almost totally useless (cf. Van Buren 1974: 293; Lipinska 1975: 7ff.). A survey of some models as applied to contrastive studies and their evaluation in terms of their suitability as frameworks for contrastive studies will be presented in Chapter VI. It is important to note here that a taxonomy of contrastive studies based on criteria external to comparisons themselves can lend credence to linguistic theories and models rather than to contrastive studies. Therefore, external criteria will not be considered in the taxonomy which we are going to suggest later in this chapter. An interesting though incomplete taxonomy of contrastive studies is presented by Di Pietro (1971: 17 — 19), who divides contrastive studies

24

Chapter

III

into autonomous and generalized, on the one hand, and into taxonomic and operational, on the other. In autonomous contrastive studies no conscious, explicit reference is made to any universal, underlying structure which the compared languages might share. In such contrastive studies each language is described independently and in its own right. In generalized models, explicit reference is made to those layers of structure which the compared languages share, not only on account of their typological or genetic similarity, but mainly because of the universal grammar which is believed to underlie all human languages. The division into taxonomic and operational models pertains only to generalized models. Taxonomic models are restricted to stating similarities and differences across languages and to stating their "hierarchical importance" Operational models seek to formulate a series of conversions performed on the source language in order to produce the forms of the goal language (Di Pietro 1971: 18). This procedure would lead to the formulation, in linguistic terms, of the steps which would have to be taken by the learner to acquire a foreign language. Ideally such models would lead to the formulation of algorithms of foreign language acquisition. From the pedagogical point of view, autonomous contrastive studies have not been of much use in the area of language teaching since neither the descriptive apparatus that they employ nor the results which they yield can be easily related to any psychological or pedagogical reality. On the other hand, operational models are extremely difficult, if at all possible, to construct since still not enough is known about both learners and grammars with relation to psychological reality. In principle, Di Pietro's taxonomy could be augmented by generative models, i.e. such models which generate equivalent constructions and sentences across languages and assign appropriate structural descriptions, including similarities and differences, to the enumerated equivalents. A detailed proposal along these lines will be presented in Chapter VIII. Since our own classification of contrastive studies is based on distinguishing various kinds of equivalence (and hence tertium comparationis), it is apropriate to mention Kühlwein's typology of equivalence, designed in connection with lexicological contrastive studies situated in the context of socio-semiotics (Kühlwein 1983). His typology embraces the following kinds of equivalence:

Towards a classification

of contrastive

studies

25

formal — based on linguistic structure; derivational-semantic — connected with the "depth" of the derivation; paraphrase procedure — which yields "regressum ad infinitum"· translation procedure — limited by truth conditions and culturespecific considerations; and functional-communicative — involving "mental processes of cognition and associative connotative components" (Kühlwein 1983: 6).

This division suffers from the following shortcomings: a) It is not based on a set of homogeneous criteria: e.g., paraphrase procedure seems to be only loosely connected with contrastive studies; b) it does not mutually relate these various types of equivalence, nor does it suggest that there is a relationship; c) it leaves out certain other types of equivalence, for example, substantive equivalence; d) it does not explicitly relate the concept of equivalence to the concept of tertium comparationis, although, admittedly, Kühlwein is certainly aware of the importance of the relationship between the two concepts, at least in the preamble to his typology. The taxonomy which we are going to present is intended to be free of these inadequacies. It is based on the assumption that various kinds of contrastive studies can be distinguished in a strict relation to various tertia comparationis adopted and, consequently, to various kinds of equivalence. The first division is drawn between text-bound and systematic (or projective) contrastive studies. It is based on the familiar distinction between la parole and la langue. Text-bound studies involve comparisons of texts in two (or more) languages and do not go beyond such texts to generalizations about grammars, i.e. rules and systems that generate those texts. Projective contrastive studies are related to text-bound contrastive studies in the same way in which the study of language is related to the study of texts. Such studies go beyond primary linguistic data found in texts in order to grasp and formulate generalizations about various aspects of the compared languages. At this point, it is useful to introduce the term 2-textj'tu:tekst/ to refer to any pair of texts, written or oral, in two languages, which are used as data in contrastive studies. Every 2-text can be described in terms of a

26

Chapter 111

binary distinction: [ +translation] (henceforth [ ± trans]). A 2-text marked as [ +trans] is such a 2-text of which it can be asserted that its constituent texts can function as translations. Such 2-texts usually provide data for qualitative contrastive studies, which constitute the main bulk of contrastive studies. 2-texts which are not translations, marked [ — trans], can be used as data for quantitative contrastive studies, which will be discussed in detail in Chapter X. Text-bound contrastive studies are corpus-restricted if no systematic generalizations outside the original data are made. Quantitative contrastive studies are necessarily corpus-restricted, even if they enable one to make statistical predictions concerning other, similar texts. Quantitative text-bound contrastive studies may also be corpus-restricted as long as they do not aim at drawing systematic generalizations about the languages of the 2-text. But they may also serve as basis for projective generalizations, if clearly stated constraints on the selection of the relevant 2-texts are formulated and implemented. The relevant 2-texts serve as linguistic data on which contrastive grammars as generalizations about differences and similarities in the compared languages are based. We can now see that tertium comparationis is in fact the reason why any two texts are brought together as a 2-text and/or why any two items in two languages are juxtaposed for comparison. Each type of contrastive studies has its own type of tertium comparationis. Within each type of tertium comparationis, it is possible to distinguish more specific subtypes, subsubtypes, etc. unique within each type. Each type of tertium comparationis is connected with a specific type of equivalence. Text-bound contrastive studies may involve statistical comparative studies, and, as was said earlier, the relevant 2-texts need not be [ +trans]. However, to prevent comparisons of incomparables, one has to establish a tertium comparationis (and consequently an equivalence). The tertium comparationis will restrict the class of texts that can undergo comparisons. Thus, it may be necessary to require that texts constituting a particular 2-text, be written in the same register or at least deal with the same topic or represent the same literary genre. Whatever requirements on the "sameness" of the constituent texts are imposed, they will determine the tertium comparationis relevant in these texts. If the compared texts are translations, no additional requirements are necessary. Statistical equivalence can be established on 2-texts which are either [ + trans] or [ — trans], but in the latter case the extra requirements, referred to above, must be met. Statistical equivalence, marked as (1) in Fig. 1 below, obtains between various systematically equivalent items (see below) which appear

Towards a classification

of contrastive

studies

27

in 2-texts and which have maximally similar frequencies of occurrence. Let us note that in order to qualify as statistically equivalent, two items across languages need not be in the strictly delimited semanto-syntactic equivalence (relation type (4) below); but, to be comparable at all, they have to be equivalent in some sense. In many instances, however, statistically equivalent constructions are not semanto-syntactically equivalent. Consider as an example the English participial constructions in such sentences as (1)

I saw Peter entering the house.

The semanto-syntactic equivalent of (1) in Polish is (2)

Widzialem Piotra wchodzgcego do domu.

as the closest approximation to an acceptable word-for-word translation of (1) (cf. Krzeszowski 1974,1979,1981 b). Yet, if we consider quantitative data, (1) and (2) will turn out to be non-equivalent on at least two counts: intralinguistic and interlinguistic (contrastive). The intralinguistic count for English and an analogous count for Polish will reveal that the frequency of occurrence of the construction NP\ V NP2 Ving X, represented by (1), in comparison with other English verb-complement constructions, is higher in English texts than the frequency of occurrence of the semanto-syntactically equivalent Polish construction NPt V NP2 Vgcy X in comparison with other Polish complement constructions appearing in Polish texts. In Polish texts such constructions as 7VP, V Subordinate S (" Widzialem, jak Piotr wchodzil" — "I saw how Peter entered") or NP] V Nom Ngen ("Widzialem wejscie Piotra" "I saw Peter's entry") are more frequent. Thus, although the nearly congruent Polish construction (2) is available, it is less frequently used than other semantically similar constructions, which are more favoured by native users of Polish. 3 The same result can be obtained cross-linguistically and more directly by looking at various translations of a given construction into another language. The equivalent construction which is most frequently relative to other, nearly synonymous constructions, will also be used most frequently in translations. Thus, statistical comparisons can be conducted both on texts which are attested as translations and on texts which are not translations but are comparable on account of being written on a similar topic, by similarly qualified authors using similar registers, etc. We can generalize the above remarks about statistical equivalence as follows: two linguistic items across languages are statistically equivalent if they occur as the most frequent translations of each other and/or if,

28

Chapter

III

in comparison with other synonymous constructions, they have maximally similar frequency of occurrence in the relevant texts. 2-texts with the feature [ +trans] consist of texts which are in the relation of translation equivalence, marked as (2) in Fig. 1. It is important to note that such translations need not be "correct" or "acceptable" In fact, they often display considerable deviations from other kinds of equivalence. In particular, translations are usually at variance with what we shall later on describe as semanto-syntactic equivalence. Such deviations are not necessarily due to errors in translation, but often have their source in various pragmatic considerations, which override the demands of semanto-syntactic equivalence. All the same, it frequently happens that particular texts which purport to be translations exhibit errors due to the translator's incompetence, negligence, and other such factors. Usually such data will be rejected by investigators exploring translation equivalents, unless pathology of translation becomes the focus of interest and translation errors become the object of study. The latter pursuit does not strictly fall within the scope of contrastive studies. More will be said about the relation between translation and semanto-syntactic equivalence in Chapter VII. Systematic contrastive studies involve comparisons of constructions, systems, and rules. Of these three types, contrastive studies of constructions are based on semanto-syntactic equivalence, which constrains 2texts for the purpose of such studies in the way referred to above. More will be said about this type of equivalence in Chapter VII. The comparison of systems requires resort to syntagmatic considerations, because any paradigmatic analysis, i. e., any analysis involving systems, must be linked to a syntagmatic analysis since isolating a particular system in a particular language requires an analysis of syntagmatic arrangements of linguistic units, i.e., of constructions. For example, in order to isolate the system of personal pronouns in English in the nominative and oblique forms, it is necessary first to investigate various constructions in which such pronouns appear and then generalize on the basis of such syntagmatic data by extracting the paradigmatic system of personal pronouns. In order to obtain an equivalent system in another language, one must investigate equivalent constructions in that language and extract the relevant paradigmatic system from the data. Therefore, equivalence of systems cannot be ascertained without the foundation provided by the equivalence of constructions in which elements of these systems appear. In contrastive practice, equivalent systems across languages are usually juxtaposed on the basis of the investigator's intuition

Towards a classification

of contrastive

studies

29

(his knowledge of the two languages), corroborated by the established grammatical tradition expressed in the common terminology. It is usually tacitly assumed that such things as pronouns in one language and pronouns in another language are comparable by virtue of the common label "pronoun" used to refer to a specific set of words in one language and a specific set of words in another language. In this way, the concept of system equivalence (3) is employed as tertium comparationis. However, system equivalence is dependent on construction equivalence, since the former can be made explicit only through the examination of constructions in which the elements of the compared systems appear, i. e. via the notion of semanto-syntactic equivalence, relating the relevant 2-texts as primary data. Likewise, any comparison of rules cannot be divorced from an implicit comparison of constructions on which these rules operate. Most rules have a construction as input and a construction as output. Therefore, semanto-syntactic equivalence also underlies rule equivalence (5). We shall return to this sort of equivalence in Chapter VI. Phonological and lexical contrastive studies are based on the type of tertium comparationis which can be called substantive (6) insofar as it is connected with the material substance outside language, with which language is joined through its phonological interface, on the one hand, and through its semantic interface, on the other (Hjelmslev 1961). In the case of phonological contrastive studies, acoustic, articulatory, and, in principle, auditory phenomena provide the substantive tertium comparationis. Most phonological contrastive studies make reference to articulatory parameters, less frequently to acoustic ones, and never to auditory parameters alone. In the case of lexical contrastive studies, the external reality, or, strictly speaking, its psychic image in the minds of language users, provides the substantial lexical tertium comparationis. In this way, lexical items across languages are compared with respect to differences and similarities concerning their reference to various elements of the reality in the world at large and the ways in which this reality is reflected in the minds of language users. Both phonological and lexical contrastive studies are mainly paradigmatic, even if, especially in various kinds of generative frameworks, syntagmatic arrangements, and consequently rules accounting for those arrangements, are also comparable. However, in the case of phonological and lexical contrastive studies, we are faced with a situation which is a mirror image of semanto-syntactic studies: paradigmatic studies are now

30

Chapter

III

central, while syntagmatic studies are secondary and impossible to conduct without ultimate reference to substantive tertia comparationis. Semanto-syntactic equivalence is not required in the case of 2-texts compared in respect of their styles or registers. Such texts need not be semanto-syntactically equivalent but must be acceptable translations, which means that they have to be pragmatically equivalent. Obviously erroneous translations must naturally be disregarded, as well as those translations which fail to convey some relevant pragmatic functions, especially if alternative, more adequate translations are available. Although constraints on the suitability of 2-texts for stylistic contrastive studies are less rigorous than the constraints imposed on 2-texts as data for syntactic contrastive studies, they are no less important and must be stated clearly and unequivocally, lest contrastive studies fail to grasp the relevant generalizations concerning the pragmatic aspect of the compared texts. Pragmatic equivalence (7) as tertium comparationis for stylistic and sociolinguistic contrastive studies is thus a relation that holds between constituent texts of 2-texts selected in such a way that they evoke maximally similar cognitive reactions in the users of these texts. Becka (1978) describes this sort of relations in the following way: Since, however, contrasting does presuppose, apart from agreements and differences under observation, also a certain common foundation, one proceeds from the fact that the fundamental functional differentiation of styles holds in roughly the same way for cultured languages and that it makes itself felt in roughly the same tendencies even though not always realized by the same means of expression. In other words: French scientific style will be characterized by analogous stylistic tendencies as its counterpart in Czech, in the same way as the basic features distinguishing the belles-lettres narrative style from descriptive style will be the same in English as, say, in Italian. This is due to the impact of social, i. e. extralinguistic communicative needs which e. g. in languages of the European cultural sphere in the given period bear on the whole an analogous character. This is why we engage in comparing discourses of belles-lettres prose in various languages, or of scientific style in various languages and so on. In this manner a certain common foundation is gained upon which contrastive analysis of styles can be built (Becka 1978: 131 - 1 3 2 ) . Pragmatically equivalent texts may, and actually often do, correspond to translations and as such provide data not only for pragmatic and

Towards a classification

of contrastive

studies

31

sociolinguistic contrastive studies but also for text-bound contrastive studies. The seven types of equivalence discussed in this chapter and the related tertia comparationis characterizing various types of contrastive studies are presented in Figure 2. All these types of equivalence and the related tertia comparationis embody a number of more specific tertia comparationis, unique in every kind of contrastive studies. For example, semantosyntactic contrastive studies employ a large number of various specific tertia comparationis at various levels of generality, corresponding to the level of abstraction at which a given contrastive study takes place. This is connected with the fact that all descriptively adequate models of language give an account of the hierarchical structure characterizing language in both the syntagmatic and the paradigmatic dimension. This hierarchical structure can be represented by means of various conventional notations, such as trees or labelled bracketings with labels of nodes or of brackets corresponding to linguistic units at various levels of abstraction. The device called Contrastive Generative Grammar, to which Chapter VIII is devoted, is based on the claim that equivalent constructions across languages are derived from identical semantic representations, which means that down to a certain level of derivation, before they are diversified by language-specific grammatical rules, they have identical representations and that somewhere in the course of the derivation they begin to differ. The exact place in the grammar where particular pairs of constructions begin to diverge are different for various pairs of constructions. Therefore, Contrastive Generative Grammar provides all the necessary tertia comparationis for semanto-syntactic contrastive studies, beginning with the universal, language-neutral, semantic representation, which provides the most general tertium comparationis for all compared constructions and ending with the Immediately Relevant tertium comparationis, i. e. that place in the derivation of equivalent constructions which immediately precedes the first diversification. Given that Contrastive Generative Grammar is based on grammars organized along five components: semantic, categorial, syntactic, lexical, and post-lexical, arranged hierarchically from the most abstract to the most concrete, it is possible to expect diversifications to occur on all subsemantic levels. According to Contrastive Generative Grammar, all equivalent constructions share a certain number of rules, before they become diversified by the first rule which is different. The input to the rule which accounts for the first diversification in each case provides the Immediately Relevant

32

Chapter

III

tertium comparationis by specifying the common grammatical categories in terms of auxiliary symbols employed in the rules. The Ultimately Relevant tertium comparationis for syntactic contrastive studies is language-neutral semantic representation, satisfying the requirement that all tertia comparationis must be expressed in terms of concepts outside the compared categories. Also, pragmatic contrastive studies are connected with a hierarchy of tertia comparationis from Ultimately Relevant tertium comparationis to Immediately Relevant tertia comparationis for every specific pragmatic phenomenon. Various sociolinguistic parameters, such as age, sex, education, occupation, etc. can serve as Immediately Relevant tertium comparationis for comparisons of specific linguistic phenomena across languages (Janicki 1985). In such instances, translations will be used as 2texts and pragmatic equivalence will the Ultimately Relevant tertium comparationis. The concept of Immediately Relevant tertium comparationis can be clearly seen in the domain of phonology. As was stated earlier, in phonology, substantive equivalence provides the Ultimately Relevant tertium comparationis in acoustic, articulatory, and possibly also auditory substance. If we focus on articulatory substance, most readily employed not only in contrastive studies but also in separate language descriptions, one can set up the Immediately Relevant tertium comparationis at various levels of "delicacy" For example, comparing consonants in two languages, one must make sure that the Immediately Relevant tertium comparationis delimits the scope of sounds called consonants and brings together for comparison all those sounds which, when articulated in isolation, require occlusion in some place of the vocal tract. At the level of Immediately Relevant tertium comparationis, the contrastive studies will recognize a number of sounds thus articulated in both languages, all of them articulated with occlusion. The contrastive studies will become more and more delicate, when more subtle distinctions are considered. The successive Immediately Relevant tertia comparationis can be evoked by such features as aspiration and its various degrees, voicing and its degrees, force and its degrees, as well as various places of articulation. These features characterize various types and subtypes of consonants in the compared languages. When enough features are evoked, one gets down to comparisons of individual sounds, which are characterized by matrices of articulatory features serving as Immediately Relevant tertia comparationis for specific pairs of sounds contrasted across languages.

Towards a classification

of contrastive

studies

33

Figure 2 only shows Ultimately Relevant tertia comparationis. Each of the seven types of constrastive studies will have its own hierarchy of Immediately Relevant tertia comparationis, which have to be stated and described relative to the relevant Ultimately Relevant tertium comparationis and to the factual data that undergo comparisons. We do not distinguish between pedagogically oriented and pure contrastive studies since we believe that this distinction is irrelevant. Whether directional or adirectional, contrastive studies may yield results relevant to teaching or other fields of application, but the potential implementations of contrastive studies do not, as a matter of principle, determine the course and the direction of the analysis. Modern studies of language and of particular languages (especially the recently flourishing cognitive approaches) offer a very broad perspective within which language is described, not as a set of semanto-syntactic objects called sentences the area roughly corresponding to items (4) and (5) in Fig. 2 — but as a symbolic organization entrenched in human experience and human society. Conducted in this broad perspective, contrastive studies yield results which are naturally relevant to teaching and other practical domains. The usefulness of these studies will be properly evaluated by applied linguists, syllabus designers and teachers. All those interested will be responsible for transforming contrastive studies into pedagogical contrastive studies, or rather for writing pedagogical grammars based on the relevant results of contrastive studies. Contrastive studies themselves will merit the epithet "pedagogical" or "applied" only to the extent to which proper use will be made of them in some practical domains. Pragmatic contrastive studies and statistical (quantitative) contrastive studies have been and will continue to be particularly fruitful in this respect.

34

Chapter III

oo U t/5 ω Ό 3 U > • Ψ-4

-*V -»h C Ο ο >1 00 ο "ο α

Ρ

Iii νa .bo

Chapter IV

Three steps in "classical" contrastive studies

By classical contrastive studies we mean studies in which the contrastive procedures have no generative format of the type suggested in Chapter VIII of this book. Classical contrastive studies, even if they are based on generative grammars of the compared languages, are still taxonomic in nature, since they are limited to producing inventories of differences and possibly of similarities between equivalent systems of the compared grammatical structures, between equivalent sentences and constructions, and between equivalent rules operating at various levels of derivations. Comparison of rules can only be conducted within those frameworks in which the concept of rule appears, as in various versions of transformational generative grammar. A classical contrastive analysis consists of three steps, not always clearly distinguishable in the analysis itself but always tacitly assumed: (1) description; (2) juxtaposition; (3) comparison, i. e., contrastive analysis in the strict sense (cf. Halliday et al. 1964: 113 — 114).

1. Description No comparison is possible without a prior description of the elements to be compared. Therefore, all contrastive studies must be founded on independent descriptions of the relevant items of the languages to be compared. The fundamental demand on such descriptions is that they should be made within the same theoretical framework. It will not do to describe one language in terms of transformational grammar and another language in terms of, say, relational grammar and then to attempt to compare them. The results of such descriptions will be incompatible and incomparable. Not all linguistic models are equally well suited as foundations of cross-language comparisons. It seems that those models which make

36

Chapter

IV

explicit references to universal categories are more suitable than those which are connected with language isolationism, inherent in many variants of structuralism. In Chapter VI, we shall discuss some problems which arise in various descriptive models as related to contrastive studies.

2. Juxtaposition This step is crucial in deciding what is to be compared with what. In classical contrastive studies, this step was based on intuitive judgments of competent bilingual informants, who determined the material to be compared. This sort of "bilingual competence", i.e., the knowledge of two languages, enables one to make decisions about whether or not element X in one language is equivalent with element Y in another language. If the two given elements are equivalent, they are eo ipso comparable. For example, anyone competent in English and in Polish intuitively knows, on the basis of his "bilingual competence" that such words as ktory, ktora in Polish and which/who in English are equivalent, given appropriate contexts. Likewise, "bilingual competence" manifests itself in judging the following pair of sentences as equivalent: (1) (2)

I want John to come. Chc§ zeby Jan przyszedl.

Such judgments are taken for granted in classical contrastive studies, so that elements recognized as equivalent are intuitively deemed to be comparable. One of the obvious weaknesses of this approach consisted in the lack of clearly stated principles underlying decisions about what to compare and why. As was pointed out in Chapter II, formal resemblance and semantic resemblance were resorted to, but both, as we have seen, led to circularities: similarity was presupposed before comparisons yielded results allowing to ascertain it. In many instances, formal resemblance (at least at the level of surface structures) so drastically contrasts with the disparity of meaning that comparisons based on formal criteria alone are reduced ad absurdum. Consider one of the early examples quoted by Stockwell et al. (1965: 40): (3) (4)

English: I said to be sure. Spanish: Dije es tar seguro. Ί said I was sure'.

Three steps in "classical" contrastive

studies

37

Although formally very similar, (3) and (4) are semantically very different. Therefore, they are incomparable since they do not share a semanto-syntactic tertium comparationis. This example shows again that formal considerations alone do not suffice in establishing comparability. Therefore, juxtapositions based on formal criteria alone, though naturally possible, are ill-conceived and must be discarded in contrastive studies. In classical contrastive studies, the investigator himself often acts as the bilingual informant and decides what to compare on the basis of his own knowledge of the two languages. Unless more explicit criteria constraining the data are applied, such a procedure often leads to arbitrary decisions, which seriously undermine the rigour required in scientific investigations. Contrastive Generative Grammar, presented in Chapter VIII, attempts to make explicit "bilingual competence" underlying intuitive judgments of bilingual informants.

3. Comparison proper We distinguish three basic areas of comparisons: 1. Comparisons of various equivalent systems across languages, such as pronouns, articles, verbs, and in phonology consonants, vowels, as well as subsystems, such as nasals, laterals, etc. depending on the degree of "delicacy" of the grammar. 2. Comparisons of equivalent constructions, for example, interrogative, relative, negative, nominal phrase, etc., and in phonology clusters, syllables, diphthongs, and various distributions of sounds. 3. Comparisons of equivalent rules (in those models where the concept of rule appears), for example, subject raising from the embedded sentence, adjective placement, interrogative inversion, passivization, etc., and in phonology assimilation, dissimilation, metathesis, etc. (cf. Sussex 1976: 7 - 1 1 ) . In each area of comparison one of three possible situations may arise: (a)

X Li = X Lj

when item X in Lj may be identical in some respects with an equivalent item in Lj. (b)

X Li Φ X y

38

Chapter

IV

when item X in L; may be different in some respects from an equivalent item in Lj. (c)

XLi = 0y

when item X in Lj has no equivalent in Lj (Krzeszowski 1976: 90). The words "in some respects" are very important. In cross-language comparisons, the relative character of identity must be remembered. Compared items can only be identical with respect to some selected property or properties which they share. For example, the systems of number of nouns in English, French, Polish, and many other European languages are in one respect identical, viz. they are all based on the dichotomy "oneness" vs. "more-than-oneness" Other, more subtle distinctions can also be made by means of numerals and quantifiers, but the grammatical systems of those languages provide morphological means to express just this dichotomy. In many other languages, the system of number is in the same respect different. In Classical Greek, Sanskrit, Lithuanian, and some other languages there is also dual number, which is employed with reference to things coming in twos, usually various body parts coming in pairs. In such languages the system consists of three elements, expressing "oneness", "twoness" and "more-than-twoness" (cf. Hockett 1958: 234). The system of number is sometimes based on the distinction "oneness" "fewness" "manyness", as in most Melanesian languages, the West Indonesian language Sanir and, in the case of personal pronouns, also in Fijian (cf. Hjelmslev 1961: 53; Hockett 1958: 234). In Chinese, nouns have no plural inflection at all, and any concept of plurality is expressed, if necessary, by means of quantifiers and numerals. In contrast with any language in which nouns are inflected for number, Chinese represents the third possibility, i.e., situation (c), distinguished above in which no equivalent form can be attested. With regard to comparisons of constructions we deal with the same three possibilities. An equivalent construction may be identical in some respects and/or different in some respects, or there may be no equivalent construction at all. To illustrate, let us consider passive constructions in English, Polish, German, French, and a number of other languages. These constructions in all these languages are identical in that they consist of an auxiliary verb, usually an equivalent of 'be', followed by past participle forms of transitive verbs. In all these instances the passive constructions are analytic:

Three steps in "classical"

(5) (6) (7) (8)

contrastive

studies

39

English: The song was first recorded by Maria Callas. Polish: Piesn zostala nagrana po raz pierwszy przez Mari§ Callas. French: Pour la premiere fois la chanson a ete enregistree par Maria Callas. German: Zum ersten Mal wurde das Lied von Maria Callas auf Schallplatte gesungen.

In other respects, passive constructions differ in all these languages. For example, French has a complex system of conjugation of verbs where two verbs correspond to one auxiliary verb in English. German requires verb-final word order, etc. New Testament Greek and Japanese passives are different from English passives. They do not involve any auxiliaries but are formed synthetically: (9)

Japanese: Kinoo zyunsa ga sanni korosaremasita. 'Yesterday three policemen were killed'

New Testament Greek and Japanese passives, in comparison with English passives, represent situation (b) where an item in one language is different in some respects from an equivalent item in another language. Situation (c) occurs when we compare English passive constructions with equivalent sentences in such languages as Sumerian and Basque. In those languages, in which no formal distinction is made between transitive and intransitive sentences, there is, consequently, no distinction between active and passive constructions. These languages may be said to have no passive at all (cf. Milewski 1965: 240). 1 In such instances, contrastive studies must state what other means, if any, are employed to convey the semantic content of passive sentences. Within theoretical linguistic frameworks in which the concept of rule is employed, one can also compare rules. Again, as in the case of systems and constructions, three possibilities can be distinguished: A rule in one language may be in some respects identical with an equivalent rule in another language, which means that both operate on identical inputs and that they have identical outputs. For example, the rule of subject raising from embedded object clauses operates identically in English and Latin, rendering the construction which is known as accusativus cum infinitivo: (10) (11)

Latin: Video magistrum ambulare. English: I see the teacher walk.

40

Chapter

IV

In some respects, however, these rules are different in the two languages since they operate on certain verbs in Latin but not on their English lexical equivalents. In Polish the rule does not operate at all. Two rules are equivalent if they operate on identical inputs. If equivalent rules also yield identical outputs, they may be said to be congruent (cf. Marton 1968 b). For example, extraposition of the subordinate clause from the subject operates identically in English and in French: (12)

P R O N O M (itIii) (S) Pred => P R O N O M (itjil) Pred (S)

as in (13) (14)

That the wine turned to vinegar is no surprise. => It is no surprise that the wine turned to vinegar. Que le vin se soit transforme en vinaigre n'est pas surprenant. => II n'est pas surprenant que le vin se soit transforme en vinaigre.

In Polish and Italian, the respective rules are non-congruent with their English and French equivalents as they yield different results. Notably in Polish and Italian the underlying pronouns do not appear in the extraposed sentences: (15) (16)

Ν on e sorprendente che il vino sia diventato aceto. Nie jest niczym dziwnym, ze wino skwasnialo.

Due to the lack of explicitly stated tertia comparationis expressed in some universal terms (such as a semantic representation of the compared items), typical classical contrastive studies were directional: depending on the aims of a particular contrastive analysis, one could start with a description of linguistic forms in L, and match them for comparison with equivalent items in L 2 . Or alternatively, one could begin with a description of linguistic forms in L 2 a n d look for their equivalents in L u Presumably an exhaustive contrastive study covering the entire grammars of both languages has to be bi-directional. Typical directional contrastive studies would bear such titles as {System XI Construction Y) in L, and itsj their equivalents in Lh for example, "English modal auxiliaries and their equivalent constructions in Polish" (Kakietek 1980), "On some properties of action nominals in Polish and their English equivalents" (Lewandowska 1975), "Tags in English and equivalent constructions in Polish" (Oleksy 1977), "The equivalents of the Finnish passive voice in English" (Karttunen 1977), etc. Establishing semantic tertia comparationis creates the possibility of adopting an alternative approach, which consists in selecting a concept

Three steps in "classical"

contrastive

studies

41

and examining the ways in which it is realized through various grammatical means in the compared languages. Typically, such contrastive studies have titles of the form Ways of expressing a category X in L, and Li. For example, "The expression of future in English and Serbo-Croatian" (Kalogjera 1971), "Ways of expressing cause in English and Polish" (Danilewicz 1982) or "Directives in English and Finnish" (Markkanen 1985). Either organization of contrastive studies can be adopted in both "theoretical" and "applied" studies. There are cases of non-directional, applied (pedagogical) contrastive studies (Sharwood-Smith 1975) as well as non-directional theoretical contrastive studies (Kryk 1987). Likewise, there are examples of directional applied contrastive studies (Karttunen 1977) and of directional theoretical contrastive studies (Kakietek 1980). Thus, contrary to what Fisiak et al. (1978) claim, the distinction between "theoretical" and "applied" contrastive studies does not seem to be correlated with the distinction between directional (horizontal) and adirectional (vertical) contrastive studies. Having outlined the scope of grammatical contrastive studies covering comparisons of systems, constructions, and rules, we are now in a position to suggest a set of operational procedures for conducting such studies. Ideally, the procedures should be algorithmic, i.e., they should be possible to follow automatically. The present state of the art does not lend itself to working out such an algorithm. The theoretical requirements for such an algorithm will be described in detail in Chapter VIII. At this point, we wish to sketch a more practical procedure, in which a number of steps have to be taken intuitively, which means that the contrastive analyst will have to employ his knowledge of the compared languages in order to make certain decisions. An overall contrastive theory would have to provide explanation for such choices in terms of explicit rules. Beginning with comparisons of systems, we isolate a system in L, and, having described it, we look for an equivalent system in L2, providing there is an available suitable description of the system. Suppose we set about comparing the systems of personal pronouns in English with the equivalent system in Polish. The English system consists of the following items: / you heI she I it

we you they

42

Chapter

IV

The equivalent Polish system looks as follows: ja ty on/ona/ono

my wy onijone

Comparing the two systems, we immediately notice that in some respects they are identical; namely, in both, distinctions are made between the first, second, and third person pronouns. These grammatical distinctions are based on the semantic distinctions between speaker, hearer, and the rest of the world. Furthermore, in both, distinctions in the systems are made between singular and plural pronouns, although here we also notice some differences (see below). Finally, we also note that in the third person singular, distinctions are made between masculine, feminine, and neuter pronouns. This is where the similarities between the two systems end. We then proceed to look for differences, which are also quite conspicuous. They involve the lack of distinctions in English between singular and plural second person pronoun you in contrast to the distinction made in Polish between the singular ty and the plural wy. Another difference consists in the distinction between virile and non-virile gender in the third person plural in Polish, which contrasts with the lack of the parallel distinction in English. The Polish grammatical distinction oni vs. one reflects the semantic distinction: oni refers to groups of people in which there is at least one male person, while one substitutes for all other nouns in the plural. Comparisons of systems are only of limited utility; they do not yield any information about the ways in which elements of these systems function in well-formed sentences and about their appropriateness in particular communicative situations. Without information pertaining to the semantics and pragmatics of the compared systems, contrastive studies remain drastically incomplete. For one thing systematic correspondences, established on the basis of comparisons of paradigms, are seldom paralleled by syntagmatic correspondences. Thus, although he/she/it systematically correspond to oni ona/ono, respectively, in actual sentences correspondences are often quite different due to the fact that the English system of gender is based on different semantic principles and also due to the existence of various idiosyncratic properties of particular nouns. The gender of nouns in Polish is "grammatical", and every noun belongs to one of the three genders on the basis of morphological properties. In English, the gender is "natural", which means that it is determined by the sex of the objects

Three steps in "classical" contrastive

studies

43

to which particular English nouns refer. The situation is complicated by numerous exceptions in both languages. These exceptions constitute a numerous and diversified class of idiosyncracies (for some analytic details see Fisiak et al. 1978: 60 — 63). Since in sentences the choice of a particular pronoun is determined by the gender of the antecedent noun, the apparent similarity of the systems of pronouns in Polish and English is misleading and often leads to learners' errors. For example, in many cases the English pronoun it corresponds in actual texts to the Polish pronouns on 'he' and ona 'she' rather than to ono 'it' Such is the case in (17) (18)

Zgubilem swoj stary portfel. On byl juz dose zniszczony. I lost my old wallet. It was already rather worn out.

Polish learners of English may, and actually often do, use 'he' rather than 'it' in such contexts as (18). Since, with some marginal exceptions, the selection of the pronoun is completely determined by the grammatical gender of the antecedent nominal, it is impossible to determine which Polish pronoun should be used as the equivalent of the corresponding English pronoun without examining the larger context containing the antecedent. In many instances, English personal pronouns appearing in actual sentences have no direct match in Polish, where person is expressed through verbal inflections alone. This in turn leads to many cases of redundant use of pronouns in Polish by foreign learners, i.e. to the socalled "errors of abundance" (see Chapter X). Moreover, comparisons of systems isolated from contexts cannot, of course, account for differences in the use of various elements of those systems in actual communicative situations. We cannot go into all the details at this point, but it should suffice to note that the circumstances in which you in English is used are not at all parallel to those in which ty or wy is used in Polish. Very often sentences with you in English correspond to Polish third person constructions with Pan 'sir', Pani 'madam', Panstwo 'ladies and gentlemen' The switch to second person forms requires the fulfillment of rigorously defined sociolinguistic conditions, such as obtain between familiars, friends, peers, etc. Only when one describes the particulars concerning differences in the contextual distribution of the compared forms and the extralinguistic factors determining their usage, can one say that a contrastive analysis is reasonably complete and has pedagogical relevance. 2 From the methodological point of view, situation (c) described above, in which an item X in Lj has no equivalent in Lj presents a problem: if there is no equivalent to compare, is it still possible to compare? The

44

Chapter

IV

problem arises most sharply in the comparison of systems. Such is the case with English articles, which cannot be juxtaposed with any single system in a number of languages. In order to see what articles can be compared with, we have to resort to the examination of construction equivalents to see through what other means, if any, the semantic content of articles is expressed. Without going into detail, let us assume that the basic semantic distinction that the English articles express is that between definiteness and indefiniteness. (In fact the problem is much more complex, but for the sake of illustration of the methodological problem in contrastive studies, we will take this simplified view of the semantics of English articles). In many instances, Polish constructions contain no indication of any distinction between definite and indefinite noun phrases: (25) (26)

Kupilem ksigzkg. I bought a book.j I bought the book.

Sometimes, however, definiteness and indefiniteness in Polish is expressed by means of definite and indefinite pronouns appearing in the function of noun determiners: (27) (28) (29) (30) (31) (32)

Nawet polubilem tego cziowieka. I even began to like the man. Jakis czlowiek tu chce sig ζ tob@ widziec. There is a man here to see you. Raz zyl pewien krol w odleglej krainie. Once there lived a king in a far-away land.

Word order, too, may signal definiteness and indefiniteness in Polish. Szwedek (1981: 53 — 54) convincingly demonstrates that (33)

W pokoju siedzial chlopiec. 'In room was sitting boy' Chlopiec wyszedl 'Boy went out'

can be rendered in English as (34)

There was a boy sitting in the room. The boy went out.

Indefinite and definite articles in English express what in Polish is expressed by means of the word order in (33) in contrast to, for example, (35)

Widzialem na ulicy siedzqcego chlopca. Ρο chwili wstal i podszedl do mnie chlopiec.

Three steps in "classical"

contrastive

studies

45

which is either incoherent or could possibly mean that a different boy is referred to in the second sentence of (35), in which case (35) could be rendered in English as (36)

I saw a boy sitting in the street. After a while a (different) boy got up and walked up to me.

In the same work, Szwedek explores some ways in which sentence stress is involved in the expression of coreference connected with definiteness. Thus, we see that in order to find Polish equivalents of English articles we have to go beyond a contrastive analysis of systems and broaden the scope of investigations to cover the territory of text grammars and discourse analysis. Our subsequent discussion will provide more examples to support the, thus far rather tentative, conclusion that an adequate contrastive analysis must reach beyond the domain of sentence grammars towards text grammars and pragmatics.

Chapter V

Contrastive studies at various levels of linguistic analysis

Most theoretical orientations, notwithstanding some terminological differences and disagreements in matters of detail, distinguish the following components of language, which correspond to various levels of analysis: phonetic/phonological, morphological, syntactic, lexical, semantic and pragmatic. The terms mentioned here do not constitute a coherent system in a definite metalanguage. They can be found in various textbooks on linguistics which refer to various components of language, and they are not always compatible with other systems. However, regardless of their theoretical orientation, these levels must be recognized and considered by an adequate theory of language, even if some theories seem to favour a particular component over other components (for example, transformational-generative grammars focus on syntax). Particular theories differ with respect to how they subdivide these levels as well as to how, if at all, they relate the levels to the world at large, i.e., to extralinguistic reality. There is also a rather fundamental disagreement as to whether these components are autonomous, as most structuralist and transformational theories claim, or whether the boundaries between them are fuzzy, as the proponents of the so-called cognitive grammars claim (e.g., Langacker 1987: 7). In some theoretical frameworks, the syntactic component of language is subdivided into phrase-structure rules and transformations, while other theories do not recognize the transformational component at all (Chomsky 1965 vs. Hudson 1976). In some theories, a distinction is drawn between sentence semantics and word semantics (e.g., Bartsch — Vennemann 1972), while in other theories the semantic component is integrated with the syntactic component to form the so-called semanto-syntax (cf., for example, generative semantics as formulated by McCawley 1968; Lakoff 1971; Postal 1970). In this book, we shall assume there is a distinction between sentence semantics and word semantics. Consequently, problems connected with sentence semantics will be dealt with in connection with lexicon. Thus,

48

Chapter V

we are left with three basic components of language (and corresponding "levels" of analysis): phonological (including morphophonology), grammatical (including sentence semantics) and lexical (including word semantics). It is evident that an analysis of these three components represents a drastically limited view of language since it does not treat language as a multifunctional system, and therefore does not offer any insights into how language actually functions in extralinguistic settings. In other words, the approach to linguistic analysis through the components distinguished above provides a framework to analyse but does not provide insights into how this structure is used to perform its numerous functions. Linguists do not always agree on whether or not they should include the functional aspects of language in their analyses (cf. different versions of structuralism discussed in Apresyan 1966, 1971). Many structuralist models of language, as well as Chomsky's standard theory, treat language in complete abstraction, divorced from concrete situations in which the language systems are realized. As a matter of fact, the long-standing tradition originating with de Saussure and labelled as structuralism (in most versions) explicitly abstracts language from its concrete realization and thus adopts de Saussure's sharp distinction between la langue and la parole, claiming that linguistics should deal only with the former. One can conduct contrastive analyses of such abstract systems, and this approach basically characterizes "theoretical" or pure contrastive studies. Practical validity of such studies, however, is seriously limited. Moreover, many linguists, opposing the structuralist tradition, claim that such limited approaches to language overlook the very essence of what language is all about: its communicative function (cf. Hymes 1972: 292; Searle 1971: 29—40). Linguists of this persuasion include in their investigations the relations between language as a system of signs and situations in which it is used. This area of investigation is usually called pragmatics (cf. Morris 1938). The importance of pragmatics will become particularly clear in the chapter devoted to pedagogically oriented contrastive studies, but in order to make the discussion of various aspects of language reasonably complete, we shall also consider pragmatics in the present section. Therefore, we add pragmatics to the previously distinguished areas phonology, grammar, and lexicon. An adequate view of language must consider the mutual interdependence among these components and possibly the lack of clear-cut boundaries between them. On the other hand, whatever rules operate at these levels must be focused on and stated separately for each level. Linguists

Various levels of linguistic

analysis

49

have a fairly clear idea of the nature and form of rules operating at the phonological and syntactic level. It is not equally clear what kind of rules can be formulated for the lexical level and even less clear at the pragmatic level. All the same, there is evidence that at all these levels there are certain facts which can be formulated in terms of rules, while other facts simply have to be listed. To the extent to which such regular facts do exist, the respective rules can be described and compared at each level of linguistic analysis. That rules operate in all four components and that these rules are to some extent independent can be best seen if we examine some deviant utterances and try to identify the source of deviance in terms of rules which have possibly been violated: (1) (2) (3) (4)

My father distrusts psychology/psai'kohd^i/. — phonological deviance Distrust my father psychology? — syntactic deviance My father distrusts psychologies. lexical deviance My father fucks up psychology. possible pragmatic deviance in a specific situation

In (1), the phonological rule which bars initial clusters consisting of a stop followed by a sibilant has been violated. Such a violation is likely to occur in the speech of foreigners attempting to speak English, particularly, if there is no analogous constraint in their native language. In (2), the transformational rules of question formation have been misapplied. The entire verb instead of just the tense has been inverted and neither the agreement transformation nor the Jö-support transformation have been applied. Whether or not these failures may be due to the interference of the native language is not a relevant issue at this point, but whatever rules have been violated in (2) belong to the syntactic component of the language. In (3), there appears a lexical item which is not part of the English lexicon. It is not clear whether this case can be handled in terms of a rule that has been violated. One could surmise, however, that false analogy is at work here. Due to this sort of overgeneralization psychologies is formed by analogy with physics, mathematics, etc. In (4), the deviation is not overtly detectable in the structure of the sentence itself or in the ill-formedness of its constituents. The sentence in (4) can only be seen as inappropriate with reference to a specific situation in which it is uttered, even if phonologically, syntactically and lexically it is perfectly well formed. In the sections which follow, we shall present an overview of topics which have to be dealt with within various components of language in a reasonably complete contrastive analysis.

50

Chapter V

1. Topics in phonological contrastive studies Contrastive studies at the phonological level are strictly determined by the theoretical framework adopted in the description of the compared languages. The specific linguistic model determines not only what to compare: segments, features, rules, or representations; it also delimits the upper and the lower bounds of such studies, i.e., morphophonemic alternations and phonetics, respectively. The model, furthermore, determines the degree of refinement of the analysis or the number of details covered by the analysis. Finally, the model determines the level of abstraction at which the contrastive analysis is most usefully performed. For example, Gussmann, within the generative framework of abstract phonology, claims that the level of underlying representation is the only level where features and segments may be legitimately compared (Gussmann 1978: 158) He, consequently, maintains that comparisons on the level of phonetic representation seem to be superfluous since phonological rules will provide the necessary specifications of phonetic properties of morphemes and the ways in which phonetic representations differ from phonological ones. Thus, the aspiration of /1/ in team will be described by an appropriate rule in the same way as the fact that /t/ is realized phonetically as [s] in partial and as [c] in departure·, the fact that the Polish /t/ in trzeba 'it is necessary' is always alveolar and that it is realized phonetically as [6] in locie 'flight' (dat. sing.) and as [c] in lec§ Ί fly' will be stated similarly. Therefore, Gussmann concludes: We believe that there is absolutely no need for separate comparison of phonetic representations since at best it would merely repeat information provided by the comparison of rules We conclude that a contrastive phonological study of two languages should comprise a comparison of features and segments at the level of underlying representation, morpheme structure conditions and of phonological rules (Gussmann 1978: 159) Gussmann's radical view is not shared by a number of outstanding contrastive linguists. Notably, Fisiak (1975) stresses the relevance of contrastive studies at the level of phonetic representations, particularly for pedagogical purposes: applied pedagogically oriented PCS (phonological contrastive studies) will contain maximum information about the low phonetic rules (e. g.

Various levels of linguistic analysis

51

voice assimilation) and phonetic features and segments with very little (if at all) abstract ("deep") phonological information (Fisiak 1975 a: 344). Scepticism about psycholinguistic reality of abstract phonology and about possible pedagogical uses of contrastive studies limited to underlying representations and rules is also voiced by Awedyk (1976: 65 ff. 1979: 125 ff., 131). It seems that the reductionism advocated by Gussmann, though possibly admissible in some theoretical frameworks of limited psycholinguistic and pedagogical relevance, cannot be accepted in any descriptions of a natural language aspiring to the status of adequate renderings of psycholinguistic reality. Particularly unwelcome is the reductionism which breaks language down into rules, their inputs, and/or their outputs; the devastating consequences of such approaches are best seen in contrast with cognitive approaches to language. Since language is a complex organization of symbols at a number of hierarchically arranged levels of analysis, with each level performing a distinct communicative function, one must carefully describe each level in the search for what contributes to the overall functional mechanism of language. It will not do to reduce the description of one level to properties of another level and to the rules of combination since such a procedure deprives the object of description of its most relevant properties. In the following paragraphs, we shall survey the problems which arise at various levels of linguistic analysis, particularly in the context of contrastive studies. In doing so, we shall highlight the shortcomings of reductionism in order to justify the claim that an adequate contrastive analysis must be conducted independently at every level of language organization since it is possible and necessary to focus on what each level uniquely contributes to the organization of linguistic communication. In the description of speech sounds, it is essential to focus as much on their mutual relationships and the functions which they perform in the phonological organization of language as on their substantial properties, insofar as these properties are involved in the language symbolism and hence in communication. Naturally, an adequate contrastive analysis will embrace all these aspects of speech sounds. In what follows, we provide an overview of topics which have to be dealt with within various components of language in a reasonably complete contrastive analysis. It must be noted, however, that due to the limitations of current linguistic frameworks, as well as due to the limi-

52

Chapter V

tations of contrastive studies themselves, only some of these topics have been worked out so far; many are still waiting to be dealt with, and some cannot be handled before more complete linguistic descriptions become available. Contrastive studies concerning speech sounds and their arrangements embrace segmental and suprasegmental phenomena.

1.1. Segmental phenomena In classical contrastive studies, phonemes and their variants can be compared across languages either in isolation from other sounds or as patterns of linear arrangements. According to Lado, three types of questions can be asked about individual sounds compared in isolation: (1) Does the native language have a phonetically similar phoneme? (2) Are the variants of the phonemes similar in both languages? (3) Are the phonemes and their variants similarly distributed? The first of these questions concerns a more general problem of establishing equivalent sounds across languages. Phonetic similarity as tertium comparationis usually means similarity of articulation in terms of manner and place. For the purposes of contrastive studies, this kind of similarity is usually the only relevant thing to consider. But it is also possible to consider acoustic similarity, which could be described with reference to the structure of spectrographic images of sounds used in the compared languages. Similarity of sounds could also be described with regard to the auditory aspects of sounds, i. e., the ways in which they are heard by listeners. This latter aspect of phonetics may have relevance in contrastive studies inasmuch as sounds articulated in a different manner may have similar acoustic properties and may be perceived as alike by foreigners (Lamminmäki 1979: 180 ff.). For example, some renderings of English dental fricatives /Θ/ and /δ/ may be confused with labio-dental fricatives /f/ and /v/, especially by those listeners in whose native language the first two are missing. In spite of the different places of articulation acoustic properties of /Θ/ and /f/ and of /δ/ and /v/, respectively, are often not sufficiently distinct to ensure proper discrimination of these sounds (cf. Jassem 1983): The two fortis fricatives /f/ and /Θ/ are acoustically and perceptually very similar. This also goes for the two lenis fricatives /v/ and /δ/ (Jassem 1983: 222).

Various levels of linguistic

analy.

53

In the majority of cases, phonological contrastive studies are confined to the examination of the articulatory features of speech sounds. Consequently, phonetic similarity usually means similarity of articulation, one of the possible substantial tertia comparationis. Therefore, one can identify equivalents of the English stops /p, t, k/ in various languages on the basis of the identity of place and manner of articulation, notwithstanding the fact that certain variants of these sounds have different articulatory features in various languages. For example, the English /p, t, k/ are aspirated in word and syllable initial positions in front of stressed vowels. Their Polish equivalents are not thus aspirated. In some languages, such as Sanskrit, there is a set of aspirated stops contrasting with unaspirated ones. In a language in which aspirated [p] contrasts with unaspirated [p], the two belong to two different phonemes, even if in some sense they represent one sound, described as a bilabial stop. The perception of the two sounds as belonging to one common type seems to be connected with the importance of particular features in determining category membership. Place of articulation seems to be the most important in the articulation of stops. Typically, the areas involved are: labial, dental-alveolar, and velar. They are respectively associated with the articulation of three basic types of stops [p], [t] and [k], and their voiced, or perhaps weak, counterparts [b], [d] and [g]. It appears that no other articulatory property, like aspiration, palatalization, or labialization, which may accompany the articulation of these consonants, is capable of changing their status as some type of p, t, k, etc. Whether aspirated, palatalized or labialized they remain ρ, t, k, etc., respectively, even if these secondary features bring about phonemic contrasts. To some extent such intuitive, "naive" and "natural" judgments concerning the categorization of the sounds in question, regardless of the language, may be supported and reinforced by the alphabets and the spelling conventions in various languages, whereby various kinds of labial stops are spelled as " p " and " b " no matter what other features they have. Thus, at least in the case of stops, the spelling conventions seem to highlight whatever psychological reality can be attributed to this sort of categorization. To some extent what was said above is also true of fricatives. Barring problems with sound discrimination, as in the case of /f/ and /Θ/, etc., discussed above, in a number of cases fricatives are identified as /f/, /Θ/, and /x/, as long as they are respectively articulated in the major articulatory areas, viz., labial, dental, and velar. Minor articulatory differences, accounting for such sounds as the Spanish bilabial [φ], differing from the English labio-dental [f], or for intradental versions of [Θ], are not signi-

54

Chapter V

ficant enough to provide grounds for recognizing such articulatory variants as belonging to other types of sounds. Consequently, labial fricatives, whether they are bilabial as in Spanish or labio-dental, as they usually are in English, will be recognized as equivalent on the basis of the identity of their primary articulatory features, i. e., the labial area and the friction accompanying the articulation (cf. Kopczynski 1973). The lack of phonetic equivalents of some foreign sounds in one's native language is said to be a source of major difficulties in learning a foreign pronunciation. Phonological contrastive studies immediately reveal these differences. For example, in comparing Polish and English speech sounds, one finds that in English there are no equivalents of such Polish sounds as /te/ in cienki 'thin', jdzj in dzien 'day', /e/ in sien 'hall', jzj in ziemia 'earth' or /x/ in chata 'cottage' Likewise, in Polish there are no equivalents of the English /Θ/ as in thin, /δ/ as in though, or /h/ as in ham. Such lists of differences in the phonemic inventories can be found in any contrastive studies at the phonetic level (cf. Moulton 1962: 27; Delattre 1965: 155-203; A g a r d - D i Pietro 1965: 2 8 - 3 4 ; Stockwell - Bowen 1965: 1 6 - 1 9 ; Kopczynski 1973; Krzeszowski 1970: 54). It is important to note that lists of non-equivalent phonemes are determined by the specific phonological analyses of the data in the compared languages. For example, there exist two different analyses of Polish consonants, resulting from two different inventories not only of consonants but also of vowels. Under one analysis most palatal consonants are recognized as phonemically distinct from non-palatal ones. For example, Stieber (1966: 117) recognizes the following 35 consonant phonemes in Polish: ρ

b f

p'

b ' f

v'm'

t

d

c* 3

s

ζ

c

3

s

ζ

c

3

s

ζ

k

g

ν

m

η

r

I

I

ή**

χ

k'g' j

*

In Stieber's transcription [c] stands for the dental voiceless affricate rather than for the IPA's palatal voiceless plosive. ** In Stieber's transcription ' marks both palatal and palatalized sounds.

Various levels of linguistic analysis

55

The alternative analysis classifies [p1, b', f , v', m1] as positional variants of the corresponding non-palatal consonants, i.e., [p, b, f, v, m], respectively, automatically reducing by five the number of non-equivalent phonemes in English (cf. Jassem 1962: 7 - 1 5 , 1 9 6 6 : 8 7 - 8 8 ; Biedrzycki 1964: 26; Puppel et al. 1977: 90 ff.). U n d e r this analysis, the five palatal, or rather palatalized, consonants appear only in f r o n t of the high f r o n t vowel jij and the semivowel /j/, while the non-palatalized variants appear everywhere else. This analysis of consonants demands a re-analysis of vowels. U n d e r the first analysis we deal with a five-vowel system, consisting of /i, e, a, o, u/, where j\j has two variants (allophones): a higher, tenser and more fronted [i], as in piwo / p'ivo/ 'beer' and a more retracted, relaxed and lower [i], as i n p y t a c j p i t a t e / The first variant appears after palatal consonants while the second variant appears after other consonants. The second analysis, which recognizes some palatal consonants as positional variants of the non-palatal ones, requires the recognition of [i] and [i] as phonemically distinct to provide phonemic contexts for the mutually exclusive variants of the consonants. Thus, the palatalized variants appear in f r o n t of j'xj and the semivowel /j/, while the nonpalatalized variants appear elsewhere. Consequently, under the two analyses we will have the following different transcriptions of words in which the relevant sounds appear: Analysis I bicjb 'itej vs. bycjbitel 'beat' 'be' Analysis II bicjbite! vs.

byc\bite\

U n d e r the first analysis, the English vowel phoneme /i/ will have no phonemic equivalent in Polish, since the Polish vowel phonemes constitute the set: u e

ο a

in which there is only one high non-back vowel. U n d e r the second analysis there are two high non-back vowels in Polish, i. e., j\j and /i/: i u e a ο

56

Chapter V

Following this analysis, it is possible to match the Polish /i/ of bic /bite/ with the English /i:/ of beat as equivalent, and similarly the Polish j\j of bycjbitQj with the English /i/ of big. A contrastive study based on the first analysis will reveal a larger number of Polish consonant phonemes which have no phonemic equivalents in English than a contrastive study based on the second analysis, inasmuch as the first analysis results in the recognition of five palatal consonants as distinct phonemes, which the second analysis interprets as positional variants (allophones). On the other hand, a Polish-English contrastive study based on the second analysis will reveal a smaller number of non-equivalent Polish vowel phonemes in comparison with their English counterparts, since both the English f\:j and /[/ will now have their phonemic equivalents in Polish. Thus, under the second analysis, the Polish and English phonemic systems will appear more similar than under the second analysis, since the first analysis reveals more differences in the inventories of both vowels and consonants. Alternative analyses are also possible for English. For example, some phonologists would claim that the vowel sound in beat is actually a diphthong consisting of a variant of j\j followed by the semivowel j]j (e.g., Jassem 1954: 74, 1983: 184). Consistently, the vowel sound in food is phonemically interpreted as /uw/. This approach allows us to reduce the number of English vowel phonemes to the following seven u e a ο ε a Obviously, if this sort of analysis of the English data is adopted as a framework for contrastive studies, the number of non-equivalent Polish phonemes will be smaller than if an alternative analysis, recognizing twelve simple vowel phonemes in English is adopted (see, e.g., Gimson 1962: 86 ff.). The fact that such divergent results of competing analyses are possible is the sign of weakness in classical phonological theories. The deficiency consists in the lack of theoretically motivated and empirically verifiable criteria permitting one to decide which of the descriptively adequate models is to be preferred over the alternative descriptions, assuming that the models in question are descriptively adequate. (For a discussion of weaknesses of classical phonology, see Postal 1968: 3 — 18). In other words, classical phonological theories do not achieve the level of explan-

Various levels of linguistic

analysis

57

atory adequacy. Hence, one or another mode of description can be given priority only on either intuitive or purely arbitrary grounds. The foregoing discussion leads to the conclusion that Lado's first question pertaining to phonological contrastive study, i.e. "Does the native language have a phonetically similar phoneme?", cannot, in many cases, be answered in unequivocal terms due to the lack of clearly stated criteria determining what constitutes phonetic similarity and due to the existence of alternative phonemic analyses resulting in alternative inventories of phonemes, which in turn leads to alternative inventories of differences and similarities across languages. The failure to answer the first question satisfactorily is naturally connected with similar difficulties in supplying answers to the next two questions. Failure to identify equivalent phonemes across languages automatically rules out any possibility of comparing equivalent allophones of those phonemes as well as the distribution of allophones. A phonological contrastive study seems to be better off if the concept of phoneme is bypassed, at least in those cases when exact matching of equivalents is impossible. In such cases, the contrastive study involves comparisons of articulatory (and possibly acoustic) properties of sounds regardless of their place in the overall system. This is presumably what Lado himself anticipated when he wrote: In comparing the sound system of a foreign language and native language I find it good safe practice to take up each phoneme separately regardless of any general patterns of difference I may have observed (Lado 1957: 13). We shall take up the problem of equivalence at the phonetic level and of the ensuing problems in the chapter devoted to structural contrastive studies. At this point, let us only observe that contrastive studies limited to a comparison of equivalent phonemes across languages are inadequate as they often obscure conspicuous differences in the realization of these phonemes in particular phonetic contexts. This problem will also be taken up in due course. Segmental phonological studies are not confined to paradigmatic listings of sounds and their equivalents in the compared languages. A complete account also includes comparisons of syntagmatic relations between sounds, i.e. various combinations of sounds called diphthongs and consonantal clusters. Contrastive studies must list similarities and differences between specific constraints operating in the compared lan-

58

Chapter V

guages with respect to admissible combinations of sounds as they constitute syllables, morphemes, and words. Let us consider clusters as a characteristic phenomenon in syntagmatic relations between speech sounds. Insofar as the number of sound types in specific languages is finite and, insofar as there are limitations on the number of consonants in a cluster, the number of clusters is finite in a given language. Hence, one may find it superfluous to look for any rules governing the structure of clusters since all clusters can be listed. Yet, it is tempting to formulate generalizations (and consequently rules) generating possible clusters and ruling out impossible ones. (See, for example, Gussmann 1980: 54.) Syntagmatic phonological contrastive studies will also include comparisons of syllabic structures. For example, Awedyk (1974: 94) analyzes structures of syllables in Polish and English and draws the following conclusions from his contrastive studies: 1. Concerning the structures of nuclei. In English the nucleus of the syllable may be either simple or complex. In Polish it is always simple. The English nucleus may consist of the resonants /r, 1, n, m/. In Polish the nucleus must consist of a vocalic phoneme. 2. Concerning the structure of the onset. English onsets may consist of from zero to three positions. Polish onsets may contain from zero to four positions. In English only /s/ can occupy position 0 3 (the first one in a sequence of three consonants). Different constraints affect other positions. 3. Concerning the structure of the coda. In both English and Polish codas may consist of from zero to four positions. In English /s, t/ occur only after the morpheme boundary. 4. Concerning the structure of the interlude. In English the interlude may consist of from zero to four positions. In Polish the interlude consists of from zero to five positions. It is worth noting that Awedyk's results are influenced by the kind of descriptions that he adopts for the two languages. Whereas for Polish he adopts Jassem's analysis referred to above (Jassem 1954), for English he adopts Gimson's description recognizing /i:/ and /u:/ as monophthongs rather than as diphthongs /ij/ and /uw/, respectively, as in Jassem (1954, 1983). If Awedyk had adopted this latter analysis, he would have come out with different numerical results concerning the number of complex and simple nuclei in English (cf. Awedyk 1974: 91). Likewise, his claim

Various levels of linguistic analysis

59

that "in Polish it [the nucleus] is always simple" (Awedyk 1974: 94) would have proved to be false if he had adopted Biedrzycki's view that the Polish nasal vowels [?] and [q] are, in phonological terms, nasal diphthongs /eg/ and /οη/, respectively (Biedrzycki 1964). As diphthongs, they constitute complex nuclei in such Polish words as k§s/kegs/, wpz/v053/ and

konski/kojiskij.

Awedyk's analysis is another example of the extent to which the results of contrastive studies are determined by the descriptive frameworks adopted in the analysis of the materials in the languages that undergo comparison. This sort of dependence also has a bearing on the results of contrastive studies vis-ä-vis pedagogical applications. Both these aspects will be taken up in the appropriate parts of the book, i. e. in Chapters VI and XII.

1.2. Suprasegmental phenomena Phonological contrastive studies must also embrace suprasegmental phenomena, stress and intonation. There exist numerous descriptions of English stress, rhythm and intonation employing a great variety of notations, which use numerals, strokes, accents, dots, and other graphic devices. The situation is further complicated by the lack of parallel descriptions for other languages. For these reasons any contrastive studies of these phenomena are still very much a pioneering endeavour. (But see the studies by Ozga 1974; Mackiewicz-Krassowska 1974; Woloszyk 1974; Marek 1974; Mieszek 1974; Varga 1975; Stockwell - Bowen 1965; A g a r d - D i Pietro 1971). The lack of rigorous criteria in the description of suprasegmentals is even more evident than it is in the case of segmental phonemes. Therefore, the consequences in the realm of contrastive studies are even more conspicuous: there is a lack of extensive and reliable contrastive studies of suprasegmental phenomena. The available descriptions and comparisons suffer from lack of uniformity, rigour, and reliability. The problems are compounded by the multiplicity of notational conventions, which still only conceal more deeply grounded differences in the description of linguistic data. For example, Stockwell — Bowen (1965: 21), following Trager —Smith (1957: 37), distinguish three levels of English stress, which they call "strong", "medial", and "weak" Gleason (1955: 4 0 - 4 2 ) , Francis (1954: 152), and many others distinguish four levels of stress in English.

60

Chapter V

Following the latter group, Fisiak (1978) speaks of four levels of stress in English and distinguishes also four levels of stress in Polish: Both in English and Polish at least four degrees [of stress] can be recognized (Fisiak et al. 1978: 242). Lado distinguishes as many as five levels of stress in English (Lado 1957: 28). Opinions concerning other languages are less diversified, perhaps because less attention has been paid to matters of stress in terms of structural phonemic analysis, widely practised in the United States. Still, in one of very few works devoted to suprasegmentale in Polish only two levels of stress are mentioned, viz., "main" and "subsidiary" (Dtuska 1976: 2 0 - 2 8 ) . Thus, Dluska tacitly contradicts both Fisiak et al. (1978) and Ozga (1974: 133). In the latter paper, reference is made to at least three levels of stress, though the author does not explicitly mention any definite number. In this situation no genuine contrastive studies of stress can be attempted. Whatever has been done so far consists in juxtaposing independent descriptions of the data in two languages, without the slightest attempt to find any tertia comparationis nor to formulate any generalizations about differences and similarities of the compared phenomena. Fisiak's two-page contrastive study of stress in English and Polish ends with the significant sentence: Because the stress in English may fall in so many different places in the word, naturally the rules of stress assignment have to be more complicated there than in Polish (Fisiak et al. 1978: 243). Future contrastive studies in this domain will have to discover both the rules in the two languages and the degree of their "complication" For the time being, it seems, it is best to refrain from any further comments on the subject. The state of the art in the domain of intonation is no better, which is not surprising, since all suprasegmental phenomena, i.e., intonation, and pauses, are closely interrelated (Ozga 1974: 127). Again there is no agreement as to the number of basic patterns. Whereas MackiewiczKrassowska (1974: 138 — 139) assumes six basic tunes in both Polish and English, Fisiak et al. (1978: 244) speak of twelve patterns in the two languages. Fisiak's analysis is confined to listing examples of the twelve patterns in the two languages with the indication of their meanings in terms of such qualifying epithets as "cool", "reserved", "dispassionate",

Various levels of linguistic analysis

61

"calm" "serious", "querulous" etc. The review ends with a rather obvious conclusion that there is no one-to-one correspondence between an intonation pattern and an emotional attitude either in English or in Polish (Fisiak 1978: 250). He immediately adds that Intonation is assigned to particular types of syntactic constructions by well defined rules which use both semantic and syntactic information (Fisiak 1978: 250). Unfortunately, the analysis ends there and not a single rule is presented. More detailed analyses of intonation in Polish and English are presented in a series of papers devoted to declarative sentences, interrogative sentences, imperative sentences, and compound sentences, respectively, by Mackiewicz-Krassowska (1974), Woloszyk (1974), Marek (1974), and Mieszek (1974). Although all these papers contain appeals for more research, the authors are able to draw some specific and definite conclusions from their contrastive studies. Thus Mackiewicz-Krassowska (1974) concludes her contrastive study of declarative sentences with the following statement: As this brief analysis of Polish and English declarative sentence intonation shows, the falling intonation is used in Polish for the majority of cases. In Polish more words are uttered on a level tune which evokes the effect of monotony to some non-Polish speakers. Moreover, an average Pole's pitch of voice does not go as high as that of an Englishman is. This also accounts for the relative paucity of those characteristic ups and downs heard in an English utterance. As we have seen from the above examples, the pitch patterns of English and Polish statements are on the whole different. The only similarity in the intonational pattern can be noticed in straightforward statements and the patterns used for enumeration. Even then, the similarity is only apparent, as the Polish tunes, the high fall especially, are lower than those of English (Mackiewicz-Krassowska 1974: 143 — 144). With respect to interrogative sentences Woloszyk has discovered that the intonation of questions in English and Polish has some certain common features. The accentual function of intonation may be quoted as the example, since there are certain common rules, one of which may be the change of pitch on the stressed syllable (Woloszyk 1974: 157).

62

Chapter V

This is again where firm conclusions end. Everything else appears to be rather muddy: English intonation, however, in its non-accentual function is to a certain degree [emphasis my own] fixed. Polish intonation depends largely [emphasis my own] on the speaker's emotional attitude and sometimes on his origin (Woloszyk 1974: 157). The comparative problem which arises here, but which Woloszyk does not tackle, is this: since in both English and Polish intonation is to some extent fixed (and consequently to some extent unstable or varied), what exactly is the extent of the variation in both languages and what exactly determines the degree of stability of intonation patterns in both languages? Further contrastive studies must provide answers to these questions. Since there are still many unresolved problems in the description of intonation in both languages, the answers may not be readily forthcoming. For example, opinions still vary as to whether English has a special tune characterizing questions, as is claimed in Gimson (1962: 256 ff.). Yet many questions have a falling intonation, in spite of theoretical predictions but in keeping with emotional attitudes of the speakers. In Polish, too, all types of questions which usually end in a rise may end in a fall when there is a great emotional load, and vice versa those usually ending in a fall may end in a rise when the speaker wants to express his feelings fully (Woloszyk 1974: 158). Marek's and Mieszek's papers end with similar notes of uncertainty. Marek states that: Intonation is not an independent system which can modify or change the meaning of a sentence with which it occurs. It reflects the underlying, intended [emphasis my own] meaning and is subordinate to it There is no separate 'normal' intonation pattern for imperative sentences. Each command will have that intonation which can bring about the desired meaning (Marek 1974: 178). Mieszek ends her contrastive study of compound sentences in English and Polish with the following words: It is evident, however, that various kinds of intonation patterns expose different feelings and connotations that add considerably to the literal

Various levels of linguistic

analysis

63

contents of the sentence. Therefore the expressive function of intonation is relevant in any case (Mieszek 1974: 187).1 It seems that the research in this area can advance substantially only if pragmatic and cognitive functions of intonation are incorporated into the description. Specifically, one needs to consider the basically metaphorical and iconic aspect inherent in the orientation U P and D O W N as involved in the explication of such concepts as U N K N O W N ( = U P (in the air)) vs. K N O W N ( = D O W N (within my reach)), U N C E R T A I N ( = UP) vs. CERTAIN ( = DOWN), G O O D ( = UP) vs. BAD ( = DOWN), etc. Systematic research into the iconic relationships between these metaphorical concepts and the pitch level in sentences may provide answers to some of the hitherto mysterious problems and, consequently, cause a breakthrough in both descriptive and contrastive linguistics. Some preliminary suggestions along these lines and concerning possible relationships between orientational metaphors and syntax have already been made by Lakoff—Johnson (1980: 126 ff.).

2. Topics in syntactic contrastive studies Regardless of the adopted model, syntactic contrastive studies must embrace basic sentence patterns, major sentence constituents, i. e., noun phrase and verb phrase, various functions of noun phrases as subjects, objects, complements, and modifiers and of verb phrases as predicates. The analysis of sentence constituents must embrace both their internal and external structure. The internal structure of a constituent embraces the hierarchical relations which occur between subconstituents of a particular constituent. For example, in English, but not in Polish, the constituent called noun phrase must contain an obligatory subconstituent called determiner, which in turn can be analysed into subsubconstituents, etc. In Polish the presence of a determiner in the noun phrase is optional, and, insofar as this is so, the internal structure of noun phrases in English and Polish can be said to be different. Another difference relating to the internal structure of constituents may concern word order. The order of subconstituents of both noun phrases and verb phrases in English is comparatively more rigid than in Polish. Thus, whereas in English only my first son is grammatical, while *first my son is not, in Polish both moj pierwszy syn and pierwszy moj syn are grammatical. Yet another difference

64

Chapter V

concerning internal structure of constituents may involve co-occurrence restrictions on subconstituents. For example, within the English determiner, possessives and demonstratives, in most dialects of English recognized as standard, are mutually exclusive, which means that they cannot both occur as determiners in the same noun phrase. N o such restriction exists in Polish. Therefore, in English * these his children is ill-formed, while in Polish te jego dzieci is well-formed. The external structure of a constituent refers to the hierarchical relationship that it enters with other constituents to combine into more complex linguistic forms. For example, nominal subordinate clauses can appear as constituents of main clauses, namely as subjects, direct, indirect and prepositional objects, subject and object complements, modifiers and appositives. Such nominal clauses can also be conjoined with other nominal clauses to form coordinated nominal clauses. What is relevant from the contrastive point of view is the fact that there may be different restrictions on the occurrence of nominal clauses in some of these functions across languages. For example, in English, unlike in Polish, certain types of nominal subordinate clauses can function as objects of prepositions. Compare: She was right in what she said. *Ona miala racjg w czym powiedziala. They couldn't agree on whether they should go. *Oni nie mogli zgodzic si§ na czy powinni tam isc. In such cases, Polish (and many other languages) requires a supporting pronoun or a noun phrase into which the nominal clause is embedded as a relative clause: Ona miala racj§ w tym, co powiedziala. Oni nie mogli sie zgodzic na to, czy powinni tam isc. The distinction between internal and external structures suggests one possible strategy in looking for syntactic contrasts across languages. Fisiak et al. (1978) distinguish the following types of differences in comparing equivalent sentences (constructions) across languages: (a) structural, (b) categorial, and (c) functional. Structural differences occur when in L; there exists a syntactic structure which has no congruent counterpart in Lj. In transformational terms (see Chapter VI), these differences are due to differences in the application of rules generating the equivalent structures. A structural difference occurs

Various levels of linguistic analysis

65

whenever a particular rule of grammar is not shared by the grammars of Lj and Lj. For instance, in Polish (and also in Spanish and other languages), but not in English (and German), the subject need not be specified as a personal pronoun but may be expressed only by means of the verb inflection: On napisai list. Napisal list. He wrote a letter. * Wrote a letter. Ona napisala list. Napisala list. She wrote a letter. * Wrote a letter. Oni napisali list. Napisali list. They wrote a letter. * Wrote a letter. In addition Polish has a number of overtly subjectless constructions, which as a rule correspond to English constructions with overt subjects: Padalo wczoraj. It rained yesterday. But not: * Rained

yesterday.

Bylo gorgco. It was hot. But not: *Was hot. Slyszy sig ο tym duzo. One hears about it a lot. But not: * Hears about it a lot. The statement of structural differences of this sort must be enhanced by the statement of categorial contrasts, which occur whenever in equivalent sentences corresponding elements belong to different grammatical categories. Such is the case in the following pair of equivalent sentences: Czujp sip zle. (Adverb) I feel bad. (Adjective) Fisiak et al. (1978: 21 ff.) also note the following categorial contrasts: Main verb Mog? tam pojechac.

vs.

modal verb I can go there.

Noun phrase wierzcholek gory I often play harp.

vs.

prepositional phrase the peak of the mountain Czgsto gram na harfie.

Categorical contrasts are inherently connected with functional contrasts, which occur whenever equivalent categories perform different

66

Chapter V

functions in equivalent sentences. 2 Noun phrases and verb phrases are the two major sentence constituents in a large number of languages. Verb phrases typically function only as predicates, but noun phrases can perform a variety of functions. Therefore, a large number of functional contrasts can be expected if various functions of noun phrases are compared across languages. Considering Polish and English and restricting the illustrations to subjects and objects, we find the following contrasts: Subject in Polish vs. Markowi (dat.) zgingiy pieni^dze (nom.). Ksi^zka (nom.) jest mi (dat.) potrzebna.

direct object in English Mark lost his money.

Indirect object in Polish Chce sig jej spac. Brakuje mi czasu.

vs.

subject in English She is sleepy. I lack time.

Indirect object in Polish On jej ufa. Jan im wierzy.

vs.

direct object in English He trusts her. John believes them.

Indirect object in Polish vs. Markowi popsul si§ samochod.

I need the book.

possessive attribute in English Mark's car broke down.

'To Mark broke down car' Adverbial phrase in Polish vs. Umarl lekkg. smiercig.. 'died-he light death-with'

direct, usually cognate object in English He died an easy death.

The three types of syntactic contrasts: structural, categorical, and functional, varied as they are, can be stated regardless of the theoretical framework adopted in the description of the compared languages, as long as the description is conceptually consistent in both languages. In this respect, contrastive analysts rely on their intuition, which prompts decisions not only about which sentences and constructions are equivalent in the compared languages but also which elements (constituents) of the compared constructions are structurally, categorically, and functionally equivalent. Necessarily, since intuition is the guiding principle, these decisions are often arbitrary and unmotivated on other, more objective

Various levels of linguistic analysis

67

grounds. For instance, among the examples discussed by Fisiak et al. (1978: 27 ff.) and quoted above, there are some which could not be recognized as semanto-syntactically equivalent if more rigorous criteria were applied and, consequently, if a more constrained view of semantosyntactic equivalence were to be employed. For example, (1) (2)

(3)

John lost his keys. would not be recognized as equivalent to Klucze zgingly Janowi. since the former is not the closest approximation to an aceptable word-for-word translation of the latter. In fact John's keys vanished.

is a more likely candidate, despite a subtle difference in the meaning: in the Polish sentence the subject seems to be more directly affected by the disappearance of the keys than the subject in the English sentence (3). But one also notes a difference in the meaning between (1) and (2): in the former the subject seems to be more directly responsible for the disappearance of the keys than the subject in (2), a more passive experiencer of the event. Considering these differences in the meaning, which occur in both cases, the fact that no other translations are readily available, and the fact that (3) is a closer approximation to (2) than is (1) in terms of word-for-word rendering, (3) rather than (1) would have to be recognized as semanto-syntactically equivalent with (2). By the same reasoning, (4) would be the Polish equivalent of (1): (4)

Jan zgubil klucze.

It is worth noting that (4) does not differ semantically from (1) in the same way in which (2) does, which provides another argument for why only (4) rather than (2) can be recognized as the semanto-syntactic equivalent of (1). Decisions concerning equivalence must be based on rigorous criteria, if equivalence is to be defined in terms of an explicit device, enumerating semanto-syntactically equivalent sentences. A device of this sort is discussed in Chapter VIII. For pedagogical and other practical purposes, it is perfectly possible to compare translationally or perhaps statistically equivalent sentences, such as (1) and (2). The fact that (1) is often found as a translation of (2) cannot fail to have pedagogical relevance. All the same, the translational equivalence of (1) and (2) cannot serve as a foundation for drawing syntactic generalizations about functional contrasts across languages since translational equivalence does not guarantee

68

Chapter V

semanto-syntactic equivalence (see Chapter III), hence it cannot provide reliable data for contrastive generalizations of a semanto-syntactic nature. Syntactic contrastive studies occasion certain theoretical and methodological problems. The crucial problem consists in the question whether syntactic contrastive studies can be projective, i.e., whether they can be generalized in terms of rules, or whether they must remain reduced to listing odd and unpredictable contrasts. The problem is connected with linguistic universale insofar as syntactic contrasts across languages begin in exactly that place where syntactic universale end. Thus, the scope of syntactic contrastive studies is negatively delimited by the scope of syntactic universals. This relationship is aptly perceived and discussed by Preston (1975), who argues within Chomsky's theoretical framework that universal grammar will eventually leave only idiosyncratic odds and ends and irregularity behind in particular grammars better grammars, in the generative sense, provide less and less detail for contrastive analysis (Preston 1975: 69). According to Preston, "contrastive analysis does not constitute a part of theoretical linguistics" (Preston 1976: 69) and has no contribution to make to a general theory of language: Contrastive linguistics is not, then, a handmaiden to theoretical grammar, providing insights into universal characteristics of human language (Preston 1975: 67). These words are drastically at variance with Zabrocki's view on the subject. Though Zabrocki also believes that contrastive study "does not aim at creating any original explanatory theory" (Zabrocki 1976: 108), he still claims that it can and should contribute to a) linguistic typology, b) general linguistic theory, c) grammatical descriptions of particular languages (Zabrocki 1976: 108). The two conflicting claims are typical of the state of uncertainty beclouding the area and its theoretical status vis-ä-vis other domains of linguistics. It is also worthy of note that controversies of this sort should arise in connection with syntactic contrastive studies, since syntax is the focus of interest in transformational-generative accounts of language, which promote and attempt to answer all kinds of questions about universal grammar. In subsequent chapters we shall attempt to reconcile these conflicting views on the nature of syntactic contrastive studies.

Various levels of linguistic

analysis

69

3. Topics in lexical semantic contrastive studies3 Earlier in this chapter, we drew the distinction between sentence semantics and word semantics. The former deals with that part of the meaning of a sentence which remains constant regardless of the specific lexical items which appear in the sentence. This aspect of the meaning of sentences serves as tertium comparationis for syntactic contrastive studies. Word semantic deals with the specific meanings which particular lexical items contribute to the overall meaning of sentences. The categorial and subcategorial status of sentence constituents, morphological markers (inflections), word order, function words, and intonation, all contribute to sentence semantics in that they are more or less systematically and predictably correlated with such phenomena as declarative, interrogative, and imperative forms of sentences, tense, sentence connection, embedding, topic/comment relations, focus formation, and presuppositions. For example, a prototypical statement has the form of a declarative sentence, a prototypical question has the form of an interrogative sentence, a prototypical expression of the past time assumes the form of the past tense, etc. All these phenomena, insofar as they deal with formmeaning relations, fall within the domain of sentence semantics. Word semantics deals with idiosyncratic properties of articular lexical items appearing in sentences and with various systematic relationships between various categories and subcategories of words insofar as these words constitute paradigmatic sets associated through some common properties (such as kinship terms, colour terms), which have no bearing on the syntactic well-formedness of sentences. The latter properties of words are also considered as part of dictionary information, usually in the form of information concerning syntactic contexts in which particular lexical items can appear. This sort of information does not, strictly speaking, fall within the domain of word semantics. Let us note that the word dictionary has at least two senses: (a) practical, in which it denotes "a reference book on the living-room or library shelf' (Leech 1974: 202); (b) theoretical, in which it denotes "a hypothesis about, or, 'the inbuilt dictionary' which every one of us carries around as part of his mental equipment as a speaker of a language" (Leech 1974: 203). Let us observe that this second sense, in fact, embraces two senses, viz., (1) "a hypothesis" itself, which must have the form of a set of statements in some metalanguage, and (2) the mental objects to which the "hypothesis" refers. In the present work, whenever we use the

70

Chapter

V

word dictionary, we are referring to the first two of the three senses, i.e., dictionary as a reference book and dictionary as a set of linguistic statements, i. e. "a hypothesis" The branch of linguistics which deals with words, called lexicology, offers a large number of definitions of "word", none of which seems to be obviously better than the next. In view of this, for the purposes of the present work, we shall assume the following: just as in the case of the sentence, which cannot be given a satisfactory one-sentence definition but rather must be defined in terms of a device called grammar, which generates sentences in a particular language, "word" cannot be defined without constructing a dictionary of that language. Thus, while sentences are objects generated by the grammar of a language, words are objects which are listed in the dictionary to the left of each lexical entry. All the same, before compiling a dictionary, the investigator must delimit the scope of his data in such a way as to conform to a native speaker's intuitions (unfortunately very misleading!) about what constitutes a word in a given language, i.e., about what qualifies as a potential lexical entry in the dictionary that he is compiling. Therefore, some tentative, working definition is necessary to ensure a certain degree of consistency in compiling lexical entries. It follows that the notion "word" can be described from two points of view. From the point of view of a linguistic theory the word is a linguistic unit which is listed in the dictionary of a particular language as the first element of each lexical entry. From the diagnostic point of view, while providing criteria for identifying words in texts (both written and spoken), it is possible to describe the word in a variety of ways. In the present work we "define" the word after Arnold (1973: 30) and Lyons (1968: 203) in a definition which is a synthesis of the definitions by the two authors: a word is the smallest significant unit of a given language, which is internally stable (in terms of the order of component morphemes), but potentially mobile (permutable with other words in the same sentence). This definition makes it possible to distinguish between the word and the phrase (not the smallest significant unit), the word and the morpheme (not positionally mobile within a word), as well as the word and the phoneme (not significant). In this way, the definition isolates lexicology from syntax (phraseology), morphology, and phonology. It is needless to say that all these areas are mutually interrelated, and that in actual analytic practice it is often difficult to draw clear-cut boundaries. Espe-

Various levels of linguistic

analysis

71

daily, the boundary between lexicology and syntax is fuzzy, which creates certain problems to which we shall presently return. The crude definition of lexicology given above defines it as the study of words. Considering the definition of words as significant units of language, it is inevitable that what lexicology must be concerned with is, among other things, meaning of words. Now, the study of meanings of words is impossible without examining linguistic and extralinguistic contexts in which words appear and in which they assume various senses within the limits characterizing a particular word. Therefore, any productive approach to meanings of words must be through the contexts in which words appear. The definition of the word given above allows it to include among words not only compounds such as blackboard or typewriter, in which constituent morphemes cannot be permuted within a given sentence, but also set phrases (fixed expressions) of various degrees of conventionality, ranging from such non-motivated phraseological fusions as red tape (bureaucracy) or kick the bucket (die), which are equivalents of words, to highly motivated phraseological collocations such as alarm clock and night-school, all of which express certain integrated notions. Some of these expressions happen to have one-word equivalents in another language (cf. Polish biurokracja, budzik, etc.). All such combinations of linguistic units, which also function as words (red, tape, night, school, alarm, clock, kick, the bucket), and which express fixed integrated notions naturally fall within the scope of lexicology, and it would be useless and impractical to pretend that they do not (see also Nowakowski 1977: 38). However, if one accepts the view that lexicology has to deal with compounds and fixed expressions of the type exemplified above, one faces a formidable task of delimiting the upper bound of lexicology, separating it from syntax. The basic problem is to what extent constraints on collocations of particular lexical items in syntactic constructions are subject to listing in a dictionary and to what extent they can be stated in terms of rules. This in turn is connected with a more general problem of what may be called "precision" of grammars. Early transformational generative grammars, written in the framework of Chomsky's Syntactic structures (1957), were extremely crude in that they imposed no constraints on the co-occurrence of various content words (nouns, verbs, adjectives, adverbs) in syntactically well-formed preterminal strings. These models generated both, semantically acceptable sentences, such as John admires sincerity, and semantically unacceptable Sincerity admires John. However, Chomsky realized that an adequate grammar should be en-

72

Chapter V

dowed with means for eliminating such semantic anomalies. One way to do this was to subcategorize verbs and nouns in such a way as to allow only some verbs to occupy positions after abstract nouns functioning as subjects of these verbs, and only some verbs in positions before human nouns, etc. This led to the formulation of the theory of selection restrictions expounded in Chomsky (1965). The elimination of all possible semantic anomalies through the operation of selection restrictions requires the introduction of an appallingly large number of theoretical concepts called "semantic markers" This fact is noted by Bolinger, who writes: If we are to account for the fluent speaker's ability to recognize an anomaly as well as an ambiguity through the markers at his command, then the number is indeed legion (Bolinger 1965: 564). All the same, there is no doubt that contextual features, strictly subcategorizing verbs in terms of syntactic categories with which verbs can co-occur in sentences, are syntactically relevant. Strict subcategorization rules eliminate such obvious syntactic anomalies as *John found sad, *John elapsed that Bill would come, or *John compelled (Chomsky 1965: 148 — 149). However, neither contextual strict subcategorization features nor markers (in whatever number) are helpful in accounting for ill-formedness of a large number of collocations such as * carry a grudge as opposed to bear a grudge or *strong rain as opposed to heavy rain, or *take hatred as opposed to take a liking. On the other hand, McCawley claims that markers have no syntactic relevance at all, since the matter of selectional restrictions should be totally separate from the base component and the base component thus be a device which generates a class of deep structures without regard to whether the items in them violate any selectional restrictions (McCawley 1968: 135). This claim leaves collocations totally outside the scope of any grammar: lexical anomalies, such as *take hatred or *carry a grudge are syntactically well-formed, and the theory does not provide any framework for considering them semantically ill-formed, either, unless lists of collocations are systematically incorporated into the theory to be drawn from at the same level at which other lexical items are inserted. We shall take up this problem in a little while. Let us state, at this point, that notwithstanding theoretical problems with finding a place for collocations in an overall account of language, they do fall within the scope of

Various levels of linguistic analysis

73

lexicology, constituting its upper-bound and being an area which shades into syntax in a non-discrete (fuzzy) way. Thus lexicology is concerned with formal and semantic properties of the following phenomena: one-morpheme words, constituting the lower bound of lexicology (e.g., table, man, dog, radio, etc.), complex words (e.g., engulfj writer, disagreeable, etc.), complex-compound words (e.g., blackboard, armchair), compound-complex words (e. g., typewriter, radioannouncer), phraseological fusions or idioms, i.e., poorly motivated or apparently quite unmotivated combinations of words which are semantically integrated (e. g., red tape, kick the bucket), phraseological unities, i.e. partially motivated combinations of words (e.g., show one's teeth, to wash one's dirty linen in public), and phraseological collocations, i.e. combinations of words characterized by lexical valency but highly motivated from the semantic point of view, constituting the upper bound of lexicology (e. g., bear a grudge, bear malice, take a liking, make a decision, take a break, have breakfast). Both formal and semantic properties of these units can be the object of contrastive studies. Let us look at some of these properties from the contrastive perspective under the following headings: (1) simple words; (2) word formation; (3) phraseological fusions, unities, and collocations; (4) semantic relations; (5) emotive, axiological and stylistic charge. 4 3.1. A simple word A simple word is a word which consists of a single base with or without inflections. A large number of simple words in English correspond to simple words in other languages, at least in some of their senses, for example English dog — Polish pies, E. bed, P. lozko, E. room — P. pokoj, etc. In some cases, a simple word in English corresponds to a polymorphic word in Polish and vice versa, for example handle u-chwyt, floor podloga. Such comparative statements are theoretically uninteresting, though it is by no means obvious on what grounds particular words in one language should be matched with their equivalents in another language. We take up this problem towards the end of this section. 3.2. Word formation Word formation involves derivation by means of affixation, back-formation, word-composition, shortening, acronymy, and some minor types, such as sound interchange, distinctive stress, and sound imitation (see

74

Chapter V

Arnold 1973: 93 ff.). The investigator faces a vast area here, since lexical equivalents across languages may display a considerable range of differences with respect to particular means employed in the formation of words. No systematic correspondences seem to be in view, since a onemorpheme word in one language may correspond to a compound word in another language or a complex word in one language may correspond to a simple word in another language, and so on, in all possible combinations. Below are some examples of such correspondences: English

Polish

simple word

complex word

seat bitter darn moral

simple word

siedzenie gorzki cerowac moralny

compound word

hedge porcupine

zywoplot jezozwierz

complex word

simple word

rubber whiteness flight poster gift

compound word rainbow armchair father-in-law man-of-war hedge-hog moon-calf

compound word telltale sightseeing book-keeping ear-ring ash-tray knee-cap

guma biel lot afisz dar

simple word tecza fotel tesc okrgt jez kretyn

complex word plotkarz zwiedzanie ksi§gowosc kolczyk popielniczka rzepka

Various levels of linguistic analysis

75

English

Polish

complex word perlocator Christmas ratter

phrase maszynka do kawy Boze Narodzenie czlowiekjpies lapigcy szczury

complex word foreigner Easter

compound word cudzoziemiec Wielkanoc

Since formal comparisons of individual lexical items do not seem to lend themselves to any significant generalizations, contrastive studies of word formation are better off if they are based on some conceptual framework. Such an analysis could involve a comparison of various means employed in the derivation of nomina actions, nomina agentis, nomina loci, and also of adjectives of intensity, inclination, possibility, ability, or of verbs of process, causation, instrument, and so on (cf. Sehnert —Sharwood-Smith 1974, as an example). As a matter of fact, any aspect of the meaning can serve as a basis for cross-linguistic comparisons. We thus obtain contrastive studies of various words expressing such concepts as colours (Duczmal 1979; Schmitz 1983), kinship (Kalisz 1976), modality (Aarts — Wekker 1982), or vision (Pasanen 1978). 3.2.1. Six possibilities for formal and semantic similarity All types of words and their equivalents may exhibit various degrees of formal and semantic similarity. Lado distinguishes six possibilities: 1. Similar in form and meaning — there are a number of words having reasonably similar pronunciation and/or spelling in the compared languages. In some cases, they are borrowings, in other cases they can be traced back to the same etymological source. Among examples from English and Polish, one finds such words as map — mapa, lamp — lampa, hotel hotel, nose nos, pilot pilot, minute minuta, second Sekunda, son syn, and many others. Naturally enough, the spelling conventions often conceal considerable differences in the pronunciation, as in the case of such graphically similar words as psychiatrist [sai'kaiatrist] psychiatra [psixi'atra] or pilot ['paibt] pilot ['pilot]. Conversely, differences in the spelling may conceal similarities in the pronunciation, as in the case of such words as combine ['kambain] kombajn [kombajn] or Shakespeare [Jeiksipis] — Szekspir [Jekspir], etc.

76

Chapter V

2. Similar in form but different in meaning — formally similar words may be only partially similar in meaning or may have quite different meanings. For example, the English word dog and the Polish dog are partly similar in meaning, but the Polish word dog has a narrower range as it denotes only one breed of dog; the word has undergone the semantic process called specialization. The same process has affected such words as toast [toast] from toast [taust], rekord [rekord] from record ['rekord], prezent [prezent] from present ['prezant], and many others. Some words belonging to this group are similar in form but totally different in meaning, e. g., actually (really) — aktualnie (at present), desk (pulpit) — deska (board), rower (one who rows) — rower (bicycle). In his numerous contrastive analyses of German, French, and Italian, Schwarze calls such pairs "false friends" {i falsi amici) since they often lead to errors in foreign language usage (Schwarze 1979: 230). A detailed study of "false friends" in Polish and English, or as the author, less metaphorically, calls them, "deceptive words", can be found in Welna (1977). 3. Similar in meaning but different in form — a large bulk of equivalents across many languages are of this type, at least as regards the primary meaning of particular words. Among them one finds such pairs as table — stol, chair — krzeslo, pen — pioro, tree — drzewo, etc. Although the primary and perhaps some further senses of particular words in two languages correspond, it is not necessarily the case with other meanings of particular equivalents. Consider as an example the word tree and its Polish equivalent drzewo. Their primary meanings correspond in that both words refer to "perennial plant with single woody self-supporting stem or trunk usually unbranched for some distance above ground" (The Concise Oxford Dictionary). In both languages, however, the two words have some further meanings which do not correspond. For example, in Polish drzewo may also refer to wood or timber. Such "differential contrasts" may also lead to serious learning problems (cf. Krzeszowski 1970: 81; Schwarze 1979: 227 ff.). 4. Different in form and in meaning — they are words with "strange" meanings since, as Lado puts it, they "represent a different grasp of reality" (Lado 1964: 217). The American English first floor is different in form from apparently similar phrases in many languages including the Spanish primer piso and the Polish pierwsze pi§tro, which are also different in what they refer to as "first" The Spanish primer 'first' and the Polish pierwszy 'first' do not refer to floor number one at the ground level but

Various levels of linguistic

analysis

77

to floor number one above the ground level so that the Spanish and the Polish expressions refer to what American English calls second floor. 5. Different in their type of construction. Differences in this area concern the morphological structure of words. For example, English has lexical items consisting of a verb and a particle modifying it. Such verbs often carry idiomatic meanings, which cannot be induced from the meanings of the constituents. They are such lexical items as call up (telephone), call on (visit), give up (resign), take in (cheat), take off (start), and so on (see Hill 1968). Such two-word verbs have no formal counterparts in most languages, including Polish. In some, less idiomatic instances the verbal particles correspond to Polish prefixes as in put vs. put on — klasc vs. na-kladac write vs. write out — pisac vs. wy-pisac call vs. call off — wolac vs. od-wolac go vs. go out — chodzic vs. wy-chodzic (Krzeszowski 1970: 82) 6. Similar in primary meaning but different in connotation. A word may be harmless in connotation in one language but its equivalent in another language may be offensive or taboo. For example, the Polish word krwawy has no offensive overtones while its English equivalent bloody is a swear word. Names of various body parts connected with sex and defecation in many languages assume additional functions as swear words. Apparently, however, some languages, such as Finnish, seem to be fairly insensitive to this sort of semantic reorientation. Sometimes a seemingly innocent word, such as kurczg 'chicken', assumes the additional function of a swear word, in this case by phonetic analogy with the much more effective kurwa 'whore, bitch' In any case, such words may be the source of learning problems as their improper use constitutes sociolinguistic goofs. 5 3.3. Phraseological fusions, unities, and collocations Another area of investigation covers what we have called phraseological fusions, unities and collocations. Any differences between the three kinds of units can be attributed mainly to varying degrees of motivation in the combination of words within particular units: presumably, fusions are semantically unmotivated, unities are partially motivated, and collocations are highly motivated. 6 Since clear-cut boundaries between these

78

Chapter V

types of phrases are impossible to draw, we shall discuss them jointly under the general term phraseological units (cf. Ginzburg et al. 1966: 100 ff.). Phraseological units are intermediate between compound words and free phrases (loose phrases) in that like compound words they are severely constrained with regard to the co-occurrence of their constituents, but from the syntactic point of view they are phrases (cf. Arnold 1973: 148 ff.). In many cases, phraseological units are semantically and functionally equivalent to single words in more than one language, which is another reason why they contribute to the domain of lexicology rather than syntax. Phraseological units may exhibit considerable differences across languages, both with respect to lexical congruity and with respect to syntactic congruity. The following examples illustrate the lack of the two types of congruity: a. Lack of lexical congruity English

Polish

small hours

wczesne godziny 'early hours' wieczne pioro 'eternal pen' m^z i zona 'husband and wife' orzel czy reszka 'eagle or tails' gwaltowna burza 'violent storm'

fountain pen man and wife heads or tails heavy storm

The lack of lexical congruity in those pairs of equivalent phraseological units resembles the lack of congruity of constituents of compounds and their phraseological equivalents in Polish: scare crow wrist-watch night-school

strach na wroble (sparrows) zegarek na (hand) szkola wieczorowa (evening)

Various levels of linguistic analysis

b. Lack of syntactic congruity Ν + Ν window curtains Ν + Ν brain trust N's + Ν cat's paw VP + NP influence somebody

79

Ν + Adj zaslony okienne Ν + Ν + genitive trust mozgow Adj 4- Ν kocia lapa V + Prep + NP wplyngc na kogos

In addition to comparisons of various syntactic patterns characterizing equivalent phraseological units across languages, it is also possible to examine and compare various degrees of valency of lexical items in compounds and phraseological units. For example, the English phrase calf love appears to be more lexically stable than its Polish equivalent cielgca milosc inasmuch as in Polish one also finds cielgce lata (calf years), cielgcy rozum (calf mind), cielgcy mozg (calf brain), and cielgcy zachwyt (calf delight) as conventional phrases, which suggests that the adjective cielgcy in this particular metaphorical sense collocates more freely with other lexical items than its English equivalent calf does. 3.4. Semantic relations between lexical items The next area of investigation embraces various types of semantic relations between lexical items and provides a convenient dimension for cross-language comparisons. Among others the following types of semantic relations can be described and compared: (a) polysemy; (b) homonymy; (c) synonymity; (d) antonymity; (e) semantic fields. Let us look briefly at each of these types of relations. 3.4.1. Polysemy Most content words in everyday use in all natural languages have more than one sense. Any specification of senses of a word in any dictionary is a recognition of the polysemic character of that word. Any specification of various equivalents in L2 of a lexical item in L, is not only a recognition of the polysemic nature of words but also an exercise in lexical contrastive studies. Insofar as bilingual dictionaries must contain such specifications, the phenomenon of polysemy is among the most crucial in lexical con-

80

Chapter V

trastive studies. This is reflected in the fact that polysemous words constitute networks of interconnections in any sizeable dictionary: for every polysemous word in L^ there exists a set of equivalent items in L2, with each such item having a set of equivalent items in L,, etc. For example, the English word table has the following twelve equivalents in Polish: 1. stol; 2. towarzystwo przy stole, biesiadnicy,; 3. plyta; 4. napis wyryty na kamiennej tablicy, ~ s prawa, ustawy; the ten ~ s dziesigcioro przykazan; 5. plaskowyz; 6. karnisz; 7. plaszczyzna drogiego kamienia, klejnotu; 8. dlon; 9. blaszka kostna; 10. tablica; tabela; spis; wykaz; 11. ~ s tryktrak; 12. stol przewodniczpcego parlamentu angielskiego (Jan Stanislawski, The Great English-Polish Dictionary). The Polish word plyta, which is one of the equivalents of table, has the following equivalents in English: 1. slab·, plate; sheet·, board·, 2. record; 3. table-land; 4. wrest-block; wrest-plank (np. fortepianu); plaque (pamiatkowa) (Jan Stanislawski, The Great Polish-English Dictionary). In turn, the English word board, one of the equivalents of plyta, has eight equivalents in Polish, and so on. To a large extent the same is true of 3.4.2. Homonymy Homonymy is the relation between lexical items which are formally identical as in the case of the Polish zamek\ 'castle', zamek2 'lock', 'zip', zamekj 'halving' (archaic) and zamek4 'hinge ligament' (in muscles), where lock has eight equivalents in Polish. The presentation of lexical stocks in two languages in terms of such networks of interconnections is possible for a given finite set of lexical items in both languages, but it is a futile procedure if textual settings in which particular lexical items appear are not considered. For the same reason, paradigms of grammatical forms are meaningless and do not constitute a grammar of a language, if they are not seen as elements of structures into which they can be inserted. 3.4.3. Synonyms Like almost every notion in lexicology, the term "synonym" has no clearly defined scope. Roughly speaking, synonyms are the words which are different in phonic/graphic form but similar in denotational (referential) meaning and interchangeable at least in some contexts. This

Various levels of linguistic analysis

81

definition of synonyms makes use of the well-known distinction between denotational and connotational meaning, which itself is not clear, as we shall see later. It also relates to the notion of context. This is so because, firstly, it is impossible to talk about synonyms of individual words as such. Usually a particular sense of a word is synonymous with one or more senses of another word. Moreover, it is impossible to match words as synonyms if they are examined in isolation from the contexts in which they can appear. For example, the verb to read in the context read dreams is a synonym of to interpret while in the context read one a lesson it is a synonym of to give. The matching of synonyms is thus basically no different from the matching of lexical equivalents across languages insofar as in both cases the selection of the appropriate sense, essential in choosing the appropriate synonym or the appropriate equivalent, is determined by the context in which the investigated word appears. Secondly, the similarity of meaning involves only that part of the overall lexical meaning which is called denotation (reference) in contrast to connotation in the well-known distinction. Presumably, words like live, dwell, and reside in their relevant senses are denotationally similar, but differ with respect to their stylistic values, i. e. with respect to their connotations: live is said to be neutral, dwell is poetic, while reside is formal. Thirdly, synonyms are interchangeable only in some contexts, while in other contexts they are not only not interchangeable but also sometimes turn out to be antonyms. Such is the case with the words exceptional and abnormal. In the context of the weather is , the two words are interchangeable, but in the context my son is , the two words are near antonyms (Quirk 1962: 120).

3.4.4. Antonyms Antonyms involve similar problems. As in the case of synonyms, one can only talk about a specific sense of a particular word being antonymous to a specific sense of another word. For example, the word good in most senses and in most contexts is antonymous to bad as in a good girl vs. a bad girl, to be good at something vs. to be bad at something, etc. Yet, good in the sense 'not less than a certain amount' as in a good three miles or a good way is not an antonym of bad since in a bad three miles and in a bad way the word bad is an antonym of good in its primary sense, i.e., 'having the right quality' The antonymous polarity in the second sense is in fact impossible to express by means of a single word, which, if it

82

Chapter V

existed, would have to express the notion 'less than a certain amount' The nearest possible candidate would be the word little as in a little three miles and a little way. Contrastive studies of synonyms and antonyms do not basically differ from such studies of other words, since in all cases the essential problems are connected with the selection of appropriate senses on the basis of relevant contexts.

3.4.5. Semantic fields Semantic fields provide grounds for yet another way of grouping words. Those words which share a common concept are said to constitute semantic fields (Trier 1931). The basis for grouping is always extralinguistic since words are grouped in semantic fields because things which they refer to are connected with extralinguistic reality (Arnold 1973: 206). Some examples of semantic fields are: colours, kinship terms, pleasant and unpleasant emotions, military ranks, educational terms, gastronomical terms, vehicles and their parts, being at a place, leaving, sports, games, etc. Contrastive studies of words constituting semantic fields in various languages are a rewarding activity and are so often performed that they are sometimes identified with lexical contrastive studies in general (e.g., Duczmal 1979; Sehnert 1975; Kalisz 1976; Tomaszczyk 1976; Schmitz 1983). The fifth type of comparison involves stylistic-emotive charge of lexical items, i. e., the connotational component of their meaning. Synonyms or groups of words with similar denotational meaning usually differ with respect to their connotations. The stylistic charge of lexical items can be described in terms of features falling into keys — oratorical, deliberative, consultative, casual, and intimate — and in terms of parameters superimposed on keys — status, technicality, dignity, and conformity (Gleason 1965: 358 ff.). The emotive charge of lexical items can be described in terms of markedness with respect to such features as "appreciative" or "positive", "depreciative" or "negative" and "neutral" The number of parameters along which lexical items can be described has never been definitely established, 7 which obviously creates problems for lexical contrastive studies. Contrastive studies of synonyms and semantic fields, enhanced by the distinctions suggested above, are a rewarding as well as amusing area of investigations (cf. Lawendowski 1974).

Various levels of linguistic analysis

83

The area of lexical contrastive studies bristles with problems, both theoretical and practical. By definition, lexicology deals with those elements of language which cannot be generalized in terms of rules within the domain of grammar. However, as has been repeatedly pointed out, no strict boundaries between lexicology and grammar (syntax) can be drawn. This fact creates a challenge for linguists, who, so far, have been unable to deal satisfactorily with the uncertain linguistic status of phraseological units, which have syntactic properties of loose phrases but, like words, are much more severely constrained with respect to the selection and order of their constituents. Since these constraints are not dependent on the syntactic structure of the units in question, and since they are not systematic, like other unpredictable phenomena they are subject to listing and thus fall within the domain of lexicology. Yet it is impossible to ignore the systematic and predictable character of some words as regards their inner structure. As Halle suggests, a linguistic theory should account for the native speaker's ability to distinguish between well-formed combinations of morphemes constituting words and ill-formed combinations of morphemes. Halle argues that lexical insertion transformations take items from the dictionary rather than from the list of morphemes a large part of the dictionary is stored in the speaker's permanent memory and he needs to involve the word-formation component only when he hears an unfamiliar word or uses a word freely invented (Halle 1973: 16). Being systematic in nature, word formation can be handled in terms of rules generating words (cf. Saumjan's (1968) concept of 'word-generator'). Such rules can be contrasted with analogous rules in another language according to principles which are not different from those that govern syntactic comparisons. Comparisons of this sort may turn out to be theoretically rewarding. Still one must bear in mind that in a large number of cases, language users treat derived lexical items as indivisible (though potentially analysable) entities and thus subject to listing rather than to generating in terms of rules. Therefore, it appears that the area of word-formation is a border area between those linguistic phenomena which are subject to listing and those that are subject to enumerating in terms of a finite set of rules generating an infinite set of lexical items. Also "vertical" and "horizontal" relations, as grasped by Nowakowski's lexicon (Nowakowski 1977: 27), may turn out to have a more systematic character than has been previously ascribed to them. However, as in the case of word-formation, any rules which might underlie such

84

Chapter V

relations are probably dormant in language users and are activated only in those special instances when the need to employ those rules arises. Among "vertical" relations, Nowakowski distinguishes between hyponymy and hyperonymy (see also Lyons 1968: 453 ff.), and among "horizontal" relations those of paraphrase, synonymy, oppositeness, reference, presupposition, and metaphorization. As we have suggested earlier on, all these phenomena can be investigated in the contrastive dimension, regardless of whether in particular languages they are handled in terms of rules or in terms of lists. The overlapping of grammar and lexicology is also evident in the description of lexical items themselves. Each lexical item is described in terms of a quadruplet of features: LE = , in which LE stands for "lexical entry", / is a specification of phonological features, s is a specification of syntactic features, m is a specification of semantic features, and r is a specification of relational features. Phonological features are specified as distinctive features of segments (cf. Fisiak 1976: 120) or abbreviated as a phonological transcription. Syntactic features specify syntactic contexts into which particular lexical items can be inserted. Each word is categorized as a noun, verb, adjective, adverb, preposition, pronoun, etc. Certain words are further subcategorized either in terms of inherent features, such as [ +countable], [ + common], [ +abstract], etc. or in terms of contextual features, which specify actual syntactic contexts into which particular subcategories of words fit. For example, certain verbs (traditionally called "transitive") can occupy positions immediately preceding noun phrases while other verbs (traditionally called "intransitive") do not fit such contexts. The information conveyed in terms of inherent and contextual features can reach a considerable level of "delicacy" and shades into information of semantic rather than syntactic sort. Thus, whereas in (1)

*John elapsed the time

the verb elapsed has been inserted in defiance of the feature subcategorizing it as intransitive, i. e. one which does not tolerate an object noun phrase, in (2)

*John drank the table

the verb appears in the context of the object noun phrase with the feature [ — liquid], thus violating a selection restriction of semantic rather than syntactic nature (cf. McCawley 1968: 133 ff.). Therefore, information of this sort ought to be contained in the specification of semantic features

Various levels of linguistic

analysis

85

which characterize particular lexical items within subcategories specified independently by syntactic features . Finally, relational features, relate a specific lexical item to other members of the same subcategory with respect to synonymity, hyponymy, entailment, etc. Thus, lexicon overlaps with syntax by virtue of ; but it also overlaps with the outer reality by virtue of , i. e., semantic features which refer to properties of things in the world at large and which eventually are responsible for the distinctions in the meaning of particular words. But such distinctions, in most cases, are based on our knowledge of the world and thus constitute encyclopedic rather than linguistic knowledge. One can expect a great deal of fuzziness in this area as well. But encyclopedic information turns out to be as much a part of the meaning of words as linguistic information. As Langacker puts it The distinction between semantics and pragmatics (or between linguistic and extra-linguistic knowledge) is largely artifactual, and the only viable conception of linguistic semantics is one that avoids such false dichotomies and is consequently encyclopedic in nature (Langacker 1987: 154). Practising lexicographers have been well aware of the problem when they did not hesitate to include "pragmatic" information in their descriptions of various lexical items. Without "pragmatic" knowledge, a great number of words could not be defined in a sufficiently precise way to warrant the necessary contrasts with other lexical items within the same semantic sphere. This indeed happens on those occasions when the lexicographer disregards knowledge of the world in attempting to define a lexical item. Such is the case with cat and dog, respectively defined as "a domesticated carnivore, bred in a number of varieties" and as "a domesticated carnivore, bred in a great number of varieties" (The Random House Dictionary of the English Language. College Edition). If one wishes to distinguish between the two species of animals, as indeed one ought to in a decent dictionary, one has to make a more liberal use of our knowledge of the world as in the following definitions of cat and dog, respectively: "a small, four-footed, flesh-eating mammal, often kept as a pet or for hunting rats and mice ..., a domesticated mammal, related to wolves, foxes, and jackals, that is kept as a pet, for hunting, and for guarding property" {The World Book Encyclopedia Dictionary). From the contrastive point of view, one of the thorniest problems is how to establish equivalence of lexical items across languages. One of the meth-

86

Chapter V

ods involves the decomposition of lexical items into features and considering as equivalent those lexical items which can be decomposed into identical sets of features. Some difficulties connected with this method transpire from Di Pietro's analysis of flesh and meat in English and their equivalents in various languages. One rather obvious difficulty consists in the fact that each feature resulting from the decomposition must be regarded as a "theoretical" construct with little, if any, independent motivation. Decomposition of lexical items into "atomic" features in fact means a reduction of the concepts associated with these lexical items into some more fundamental concepts, which are presumably universal and can thus serve as tertium comparationis for establishing equivalence between lexical items across languages. Presumably, then, equivalent words would have identical matrices of "features", i. e., the atomic concepts obtained from the decomposition. Adopting this procedure results in the necessity of constructing a special "theory" for every pair of equivalent items across languages. Every feature resulting from the decomposition is a "theoretical" construct with little, if any, independent motivation. The problem which arises is analogous to the one encountered in connection with the recognition of semantic markers, discussed earlier on, whose "number is indeed legion" (Bolinger 1965: 564). What is even worse, there seems to be no upper limit to this number since the process of decomposition of words into components of meaning appears to have no end, and it, moreover, often leads to circularities. The sort of problems that one encounters when adopting the decompositional procedure can be easily seen when a standard example involving the words man, boy, woman, and girl is considered. These words are usually decomposed into the features [male], [female], [young], and [adult], according to the following matrix:

[adult] [young]

[male]

[female]

man boy

woman girl

This sort of framework can hardly serve as tertium comparationis for lexical contrastive studies for the following reasons: Firstly, none of the features resulting from the decomposition is a primitive concept: they can all be decomposed further. For example, The Concise Oxford Dictionary defines young as "that has lived a relatively short time; not mature [italics my own] or fully developed" Further decompositions usually lead to circularities: mature is defined as "com-

Various levels of linguistic analysis

87

plete in natural development or growth; fully developed in body and mind" and finally, to complete the circle, develop is defined as "to grow into a fuller, higher, or maturer condition" Likewise, the definition of adult resorts to the same circularity: "grown up, having reached the age of maturity'''' Such examples are a matter of rule rather than exception. Since the features resulting from decomposition are not primitive concepts, they are usually ambiguous, which renders them useless as potential tertia comparationis. For example, young, in addition to the sense described, has some further senses, such as 'newly begun or formed; not advanced: The evening is young', 'pertaining to or suggestive of youth or early life: young for her age\ 'vigorous or fresh, youthful' Although these senses are related through "family resemblance", they do not necessarily correspond to a set of parallel equivalents in another language. For example, young corresponds to mlody in Polish in its most prototypical sense "that has lived a relatively short time', but not always in the sense 'being in the early period of growth' when it refers to plants or feelings. In this sense, young may correspond to the Polish adjective swiezy. The polysemous character of such "features" and the various connotations, inseparable from their basically lexical character make them very questionable as possible tertia comparationis for lexical contrastive studies. Secondly, the reductions inherent in decompositional analyses of this sort do not seem to reveal accurately the meaning of the analysed lexical items insofar as they leave some area of the meaning unaccounted for. That boy is not merely 'young male' is well seen in the collocation young boy, which is not tautologous even if the componential analysis suggests that the adjective young, when placed in front of a noun containing the feature (young) as one of its components, should produce tautology. Similarly, the expression young man would be impossible since the analysis suggests the incompatibility of the adjective young with a noun containing the feature (adult) as one of its components. That fact that such expressions as young man are semantically well-formed demonstrates the presence of some aspects of meaning which escape description in terms of the componential analysis of this sort. What is even worse, such analyses lead to faulty predictions. Thirdly, because of the presence of areas of meaning which escape componential analysis, the matching of feature matrices will also lead to faulty predictions in the domain of lexical contrastive studies. There is no doubt, for example, that the features (young) and ( m a n ) are somehow present in such lexical items as boy, lad, youth, urchin, and master. The most appropriate ways in which these words can be matched with their equivalents

88

Chapter V

in a foreign language cannot depend on such features as and alone. A great deal more is at stake and a host of other features would have to be postulated to account for contrasts between these words. The nature of these additional features, especially with reference to the emotional aspects of the meaning of such words has never been made sufficiently clear to provide a basis for contrastive studies. Fourthly, the existence of various fuzzy concepts casts further doubts on the componential analysis of the sort described above. The componential analysis rests on the classical view of sufficient and necessary conditions, determining category membership, whereby an element is a member of a category if and only if it is characterized by a given set of features which define a given category. The presence of these features creates both necessary and sufficient conditions of category membership. This approach may work reasonably well when categories with clear-cut boundaries are considered. For example, the category "bird" can be defined in terms of two features (feathered) and (vertebrate). Anything to which these two features can be attributed is a bird. Thus, the two features are sufficient and necessary as conditions of membership in the "bird" category. This procedure cannot be applied to various concepts which have no clear-cut boundaries and which shade into other concepts in a non-discrete way. For example, such English words as forest, wood, woodland, grove, copse and spinney express non-discrete concepts. The differences between them are a matter of degree. The American Heritage Dictionary defines spinney as "a small grove", and the same dictionary defines grove as "a small wood", while wood is defined as "a dense growth of trees; forest" Clearly, then, since the adjective small expresses an inherently relative notion, the differences between these concepts are a matter of degree and as such cannot be stated in terms of necessary and sufficient conditions. One faces the same sort of problem with certain polysemous lexical items such as game. There is no set of "precise necessary and sufficient conditions for something to be game" (Lakoff 1982: 14). Various games are related through "family resemblance" which allows us to identify such diverse activities as ring-around-the-rosie, solitaire, Monopoly, and football as belonging to the category "game" (Wittgenstein 1953 [1972]: 50 ff.). In view of these problems, componential analysis cannot be adopted either as a method of analysis of fuzzy concepts or as a provider of tertia comparationis for lexical contrastive studies. Bilingual dictionaries are practical results of lexical contrastive studies. Such dictionaries consist of lexical items in one language and their

Various levels of linguistic analysis

89

equivalents in another language, usually selected on the basis of linguistic intuitions of the compilers. A more rational procedure would be to match lexical items in two languages on the basis of detailed analyses of numerous equivalent texts written or spoken in both languages. This sort of analysis would provide a rigorous way of determining the number of senses of polysemous words: a careful comparative analysis of translationally equivalent texts would reveal the number of ways in which a particular item in one language could be rendered in another language. The number of these renderings would correspond to the number of senses of that lexical item (Manczak 1970). Such an analysis would provide statistical grounds not only for matching equivalents but also for listing them in the order from the most frequent to the least frequent in all those numerous instances when a lexical item in one language has more than one equivalent in another language. In any case, no decomposition of lexical items into semantic features would be required to guarantee correct matching of equivalents. (For more arguments against decomposition see Wierzbicka 1980: 70 ff.). Thus, a large bulk of any bilingual dictionary is a juxtaposition of lexical equivalents across languages on the basis of the identity or at least similarity of referential meanings, such as cat — hot, table — stol, wireless — radio, walk — chodzic, yellow — zolty, small — maly, often — czgsto, diet — sejm, primary school — szkola podstawowa, M.A. thesis — praca magisterska, power plant — elektrownia, etc. Any person using a bilingual dictionary certainly expects to find this sort of information in the first place. Yet, he is seldom happy if his dictionary does not go beyond listing equivalents, even if, or perhaps especially if, more than one equivalent is given for a particular item, as indeed is the case in the vast majority of cases. Most dictionary users inevitably wish to be able to find information about the use of particular lexical items in both situational and linguistic contexts. They also welcome information about the possibilities of forming derived lexical items on the basis of those listed in the dictionary. A dictionary which fails to provide this sort of information is considered to be inferior to a dictionary in which such information can be found (Tomaszczyk 1979: 103-119). Here we arrive at a difficult problem: how much information about contexts in which particular words appear should be included in a dictionary? The answer partly depends on the needs of the prospective users, different in the case of native and non-native users. Ideally, every bilingual dictionary should consist of four parts: 1. L, *-*• L2, 2. L 2 Verb + N P iNPs,n g ]

(3)

N P (.NPpiurJ

(4) (5) (6) (7) (8) (9) (10) (11) (12)

NP sing —> Τ + Ν + 0 NPplur—» Τ + Ν + S' Τ — the Ν — man, ball, etc. Verb — Aux + V V —> hit, take, walk, read, etc. Aux — C(M)(have 4- en)(be + ing) Μ —> will, can, may, shall, must fpres

Η

[past

(Note: Rule (12) is not explicitly formulated on p. I l l , but it is referred to throughout the book.) Polish PS-rules (after Polahski 1971: 199): (1) (2) (3)

Zdanie —• GN n o m + G W GN n o m Przyrn nom + R n o m GW —• C + GN a c c

130 (4) (5) (6) (7) (8)

Chapter

VI

GN a c c —> R acc Przym —• znajomy, maly Rnom —*• chlopiec, uczen C —• widzi, wola, lubi Race —• kolegg, ojca. In these rules Zdanie = Sentence GN n o m = Noun Phrase in the nominative GW Verb Phrase Przym nom = Adjective in the nominative Rnom = Noun in the nominative Racc = Noun in the accusative

The sets of rules such as those above cannot easily be compared if only because they do not represent the same degree of accuracy. For example, the Polish set does not provide any information about tenses, nor does it mention modal verbs. Moreover, both these grammars can be further refined to account for a greater number of sentence types. For example, the English rules do not generate sentences without objects or sentences with adverbs, such as John smiled, Mary danced yesterday, etc. The Polish rules have analogous limitations. In addition, it is unable to generate sentences with objects in cases other than the accusative as in Jan pomaga dzieciom (dative) 'Jan helps (to) the children' Such trivial inadequacies can be easily eliminated as new rules are added to both grammars to make them generate whatever number of sentence types are used in a particular language. As a result, both the sets of PS-rules can be brought to an approximately comparable level of refinement. Less trivially, even if reasonably complete PS-rules for the languages to be compared were formulated, one would still confront formidable problems connected with comparability. Since all the rules concern only the syntactic structure of sentences, there is no principled way of establishing the necessary tertium comparationis, insofar as no semantic considerations ever enter into play in the PS-rules formulated in Chomsky's early model. It will be remembered that in Chapter III we argued that only a semantic tertium comparationis in syntactic contrastive studies is viable. For this reason, no contrastive studies of PS-rules have ever been successfully performed (cf. Dingwall 1964 b). Instead contrastive analysts compare basic sentence patterns in a fashion resembling structural contrastive studies (Stockwell et al. 1965; Lipinska 1972, 1973; Fisiak et al. 1978). Particular basic sentence patterns across languages are brought

Linguistic models and contrastive

studies

131

together for comparison (juxtaposition) in an intuitive manner on the basis of the judgments of the analyst himself, who associates particular patterns across languages on the basis of either their formal similarity (again a vicious circle since similarity is to be revealed by the contrastive studies and cannot be used as tertium comparationis) or by means of their semantic similarity, never explicitly acknowledged. Thus, the problems with comparability which arise in connection with comparisons of basic sentence patterns are identical to those that are inherent in structural contrastive studies. Stockwell et al. (1965: 32) distinguish six basic sentence patterns in English and Spanish. They compare them mainly on the basis of formal resemblance (committing the error of circularity). They resort to translation equivalence — loosely connected with semantic equivalence (cf. Chapter III) — only in those instances when structural similarities are not matched by semantic similarity or similarity of usage. Disparities of this kind occur mainly in the area of verbal complementation. For example, English sentences: Ε (a) (b)

I wanted him to come I asked him to come

containing verbs of communication can only be rendered in Spanish with two clauses, each containing its own finite verb: S (a) (b)

Queria que el viniera Le pedi que viniera

On the other hand, a sentence like Ε (c)

I beg you not to talk any more about politics,

which contains a verb of suasion, can be rendered in Spanish by means of either a complement construction or a clausal one: S (c)

Les ruego que no discutan mäs de politica Les ruego no discutir mas de politica

However, a construction such as: Ε

I said to be sure

and its Spanish equivalent: S

Dije estar seguro Ί said I was sure'

express completely different meanings. Stockwell et al. (1965: 40) quote this example to show that although similar patterns generally express

132

Chapter

VI

similar meanings, in specific instances structurally different constructions are required. The examples quoted above demonstrate the lack of consistency in deciding about criteria of comparability. Like in structural syntactic contrastive studies, many basic sentence patterns are juxtaposed for comparison because they resemble each other formally and in this way what should result from the comparison proper, i.e. the statement of formal similarities and differences, is anticipated in juxtaposition, as it illegitimately serves as tertium comparationis. The rare occasions when the semantic tertium comparationis is referred to are confined only to those instances when there is a discrepancy between meaning and form across languages. Such discrepancies, too, should be stated as results of comparison proper. Thus, what characterizes both structural contrastive studies and generative transformational contrastive studies is the confusion between the principles of juxtaposition and the results of comparison proper. There are, however, further problems and limitations inherent in transformationally based contrastive studies, to which we shall now turn attention. A typical contrastive study, within the generative transformational framework, usually also contains a detailed description of particular sentence constituents in the compared languages, often described in terms of PS-rules. In Stockwell et al. (1965: 64 ff.) one finds descriptions and comparisons of noun phrases and their constituents, verb phrases and their constituents, determiners, etc. For instance, the rule expanding N P as suggested by Stockwell et al. is identical in both languages: N P —> D + Ν + N U M where D — determiner, Ν — noun, N U M — number. The rules rewriting N U M are also identical: NUM The rules rewriting determiners are different in the two languages:

S:

Linguistic

models and contrastive

Studie.

133

This is as far as the analysis goes within the purely generative framework. In fact, what we are dealing with here is not a comparison of rules but rather constructions generated by those rules. The rules are merely a convenient algebraic device to describe the constituent structure of determiners in both languages. Unfortunately, such an analysis does not reveal much. Therefore, the authors proceed to analyse the relevant material in a more detailed way, but in doing so they abandon the generative framework and adopt a combination of traditional and structural principles, involving comparisons of systems and constructions rather than of rules. They juxtapose the articles in the two languages in the following table: Definite Articles SPANISH singular

ENGLISH plural

masc.

el

los

fem.

la (el)

las

neut.

lo

(singular or plural) before a consonant

before a vowel

[di]

[diy]

Some obvious differences concerning the gender and number distinctions in Spanish and their absence in English can be seen immediately. There are, however, other differences which can only be revealed when constructions in which these systems are employed are compared. In specific constructions, various articles in Spanish correspond to various articles in English and vice versa. The comparison of systems proves to be totally inadequate as one faces the following equivalence possibilities, analogous to those encountered in the comparison of pronouns (see Chapter IV): Ε

S

(1)

the

el

(2)

a

un

0

0

(3)

134

(4) (5) (6) (7) (8) (9)

Chapter

VI

Ε

S

the the a a

un

0

0

0

el

0

el un

The table shows that in six out of nine cases there is a deviation from "systematic" correspondence the — el, a — un, 0 — 0. In actual equivalent sentences, considerable discrepancies occur, and they can be accounted for only in terms of conditions of usage. Stockwell et al. state them in a traditional, nonformal way. For example, (7) is described thus: Pattern (7) above is that in which English has a matching Spanish 0. This contrast occurs most frequently after a linking verb when the following noun shows mere identification, as in the following examples: He's a philosopher He's a friend of mine He's a relative of the chief

Es filosofo Es amigo mio Es pariente del jefe, etc. (Stockwell et al. 1965: 68).

The example concerning the determiner in English and Spanish demonstrates the inadequacy of a contrastive study limited to a study of PSrules generating the constituent called Determiner. To be complete, a contrastive study of determiners must be expanded to cover both the study of the entire systems and, moreover, the study of ways in which elements of those systems are used in actual utterances. This is so because the PS-rules, no matter how detailed, do not provide access to the nature of paradigmatic relations between particular elements of the systems. Specifically, the rules provide no clues as to when elements of the systems can be used and to what choices are appropriate in those cases in which rules contain various options. Such is the case with the rule rewriting Determiner in Spanish and English, where the indication "in specified environments" cannot be made a working part of the rule without reference to extralinguistic situations attending the specific utterances rather than to intralinguistic (textual) factors. In contrastive studies based on generative transformational grammars, comparisons of various transformations constitute an essential part of these studies. In contrastive studies based on early models of transfor-

Linguistic models and contrastive

studies

135

mational grammar (Chomsky 1957), T-rules were used to account for structural relationships between kernel sentences and transforms. Kernel sentences, i. e., sentences generated by PS-rules and obligatory transformations, were simple, declarative, affirmative sentences in the active voice. Interrogative, negative, passive, complex, and compound sentences were derived through T-rules from kernel sentences. Expectedly, T-rules operating in various languages displayed differences and similarities and in this way yielded themselves to contrastive studies. In short, they were assumed to be comparable (cf. Dingwall 1964: 158 — 159). According to Marton, two T-rules in two languages are essentially identical if they "operate on two congruent structures in the same way and consequently result in congruent transforms" (Marton 1968 b: 59). Congruent structures (strings) are those structures (strings) which underlie congruent sentences. Congruent sentences, in turn, are defined as sentences in two languages, which are semantically equivalent and consist of equivalent items belonging to the same word classes and having the same syntactic functions in the respective sentences. An additional requirement is that in congruent sentences all constituents should appear in the same linear order (cf. Marton 1968 b: 56 — 57; Krzeszowski 1967: 37). Although articles and auxiliaries are disregarded in this definition of congruence, there still remain considerable problems with other function words and grammatical morphemes. Marton quotes the following pair of strings as congruent in Polish and English: Pol: Eng:

# d z i e c k o + Neut + Sing + 3rd Pers + Past 4- Perf + isc — do — kino + Neut + Sing 4- G e n i t # # the — child + 02 — Past + go — to — the — cinema + 02 #

which underlie two presumably congruent sentences: Pol: Eng:

Dziecko poszlo do kina. The child went to the cinema.

The quoted strings do not consist of the same number of identical elements. Specifically, in Polish, but not in English, there appear formatives representing gender, third person, aspect, and case. On the other hand, the English string contains articles, which have no equivalents in Polish. Thus, the two strings can be considered as congruent only if all function words and grammatical morphemes are disregarded. The definition of congruent strings would then have to be reformulated as follows: Congruent strings are such strings which consist of the same number of lexical words, representing equivalent grammatical categories, arranged

136

Chapter

VI

in the same linear order and underlying semantically equivalent sentences. By virtue of this reformulated definition, the two strings and the two sentences quoted above could indeed be recognised as congruent. Having defined congruent strings and sentences in the way suggested above, one can accept Marton's definition of "identical" transformations as those that "operate on two congruent structures in the same way and consequently result in congruent transforms" (Marton 1970: 59). As an example, consider one of Marton's cases, viz., the transformation which derives attributive adjectives from nouns. In both Polish and English it adds an adjective forming suffix to a noun. A variant of this transformation derives the adjectives in question from kernel strings Ν has N' and Ν ma Νrespectively, and this transformation is identical (congruent?) in both languages in that it derives congruent transforms from congruent structures: Eng: Pol:

the peasant has a beard => the bearded peasant chlop ma brodg => brodaty chlop

Marton also distinguishes "similar" transformations across languages. Like identical transformations, similar T-rules operate on congruent structures but do not yield congruent transforms. Thus, the Polish expressions niebieskooka dziewczyna, siwowlosy mgzczyzna, jednooki zebrak and czteromotorowy samolot, according to Marton, are not fully congruent with their English equivalents a blue-eyed girl, a white-haired man, a one-eyed beggar, a four-engined aircraft because of the differences in the distribution of stresses (one primary stress in Polish against two stresses in English, compound word in Polish against a hyphenated word in English).11 Still, because they are derived from congruent structures, they involve what Marton calls "similar" transformations. Marton's proposal is an important step in making such fundamental concepts in contrastive studies as equivalence and congruence more precise. Within the adopted framework of the early transformational grammar it was impossible to make these notions fully explicit. Later developments in linguistic theory led to the reformulation of the place of semantics in the model and rendered it possible to couch the concepts of equivalence and congruence in more rigorous terms (see Chapter VIII). In 1965, Chomsky presented a new version of his theory, which became known as the standard theory. In the standard theory, all transformations were made to be obligatory and were triggered off by various abstract symbols implanted in deep structures underlying particular sentences. Such symbols as Q for questions or Neg for negative, generated by PS-

Linguistic models and contrastive

studies

137

rules, indicated that the appropriate T-rules were to apply to the strings in which these symbols appeared to yield surface structures of interrogative, negative, passive, etc., sentences. The standard theory incorporated semantics as one of the components of the grammar. The semantic component was interpretative in that it assigned semantic interpretations to the deep structures generated by the central syntactic component, or more strictly, the part of the syntactic component referred to as the Base (PS-rules and the lexicon). All the necessary aspects of the meaning of various types of sentences were potentially present in the form of symbols indicating, on the one hand, what sort of meaning is to be assigned to the sentence by the interpretative semantic component, and on the other, what T-rules are to be applied to yield the surface structure of the sentence. Thus, transformations did not, in principle, change the meaning of strings on which they operated because all the relevant meanings were determined at the level of deep structures, which constituted inputs to semantic interpretations and to transformations. In contrast to earlier models, in the standard theory PS-rules did not account for only simple, declarative, affirmative sentences. They provided deep structures of all types of sentences. As was said above interrogative, negative, and passive sentences contained appropriate abstract symbols, such as Q (Question), Neg (Negation), or Pass (Passive) in their deep structures. These symbols triggered off the appropriate T-rules and semantic rules. Complex sentences were generated by means of the recursive S in the rule rewriting NP: NP

(Det) Ν (SO (for details see Chomsky 1965: 107 ff.)

By way of example, let us see how various types of questions are handled in a comparative study conducted within the framework of the standard theory. All types of interrogative sentences are derived from underlying basic strings containing the abstract element Q as a presentence morpheme, which triggers off the interrogative transformation. Since there are several types of questions, including w/z-questions, tag questions, and echo questions, and since many questions may occur in either dependent or independent clauses, they require various T-rules to account for them. These rules and details of their applications differ across languages. The illustrative survey that follows concerns questions in English and Polish and is based on Fisiak et al. (1978: 180 ff.). The rule which ultimately accounts for the surface structure of English interrogative sentences, the interrogative transformation, inverts the sub-

138

Chapter

VI

ject NP and the auxiliary constituent called tense and the modal and/or have and/or be, whatever the case may be. If no modal or have or be follow tense, another transformation, called do-support, must be applied. These rules yield surface structures of such sentences as Has he gone there? Will you do it? Are we going to dance? Do you speak English? derived from the following respective strings Q Q Q Q

+ + + +

NP NP NP NP

+ + 4+

present present present present

+ + + +

have + gone + there will + do + it be + going + to dance speak + English

The interrogative transformation yields the following strings: present present present present

+ + + +

have + NP + gone + there will + NP + do + it be + NP + going + to dance NP + speak + English

The last string illustrates the situation in which the transformation dosupport must be applied to eliminate the case of the "dangling" tense. In Polish, inversion is optional, and when it does apply, it also moves the main verb with tense attached to it. Therefore, Polish has no transformation which would parallel the English do-support. In contrast to English, the Polish rule which preposes auxiliary or the main verb is ordered after the rule which attaches the tense affix to the auxiliary or the main verb, with the result that in Polish sentence tense never "dangles", and, consequently, do-support is not required. Moreover, since in Polish, word order is generally more free, at least as with regard to grammatical constraints, and the question word czy can be optionally omitted, and a much larger number of arrangements of words in questions is possible than in English: Czy Piotr bgdzie jutro tutaj? 'Whether Peter be-future tomorrow here?' Czy bgdzie Piotr jutro tutaj? 'Whether be-future Peter tomorrow here?'

Linguistic models and contrastive

studies

139

Gdzie b§dzie Janek spac? 'Where be-future Janek sleep?' Gdzie Janek bgdzie spac? 'Where Janek be-future sleep?' Czy on czytal t§ ksiqzkg? 'Whether he read that book?' (Czy) czytal on t§ ksigzk§? '(Whether) read he that book?' The interrogative transformations in the two languages are formulated as follows: English: Q + N P + Τη + V => Tn + N P + V => i/o + Tn + N P + VP through inversion through do-support and Q-deletion Polish:

Q + NP, + Τη + V + NP 2 => Q + NP, + V Tn + NP 2 through Affix-placement or => (Q) + V - Tn + NP, + NP 2 through inversion and interrogative-deletion (optional)

Abstracting from the accuracy of these rules, they show that tense in Polish does not dangle, and, consequently, ί/o-support is not necessary. Furthermore, ß-deletion in Polish is optional. In those cases when Q is not deleted, it is realized lexically as czy. In English Q is obligatorily deleted in independent questions, and thus we get Does he drink? and not * Whether does he drink? In dependent questions, the lexical marker of yesjno (general)-questions is realized in both the languages, as czy in Polish and as if/whether in English: Zapytaj go, czy chce pic. Ask him if/whether he wants to drink. This sample contrastive study performed in the framework of the standard theory, rather inconsistently applied and augmented by remarks and observations couched in traditional terms, reveals some weaknesses of the standard theory as a model founding contrastive study: On the

140

Chapter

VI

one hand, its formal apparatus is rather complex, and it tends to obscure rather than to highlight the relatively simple structural differences between questions in the two languages (these differences can be stated more clearly in traditional terms). On the other hand, the standard theory pays insufficient heed to certain matters of detail, such as gender, number, or case, especially in highly inflected languages, such as Polish. Details of this kind may be trivial in the context of a theory which seeks to formulate universal principles of language, but they may be of utmost importance in establishing cross-language contrasts, and in fact they constitute the main body of grammatical contrastive studies. One might say that crosslanguage differences, and not linguistic universale, are what contrastive studies are all about. Fisiak et al. are well aware of these dilemmas inasmuch as they allow their contrastive study to be eclectic rather than consistent within the framework of the standard theory. In the Preface to their analysis, they admit the following: The authors, however, have tried to avoid formalizations as much as possible, and have not hesitated to be eclectic and to use traditional interpretations to explain the facts which have not yet been satisfactorily explained by transformationalists (e. g. gender, number, case and aspect, to name but a few) or whose explanation, tentative as it is, would require the introduction of a highly abstract and theoretical apparatus, thus contributing to an already complicated picture of language structure (Fisiak et al. 1978: 5). The most serious limitation of the standard theory as a framework for contrastive studies in the domain of syntax is that it does not provide a clearly delimited semantic tertium comparationis against which such studies could be conducted. The semantic component in the standard theory, being interpretative, is peripheral. The relevant semantic information, though certainly available, is diffused in the device called the lexicon and in the "projection" rules assigning semantic interpretations to sentences and their constituents. On the other hand, the Base, generated by PSrules and lexical insertions, as formulated in the standard theory, cannot serve as a tertium comparationis inasmuch as it is syntactic in nature. It is required of a tertium comparationis that it should consist of properties outside the compared material (Krzeszowski 1984: 305; Janicki 1985: 15). Therefore, a syntactic base cannot serve as a tertium comparationis for syntactic contrastive studies, if only because its shape is determined by syntactic properties of specific languages. As a result, Bases in specific

Linguistic models and contrastive

studies

141

languages must be different in precisely those ways which contrastive studies seek to discover and describe. Juxtaposing such Bases for comparison would illegitimately anticipate the stage of comparison proper since such a juxtaposition would be founded on the assumption that the Bases are in some respect similar, and this could only be a syntactic similarity. Thus highlighting similarity and suppressing difference would, on the one hand, defy the very purpose of contrastive studies, and, on the other, more seriously, would entangle the procedure in the circularity whereby syntactic similarity would be used as tertium comparationis for a procedure which seeks to establish syntactic similarity and difference, etc. It is mainly due to these problems that the standard theory did not enjoy popularity among contrastive analysts. 12 However, later developments in the transformational theory, such as Case Grammar and Generative Semantics, offered much better perspectives for solving these thorny problems. Unfortunately, they also introduced a host of new problems to cope with. Fillmore's Case Grammar (Fillmore 1968), notwithstanding its obscurities and theoretical shortcomings (cf. Robinson 1970; Nilsen 1972), provided a very convenient framework for numerous contrastive studies (e.g., König 1970; Krzeszowski 1971; Di Pietro 1971; Konderski 1973; Boas 1977; and many others). In contrast with Chomsky's PS-rules generating deep structures of only English sentences, Fillmore's rules were designed to generate universal structures, from which surface structures of sentences in all languages could be derived. The elements of these structures were unordered and involved nominal phrases and verbs. The semantic functions of noun phrases were represented by so-called cases (or more appropriately "roles"). Fillmore assumed that the rules of the Base conceived in this way were universal and could be stated in terms of modified predicate calculus: In their deep structure the propositional nucleus of sentences in all languages consists of a V and one or more NPs, each having a separate case relationship to the Ρ (and hence to the V) (Fillmore 1968: 51). Each proposition is preceded by Modality (M), so that the first rule of the universal Base is formulated as S —> Μ + Ρ

(Fillmore 1968: 24)

This rule reflects the general assumption that every sentence, on the semantic plane, consists of the propositional core, symbolized as Ρ and non-propositional semantic material, symbolized as M. The concatena-

142

Chapter

VI

tion symbol " + " appearing in the rule is somewhat misleading since the two elements cannot be seen as appearing in linear order from left to right in what purports to be a universal base where elements cannot be linearly ordered. Neither does Μ stand in any one-to-one correspondence to a particular sentence constituent or constituents since Μ can be realized by a variety of syntactically heterogeneous means. In Fillmore's deep structure, Μ represents such divergent categories as tense and aspect of the verb appearing in P, sentence modality (declarative, interrogative, or imperative), as well as modality expressed by various modal verbs, such as English shall, will, can, may, etc. Fillmore does not provide any indications as to how these various semanto-syntactic phenomena are generated in the Base, nor what rules map the semantic structure of Μ on surface realizations. 13 The propositional core Ρ states the event and its actors, i.e., nominal concepts involved in the event and characterized as particular "cases": Agentive (A), Dative (D), Factitive (F), Locative (L), Objective (O), Instrumental (I), and perhaps a few more. The number and the nature of these "cases" have never been determined or limited either by Fillmore himself or by his followers, although various revisions of the original proposal have been suggested. 14 Regardless of the number of postulated cases, Ρ can be expanded according to the following schema: P-^V + C,+

+cn

where particular expansions can be realized as V + A, V + Ο + A, V + D, V + O + I + A, etc. In the dictionary, particular verbs are represented in terms of "frame features" indicating sets of case frames into which particular verbs can be inserted. A given verb can appear in more than one frame. For example, open in the door opened appears in the frame O, while in John opened the door the same verb appears in the frame Ο + A. In the wind opened the door the verb open is assumed to appear in the frame Ο + I, whereas in John opened the door with a chisel, the frame is Ο + I + A. Fillmore further argues that subject and object are notions which are only relevant in the surface structure of sentences, and that, in most cases, the NP which becomes the subject in the surface structure is not marked as such in the deep structure. Instead the following general rule applies: If there is an A, it becomes the subject; otherwise, if there is an I, it becomes the subject; otherwise, the subject is Ο (Fillmore 1968: 33).

Linguistic models and contrastive

studies

143

This rule can be augmented to cover other "cases" as in Langendoen (1970: 59 ff.). A modified version of Fillmore's model was used by Di Pietro in his monograph Language structures in contrast (Di Pietro 1971: 54 ff.), which was the first book entirely devoted to various theoretical and methodological issues in contrastive studies, rather than to a contrastive analysis of a specific pair of languages. Di Pietro assumes that there is a deep structure shared by all languages and that the differences between languages can be found at any number of stages between the deep and the surface structures It follows that languages are similar to each other in proportion to the number and hierarchical ordering of rules shared in the intermediate levels. Thus, Spanish shares more rules with Italian than either does English or German. The four languages all share more of such rules than any of them does with Swahili. Finally, Spanish, Italian, English, German, and Swahili all share the deepest level of structure because they are all human languages and are all equally accessible to human learners (Di Pietro 1971: 27). In Di Pietro's deep structure all languages of the world share such basic elements as SENTENCE, MODALITY, and PROPOSITION as well as basic case relationships, i. e., agent, dative, object, instrument, etc. Language-specific rules operating on various levels of derivation account for surface-structure similarities and differences between particular pairs of languages. Di Pietro's model differs from both Chomsky's standard theory and Fillmore's case grammar. In Di Pietro's model, in contrast with Chomsky's, those deep structure categories which are later realized as NP's are not arranged in any linear or hierarchical order analogous to the order in which NP's in Chomsky's deep structures appeared. Di Pietro's model differs from Fillmore's in that the former contains language-neutral, "universal" categories or, as Di Pietro calls them, "syntactic primes", i.e. Names and Verboids, in addition to S(entence), M(odality), P(roposition), and C(ase) CAT(egory). Di Pietro defines "Name" as a "unit of arrangement which refers to that property of human languages not only as verbs but also as adjectives and possibly as adverbs. The other universal categories in Di Pietro's deep structures are arranged hierarchically in the following fashion:

144

Chapter

[time]

VI

V

CCATlObjJ

CM

[past]

[drink, boirei

CCATtagentJ

Ν

[wine, vin]

CM

Ν

[John, John]

where CM is a case marker and where the linear ordering of constituents is irrelevant (Di Pietro 1971: 61). Unlike in Fillmore's version of case grammar, in Di Pietro's model the specifics of case, i. e. agent, dative, instrumental, etc. are provided by the semantic projection rules and are not generated by the syntactic rules, which is consistent with Di Pietro's claim that his base rules are syntactic rather than semantic. We shall comment on this point later. At this moment let us observe that Di Pietro's hierarchical arrangement of "syntactic primes" as well as his recognition of S as a syntactic prime in this particular framework raise certain doubts. If Noun Phrase and Verb are dismissed from the universal base on the grounds that they are too "suggestive of the surface restrictions of particular languages" (Di Pietro 1971: 56), then S should be dismissed on the same grounds since surface structures in particular languages observe numerous constraints characterized by grammars of those languages. In the standard theory, S directly dominated NP and Predicate Phrase, which reflected the assumption that in English every sentence in its surface structure consists of a noun phrase followed by a predicate phrase, and that they appear in the specified order unless transformed otherwise. This sort of deep structure was a direct result of the analysis of every sentence into two immediate constituents expressed by categories related to the reality of the surface structure of English sentences. Ad-

Linguistic models and contrastive

studies

145

mittedly, not all sentences could be analysed in this way, but at least every sentence could be "untransformed" to restore the basic underlying shape. It was, furthermore, known that every noun phrase contained a noun as one of its constituents and every predicate phrase contained a verb or be. Thus, the categories appearing in the deep structure were not different from those appearing in the surface structure; consequently, the initial element S was justified in that it dominated all the categories which appeared in the surface structure of sentences in addition to whatever symbols were necessary to trigger off appropriate transformations mapping deep structures into surface structures. In Di Pietro's deep structure, S no longer dominates those categories which in the surface structure are constituents of sentences. Instead it dominates a number of "syntactic" units which in various languages are realized as various surface-structure categories in a wide variety of ways. For example, Modality cannot be associated with any single type of expansion. As a matter of fact, in purely syntactic terms, it cannot be expanded at all, at least in terms of PS-rules, as it incorporates various semantic elements which have highly heterogeneous syntactic realizations, such as time, aspect, question, negation, etc. Moreover, it is possible to envisage a situation in which a particular deep structure will be realized as one surface-structure sentence in some language and as two or possibly more surface-structure sentences in another language. Non-restrictive relative clauses constitute a good example: John is a coward. He ran away. John, who is a coward, ran away. Whether or not the two phrase markers underlying the two sentences come under the domination of one S is a matter which cannot be decided at the level of deep structure suggested by Di Pietro. A single Sentoid(?) may be realized as two surface-structure sentences, or two Sentoids may be realized as one surface-structure sentence without affecting the basic meaning. The common deep structure correctly reflects this situation, but it contains no indication of the number of sentences which are its surfacestructure realizations. Sentences with concessive or resultative clauses, which Di Pietro associates with Modality, furnish another example illustrating the point in question. Di Pietro's deep structure, which purports to be syntactic in nature, does not distinguish between the following sentences: If it rains, I'll take my umbrella. It will rain, and I'll take my umbrella. (understood conditionally rather than sequentially)

146

Chapter

VI

Di Pietro's syntactic deep structure does not distinguish between the two sentences on semantic grounds as the two sentences express the same semantic content. Yet, a purely syntactic base, such as Di Pietro's base purports to be, ought to be capable of distinguishing between complex and compound sentences. The difficulty results from the fact that although Di Pietro assumes that his deep structure contains syntactic primes, in reality it is a construct representing semantic relations underlying surface structures of sentences. Such descriptions of terms appearing in Di Pietro's deep structures as "[things] which represent man's universal need to label objects, ideas and sentiments which are subject to cognitive powers" (Di Pietro 1971: 56) clearly point to the semantic nature of the objects thus described. The difficulties alluded to above seem to suggest that if syntactic contrastive studies are to be conducted whithin the framework of a generative transformational grammar, it is necessary to adopt a version of the theory (possibly a notational variant) in which distinction is made between the semantic structure as input to transformations and the syntactic structure as output of transformations. Consequently, a consistent difference must be made between semantic primes of semantic representations and surface structure categories. In his attempt to construct the universal base, Di Pietro is only half successful inasmuch as he introduces semantically grounded terms into his syntactic model of the base. For the purposes of syntactic contrastive studies, in order to introduce a semantic tertium comparationis, it is necessary to go further and to postulate an universal semantic base, free of all syntactic relations, which are inevitably language-specific. The existence of such an universal semantic base would be the initial postulate for the construction of contrastive grammars of various languages (see Chapter VIII).

Chapter VII

Towards a theory of semanto-syntactic equivalence1

Syntactic contrastive studies, even if based on transformational generative grammars, remain taxonomic in nature as long as they aim at producing inventories of differences and possibly similarities either between parallel places of grammatical systems or between parallel constructions or even between parallel rules operating at various levels of derivation. Admittedly, applying the generative-transformational framework allows the investigator to grasp differences and similarities at those levels of representation which have been made available thanks to the developments in generative-transformational grammar, i. e., the levels which have been sometimes referred to as "deep structure" However, numerous contrastive studies conducted within the generative-transformational framework, discussed in the previous chapter, do not aim at providing an explicit format for contrastive studies where explicitness would not refer merely to the way in which the materials compared are presented individually but to the contrasting technique itself. Although "translational equivalence" was often adopted as the criterion of comparability, no precise definition of the concept emerged from those studies. Equivalent constructions (and sentences) were said to be at least sometimes mutually translatable, and the relation which holds between such objects was called "textual equivalence" (Catford 1965: 27 — 34). In Chapter II, we pointed to certain difficulties connected with adopting translation equivalence as a basis for syntactic contrastive studies. We postulated that it is necessary to distinguish between translation as a concept in the domain of performance and semanto-syntactic equivalence as a concept in the domain of competence. In this chapter, we propose to elaborate on the concept of semanto-syntactic equivalence in order to provide grounds for a more rigorous definition. We shall first put forward a hypothesis concerning semanto-syntactically equivalent sentences, and we shall provide a preliminary justification for this hypothesis. We shall then discuss and clarify certain difficulties that arise from the hypothesis. Finally, we shall rephrase the hypothesis as an initial postulate (axiom), using it as a basis

148

Chapter

VII

for a theory of semanto-syntactic equivalence, which we shall outline in Chapter VIII. In order to discover equivalents across languages, one has to rely on the authority of a competent bilingual informant, usually the investigator himself. The informant's judgments are based on his intuition, which is connected with his competence in the two languages. Although the notion of semanto-syntactic equivalence is intrinsically connected with the meaning of the linguistic objects compared, a complete constrastive study must also give an account of their form. If we characterize various constructions in any two languages in terms of equivalence and formal correspondence, we shall observe that the compared sentences may be: (a) equivalent, formally different, (b) equivalent, formally similar, (c) non-equivalent, formally similar, (d) non-equivalent, formally different. In this chapter we are mainly concerned with situations (a) and (b). We shall work on the assumption that the first term in each pair refers to the deep structures of the compared sentences while the second term refers to their surface structure. Moreover, we shall try to justify the following hypothesis: Equivalent sentences have identical deep structures even if on the surface they are markedly different. Before we provide some justification for this hypothesis, let us specify what is meant by "deep structure" in the standard theory. The argument which follows will show that "deep structure" as conceived in the standard theory is too "shallow" to accomodate the above hypothesis. Lakoff (1968) derives a definition of deep structure from Katz —Postal (1964) and from Chomsky (1965). Deep structure in the standard theory is viewed by Lakoff as a level of linguistic analysis where (a) basic grammatical relations between fundamental grammatical categories are defined; (b) selectional restrictions and co-occurrence relations are stated; (c) appropriate grammatical categories receive lexical representations; (d) inputs to transformational rules are provided. Condition (b) implies that the selectional restrictions and other cooccurrence relations stated at the level of deep structure make it possible to formulate correct generalizations about selectional restrictions and cooccurrence relations among the elements of the sentence, and that they remain constant in the surface structure. Condition (c) emerges from the assumption that the level of deep structure directly determines the semantic interpretation of the sentence. Since semantic interpretation involves semantic rules defined both in terms of semantic content of lexical items and by grammatical relations, conditions (b) and (c) are interdependent. Nevertheless, the semantic

Towards a theory of semanto-syntactic

equivalence

149

component as viewed by Katz—Pastal (1964) involves rules of a different order ("projection rules") than the rules of the Base. This definition of deep structure allows one to make decisions as to the identity of deep structures of synonymous constructions which are not identical on the surface. By demonstrating that the same selectional restrictions and other co-occurrence relations are in effect for synonymous, though superficially different, constructions, it is possible to infer that such constructions are essentially identical at the level of deep structure, in accordance with the restrictions as stated in (b). As we shall see later, the identity of selectional restrictions and co-occurrence relations is a necessary, but not a sufficient, condition for two synonymous constructions to have a common deep structure. At this stage, however, let us look at how Lakoff demonstrates the identity of deep structures of a pair of constructions represented by (1) (2)

Seymour sliced the salami with a knife. Seymour used a knife to slice the salami.

Before we continue with our task, let us introduce the Polish equivalents of (1) and (2) in order to be able to subject them to the same procedure to which Lakoff subjects (1) and (2). The Polish equivalents of (1) and (2) are, respectively (3) (4)

Seymour pokrajal salami nozem. Seymour uzyl noza aby pokrajac salami.2

In (3), the instrumental inflection in the word nozem is an equivalent of the preposition with in (1) since the former like the latter has the purposive, instrumental sense (Lakoff 1968). Also in (4) the verb uzyc is an equivalent of use in (2) as both appear in the instrumental sense. The surface structures of the two pairs of sentences can be represented as (5) (6) (7) (8)

NP, NP, NP, NP, -

V NP 2 - with - NP 3 use - NP 3 to - V NP 2 V NP 2 - Inst - NP 3 uzyc NP 3 - aby - V NP 2

where Inst in (7) stands for the instrumental inflection. Lakoff demonstrated that the same set of selectional and co-occurrence restrictions holds in (5) and (6), and he thus proved the identity of the deep structures underlying the two constructions. It appears that a parallel set of restrictions applies with reference to (7) and (8):

150

Chapter VII

I. In Polish as in English the verb represented by V must be characterized by the feature [ + Activity] for the sentence to be grammatical: (9 a) b) (10 a) b) (11 a) b)

Albert computed the answer with a sliderule. Albert used a sliderule to compute the answer. Albert obliczyl (uzyskal) odpowiedz suwakiem. Albert uzyl suwaka aby obliczyc odpowiedz. * Albert znal odpowiedz suwakiem. * Albert uzyl suwaka aby znac odpowiedz.

II. Both in English and in Polish NP, must be [ + animate]: (12 a) b) (13 a) b) (14 a) b)

Peter hit Oliver with a hammer. Peter used a hammer to hit Oliver. Piotr uderzyl Olivera mlotkiem. Piotr uzyl mlotka aby uderzyc Olivera. *Cios uderzyl Olivera mlotkiem. *Cios uzyl mlotka aby uderzyc Olivera.

III. In none of the constructions can NP 2 have the same referent as NP 3 : (15 a) b) (16 a) b) (17 a) b)

He unlocked the door with a master-key. He used a master-key to unlock the door. Otworzyl drzwi wytrychem. Uzyl wytrycha aby otworzyc drzwi. * Otworzyl drzwi nimi samymi. * Uzyl drzwi aby je otworzyc.

IV. In none of the constructions can NP] have the same referent as NP 3 : (18 a) b) (19 a) b) (20 a) b)

Anatol broke the pane with the scared soldier. Anatol used the scared soldier to break the pane. Anatol wybil szyb§ przerazonym zolnierzem. Anatol uzyl przerazonego zotnierza aby wybic nim szybe. ?Anatol wybil szybg sobg. * Anatol uzyl siebie aby wybic szybg.

V. Questions which are derived from (7) and (8) are both ambiguous in the same way as are the equivalent questions in English: (21 a) b) (22 a) b)

Did Did Czy Czy

Seymour Seymour Seymour Seymour

slice the salami with a knife? use a knife to slice the salami? pokrajal salami nozem? uzyl noza aby pokrajac salami?

Towards a theory of semanto-syntactic

equivalence

151

All the four questions are ambiguous in at least two senses. First, they may be questions asking whether the actual action of slicing took place. Second, if the action is presupposed, they may ask for confirmation that the instrument used for the slicing of the salami was a knife. VI. Negative constructions based on (7) and (8) are also ambiguous in the same way as are the equivalent constructions in English: (23 a) b) (24 a) b)

Seymour Seymour Seymour Seymour

did not slice the salami with a knife. did not use a knife to slice the salami. nie pokrajal salami nozem. nie uzyl noza aby pokrajac salami.

All four sentences may mean that the action of slicing did not take place or, if the action is presupposed, that the instrument used was not a knife. VII. In Polish as in English the verb phrase functioning as the complement of uzyc cannot be negated: (25 a) b) (26 a)

*I used the knife not to slice the salami. */ used the knife to not slice the salami. *Uzylem noza aby nie pokrajac salami.

The fact that Polish equivalents of (5) and (6) are subject to a parallel set of selectional and co-occurrence restrictions provides preliminary justification for the hypothesis that equivalent sentences have identical deep structure. In his paper Lakoff did not attempt to formulate any explicit rules representing relations between various grammatical categories in the deep structure. He contented himself with some inferences about deep structure, the most significant one being that Deep structures must be somewhat more abstract (further removed from the surface) than previous research in transformational grammar has indicated (Lakoff 1968: 24). Chomsky questioned the validity of Lakoff s argumentation supporting the conclusion quoted above and claimed that different deep structures can "express the required selectional and grammatical relations in a unified way" (Chomsky 1972: 84). Therefore, according to Chomsky, there is no need to postulate a more abstract deep structure where these generalizations could be stated. Nevertheless, for independent reasons, many linguists also expressed the conviction that deep structure must be

152

Chapter

VII

more abstract, and that it should incorporate semantics. It was claimed that semantics is not only interpretive but also generative, and that, consequently, semantic representation and syntactic representation are of essentially the same formal nature (McCawley 1968: 167 — 169). The formal apparatus used in this uniform representation is the general framework of symbolic logic or, more specifically, predicate calculus, where propositions correspond to deep structure sentences, functions correspond to verbs, and arguments to noun phrases or nouns. Fillmore's base, expressed in terms of case relations of noun phrases to verbs, can be viewed as a possible implementation of the logical notation in the description of a natural language. Before we explore the possibilities of applying such a semantic base as tertium comparationis for syntactic contrastive studies (see also the account of Di Pietro's attempt in the previous chapter), let us consider various problems which arise in connection with the hypothesis that equivalent sentences have an identical deep structure and the possible implications that this hypothesis, if true, could have for the nature of deep structure. For the sake of brevity we shall refer to the above formulated hypothesis as the "identical-deepstructure hypothesis" However, it should be remembered that when this hypothesis was first formulated (Krzeszowski 1971: 38), the term "deep structure" was used to mean a more abstract level of representation than the level of deep structure as formulated in the standard theory. This more abstract level could be more appropriately called "semantic structure" or "input structure", where "input" would mean "input to syntactic derivations" The most elaborate criticism of the identical-deep-structure hypothesis comes from Bouton (1976). He points to two difficulties involved in its original formulation. The first difficulty is that "tense and aspect contained within them [two identical semantic structures] must be identical also" (Bouton 1976: 154). In a great number of Polish and English sentences, including those discussed above, i.e., (1) and (3), aspect is apparently different since in Polish but not in English, completed (perfect) vs. incompleted (imperfect) aspect of the verbs involved in the sentence is obligatorily expressed. Thus, not only (1) and (3) but also (2) and (4) are not equivalent since the Polish sentences, in contrast with their English counterparts, contain information that "Seymour had finished his work, having cut all the salami that he had intended to cut" (Bouton 1976: 154). The incompletion of the action of cutting would have been expressed by means of a different aspectual form, viz. uzywac, as in (27)

Seymour uzywal noza aby pokrajac salami.

Towards a theory of semanto-syntactic

153

equivalence

which would presumably mean "Seymour was cutting salami and did not finish" Despite certain inaccuracies in Bouton's interpretation of the Polish aspectual forms, to which we shall return shortly, the point that he is making is clear: since in contrast with English, Polish must specify the feature [ ± completion] on the verb, Polish and English sentences must have different semantic structures in spite of the fact that they are attested to be equivalent. Therefore, the hypothesis that equivalent sentences have an identical semantic structure is wrong. The other difficulty that Bouton saw in the identical-deep-structure hypothesis involved active and passive constructions in English and their Finnish equivalents: (28) (29) (30) (31)

Tom broke the window. (28F) The window was broken by Tom. Someone broke the window. (30F) The window was broken by someone.

Tom rikkoi Ikkuna

ikkunan.

rokotiin.

According to Bouton, since (28) and (29) are mutually translatable with (28F) and, and (30) as well as (31) with (30F), the two respective sets constitute textual equivalents in the two languages. Since, however, (28) and (29) cannot be derived from the same deep structure and neither can (30) and (31), one or the other of the two English sentences in both of the triplets must be derived from a different structure, thus contradicting the hypothesis that translation equivalents have identical deep structures. Bouton's first example functions as evidence against the identicaldeep-structure hypothesis only on the assumption that aspect, like tense, must be represented in the semantic structure of sentences. As Bouton does not support this assumption with any arguments, it must be taken that the assumption has the character of an axiom and as such requires no proof. Under this particular axiom, it may be surmised that any semantic distinction must be reflected in the semantic structure of sentences as is indeed the case in the so-called generative semantics. However, an alternative way of viewing semantic phenomena, namely that not all semantic distinctions are to be included in the semantic representation of sentences, should also be allowed. Such an alternative might exclude aspect (though not necessarily tense) from the universal stock of semantic distinctions to be represented in the semantic structure of sentences. It is quite possible to adopt the initial view that semantics falls into sentence

154

Chapter

VII

semantics and word semantics (see, for example, Bartsch — Vennemann 1972). This distinction is drawn in the following manner: The sentence semantics of a sentence represents the meaning of the surface syntax of a sentence in logical syntax without reference to the meaning of any particular word of the sentence (Bartsch — Vennemann 1972: 39). The following aspects of sentences contribute to their semantics: the categorial and subcategorial status of constituents, morphological markers (inflection), word order, and intonation. These markers serve as exponents of such semantic information as declarative, interrogative, tense, sentence connection, sentence embedding, topic/comment relations, focus formation, and presuppositions. What exactly constitutes the domain of sentence semantics and what does not is a moot point, but at least it is possible that the aspect of Polish verbs should not be included there but rather in the domain of word semantics, which is treated in the lexicon. On the other hand, the English durative and perfective "aspects" should rather be represented at the level of sentence semantics. Some of the reasons motivating different treatment of aspect in Polish (and other Slavonic languages), in comparison with the treatment of "aspect" in English, are the following: 1. Aspect in Slavonic languages is an intraword phenomenon while in English both durative and perfective require the presence of auxiliary verbs. 2. In Slavonic languages changes in the aspect do not affect the syntactic structure of the sentence while in English sentences expressing aspect have a different structure in questions, negations, etc., from sentences without aspect. 3. Every Slavonic verb is inherently completed or incompleted while no English verbs are inherently "durative" or "perfective", just as in the majority or cases, nouns are not inherently singular or plural and need not be listed separately in the lexicon as plural. 4. There is a high degree of unpredictability concerning the completed and incompleted forms of verbs in Slavonic languages, and these forms have to be listed in the lexicon. Rules, if at all possible to formulate, abound in exceptions. In English all durative and perfective forms obey a simple syntactic rule: Aspect —> (be 4- ing) (have + en).

Towards a theory of semanto-syntactic

equivalence

155

If the distinction between sentence and word semantics is drawn, and if the Polish aspect is treated in the lexicon, the Polish and English "Seymour sentences", objected to by Bouton, and other equivalent Polish and English sentences would still have identical semantic structures at the level of sentence semantics. In addition to sharing identical sentencesemantics structure, equivalent sentences would contain equivalent lexical items drawn from a bilingual dictionary where particular lexical entries in LI would correspond to lexical entries in L2 according to complex networks of grammatically and semantically conditioned relations. Such a dictionary would have to contain as its entries on both sides the same lexical entries which appear in respective monolingual dictionaries, although the organization of the definitions and the information included in the bilingual dictionary would not be a simple juxtaposition of the respective monolingual dictionaries, but rather a principled network of matchings of lexical items across languages (cf. Nowakowski 1977: 32 — 33). Completed and incompleted verbs in Slavonic languages would constitute separate lexical entries in monolingual dictionaries of Polish, as part of the overall account of the Polish language, as well as in the bilingual Polish-English dictionaries, as part of the overall account of equivalence relations between Polish and English sentences (and texts). Thus such Polish verbs as uzyc (perfective) and uzywac (imperfective) as well as pokrajac jpokroic (perfective) and krajacjkroic (imperfective) would be entered separately in the Polish part of the dictionary. Marginally, let us note that the difference between Seymour uzyl noza aby pokrajac salami and Seymour uzywal noza aby pokrajac salami does not consist in one action being completed in contrast with an incompleted one but rather in the iterative (repetitive) character of the action expressed by the verb. Thus, Seymour uzywal noza aby pokrajac salami would suggest "using a knife more than once on separate occasions" and could be rendered in English as "Seymour used to use a knife to slice the salami" On the other hand, uzywac may express the concept of duration as in Seymour uzywal noza przez trzy godziny 'Seymour was using knife for three hours' However, both kroic and pokroic will correspond to the English verb slice. (Of course, other English equivalents, such as cut, carve, etc., must also be included, but this fact is irrelevant to the problem of aspect correspondences.) The fact that both the imperfect verb kroic and the perfect verb pokroic correspond to only one form in English, which is aspect neutral, does not ruin our theory of equivalence, if the notion of

156

Chapter

VII

markedness is employed to constrain equivalence along the lines expounded below. The notion of markedness is involved in the Polish aspect in the following way: the so-called perfect verbs in Polish enter into oppositions with the imperfect ones, with perfect verbs being marked and imperfect ones being unmarked. This means that an imperfect verb can be used in all those situations in which a corresponding perfect verb can be used but not vice versa, as in (32) (32E)

Seymour krajal salami przez caly dzien. (unmarked) Seymour sliced the salami all day. (unmarked)

(33) (33E) (34) (34E)

Seymour Seymour Seymour Seymour

(35) (35E)

* Seymour pokrajal salami caly dzien. (marked) ?Seymour sliced up the salami all day. (marked)

krajal salami do skutku. (unmarked) sliced the salami to the end. (unmarked) pokrajal salami do skutku. (marked) sliced up the salami to the end. (marked)

Thus, the imperfect verb, being unmarked, can express both completed and incompleted actions, while the perfect verb, the marked one, can only refer to completed actions. In English, the completion of the action can be optionally marked by means of the particle up added after the verb as in (34E), which is then equivalent to (34) in that both are marked for completion. (32) and (32E) as well as (33) and (33E), which are respectively equivalent, are not marked for the feature [ ± completion]. Note, however, that (36)

Seymour pokrajal salami, (marked [ + completion])

and (36E)

Seymour sliced the salami, (unmarked)

are also equivalent, while (37)

Seymour bezskutecznie krajal salami, (marked [— completion])

and (37E)

Seymour sliced up the salami, (marked [ + completion])

are not equivalent. Under no circumstances can (37) and (37E) be considered as equivalent since they have conflicting marking polarity.

Towards a theory of semanto-syntactic

equivalence

157

The situation is somewhat more complex with respect to the verbs uzyc, zuzyc, and zuzywac. In this case, too, we deal with the opposition marked vs. unmarked, but in addition to the feature [ + completion] the feature [ + iterative] is also involved. In every case it is possible to mark polarity by three types of specification. A lexical item W may be positively specified for a feature F as [ + F], negatively specified for F as [ — F] or remain unspecified [ ± F], We will say that if W is either positively or negatively specified for F, it is thereby marked. Otherwise, it is unmarked. Adopting this marking convention, we can discover the following system of marking for uzyc, zuzyc, and uzywac, and their English equivalents: (38) [ + completion] [—completion] [ + iterative]

[— iterative]

marked

unmarked

zuzyc 'use up' uzywac 'use' miec zwyczaj uzywac 'use to use' (chadzac 'use to go') zuzyc 'use up'

uzyc uzyc uzyc (chodzic) 3 uzyc

(38) can be rearranged to spell out the features accounting for markedness in the following way: zuzyc

+ completion — iterative

use up

+ completion — iterative

miec zwyczaj uzywac

— completion + iterative

use to use

— completion + iterative

uzyc

± completion + iterative

use

+ completion ± iterative

uzywac

— completion + iterative

Note: the form miec zwyczaj uzywac, literally 'have the habit to use' is synonymous with *zwykngc uzywac. The latter, however, is defective as it never appears in the infinitival forms, In finite forms, such as zwykl uzywac 'used to use', it will appear in our examples as marked for [— completion, + iterative]. This system of marking allows us to formulate the following constraint on equivalence of lexical items: (C,) N o two lexical items will be assigned the status of equivalence if they have conflicting marking polarity with respect to a specific feature F.

158

Chapter

VII

In order to determine whether or not a lexical item Wj in a language L p (W L ) is equivalent to a lexical item Wj in a language L q (W L ) the constraint (Ci) has to be applied in the following way:

wqL

WPL

Equivalence

polarity

+ +

+

±

+

± ±

± +

+

Yes Yes Yes Yes Yes Yes No No

Constraint (C t ) will permit us to assign the status of equivalence to the following pairs of sentences: marking polarity: + c = ± completion + i = + iterative (40)

Seymour used the knife Seymour uzyl noza

± c ± i ± c ± i

(41)

Seymour used the knife Seymour uzywal noza

± c ± i - c ± i

(42)?

Seymour used the knife Seymour zwykl uzywac noza

± c ± i - c + i

(43)?

Seymour used the knife Seymour zuzyl ηόζ

+ c ± i + c - i

(44)

Seymour used to use the knife Seymour zwykl uzywac noza

- c + i - c + i

(45)

Seymour used to use the knife Seymour uzywal noza

- c ± i - c ± i

(46)?

Seymour used to use the knife Seymour uzyl noza

- c + i ± c ±i

(47)

Seymour used up the knife Seymour zuzyl ηόζ

+ c - i + c - i

(48)?

Seymour used up the knife Seymour uzyl noza

+ c - i ± c + i

Towards a theory of semanto-syntactic

equivalence

159

The following pairs, because of their conflicting polarity, will not be assigned the status of equivalents: (49)

Seymour used to use the knife Seymour zuzyl ηόζ

- c + i + c - i

(50)

Seymour used up the knife Seymour zwykl uzywac noza

+ c - i - c+ i

(51)

Seymour used up the knife Seymour uzywal noza

+ c - i - c ± i

As is to be expected, those sentences that share both polarities seem to be better equivalents than those that share only one polarity. The least acceptable as equivalents are those pairs which do not share any polarity, even if no conflict of polarity occurs. Such is the case with (42), (43), (46), and (48). It is possible that further constraints are needed in those cases when more than one polarity is involved. It may be required, for instance, that in such cases equivalent lexical items should share at least one polarity. Such a more severe constraint would rule out (43), (46), and (48) as equivalents. However, this problem has no bearing on the present discussion. In view of C t , the fact that some of the examples discussed by Bouton carry the implication that the action expressed by the verb was completed, while the English equivalents carry no such implication, does not lead to the conclusion that the relevant sentences have different semantic structures. At the level of sentence semantics, they can be said to have identical semantic representations. As long as at the level of word semantics these sentences do not violate Q , they can be recognized as equivalents despite noticeable differences in the meaning at the level of word semantics. Let us now turn to the second of Bouton's objections to the identicaldeep-structure hypothesis. According to Bouton, it can be saved only if one of the two mutually exclusive situations can be shown to be true: (a) English passive and active sentences can be derived from the same deep structure. In that event (28) and (29) would be considered as paraphrases, with (28 F) as their Finnish equivalent; all these sentences could then be derived from the same semantic structure, (b) The Finnish sentences such as (28F) and (30F) could be considered as ambiguous, each being derived from two different semantic structures, such as those underlying the English sentences (28) and (29), respectively. The first solution is impossible in view of the fact that ever since the publication of Katz —Postal (1964), English active and passive sentences

160

Chapter

VII

have been repeatedly demonstrated to be derived from different underlying structures. The second solution, according to Bouton, is also impossible as the Finnish verbs in (28) and (30), as well as in other similar sentences, observe selectional restrictions which correspond exactly to the selectional restrictions of English active sentences. They do not observe any selectional restrictions that characterize passive sentences. Thus, in the first place, Finnish verbs are not subject to any stativity restrictions, which characterize English passive sentences. In the second place, intransitive verbs can occur in both English active sentences and in Finnish sentences, but not in English passives, so that we can have: (52) (53)

Sunnuntaina mennaan kirkoon. 'On Sunday people go to church' Suomessa hudetaan paljon. 'In Finland one skis a lot'

But not (54) (55)

* Was laughed loudly by Bill, (from Bill laughed loudly.) *It sat here all day. (from Someone sits here all day.)

Therefore, Bouton concludes that the identical-deep-structure hypothesis fails since the translation equivalents (28), (29) —(28F), and also (30), (31) —(30F) cannot be shown to have identical deep structures: either (28) and (29) have two different deep structures, in which case (28F) has the identical deep structure with only one of them, or, quite implausibly, (28F) and (30F) must be recognized as ambiguous. The latter solution is impossible to adopt as the Finnish sentences observe selectional restrictions which characterize active sentences and do not observe any of those restrictions which characterize passive sentences. Therefore, (28F) and (30F) cannot be interpreted as ambiguous and can each have only one deep structure. Since, however, they each correspond to at least two English sentences, each having a different source, the identical-deepstructure hypothesis fails as at least one of the translation equivalents of both (28F) and (30F) is derived from a different deep structure. It must be admitted that Bouton's reasoning is impeccable and that the sentences in question cannot be claimed to have an identical deep structure as established through the examination of selectional restrictions since these restrictions are quite obviously different for active and passive sentences. However, the problem must be seen somewhat differently and must involve the decision as to what in Finnish corresponds to the English

Towards a theory of semanto-syntactic

equivalence

161

sentences (28), (29), (30), and (31), respectively. In order to answer this question, it will be necessary to examine the semantic content of English passive sentences and the ways in which the same content, if at all, is rendered in Finnish. Such problems can be formulated and answered within linguistic models in which deep structure is not an independent, well-defined, and well-delimited level of analysis, as in the standard theory, but only within such models in which sentences are derived from semantic representations of a more abstract nature. Bouton's reasoning, as well as arguments presented by other authors, including the present one (cf. Van Buren 1974; Krzeszowski 1974, 1979, see also the previous chapter) show that the deep structure as formulated in the standard theory cannot serve as a possible tertium comparationis for conducting syntactic contrastive studies. It should be noted in this connection that if two sentences across languages do not observe the same set of co-occurrence and selectional restrictions, it does not necessarily follow that they do not share an underlying structure at some level of linguistic representation. It only means that they do not share that level of analysis at which these restrictions are stated, i.e. the level of the standard theoretical deep structure. Therefore, if two sentences do share the same set of cooccurrence and selectional restrictions, they share deep structure and, in all probability, also more abstract levels of representation, ipso facto being equivalent. However, the fact that two sentences do not share the same deep structure does not rule out the possibility of their sharing a more abstract semantic structure and thereby being equivalent, as well. Nevertheless, considering the constraints on equivalence formulated below, we shall try to show that the technique of establishing equivalence through testing co-occurrence and selectional restrictions obtaining in pairs of sentences across languages is valid, even if more abstract levels of representation are accepted as tertium comparationis for syntactic contrastive studies. At this point, it is necessary to make a digression in order to recall the essential distinction drawn in Chapter II between translations and semanto-syntactic equivalents. The term "equivalent sentences" refers to those sentences in two or more languages which have identical input structures; this is not synonymous with the term "translations" in the actual translation practice. The ability to recognize equivalents is a part of a bilingual person's competence, while ability to translate is a part of translation performance. In actual translation practice, sentences which are translations often differ in their meaning, not necessarily due to errors

162

Chapter

VII

in translation, but owing to a number of grammatical, stylistic, and pragmatic reasons, which result in frequent instances of translations being semantically non-equivalent. For instance, a line from Pygmalion: (56)

Nobody's going to touch you

is translated by Florian Sobieniowski as (56P)

Nie böj si§ nie damy ci krzywdy zrobic. 'Don't fear, we won't let anybody hurt you'

(56) and (56P) are obviously semanto-syntactically non-equivalent, but (56P) does appear as the actual translation of (56). At this point we do not intend to discuss all the implications of examples such as (56) and (56P), especially for text-bound contrastive studies (cf. Chapter III). In this chapter we are not dealing with such obvious deviations from semanto-syntactic equivalence. They usually result from the translator's conscious attempt to preserve invariables other than semanto-syntactic in his translation and only marginally from faulty competences. In any event, such conscious deviations from equivalence do not constitute counterexamples to the identical-deep-structure hypothesis. On the contrary, they confirm the empirical reality of equivalence as manifested in the translator's awareness that his translation does not convey all the invariables of the original text. In Chapter III, we said that semanto-syntactically equivalent sentences are those which are the closest approximations to acceptable word-forword translations and their synonymous paraphrases (if indeed such paraphrases exist). Let us rephrase this description of equivalent sentences in the form of the following Constraint 2 (C2): (C2) No two sentences will be assigned the status of semanto-syntactic equivalents unless they are the closest approximations to acceptable word-for-word translations and their synonymous paraphrases, if such exist. The two constraints seriously restrict the scope of data to be accounted for by a possible theory of equivalence. Furthermore, since it will be required that equivalent sentences have an identical deep structure, the semanto-syntactic equivalence will be constrained in the following two ways: (1) On the surface it will be constrained lexically (by C t ) and structurally (by C2).

Towards a theory of semanto-syntactic

equivalence

163

(2) In the sentence-semantic structure it will be constrained by the requirement that equivalent sentences (in the sense of (1)) have identical sentence-semantic structure. In view of these constraints on semanto-syntactic equivalence, Bouton's examples involving English active and passive sentences and their Finnish counterparts no longer constitute counterexamples to the identical-deep-structure hypothesis. More precisely, these examples do not jeopardize its empirical validity. Only (28) and (28F) and (30) and (30F), respectively, are equivalent in the constrained sense, (29) and (31) are not equivalent to (28F) and (30F), respectively, since they are not the closest approximations to grammatical word-for-word translations and, as was said earlier on, they manifest different co-occurrence and selectional restrictions. Therefore, (28) and (28F) will be derived from some identical semanto-syntactic structure and so will (30) and (30F), while (29) and (31) must be derived from some different structures. In English (28) and (30) are derived from a similar semanto-syntactic structure, and so are (29) and (31). The only difference in both pairs is that in (28) and (29), but not in (30) and (31), the agent is specified. Thus, on the one hand we have: (28)

Tom broke the window.

(29)

The window was broken by Tom.

with the specified agent, and (30) (31)

Someone broke the window. The window was broken (by

someone).

with an unspecified agent. On the other hand, the four sentences can be grouped pairwise, according to the underlying semanto-syntactic structure, thus: (28)

Tom broke the window.

(30)

Someone broke the window.

realizing the semantic structure. SI: A G E N T VERB PATIENT and in contrast with (29) (31)

The window was broken by Tom. The window was broken by someone.

realizing the semantic structure

164

Chapter

VII

S2: PATIENT STATE PREPOSITION AGENT. 51 and S2 are, of course, drastically oversimplified, but they adequately represent the relevant difference in the semantic content of the discussed sentences. Only SI can be realized in Finnish, but unlike in English, in Finnish SI can be realized in the surface structure exemplified by (28F) if Agent is specified and only as (30F) if Agent is not specified. Thus in Finnish, if Agent is unspecified, it is obligatory deleted and the obligatory "impersonal" transformation yields (30F). If Agent is specified, sentences like (28F) result (cf. Kudzinowski 1978: 90: "The Finnish passive is not in fact an impersonal class; one always associates it with some personal agent" [translation is my own]). 52 is never realized in Finnish while the translations such as (31) — (30F) constitute further examples of non-equivalent sentences functioning as translations. It might be added that in Polish the "impersonal" transformation is optional, so that both (57)

Ktos wybil okno. 'Someone broke the window'

and (58)

Okno wybito. 'Window broken (by someone)'

are possible. However, unlike in Finnish, in Polish realizations of S2 are also possible: (59) (60

Okno zostalo wybite przez Tomka. 'Window was broken by Tom' Okno zostalo wybite (przez kogos). 'Window was broken (by someone)'

It may also be worth noting that the English passives, having no equivalents in Finnish, are not the only instance of semanto-syntactic gaps. The so-called perfective tenses, which also exhibit properties of stative constructions, are, for the most part, rendered in other languages by means of constructions in which similar forms either do not appear at all (as in Finnish and Polish), or they do not observe the restrictions characterizing stative verbs (as in German and French). Therefore, German sentences containing "Perfekt" tenses and French sentences containing "passe compose" must be derived from semantic structures which

Towards a theory of semanto-syntactic

equivalence

165

underlie agentive constructions rather than from semantic structures underlying Stative constructions, as in the case with the English presentperfect tense and possibly other perfective forms. The problem with the English aspect (as well as with the passive sentences in English and their Finnish translations) suggests another important difference between translations and semanto-syntactic equivalents, which consists in the fact that whereas some kind of translation of any given sentence into another language is always available (even if in the process of translation a great number of various linguistic properties at various levels are not preserved), equivalent sentences across languages are not always available due to the existence of semanto-syntactic gaps in various languages. Of all possible sentence-semantic structures each language realizes only a subset proper to that language. For instance, English passive sentences do not have equivalents in Finnish; likewise sentences containing stative (perfective) tenses can hardly be considered as equivalent to any sentences in other languages. Such a constrained view of equivalence lends further rigour to contrastive studies but also severely limits their usefulness, at least in their present form. The constraints on equivalence (C, and C 2 ) and the distinction between semanto-syntactic equivalence and translation allow us to dismiss all the examples discussed above, including Bouton's active and passive sentences in English and their Finnish translations, as irrelevant to the identicaldeep-structure hypothesis. All the remaining examples discussed by Bouton can be dismissed on similar grounds. For example, the English translation (61)

Where have you been?

of the Persian question (61 Pr)

Az koja miyaid?

is certainly non-equivalent. The closest approximation to the grammatical word-for-word translation would have been (61')

Where do you come from?

and (6Γ) rather than (61) should be regarded as the semanto-syntactic, though not necessarily translational or statistical (see Chapter X), equivalent of (61 Pr). Finally, to invalidate the identical-deep-structure hypothesis, Bouton quotes an extremely interesting example of answers to negative questions in English in contrast with other languages, including Korean. 4 English

166

Chapter

VII

negative questions are answered yes if the action mentioned in the question actually took place and no if it did not. The answers to equivalent questions in Korean correspond, respectively, to no and yes. In Korean (and in Polish) the answerer affirms or denies the existence of the situation to which the question refers. On the basis of these facts Bouton concludes answers of opposite polarity, which seem to be responding to different aspects of interrogative situation and in that sense have a different meaning from each other, should not be assigned to a common underlying structure merely because they are translation equivalents (Bouton 1976: 158-159). They should not be assigned a common structure indeed! And there is nothing in the present proposal that would induce anyone to take such an unmotivated step. Unfortunately, my knowledge of Korean is too limited to enable me to analyse self-made examples, and I have no access to native informants. Therefore, I shall confine the subsequent discussion to the Polish material, trusting that the situation in Korean is similar in the relevant respect and as such does not require a separate treatment from the point of view of the theory of semanto-syntactic equivalence. Bouton's examples concerning answers to negated questions in English (and perhaps in Korean too) does not seem to be complete. The dialogues (62)

A: Didn't you go to school today? B: Yes. (if Β did go to school)

and (63)

A: Didn't you go to school today? B: No. (if Β did not go to school).

do not sound like genuine dialogues, and especially in the answer in (62) appears to be vague and is likely to be followed by A's Yes, what?, meaning Did you or didn 't you go to school? In Polish, likewise, the plain Tak 'Yes' and Nie 'No' are insufficient as answers: (62P)

(63P)

A: Nie poszedles dzis do szkoly? B: Nie, poszedlem. (if Β did go to school) 'No, I went' A: Nie poszedles dzis do szkoly? B: Tak, nie poszedlem. (if Β did not go to school) 'Yes, I did not go'.

Towards a theory of semanto-syntactic

equivalence

167

There are two ways of handling these examples, depending on what one expects as correct answers in (62P) and (63P): 1. In my own idiolect the answers Tak 'Yes' to the question in (62P) and the answer Nie 'No' in (63P) are also possible. Therefore, in my idiolect answers in (62P) and (63P) could be equivalent to those in (62) and (63), respectively. Otherwise, in those cases when Nie 'No' corresponds to Yes in English and Tak 'Yes' to No in English, we would have to recognize again the presence of semanto-syntactically non-equivalent translations. In view of the distinction between semanto-syntactic equivalence and translation, these instances would not constitute counterexamples to the identical-deep-structure hypothesis, especially in consideration of the fact that that word-for-word lexical equivalents would also be available. 2. In the case of those speakers in whose idiolects only Nie in (62P) and only Tak in (63P) are acceptable as answers, the following solution may be offered. It is true, as Bouton suggests, that The difference in the response in the two languages seems to be that the English speaker affirms or denies the existence of the situation about which the question asks; the Korean [and the Pole] affirms and denies the accuracy of the statement underlying the question (Bouton 1976: 158). These affirmations and denials are in fact present in more complete English dialogues, in which complete answers could be, respectively, Yes, I did (meaning: Yes, I did go to school today) and No, I didn't (meaning: No, I didn't go to school today) with the ellipses of the repeated material. In Polish, indeed, the affirmation and the denial may refer to the form of the question itself and to the statement implied in the question rather than to the actual situation. Therefore, the Polish answer actually means Nie to nieprawda, ze nie poszedlem do szkoly, bo poszedlem 'No, it is not true that I did not go to school because I did go' and Tak to prawda, ze nie poszedlem do szkoly, bo nie poszedlem 'Yes, it is true that I did not go to school because I didn't' In real conversations, nobody ever gives such full answers unless there is a reason motivating a complete answer. However, such full answers, which presumably underlie the short ones, do constitute equivalents across the languages. The shorter answers may not be equivalent, but often function as translations. Competent bilingual informants are well aware of this lack of semantic equivalence. However, as was repeatedly emphasized before, the identical-deep-structure hy-

168

Chapter

VII

pothesis is not intended to account for translation but for semantosyntactic equivalence. The discussion of Bouton's counterexamples shows that the hypothesis can be defended if certain constraints are imposed on equivalence, and if translation performance is distinguished from bilingual informant's competence in recognizing semanto-syntactic equivalents across languages. Therefore, it is natural to conclude that any theory of semantosyntactic equivalence which would assign different semantic structures to equivalent sentences would be false since it would fail to explain the bilingual informant's competence in distinguishing semanto-syntactically equivalent sentences from non-equivalent ones. In the next chapter, we shall outline a theory based on the asumption that all semanto-syntactically equivalent sentences have identical semantic structures. Thus, the identical-deep-structure hypothesis will serve as the initial postulate (axiom) of the theory. To avoid terminological confusions, we are going to abandon the term "deep structure" (which ought to be used only in the context of Chomsky's standard theory), and we shall use the self-explanatory term "semantic structure" (referring to sentence semantics) or the term input structure. It ought to have become clear by now that Chomsky's deep structure is a level of representation considerably "shallower" or less abstract than the level of representation which has emerged from the aforegoing discussion.

Chapter VIII

Contrastive Generative Grammar

At this stage we are ready to introduce the concept of "Contrastive Generative Grammar" as a device explicating semanto-syntactic equivalence as described in the previous chapter. 1 Since equivalent sentences are derived from identical semantic representations, a claim which we take to be axiomatic, it necessarily follows that whatever structural differences between them occur must be due to the operation of various language-specific rules at various levels of derivation in the respective grammars of particular natural languages. A generative grammar of a particular natural language may be conceived as a diversifying device, whose input are semantic representations of sentences (or, briefly speaking, input structures), generated by the universal semantic component, and whose output are well-formed sentences characterizing a particular language. The level at which two particular equivalent sentences are diversified for the first time marks the extent to which such sentences are similar: the lower (i.e., the less abstract) the level of the first diversification, the more similar the equivalent sentences are since they share more rules in their respective derivations (see also Di Pietro 1971: 27). Let us look at some examples. First, consider the following two sentences in English and Polish, respectively: (1) (2)

They painted the fence white. Pomalowali plot na biato.

These two equivalent sentences are first diversified at the level of the categorial component, i.e. the level at which the rules assigning grammatical categories to various portions of semantic representations operate. Both (1) and (2) have identical semantic structure consisting of the roles Agent and Patient and of a Predicate. 2 The categorical component assigns the categories Noun Phrase to Agent and Patient in both languages. In contrast with Polish, however, a part of the English Predicate is assigned the category Adjective, eventually realized lexically as white whereas in

170

Chapter

VIII

Polish the corresponding part of the Predicate is assigned the category (Prepositional) Noun Phrase. Thus, the first diversification of (1) and (2) occurs at the level of the categorical component. The sentences (3) (4)

He was asked a lot of questions. Zadano mu mnostwo pytah.

are not diversified at the level of the categorical component since in both of them there occurs an equal number of major grammatical categories as realizations of various semantic roles and predicates. Thus, he of (3) corresponds to mu of (4), a lot corresponds to mnostwo, of questions corresponds to pytan, and was asked to zadano. Assuming that both sentences are derived from an underlying structure which, when spelled out in English would read as (5)

Somebody asked him a lot of questions

where no diversification occurs yet, we observe that (3) and (4) are diversified for the first time at the level of syntactic transformations. In order to generate (3) it is necessary to apply the English passive transformation, and in order to generate (4) the Polish 'impersonal' transformation must be applied. Finally, a pair of sentences like (6) (7)

One of the signatures was illegible. Jeden ζ podpisow byl nieczytelny.

are diversified for the first time at the level of lexical insertions. In languages which are genetically closely related (Slovak and Polish, Spanish and Italian, etc.) diversifications at even later stages are possible. Di Pietro (1971: 27) quotes a poem entitled La gloria del momento by Juseph Tusiani, which is written in such a way that it can be either Spanish or Italian. The only diversifications occur at the level of phonological rules: the poem sounds different when read aloud. In writing, only certain accentual marks identify it as Spanish. Contrastive Generative Grammar is to explicate contrasts at various levels of derivation and provide an explicit account of semanto-syntactic equivalence. In order to fulfil these tasks, Contrastive Generative Grammar must be capable of achieving some more specific aims. We state these aims in the form of five postulates characterizing it:

Contrastive

Generative

Grammar

171

(1) If L; L n is a set of natural languages, Contrastive Generative Grammar must recursively enumerate sentences in any L; and Lj; this means that for every sentence, the grammar must decide whether or not the sentence has been generated either by G; or by Gj, where Gj and Gj are generative grammars of Lj and Lj, respectively. (2) For each sentence in L, and Lj, Contrastive Generative Grammar must assign one or more structural descriptions (each ambiguous sentence must receive as many structural descriptions as there are ways in which it can be disambiguated). These first two postulates characterizing Contrastive Generative Grammar entail a definition of generative grammar for a particular language Lj (cf. Chomsky 1965: 3ff.). The remaining three postulates define Contrastive Generative Grammar as a device different from something which is merely a collection of generative grammars of η languages: (3) For each pair of sentences in L, and Lj, Contrastive Generative Grammar must determine whether these sentences are equivalent. At this point, we recall again the distinction between semanto-syntactic equivalence and translation drawn and described earlier (see Chapters II and VII). It is to be expected that semanto-syntactically equivalent sentences will have identical parts in their structural descriptions, which follows from the assumption about the universality of input structure and from the empirical observations that there exist similarities between specific grammars. Therefore, Contrastive Generative Grammar must also do (4): (4) For each pair of equivalent sentences in Lj and Lj? it must specify those parts of the equivalent structural descriptions which are identical and those which are not. In other words, it must note the level of derivation at which the first diversification occurs. It will do so by scanning the derivation, beginning with the semantic input and moving by the successive strings, generated by successive rules, where any output of a rule (except the terminal ones) constitutes input to a successive rule. Let L 0 be original input, common for both Lj and Lj, and let R n and R m be sets of rules in L; and Lj? respectively. Then for Lj, each I p _, will be an input to a rule R p whose output is O p and each O p will serve as input Ip to a rule R p + 1 , unless O p is terminal. In this way I 0 will be the input to R,, whose output will be O,, which will be the input i! to R 2 , whose output will be 0 2 , etc. Similarly, for Lj? each output O q of R q will serve as input I q to a rule R q + i , unless O q is terminal.

172

Chapter

VIII

In order to fulfill postulate (4) Contrastive Generative Grammar will have to inspect the derivational history of each equivalent sentence, beginning with the semantic input I c . If a particular Ioi for a sentence S p in Li also serves as input for a sentence S p in Lj, then the two sentences are equivalent in accordance with postulate (3). The grammar then scans further rules operating on Ioi to stop at the place where the output of a rule R k in Lj is not identical with the output of a rule R k in Lj. It will thus note that place in the derivation where the first diversification occurs. (5) For each pair of pairs of sentences in L; and Lj, Contrastive Generative Grammar may also determine the degree of similarity according to the relation "more similar", where this relation is equivalent to the relation "diversified at a lower level" Thus, for each pair of pairs of sentences S p in L; Sq in Lj and St in L; Sk in Lj, where the double arrow represents equivalence in terms of the identity of input structures, Contrastive Generative Grammar must determine which of the two pairs is more similar, i.e., diversified at a lower level of derivation. Postulate (5) cannot be fulfilled without having all the rules of the grammar extrinsically or intrinsically ordered. Since extrinsic ordering of the rules has been shown to be superfluous in explaining facts about natural languages (see, for example, Koutsoudas 1972: 88 — 96), the full realization of this postulate is dependent upon the construction of a grammar in which all rules would be intrinsically ordered. It is not clear whether such a grammar can be constructed at all. Although the existence of postulate (5) has no bearing either on the theory of semanto-syntactic equivalence or on the theory of translation, it may prompt research strategies for constructing hierarchies of difficulties in second-language learning, e. g., it may clarify some problems involved in negative and positive transfer (see Chapter IX). It may also make it possible to calculate a "similarity index", i. e., the ratio of identical and different rules employed in various languages (cf. Di Pietro 1971: 27). In order to fulfill these postulates, Contrastive Generative Grammar must be based on suitable theories of the languages to be contrasted. Thus, in order to fulfill postulate (3), it must be based on a model of linguistic description in which the semantic input is universal. In order to fulfill postulate (4) it is, furthermore, required that all generative grammars of particular natural languages should be based on categoryneutral semantic input represented in terms of a metalanguage free from

Contrastive

Generative

Grammar

173

such notions as noun phrase, verb phrase, tense, modal, etc., since these grammatical categories are neither universal nor semantic. Postulate (4) can be fulfilled only if all generative grammars introduce grammatical categories at the same level of derivation, i.e., the level intermediate between the semantic representation and syntactic transformations. It is, therefore, to be anticipated that the organization of each generative grammar will require at least five levels of representation: (1) semantic, where the fundamental semantic relations, i. e., the meaning of sentences is represented in the form of a universal, category-neutral semantic input to sentence derivation; (2) categorial, where languagespecific rules assign various categories, such as noun phrase, verb, adjective, tense, modal, etc., to various portions of the semantic representation; (3) syntactic, where syntactic transformations arrange the major syntactic categories (nouns, verbs, adjectives, and perhaps adverbs) in the linear order in which they appear in actual sentences and introduce some of the minor categories (prepositions, auxiliaries, and perhaps some adverbs); (4) lexical, where lexical items are inserted from the dictionary, in which lexical entries are defined in terms of syntactic frames of grammatical categories; (5) post-lexical, where "cosmetic" transformations arrange minor syntactic categories in the linear order in which they appear on the surface, provide words with inflectional endings, and introduce word boundaries. Anticipating our discussion in Chapter IX, let us note here that not all languages seem to require an equal number of rules at these five levels in order to derive sentences from the semantic representations. For instance, there are languages remarkably poor in inflections (Chinese, English), in which the number of "cosmetic" transformations seems to be negligible in comparison with highly inflected languages (Russian, Polish, Latin). Various "reduced" registers (telegraphese, baby-talk) and pidgins seem to function without many syntactic transformations which account for complex structures and multiple embeddings. Such languages are derivationally "shallower" than fully elaborated codes. It is possible that in such languages, lexical entries are inserted directly into semantic structures rather than into outputs of transformational rules, which account for most structural complexities of fully elaborated codes. Another feature of Contrastive Generative Grammar is that it must be based on text-grammars in order to remove ambiguity. No sentencegrammar (i.e. a grammar generating sentences rather than their sequences and longer stretches of texts) is capable of serving as a foundation for Contrastive Generative Grammar even in principle, because it does

174

Chapter

VIII

not provide grounds for determining whether or not particular sentences across languages are equivalent in those cases when they are ambiguous. For example, given an ambiguous sentence in a language Li? Contrastive Generative Grammar will be unable to match it with an equivalent sentence in Lj without analysing the ambiguity of the sentence in L;. The analysis may reveal that a sentence Sk in Lj is in η ways ambiguous, and that a sentence Si in L, is in m ways ambiguous. The analysis may also reveal that some of the η readings of Sk are equivalent to some of the m readings of S; in some contexts but not in others. The determination of which specific reading is involved cannot be effected without inspecting stretches of texts longer than a sentence. Suppose, for example, that Contrastive Generative Grammar confronts the English sentence (8)

Drinking water can be dangerous.

which is ambiguous in at least two ways. One reading roughly corresponds to (9)

The drinking of water can be dangerous.

while the other meaning corresponds to (10)

Water for drinking can be dangerous.

In Polish (8) cannot be rendered by means of one, equally ambiguous, sentence. One of the following two must be chosen: (11)

Woda do picia moze bye niebezpieczna.

(12)

Picie wody moze bye niebezpieczne.

Without inspecting some disambiguating contexts in which (8) appears, and which stretch beyond sentence boundaries, Contrastive Generative Grammar will be unable to choose either (11) or (12) as the equivalent of (8).3 In order to fulfill postulate (3), Contrastive Generative Grammar must, furthermore, be based on a generative grammar which accomodates the notions of reference and co-reference. The notion of reference must be understood as an association obtaining between noun phrases of a natural language and mental entities present in the language-user's mind. In

Contrastive

Generative

Grammar

175

languages without noun phrases (if such indeed exist) 4 , this association obtains between some other appropriate syntactic categories and respective entities in the language-user's mind. This concept of reference follows from the realization of the fact that the relation between syntactic categories and what they refer to in the world at large is a composite one. 5 Thus, it is necessary to distinguish noun phrases themselves, entities in the world at large, and an intermediate set of abstract entities inhabiting the language-user's mind. The composite character of reference consists in the fact that noun phrases are not directly associated with entities in the world, but they refer to an intermediate set of conceptual mental entities, which in turn can be related to entities in the world at large. This concept of reference provides a uniform way of treating fictitious and non-fictitious, living and dead, existing and extinct characters, objects, and events as there is no syntactic evidence in support of the view that these various kinds of phenomena require a fundamentally different semantic treatment within the realm of linguistics (cf. Sampson 1970b: 13). The differences can be formulated only in non-linguistic terms and consist in the fact that some concepts can be associated with objectively and currently existing phenomena in the world at large while some cannot. This view of reference enables us to associate mental entities we shall call them referents, with a set of integers from 1 to η in such a way that each referent is labelled with a unique integer, the cardinality of the set of referents formed in this way being definite at any given moment for an individual language user and determined by his former linguistic career. It becomes possible to associate each referent with one or more linguistic expression by attaching respective indices to those portions of derivations which underlie these expressions. 6 Information of this sort is indispensable for discovering equivalent sentences across languages; it is, therefore, relevant and necessary in constructing Contrastive Generative Grammar (for more details see Krzeszowski 1979: 28 ff.). Let us look at some English and Polish sentences in order to see how the equivalence relation between them can be established only if larger stretches of texts over which referential indices are established are taken into account. For clarity, we shall use only examples in which the relevant portion of the text appears in the form of a sentence immediately preceding the examined sentence, although this is by no means the only possible situation. Let us consider sentences containing referentials, i.e., words which refer to other words in the text. On numerous occasions, Contrastive Generative Grammar will be unable to decide whether two sentences containing referentials are equivalent without finding which

176

Chapter

VIII

words in the neighbouring sentences the referentials refer to, i. e. without establishing co-reference. Confronting sentences like (13)

They do not need mending yet (Krzeszowski 1979: 21).

Contrastive Generative Grammar is incapable of deciding on the correct choice of the Polish equivalent without examining the antecedent of they which may occur in the previous sentence. If the antecedent of they is a word such as clothes, then depending on which of the possible four lexical equivalents of clothes in Polish is selected in the preceding sentence, one of four personal pronouns will have to be selected: (14) (15) (16) (17)

Ten ubior (masc.) jest calkiem nowy. On nie wymaga jeszcze naprawy. Ta garderoba (fem.) jest jeszcze calkiem nowa. Ona nie wymaga jeszcze naprawy. To ubranie (neut.) jest jeszcze calkiem nowe. Ono nie wymaga jeszcze naprawy. Te ciuchy (plur.) sg. jeszcze nowe. One nie wymagajg. jeszcze naprawy.

In order to select the correct equivalents of sentences containing referentials, Contrastive Generative Grammar must match those sentences with other sentences appearing in the same text and containing words having the same reference as the referentials for which it seeks to find equivalents. This procedure can only be effected if the information concerning the referential identity of particular fragments of texts is contained in the semantic inputs to those texts. This kind of information cannot be presented at the level of syntactic representations since the notion of reference is semantic in nature (Hiz 1968: 5 ff.). It follows that Contrastive Generative Grammar cannot enumerate equivalent sentences in two or more languages and thus fulfill postulate (3) if the semantic structure of these sentences does not contain information concerning coreferentiality of nominal concepts not only within the boundaries of sentence derivation but also within the boundaries of a particular text. In brief, a linguistic model which is to serve as the foundation for Contrastive Generative Grammar must fulfill the following requirements: 1. It must contain a semantic base, generating universal semantic representations as inputs to language-specific grammatical rules. The

Contrastive

Generative

Grammar

177

semantic base serves as tertium comparationis for establishing semantosyntactic equivalence. Being universal, it must be free of languagespecific syntactic categories appearing at less abstract levels of representation. 2. The semantic input must contain information about reference since reference is a relevant concept in establishing semanto-syntactic equivalence of sentences across languages. The model cannot be restricted to generating sentences, but must be capable of generating longer stretches of texts (sequences of sentences), relevant in establishing reference and co-reference. In Krzeszowski (1974, 1979), I suggested a model of linguistic description, which, in my view, met the requirements of Contrastive Generative Grammar and, in contrast to other currently available linguistic frameworks, could serve as the foundation for it. Specifically, the model was based on a conversion grammar, that is, a device which enumerates wellformed sentences and their sequences by mapping an infinite set of semantic inputs onto an infinite set of well-formed sentences and their sequences by imposing conditions of well-formedness upon semantic inputs in terms of well-formedness of sentences, which constitute the output of the grammar. Such a concept of output-monitored generative grammar presupposes an infinite number of semantic inputs representable in terms of, for example, graphs arranged in a finite number of patterns but endowed with the property of self-embedding to ensure the infinity of their combinations. Such graphs serve as inputs to a finite number of conversion rules, whose purpose is to filter out ill-formed configurations of graphs and convert well-formed configurations of graphs into surface structure sentences and their sequences. In terms of a framework of this kind, well-formedness of graphs is determined by their suitability to serve as inputs to conversion rules, which have to be constructed in such a way as to guarantee the enumeration of all and only well-formed sentences and their sequences in a given language. Conversion rules select certain configurations as inputs to derivations and assign grammatical categories to particular portions of these configurations. In the present framework the expression "conversion rules" embraces all types of rules required in a grammar to convert semantic inputs (configurations of graphs) into sentences of a particular language. These rules will include universal semantic categorical rules, language-specific syntactic categorical rules, syntactic transformational rules, lexical insertion rules, post-lexical transformational rules, and morphographemic/morphophonemic rules.

178

Chapter VIII

All inputs to all conversion rules (derivations) are constructed according to the following general patterns, which we shall call the original configuration:

(18)

·Α

•ρ

In the original configuration, the dots are uniquely associated with roles, which are primitive, axiomatic notions, solely characterized by their positions on the original configuration. These roles are: the uppermost A(gent), the one below A P(atient), the one below Ρ R(esident), the three roles below R fL (from Locus), atL (at Locus) and tL (to Locus) from left to right, respectively. Both fL and tL are mutually exclusive with atL, when dominated by the same R. In other words, neither fL nor tL can be dominated by the same R as atL. This general pattern is characterized by the property of self-embedding, which means that some configurations may function as referents in other configurations. A configuration which functions as a role in another configuration is called a subconfiguration. Subconfigurations are assigned role labels according to the pattern of the original configuration. In other words, the position of each dot can be occupied by a referentially indexed subconfiguration as one of its elements, etc., in principle ad infinitum. All subconfigurations receive not only role labels, but also referential indices. Thus, it is possible to have the following configuration:

Contrastive

Generative

Grammar

179

(19)

In (19) there is a subconfiguration in the position of Resident, and it contains another subconfiguration in the position of Patient. All configurations which constitute inputs to derivations are structured according to the original configuration. Every configuration which is accepted by the conversion rules as an input to derivations consists of at least one nuclear subconfiguration and of a certain number of nonnuclear subconfigurations and elementary roles. A nuclear subconfiguration is a subconfiguration which contains no subconfigurations. An elementary role is a role which is not a subconfiguration. A nuclear role is a role within a nuclear subconfiguration. In (19) 3 is a non-nuclear subconfiguration and Resident; 1 (Agent), 2 (Patient), 4 (Agent), 11 (Resident), 12 (at Locus), 13 (from Locus), and 14 (to Locus) are nuclear roles; and 5 (Patient) is a nuclear subconfiguration. Universal semantic categorical rules operate upon initial configurations and assign role labels to elementary roles and to subconfigurations, erasing those elementary roles which are not referentially indexed. The universal categorical rules assign roles to dots and circles of every con-

180

Chapter

VIII

figuration in accordance with the patterning of the original configuration. They also assign some universal semantic categories, such as Manner (to the Locus directly dominated by each nuclear subconfiguration), Time (to the Locus dominated by the outermost subconfiguration), and either Time or Place to every Locus of every subconfiguration, except the nuclear and outermost subconfigurations. These rules, moreover, assign an index to every referentially non-empty role: α — β—

γ —

δ — ε —

φ —

which refers to the speaker T , is always assigned to the A within the outermost subconfiguration and to all coreferential nodes; which refers to the content of the speech act uttered by a, is always assigned to Ρ and R (which are later conflated by another rule) within the outermost subconfiguration; which refers to the addressee of the content of the speech act ('you'), is always assigned to the tL within the outermost subconfiguration and to all coreferential nodes; which refers to the speech act, is always assigned to the R labelling the outermost subconfiguration; which refers to the location of the speech act in time, is always assigned to the Time Locus (TL) dominated by the outermost subconfiguration (within every configuration) as well as to all tTL's, fTL's, and atTL's; which refers to an unidentified but specific person (not T , not 'you'), that is, to an unidentified third person, may be assigned to any role which has not been mentioned above.

Moreover, indices from 1 to η (and a few others which we shall ignore here to simplify the description of the framework) may be assigned to all the remaining, referentially non-empty roles in all configurations underlying a particular text. All roles with no referential indices are erased, while coreferential roles occurring in the same subconfiguration are conflated (like PRß above). The resulting objects can be represented as branching trees, such as those used in the standard theory. Thus, the first universal rules filter out a large number of all possible configurations and in doing so they specify a subset of all configurations, which can serve as inputs to subsequent rules. They also construct diverging trees (which can also be represented as labelled bracketed strings), serving as inputs to language-specific categorical rules. For example, the text in (20): (20)

I stroke a dog. The dog was small.

Contrastive

Generative

Grammar

181

is derived from an initial configuration, which after the application of all the relevant universal semantic rules can be represented in the form of the following branching (diverging) tree: Σ

(21)

Αα Δ Ra Δ Aa Δ

Ρ3Δ tTL6R3 AatL5AtTLEA tLyA atTLc

where Σ stands for 'sequence of sentences within one performative complex' — in this case declarative, and where A's correspond to traditional predicates. (21) can also be represented as a labelled bracketed string (22):

(22)

Z(R 8 (A a Δ PRß(R a Δ t U R J A . Δ P3] Δ tTL c R 4 [R 3 Δ atL 5 ] Δ tTL c ) Δ tLY)) Δ atTL e )

which should be interpreted as referring to the following situation: The sequence of sentences (Σ) places the speech act (R s ) at the moment of speaking (atTL c ), that is, 'now', and it embraces the content of the speech act (PRß) directed by the speaker (A a ) to the addressee (tL 7 ). Thus, (22) can be spelled out as: I(A a ) declare that I(R a ) assent to the proposition (tL,) that there occurred the event (R 2 ) of I(A a ) stroking a dog (P3) at a certain time before 'now' (tTL,.) and the event (R 4 ) of the dog (R 3 ) being in the state of smallness (atL 5 ) at the time before 'now' (tTL c ). It must be noted that the content of the two nuclear subconfigurations, indicated by the square brackets, when spelled out in ordinary English words, is restricted to I stroke a dog (R 2 ) and The dog be small (R 4 ),

182

Chapter

VIII

respectively. This means that neither sentence modality nor tense and aspect are included since they are, whenever necessary, present elsewhere in the appropriate strings but outside nuclear subconfigurations. We emphasize this point here since it will be crucial in understanding the line of reasoning in the next chapter. Subsequent language-specific, categorical, and transformational rules assign grammatical categories to labelled objects such as (21). The process is too complicated to be described briefly here, and in fact the details of the derivation need not concern us. Suffice it to say that these rules produce labelled bracketed strings (or trees) with all grammatical categories and the relevant syntactic structure in a given language. It is into those strings that lexical items from the dictionary are inserted. At this stage of the derivation I stroked a dog would look as follows (the places where lexical items are inserted have been indicated: (22)

ZiRjiA. Pres-V PRß(Ra V tLNP,(S(RNP2((Def)-NP-(N3)) Past-V-ed stroke the boy (Indef) — NP — (N4))) # ))tLT) a dog

Semantic representations in terms of nuclear subconfigurations, corresponding to simple logical propositions and various indexed roles and their combinations, provide a universal framework whereby sentence semantics of all types of languages can be represented. It is essential to notice that all types of sentences, complex and compound, in various moods, aspects, and tenses, involve nuclear subconfigurations. Thus, the nuclear subconfiguration (Αα Δ P3), appearing in I stroked the dog is also involved in Did I stroke the dog? I might stroke the dog. I have been stroking the dog. I didn't stroke the dog. I said I would stroke the dog., etc. while the elements underlying these various forms appear elsewhere in the relevant strings, but outside the nuclear subconfiguration. For example, past, present and future tense are derived from tTL c , atTLB, and fTLE, respectively.7 Thus (23) (24) (25)

John stroked a dog. John strokes a dog. John will stroke a dog.

would, respectively, have the following tense specification: (26) (27) (28)

E(R 5 (A a Δ PRp(R a Δ tL 1 (R 2 (A 3 Δ P4) Δ tTL£) Δ tLy)) Δ atTL £ ) tL,(R 2 (A 3 Δ P4) Δ atTLE tL,(R 2 (A 3 Δ P4) Δ fTL e

Contrastive

Generative

Grammar

183

English sentences containing verbs in the so-called continuous aspect are derived from configurations containing a nuclear subconfiguration functioning as at L; for either AR or PR co-referential with A or Ρ within the nuclear subconfiguration. (It will be remembered that complex roles like AR or PR are derived from co-referential roles within the same (sub)configuration by virtue of a universal semantic rule conflating coreferential roles within the same configuration). The entire subconfiguration, in turn, functions as R in one of the three TL E 's to account for the distinction between present, past, and future continuous forms. Thus the sentences (29) (30) (31)

John is stroking a dog. John was stroking a dog. John will be stroking a dog.

are derived from (32) (33) (34) where

tl,(R 2 (AR 3 Δ atL 4 (A 3 Δ P 5 )) Δ atTL c ...) tl,(R 2 (AR 3 Δ atL 4 (A 3 Δ P 5 )) Δ tTL E ...) tl,(R 2 (AR 3 Δ atL 4 (A 3 Δ P 5 )) Δ fTL E ...) 1 2 3 4 5

— — — — —

John/is/was/will be stroking a dog. John be stroking a dog. John John stroke a dog. a dog.

When spelled out in ordinary English words (32) would look as follows: (35)

John be in the state of [John stroke the dog] at now.

(33) and (35) can be represented analogously. Sentences containing the so-called perfect aspect are derived from configurations containing a nuclear subconfiguration that functions as R in tTL e , and the subconfiguration formed in this way functions as R in atL co-referential with either A or Ρ within the nuclear subconfiguration. Therefore, such sentences as (36) (37) (38)

John has stroked a dog. John had stroked a dog. John will have stroked a dog.

184

Chapter

VIII

are derived, respectively, from: (39) (40) (41)

tl 1 (R 2 (R 3 (R 4 (A 5 Δ P 6 ) Δ tTL £ ) Δ atL 5 ) Δ atTL e ) tTL e ) fTL £ )

where

1— 2— 3— 4— 5— 6—

John hasjhadjwill have stroked a dog. John have stroked a dog. John stroked a dog. John stroke a dog. John. a dog.

Generally speaking, the meaning of sentences with verbs in the Perfect Aspect involves the indication of the current relevance of some past action, which is formally represented in the fact that the subconfiguration referring to such actions is placed not only in the tTL c but also in atTL E . Consequently, the meaning of such sentences as (36) can be explicated as: (42)

The fact that John stroked a dog is at John ( = relevant for John) at now.

(42) is the spelling out of the semantic representation (39) underlying the sentence (36). Declarative sentences and their sequences are derived from configurations beginning in: (43)

A a Δ PR p (R a Δ {fL/tL} (Χ)) Δ tLY

where the first Δ corresponds to the performative verb DECLARE, and the second Δ corresponds either to ASSENT TO or DISSENT FROM, depending on whether tL or fL is selected (for details and justification of these terms see Krzeszowski 1979: 48 ff.)· The version with ASSENT TO underlies all declarative positive sentences. The version with DISSENT FROM underlies sentences with "modal" negation as distinct from ordinary negation relating to one of the elements of the sentence. "Modal" negation can be signalled by means of the surface performative verb doubt, as in (44)

I doubt that John stroked a dog.

The superordinate sentence I doubt that can be paraphrased with the negative morpheme not or η 't overtly present, as in

Contrastive

(45)

Generative

Grammar

185

I don't think that John stroked a dog.

On the other hand, the performative complex (involving the two A's) can be deleted altogether, and the sentence can be realized as (46)

John didn't stroke a dog.

In all three cases, the resulting sentences are ambiguous in the same variety of ways as all other negative sentences. From the point of view of the performative analysis, however, it is more important to note that sentences such as (46), that is, all negative sentences without performative verbs present in the surface structure, are also ambiguous between the "modal" reading of the negation and the "pure" negation. The "modal" negation involves the act of denial on the part of the utterer of the sentence and is represented in the semantic structure by the second performative DISSENT FROM, with R a as its subject. This reading of negative sentences can be paraphrased with the use of the surface-negative performatives, such as doubt and don't think. The "pure" negation consists in negating an element within the nuclear subconfiguration (proposition). In this case, the second performative is ASSENT TO, with R 3 as its subject. Such readings of negative sentences cannot be paraphrased with the use of overt negative performatives. The distinction made here reflects the difference in the attitudes of the speaker to the contents of every sentence that he utters. The speaker either believes that what he says is true, or that it is false. Therefore, any negative sentence can either be an expression of doubt concerning the truth of the proposition contained in that sentence or a plain report concerning the non-occurrence of events, actions, or facts expressed by that sentence. Observe that the two readings of every negative sentence such as (46) differ with respect to presuppositions involved in each reading. Thus (46) can be paraphrased either as (47)

I think it is false that John stroked a dog. (Someone says he did.)

or as (48)

I think it is true that John didn 7 stroke a dog. (Someone says he did not, or, in fact, John did not stroke a dog.)

Sentences containing "pure" negation are derived from subconfigurations containing R(X) Δ fL(Y), where the role to be negated is specified as Ri(X) which is co-referential with an element within Y, e.g.,

186

(49)

Chapter

VIII

R, Δ fl2(R3(A, Δ P4) Δ tTL £ )

where X is null since R] is elementary, 1 — John, 2 — John stroked a dog, 3 — John stroke a dog, 4 — a dog. This configuration is involved in the derivation of (50)

John did not stroke a dog.

with a special emphasis on the word John. It is also possible to have (51)

R 4 Δ fL 2 (R 3 (A, Δ P4) Δ tTLJ

which underlies (52)

John did not stroke a dog.

Another input underlying 'pure' negation is represented by (53)

R,(R 2 (A 3 Δ P4) Δ tTL £ ) Δ fL,(R 2 (A 3 Δ P4) Δ tTL e )

where 1 — John stroked a dog, 2 — John stroke a dog, 3 — John, 4 — a dog. (53) underlies (54)

John did not stroke a dog.

uttered with "normal" stress and intonation. When spelled out in ordinary English words, (49), (51), and (53) can be represented as (55) (56) (57)

John not involved in ^(ai John stroked P4 a dog) John stroked P4 a dog) R4 a dog not involved in ^(ai , (John stroked a dog) not involved in ^John stroked a dog). R

R1

Note that the subconfigurations accounting for negated elements in sentences resemble the subconfigurations underlying sentences with / doubt that associated with the performative DISSENT FROM insofar as the contextual features characterizing the respective A's involve R's with the related fL's. This is a formal reflection of the meaning of negation, which always relates either to a disassociation of an element from the truth involving that element, as in (49), (51), and (53), or the disassociation of the speaker from a certain truth, as in the sentences of the type / doubt that discussed earlier on and derived from a version of (43). Positive questions and commands, as well as their sequences, are derived from configurations with COMMAND as the first performative.

Contrastive

Generative

Grammar

187

In the case of positive questions, the second performative is either SPECIFY or YES/NO, depending on whether a specific or a general question is involved. The following configuration underlies all specific (wh-)questions, that is, questions about a particular role: (58)

A a Δ PRp(AY Δ PRj Δ atLj(X)) Δ tLY

which, when spelled out in less abstract terms, gives

(59)

/ IMP (to) you (you SPECIFY (SOMETHINGi/SOMEBODY;) in (X)),

where X is a subconfiguration containing a role i, co-referential with SOMETHINGj/SOMEBODYj. General questions ask for the confirmation or the denial of a subconfiguration underlying a certain statement. The appropriate configuration is as follows:

(60)

Αα Δ PRp(AY Δ P ; (Rj Δ atL k (X))) Δ tLY

with the performative complex involving C O M M A N D and YES/NO. When spelled out in less abstract terms, (60) can be represented as

(61)

/ IMP (to) you SOMETHING,, (you YES/NO (SOMETHING; (SOMETHINGj/SOMEBODYj) be in (X))),

where SOMETHINGß is the direct object of the first performative and introduces the subconfiguration containing the content of the command; SOMETHING; is the direct object of the second performative and introduces the subconfiguration underlying the proposition to be questioned; SOMETHINGj/SOMEBODYj is an element of the questioned subconfiguration X (proposition), co-referential with an element within X and constituting the focus of the question, signalled on the surface by a special stress. Although a complete description of Contrastive Generative Grammar is available elsewhere (Krzeszowski 1979), we have taken the liberty of presenting the appropriate semantic inputs in considerable detail since in the chapter that follows we are going to explore a possible application of this model in the prediction of certain errors that foreign learners make in attempting to learn and use the target language. The explanation of the possible linguistic causes of these errors crucially depends on the proper understanding of how, according to Contrastive Generative Grammar, sentences are derived from underlying representations.

Chapter IX

Error analysis, interlanguage, and Contrastive Generative Grammar

Prediction of errors made by learners attempting to use a foreign language was one of the motivating factors in contrastive studies, especially in the United States. Lado hoped that contrastive studies would reveal similarities and differences between the native (source) language and the foreign (target) language, which would in turn make it possible to predict whether positive or negative transfer is likely to take place: If the expression, content, in the native and the new Actually no learning takes unit or pattern and merely

and association are functionally the same languages, there is maximum facilitation. place since the student already knows the transfers it (Lado 1964: 40).

In other situations, negative transfer or interference would presumably take place. Lado defined T R A N S F E R , either positive or negative, as the extension of a source-language habit into the target-language, with or without the awareness of the learner. He expected that similar places would lead to cases of positive transfer, which would facilitate learning, while different places would cause negative transfer or interference with the targetlanguage patterns, which would lead to errors having their source in the source language (Lado 1964: 222). Thus, analysing foreign language learners' errors would provide empirical verification of contrastive predictions. Empirical research quickly invalidated this simplistic view: many similar places across languages turned out to be troublesome and were erroneously handled by foreign learners, and, conversely, different items across languages would be assimilated in a more expeditious way. Briere (1968) demonstrated the relative independence of hierarchies of difficulties and of linguistic predictions. Politzer (1968: 35 — 43) produced examples of parallel structures in French and English which cause enormous learning difficulties. The inadequacy of contrastive methods was partly to

190

Chapter

IX

blame, but in many instances these experiments revealed the basic nonidentity of linguistic predictions based on the study of competence and errors accessible through the study of performance, which, as is well known, is affected by various other factors, escaping purely linguistic approaches. Let us quote as another example the instance of the Russian verb form ponjal, notoriously mispronounced by Poles as ponjäl, in spite of the fact that the equivalent Polish word 'pojgl has the stress on the first syllable, i.e., exactly where it falls in Russian. It was, moreover, realized that if an error is a deviation from some norm, it depends on the norm whether a particular linguistic form is recognized as an error. If a Pole says I goes, he makes an error. The same form in Black English may be considered perfectly acceptable (cf. Arabski 1979: 12). Since the relation between contrastive studies and traditional error analysis turned out to be more complex than Lado originally expected, the latter continued to flourish as a more direct and more reliable source of information about the learner's problems than contrastive studies, which only indirectly and often inaccurately predicted certain types of errors among all those that foreign-language learners make, namely only those which have their source in negative transfer from the source language. Traditional error analysis was thus an ad hoc attempt to deal with the practical needs of the classroom teacher. It was confined to impressionistic collections of "common" errors and their classification into various categories, such as phonetic errors, grammatical errors, stylistic errors, etc. In addition, traditional error analysis also attempted to discover frequencies of occurrence of various errors in texts. Conceived in this way, it was more immediately useful in determining the sequence of presentation of target-language items, in deciding on the relative degree of emphasis and practice, and in devising remedial exercises than was contrastive studies. More sophisticated traditional error analysis would attempt to analyse the source of errors (e. g., source-language interference, overgeneralization, etc. (see Arabski 1979: 31) and to evaluate the "seriousness" of the error in terms of communication disturbances or the distance from the norm (Olsson 1972; Enkvist 1973, 1977; James 1974, 1977). Three major reasons determined the practical usefulness of traditional error analysis vis-ä-vis contrastive studies: 1. Traditional error analysis did not suffer from the limitations of contrastive studies, which restricted the latter to errors caused by interlin-

Error analysis

191

gual transfer; it discovered many other types of errors, like those resulting from overgeneralizations, faulty teaching and learning strategies. 2. Traditional error analysis pointed out actual, attested difficulties, most directly connected with language usage while contrastive studies was concerned with the study of competence. 3. Traditional error analysis was not beset by formidable theoretical problems, which plagued contrastive studies, for example the problem of equivalence. This is why traditional error analysis, originally intended as a method of testing constrastive predictions and enhancing the power of the pedagogical applications of contrastive studies (cf. Banathy — Madarasz 1969: 77 — 92), became an autonomous field of study, more useful than contrastive studies in syllabus planning and in designing materials for foreignlanguage instruction. A new approach to error analysis emerged when the focus was shifted from teaching to learning, as a consequence of the realization that learning strategies do not necessarily correspond to teaching strategies. This observation was based on the fact that learners persist in making certain types of errors regardless of the amount of time and effort invested in the teaching of relevant linguistic forms, and that the order in which the learners master various linguistic forms does not always correspond to the order in which these forms are taught. Corder (1967 [1974]: 24) drew the important distinction between errors of performance (slips, lapses) and systematic errors of what he called "transitional competence", concentrating his attention on the latter. He noticed that systematic errors are merely a realization of the learner's transitional competence in the target language, and as such they are evidence of the learner's success rather than of his failure. The process of acquiring transitional competence is not fundamentally different from the process whereby the child acquires his native tongue in that both these processes involve the formation of hypotheses about language rules and the subsequent testing of these hypotheses. As an illustration Corder quotes the following dialogue: Mother: Did Billy have his egg cut up for him at breakfast? Child: Yes, I showeds him. Mother: You what? Child: I showed him. Mother: You showed him?

192

Chapter

IX

Child: I seed him. Mother: Ah, you saw him. Child: Yes, I saw him. (Corder 1974: 2 5 - 2 6 ) In this dialogue the child has tested three hypotheses: 1. Concord: subject verb in the past tense; 2. The meaning of the verb to show different from the meaning of to see; 3. Irregular past of to see. Having received negative reinforcement on all the three hypotheses, the child has formulated new hypotheses about the respective rules, which eventually yielded linguistic forms conforming to the norms of standard English. However, the original forms used by the child in the quoted dialogue were manifestations of the child's transitional competence in his native language. According to Corder, in learning a foreign language the learner has to test the following hypotheses: Are the systems of the new language the same or different from those of the language I know? And if different, what is their nature? (Corder 1974: 27). In testing such hypotheses, the learner goes through successive transitional competences, which Nemser (1971 [1974]: 55) called "approximative systems" According to Nemser "an approximative system is a deviant linguistic system actually employed by the learner attempting to utilize the target language" (Nemser 1974: 55). Nemser's description of approximative systems was based on three assumptions: 1. Learner speech at a given time is the patterned product of a linguistic system, L a (approximative system), distinct from Ls and Lt and internally structured. 2. L a 's at successive stages of learning form an evolving series, L a j n, the earliest occurring when a learner first attempts to use Lt (merger, the achievement of perfect proficiency, is rare in adult learners). 3. In a given contact situation, the L a 's of learners at the same stage of proficiency roughly coincide, with major variations ascribable to differences in learning experience (Nemser 1974: 56). Implicit in Nemser's presentation of L a 's is an additional assumption that at some stage of development, L a 's of a particular learner cease to

Error analysis

193

evolve and the system becomes stable at the most advanced stage available to a given learner. If this is true, such a stabilized approximative system can be studied and described by the same methods in which other natural languages are described. We shall elaborate on the significance of this conclusion later in this chapter. Selinker (1972 [1974]) also hypothesized "the existence of a separate linguistic system based on observable o u t p u t which results f r o m the learner's attempted production of a T L n o r m " (Selinker 1974: 35). Selinker called such a system "interlanguage" He observed that there are certain linguistic items, rules and subsystems, which foreign learners tend to keep in their interlanguage regardless of their age and the a m o u n t of time devoted to explanation and instruction in the target language. Selinker designed the term "fossilization" to refer to this phenomenon. It seems that Selinker's fossilized interlanguage corresponds to Nemser's stabilized approximative system. In attempting to describe interlanguage, Selinker identified five m a j o r (and a few minor) processes determining its formation. In our account of these m a j o r processes, we shall augment Selinker's original illustrations with our own examples. The following processes f o r m an interlanguage: 1. Overgeneralization of target language rules, which consists in extending the use of certain linguistic forms to contexts in which they are not appropriate, or in which they result in ungrammaticalities due to false analogy with other forms. F o r example: (1) (2)

What did he intended to say? (Selinker 1974: 38) I didn't took a big trip.

in which the past tense form is used where it could be logically expected by analogy with the forms in declarative sentences. Also (3)

After thinking a little I decided to start on the bicycle as slowly as I could as it was not possible to drive fast.

in which the speaker falsely extends the use of the verb to drive to all types of vehicles, including bicycles, which one normally rides. 2. Transfer of training, which has its source in the way in which drills and exercises are constructed and ordered. F o r example, Serbo-Croats (and also Poles) have considerable problems with the he-she distinction, and they tend to use he on almost every occasion, even when she is called

194

Chapter

IX

for. The source of this persistent error cannot be attributed to the transfer from the source language since in both Serbo-Croatian and in Polish, unlike in Finnish and Hungarian, the analogous sex-based distinction is codified in the grammar. Selinker attributes this error to the fact that "textbooks and teachers in this interlingual situation almost always present drills with he and never with she" (Selinker 1974: 39). My own examples involve (4)

John is a sitting

in which the article a erroneously appears in front of the present participle as a result of the prior overpractice of the pattern John is a boy, John is a pupil, etc. and (5)

Housework is the work which is doing at home

in which the overpractised present continuous effectively interferes with the less intensively practised passive forms. 3. Strategy of target-language learning. The general strategy is that of simplification (cf. Widdowson 1975: l l f f . ) . Foreign-language learners tend to reduce the target language to a simpler system and make errors typically involving omission of various grammatical formatives, articles, and sometimes function words. For example, (6) (7) (8)

Swimming is 0 very interesting and nice kind of sport. It is not to 0 done. This man kill-0 the most dangerous lion.

Another manifestation of this type of strategy is the coalescence of aspect, as in (9) (10)

Shakespeare have been born in 1564. I have read many plays by Shakespeare.

in which the simple and perfect aspects coalesce to produce a simpler system of tenses. 4. Strategies of communication involve yet another form of simplification. The errors in this domain usually have their source in the learner's conscious or unconscious conviction that one cannot waste too much time reflecting on how to say things even if one knows how to say them properly. This leads to ignoring many grammatical phenomena which do not affect communication. These include the plural number of nouns,

Error analysis

195

certain function words, and certain conjunctions. A spectacular example was uttered by a fluent speaker of English, under considerable strain, when he was trying to get a ride when hitch-hiking in Sweden: (11)

Do you going nach

Stockholm?

The strategy of communication influenced by the lack of time for reflection caused by the immediate necessity to get a lift, prevented the speaker from making use of the grammatical resources that he had already acquired. The example involves other processes, such as false analogy, transfer from German, etc. 5. Transfer from the source language, which is the familiar process of interference of the source language patterns. Examples are numerous and varied. Among them one finds (12)

I bought yesterday five electric bulbs.

in which the Polish word order has influenced the speaker's production in English. Also (13)

This things sometimes haven't in the shop.

where the Polish word order is accompanied by the faulty use of haven't, which must be attributed to the analogous use of the verb miec in Polish negative sentences. Using Selinker's ideas as a starting point for his own interesting speculations, Widdowson claimed that all of the processes which Selinker refers to are tactical variations of the same underlying simplification strategy and in general error analysis is a partial account of basic simplification procedures which lie at the heart of communicative competence (Widdowson 1975: 12). According to Widdowson, simplification is a result of an attempt to adjust the language behaviour in the interests of communicative effectiveness. Simplification may, therefore, affect any stage of interlanguage, and it may involve a movement away from the reference norm of the standard language (or even a particular interlanguage) so as to arrive at forms of speaking judged to be dialectally appropriate in certain contexts of use (Widdowson 1975: 15). Simplification presupposes the existence of non-simplified or more complex forms of language. "Simplified" is a property of language by no

196

Chapter

IX

means restricted to the foreign learner's interlanguages. It also characterizes such types as pidgins and Creoles (De Camp 1968; Valdman — Phillips 1975), "baby-talk", "lover's talk", and "foreigner talk" (Ferguson 1964, 1975) and various other types of "reduced registers", such as "telegraphese", technical descriptions, and the language of the language of instruction (Corder 1975: 5 - 7 ) . It is interesting to acknowledge some results of research into the structure of all these "simplified" codes. It appears that all of them share some structural similarities, such as an extremely simple, or non-existent, morphological system, a comparatively rigid word order, a reduced system of pronouns, a small number of grammatical function words, a reduced use of the copula, and the absence of articles (and to a smaller extent of deictic words). According to Valdman (1975: 23): "pidgins are derivationally shallower [italics supplied] than other natural languages and reflect more closely universal cognitively-based deep structures" In view of the overt similarities between all types of "simplified" codes, Valdmann —Phillips' contention can be extended to cover them all. We shall presently return to this observation in the context of Contrastive Generative Grammar. Corder (1975) expresses an unorthodox view on the role of simplification in language learning, taking Valdman — Phillips' reasoning to its logical conclusions. He suggests that it might be possible to regard "standard" codes as "elaborated" forms of "basic" simple codes such as pidgins, Creoles, interlanguages, and all types of "reduced" registers. His proposal assumes the existence of some universal process of elaboration or complication involved in all types of language learning. Though universal, the process would, of course, involve language-specific "complication rules" The development of a pidgin into a Creole would be a case of progressive complication. Another case would be the development of approximative systems into an interlanguage closely approaching the target language. All types of "reduced" registers used in appropriate situations would be instances of fossilized intermediate approximative systems or institutionalized, stereotyped stages in the process of complication towards the standard version of the language. Contrastive Generative Grammar as a device generating equivalent constructions seems to be well suited for providing a fairly explicit account of "reduced" codes, including interlanguages, vis-ä-vis fully elaborated codes. It can provide a linguistic account of how a simplified code can be related to a more complicated code, by spelling out formal complication routes by which the latter can be derived from the former. 1 It will

Error analysis

197

be remembered that Contrastive Generative Grammar involves two kinds of lexicalizations (lexical insertions). Minor lexicalizations insert function words and grammatical formatives, and they constitute part of syntactic transformations. Major lexicalizations take place after syntactic transformations and insert content words. In fully elaborated codes, all major lexicalizations take place after all the relevant syntactic transformations. In various types of "simplified" codes, notably also in interlanguages, major lexicalizations may occur at earlier stages of derivation. In extreme cases major lexicalizations may occur at the level of semantic representations. In such cases, "cosmetic" transformations, accounting for agreement, concord, government, and generally for inflections, are either completely left out, or they apply to a limited extent. Major lexicalizations, as it were, "petrify" the syntactic structure of constructions since, at least in Contrastive Generative Grammar, once major lexical items are inserted, no syntactic transformation can apply. Therefore, the degree of syntactic complication (elaboration) of a construction in an interlanguage directly depends on the stage of derivation at which major lexicalizations take place. Early lexicalizations result in the syntactically simplest constructions, characterized by the absence of function words (inserted by syntactic transformations which in such cases cannot apply) and a rigid word order, often different from the way in which words are arranged in fully elaborated codes. Within the framework of Contrastive Generative Grammar, complication can be defined as the gradual shift of the place at which major lexicalizations occur, from the deepest level of representation, i.e., the semantic level, to the level which constitutes input to "cosmetic" transformations, and where major lexicalizations occur in fully elaborated codes. Among the first sentences in child speech, Brown (1970: 220) quotes Adam put, Eva read, Mummy lunch, dog bite, hit ball, etc., which represent various relations such as agent —action, action —object, agent —object, etc. Such examples illustrate extreme cases of early lexicalizations occurring in the simplest codes, whereby lexical items are inserted directly into semantic representations, or more specifically into nuclear subconfigurations, thus causing "petrification" of their structure. In these cases, syntactic transformations do not apply, so that all other distinctions represented by the material outside the nuclear subconfiguration is left out of the derivation (cf. Chapter VIII). In more elaborated codes, more non-nuclear material is involved in the derivation of sentences. As an example, let us consider w/z-questions in English. It will be recalled that in the framework of Contrastive

198

Chapter

IX

Generative Grammar all w/j-questions are derived from configurations marked in Chapter VIII as (58) and (59), which we now repeat for convenience as (14)

A a Δ PRß(AY Δ PR^ Δ atLj(X)) Δ tLY

and (15)

I IMP (to) you {you SPECIFY (SOMETHING./SOMEBODYj) in (X)).

A full specification of the semantic representation for a sentence like (16)

What did John touch?

can be formulated as (17)

Αα Δ PRß(AY Δ PR; Δ atL 2 (R 3 (A 4 Δ Ρ,) Δ tTL e ) Δ tLY) Δ atTL e

where

1 — the object touched by John about which the question is asked (SOMETHING^SOMEBODYO 2 - JOHN touched SOMETHING/SOMEBODY 3 - John touch SOMETHING/SOMEBODY (nuclear subconfiguration) 4 — John

In ordinary English words (17) can be spelled out as (18)

AaI

command tLyyou atTu that A7you specify in atL2 the proposition involving PRisomething/somebody Ri[John touch something/somebody] at tTLi:some time before now.

In order to convert (17) into the fully elaborated question, several operations must be performed by the rules at various levels. 1. At the categorical level, appropriate categorical rules will assign grammatical categories to various portions of (17). By virtue of these rules, the categories Σ and S (in the case, the Σ-sequence consists of just one S-sentence) will be assigned to R's placed in TL's, various roles will be assigned the categorical status of NP's; some Δ'β will be categorized as verbs while other Δ'β will be erased; tTLE's and atTL c 's will be, respectively, assigned the status of past and present grammatical formatives; and the interrogative morpheme Q will be adjoined to the left of PR, (for details see Krzeszowski 1979: 62 f f ) . As a result of these operations, (17) becomes (19):

Error analysis

(19)

199

I(R 5 (A a V PRß(AY Δ V Q-PR, V atLNP 2 (S(RNP 3 (ANP 4 V PNP,)past)#)) tL Y )present)

2. At the level of syntactic transformations, the following operations will be involved: (a) the deletion of the PRj co-referential with a role in the nuclear subconfiguration. In our example, PR, and P) are involved, so the former is deleted together with its verb. Simultaneously, Q is shifted to be adjoined to that role in the nuclear subconfiguration which is co-referential with the deleted PR. In our example Q is shifted to P, in the nuclear subconfiguration. Thus (19) becomes (20): (20)

E(R5(Aa

V

PR P(AY

V

atLNP2(S(RNP3(ANP4

V

Q-

PNP,)past))#)) tLj)present) (b) adjoining the tense morphemes of each Σ- and S- subconfiguration to the verb within that (sub)configuration. Thus (20) becomes (21): (21)

E(R 8 (A a present-V PRß(AY V atLNP 2 (S(RNP 3 (ANP 4 past-V QP N P , ) ) # ) ) tL,))

(c) preposing the element with Q to the initial position within the nuclear subconfiguration. Thus (21) becomes (22): (22)

E(R 5 (A 2 present-V PR ß (A T V atLNP 2 (S(RNP 3 (Q-PNP, past-V ANP 4 ))#)) tLY))

3. At the level of "cosmetic" transformations (after the insertion of lexical items) introducing do to support the tense in those cases when the V is not an auxiliary verb be or have. In this way (22) becomes (23): (23)

I(R 8 (A a present-V PR ß (A a V atLNP 2 (S(RNP 3 (Q-PNP, past-do V ANP 4 ))#)> tLY)).2

The successive application of all these operations (categorical, syntactic, and "cosmetic") mentioned in (1), (2), and (3) yields an increasingly complex or "thick" structure as more and more rules are involved in the derivation of the question. However, at each of these steps the derivation can be terminated by the insertion of lexical items. In such cases, which might be metaphorically called "premature lexicalizations", 3 we obtain structures which are derivationally shallower as they involve fewer rules than the derivationally fuller structures, and these shallower structures resemble underlying semantic representations more closely. Thus, we can

200

Chapter

IX

hypothesize that due to such "premature" lexicalizations the following questions will be generated: (24)

(a) John touch?

(b) John touch what?

(c) John touched what? (d) What John touched? (e) What touched John?

(lexical insertions at the semantic level or at the categorical level before the rule adjoining Q to the node co-referential with the deleted PR) (lexical insertions before the rule adjoining tense to the verb and the rule preposing the questioned element) (lexical insertions before the rule preposing the questioned element) (lexical insertions before the rule preposing tense) (lexical insertions before the rule introducing do)

Since the rules operating on each level are intrinsically ordered, the rule adjoining tense may also operate after the rule preposing the questioned element. Therefore, one may also expect sentences such as (f) What John touch?

(lexical insertions after the rule preposing the questioned element but before the rule adjoining tense to the verb)

Thus, Contrastive Generative Grammar generates questions of various degrees of derivational complexity. Some of these questions correspond to questions actually attested in various "reduced" codes, including child's speech and interlanguages. In the materials studied by Brown (1970) and Ravem (1974), one finds types (a), (d), (e), and (f). Ozga (personal communication) finds types (d) and (e) to be frequent among Polish learners of English. It is perhaps slightly puzzling to realize that none of the researchers found intermediate hypothetical questions representing patterns (b) and (c), that is, those in which lexicalizations take place immediately before the rule preposing the questioned element, except in the case of that Brown calls "occasional" questions. 4 Although Contrastive Generative Grammar generates such structures as (b) and (c) as potential sentences, it cannot explain why they do not actually occur in the examined corpus. However, since the semantic material triggering off the rule which introduces Q into derivations actually precedes in the linear order from left to right the relevant nuclear subconfiguration in

Error analysis

201

the semantic representation, and since after being adjoined to the questioned element in the nuclear subconfiguration, it is subsequently preposed again, the latter two steps could be ignored altogether, while Q would simply attract the questioned element from the nuclear subconfiguration. In view of the evidence presented by Brown, Ravem, and Ozga, the appropriate rules could be reformulated in the following way: instead of two rules — one that adjoins Q to the nuclear role, coreferential with the deleted PR; and one that preposes that role to the initial position within the nuclear subconfiguration one rule could be postulated. It would perform three operations at once: it would delete the PR; to which Q is originally adjoined, it would move Q to the initial position within the nuclear subconfiguration, and it would prepose that role within the nuclear subconfiguration which is coreferential with the deleted PR ; to adjoin it to Q. In this way, the stage at which Q is placed in the nuclear subconfiguration in a position other than the initial one would be eliminated, and the grammar would not generate (b) and (c) as intermediate constructions between (a) and (d). Although this solution lacks internal motivation, it is at least plausible in the framework of Contrastive Generative Grammar, and it is motivated externally by the empirical evidence furnished by Brown, Ravem, and Ozga. This example reveals one of the weaknesses of Contrastive Generative Grammar. It can accomodate many facts in the domain of performance in an interlanguage, but it cannot predict those facts in an internally motivated way. In other words, Contrastive Generative Grammar in its present form, is too powerful. It is capable of generating a number of constructions which are not actually found in performance. However, it seems that many generative grammars suffer from the same shortcoming. In terms of error analysis, the above discussion concerns various errors which result from "premature" lexicalizations, and which consist in inserting lexical items into syntactic structures which have not been fully elaborated (in comparison with fully elaborated codes). Such errors can be defined as those forms in an interlanguage which are different from the norm formulated in terms of generative rules characterizing the fully elaborated version of target language, and which result from the omission of various syntactic rules which "complicate" the structure. Contrastive Generative Grammar accounts for the fact that these "errors" are similar across various "reduced" codes and across interlanguages, regardless of the source languages: lexical items are inserted at early stages of sentence derivation into underlying structures which are similar in proportion to the distance from the surface at which such lexicalizations take place.

202

Chapter

IX

Thus, Contrastive Generative Grammar can at least predict that errors resulting from such "premature" lexicalizations, involving the absence of articles, function words, and other formatives derived from the material outside the nuclear subconfiguration, will be similar in various interlanguages. Another type of deviation from target-language norms that may be handled by Contrastive Generative Grammar consists in the instances of inserting lexical items of the target language into structures generated according to the rules of the source language. For example, the grammar of the Polish interlanguage of English may generate such sentences as (25)

I wanted that Peter opened the door

in which English lexical items have been inserted into the structure generated by the Polish syntactic rules which generate such sentences as (26)

Chcialem, zeby Piotr otworzyl drzwi.

Errors of this sort are traditionally described in terms of transfer of patterns from the source language and are to a large extent languagespecific. They result from substituting complication routes of the source language for complication routes of the target language and can be predicted by Contrastive Generative Grammar on the basis of a comparison of particular equivalents it enumerates. As a model of bilingual competence, Contrastive Generative Grammar cannot make any claims about actual occurrence of these various forms in the performance of particular interlanguage users, nor can it pretend to be a model of psycholinguistic processes which attend on the actual production of various deviant utterances in particular interlanguages (cf. Chesterman 1980: 2 1 - 2 3 ; James 1980: 54). It can, however, predict potential errors and suggest falsifiable hypotheses about the occurrence or non-occurrence of particular forms in particular interlanguages (cf. Ravem 1974:128 ff.). The formulation and verification of such hypotheses falls within the domain of error analysis proper.

Chapter Χ

Quantitative contrastive studies1

Quantitative contrastive studies belong to text-bound contrastive studies. As was pointed out in Chapter II, quantitative contrastive studies may, but do not have to, use translations as their primary data. It is possible to conduct quantitative contrastive studies on texts which are not translations, and which are not systematically equivalent nor even equivalent in any sense of the word "equivalent" (cf. Chapter II). The constituent texts of 2texts used for quantitative contrastive studies may be chosen for comparison only on the grounds that they represent the same register, or the same style, or the same literary genre, or on any other grounds which provide the common platform of reference motivating the comparison. Using such texts, one can look for statistical equivalence, represented by certain fixed patterns of frequencies of occurrence of various linguistic forms characterizing a particular register, style, genre, etc. These patterns of frequencies characterize specific "styles" in particular languages by following a certain statistical norm. Deviations in plus or in minus from such a norm may be connected either with non-native performance, as we shall see later, or they may be intended, for example, in parodies. 2 Systematic contrastive studies are incomplete and inadequate unless supported by quantitative data at all levels of linguistic analysis. 3 As an example let us consider the case of personal pronouns across languages. In Chapter IV, systematic equivalence was established between English and Polish paradigms of personal pronouns: / — ja, you ty/wy, they — onijone, etc. Correct, as far as it goes, this generalization is inadequate for several reasons. For example, it does not say anything about the appropriateness of specific pronouns in specific situations. Above all, it does not say anything about the relatively low frequency of occurrence of personal pronouns in Polish texts in contrast to a relatively high frequency of occurrence of personal pronouns in English texts. Systematic contrastive generalizations prove to be even less adequate in all those numerous cases in which a particular construction in one language has more than one nearly synonymous equivalent in another language.

204

Chapter Χ

Some of these equivalents may turn out to be nearly congruent (cf. Chapter VI). In such cases, systematic contrastive studies would predict no learning problems but rather something in the nature of positive transfer. In reality, however, different frequencies of occurrence of such congruent forms in texts could result in errors consisting in using congruent forms either too frequently or too rarely. In the former case, we would be dealing with errors of abundance, in the latter case, with errors of avoidance. In either case the frequency of use by non-native users of the language would be different from the frequency of use by native users. Let us consider the following example. The French construction apres etre revenu can be rendered in Czech (Becka 1978: 129) and in Polish in a number of ways, none of which is a congruent equivalent of the French original: Czech Polish French Active Past Participle: Czech Polish French Subordinate Clause: Czech Polish French Main Clause: Czech Polish French Prepositional Phrase:

po navratu po powrocie apres le retour vrativ se wrociwszy (si§) etant revenu kdyz se vrätil kiedy wrocil quand il etait revenu vrätil se a... wrocil i il etait revenu et

'after the return'

'having returned'

'when he returned'

'he returned and

Becka discusses this example in the context of quantitative contrastive stylistics and observes that "In the translation process a grammatical phenomenon has become a stylistic one" (Becka 1978:129). As a stylistic issue, it is primarily a matter of the frequency of use. All these constructions in the three languages are nearly synonymous and can be arranged in pairs which exhibit a certain degree of formal resemblance. Systematic contrastive studies would assign to those pairs the status of equivalence and would discover formal differences resulting from the application of non-identical rules involved in the derivation of these constructions. Some valid generalizations could thus be made about systematic differences and similarities of specific constructions across languages. Such systematic analyses would provide no clues to the ways in which native speakers employ these equivalent constructions in actual texts. The fact that there is an equivalent or even a congruent construction in a target language does not necessarily

Quantitative

contrastive

studies

205

mean, and in fact rarely does, that it is employed with equal willingness by native users of the language. In reality, frequencies of occurrence of equivalent constructions are often quite different. For example, a native user of French would seldom use apres le retour in the contexts in which Czech would have po navratu or Polish would have popowrocie. Although in terms of systematic contrastive studies such pairs would constitute semanto-syntactic equivalents, in statistical terms they would probably have to be considered as non-equivalent. In her contrastive analysis of simple sentences in Czech and English, Duskovä notes that One of the features in which Czech and English differ is the tendency of English to what has been called complex condensation, viz. the use of non-finite verb-forms where Czech has subordinate clauses (Duskovä 1978: 83). The differences, according to Duskovä, may be due to systematic differences in those instances in which English possesses certain grammatical forms that Czech lacks (e. g., the gerund), or they may be due to more restricted uses of such Czech forms as infinitives, participles, as well as perceptive and causative verbs. In quantitative terms, "more restricted uses" may suggest lower frequencies of occurrence in texts although this is not always the case. For instance, the fact that in Polish infinitives do not appear as object complements in the construction called accusativus cum infinitivo nor in their passive counterparts called nominativus cum infinitivo means that infinitives in Polish have "more restricted uses" or, strictly speaking, that they have a more limited distribution. But it does not mean that infinitives are less frequent in Polish texts than in English texts. As a matter of fact, according to my own investigations, about every 65th word in a continuous Polish text is an infinitive, while only about every 90th word in a continuous English text is a marked infinitive (with to). Therefore, Duskova's suggestion that the statistical tendency in English to complex condensation may be due to systematic differences must be regarded with caution. Whether or not there is a direct correlation between the number of various infinitival constructions and the frequency of occurrence of infinitives in continuous texts is a matter that remains to be investigated. 4 Quantitative contrastive studies provide means of approaching the poorly investigated phenomena connected with errors of avoidance and errors of abundance, which cannot be handled by systematic contrastive studies since the very concepts of avoidance and abundance are quantitative

206

Chapter Χ

rather than qualitative in nature. Interestingly enough, it has been observed that differences in the frequency of use of certain constructions are connected with native vs. non-native use of a language. For example, Graczyk (n. d.) noted the tendency of Polish learners of English to overuse extraposed subjects as in It is a fact that it is hard to convince people that there is much sport in boxing. Although grammatical, this sentence suffers from clumsiness, which could be remedied by writing Boxing may jean be a good sport though few people know/admit it. According to Graczyk, one finds fewer extraposed subjects in compositions written by native speakers of English. This sort of overuse of a favourite construction could be labelled "error of abundance" One can easily envisage the opposite tendency to avoid certain types of constructions. If a foreign-language learner manifests this tendency in his writing and/or speaking, he is committing an "error of avoidance" One such error, involving a comparatively low frequency of a certain type of nominals in texts written by Polish learners of English is discussed below. In either case, the learner deviates in plus or in minus from a certain statistical norm which characterizes native performance in a particular language. To ascertain such an error, one has to perform a quantitative contrastive study of texts written by native users of a particular language and by a non-native user of the same language and compare the frequencies of use of the investigated forms. In other words, one has to conduct a quantitative contrastive study of texts written in the target language (by native users) and texts written in an interlanguage (by foreign-language learners). Whatever differences in frequencies of occurrence of various linguistic forms in native and non-native productions will be thus discovered, are of great interest in themselves, as they can prompt teaching strategies that could lead to the elimination of both types of frequency errors. However, it is at this point that one begins to wonder what possible impact native frequency patterns might have on foreign-language performance. It would seem plausible to assume that the performance in a foreign-language is determined by the frequency patterns of distribution of systematically equivalent forms in source-language texts. One could hypothesize that this is indeed the case and then proceed to verify this hypothesis in the course of statistical research. Let us take a look at one such study, whose results were at first surprising in that they apparently contradicted the contrastive predictions but in the end provided a reasonable explanation. In the course of the research, I investigated frequencies of occurrence of Gerundive and Action Nominals in M.A. theses on linguistics written by two Polish students (both nonnative, average students of the English Department at the University of

Quantitative

contrastive

studies

207

Gdansk) and in a Diploma thesis written by a native user of English. For convenience, I shall refer to the two theses written by the Polish students as PEl and PE2, respectively, and to the thesis written by the British student as BE. For the purpose of this particular research, no distinction was made between Gerundive Nominals and Action Nominals ending in -ing as described by Lees (1963), so the two kinds of nominals were described jointly. However, Action Nominals ending in suffixes other than -ing, as well as full nouns ending in -ing (Regular Gerunds in Lees' misleading terminology), were excluded. The frequency of occurrence of gerundive and action nominals were calculated in relation to the number of sentences constituting the three theses (citations and illustrative examples had been omitted from the count). Interestingly enough, the average length of sentences was approximately the same in each of the three theses and amounted to 23.9 words per sentence in BE, 24.3 words per sentence in P E l , and 24.6 words per sentence in PE2. Therefore, the relatively high number of gerundive and action nominals in the thesis written by the native speaker (BE) can by no means be attributed to the hypothetically greater length of sentences, which, perhaps, might have been expected in a thesis written by a native user of English. On the contrary, it turns out that the sentences in BE tended to be slightly shorter though the difference had no statistical significance. The frequency of occurrence of gerundive and action nominals in BE was significantly higher than in PEl and PE2. Whereas in BE the number of such nominals averaged 0.26 per sentence, in PEl it amounted to only 0.09 and in PE2 to even less, viz., 0.07. Characteristically, PE2 was graded as the lowest passing ("satisfactory") while PEl was evaluated one grade above the lowest passing ("good"). Table 2 contains the relevant figures. Table 2. Comparison of English written by a native speaker (BE) and by two non-native speakers ( P E l , PE2)

Number of sentences Number of gerundive and action nominals Gerundive and action nominals per sentence Words per sentence

BE

PEl

PE2

300 78 0.26 23.9

231 21 0.09 24.3

341 25 0.07 24.6

In order to see to what extent the frequency of occurrence of equivalent Polish nominals in Polish continuous texts might have negatively influenced the performance of the two Polish students in English (both

208

Chapter Χ

committed evident errors of avoidance), a quantitative contrastive study of translationally equivalent (cf. Chapter II) texts in English and Polish was undertaken. As in English, one encounters problems with the classification of nominals in Polish. The so-called nomina actionis in Polish basically fall into two groups: substantiva verbalia and substantiva deverbalia. Subs t a n t i a verbalia differ from substantiva deverbalia in that the former are morphologically regular as they take regular suffixes -anie, -enie, -cie and can be formed from nearly all verbs (with very few exceptions, which can be easily listed). The deverbalia are formed from a fairly limited number of verbs and abound in alternative forms which do not obey any obvious rules. In comparison with English, substantiva verbalia are closer to Lees' Gerundive Nominals as they are more "verby" in nature while the deverbalia are closer to Lees' Action Nominals and Regular Gerunds by being more "nouny" 5 Since the Polish nomina actionis fall into two groups while corresponding English nominals fall into three groups, from the point of view of quantitative contrastive studies, there arises the problem of what to count and what to compare. The situation can be visualized thus: Polish nominals: Substantiva verbalia (verbal substantives) — regular pisac mowic

pisanie mowienie

Substantiva deverbalia (deverbal substantives) — irregular pisac mowic

pisarstwo mowa

Lees' English nominals: Gerundive nominals — regular write speak

writing speaking

Action nominals — regular write discuss

the writing of irregular discussion

Regular gerunds — regular fill

— one filling, two fillings

Quantitative

contrastive

studies

209

It appears that gerundives are the least "nouny" while full gerunds are the most "nouny" (cf. Ross 1973). In Polish, verbal substantives are less "nouny" than deverbal substantives. The comparative problem consists in the fact that, on the one hand, Polish verbal substantives seem to combine some characteristic features of both English gerundive nominals (teaching English) and action nominals (the teaching of English) while on the other hand, Polish deverbal substantives combine some features of English action nominals and of regular gerunds. The situation is further complicated by the fact that within the Polish verbal substantives (in -anie, -enie, -cie) it is possible to distinguish between at least two subcategories, of which one is more "nouny" than the other. Thus, alongside such verbal nouns as czytanie 'reading' and mowienie 'speaking', we also have such regular nouns as wprowadzenie 'introduction', uzycie 'currency; use', which resemble English irregular action nominals, but unlike the latter they retain regular suffixes. The most "nouny" Polish nominals, i.e. full nouns called deverbal substantives, have irregular and unpredictable suffixes of various sorts, which accompany morphophonemic and semantic shifts. In that latter respect, they resemble English irregular action nominals rather than full gerunds, which regularly take the suffix -ing and form regular plurals with -s. If we assign numerical indices to the three kinds of nominals in Polish and English, indicating the rank on the scale of "nouniness", we shall obtain the following situation: Polish: 1. Deverbal substantives (irregular) 2. Verbal substantives, (regular) 3. Verbal substantives 2 (regular) English: 1. Regular gerunds 2. Action nominals (regular, irregular) 3. Gerundive nominals (regular) Thus, at the level two of "nouniness" we have a regular morphological situation in Polish and an irregular situation in English. At the level one of "nouniness" we have an irregular morphological situation in Polish and a regular situation in English. Unfortunately, one cannot rule out the possibility that the two taxonomies are based on erroneous criteria. For example, Lees' distinction between gerundive and action nominals rests on a mixture of morphological, syntactic, and semantic properties.

210

Chapter Χ

As a result, action nominals comprise both regular -ing of formations and irregular ones. For the purposes of the present study, it was necessary to unify the criteria by restricting them to morphological ones alone. This meant that in both Polish and English only regular formatives were considered. These included all Polish forms in -anie, -enie, -cie and all English gerundive nominals as well as those action nominals which take the regular suffix -ing. Lees' Regular Gerunds (full nouns) were excluded from the count, following another morphological criterion, viz. their ability to form plural by taking the plural inflection -s. This property places Less' Regular Gerunds outside all verbal forms, on the very top of the scale of "nouniness" On the other hand, no such elimination was possible in Polish since only semantic criteria allow one to distinguish the more "nouny" from the less "nouny" of the Polish verbal substantives. In view of the results of the quantitative contrastive study of PE1, PE2, and BE, one would have expected similar results from the quantitative contrastive study of a Polish text written by a native user of Polish and a text written by a native user of English. In this way the scanty use of gerundive and action nominals in the Polish-English interlanguage (PE1 and PE2) would be accounted for by the negative transfer of the presumably lower frequencies of occurrence of nominals in Polish texts. However, it turned out that the hypothesis about the negative transfer from Polish to English, in the case of the frequencies of nominals, is not easily verified. Two texts were compared with regard to the frequency of occurrence of English gerundive and action nominals ending in -ing and of Polish substantiva verbalia. The first 300 sentences of John Lyons' Introduction to theoretical linguistics were examined and compared with a very "Polishsounding" version of the same book translated from the original by Krzysztof Bogacki under the title Wst§p do j§zykoznawstwa. In the original version, 60 gerundive and action nominals were found, which means that the average occurrence of such nominals was about 0.2 per sentence and corresponded to the results obtained from the previous study. In the Polish text, as many as 208 regular nominals were found, which gives the average occurrence as high as 0.7 per sentence, which is significantly higher than in English. A closer quantitative look at the data helps to understand this unexpected result. It turns out that only about 80 of the total 208 words in -enie, -anie, -cie constituted the less "nouny" words of rank 1 on the scale of nouniness. These averaged to 0.26 per sentence, that is, almost as many as in the case of English

Quantitative

contrastive

studies

211

nominals. Interestingly enough, among them there were many words which corresponded to English full nouns (Lees' Regular Gerunds) rather than to gerundive nominals or to action nominals. They were such words as wyzwolenie sig 'emancipation', zepsucie 'corruption', zalozenie 'establishment', etc. As a matter of fact only as few as 24 of the Polish substantiva verbalia corresponded to English nominals ending in -ing, which constitutes only about 10 percent of all occurrences of substantiva verbalia in Polish. This situation offers a hint suggesting factors which influenced the Polish-English interlanguage in which PE1 and PE2 were written. It appears that in the preponderant majority of cases, Polish substantiva verbalia correspond to English full nouns (Regular Gerunds) or to irregular action nominals as their lexical equivalents. Therefore, the Polish learners of English associate Polish substantiva verbalia with English full nouns rather than with -ing forms of gerundive and action nominals as a matter of a general principle, following such equivalents as dodanie 'addition', stosowanie 'use', odrzucanie 'refusal', porownanie 'comparison', spostrzezenie 'insight', uzycie 'currency', omawianie 'discussion', etc. Once such an association is formed, it is easily overgeneralized to other instances. This overgeneralization is manifested in the learner's attempts to find a lexical equivalent for every Polish substantivum verbale rather than follow a possible transfer rule from a grammatical rule deriving substantiva verbalia in Polish to a grammatical rule deriving gerundive and action nominals in English. Presumably, no such transfer rule has been formed in the case of the authors of PE1 and PE2. The absence of such a transfer rule of grammar and the overgeneralization of the lexical transfer account for the non-native-like performance in the case of PE1 and PE, manifested as errors of avoidance affecting the frequencies of occurrence of gerundive and action nominals. The quantitative contrastive studies described above leads to the following general conclusion concerning quantitative contrastive studies: 1. Quantitative contrastive studies are strictly dependent on systematic contrastive studies in that the latter provide criteria for identifying linguistic forms to be counted in actual texts. For instance, the decision about the system equivalence of the Polish forms in -enie, -anie, -cie and the English forms in -ing is only possible to make when systematic analyses of Polish and English noun-forming suffixes are taken into consideration. The results of such qualitative analyses provide structural categories, which are then counted in the course of quantitative contrastive studies.

212

Chapter Χ

2. The quantitative contrastive studies presented above suggests that the hypothesis about the native vs. non-native performance, as manifested by different frequencies of occurrence of certain forms is valid. In this case, quantitative contrastive studies consists in the quantitative comparison of texts produced by native users of a target language and nonnative users of that target language; the latter presumably produce texts in an interlanguage. The deviations from the native norms are either in minus (errors of avoidance), as in the case of English nominals used by Polish students, or in plus (errors of abundance), as in the case of extraposed subjects, so frequent in the Polish English interlanguage texts. Such quantitative contrastive studies might be called intralinguistic inasmuch as they involve a target language and an interlanguage associated with that target language. If texts produced by native users in two different languages are compared, we deal with interlinguistic quantitative contrastive studies. It would seem that its results in this area cannot explain certain transfer phenomena if these results are considered in isolation form other results of both a qualitative and quantitative nature. Transfer appears to be determined by a complexity of factors and cannot be accounted for by one particular kind of contrastive studies. 3. There seems to exist a need for massive statistical research involving both intralinguistic and interlinguistic quantitative contrastive studies, which could eventually lead to a better understanding of the complex phenomena of transfer, as well as those phenomena which characterize native and non-native performance in a foreign language. The practical validity of this kind of research is easy to appreciate. One result might be a quantitative description of "individual style" at the plane of interlanguage. One can expect that the Polish interlanguage of English is quantitatively different from the Hungarian interlanguage of English, etc. Each national interlanguage of English is probably characterized by a unique matrix of frequencies of occurrence of various linguistic forms. The empirical verification of such hypotheses, among other concerns, will keep contrastive studies alive for some time to come.

Chapter XI

Cognitive linguistics and contrastive studies

Cognitive linguistics is a field in modern linguistics which stands in opposition to the transformational generative tradition. 1 Rather early in its history, that is at the beginning of the 1960s, the originally uniform theory created by Noam Chomsky began to branch off as a result of the criticism expounded in the works of such linguists as Lakoff, Ross, McCawley, Fillmore, and many others (cf. Newmeyer 1980: 133 ff.). While Chomsky subjected his own standard theory to endless revisions and extensions, the dissidents broadened the scope of their linguistic investigations, and, having gone through successive stages of generative semantics (Lakoff 1971), global rules (Lakoff 1970) and fuzzy grammar (Lakoff 1973), they eventually reduced all generative grammars to the status of an interesting but marginal enterprise within linguistics. Under the inspiration of various cognitive sciences, notably psychology, anthropology, ethnography, and philosophy, cognitive linguistics has come into being in the works of Lakoff (1982, 1986), Langacker (1986), Lindner (1981, 1982), Brugman (1981), Fillmore (1984a, b) and many others. Cognitive linguistics is based on the assumption that human language cannot be viewed as an abstract system isolated from socio-psychological settings in which it is used, but, on the contrary, it reflects the way people experience the world: A wide variety of experimental factors perception, reasoning, the nature of the body, the emotions, memory, social structure, sensorimotor [sic!] and cognitive development, etc. determine in large measure, if not totally, universal structural characteristics of language (Lakoff 1977: 237). Moreover, cognitive linguistics is based on radically different ontological foundations in comparison with those on which structuralism and its continuations are based. Cognitive linguistics rejects the classical, Aristotelian view of the world as consisting of discrete categories; it thus

214

Chapter XI

rejects the most essential axioms of the classical theory of knowledge on which traditional linguistics was based. Specifically, cognitive linguistics rejects the view that the world can be described in terms of categories defined by means of necessary and sufficient conditions which must be fulfilled by all the members of a given category. The classical theory does not admit intermediate cases: everything either is or is not a member of a given category, depending on whether it fulfills the required conditions. Classical categorization admits complex categories as logical combinations of simpler categories. Thus complex categories can be always reduced to simple ones. This sort of reductionism was applied in the description of all aspects of language and manifested itself in reducing complex signs to simple signs (morphemes) and even to a finite set of submorphemic elements (phonemes). Reductionism, inherent in classical categorization, was connected with set theory, in terms of which all levels of language were analysed. For example, in phonology, matrices of distinctive features are sets of conditions whereby category membership is defined while various kinds of brackets represent various logical operations on those sets. Likewise, any generative grammar defines sets of sentences, and every sentence is an ordered set of phonological features, etc. The classical theory of categorization is firmly rooted in the tradition of human thinking as it satisfies man's needs for orderliness. Questioning it might be considered as an attempt to question the very existential foundations of many scientific disciplines, mainly those concerned with human knowledge, that is, cognitive sciences: philosophy, psychology, anthropology, linguistics, and artificial intelligence. Yet, in all those disciplines the classical theory of categorization has been challenged (cf. Lakoff 1982: 14). With regard to philosophy and linguistics, Wittgenstein (1953 [1972]: 50 ff.) pointed out that classical categorization cannot adequately cope with various concepts, such as "game", inasmuch as there exists no set of necessary and sufficient conditions characterizing this concept, which must embrace various heterogeneous activities like chess, bridge, football, solitaire, or hide-and-seek. Particular "games" mutually resemble one another in varying degrees, like members of one family, but one cannot identify a feature, leave alone a set of features, which would be common to all games. It is only the "family resemblance" that can serve as the basis on which one can classify all games as belonging to one natural category.

Cognitive

linguistics

and contrastive

studies

215

Eleanor Rosch, a psychologist, investigated natural categories and found that not all members of a given category are equally representative. For example, in our culture robin is more representative or prototypical in the category of birds than chicken, penguin or ostrich. Thus, according to Rosch, the degree to which some creature fits the category bird is a function of humanly-relevant properties — perceptual, functional, motor, intentional, etc. which it shares with prototypical birds, such as sparrows, robins, etc. Moreover, she discovered psychologically "basic" categories, which from the set-theoretical standpoint are not fundamental. The experiments have shown that such concepts as D O G and C H A I R are basic while M A M M A L and F U R N I T U R E are superordinate, and R E T R I E V E R and R O C K I N G - C H A I R are subordinate (Rosch 1973; Rosch —Lloyd, quoted after Lakoff 1982: 17 — 19). Basic level categories were found to "have maximal clusters of humanly-relevant properties" (Lakoff 1977: 9), a fact which manifests itself in having visual prototypes for D O G and C H A I R but not for M A M M A L and F U R N I T U R E . Lakoff additionally insisted that "thought, perception, the emotions, cognitive processing, motor activity, and language are all organized in terms of the same kind of structures" (Lakoff 1977: 11), which he called gestalts. He deliberately left the concept vague but suggested that gestalts are "at once holistic and analyzeable (sic!)" (Lakoff 1977: 11) and though they have parts, they are not reducible to parts since they have additional properties, which cannot be seen in their parts examined in isolation from the gestalt. 2 In the framework of cognitive linguistics, gestalts are idealized cognitive models of reality as it is reflected in the minds of language users. As idealizations, such models necessarily involve oversimplifications, metaphorical understandings, and both expert and folk theories of reality. Idealized Cognitive Models define natural categories, which are not objective, but are strictly connected with human experience of reality. They contain the usual propositional content (the sort of information that one is likely to find in a dictionary) as well as mental images of a visual, or more generally, sensory and kinaesthetic type. Words in language are not defined relative to relevant fragments of "objective" reality but rather relative to such models (cf. Lakoff 1982: 48 ff.). The concept of Idealized Cognitive Model corresponds to Langacker's concept of "abstract domain" (Langacker 1983: 55 ff.). One way to represent them is by means of "mental schemes", which we shall introduce later in this chapter. Further research has revealed that natural categories, characterized by lack of clear-cut category boundaries (fuzziness, ability to stretch), inter-

216

Chapter XI

nal prototypical structure, family resemblance of category members, and gradience (a member may be more or less representative of the category), permeate natural languages. Consequently, language cannot be described adequately in terms of strict rules which can only handle prototypical phenomena but are less and less effective when non-prototypical phenomena are involved. For example, it turned out that the fundamental phonological category, phoneme, is also a natural category (Jaeger 1980, quoted after Lakoff 1982: 94). Classical phonology was incapable of handling the well-known phenomenon of phonemic overlap, which involves situations in which many non-prototypical variants of some phonemes belong to a given category because of family resemblance, and because of functional considerations. These non-prototypical variants are often phonetically quite remote from prototypical variants. For example, the English apico-alveolar strong stop /1/ has several variants, of which only the most prototypical ones sound like [t]. The prototypical English [t] appears in syllable initial positions in front of stressed vowels. Other variants of ft/, in other positions, are sometimes quite remote from the prototype. For example, between a stressed and an unstressed vowel, as in the word water, jtj may be realized in a very non-prototypical way as a voiced stop [d] or even a 'flap' [J], resembling the [J] in the word right, [J] is also a variant of the English phoneme /r/. This example, in cognitive linguistics, is not treated as an inconvenient case, which the phonological theory cannot cope with, but as a natural situation characterizing all languages, in which categories have fuzzy boundaries. Eventually, the [i] in water is classified as a variant of jtj and the [α] in right as a variant of /r/, not because of their inherent structural properties, but because they are perceived as such by native users of English. They are thus assigned to appropriate categories not on structural grounds, which in this case fail, but on cognitive grounds. Ross (1973, 1975, 1981) has demonstrated that virtually all grammatical categories exhibit prototypical effects. So we have more and less prototypical nouns, verbs, adjectives, sentences, prepositions, etc. Ross has also shown that all types of grammatical constructions, such as passive, relative, coordinative, etc., also display prototypical effects. In comparison with the generative tradition, cognitive linguistics embraces a much larger scope of linguistic data, including the traditional domain of grammar, that is, morphology and syntax. In terms of cognitive linguistics, one may say that transformational grammar dealt only with the most prototypical phenomena, those that are best motivated in our experience, while it completely ignored all kinds of non-representative

Cognitive linguistics and con t ras live

studies

217

examples, so vital in everyday use of language. This restriction in the scope is particularly evident in the case of figurative meanings, about which generative grammars have virtually nothing illuminating to say. Transformational grammar generates only those sentences which are prototypical for a given category, which means that it ignores all those numerous interpretations in everyday uses of language, which are metaphorical extensions of prototypical senses. Metaphor has been placed in the very focus of attention of cognitive linguistics since human experience and human thinking are largely metaphorical (Lakoff—Johnson 1980: 229 ff.), which cannot fail to influence language. In Lakoff s estimation, "literal" senses of sentences amount to only 2 to 5 percent of all senses used in everyday communicative acts. This is the area embraced by the so-called core grammar consisting of about 20 rules (Lakoff 1983: 38). This situation results from the fact that, in Lakoff s words: "There appears to be a continuum between productive constructions and completely frozen constructions" (Lakoff 1983: 38). It is perhaps needless to add that the limitations of core grammars, pointed out above, can be attributed a fortiori to Contrastive Generative Grammar as it adds its own difficulties connected with semantic equivalence to those which are already inherent in the nature of generative grammars themselves. The emergence of cognitive linguistics has created the necessity of revising the work in contrastive studies. The first attempt in that direction was made by Schlyter (1980), who produced an excellent contrastive study of various nouns and verbs in German, French, Spanish, English, etc. in terms of the prototype theory, coming up with a number of insightful and valuable observations, unavailable outside the cognitive framework. In a brief section of his thesis, Kalisz (1981) attempted to restate the concepts of congruence and equivalence in terms of "partial pattern matching" as discussed in Lakoff (1977). This latter concept is based on the assumption that linguistic forms can be characterized by clusters of pragmatic, semantic, and syntactic properties (parameters). Various constructions, it appears, exhibit various degrees of correspondence of such parameters. This kind of correspondence may be referred to as matching, which extends over a continuum from full matching, through partial matching, to no matching at all. Kalisz tries to employ the concept of partial matching in explicating the familiar relations of equivalence and congruence, crucial in all contrastive studies based on structural and transformational models. He maintains that

218

Chapter

XI

Equivalence between two given structures is a matter of degree of partial pattern matching of properties. One can talk about a degree of syntactic equivalence even if lexical properties do not match, a pragmatic equivalence when the two structures produce the same perlocutionary effect in spite of their syntactic and lexical properties etc. The higher degree of matching of syntactic, semantic and pragmatic properties reflects the higher degree of overall equivalence between two or more constructions (Kalisz 1981: 45 — 46). As originally conceived in structural and transformational terms, congruence and equivalence referred to formal and semantic identity of the compared constructions (Krzeszowski 1967: 37; 1971: 38; Marton 1968b: 56 — 57). The separation of semantic and syntactic representations, as formalized in the standard theory, led to the "identical-deep-structure hypothesis" (cf. Chapter VII), which, as we remember, claimed that equivalent constructions and sentences have identical semantic representations, even if on the surface they are markedly different (Krzeszowski 1971: 38, 1979: 11). Formal diversifications occurred at various levels of derivations in the subsequent derivational histories of equivalent constructions. More similar constructions shared more rules and were diversified later, that is, at some level closer to the level of surface representations. The distance from the surface at which the first diversification occurred provided grounds for measuring the degree of similarity and difference of the compared constructions (Di Pietro 1971: 27; Krzeszowski 1974, 1979: 13). The formal device which was to accomplish this task is called Contrastive Generative Grammar. Contrastive Generative Grammar rendered the concept of congruence somewhat superfluous since all equivalent constructions were congruent by definition at some level of representation (if only at the semantic level, at which congruence was guaranteed as a matter of the initial postulate). Therefore, the idea of degrees of syntactic similarity is not new with Kalisz's suggestion. Kalisz's proposal requires at least two amendments. Firstly, the requirement that pragmatically equivalent structures must produce the same perlocutionary effect should be relaxed through substituting "maximally similar cognitive effects" for "the same perlocutionary effects" Not all speech acts have perlocutionary effects, so that under Kalisz's proposal a number of speech acts such as statements, rhetorical questions, etc. would be unaccounted for. Kalisz's proposal is too restrictive also because it contains an untenable and, I am sure, unintended, implication that

Cognitive linguistics and contrastive

studies

219

only those utterances which happen to be produced simultaneously in the same setting and with the same perlocutionary effect are equivalent. Even if restricted to perlocutionary effects, the requirement concerning their "sameness" rules out any prospects of generalization and renders the proposal uninteresting. It is obvious that a given perlocutionary effect can occur only once. Instead, it is reasonable to consider identity of effects as a special case of similarity, where identity is to be viewed as the highest theoretically conceivable degree of similarity. Secondly, the concept of partial pattern matching as envisaged by Kalisz in the context of contrastive studies seems to require some constraining. According to Kalisz, "Equivalence between two structures is a matter of degree of matching properties. Thus, it reflects a degree of partial pattern matching of properties" (Kalisz 1981: 45). This claim must be interpreted as suggesting that given a set S of pairs of equivalent constructions, C's and K's, one can arrange these pairs on the scale provided by the varying numbers of certain matching properties. Let us assume that S contains the following pairs of equivalent constructions, characterized by some properties which match and some which do not match. Let P's stand for those properties which match, and let p's stand for those properties which do not match: S = C,(PPPP) C 2 (pPPP) S = K,(PPPP) K 2 (pPPP)

C 3 (pppP) K 3 (pppP)

C 4 (pppp) K 4 (pppp)

The pairs of constructions in S display various degrees of similarity, with Q and K, being identical in all relevant respects and with C 4 and K 4 being different in all relevant respects. Considering this latter case, the question should be asked: on what grounds are they compared? Note that we have artificially limited the number of properties (parameters) to four, providing the upper bound on the scale of similarity with the value of 4 and the lower bound having the value of 1. N o such limitations can be arbitrarily imposed on actual constructions. However, if we do impose some more or less arbitrary limits on the number of properties considered, we still face the following problem: if two linguistic forms match with regard to one property, can they be said to exhibit partial pattern matching? From the set-theoretical point of view the answer seems to be "yes" After all, one property out of four is a part of the total four, and if it matches, which can be expressed in terms of set intersection, it is involved in partial pattern matching.

220

Chapter XI

However, consider the following examples from English and Polish: (1) (2)

All visitors are kindly requested to leave the boat Proszg siadac. 'Please sit down'

immediately.

(1) and (2) evidently share at least one property request. But can they be said to be equivalent, pragmatically or otherwise, merely because they exhibit this extremely low degree of pattern matching? Consider, moreover: (3)

Spieprzajcie stqd. 'Get the hell out of here'

and (4)

Statek zaraz odplywa. 'The boat departs in a moment'

In certain cirumstances (1) and (3) may have identical perlocutionary effects (visitors leaving the boat); yet, it is doubtful whether anybody would consider them as pragmatically equivalent. On the other hand, (4) may in some situations be a better equivalent of (1), even if it exhibits even less pattern matching with (1) than do either (2) or (3). The discussion so far leads to the following conclusions: (a) The mere number of shared properties is not decisive in establishing pragmatic equivalence, because it does not seem to be correlated with the degree of similarity as grasped by partial pattern matching. (b) The identity of perlocutionary effects does not guarantee that the corresponding utterances are pragmatically equivalent. (c) Pragmatic equivalence appears to be dependent on elements of extralinguistic contexts attending the production of equivalent utterances (see also Janicki 1985: 15ff.). Thus, pragmatic equivalence depends on the status attributed to various properties rather than on their number, and the status may vary according to the effects that the speaker seeks to achieve. Given a highly onomatopoeic poem such as "The Bells" by Edgar Allan Poe, the translator will seek to retain as many acoustic effects, imitating various kinds of bells, as possible. He will, therefore, attribute privileged status to the phonetic properties of the poem. A pragmatically equivalent translation of such a text will focus on the phonetic substance of the poem, perhaps

Cognitive

linguistics

and contrastive

studies

221

to the detriment of semantic and most probably syntactic properties. 3 On the other hand, in the translation of legal documents, pragmatic equivalence will be determined by maximum semanto-syntactic accuracy. Pragmatically equivalent legal texts will probably approach semantosyntactic equivalents as defined in Chapters III and VII. Briefly speaking, pragmatic equivalence of texts can only be seen as a realization of a number of conditions which are necessary to obtain particular cognitive effects evoked by those texts. The conditions vary depending on the kind of text, its internal structure and its cognitive function. Each of these factors may determine the privileged status of a certain property or a cluster of properties of the text, which then serve as tertium comparationis for pragmatic equivalence in a specific instance. In view of the above considerations, I would like to claim that if two linguistic forms across languages belong to one category within the domain of contrastive studies 4 , the properties crucial in determining category membership are of unequal status as category determinants. In other words, some properties are more privileged than others in assigning two linguistic forms in two languages to one category within the domain of contrastive studies. Moreover, the properties which are critically important in determining pragmatic equivalence constitute a gestalt in the sense described by Lakoff (1977). In the case of (1), tlu· most important parameters constituting the relevant gestalt are: polite request to leave the boat.

Other parameters, such as syntactic congruity or lexical congruity, in this case, play a less important role. The three relevant parameters are: type of speech act (request), modality (politeness), and anticipated perlocutionary effects (visitors leaving the boat). Focusing on any one of these three elements of the gestalt in the rendering of (1) in Polish results in the distortion of the original pragmatic function. Thus in (2) the anticipated perlocutionary effect is different, while in (3) the modality is different (rudeness rather than politeness), which in turn may influence the perlocutionary effect. This explains why (4) ranks higher on the scale of pragmatic similarity than does either (2) or (3): (4) is neutral with respect to politeness, and at the same time it does not rule out the possibility of being interpreted as a request with the perlocutionary effect such as anticipated in (1). Naturally, (5) and perhaps (6) would rank even higher on the same scale:

222

(5) (6)

Chapter

XI

Uprasza si§ wszystkich gosci ο natychmiastowe opuszczenie statku. (Impersonal construction) Wszyscy goscie sq proszeni ο natychmiastowe opuszczenie statku. (Passive construction)

(5) and (6) are the most accurate equivalents of (1) in all respects: syntactic, semantic, and pragmatic. 5 At this stage I would like to suggest that (5) and (6) are the most prototypical equivalents of (1), while (3) is less prototypical. At the same time (4) is a poor representative of the category of equivalents of (1), and (2) is probably outside the category. We thus note that linguistic forms may have more (or less) prototypical equivalents in other languages. Let us consider one more example in order to explore some further consequences of the prototype theory of equivalence. In her study of English prepositions and particles, Brugman (1981) distinguishes nearly 100 senses of the preposition-particle-prefix (ppp) over and among them, the following 20 non-metaphorical senses: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

The plane flew over the town. The plane flew over the hill. The helicopter is hovering over the town. The helicopter is hovering over the hill. He is walking over the hill. Sam lives over the hill. The glider flew over the wall. Harry jumped over the wall. Harry jumped over the cliff. The wall fell over. Sam turned the page over. Sam rolled the log over. The power line stretches over my yard. She spead the tablecloth over the table. The guards were posted all over the hill. He walked all over the hill. She held the veil over her face. He drove over the bridge. He's over. The water overflowed.

Each of these senses is associated with a specific image schema indicating trajectors (the objects situated somewhere or moving), landmarks

Cognitive

linguistics and contrastive

studies

223

(the places at which trajectors are situated or relative to which trajectors move), and paths, whenever they are involved. (For details see Langacker 1987) These examples do not include metaphorical extensions of the ppp over in such sentences as He has the authority over a staff of hundreds, as an extension of 3 or He turned the question over in his mind as an extension of 12. All these senses are related by virtue of family resemblance and can be chained in such a way that the most prototypical ones, that is, 1 and 2, are situated in the centre while the progressively less prototypical ones grow out in various directions. Towards the edges of the chaining, one finds those senses which bear the least resemblance to the prototypical ones, but which are locally more similar. The relevant part of the chain is presented in (7):

(7)

©

Θ

Θ

©

In addition, senses 2, 3, 5, 6, 8, 9, 12, 14, 16, 18, and 20 have metaphorical extensions, which are not indicated on the chain. In attempting to compare this material with the relevant material in another language, say Polish, the investigator faces the usual question: what is the equivalent of over in Polish? Naturally, given at least twenty senses of over in English one cannot expect a single word equivalent in Polish. In search of the equivalents Brugman's examples were submitted to a group of Polish students of English with the request to translate them into Polish. There were 25 respondents, all well advanced in English. The purpose of the exercise was to find out how the ppp over would be rendered in Polish. The results are presented in Table 3.

224

Chapter XI

Table 3. The preposition-participle-prefix over translated into Polish by 25 respondents English senses 1 2a 3 4 5

6

7 8b 9"

10b

12 13c 14 15

16

Polish equivalents

Number of answers

nadjponad nadjponad nad/ponad nadjponad po przez na instrumental case nadjponad, woköl za na obok przy po drugiej stronie niedaleko nad/ponad przez przez accusative case przez accusative case ζ po poprzez other means (phrases, prefixes) accusative case na drugp strone accusative case na drugg strone nadjponad przez na po na wokol na obszarze po przez accusative case wzdluz i wszerz wokol

25 24 25 25 10 9 2 2 1 1 18 3 1 1 1 1 24 1 16 11 10 6 6 1 1 24 23 2 22 3 19 4 25 13 10 1 1 15 4 4 1 1

Cognitive linguistics and contrastive studies

English senses 17d

18" 19a 20

a b c d

Polish equivalents

Number of answers

na accusative case nad przez accusative case po other means wy- (prefix) prze- (prefix)

15 8 1 21 2 1 24 18 7

225

One respondent provided no answer. In two cases two answers were provided: accusative case and przez. Two respondents provided no answer. In one case a completely irrelevant answer was given. 10 and 19 were nearly always rendered as pseudotransitive verbs przewrocic sig i skonczyc sif, respectively.

These results are interesting for several reasons. Firstly, if we coalesce nad and ponad as free variants, at least in the contexts in question, we see that the most prototypical senses of over (1, 2, 3, and 4) are invariably rendered as nad/ponad, which must be recognized as the most prototypical equivalent of over. Over 7, rendered as przez, comes second by a very narrow margin of one. (Po)nad as the most prototypical equivalent of over deserves the first mention in an English-Polish dictionary although Stanislawski has 'na', which is the only equivalent of the less prototypical over 14. Secondly, in less prototypical senses, the Polish equivalents of over vary over a considerable range of prepositions, prefixes, and certain other means such as inflections and special forms of verbs. Thirdly, the Polish equivalents of non-prototypical senses of over evoke varying degrees of agreement among the respondents. On the one hand, in the case of 8, 9, or 15, there occur considerable divergencies, on the other hand, we deal with complete unanimity in the case of 14. Although the reasons for these divergencies remain to be investigated, one can surmise that they are partly due to the gaps in the linguistic competence of the respondents, but also, at least to some extent, to their different cognitive processing of the situations are depicted by the examples (especially in the case of 15 and 17). Fourthly, less prototypical senses of over may have very prototypical equivalents in Polish alongside less prototypical ones. For instance, po

226

Chapter XI

calym wzgorzu appears to be the most prototypical equivalent of all over the hill in 16 while przez wzgorze and wzdluz i wszerz wzgorza are less prototypical equivalents of over 16. Fifthly, nad, the Polish equivalent of the five most prototypical senses of over, itself has a number of senses, many of which have metaphorical extensions and many other senses which do not even correspond to over. I have been able to distinguish at least 14 less basic senses of nad as a preposition and as a prefix. The eight basic senses of nad are exemplified as follows: 1. Balon unosi si§ nad miastem. 'The balloon is hovering over the town' 2. Samolot przelecial nad miastem. 'The plane flew over the town' 3. Wierzba pochyla sie nad wodg. 'The willow leans over the water' 4. Balon unibsl sie nad wysp§. 'The balloon rose over the island' 5. Balon przylecial nad wyspg. 'The balloon flew to (the area) over the island' 6. Usiedli nad wodg. 'They sat near the water (front)' 7. On mieszka nad morzem. 'He lives near the sea' 8. Wyjechali nad morze. 'They went to the seaside' Senses 1 and 2 are the most prototypical and constitute the core of the following chaining: (8)

®—@—@—®—®—CS>—@—® In addition there are the following senses:

9. 10. 11. 12. 13. 14. 15. 16.

On nadbiegl. 'He came running' Nadbudowali strych. 'They built the attic above' Nadwozie 'car body' Nadkomisarz 'supercommissar' Nad snieg bielszy 'whiter than snow' Nadcz/ow/e/c 'superman' Nadcisnienie 'hypertension' Nadspodziwany 'unexpected'

Cognitive linguistics and contrastive

17. 18. 19. 20. 21. 22.

studies

227

Nad/ac 'pour out a little' Nad ranem 'before dawn' Sqdownictwo nad chlopami 'peasant jurisdiction' Prace nad siownikiem 'work on a dictionary' Zachwyt nad urodg. 'delight at someone's beauty' Nadbutwiec 'start being affected by rot'

Allowing for possible modifications and further extensions, the basic chaining can be augmented thus:

In addition to the extensions mentioned here, many of which become metaphorical towards the end of the chain (e.g., 13 onwards), some of the spatial senses have direct metaphorical extensions, for example, 3 — wyrastal nad przecigtnosc 'he grew above the mediocrity', 4 — pochylal si§ nad kazdym bliznim 'he leaned over his every neighbour' The chaining of the senses of nad is quite complex, but all these senses exhibit family resemblance. Through relating all those divergent senses by means of family resemblance, we obtain a fairly coherent view of otherwise disparate phenomena. It will be noted that each of these senses of nad will have its more and less prototypical equivalents in English. 1, 2, and 3 will prototypically

228

Chapter XI

correspond to the English 'over' while 4 corresponds to 'above', etc. On the basis of these data one can state that over and nad are prototypical equivalents in the two languages, and that the relevant spatial senses, that is, 1, 2, 3, and 4 in English and 1, 2, and 3 in Polish exhibit complete pattern matching with respect to image schemas associated with those senses. To provide a complete contrastive study of the area of senses covered by over and by nad, a similar procedure would have to be adopted for each sense individually. Consequently, each sense would have to be explored with the prospect of finding its prototypical and less prototypical equivalents in the other language, until the entire area could be thus explored. The resulting chainings would be mutually interconnected by a complex network of links representing the equivalence of specific senses. Such findings could be additionally supported by the study of image schemas associated with particular linguistic forms. For example, the research could either confirm or refute the prediction which seems to ensue from the foregoing discussion that the image schemas evoked by over in its four most prototypical senses and by nad in its three most prototypical senses are centred around the following image schema:

where the trajector (TR) is either stationary or moving, but at a certain time either its fragment or its complete body finds itself in the position indicated by the black dot over the landmark (LM) in (10). Dashed lines represent these potential paths (P's). The image schema is a gestalt, against which the most prototypical senses of over and nad are centred. Apart from this gestalt, there are other gestalts which serve as centres for other senses of over and nad. These other centres attract other equivalents, etc. The cross-language landscape of senses can, therefore, be seen as a multifocal space with various linguistic forms in both languages, interconnected by identical gestalts. Every instance of the use of over involves focusing on the relevant sense associated with the relevant gestalt. In the case described here we have been dealing with linguistic phenomena which exhibit a great deal of resemblance. The very fact that the

Cognitive linguistics and contrastive

studies

229

most prototypical equivalents of the most prototypical senses of the English preposition over correspond to a preposition in Polish suggests a high degree of pattern matching. Yet, when less prototypical (and more conventionalized) senses are considered, or when typologically more distant languages are involved, one can expect a smaller degree of similarity (less pattern matching). Consider over 9, which is connected with the image schema (11):

(11)

LM

A small majority of ten respondents rendered over 9 as 'przez', misconstruing the image schema and visualizing it as

TR

(12)

Another group (6 respondents) apparently had a different image, namely something like (13), when they rendered over 9 as the accusative case 'przeskoczyl ska!?': TR

(13)

LM

230

Chapter XI

Finally, another group (6 respondents), by rendering over 9 as 'zeskoczyl ζ cliffu' imagined something in the nature of (14) and thus obtained the highest degree of pattern matching: (14) LM

Disregarding other, less numerous possibilities, we face at least three distinct gestalt interpretations of the situation involved in over 9 and three grammatical, though hardly accurate, translations of over 9 into Polish. These divergent interpretations are due to the fact that Polish has no way of expressing the situation depicted in (11) by grammatical means similar to those which are needed to translate the preceding eight senses of over. Therefore, any attempt to use an equivalent preposition or at least an inflectional ending or a prefix yields results which only partially match the original with respect to the relevant image schema. The relatively conventionalized English expression to jump over the c/i#"associated with (11) can be more fully matched with the Polish periphrastic (hence highly motivated!) expression skoczyc w dol przez kraw§dz urwiska. Full matching of the image schemas in this case is bought at a price: the degree of syntactic pattern matching is now lower, and the degree of conventionality of the equivalents is unequal since the English conventional expression is rendered in Polish by means of a less conventionalized and more productive collocation of words. To conclude, in contrastive studies conducted in terms of the prototype theory, gestalts and partial pattern matching, equivalent forms exhibit various degrees of pattern matching. Thus, similarity (and difference) can be evaluated by means of a gradient scale. The upper bound of the scale is delimited inherently by complete pattern matching of semantic and/or syntactic properties. This situation is most likely to occur when prototypical equivalents are involved, although, by no means, need this always be the case. The lower bound of the scale is not delimited by the matching patterns themselves since there is no a priori way of deciding on the necessary minimum of similarity required for the recognition of two linguistic forms as matching. Therefore, the lower bound of the scale is delimited cognitively through the bilingual informant's recognition of two linguistic forms in two languages as being in one category within the

Cognitive linguistics and contrastive

studies

231

domain of contrastive studies. Thus, the decision is made not on the basis of the inherent properties of the compared categories but on the basis of what Lakoff calls "background framing" (Lakoff 1982: 25). Notwithstanding the fact that category boundaries may be fuzzy, their area is delimited by the extent to which their various non-prototypical senses stretch. Such delimitations are also based on cognitive grounds.

Chapter XII

"Theoretical" and "applied" contrastive studies

In 1980, Carl James published his Contrastive analysis, the second book, after Di Pietro (1971), devoted to the theory and practice of contrasting languages. In contrast to Di Pietro's, James' book is much more pedagogically oriented. James touches on many problems which have a direct bearing on teaching foreign languages, though he addresses many controversial theoretical problems beclouding contrastive studies without claiming to provide definite solutions. In addition to a special chapter devoted to "Pedagogical Exploitations of Contrastive Analysis", the book abounds in digressions elsewhere whenever the author thinks that he has something relevant to say about language pedagogy. In addition to James' book, excellent accounts of the relevance of contrastive studies to teaching foreign languages can be found in numerous published works, such as N i c k e l - W a g n e r (1968), Catford (1968), Lee (1968a, b), Marton (1976), Fisiak (1981), Sanders (1981), and very many others. Since James' book is the most extensive treatment, so far, of the numerous controversial problems connected with the distinction between "theoretical" and "applied" contrastive studies, I hope its author will not object to my treating it as a take-off ground for my own subsequent thoughts. To a large extent this chapter will be parasitic on James' book, inasmuch as it will mostly contain ideas which the book has inspired. The most conspicuous feature of James' monograph is its eclecticism or, to use Sharwood-Smith's less emotive term, unorthodoxy (SharwoodSmith 1976: 50). James does not commit himself to any specific linguistic theory nor indeed to any mode of contrastive analysis but provides the reader with a large array of models, opinions, and proposals, while only occasionally extending moderate criticism. James' criticism is not, however, anchored in any system of criteria of empirical or theoretical adequacy. This rather liberal attitude to various proposals is to be applauded, but it has its dark sides too. Most importantly it shows in James' attitude to the very nature of contrastive studies, which in turn has a tremendous impact on the organization of the book. James seems to be unable to

234

Chapter

XII

decide whether contrastive studies is a part of applied linguistics or a branch of linguistics and what the mutual relations between contrastive studies, psychology, and pedagogy are. The problems which James faces are the same ones that lie at the heart of many linguistic debates. As such these problems cannot be avoided in any work attempting to provide a synthetic view of contrastive studies since the field is so closely connected with, and so obviously dependent on, theoretical linguistics. The first chapter promises a neat division into "pure" and "applied" linguistics and disposes of the views denying the existence of the latter as a theoretically-based discipline (Corder 1973: 10). James' main argument against considering an applied linguist as a consumer rather than producer of theories is that in order to consume something one must be selective and the selections must be guided by a theory of relevance and applicability. James' next argument is that unlike "pure" linguistics "applied" linguistics (including contrastive studies as its central component [p. 8]) is a hybrid discipline, embracing in some relevant aspects psychology and sociology, and, therefore, it is a discipline distinct from "pure" linguistics. It remains unclear how James' intends to show that by being a hybrid, "applied" linguistics is ipso facto a theoretical discipline. Being a hybrid only shows its different status vis-ä-vis "pure" linguistics. It seems that James' argument in favour of the theoretical status of "applied" contrastive studies rests on the use of the notion "a theory of relevance and applicability" One would, therefore, expect that this notion would be thoroughly discussed as a potential provider of the necessary criteria of adequacy of contrastive studies. Unfortunately, James' book falls short of these expectations. In the remainder of the book, James does not maintain his initial stand on the status of contrastive studies: occasionally recalling Fisiak's distinction between "theoretical" and "applied" contrastive studies, he announces at one place: "From now on I shall intend 'applied CA' whenever I use the term CA" (p. 8). Yet, the relevance of this distinction remains obscure as James continually refers to both theoretical and applied aspects of contrastive analysis. The relevance of both of these in language pedagogy is well demonstrated, even in the chapters dealing with supposedly "purely" linguistic aspects of contrastive analysis, i. e., chapters 3, 4, and 5, which deal with linguistic components, of contrastive analysis, and its microlinguistic and macrolinguistic aspects respectively. It appears that James' problems begin the very moment he declares that contrastive analysis is merely a study of interlanguage (p. 3), thus

'Theoretical"

and "applied" contrastive

studies

235

bluntly equating the two concepts, which, even though to some extent related, are in fact products of completely different approaches to language and are in no way identical. If contrastive analysis has the form of a generative grammar, then indeed it may serve as a model of interlanguage along the strictly confined lines suggested above in Chapter IX. But James insists that contrastive analysis should be a study of an interlanguage as a dynamic, changing structure. In his view, contrastive analysis cannot be a synchronic endeavour but must have the capacity to account for the changing nature of an interlanguage, that is, it must embrace the diachronic dimension. Let us marginally observe that the word "diachronic" is not particularly fortunate to use in this context. If contrastive analysis is to account for the changing nature of an interlanguage within an individual, then the word "dynamic" is probably more appropriate. "Diachronic" would be fitting if someone attempted to describe the changes which affect various interlanguages, and which take place in time for a whole given population of foreign-language learners. But, by his own admission, this is not what James means. Nonetheless, even if one accepts this narrower concept of "diachronic" meaning "changing", "evolving", one cannot help wondering about the nature of interlanguage and its presumably changing status. At this point, it is necessary to distinguish between approximative systems in the sense of Nemser (1971) and interlanguage in the sense of Selinker (1972). Such a distinction seems all the more necessary since the differences between the two concepts are implicit in the cited papers, though to my knowledge they have never been explicitly discussed. The two terms are often used as synonyms alongside Corder's transitional competence (Corder 1967, 1974: 25). An approximative system in Nemser's sense indeed refers to something elusive, continually changing in the direction of the target language and extremely difficult to describe except as a dynamic structure having a distinctly temporal dimension. In contrast to this, interlanguage in Selinker's sense appears to be much more stable, a kind of fossilized approximative system, not subject to changes, except those that characterize any language of a mature individual. Therefore, interlanguage is no less amenable to a synchronic description than any other language. More generally, any frozen competence, whether in one's native or in a foreign language (or interlanguage), is equally easy or equally difficult to describe diachronically, synchronically, or otherwise. Now, if this distinction is valid, it is obvious that the research into the emergence and evolution of approximative systems is a different matter

236

Chapter

XII

from any study of a particular interlanguage. From the psycholinguistic and applied linguistic point of view, the study of approximative systems is certainly more interesting, more rewarding, and more tempting. But a linguist is also entitled to describe any interlanguage within the framework of an available linguistic theory and the two fields must be distinguished. There is no doubt that one can conduct contrastive psycholinguistic research and within such studies investigate the rise and development of approximative systems. It is equally possible to investigate interlanguages and compare them with both the source and the target language. When taken to logical conclusions, Corder's stand on transitional competence leads to a complete annihilation of the traditional field of error analysis but certainly not of contrastive analysis! Error is no longer a valid linguistic concept as it can only be conceived in microsociolinguistic terms. The so-called error analysis turns out to be a contrastive analysis of interlanguage and the target language. The question arises whether such "errors" are predictable. The prediction of errors has always been a major concern of contrastive analysis so naturally James, too, expresses his concern in this respect. But in view of what we said above, the simple question whether contrastive analysis can predict errors has acquired an unexpected depth since now the question concerns no more nor less than the form of interlanguage and can be rephrased thus: "Is the form of interlanguage predictable?" "Are there any constraints on the form of interlanguage, and if so what are they?" Such questions touch on the very essence of modern linguistic theory. They are questions about the boundaries of language universals. If contrastive analysis is able to cope with such questions, it no longer deserves the humiliating title of the Cinderella of linguistics but should be allowed in the royal chambers of human knowledge. James does not draw the distinction between approximative systems and interlanguage along the lines presented above. Instead, he identifies interlanguage with the dynamic, changing structure, which we would prefer to call an approximative system. In James' view, contrastive analysis ought to be a study of such dynamic objects. In this sense, Fisiak's distinction between applied and theoretical contrastive studies becomes vacuous since contrastive analysis by definition must be concerned with processes of foreign-language learning and acquisition, that is, with the domains traditionally associated with applied linguistics. This is what James himself readily admits: In a sense, the contrastivist continually transcends his own competence, in that he is first and foremost a linguist, whose proper concern is

'Theoretical"

and "applied" contrastive

studies

237

with structure, and yet he presumes to draw conclusions about a mode of human behaviour, learning. He seems to act thus out of a conviction that his CAs possess some sort of psychological reality (James 1980: 178). Such a view of contrastive analysis seems to leave no room for "theoretical contrastive analysis" unless one is ready to accept the rather uncomfortable conclusion that such analyses are concerned with the study of interlanguages while "applied" analyses deal with the study of approximative systems. There is no obvious reason why "theoretical contrastive analyses" should be thus restricted. James' eclecticism is very well seen in his attempt to match the outdated behaviourist psychological model of language based on the StimulusResponse relation with modern linguistic theory, which he does in the chapter devoted to the psychological basis of contrastive analysis. The impossibility of reconciling the two approaches is manifested in the contradiction with which James concludes the chapter. On the one hand he writes: 'Transfer' is the psychological cornerstone of CA. I have shown how it is manifested in L2 learning. The concept of transfer originates in behaviourist psychology, which has been superseded by cognitive psychology (James 1980: 25). But immediately, on the same page, he adds: The contrastive analyst is not, and need not inspire (sic!) to become, a psycholinguist. It is the contrastive analyst's duty to chart the linguistic (structural) routes in L2 learning. His findings and those of the psycholinguist will be complementary, but their instruments and methods must be different. Thus, one remains in the dark: are contrastive analysis and psychology autonomous, or does psychology affect the conceptual framework of contrastive analysis? Furthermore, are contrastive analysis and psycholinguistics complementary, or is the former based on psychology, as the title of the chapter suggests. These are not merely questions of terminology. The neglect of this problem leads to numerous theoretical and practical difficulties, especially in connection with the often discussed "psychological reality" of linguistic models. One of the most controversial problems connected with the pedagogical applications of contrastive analysis is how cross-language differences are

238

Chapter

XII

related to learning difficulties. James devotes a fair amount of space to this vital matter but does not seem to be able to arrive at any definite conclusions. Due to some editorial mishap, the subchapter entitled "Scales of Difficulty" appears twice, first as a subchapter of Chapter 6 (James 1980: 147 — 148), then lengthy fragments are repeated virtually verbatim as a subchapter in Chapter 7 (pp. 187 ff.). When James first quotes Stock well, Bowen and Martin's scale of difficulty, he seems to be rather sceptical about its validity. Having evoked Tran-Thi-Chau's criticism of the scale (Tran-Thi-Chau 1975), he concludes: The scale is, of course, subject to empirical validation, though when one attempts this a whole array of other complicating factors motivation, aptitude, teaching or learning style, etc. enters the picture (James 1980: 148). These "complicating factors" indeed reduce the scale to a mere abstraction, weakly reflecting didactic reality. But in the course of the forty pages separating the two references to the scale of difficulty, James seems to have changed his mind. Having quoted even more criticism, notably by Nickel (1971) and Nickel — Wagner (1968), he unexpectedly and without any justification concludes: There is undoubtedly substance in all these criticisms of the scale of difficulty. But all of them can be answered to some extent without totally abandoning the conceptual framework it is based on (p. 192). Unfortunately, this is where the book ends, and James does not show how these criticism can be answered, nor whether indeed the framework can be defended. In addition to the critical comments that James himself quotes, one may add that the proposed scales are based on criteria which completely ignore semantic and communicative values of the relevant constructions. It seems that since the role that various grammatical forms play in communication is a distinctly motivating factor, it is impossible to design a sensible hierarchy of difficulty without considering it. There is no doubt that those forms which have low communicative value are less interesting to the learners and subjectively more difficult. This is but one aspect of why the original scales do not promise to provide a particularly useful framework on which the preparation of materials for teaching could be based. Another aspect is the following. There is no doubt that in a large number of instances places which are identical across languages are easier to learn than those that are not.

'Theoretical"

and "applied" contrastive

studies

239

As a matter of fact, the learner does not have to learn them at all but only to realize that they are identical. This truth can be verified by anyone who has tried to learn two languages, one genetically close to his native language and one genetically remote. The language which is genetically close has extensive familiar areas, which require little or no learning. For example, a Pole attempting to learn Slovak will find this language considerably easier to acquire than a Portuguese learning Slovak. This is particularly well seen in the case of receptive skills, listening and reading. All Poles and Slovaks living in the border areas daily experience the benefits of this situation. To various degrees, this concerns such languages as Bulgarian and Macedonian, Dutch and German, Swedish and Norwegian, Finnish and Estonian, and numerous other languages, representing various families. The same languages, which seem so easy to people using genetically closely related languages, may appear to be enormously difficult to people speaking genetically remote languages. The truth of these claims is so obvious that it requires no further comments or illustrations, but the only way in which it is possible to account for these facts is by saying that some languages indeed share more common properties than others. Therefore, for every language, it is possible to arrange all other languages in the ascending order of difference, and, at least in principle, of difficulty. But this type of scale would only define one kind of difficulty, i. e. the wholesale, overall degree of difficulty of particular languages in comparison with a given language. Specific grammatical phenomena could not be plotted on a similar hierarchy in which difference and difficulty would be strictly correlated, due to the reasons mentioned earlier on. James is certainly aware of these problems when he mentions them twice in his book. Unfortunately, in this case he contradicts himself again. Having referred to Lado's observation that what is the same in a foreign language requires no learning on the part of the learner inasmuch as he already knows it (Lado 1957: 7), James proceeds to draw the following conclusion: A bizarre conclusion from this claim is that since all languages have something in common — the 'linguistic universale' — each of us knows at least parts of languages we have never heard or read. This is however a reductio ad absurdum, not to be taken seriously (James 1980: 17). Yet, towards the end of his book, James seems to have changed his mind on this issue as well since he thus writes about identical places across languages:

240

Chapter

XII

Our assumption is that the LI L2 identities will not have to be learned by the L2 learner, since he knows them already by virtue of his LI knowledge. Thus, though I have never attempted to learn Icelandic, some aspects of this language are nevertheless known to me in advance: on the one hand, I 'know' those features of Icelandic that are universal, and those that are shared by it and my native language. This is not an absurd claim (p. 151). Numerous observations as well as some experiments support James' altered view on the issue. One might add that the native language is not solely responsible for the facilitation which is involved here. Polyglots over centuries have unanimously agreed that each successive language is easier for them to learn than the previous language. This phenomenon can only be explained by a theory claiming the existence of identical places across languages, which do not have to be learned: the more languages one knows the less one has to learn in each successive language. In other words, in each new language a polyglot finds less and less of what he still does not know. Yet, in specific instances "identical" does not necessarily mean easier to learn, where by "easier to learn" one reads "not conductive to errors" In many cases, learners expect things to be different in a foreign language. In Chapter IX we quoted the example of Polish learners of Russian habitually misplacing the stress in the word ponjal ('he understood') in spite of the fact that the equivalent Polish word is stressed identically. Identical areas are also occasionally causes of errors of avoidance: anticipating things to be different in the foreign language, the learner will avoid identical constructions fearing an error. For example, many Polish learners of English shun the phrase from time to time under the influence of the congruent Polish phrase od czasu do czasu. These and many other examples show that cross-language identity and difference cannot be uniquely associated with ease and difficulty, respectively. A workable scale of difficulty must be based on the investigation of various other parameters in addition to the scales of difference. One of the burning problems in the theory of contrastive analysis concerns the selection of the most suitable linguistic model. The following alternative presents itself: either select the best theory, regardless of whether or not it can be usefully employed in contrastive analysis or select the model which is best suited for it. Thus, the contrastive analyst faces a rather dramatic choice: unless fortune has it that one theory is best by both these counts, he will be unfaithful either to his linguistics

'Theoretical"

and "applied"

contrastive

studies

241

or to his contrastive analysis. James suggests two solutions to this dilemma: 1. Provide the fullest descriptions of LI and L2 in terms of models which can best achieve this task, and then translate both into a form which is model-neutral, a sort of a contrastive-analysis variant of "etalon language" used in translations (Mel'cuk 1963: 62). 2. In the description concentrate on one of the two L's involved, namely the one which is to be learnt, following Filipovic's procedure whereby his contrastive analysis of Serbo-Croatian and English is descriptively biased towards English as the target language (Filipovic 1975). There exists a third possibility, and this involves the concept of prototypes and partial pattern matching as described in Chapter XI of the present work. The comparative procedure would not require of any particular category or construction in LI to be an ideal realization of some prototype. The equivalents in L2 would not have to be ideal realizations of the prototype either. But contrastive analysis would reveal that such equivalent pairs share some properties but not others, with one or the other item being closer to the prototype by virtue of sharing more features with it. This procedure differs from James' solution 1 in that the two descriptions would not be independent but would both be anchored in the description of the prototype. James faces this dilemma practically when he compares advantages and disadvantages of generative phonology in comparison with taxonomic phonology. James opts for taxonomic phonology, committing himself in the following way: Given the choice between taxonomic and generative phonology, while accepting that the latter is probably more powerful for 'pure' linguistic purposes, we should, as Burgschmidt and Götz (op. cit.: 199) do, opt for the former and weaker, for the simple reason that it is more practical and concrete (James 1980: 82). But even if one accepts taxonomic phonology one still faces alternative possible descriptions of the same data in one language. In Chapter V, we pointed out that for Polish one encounters a number of different analyses of phonological systems, leading to different inventories of phonemes. The Polish palatalized obstruents [p', b', t', d \ Γ, v'] can be analysed as separate phonemes or as positional variants (before /i/) of non-palatalized homorganic obstruents. The analysis depends on whether [i] is recognized as an allophone of /i/, mutually exlusive with [i], or whether [i] and [i]

242

Chapter

XII

are considered to be in contrast. In this latter case, it becomes necessary to postulate contrast between palatalized and non-palatalized obstruents. The two analyses are mutually incompatible, but the second one is preferable because it allows to reduce the number of Polish obstruents to non-palatalized ones only and to consider the palatalized obstruents as positional variants of the non-palatalized ones. This reduction is accomplished at the expense of having only one more vowel phoneme, viz. /[/ in contrast with /i/. The principle of economy, so much valued in taxonomic phonology, is decisive in this case. However, such an analysis conceals a number of important phonetic (allophonic) differences between Polish and English. One of these differences consists in the fact that in English, unlike in Polish, obstruents are not palatalized in front of high front vowels. Such an analysis will fail to predict negative transfer from Polish unless the contrastive analysis is continued at the subphonemic level. This indeed seems necessary in view of the fact that the most conspicuous deviations from the norms of the target language occur at the level of phonetic realizations of equivalent phonemes. The example evoked above again shows that even the selection of one linguistic model does not guarantee a uniform description within that model, and the contrastive analyst must be continually on the watch to seek a balance between the demands imposed by the accuracy of description and the accuracy of comparison. Still, James' readiness to sacrifice "purely" linguistic merits, such as accuracy and economy, in favour of such qualities as "more practical and concrete" (p. 82) leaves one with a certain amount of discomfort, even if this position is probably practically justified. And yet Gussmann (1984: 29) writes: "we reject any a priori assumption that some type of phonology may be better suited to contrastive ends than others and maintain instead that whatever phonology is judged to be most adequate in general terms should also be used in contrastive studies Gussmann's assumption may well point to one of the reasons why "pure" linguists have little sympathy for the dilemmas of contrastive analysts: confidently preoccupied with the descriptive adequacy of their models "pure" linguists turn a blind eye to the practical needs of applied linguists. The greatest merit of James' book is that it boldly faces all these difficult problems of which the author is profoundly aware, Instead of offering pseudosolutions and formulating sweeping and unmotivated claims, he leaves many of those controversial problems open. Thus, to a large extent the relevance of contrastive analysis to language teaching remains a mystery. But this is certainly not James' fault. We still know

'Theoretical"

and "applied" contrastive

studies

243

far too little about the nature of foreign language learning in general and about the nature of possible contrasts in particular. For example, we still do not know whether there is such a thing as an impossible contrast. If contrastive analysis could specify possible contrasts in possible human languages, and if it could rule out impossible contrasts, it would be related to general linguistic theory in a natural way. But general linguistic theory is still far from being capable of performing this sort of task. It is not even possible to formulate reasonable research strategies along these lines. For the time being, contrastive analysis must be limited to predicting potential errors, even if one builds it, as James does, on psycholinguistic foundations. Actual performance must still remain out of reach. All these dilemmas and controversies show that contrastive analysis remains a living discipline, and that interest in it, which has not faded in the past few years, is well founded in the wealth of theoretical and practical problems that still remain to be investigated.

Notes

Introduction Actually, we arc still oversimplifying matters in as much as "similar" cannot be interpreted as an absolute value but only as relative to a feature or a set of features, or more generally as "similar in some respect or in some respects" (see Chapter IV). Furthermore, we do not analyse the concept "language element", which we allow to subsume such various structural concepts as "phoneme" "morpheme" "syntagm" or "a syntactic structure" For the purpose of argumentation, we assume that all the problems inherent in analysing language into such "elements" have been satisfactorily solved.

Chapter I In fact, what is compared are particular aspects of languages, such as various grammatical systems, constructions or rules, not languages in their entirety. "Comparison of languages" is an abbreviation embracing such specific comparisons. For details see Chapter IV.

Chapter II 1. This chapter is largely based on Krzeszowski (1984). 2. One of the few exceptions is provided by James (1980: 90), who mentions "substantive" tertia comparationis for phonological and lexical contrastive studies, as well as formal (surface structure), semantic (deep structure) and translation equivalence as the best tertia comparationis for contrastive studies, provided it embraces both semantic and pragmatic equivalence (James 1980: 178). No attempt is made, however, to relate translation equivalence as tertium comparationis to other kinds of possible tertia comparationis within the overall landscape of contrastive studies. Substance as tertium comparationis is also briefly mentioned in Rusiecki (1976: 37). 3. Strictly speaking, squares constitute a proper subset of rectangles, but this does not, of course, rule out the possibility of comparison.

Chapter III 1. Largely based on Krzeszowski (1989). 2. The terms "pure" and (for example, pedagogically) "oriented" seem to be more fitting as expressions of the teleological difference between the two kinds of contrastive activities. 3. No extensive statistical data are available to support these impressionistic observations. However, Becka (1978: 135) has found that in a sample of 100 sentences from a Czech translation of Three Man in a Boat by J. K. Jerome, not a single occurrence of participle/ gerund construction is to be found as a condensation of a subordinate (complement) clause. My own quick count of the corresponding fragment in the Polish version of the text has yielded the same result. These findings contrast sharply with English, as in the original text eleven instances of the participle/gerund construction can be found.

246

Notes

Chapter IV 1. In her exhaustive survey of passive constructions in various languages Siewierska (1984: 43—44) considers Basque sentences with the so-called intransitive subject and the intransitive auxiliary d-a (a form of izati 'be') to be cases of passives. If one accepts her analysis, Basque and English would constitute situation (b). However, Siewierska considers the broadest possible sense of the concept "passive" Under narrower senses, it might be required of passive sentences to have special nominal morphology as well as syntactic markers as features minimally required of passive sentences in contrast to active ones. This, as Siewierska's examples show, is not the case in Basque. If this contrast is required of passives, English and Basque would indeed represent situation (c). 2. A detailed sociolinguistic analysis of the forms of address in English and Polish can be found in Ludkiewicz (1985).

Chapter V 1. Characteristically, the most extensive contrastive study of intonation known to me ends with a similar plea for more research in the domain of emotive aspects of intonation: A contrastive analysis of emotional-attitudinal intonation would also be worth making. The difficulties of such an analysis would be considerably greater The different available descriptions do not examine the same sets of attitudes; Moreover, in the attitudinal-emotional functioning of language the role of paralinguistic features (which are largely undescribed) is increased (Varga 1975: 133). 2. In fact many examples discussed in Fisiak et al. do not involve semanto-syntactically equivalent sentences, i.e., closest approximations to acceptable word-for-word translations (cf. Chapter VII). We shall return to this notorious problem below. 3. This section is a considerably expanded version of Krzeszowski (1981c). 4. Outside the scope of our investigations remain, for the time being, some further phenomena which could also be studied contrastively, for instance diachronic distinctions (archaisms), geographical distinctions (regionalisms), metaphorizations (see, however, Chapter XI), and morphophonological nets in what Nowakowski (1977: 27) calls the LEXICON. 5. In fact, Lado mentions one more "pattern of difficulty" connected with geographical distribution of certain lexical items, but since it is discussed within the context of one language, not as a contrastive phenomenon, we will not discuss it here. 6. Skorupka divides phraseological units into 'zwi^zki frazeologiczne stale, fyczliwe i luzne' which roughly corresponds to the tripartite division introduced above (Skorupka 1967: 6 - 7 ) . 7. Cf. Preston (1979), where as many as 40 categories of language variation are distinguished. 8. I wish to thank Ms. Katarzyna Mostowska for kindly providing Polish pragmatic equivalents of Fillmore's examples.

Chapter VI 1. For detailed surveys of the classical tradition in grammar see Dinneen (1967: 70 ff.) and Lyons (1968: 44 ff.). 2. The so-called Saxon genitive is a form intermediate between an inflectional ending and a function word (Krzeszowski 1980: 140).

Notes

247

3. One of the earliest fully comprehensive comparative grammars of two modern languages, written for pedagogical purposes, is Nicolas Salmon's Grammaire Angloise comparee avec La Grammaire FratiQoise published in London in 1797. 4. For historical accuracy it must be noted that most structuralists distinguished between two kinds of meaning: structural and lexical. For example, Francis defined structural meaning as "The meaning which a linguistic structure has over and above the lexical meanings of the words it contains" (Francis 1954: 595) and lexical meaning as "Meaning of a morpheme or word apart from the meaning it acquires by virtue of its position in a larger structure; 'dictionary meaning'" (ibid). Ignoring the problem of circularity and indeterminacy involved in these "definitions", whenever we talk about "meaning", we shall focus on "lexical" rather than "structural" aspects of what to us seems an overall, integrated concept of meaning. 5. Based on Jassem (1954). 6. Based on Jassem (1962, 1966). 7. For a detailed study of such matters in Georgian and East Armenian see Kramsky (1978). 8. Halle (1964: 326) distinguishes four degrees of narrowing of the vocal tract: The most extreme degree of narrowing, termed contact, is present when two opposite parts of the vocal tract touch. Stop consonants such as [p] [d] or [k] are articulated with contact at different points of the vocal tract. A less extreme degree of narrowing, termed occlusion, is one capable of producing turbulence. Occlusions are characteristically involved in the production of fricatives such as [v] [s] or [s]. The next degree of narrowing, termed obstruction, is exemplified in the articulation of glides such as [w] or [j]. The fourth degree of narrowing, termed constriction, is that manifest in the articulation of diffuse ("high") vowels such as [i] or [u], 9. Some phoneticians consider English final consonants in match/mxt$l and bridge/bridal to be clusters stop + sibilant rather than single segments called affricates. Some arguments in favour and against either view can be found in Gimson (1962: 166fT.). See also Jassem (1983: 222 — 223) for the view maintaining that /tf/ and /d$/ are segments in contrast to his earlier view that they are clusters (cf. Jassem 1954: 38 fT.). Structural phonology abounds in such alternative analyses. In Chapter V, we presented two possibilities in the analysis of the Polish vowels. Such alternative analyses have an obvious and immediate bearing on the results of contrastive studies. 10. In the most recent versions of transformational generative grammar, the concept of rule no longer occupies the central position. Instead, various universal principles and parameters are in the focus of attention, cf. e.g., Chomsky (1982). Since similarities rather than differences (contrasts) preoccupy linguists of this persuasion, we shall not, at present, evaluate these approaches in terms of their relevance to contrastive studies. 11. Curiously enough, Marton's definition of congruence is silent on matters of stress and other suprasegmental phenomena, which have turned out to be of critical importance in establishing cross-language correspondences. It must also be noted that later models promoted the so-called lexicalist hypothesis whereby the discussed items were listed in the dictionary (or derived by word-formation rules) rather than derived transformationally (Chomsky 1972). 12. Van Buren (1974: 303 ff.) expertly discusses some further difficulties connected with implementing the standard theory in contrastive studies. 13. For some ideas on how to incorporate various elements of Μ in the case grammar see Dillon (1973: 271 - 2 7 9 ) and Stockwell et al. (1973: 27 fT.). 14. The revisions represent two major trends: increasing the number of cases (e.g., Nilsen 1972) and limiting the number of cases, usually to four (e.g., Anderson 1971; Krzeszowski 1974, 1979).

248

Notes

Chapter VII 1. Based on Krzeszowski (1971, 1981b). 2. At this point we ignore the question of aspect in Polish. We shall presently return to this problem, which slightly complicates the issue in question. 3. Since the verb uzyc lacks the synthetic form of the iterative, we provide another example to Fill the existing systematic gap of the verb in question. 4. A detailed discussion of the analogous English-Polish contrast can be found in Cygan (1974).

Chapter VIII The account of contrastive generative grammer in this chapter is a summary of Krzeszowski (1974, 1979: 1 9 - 2 0 ) . 2. This representation is oversimplified since the predicate can be further analysed. In the subsequent part of this chapter, we shall outline a way in which sentences such as (1) and (2) can be analysed semantically in a more detailed way. 3. In Krzeszowski (1979) I erroneously analysed (a)

The invitation of the doctor surprised John

and its Polish translations (b) (c)

To, ze doktor zostal zaproszony, wywolalo zdumienie Johna To, ze doktor zaprosil kogos, wywolalo zdumienie Johna

as semanto-syntactically equivalent. In view of the restricted definition of semantosyntactic equivalence, I now see that neither (b) nor (c) can be recognized as semantosyntactic equivalents of (a). The only possible such equivalents of (a) in Polish is (d)

Zaproszenie

doktora zdumialo

Johna.

which is ambiguous in the same way as (a), so the problem of disambiguation does not arise in this case since contrastive generative grammar will correctly match (a) and (d) as a semanto-syntactic equivalent on the basis of the identity of their (at least) double semantic inputs. Naturally, this example does not affect the gist of the argumentation presented here. 4. Reportedly, Navaho (Apresjan 1966, 1971: 5 7 - 5 8 ) and Nootka (Hockett 1958: 2 2 4 225) are void of the distinction between verbs and nouns. On the conceptual level "things" in Navaho do not exist in a three-dimensional space but rather only in time, so that everything is in the state of continual "verbing", a dynamic process resembling Heraclitean panta rhei. In a fictitious language called Tlön, described by J. L. Borges in one of his Ficciones, entitled "Tlön, Uqbar, Orbis Tertius", one encounters the same situation as regards the dynamic status of concepts which in most languages are confined to three-dimensional space and expressed as "nouns" rather than as verbs. 5. The view of reference adopted here is compatible with the concept of ICM's (Lakoff 1982: 48 ff. 1986: 68 ff.), relative to which meanings of words are defined, rather than relative to elements of the world at large. For details see Sampson (1970a, b) and, in a somewhat different context Ludskanov (1972, particularly pp. 78 ff.). 6. Such an understanding of the notion of reference as a composite association makes it possible to explain why a semantic interpretation of sentences like (a)

I spoke to a man who spent three years on Mars

is possible even if there exists no phenomenon in the outer reality with which a man of (a) could be associated. Nevertheless, there exists a relation of reference between the modified noun of (a) and some mental image of a man present in my mind. The

Notes

249

potential multiple association of one referent with more than one linguistic expression can be seen in the possibility of associating such forms as the man whom I met yesterday, Mr. McClusky or my greatest enemy with a single referent. It may be argued, following Kurylowicz (1971: 9), that "future" is different from "past" and "present" in that it is a sphere of time embracing events which are not real in the sense that they only exist as predictions, that is, as acts of the speaker's mind. The modal character of the future tense is well grasped by Boyd and T h o m e ' s performative analysis of modal verbs in English (Boyd —Thome 1969). In terms of contrastive generative grammar, could formulate the appropriate representation alternative to the one suggested here, which derives the future tense from fTL^

Chapter IX Contrastive generative grammar does not attempt to account for actual processes of sentence production by the foreign language learner inasmuch as it is an account of competence, not of performance. Therefore, the descriptions of equivalent constructions across fully elaborated codes, as well as across fully elaborated codes and "reduced" codes cannot be interpreted as descriptions of what the learners actually do when they produce or understand sentences (cf. Chesterman 1980: 18 ff.). However, the descriptive account furnished by contrastive generative grammar may suggest some strategies in psycholinguistic research whereby the accuracy of such descriptions may receive additional support. For some evidence to that effect see below. 2. Another "cosmetic" transformation will ensure agreement of the verb with the subject N P if the present tense is involved {do becomes does in the singular). 3. "Premature" by comparison with the situation in the fully elaborated code. 4. "Occasional" questions contain the Wi-word in the position before the preposing and are uttered with heavy stress and rising intonation. "If someone said 'John will read the telephone book', one might respond, 'John will read what Τ this response is an occasional form" (Brown 1970: 133). Therefore, such questions must be derived from other sources than those that underlie normal w/i-questions, and, consequently, they cannot be considered as representing an intermediate stage in the derivation of normal H'A-questions.

Chapter X 1. Based on Krzeszowski (1981d). 2. A parody typically involves a high concentration of forms which are already outstandingly frequent in the text(s) which serve(s) as the object of parody. One can envisage studies aimed at a statistical account of parody. 3. For an excellent phonological quantitative contrastive study see Kramsky (1978). 4. Such a correlation is highly plausible as Polish employs infinitives in certain ways in which English does not, for example as categorical imperatives in trzymac sig mocno, chlopaki 'hold on, boys' or in such constructions as trzeba nam myslec po nowemu 'it is necessary that we think in new ways' (lit. 'needs us to think anew'). Without more extensive investigations in more than two languages, it is impossible to demonstrate a direct correlation between more restricted contexts in which particular forms appear at the level of language description and less frequent occurrences of these forms in continuous texts even if such correlations can be hypothetically expected. 5. A very detailed systematic contrastive study of morphological, syntactic, and semantic properties of Polish nominals in contrast with English ones can be found in Lewan-

250

Notes dowska (1975) and need not be recapitulated here. A detailed description of Polish nominals can be found in Puzynina (1969) and Grzegorczykowa (1979). The equivalent English phenomena are described in Lees (1963) and Fraser (1970).

Chapter XI 1. The present chapter is a considerably expanded version of Krzeszowski (1986b). 2. According to Lakoff (1982), gestalts exhibit various other characteristics which need not concern us at this point. 3. Sometimes a pragmatically valid translation will resemble the original only by the rhythmic pattern, as in the case of nursery rhymes or various fixed expressions which do not convey any meaning outside a strictly limited historical and cultural context of a given nation or a group of people. Such in the case with the famous couplet from one of Mickiewicz's poems: Emilia Plater Dziewica-bohater which is a distorted, folk version of two lines from Smierc pulkownika (The colonel's death), functioning very much like a nursery rhyme with distinctly comic overtones parasitic on a historical allusion. An inspired translator once extemporized the following English version: Helen Shapiro virgin-hero achieving the comic effect which unexpectedly issues from the originally pompously sounding poem and not bothering to find an English historical equivalent of the said hero or trying to explain her place in Polish history. It is this comic effect, reinforced by the rhyme in which the word 'hero' matches the name of a girl, which received the privileged status and determined pragmatic equivalence in this particular case. 4. Two linguistic forms belong to one category within the domain of contrastive studies if there is a tertium comparationis whereby they can be compared. The common tertium comparationis delimits the category membership in the following way: saying that a form A in LI is in some way equivalent to form Β in L2 entails that there is such a tertium comparationis whereby they can be compared. 5. Kalisz (1986: 1251) suggests that (6) may be a better (pragmatic?) equivalent of (1) than (5) inasmuch as (6), like (1) is a passive sentence while (5) is an impersonal sentence. Thus, other things being equal, (6) shares one more property with (1) than does (5). Kalisz seems to be confusing pragmatic equivalence with semanto-syntactic equivalence (for which he has designed the unfortunate terms "overall equivalence"); (6) is certainly the semanto-syntactic equivalent of (1), being the closest approximation to acceptable word-for-word translation of (1) (cf. Chapter VII). It does not necessarily follow that it is the best pragmatic equivalent since passive constructions in Polish do not enjoy the same stylistic status as they do in English. Instead, Polish makes ample use of impersonal constructions such as (5). In this case stylistic considerations may override syntactic considerations, so that (5) may be a better pragmatic equivalent of (1). Again the privileged status of properties rather than their mere number seems to be decisive.

References Abbreviations IRAL ISB JESL L.A.U.T. MSLL PAKS

PSiCL PSML PTJ SAP SSLA YSCECP

International Review of Applied Linguistics in Language Teaching Interlanguage Studies Bulletin (Utrecht) Journal of English as a Second Language (Philadelphia, Pennsylvania) Linguistic Agency. University of Trier Monograph Series on Languages and Linguistics (Georgetown University. School of Languages and Linguistics) Projekt für Angewandte Konstrastive Sprachwissenschaft (Institut für Literatur- und Sprachwissenschaft. Lehrstuhl Anglistik: Linguistik, Universität Stuttgart) Papers and Studies in Constrastive Linguistics (Adam Mickiewicz University, Poznan, Poland) Prague Studies in Mathematical Linguistics (Academia, Publishing House of the Czechoslovak Academy of Sciences) Polskie Towarzystwo Jezykoznawcze [Polish Linguistic Association] Studia Anglica Posnaniensia. An International Review of English Studies. Adam Mickiewicz University, Poznan Studies in Second Language Acquisition. Indiana University The Yugoslav Serbo-Croatian English Contrastive Project. (Institute of Linguistics. Faculty of Philosophy. University of Zagreb, Yugoslavia: Center for Applied Linguistics, Washington D.C., U.S.A.)

Aarts, Flor —Herman Wekker 1982 "Writing a contrastive grammar of English and Dutch", PSiCL 14: 25 — 43. Agard, Frederick B. —Robert J. Di Pietro 1965 The sounds of English and Italian (Chicago — London: University of Chicago Press). Allen, J. P. B . - S . Pit Corder (eds.) 1974 Papers in applied linguistics (Edinburgh Course in Applied Linguistics I) (Oxford: Oxford University Press). Anderson, John M. 1971 The grammar of case: towards a localistic theory (Cambridge Studies in Linguistics) (Cambridge: Cambridge University Press). Anderson, Stephen R. 1974 The organization of phonology (New York: Academic Press). Apresjan, Jurij D. 1966 Idei i metody sovremennoj lingvistiki [Concepts and methods in contemporary linguistics] (Moskva: Izdatelstvo Prosvescenie). 1971 Koncepcje i metody wspoiczesnej lingwistyki strukturalnej [Concepts and methods in contemporary linguistics]. Translated by Zygmunt Saloni (Warszawa: Panstwowy Instytut Wydawniczy). Arabski, Janusz 1979 Errors as indications of the development of interlanguage (Katowice: Uniwersytet Sl^ski).

252

References

Arnol'd, I. W. 1973

The English word: leksikologija sovremennogo anglijskogo jazyka [Lexicology of modern English] (Moskva: Vyssaja Skola). Awedyk, Wieslaw 1974 "The syllabic structures of English and Polish", PSiCL 2: 8 9 - 9 5 . 1976 "Some remarks on generative contrastive phonology", PSiCL 4: 53 — 60. 1979 "Towards pedagogical contrastive phonology", PSiCL 10: 1 2 5 - 1 3 3 . Bach, Emmon 1968 " N o u n s and noun phrases", in: Bach —Harms (eds.), 91 — 122. Bach, Emmon —Robert T. Harms (eds.) 1968 Universals in linguistic theory (New York: Holt, Rinehart and Winston). Banathy, Bela H. — Paul H. Madarasz 1969 "Contrastive analysis and error analysis", JESL 4.2: Π — 92. Bartsch, Renate —Theo Vennemann 1972 Semantic structures (Frankfurt am Main: Athenäum). Becka, J. V. 1978 "Application of quantitative methods in contrastive stylistics", PSML 6: 129-147. Berault, Peter 1688 A new, plain, short, and compleat French and English grammar (London). Biedrzycki, Leszek 1964 "Fonologiczna interpretacja polskich glosek nosowych" [A phonological interpretation of Polish nasals], Biuletyn PTJ 23: 25 — 46. Bloomfield, Leonard 1933 Language (New York: Holt). Boas, Hans Ulrich 1977 "Some remarks on case grammars as bases for contrastive studies", PSiCL 7: 2 1 - 3 2 . Bolinger, Dwight 1965 "The atomization of meaning", Language 41: 555 — 573. 1966 "Transformation: structural translation", Acta Linguistica Hafniensia 9.2: 130-144. Borkowski, G —L. Mickelsen 1963 "A contrastive study of the impersonal sentences of Polish and Russian" (Paper presented at the annual meeting of the Linguistic Society of America, Chicago, December 28 — 30). Bouton, Lawrence F. 1976 "The problem of equivalence in contrastive analysis", IRAL 14.2: 143 — 163. Boyd, Julian —James P. Thorne 1969 "The semantics of modal verbs", Journal of Linguistics 5: 57 — 74. Briere, Eugene J. 1968 A psycholinguistic study of phonological interference (The Hague: Mouton). Bross, J. S. 1962 "Problems of equivalence in some German and English constructions", Mechanical Translation 7: 8 — 16. Brown, Roger 1970 Psycholinguistics: selected papers (New York: MacMillan). Brugman, Claudia 1981 Story of over [M.A. thesis] (Bloomington, Indiana: Indiana University Linguistics Club).

References

253

Burgschmidt, Ε. —D. Götz 1974 Kontrastive Linguistik, Deutsch-Englisch: Theorie und Anwendung (München: Hueber). Catford, John C. 1965 A linguistic theory of translation (Oxford: Oxford University Press). 1968 "Contrastive analysis and language teaching", MSLL 21: 159 — 173 (Georgetown University Press). Chesterman, Andrew 1980 "Contrastive generative grammar and the psycholinguistic fallacy", PSiCL 11: 1 7 - 2 4 . Chomsky, Noam 1957 Syntactic structures (The Hague: Mouton). 1965 Aspects of the theory of syntax (Cambridge, Mass.: M.I.T. Press). 1972 Studies on semantics in generative grammar (The Hague — Paris: Mouton). 1975 Reflections on language (New York: Pantheon). 1982 Some concepts and consequences of the theory of government and binding (Cambridge, Mass. — London: M.I.T. Press). 1984 Modular approaches to the study of the mind (San Diego: San Diego University Press). Chomsky, Noam —Morris Halle 1968 The sound pattern of English (New York: Harper and Row). Coles, Elisha 1675 Syncrisis: a very important introduction about what is similar and what is different and how it facilitates learning (London). Comrie, Bernard 1976 "The syntax of causative constructions: cross-language similarities and divergencies", in: M. Shibatani (ed.), Syntax and semantics (New York: Academic Press), 261 —312. Corder, S. Pit 1967 "The significance of learners' errors" IRAL 4: 161 — 170 [1974] [reprinted in: J. C. Richards (ed.) Error analysis. London: Longman 19 — 27]. 1973 Introducing applied linguistics (Harmondsworth: Penguin Education). 1975 "'Simple codes' and the source of the second language learners' initial heuristic hypothesis", SSLA 1.1: 1 — 10. Cygan,Jan 1974 "Negation in English and Polish", PSiCL 2: 2 9 5 - 3 2 9 . Danilewicz, Tadeusz 1982 Ways of expressing cause in English and Polish [unpublished doctoral dissertation, Uniwersytet Sl^ski], De Camp, David 1968 "The field of creole language studies" SAP 1: 2 9 - 5 1 . De Geest, W. 1979 The relevance of relational grammar to contrastive linguistics. L.A.U.T. Series B. no. 39. Delattre, Pierre 1965 "Comparing the consonantal features of English, German, Spanish and French" IRAL 2: 1 5 5 - 2 0 3 . Dillon, George L. 1973 "Perfect and other aspects in a case grammar of English" Journal Linguistics 9: 271 - 2 7 9 . Dingwall, William O. 1964a Diaglossic grammar [unpublished Ph.D. dissertation, Georgetown University Washington D.C.].

254 1964b

Reference.

"Transformational generative grammar and contrastive analysis", Language Learning 14: 147—160. Dinneen, Francis P. 1967 An introduction to general linguistics (New York — Chicago — San Francisco—Toronto—London: Holt, Rinehart and Winston). Di Pietro, Robert J. 1971 Language structures in contrast (Rowley, Mass.: Newbury House). Dluska, Maria 1976 Prozodia jgzyka polskiego [Prosody of the Polish language] (Warszawa: Panstwowe Wydawnictwo Naukowe). Duczmal, Stanislaw 1979 "A contrastive semantic analysis of colour adjectives in Polish and English", PSiCL 9: 1 8 1 - 1 9 1 . Duskovä, Libusa 1978 "The simple sentence in Czech and English", PSML 6: 83 — 92. Enkvist, Nils Erik 1973 "Should we count errors or measure success?", in: Jan Svartvik (ed.), Errata: Papers in error analysis (Lund: Gleerup), 16 — 23. 1977 "Contextual acceptability and error evaluation", in: Papers from the conference on contrastive linguistics and error analysis. Stockholm and Abo Akademi Foundation), 1—26. Ferguson, Charles A. 1964 "Baby-talk in six languages", in: The ethnography of communication ( = American Anthropologist 66(6), pt. 2) (Washington D.C.: American Anthropological Association), 103 — 104. 1975 "Towards a characterization of English foreign talk" Anthropological Linguistics 17: 1 — 14. Filipovic, Rudolf (ed.) 1971 The Yugoslav Serbo-Croatian English contrastive project (Zagreb). 1975 Contrastive analysis of English and Serbo-Croatian (Zagreb: Institute of Linguistics). Fillmore, Charles 1968 "The case for case", in: Bach —Harms (eds.), 1 —88. 1984a "Remarks on contrastive pragmatics", in: Jacek Fisiak (ed.), 119 — 141. 1984b Frames and the semantics of understanding [unpublished ms. Berkeley: Department of Linguistics, University of California]. Fisiak, Jacek 1971 "The Poznan Polish-English contrastive project", in: Filipovic (ed.), 87 — 96. 1975a "The contrastive analysis of phonological systems", Kwartalnik Neofilologiczny 22: 341—351. 1975b Wstgp do wspolczesnych teorii lingwistycznych [Introduction to contemporary linguistic theories] (Warszawa: Wydawnictwa Szkolne i Pedagogiczne) (revised edition 1978). 1976 "Generative phonological contrastive studies", Kwartalnik Neofilologiczny 23: 1 1 9 - 1 2 4 . 1983 "Present trends in contrastive linguistics", in: Kari Sajavaara (ed.), CrossLanguage Analysis and Second Language Acquisition 1 ( = Jyväskylä Cross Language Studies 9), 9 - 3 8 . 1984 "On the roots of contrastive linguistics", Folia Linguistica 18: 139—153. Fisiak, Jacek (ed.) 1980 Theoretical issues in contrastive linguistics (Amsterdam: Benjamins).

References 1981

255

Contrastive linguistics and the language teacher (Oxford —New York —Toronto — Sydney — Paris — Frankfurt: Pergamon). 1984 Contrastive linguistics. Prospects and problems (Berlin —New York —Amsterdam: Mouton). Fisiak, Jacek —Maria Lipinska-Grzegorek — Tadeusz Zabrocki 1978 An introductory English-Polish contrastive grammar (Warszawa: Panstwowe Wydawnictwo Naukowe). Francis, W. Nelson 1954 The structure of American English (New York: Ronald Press). Fraser, Bruce 1970 "Some remarks on the action nominalization in English", in: R . J a c o b s — P. Rosenbaum (eds.), Readings in English transformational grammar (Waltham, Mass. —Toronto —London: Ginn), 83—98. Fries, Charles C. 1945 Teaching and learning English as a foreign language (Ann Arbor: University of Michigan Press) [ l4 1960], 1952 The structure of American English: an introduction to the construction of English sentences (London: Longmans). Gazdar, Gerald 1979 Pragmatics: implicature, presupposition, and logical form (New York —San Francisco —London: Academic Press). Ginzburg, R. S . - S . S. K h i d e k e l - G . Y. K n y a z e v a - A . A. Sankin 1966 A course in modern English lexicology (Moscow: Higher School Publishing House). Gimson, A. C. 1962 An introduction to the pronunciation of English (London: Edward Arnold). Gleason, Henry Α., Jr. 1955 Introduction to descriptive linguistics (New York: Holt) [revised edition 1961 ]. 1965 Linguistics and English grammar (New York — Chicago — San Francisco — Toronto —London: Holt, Rinehart and Winston). Gniadek, Stanislaw 1979 Grammaire contrastive franco-polonaise (Warszawa: Panstwowe Wydawnictwo Naukowe). Graczyk, Ireneusz n.d. On writing English as a second language: verb phrase in advanced learners' written English, I (Ms. Uniwersytet Lodzki). Grzegorczykowa, Renata 1979 Zarys slowotworstwa polskiego: slowotworstwo opisowe 3 [An outline of Polish word-formation: descriptive word-formation] (Warszawa: Panstwowe Wydawnictwo Naukowe). Gussmann, Edmund 1978 Contrastive Polish-English consonantal phonology (Warszawa: Panstwowe Wydawnictwo Naukowe). 1980 Introduction to phonological analysis (Warszawa: Panstwowe Wydawnictwo Naukowe). 1984 "Contrastive analysis substantive evidence and the abstractness issue" in: Stig Eliasson (ed.), Theoretical issues in contrastive phonology (Heidelberg: Groos), 2 7 - 3 6 . Halle, Morris 1964 "On the bases of phonology", in: Jay A. Fodor — Jerrold Jay Katz (eds.). The structure of language (Englewood Cliffs, N.J.: Prentice-Hall, 324 — 333. 1973 "Prolegomena to a theory of word formation", Linguistic Inquiry 4.1:3 — 16.

256

References

Halliday, Michael A. K. 1970 "Language structure and language function", in: John Lyons (ed.), New horizons in linguistics (Harmondsworth: Penguin), 140—165. Halliday, Michael A.K. —Angus Mcintosh — Peter Strevens 1964 The linguistic sciences and language teaching (London: Longmans). Harris, Zellig S. 1946 " F r o m morpheme to utterance", Language 22: 161 — 183. 1951 Methods in structural linguistics (Chicago: University of Chicago Press). Hewes, John 1624 A perfect survey of the English tongue taken according to the use and analogie of the Latine (London). Hill, L. A. 1968 Prepositions and adverbial particles (London: Oxford University Press). Hiz, Henry 1968 Referentials [Ms. University of Pennsylvania, Philadelphia, Pa.]. Hjelmslev, Louis 1961 Prolegomena to a theory of language. Translated by J. Whitfield (Madison: University of Wisconsin Press). Hockett, Charles F. 1958 A course of modern linguistics (New York: MacMillan). Howel(l), James 1662 A new English grammar (London). Hudson, Richard A. 1976 Arguments for a non-transformational grammar (Chicago —London: University of Chicago Press). Hymes, Dell 1972 "On communicative competence", in: J. B. Pride—J. Holmes (eds.), Sociolinguistics (London: Penguin), 269 — 293. Ivir, Vladimir 1969 "Contrasting via translation", YSCECP B. Studies 1: 1 3 - 2 5 . 1970 "Remarks on contrastive analysis and translation", YSCECP B. Studies 2: 14-25. Jaeger, Jeri 1980 Categorization in phonology: an experimental approach [Unpublished Ph.D. dissertation, Berkeley: University of California]. Jakobson, Roman —Morris Halle 1956 Fundamentals of language ('s-Gravenhage: Mouton). James, Carl 1974 "Linguistic measures for error gravity", Audio-visual Language Journal (Birmingham): 3 — 9. 1977 "Judgments of error gravities", English Language Teaching Journal 21: 116 — 124. 1980 Contrastive analysis (Harlow, Essex: Longman). Janicki, Karol 1985 "Tertium comparationis in contrastive sociolinguistics", Nordlyd 10: 7 — 27. Jassem, Wiktor 1954 Fonetyka jezyka angielskiego [Phonetics of the English language] (Warszawa: Panstwowe Wydawnictwo Naukowe). 1962 "The distinctive features of Polish phonemes", Speech translation laboratory. Quarterly progress and status report 1 (Stockholm: Royal Institute of Technology), 7 — 15. 1966 "The distinctive features and the entropy of Polish phoneme system", Biuletyn PTJ 24: 8 7 - 1 0 8 .

References 1983

The phonology Naukowe).

257

of modern English (Warszawa: Panstwowe Wydawnictwo

Jones, Daniel 1950 The phoneme (Cambridge: HefTer). Kakietek, Piotr 1980 English modal auxiliaries and their equivalent constructions in Polish (Katowice: Uniwersytet Sl^ski). Kalisz, Roman 1976 "On the kinship terms in English and Polish", PSiCL 5: 2 5 7 - 2 7 0 . 1981 The pragmatics, semantics and syntax of English sentences with indicative that complements and the Polish sentences with ze complements. A contrastive study (Gdansk: Uniwersytet Gdanski). 1986 "More on pragmatic equivalence", in: D. Kastovsky — A. Szwedek (ed.) (Berlin —New York —Amsterdam: Mouton de Gruyter), 1247 — 1255. Kalogjera, Damir 1971 "The expression of future time in English and Serbo-Croatian", YSCECP Reports 2: 1 2 0 - 1 3 4 . Karttunen, Kimno 1977 "The equivalents of the Finnish passive voice in English", Jyväskylä Contrastive Studies 4: 1 0 9 - 1 2 5 . Kastovsky, Dieter —Aleksander Szwedek (eds.) 1986 Linguistics across historical and geographical boundaries 1 — 2 (Berlin — New York —Amsterdam: Mouton de Gruyter). Katz, Jerrold J . - P a u l M. Postal 1964 An integrated theory of linguistic descriptions (Research Monograph No. 26) (Cambridge, Mass.: The M.I.T. Press). Kielski, Boleslaw 1957 Struktura jqzyköw francuskiego i polskiego w swietle analizy porownawczej [A comparative analysis of the structure of French and Polish] (Lodz: Lodzkie Towarzystwo Naukowe). Kohler, Klaus 1971 "On the adequacy of phonological theories in contrastive studies", in: Gerhard Nickel (ed.), Papers in contrastive linguistics (Cambridge: Cambridge University Press), 83 — 89. Kohn, Kurt 1974 Kontrastive Syntax und Fehlerbeschreibung (Kronberg, Taunus: Scriptor). Kolbert, Jack —Harry Goldby 1958 The first French handbook for teachers in elementary schools (Pittsburgh). Konderski, Stefan 1973 "Some surface and deep aspects of case in Polish and English", PSiCL 1: 65-77. König, Ekkehard 1970 "Transformational grammar and contrastive analysis", PAKS 6 (Stuttgart), 43-58. Kopczyhski, Andrzej 1968 English and Polish consonant phonemes [unpublished doctoral dissertation. Uniwersytet im. A. Mickiewicza, Poznan], 1973 "The Polish and English fricatives a problem in phonological equivalence", PSiCL 1: 9 3 - 9 8 . 1980 Conference interpreting: some linguistic and communicative problems (Poznan: Uniwersytet im. A. Mickiewicza. Seria filologia angielska nr. 13). Koutsoudas, Andreas 1972 "The strict order fallacy", Language 48: 8 8 - 9 6 .

258

References

Kramsky, Jiri 1978 "Quantitative analysis of near-identical phonological systems", PSML 6: 9-38. Kryk, Barbara 1987 On deixis in English and Polish. The role of demonstrative pronouns (Frankfurt am Main: Lang). Krzeszowski, Tomasz P. 1967 "Fundamental principles of structural contrastive studies", Glottodidactica 2: 3 3 - 4 0 . 1968 An outline of American English phonetics (Warszawa: Panstwowe Wydawnictwo Naukowe). 1970 Teaching English to Polish learners (Warszawa: Panstwowe Wydawnictwo Naukowe). 1971 "Equivalence, congruence and deep structure", in: Gerhard Nickel (ed.), Papers in contrastive linguistics (Cambridge: Cambridge University Press), 37-48. 1974 Contrastive generative grammar: theoretical foundations (Lodz: Uniwersytet Lodzki). [1979] [Reprinted: Tübingen: Narr.] 1976 "On some linguistic limitations of classical contrastive analyses", PSiCL 4: 8 9 - 9 5 [reprinted in: Jacek Fisiak (ed.) 1980, 1 9 3 - 1 9 9 ] . 1980 Gramatyka angielska dla Polakow [English grammar for Poles] (Warszawa: Panstwowe Wydawnictwo Naukowe). 1981a "Tlumaczenie j a k o czynnosc pragmatyczna" [Translation as a pragmatic activity], in: Materialy ζ IV Sympozjum Instytutu Lingwistyki Stosowanej Uniwersytetu Warszawskiego, Jachranka 3 — 5 listopad 1976 (Warszawa: Wydawnictwa Uniwersytetu Warszawskiego). 1981b "The problem of equivalence revisited", IRAL 19: 1 1 3 - 1 2 8 . 1981c "What do we need lexical contrastive studies for?", PSiCL 3: 1 3 3 - 1 4 8 . 198Id "Quantitative contrastive analysis", Studia Linguistica 35: 43 — 51. 1984 "Tertium comparationis", in: J. Fisiak (ed.), 301 —312. 1985 "The so-called 'sign theory' as the first method in contrastive linguistics", in: Ursula Pieper —Gerhard Stickel (eds.), Studia linguistica diachronica et synchronica (Berlin —New York —Amsterdam: Mouton de Gruyter), 485 — 501. 1986a "An Elizabethan contrastive grammar of Spanish and French", in: D. K a s t o v s k y - A . Szwedek (eds.), 1 3 0 3 - 1 3 1 0 . 1986b "Prototypes and equivalence", PSiCL 21: 5 - 2 0 . 1989 "Towards a typology of contrastive studies", in: Wieslaw Oleksy (ed.), 55 — 72. (in prep.) "Early contrastive studies in England" Kudzinowski, Czeslaw 1978 Gramatyka jgzyka finskiego [Grammar of the Finnish language] (Poznan: Uniwersytet im. A. Mickiewicza). Kufner, Herbert L. 1962 The grammatical structures of English and German (Chicago: University of Chicago Press). Kühlwein, Wolfgang 1983 "Sociosemiotics across cultures", in: Proceedings of a symposium held at St. Patrick's College, Drumcondra, Dublin, 8 — 9 July 1983: 5 — 19. Kurylowicz, Jerzy 1971 "Podstawowe kategorie morfologiczne" [Fundamental morphological categories], Biuletyn PTJ 28: 3 - 1 2 .

References

259

Lado, Robert 1957 Linguistics across cultures (Ann Arbor: University of Michigan Press). 1964 Language teaching: a scientific approach (New York: McGraw-Hill). Lakoff, George 1968 "Instrumental adverbs and the concept of deep structure", Foundations of language 4.1: 4 — 29. 1970 "Global rules", Language 46: 6 2 7 - 6 3 9 . 1971 "On generative semantics", in: D. Steinberg —L. Jacobovits (eds.). Semantics (Cambridge: Cambridge University Press), 232 — 296. 1973 "Fuzzy grammar and the performance/competence terminology game" Papers from the ninth regional meeting of the Chicago Linguistic Society (Chicago: University of Chicago Press), 271—291. 1977 "Linguistic gestalts", Papers from the thirteenth regional meeting of the Chicago Linguistic Society (Chicago: University of Chicago Press), 236 — 287. 1982 Categories and cognitive models. L.A.U.T. Series A. no. 96. 1987 Women, fire, and dangerous things: what categories reveal about the mind (Chicago: University of Chicago Press). Lakoff, George —Mark Johnson 1980 Metaphors we live by (Chicago —London: University of Chicago Press). Lakoff, Robin 1972 "Language in context" Language 48: 907 — 927. Lamminmäki, Reijo 1979 "The discrimination and identification of English vowels, consonants, junctures and sentence stress by Finnish comprehensive school pupils" Jyväskylä Cross-Language Studies 7: 165 — 231. Langacker, Ronald W. 1987 Foundations of cognitive grammar 1 (Stanford: Stanford University Press). Langendoen, Terence D. 1970 Essentials of English grammar (New York: Holt, Rinehart and Winston). Lawendowski, Boguslaw 1974 A hypothesis concerning emotive aspects of lexis in English and Polish [Ms., Uniwersytet Warszawski]. Lee, William R. 1968a " H o w can contrastive linguistic studies help foreign language teaching?", YSC EC Ρ Β. Studies 5 7 - 6 6 . 1968b "Thoughts on contrastive linguistics in the context of language teaching" MSLL 21 (Washington D.C.: Georgetown University Press), 1 8 6 - 1 9 4 . Leech, Geoffrey 1974 Semantics (Harmondsworth: Penguin). 1983 Principles of pragmatics (London —New York: Longman). Leech, Geoffrey —Jan Svartvik 1975 A communicative grammar of English (London: Longman). Lees, Robert B. 1963 The grammar of English nominalizations (The Hague: Mouton). Lewandowska, Barbara 1975 "On some properties of action nominals in Polish and their English equivalents", PSiCL 3: 1 6 7 - 1 7 6 . Lindner, Susan 1981 A lexico-semantic analysis of verb-particle constructions with up and out (University of California, San Diego Ph.D. dissertation) (L.A.U.T. Series A, no. 101).

260 1982

References

"What does go up doesn't necessarily come down: the ins and outs of opposites", in: Proceedings of the eighteenth regional meeting of the Chicago Linguistic Society (Chicago: University of Chicago Press), 305 — 323. Lipmska, Maria 1972 Basic sentence patterns in English and Polish [unpublished ms., Uniwersytet im. A. Mickiewicza, Poznan]. 1973 "Some differences between English and Polish on the level of basic sentence patterns", PSiCL 1: 3 9 - 4 6 . 1975 "Contrastive analysis and the modern theory of language", PSiCL 3: 5 — 62. Lipinska-Grzegorek, Maria 1977 Some problems of contrastive analysis: sentences with nouns and verbs of sensual perception in English and Polish (Edmonton, Canada —Champaign, USA: Linguistic Research). Liston, Jerry L. 1970 "Formal and semantic considerations in contrastive analysis", Y SC EC Ρ Β. Studies 2: 2 7 - 4 9 . Ludkiewicz, Anna 1985 American and Polish address system [unpublished M.A. thesis, Uniwersytet Gdanski]. Ludskanov, Alexander 1972 Mensch und Maschine als Übersetzer (Halle (Saale): Niemeyer). Lyons, John 1968 Introduction to theoretical linguistics (Cambridge: Cambridge University Press). Mackiewicz-Krassowska, Halina 1974 "Intonation of English and Polish declarative sentences", PSiCL 2: 137 — 144. Mansion, J. E. 1928 French reference grammar (London — Edinburgh). Manczak, Witold 1970 Ζ zagadnien jgzykoznawstwa ogolnego [Problems in general linguistics] (Wroclaw, Warszawa, Krakow: Ossolineum). Marek, Boguslaw 1974 "Intonation of imperative sentences and requests in Polish and English", PSiCL 2: 1 6 1 - 1 7 9 . Markkanen, Raija 1985 "Directives in English and Finnish", Jyväskylä Cross-Language Studies 11: 15-42. Marton, Waldemar 1968a Noun modification in Polish and English [unpublished doctoral dissertation, Uniwersytet im. A. Mickiewicza, Poznan], 1968b "Equivalence and congruence in transformational contrastive studies", SAP 1: 5 3 - 6 2 . 1970 "English and Polish nominal compounds: a transformational contrastive study", SAP 2: 5 9 - 7 3 . 1976 Νowe horyzonty nauczania jgzykow obcych 2 [New horizons in teaching foreign languages] (Warszawa: Wydawnictwa Szkolne i Pedagogiczne). Matisoff, James A. 1979 Blessings, curses, hopes and fears: psycho-ostensive expressions in Yiddish (Philadelphia: Institute for the Study of Human Issues). McCawley, James D. 1968 "The role of semantics in a grammar", in: E. Bach —R. T. Harms (eds.), 124-169.

References

261

Meech, S. B. 1935 "Early applications of Latin grammar to English" PMLA 1: 1012 — 1032. Mel'chuk, Igor A. 1963 "Machine translation and linguistics", in: O. S. Akhmanova et al. (eds.), Exact methods in linguistic research (Berkeley: University of California Press), 4 4 - 7 9 . Miege, Guy 1687 The grounds of the French tongue or a new French grammar according to the present use, and modern orthography (London). Mieszek, Aleksandra 1974 "Intonation of compound sentences in English and Polish", PSiCL 2: 181 — 188. Milewski, Tadensz 1965 Jgzykoznawstwo [Linguistics] (Warszawa: Panstwowe Wydawnictwo Naukowe). Montague, Richard 1974 Formal philosophy: selected papers (edited by R. H. Thompson) (New Haven: Yale University Press). Morgan, Bayard Q. —F. W. Strothmann 1952 Shorter German reading grammar (New York —Boston: Ginn). Morris, Charles W. 1938 Foundations of the theory of signs (Chicago: University of Chicago Press). Moulton, William G. 1962 The sounds of English and German (Chicago: University of Chicago Press). Nemser, William 1971 "Approximative systems of foreign language learners" IRAL9: 115 — 123. [Reprinted 1974 in: J. C. Richards (ed.), Error Analysis. London: Longman, 55-63], Newmeyer, Frederick J. 1980 Linguistic theory in America (New York — London — Toronto — Sydney — San Francisco: Academic Press). Nickel, Gerhard 1971 "Variables in a hierarchy of difficulty" The Pacific Conference on Contrastive Linguistics and Language Universals Papers (Honolulu: University of Hawaii), 1 8 5 - 1 9 4 . Nickel, Gerhard —Karl Heinz Wagner 1968 "Contrastive linguistics and language teaching", IRAL 6: 233 — 255. Nilsen, Don Lee Fred 1972 Toward a semantic specification of deep case (The Hague: Mouton). Nowakowski, Miroslaw 1977 "The lexicon and contrastive language studies" PSiCL 6: 25—42. Oleksy, Wieslaw 1977 "Tags in English and equivalent constructions in Polish", PSiCL 7: 9 5 109. 1983 "Pragmatic equivalence in contrastive studies: requests in Polish and English", Jyväskylä Cross-Language Studies 10: 79 — 93. Oleksy, Wieslaw (ed.) 1989 Contrastive pragmatics (Amsterdam/Philadelphia: Benjamins). Olsson, M. 1972 Intelligibilty: a study of errors and their importance (Gothenburg School of Education Research Bulletin 12, Göteborg). Ozga, Janina 1974 "Stress in English and Polish — an introduction to a contrastive analysis", PSiCL 2: 1 2 3 - 1 3 5 .

262

References

1976

"The relevance of the notion 'basis of articulation' to contrastive phonetics", PSiCL4: 6 1 - 7 4 . Pasanen, Maija-Liisa 1978 "Finnish and English verbs of vision", Jyväskylä Contrastive Studies 6: 96 — 122.

Peters, Stanley (ed.) 1972 Goals of linguistic theory (Englewood Cliffs, N.J.: Prentice-Hall). Pickbourne, James 1789 A dissertation on the English verb (London). Pike, Kenneth L. 1947 Phonemics: a technique for reducing language to writing (Ann Arbor: University of Michigan Press). Polanski, Kazimierz 1966 "Glowne typy struktur zdaniowych w j?zyku polskim" [Main types of sentences structures in the Polish language], Zeszyty Naukowe Wyzszej Szkoly Pedagogicznej w Katowicach 31 (Prace j?zykoznawcze 11), 83 — 99. 1971 "Gramatyki generatywne a metoda transformacyjna" [Generative grammars and the transformational method], in: A. M. Lewicki (ed.), Problemy skladni polskiej [Problems of Polish syntax] (Warszawa: Panstwowe Wydawnictwo Naukowe). Politzer, Robert L. 1968 "An experiment in the presentation of parallel and contrasting structures", Language Learning 18: 35—43. Postal, Paul 1968 Aspects of phonological theory (New York: Harper and Row). 1970 "On the surface verb 'remind'", Linguistic Inquiry 1: 37 — 120. Preston, Dennis R. 1975 "Contrastive analysis — the outlook from modern grammar", PSiCL 3: 63-72. 1979 "Introduction", in: D. R. Preston —R. W. Shuy (eds.), Varieties of American English. A reader (Washington D.C.: International Communication Agency), 1—13. Puppel, Stanislaw—Jadwiga Nawrocka-Fisiak — Halina Krassowska 1977 A handbook of Polish pronunciation (Warszawa: Panstwowe Wydawnictwo Naukowe). Puzynina, Janina 1969 Nazwy czynnosci we wspolczesnym jgzyku polskim [Names of activities in contemporary Polish] (Warszawa: Panstwowe Wydawnictwo Naukowe). Quirk, Randolph 1962 The use of English (London: Longmans). Ravem, Roar 1974 "The development of w/i-questions in first and second language learners", in: J. C. Richards (ed.), Error analysis (London: Longman], 134 — 155. Riley, Philip 1981 "Towards a contrastive pragmalinguistics", in: J. Fisiak (ed.), 121 — 146. Robinson, Jane J. 1970 "Case, category, and configuration", Journal of Linguistics 6.1: 57 — 80. Rosch, Eleanor 1973 "Natural categories", Cognitive psychology 4: 328 — 350. Rosch, Eleanor —Β. B. Lloyd (eds.) 1978 Cognition and categorization (Hillsdale, N.J.: Lawrence Erlbaum). Ross, John Robert 1973 "Nouniness", in: Osamu Fujimura, Three dimensions of linguistic theory (Tokyo: TEC Corporation), 137-258.

References 1975

"Clausematiness", in: Edward Keenan (ed.), Formal semantics language (London: Cambridge University Press), 422 — 475. "Nominal decay", Department of Linguistics, M.I.T.

263 of natural

1981 Rubach, Jerzy 1983 "Applied contrastive linguistics: in search of a framework", Jyväskylä CrossLanguage Studies 9: 141 — 156. Rülker, K. 1973 "Zur pragmatischen Invarianz bei der Translation" in: A. Neubert — O. Kade (eds.), Neue Beiträge zu Grundfragen der Übersetzungswissenschaft (Leipzig: Athenäum), 29 — 35. Rusiecki, Jan 1976 "The development of contrastive linguistics", ISB 1: 12 — 44. Salmon, Nicolas 1797 Grammaire angloise comparee avec la grammaire franfoise (London: W and S Spilsbury). Sampson, Geoffrey 1970a Towards a linguistic theory of reference [unpublished Ph.D. dissertation, Yale University], 1970b Natural language and the paradox of the liar [ms.]. Sanders, Carol 1981 "Recent developments in contrastive analysis and their relevance to language teaching", in: J. Fisiak (ed.), 21 —32. Sapir, Edward 1921 Language: an introduction to the study of speech (New York: Harcourt, Brace and World) [reprinted in 1949], Saumjan, Sebastjan 1968 Osnovanija porozdajuscej grammatiki russkogo jazyka [Foundations of a generative grammar of the Russian language] (Moskva: Nauka). Schlyter, Suzanne 1980 Vagheit, Polysemie und Prototypentheorie [Ms. Universität Konstanz]. Schmitz, John Robert 1983 "Color words in English and Portuguese: a contrastive semantic analysis" PSiCL 17: 3 7 - 4 9 . Schwarze, Christoph 1978 "Sprachtheoretische Grundlagen der Sprachvergleichsgrammatik", in: C. Schwarze (ed.), Kasusgrammatik und Sprachvergleich (Tübingen: Narr), 93-112. 1979 "Lessicologia contrastiva e apprendimento lessicale", in: C. Schwarze (ed.). Italienische Sprachwissenschaft (Tübingen: Narr), 227 — 243. Searle, John R. 1969 Speech acts: an essay in the philosophy of language (Cambridge: Cambridge University Press). Searle, John R. (ed.) 1971 The philosophy of language (Oxford: Oxford University Press). Sehnert, James 1975 A contrastive gastronomic terminology in English and Polish [Ms. Uniwersytet im. A. Mickiewicza, Poznan], Sehnert, James —Michael Sharwood-Smith 1974 "The verbalization of instrumentals in English", SAP 5: 37—44. Selinker, Larry 1972 "Interlanguage", IRAL 10: 2 1 9 - 2 4 3 . [Reprinted 1974 in: J. C. Richards (ed.), Error Analysis. London: Longman],

264

References

Sharwood-Smith, Michael 1975 "Aspects of future reference in English and Polish", PSiCL 3: 9 0 - 9 9 . 1976 "Pedagogical grammars", IS Β 1: 45 - 57. Siewierska, Anna 1984 The passive. A comparative linguistic analysis (London —Sydney —Dover — New Hampshire: Croom Helm). Skorupka, Stanislaw 1967 Slownik frazeologiczny jgzyka polskiego [Phraseological dictionary of the Polish language] (Warszawa: Wiedza Powszechna). Snook, Roger L. 1971 "A stratificational approach to contrastive analysis", in: G. Nickel (ed.), Contrastive linguistics, 17 — 36. Spalatin, Leonardo 1969 "Approach to contrastive analysis", YSCECP B. Studies 1: 2 6 - 3 5 . Stanislawski, Jan 1950—1951 Gramatyka angielska dla zaawansowanych. Szczegölowa analiza rozbieznosci zachodzgcych migdzy gramatykg angielskg a polskg [English grammar for advanced learners. A detailed analysis of divergencies between English and Polish grammars], 1 —2. (Warszawa: Ksi^zka i Wiedza). Stieber, Zdzislaw 1966 Historyczna i wspolczesna fonologia j§zyka polskiego [Historical and contemporary Polish phonology] (Warszawa: Panstwowe Wydawnictwo Naukowe). Stock well, Robert P.—J. Donald Bo wen 1965 The sounds of English and Spanish (Chicago —London: University of Chicago Press). Stockwell, Robert P . - J . Donald B o w e n - J o h n W. Martin 1965 The grammatical structures of English and Spanish (Chicago — London: University of Chicago Press). Stockwell, Robert P.— Paul Schachter — Barbara Hall Partee 1973 The major syntactic structures of English (New York: Holt, Rinehart and Winston). Suomi, Kari 1980 Voicing in English and Finnish stops (Turku: Publications of the Department of Finnish and General Linguistics of the University of Turku). Sussex, Roland 1976 "The measurement of contrast in contrastive linguistics", PSiCL 5: 5 — 17. Szwedek, Aleksander 1981 Word order, sentence stress and reference in English and Polish (Edmonton, Alberta: Linguistic Research) [first published Bydgoszcz: Wyzsza Szkola Pedagogiczna, 1975]. Tomaszczyk, Jerzy 1976 "On establishing equivalence between lexical items of two languages", PSiCL 5: 7 7 - 8 1 . 1979 "Dictionaries: users and uses", Glottodidactica 12: 103 — 119. Trager, George L. — Henry Lee Smith 1957 An outline of English structure (Washington D.C.: American Council of Learned Societies). Tran-Thi-Chau 1975 "Error analysis, contrastive analysis, and students' perception: a study of difficulty in second-language learning", IRAL 13: 119 — 143. Trier, Jost 1931 Der deutsche Wortschatz im Sinnbezirk des Verstandes. Die Geschichte eines sprachlichen Feldes (Heidelberg: Winter).

References

265

Twaddell, W. Freeman 1963 "On defining the phoneme", in: Martin Joos (ed.), Readings in linguistics 3 (New York: American Council of Learned Societies), 55 — 80. Valdman, Albert —John S. Phillips 1975 "Pidginization, creolization and elaboration of learner systems", Studies in Second Language Acquisition 1: 21—40. Van Buren, Paul 1971 "Some proposals concerning contrastive analysis" [Paper read at the 2nd Conference of Polish-English contrastive studies at Karpacz, December 16-18], 1974 "Contrastive analysis", in: J. P. B. Allen —S. P. Corder (eds.), Techniques in applied linguistics (Edinburgh Course in Applied Linguistics 3) (Oxford: Oxford University Press), 2 7 9 - 3 1 2 . 1976 Review of Tomasz P. Krzeszowski, Contrastive generative grammar- theoretical foundations, ISB 1: 2 5 0 - 3 2 9 . Van Dijk, Teun 1972 Some aspects of text grammars: a study in theoretical linguistics and poetics (The Hague —Paris: Mouton). Varga, Laszlo 1975 A contrastive analysis of English and Hungarian sentence prosody (Budapest: Linguistic Institute of the Hungarian Academy of Sciences; Washington, D.C.: Center for Applied Linguistics). Welna, Jerzy 1977 "Deceptive words. A study in contrastive lexicon of Polish and English", PSiCL 7: 7 3 - 8 4 . Widdowson, Henry 1975 "The significance of simplification", Studies in Second Language Acquisition 1: 1 1 - 2 0 . Wierzbicka, Anna 1980 Lingua mentalis. The semantics of natural language (Sydney —New Y o r k London—Toronto—San Francisco: Academic Press). Wilkins, John 1668 An essay towards a real character and a philosophical language (London: Printed for SA: Gellibrand and for John Martyn. Printer to the Royal Society). Wittgenstein, Ludwig 1953 Philosophical investigations, translated by G. Ε. M. Anscombe (Oxford: Blackwell and Mott). Woloszyk, Alicja 1974 "Intonation of interrogative sentences in English and Polish" PSiCL 2: 145-159. Wojcik, Tomasz 1975 Gramatyka jgzyka rosyjskiego. Studium kontrastywne [Russian grammar: A contrastive study] (Warszawa: Panstwowe Wydawnictwo Naukowe). Yngve, V. H. 1975 " H u m a n linguistics in face-to-face interaction", in: A. Kendon et al. (eds.), Organisation of behaviour in face-to-face interaction (The Hague: Mouton), 47-62. Zabrocki, Tadeusz 1976 "On the so-called 'theoretical contrastive studies'", PSiCL 4: 9 7 - 1 0 9 .

Index of subjects Figures in italics indicate detailed treatment of a topic.

Abstract domain 215 Aceusativus cum infinitive 39, 205 Acronymy 73 Action 109, 185, 197 completed and incompleted, in Polish 152-159 momentary 110 Active voice 111 sentences 246 Actors 142 Address 94, 246 Addressee 1 8 0 - 1 8 1 Adjective 71, 75, 79, 84, 87, 109, 115, 130, 169, 173 derived from nouns 136 , prototypical 216 Adverb 71, 84, 109, 115, 143, 173 Adverbial phrase 66 Affixation 18, 73 Affricates 123, 124, 125, 247 in Polish 1 2 3 - 1 2 5 see also Consonants Age see Sociolinguistic factors Agent 110, 143, 144, 163, 164, 169, 178-187 in the original subconfiguration 178 unspecified 163, 164 Agentive case 142 Agreement 197 subject-verb 249 see also Concord Aliteration 18 Allophones 55, 56, 120, 126, 241 Ambiguity 173 — 174 American English 76, 98, 114 American structuralism see Structuralism Analogy 49, 195 Anthropology 213, 214

Antonyms 81—82 Antonymity 79 Approximative systems 192 — 193, 196, 236, 237 fossilized 235 Archaism 246 Arguments as noun phrases 152 Articles 16, 37, 135 comparison of 44 —45 erroneous a in front of present participle 194 in congruent strings 135 in reduced registers 196 in Spanish and English 133 — 134 Articulation 111 of vowels 112 Articulatory basis 127 Artificial intelligence 214 Artificial languages see Language Aspect 140, 141, 145, 152, 165 coalescence of simple and perfect 194 completed vs. incompleted in Polish 152-155 continuous 183 durative 1 5 4 - 1 5 5 iterative 155, 248 perfective (perfect) 1 5 4 - 1 5 5 , 183, 184 Aspiration 32, 53, 125 Association in pragmatic contrastive studies 101 -103 Auxiliaries in congruent strings 135 Auxiliary Verbs see Verbs Baby-talk 196 Back-formation 73 Background framing 231 Base 137, 140, 142, 149

268

Index of subjects

semantic as tertium comparationis 152, 176-177 universal rules of 141 Basic sentence patterns 63, 128, 130 in Spanish and English 131 — 132 Basque 39, 246 Bilingual competence 161, 168, 202 informant 104, 148, 230-231 Black English 190 Borrowings 75, 90 Bracketed strings 180, 181, 182 Branching trees 180-181 Bulgarian 239 Case 109, 118, 140, 247 accusative in Polish 224, 225, 229 in case grammar 142 — 144 in Polish and English 135 instrumental in Polish 224 Case grammar as a framework for contrastive studies 141-146 see also Grammar Categories 241 classical 214 complex vs. simple 214 discrete 214 natural 214-216 Categorial component 169, 170 Chinese 9, 38, 109, 128, 173 Classical categorization 214 Clauses 137 as complements, as objects, as objects of prepositions, as subjects, nominal, relative 64 concessive, relative, resultative 145 dependent and independent 137 main 204 subordinate 204, 205, 245 Clusters of consonants 57, 58, 120, 247 of pragmatic, semantic and syntactic properties 217 Coda 58 Cognate object see Object Cognitive

grammar see Grammar Cognitive linguistics see Linguistics Cognitive sciences 213, 214 Collocations 87, 102 —, more and less conventional 102 Colour terms 69, 75, 82 Commands 186, 187 Comment 154 Communicative acts 92, 93, 97, 217 Communicative competence 91 Comparison 83, 131, 203, 245 accuracy of 242 of constructions 28, 37, 38-39, 133, 245 of languages 245 of phonemic inventories 112 of rules 28, 37, 39-40, 133 of speech sounds 11, 122 — 127 of systems 28, 37, 41, 42-44, 133, 245 Comparison proper 11, 35, 37—45, 124, 132, 141 Competence 128, 147, 148, 190, 225 bilingual 202 frozen 235 grammatical 91 transitional 191, 192, 235 vs. performance 128, 249 Complement 63, 110 object 205 Complementary distribution of plural endings 119 Complication 196, 197 routes 196, 202 see also Elaboration Concepts basic, subordinate and superordinate 215 Concord 197 subject-verb 192 see also Agreement Condensation of subordinate clauses 205, 245 Configuration 177, 183, 187, 198 as input to derivation (conversion rules) 177 - , initial 179, 181

Index of subjects nuclear 179 original 178 Congruence 136, 217, 218, 247 Congruent strings definition 135 — 136 Conjunction 109 Connotation 77, 81, 82, 87 Consonants 32, 37, 55, 56, 122, 1 2 3 - 1 2 5 , 126 affricates 123 fricatives 52, 53, 54, 1 2 2 - 1 2 3 laterals 1 2 2 - 1 2 3 nasals 122—123 sibilants 49, 1 2 2 - 1 2 3 stops 49, 53, 122, 125 Constructions 28, 147, 241 avoidance of identical 240 comparisons of 28, 35, 218, 245 congruent 27, 204, 240 coordinative 216 differences in frequencies across languages 206 equivalent 147, 203, 204, 205, 218, 219, 249 impersonal vs. passive 250 infinitive as imperative in Polish 249 interrogative 37 negative 37 ambiguity of in English and Polish 151 passive 246, 250, see also Passive relative 37, 216 semantic and communicative values of 238 Contact 247 Contrast 2, 243, 247 categorial 65 explicated by Contrastive Generative Grammar 170 functional 66, 67 impossible 243 phonemic 1 2 0 - 1 2 7 pragmatic 101 socio-cultural 101, 103 structural 66 see also Difference Contrastive analysis 11, 35, 45, 234, 235, 236, 240, 241, 242, 243, 246

269

see also Contrastive studies Contrastive Generative Grammar 3, 12, 31, 37, 169-187, 196, 197, 200, 201, 202, 217, 218, 248, 249 five postulates 171 Contrastive grammar see Grammar Contrastive linguistics 1, 9—13, 68 and descriptive linguistics 10 "theoretical" vs. "applied" 10 vs. contrastive analysis and contrastive grammar 11 see also Contrastive analysis and Contrastive studies Contrastive studies 48, 140, 242, 245, 247, 250 and cognitive linguistics 213 — 231 and linguistic models 107—145 and prediction of errors 189—191 autonomous 24 classical (three steps) 35 — 45 classification of 23—33 corpus restricted 26, 34 directional 23, 33, 40, 41 generalized 24, 108 generative transformational 107, 108, 127- 146 grammatical 4, 140 history of 1 —3 lexical 6, 24, 2 9 - 3 0 , 69-90, 245 of intonation 246 of speech sounds 120—127 operational 24 paradigmatic 29 phonological 21, 2 9 - 3 0 , 50-63, 245, 247, 249 pragmatic 32, 33, 90-105 projective (systematic) 25, 28, 34, 67, 203, 204, 205, 211, 249 qualitative 26, 34 quantitative 26, 34, 203-212, 249 structural 107, 108, 113-127, 130, 131, 132 subphonemic 11, 122, 125 syntactic 16, 18, 20, 31, 63-68, 69, 101, 140, 146, 147 taxonomic 24, 35, 147 - , text-bound 25-26, 31, 34, 162, 203

270

Index of subjects

"theoretical" vs. "applied" 23, 33, 41, 48, 5 0 - 5 1 , 233-243 traditional 107, 108-113 see also Contrastive analysis and contrastive linguistics Conventionality scale 103 Conversational greasers 98 Co-occurrence 71, 78 restrictions 148-151, 161, 163 Cooperative principle 91 Coreference 45 Creole 196 Czech 2 0 4 - 2 0 5 , 245 Dative case 111, 142, 143 Deceptive words see False friends Decomposition 86, 87, 89 Deep structure 136, 141, 168, 196 as input to transformations 137 as tertium comparationis 161, 245 identical deep structure hypothesis 152-168 in case grammar 142 —146 in the standard theory 148—151, 152, 161, 168 Degree 109 Definiteness 44 — 45 Deictic words in reduced registers 196 Denial 185 Denotation 80, 81, 82 Derivational contrast as a structural signal 115 history 172, 218 Description (as a step in contrastive studies) 11, 3 5 - 3 6 Descriptive adequacy 242 Determiner 63, 117, 132, 133, 134 in Spanish and English 133—134 Devoicing 125 Dictionary 69-70, 80, 83, 90, 182, 215, 247 bilingual 79, 8 8 - 9 0 , 155 context in 8 9 - 9 0 English-Polish 225 in Contrastive Generative Grammar 182 —, monolingual 155

Difference 64, 147, 189, 237, 239, 247 categorial 64, 65 — 66 degree (scale) of 218, 239, 240 functional 64, 65 — 67 in form 118, 120 structural 64 — 65 see also Contrast Difficulty 172, 239 hierarchy of 172, 189, 239 learning 238 pattern of 246 scale of 246 Diphthongs 37, 57, 120, 126 Disambiguating context 174 Discourse 100 analysis 45 definition 93 Distinctive features 121, 122, 125, 214 Distinctive stress 73 Distribution 120 limited 205 Distributionalism see Structuralism Diversification 169, 170, 218 Doubt 185 Drills in transfer of training 193 — 194 Dutch 239 Educational terms see Semantic fields Elaboration 196, 197 Ellipsis in answers to questions 167 Embedding 69 Emotions pleasant and unpleasant 82 English 1, 2, 9, 36, 38, 39, 40, 41, 42, 43, 44, 45, 4 7 - 1 0 5 passim, 108, 111, 112, 116, 117, 118, 119, 128, 1 2 9 - 1 3 4 , 1 3 7 - 1 3 9 , 141, 143, 1 4 9 - 1 5 1 , 1 6 9 - 1 8 7 passim, 189, 194, 1 9 7 - 2 0 3 , 2 0 5 - 2 1 2 , 217, 220, 2 2 2 - 2 2 5 , 227, 228, 230, 241, 242, 245, 246, 248, 249 modern standard southern 122 phonemes 120—127 standard 192 — word order 63 — 64

Index of subjects Entailment 85 Equivalence 5, 16, 21, 24 — 33, 67, 123, 126, 133, 136, 167, 191, 204, 217, 218, 219 contraints on 156, 162, 165, 168 construction 29, 34, 38-39, 149-151 contextual 17 derivational-semantic 25 empirical reality of 162 formal 25, 100 functional-communicative 25 pragmatic 18, 30, 32, 34, 99, 100, 1 0 4 - 1 0 5 , 218, 220, 221, 245, 250 prototype theory of 222, 230 rule 34, 39-40, 129 semantic 16, 17, 21, 121, 131, 217 semanto-syntactic 19, 27, 28, 29, 30, 34, 6 7 - 6 8 , 104, 147-168, 169, 170, 177, 218, 245, 248, 250 substantive 32, 34 statistical 26 - 28, 34, 203 system 29, 34, 38, 211 textual 147 translational 6, 17, 18, 28, 34, 67, 131, 147, 245 Equivalents 71, 148, 165, 176, 202, 203, 228, 230, 241 congruent 204 phonemic 54, 57 lexial 74, 76, 79, 80, 81, 8 5 - 9 0 , 155, 211, 223 matching of in traditional contrastive studies 110 pragmatic 98, 9 9 - 1 0 0 , 218, 220, 250 prototypical 222, 225 — 230 semanto-syntactic 162, 165, 168, 205 situation 99 socio-cultural 101 statistical 165 structural description of 171 textual 89 translation 18, 89, 100, 160, 165, 166, 203 Equivalent sentences see Sentences Error analysis 189, 190, 201, 202 annihilation of 236 — vs. contrastive studies 190 — 191

271

Errors 8, 187, 190, 191, 201, 202, 240 frequencies of 190 of abundance 43, 204, 205, 206, 212 of avoidance 204, 206, 208, 211, 212, 240 overgeneralization as a source of 191 performance vs. competence 191 potential 202 prediction of 189, 190, 191, 206, 236, 243 translation 28 Estonian 239 Ethnography 213 Event 142, 185, 249 Exercises 193 in transfer of training 193 — 194 remedial 190 translational 2 Explicitness 147

Facilitation 189 Factitive case 142 False friends 76 Family resemblance 87, 88, 214, 216, 223, 227 Features 245 articulatory 111 contextual 84, 186 decomposition into 86 distinctive 121, 122, 125, 214 inherent 84 phonological 50, 214 relational 84, 85 relevant and redundant 120, 121 — 127 selection 7 semantic 84, 85, 8 6 - 8 8 syntactic 84, 85 Fijan 38 Finnish 77, 1 5 9 - 1 6 5 , 194, 239 passive sentences in 159 — 165 Fixed expressions 71 Flap 126 Focus 69, 154, 187 Foreign language 239, 240 acquisition 236 learning 8, 236, 237, 243 - teaching 2, 13, 233, 242

272

Index of subjects

Foreign learners 187, 189, 190, 193, 235, 238, 239, 249 Foreigner talk 196 Formal correspondence 16, 17, 148 Formality 93 Formulaic expressions 98 — 99, 101, 103, 104-105 Fossilization 193 Free variants 121 French 38, 39, 40, 76, 9 4 - 9 6 , 110-111, 112, 164, 189, 2 0 4 - 2 0 5 vowels compared with English 112 Fricatives 122, 124, 247 in English 122 in Polish 1 2 2 - 1 2 3 Friction 54 Fronting 125 Function words 9, 18, 69, 109, 115, 194, 197, 246 as a structural signal 115 auxiliaries, coordinators, interrogators, prepositions, qualifiers 115 in congruent strings 135 in reduced registers 196 noun determiners 115, 116 omission of 194, 201 Future 249 Fuzzy grammar see Grammar Fuzzy phenomena 109 boundaries of categories 231 boundaries of phonemes 216 Gender 43, 133, 140 in Polish and English 135 in Spanish and English 133 virile vs. non-virile 42 Generative grammar see Grammar Generative phonology see phonology Generative semantics 47, 141, 213 Generative transformational models 107, 147 Georgian 247 German 38, 39, 65, 76, 111, 116, 117, 118-119, 143, 164, 217, 239

"Perfect" tenses in comparison with English 111 transfer from 195 Gerund 205, 245 regular 207, 208, 209 Gestalts 215, 221, 228, 230, 250 Glides 247 Global rules see Rules Glottalization 125 Government 197 Gradience 216 Grammar 48, 70, 83, 84, 194, 202, 216 as a diversifying device 169 case 141 cognitive 47 comparative 247 contrastive 3, 26, 146 conversion 177 core 217 descriptive 108 foreign language 107 fuzzy 213 generative 3, 127, 171, 201, 214, 217, 235 generative format of contrastive studies 235 relational 35, 107 stratificational 107 text 45, 128, 174 vs. sentence grammar 173 traditional 107 transformational generative 35, 47, 71, 127-146, 213, 217, 247 universal 24, 68, 108, 109 Grammatical rules see Rules Graphs 177 Greek 38, 39 scholars 108 Hearer 94, 128 Historical linguistics 9 Homonymy 79, 80 Hopi 109 Hungarian 194, 212 Hyperonymy 84 Hyponymy 84, 85

Index of subjects Icelandic 240 Idealized Cognitive Models 215, 248 Idioms 73 Illocutionary acts 93 force 97 structure 93 — 96 Image schema 222, 228, 229 Immediate constituents 144 Imperatives 249 see also Sentences Indefiniteness 44 Infinitive 205, 249 —, marked in English 205 Inflection 9, 18, 69, 73, 154, 197, 246 as a noun marker 109 as a source of errors 8 as a structural signal 115 plural in German and English 118-119 Input structure 169, 171, 172, 187, 248 see also Semantic structure Instrumental case 142, 143 inflection in Polish 149 — 151 Interference 498, 195 Interjections 109 Interlanguage 189, 193, 1 9 6 - 1 9 7 , 2 0 0 - 2 0 2 , 206, 212, 2 3 4 - 2 3 7 Polish-English 211 Interlude 58 Intonation 154, 246 cognitive function 63 rising in occasional questions 249 Italian 40, 76, 143, 170 Japanese 39, 98 Juxtaposition 11, 35, 36-37, 60, 89, 116-117, 119-120, 122-124, 1 3 2 - 1 3 3 , 141, 155 of basic sentence patterns 131 — 132 of phonemes 122—123 of lexical items 89, 117, 155 Kernel sentences see Sentences Keys 82 Kinship terms 69, 75, 82

273

Knowledge encyclopedic vs. linguistic 85, 90 Korean 165 — 166 Labialization 53 Landmark 222, 228, 229 Language 3, 217, 235, 243 artificial 128 as a hierarchical structure 15, 31 aspects of in comparing 245 behaviorist model of 237 categories of variation 246 communicative function of 48 etalon 241 genetically close and remote 239 modular view of 91 native 191, 240 second, learning, hierarchies of difficulties 172 source 107, 189, 190, 201, 202, 236 patterns 195 transfer from 195 target 90, 107, 187, 189, 190, 191, 193, 196, 201, 202, 204, 206, 212, 235, 236 deviations from the norms of 201-202, 242 strategy of learning in forming interlanguage 194 Langue 25, 127 Laterals 37 in English 122 in Polish 123, 124, 125 Latin 1, 2, 39, 40 Lexical congruity 78 — 79 entry 70, 84, 90, 155, 173 insertion 140, 170, 197, 198, 200, 202 item 69, 71, 72, 75, 77, 79-90, 148, 157, 182, 197, 198, 201, 202, 246 equivalence of 85 — 90, 155, 158-159 representation 148 Lexicalist hypothesis 247 Lexicology 15, 70, 71, 83, 84 Lexicon 47, 48, 91, 112, 137, 140, 154, 155. 246 Linguistic models 107 —, choice for contrastive analysis 241, 242

274

Index of

subjects

generative transformational 107, 147 relational 107 stratificational 107 structural 107 Linguistic universals 9, 68, 140, 236, 239, 240 Linguistics 91, 214, 234 applied 234 broader and narrower sense 91 "code" vs. " h u m a n " 91 cognitive 213 — 231 theoretical 234 typological 9 - 1 0 , 68 vs. contrastive analysis 241 Listening 239 Lists 84 Locative case 142 Locus 1 7 9 - 1 8 7 in the original configuration 178 Lover's talk 196 Macedonian 239 Manner 1 8 0 - 1 8 7 as a universal semantic category 180 Markedness 82, 1 5 6 - 1 5 7 M a r k i n g convention 157 Marking polarity 156 — 159 conflicting 1 5 7 - 1 5 9 Meaning 132, 148 figurative 217 in establishing equivalence 117, 119, 166 in the deep structure 137 in linguistic analysis 113 structural vs. lexical in American structuralism 247 vs. form 132 Melanesian 38 Mental scheme 215 Metalanguage 172 M e t a p h o r 217 Metaphorical concepts 63 Metaphorical extension 217, 223 of spatial prepositions 227 Metaphorization 84, 246 Military ranks 82 M o d a l s 173 see also Verbs

Modality 75, 221 in case g r a m m a r 141, 143 — 144 - , sentence 142, 145, 182, 247 Modifier 63, 110 M o o d 109, 182 M o r p h e m e 70, 83, 214, 245 negative 184 pre-sentence Q 1 3 7 - 1 3 9 , 198, 200-201 tense 198 well-formed and ill-formed combinations of 83 M o r p h o l o g y 70, 112, 216 nominal 246 —, simplified in reduced registers 196 Morphophonological nets 246 Names 143 Narrowing 247 Nasals 37, 122 in English 1 2 2 - 1 2 5 in Polish 1 2 2 - 1 2 5 Native language see Language N a v a h o 248 Necessary and sufficient conditions 214 Negation 145, 154, 184 pure vs. modal 185, 186 Negative constructions see Constructions Nomina agentis and loci 75 actionis 75, 208 see also Nominals Nominals 206 action and gerundive 206 — 211 average frequency of occurrence in English and Polish 210, 2 4 9 - 2 5 0 Nominative 130 Nominativus cum infmitivo 205 N o o t k a 248 Norwegian 239 N o u n 71, 72, 84, 109, 111, 115, 116, 130, 134, 145, 154, 173, 217, 248 as a universal category 116 full 207, 211 prototypical 216 —, structural definition 116

Index of subjects verbal vs. regular 209 Nouniness 2 0 8 - 2 1 0 Noun phrase 63, 65, 66, 84, 130, 132, 138, 141, 144, 145, 169, 170, 173, 174, 175, 198 Nucleus 58 Number 109, 111, 132, 133, 140 in Spanish and English 133 of nouns 109, 118-119, 132, 194 Numerals 38 Nursery rhymes translation of 250 Object 63, 109, 116, 128, 198 cognate 66 direct 66, 187 in case grammar 142 indirect 66 , prepositional 111, 116 Objective case 142, 143, 144 Obstruction 247 Obstruents 141, 142 Occlusion 32, 124, 247 Onset 58 Overgeneralization 211 as a source of errors 191 in forming interlanguage 193 Palatalization 53 Paralinguistic expressions body movement, gestures, grunts, features, mimicry, noises 92 — 93 features 246 Paraphrase 84, 159, 162 Parody 203, 249 Parole 25, 127 Parsing 109-110 Participant states 93 Participle 205, 245 past 38 active 204 present 194 Particle 109, 222, 223, 224 Parts of speech 109, 114-116 Passe compose in French 164 Passive 3 8 - 3 9 , 110, 111, 129, 137, 159-160, 163, 165, 194, 216, 246, 250

275

in Finnish 164 see also Constructions and Sentences Path 223, 228, 229 Patient 163, 164, 169, 179-187 in the original configuration 178, 179 Pattern matching 229, 230 complete (full) 228, 230 partial 2 1 7 - 2 2 1 , 241 Pedagogy 234 Perfekt in German 164 Performance 128, 147, 190, 243 in interlanguage 201, 202 native 206, 212 non-native 203, 206, 211, 212 translation as 147 vs. competence 128, 249 Performative 186 complex 185, 187 negative 185 verb 184, 185 Perlocutionary effect 2 1 8 - 2 2 1 Persian 165 Person 109, 135 in Polish and English 135 Philosophy 213, 214 Phoneme 70, 113, 126, 127, 214, 241, 245 as a natural category 216 comparison of 52 — 57 equivalent 52 — 57, 242 definitions 120—122 variants of 113 Phonemic inventories 112, 241 comparisons of 112 Phonemic overlap 126, 216 Phonetic representation 50 Phonetic similarity see Similarity Phonetic substance 220 Phonological interface 29 Phonological representation 50 Phonological rules see Rules Phonology 32, 48, 70, 91, 114, 214, 242 classical 216 generative 121, 126, 127 generative vs. taxonomic 127, 241, 242 —, segments in 121, 247

276

Index of subjects

Phrase 70, 71, 75 Phrase structure rules see Rules Phraseological collocations 72, 73, 77 fusions 73, 77 units 7 8 - 7 9 , 83, 246 Phraseology 70 Pidgin 196 Place as a universal semantic category 180-187 Plural 111 inflection in German and English 118 of regular gerunds 209, 210 Polish 36, 3 8 - 4 5 , 4 7 - 1 0 5 passim, 108, 119, 1 2 9 - 1 3 1 , 1 3 7 - 1 3 9 , 1 4 9 - 1 5 1 , 1 6 9 - 1 9 4 passim, 195, 2 0 3 - 2 1 2 , 220, 222-226, 229-230, 240-242, 245-249 interlanguage of English 202 phonemes 122, 127 word order 63 — 64 Politeness 91, 221 Polyglots 240 Polysemy 79-80, 87, 88 Possessive attribute 66 Pragmalinguistics 92 Pragmatics 18, 45, 48, 85, 97, 99, 112 contrastive 91, 92 definitions 90-92 general vs. sociopragmatics 91 "small" facts vs. "large" facts 97, 100, 101 Prague School 121 Predicate 63, 110, 169, 170, 181, 248 calculus 152 phrase 144, 145 Prefix 77, 2 2 2 - 2 2 6 "Premature" lexicalization 199-200, 201, 202, 249 Preposition 84, 109, 164, 173, 2 2 2 - 2 2 4 , 226 prototypical 216 with as an equivalent of instrumental inflection 1 4 9 - 1 5 1 Prepositional phrase 65, 204 Presupposition 69, 84, 152, 154, 185

Pronoun 29, 37, 84, 109 comparison of 42—43 definite and indefinite 44 frequency in texts 203 in reduced registers 196 personal 16, 28, 41 - 4 3 , 65, 176, 203 Pronunciation 75 Preposition 187 in case grammar 141 — 144 Prototypes 241 Prototypical effects 216 Psychological reality 237 Psycholinguistics 11, 237 Psychology 213, 214, 234 vs. contrastive analysis 237 Quantifiers 38 Question 49, 69, 137, 145, 154 ambiguity of in English and Polish 150-151 answers to negative in Korean 166-167 dependent and independent 139 echo 137 general 187 in Polish and English 137—140 intonation of 61 —62 negative in Persian 165 occasional 200, 249 positive 186, 187 rhetorical as a speech act 218 tag 137 wh- 97, 137, 187, 197-201, 249 Question word czy in Polish, optional omission of 138 ij]whether in English 139 Reading 239 Received Pronunciation 122 Standard 122 Receptive skills 239 see also Listening and Reading Reduced codes 200, 201 registers 172, 196 Reductionism 214 Reference 84, 174, 177, 248

Index of subjects and coreference 174, 176—177 composite character of 174 Referential index 178, 180-187 Referentials 175-176 Referents 175, 249 Regionalisms 246 Relational grammar see Grammar Request 2 2 0 - 2 2 1 Resident 179-187 in the original configuration 178, 179 Retraction 125 Rheme 110 Rhyme 18, 250 Rhythm 18, 59, see also Suprasegmental phenomena Roles 7, 48, 49, 169, 170, 178, 198, 204 complex 180, 183 coreferential 187 elementary 179, 186 in case grammar 142 in Contrastive Generative Grammar 169-187 nuclear 201 social 93 transfer of 211 Roman scholars 108 Rule 7, 28, 31, 68, 83, 84, 147, 169, 171 - 1 8 7 , 192, 193, 198, 201, 216, 217, 218, 245, 247 adjective placement 37 aspect 154 comparisons of 28, 35, 245 complication 196 congruent 40, 135 conversion 177—178 equivalent 40, 135 exceptions to, in traditional grammars 109 identical and different 172 in generative grammars 127, 128 — 146 input to 171-172 interrogative inversion 37 lexical insertion 177 morphographemic/morphophonemic 177 ordering 172, 200 , output from 171—172

277

passivization 37 phonological in Contrastive Generative Grammar 170 phrase structure 47, 128 — 136, 141, 145 post-lexical 177 projection 140, 144, 149 semantic categorical 177, 179 subcategorization 72 subject-raising 37, 39 syntactic categorical 177, 180, 182, 198, 202 syntactic transformational 177 transformational 127, 128, 129, 134, 135,137,148 wh- formation 198-199 word formation 247 Russian 190, 240 Samir 38 Sanskrit 53 Saxon genitive 246 Second language learning see Language Segment 50 in phonology 121, 247 Selection features 7 restrictions 72, 84, 148-151, 160, 161, 163 Self-embedding 177, 178 Semantic anomalies 72 fields 79 being at a place, colors, educational terms, emotions, gastronomical terms, kinship terms, leaving, military ranks, sports and games, vehicles 82 interface 29 interpretation 148, 248 markers 72, 86 relations 73, 84 horizontal and vertical 83, 84 representation 4, 6, 31, 32, 40, 152, 161, 169, 173, 182, 184, 197, 218 as nuclear subconfigurations 182 — of wA-questions 198

278

Index of subjects

structure 146, 152, 159, 163, 164, 165, 168, 169 as input to transformations 146, 152 identical in equivalent sentences 155, 168, 169, 1 7 6 - 1 7 7 Semantics 6, 85, 91, 92, 118, 136, 152 and deep structure 152 definition 92 interpretative vs. generative 152 in the standard theory 137 sentence vs. word 18, 47—48, 69, 1 5 4 - 1 5 5 , 159, 162 vs. pragmatics 85, 92 Semanto-syntactic gaps 164 Semanto-syntax 47, 100 Semivowels 55 Sense 223, 226, 230 metaphorical 217, 222, 227 prototypical 223, 225, 229 spatial 227, 228 Sentence 69, 70, 128, 130, 137, 145, 207, 210, 214 active 1 5 9 - 1 6 0 , 163 affirmative 93, 128, 135 ambiguous in Finnish 159—160, 171, 185 as output of generative grammar 169 average length of, in English and Polish 207 complex 129, 135, 146, 182 compound 129, 135, 146, 182 congruent 135 connection 154 embedding 154 enumeration of, in generative grammars 128 equivalent 6, 31, 65, 66, 136, 147, 148, 1 4 9 - 1 5 1 , 157, 169-187, 218-222, 246 hierarchical structure of 109—110 imperative 69, 142 interpersonal function of 18 interrogative 69, 129, 137, 142, 154 intonation of 61 kernel 128, 129, 135 negative 129, 137, 185, 195 - , passive 129, 137, 1 5 9 - 1 6 0 , 163, 165

positive 184 production of 249 prototypical 216 simple 128, 135, 205 stress 45 superordinate 184 various stages of derivation 201 Sentoid 145 Serbo-Croatian 194, 241 Set intersection 219 theory 214 Sex see Sociolinguistic factors Shortening in word formation 73 Sibilants 122, 124 in English 122 in Polish 1 2 2 - 1 2 3 Signs 214 simple see Morphemes Similarity 189, 247 as tertium comparationis 141 degree of 172, 2 1 8 - 2 2 1 , 230 formal 131 index 171 in lexical studies 75 — 77 of usage 131 phonetic 52 — 53, 57 pragmatic 221 structural 131 syntactic 218 Simplification 194, 196 strategy 195 Simplified codes 196, 197 Situation 93 Situational setting see Tertium comparationis Slovak 170, 239 Social conditions see Sociolinguistic factors Sociolinguistics 11 Sociolinguistic factors age, sex, temperament, social conditions, verbal skills 104 Spanish 36, 53, 54, 65, 76, 77, 1 3 1 - 1 3 4 , 143, 170, 217 Speaker 94, 128, 181, 185 Specialization 76

Index of subjects Speech act 92, 99, 100, 180, 181, 218, 221 Spelling 75 Sports and games see Semantic fields Standard theory 7, 48, 136-137, 140, 144, 148, 152, 161, 180, 213, 218 and contrastive studies 247 Statement 69 as a speech act 218 State in the semantic structure of passive sentences 164 Stimulus in stimulus-response relation 237 Stops 122, 124, 125, 247 in English 122, 125 in Polish 1 2 2 - 1 2 3 , 125 Strategies of communication in forming interlanguage 194 Stratificational grammar see Grammar Stress 240, 247 heavy in occasional questions 249 in compound nouns 136 special 187 see also Sentence Structural description 171 models 107 signals 115-117 Structuralism 36, 48 American 4, 113 — 127 vs. cognitive linguistics 213 Style 203 individual 212 in quantitative contrastive studies 203 Subconfiguration 178 — 187 nuclear 181 - 1 8 5 , 1 9 7 - 2 0 2 Subject 63, 65, 66, 109, 110, 116, 128, 249 extraposed 206 in case grammar 142 intransitive 246 of the passive 111 Substantia deverbalia 208-211 verbalia 208 - 211 Sufficient and necessary conditions 88 Suffix 117

279

-anie, -enie, -cie in Polish 208 — 211 -ing in English 207—211 Sumerian 39 Suppletion 7 Suprasegmentals see Suprasegmental phenomena Suprasegmental phenomena as a structural signal 115 intonation 59, 60-63, 69 pauses 60 rhythm 59 stress 59-60 Surface structures 4, 8, 127, 137, 138, 1 4 1 - 1 4 6 , 148, 185, 245 Swahili 143 Swedish 9 4 - 9 6 , 239 Syllable 37, 58 Symbolic logic 152 Synonymity 79, 84, 85 of constructions 149 Synonyms 80 — 82 Syntactic congruity 78 frames 173 markers 246 primes 146 names, verboids 143 — 144 structure 181, 201, 245 and aspect of verbs 154 as output of transformations 146 Syntagm 245 Syntax 68, 70, 73, 85, 91, 112, 216 Systems 28, 147 comparison of 28, 35, 245 of phonemes 120 Tautology 87 Telegraphese 196 Temperament see Sociolinguistic factors Tense 69, 109, 138, 142, 152, 153, 173, 182, 198, 200 future 182, 249 "dangling" 1 3 8 - 1 3 9 past 69, 144, 182, 192, 198, 249 perfective forms of 164, 165 present 182, 198, 249 present continuous 110, 194 —, present perfect 165

280

Index of subjects

Tertium comparationis 4, 15 — 21, 24 — 33, 37, 40, 60, 69, 8 6 - 8 8 , 1 0 0 - 1 0 1 , 110, 113, 1 1 6 - 1 1 7 , 119, 126, 1 3 0 - 1 3 2 , 1 4 0 - 1 4 1 , 221, 250 acoustic 52, 124 articulatory 5 2 - 5 3 , 123 auditory 52, 124 deep structure as 161 formal and semantic 16, 20 — 21, 37, 132, 146 immediately relevant 31 — 33, 123 — 125 phonetic similarity as 52 semantic features as 86 situational setting as 100—101 substantive 2 9 - 3 0 , 123, 126, 245 ultimately relevant 32 — 33, 123 — 124 universal base as 127 Text 173, 1 7 4 - 1 7 5 , 177, 190, 203, 205, 211, 220 as a translation 203 as object of parody 249 contrastive study of 206 in interlanguage 212 Text grammar see Grammar Theme 110 Theories choice of for contrastive analysis 240 expert and folk 215 linguistic 233, 2 3 6 - 2 3 7 , 243 Time 145, 1 8 0 - 1 8 7 as a universal semantic category 180 Topic 154 comment relations 69, 154 Traditional grammars see Grammars Trajector 222, 2 2 8 - 2 2 9 Transfer 212, 237 lexical 211 negative 2, 97, 172, 189, 210, 242 of training, in forming interlanguage 193-194 positive 172, 189, 204 unpredictability of, by contrastive studies 212 Transformational generative grammar see Grammar Transformation 47, 128, 134, 136, 145, 146

agreement 49 and the meaning in the standard theory 137 "cosmetic" (post-lexical) 173, 197, 198, 249 do-support 49, 138 impersonal 164, 170 interrogative 137 — 139 lexical insertion 83 passive in English 170 syntactic, in Contrastive Generative Grammar 170, 197, 198 Translation 1 7 - 1 8 , 26, 28, 30, 147, 204, 241, 248, 250 as primary data for quantitative contrastive studies 203 as performance 161, 168 erroneous 30, 161 — 162 of legal texts 221 of nursery rhymes 250 statistical data 245 vs. equivalence 18 — 19, 147, 161, 164, 165, 167, 171 word-for-word 19, 20, 27, 162, 163, 165, 246, 250 Transforms 135 Tree as representation of structure 181 Trill 126 Truth conditions 92 2-text 25-26, 28, 30, 34, 203 Typological linguistics see Linguistics Underlying structure 4, 24, 127, 160, 161, 165, 170, 187, 201 Universal concepts 112 principles and parameters 247 Universal base as input to Contrastive Generative Grammar 172, 177 as tertium comparationis 127 hypothesis 6 - 7 , 144, 146 Universal grammar see Grammar

Index of subjects Variants of phonemes prototypical vs. non-prototypical 216 free, of Polish nadjponad 225 Variphones 120 see also Free variants Vehicles see Semantic fields Verb 2, 37, 38, 71, 72, 75, 84, 109, 115, 141, 142, 144, 145, 150, 155, 163, 173, 192, 198, 200, 208, 217, 248, 249 auxiliary 3 8 - 3 9 , 173 auxiliary, be 38, 145 auxiliary, do 249 causative 205 copula, absence of, in reduced registers 196 defective in Polish 157 equivalent in Polish and English 156-159 finite 128, 131 imperfect 156 inflection of 65 intransitive 84, 160 in Finnish 160 linking 134 main 65, 138 modal 65, 138, 142, 247 non-finite 205 of communication 131 perceptive 205 , perfect 156

281

prototypical 216 Stative 164 transitive 38, 84 two-word 77 Verb phrase 63, 66, 130, 132, 173 as complement 151 as predicate 151 Verbal particles 77 Verbal skills see Sociolinguistic factors Verboids 143 Vocal tract 32, 247 Voicing 32, 125 Vowels 37, 54, 55, 126, 242, 247 stressed and unstressed 216 Word 7 0 - 7 1 , 7 3 - 9 0 , 205, 210, 211, 248 complex 74 — 75 composition of 73 compound 71, 74 — 75, 78 content 79 definition of 70 - 71, 248 formation 73 — 77, 83 generator 83 lexical 135 order 9, 10, 4 4 - 4 5 , 63, 69, 115, 154 as a structural signal 115 in reduced registers 196 , transfer from Polish 195 polymorphic 73 polysemous 80 Yiddish 98

Index of authors

Aarts, Flor 75, 251 Aclfric 1 Agard. Frederick B. 54, 59, 107, 251 A k h m a n o v a , Olga S. 261 Allen, J. P. B. 251, 265 Anderson, J o h n M. 247, 251 Anderson, Stephen R. 125, 251 Apresjan. Jurij Y. 48, 113, 248, 251 Arabski, Janusz 190, 251 Arnold. I. W. 70, 74, 78, 82, 252 Awcdyk. Wieslaw 51, 58, 59, 252 Bach, E m m o n 6, 252, 254, 260 Banathy, Bela H. 190. 252 Bartsch, Renate 47, 154, 252 Baudouin de Courtenay, Jan 121 Becka. J. V. 30, 204, 245, 252 Berault, Peter 112, 252 Biedrzycki, Leszek 55, 59, 252 Bloomficld, Leonard 113, 252 Boas, H a n s Ulrich 141, 252 Bogacki. Krzysztof 210 Bölingen Dwight 17. 72, 86, 252 Borges, Jorge Luis 248 Borkowski. G . 129, 252 Bouton, Lawrence F. 1. 152, 153, 155, 159, 160, 161, 163, 165, 167, 168, 252 Bowen, J. D o n a l d 54, 59, 120, 238, 264 Boyd, Julian 249, 252 Briere, Eugene J. 189, 252 Bross, J. S. 129, 252 Brown, Roger 197, 200, 201, 249, 252 Brugman. Claudia 213, 222, 223, 252 Burgschmidt. Ε. 253 C a t f o r d , John C. 6, 147, 233, 253 Chesterman, Andrew 1, 202, 249, 253 C h o m s k y , N o a m 7, 47, 48, 68, 71, 72, 91, 126, 128, 129, 130, 135, 136, 137, 141, 143, 148, 151, 168. 171, 213, 247, 253 Coles, Elisha 2, 253 Comrie, Bernard 1, 253

Cordcr, S. Pit 191, 192, 196, 234, 235, 236, 251, 253, 265 Cygan, Jan 248, 253 Danilewicz, Tadeusz 41, 253 De C a m p , David 196, 253 De Geest, W. 107, 253 Delattrc, Pierre 54, 253 De Saussure, Ferdinand 48 Dillon, George L. 217, 253 Dingwall, William O. 11, 129, 130, 135, 253 Dinneen, Francis P. 246, 254 Di Pietro, Robert J. 1, 23, 24, 54, 59, 86, 107, 108, 141, 143, 144, 145, 146, 152. 169, 170, 172, 218, 233, 251, 254 Dluska, M a r i a 60, 254 Duczmal, Stanislaw 75, 82, 254 D u s k o v a , Libusa 205, 254 Eliasson, Stig 255 Enquist, Niels Erik 190, 254 Ferguson, Charles A. 196, 254 Filipovic, Rudolf 241, 254 Fillmore, Charles 6, 7, 90, 91, 97, 98, 99, 100, 101, 103, 104, 141. 142. 143, 144, 152, 213, 246, 254 Fisiak, Jacek 3, 10, 11, 15, 23, 41, 43, 50, 51, 59, 60, 61, 64, 65, 67, 84, 107, 108, 113, 121, 126, 130, 137, 140, 233, 234, 236, 246, 254, 255, 258, 262, 263 F o d o r , Jay A. 255 Francis, W. Nelson 59, 109, 115. 116, 121, 247, 255 Fräser, Bruce 250, 255 Fries, Charles C. 2, 113, 114, 117, 255 G a z d a r , Gerald 92, 255 G i m s o n , A. C. 56, 58, 62, 247, 255 Ginzburg, R. S. 78, 255 Gleason, Henry A. Jr. 59, 82, 121, 255 G n i a d e k , Stanislaw 107, 255 Goldby, H a r r y 112, 257 G ö t z , D. 253

284

Index of authors

Graczyk, Ireneusz 206, 255 Grzegorczykowa, Renata 250, 255 Gussmann, Edmund 50, 51, 58, 242, 255 Halle, Morris 83, 125, 126, 247, 253, 255, 256 Halliday, Michael A. K. 6, 17, 18, 19, 35, 256 Hall-Partee, Barbara 264 Harms, Robert T. 252, 254, 260 Harris, Zcllig 113, 114, 256 Hewes, John 2, 256 Hill, L. A. 77, 256 Hiz, Henry 176, 256 Hjelmslev, Louis 18, 29, 38, 256 Hockett, Charles 38, 248, 256 Holmes, J. 256 Howel(l), James 2, 256 Hudson, Richard A. 47, 256 Hymes, Dell 48, 91, 256 Ivir, Vladimir 16, 17, 256 Jacobovits, Leon A. 259 Jacobs, Roderick A. 255 Jaeger, Jeri 216, 256 Jakobson, Roman 121, 125, 256 James, Carl 1, 11, 15, 18, 19, 91, 190, 202, 233, 234, 235, 236, 237, 238, 239, 241, 242, 245, 256 Janicki, Karol 32, 104, 140, 220, 256 Jassem, Wiktor 52, 55, 56, 58, 120, 121, 125, 247, 256 Jerome, K. Jerome 245 Johnson, Mark 63, 217, 259 Jones, Daniel 121, 257 Joos, Martin 265 Kade, O. 263 Kakietek, Piotr 40, 41, 257 Kalisz, Roman 75, 82, 217, 218, 219, 250, 257 Kalogjera, Damir 41, 257 Karcewski, Sergej 121 Karttunen, Kimmo 40, 257 Karttunen, Lauri 92 Kastovsky, Dieter 257, 258 Katz, Jerrold J. 148, 149, 159, 255, 257

Keenan, Edward 263 Kendon, A. 265 Khidekel, S. S. 255 Kielski, Bolestaw 107, 257 Knyazeva, G. J. 255 Köhler, Klaus 127, 257 Kohn, Kurt 257 Kolbert, Jack 112, 257 Konderski, Stefan 141, 257 König, Ekkehard 141, 257 Kopczyhski, Andrzcj 18, 54, 125, 257 Koutsoudas, Andreas 172, 257 Kramsky, Jiri 247, 249, 258 Krassowska, Halina 262; see also Mackiewicz-Krassowska, Halina Kryk, Barbara 41, 258 Krzeszowski, Tomasz P. 3, 7, 9, 18, 20, 27, 38, 54, 76, 77, 95, 107, 120, 121, 125, 135, 140, 141, 152, 161, 175, 176, 177, 184, 187, 198, 218, 245, 246, 247, 248, 249, 250, 258, 265 Kudzinowski, Czeslaw 164, 258 Kufner, Herbert L. 107, 116, 119, 120, 258 Kühlwein, Wolfgang 12, 24, 25, 258 Kurytowicz, Jerzy 249, 258 Lado, Robert 16, 52, 57, 60, 75, 76, 189, 190, 239, 246, 259 Lakoff, George 47, 63, 88, 109, 148, 149, 151, 213, 214, 215, 216, 217, 221, 231, 248, 250, 259 Lakoff, Robin 1, 259 Lamminmäki, Reijo 52, 259 Langacker, Ronald 47, 85, 109, 213, 215, 223, 259 Langendoen, D. Terence 143, 259 Lawendowski, Boguslaw 82, 259 Lee, William R. 233, 259 Leech, Geoffrey 69, 91, 97, 259 Lees, Robert B. 207, 208, 209, 210, 211, 250, 259 Lewandowska-Tomaszczyk, Barbara 40, 249, 250, 259 Lewicki, Andrzej Maria 262 Lewis, Mark 2 Lindner, Susan 213, 259 Lipinska-Grzegorek, Maria 3, 7, 15, 16, 23, 107, 130, 255, 260 Liston, Jerry L. 16, 260

Index of authors Lloyd, Β. Β. 215, 262 Ludkiewicz, Anna 246, 260 Ludskanov, Alexander 248, 260 Lyons, John 70, 84, 210, 246, 256, 260 Mackiewicz-Krassowska, Halina 59, 60, 61, 260 Madarasz, P. H. 190, 252 Mansion, J. A. 110, 260 Manczak, Witold 89, 260 Marek, Boguslaw 59, 61, 62, 260 Markkanen, Raija 41, 260 Martin, J. W. 238, 264 Marton, Waldemar 40, 129, 135, 136, 218, 233, 247, 260 Matisoff, James A. 98, 260 McCawley, James D. 47, 72, 84, 152, 213, 260 Mcintosh, Angus 256 Meech, S. B. 3, 260 Mel'chuk, Igor A. 241, 260 Mickiewicz, Adam 250 Micklesen, L. 129, 252 Miege, Guy 112, 261 Mieszek, Aleksandra 59, 61, 62, 63, 261 Milewski, Tadeusz 39, 261 Montague, Richard 92, 261 Morgan, Bayard Q. I l l , 261 Morris, Charles W. 48, 261 Mostowska, Katarzyna 246 Moulton, William G. 54, 261 Nawrocka-Fisiak, Jadwiga 262 Nemser, William 192, 193, 235, 261 Neubert, Albert 263 Newmeyer, Frederick J. 213, 261 Nickel, Gerhard 107, 233, 238, 257, 258, 261, 264 Nilsen, Don Lee Fred 141, 247, 261 Nowakowski, Miroslaw 71, 83, 84, 155, 246, 261 Oleksy, Wiestaw 40, 99, 100. 102, 258, 261 Olsson, M. 190, 261 Ozga. Janina 59, 60, 127, 200, 201, 261 Pasanen, Maija-Liisa 75, 262 Phillips, John S. 196, 265 Peters, Stanley 262

285

Pickbourn, James 2, 262 Pieper, Ursula 258 Pike, Kenneth L. 121, 262 Poe, Edgar Allan 220 Polanski, Kazimierz 129, 262 Politzer, Robert L. 16, 189, 262 Postal, Paul 47, 56, 148, 149, 159, 257, 262 Preston, Dennis R. 7, 8, 68, 104, 246, 262 Pride, J. B. 256 Puppcl, Stanislaw 55, 262 Puzynina, Janina 250, 262 Quirk, Randolph 81, 262 Ravem, Roar 200, 201, 202, 262 Richards, Jack C. 253, 261, 262, 263 Riley, Philip 92, 93, 94, 95, 96, 97, 99, 100, 101, 262 Robinson, Jane J. 141, 262 Rosch, Eleanor 215, 262 Rosenbaum, Peter S. 255 Ross, John Robert 209, 213, 216, 262 Rubach, Jerzy 127, 263 Rülker, Κ. 18, 263 Rusiecki, Jan 245, 263 Sajaavara, Kari 254 Salmon, Nicolas 247, 263 Sampson, Geoffrey 175, 248, 263 Sanders, Carol 233, 263 Sankin, Α. Α. 255 Sapir, Edward 113, 114, 263 Saumyan, Sebastjan 83, 263 Schachter, Paul 264 Schlyter, Suzanne 217, 263 Schmitz, John Robert 75, 82, 263 Schwarze, Christoph 12, 76, 263 Searle, John 48, 263 Sehncrt, James 75, 82, 263 Selinker, Larry 193, 194, 195, 235, 263 Sharwood-Smith, Michael 41, 75, 233, 263. 264 Shibatani, Masayoshi 253 Shuy, Roger W. 262 Sicwierska, Anna 1, 246, 264 Skorupka, Stanislaw 246, 264 Smith. Henry Lee 59, 264 Snook, Roger L. 107, 264

286

Index of authors

Sobieniowski, Florian 162 Spalatin, Leonardo 16, 18, 264 Stalnakcr, Robert C. 92 Stanislawski, Jan 80, 107, 264 Steinberg, Danny D. 259 Stickel, Gerhard 258 Stiebcr, Zdzislaw 54, 264 Stock well, Robert 20, 36, 54, 59, 107, 120, 130, 131, 132, 134, 238, 247, 264 Strevens, Peter 256 Strothmann, F.W. I l l , 261 Suomi, Kari 125, 164 Sussex, Roland 37, 264 Svartvik, Jan 97, 254, 259 Szwedek, Aleksander 44, 45, 257, 258, 264 Thomason, Richmond H. 92 Thorne, James 249, 252 Tomaszczyk, Jerzy 82, 89, 264 Träger, George L. 59, 264 Tran-Thi-Chau 238, 264 Trier, Jost 82, 264 Trubetzkoy, Nikolaus S. 121

Tusiani, Juseph 170 Twaddell, W. Freeman 121, 265 Valdman, Albert 196, 265 Van Buren, Paul 1, 3, 12, 23, 107, 161, 247, 265 Van Dijk, Teun 128, 265 Varga, Laszlo 59, 246, 265 Vennemann, Theo 47, 154, 252 Wagner, Karl Heinz 233, 238, 261 Wekker, Herman 75, 251 Welna, Jerzy 76, 265 Whitfield, Francis J. 256 Widdowson, Henry 194, 195, 265 Wierzbicka, Anna 89, 265 Wilkins, John 109, 265 Wittgenstein, Ludwig 88, 214, 265 Woloszyk, Alicja 59, 61, 62, 265 Wojcik, Tomasz 107, 265 Yngve, V. H. 91, 265 Zabrocki, Tadeusz 3, 8, 68, 255, 265

Jacek Fisiak (Editor)

Contrastive Linguistics Prospects and Problems 1984.14.8 χ 22.8 cm. Χ, 449 pages. Cloth. DM 1 5 8 ISBN 90 279 3260 3 (Trends in Linguistics. Studies and Monographs 22) This collection of papers covers various aspects of contrastive linguistics. Among the topics treated are contrastive phonology, syntax, text linguistics, pragmatics and applications of contrastive linguistics to language teaching. The languages investigated include English, German, Polish and Irish. Contents: Robert D. Borsley: Free relatives in Polish and English - Stefan Dyia: A note on inversion/conjunct ascension constructions in Polish and English - Nils Erik Enkvist: Contrastive linguistics and text linguistics - Claus Fierch and Gabriele Kasper: Ja und? - og hva' sä? - a contrastive discourse analysis of gambits in German and Danish Rudolf Filipovic: What are the primary data for contrastive analysis? - Charles J. Fillmore: Remarks on contrastive pragmatics - Maria Grzegorek: English sentences with introductory there and their Polish counterparts - Edmund Gussmann: Abstract phonology and contrastive analysis Raymond Hickey: Towards a contrastive syntax of Irish and English - George M. Horn: Constraints on transformations: evidence from contrastive analysis - Ewa Jaworska: On the structure of adverbial subordinate constructions in English and Polish - Andrzej Kopczynski: Problems of quality in conference interpreting - Tomasz P. Krzeszowski: Tertium comparationis - Wolfgang Kühlwein: Pedagogical limitations of contrastive linguistics - Lewis Mukattash: Contrastive analysis, error analysis and learning difficulty - Wieslaw Oleksy: Towards pragmatic contrastive analysis Jerzy Rubach: Rule ordering in phonological interference - Kari Sajavaara: Psycholinguistic models, second language acquisition, and contrastive analysis - Michael Sharwood Smith: Learnability and second language acquisition - Aleksander Szwedek: Some problems of contrastive analysis and text linguistics - Tadeusz Zabrocki: On the nature of movement rules in English and Polish.

mouton de gruyter Berlin · New York

m

Niels Davidsen-Nielsen

m Tense and Mood in English mj A Comparison with Danish m

m m m

m m mj m m

1990. X, 224 pages. Cloth. ISBN 311 0125811

(Topics in English Linguistics l)

This monograph deals with the grammatical realization of expressions of time and modality in English. Tense is interpreted as a broad category with eight members, the perfect and future constructions being included. Mood is assumed to comprise not only morphologically signalled constructions like the subjunctive and the imperative, but also syntactically signalled constructions with modal auxiliaries. This work has both an analytical and a descriptive objective. One of the purposes is to provide a descriptive basis on which reliable and efficient teaching material may be prepared. By contrasting English with Danish, a language in which the forms of expression of modality and time are relatively similar, important differences are brought to light.

m m m m

m m

mouton de gruyter Berlin · New York