Universalism versus Relativism in Language and Thought: Proceedings of a Colloquium on the Sapir-Whorf Hypotheses 9783110805826, 9789027977915

Citation preview

UNIVERSALISM VERSUS RELATIVISM IN LANGUAGE AND THOUGHT

Contributions to the Sociology of Language

11

Joshua A. Fishman Editor

MOUTON · THE HAGUE • PARIS

Universalism versus Relativism in Language and Thought Proceedings of a Colloquium on the Sapir-Whorf

Hypotheses

R i k Pinxten

Editor

MOUTON · THE HAGUE · PARIS

ISBN: 90 279 7791 7 © 1976, Mouton & Co., Herderstraat 5, The Hague, The Netherlands Jacket design by Jurriaan Schröter Printed in the Netherlands

Preface

Fantasy is still a swear word in the social sciences; nevertheless, it might well be the only useful means of confronting well-rooted intellectual traditions. Let us therefore hope that we have displayed enough of it in our approach to the Whorfian problems. In the late winter 1971 some scientists and philosophers of our university in Ghent formed a subgroup within Communication and Cognition to work on the hypothesis of B. L. Whorf. Maybe one of the most important aspects of his statements is their vagueness, which sometimes appears to be a drawback to serious research, while at other times it seems to enable innovations and progression. The former is in many ways characteristic of much of the empirical work done on the subject hitherto (hard data still seem to be either lacking or uninterpretable). It is our conviction that this volume can serve as an example of the innovating effect of vagueness. Where progress was noted, it was due to the explicit and deliberate interdisciplinarity of the approach of all members of the group. Instead of constantly stressing the linguistic aspects of the problems, the group gradually came to look at them from a less disciplinarily bound distance: questions were asked about the nature, features, and boundary of the language and cognitive systems, and about the kind of relationship that could be established between the two. Of course, we did not answer all of the questions, but we were able to restate some of them in a more adequate way. On March 16 and 17, 1973, we organized and international colloquium on the subject. The following persons agreed to take different responsibilities for the organization of it: P. Burghgraeve, F. Vandamme, E. Vermeersch, and R. Pinxten. They were backed by the sincere enthusiasm of the members of the Committee of the Colloquium, comprising L. Apostel, Μ. De Mey, Η. Gaus, J. Kruit-

VI

Preface

hof, R. Thibau, H. van den Enden, P. van Geert, and E. Verhofstadt. Many internationally known scholars — mainly linguists, of course — were interested and were kind enough to offer their collaboration, thus enlarging our local research group in a magnificent way. Both the quality of the papers they presented and the intensity and cordiality in the more informal contacts between all participants guaranteed the success of the initiative. Finally, the publication of the papers of the colloquium was justified by the conviction that the colloquium had brought some new outlooks on the problems of relativism. The organizers considered that probably for the first time in the history of linguistic relativism (going back to Herder and Von Humboldt), relativists and universalists were more systematically and purposefully confronted with each other (most of all due to the profound insights of Schaff, Gipper, and Robins). A second novelty originated from our group: the problem of linguistic versus cognitive primacy was discussed in the light of the foregoing polarization. One should keep in mind the controversiality of this subject to linguists (e.g. Chomsky) and psychologists (e.g. Piaget). The papers are arranged here in the order of their presentation in the colloquium. Those articles contributed by authors unable to attend the colloquium are given in a separate part of the volume. The colloquium and its publication were made possible for Communication and Cognition by grants from the Faculty of Letters and Philosophy of the Rijksuniversiteit Gent, as well as from the Belgisch Centrum voor Onderzoek der Logika. Moreover, the kind help of many colleagues and staff members of the university must be acknowledged. As regards this publication in the Mouton series, we want to express our deepest gratitude to Professor Dr. J. A. Fishman who showed great interest in the papers, expressed encouraging enthusiasm about them, and agreed to have them published in the series he edits. Our gratitude goes to E. De Vlieger, M. De Sloovere, and N. Vanden Driessche for their assistance in preparing the manuscript. A last word goes to all the participants who manifested their sincere belief in the need for and the appropriateness of the interdisciplinary approach by contributing to this volume or by attending the colloquium. Let us hope, for the sake of interdisciplinarity, that this work can have some impact on the study of the languagethought relation in the future. THE EDITOR

List of contributors

AUSTRIA

Adam Schaff (Prof. Dr.)

European Coordination Centre for Research and Documentation in Social Sciences Grünangergasse 21; P.O. Box 974 Vienna — I

BELGIUM

Seminary of Logic and Rijksuniversiteit Gent Blandijnberg 2, Ghent

Epistemology

Seminary of Logic and Rijksuniversiteit Gent Blandijnberg 2, Ghent Aspirant Ν F. W.O.

Epistemology

Marc De Mey (Dr.)

Seminary of Logic and Rijksuniversiteit Gent Blandijnberg 2, Ghent

Epistemology

Flip G. Droste (Prof. Dr.)

Catholic University of Louvain Tiensestraat, Louvain

Rik Pinxten (Drs.)

Seminary of Contemporary Philosophy and Philosophic Anthropology Rijksuniversiteit Gent Blandijnberg 2, Ghent Aspirant N.F.W.O.

Karel Boullart (Dr.)

Paula Burgghraeve (Drs.)

VIII

List of

contributors

Fernand J. Vandamme (Prof. Dr.)

Seminary of Logic and Epistemology Rijksuniversiteit Gent Blandijnberg 2, Ghent; Vakgroep filosofie Hogeschool Tilburg, The Netherlands

Paul Van Geert (Drs.)

Seminary of Genetic and Psychology Rijksuniversiteit Gent Blandijnberg 2, Ghent Aspirant N.F.W.O.

Etienne Vermeersch (Prof. Dr.)

Seminary of Contemporary Philosophy and Philosophic Anthropology Rijksuniversiteit Gent Blandijnberg 2, Ghent RUCA, State University of Antwerp Antwerp

Differential

G R E A T BRITAIN

Robert H. Robins (Prof. Dr.) School of Oriental and African Studies University of London London, WC1E 7HP Heinz P. Walz (Prof. Dr.)

Department of Linguistic and Regional Studies University of Surrey Guilford Surrey GU2 5XH

ISRAEL

Nathan Stemmer (Prof. Dr.)

Department of Philosophy Bar-Ilan University Ramat-Gan

List of

contributors

ITALY

Max Beluffi (Prof. Dr.)

Instituti Psicbiatrici Milano

Provinciali

WEST GERMANY

Helmut Gipper (Prof. Dr.)

Institut für Allgemeine Sprachwissenschaft Westfälische Wilhelmsuniversität Münster

THE NETHERLANDS

mo Kaa

University of Postbus 3295,

Amsterdam Amsterdam

UNITED STATES OF AMERICA

Ervin Laszlo (Prof. Dr.)

Center of International Princeton University

Studies

John Cipolla (Drs.)

Center for the Advanced Communication University of Iowa Iowa City, Iowa 5224Q

Study

of

Contents

PREFACE BY THE EDITOR LIST OF CONTRIBUTORS

V VII

PART I: LINGUISTIC AND LOGICAL APPROACHES 1.

Generative grammar and the concept of innate ideas, by Adam Schaff

3

1. The school of generative grammar and structuralism in linguistics; 2. Generative grammar and neonativism; 3. Discussions on neonativism in connection with generative grammar; References.

2. 3.

Presupposition, truth and grammatically, by Flip G. Droste

57

Language, logic and thinking by Fernand J. Vandamme

69

Introduction; 1. The reductionism of T h o m and some parallelism between T h o m and Piaget; 2. Some data on the relation between thinking and language; 3. Logic and language; 4. Conclusion; Acknowledgement; References.

4.

The logic of non-European linguistic categories, by mo Kda

85

PART II: PSYCHOLOGICAL AND ANTHROPOLOGICAL APPROACHES 5.

The current relevance of the Sapir-Whorf hypothesis, by Robert H. Robins

99

Contents

XII 6.

Sapir-Whorf h y p o t h e s i s and brain a c t i v i t y , by Heinz

7.

P.

Walz

109

E p i s t e m i c universals. A c o n t r i b u t i o n t o c o g n i t i v e anthropology, by Rik

Pinxten

117

1. Introduction: 1.1 Intuitions; 1.2 Analysis: 1.2.1 Epistemic phenomena; 1.2.2 Economy of research; 2. Definitions of some basic notions: 2.1 What about universals?; 2.2 What epistemic is about; 2.3 A space among spaces; 3. The problem and the method; 4. Piaget's concept of space: 4.1 Bask assumptions of Piaget; 4.2 The model: 4.3 Interpretation and use of Piaget's model: 4.3.1 Stress on constitutional aspects; 4.3.2 Procedure of comparison; 4.3.3 The problem of the sources; 4.3.4 Pretensions of the work; 5. In search for an epistemic universal: 5.1 Chinese space; 5.2 African space; 5.3 Indian space; 5.4 Pre-Socratic space; 6. Conclusions: 6.1 Interpretation of the results; 6.2 Epistemic universals; Notes; References. 8.

A critical analysis o f S c h a f f s v i e w s in c o n n e c t i o n w i t h t h e r e l a t i o n b e t w e e n language and t h o u g h t , by Paula Brugbgraeve

177

1. The equation of thinking to human thinking; 2. Argumentation in favour of the unity between language and thinking; 3. The central characteristic of thinking: the categorial use of language; 4. An heuristic approach to the notion of thinking; 5. Experiments with animals: 5.1 Generalisation and discrimination; 5.2 Trial and error; 5.3 Conceptual level process; 6. The role of language in a thought process; References. 9.

L a n g u a g e and m e a n i n g f u l i n t u i t i o n o f reality. A general c o n t r i b u t i o n to t h e p r o b l e m o f linguistic relativity, by Paul Van Geert 1. Introduction: the problem of the elephant; 2. Meaningful intuition of reality: general definition; 3. Preparatory experiments: 3.1 Enchanted world game; 3.2 Choosing alternatives; 4. The structure of MIR: semantic matrices: 4.1 General notions; 4.2 Semantic matrices; 4.3 Nature and function of semantic matrices; 4.4 An explicitation of the structuring of MIR by means of semantic matrices; Appendix.

193

Contents

XIII

PART III: EPISTEMOLOGICAL AND METHODOLOGICAL ASPECTS OF RELATIVISM 10.

11.

12.

Is there a linguistic relativity principle?, by Helmut Gipper

217

Epistemologie et hypotheses de Whorf, par Etienne Vermeersch

229

La verification psycholinguistique de la psychiatrie et l'anthropoanalyse, par Max Beluffi 241

13. The reduction of Whorfian relativity through a general systems language, by Ervin Laszlo 257 Introduction; 1. The fragmentation of the common-sense world picture; 2. The fragmentation of the scientific-world picture; 3. The reunification of scientific-world pictures through general-systems theory; 4. Current effects of the reduction of Whorfian relations through general-systems language.

14.

Incommensurability of theory and untranslatability of languages, by Marc De Mey 269

15. A l'occasion de l'hypothese Sapir-Whorf: L'incompatibilite des systemes. Remarques generales, par Karel Boullart 277

PART IV: EXTRACOLLOQUIAL PAPERS 16. Nonlinguistic factors in language, by Nathan Stemmer

295

17. Mass media as language. The Sapir-Whorf hypothesis and electronic media, by John Cipolla 303

PART ONE

Linguistic and Logical Approaches

ADAM SCHAFF

1

Generative Grammar and the Concept of Innate Ideas

The present paper owes its origin to an analysis of the following issues: — What is the place of the concept of innate ideas in the system of generative grammar? — How is the adoption of that concept as one of the basic assumptions of that system being substantiated? — What critical comments suggest themselves in connection with that assumption? When reverting to the discussion of the problem of innate ideas in the system of generative grammar, I must begin with an explanation, be it alone for the facts that discussions on this issue have been both numerous and comprehensive (see Hook 1969; Chomsky 1967b), that both the advocates and the adversaries of the hypothesis, stating that certain grammatical structures are innate to all specimens of the species Homo sapiens and are in this sense universal, have formulated their principal pros and cons, thus formulating their standpoints with sufficient precision; and that the controversy can be settled neither by philosophers nor philosophizing linguists, but by natural scientists who, at the present day level of their knowledge of the genetic code and the related inherited mechanisms and man's innate structures of thinking and acting, can at most declare their predilections for certain hypotheses (as has been done by such prominent scientists as Jacques Monod (1970: 144ff) and Fran?ois Jacob (1970: 343), but cannot yet provide decisive arguments. It seems, accordingly, that we have to wait patiently for further advances in natural science, and especially further results in molecular biology. Philosophical considerations, including metatheoretical reflections on linguistics, inevitably run the risk of becoming, in this situation, a purely verbal controversy. But one exception may be raised here: Such reflections

4

Adam

Schaff

may help the controversy to become clearer by bringing out the latent premisses referred to and the related courses of reasoning, and/or by stimulating specialized researches by way of posing new questions and thus indicating new paths of research. It seems that the problems raised below satisfy these conditions at least in part. This may be an illusion, but my subjective conviction at least justifies my late contribution to earlier discussions.

1. THE SCHOOL OF GENERATIVE GRAMMAR A N D STRUCTURALISM IN LINGUISTICS

The following comment, obvious as it is, seems to be in place just at the outset of this paper: When it comes to the controversy over the underlying assumptions of generative grammar I will be interested — in the sense of the field in which I may intervene — in general theoretical, or rather philosophical, aspects of that trend in linguistic reflections. These cannot, of course, be separated from specialized linguistic analyses, in which linguists are solely competent. But it may be the case that there is a specific remainder under consideration which, by being situated in the border area of linguistics and logic, makes it possible for a philosopher to interfere without exposing himself a priori to an objection of noncompetence, provided that he approaches the issue qua philosopher. This is the more so if certain elements of a theory are clearly philosophical in nature, as is the case here, when it is said that specified grammatical structures have the status of innate ideas. What has been said in this paragraph explains the nature of the comments that are to follow. Transformational generative grammar, associated mainly with the name of Noam Chomsky, has its origin in American structural linguistics, namely the school of descriptive linguistics, and is, as is commonly known, an important branch of structuralism in linguistics. But if we compare the principles of generative grammar with those of the Prague school of phonology, which marked the development of structural linguistics as the dominant trend in modern linguistic research, then it is difficult to understand why these two schools bear the common name of 'structual linguistics'. This astonishment, in the past supposed to give rise t o philosophical issues, is proper not only to philosophers who engage in metatheoretical reflections. When it comes to the alleged homogeneity of linguistic structuralism similar scepticism was voiced long ago by

Generative grammar and the concept of innate ideas

5

A. Martinet (1955: ch. 3), who thinks that the term structuralism has become a label of almost any no'ntraditional trend in linguistics. When such scholars as E. Benveniste (1962), A. Martinet (1960), and Ν. M. Gukhman (1962) strive to find out what is common to the various trends of structural linguistics, it turns out that differences dominate, and the common points are mainly negative in nature, namely the rejection of the concepts of the neogrammarians. The same applies t o the school of transformational generative grammar which stands so far apart even if compared with American descriptive linguistics, from which it has developed, that it deserves a separate treatment and a separate name. The opinion of Paul M. Postal is very characteristic in that respect: 'It must be said that, within the tradition of modern structural linguistics which developed the ideas of immediate constituent grammar, little attention has been directed toward the task of precisely characterizing phrase structure as a generative theory. Rather, in line with the highly methodological and procedural orientation of modern linguistics, most effort has been directed toward the goal of formulating procedures of segmentation, classification, and substitution which could be used to discover the grammatical analyses of arbitrary sentences in arbitrary languages. This naturally led to the focussing of much attention on the notion of structural description, but almost none on the finite devices or linguistic rules which the speaker must learn which can assign grammatical analyses to infinite sets of sentences. There was before Chomsky's work, thus, little interest in the goal of specifying exactly the character of the notions linguistic rule, grammar, and so forth . . . There are also dozens of articles in modern linguistics attempting to describe abstractly the nature of such elements as phoneme, morpheme, word, constituent, and other elements of structural descriptions. But for the most part one looks in vain prior to 1956 for works specifying the abstract character of linguistics rules.' (1964: 141; for a similar though not so radical, interpretation see Ruwet 1967: 63—64, 77—78). This statement is interesting for two reasons: First, it contributes an additional element to the problem of the meanings which underlie the ambiguous term structuralism, and, second, it introduces us to the main issue, namely an answer to the question about the issues which are focal for, and characteristic of, the school of generative grammar. The differences between the set of concepts used in transformational generative grammar and those current in other schools of

6

Adam

Schaff

linguistic structuralism are linked with the differences in the subject matter of study and the research methods applied. All schools of linguistic structuralism may be said to study language understood as a system; all of them, too, strive for one and the same goal, i.e. finding out the structure of that system. Closer inspection shows, however, that this statement is so vague a generality that is obscures the most interesting and the most essential characteristics of each of those schools, namely the differences in the interpretation of the statement that language is a system, and hence the difference in establishing the structure of what they are supposed to investigate. The phonological school, which was the most precise of all in defining the subject matter and the method of research, intends to offer a structural description of the phonic aspect of language by starting from oppositions between phonemes, i.e. the smallest phonic units which are devoid of meaning. The descriptivists strive to describe structurally the various linguistic phenomena by way of segmentating and classifying morphemes as the smallest meaningcarrying units of language. Both schools, differing from one another as they do, start from the requirements formulated by de Saussure, who in fact confined the methods of linguistic analysis to segmentation and classification, and believed the construction of sentences to be independent of the system of language (langue) and to be in the sphere of the mechanism of speech {parole). (On this issue see Chomsky 1968: 17.) Now in this respect the school of transformational generative grammar consciously opposes traditional structuralism by concentrating on sentences and the grammatical rules of sentence formation. 'The grammar, thus is a device that (in particular) specifies the infinite set of well-formed sentences and assigns to each of these one or more structural descriptions. Perhaps we should call such a device a generative grammar to distinguish it from descriptive statements that merely present the inventory of elements that appear in structural descriptions, and their contextual variants' (Chomsky 1969a: 9). Chomsky and his school point emphatically to the differences between the two types of structuralism; they see a demarcation line between the taxonomic model of grammar, associated with modern structural linguistics, and the transformational model which, according to Chomsky, comes closer to traditional grammar, or, as Chomsky (1969: 11) sometimes says pointedly, to philosophical grammar. Chomsky's point is not to give a correct description of an

Generative grammar and the concept of innate ideas

7

utterance by making an orderly list of its elements (he does not disclaim this task, nor does he reject a phonological analysis, but he does not want to confine linguistic theory, i.e. grammar, to this task alone), but to find out how people, who communicate with one another by means of language and have a limited repertory of linguistic means at their disposal, can construct and understand an infinite set of grammatically correct sentences. When striving to answer this question he does not find inspiration in modern structuralism because that school has never been interested in this issue. He does find inspiration in philosophical tradition, for even if that problem was often there being formulated in an obscure and ambiguous manner (compare W. van Humboldt's concept of 'Form der Sprache', which Chomsky interprets as a 'generative process' (1969a: 17); it was treated as an important one. As Chomsky's sally into the history of philosophy, made in his Cartesian linguistics (1966a) has shown, the development of the philosophy of language from Descartes to Wilhelm von Humboldt was in many respects a forerunner of transformational generative grammar. The stressing of the opposition of the two types of structuralism, and his dissociation from the trend, which he terms 'modern structuralism', found a vigorous manifestation in Current issues in linguistic theory an expended version of Chomsky's paper read at the Ninth International Congress of Linguists in 1962, the paper which carried Chomsky's sui generis declaration of faith. The passage in question deserves quoting not only because of its importance for the development of the school of transformational generative grammar, but above all because it clearly defines Chomsky's point of departure, with all its consequences for the logical construction of the conceptual apparatus of his theory. 'Modern linguistics is much more under the influence of Saussure's conception of langue as an inventory of elements (Saussure, 1916, 154 and elsewhere, frequently) and bis preoccupation with systems of elements rather than the system of rules which were the focus of attention in traditional grammar and in the general linguistics of Humboldt. In general modern descriptive statements pay little attention to the 'creative' aspect of language; they do not face the problem of presenting the system of generative rules that assign structural descriptions to arbitrary utterances and thus embody the speaker's competence in and knowledge of his language. Furthermore, this narrowing of the range of interest, as compared with traditional grammar, apparently has the effect of making it impossi-

8

Adam

Schaff

ble to select an inventory of elements correctly, since it seems no inventory (not even that of phonemes) can be determined without reference to the principles by which sentences are constructed in the language. To the extent that this is true, 'structural linguistics' will have suffered from a failure to appreciate the extent and depth of interconnections among various parts of a language system. By a rather arbitrary limitation of scope, modern linguistics may well have become engaged in an intensive study of mere artifacts' (1969a: 23—24; my italics). This is not only a criticism of structuralism which, be it alone because of the time of its birth, we might term traditional, but also a definition of Chomsky's own standpoint, even if that definition be by negation only. Before we try, by reference to the above formulations, to reconstruct the inner logic of the idea of transformational generative grammar (which is necessary if we are correctly to understand the place and the role, in that grammar, of the theory of innate ideas, which is the main goal of this analysis), it seems to the point to make one more comment on the method used by the generative grammar school, which will help us better to answer not only the 'what?' question, but also the 'how?' question concerning the subject matter of the researches carried out by that school. As mentioned above, Chomsky explicitly stresses the links between his ideas, on the one hand, and traditional grammar and the issues of the philosophy of language (which because of its origin he has termed Cartesian), on the other. He is certainly right in doing so, but this does not suffice if we are to comprehend those filiations of ideas which are essential for the comprehension of the meaning and the methods of transformational generative grammar. What is meant here above all is that grammar links with modern mathematical logic, in particular with the school of the logical analysis of language in its British and continental forms (in the latter case, mainly the Vienna Circle and the Lvov Warsaw school of logic). The comprehension of these filiations of ideas and the impact of those ideas upon that trend in language analysis which has been taken up by Chomsky is a sine qua non of the comprehension and appreciation of the difference between generative grammar and linguistic structuralism in its form to date: while the method used in traditional structuralism, especially in American descriptivism, has been markedly inductive, that used in transformational generative grammar is deductive. (An explicit reference to this fact is made by Shaumian 1962: 406.) The latter method consists in the construction

Generative grammar and the concept of innate ideas

9

of models applicable to the interpretation of infinite sets of elements (those elements in this case being sentences). The procedure resembles the axiomatic method: certain assumptions are adopted as axioms on which the whole structure is based. In other words (and f r o m a somewhat different point of view), transformational generative grammar strives not only to describe sentence structure in the sense of describing the relationships between its elements (morphemes or phonemes), but also to answer the question of how a given sentence can be derived from other sentences and from which other sentences it is derivable (which refers to the structure of relations between sentences). Hence the stress on syntax and the criticism of traditional linguistic structuralism as a trend which neglected, or failed to pay due attention to, the problems of syntax. 'Thus, in this terms, sentence formation is not strictly a matter of langue, but is rather assigned to what he (Saussure) called parole, and thus placed outside the scope of linguistic proper; it is a process of free creation, unconstrained by linguistic rule except insofar as such rules govern the forms of words and the patterns of sounds. Syntax, in this view, is a rather trivial matter. And, in fact, there is very little work in syntax throughout the period of language linguistics' (Chomsky 1968: 17). But there have been trends which specialized in those problems exactly and which must have affected Chomsky's ideas:-They were those trends in the philosophy and the logic of language which in a certain period (prior to the demonstration by Alfred Tarski that the problems of truth can be introduced into deductive sciences without the risk of contradictions, which paved the way for the inclusion by logical positivists of semantic issues in their analyses of the language of science) were solely confined to an analysis of the logical syntax of language. Two names come to the fore if we consider the forerunners of transformational generative grammar from the point of view of the modern logic and the philosophy of language: Rudolf Carnap and Kazimierz Ajdukiewicz. Without taking into consideration Die logische Syntax der Sprache, Carnap's fundamental work, and the works published by Ajdukiewicz in Erkenntnis (1934a,b; 1935), we could not, in my opinion, understand the mental work done by Chomsky in his formulation of his concept of transformational generative grammar. Since this analysis of filiations of ideas in Chomsky's conceptions (apart from those to which he himself points in the sphere of the philosophy of language) is to be continued below, a reservation must

10

Adam

Schaff

be made concerning a certain basic issue. The originality of a given system is not refuted by the demonstration that certain elements of that system used to occur in some form earlier in the history of human thought. The concepts and ideas which come here in question as a rule did not (and could not) occur in the past in the same form, and a fortiori could not play the same role in the structure of the whole system, as in the case of the modern theory which is the subject matter of this analysis. Further, which is the most important, the originality and the scientific significance of a theory is not a function of the novelty of its elements (at least, it is not only a function of the ingenuity of the ideas into which that theory can be broken up), but in most cases it is a function of the novel and creative nature of the combination of those elements into a whole, into a definite system. This is why it is not paradoxical to state that in the case of many theoretical systems whose role in the history of science was novel and creative, most, or perhaps even all basic ideas of those systems had been earlier formulated in some way, and hence were not original in the radical sense of the word. This fact is emphasized here because, in my opinion, the basic concepts and ideas in Chomsky's theory (as stated above, his techniques,of the description and analysis of language are not assessed here) are — in the radical sense of the word — neither new nor original, while his theory as a whole I hold to be both novel and valuable. Let us begin with the distinction between competence and performance a distinction which is fundamental in the system of generative grammar. Its forerunners are the theory of langue and parole in de Saussure's Cours (see Katz 1966: 116), and the distinction between code and message in communication theory. Another concept which is of fundamental importance for transfromational generative grammar is the distinction between deep structure and surface structure, which in turn has its forerunner in the theory of double articulation, as developed in French linguistics. (Roman Jakobson [1970] says that priority in this respect goes to the paper by Bubrix [1930].) The theory of language rules (rules of sense and rules of syntax), non existent, as Chomsky points out, in traditional linguistic structuralism and so important in the set of concept used in generative grammar, was being developed in detail by logical positivists, the works of Kazimierz Ajdukiewicz being of special significance in that field. The concept of universal grammar is to be found in the works by representatives of what Chomsky calls 'Cartesian linguistics'. Of

Generative grammar and the concept of innate ideas

11

course, beginning with Descartes is quite arbitrary, because the principle of nihil novi sub sole, applicable to all philosophical doctrines, holds for Descartes, too, which in this case is confirmed by Raymond Lull's idea of calculi universales. Likewise, the concept of innate linguistic structures has its roots in the theory of innate ideas of genetic rationalists, which Chomsky himself emphasizes. Other items could be added to this list. This is mentioned here to absolve me of the duty of making excursions into history when I come to discuss issues which, in my opinion, are essential for the comprehension of the set of concepts used in transformational generative grammar. Here is the attempt to reconstruct that set of concepts. The starting point is the definition of .language as a set (finite or not) of sentences, which are constructed from elements whose set is finite (Chomsky 1957: 13). (From the point of view of filiations of ideas it is worth noting that such precisely was the standpoint formulated by Ludwig Wittgenstein in his Tractatus Logico-Philosophicus, 4.001: 'The totality of propositions is the language'.) The grammar of a given language is a device which serves to produce sentences which have a specified structure and are in this sense grammatical (Chomsky 1957: 13; see also 1969a: 9). Thus the task is not so much to describe the structures of statements actually made, but to simulate and to explain the behavior of a language user who in the light of his limited experience with a given language can produce and understand infinitely many new sentences in that language. This idea is fundamental for the definition of generative grammar as distinct form traditional grammar. Only a grammar so conceived can explain linguistic creativity, understood as the ability to produce infinitely many sentences from a finite set of elements. The point is, obviously to produce grammatical sentences, i.e. sentences which comply with the rules of a given grammar. Hence grammar may be defined as a set of rules (identical with the concept of device) which makes it possible to produce in a given language an infinite set of grammatical sentences and to describe their structures (Chomsky 1964a: 119—120). Generative grammar, when assimilated by a person who has the command of the language, coincides with what de Saussure termed langue. This brings us to the distinction between competence and performance (with its analogues langue versus parole and code versus message), which is fundamental in the system of generative grammar (Chomsky 1969a: 10). The person who has assimilated the system of

12

Adam

Schaff

a given language, which defines both the phonetic form of a given sentence and its semantic content (i.e. the rules which assign specified meanings to specified sequences of sounds) (Chomsky, 1967a: 397—398), has thereby acquired linguistic competence. This breeds a new look at grammar. The grammar of a language, as a model of idealized competence, establishes a certain relation between sound and meaning — between phonetic and semantic representations' (Ibid.: 398). The linguistic study of a given language is intended to discover that grammar which is a specific theory of that language by formulating the rules of that grammar (Chomsky 1966b: 11), whereas the general theory of linguistic structures is a specific theory of grammars intended to find out those general conditions which each such grammar must satisfy There are three possibilities (ordered here by the strength of the requirements posed to linguistic theory concerning its relation to the various grammars): (1) providing a method of discovering grammars, i.e. their structures, on the basis of respective sets of utterances (discovery procedure) ·, (2) providing a method of deciding which of the possible grammars is the best {decision procedure)·, and (3) providing a method of evaluating which of given grammars is better (evaluation procedure). Chomsky is in favor of the third possibility, which is the weakest of the three as far as research tasks are concerned, but is realistic and scientifically important (1957: 49—53). The above definitions of 'grammar' and 'competence' in transformational generative grammar result in the distinction of surface structures and deep structures in that grammar. Grammar is supposed to offer a description of those sentences which are specified by it as acceptable in the light of the rules of a given language, i.e. those sentences which are grammatically correct. But in order to make such a description it does not suffice to confine oneself to the phonetic form of a given sentence which is determined by its surface structure, since the latter does not suffice to find out certain, semantically important, grammatical relations within that sentence (these relations are termed the deep structure of that sentence). This phonetic and the semantic distinction, which strikingly resembles the concept of the double articulation of a sentence, is the foundation of transformational grammar. That grammar consists of three component parts : (1) the syntactic element (SD — syntactic description, which has a surface structure and a deep structure); (2) the semantic element, which, like Ajdukiewicz's rules of sense, assigns a semantic interpretation to a deep structure; and (3) the phonological element,

Generative grammar and the concept of innate ideas

13

which assigns a phonetic interpretation to a surface structure (Chomsky 1966b: 13—17). When analyzing the history of 'Cartesian linguistics', and especially the principles of the Port Royal Grammar, Chomsky illustrates this distinction by referring to the history of what he terms philosophical grammar (1966a: 31—42). Another important concept, already mentioned above when linguistic theory as distinct from grammar was described, is that of the universal nature of grammar. As has been said above, linguistic theory is to find out those general conditions which every grammar must satisfy. Now, following the rationalist principles of 'Cartesian linguistics', Chomsky accepts not only the claim that such a grammatica universalis does exist but also the claim that deep structures are common to all language (1966a: 35, 59 et passim). In another formulation of his theory Chomsky also postulates universal phonetics and universal semantics as parts of universal grammar (1967a: 402 ff). But the syntactic element, especially the claim about the universal nature of deep structures (surface structure being by definition that element which differentiates between the various natural languages) remains the focal factor, which is almost the most interesting one from the point of view with which we are here concerned. 'Universal grammar might be defined as the study of the conditions that must be met by the grammars of all human languages . . . so defined, universal grammar is nothing other than the theory of language structure' (Ibid.: 407). And now comes the last — last in the sense of the logical order of element of his theory — components of the set of concepts in generative grammar, the components which, in view of the goal of this analysis, must be of special interest here: AD, i.e. the acquisition device (the mechanism of assimilating a language) and AM, i.e. the acquisition model (the model of assimilation of a language). The model is a formalized representation of what actually takes place when a person assimilates a language by means of the acquisiton device. In order clearly to visualize the problem and Chomsky's standpoint on this important issue let us quote his words from one of his later, synthesizing works: 'The process of abstraction can be carried one step further. Consider an acquisition model AM that uses linguistic data to discover the grammar of the language to which this data pertains. Linguistic data -> AM -»· Grammar Just how the device AM selects a grammar will be determined by its

14

Adam

Schaff

internal structure, by the methods of analysis available to it, and the initial constraints that it imposes on any possible grammar. If we are given information about the pairing of linguistic data and grammars, we may try to determine the nature of the device AM. Although these are not the terms that have been used, linguistics has always been concerned with this question. Thus modern structural linguistics has attempted to develop methods of analysis of a general nature, independent of any particular language, and an older and now largely forgotten tradition attempted to develop a system of universal constraints that any grammar must meet. We might describe both'these attempts as concerned with the internal structure of the device AM, with the innate conception 'human language' that makes language acquisition possible' (1967a: 401; my italics). At the end of the same paragraph Chomsky refers the reader to a footnote which in view of its significance also deserves being quoted. 'The existence of innate mental structure is, obviously, not a matter of controversy. What we may question is just what it is and to what extent it is specific to language' {Ibid.: 439). We have thus come to face the main subject matter of our analysis: the problem of innate ideas, or innate structure, on which the hypothetical deductive system of generative grammar is based. Before proceeding critically to analyze the appropriate claims of the founders and supporters of transformational generative grammar we shall try to reconstruct them with all possible precision and to make clear what place they hold in the ideas of representative of generative grammar.

2. GENERATIVE GRAMMAR A N D NEONATIVISM

What has been termed above an analysis of the set of concepts used in generative grammar is in fact a tentative reconstruction of the inner logic of that theory. Of course, this has been done with a definite purpose, which is determined by the system of reference of the concepts under consideration and their logical ordering. Chomsky proceeds in a similar manner by analyzing the inner logic of the concepts used in generative grammar when the problem of linguistic competence is adopted as the starting point (1966b: 18-19).

Generative grammar and the concept of innate ideas

15

It is important to realize that in the conceptual system of generative grammar the idea of innate structures, inherent in the acquisition device of a given language, i.e. that 'device' or 'black box' which at the input has acts of speech and at the output has the generative grammar of that language, is not accidental, but necessary. Without that idea it would not be possible to substantiate the statement on the universal nature of grammar, and further, if we go in the reverse direction as compared with the analysis made previously, the universal nature of deep structures, the specific interpretation of linguistic competence, and finally, the specific concept of grammar and that of linguistic theory. As has been said, generative grammar is constructed as a hypothetical deductive model, and its structure resembles constructs in mathematical logic rather than traditional linguistic models. In the dedecutive systems of mathematics and mathematical logic a given calculus is based on the adopted axioms and transformation rules. An axiom is an assumption which is one of the starting points of a given calculus. The statement on innate linguistic structures whose nature is that of innate ideas, is an assumption and function are an axiom. Such at least is the logical structure of the system. The belief held by the founders of generative grammar that this is not an assumption but a theorem which they try to prove, and the actual role of that statement in the logical structure of the system are two different things. It must be emphasized here that the claim that linguistic universals exist is a necessary component of generative grammar in the same way as the claim that linguistic differentials exist is a necessary component of the Sapir-Whorf hypothesis, i.e. the school of linguistic relativism. In view of the importance of this statement in the system of generative grammar and because of our interest in that statement, it is imperative first to reconstruct precisely the opinions of the founders of generative grammar on the issue. This is even more important because Chomsky in his successive works reverts to the basic concepts of his theory. We shall accordingly begin with what, in my opinion, are Chomsky's basic formulations on the subject of innate ideas; in doing so we shall follow the chronological order of the publication of his works. We begin with the review of B. F. Skinner's work, Verbal Behavior, published in 1957. In his review (1964b) Chomsky sharply criticized Skinner's attempt to interpret behavioristically verbal behavior and in particular the process of learning a language, and in this

16

Adam

Schaff

connection advanced his idea of LD (learning device), which he later came to term AD (acquisition device). In discussing Skinner's method of analysis Chomsky is alarmed not only by Skinner's confinement to the observation of input and output data, but above all by the fact that analysis is restricted to external factors without consideration of the inner structure of human organism and the way in which that organism processes the information received. 'These characteristics of the organism' Chomsky says, 'are in general a complicated product of inborn structure, the genetically determined course of maturation and past experience' (1964B: 548). Chomsky is still cautious in advancing his concept of innate structures. He does not reject the statements on imitation, on the reinforcement of certain characteristics by repetition, etc., in the process of learning in animals and in children, but he adds that there are indications that observable responses to sign stimuli 'are genetically determined and mature without learning'; that in the case of a child who begins to speak we cannot exclude the possibility that his ability to select from among the sounds he has heard, those which are phonologically important and may develop largely independently of reinforcement, through genetically determined maturation: and that it cannot be excluded that the development of the brain accounts for the fact that at a certain age a child produces the rules of the grammar of a given language from the sentences he has heard in that language (Ibid. ·. 563—564). Finally, in opposing the theory that a person understands new sentences by comparing them with previously acquired patterns, Chomsky advances his concept of generative grammar. But in this, too, he is still extremely cautious. In order to formulate meaningful sentences and in order to understand them one has to assimilate the grammar of the language in question. How this is done, and, especially, how this is done by a child, who in a short time achieves what in the case of a theorist requires much work, and that independently of that child's intelligence, is still not known. But he concludes: 'The fact that all normal children acquire essentially comparable grammars of great complexity with remarkable rapidity suggests that human beings are somehow specially designed to do this . . . The study of linguistic structure may ultimately lead to some significant insights into this matter. At the moment the question cannot be seriously posed, but in principle it may be possible to study the problem of determining what the built-in structure of an information-processing

Generative grammar and the concept of innate ideas

17

(hypothesis-forming) system must be to be enable it to arrive at the grammar of a language from the available data in the available time' (Ibid.·. 5 7 7 - 5 7 8 ) . Thus, so far we see cautious hypotheses and no less cautious postulation of a program of research. The situation then changes rapidly, and that not on the strength of any new results obtained in the course of the postulated research, but rather in the sphere of philosophical reflection which, through the assimilation of rationalist principles, provides a convenient basis for the construction of a model of generative grammar as the universal grammar. Chomsky's successive works not only show the growing conviction that the concept of innate structures is good for the interpretation of linguistic facts, but also reveal ever stronger reliance on inspiration drawn from philosophical sources. In his paper 'Explanatory models in linguistics' (1962) we find, next to developed and more vigorously formulated opinions on grammar as linguistic theory, a more radical opinion on the learning device which, it is now assumed, must include a theory of linguistic structure. Ά general theory of linguistic structure of the sort just outlined would, in this way, provide an account of a hypothetical language-learning device and could thus be regarded as a theoretical model of the intellectual abilities that the child brings to language learning' (1962: 535). Since the statement that the child contributes his intellectual abilities to the process of language learning is vague and might be interpreted in various ways, another explanation follows, which contains in nuce the formulation of the concept of innate structures. 'It seems to me that the relative suddenness, uniformity and universality of language learning, the bewildering complexity of the resulting skills, and the subtlety and finesse with which they are exercised, all point to the conclusion that a primary and essential factor is the contribution of an organism with highly intricate and specific initial structure' {Ibid.: 536; my italics). This is a brief formulation not only of the concept which will often return in Chomsky's later works, but also the indication of the facts which induced Chomsky to accept it. In his paper read at the Ninth International Congress of Linguists (in extended form published as Current issues in linguistic theory, 1969a). Chomsky presented a more radical version of that concept and referred, this time in an expanded form, to W. von Humboldt's philosophy of language, in particular von Humboldt's well-known

18

Adam

Schaff

idea that a language cannot be taught, but that it is merely possible to incite a person's mind spontaneously to develop his language, and also von Humboldt's opinion that mankind has only one language (1969a: 17 ff). Chomsky's standpoint on the learning device is already precisely formulated. 'The learning model Β is a device which constructs a theory G (i.e., a generative grammar G of a certain langue) as its output, on the basis of primary linguistic data (e.g., specimens of parole), as input. To perform this task, it utilizes its given faculte de langage, its innate specification of certain heuristic procedures and certain built-in constraints on the character of the task to be performed. We can think of general linguistic theory as an attempt to specify the character of the device B' {Ibid. ·. 26; my italics). In his paper under consideration, which was clearly conceived as a program of research, Chomsky formulated his opinion that the limited nature of the taxonomic model is linked with a narrowed concept of the nature of cognitive processes in man, and that the return to traditional views, made more precise by methods used in modern linguistics, would broaden our views on learning processes in man {Ibid.: 27). This program has been carried out in his Cartesian linguistics (1966a). We disregard here the historical data quoted in Cartesian linguistics: It is obvious that in the works of rationalists we can find support for rationalist ideas (even though Chomsky has found in their works, especially in the Port Royal Grammar, ideas that came to be forerunners of certain technical aspects of the methods used in generative grammar). What is more important and more valuable for us is the use made by Chomsky of those historical data, the way he generalizes them to use them as evidence for his statements. In Chomsky's opinion, the focal doctrine of Cartesian linguistics is that which says that general characteristics of grammatical structures are common to all languages. Thus there are certain grammatical universals which restrict the variety of natural languages, and the study of these universals is what is termed grammaire gemrale. These universals cannot be learned. 'By attributing such principles to the mind, as an innate property, it becomes possible to account for the quite obvious fact that the speaker of a language knows a great deal that he has not learned' (1966a: 60). Likewise, an analysis of the views held by Leibniz, Schlegel, and von Humboldt is an opportunity not only for stating that according to them language capacity is innate in man, but also for formulating

Generative grammar and the concept of innate ideas

19

the opinion, for instance, of von Humboldt so that it can be incorporated, lock, stock, and barrel, in the system of generative grammar. 'In short, language acquisition is a matter of growth and maturation of relatively fixed capacities, under appropriate external conditions' {Ibid.: 64). What is at stake here is not so much the opinions of past thinkers, but what is being read into them and what is the context in which quotations from their works are located. That this is so can best be seen in the case of von Humboldt, in whose works one can find, as it turns out, what one looks for. It is common knowledge that the school of linguistic relativism, which is antipodally opposite to the universalism of generative grammar, has been drawing inspiration, and rightly so, precisely from von Humboldt's works, and is even commonly known as Neo-Humboldtianism. These excursions into the history of philosophy prove beyond doubt that ideas born in speculative philosophy can work as mental stimuli in the development of science as they have a heuristic value in that respect. To state this is certainly pleasing to a philosopher who in spite of himself may feel frustrated under the impact of positivist trends in modern science. It is also an excellent illustration of the migration of ideas in time: theories and ideas which at one time were en vogue and were later abandoned, apparently for good, come back to the arena of science when new discoveries place them in a new intellectual context and bring out such aspects of these ideas which were either unnoticed or eclipsed by other factors. This applies to nativism, which explains our interest in the issue. Cartesian linguistics was published in 1966, and the paper on 'The formal nature of language' (included as an appendix to a book by Eric Lenneberg) appeared one year later. This paper sums up succinctly the theory of generative grammar and the results obtained until then by the application of that method. The problem of learning device recurs inevitably, this time approached from a different view point as AM (acquisition model). Chomsky states that the trend, common in modern structural linguistics, to develop methods of linguistic analysis that are general enough to be independent of any single natural language find support in a much older tradition, and concludes: 'We might describe both these attempts as concerned with the internal structure of the device AM, with the innate conception of 'human language' that makes language acquisition possible' (1966a: 401). When referring later, especially in his phonological considerations, to the issue of that device which makes it possible to acquire a

20

Adam

Schaff

language by means of innate structures, Chomsky formulated an idea which is worth quoting, since it contains an attempt to substantiate the concept of innate ideas, the concept to which we shall have to revert later. In describing the phonological component of a universal grammar Chomsky says that the assumption made concerning a universal grammar restricts the class of possible human languages to a special subclass of conceivable languages, and goes on as follows: 'The evidence available to us suggests that these assumptions pertain to the language acquisition device AM [. . .]; that is, that· they form one part of the schematism that the child brings to the problem

of language learning. That this schematism must be quite elaborate and highly restrictive seems fairly obvious. If it were not, language acquisition, within the empirically known limits of time, access and variability, would be an impenetrable mystery. Considerations of the sort mentioned in the foregoing discussion are directly relevant to the problem of determining the nature of these innate mechanisms, and, therefore, deserve extremely careful study and attention' {Ibid.: 4 1 6 ; my italics). The standpoint has thus been described with full clarity, but we still can have doubts as to how these 'innate mechanisms', 'innate schematisms', and 'innate structures' are to be interpreted, the more so as in all the cases quoted above the statements in question are marginal to more or less specific issues of generative grammar, which are the main subject of the papers quoted. Another difficulty is that in his works marked by precision of formulations that comes close to mathematical logic, Chomsky leaves a wide margin of obscurity and ambiguity when it comes to such expressions as 'acquisition device' and all expressions in which the term 'innate' occurs (i.e. innate structures, mechanisms, schematisms, etc.). Assistance inthis field is offered by Chomsky's book which carries his somewhat popular lectures; this rather popular approach makes Chomsky formulate his intentions with precision, also when it comes to marginal or vague concepts and ideas introduced in his other works, especially those ideas which are clearly philosophical in nature. The book in question is Language and mind, published in 1968 and thus chronologically later than the works quoted here previously. We do not mean in this connection his historical comments on philosophical grammar and its rationalist standpoint, an issue known from his earlier works. The most important is Chomsky's exposition of his own standpoint, which includes a commentary to his earlier

Generative grammar and the concept

of innate

ideas

21

argumentation. The most significant and illuminating in this respect is Lecture Three, in which the generative grammar approach is explicitly defined as a specific return to rationalism: 'Because to a significant extent it is the traditional rationalist approach, now amplified and sharpened and made far more explicit in terms of the tentative conclusions that have been reached in the recent study of linguistic competence' (1968: 70). At this point we are concerned only with a faithful reconstruction of Chomsky's opinion on innate ideas, in order to avoid criticism of their misrepresentation. From this point of view it is inessential what Locke and Descartes really thought on the subject; this problem can be left for historians of philosophy. The important issue is what Chomsky thinks on the issue. In this connection his declaration in favor of genetic rationalism is significant, even more so since — as a lawyer would say — we have to do with an authentic interpretation of earlier statements, which were not at all clear. It is only in the light of this declaration in favor of rationalism that the formulation of the problem, in which the existence of innate structures is adopted as an assumption, becomes comprehensible. 'Insofar as we have a tentative first approximation to a generative grammar for some language, we can for the first time formulate in a useful way the problem of origin of knowledge. In other words, we can ask the question, What initial structure must be attributed to the mind that enables it to construct such a grammar from the data of sense?' (Ibid. : 68, my italics). It is important, for the further analysis of the problem, to establish that the very fact of the existence of such structures (which, as we have seen above, are intricate, as they contain grammar in the sense of linguistic theory) is beyond dispute: it is adopted as an assumption. According to Chomsky, it is the limits and the properties of those structures which are disputable and require empirical criteria. But, for all the explicit assurances that these are empirically decidable problems (Ibid.·. 69), this is not always clear. This is so because we are told that: a) such an innate structure is a generic characteristic and is independent of individual intelligence, since those grammars which are in fact constructed by the speakers of a language do not differ much from one another (Ibid. ·. 68—69); b) the claim that these structures are universal applies also to the dialects of one and the same language, and even to different lan-

22

Adam Schaff

guages which do not reveal any affinity (Ibid.: 69); c) 'we must postulate an innate structure that is rich enough to account for the disparity between experience and knowledge, one that can account for the construction of the empirically justified generative χ grammars within the given limitations of time and access to data' (Ibid.·. 69); and d) such a postulated structure should be neither too comprehensive nor too restrictive, in order not to exclude any language: 'There is, in other words, an upper bound and a lower bound on the degree and exact character of the complexity that can be postulated as innate mental structure' (Ibid.: 69). We shall still revert to these issues. At this moment it is to be noted merely that when reference is made to 'empirical conditions', what is meant in fact are certain findings or postulates about that 'innate structure', but not its existence, which has been assumed (obviously in the light of certain premises) and is not subject to dispute. The fact that the concept of innate structure has been adopted as an assumption of the system is due to the conviction that the rationalist conception is the only acceptable one when it comes to the analysis of the problem of language acquisition (Ibid.: 76). Such being the case, empirical research can pertain only to the properties of that structure (for instance, the psychologists are assigned the task 'to discover the innate scheme that characterizes the class of potential languages — that defines the 'essence' of human language') (Ibid.: 76). This tour of Noam Chomsky's works has enabled us to establish beyond dispute, as it seems, what he means by his access to genetic rationalists and what role is played-by the concept of 'innate structure' in the system of generative grammar. We have traced the history of the conception of innate ideas as related to generative grammar, first, in order to show that it is an element in the system of that grammar; second, to avoid a possible objection of being inexact in reporting on the opinions which will be the subject matter of a dispute; and third, because this aspect of the issue seems to have been insufficiently treated in scholarly literature. For the same reasons, reference will here be made to two more works, one of which has been written by Jerrold J. Katz (1966), a member of the school of generative grammar, so that his opinions on the concept of innate ideas may be treated as representative of that school. The other work comes from Eric H. Lenneberg (1967), an author who is closely linked with that way of thinking and who

Generative grammar and the concept of innate ideas

23

follows a similar path of reasoning in the study of the biological foundations of language. As the title of his book shows, Katz has written a study on the philosophy of language based on the concept of generative grammar. As for the latter aspect of the issue he does not contribute anything new, and the originality of his book consists in the taking up of certain classical philosophical problems and their tentative solution in the light of linguistic analysis conceived in a specific manner. But this is of less interest here, and hence we shall confine ourselves to that aspect of his work which is concerned with the issue of innate ideas. Katz is right in linking this problem with the old controversy between the empiricists and the rationalists. He is of the opinion that in this case the key issue is the role of innate ideas in the process of language acquisition; he thinks that the existing body of knowledge in linguistic theory makes it possible to settle the dispute between the empiricists and the rationalists. The fundamental problem, as Katz claims in accordance with the guiding principles of the system of generative grammar, is the birth of linguistic competence as the child develops from infancy to the moment when he begins to speak a given language fluently (i.e. when under the impact of utterances he has heard in a given language he assimilates the rules of constructing, using, and understanding sentences in that language). The criterion for settling the dispute between the two rival hypotheses would refer to the effectiveness of either hypothesis in solving the problem of the origin of that competence: 'Whichever hypothesis thus provides the most fruitful model of the internal structure of the language acquisition device will be accepted as the best hypothesis' (1966: 247). The answer is as follows: the empiricist hypothesis is based, in principle, on an inductive generalization, whereas the rationalist one states that 'the language acquisition device contains a stock of innate, ideas that jointly specify the necessary form of language (realized in any actual natural language) and thus the necessary form of a speaker's internal representation of the rules of languages' {Ibid.: 247—248). But since no rationalist has ever explicitly formulated the concept of innate ideas it would be difficult to speak about any specified rationalist hypothesis on this issue. Now this has been done by the theory which underlies the foundations of generative grammar. The necessarily strong hypothesis states that 'the language acquisition device contains, as innate structure,

24

Adam

Schaff

each of the principles stated within the theory of language' {Ibid.·. 269). The list of these principles is imposing: language universals which determine the form of linguistic description; the form of the components (phonological, syntactic, and semantic) of that description; the formal nature of each of these components; the set of universal constructs which underlie the laws of the various descriptions; and the methodology of the choice of optimum linguistic descriptions. But it turns out that this is not enough, that we have to do with an open hypothesis in the sense 'that it asserts that anything which, for good empirical reasons, is found to be part of the theory of language is eo ipso facto part of the language acquisition device, and hence part of the child's native language-apparatus, his innate ideas about language' {Ibid.: 269—270). To put it briefly, all that which is connected with learning, using, and understanding language is innate. This is not an exaggeration: Katz really thinks so and adds that innate ideas are part of the system which organizes experience, a system whose existence 'has been hypothetically inferred from the linguistic performance of speakers intheir acquisition and use of language' {Ibid.·. 270). Now it just remains to state that the rationalist hypothesis enables us to explain why every natural language has the properties ascribed to it by the laws of the theory of language (this could not be otherwise, since this has been an assumption); and even more, why not only those languages which have been studied so far, but also those which still be studied in the future {sic) {Ibid.: 270) comply with those laws of the theory of language. We now proceed to report on the final, synthetic, conclusions to be drawn from this reasoning. As compared with Chomsky's statements quoted above they are more pointed: They state more fully and more clearly the consequences of the ideas to be found in Chomsky's works in an abbreviated form. But this is just the value of these conclusions, since their nature will help us later to reflect on Chomsky's claims. The objection that this is a radicalization typical of philosophical discussions, whereas Chomsky is concerned with constructing a linguistic theory, even if it be based on certain philosophical premises, can be warded off by the explanation that we have to do with the work of one of Chomsky's collaborators which thus has its origin in the same intellectual circle and has never met with any reservation, as to the formulation of the issues under consideration here, on the part of other representatives of the school of generative grammar. Katz's final reasoning on the issue of innate ideas in the sphere of

Generative grammar and the concept of innate ideas

25

language is as follows. It may not be assumed that the concurrence of certain properties in all languages, i.e. the universal nature of that properties, is just accidental. This universality must be explained by the indication of its cause. Since we have to do with specifically linguistic laws we have to disregard in the explanation such invariants which are connected with the purely biological community of the species Homo sapiens (e.g. the fact that all human beings live on the surface of the earth, that they breathe, etc.). We also have to exclude the geographical, psychological, cultural, and sociological factors, since these are variable, and as such cannot explain the occurrence of invariant characteristics of language. What then? And here we find the passage which offers Katz's answer to the question and which is so significant for further discussion that it is to be quoted in extenso to ward off any possible objections that Katz's views are described inaccurately, or perhaps even contrary to his intentions. We have here to do with a statement which with exceptional clarity, and dotting all the i's, expounds the viewpoint of the school of generative grammar on the problem under consideration. 'Hence, by this process of elimination, the only thing left that can provide the invariant condition that we want to connect with the universal features of language as their causal antecedent is the common innate endowment of human language learners, i.e. some component of their specifically human nature [. . .] it is also the differentia between language at the human level and its absence at the lower levels of the animal kingdom. [. . .]' 'Now, if we bring in the rationalist hypothesis, we can obtain an explanation of this regularity. This hypothesis says that part of the generic endowment of a human being is the full set of linguistic universals /I/—/V/ and that these universals constitute the internal structure of the device which the nonverbal infant utilizes to become a fluent speaker on exposure to a sample of the utterances of a language. Since the innately given language acquisition device incor-* porates the linguistic universals, we can explain why the linguistic universals are necessary features of any language that is spoken by a human being. Namely the linguistic universals are found in each and every language because, in acquiring a language, they are emplanted in the speaker's internalization of the rules of his language by the device that accomplishes its acquisition. The very mechanism which the child uses to acquire fluency in a natural language introduces them as the framework in which his linguistic experience is organized in the form of linguistic rules' (Ibid.: 273—274).

26

Adam Schaff

This passage is the quintessence of the issue under consideration as seen by the school of generative grammar. We shall revert to it later when we come to criticize the standpoint described here with such clarity. At that point the continuation of this passage, in which Katz opposes the expected objection that this explanation is a petitio principii, will be quoted, too. That continuation is no less interesting for the problem being analyzed than what has been quoted above, but since for the time being we confine ourselves to a presentation of views, we only mention its existence. To conclude the review of the opinions of the representatives of the school of generative grammar let us have a closer look at Eric Lenneberg's book which has been mentioned above. Its title, Biological foundations of language (1967), reflects well the goal and the approach of its author. Lenneberg is much more cautious in his formulations than are the above-mentioned promoters of generative grammar, but he comes close to their reasoning. It was not accidental that Chomsky joined his synthetic exposition of the doctrine of generative grammar (cf. 'The formal nature of language' as referred to above) as an appendix to Lenneberg's book. He apparently thought, and rightly so, that the book represented a trend which is at least convergent with his own. This is why, to make the picture complete, the book must be analyzed here. Lenneberg says, in fact, the same that Chomsky and Katz say on the issue under consideration, but he does not say it in the same manner. Chomsky constructs a deductive model which, among other things, is to explain what he terms language acquisition; in his model the assumption that the rules of generative grammar are innate in man contributes to a mathematical elegance in the solution of the problem, and substantiates the claim that the transformational grammar which is based on those rules is truly universal. Lenneberg approaches the issue at a different angle, which changes his viewpoint and also affects his specific fagon de parier: He speaks as a biologist. But the basic question is the same: Why is language ability a generic trait of Homo sapiens? And the answer is the same: this is so because it is an innate ability. What is more, Lenneberg refers to the biological community of the human race and says that deep structures (which he terms latent structures) are universal in nature and that linguistic structures, in fact identical with generative grammar, are innate. Let us begin with the conclusions which are pertinent to our issue, to be found on the last pages of Lenneberg's book.

Generative grammar and the concept of innate ideas

27

At one time he says in the last section entitled 'Innate mechanisms', the concept of innateness was banned and believed to be unscientific. This in now over. Biology now is doing nothing else than investigating the innate natural of various forms. 'The discovery and description of innate mechanisms is a thoroughly empirical procedure and is an integral part of modern scientific inquiry' (1967: 393). This applies to language, too. It is only owing to the rules which underlie syntax that we obtain results which are in agreement with logic by having at our disposal sentences as the initial data of mental operations. This proves, in Lenneberg's eyes, that these rules are an innate property of human mind. 'When we say rules must have been built into the grammatical analyzer, we impute the existence of an apparatus with specific structural properties or, in other words, a specific internal organization' {Ibid.). Attention is drawn first to these conclusions, which show a striking resemblance to Chomsky's views, for the following reasons: (1) they reveal an aspect of the problem which is different from those discussed so far and is important for later analysis; and (2) it is easier to reconstruct Lenneberg's views on the issue under consideration by basing them on these final conclusions. This will make it easier to bring out both what is common to the two approaches and what is specific to Lenneberg's standpoint. When investigating the problems of language and speech Lenneberg takes up the biological aspect of the issue and consequently, which is self-evident, is interested mainly in the role of human organism in the development of speech. This is not only admissible, but necessary if that intricate process is to be studied. Lenneberg can also refer to the fact that he pays due attention to the role of the social factor in the development of speech (which he doeS mainly in chapter 9). Thus when he emphasizes the importance of the process of maturation and the fact that certain functions of human organism are genetically conditioned, this is substantiated by results of biological research and is subject only to experimental verification of refutation when it comes to particular statements. Let us bring out a comment made by Lenneberg in connection with what he says on the role of the maturation of human organism in the development of its specified functions, a comment which sheds bright light on the substantiation of his statements on innate language structures. He strongly emphasizes the difference between the potentiality of certain behavior in the process of the maturation of human organism, and a given act of such behavior.

28

Adam

Schaff

'The aim of these comments is to direct attention to potentialities of behavior — the underlying matrix of behaving — instead of to a specific act. If we find that emergence of a certain behavior may be partially or wholly attributed to changes within the organism rather than to causative changes in the environment, we must at once endeavor to discover what organic changes there are. Unless we can demonstrate a somatic basis, all our speculations are useless' (Ibid.: 127; my italics). In my opinion this statement and this postulate can be applauded only if we refer to a biological interpretation of language. They will be tentatively applied below· to those statements in the theory of generative grammar which explicitly refer to such a biological interpretation (especially when it comes to the acquisition device). Unfortunately, Lenneberg himself fails to comply with his postulate, when in chapter 5, 'Neurological aspects of speech and language', he takes up the issue of innate mechanisms and thereby paves the way for the radical formulations to be found in chapter 9. Animals, he says, function like machines, and their internal structure is not a matter of chance; 'the internal structure is programmed into the ontogenetic process' (Ibid.: 220). If this internal structure is termed innate mechanism, and if the ways of functioning determined by those mechanisms are termed innate behavior, then there is nothing more trivial, from the scientific viewpoint, than the statement that they are there. So far no objection can be raised against this reasoning as long as it remains in the sphere of general considerations and as long we bear in mind the above-quoted postulate concerning the scientific meaningfulness of such hypotheses. But now we pass from the sphere of general considerations t o a very specific issue, namely that of language. 'There is, then', Lenneberg says, 'nothing unscientific about the claim that a species specific behavior pattern, such as language, may well be determined by innate mechanisms' (Ibid.: 221). By what innate mechanisms? This question suggests itself on the basis of Lenneberg's own postulates, which have such venerable predecessors as Peirce and the logical positivists who claimed (and the claim is substantiated if given an appropriate framework) that the meaning of a given statement equals the method of its verification, and hence that it is meaningful only on the condition of being verifiable. But here the answer is that 'it would be presumptuous to try to explain the nature of the innate events that control the operation of language' (Ibid.). Lenneberg mentions some neurological

Generative grammar and the concept of innate ideas

29

mechanisms which may participate in language processes, but he concludes that 'how those phenomena interact to elaborate language remains a mystery' (Ibid.). His further statement that known facts make us assume that 'there is just one peculiar mode of neural activity for aural-oral communication in man' (Ibid.) refers to a different sphere of problems and can be readily accepted in such a general formulation. But this does not refer to the problem of 'innate mechanisms', concerning which we can only according to Lenneberg's own words, form general hypotheses which, however may not be treated as verified since the problem remains a 'mystery'. And what is said next on this matter? Reference to molecular biology does not yield any spectacular results: 'Considerations of this type show that it is possible to talk about language in connection with genetics without having to make shaky assumptions about 'genes of language'. It is true that we do not know what the direct relationships are between man's complement of genes and his mode of communication, we merely wish to outline the theoretical possibilities for relating the two' (Ibid.: 244). What are the final conclusions after these cautious reservations? In chapter 9, which is a summary, we find the statement that the process of maturation results in a stage of language-readiness, which, under the impact of the environment that provides the raw material in the form of the language used by adults, develops into actuality, i.e. the command of the language. But the conclusion is surprising. 'The presence of the raw material seems to function like a releaser for the developmental language synthesizing process. The course of language-unfolding is quite strictly prescribed through the unique maturation a path traversed by cognition, and thus we may say that language-readiness is a state of latent language structure. The unfolding of language is a process of actualization in which latent structure is transformed into realized structure. The actualization of latent structure to realized structure is to give the underlying cognitively determined type a concrete form.' (Ibid.·. 375—376) And the footnote to this statement disperses all doubts as to the meaning of these formulations. 'This formulation might be regarded as the biological counterpart to what grammarians have for centuries called universal and particular grammar. Latent structure is responsible for the general type of all features of universal grammar; realized structure is responsible both for the particularities of any given statement as well as these

30

Adam Schaff

aspects that are unique to the grammar of a given natural language.' {Ibid.: 375-376) Not only does Lenneberg declare himself to be in solidarity with that type of reasoning which is characteristic of the school of generative grammar, but he also subscribes, starting from biological assumptions, to the program of universal grammar. Here again he must be quoted in full in order to disperse possible doubts that his views are distorted, the more so since to my knowledge, Lenneberg disclaims that interpretation of his statements which, in my opinion, follows from his writings quoted above. 'The language .potential and the latent structure may be assumed to be replicated in every healthy human being because they are a consequence of human-specific cognitive processes and humanspecific course of maturation. In other words, universal grammar is of a unique type, common to all man, and it is entirely the byproduct of peculiar modes of cognition based upon the biological constitution of the individual. [. . .] Because latent structure is replicated in every child and because all languages must have an inner form of identical type [....] every child may learn any language with equal ease'. {Ibid.: 377; my italics) Hence language behavior differs from other fields of culture: It is not transmitted by tradition, but all the time formed anew by autonomous individuals. Social contacts activate that process, but 'The individual is seen as functioning by virtue of his own power supply, so to speak; he constructs language by himself (provided he has the raw material to do it with) t and the natural history of his development provides for mechanisms by which he will harmonize his function with that of the other equally autonomously functioning individuals around him . . .' {Ibid.: 378) This explains the final conclusions with which we started to relate Lenneberg's views on innate mechanisms and language structures. This also concludes the review of the opinions of generative grammarians on innate ideas. The review was somewhat pedantic, but as has been mentioned above, indispensable as well, since the views which are known mostly in a summary form had to be explained in full, and since it was imperative to do everything to avoid a distortion of those views and the intentions behind them. We now proceed to analyze the discussions on the issue under consideration and to formulate our own views on this matter.

Generative grammar and the concept of innate ideas

31

3. DISCUSSIONS ON NEONATIVISM IN CONNECTION WITH GENERATIVE GRAMMAR

As has been said earlier, the concept of innate ideas in the generative grammar version has become the subject matter of numerous discussions, with two types of participants: philosophers and theoretical linguists who in principle reject Chomsky's conception of innate language structures, and biologists who are less hostile to it. Let us begin with the latter group, not only because they contribute new arguments to Chomsky's theory, but above all because the controversy under consideration can be settled by natural scientists alone. What is decisive is not whether a given theory of natural phenomena does, or does not, find a place within the framework of one's philosophical views, but how things are in fact. And the issue in question is not only par excellence in the field of natural science, but it is also empirical, if decidable at all. Note Lenneberg's word quoted above that if we refer in explaining human behavior (hence language behavior as well) to changes taking place in human organism itself (which is to cover the properties of that organism), then we have to state with what changes (and/or properties) we are concerned, since otherwise what we engage in is mere speculation. It is obvious that this requirement is intended to ensure the empirical nature of appropriate statement, and consequently to describe an appropriate crucial experiment. Chomsky, Katz, and Lenneberg frequently emphasize in their works that their claims are empirical in nature. If we ask at the outset the question whether natural science, and molecular biology in particular, can at its present stage settle the dispute under consideration, then the answer must be in the negative (see Lenneberg 1964: 76). This does not change the fact that the problem itself is empirical, since the fact that it cannot be decided by science at its given stage merely means that the development of science is still insufficient, not that the problem is necessarily empirically undecidable. If biology can not decide the controversy now, then this can not be done by present-day science. Hence the various statements may have the status of hypotheses, and this is how generative grammarians understand them. But the opinion of biologists, even if it cannot be decisive for the time being, is nevertheless of special importance in view of the significance of biology as the science which can be expected to solve the problem in the future. Reference has been made earlier to the standpoint taken on the

32

Adam Schaff

issue by such eminent biologists as Fran£ois Jacob and Jacques Monod. These two names have been selected not only because Of their scientific status, but also because both of them, and Monod in particular, have made explicit statements on the problem under consideration. But, to be more specific, we have to say that their statements are differentiated: Jacob is very restrained and practically does not express his own opinion, although he mentions the issue, whereas Monod is clearly in solidarity with the hypothesis on the existence of innate language structures. This differentiated attitude of two close collaborators, manifested in their works which have appeared almost simultaneously, is characteristic, too. Jacob's statements are cautious, to say the least. In his opinion the development of the nervous system in animals is accompanied by a weakening of the rigorous nature of heredity (1970: 337ff). The genetic program which underlies the development of every organism has two parts — the closed one which is strictly defined, and the open one which leaves to the individual some freedom of response. The closed part of the program strictly determines certain structures, functions, and properties; the open part determines only. The closed part imposes something, the open one makes a choice possible. The trend of evolution increases the importance of the open part. In the case of man the latter plays such a significant role that we might speak about 'free will'. 'Mais la souplesse n'est jamais sans limites. Meme lorsque le programme ne donne ä l'organisme qu'une capacite, celle d'appendre par example, il impose des restrictions sur ce qui peut etre appris, sur le moment ou doit avoir lieu l'apprentissage et dans quelles conditions. Le programme genetique de l'homme lui confere 1'aptitude au language. II lui donne le pouvoir d'apprendre, de comprendre, de parier n'importe quelle langue. Encore l'homme doit-il, ä une certaine etape de sa croissance, se trouver dans un milieu favorable pour que se realise cette potentialite. Passe un certain age, trop longtemps prive de discours, de soins, d'affection maternelle, l'enfant ne parlera pas. Mernes restrictions pour la memoire. [. . .] Mais cette frontiere entre la rigidite et la souplesse du programme, on ne l'a encore guere exploree {Ibid.·. 338—339). On reading this passage, which is a coherent part of a reasoning, we may not conclude that there are any leanings towards the idea of innate language structures. On the contrary, we have to link that issue with the open part of the genetic program, to use Jacob's terminology. Jacob declares himself firmly in favor of the claim that

Generative grammar and the concept of innate ideas

33

aptitude au langage (faculte de langage in de Saussure's terminology) as a potentiality of learning and understanding any language is genetically conditioned. No one can reasonably deny this, as no one can reasonably deny that a healthy child has an innate faculty of manual operations which makes it possible for him, for example, tc play tennis. But although all analogies are risky and the differences between these two examples are considerable, the gap between the statement on an innate aptitude au langage and that on man's innate generative grammar is similar, though not so big, to the gap between the statement on man's innate manual abilities and that on man's innate structures of playing tennis. Jacob goes on to analyze the binding force of the genetic code in the case of human behavior. As living beings liberate themselves from the rigid nature of the program imposed by the genetic code Jacob sees the role of symbols — when living beings begin to use signs — as filters between the organism and its surrounding. This leads to situations in which integration on the level of a single organism and on that of relations between organisms is based not on an interaction of molecules, but on the exchange of coded information. According to Jacob, culture is the second genetic system, which rises above heredity, and this is why the code of these now integrating units transcends the limits of schemata of biological explanation. Hence the study of man and society is not reducible to biology, even though it cannot do without it {Ibid.: 340—343). And at this point Jacob reverts to the issue we are concerned with. 'De tous les organismes, c'est l'homme qui possede le programme genetique le plus ouvert, le plus souple. Mais ou s'arrete la souplesse? Quelle est la part du comportement prescrite par les genes? A quelles contraintes de l'heredite l'esprit humain est-il soumis? De toute evidence, de telles contraintes existaient a certains niveaux. Mais oil tracer la limite? Pour la linguistique moderne, il y a une grammaire de base, comme a toutes les langages: cette uniformite refleterait un cadre impose par l'her0dite a I'organisation du cerveau. [. . .] Mais alors, quelle est la rigidite de ce cadre? Quelles sont les restrictions imposees a la plasticite de l'esprit humain par le programme genetique?' {Ibid.: 343; my italics). Here again the formulations are very cautious; they are more like questions which imply that they cannot be answered rather than standpoints that could be interpreted as answers. Let us not be misled by the declarative form of the statement on modern linguistics (which clearly alludes to the school of generative grammar). Jacob

34

Adam Schaff

merely quotes opinions of linguists, as he later quotes neurophysiologists who think that dreams are an innate feature, or anthropologists who think that aggressiveness is such a feature. Jacob does not take any standpoint on those issues, except for the question with which he concludes his analysis and which reflects his conviction that for the time being science cannot provide an answer. It is otherwise in the case of Jacques Monod. His book Le basard et la necessite is in the border area between biology and philosophy, and if we are to judge from his manner of writing and his readiness to risk generalizations, it is rather a philosophical work. This is certainly one of the most philosophically stimulating books which I have read for years. Monod does not share Jacob's cautious formulations and firmly backs Chomsky's idea of innate structure. But, while being so firm, he always adds a formula like 'in my opinion' or an equivalent one, which shows clearly that what he has in mind is not a theoretical statement established in modern biology, but a hypothesis toward which he inclines, of course on the basis of biological data. The difference is obvious and remarkable, even if we fully respect Monod's intuition in that field. To begin with the fundamental issue, namely the acceptance of Chomsky's claim, let us see whether, and how far, Monod is in its favor. 'On sait que, selon Chomsky et son ecole, sous l'extreme diversite des langues humaines, l'analyse linguistique en profondeur revele une "forme" commune a toutes les langues. Cette forme doit done, d'apres Chomsky, etre consideree comme innee et caracteristique de l'espece. Cette conception a scandalise certains philosophes ou anthropologistes qui y voient un retour a la metaphysique cartesienne. A condition d'en accenter le contenu biologique implicite, cette conception ne me choque nullement. Elle me parait naturelle au contraire, des lors qu'on admet que revolution des structures corticales de l'homme n'a pu manquer d'etre influencee, pour une part importante, par une capacite linguistique tres tot acquise a I'etat le plus fruste. Ce qui revient a admettre que le langage articule, lors de son apparition dans la lignee humaine, n'a pas seulement permis revolution de la culture, mais a contribue de fa$on decisive a revolution physique de l'homme. S'il on a bien ete ainsi, la capacite linguistique qui se revele au cours du developpement epigenetique du cerveau fait aujourd'hui partie de la "nature humaine" eile meme definie au sein du genome dans le langage radicalement different du code genetique' (1970: 150—151).

Generative grammar and the concept of innate ideas

35

On another occasion Monod makes a still stronger statement on the same subject. 'Les decouvertes modernes [reference is made' here to the role of sensory analysers in spatial perception — A.S.] donnent done raison, en un sense nouveau, a Descartes et a Kant, contre l'empirisme radical qui cependant n'a guere cesse de regner dans la science depuis deux cent ans, jetant la suspicion sur toute hypothese supposant Γ "inneite" des cadres de la connaissance. De nos jours encore certains ethologistes paraissent attaches a l'idee que les elements du comportement, chez Panimal, sont ou bien innes ou bien appris, chacun de ces deux modes excluant absolument l'autre. Cette conception est entierement erronee comme Lorenz l'a vigoureusement demontre. Lorsque le comportement implique des elements acquis par I'experience, ils le sont selon un programme qui, lui, est inne, e'est-a-dire genetiquement determine. La structure du programme appelle et guide I'apprentissage qui s'inscrira done dans une certaine "forme" preetablie, definie dans le patrimoine genetique de I'espece. C'est sans doute ainsi qu'il faut interpreter le processus d'apprentissage primaire du langage chez l'enfant' {Ibid.: 167—168; my italics). As a biologist — the point to be emphasized if the picture is to be complete — Monod dissociates himself from rationalism by stressing that genetically all that which is to be found in living beings, including innate mechanisms (from instinctive behavior of the bee to the innate structures of human cognition) originates from phylogenetic, and not ontogenetic, experience {Ibid.: 169). Even with this reservation in favor of empiricism, Monod's statements fully support Chomsky's claim, with the provision, however, as made above, that this is merely his agreement to accept Chomsky's hypothesis, but in no way a statement that we have to do with a scientifically verified thesis. Moreover, which is not insignificant, Monod refers to an unspecified 'form' (which is a Humboldtian term rather than that of contemporary generative grammarians) without taking a stand on the issue of 'linguistic universals', which are said to be part of the 'acquisition device' innate in man. This problem has been taken up by Roman Jakobson, who in his synthetic study 'Linguistics' (1970) also tackles the issue of the relationship between linguistics and biology. Even though he firmly dissociates himself from the biological trend in the interpretation of language phenomena, he advances, which may seem paradoxical, the strongest (in my opinion) arguments in favor of the genetic endowment of man as far as language is concerned.

36

Adam

Schaff

Jakobson says that over the recent decades various universale have been discovered in phonology and grammatical patterns; it has also been found that the child has innate dispositions to learn any language, but that it would be pure speculation to infer from these dispositions .that any innate linguistic universals exist. To explain those universals which are known it suffices to refer to the inner logic of linguistic structures and there is no need to argue about 'genetic instructions'. Argument is also denied by the existence of syntactic structures which are unknown in certain languages, the existence of argots, individuel poetic languages, etc. (Ibid.: 438). While firmly emphasizing the societal aspect of language learning in the child, opposing the claim that the language of adults is just the raw material for the child, and stating that all available data refute the statement that there is no need to teach the child a given language, Jakobson proceeds to discuss the problem of man's genetic endowment when it comes to the foundations of language. And now he, an opponent of biologism in the study of language, raises issues which are not philosophical speculations, but refer to facts that make us analyze in depth the correlation between the language of the genetic code and human language. It turns out that the language of the genetic code (genetic information as transmitted in the four-element language of nucleic acid molecules) is analogical in its structure to human language. The chemical information which determines heredity is recorded along the chromosomes in an alphabet which resembles Morse's alphabet. Since the letters of an alphabet merely are substitutes for phonemes, and Morse's alphabet is secondary with respect to the letters of the ordinary alphabet, the elements of the genetic code may be compared to phonemes. This results in the significant generalization. 'We may state that among all information-carrying systems, the genetic code and the verbal code are the only ones based upon the use of discrete components which, by themselves, are devoid of inherent meaning but serve to constitute the minimal senseful units, i.e. entities endowed with their own, intrinsic meaning in the given code' (Ibid. ·. 438). The second striking similarity between the two information systems is that they use binary oppositions between their basic elements (phonemes in one case, and the four elements of the nucleic code in the other). Third, in both cases we have to do with a hierarchical design as the principle of integration of both verbal and genetic information.

Generative grammar and the concept of innate ideas

37

Fourth, there is a strict colinearity in the temporal sequence of the encoding and the decoding of both languages. And here is Jakobson's conclusion: 'How should one interpret all these salient homologies between the genetic code which "appears to be essentially the same in all organisms" and the architectonic model underlying the verbal codes of all human languages and, nota bene, shared by no semiotic systems other than natural language or its substitutes? The question of these isomorphic features becomes particularly instructive when we realize that they find no analogue in any system of animal communication [. . .]. Now, since "heredity, itself, is fundamentally a form of communication", and since the universal architectonic design of the verbal code is undoubtedly a molecular endowment of every Homo sapiens, one could venture the legitimate question whether the isomorphism exhibited by these two different codes, genetic and verbal, results from a mere convergence induced by similar needs, or whether, perhaps, the foundations of the overt linguistic patterns superimposed upon molecular communication have been modeled directly upon its structural principles' (Ibid.·. 440, my italics). This is a question of great importance, which shows that we can move far along the path of speculation concerning man's genetic endowment in the sphere of linguistic structures. Here, too, the question has merely been posed, but its underlying evidence makes us reflect on a more general hypothesis. What are the results of those considerations on which the school of generative grammar could and should have counted most? To put it briefly, it has been shown that the claim concerning the existence of innate linguistic structures remains a hypothesis, and that for the time being there are neither possibilities nor grounds for admitting that hypothesis as an accepted theory. Hence the important conclusion for the generative grammarians that their hypothesis has not been disproven, and hence cannot be just dismissed from further considerations; but hence the no less important conclusion that this is a hypothesis which, for the time being, cannot be verified empirically from the biological point of view (and no one has so far taken the trouble to verify it in other ways to be mentioned below). This gives rise to some polemic remarks in connection with Katz's statement quoted above, and to some general comments which are marginal to the main issue under consideration but are worth making in view of their philosophical significance.

38

Adam

Schaff

Katz's statement, quoted above in full, sums up — in the radical version — considerations on innate linguistic structures. Concerning the problem of how a child assimilates a given language Katz declares himself for the rationalist hypothesis, and his answer is: He does so by means of an innate acquisition device which contains linguistic universals that occur in all languages. The weak point of his alleged proof lies in the fact that what has to be proved (this is the only way of settling the controversy between the rationalists and the empiricists, as Katz suggests in his book) has been here adopted as an assumption. In fact, the rationalist hypothesis offers a simple solution of an intricate problem: all languages have a c o m m o n stock of linguistic universals (quite comprehensive at that, as it covers all rules of generative grammar), and this is due to the fact that those h u m a n beings w h o assimilate those languages have these universals as their innate features. Hence we can assimilate any language, since it includes the stock of linguistic universals that is c o m m o n to all languages, and these universals occur in those languages because they are innate features of those human beings who assimilate those languages. Apart f r o m the fact that what had to be proven is being assumed, the reasoning arouses doubts as to its logical correctness. Katz seems to realize this clearly and adds the passage whose quoting has been announced earlier. 'This explanation, it should be noted, is neither vacuous nor a petitio principii. To say that the language spoken by any h u m a n has the universal features of the natural languages because the rules t h a t define his linguistic competence are modeled on inborn archetypical representations of these features would be vacuous if there were no independently arrived at statement of these archetypical representations. But there is such a statement. Thus, the explanation is not vacuous because it is formulated in terms of the independently arrived at specification of the linguistic universals given in the theory of language. On the other hand, this explanation is not circular because the justification of the theory of language nowhere presupposes it. Note also that, although no account of the orgin of such innate ideas is given, there is no question to be begged because there is no account of the origin of the principles of associative learning which the empiricist takes as inborn' (Katz 1966: 274). Piling arguments does not strengthen Katz's position, b u t rather testifies to his embarrassment, for, — in m y opinion, none of his arguments holds. 1. It is n o t true that t h e statement of the existence of innate

Generative grammar and the concept of innate ideas

39

linguistic structures (Katz this time refers to 'inborn archetypical representations', but this difference in terminology does not change the essence of the issue) has been arrived at 'independently'. Independently of what? Should this 'independence' consist in the fact that, as Katz suggests, the listing of the linguistic universals has been made 'independently' within the theory of language, then the proof would be as scientific as the various proofs of the existence of God. 'Independence' in this case is to the point if the existence of appropriate innate structure be established independently of the theory of language, i.e. within the framework of another theory, in this case a biological theory being the only possibility. Now, as we have seen, this is not so, and hence the theory of generative grammar, when defending the concept of innate linguistic structures (which it must defend if it is to survive in its present form), has found itself in the position of Baron Münchhausen, who saved himself from sinking in a bog by pulling himself out of that bog by his own bob-wig. 2. The reasoning is vitiated not only by a petitio principii, but also by a circulus vitiosus, since the linguistic theory (i.e. the metatheory of generative grammar) which Katz defends assumes — contrary to what he claims — reference to the concept of innate structures, because, as we have seen earlier, the claim as to the existence of a universal grammar cannot be defended without this assumption. 3. If the supporters of the associative theory of learning in fact refer to some innate features and do not give any proof of their claim (especially as far as the origin of those innate features is concerned), then this demonstrates the weakness of their theory, but does not exclude the possibility of criticizing the weak points in the theory advanced by their opponents. Note that in the case of generative grammar the point is not to account for the origin of innate ideas, to which generative grammarians refer, but to prove their existence, which can be expected of natural science only. The impossibility of settling the issue at the present level of development of natural science, and of molecular biology in particular, is thus decisive for the assessment of the concept of innate linguistic structures, with all the consequences of this fact for the models constructed in generative grammar. The hypothesis may not be rejected, and it keeps its heuristic value by being a stimulus to appropriate research. But neither may we deny the right to disagree to such thinkers as Roman Jakobson, who treat the issue as an 'utterly speculative and sterile question' (1970: 436). As mentioned previously, some comments of a broader signifi-

40

Adam

Schaff

cance suggest themselves in connection with the biological aspect of the issue. First, why did the founders of generative grammar resort t o the nativist hypothesis in order to develop their ideas? The recurrent argument, that no other theory explains the fact that the child learns any language in a very short time and even under unfavorable conditions and that otherwise belief in miracles would be the only satisfactory explanation, is n o t convincing, especially if we consider the otherwise well-known logical precision in reasoning on the part of those w h o advance the nativist hypothesis. The simplest way out would be to abstain f r o m statements which they were in no way b o u n d to make while constructing the specified structural model of language and developing the transformation calculus based on that model. Now if they decided t o adopt a biological interpretation (their declaration of access t o rationalist tradition means merely couching their views in philosophical terms), t h e y did so because the neonativist ideas, in connection with recent advances in molecular biology, comply in a way with the spirit of our times. A general interpretation of the problem has been offered by Fran£ois Jacob, who has based his history of biology, interpreted as la logique du vivant (which is the title of his book), precisely on this methodological conception. His idea is as follows: theory is always ahead of research practice, especially in experimental disciplines. In the case under consideration this idea is significant also because of the theoretical implications of the specific rebirth of nativism, especially in connection with research in molecular biology on the genetic code. The importance of Jacob's comments justifies quoting him in full: 'Pour q u ' u n object soit accessible a l'analyse, il ne suffit pas de l'apercevoir. II faut encore q u ' u n e theorie soit prete a l'accueillir. Dans 1'echange entre la theorie et l'experience, c'est toujours la premiere qui engage le dialogue. C'est eile qui determine la f o r m e de la question, done les limites de la reponse. " L e hasard ne favorise que les esprits prepares", disait Pasteur. Le hasard, ici, cela signifie que l'observation a ete faite par accident et non afin de verifier la theorie. Mais la theorie etait deja la, qui permet d'interpreter l'accident' (Jacob 1970: 24). This is to be interpreted so that what we perceive in the world around us and what questions we ask about it depends on the theory we have at our disposal (which theory, in turn, is a product of our knowledge at the given stage). Indirectly, this helps us understand the biological trend in the interpretations of certain linguistic facts,

Generative grammar and the concept of innate ideas

41

among other things the interpretations advanced by generative grammarians. The achievements in the field of molecular biology, with the striking discovery of the genetic code, make us revert both to the issue of heredity and that of genetic endowment with which every man is born, and consequently, the problem of the relationship between what is innate and what is acquired societally. At any rate, the comforting conviction that nativism is dead, that, 'as even Locke said', man is born as a tabula rase, belongs to the past. The problem has been revived and will have to be assimilated anew by philosophers, which may give rise to many troubles that will have to be taken into account deliberately. All this evokes two comments. The first is on the fortunes of the various ideas and theories which seemed to be dead and which, under new conditions, start living again. Early in this century, Ludwik Krzywicki, the prominent Polish Marxist scholar, wrote a fine book on Migrations of Ideas, which unfortunately is little known abroad because of the language barrier. He analyzed there, on the example of social ideas, the migrations of ideas in time and in space, and showed that under new conditions old ideas acquire new aspects and sometimes breed novel concepts. This obviously applies to Chomsky's reference to the history of what he termed Cartesian linguistics. It turns out that the evolution of ideas, especially in science, does not follow a straight line to break off at a certain point: it rather follows a spiral, which makes returns possible, but then at a higher level of general development, and the development of human knowledge in particular. Second, this fact is not only to be stated, but also predicted in a scientific manner. In some cases, especially if such returns have social implications, be it only indirect ones, they must be not only predicted, but met properly. This undoubtedly holds for neonativism, i.e. the rebirth of interpretations of various facts in the light of the nativist theory. The idea of innate linguistic structures is only an example: We have the well-known facts that children learn languages almost casually, we have the new structure of science, connected with advances in molecular biology, and the result is the concept of innate linguistic structures, because this theory now lends itself well to the integration of problems connected with the functioning of the human organism. We may, and we ought to ponder this theory, even though it may seem shocking, but we also have to realize that we have here to do with a very mild case as far as its social implications

42

Adam

Schaff

are concerned. Reference is made in it to a universal genetic endowment of mankind, i.e. to what unites human beings, be it alone because of the common biological fortunes of the species Homo sapiens. But the genetic code, or the plasma language of heredity, may, perhaps, record also factors by which human beings differ among themselves, be it alone for the climatic, geographical, social, etc., differences in their history. Could then there not be a genetic substantiation cf some forms of racialism? This could not be exluded. At any rate the facts now observed, although quite different in nature, make us take into philosophical consideration that aspect of the issue, too. As mentioned above, biological reflections form only one aspect of the discussion of the concept of innate ideas within the framework of generative grammar. This aspect of the discussion is, as we have seen, an important one, but it .has been pushed into the peripheries since the discussions reflected in a number of publications (two principal ones of which were mentioned in the introduction to this paper) have been mainly concerned with philosophical issues. Before these discussions are summed up here, my intention is to eliminate one aspect which, in my opinion mentioned earlier, is irrelevant to the problem under consideration: the aspect to be disregarded is that of the history of philosophy. Who correctly interprets the works of Locke and Descartes and what these and other prominent thinkers had to say on the subject of innate ideas had little, if any, bearing on the settling of the controversy which is connected with the present-day state of science. These are matters of interest to a historian of culture, but they can well be disregarded here and left to historians of philosophy in general, and of the philosophy of language in particular. (Chomsky is right in demanding that the history of ideas and the study of facts be separated from one another, see 1967c.) Not all arguments and counter-arguments will be quoted here either. They have been published and are widely accessible; this holds both for criticism and for Chomsky's two papers (1967c; 1969b), which are at least partly his reply to such criticism (they have not been mentioned here because, apart from offering somewhat more radical formulations, they do not contribute anything new to his statements, and some replies to criticism can be understood only against the background of the discussion as a whole). I, while raising here my own objections and making my own suggestions, will quote

Generative grammar and the concept of innate ideas

43

out of the arguments formulated in the discussion only those two which I consider to be most important: The one advanced by Nelson Goodman that the lack of an alternative hypothesis does not prove that the hypothesis which is the object of a controversy is in any way substantiated, and the one formulated by Sidney Hook and Thomas Nagel that there is little precision in the distinction between 'knowledge' and 'capacity' when Chomsky speaks about innate linguistic structures which convey to the child, endowed with the appropriate acquisition device or learning device, the knowledge of the rules of generative grammar. Let us begin with those arguments which have been stated by others and with which I am in solidarity. Goodman's argument is a reply to the following type of statements which recur in the writings of Chomsky and his collaborators: now we face the difficult problem of explaining how the child learns a language and we have no hypothesis alternative to ours, hence our hypothesis is true. Of course, this has never been so worded, since that would be too naive, but the idea was implied by many statements. This, of course, is erroneous, and has to be eliminated at the outset when we proceed critically to analyze Chomsky's views. Since, Goodman has already done this, it suffices to quote his words. 'Let us now assume that for certain remarkable facts I have no alternative explanation. Of course, that alone does not dictate acceptance of whatever theory may be offered; for that theory might be worse than none. Inability to explain a fact does not condemn us to accept an intrinsical repugnant and incomprehensible theory' (Goodman 1967: 27). In his reply Chomsky (1968) concerns himself only with the objection that we have to do with a repugnant (an emötional label which does not contribute anything to the discussion) and incomprehensible theory (since it follows from Goodman's later words that he comprehends Chomsky theory even though he criticizes it for the lack of precision in certain formulations). In my opinion Chomsky is right in rejecting these two objections, but he has neither taken up the principal objection nor replied to it. The conclusion to be drawn from the objection is in a way trivial, but it is not in the least trivial when it comes to the assessment of Chomsky's views: They just include a hypothesis that requires empirical verification but may not be accepted merely on the grounds that there is no other theory that would explain the same problem in a satisfactory manner. All arguments other than that concerned with empirical verification are,

44

Adam

Schaff

so to say, auxiliary, but they do not settle the controversy. Since Chomsky's many statements seem to indicate that he agrees with this, the point is to indicate the manner of an empirical verification of his theory, a specific crucial experiment conceived for the case under consideration. Of course, the burden of the proof could be passed to the opponents who would then have to adduce facts which disprove the theory (this would be in line with the procedure used in penal cases, and Chomsky (1969: 84) seems to favor it on certain occasions, but from the scientific point of view such an evasion cannot replace verification without which opinions remain a hypothesis. Sidney Hook (1969: 162ff) raises another issue and his argument is quite pertinent: If it is said that the child 'is born with knowledge', then we have to state precisely what is meant by 'knowledge'. If this is neither a skill nor biological processes inherited together with cellular plasma, then Platonic anamnesis is the only idea left to explain Chomsky's conception. The same path of argument is followed by Thomas Nagel (1969: 172ff) who makes a distinction between 'innate capacity' and 'knowledge' to demonstrate that Chomsky, despite his misleading terminology, does not in fact go beyond 'innate capacity'. Let us agree that this lack of precision contributes to a theoretical confusion, but if we are convinced that we have to do with a research hypothesis, we should not reject it even while we pin down some inaccuracies in formulations. But such a hypothesis ought to be subjected to a critical analysis. With what are we to begin such a critical analysis when it comes to the concept of innate ideas as advanced within the framework of generative grammar? It seems that we ought to begin with the statement that, in the light of the present-day state of science, biology in particular, the hypothesis is legitimate. Hence it may not be rejected merely because of the opinions of one or another philosophical school. But, on the other hand, it is legitimate to demand that this hypothesis be verified empirically, or that, at least, an approach to such a verification be indicated. (We adopt here the convention that the term verification will be used to denote a testing procedure, which may result either in verification in the strict sense of the word, or in a failure or disproving (falsification), if an appropriate crucial experiment has been designed. R. Carnap also was inclined to accept the view that the empirical nature of a statement is connected with its verifiability

Generative grammar and the concept of innate ideas

45

or falsifiability; this compromise formula was adopted following the controversy in positivist circles, with Popper as the main opponent, over the possibility of verification.) In this respect, a philosopher may prove useful by posing appropriate questions and, as far as possible, suggesting experiments, research methods, etc. In my opinion, the neglect of the empirical aspect of the issue on the part of generative grammarians (for all their oft-repeated claims that their theory is empirical in nature) may be interpreted as a consequence of the inclination to construct a hypothetical deductive model. It might be replied that generative grammarians by making use of their transformation model can answer questions in the face of which taxonomical descriptivists stand helpless (Shaumian 1962). This is certainly true, and this is certainly an empirical argument. But this does not prove the existence of a universal grammar, nor, a fortiori, is it an empirical verification of the statement that there are innate linguistic structures which coincide with the system of the rules of generative grammar. Is such a verification procedure at all possible? Yes, and in two ways at that. First and above all, we mean the verification of this hypothesis by molecular biology. The point, as E. Lenneberg rightly says in his Biological Foundations of Language, is not to find the gene of language. The problem is much more subtle and much more intricate, but if it is true that not only the faculte du langage, but also the whole of generative grammar, being universal, is innate and belongs to the genetic endowment of every human being, then, if we are to follow Lenneberg, it must be a verifiable feature in the sense that it should be possible to indicate its somatic background. Hereditary features are recorded in the language of the genetic code, i.e. in the form of specific combinations of the four elements — nifcleic acids — along the chromosomes. Molecular biologists have succeeded, by using elaborate research methods, to decode a number of records in the said code, thus solving various puzzles of the human organism. It may happen that in the future they will also be able to answer the question we are concerned with, either in favor of the generative grammarians by verifying their hypothesis, or against them, by disproving it. In such a case Rome would speak and the controversy would be settled. For the time being they are unable to provide such an answer, and hence the issue must remain pending. Second, there are some chances of verifying the hypothesis on the ground of the activity of generative grammarians (and their opponents), i.e. within the domain of language. I have to admit that I

46

Adam

Schaff

fail to grasp why this self-evident test has never been made (nor, by the way, can I comprehend why after the original euphoria over the Sapir-Whorf hypothesis concerning linguistic universals the interest in this hypothesis has vanished in American linguistic circles, even though the hypothesis has never been disproven, and, to make matters worse, no serious attempt to disprove it has ever been made. On another occasion several years ago, I discussed this unwillingness to undertake an attempt to verify the Sapir-Whorf hypothesis.) It seems that it is easier to formulate general linguistic hypotheses that are philosophical in nature than to try to verify them empirically by sophisticated methods. This is certainly so when it comes to interests manifest in linguistic circles, although their members might be expected to have a different approach to the issue. The empirical verification which is suggested here is simple in its general idea, but presumably much less so when it comes to putting it into effect. If it is true that deep structures are universal and that surface structures are derived from the former by appropriate transformation rules which are explicitly formulated in transformational generative grammar, then it would be enough to know these rules concerning the three language components — phonetics, semantics, and syntax — to be able, following an appropriate descriptive analysis of a given language, freely to move from deep to surface structures, and vice versa. At any rate, nothing would prevent an appropriately programmed computer of adequate power, in view of its great operational speed and its vast memory, from so* moving between the deep and the surface level. Assume now that an adequate computer is programmed to move so between deep and surface structures and vice versa not in one language, but in two, or three, or four languages. In the first stage of the experiment we could confine ourselves to one family of languages (e.g. the Indo-European), and — to avoid other complications — to languages at a similar level of development as far as linguistic history, the socioeconomic and cultural levels of the users of these languages, etc., are concerned. If the hypothesis stating that deep structures are universal is true, then in this case the fact can be most easily verified. The computer which can move between the two levels of linguistic structures is given the following task: analyze a given text, e.g. in French, pass from its surface structures to its deep structures, and .then move in the other direction, but, for examples in the field of the English language, thus translating the French text

Generative grammar and the concept of innate ideas

47

into English. In plain terms, this would just be a translation from one natural language to another natural language (both covered by the program of a given computer), the only provision being that the intermediate stage would be that of deep structures, which would have to be, if not identical, at least of the same type for all the languages involved. The universal grammar hypothesis would be tenable only if this experiment proved successful. The result of this simple experiment, consisting of translation through the intermediary of deep structures, would be the first empirical test for the hypothesis in question. I believe that such an experiment would succeed. Even in this simplest case there would be difficulties, especially with the semantic component — the rules of sense, which assign meanings to sequences of sounds — mainly in the cases of polysemy, but this is just the issue at stake, and recourse to deep structures is expected to solve such problems. The problem of the semantic component has so far not been solved in generative grammar, but this must be done in the next future, if generative grammar is to be considered seriously as a hypothesis. Thus, even if the suggested experiment cannot be carried out at once, it must be taken into account as a program for the future. This will stress the empirical nature of the generative grammar hypothesis and point clearly to what the crucial experiment is to be. But, as mentioned above, this would be merely the first, and the easier, part of the experiment. Part two would consist in a similar translation test that makes recourse to the intermediary of deep structures, but this time it would involve languages which are in different language groups remote from one another both genetically and culturally. For instance, translation could be made from English into Chinese, into Hopi, into one of the Eskimo languages, into a language of Australian aborigines, etc. The task would be immense, even if we consider the preparatory stage of describing such languages structurally and working out appropriate computer programs. Even if we disregard the rather trivial issue that these in most cases would be lexically (and hence conceptually) poor languages, the point would be to describe their respective structures so to say, from the inside. This metaphor is intended to convey something which is both simple and extremely important. When we face the structure of a language, especially of a culturally remote one, we may be inclined to make that description comply with an otherwise known model of the structure of another language. As demonstrated by the well-

48

Adam Schaff

known examples provided by Malinowski ( 1 9 5 3 ) , this procedure may easily yield nonsensical results, but it may also suggest that a given theory, applied in such a way, works well. We can, and we ought, to approach the issue at a different angle, by making an autonomous analysis of the semantic and syntactic components of a given language to serve as a basis of its structural description which would cover phonetics, syntax, and semantics. Such work being done, the proof would be completed if a translation through the intermediary of deep structures (which would have to be the same, or at least greatly similar) really succeeded: the proof would mean both the verification of the hypothesis assuming the existence of a universal grammar, and the disproving of the hypothesis assuming linguistic relativism. F o r the time being we can only have various, often incompatible, beliefs as to the results of such an experiment, but the controversy can be settled only with reference to its outcome, if and when such an experiment is carried out. But even this would not yet prove the existence of innate linguistic structures (since that can be done only in the sphere of natural science, and molecular biology in particular), but it would prove that something like a universal generative grammar does exist. Such a proof would make the hypothesis on innate linguistic structures much more probable, and at any rate it would be a big step toward testing that hypothesis empirically. So much for the preliminary statement that we have to do with a hypothesis which is legitimate, but which requires empirical verification if it is not to be treated merely as a deductive model based on speculative assumptions. This requirement also suggests the program for empirical tests. But what can be said about the present shape and meaning of the hypothesis in question? We shall now quote Nelson Goodman's and Sidney Hook's, arguments which open the list of questions and objections with reference to the present form of the hypothesis on innate linguistic structures, i.e. before its verification or disproving following appropriate experimental tests. Hence, first, the fact that we have no alternative theory that would be satisfactory from the scientific point of view, can not be taken as the confirmation of the hypothesis under consideration, which does not comply with all the requirements that a legitimate theory ought to satisfy. Second, if, following Chomsky's reasoning, we say that the child's innate acquisition device (i.e. a kind of a black box which for a given

Generative grammar and the concept of innate ideas

49

input has a given output — in this case the knowledge of a given language) gives him the knowledge of a given language, then this may be understood in two ways: a) In the sense that this mysterious acquisition device is identical with de Saussure's faculte du langage, i.e. that the child has an innate capacity to learn a language (without explaining how this occurs, i.e. taking this learning mechanism to be unknown). This would rank the acquisition device on par with other innate capacities (from instinctive actions to such which require learning, e.g. on the basis of innate manual abilities), thus making the hypothesis something trivial and thereby destroying the intricate concept of generative grammar; and b) in the sense that the statement on the existence of such an acquisition device is equivalent to the statement that every child has an innate knowledge of a set of linguistic rules which make up generative grammar. Thus in this case we would have to do not with a capacity to learn a language, but with its ready-made knowledge. This issue has not been formulated by Chomsky and his followers with sufficient precision. On the one hand, it would seem that they mean interpretation (b), without which, as has been said above, the entire hypothesis would become trivial and of little interest, but, on the other hand, the course of their reasoning does not justify going beyond interpretation (a) (for instance, in his paper in Synthese Chomsky protested against being ascribed the intention to defend the theory of instinctive actions), which makes the whole discussion questionable. I will now add some more criticism. Third, it must be emphasized that when the advocates of the generative grammar theory use the term universal grammar they in fact resort to a fiction. By using the term 'fiction' I want to do the i in order to resist the psychological pressure of a hypostasis: The fact that there is a term 'universal grammar' does not imply in the least that something like a universal grammar — a set of generative grammar rules, which is universal, i.e. valid in all languages in the form of deep structures — does in fact exist. As long as its existence is not proved we must treat the concept 'universal grammar' as a fiction, if only because the relevant hypothesis, namely that of linguistic relativism, has never been disproven and is supported by a much more comprehensive body of data than is the generative grammar hypothesis. It may even be said that, in view of the hypothetical deductive nature of the model of generative grammar, its authors are

50

Adam Schaff

specifically nonchalant on the issue. This was manifest in Chomsky's reply to Hiz (at a symposium organized by Sidney Hook) when tlie latter objected that Chomsky's constructions are not based on an additional analysis of various languages. Chomsky rejected this argument as groundless by claiming that, for instance, Matthew's study of the Hidatsa language is worth more than a thousand superficial papers on various languages. While the assessment of the value" of such papers is to be left to experts, let it be noted that no one suggests that superficial papers be written. What is needed is good studies in depth, but it certainly is not true that we may rest satisfied with a single study, even though it be perfect. In this respect W. van Humboldt, to whose theory of 'language form' Chomsky so willingly refers, was of a quite different opinion, and rightly so. And when, in Chomsky's reply to Hiz's objection, we read: 'If someone feels that the base of data is too narrow, what he should do is show that some of the material omitted refutes the principles that have been formulated . . .' ( 1 9 6 9 b : 84), then we must disapprove of such a standpoint. As has been mentioned on one occasion earlier, this holds for a penal trial, in which all guilt has to be proved, but in science it is otherwise: The fact that a hypothesis has not been refuted does not mean that it has been verified. Anyone who advances a hypothesis must verify it if he wants it to become an accepted scientific theory. But all these reservations against a hypothesis are merely an introduction to the main argument to be presented here. The point is that even if we prove the existence of a universal grammar (e.g. by means of the experiment suggested above, or in any other way), it would not follow therefrom, as already has been said earlier, that such a grammar be a specific product of certain innate linguistic structures. Such a fact could well be a result of some other causes. The existence of innate structure can be proven only in the sphere of natural science, biology in particular. But should biologists or other natural scientists prove the existence of innate linguistic structures that would correspond to Chomsky's acquisition device, we still could not conclude that a universal grammar exists, unless it be proven additionally that those innate linguistic structures are the same in all (healthy) human beings. For as far as the specific records in the genetic code are concerned we know that such records differ, which accounts for various differences not only between individuals but between fairly stabilized groups (compare such anthropological features as the

Generative grammar and the concept of innate ideas

51

pigmentation of the skin, etc.)· Hence, to verify Chomsky's claim we would have to prove not only that an innate acquisition device in the sphere of language exists as a feature of the human species, but also that this device is the same in all humans. Fourth, the concept of innate linguistic structures, and consequently the whole concept of generative grammar, is burdened with lack of clarity when it comes to the relationship between language and thinking. We face here one of the 'immortal' problems in the philosophy of language, without the solving of which, however, we are not in a position to solve any general problems in the theory of language, which Chomsky's hypothesis claims to be. According to the concept of generative grammar, deep structures are universal, the same in all languages, which is to substantiate the claim that a universal grammar exists. But the semantic component of language cannot be separated from deep structures. The question arises, what does it mean that deep structures are innate (are given by the language acquisition device, which is innate in all humans)? This can be interpreted in two ways: a) deep structures are innate together with the semantic component; b) deep structures are innate as purely linguistic entities, to which, in the process of genetic development, the semantic component (i.e. the specific rules of sense) is somehow adjusted. In the former case we have to do with the statement which identifies language and thinking. The situation is then clear from the point of view of the hypothesis that assumes innate ideas: language as a phonetic-and-semantic whole is innate and we are accordingly relieved of all troubles of explaining the relevant facts; what we have to do is merely to describe them. This is a convenient situation (of course, on the 'trifling' condition that the hypothesis has been verified) — one would even say, too convenient — as it relieves us of many difficult problems, which are then accounted for by the innate nature of the relevant phenomena. On the other hand it requires acceptance of the hypothesis that assumes the identity of language and thinking, a hypothesis which is not acceptable even to the staunchest supporters of the idea of their inseparable unity (W. von Humboldt, so cherished by Chomsky, defended the statement that we think in the way we speak, but we speak in the way we think; he would accordingly most emphatically disclaim the statement that language is the same as thinking), not to speak of all those who either approach the problem from the genetic point of view and indicate certain forms of thinking (in the sense of problem solving) at the

52

Adam Schaff

preverbal stage of development of the animal kingdom, or even plainly claim that there are forms of nonverbal thinking (mathematical, musical, etc., thinking). It follows from certain statements made by Chomsky (and especially by Lenneberg) that he rejects the hypothesis on the identity of language and thinking. This leaves the second interpretation.· only linguistic structures are innate. But what does this mean, and what are its consequences for the issues we are concerned with? If the semantic component is not innate, then are both the phonetic and the syntactic component innate, or is only the phonetic component innate? We could hardly imagine a syntax without the semantic aspect of language. But should only the phonetic component be innate (suppose we have proven that), then this leaves us in the absurd situation in which language sounds are dissociated from meanings; moreover, this assumption does not solve the problems we are interested in. In order to arrive at the generative grammar postulated by Chomsky we must have at our disposal of all three its components; and if that generative grammar is given to every human through the intermediary of innate linguistic structures, then we must assume that it is in them that the identity of language and thinking is manifested — the conclusion which we wanted to avoid. We thus face considerable difficulty resulting from the hypothesis suggested to us. The difficulty requires that the authors of the hypothesis at least make their standpoint more precise; it also requires some answers to the questions which arise when the problem is formulated with sufficient precision. Fifth, there is the problem of the social conditionings of language. That language is a social phenomenon is now, in the light of modern linguistics in general and not only its specialized branches such as sociolinguistics, just a trivial statement. The same is true of the more general statement that man is a product not of nature only, but of society as well (this statement was first advanced by Marx in his criticism of Feuerbach's anthropology, but it has since become so widely accepted in science that the realization of its Marxian origin has been lost). Chomsky, and especially Lenneberg, emphasize that although the language acquisitition device is innate, nevertheless we do not have to do with a purely biological process, since an appropriate input in the form of acts of speech which a child hears in his milieu (i.e. the impact of the social factor) is required for the mechanism of the acquisition device to start working. This is why the supporters of generative grammar say that they accept the role of the

Generative grammar and the concept of innate ideas

53

social factor, and hence the objection that they stand for biologism is groundless (cf. the criticism formulated by R. Jakobson in his article in 'Linguistics', 1970). Yet the problem is much more complicated, and the objection that they disregard the role of the social factor when analyzing language seems to be justified when it comes to the concept of innate linguistic structures. Let us first see to what the role of the social factor, which grammars admit, led to in the linguistic theory. They reduce it to a specific input which, in the form of the stimulus provided by the acts of speech of the child's milieu, sets the language acquisition device in motion. Present-day supporters of the theory stating that language is based on innate ideas thus differ from their predecessors in the eighteenth century who believed that a child totally isolated from all influence of a human milieu would begin to speak spontaneously as a result of the maturation processes; their controversies were confined only to the issue which language such a 'savage' child would speak (they were mostly inclined to think that it would be Hebrew, which as the language of the Bible was believed to be the original language of mankind). Today we know quite well that a child totally isolated from appropriate influence of all human milieu would not develop his language capacity, but would, on crossing a certain age limit, remain a homo alala for ever. This is also confirmed by the cases of 'savage' children, recorded in greater or lesser detail; facts show that 'wolf children do not behave like Mowgli of The Jungle Book, who knew the languages of all animals, but become imbeciles, and incurably so, once they cross a certain physiological age barrier. No serious researcher, even if he accepts the claims of genetic rationalism concerning language and thinking, will now repeat the eighteenth century errors about a supposed spontaneous emergence of speech in the child merely as a result of maturation processes. But does the statement that an appropriate input in the form of acts of speech heard in the child's human milieu, as the issue is now being formulated by generative grammarians, qualitatively change their standpoint as compared with the naive eighteenth century beliefs? No, it does not. For what is being claimed in that respect by generative grammarians? It follows clearly from many statements by Chomsky and Lenneberg (including those quoted above) that they mean only a stimulus, which is analogous to stimulation in chemical processes. We find in those statements that the acquisition device works if excited by

54

Adam Schaff

external stimuli, regardless of differences between the levels of intelligence of the various persons and regardless of the type of input (i.e. regardless of how his milieu speaks the child develops the correct generative grammar of a given language). Thus what is meant is clearly just the stimulation of innate mechanisms, with the provision that the stimulation works toward a specified language (following the stimulus of the acts of speech heard by the child), and not in general. But we have to say that there is little difference between this and the corresponding view of the eighteenth century: The language mechanism develops like innate physiological mechanisms. We even find the statement that a human being does not learn a language, as he does not learn digestive processes: his language simply develops. This may be an elegant explanation (in the mathematical sense of 'elegance') because of its 'economy', but it is unfortunately erroneous. It is not true that the quality of the input (the acts of speech heard by the child) does not affect the output (the language of the individual in question): if a person is brought up in a milieu which speaks a given language ungrammatically (by the way, how does this happen if the result is supposed to be always the same, i.e. in the form of a correct generative grammar, independent of the quality of the input?), then he will speak that language ungrammatically himself. Usually this is a life sentence, which refutes the claim that every child develops a generative grammar which is universal as it is imbedded in innate linguistic structures. Such defects in linguistic upbringing can be eliminated in the adult age only with greatest difficulty, and even that not in all cases (compare G. B. Shaw's Pygmalion). But even if we disregard the naivety of the standpoint analyzed here we still have to discuss the vast sphere of problems connected with the social factor in the field of language. We have to say plainly that when we refer, in analyzing the development of the language faculty in man, to the social factor we do not mean only the speech sounds heard by the child (since then a loudspeaker from which the child would hear sounds of even grammatically incorrect speech, as the generative grammarians claim, would suffice), but human speech as related to man's social actions. The theory of generative grammar, and especially the hypothesis which assumes the existence of innate linguistic structures, unduly simplifies the issue of the social factor, if it notices it at all. By doing so it eliminates from its field of vision (1) all problems of language in the context of social actions (i.e. the field of study of sociolinguistics), and (2) the problems of the relationship

Generative grammar and the concept of innate ideas

55

b e t w e e n language a n d thinking, which also requires t h e c o n t e x t o f social actions. N o w t h a t w e revert t o t h e starting p o i n t o f this analysis, w h e r e it was said t h a t t h e h y p o t h e s i s o f innate linguistic s t r u c t u r e s is legitim a t e f r o m t h e p o i n t o f view o f present-day s c i e n c e b u t r e m a i n s an e m p t y s t r u c t u r e w i t h o u t an empirical verification, w e m a y add t h a t it also requires at least m o r e precision in c e r t a i n f o r m u l a t i o n s b e f o r e its verification c a n b e u n d e r t a k e n . To

conclude,

I say t h a t I a m c o n v i n c e d o f t h e usefulness

of

t r a n s f o r m a t i o n a l generative g r a m m a r for t h e d e s c r i p t i o n o f linguistic structures, even if t h e claim o f universality be s h a t t e r e d

together

with t h e h y p o t h e s i s o f innate linguistic s t r u c t u r e s . Making t h e foundations o f generative g r a m m a r less c o m p r e h e n s i v e will p r o b a b l y n o t please its founders, w h o are m u c h m o r e a m b i t i o u s w h e n it c o m e s t o t h e t h e o r y o f language, b u t this does n o t belittle t h e i m p o r t a n c e o f w h a t t h e y have really achieved.

REFERENCES Adjukiewicz, K. (1934a), 'Das Weltbild und die Begriffsapparatur', Erkenntnis, 4. — (1934b), 'Sprache und Sinn'. Erkenntnis, 4. — (1935), 'Die wissenschaftliche Weltperspektive', Erkenntnis, 5. Benveniste, E. (1963), ' "Structure" en linguistique', in R. Bastide (ed.), Sens et usages du terme 'structure' dans les sciences humaines et sociales. The Hague — Paris, Mouton (Janua linguarum, Series minor 16). 2nd ed.: 1972. Bubrix, D. (1930), 'Neskolko slov ο potoke rechi', Biuleten Lojkfun, 5. Carnap, Rudolf (1956), Die logische Syntax der Sprache. Chomsky, Noam (1957), Syntactic structures. The Hague — Paris, Mouton (Janua linguarum, Series minor 4). 10th ed.: 1972. — (1962), 'Explanatory models in linguistics', in E. Nagel, P. Suppes and A. Tarski, Logic, methodology, and the philosophy of sciences. Proceedings of the 1960 International Congress, pp. 528—550. Stanford, 111., Stanford University Press. — (1964a), 'On the notion "rule of grammar" ', in J . A. Fodor and J . K. Katz " (eds.), The structure of language, Englewood Cliffs, N. J., Prentice-Hall. — (1964b), Ά review of B. F. Skinner's Verbal behavior', in J . A. Fodor and J . K. Katz (eds.), The structure of language. (See above under 1964a). — (1966a), Cartesian linguistics New York, Harper and Row. — (1966b), Topics in the theory of generative grammar. The Hague — Paris, Mouton (Janua linguarum, Series minor 56). 3rd ed.: 1972. — (1967a), 'The formal nature of language', Appendix A in Eric N. Lenneberg, Biological foundations of language. New York. — (1967b), 'Symposium on innate ideas', in Noam Chomsky, Hilary Putnam and Nelson Goodman (eds.), Synthese, 17

56

Adam

Schaff

— (1967c), 'Recent contributions to the theory of innate ideas', in Noam Chomsky, Hilary Putnam and Nelson Goodman (eds.), Synthese, 17. — (1968), Language and mind. New York. — (1969a), Current issues in linguistic theory. The Hague — Paris, Mouton (Janua linguarum, Series minor 38). 5th ed.: 1970. — (1969b) 'Linguistics and philosophy', in S. Hook (ed.), Language and philosophy. New York, University Press. Fodor, Jerry Α., and Katz, Jerrald J., eds. (1964), The structure of language. Englewood Cliffs, N. J., Prentice-Hall. Goodman, Nelson (1967), 'The epistemological argument', in Noam Chomsky, Hilary Putnam and Nelson Goodman (eds.), Synthese, 17. Gukham, N . M . (1962), 'Istoricheskiye i metodologicheskiye osnovy strukturalizma, in V.A. Zvegintsev (ed.), Novoe ν lingvistikvo, Series 11 (Moskva). Hook, Sidney, ed. (1969), Language and philosophy. New York, University Press. Jacob, Francois (1970), La logique du vivant. Paris. Jakobson, Roman (1970), 'Linguistics', in Main trends of research in the social and human sciences, Part one: Social sciences. The Hague — Paris, Mouton — Unesco. Also in Tendances principales de la recherche dans les sciences sociales et humaines, Premiere partie: Sciences sociales. The Hague — Paris, Mouton - Unesco, 1970. Katz, Jerrold J. (1966), The philosophy of language. New York. Lenneberg, Eric H. (1964), Ά biological perspective of language', in Eric H. Lenneberg, ed., New directions in the study of language. Cambridge, Mass., MIT Press. — (1967), Biological foundations of language. New York, John Wiley. Martinet, A. (1955), Economie des changements phonetiques. Bern. — (1960), Elements de linguistique generale. Paris. Monod, Jacques (1970), Le hasard et la necessite. Paris. Postal, Paul M. (1964), 'Limitations of phrase structure grammar', in Jerry A. Fodor and Jerrald J. Katz (eds.), The structure of language. (See above). Ruwet, Nicolas (1967), Introduction a la grammaire generative. Paris. Shaumian, S. K. (1962), 'Teoreticheskiye osnovy transformasionnoy grammatiki', in V.A. Zviegintsev, Novoe υ lingvistike, Series II (Moskva). Skinner, B. F. (1957), Verbal behavior.

FLIP G. DROSTE

2

Presupposition, Truth and Grammatically

The question our topic raises is whether and to what extent presupposition, truth, grammaticality, and their possible interrelation have any bearing on the Sapir-Whorf hypothesis. Although it seems a bit far-fetched to fully converge both themes, there are points of common interest, nonetheless. A negative answer to the question could state that our three notions pretend to be universal linguistic qualities and, as such, negate the Whorfian hypothesis. But the answer can be positive as well, since it is argued that certain types of deviance cannot be accounted for on the level of competence but are performance phenomena. And as soon as language-specific qualities are at stake, the hypothesis could have a greater impact than has been acknowledged so far by transformational theorizing. It is a truism to state that presupposition, truth, and grammaticality are mutually related in one way or another. The exact nature of these relations is, however, problematic, and discussions as to where these mechanisms should be accommodated in a grammatical model show fundamental differences of opinion. We have no intention of trying to solve all the problems related to this subject. We would rather make some general suggestions on the relevance of presupposition and truth for a delimitation of deviance, especially in those areas where grammaticality cannot cope with this phenomenon. In order to proceed in this direction it is necessary first to delimit the three mechanisms under discussion. We shall do this in an off-hand way, i.e. we shall only circumscribe them in so far as it is needed for operational reasons. As regards presupposition, we follow the line of Frege-StrawsonKeenan, which implies a refutation of the Russell-Quine doctrine. We also abstain from commenting upon those aspects of presupposition

58

Flip G. Droste

which regard the speaker, the language, the addressee, and the like. Presupposition as it is conceived here, relates to specific reference (cf. Fillmore 1969: 120 f f ; Keenan 1971: 48, sub 2) as may become apparent from a random example: (1) This will be organized by a brother of mine (2) This will be organized by my brother In (1) it is presupposed that the speaker has more than one brother, whereas (2) has the presupposition that the speaker has only one brother (or, at best, that one specific brother is involved). The presuppositions, so it seems, have to be related to a . . . of mine and my, respectively, or, rather, to the structures underlying these surface phenomena. Presupposition not only relates to embedded relative clauses such as the above examples, but also to dependent clauses. In Keenan's well-known utterance: (3) Mary loves the puppy she found a presupposition is valid, viz. Mary found a puppy, as can be demonstrated by negating the utterance: under negation, indeed, the presupposition remains intact whereas the logical consequence becomes negative as well (cf. Keenan 1971: 255—256). It is clear that we delimit our field in essential respects. In the first instance we regard only those aspects of presuppositions relating to statements, or rather to the contents of a statement, i.e. to the proposition. Presuppositions relating to lexical items are left out of consideration. We do not deny that the imperative open presupposes the 'closedness' of its object (cf. Fillmore 1969: 121), nor that the noun bachelor presupposes its referent to be 'adult'. Our subject, however, has no direct relation to these lexical phenomena as it is syntactic on the one hand and referential on the other. As to the truth problem, we should distinguish presupposition and truth in an unambiguous way. Indeed, it can be argued that presupposition implies that the clause it is related to is regarded [+ true] by the speaker and is presented as such to the addressee. In order to avoid confusion, we shall no longer use the term [± true], but [± valid] in relation to presuppositions. The reason for this terminological change is the ambiguity of the term [True] in linguistic as well as in philosophical treatises on this subject. Truth, then, is not implied or presupposed in an utterance, but it is the result of semantic functioning. Whether (3) is a [+ true]

Presupposition,

truth and

grammatically

59

statement has nothing to do with its presupposition, as can be easily proved. If Mary did not find a puppy, i.e. if the statement is not in harmony with the presupposition of the dependent clause, the proposition does not suddenly become [— true]. At best we can call it inappropriate (cf. Strawson 1966: 29 ff). The statement as such, however, may very well prove [— true], e.g. when it becomes apparent that Mary is in the habit of pinching her puppy in a nasty way. The truth of a statement, in other words, should be measured along the lines indicated by Wittgenstein (1968: 43) and Tarski (1956: 155). Therefore, the [± truth] of an utterance such as (1) and (2) can only be judged by comparing it to the actual situation it refers to. If, for example, one states in a temporary wave of wishful thinking: (4) McGovern has won the 1972 elections in the U.S.A. the statement must ultimately be rejected and considered [— true]. This, apparently, has nothing to do with presuppositions, but only with the inconsistency of reality and conceptuality, the latter being the way in which this reality has been presented by linguistic means. Whereas presupposition is directly related to the attitude of the speaker, truth relates to the object discussed. Since the latter, the object in the extralinguistic world, may as well be situated in reality as in an imaginary world, it seems justified to distinguish two performative frames, viz.: (5) I tell you that I consider it [+ True] that (...) (6) I tell you that I consider it [+ True] in world Y that

(...).

The latter mechanism is essential for the creative function of language, one of whose fundamental goals is to relate ideas in a novel way, and not only to put real facts in a predetermined strait jacket. In an utterance such as: (7) To me she is the most beautiful woman in the world the [+truth] of the propostition (she . . . world) does not depend directly on the presupposition, as has been conjectured by Morgan (1969: 168 ff); since it has been embedded in the frame (6) the addressee can only accept the statement as a [+ true] one. The same goes for fairy tales, for modal frames, for performatives such as dream, hope, seem, as well as for specific tense frames. In short, with (6) a universe of interpretation different from the unmarked one of

60

Flip G. Droste

(5), has been introduced (cf. also Seuren 1972: 353ff). Our third notion, grammatically, should first of all be related to the unmarked performative frame. Only those sentences generated according to the rules of an (idealized) grammar are [+ grammatical]. This implies that the rules of grammar only generate grammatical structures — a merely technical term, this — so that grammatically deviant strings receive their structural descriptions only derivatively (Chomsky 1965: 227). It is next to impossible that a grammar of natural languages can be made so powerful as to exclude the generation of a sentence such as (4). For that reason, it seems justified not to identify deviance and [— truth]. How can one then judge metaphors and deviances of this type, of which it has been stated that they 'break selectional rules' and 'are apparently interpreted by a direct analogy to well-formed sentences'? (Chomsky 1965: 149). Consider a deviant string such as: (8) (Dreaming) when dawn s left hand was in the sky In our opinion, the clause dawn's left hand should, in the first place, be generated within the performative frame of (6) in which it is not deviant, since dawn has been personified. This implies indeed a breaking of selectional rules as they are given in the grammar, but it also implies a correct extension of the rules for world Y (cf. also van Dijk 1972: 258). The metaphorical character of the clause results f r o m the comparison of the F-world P-marker and the matching P-marker generated by the (idealized) real-world rules. The transfer rules, introduced by Katz (1964: 411), are in fact rules of comparison relating a F-world structural description and its nondeviant counterpart. Although we consider utterances such as (8) nondeviant within their proper universe of interpretation, they must nonetheless be judged [— gramm]. Otherwise it becomes manifestly impossible to qualify an expression as a metaphor, and no figurative language usage could any longer be marked as such. On the other hand, we agree that this limited use of the term grammaticality is only a matter of convenience, and an extension of the term could be defended as well. Within our framework, then, (8) is [— gramm] and also [— deviant]; with this set of terms it can be distinguished from sentences such as (3) which are [— deviant] but [+ gramm]. We shall now try to relate the three linguistic notions grammaticality, truth, and presupposition. Grammaticality is a notion central to all linguistic functioning, whereas presupposition and truth relate

Presupposition,

truth and

grammatically

61

to the central linguistic mechanisms as well as to the social and/or ontological frame each linguistic action is embedded in. Whereas presupposition relates to a fact prior to the linguistic act, truth relates to facts resulting from the latter. Moreover, presupposition concerns the speaker's attitude, but truth is concerned with reality. Schematically this can be expressed as follows:

PRESUPPOSITION

GRAMMATICALITY TRUTH

speaker

syntax

world

In how far can presupposition be related to specific reference, as it has been argued on several occasions? (Cf. Kiparsky 1970: 167; Droste 1974). This may be illustrated with an example: (9)The poor man was lost upon bearing her voice again Whether this statement is [+ true] or [-true] cannot be settled on this abstract level; it depends on the facts presented by reality. It is however, hardly conceivable to regard the man under discussion as being [+ true], that is, there is no reason whatsoever to relate the presupposition of the man's real existence with truth. There is indeed a presupposition, but this can best be circumscribed as follows: to the speaker the delimitation of the intended man, out of the class of all possible men, is sufficient, and therefore he can be marked linguistically as [+ spec]. In sentence (3), the presupposition can also be related to specific reference. There is no reason, actually, to state that the dependent clause Mary found a puppy is regarded [+ true]: the truth of this clause is simply not discussed. What is implied, however, is that the speaker takes it for granted that he does not have to explain what preceded Mary's love. He can, in other words, refer to Mary's finding a puppy, since he accepts the validity of this fact and since this fact is preexisting in relation to this novel statement. As to the famous topic (cf. Strawson 1950; Russell 1957): (10) The present king of France is bald

62

Flip G. Droste

we can state: a. it is a sentence and, therefore, a [+ gramm] string, since no rule of grammar has been broken; b. whether the statement (i.e. the contents of S) is [+ true] or [— true] depends on the situation it refers to: for the second half of the seventeenth century it seems [+ true], while it seems [— true] for the period around 1840. Uttered in our days it seems neither [+ true] nor [— true], since there is no present king of France, so we could call the statement inappropriate (cf. also Droste 1973a: 25ff); and c. the latter judgement, however, makes it clear that there exists at least an indirect relation between presupposition and truth. For in order to decide whether (10) is [+ true] or [inappropriate], we should know whether the subject under discussion (the present king of France) does or does not exist. A statement, therefore, can be judged as to its truthfulness if and only if its presupposition(s) is (are) valid. The statement should have specific reference, otherwise one discusses objects whose existence hangs in the void, (of course the term existence can relate to a real world as well as to an imaginary one, the F-frame, as mentioned above.) The above circumstantial evidence was necessary to tackle a problem which seems only remotely related to it. We have in mind unusual strings (marked with * for deviance) such as: (11)*

(12)* (13)* (14)* (15)* (16)*

A man is standing in the garden (Een man Staat in de tuin) (Ein Mann steht im Garten) He is the baker (In reply to: Who is there?) A ring is at the bell Lord-mayor is he The train I missed Other problems are there

It is to be mentioned first, that the strings (11) through (16) are unmarked, that is no special situation nor a specific intonation is to be related to them. In the latter case, it is possible t o consider some of them acceptable, and it stands to reason that acceptability is one of the parameters to decide on the highly theoretical delimitation of grammatical correctness. The question to be raised, however, is whether these strings are only abnormal in one way or another, or whether they are [— gramm] in the restricted technical sense. If we concentrate on (11),

Presupposition,

truth and

grammatically

63

considering this string representative for the whole group, there seems to be no rule violation on the level of the base component, i.e. the normal branching rules have functioned to generate a syntactic pattern which in no way deviates from those underlying sentences that offer no problem·.

The situation becomes opaque, however, if we make the insertion rules work on this (context-free) P-marker, since neither a subcategorization rule nor a selectional restriction is being violated. The basic rewrite rules, at least as they function in the standard theory (Chomsky 1965), then, oblige us to consider the generation of (11) a fully correct one, and (11) has to be regarded [+ gramm] in consequence. On the level of truth there is no reason to regard these strings as deviant either. If it is indeed correct that speaker and addressee notice a man standing in the garden, the statement is to be considered [+ true]; the same goes for the other strings (12) through (16).

It seems that the deviance of the above strings in one way of another relates to presupposition, or rather to specific reference. That presupposition, at least on the syntactic level considered here, is the same as specific reference can be proved with some other examples. If we negate the following sentence: (17) They noticed the man standing in the garden the presupposition remains valid. In (18), however, the dependent clause can be interpreted as a valid presupposition as well as a mere implication: (18) They did not notice a man standing in the garden Clearly the validity of the presupposition in (17) should be related to the specified character of the NP, expressed in the determiner the. Up to now our discussion has concentrated upon syntactic claims: which possible combinations can be generated byG? This approach

Flip G. Droste

64

implies that the rules only regard the linguistic competence. The problem we are confronted with in the deviant strings is, however, to be solved on another level, since they relate t o the question: which structures are to be used under which circumstances? And this, so it seems, implies the organization of performance, since circumstances ask for a choice. The organization of performance as necessary on this level is, first of all, attuned to perception. The Prague adagium of communicative dynamism (cf. Kern 1972: 4 6 9 f f ) demands a surface structure presentation which goes f r o m the known to the unknown, f r o m the specified t o the unspecified. In fact we meet here with an ancient presentative apriorism about the subject as the known constituent and the predicate as the new one. It seems, however, that this construction, based first of all on surface structure phenomena, is responsible for a confusion of linguistic levels, viz. — the performance level on which the way of presentation is organized — the competence level on which functional relations are being organized The above confusion is responsible for the ambiguous terminology of Paul (1920: 124) who tried to operate with notions such as grammatical and psychological subject. It seems to be on the safe side, therefore, to agree with Fillmore (1968: 17ff) who states that the category subject is to be confined to surface phenomena. Summarizing these somewhat disperse statements, we may conclude that the deviance of strings such as (11) through (16) should n o t be accounted for on the level of competence — we must consider t h e m [+ gramm] — but on the level of performance, where the lack of specific reference explains the incorrectness. The two following theses present themselves as a theoretical explanation: a. the first place in an utterance must be occupied by a constituent with specific reference; and b. if no constitutent of the utterance has an overt specific reference it has to be substituted by a marker of [— spec] character, such as it, there (the so-called expletive constituents). On the level of perception this implies that a first constituent in an utterance — except for substitution markers — is interpreted as [ + spec.], that is, the addressee assumes that the presupposition is a valid one. Therefore statements such as: (19) A fir is an evergreen

Presupposition,

truth and

grammatically

65

can only be interpreted as a categorial statement, and categorial articles should be interpreted as [+ spec.], (cf. also Heidolph 1970: 82).

Where Chomsky states (1965: 126) that 'all' languages (German, Russian, Mohawk) preserve the order subject-verb-object, he is right in so far as subject means the presupposed constituent of the utterance. But he is wrong is so far as he assumes that deep structure qualities are involved, since base rules do not decide on matters of specific reference. We shall try to explain the consequences of the above suggestions with additional evidence (cf. also Kraak and Klooster 1968: 105ff). (20)* (21)*

De Amerikanen hebben een gedenkteken opgericht (The Americans have erected a memorial) Een gedenkteken is opgericht door de Amerikanen (A memorial has been erected by the Americans)

It is clear that a passive transformation is blocked here, but there are other situations in which this transformation is obligatory: (22)* (23)*

Amerikanen hebben bet gedenkteken opgericht (Americans have erected the memorial) Het gedenkteken is opgericht door Amerikanen (The memorial has been erected by Americans)

It might be asked whether it is not possible to cope with referential problems as they are presented here on the level of competence. A referential constraint, for example, could possibly prevent the fronting of a [— spec] constituent. This solution, however, raises other problems, as can be proved with an example f r o m the Kiparskys (1970: 167): (24) The Upi reported that Smith bad arrived Whether Smith has indeed arrived is uncertain, and utterance (24), therefore, is ambiguous. It is, however, impossible to disambiguate the underlying S, since the [+ spec] or [— spec] character of the dependent clause only relates to the attitude of the speaker and has nothing to do with deep structure phenomena. It depends, in other words, solely on the speaker's conviction whether the dependent clause is considered a [+ valid] presupposition. In a way this has been acknowledged by Chomsky, where he revised his standard theory (1971: 209ff) and stated that surface structure determines, at least in part, the scope of logical elements and, consequently, the way in

66

Flip G. Droste

which a sentence should be interpreted semantically. The transformation of (24) into: (25) That Smith had arrived was reported by the Upi not only depends on the attitude of the speaker, but it must be considered highly improbable that a deep structure mechanism could be developed preventing such a transformation, since, in that case, the deep structure representations should also formalize the speaker's convictions, etc. That we are, indeed, confronted with a subjective notion as the speaker's conviction can be proved by a comparison of (26) and (27). Both have the same truth value and it could be argued, therefore, that they have the same underlying structure. Their presuppositions differ, however, in that the former sentence is ambigous, whereas the latter is [+ valid]: (26) Did you notice a man standing in the garden? (27) Did you notice the man standing in the garden? In order to cope with this problem in which presupposition, grammaticality, and truth are clearly interrelated, we refer to a solution advocated elsewhere (Droste 1973b: 58ff): On the level of deep structure, only base sentences are generated, fundamental to, for example, (20) and (21) as well as to (22) and (23). Base sentences are abstract in the sense that they do not decide on the ultimate shape of the utterance, that is on the practical order of its constituents, but on the ideal order at best. The abstract base S is mapped into a S also abstract, since it represents a class of U's — in relation to the speaker's attitude, that is, more specifically, in relation with the validity of his presupposition(s). If he wants to present a certain constituent as positively specified, he can front it without difficulty, and in cases of overt ambiguity, the fronting may even become obligatory. Consider the well-known example: (28) Two languages are known to everybody

in this room

All this implies that developing a grammar as a device for explaining an idealized competence is indeed a rather abstract occupation, since (a) the decision whether a competence string is [+ gramm] or [— gramm] always depends on the interpretation of material derived from language usage (i.e. performance), and (b) it is not only the competence grammar whose rules decide on the correctness of a string as it is proved, for example, by (12) through (16). Interesting, therefore, is a tentative solution of this kind of

Presupposition,

truth and

grammatically

67

problem by Ruwet, subscribing to a proposal of Klima. Facts of the above nature, up until now described in terms of transformations, are there explained by so-called perceptual strategies, i.e. performance constraints solving problems of structural ambiguity (Ruwet 1973). Our examples (20) through (23) affirm that only those readings which correspond to psychological mechanism of perception and explanation are acceptable. Returning now to our point of departure we may state that: a. the notion [± gramm] is not sufficient to fully decide on the deviance of a S; b. the notion [pres] should be introduced in order to distinguish between a deviant and a non-deviant [+ gramm] string; and c. the notion [± true] cannot be measured on the deep level of base sentences, but only on the level of sentences, whereas only these sentences whose presuppositions count as valid can be measured as to their [± true] value. If the latter is not the case, the [± true] question cannot be solved, since the statement, then, is inappropriate, at least where the framework of world Y has been taken into consideration. As mentioned in our introduction, the Sapir-Whorf hypothesis does not seem compatible with the theorems of transformational grammar, since the latter presupposes competence to be an innate, general mechanism, equally valid for all natural languages. As soon as it is demonstrated, however, that notions such as deviance not only depend on universal mechanisms such as grammaticality — as soon as the impact of personal experience, attitude of the speaker, etc., are acknowledged — the problem is open to discussion again. The influence of language-specific phenomena on the cultural outlook of the language user, that is the hypothesis in its relativistic version, may then prove less refutable than has been supposed in the last decades.

REFERENCES Chomsky, N. ( 1 9 6 5 ) , Aspects of the theory of syntax. Cambridge, Mass. — ( 1 9 7 1 ) , 'Deep structure, surface structure, and semantic interpretation', in D. D. Steinberg and L. A. Jakobovits (eds.), Semantics. Cambridge, Mass. Dijk, T. A. van (1972), Some aspects of the text grammars, A study in theoretical linguistics and poetics. The Hague — Paris, Mouton (Janua linguarum, Series maior 63). Droste, F. G. (1973a), 'Model theory, logic and linguistics', Linguistics, 105. — ( 1 9 7 3 b ) , 'Qualifying and relational terms as linguistic universals', Linguistics, 118.

68

Flip G. Droste

— (1974), Ά note on paraphrasing', Linguistische Berichte, 29. Fillmore, Ch!" (1968), 'The case for case', in E. Bach and R. T. Harms (eds.), Universals in linguistic theory. New York. — (1969), 'Types of lexical information', in F. Kiefer (ed.), Studies in syntax and semantics. Dordrecht. Frege, G. (189?), 'Über Sinn und Bedeutung', Ζ. Philos. und Kritik, NF 100. Heidolph, Κ. E. (1970), . . . in H. Steger (ed.), Vorschläge für eine strukturelle Grammatik des Deutschen. Darmstadt. Katz, J . J. (1964), 'Semi-sentences', in J. A. Fodor and J . J . Katz, The structure of language. Englewood Cliffs, Ν. J., Prentice-Hall. Keenan, Ε. L. (1971), 'Quantifier structures in English', Foundations of language, 7. Kern, R. (1972), 'Zur Worstellung', Cahiers Inst. Ling. Louvain, 1. Kiparsky, P. and C., 'Fact', in M. Bierwisch and Κ. E. Heidolph (eds.), Progress in linguistics. The Hague — Paris, Mouton (Janua linguarum, Series maior 43). Kraak, Α., and Klooster, W. G. (1968), Syntaxis. Culemborg — Köln. Morgan, J . L. (1969), 'On the treatment of presupposition in transformational grammar', Papers 5th Reg. Meeting Chicago Ling. Society. Chicago. Paul, H. (1920), Prinzipien der Sprachgeschichte, 5th ed. Halle. Russell, Β. (1957), 'Mr. Strawson on referring', Mind, 64. Ruwet, N. (1973), 'How to deal with syntactic irregularities', in F. Kiefer and N. Ruwet (eds.), Generative grammar in Europe. Dordrecht. Seuren, P. A. M. (1972), 'Taaluniversalia in de transformationele grammatika', Leuvense Bijdragen, 61. Strawson, P. F. (1950), O n referring', Mind, 59. — (1966), Introduction to logical theory, 5th ed. London. Tarski, A. (1956), Logics, semantics, metamathematics, Oxford. Whorf, B. L. (1956), 'Science and linguistics' [1940], and 'Languages and logic' [ 1 9 4 1 ] in Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J . B. Carroll. Cambridge, Mass., MIT Press. Wittgenstein, L. (1968), Philosophical •investigations, translated by G.E.M. Anscombe. Oxford.

FERNAND J. VANDAMME

3

Language, Logic and Thinking

INTRODUCTION

Fishman (1965: 323) starts his systematization of the work done in the field of the Whorfian hypothesis by describing the hypothesis as formulating the view that the language characteristics have determining influences on cognitive processes. In these terms the whole discussion on the interpretation of how language determines thinking, and on the type of characteristics which determine language, stays open. Whorf was well aware of the problem. In his article 'Languages and logic' (Whorf 1941: 339) we find the assertion that languages determine tl;e way we dissect nature. But on the same page he remarks that language is in some sense a superficial embroidery upon deeper processes of consciousness. In this respect he says: 'It is the grammatical background of our mother language, which includes not only our way of constructing propositions but the way we dissect nature and break up the flux of experience into objects and entities to construct propositions.' The importance of this for science and logic he explains as follows: 'This fact is important for science, because it means that science can have a rational or logical basis even though it be a relativistic one and not Mr. Everyman's natural logic. Although it may vary with each tongue and a planetary mapping of such variation may be necessitated, it is, nevertheless, a basis of logic with discoverable laws. Science is not compelled to see its thinking and reasoning procedures turned into processes merely subservient to social adjustments and emotional drives.' But against this dominance of language on science, on logic and on thinking, Whorf immediately adds the following: 'The tremendous importance of language cannot, in my opinion,

70

Fernand J.

Vandamme

be taken to mean necessarily that nothing is back of it of the nature of what has traditionally been called " m i n d " . My own studies suggest to me that language, for all its kingly role, is in some sense a superficial embroidery upon deeper processes of consciousness, which are necessary before any communication, signalling or symbolism whatsoever can occur, and which also can at a pinch, effect communication (though not true agreement) without language's and without symbolisms aid.' When comparing these two assertions, we see that awareness of the problem, mentioned in the beginning, can be f o u n d in Whorf's own work: In which sense and t o what extent is it true to say that language influences or determines processes of the mind and to what extend is the language itself a product of such processes of mind? This problem of 'what determines the structure of language' — a problem which arises when one forwards the determination of science, logic, and thinking b y language — comes into different light when taking into account the data on the language structure of the DNA. In his interesting survey of linguistics and its relation to other sciences, Jakobson (1970: 526) remarks that George and Muriel Beadle's thesis that the code of the DNA is a language — the oldest language — is fully justified. 'Le titre de l'ouvrage de George et Muriel Beadle (1966) The language of life n'est pas une simple expression figuree et l'extraordinaire degre d'analogie entre le systeme de l'information genetique et celui de l'information verbale justifie pleinement la these directrice de cet ouvrage: " L e dechiffrement du code de l'ADN a revele que nous possedons un langage beaucoup plus ancien que les hieroglyphes, un langage aussi ancien que la vie lui-meme, un langage qui est le plus vivant de t o u s " . ' In his article 'Topologie et signification', Rene T h o m (1968: 22) underlines also the language-like structure of the DNA. His explanation of this structure is important: 'En raison des contraintes qu'impose la viabilite globale du systeme, la chaine d'ADN doit s'organiser en segments relativement autonomes et stables les segments "significatifs" qui peuvent d'ailleurs presenter une hierarchie de subordination fonctionelle, t o u t c o m m e notre langage se decompose en phrases, en m o t s et en lettres . . .' In other words, according to T h o m , the language-like structure of the DNA is caused by the requirements for the global viability of the system. In the framework of T h o m it is interesting t o investigate (a) which are the determinants of the language structure in general, and (b)

Language, logic and thinking

71

what can be learned from it in regard to the Sapir-Whorf hypothesis. We will take here the opportunity to refer to some resemblances between Thom and Piaget. Subsequently we will look at some facts relevant to the understanding of the relation between intelligence and language. In this light we like to expose our view on the relation between logic and language.

1. THE REDUCTIONISM OF THOM A N D SOME PARALLELISM BETWEEN THOM A N D PIAGET

Thom (1968: 224) argues that all interaction can be reduced to a phenomenon of resonance. Therefore the characteristics of interaction can be deduced from a mathematical theory of this physical phenomenon of resonance. In the history of the human striving for the realization of the unity of sciences, reductionism plays an important role. Kemeny and Oppenheim (1956) define reductionism as follows: Given two theories T1 and T2, T2 is said to be reduced to T1 if, and only if: a. the vocabulary of T2 contains terms not in the vocabulary of T l ; b. any observational data explainable by T2 are explainable by T l ; and c. T l is at least as well systematized as T2. An interesting discussion and evaluation is found in Oppenheim and Putman (1958). We have microreduction if T l (in the definition above) deals with the parts of the objects dealt with by T2. In Thorn's work we find a splendid attempt at the reduction of all types of interactions — group, individual, biological, communicational, chemical, etc. — to the characteristics of resonance. In this framework we can understand the following assertion about the characteristics of coding: Ί1 faut voir la [in the theory of resonance] l'origine des codages dans les phenomenes de transmission. Comme toute interaction repose en derniere analyse sur un phenomene de resonance, il importe, si Ton veut que l'operation soit sure et comporte le moins de "bruit" possible, que les resonances utilisees soient tres stables et aigues. Ceci impose l'utilisation des dispositifs isomorphes dans la source et le recepteur et une morphologie repetitive du message' (Thom 1968: 224). Still more important in this connection is the following interpretation of Thom on 'signification' in general or on the signification

72

Fernand J.

Vandamme

of a word or a sentence in particular: '. . . la signification exprime la possibilite pour un systeme d'adopter sous l'influence de perturbations externes des regimes correcteursqui annulent l'effet de la perturbation.' The signification expresses the possibility for a system to adopt the reactions which cancel the effect of the perturbation by external influences. Thorn explains this by the following remarks: Ά l'audition de l'ordre, la dynamique cerebrale subit un stimulus specifique S, qui la met dans un etat instable d'excitation; cet etat evolue ensuite vers la stabilite par sa capture par l'attracteur A, dont l'excitation engendre par couplage aux motoneurones 1'execution motrice de l'ordre' (Thorn 1968: 225). His assimilation of the p h e n o m e n o n of comprehension with dynamic resonance is also interesting in relation to his relating of signification t o resonance. The absence of signification of a text is never total. There are always at least more or less hazy resonances, but they cannot always attract concepts. 'Comme le m o n t r e bien l'assimilation du phenomene de comprendre a une resonance dynamique, l'absence de signification d'un texte n'est jamais totale: il se forme toujours des resonances plus ou moins fluctuantes, mais qui ne peuvent attacher l'esprit' (Thorn 1968:225). If we consider, as T h o m does, signification as a reaction scheme for canceling the effect of perturbations, then it seems worthwhile to compare this with Piaget's point of view on it. Piaget explains the structures he introduces by the striving of the system towards an equilibrium by autoregulation. The functional factors in this process are assimilation (a process whereby the behavior changes and incorporates new objects) and accomodation (the adaptation of the assimilation schemes to the several objects) (Piaget 1969: 56). In this respect the Piagetian notions are somewhat more specific than Thorn's. The problem is, can the notion of accomodation and assimilation be formulated in the resonance theory? We will not discuss this here. It is easy to foresee that in Thorn's framework, the grammatical categories will also be the result of interactions. In fact, this is what T h o m argues: 'En effet, l'interaction des logoi, qui s'exprime dans la syntaxe a l'aide des categories grammaticales, tire son origine des schemas d'interaction spatiale lies aux catastrophes elementaires' (Thom 1968: 231). In order to understand this point of view we must understand how T h o m interprets the notion 'logos' and 'elementary catastrophe'.

Language, logic and

thinking

73

'Logos' is the dynamical interpretation of the stability of a spatial form: 'la stabilite d'une forme spatiale, pour laquelle on recherchera une interpretation dynamique' (Ibid.: 226). An elementary catastrophe is a logos of a higher category, which determines the organization of two logoi which enter into conflict with each other. 'Lorsque, plusiers logoi sont definis sur le raeme substrat, ils finissent par entrer en conflit (et ici nous retrouvons Heraclite); mais tres souvent, le conflit entre ces differents logoi s'organise specialement suivant une configuration structurellement stable, elle-meme regie par un logos hierarchiquement superieur. Ce phenomene qu'on peut decrire et expliquer algebriquement peut etre qualifie de "catastrophe"' (Ibid.: 228). The next step is his hypothesis that the grammatical categories reflect the environmental catastrophes. L'interaction des logoi, qui s'exprime dans la syntaxe a l'aide des categories grammaticales, tire son origine, selon nous, des schemas d'interaction spatiale lies aux catastrophes elementaires' (Ibid.: 231). This is an important point of view in the discussion which interests us, viz. the problem of the influence of language on thinking and the problem of how language structure came about. Before developing this we want to turn to a comparison of Thorn's notion of logos, and Piaget's notion of structure. Like Thom, Piaget accentuates the dynamism in the stability. While Thom approaches this in terms of dynamical topology and so gets his 'logos', Piaget introduces the notion of structure for the explanation of the dynamicity in the stability. He stresses in this the dynamism of the structure. In the structure, the dynamicity is guaranteed by the characteristics of transformation and autoregulation (Piaget 1969: 8, 1 2 - 1 8 ) . This is so important for him that in a discussion on Foucault he concludes because he finds no transformations in Foucault's systems, that there are no structures in Foucault's approach (Ibid.: 117). Tcv Piaget, Foucault is 'a structuralist without structures', as the title of Piaget's chapter on Foucault's work declares (Ibid.: 111). (A criticism of Kuhn's paradigma theory can certainly be made along the same lines.) For Piaget the notion of levels and structures of higher levels, viz. structures which regulate the interaction between structures of a lower level, is also crucial (Ibid.: 15—16; 58). The resemblances of Thorn's notion of 'catastrophe' and Piaget's higher order structures is clear. Piaget explicitly relates what he calls confederation between structures of higher and lower levels (see

74

Fernand J.

Vandamme

above) t o autoregulation, a characteristic of each dynamical stability in Thorn's terminology. An important difference between Piaget and T h o m needs t o be mentioned here. Piaget considers each structure as closed. This does not exclude the possibility of constructing new structures; however, the original closed structure is kept intact in the process, and new characteristic laws are added {Ibid.: 8—16; 58 note 8). This approach, therefore, is contrary to the microreduction approach (see above) in which the properties, laws, and structures of the whole are explained by the laws, properties, and structures of the parts. In our view, this approach is also contrary to the synthesis hypothesis in which the structures are incompletely closed. This means that the interaction between structures, for instance by their belonging to higher order structures, can modify in part the original structure, without this changing the relative independence of each structure. Thus the autoregulation of each structure is n o t as absolutely interpreted as in Piaget's view. The introduction of incompletely closed structures, which implies graduations of closures with complete closure only in extreme cases, seems to us much more fructuous. Thorn's logos seems t o be more near our own point of view. As we have already said the next step of T h o m was to argue that grammatical categories reflect surrounding catastrophes. He illustrated this by showing how the elementary structures of phrases are related t o some elementary catastrophes in the environment (1968: 232—234). He thinks that in analysing these elementary catastrophes a certain universality of grammatical categories may be inferred, e.g. t h e logos of the verb is hierarchically higher than a noun since it organizes the conflict of the noun. He also thinks that the verb-noun distinction is universal. We can question, however, whether T h o m in his analyses has not been dazzled by the Indo-European languages f r o m whence all of his examples come. That his approach at least for the Indo-European languages has been fruitfull is easy to illustrate. Jakobson — in a lecture at the Franqui foundation (Leuven 1972) — mentioned how his study of aphasie confirmed some aspects of Thorn's typology of natural languages. The same, we believe, may be said of our study on simulation of natural language. We think, however, that prudence is needed as far as conclusions on universality are concerned. T h o m , in accordance with his resonance theory, accentuates that only configurations which are compatible

Language, logic and

thinking

75

with the internal structure of each logos can be present. As a consequence only certain configurations of catastrophes have become archetypes. Ί1 est clair que dans des cas concrets, oil le logos des actants est plus complexe, seules les configurations compatibles avec la structure interne a chaque logos pourront se presenter. De la vient que seuls certains types de configurations catastrophiques ont pris valeur d'archetypes; et a cet egard, les structures syntaxiques forment un ensemble encore plus restreint que les structures actantielles biologiques' (Ibid.: 334). In the framework of the resonance theory, the question arises about whether in each man, in each culture, the same logoi are necessary present. If in some cultures · other logoi are present or possible, which does not seem impossible, then it is possible or maybe even necessary that other archetypes are choosen. The second short remark we want to make about Thorn's conclusions concerns the verb-noun distinction. The verb, he says, introduces a space of interaction U and spectral figures (subject, object, destination, etc.). The noun, however, also indicates a space of interactions and spectral figures (Ibid.: 234—235). What is then the fundamental difference between the noun and the verb? Is it a question only of the hierarchy of the spaces of interactions? If this is the case then the universality of the fact that the logos of the verb is hierarchically higher than the logos of the noun is trivial, for the higher one, is the verb by definition. Returning now to the main line of thinking, what is then the final conclusion of Thom? It is that the origin of conceptual thinking — language, etc. — is another example of the principle of creation of the organ by its function. But how can one explain the retroactivity, viz. the possibility of a fact influencing its antecedents? In other words, how can one explain biological finality? Thorn's answer is continuity: 'Si l'on considere que le logos d'une espece biologique definit une figure continue de l'espace-temps il est normal de penser que les variations continues de cette figure au cours de revolution s'effectueront conformement a un principe variationel excluant les discontinuites des angles de cette figure: mais le lissage ainsi opere peut s'effectuer dans le sens du passe aussi bien que de l'avenir' (Ibid.: 237). This problem of necessity and finalism is also an important problem for Piaget. His solution is that the necessity of the constructions is one which is there only when the end is reached; it is a result

76

Fernand J.

Vandamme

of the search for equilibrium (in the sense of autoregulation) on the basis of the material won by abstraction (Piaget 1969: 55—56). In other words, the necessity is one of continuity, but, for Piaget, continuity is on the basis of accomodation and assimilation. Another way in which Thorn's approach is important for the Whorfian problem is the following. The biological human forms, human thinking,'human language are forms: logoi: structures which result f r o m the human interaction with other forms. The whole of these interactions can be described in terms of a general resonance theory. What we gain is the insight of a framework for describing general interaction. This framework is analogous, as far as some basic points of view are concerned, to the Piagetian one. Since it would be interesting in this framework to bring together facts of the several types of interactions, viz., world thinking, thinking language, etc., and then to make inferences on this basis on the Whorfian problem, we will a t t e m p t to do so, although rather informally. Something which is already gained in this approach is that if forms eventually make up different subclasses of catastrophes, then a general theory of dynamics which can describe the several possibilities of catastrophes would mean an important trespassing of the Whorfian relativism as applied to science in its dependency on language (Whorf 1940; 1941), for all the possible structures of language could at least theoretically be taken into account. Thorn's point that the grammatical categories reflect their environment goes also in the direction of the Whorfian approach — the relativity of language to the context. However, the universality of the categories is not thereby excluded, since certain aspects of the context in which language is created are universal. This point was also stressed by H. Putnam. Hereby, in a position between Chomsky's a priori universalism and the Whorfian absolute relativism, we have the possibility of generality or even universalism, b u t only on the basis of generality or universalism of certain contexts, certain functions.

2. SOME D A T A ON THE RELATION BETWEEN THINKING A N D LANGUAGE

If T h o m is right, then the structure of language results f r o m the catastrophes — the interaction between several forms. In consequence the same structure can result f r o m several different types of interaction. Thus we get the possibility of analogous structures

Language, logic and

thinking

77

between DNA and a natural language. It is, of course, interesting to know more exactly which conflicts between forms give rise to the language structure. Does thinking play an important role in the formation of natural language? What is the role of thinking on the interaction of logoi which are expressed in the grammatical categories of a natural language and what is the role of thinking in the process of interferring this interaction of logoi from the spatial interactions (Thom 1968: 231)? Can thinking be differentiated from natural language? There are some arguments which we think would indicate that to a certain extent both can be separated. We will here introduce some of these arguments. When treating signification and resonance Thom talks about the imperatives. He remarks that the definition of execution of the order exists independently and before the word. Education has connected the definition of the execution of the order with the phonetic recognition of the word (see above). But is not this, a first form of (conceptual) knowledge? And this can certainly be generalized. In Thorn's resonance theory is it not necessary that the logos is present at least partly before the resonance can take place? This explains why the interaction of logoi can be, and need to be, expressed in grammatical categories in order to make the resonance possible. But, for this, he says, the logoi themselves must be based on the spatial interactions at least partly independent of grammatical categories. The problem, and that which seems to be the crucial point of Thorn's approach, is: how can this happen? Piaget also believes that intelligence precedes language phylogenetically as well as ontogenetically. Facts on the deaf-and-dumb and on the intelligence of the higher apes indicate this (Piaget 1969: 82, 76—80). But, Piaget remarks, even if it is the result (as it is in part in the Thomsian approach) of the partly structural knowledge, is natural language important for further structuring of the intelligence. That intelligence plays an important role in the construction of language is evident if we think at the role of abstraction which is necessary for bringing together several tokens into the same type — abstraction without which language would be impossible. But even if it is agreed that language and thinking or intelligence have a different origin, e.g. if intelligence precedes language, the question is still open about their interrelation when both exist. Are they inseparable, identical? Piaget denies this. The studies of M.D.S.

78

Fernand J.

Vandamme

Braine, Inhelder, Sinclair, and Bovet give information on their interrelation. Consider, for instance, the following important conclusions of H. Sinclair (Piaget 1969: 8 3 - 8 4 ) . Sinclair saw that children who have reached the intelligence stage of conservation have a different language use than children who have not reached that level. The last group uses more scalar oppositions, while the second group uses more vectors (gradual oppositions: older than, etc.). Language lessons (which were only fructuous in one out of ten cases) t o learn to use vectors, did not help make the transition to the stage of conservation. In fact most of the children were able to understand sentences with vectors, but could not use them. This clearly indicates how intelligence is relatively independent on language development while its evolution strongly influences the evolution of language. By the way it is interesting that children w h o were n o t able t o use vectors were able to understand them. This is another confirmation of the thesis that we elsewhere elaborated (Vandamme 1972) that language recognition must be differentiated f r o m language production. Brown (1970), in a study on the development of the vocabulary of the child, tried to explain the general tendences. He observed that usually the vocabulary of a child develops f r o m more concrete to more abstract. Psychologists, as he remarks, believe in general that mental development is f r o m the abstract to the concrete. This opposite evolution of intelligence and vocabulary indicate strongly the relative independency of intelligence on language. In an article on inferences f r o m linguistic data to nonlinguistic ones, Greenberg (1956: 473) points to the following: 'Viewed historically, there is a greater plausibility in the notion that the stock of EMU's [elementary measuring units] tends to adapt to nonlinguistic changes rather than to linguistic shifts. This is certainly true in technology, but even in philosophy, science and religion it seems likely that the appearance of new EMU's is chiefly a response to needs which have arisen at the particular point of development of the subject matter itself . . .' And so we can go on enumerating facts which indicate the independency, and primacy of thinking or intelligence above language. On the other side it is clear that language plays an important role in the life of mankind. Without communication no system can subsist and language is an efficient type of communication. It is also needless to repeat the influence particular language facts have at least heuristically on intelligence, attention attraction, etc. When looking

Language, logic and

thinking

79

at the history of science, we repreatedly observe that when two types of phenomena are related to one another, one strongly tries to reduce the one. Applied in this case, it is the problem of the reduction of intelligence t o language or vice versa. We always welcome attempts at reduction, because the more reduction there is, the more interaction between separated phenomena is brought into the open. What strikes us in the matter is that we usually find partially successful reduction in the several directions. Intelligence is partially successfully reduced to language (Whorfian hypothesis), while others reduce language to intelligence (Chomsky according to Piaget, Lakoff, etc.). (As far as the relation between language and economics is concerned, we find analogous tendences: the reduction of language to economy [Martinet, Apostel, even de Saussure, etc.], the reduction of economics to language [Talcott Parsons, Rossi-Landi, etc.]. This is dealt with more at length in Vandamme 1972b.) The partial success of the opposite reductions seems to me to suggest that we need not only reduction although this is better than nothing, but synthesis as well. We need the confrontation and interrelation of the theories of the several systems of phenomena now seen as relatively independent units in an equilibrium by autoregulation — to put it in Piaget's terms — or as two different logoi in a catastrophic relation with each other in the terms of Thorn. Our point is that the understanding of the interrelationships between the several structures (the catastrophic relation) makes possible a better understanding of each logos. This point of view, I think, is what is characteristic for the synthesis idea and what makes more fructuous the study of several systems, in this case language and intelligence or thinking.

3. LOGIC A N D L A N G U A G E

In the last century and in particular since Carnap, much has been said on the relationship of logic and language. Carnap was without doubt one of the most illustrious proponents of the linguistic theory of logical truth. In this theory, logic is true by virtue of the linguistic rules. It is also true that an important list of scholars, such as Dewey and Quine, have strongly critized this. (We discuss this more thoroughly in 'Taal en logica' — 1973). The opposite tendency also exists. There are people who try to base language on logic, as is clearly the case with Lakoff. Piaget

80

Fernand J.

Vandamme

attributes this also to Chomsky and comments: 'It is now more or less clear that language is not the base [the origin] of logic and Chomsky rightly bases language on logic' (Piaget 1969: 84). Piaget also argued ardently that language is important too in the formation of the logical system. In his 'Les operations logiques et la vie sociale' (Piaget 1945), he says: 'The individual functions as well as the collective functions are necessary in the explication of the conditions necessary for the logical equilibrium: 'Les fonctions individuelles et les fonctions collectives s'appellent les unes sur les autres dans l'explication des conditions necessaires a 1'equilibre logique. Quand a la logique elle-meme, elle les depasse toutes deux puisqu'elle releve de Tequilibre necessairement ideal auquel elles tendent l'une et l'autre. Ce n'est pas a dire qu'il existe une logique en soi, qui commanderait simultanement les actions individuelles et sociales, puisque la logique n'est que la forme d'equilibre immanente au processus de developpement de ces actions memes . . .' (Piaget 1945: 32—33). He points also out that language plays an important role as one of the collective functions. For instance, in the conservation of totalities in which the complete reversabilities operate, the individual symbolism is not strong enough: 'D'autre part, comment l'individu parviendra-t-il a conserver les totalites sur lesquelles portent ses operations, c'est-a-dire a atteindre la reversibilite complete? Les processussensori-moteurs ne sauraient, en effet, suffire a expliquer la reversibilite, puisqu'ils sont essentiellement irreversibles et ne parviennent a s'inverser qu'en partie et sans doute sous Taction de facteurs d'ordre superieur. La reversibilite complete suppose le symbolisme, car ce n'est que par une reference a revocation possible des objects absents que l'assimilation des choses aux schemes de Taction et Taccomodation de ceux-ci a celles-la parviennent a une equilibre permanent et constituent ainsi un mecanisme reversible. Or, le symbolisme des images individuelles est beaucoup trop fluctuant pour conduire a ce resultat. Un langage est done necessaire et nous retrouvons ainsi les facteurs sociaux.' So language is important for the process of 'objectivation' — of making a notion or a concept more stable, more durable. In modern methodology too, much objectivity is interpreted as intersubjectivity (Nagel 1967; Popper 1962; Vandamme 1972c). This means that the social aspect of knowledge is more accentuated. Communication, and especially language, plays an important role in this intersubjectivity. Thus the importance of studying the inter-

Language, logic and

thinking

81

relations between logic and language is accentuated. But, as we already said before, the whole reduction of systems with its problem of the direction of reduction is only interesting as far as it brings information on interrelations between systems. Therefore, the discussion about whether language is based on logic or logic on language is, as it seems to us, not the interesting question. For even if the reduction in both directions may be partially successful, the important thing is that we have to do with two distinct, relatively independent systems which are in constant interaction with each other, and their forms result from this interaction between systems. In other words, at each moment, although the interaction may motivate change, the independency, the distinct type of reactions is conserved. Let us accept the hypothesis that logic and language are two relatively, independent systems, although through communication (language) there is influence on the construction of one's logic and vice versa. The primary function of the language system is communication of ideas and feelings. The primary function of logic and thinking is the systematization of the ideas and concepts. As we noted already, the social functions are important for an individual's thinking. Therefore it will be clear that it is highly probable that an individual will communicate about the most important cognitive differentiations he makes. As language is one of the most important types of communication, it is probable that he will communicate these differentiations in language. And language is always an open system in at least two senses: 'New permitted combinations with permitted elements can be constructed', and 'New permitted elements can be introduced'. This means that it will always be possible to introduce new differentiations if the community wants it. Taking into account the importance for human knowledge of the objectivity or, in modern terms, of intersubjectivity, it is clear that this communication possibility by language (language being the most efficient communication system of thoughts) will be important. The opposite is also true, viz. if one wants to, it is possible not to take into account some differentiations which can be made. We only have to look at the sciences to find plenty of examples of this latter possibility. When there is a scientific reconstruction of a term, in most cases some information about the term or some differentiating characteristics of the term are not taken into account; sometimes these are even replaced by other differentiations. The appearances of

82

Fernand J.

Vandamme

scientific reconstruction of terms in science are, obviously, immense. So we could say that language determines a sphere of possible constructions of differentiations which can be communicated. Language is a device for communicating such constructions. The specific sciences are possible constructions inside this sphere. Eventually they can force the language sphere to be widened, or they can neglect possible differentiations. This neglect can even be useful. Another important point is that the combinations of the characteristics communicated through the natural language are not restricted by the natural language: Natural language permits contradiction. It never restricts the possible combinations of the differentiations, characteristics, or relations which it communicates. It does put restrictions on the combinations of word categories, in other words, on communication categories, but never on the content of what is communicated. For illustrations on this and a more elaborated exposition see Simulation of natural language (Vandamme 1972a), and also O n negation. An interdisciplinary study' (Vandamme 1972d). Even if one considers the grammatical categories as forms based on the spatial interaction schemes (Thom), it does not follow that these categories do not get a relative independency whereby the interrelations of the content are not restricted by the spatial forms. (Also it does not follow that the categories cannot become relatively independent on the spatial interaction schemes from which they originated.) The accent here is on restriction. The grammatical categories indicate what type of conflict the speaker sees in the content of the terms, but this is positive information, not a restricting operation.

4. CONCLUSION

We believe that here we may be brief. Many studies on the problem of the relation between knowledge, science, and language are made with the goal of identifying them, reducing the one to the other, or indicating a kind of priority. What we believe to be more fructuous is the study of (a) the feedback relations between them, (b) their interactions analogous with other structures and systems, and (c) their determination by these, not forgetting that the systems are relatively independent.

Language, logic and

thinking

83

ACKNOWLEDGEMENT

It was Roman Jakobson who attracted my attention to the topological approach to language by Rene Thorn in his article 'Topologie et signification', Lage de la science, 1 (1968), No. 4. We are very grateful to him for this. We like to thank Mrs. C. S. M. Merkx-Van der Kamp for her indispensable help in collecting the relevant literature on this topic.

REFERENCES Beadle, G. and H. (1966), The language of life: An introduction to the science of genetics. New York. Brown, R. (1970), 'How shall a thing be called?', in R. Brown (ed.), Psycholinguistics. Selected papers. Carroll, J. B., ed. (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf Cambridge, Mass., MIT Press. Fishman, J. A. (1965), Ά systematisation of the Whorfian hypothesis', Behavioral Science, 5. Greenberg, J. H. (1956), 'Concerning inferences f r o m linguistic to nonlinguistic data', in H. Hoyer (ed.), Language in culture. Chicago. Also in Sol Saporta and J. R. Bastion (eds.), Psycholinguistics. New York, 1961. Jakobson, R. (1970), 'La linguistique' in Tendances principales de la recherche dans les sciences sociales et humaines, l e partie: Sciences sociales. Paris — The Hague, Mouton — Unesco. Also 'Linguistics' in Main trends of research in the social and human sciences, Part one: Social sciences. Paris — The Hague, Mouton — Unesco, 1970. Kemeny, J. G., and Oppenheim, P. (1956), 'On reduction', Phil, studies, 7: 6-19. Nagel, Ε. (1961), The structure of science. London, Routledge and Kegan Paul, Ltd. Oppenheim, P., and Putman, H. (1958), 'Unity of science as a working hypothesis', Minnesota studies in the philosophy of science, 2, 3—36. Piaget, S. (1945), 'Les operations logiques et la vie sociale', in Melanges d'etudes economiques et sociales, offerts a M.M.E. Folliet et L. Hersch. Geneve. — (1969), Structuralisme. Meppel (The Netherlands), Boom. Popper, K. (1934), The logic of scientific discovery. London, Hutchinson. Thom, R. (1968), 'Topology et signification', L'age de la science, 1: 4. — (1969), Sur la typologie des langues naturelles. Essai d'interpretation psycholinguistique. Vandamme, F. (1972a), Simulation of natural language: A first approach. The Hague — Paris, Mouton (Janua linguarum, Series maior, 50). — (1972b), Taal en economie. Tilburg (The Netherlands), Gianotten.

84

Fernand J.

Vandamme

— (1972c), Een inleiding tot wetenschapsfilosofie (mimeographed). — (1972d), On negation. An interdisciplinary study. — (1973), 'Taal en logica', Communication and Cognition (Ghent), 6. Whorf, B. L. (1940), 'Science and linguistics', Technol. Review, 42: 2 2 9 - 2 3 1 . Also in Carroll, J. B. (ed.), 1956. — (1941), 'Language and logic', Technol. Review, 43: 2 5 0 - 2 5 1 . Also in Carroll, J.B. (ed.), 1956.

MO KÄA

4

The Logic of non-European Linguistic Categories

Centuries of misunderstanding, strife, and bloodshed have resulted from the European not knowing what 'primitive' peoples were really saying and thinking. Because they do not think the same, the European has often considered non-Western people as 'savages' and 'wild men', not much better than animals. There are basic differences in ways of thinking between civilized and primitive peoples, but it requires no bloodshed to try to understand them. It is almost always the case in anthropology and linguistics that the data on primitive peoples is analysed from a Western point of view. The aim of this paper is to give voice to the viewpoint of the people involved, especially where it concerns their own language, thinking, and culture. The words 'Indian', or 'American Indian', or 'Amerindian', as I use them, refer to the people, viewpoints, languages, and cultures native to the Americas before the arrival of the Europeans, and which can still be found to some small extent in the modern-day descendants of these peoples. I use the words 'European', 'Western', 'civilization', etc., as broad generic terms referring to the languages, cultures, thinking, etc., which originated in Europe and now includes that which dominates the Americas, having largely replaced the original Amerindian languages, cultures, and races. Indo-European and Semitic languages feature a system of classification of nouns based on distinction of sex: masculine, feminine, and sometimes neuter. In American Indian languages, however, sex distinction is not usually the subject of classification. The Algonkian Family of languages, for instance, features a system of gender based on distinctions between what linguists have called 'animate' and 'inanimate' gender. Bloomfield's definition of Algonkian 'animate' includes 'all persons, animals, spirits, and large trees, and some other

86

mo Κάα

objects, such as tobacco, maize, apple . . ., feather, bird's tail, kettle, pipe for smoking, snowshoe' (Bloomfield 1946: 94). In addition to 'animate' and 'inanimate', various other European designations for these categories include 'rational' and 'irrational', 'having a soul' ('souled') and 'soulless', 'living' and 'lifeless', 'alive' and 'dead' [Italian: animato/inanimato; Dutch: levend/levenloos bezield/ onbezield; German: belebt/unbelebt, beseelt/unbeseelt, lebend/tot; etc.] J osselin de Jong 1913: 155, 163). The basic assumption of the European is that Indians classified all material phenomena into two well-defined categories on the basis of whether it was 'living' or 'nonliving'. Therefore, one of the first questions that occurs to the European mind is: How could such 'inanimate objects' as feather, pipe, kettle, and so forth, possibly be classified as 'animate'? And perhaps he concludes that Indians must not be logical about it. Let us see what these terms mean. Animate/inanimate. The English dictionary (C.O.D.) definition of animate is 'living'; the definition of inanimate is 'destitute of life; not endowed with animate life, as inanimate nature (outside the animal world)'. According to the definition the meaning of 'animate' as 'living' would include persons, animals, and 'living plants'; but the meaning of 'inanimate nature (outside the animal world)' also includes 'living plants', as well as mineral phenomena, etc. Thus the terms 'animate' and 'inanimate' are somewhat ambiguous concerning the status of plants, and this makes the English designations of the Algonkian gender categories somewhat confusing. Living/lifeless. 'Living' most likely includes living persons, animals, and plants; and 'lifeless' would include mineral matter as well as objects composed of dead plant and animal matter. Alive/dead. 'Alive' would include only living persons, animals, and plants: 'dead' would include all dead persons, animals, plants and mineral matter. Souled/soulless: 'Having a soul' would, according to many Europeans, include only human beings, but others would extend this to include animals as well, or perhaps only some animals; 'soulless' would include minerals, plants, and possibly, or probably animals. The status of animals, here, might be somewhat doubtful in the European mind. Rational/irrational. 'Rational' beings would include human beings, and might include some of the more 'intelligent' animals, but might exclude animals altogether, and in the minds of some it would

The logic of non-European

linguistic

categories

87

exclude some human beings as well. It appears that the different European designations of Algonkian gender not only do not agree with each other, but they are in themselves ambiguous. There is, nevertheless, one outstanding characteristic that they all have in common: each pair of designations is an absolute polarity of positive and negative, black and white categories diametrically opposed to each other: animate/inanimate, rational/irrational, souled/soulless, living/lifeless, alive/dead. This absolute logic of opposing categories is a complete contradiction to the ambiguity and vagueness as to what it actually is that is being categorized! One might ask whether this is 'rational' or 'irrational'. It is important to keep in mind that these labels reflect the way Europeans have interpreted the Algonkian categories, which is based on their own way of looking at things, in other words, the European World View. Let us examine some aspects of this world view which may influence their interpretation. Western man has transformed nature and created an artificial man-made world which he calls 'civilization'. This civilized world consists almost overwhelmingly of 'lifeless, inanimate' objects. This artificial world, which is his 'natural habitat', may color his concept of what is 'natural', and what is 'nature'. Western man seems to divide existence into a kind of polarity of civilization and nature, the latter including plants, animals, and people living 'in the natural state'. He speaks of 'the world of nature' as of literally another world, a sort of wild, unknown, and perhaps hostile land existing beyond the limits of known civilization. He speaks of 'going out into nature' as of going on a long trip or safari into the unknown, and of going 'back to nature' as a retreat from civilization, or even as going backwards. The Western mind may be more likely to consider automobiles and trucks in busy traffic, and the wheels of industry as much more dynamic manifestations of 'life', than he would some of the less noisy and more subtle natural phenomena which he might call* 'inanimate nature'. The Western anthropologist says that 'to primitive man all nature seemed to be endowed with life' (Josselin de Jong 1913: 161). Notice that 'endowed with life' means'given life', as if 'life* were somehow not a natural characteristic of 'all nature'. This reflects a concept of nature as bisected between 'living' and 'nonliving' phenomena. His relatively natureless cultural context, therefore, may color and influence his interpretation of the Algonkian gender categories.

88

mo Κάα

In Algonkian languages an animal, a tree, tobacco, or a kettle would be referred to with the same personal pronoun form (inflection) used to indicate a person, in other words, the equivalent of 'he' or 'she' (Yegerlehner 1954: 290). To the European mind, however, tobacco, kettle, and so forth are considered 'objects', and along with tree and animal would each be referred to with the impersonal pronoun 'it'. Relatively speaking, thus, some things which are more personalized by Indians are more 'objectified' by Europeans. Because of his overwhelmingly material culture which places an emphasis on the possession of material goods, the European seems to have a bias for the category of 'inanimate objects'. But it is not only man-made things that are considered 'objects' to be possessed or manipulated: plants, animals, human slaves, land, nature, the whole earth have all been 'objectified'. Thus, 'objectification' appears to be an active principle in the European mind, and may therefore influence his interpretation of the gender of nouns in Algonkian languages. I would now like to review the various points which may bear on our understanding or misunderstanding of Algonkian gender: (1) the vagueness and ambiguity of the European terms, the absolute character of the European labels, whether the logic involved is rational or not, (2) the Western tendency to objectify phenomena, and (3) the alienation from nature which is implicit in the European world view. Let us now consider the American Indian factors, concepts, and world view which are more relevant to our understanding of Algonkian gender. The Indian was completely involved in nature; mature was his life and nature was life itself. The Indian was an 'outdoors' person; 'indoors' for him was usually only a small and often only temporary shelter for sleeping or against inclement weather, and was not primarily a storehouse for material objects, of which he possessed few. Since he was more active, all other phenomena seemed more active as well, and this more dynamic inertia was reflected in his language. The so-called 'polysynthetic' sentence structure characteristic of Amerindian languages centers around the dynamic action word or verb. Also, many concepts which European languages objectify as nouns, such as 'the lightning', 'the wave', 'the flame', Amerindian languages, such as Hopi, express as verbs, saying something like 'waving occurs', 'flaming occurs', 'lightning occurs'. In Nootka 'all words seem . . . to be verbs' (Whorf 1956: 215). What Europeans would express as 'a house'. Nootka would say something like 'it

The logic of non-European linguistic categories

89

houses', or 'housing occurs'. The Indian, thus, did not place emphasis on material phenomena. He was not 'materialistic' but activistic, and not so much impressed with living 'things' as with the living spirit. A Blackfoot Indian. Crowfoot asks and answers: 'What is life? It is the falsh of a firefly in the night. It is the breath of a buffalo in the winter time. It is the little shadow which runs across the grass and loses itself in the sunset' (McLuhan 1971: 12). Even death was not the end of life, but merely another phase of existence. Instead of saying 'He is dead', which sounds so final. Indians, such as the Hupa and Yurok, would say 'He is decayed (Curtis 1924: 42), which expresses a dynamic organic process, rather than a static material state. And Seattle (Dwamish chief) has said: '. . . the dead are not powerless. Dead — I say? There is no death. Only a change of worlds' (McLuhan 1971: 30). Coming back to the phrase: 'to primitive man all nature seemed to be endowed with life', the point is, that the primitive mind 'draws no sharp line between "living" and "non-living" phenomena (Josselin de Jong 1913: 209) as does the European mind. The Indian did not separate, sort, and classify natural phenomena into categories or lists of individual material 'objects'. Rather he saw the activity, interactivity, and interrelationship of all natural phenomena as part of a whole living organismic universe in which he himself was also completely involved. Just as among the Lacandones, 'clouds, air, rocks, animals, rivers, and stars — in short, everything — were believed to be equally alive with mankind. Foot races determined whose strength was greater than that of the wind; a mountain was climbed to conquer its spirit; when an animal was slain, its forgiveness was fervently sought' (Gallenkamp 1959: 119). Therefore, as seen from the context of the Amerindian world view, I feel that the materialistic considerations and the artificial separation implied in the designations animate/inanimate, living/lifeless, living/dead, etc., are not native American Indian concepts, but are a projection of the European world view. The concepts animate/inanimate, etc., demonstrate the logic of absolute generalizations or rigid generic categories based on the discrimination of a black/white polarity. All 'animate' phenomena must go into the 'animate' category, all 'inanimate' phenomena must go into the 'inanimate' category. This process is more appropriate to the mechanical sorting of material objects. Indians, however, were not sorting material objects, but were more

90

mo Κάα

activistic, pragmatic, and relativistic. When they had to move f r o m one camp site to another, they had to re-orient themselves each time relative to completely changed conditions. The h u n t e r on the trail of an animal had to find his way back t o the tipi relative to a constantly changing path through the pathless forest. There were many new and changing factors in the daily lives of Indians. In this naturalistic context, therefore, the Indian had a more relativistic world view. Which brings me to m y next point, which is that Algonkian gender is not based on a classification system of absolute categories. The first point is that the two Algonkian 'classes' do not represent homogeneous categories. The so-called 'animate' gender includes p h e n o m e n a which Europeans categorize as 'animals', b u t which is not a homogeneous category in any sense, for it includes human beings, quadrupeds, birds, fishes, insects, reptiles, crustacea, and so f o r t h (Joselin de Jong 1913: 158). Although this gender includes whole persona and animals, it does not include most of the human b o d y parts, b u t does include m a n y b o d y parts of animals. The 'animate' gender also includes some large trees, b u t n o t all trees: and includes various species of fruits and seeds, but not all such species. Sometimes a tree may be included in one gender, but its fruit and seeds in the other. Names of parts of plants are somewhat equally represented in both genders. The so-called 'animate' gender also includes many so-called 'inanimate' objects, such as feather, horn, kettle, metals, snowshoe, snow, thunder, sun, moon, and stars (Bloomfield 1946: 94; Joselin de Jong 1913: 1 2 4 - 2 2 3 ) . Both 'animate' and 'inanimate' phenomena are f o u n d in the so-called 'animate' category, and both 'living' and 'non-living' phenomena are f o u n d in b o t h categories. In fact they cannot even be called 'categories' in the European sense, for the word 'category' is already t o o m u c h of an absolute generalization. Algonkian must be seen f r o m the point of view of relativity. A second point is that there is n o t always agreement between one Algonkian language and another as to the gender of the same 'object'. The majority of objects of daily use, furniture, tools, means of transport, weapons, etc., are 'animate' in Blackfoot, but mostly 'inanimate' in Ojibwa and Cree. Wind is 'animate' in Cree and Ojibwa, and inanimate in Blackfoot. Earrings are 'animate' in Cree, and 'inanimate' in Ojibwa and Blackfoot; and trousers are 'animate' in Blackfoot, 'inanimate' in Ojibwa, and 'indifferent' in Cree (Joselin de J o n g 1913: 1 4 0 - 1 4 6 ) . Further points are that in the same language the same word can

The logic of non-European linguistic

categories

91

have different genders and the speaker does have some choice. And, the change of gender can sometimes, but not always give a different meaning to the word (Josselin de Jong 1913: 147, 159). Perhaps the only thing that the European can see which looks "logical' about these genders is that all whole 'animals' have the same gender, and that is apparently the rationalization for calling that gender 'animate'. But what about all those other 'things'? Algonkian gender seems to be a hopeless confusion, which he can not understand. 'Why, for instance, is showshoe "animate", and moccasin "inanimate"? he asks. It may be somewhat apparent by now that perhaps European concepts, terminology, and logic have little application to Algonkian gender. And even if the Indian himself called one gender 'living' because that was the'closest word he could think of in English, it must be understood that he does not have the same concept of that word that the European does. What is it then that the Indian really did mean, if he didn't mean what the linguist implied he did? After all, there must have been some reason or guiding principle which caused the Algonkians to use a gender system, and which continued to guide them in deciding the gender of new words, for objects introduced after the arrival of the Europeans, such as those which are added to their languages. Well, if it doesn't make much sense by Western logic and reason, let's try the Indian's logic. In considering a better designation for Algonkian gender, let us see what kind of generalization we can make or perceive. The first and most obvious generalization is that all persons and animals are included in one gender. A second generalization is that 'the names for animals are only employed as animates, while the.objects are referred to as wholes and species, but the gender must change when it becomes necessary to speak of separate members (Joselin de Jong 1913: 145). A third, and perhaps less all-inclusive generalization is that spirits and large trees tend to have 'animate' gender. A fourth generalize is that in those cases where gender choice is at the option of the speaker, choice of 'animate' gender may 'generally be traced to a particular respect paid to certain inanimate bodies either from their real or fancied proporties, the uses to which they are applied, or the ceremonies to which they are dedicated' (Ibid.: 159). A fifth generalization is that all the phenomena in the 'animate' category inflect the verb with the more personal pronoun form. A sixth generalization is that the 'animate' gender surpasses by far the 'animate' gender in the number of nouns (Ibid. : 151).

92

mo Κάα

The fact that this gender is the numerically smallest gender as well as the hint of 'particular respect', suggest that this gender is a grouping of specially chosen phenomena. Keeping these six points in mind, and knowing something of the Indian mind, I am going to suggest a generalization which I feel has a better conceptual fit to this gender than does the word 'animate' or any of its other European equivalents. I will call this gender: more spiritually relevant and personalized beings, things, and phenomena. Let me emphasize that this is not a label, but a generalized concept. Although it may seem somewhat vague now, I will try to clarify it. This concept' is not as easy to put one's finger on as if it were called 'animate'; you cannot give it a number designation, or locate it on a Cartesian coordinate system. It is not one pole or a polarity; it's not a positive which has a negative. It may evade European logic and perhaps European logic may even deny it. If we must have a shorter designation, we could refer to it temporarily as the 'more special' gender, and the other gender as the 'more common' one. But 'more spiritually relevant phenomena' does not mean that everything in this category has a spirit or soul, and everything in the other category has no spirit or soul. The point is that each thing within this gender has some special characteristic(s) or personality trait, or something unique to itself for which reason it merits this gender in the collective consciousness of the Indian. The reasons are various and are a matter of relative emphasis. Due to time limitations I will concentrate on only a few of those phenomena whose inclusion in the 'animate' category tends to escape European logic. Notice that in the designation 'more spiritually relevant and personalized phenomena' that I did not say 'personified'; I said personalized. The Indian did not categorize phenomena because he was much more impressed with the different personalities of nature. Okute (or 'Shooter'. Teton Sioux) says: 'The reason why Wakan tanka does not make two birds, or animals, or human beings exactly alike is because each is placed here . . . to be an independent individuality and to rely on itself . . . From my boyhood I have observed leaves, trees, and grass, and I have never found two alike. They may have a general likeness, but on examination I have found that they differ slightly' (McLuhan 1971: 18-19). Trees are very 'spiritually relevant beings' to the Indian. He does not 'personify' nature in the sense that when he looks at a tree he sees a person or human features. When he looks at a tree, he sees a tree, of course. He sees a tree and appreciates its unique 'tree-ness'. .

The logic of non-European linguistic

categories

93

To the European mind the tree is an 'object', a thing: It is an inpersonal 'it'. But the Indian does not see the tree as an 'object'. He sees the tree as much a 'living being' as himself — not a human being, but its own being. And he respects it just like he respects a human being. He respects the tree because he knows the tree, he knows its nature. He knows the seasons of the tree, he knows its leaves, flowers, and fruit, its bark and sap. He knows the sight, sound, touch, taste, and smell of the tree. He can tell one tree from another, and each tree is an individual to him. He knows that just as each person has his own personality, so does each tree. The Indian does not personify nature, he merely knows its personality and he respects it. The linguist was puzzled about why snowshoe was 'animate', whereas moccasin was 'inanimate' (Joselin de Jong 1913: 145). But let us see why snowshoe might be relatively 'more special' than moccasin. Algonkian earth is either 'animate' or 'inanimate' (Ibid.: 137), but snow is 'animate'. Thus snow is in some way more special than earth. After all, snow is sent down from above by the Great Spirit. It doesn't wash away like the rain ('inanimate'), but stays a little longer, covers the earth, transforming the whole world of nature. And snowshoes can carry the Indian up above the ground and let him walk on top of the snow which was sent down by the Great Spirit. And so, relatively speaking, snowshoes are 'more spiritually relevant things' than moccasins (Ibid.: 138). Perhaps something like this escapes European logic because snow to him has such completely different associations. After all, what is snow to a civilized man? Snow is something which he transforms and shovels away so he won't have to use any snowshoes. Snow seems to be more of a nuisance to him. What nature and natural phenomena mean to you depends very much on your particular world view. In the materialist world snow is more 'objectified' matter, in the activistic world snow is 'more spiritually relevant phenomena'. The linguist says that stones are 'animate'. Let's see what it is that's more special about stones. In a universe so full of activity and movement, stones have a very unique character trait: They are so silent. And the Indian believed profoundly in silence. Silence is 'the absolute poise or balance of body, mind, and spirit'. The fruits of silence are 'self-control, true courage or endurance, patience, dignity, and reverence. Silence is the comcr-stone of character.' The old chief Wabasha (Santee Sioux) said: 'Guard your tongue in youth, and in age you may mature a thought that will be of service to your people' (Seton 1963: 17). Stones must be very wise indeed.

94

mo Κάα

Can you imagine what it means t o a medicine man to have those very silent and very wise stones 'talk' to him in his dream? Brave Buffalo (Sioux) says: 'In my dream one of these small round stones appeared to me and told me that the maker of all was Wakan tanka, and that in order to honor him I must h o n o r his works in nature. The stone said that in m y search I had shown myself worthy of supernatural help. It said that if I were curing a sick person I might ask its assistance, and that all the forces of nature would help me work a cure' (McLuhan 1971: 16). A stone is the altar of sacrifice t o the manitous, a very 'specially relevant spiritual phenomena'. But what is a stone t o the Western mind? A stone is a dead object, something y o u step on, or kick out of the way, or throw at someone to express your hostility. 'Sticks and stones can break your bones.' But, why did the linquist say that Indians think stones are 'animate', if the trait which is most outstanding in the mind of the Indian, is their very 'inanimate-ness' (Josselin de Jong 1913: 221)? Perhaps the European has a more animistic mind. The list of 'more spiritually relevant phenomena' goes on, but time is short. Let us summarize this special gender of phenomena. These are things which have more special significance to the individual Indian, to the collectivity of the tribe, in the American Indian world view, or to the primitive mind in general; things which to the Indian mind play a more active role in his thought, feeling, emotion, and to his biological, social, or cultural life; things which he feels have more personality, or more relevant personality, and about which he has a more personal feeling than something which does n o t have this gender. 'Animate' refers more t o the potential for self-directed mechanical movement, like animals and men. The Indian did not attribute such qualities t o 'inanimate' things. Men and animals, of course, as beings who can move about of their own free will, do have more opportunities to express more varied personalities than 'inanimate' things. But the Indian also allowed each 'inanimate' thing t o express its own unique personality in the way which was most natural or unique t o it. Whether it was 'animate' or 'inanimate' was not his concern. A stone is more special because it has more personality to the Indian, and is perhaps even wiser than his own brain, which has more common gender. Flowers are beautiful to look at, b u t perhaps they are mere ostentations and vain, and therefore of more c o m m o n gender than a stone. There is plenty of logic in the mind of the Indian, and it is relative indeed. In fact, the naturalistic beliefs of

The logic of non-European

linguistic

categories

95

Indians were based much more on relative logic, than on a fixed body of absolute dogma. We have seen how the same words can have almost completely opposite associations or include different concept fields which are highly relative to the particular culture and world view. A few such words are 'nature', 'life', 'death', 'tree', 'being', 'personality', 'snow', and 'stone'. My basic assumption has been the relativity of language, thought, and culture, and I believe I have somewhat demonstrated the validity of this assumption. I think I now have presented enough evidence on which to base my argument and make my point. Western linguists have applied their own labels to the genders of American Indian languages, but these labels tell much more about the Western mind and world view than they tell about the American Indian mind or world view. They are a form of projection. The error of such projection is that, completely believing his own labels, the European imagines he is discussing Indian concepts when in reality he is discussing his own. The civilized linguist is doing exactly what he says the primitive 'savage' does with words, that is, he imagines that the word he uses is equivalent to the phenomena he is discussing. The linguist's labels and logic have neither described, explained, nor clarified Algonkian gender; rather they have generated misunderstanding in the European mind, and have misinterpreted and misrepresented the Indian. It certainly appears much more logical and rational when the American Indian's grammatical concepts are seen from the point of view of his own thinking, culture, and world view. Ϋοιι. . . assumed we were savages. You didn't understand our prayers. You didn't try to understand. When we sang our praises to the sun or moon or wind, you said we were worshipping idols. Without understanding, you condemned us as lost souls . . . Walking Buffalo (McLuhan 1971: 23)

REFERENCES Bloomfield, L. ( 1 9 4 6 ) , 'Algonquian', in H. Hoijer (ed.), Linguistic structures of native America, pp. 85—129. New York, Viking Press (Viking Fund Publications in Anthropology, 6).

96

mo Κάα

Curtis, Ε. S. (1924), The North American Indian. Massachusetts, Pliniton Press. Gallenkamp, Ch. (1959), Maya. New York, Pyramid Publications. Josselin de Jong, J. P. B. de (1913), De waardeeringsonderscheiding van 'levend' en 'levenloos'. Leyden. McLuhan, T. C., ed. (1971), Touch the earth: A self-portrait of Indian existence. New York. Seton, Ernest Thompson and Julia M. (1963), The gospel of the Redman. Santa Fe, New Mexico. Whorf, B. L. (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J. B. Carroll. Cambridge, Mass., MIT Press. Yegerlehner, John (1954), 'The first five minutes of Shawnee laws in multiple stage translation', International Journal of American Linguistics, 20: 4 (Oct.), 2 8 1 - 2 9 4 .

PART TWO

Psychological and Anthropological Approaches

ROBERT Η. ROBINS

5

The Current Relevance of the SapirWhorf Hypothesis

It is important to realize the change that has taken place in the dominant orientation of linguistic science since the Sapir-Whorf hypothesis was first put forward and since the period in which it was most actively discussed. Sapir's most definitive statement of his position appeared in his 1929 article, 'The status of linguistics as a science', and his influence on Whorf was at its strongest in the early 1930s, after Whorf had enrolled in his classes at Yale (Carroll 1956: 16-18). Whorf died in 1941 at the relatively young age of forty-four. His four papers that most directly set out his views on the relations between language and thought were published between 1939 and 1941, but first attracted widespread attention on their republication under the semi-official auspices of the United States Department of State, with the title Four articles on metalinguistics. Hoijer's application of Whorfian ideas to the relations between language structure and aspects of culture among the Navajo appeared in 1951. In 1953 a conference on this general theme was held, and its proceedings were edited by Hoijer, to be published as Language in culture in 1954. In 1953 Hoijer published an article, 'The relation of language to culture' in Kroeber (1953), and in the same year two critical articles appeared, Feuer's 'Sociological aspects of the relation between language and philosophy' and Lenneberg's 'Cognition in ethnolinguistics'. In 1956 Whorf's writings over the whole field of linguistics were published under the editorship of J. B. Carroll, as Language, thought and reality, together with an excellent introduction by the editor. By 1967 this book had run through eight reprintings. Carroll (1956: 23) gives what is perhaps the best summary statement of the whole Sapir-Whorf hypothesis, as 'the principle of linguistic relativity, which states, at least as a hypothesis, that the structure of a

100

Robert Η.

Robins

human being's language influences the manner in which he understands reality and behaves with respect to it'. Such a viewpoint had its origin in the Romantic reactions of such thinkers as Johann Georg Hamann (1730—1788), Johann Gottfried Herder ( 1 7 4 4 - 1 8 0 3 ) , and Wilhelm von Humboldt ( 1 7 6 7 - 1 8 3 5 ) to the universalistic ideas of the seventeenth and eighteenth centuries (Penn 1972). In its modern form it was expounded and most actively debated during the period of the rising and the full tide of structuralism and formalism in linguistics. At this time the emphasis was on linguistic diversity, and, at least as far as formal structures in grammar were concerned, no limit could be set on the way languages might differ from one another (Joos 1958: 228). During the 1940s and 1950s relatively little interest was expressed in semantic questions on the part of linguists; this was particularly the case among American scholars and was probably due to a misinterpretation of Bloomfield's behaviourist position in regard to the statement of meaning. In their different ways, B. L. Whorf and J. R. Firth were the major exceptions during this period by reason of their active and professed concern with semantic questions within linguistic theory. It is well known that since 1960 transformational-generative linguists, who, deservedly or not, dominate the field in linguistics today, have become actively and vitally concerned with semantics, the problems of linguistic universals, and the underlying unity of language structure in relation to the human mind. But there has been little interest on the part of transformational-generative linguists in relating their discussion of these topics to the Sapir-Whorf hypothesis. There seem to be no references to Whorf in Chomsky's writings on semantics; there are none in Fillmore and Langendoen's Studies in linguistic semantics (1971), very few in Steinberg and Jakobovits's Semantics (1971), and none in Lakoff's Linguistics and natural logic (1971), although natural logic and natural metaphysics are a main theme of Whorf's hvpothesis. In critical discussions on the Sapir-Whorf hypothesis since 1950, there have emerged a strong and a weak form of it, that may be designated linguistic determinism and linguistic relativity, respectively. By the former interpretation our cognition and thought processes are totally determined by the structure of the language that we speak; by the latter it is merely claimed that the structures of different languages do exert some influence on the thinking and categorization of speakers (Penn 1972: 28—32).

The current relevance of the Sapir-Whorf

hypothesis

101

Among recent references, Bach (1968: 121—122) argues for the rejection the of hypothesis 'in its strongest form'; but Lyons, who of the transformationally inclined linguists has devoted more attention to Whorf than others have, while equally rejecting linguistic determinism, has in two places expressed general support for the weaker interpretation of the hypothesis as entirely conformable to the facts of linguistic observation (Lyons 1963: 39—43; 1968: 4 3 2 - 4 3 3 ) . Certainly the strong form of the hypothesis appears to be unacceptable. On a priori grounds one can contest it by asking how, if we are unable to organize our thinking beyond the limits set by our native language, we could ever become aware of these limits. Empirically the admitted possibility of translation between languages of diverse structures spoken by people of different cultures is scarcely compatible with total linguistic determinism, while the equally admitted difficulties involved in translation afford solid support for the validity of linguistic relativity. Adopting a physical metaphor, it would seem best to liken language not to a tramline nor to an open road, as far as thought and categorization are concerned, but to sets of grooves or ruts, along which it is easier and more natural to direct one's thinking, but which with some effort can be overcome. Indeed Whorf himself tacitly admits as much when he points out that it is Newtonian physics and Euclidean geometry, the physics and geometry of 'common sense', that are the product of 'Standard Average European' (SAE) languages, not Einsteinian physics and post-Euclidean geometries. Yet it is in terms of the latter two that modern science expresses itself (Whorf 1956: 59, 152—153, 241). It is a great pity that Whorf died so soon after he had put out the hypothesis by which he is known today. Though it had been expressed in general terms by Sapir a decade earlier and by Humboldt more than a century before that, it was Whorf who specifically directed it toward the technical language (or the metaphysics) of the natural sciences. He was not present to respond to criticism and take part in further discussion in the years when the debate that he had aroused was keenest. In Whorf's own writing we can find evidence for his having espoused both the strong, deterministic interpretation of his hypothesis and the weaker, relativistic interpretation. The stronger interpretation is expressed in such passages as: 'We cut nature up, organize it into concepts, and ascribe significances as we do, largely because we are parties to an agreement to organize it in this way — an agreement that holds throughout our speech community and is

102

Robert Η. Robins

codified in the patterns of our language. The agreement is, of course, an implicit and unstated one, but its terms are absolutely obligatory; we cannot talk at all except by subscribing to the organization and classification of data which the agreement decrees' (1956: 213—214). (Perhaps the adverb largely provides an escape clause, but otherwise the implication is unmistakable.) 'Segmentation of nature is an aspect of grammar' (Ibid.: 240); 'Every language is a vast pattern-system, different from others, in which are culturally ordained the forms and categories by which the personality not only communicates, but also analyses nature, notices or neglects types of relationship and phenomena, channels his reasoning, and builds the house of his consiciousness (Ibid.: 252). On the other hand a weaker interpretation, whereby the possibility of deliberately channelling one's conceptualization along lines other than those immediately prescribed by one's own language (orby languages of the SAE historical mould) is admitted, appears in such places as these: Ά system of thought in which categories different from our tenses are natural' (Ibid.: 159), but not impossible; Indo-European languages 'give structural priority' to subjectpredicate type sentences, and 'to rid ourselves of this way of thinking is exceedingly difficult' (Ibid.: 237—238), but again, not impossible And, of course, when he wrote: 'Newtonian space, time, and matter are no intuitions. They are recepts (sic) from culture and language. That is where Newton got them' (Ibid.: 153), he was entirely aware, as is every informed person, that scientists no longer feel constrained to talk in these terms when they are engaged in scientific discourse. Nor is Whorf alone in making such statements. They are almost a commonplace in modern philosophy, as when Russell wrote: ' "Substance", in a word, is a metaphysical mistake, due to the transference to the world-structure of the structure of sentences composed of a subject and a predicate' (1946: 225). The weaker interpretation, linguistic relativity, while being much more readily acceptable than the stronger, does not rob Whorf's ideas and insights of their significance in interdisciplinary studies of communication and cognition such as constitute the theme of the Colloquium for which this paper is offered as a contribution. Sapir's use of the verb predispose also implies the weaker interpretation: 'We see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices for interpretation' (Sapir 1929: 210). The same thing is true of Hockett's remark in the proceedings of the 1953 conference: 'Lan-

The current relevance of the Sapir-Whorf

hypothesis

103

guages differ not so much as to what can be said in them, but rather as to what it is relatively easy to say' (in Hoijer 1954: 122). The important changes in linguistics since 1960, after the impact of transformational-generative theories, are, as far as the Sapir-Whorf hypothesis is concerned, the present emphasis on the underlying sameness, or at least the similarity, of languages, the renewed emphasis on linguistic universals as evidence of the unity of the structure of the human mind as species-specific to man (e.g. Chomsky 1968), and the recognition, in one or another form, of two levels of syntactic structure — deep structure and surface structure. One could say that in Whorf's time the structures of the structuralists and of Whorf's linguistic writings were surface structures. Both the major divisions of transformational-generative theory today, interpretive semantics (Chomsky and others) and generative semantics (Lakoff and others), see linguistic universality in deep structures and linguistic diversity in surface structures. In both versions of the theory, thought and its linguistic expression are in principle distinguished, though generative semanticists regard the semantic representation of a sentence and its deep structure as the same thing, whereas interpretive semanticists regard the meaning of a sentence as the interpretation of its deep structure, though, latterly, with some semantic contribution from surface structure (Chomsky 1971). Any reexamination of the Sapir-Whorf hypothesis today must take into account the concepts of deep structure and surface structure. It would seem that linguistic relativity would most readily be locatable in the effects of surface structure patterns, which are the syntactic structures of sentences as they are produced and as they are heard and read. In 1967 Lenneberg refers to a selectivity in language, whereby different languages tag different concepts among those available to the human mind. This selectivity, however, 'does not cripple or bind the speaker because he can make his language, or his vocabulary, or his power of word-creation, or his freedom in idiosyncratic usages of words do any duty that he chooses, and he may do this to a large extent without danger of rendering himself unintelligible because his fellow men have similar capacities and freedoms which also extend to understanding' (Lenneberg 1967: 554). This involves some effort, and is a deliberate movement of one's descriptions and conceptualizations out of the habitual grooves of one's language; it is one part of the creativity of human language on which transformational-

104

Robert Η. Robins

generative linguists rightly lay so much stress. The developments in the present century in geometry and physics must be reckoned as important results of just such an effort by SAE speakers. Do different surface structures make certain conceptualizations easier or more natural, though not, of course, obligatory? Perhaps we should test the Sapir-Whorf hypothesis in this form by reference to various typologically diverse languages, comparing the ways in which their habitual overt patterns of surface syntactic relations differ from the underlying, probably universal, relations between entities of thought and the elements of deep structure syntax. A possible field might be the English and general SAE semantic polyvalence of genitival noun phrases. John's portrait, Johannis pictura, le portrait de Jean, Johanns Bildnis, and das Bildnis von Johann may all refer, according to the context, to a portrait owned by or painted by John, or to a portrait representing John. In Latin grammar we are familiar with this difference in meaning being explained in terms of distinction between the subjective and the objective genitive construction. The semantic relations involved can be expressed in different sentences more closely representing the deep structures involved: John owns (or makes) a portrait, and someone portrays John; Johannes picturam habet (or facit) and aliquis Johannem depingit, etc. Several Pacific languages, for example Hawaiian and Maori, make a surface distinction among genitival phrases, separating what is usually referred to as Α-possession from O-possession, these terms being derived from the characteristic vowel of the possessive particle and the possessive pronouns. These formal differences broadly correlate with a semantic-cultural distinction, so that most nouns and noun phrases belong either to the Α-class or to the O-class. The distinction is, grosso modo, between deliberately acquired possessions (in the widest sense) and deliberate actions over which the possessor stands in some sort of dominant position, on the one hand, and inherited or involuntarily acquired possessions, feelings, and attributes, and things to which the prossessor stands in a subordinate or passive position, on the other. This is a general formal-semantic distinction maintained in Pacific languages, and it must be presumed to have a long cultural history behind it. Of particular relevance to the theme of this paper is the fact that certain nouns and nominal phrases may take either Α-class or O-class possessives, with different meanings: e.g. Hawaiian ka? ulei (my flower necklace, that I give away, or make t o sell); ko? u lei (my

The current relevance of the Sapir-Whorf

hypothesis

105

flower necklace, that I wear). In cases like those cited above, the underlying distinction between subjective and objective genitive is reflected in a surface distinction between subjective and objective genitive is reflected in a surface distinction between Α-possession and 0-possession: Hawaiian: ka-.na mele kona mele

— his song, composed by him — his song, composed about him or in his honour

ka-.na ki? i ke ki?i a Keoni

— his portrait — John's portrait, that he owns or that he painted. — his portrait — John's portrait, the portrait depicting him

kona ki?i ke ki? i ο Keoni Maori: te waiata a te ίαηαΐα te waiata ο te ΐαηαία

— the man's song, composed by him — the man's song, composed or sung about him

te j ukapuka a Ra:niera— the book of (by) Daniel te pukapuka ο Hopa — the book of (about) Job In cases like these, where the deep structure syntax determines the selection of Α-possession or O-possession surface forms with the same noun, can one establish any correlative cognitive, emotional, or behavioural differences in speakers' responses to such phrases, as compared to responses to genitival phrases in SAE languages on the part of SAE speakers? The different patterns favoured by the surface structures of different languages are plain to see. Structuralists of the 1940 s and· 1950 s may have erred by ignoring underlying inherent universalities. Given the cultural circumstances and requirements, underlying universalities would account for the human ability ultimately to say anything in any language. But the most regularly employed surface syntactic structures would lay down the preferred and most readily available modes of describing and therefore of conceptualizing the world of our experience. Hjelmslev expressed such an opinion in 1935: 'Toute idee peut etre exprimee dans toute langue, mais non

106

Robert Η. Robins

dans tout systeme' (1935: 135). In terms of the generative semantics version of transformational generative theory, the influence of surface structure would also include the different ways in which conceptual components are brought together in the lexical items inserted during the transformational stage of the grammar. Although according to this version of the theory deep structures and semantic representations are the same thing, some semantic representations would be more readily available to a speaker by reason of the most usual lexical combinations of concepts in the vocabulary of a particular language. I am well aware that I have asked rather vaguely formulated questions and made rather vaguely worded suggestions for possible lines of research, but this is perhaps one of the ways in which the Sapir-Whorf hypothesis could be reconsidered today.

REFERENCES Bach, E. (1968), 'Nouns and noun phrases', in E. Bach and R. T. Harms, Universals in linguistic theory. New York. Carroll, J. B., ed. (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf. Cambridge, Mass., MIT Press. Chomsky, N. (1968), Language and mind. New York. —(1971), 'Deep structure, surface structure, and semantic interpretation', in D. D. Steinberg and L. A. Jakobovits (eds.), Semantics. Cambridge, Mass. Feuer, (1953), 'Sociological aspects of the relation between language and philosophy', Philosophy of science, 20: 85—100. Fillmore, C . J . and Langendoen, D. T., eds. (1971), Studies in linguistic semantics. New York. Hjelmslev, L. (1935), La categorie des cas. Aarhus (Denmark). Hoijer, H. (1951), 'The cultural implications of some Navajo linguistic categories', Language, 27: 111—120. — (195 3), 'The relation of language to culture', in A. L. Kroeber (ed.), Anthropology today, pp. 554—573. Chicago. — ed. (1954), Language in culture. Chicago. Joos, M., ed. (1958), Readings in linguistics. New York. Kroeber, A. L., ed. (1953), Anthropology today. Chicago. Lakoff, G. (1971), Linguistics and natural logic. Ann Arbor, Mich. Lenneberg, Ε. Η. (1953), 'Cognition in ethnolinguistics', Language, 29: 463-471. — (1967), Biological foundations of language. New York, John Wiley. — (1971), 'Language and cognition', in D. D. Steinberg and L. A. Jakobovits (eds.), Semantics. Cambridge. Lyons, J. (1963), Structural semantics. Oxford (Publications of the Philological Society, 20). — (1968), Introduction to theoretical linguistics. Cambridge, Mass.

The current relevance of the Sapir-Whorf hypothesis

107

Penn, J. M. (1972), Linguistic relativity versus innate ideas. The origins of the Sapir-Whorf hypothesis in German thought. The Hague — Paris, Mouton (Janua linguarum, Series minor, 120). Russell, B. (1946), History of Western phylosophy London, George Allen and Unwin Ltd. Sapir, E. (1929), 'The status of linguistics as a science', Language, 5: 207—214. Steinberg, D. D., and Jakobovits, L. Α., eds. (1971), Semantics. Cambridge, Mass. Whorf, B. L. (1950), Four articles on metalinguistics. Washington. — (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J. B. Carroll. Cambridge, Mass., MIT Press.

6

HEINZ P. WALZ

Sapir-Whorf Hypothesis and Brain Activity

This paper will attempt to establish a correlation between the SapirWhorf hypothesis and cerebral phenomena, and thus indicate — for better or worse — a new apporach to the hypothesis. Stated in very simple terms: What happens to the brain if it is impinged upon by language, i.e. if we hear and recognize language, leaving out at this stage the totally different proposition of what happens if we speak language, i.e. create language? Or, simpler still, what does language do to our brain? I should like to put before you the results of experiments which might help to answer our question. They are based on the fact that the human brain emits without interruption electrical waves which so far have never been systematically investigated in connection with normal linguistic behaviour. These waves can be tapped from the human skull by electrodes which are fed into a highly complex aggregate of instruments and can be recorded graphically. As such they produce a large variety of forms which offer many difficulties for their interpretation. For the purpose of our linguistic spectra the most important shapes and their possible significance are roughly: nerve tissue in state of excitation nerve tissue stimulated

nerve tissue actively responding to stimulus

nerve tissue in state o f relaxation

110

Heinz P. Walz

The short time at my disposal does not permit me to be more specific and the above images are offered for basic guidance. One of the experiments we made was the following We asked somebody to select a German sentence, and then we had this translated into English. Here are the two sentences, the German f r o m a novel of Thomas Mann and syntactically involved as German sentences go, with the much less complicated English equivalent: Und die kleine Antonie, achtjährig und zart gebaut in einem Kleidchen aus ganz leichter changierender Seide, den hübschen Blondkopf ein wenig vom Gesichte des Grossvaters abgewandt, blickte aus ihren graublauen Augen angestrengt nachdenkend und ohne etwas zu sehen, ins Zimmer hinein, wiederholte noch einmal 'was ist das?' sprach darauf langsam 'ich glaube, dass mich Gott . . fügte, während ihr Gesicht sich aufklärte, rasch hinzu, '. . . geschaffen hat samt allen Kreaturen . . .' The English rendering is: Little eight year Antonia, in her light shot-silk frock, turned her head away from her grandfather and stared aimlessly about the room with her blue grey eyes, trying hard to remember. Once more she repeated 'What is it?' and went on slowly: Ί believe that God . . .' and then, her face brightening, briskly finished the sentence '. . . created me together with all living creatures . . .' Both sentences were played through instruments to a Subject who had no knowledge of the material. Let us stress the fact that the sentences had the identical t h o u g h t content and now only differed grammatically. How did the subject's brain react to the German syntax and again to the English syntax? On the next page are the brain wave recordings. They reveal even to the naked eye that the German version evokes considerably more brain activity than the English equivalent which produced an almost placid movement. The graph before you, only a segment of many yards of recorded matter, is representative of the overall trend and of all subsequent experiments with other sentence material. Now the relevant point here is that the subject was a man who for some sixty-five years had been attuned to the complicated grammar of his native German language. Yet when he was exposed to it his brain was forced to react more laboriously than when he was exposed to a language whose less complicated grammar obviously

Sapir-Whorf hypothesis and brain activity

111

112

Heinz P. Walz

produced a much milder impact on his brain, even though it was not his native language. Thus in our graph the two segments which are most concisely identical in thought content but differ most strikingly in the syntactic handling of their content show a state of persistent high tension and activity in the German wave record and one of placidity in the English equivalent. German: 'blickte aus ihren graublauen Augen angestrengt nachdenkend und ohne etwas zu sehen, ins Zimmer hinein' English: 'and stared aimlessly about the room with her blue grey eyes trying hard to remember' Two more examples confirm the finding of a basic difference in brain reaction to variables in syntactic stimuli:

(1)

John

gave

the

(2)

Hans

hat

letzte

(3) (4)

book

to

Woche

a

das

girl

last

Buch

week

einem

Mädchen

geschenkt

There is increasing agreement in the realisation of what is essential in modern art. Es Bahnt sich in der Erkenntnis des Wesentlichen der modernen Kunst eine Ubereinstimmung an.

The sentence 'John gave . . .' (1) produces similar frequencies in English and German, but the latter's (2) much higher spike record indicates greater nerve activity. In the sentence 'There is increasing. . .', the English version (3) has no syntactic complexities and thus we have a sequence of waves of relaxation. The German version (4) is syntactically exceptionally tense and this is reflected in the brain wave recording. Stressing once more the fact that the messages in each pair have the same thought content and that the difference lies in the grammatical rendering it appears that the purely structural impact causes

Sapir-Whorf hypothesis

and brain

activity

113

the brain to react in terms of varying degrees of tension, of electrical activity, altogether of cerebral behaviour. Our approach and our experiments are still too recent and too novel to invite generalisations and theories; many parameters will now have to be investigated to establish the part of various artifacts. Some pioneer tests seem to indicate against our expectation that, among others, psychological and auditory factors do not mask results appreciably, whereas age offers significant variables. Altogether, though, we feel we have opened new vistas for the reinforcement or the modification of the Sapir-Whorf hypothesis. I will now put before you another aspect of linguistic brain research which ought to have a bearing on our hypothesis, i.e. the location of grammar in the brain. Without entering into a discussion on the definition of grammar I think we can assume grammar to be the most important integrating parameter of language. As such, then, its manipulation of the language of which it is part will give this language individuality. Now, in physiological terms grammar is an item apart of aphasia will reveal that almost all language behaviour is prone to physiological deficits: semantics, reading, writing, and verbal cognition all can suffer considerable impairment in consequence of pathological conditions. The one linguistic parameter which seems to be immune is grammar. Once grammar is absorbed by the child at the age of about five and applied, it remains inviolate for the rest of the persons's life while all other aspects can degenerate. As linguistic impairment" is normally caused by neurophysiological is located in the brain in an area outside the cortex. My study of aphasia has convinced me that this is so and that the area that houses grammar is the subcortex, and here the Limbic System. But if so, then there immediately arise a number of conditions and inferences which might have a bearing on Sapir-Whorf. Let us briefly see how the brain works. The following diagram, based on the theories of Norman Geschwind, is compressed into a simplified record of only those aspects which have a bearing on our linguistic approach. According to this the Non-Limbic portion of the brain receives sensory impacts which it passes on via the relevant cortices to the Association Cortex then to the Temporal Lobe and finally to the Limbic System which projects the conditioned impacts back to the

114

Heinz P. Walz

NON-LIMBIC SYSTEM

LIMBIC SYSTEM

Sensory Organs eyes, ears, skin, etc.

Thalamus, Hypothalamus, Amygdala, etc.

visual cortex

auditory cortex

somesthetic cortex

Association Cortex Temporal Lobe

Association Cortex for, possibly, cognitive categorisation and/or application. The Limbic System appears to have a number of attributes, however, which might be of relevance to our investigations on the role of grammar: 1. The Limbic System is the seat of primary sensations, i.e. fear, rage, aggression, taste, smell, hunger, satiation, consciousness, and sex. 2. Phylogenetically, i.e. in terms of human evolution, the NonLimbic System developed long after the Limbic System. 3. Ontogenetically, i.e. in terms of the individual's growth, the Limbic System is almost fully developed at the age of about four whereas the Non-Limbic System is then still completely underdeveloped. 4. The Limbic System is considerably more consolidated than the Non-Limbic. One authority calls some of its attributes 'soldered in', i.e. almost immutable. Now, what are the implications of all this in terms of Sapir-Whorf? 1. There is an area of the brain which houses the majority of primordial attributes. This area was there before the human skull developed brain areas for the more sophisticated manifestations of human behaviour. ~ If grammar is located in the older area — and I maintain there are many convincing reasons for this — then grammar was there before the sophisticated attributes of language. These, in particular verbal cognition and semantics, would, in other words, be dependent on this, a primordial language feature, which as we said is the basic integrative force of all language behaviour. Or, put differently, the primordial part of the brain projects onto the sophisticated part conditions which are irrational and concommitant with all the other irrational foundations of human behaviour — hunger, fear, sex, etc. — and does so, then, in language, too.

Sapir-Whorf hypothesis and brain activity

115

2. The phylogenetic situation is now ontogenetically duplicated in each new human being inasmuch as few or no linguistic skills are developed before the primordial factor is fully operative. Only then will the child venture into large scale rational levels of language behaviour, and thus automatically condition language expression in a preconceived form of primordial grammar. 3. This preconceived form is not only firmly consolidated at an early age but once it is soldered in it remains immutable and inviolate, and thus will continue conditioning the subject's language in one strictly limited way throughout his life. These speculations, inferred from cogent premises lead up to the question of universals, of innate or acquired Engrams, of the function of grammar as a purely grammatical integrating force or as a structure, and it is hard to resist the temptations offered by these aspects in correlation to biological factors. As yet there seems to be no conclusive evidence to justify any theory going further than the hypotheses advanced above. In fact, our experiments and extrapolations are still in the exploratory stage and it would be scientifically irresponsible to venture into more speculations. I believe, however, that the neuro-physiological aspect should seriously be taken into account. If its findings are proved to be tenable then it might help to lift the Sapir-Whorf hypothesis out of much of its speculative uncertainty and expose it, for better or worse, to the test of experimental validity.

RIK PINXTEN

7

Epistemic Universals A Contribution to Cognitive

Anthropology

1. I N T R O D U C T I O N

1.1

Intuitions

The works of Whorf had a frustrating effect upon me: he seemed to reject several philosophical problems as nonsensical (e.g. a whole lot of epistemological questions), while restating others in a different way. Perhaps the most attractive side of his theory, though, is its generality and its vagueness. In the course of discussion in our Communication and Cognition seminars, 1 questions arose on Whorf's intuitions as well as on the fruitfulness of his method. In a first approach (Pinxten 1972), I stressed the methodical aspect by analyzing and reformulating the Whorfian hypothesis in order to make empirical control easier. In this article, an attempt is made to reconsider his intuitions. It seems most improbable to me that the systems of man's knowledge profoundly differ, relative to the linguistic system he uses. This incredulity rests upon two intuitions: — the hierarchical structure of knowledge systems (see below, 1.2.1). The terms 'universal versus relativistic' should then be used as a mutual relation, in analogy with the linguistic terms 'deep structure versus surface structure' (cf. Vandamme 1968); — the identity of a whole set of fundamental physical and biological determinants for all human beings. Let us examine the latter point more closely. All peoples (cultural communities) on earth live in environmental conditions that vary according to the same criteria; only the specific values of these criteria differ from place to place. The same physical, chemical, and biological laws hold for every human being. 2 Man is universally

118

Rik Pinxten

determined by the same organic and environmental laws. Moreover, one can postulate that the perceptual, mental, and motor capacities of every newborn human being are universally identical, that is, are identical to some degree and largely independent of cultural differences. The totality of universalities is quite impressive: each baby possesses an elaborate neurological system to perceive and process information up to a certain nonrandom degree, and he acts upon his environment with the same movements. Most of all, the uniformity of properties of received information (though not necessarily perceived as such) must be — initially — quite extensive. This set of invariants in a highly variable surrounding defines two interacting systems: man (self, body) and environment. Language and mind — two human subsystems — are the results as well as the instruments of this interaction. The problem, then, is to determine which one is primary to the other; which one is subject/object of the other, the latter being a direct instrument 3 of the interaction with environmental phenomena. A possible answer can be found in general structural identities and specific differences between them. The latter is the domain of Whorf: he noticed several grammatical divergences between two languages and began searching for the corresponding cognitive components, which were probably found by him and his followers 4 . Hence, he formulated a hypothesis on the relationship between peculiarities of the linguistic and cognitive systems. Finally, he generalized his assumption to a hypothesis about the relation between language and thought. Since concrete grammars seem to differ quite largely, but since, on the other hand, the environment as well as the initial biophysiological equipment of man are universally comparable and to some degree uniform, I conclude that it may be interesting to try to search for universal elements in the semantic (cognitive or representational) subsystem. It seems probable that the internal map that represents the universal physical phenomena mentally, with the aid of uniform neurophysiologicai apparatus, might have characteristics that are universal. Each newborn baby can learn to know his environments by the limited capabilities of his hands and head (mouth). This holds for all babies in exactly identical way. 'Cognition' is a rather confusing term in contemporary psychology. Moreover, this study will be restricted to some elements of the cognitive map or natural knowledge system 5 of man — the descriptive and predictive cognitive categories one uses in everyday interaction with the environment.

Epistemic universals

119

For these reasons, we introduce the term epistemic universals, rather than cognitive universals. In dealing with epistemic universals, one has to analyze to what extent general categories, like the concepts of space, time, object, causality, and universe, are identical for all human beings. The same should be done for several cognitive operators and functions, for example, the inference processes involved in conceptual behavior, the memory processes, etc. It is my intuition that these epistemic phenomena are indeed universal up to a strictly definable degree, and thus form a universal subsystem of human interaction with the environment. If this intuition should be validated, then at least the hypothesis of the absolute primacy of language upon thought (synchronically and diachronically) can be rejected; at most, the reverse could be taken into consideration. 1.2 Analysis Let us now leave intuition and turn to the analysis and the foundation of the propositions made hitherto. What are the implications of these intuitions with regard to the problems Whorf put forward? Attacking the problems involved in the study of the languagethought relation, two positions remain possible: the relativistic and the universalistic. A choice must be made. Complementing this choice, another decision is required: shall we be concerned with the analysis of linguistic or of epistemic (cognitive) phenomena? It looks preferable to fulfill this complicated task in as nearly a scientifically decent way as possible that is, not completely arbitrarily. 1.2.1 Epistemic

phenomena

My preference for the analysis of cognitive, rather than linguistic, phenomena is purely pragmatic. Being engaged in the study of natural knowledge systems, I am more acquainted with psychological facts than with linguistic facts. In the course of research two fundamental characteristics of the study of thought stood out. First, it is highly recommendable to presuppose a hierarchic structuring of the natural knowledge system of man. This opinion is held by many contemporary cognitive psychologists who are trying to

120

Rik

Pinxten

surpass the study of isolated psychological features and to integrate diverse facts in a total view of conceptual behavior as a system (cf. Piaget and Inhelder 1947; Miller, Galanter, and Pribram 1960). Fundamental concepts in the hierarchy determine to a high degree the structuring and functioning of more specific categories. This does not necessarily imply that the higher, more specific features should be in any way, mathematically or logically, reducible to the more fundamental ones. This point of view would supplement the previous position. (It would be advocated by Piaget, but probably not by the authors of the Millerwork.) The main difficulty lies in the detection and description in a nonarbitrary way, of the most 'primitive' concepts — those that form the basis to build upon. Second, the cross-cultural studies on thinking supplied me with arguments in favor of my intuition. 6 In general, all modern anthropologists I have read until now (including great names such as P. Radin, E. Boas, and H. Frankfort) agree on one point: 'primitivity' in culture or thinking of other civilizations is a Western bias. It is common knowledge that differences exist, of course, but I gather they are not fundamental. An important work in this area crossed my way when my analysis was already begun: Northrop (1964: especially 206—216) formulated his theoretical aspirations for a general anthropology by combining epistemological considerations (e.g. from Whitehead) and cultural anthropological data. However, he was satisfied with mere analogies between different cultures to furnish the proposed theoretical frame. Still, more accurate methods of observation and experimentation should result in more precise statements than those given by the authors mentioned. Such improvement of results is now reached, thanks to the laborious work of a team of psychologists-ethnologists, composed of M. Cole, J. Gay, J. Glick, and D. Sharp (1971), along with others. These scientists were curious about the nature of knowledge — first of all mathematics — in other civilizations. Therefore they read through the rather limited ethnological literature on categorizing and knowledge (with a main input by Levy-Strauss). By comparing data and procedures of different cultures, they concluded that the construction of a general theory of thinking out of ethnological facts had hardly begun. Moreover, despite the cross-cultural material brought by psychologists surrounding Piaget specific modifications of experimental methods are nearly totally lacking. Cole, et al., systematically tried to 'vary the content and context of the experiment, while maintaining the central principle' (1971: 23).

Epistemic universals

121

To prepare experimentation with the Kpelle (the primitive natives of Liberia, only part of whom is westernized), the geographical and ecological environment, as well as the economical and cultural habits are extensively reported. Then, using an elaborate procedure upon this frame, supplemented with samples of Kpelle who answered questionnaires modeled on the experiment to be done, the authors came to the general conclusion that the same abstract categories and operators, the same forms of classification, appear in 'primitive' Kpelle thought as in Western civilization. The difference between them does not consist in which categories are used, but in the way they are used. Some abstractions are less often referred to, or used in other situations, than in our civilization. The numerous comparative experiments of the authors confirm these conclusions. Cultural relativism can thus only refer to cultural differences in the use of abstractions of perceptual or memory data. Hence, we could jump to the position that the abstractions themselves can be regarded as universal. The latter position will be advocated here. Still, many problems are to be solved. First, and most important, what are those abstractions? Take, for instance, the concept of space: is the whole of our concept of space universal — present in all possible knowledge systems — or are there only common traits in all actual space concepts? How can they be detected? This particular abstraction, and I agree that it is quite complex, is purposely not investigated by Cole, et al. I think it necessary to try to delineate the abstractions that seem fundamental to any natural knowledge system. 1.2.2 Eco no my of research The choice of a 'universalistic' approach is primarily 7 based on the principle of economy of research. The falsification or confirmation of the relativistic option implies a cumbersome work: one must determine for every two linked particularities (one cognitive, the other linguistic) which one is primary to the other. Furthermore, in order to eliminate artifacts and arbitrariness, one must investigate the position of the phenomena at hand relative to parallel or more comprehensive phenomena. The relationship that is thus first stated could be a mere consequence of a common influence on both phenomena. Some linguistic as well as cognitive features can apparently bear some relation to each other, which is then only fully described in the frame of common memory influences. For example,

122

Rik Pinxten

the process of chunking information units is first experienced with regard to linguistic elements (through grammatical rules), later with nonverbal material (stimuli). It is, however, possible that the phenomenon of chunking as such, and maybe the specific rules, are due to memory characteristics (short-term memory) and not specifically to language or thought. Finally, one ought to repeat the initial examination until the total spheres of language and thought are completely incorporated in the research. Only then would the results be reliable. In this respect, it is not surprising to me that the empirical studies on the subject (Carroll, Casagrande, etc.) bear little results. A nearly literal reproduction of this casehistory of research is given, in my opinion, by the early work of Lenneberg on color classification.8 The study of universal categories or relations is more risky. The hypotheses are generally stronger. On the other hand, the results, if confirmative, are obviously much more interesting. In combination with the first choice 1.2.1, the following program results: the investigation of whether the most fundamental cognitive (epistemic) concepts are universal. A supplementary choice is obtruding: what is fundamental in natural knowledge? The rare studies on semantic universale (done from the linguistic perspective) do not throw any light on the question and are bound to leave the hopeful reader with a feeling of disappointment. Casagrande (1963) is satisfied just to state that semantic universals could be important with regard to a theory of anthropological universals and that, consequently, they should be searched for. I chose to investigate the universality of the (or a) concept of space. The reasons for this choice and the meaning of the term 'space' will be made explicit.

2. DEFINITIONS OF SOME BASIC NOTIONS

2.1 What about universals? In order to characterize aspects of language or thought as universal, it is required that one has a clear notion of the meaning of the label 'universal'. In this article a property or a relation will be called universal if and only if it appears under the same form in each and every culture. This means, that — the respective characteristics must appear in all civilizations; — they must be found in the same form — they must be translatable or substitutable by one another. When searching for cognitive

Epistemic

universah

123

universale, the results ought to be cognitive features t h a t are independent of linguistic or cultural differences, and that are, by consequence, primary to culturally determined or conditioned features; — only the f o r m of appearance must be indentical, because it is m y opinion (resulting f r o m discussions with Vermeersch) t h a t formal features are the sole material of scientific research. So-called 'contents' can only be studied with regard to their formal aspects and thus are, scientifically speaking, probably only less clear and more complicated aspects of form. A few restrictive remarks must be made on this convention. First, it must be regarded as a limit of research, which can only be neared asymptotically, that the work will be incomplete, since only a f e w knowledge systems f r o m a few different civilizations can be studied here. Second, within the scope of the article, I shall treat psychological, and only then epistemic, universals. Furthermore, the t h e o r y I will draw upon is Piaget's. T h e reasons of this choice will be given below under 4.3. In view of these important restrictions the notion will be kept in its strongest form. F r o m the m o m e n t a feature fails to occur in one knowledge system, it will be considered as specific to one or more civilizations, b u t not as universal. If it is present under a d i f f e r e n t f o r m , I shall feel obliged to restate its general definition in order t o enable its integration in the transcultural framework. I am aware of the danger that this point of view could leave us with nothing b u t trivialities. Much will depend on the richness and unambiguousness of the material that can be f o u n d for the purpose.

2.2 What epistemic

is about

As mentioned above, the approach will be psychological, instead of linguistic. Stress will be laid on characteristics that are in a way purely cognitive, that is, p h e n o m e n a t h a t constitute the natural knowledge system of man, informational units that are selected f r o m multivariate perceptions and actions and are at least in part processed and stored in memory. Since the focus is on the cognitive m a p or knowledge system of man, in other words on the result of information processing, rather than on the processing itself, the processes of attention, motivation, and remembering will not be considered explicitly. Only the constituted system as such counts, even though it be relatively rather than absolutely constant.

124

Rik

Pinxten

For the same reason, the action procedures that are based upon or make use of the units of the cognitive map (concepts) are neglected. The study appears as a search for a kind of rationale that is a condition for and a result of perception, imagination as well as of action. Something of the kind can be found in the works of Whitehead and Russell on natural philosophy, a testimony of brilliantly sharp minds. 2.3 A space among

spaces

The philosophers of the Wiener Kreis (Circle of Vienna) held that a strict distinction be made between physical and mathematical spaces, which were to be studied respectively in inductive or deductive scientific disciplines. The attitude should be extended to make a distinction between the former and the so-called psychological or anthropological space. Within the latter sort, three further discriminations are required: perceptual space, representational space,9 and action/life space. The last one refers to existential, social, and communicative aspects. Although the three spaces certainly are mutually linked in various ways, the a priori distinction between them is important in order to avoid vagueness and ambiguity. The literature on physical spaces is simply vast. The works treating mathematical spaces are nearly as numerous. The relation between the two is now well established in scientific theories. The interested reader should consult Grunbaum's basic work (1964) on this subject. On the other hand, psychologists and even certain anthropologists were puzzled by space as well. To be taken seriously on this subject in these sciences, one is still expected to treat the influence of aspects of physical environment on man at a so-called controllable level; that is one must study the perception of physical attributes in space, or, in one word, phenomenal space. The topic certainly is important. The result until now, however, has been abundance of books and articles on specific problems, such as the perceptual form and volume constancy, distance constancies, etc. and a few theories on space perception, but mainly with a complete disregard of the conceptual and/or imaginative counterparts. The studies, therefore, seem to me to be insufficient. In this article the conceptual representation of space — the epistemic space — will be treated, because, in fact, that is what it is all about. Perception implies structuring and recognition of information data; acti'on is based on a conceptual frame of reference. The

Epistemic

universals

125

differences between the interpretation of physical spatial forms of different civilizations must then be explained through the divergent conceptualization of the same physical environment (see note 9). The same goes, a fortiori, for mathematical spaces (see the particularity that all existing metrical spaces were, as far as I know, developed in the Western world). That is why the conceptual or epistemic aspect should be considered. But why ought space in itself be taken as an object of study? Here, again, philosophy and intuition enter the arena. 10 The most impressive and functionally valuable segmentation of the outer world in everyday life is the spatiotemporal frame. Living beings in general, and humans in particular, live in a partly continuous, partly cyclical conception of time. Seasons, lifetime, death, ageing are psychologically real; they are among the most important points of reference for cultural and social behavior. (It must sound trivial to make such a remark.) The same can be said about the role of spatial relations, delineation of territory, orientation, and the common use of spatial characteristics in a metaphorical way. This tendency is most visible in Western spatialized concept of time 11 — segmentation in measurable units — and in Chinese thought where the symbol of the ideal state of affairs happened to be a perfect square. Furthermore, when cognitive reference with the normal world is lost, as in schizophrenics, a remarkable deformation of spatial experience appears at the peaks of the crisis — the well-known Korsakow syndrome. A better understanding of the space concept in general could, perhaps, lead to a more correct view of the experience of the world by schizophrenics. It is understandable that such perspectives tend to motivate this work. These observations and intuitions suggest to me that space probably constitutes a major reference point or a subframe in human life. Of course, it is obvious that other referents ought to be considered as well, such as time, object, causality (or, maybe better, relatedness), and universe. These lie beyond the scope of this work. 1 2 A more precise definition of epistemic space is required; it will be given in full below, under 4.

3. THE PROBLEM A N D THE METHOD

There are four prerequisites of analysis: a problem, a method, an analytical tool, and material to be analyzed.

126

Rik

Pinxten

The latter two are considered first. The analytical tool will be an elaborate model of space by the psychologist Jean Piaget, worked out through many observations and experiments on children of different age classes. It cannot be explained in a few words. An extensive description will therefore be given in 4. The material to be analyzed consists of descriptions of knowledge systems and cultural habits of a few non-Western civilizations. Since it was impossible to gather information on all civilizations, a few among them were chosen. On the other hand, and in order to eliminate arbitrariness of the facts considered, it became my principal aim to consult, whenever possible, more than one source of information on each civilization. The following non-Western cultures will thus be treated: — Chinese civilization before Christ, that is during a period when unilateral influence on or interaction between Chinese and European cultures is highly improbable; 1 3 — Liberian and Dogo civilizations, and two studies on sub-Saharan African cultures (geographically less delineated than the former); — some Indian world views, complemented with two works on Ancient Aztecs; — two works on pre-Socratic thought. I understand that these choices will be subject to much criticism. Moreover, the material is entirely secondhand, since I had to draw completely on experiences and descriptions of others. I think it nevertheless important that a first approach in this direction should be attempted. The method that is used is quite "clear and simple; it consists of processes of matching and recognition. It is thus, evidently, a method for a theoretical approach. The model of space that serves as an analytical tool is constructed hierarchically. Each level is constituted by a spatial characteristic that is neatly distinguished from, but in itself fully implies, the concept of a previous, lower level. The method to be followed consists of the systematic matching of the subsequent spatial characteristics of the (with regard to the research) a priori framework with the spatial concepts of each non-Western civilization. The framework concepts or levels must be treated one by one in order to reveal the correspondence, or the lack of it, systematically. Each time that explicit descriptions of the 'primitive' features are not available, the contingent interpretation of implicit notions will be extensive and justified. Finally, the sum of all universal characteristics will be registered and put in a scheme. This

Epistemic

universais

127

scheme will then represent the sum of all spatial characteristics that can be found universally, or, the universal concept of space. The problem can be formulated in the following hypothesis: 'There exists a universal concept of space that is independent of cultural and/or linguistic differences. Specifically, the fundamental features of the hierarchical space concept are universal; the more complicated and higher features are particular for each culture.' The confirmation of this statement would yield a number of important consequences with regard to several anthropological questions in general, and to the Whorfian problems in particular. Its falsification would add considerable support to the relativistic position.

4. PIAGET'S CONCEPT OF SPACE

The works of Piaget used here are the following: Piaget and Inhelder (1947) 14 ; Piaget, Inhelder, and Szeminska (1948); Piaget, Bang, et al. (1964). A pedagogical booklet on topological space by Mr. and Mrs. Sauvy (1972) will be used occasionally. In the first place, many psychologists share the intuition that the human representation of space, in its most primitive and fundamental characteristics, is non-Euclidian. To quote one eminent representative, F. Allport (1955: 150) states that life space is 'capable of having at any moment any size or geometric shape whatsoever, so long as the principle of includedness within boundaries was constantly recognized and there was afforded within it a continuous region, or connected regions, in which locomotion could occur'. Allport is conscious that the terms used — 'includedness', 'boundaries' — are indeed meant to refer to topological notions. Piaget will be, to my knowledge, the first psychologist to investigate systematically the topological basis of the representation of space. A second introductory word is necessary. In the following paragraphs the terms 'constitution', 'spatial constitution system', etc., are used to denote the construction of more complex or higher level aspects of spatial representation out of, or on the basis of, more primitive units (cf. again the works of Whitehead, Russell, and Goodman). Constitution, then, does not refer to actual genetic building up of concepts. It is an abstract frame for representation; it expresses the (psycho-)logic of complex representational activities in man. 15 Each separate constitutional rule or concept must have

Rik

128

Pinxten

psychological reality. Let us now develop the Piagetian model of the representation of space. 4.1. Basic assumptions

ofPiaget

I shall mention only a few of Piaget's main suppositions, just those" that are absolutely required to understand his theory of space. In general, Piaget was struck by the strength and possibilities of mathematics, particularly those of geometry. Historically, the most general and logically simplest spatial notions developed only recently (topology), while the more sophisticated metrical space of Euclidean geometry has existed for nearly two thousand years. Piaget had the intuition that, nevertheless, the topological model was genetically primary to any other spatial representation system. His researches tended to confirm this intuition. The former does not presuppose that the mathematical and the representational spaces are identical. Only, the same basic operators are used, supplemented, in the case of the latter, by a few prelogical procedures (cf. Piaget 1964). Piaget distinguishes between different stages of development in the representation of space. The evolution of the representation of space is identical with that of perceptual space, but with a few years' lag. 16 Each stage is constituted of elements (concepts, operators) of former stages, together with new ones — more specific segmentations and combinations. The sum of all particular progressions is presented as the succession of three major spatial systems: topological, projective, and metric (Euclidean) space. The last two fully imply the first one; they are more complex and make possible a more specific description and more adequate control of spatial environment. Each new level (construct in the system) involves the constitution of one new feature. This rule is generally held, except for on the ground level, since initially one (Piaget and Inhelder, 1947) and later two operations are accepted as constitutional for the basis (Piaget 1964). 17 Genetically, the following phases can be distinguished: — from birth until two years: sensorimotor or perceptual space, with topological segmentation; no representation, no symbolization; — from two until seven/eight years: topological space concept; representation and symbolization of the topological notions gradually develop;

Epistemic

universals

129

— from seven/eight until twelve/thirteen: systematic and parallel elaboration of the projective and Euclidean space system; even here, the basic notions of projective space are genetically slightly primary to those of Euclidean space. Piaget systematically discerns both recognition and application aspects. In his observations and experiments he nearly always provided two settings — one to explore the extent to which a particular spatial construct is recognized exactly, and another one to check up to what degree some constituted feature is applied in situations where it is suitable. The correlation between both sets of results was nearly uniform; the correct application showed more difficulties and was arrived at later than the recognition (categorization). Special attention must be paid to the explanatory concepts of 'mental operation', which is used quite abundantly by Piaget. It is defined in a rather ambiguous, but at least explicit, way: 'An operation is an action that is likely to recur to its starting point and to become composite with other operations in a double way, namely in direct and inverse sense (Piaget and Inhelder 1947: 50; my translation). 18 Intuitively, this can be restated as 'interconnectable internalizations of actions'. They are mental and reversible, in opposition to concrete actions that are material and irreversible (cf. Piaget and Inhelder 1947: 565). Mental operations only appear at the level of representational space. I will use the term rarely, only if it is necessary in order to avoid misunderstanding. Elsewhere more precise and restricted labels will be searched for. 4.2 The model A. Topological space The topological representation system is gradually constituted through the emergence of only five relations (operations). Typical for this form of spatial segmentation is the restriction of scope to what Grize called 'analyses intra-figurales' (in Piaget 1964: § 2): the object that is spatially conceived provides in itself the complete field of segmentation, recognition, and application of topological notions. Elements that are mutually distinct within one whole are represented in certain spatial relationships to one another. The same goes for a second whole, etc., but the relations between the distinguished

130

Rik

Pinxten

wholes are not considered at all. The actual mi^r-figural relationships at this level of constitution are random. A . l Voisinage, translated as neighborhood, 'constitutes without any doubt the most elementary spatial relation' (Piaget and Inhelder 1947: 63; my translation). Intuitively, the relation of neighborhood has the following effect: distinctly presented elements within a whole are internally represented as interconnected with respect to their 'neighboring each other'. Elements that lie close to each other will be conceived as a cluster, apart from similar groupings, whatever the particular geometrical character of each of the elements may be. The problem of the nature of the elements need not concern us here. Psychologically speaking, they are provided by the stimuli that are presented (by the experimenter) and are at first, perceived as distinct, and afterwards, in a first approach of spatial representation, thought of as belonging together in view of their neighborhood. What they really stand for in the natural situation of everyday perception and action is another and a considerably difficult question. It need not be treated, though, to understand the processes of spatial representation as such, that is, of the relationships between the elements. A.2 Separation will be translated here as distinctness. The relation of distinctness is present to the extent that the actual elements are perceived as mutually distinguished, discriminated from each other. (Piaget and Inhelder 1947: 64) It is a mutual relationship; therefore, the term 'distinctness' seems most appropriate. It is clear this relation complements the first one. For example, when four out of five elements are represented as a group, a cluster, this state of affairs implies that a relation of distinctness is established between the cluster on the one hand and the fifth element on the other. A. 3 Relations d'ordre understood as ordering relations. They constitute the order of succession in a set of elements. Their initial constitution is quite primitive: 'exclusively on pairs of terms, of which the relative position is to be determined' (Piaget and Inhelder 1947: 64; my translation). Applying an ordering relation thus consists in the mental organization of elements in a row, determining which one preceeds which, until all elements have a specific, well-defined place.

Epistemic

universals

131

Genetically and constitutionally, this relation is in a way the synthesis of both the foregoing. The observations of Piaget pointed to this conclusion: the subject at first discriminated pairs of elements (neighborhood relation). Once the couple was conceived, the elements were mentally reseparated, keeping them in unchanged position in relation to each other, and each of them formed a pair with elements of the two neighboring pairs of which one preceded and the other followed the initial pair (Piaget and Inhelder 1947: 99-103). A primitive notion of betweenness can be recognized. It is, however, still primitive, since the direct relation of ordering yields success, but the inverse relation nearly completely fails. The relation of betweenness, however, implies that the ordering relation holds in both (direct and inverse) senses. A.4 Enveloppement will be translated as enclosure. This is a complex phenomenon, entailing some other concepts. It implies the direct relation of ordering as well as its inversion. 'Enclosure' of an element is to be conceived in opposition to elements of an outer world that are not enclosed. The fundamental cognitive instrument that is needed to distinguish between in and out, or enclosed and nonenclosed, consists in a borderline, the boundary. It can be defined as the set of all elements that are part neither of the enclosing figure nor of the outer world. Some examples may bring clarity: the eyes are enclosed in the surface or volume of the head (two- or three-dimensional), the ears are not; a ring is a circle that neatly encloses another circle; a knot is a three-dimensional form of enclosure — that is exactly what distinguishes a real knot from a false knot. Reproduction through drawings or manual activity reveals if the operation of enclosure is really mastered (Piaget and Inhelder 1947: ch. 4). When only the foregoing four relations are used together, the spatial environment is segmented and constituted topologically. Within this space elementary, qualitative homeomorphism occurs, i.e. sets of elements (or constructs) can be spatially isomorphous, using the four criterial operations (cf. Sauvy 1972: ch. 6; Piaget and Inhelder 1947: 148 ff). A last topological notion can be introduced now. A.5 Le point et le continu will be referred to under the heading of continuity and discontinuity (Piaget and Inhelder 1947: ch. 5).

132

Rik

Pinxten

It should be remembered that the term refers only to topological features. The phenomenon consists in 'the reversible composition of the whole out of elements or of elements starting with the whole; b u t the composition remains " c o n c r e t e " and not " f o r m a l " , that is, it remains finite and perceptible and not infinite and intellectual . . .' (Piaget and Inhelder 1947: 169, my translation). In fact, the whole — for example, a topological line — is conceived as continuous, while the elements — for example, the points that constitute the line — are represented as discontinuous. This state of affairs is felt as a contradiction by the subjects (children). Nevertheless, they intuitively agree on the necessity of conceiving spatial p h e n o m e n a in a scheme like this. A clear elaboration of the processing of this apparent contradiction is given by J. and S. Sauvy (1972: ch. 3.1): the concept of continuity is constructed on the basis of double or multiple neighborhoods, in itself discontinuous. Theassemblage of all consecutive neighborhoods between two (random) extremes constitutes a continuity (for example a topological line).

B. Projective

space

The f u n d a m e n t a l difference between topological spatial representation, on the one hand, and projective and Euclidean space, on the other, 'lies in the mode of coordination of figures with each other' (Piaget and Inhelder 1947: 179; my translation). In fact, with the emergence of projective relations the subject begins to structure his spatial environment on an inter-figural level (Grize, in Piaget 1964: § 2 ) , whereas the topological segmentation was restricted to the area within one figure. In projective space distinct figures are related to each other by means of projections or perspectives. In general, the notions of 'point of view' and 'difference of point of view' are introduced as fundamental. Spatial representation now implies the (mental) conservation of straight lines, curves, angles, distances, and certain specific transformations. All thes-; spatial features were, of course, in some way distinguishable in the topological space as well. However, they were not conserved, that is, they could not be properly reproduced by the individual, nor could they be used as criteria in representational spatial segmentation (cf. Piaget and Inhelder [1947: 65], who refer to the topological, rather naive constructions as products of 'intellectual realism', rather than representation). A f e w projective constructs will be explained. Their recognition in

Epistemic

universals

133

other (primitive) civilizations will be used as indices for the actual occurrence of projective spatial segmentation there. B.l First of all, two notions are acquired nearly simultaneously: the projective line and the perspective {la droite projective and la perspective) (Piaget and Inhelder 1947: 181). Following the authors, they will be treated separately here. B.la The construction of the projective line requires the establishment of the conscience of the point of view: the subject recognizes that things look different when HE changes his position with regard to them. 1 9 In other words, things will be looked at from now on as relative to at least one other thing in the perceptual field, namely the observer himself! The concept of 'projective line' can then be defined: It is a topological line (see above), but such that its elements are ordered relative to a point of view and thus appear in succession along the relation 'front χ back', in which the first element masks all the others. And, supplementary to what was just said, the projective line is the only line (and in opposition to curved lines) whose shape remains unaltered, whatever point of view one takes (cf. especially Piaget and Inhelder 1947: 199). In other words, it always remains straight (with the limit position of the point, of course), whereas curves can project as straight lines or as curves, depending on which point of view is taken. B.lb The construction of 'perspective' sets forth two a priori requirements. First the subject must be able to recognize and produce a successive series of enclosures. 20 For example, take the perspective of a row of rings, the biggest closest to and facing the observer, the smallest ones being the most remote. In order t o conceive the perspective of the row, one must be able mentally to apply the operation of enclosure to all consecutive rings: the biggest encloses the second one (second in position and in shape), which in turn envelops the third one, etc. Thus, a regularity of enclosures results. Second, the subject must become aware of different points of view and make the coordination between them. The result of this coordination is a system of points of view in which, by definition, projective lines, etc., become possible (cf. Piaget and Inhelder 1947: 220-225).

134

Rik

Pinxten

The elementary projective space, then, in fact, consists in the regrouping of the topological relations (neighborhood, ordering, etc.) within the coordination system of points of view. Piaget gives an important example (Piaget and Inhelder 1947: 223): — the topological relation of betweenness (cf. above), restated from a certain point of view, results in the projective dimension leftright. Similarly must be said: — the relation of betweenness from another point of view gives the projective dimension above-helow. — finally, in a third possible point of view, this projective dimension is constituted: front-behind. These three dimensions offer a useful frame of analysis with respect to the further research. B.2 Relativity of each perspective and the coordination system of perspectives. The possible perspectives are not in an arbitrary relation to each other. The subject interconnects all possible perspectives into a non metric coordination system, in which every particular perspective has its definite relations to all others, and to the subject. In Piagetian terms, a transition must be accounted for that goes from spatial egocentrism, where each possible is equated to the actual perspective of the child, into objective grouping of spatial features, where spatial perspectives are regarded from the point of view of other — mainly virtual — observers and from the objects relative to each other (cf. Piaget and Inhelder 1947: ch. 8, § 2, 3, 4). We would suggest the characterization of the constitution that accounts for this evolution as follows: The coordination system of perspectives appears through the systematic relating of each particular perspective to all others, especially the relativity of the proper point of view. It thus results in the segmentation of space and the specification of objects through their position in the different perspectives. The absence of such a coordination system would result in the representation of all objects either relative to one single perspective (everything is seen in front side), or relative to one or another perspective at random. The latter would result in an apparent chaos in all pictorial representation. I hope this exposition is sufficiently elaborated to add to the tool of analysis.

Epistemic

universah

135

Β. 3 Geometrical sections stand in opposition to any topological notion encountered; the topological characteristics of space prohibited per se any partition along a determined scheme. This is only possible within a space with some degree of coordinatedness (cf. projective characteristics sub 1 and 2). Consequently, the sections that are actively effected in a projective space lead to qualitatively new results. The authors introduce the term 'geometry of points of view' (Piaget and Inhelder 1947: 286): The subject experiences the objects as part of the outer world; consequently, any action of him upon them is conceived as external as well and its effect is not reducible to the subject alone, but is framed in the coordination system that comprises the objects. Piaget and Inhelder watched subjects cutting and/or drawing the sections of different geometrical forms — cylinder, prism, parallelepiped, sphere, and cone — from different angles. Some sections appeared to be more difficult than others. I shall not describe the operations involved in detail, primarily because the material to be analyzed would show up too poorly in front of such highly elaborated analytic frame. The only thing that matters is the analysis of the occurrence or absence of the correct representation of (mental; internalized) sections of geometrical shapes as such in other civilizations. 21 B.4 The constitution of surface (Piaget and Inhelder, 1947: ch. 10) is reached with the help of different and, in my opinion, overly complicated techniques. It is only through the use of precipitation and dismantling (rabattements et developpements, pp. 314 ff) that it is arrived at. It appears to me that, although these operations might shed light on the attainment of the surface construct, they certainly hamper the task considerably, even if only by the hypothesis of composed mental operations as such. Moreover, I suspect (though this is not the place to work out this criticism) that some concept of rotation has silently slipped into the very constructs of precipitation and dismantling, without being introduced explicitly. For these reasons I shall treat only the concept of 'surface' as a criterion, leaving aside the associated phenomena. The representation of projective surfaces is, in fact, much like that of the projective line. Mainly, it is more anticipative (Piaget and Inhelder 1947: 341—343). Its constitution implies the necessary coordination of — the interrelatedness of — at least two perspectives

136

Rik

Pinxten

(dimensions), while the projective line could be constructed with the help of one perspective only. It thus consists in the application of the two dimensions 'left χ right' and 'above χ below' together and in coordination with each other. This ends the story of the projective space. Let us now turn to the Euclidean features. C. The fundaments

of Euclidean

space

It is absolutely out of the question to give a complete description, let alone a frame of definitions, of the Euclidean space concept in this article. The focus will be in some fundamental concepts, while others will only be mentioned briefly. Moreover, only a few explicit definitions will be given, since these would, in principle, require an extensive treatment of the Euclidean system. In general, two qualitatively distinct sets of characteristics are considered: — transitional phenomena between projective and Euclidean space; — Euclidean constitution. C.l Transition from projective to Euclidean space. Both spaces are related to each other in a double way. First, they are both, mutually independently, constituted on topological spatial notions. Second, there exist a number of transitional operations between them. A fundamental difference, of course, resides in the metric character of Euclidean spatial constitution. This can only be understood, at least in the opinion of Piaget, Inhelder, and Szeminska (1948) in terms of the fundamental intellectual phenomenon of conservation. It therefore appears preferable to give a brief characterization of this concept. This is rather difficult, since in both of the works referred to the concept is used as a condition for spatial constitution, rather than a constitutional level itself. Probably in view of its function, it is only seldom delineated explicitly. I shall try to confront the two definitions I found and to deduce a usable characterization from them: (I) 'The discovery of a notion of conservation by the child is always the expression of the construction of a "grouping" (logical or infralogical) or of a "group" (mathematical) of operations' (Piaget, Inhelder, and Szeminska 1948: 354; my translation). The second definition more specifically deals with the conservation of the whole: (II) 'The proper characteristic of concrete operations, in the domain of logical "groupings" as well as in that of

Epistemic

universals

137

composition-out-of-parts, is precisely that the free mobility of the parts remains possible within the boundaries of a whole that is necessarily conserved as a (real or virtual) union of its elements (Ibid., 1948: 414; my translation). About the specific character of conservation we can conclude that: — conservation is not limited to specifically spatial operations; — it is a logically complex phenomenon, implying abstract operations of at least logical grouping (The concept is quite complex: it implies inversion, negation, reciprocity and correlative transformation. Interested readers may find an extensive and clear treatment of it in: Frey 1967); — it consists in the delineation of a whole that remains permanent and constant, independent of the changes that occur on the level of its constituent parts; the whole is said to be 'conserved'; — I am aware of the shortcomings of this approach to the concept. Nevertheless, I hope these lines might give some clarity as to the character of this fundamental notion, at least enough to enable understanding of the particular instances in which it occurs. C.la Affinities Piaget and Inhelder 1947: 353) are transitional operations constituting the notion of parallelism of simple straight lines. This constitution implies the constructs of straight line (cf. B.l above) of direction (perspective), and of sameness of direction (i.e. the conservation of a specific direction). Another affinity, complementary to the foregoing, is the constitution of angle. By definition, an angle appears whenever two projected lines within a plane fail to be parallel. Applying the conservation operation on angles, the concept of angle as such is arrived at. C . l b Similarities are second-level phenomena. The use of both parallels and angles together results in the recognition of certain geometrical constructs as similar (for example, all triangles are experienced as having three angles and three sides in a combination that makes them, to a certain degree, mutually interchangeable). Through the actual conservation of these similarity characteristics the main geometrical figures are constituted (circle, versus triangle, versus rectangle, etc.) (Piaget and Inhelder 1947: 386—405). When angles and parallelism fail to enable appropriate distinctions (for example, both a square and a rectangle have right angles and parallel sides), recurrence is made to a supplementing criterion — the

138

Rik

Pinxten

notion of proportion (Piaget and Inhelder 1947: 420—429). In clear terms, proportion refers to the estimation by the subject of the degree of identity or interchangeability of either two parallel lines with two other, or an angle with another. In this perspective a rectangle is distinguished f r o m a square, because of the different proportions of length and breadth. C . l c When proportions are conserved themselves, constancies of length, height, etc., are constituted. In general, constancy of size is introduced. Hence, the construction of metrical space — the quantitative spatial segmentation — can take a start. C . l d The combining of the foregoing features — conservation of size and geometrical similarities — in view of a uniform interconnection of all objects of one's environment gives 'way to the construction of frames of reference or coordinate systems' (Piaget and Inhelder 1947: 435 — my translation). In fact, the constitution of a coordinate system consists in the application of order relations (cf. above A.3) on all objects of a large frame of perspectives, applied to two or three dimensions at a time. The fundamental axes of this system are the horizontal and the vertical dimensions. Because of the familiarity of these concepts, I think it hardly necessary to try to define them elaborately. I hope the descriptions by constitutional relations given in this section are sufficient to make the constructs serve as operational tools of analysis. C.2 Some fundamental Euclidean Notions.22 In this section I shall draw exclusively on the work of Piaget, Inhelder, and Szeminska (1948). It will be short since only fundamental concepts will retain attention. I shall follow the threefold division of the authors in this book. C.2a Qualitative operations constituted with the help of conservation: — conservation of the topological relation of betweenness results in the Euclidean notion of partition or interval (p. 487). The conservation works as follows: an interval appears whenever the relative size of part of a line is conserved. Meanwhile, the actual or absolute size can vary; — the concept of interval enables the constitution of Euclidean

Epistemic

universals

139

transposition. It is the Euclidean form of ordering — of progression along a straight line through different consecutive intervals. In fact, the progression along one dimension only, either the vertical or horizontal dimension (cf. C . l d ) through the interval between two immobile points bears the notion of distance. Thus, distance consists in the occurrence of transposition in a coordinate system; — finally, when different operations of transposition are applied to two or three dimensions at a time, instead of one single dimension, the Euclidean constitutions of surfaces, or volumes are introduced in the spatial system. C.2b The constitution of the linear measure (Piaget, Inhelder, and Szeminska 1948: ch. 5,6). The constitution of the linear measure genetically rests on concrete actions of comparison, overlapping, and matching. A part of a line is laid on top of a similar one, and the relative sizes of both are compared. By this procedure the second part is measured in terms of the first one; it is judged to be visibly bigger, smaller, or equal. In still other words, the original part is considered the unit against which other, similar parts are matched. Once a unit is accepted, one can form a concatenation of a sequence of identical units. The sum of them is, again, an extended measure. It is evident that two major operations are involved here: — absolute conservation of size, in order to produce a unit of measure; — transposition of the unit along a straight line, in order to secure the sequence. C.2c The calculation of surfaces and volumes. This 'measuring' appears somewhat later, and in a somewhat different form. In fact, the calculation of surfaces and volumes does not consist in the measuring along two, or respectively, three, dimensions of the Euclidean coordination system at a time. They require the development of two specific forms of conservation — conservation of surfaces (Piaget, Inhelder, and Szeminska 1948: ch. 11, § 7) and the conservation of occupied volume (ch. 14, § 6). Scheme 1 is an attempt to depict the relationships between different spatial features. It may also give a clearer view on the hierarchic character of the final totality of Piaget's concept of space.

140

Rik Scheme 1. Relationships

4.3 Interpretation

between different

and use of Piaget's

4.3.1 Stress on constitutional

spatial

Pinxten

features

model

aspects

A most important characteristic of Piaget's model is the complete concurrence of genetic development through different phases and of gradual logical complexification. As mentioned before, each constituted spatial feature analytically implies (in a nearly axiomatic way) the foregoing one(s). This is exactly what is meant by constitu-

Epistemic

141

universals

tion. Piaget claims that this constitution both follows the revealed hard (psychological) logic and manifests itself as parallel to genetic development. I am not interested in the genetic evolution; I am only concerned with the constitutional relationships within the spatial frame of reference. My interpretation of Piaget's space concept thus shows itself as follows: the first part of the representation of space (topological space) gives a complete view on the primitive or fundamental properties of spatial orientation. 23 The foregoing paragraphs reveal the way in which Piaget's model will be used for my analysis. One can easily grasp the importance of the constitutional system described above (4.2). In connection with the foregoing i t must be claimed most explicitly that the comparison between the space concepts of our and other civilizations is regarded here as the comparison between structures and operations of two (or more) ADULT civilizations. It is thus out of the question to take into consideration the genetic aspects of Piaget's model, which would imply that non-Western man is, by definition, comparable to the Western child of some specific age, rather than to Western adults. I wish to emphasize most explicitly that such reasoning rests upon a grave misconception; as said above, Piaget's system is conceived as a hierarchic system representing the constitution of psychological spatial features. The comparison to be made thus does not imply any genetic connotations.

4.3.2 Procedure of

comparison

Taking for granted the descriptive scientific value of Piaget's model, I hold it to be a valid analysis of Western psychological spatial constitution. My intention, then, is to check to what degree this (Western, Piagetian) model of space is universal. For that purpose, a systematic matching is taken up. I shall have to look up, in all possible descriptions of non-Western civilizations, the equivalent of each Piagetian construct. I will see if each particular feature can be recognized in every culture. If a feature can indeed be found in all civilizations under the same form, it is classified as a universal (cf. definition of universal, above, under 2.1). If it fails to occur at all in at least one civilization, it is characterized as 'culturally specific', i.e. dependent upon cultural conditioning. If it fails to occur under the specific form described in the analytic frame (cf. above, 4.2), a closer examination of both

142

Rik

Pinxten

empirical data and frame concept is needed in order to establish correspondence between them. 4.3.3 The problem of the sources A practical problem is posed by the choice of the material in view of the character of the analytic frame (Piaget's model). There exist, to my knowledge, few anthropological accounts on the natural knowledge systems or on specific aspects of them in other civilizations. Most authors seem to be more intrigued by cultural, religious, and economical peculiarities, than by properly epistemic characteristics. I shall have to draw mainly upon the explicit reports of the latter. Only secondarily shall I make use of accounts on religious and cultural features in order to infer from them information on the concept of space. Thus, it is clear, the study will be limited to the investigation of epistemic phenomena of a restricted number of natural knowledge systems. 4.3.4 Pretensions of the work As to the usefulness of the work, I suppose it can serve two main purposes in a modest way. First, it is a preempirical analysis, enabling, I think, more thoroughgoing empirical testing later. The study can result in some predictions as to the universal character of space (cf. Scheme 2). Second, where empirical work, stricto sensu, is impossible (ancient civilization), I suppose the kind of research presented here might bear some interesting results (cf. in this respect the claim for 'historical anthropology' by Vernant [1969]).

5 . IN SEARCH FOR AN EPISTEMIC UNIVERSAL

The material to be analyzed can be grouped in different ways. The principle of grouping that is chosen here lays stress on elementary geographical characteristics. Civilizations are grouped together because they are found on the same continent. This criterion may look arbitrary. Maybe common stems or articulate mutual influences of two or more cultures might serve as a more convincing criterion. But, following my preference for risky hypotheses, I think my

Epistemic

universals

143

ignorance in these matters might serve me well here; indeed, if my analysis should reveal common features of spatial representation in civilizations that are deliberately considered as disparate and that are analyzed at random, the reliability of the results might be considered the more probable, as the preliminary scientific knowledge and traditional (or scholarly) prejudices will have been the less. Of course, each culture will be treated separately, in order not to result in a complementary 'continental culture'. Nevertheless, an apology seems appropriate. The present, modest approach to the problem cannot be estimated but as a first attempt. Of course, only large scale empirical research can decide on the validity of its results. The material will thus be grouped in the following way: — Chinese thinking: a work of Granet (1934) and the definitely monumental volumes by Needham (1956), of which, obviously, I could read only the relevant parts (par. 5.1, pp. 143—148); — African world views: these are extracted from articles by Jahn (1964), Griaule (1952), and Campbell (1964), and the books by Cole, Gay, et al. (1971) on Kpelle, by Agblemagnon (1969), and by Calame-Griaule (1965) on Dogo (par. 5.2', pp. 1 4 8 - 1 5 6 ) ; — American Indians: a wonderful ethnological treatise by ReichelDolmatoff (1971) on Columbian Indians (South-America), an article by Leon-Portille (1964), and a book by Sejourne (1966) on ancient Aztecs and Toltecs (Middle America), (par. 5.3, pp. 1 5 6 - 1 6 2 ) ; — two Western, pre-Socratic conceptions of space come to complete this list: an article by Brunner (1957) on Egyptian space, and a book by Vernant (1965) on Greek thought (par. 5.4, pp. 162—164). I shall treat them in the given order. 5.1 Chinese space As mentioned above, only two works were read to supply information on Chinese civilization. Needham's book gives fairly exhaustive surveys of the different periods of Chinese philosophy and science. Granet elaborates a less historical and rather more comparative anthropological study. A first difficulty consists in the search for correspondences between the two, that is the need to determine to what degree both authors are referring to the same reality. It appears that both authors treat the ideas of a number of Chinese philosophers only of the fourth and third century before Christ whose ideas are estimated to have had a widespread influence on, or to be a good

144

Rik

Pinxten

synthesis of, Chinese thought about nature, time, space, and universe, as well as on logical problems in Chinese civilization (for example, paradoxes were formulated about the situation of nature by Hui Shih; cf. Needham 1956: 191-192). 2 4 For more initiated readers, the set of philosophers Needham designates in this respect encompasses those known as Taoists, Mohists, and Logicians. Both authors give some preliminary remarks on Chinese spatial conceptualization. In the first place, they stress the concreteness of space. This refers to the general feature that space and time are not conceptualized or verbalized as such: no specific term is available in Chinese to designate the concept of space. On the other hand, many words denote specific spaces or particular combinations of space and time. Moreover, spatial terms not only have denotive value, but are also often provided with connotative value. A good example of both of these characteristics is to be found in the most general representation of space expressed by Needham (1956: 288): 'Space was not abstractly uniform and extended in all directions, but was divided into the regions south, north, east, west and centre'. The Chinese concept that represents each of these five regions is quite complex. Each region is connected with a season, and is defined in terms of many and diverse attributes of the environment. For example, west implies the color white, mountains, autumn, hunchback, and harvest. The same goes, mutatis mutandis, for all regions. 25 Let us now examine this space concept with the help of Piaget's model. A. Topological space A.l Neighborhood. This feature is most obvious in Needham's report: the five regions of space are juxtaposed so that one is always contiguous with at least one other. This characteristic can be found explicitly in the following Chinese paradox: HS 3 'The heavens are as low as the earth; mountains are on the same level as marshes' (Needham 1956: 190). The heavens are neighboring the earth, etc. No further accounts were found. However, I think the foregoing are satisfying, and do not imply sophisticated interpretation. A.2 Distinctness. This feature is also quite apparent. I shall only relate one particularly clear form of distinctness. The five regions cited in A.l are occupied by five 'qualitatively different' sorts of people. The civilized human beings lived in the central territory, the

Epistemic

universah

145

space of the chief and of the capital and the palace. This domain was distinguished from all others in the most complete way: it was the domain of civilization, in sharp contrast to the others that were domains of 'barbarians'. Not only civilizations, but human beings as well, were estimated to diverge along such criteria; when someone was banned from the central region, he was said to be banned from mankind, in order to be doomed to live the rest of his life with the 'barbarians', the 'subhumans' of another region (Granet 1934: 92). The domains of the other regions were also mutually distinguished in qualitatively quite similar ways. Another spatial concept of distinctness exists between the particular concrete segments of space within one region. They have no clear mutual interconnection, and are conceived as distinct from each other. A.3 Ordering relation; betweenness. Here I feel obliged to refer to the relation of neighborhood between the five regions, in which concept the relation of betweenness is implicit but obvious. A most interesting and strong form of ordering is supplied by the importance of the central region and the general stress on 'center' in Chinese conceptualization of space. The capital constitutes in itself the center of the central region. As mentioned above (sub A.2), it is qualitatively distinguished from the rest. Supplementarily, it can be said to have a superior position on still other ground, since the chief, who has his residence in it, is held to be the animator of spatial order. That is, through ritual ceremonies that take place at regular times, he animates the forces that keep the spatial forms and proportions intact. A.4 Enclosure and boundary. These notions were also touched on above, under the heading 'distinctness'. A beautiful example of the explicit care about these segmentation is given in the following paradox: HS 6 'The South had at the same time a limit and no limit' (Needham 1956: 191). Also paradox C: 'The boundaries of space (the spatial universe) are constantly shifting . . {Ibid.: 193). A. 5 Continuity and discontinuity. These can be recognized easily in Chinese space and time: it is said that the space-time continuum forms the 'Great Unit' (Needham 1956: 194), while particular events and positions in it constitute 'Little Units'. The problem of the paradoxical relation between continuous and discontinuous in this

Rik

146

Pinxten

respect is expressed in two statements which Needham quotes from the Mo Ching (Cs /61 and ch. 60). Granet (1934: 113) gives a description of the specific relation between both that corresponds with Piaget's: 'Un complexe de rubriques identifiees a des ensembles agissants.' Thus far, all topological features of Piaget's model have been recognizable without difficulty, keeping in mind, however, that the Chineze concept cannot be accounted for completely, since only its Piagetian characteristics are focused. B. Projective

space

B.la Projective line. No comparable concept was found. B.lb Perspective is a fundamental characteristic of Chinese space, according to both authors. Needham (1956: 194) claims: 'From the standpoint of an observer at any one [particular location] of them, the universe will look very different from that which another observer sees.' Granet (1934: 91) gives another example of perspective: all towns and all landed estates are squares, which in turn lie in a specific perspective to one another. B.2 Coordination of perspectives represents a central scheme of spatial segmentation. A first example is the division of earth in five regions (cf. A.2). Furthermore, one other example from Needham (1956: 193) may stress the importance of this notion. He quotes the book Mo Ching again (ch. 331) : CS 'Knowing that "this" is no longer "this", and that "this" is no longer "here", we still call it South and North. That is, what is already past is regarded as if it were still present. We called it South then and therefore we continue to call it South now.' B.3 Sections.

No explicit, nor even implicit, references were found.

B.4 Surface. This concept is certainly available to the Chinese. There exists a notion of ideal space that represents the pure form in which space is to be preserved by the ruling activities of the chief. The ideal space has the concrete characteristics of a square (Granet 1934: 90). Again, Needham quotes a Chinese paradox on this matter (1956: 190): HS /2 'That which has no thickness cannot be piled up,

Epistemic

universals

147

but it can cover a thousand li (square miles) in area.' The expression 'that which has no thickness . . .' can be regarded as a nearly perfect intuitive description of a surface. C. Euclidean space C.la Angles and parallels are present in the idea of the square. They are quite explicit, since it is claimed that the square is sacred, as a sort of guarantee that the proportions of angles and lines should be kept as they are. That is also why the palace, the capital, as well as the five regions of earth are conceived of as squares. The deformation or demolition of one of those squares (by earthquakes, by occupation of the territory by an enemy, etc.) is considered a sign of weakness or decadence on the part of the chief, whose main task is indeed to preserve the 'holy form'. In order to differentiate as well-formed from a deformed square, for example in the case of the palace, one must possess notions of parallel and angle, (cf. Granet 1934: 91). C . l b Proportions. There were no unambiguous references found, except for the fact that, as mentioned in the foregoing paragraph, notions of proportion can be useful when conceiving a square. Since, however, there is no logical or factual evidence of the necessity of the use of proportions, I will consider the concept as 'not found'. C.lc Geometrical figures. Apart from square (perfect space) and circle (perfect time), it is said that some oblongs appear as intermediary forms of particular interactions between space and time (Granet 1934: 90). However, these oblongs do not seem to be geometrical stricto sensu; the only example Granet (1934: 87) quotes deals with a figure in the west with a rather topological, indeed nonmetrical character, situated between a circle and a polygon: 'T'ang le Victorieux, heros celebre . . . par la fa£on qu'il avait de marcher le corps tout courbe.' C.ld Vertical/horizontal. One profound expression of this organization principle has been given already: the HS 3-paradox about the distinctness of heaven and earth, of mountains and marshes (Needham, 1956: 190). This may suffice. C.2a Transposition

as such was not reported, but the notion of

148

Rik

Pinxten

distance is used with ease, be it only applicable within the frontiers of the central region of the five regions. Because of the presence of the chief, distances are reliable within this region alone (cf. Granet 1934: pp. 93—95). 26 C.2b Linear measure. No references were found. C.2c Calculation of surfaces and volumes. Here I am left in doubt. The sole quotation about surfaces is from Needham; reference is made to the 'Ii' (HS /2) as a measure of surfaces. Since the material is too scarce to justify conclusions as to the metrical quality and the representational character of the concept (it could just be a habitual set of concrete actions), I shall place it between brackets in the frame of Chinese space. The foregoing gives an idea, and I hope a correct one, of the probable common traits between Western and Chinese space. The earlier warning still holds, though: neither a complete characterization of Chinese space was attempted, nor an exhaustive confrontation of both systems made. If my analysis should be valid, it would be clear which Western features of space are certainly not t o be looked for in Chinese minds, or which Western biases researchers must be extremely conscious of with regard to Chinese space. This could be of some help to empirical studies on the subject of the Chinese natural knowledge system. 5.2 African

space

The material I found on the African natural knowledge system is rather diverse. Moreover, most of it treats only the behavior or thinking of one or two specific tribes. An important exception on this rule was the article by M. Griaule (1952) who presented a beautiful frame of concepts, claimed to be shared by at least ten different tribes. 27 His approach encouraged me to search for a justifiable delineation of an African 'macroculture', defined by a set of elements of particular tribal and/or regional civilizations with strong structural correspondences. In one word, I searched for a valid African equivalent to the widely accepted construct 'Western (or West European) civilization'. 28 The Islamic civilizations of North Africa form a complex unit, separated, in general, from 'black Africa' by the Sahara. The Bantu

Epistemic

universah

149

civilization constitutes a homogeneous cultural complex, reaching from latitude 2° North to the Cape of Good Hope, with the exception of South-West Africa). 29 Between both regions, a central strip of sub-Saharan territory extends, enclosing different countries and civilizations. Of these I shall consider the western part including Liberia, Ivory Coast, Gold Coast, Dahomey, Guinea, Ghana, Nigeria, and Sierra Leone. The influence of different colonial rulers (French, German, English) disturbed the picture of most of the original civilizations in this area. 30 Nevertheless, according to Agblemagnon (1969), a macro civilization of original inhabitants of the region can still be defined — a set of non — Westernized tribes sharing the same language and culture (Eve language [cf. Agblemagnon 1969: 40—41]). This civilization can geographically be situated in the states of Ghana, Togo, and Gold Coast. East of it, G. Calame-Griaule (1965) studied the Dogo civilization of Nigeria. On the west side of Eveterritory, the Kpelle can be found spread out in parts of Liberia, Sierra Leone, and Gold Coast (cf. Cole, et al. 1971). Thus, at least three 'cultural regions' will be, for the sake of accuracy, mutually distinguished. M. Griaule and J. Jahn, two eminent ethnologists and pioneers in the scientific study of African language and culture, treat of the totality of the above three regions. In their writings that I studied 31 they consider the sub-Saharan civilization (Jahn 1964) as a unity where intertribal correspondences are apparent and outweigh differences. Since the authority of both scientists is widespread and their knowledge of different African cultures is uncontested, one would feel inclined to accept their approach. However, I shall proceed in another way because, (among other reasons,) it so happens that I was unable to gather any material at all about space on the central region (Eve culture), and thus was unable to subject Griaule's statements to criticism. 32 I shall work as follows: Griaule's position will be advocated implicitly in that only one report will be presented (cf. the treatment of Chinese space), primarily to avoid a long and boring explanation. The information about each level, however, will always contain the results of the analysis for both Kpelle and Dogo, compared with Griaule's statements. If a specific feature is found in the three sources of information, it is considered a sub-Saharan feature. 33 The nonmatching characteristic, will be mentioned separately. Conventionally, the parts of analysis concerning the Kpelle will be preceded by the sign K, those about Dogo by the sign D.

150

Rik Pinxten

A. Topological space A characterization by Jahn (1964) of sub-Saharan conceptions of space can serve as introduction to the subject. He claims that four major forces are conceived as constituting the African concept of universe: Muntu, Kintu, Huntu, and Kuntu. The Huntu category 'includes localizing forces, place-time (not space-time) as a unity' (p. 58). In the sub-Saharan vision, everything in the universe is motion. The Huntu forces localize this motion in place-time units that are qualitative and highly particular. Jahn's statement suggests that the space concept of this civilization is topological in essence. Particular units of space in correlation with time are constituted, each one apart from others. No coordinate system, no perspectives, are supplied. The analysis will (also) test this assumption. 34 A.l

Neighborhood

A.2 Distinctness35 K: A wonderful and unambiguous demonstration of this relation is given in Kpelle classification. Cole, Gay, et al. (1971: 61—65 and Appendix A) found that Kpelle build up a 'Chart of "things" (Ser?)' as a major instrument of classification. The 'Sei? Chart' is subdivided in a number of classes. One of these holds 'planted trees, planted vines, and planted shrubs' (p. 239). They are classed together as a subclass of 'wild trees, vines, and shrubs' for the Kpelle. The principle of the latter integration, as well as the property that defines the class of 'planted things' was given by Kpelle in a firmly debated argument: (they) '. . . said that are forest things, both because they originated in the forest, and because they are only cultivated in the forest' (p. 239). 36 The analysis shows that all 'trees, vines, and shrubs' belong together because, among other reasons, they are found on the same spot. The relation of neighboring appears to hold. D: The segmentation of the human body into twenty-two spatially and functionally distinguished parts is the Dogo's model, in analogy to which he differentiates between phenomena in the universe. The neighborhood relation of distinct parts of the body codetermine their function (e.g. the word for larynx must be translated by 'Carrefour du nez, des yeux, des oreilles et de la bouche' [p. 64]). This holds by extension, for all parts of the body and thus, mutatis mutandis, of the universe (cf. especially Calame-Griaule 1965: 58-71).

Epistemic

universah

151

Μ. Griaule (1952) gives extensive examples of the organization of space in African architecture, choreography, and cosmetics. His example of the map of the village is quite relevant in this instance: the huts are built on a curved line (circular form), each one adjacent to at least one other (cf. figure 1). Only two little round buildings (a and b in figure 1) are purposely distinct from this chain. The totality of these buildings must give shape to a scheme of the human body, where (a) and (b) are the legs — distinct from the rest — and (c) is the head, the residence of the chief of the community (cf. figure 2; Griaule 1952: 17).

Figure 1

Figure 2 (interpretation by Griaule)

37

A. 3 Ordering Relation and betweenness. K: The notion of center is, again, well-established in Kpelle thought. For example, their huts are circular. In order to construct a hut, a stick is stuck in the ground, and a rope is bound to it. Then, a perfect circle is drawn around the center. The wall is erected on this contour, while the top of the pointed roof is situated just above the center. (Cole, et al. 1971: 37). D: The liver is considered to be spatial center of all bodily parts. It also serves as the center of personality (Calame-Griaule 1965: 60—61). On the role of the specific spatial order of parts to one another, referencee was made in A l and A2. M. Griaule (1952: 15—16) gives an example of the spatial segmentation of the universe into nine fundamental parts: eight parts are conceived of as lying on a spiral, the ninth constitutes the center of the spiral. A.4 Enclosure and boundary. K: A good example is given in the classification cited above (under A.l and A.2). Many things are

152

Rik

Pinxten

gathered in the class 'town things', while others are member of the 'forest things' (Cole, et al. 1971: 238—240). Both regions are strictly distinguished by the things they do and those they do not spatially enclose. The boundary between b o t h regions is strictly delineated, for example, by task definition: different tasks must be fulfilled in the t o w n region than in the forest region. D: Calame-Griaule (1965) gives ample examples of these notions. The left hand is used to accept things, to integrate them; the right hand is connected with the 'external' world — it gives away, treats with the outer world. The members of the tribe live on a strictly delineated territory facing that of the gods. If one crosses the boundary between the t w o when pregnant, the child that is born will be a deaf-mute (the m y t h of Auna). Since oral speech is essential in Dogo life and religion, this representation of boundary speaks for itself. Finally, Griaule (1952) claims that enclosure is a main feature of African spatial representation. The totality of the village is framed in an oval (form of rice grain). The same goes f o r the spiral f o r m of the universe (1952: 17 and 16). A.5 Continuity and discontinuity. K: no explicit references were f o u n d . On the other hand, many reported concepts imply the notion of continuity, e.g. constructs of circularity and of straight and curved lines (Cole, et al. 1971: 145). D: The notion of continuity is apparent in Dogo's vision on the 'channel of spoken words' ('Le chemin de la parole', Calame-Griaule 1965: 66—68). The Dogon have a specific spatial representation of it: the word follows a stretch, beginning in the m o u t h of the speaker and ending (if the word is 'good' it at least has an end) in the ear of the listener. The word is supposed to be projected in the air in a continuous zigzagging. Figure 3 gives consecutive representations of:

A. B. C. D.

a a a a

word spoken with rage word spoken with disdain calm and regular word song (word + r h y t h m )

Figure 3

Epistemic

universah

153

Μ. Griaule (1952) reports the same representation of speech as a general feature of sub-Saharan representation. Once more, the analysis resulted in the easy recognition of the total topological system.

B. Projective space B.la Projective line. K: only one citation can begiven. In my opinion, however, it is a rather ambiguous, not to say deficient, definition of projective line. 'The term [in Kpelle language] for a path refers equally to a straight and a curved line' (Cole, et al 1971: 145). 38 What results is a notion of aline starting from the actual place of the observer. D: No explicit accounts were found. In view of the scarceness of the facts reported in this respect, I shall consider the feature as improbable in African thought, and place it between brackets in Scheme 2. B.lb Perspectives certainly form a part of the African spatial constitution system. K: Cole, et al. (1971) report an experiment to test the representation of the left-right differentiation. It was pointed out that this dimension was acquired by the Kpelle children of se'ven to eight years of age (p. 275). The correspondence with Piaget's results is striking. The authors furthermore trust Kpelle's left-right orientation enough to use it as a prerequisite in an experiment on conceptual rule learning. D: Calame-Griaule (1965) reports many examples to show the manifest use of the three projective dimensions. Left/right is associated with male/female, pure/impure, and an elaborate system in which left and right hands or feet play a determining role. In a similar way, the roles of the above/below and front/back dimensions are pointed out (p. 77). M. Griaule (1952: 14) states that 'la vibration interne de la matiere' expressed in the pictorial representation of speech, is an essential characteristic of the African conception of the world. This vibration implies alternating of left and right, and below and above. In many dance patterns he presupposes the front-back dimension of the rhythm going to and for in a circle, that is, in sum, gradually expanding away from a virtual center.

154

Rik

Pinxten

Β.2 Coordination of perspectives. K: No explicit accounts were found. Nevertheless, one can easily demonstrate that coordination is implied in many constructions where spatial organization is involved, e.g. the building of the huts (cf. above A.3). Cole, et al. (1971: 37) go even further, claiming: 'The construction of the house exhibits at least implicit knowledge of what may be termed geometric axioms and engineering rules.' Of course, everything depends on the precise delineation of the so-called axioms. 39 D: Some explicit reports were found. One of them concerns Tannonce de la pluie' (Clame-Griaule 1965: 191). This concept is represented in a frame of directions, implying a north-south orientation of a rainbow picture, and an east-west axis with a clouds-andlightning complex. When both features are evoked together in this spatial frame, the concept Tannonce de la pluie' is constituted. M. Griaule (1952: 19) claims that space is essentially represented as a system with four directions (cf. directions of the winds). The pictorial, sculptural, and choreographical representation is of the next general form. Ί Figure 4

B.3 Sections. No references were found. B.4 Surface. K: No direct references were found, except for one 'class of things' in Kpelle classification, namely 'Earth'. However, interpretation of the data seems difficult to me. The class of earth is defined by the following elements: 'large flat rock, stone, dirt, mud, sand' (p. 247) On the other hand, it is said that the earth is constituted by four types of material only (in Kpelle mind, that is): 'dirt, stone, sand, and mud' (p. 239). The differentiation between 'stone', and 'large flat rock' is probably due to the surface characteristics of the latter, as is suggested by the adjectives 'large flat'. This interpretation is, however, partly gratuitious, since, to my sorrow, the authors neglected to investigate the matter. D: Calame-Griaule (1965) does not explicitly refer to the surface concept. Still, as in K's case, the notion seems implicit in other phenomena. One of these is the spatial organization of agricultural cultivation: the land is subdivided into 'squares' of approximately equal size. I think it justified to state that the concept of surface might be used for that purpose.

Epistemic

universals

155

Μ. Griaule does not give any references at all. In view of the rather indirect and insufficient data, I shall consider the presence of the concept as improbable (between brackets in Scheme 2). C. Euclidean

space

C.la Angles and parallels. K: No references were found. D: Calame-Griaule does not give explicit reports either, b u t many drawings by the Dogon do in fact express both notions (cf. above B. 4), as does their segmentation of agricultural land. Finally, Campbell (1964), who uses data of a widely set up cross-cultural experiment on optical illusions (covering large parts of Africa), stresses the fact that angles and parallels are certainly conceived in African thought, but that right angles are less dealt with because they simply do not appear as often in the natural environment of Africa as in the carpentered world that is ours. (This would explain the low ratings by Africans on typical Western optical illusions, where geometrical notions are involved, such as the MullerLyer test. [Campbell 1964: 312—313]). Before this rather confusing state of affairs, I conclude that no sufficient empirical evidence is available to decide whether the feature does appear or not. In Scheme 2, it will be placed between brackets. C . l b Proportions. K: Cole, et al. (1971: 168) report an experiment on proportional ordering by Kpelle children of six to eight years old. The task involved choosing systematically the 'greatest' out of a set of similar objects. The selected object is then, again, to be compared with a set of other objects of a group for the same task. The subjects scored 90 percent correct answers. They seemed t o ' sort objects relationally, that is in terms of the criterion 'taller or smaller than . . .'. D: No references were found. M. Griaule (1952) does not give any data that would yield a notion of proportion. I shall, in consequence, remain with the same conclusion as above in C.la. C . l c Constancy of size and geometrical figures. K: Experiments by Cole, et al., on size discrimination tend to yield the same results as Kendler's on American subjects. Moreover, they discovered in experiments on classification that the Kpelle had clear representations of the concept of circularity (Cole, et al. 1971: 145). (Cf.

156

Rik Pinxten

also the form of the huts.) On the other hand, 'Adjectives such as round, square, and triangular simply do not exist in Kpelle' (p. 145). On the contrary the adjectives like tall and short do not only exist, but seem to hold a central position in the semantic framework of Kpelle (p. 144). D: Calame-Griaule (1965) gives one detailed example of a geometrical figure in Dogo culture: a rectangular carpet, segmented in twelve identical rectangles. Within the boundary of some of the latter big lozenges are made, symbolizing human lives. The geometrical characterization of the carpet (pp. 516—523) is obvious. It is only a pity that the author's analysis does not concern it and that no other similar examples are given. It is thus impossible to control whether the geometrical features are originally African, or if they express colonial influences. In view of the former argument it must be claimed that constancy of size probably exists in African thought, as well as some rough geometrical figures (e.g. circle and quadrangle). However, no conclusions are justified concerning more specific geometric shapes. As above (C.lb), the construct will be placed between brackets in Scheme 2. C.ld Coordination system and Horizontal/vertical. K: No references were found. D: Calame-Griaule (1965: 15) claims that the coordination system with horizontal and vertical axes plays an important role in Dogo thinking and religion (cf. especially the 'parole de la descente de l'arche' [p. 115], and similar reports)! It may be significant that this author is the only one who reports this feature in African space. On the other hand, the scarceness of empirical data might be totally responsible for the meager results here. C.2 No references were found about (a) transposition, measure, and (c) calcul of surfaces and volumes.

(b) linear

This completes the picture of African spatial representation. 5.3 Indian space With Indian space the same initial difficulties were met as in the case of African thought: is it possible to delimit Indian macrocivilizations?

Epistemic

universah

157

A severe selection was made, taking into consideration different classifications of Indian civilizations. Since the amount of literature on North American Indians seemed overwhelming, I decided to cut drastically. Information on only two groups of Indian cultures was retained: 40 — the culture of ancient Mexico (Aztec-Toltec civilization); — the contemporary primitive culture of the Amazonian Indians. The analysis of the data of these sources will be set up in complete analogy with that of the African world view. Two distinct analyses are synthesized in the way they are reported (one single report). It must be said, however, that, since both civilizations are probably historically connected in some way, but since their identity can certainly not be advocated, room will be 4eft to stress differences, both in the analysis and in the synthesis of the results. Notations will be Am for Amazonians and Τ for Toltecs-Aztecs. A. Topological space A.l and 2 Neighborhood and distinctness. Both features are abundantly represented in literature on Indian civilizations. Am: Reichel-Dolmatoff (1971) reports that the Amazonians consider moon and sun as twins, they are one and they are two at a time. The notions A.l and A.2 are beautifully expressed in this vision. Their unity is stressed by their being spatially together, at least with regard to the earthly man. Yet they are distinct as well, since the one emits light and the other does not (explained in a myth); the one appears during daytime, the other at night. In pictorial representation, the two are placed near each other with the distinguishing characteristics (1971: 44). T: Leon — Portilla (1964) recorded that Aztecs conceived of the universe as constituted of nine consecutive levels: the elements within each level are conceived as being neighbors (e.g. all stars are located as neighbors on one and the same level), while the elements of each level are strictly distinct from those of another (p. 42). Sejourne (1966: ch. 3,2). further explains that earth and sky are considered as mutually distinct. Nevertheless, they are neighboring in space, even to the degree that an ascending road that leads man after his death from earth into the sky (heaven) is said to exist. A.3 Ordering relation and betweenness. Am: The following explicit ordering was found: land is manifested in forests; water exists in the

158

Rik

Pinxten

f o r m of rivers; they stand in relation t o each other spatially through the intermediary element, air. T: The Aztec civilization gives more daring examples of the use of this constitutional concept. Leon-Portilla (1964: 41) reports t h a t the world is conceived as the center of the universe. His finding is supported by Sejourne (1966: Ch. 3,2). T h e latter gives other examples as well: A widespread figure of central importance is the five-point figure ('quinconce' p. 91): it consits of a quadrangle with an articulated f i f t h point in its center. The combination of four regio's (four angles of the quadrangle) with a fifth, central regio is claimed t o be a fundamental sign in all religious experiences of the Aztecs. Cf. the 'quinconce' is, in its simplest form, 'constitue par cinq points renfermes ou non dans un quadrilatere qui symbolisent la pierre precieuse, embleme du Soleil, du coeur humain et de la douleur'. A.4 Enclosure and boundary. Am: The sun is conceived of as the Creator of everything on earth: all created phenomena are held together by the sun on our planet (enclosure), strictly delimited and distinct f r o m the domain of the 'big lake' (chaos) at the m o u t h of all rivers, situated at the place of the sunrise. The mouths of rivers and the beaches f o r m the actual boundary of earth (Reichel-Dolmatoff 1971: 30 and 3 8 - 3 9 ) . T: Leon-Portilla (1964) gives an explicit description of a similar (or the same?) world view and stresses boundaries: 'The earth, and its vast ring of water . . .' (p. 41). The earth is a big disk, surrounded by water. A.5 Continuity and discontinuity. Am: The most clear example of actual spatial continuity was f o u n d in the picture of the cosmos of the Amazonians. The moon and sun describe a continuous elliptical orbit around the disk of the earth, appearing above the disk and disappearing under it. Moreover, still more remote f r o m the creatures of earth, a string of stars is stretched out in the universe, constituting in fact its final boundary (cf. below figure 5). T: Identically the same ideas are found among the Aztecs, supplemented by a similar ellipse of Venus (Sejourne 1965: 98—99). T h e notion of discontinuity is complementary to the foregoing, for instance, by the belief that different 'levels' of cosmological phenomena, within which continuity holds (earth, sun-moon, stars), are mutually nearly independent. They are distinct in space and

Epistemic

universals

159

qualities. Thus I think it justified to conclude that the topological notions can be easily recognized in both Indian civilizations. B. Projective

space

B.la Projective line. No references were found. B.lb Perspectives. It is already apparent from the examples given above that perspectives play an important role in Indian spatial orientation especially in the cosmological models that were set up. Am: The Amazonians have strict delineations on the above/below dimension: above is the sphere of the sun, then comes the earth, and below is situated the sphere of death and Paradise. Life on earth is created by and dependent on the sphere of the sun. The earth is said to have two dimensions (the third being that of the sun and death), subdivided in the four directions of the wind (Reichel-Dolmatoff 1971: 46). T: The same elements are found among the Aztecs. Earth is built up along two dimensions: '(it) extends horizontally and vertically' (Leon-Portilla 1964: 41). And further: 'The earth, . . . are divided in four great quadrants or sectors which, opening out from the center of the world, extend to where the water joins the heavens . . .' (idem). The four regions are loaded with symbolic meaning. B.2 Coordination of perspectives. Am: The systematic spatial situation of all cosmological elements and the subdivision of earth in four directions presuppose an integration of the perspectives used, (cf. especially examples of B.lb). T: The same argument holds here. A further example is given by the specific orientation of the temples of the sun: they are always deliberately built along the east-west axis, with a distortion of 17° to the north. 4 1 Such operation is utterly impossible without the coordination of perspectives. B.3 Sections. No references were found. B.4 Surface. Am: The following quotation gives a lead for the statement that the Amazonians do in fact possess a notion of surface: the earth has the 'shape of a large disk, an immense round plate' (Reichel-Dolmatoff 1971: 24). The picture of the universe (figure 5 below) adds weight to this opinion.

160

Rik

Pinxten

Τ: Leon-Portilla's report on the Aztecs' representation of the earth as a 'big disk in the center of the universe' (p. 41) corresponds to the foregoing, in so far as the notion 'big disk' ought to imply the notion of 'surface'. This question can be answered, I think, quite satisfactorily with the data by Sejourne. The earth is conceived of as a quadrangle, organized along a network of horizontal lines, whereas the heavenly things are constructed vertically: 'La pyramide du Soleil semble percer le ciel, alors que la domaine du serpent a plumes [=earthly things] apparaft couche a meme le sol' (p. 87). C. Euclidean space C.la Parallelism and angles. Am: Consider figure 5 below: parallelism as well as angles are implied at different instances. The disks of earth and residence the actual Sun both imply parallels themselves and are mutually parallel. The notion of disk implies angles (the border of earth and that of the Sun's residence) as well. 42 The ellipses of the Sun-and-moon orbit and the static ellipse of the stars are parallel. Where sun/moon appear above the surface of earth, and where the stars end their ellipse and the Sun's residence begins angles are formed. Sun \ '

Stars /

ζ_ Residence of the Actual Sun J Figure 5 (Reichel-Dclmatoff 1971:44)

T: As far as the Aztecs are concerned, the information on the shape of the temples (mostly pyramids) as well as the quadrangular

Epistemic

universals

161

form of most of the 'quinconces' (cf. especially Sejourne 1965: ch. 3, par. 3) are expressions of the mastering of the notions of parallel and angle (cf. the explanation of the form of the 'quinconce' as the picture of the pyramid in a two-dimensional plane [Sejourne 1965: 91]). C. l b Proportions. Here we were confronted with a systematic lack of reports. Anyone who sees pictures or photographs of Indian culture, particularly of the Aztec civilization, is struck by the highly sophisticated mastering of proportions in the buildings and the paintings However, probably because of the specific scope of the material I selected, the subject of proportions was not dealt with. Therefore, in order to keep up with my methodological principles, I shall have to consider the notion of proportion as (highly provisionally) improbable. C . l c Constancies of size and geometrical figures. Am: Reports on geometrical figures are scarce. Reichel-Dolmatoff (1971) only refers to the circle (pp. 44, 56). On the other hand a whole series of not strictly geometrical figures, built up in strict symmetry, are reported (p. 47 ff.). These represent snakes, butterflies, etc. The geometrical character of most of the decorative pictures is apparent (pp. 56, 105, and the photographs preceding p. 169). T: Sejourne also reports the mastering of the circle. Apart from that the quadrangle and the pentagon occur (cf. above). Finally, she also stresses symmetry in most of the decorations (p. 142, ch. 4, par. 5). Above all of course, there are the pyramids that give evidence of the mastering of some geometrical notions. (It is, however, impossible to delineate in this work which geometrical notions are absolutely required, which are less necessary, etc., in constructing a pyramid in bricks.) C . l d Coordination system and Horizontal/vertical. Am: As mentioned above, the coordination of terrestrial space in a framework of horizontal and vertical axes is quite well established iin Amazonian representation. Two-dimensionality holds for spatial segmentation on earth; the orientation in three dimensions is valid in cosmological speculations (cf. especially Reichel-Dolmatoff 1971: 43—44). T: Leon-Portilla gives explicit reference to the Aztec vision on earth: a disk 'which extends horizontally and vertically' (p. 41).

162

Rik

Pinxten

Sejourne gives the same impression. The sky is vertically organized and the earth is in a horizontal network, that is as far as the buildings of the Aztecs are concerned. C.2 No explicit references were found about (a) transposition, linear measure, and (c) calcul of surfaces and volumes.

(b)

A final remark is certainly justified. As mentioned before, the mere existence of pyramids and big round temples worries me with regard to this analysis. In order to test the last three (or maybe all) Euclidean notions in an adequate way, more specific information on the architecture and art of Aztecs ought to be consulted. This was utterly impossible for me. Therefore, I hold the results of this part of the analysis as highly provisional.

5.4 Pre-Socratic

space

In this section only a few remarkable correspondences and/or differences will be pointed out. No systematic procedure of recognition will be worked out because of, among other reasons, the fragmentary information on the subject. Still, I shall describe a few features because of their clarifying power. The same general topological notions seem to be present in Egyptian space that were revealed in all cultures. 43 A most striking characteristic is its strict distinction between actual life space (earth), within which segmentation and orientation are possible, and the absolute chaos that is said to surround the world of ordered space and time. The boundary between both seems to be of extreme importance in Egyptian world view. Brunner (1957: 615) states that: 'Raum und Grenze untrennbar zusammengehören'. The foregoing explanation implies the notions of boundary, distinctness, and neighborhood. A particular form of enveloppement (enclosure) is provided by the belief that several holy places can actually be situated on one and the same physical spot. The latter thus spatially encloses several virtual temples or holy territories. As to the other (nontopological) spatial features, Brunner stresses the correspondences with our (West European) space, even with our use of metrical space. His references are, however, very poor; in fact they consist in a single statement: 'Der Raum des Alltags wird ebenso empfunden wie bei uns. Die Teile dieses Raumes sind messbar,

Epistemic

universals

163

wahrnehmbar, eindeutig lokalisierbar, der Mensch kann über sie verfügen' (Brunner 1957: 615). Vernant (1969) was consulted with regard to the pre-Socratic representation of space. 44 His analysis is mainly based on an old text ('Homeric Hymn to Hestia') that gives explicit accounts on the divine couple Hestia and Hermes. Essential in these gods is their meaning: Hestia is the symbol of the space of home (foyer), of the immobile center in a changing world. Ά Hestia, le dedans, le clos, le fixe, le repli du groupe humain sur lui-meme . . .' is characteristic. Hermes is in many ways her complement in spatial and functional attributes: '. . . a Hermes, le dehors, l'ouverture, la mobilite, le contact avec l'autre que soi' (Vernant 1965: 101). It is utterly redundant to stress even the topological aspects of the spatial features of these gods. One more example of the topological characteristic of boundary can be given. The author refers to a text of Plutarchos, relating the rite of 'Expulsion of Hunger' in which spatial categories (especially boundary) play an important role. A slave is walloped and driven out; when he crosses the steps before the house, the following words must be sent out with him: 'Dehors la Faim, dedans Richesse et Sante' (Vernant 1965: 137). Some geometrical figures are explicitly associated with each divinity. Hestia has always a round pedestal and even a round temple. Hermes is exposed on a quadrangle. The features of pre-Socratic space that were presented here certainly do not satisfy. They are mere examples to test the validity of my method of matching and recognition in a field that is, most probably, more known to the West European than the features of the other cultures that were considered. This ends the analysis. Scheme 2 will be constructed, gathering all features that are universal in the first column, and those that are probable in other civilizations but not in all in the second column. 4 5 It is evident, as mentioned above, that the Greek and Egyptian spatial notions will not be integrated in this scheme. The following information is useful for the interpretation of Scheme 2: — only those features that appear in the first columns (universals) are universal, common to European, Chinese, African, and Indian spatial representation in the same form; — nonuniversal features are, by definition, considered as specific to a culture. Each can, however, have a different status:

164

Rik

Pinxten

— plainly written out: there is no doubt that the notion is present in the respective culture — between brackets: there are doubts; due to the lack of empirical evidence it was impossible to determine whether or not the feature is present. Scheme 2 Universals

Nonuniversal features Chinese

A. A.l A.2 A.3 A.4 A.5

Indian

Topological space: All features were discovered in the three cultures neighborhood distinctness ordering and betweenness enclosure and boundary continuity and discontinuity

B. Projective space B.la B.lb perspective B.2 coordination of perspectives B.3 B.4 surface C. Euclidean transitions: C.la C.lb C.lc

African

Euclidean notions: C.ld -

-

C.2b C.2c

-

surface

(surface)

surface

parallelism and angles

(parallelism and angles) (proportion) (geometrical figures)

parallelism and angles (proportion) (geometrical figures)

coordination system (H/V) transposition distance

(coordination system H/V)

coordination system H/V

-

-

-

-

-

-

-

space

-

C.2a

(projective line)

-

geometrical figures

calculation of surfaces and volumes

Epistemic

universals

165

— finally, some notions of the West European spatial concept may not be recognized in any way in one or different civilizations at all. They will be omitted (straight line).

6 . CONCLUSIONS

6.1 Interpretation

of the results

Since it is my opinion that not the actual results of the analysis on the preceding pages, but rather the analysis itself and its underlying epistemological principles are most important, I shall stress only these latter aspects of the article. Nevertheless, the first conclusions will refer to the results, in so far as they serve as an example of the intention I had in my contribution to the Whorf symposium: First, it appears that a few, and namely the most basic features of spatial representation — according to Piaget's model, of course — turn out to be universal; that is, all topological notions can be recognized with ease in the spatial constitution of all civilizations. Some projective characteristics (namely, perspective and coordination of perspectives) have the same status. This does not imply that the respective features are identical, or present in precisely the same concrete actions or symbols all over the world. Such a statement indeed could be easily falsified, as was done at different places in the text when abstraction was made explicitly from just those nonspatial connotations in the diverse cultures. Universality only refers to the presence — in whatever context — of a purely abstract epistemic operator or structure (cf. Note 18); it is meant as a minimum requirement, not as a complete, exhaustive description of the spatial feature that is concerned. Thus, the actual spatial segmentation of two different cultures can be identical in a minimal degree (the Piaget feature can be recognized in both under the same form), but they can still diverge largely from each other with regard to the possible culture-specific connotations. Further empirical testing is necessary. Second, some higher level constitutions appear in nearly all nonEuropean civilizations (e.g. surface, parallelism and angles, geometrical figures, coordination system, and horizontal/vertical). Detailed empirical research could probably decide definitively whether these features are universal, or whether my analysis of them is completely gratuitious.

166

Rik

Pinxten

Third, a set of Piagetian spatial features appears as culturespecific, with a high probability of their specificness. Some of them are not found at all in other than the European culture (e.g. section), others appear in some but not all representational frames (e.g. distance). Again, only deliberately set up cross-cultural observations and experiments on these topics can bear decisive results. Finally, one important characteristic is implicit in Scheme 2. It is clear that universals are to be found mainly in the more fundamental regions of representation — the topological notions and some of the projective notions. Differences appear on the higher levels. This result corresponds exactly with the intuitions expressed above (1.1): the fundamental epistemic phenomena are bound to be independent of cultural and linguistic differences and to be completely determined by world-wide identical patterns of perception and action of the newborn human being. Differences only occur at the level of more sophisticated representation, which would then be more dependent on cultural and/or linguistic peculiarities, specific to the civilization that is considered. A word of caution is, however, necessary. I did not prove any dependence on cultural or linguistic characteristics as yet. This work remains to be done, partly anew, partly by comparing existing data (e.g. Lenneberg's) with the results of this study. I would, on the other hand, claim to have produced evidence for the universality — the cultural independence — of the more fundamental features. Moreover, I am convinced that more empirical data are required in both constitutional areas. 6.2 Epistemic

universals

With regard to the relevance of this article to the Whorfian questions (cf. Note 48 below, as well), I shall respect some distinctions made elsewhere (Pinxten 1972) together with the general idea of R. H. Robins in his paper to this conference. Robins states that the deep structure of language could be considered as universal, and consequently, would not be subject to Whorfian relativism. I would claim, as the net result of the analysis presented above, that the same holds for cognition or, more restrictedly, for epistemic phenomena. The universal characteristics are to be considered, in my opinion at least, as the actual definition of a specific part of epistemic deep structure (however unorthodox the

Epistemic universals

167

term deep structure is used there). If not, what else could this deep structure be? In accordance with an earlier article (Pinxten 1972), it must be stated that the problem is restricted to the anthropological study of the relation between language and thought. 46 It would thus be pointed out that universal cognitive features exist, and that the conditioning influence (there can be no more question about determining influence) 47 of language would be restricted to the higher level constructs. The Whorfian hypothesis in its weakest form is then pushed up (or aside?) to the level of less fundamental, more differentiated representations of space. It most probably remains relevant (cf. the impossibility of recognizing spatial features on the higher levels in all civilizations), though perhaps on less 'vital', fundamental topics than Whorf had hoped. Instead of determining one's outlook on the world as such, it would serve as a modifying instrument in the perception and conceptualization of more specific differenciations of segmentations or structurings of the world. 48 A further consequence of this analysis, and — I gather — the most crucial aspect of my contribution, is the denial of absolute linguistic primacy, both synchronic and diachronic, to be replaced by either cognitive or epistemic primacy (as Piaget 1954 advocates), or by some sort of coexistence of fundamental cognitive and linguistic characteristics. Further study, maybe most of all in the field of language and conceptacquisition, will point out which position can be advocated validly. The work presented here is then meant to be a first contribution in this respect. During many discussions in our group of Communication and Cognition, it was obvious that some fundamental conventions ought to be found concerning the respective domain of language and thought in order to be able to start a description of the relationships between them. The foregoing pages can thus be considered to serve as a rough delineation of at least one fundamental subsystem of the natural knowledge system of man, one subsystem of thought. Of course, extension to other subsystems and serious generalizations will probably be necessary. Nevertheless, this may be one way to the definition of thought and of language, (cf. the work of the universalistic linguists), and thus for a definitive treatment of the relationship between both. But even now, with the specific results we reached, a first and most important statement concerning this relationship seems possible: the thesis of absolute linguistic primacy is to be rejected. In order to end my plea for universals on more positive terms I

168

Rik Pinxten

would like to invite the audience to subject my attempt to the most severe criticism, particularly on its scope as an instrument in the elaboration of a cognitive anthropology, of absolute knowledge about knowledge. This might represent an ultimate object of joy for thinking human beings.

NOTES 1. I am very grateful to the group of Communication and Cognition which worked on the subject for nearly two years and supplied useful information during long-lasting discussions. I would particularly like to thank the professors E. Vermeersch, L. Apostel, Μ. Van Spaandonck, and F. Vandamme, as well as my friend J. Bernabe who kindly and patiently read through great parts of the text. The discussions of the group were mainly centered around Whorf (1956) and Schaff (1966). Furthermore, some members of the group presented their own opinions on specific subjects at large. 2. This is evident. For example, the law of gravitation is as active in Europe as in Africa. Of course, one can be conscious of it or not. This does not, however, interfere with its effect, but only with its mastering. Moreover, the phenomenal aspect, the appearances, may differ from culture to culture (probably because of its different conception). They must be considered as 'different values on a specific criterion', where the criterion is the explanatory concept in the knowledge system (e.g. somewhere an earthly god with specific powers), and the phenomenon is the statement, the description of the value. 3. This must not lead to an idealistic position, as was claimed by Schaff (1966: ch. 1 and 2). There are several possibilities: language and thought can be parallel mechanisms with specific interrelations, or one of them can be primary to and completely determinating the other, or they can be partly parallel and partly independent. I shall try to substantiate the last position. The choice of parallelism in any degree seems to eliminate all suspicion of idealism, although it entails other problems, as will show my work, I hope. 4. Cf. the well-known works of Lenneberg, Casagrande, Carroll, and others. It is significant that Lenneberg changed his position lately while studying the biological foundations of language (1967). 5. The term 'natural knowledge system' is introduced in analogy with the term 'natural language' in linguistics. It points to the knowledge system used in everyday life by all members of a community (or culture?), as opposed to 'scientific knowledge system' or other, more abstract or even metasystems. 6. Piagetian genetic psychology proposed a 'prelinguistic phase' of development (from birth to two years). The existence of such a period is widely accepted. There is, however, no uniformity in opinion about its actual characteristics (cognitive or pragmatic or something else?). Therefore, opinions of this area will not be used as arguments in this article. 7. The choice between 'universalism' and 'relativism' implies no mutual

Epistemic

universah

169

exclusion, as was recently most elegantly shown by Niyekawa-Howard (1972: par. 4). Recent facts point to the universality of the way languages are learned, and t h e relativity of what is actually acquired. Still, only linguistic p h e n o m e n a are concerned. This o u t l o o k ought t o be broadened t o include cognition (cf. Chomsky 1970: 3). 8. In a first approach Lenneberg (1953) f o u n d cultural differences o n color discrimination, as such, due to grammatical d e t e r m i n a t i o n . Later (1961), t h e f o r m e r results were changed and put in a broader perspective: discrimination, color discrimination in particular, was correlated with m e m o r y characteristics. Still later, Lenneberg ( 1 9 6 7 ) was convinced t h a t t h e reported relativism only holds for superficial p h e n o m e n a , and, in consequence, not for language and t h o u g h t in general. 9. It is one of t h e merits of J. Piaget to have been t h e first psychologist t o plea f o r a strict distinction between perceptual and representational space ( 1 9 4 7 and 1964). 10. The comparison of all scores on spatial visual illusions corroborates this statement. Cf. the excellent works by Cole, Gay et al. (1971) and Segall, Campbell, and Herskovits (1970). 11. I am very grateful to J. Bernabe, who helped m e repeatedly on this subject. 12. Among others, t h e remarkable work of Nelson G o o d m a n (1968) must be mentioned here. Following the intuition of Carnap's first book, he constructed a logical 'cognitive m a p ' wherein topological notions play an important part. 13. J. N e e d h a m (1956) estimates the first real Chinese influence o n Western t h o u g h t to have been a b o u t t h e seventeenth c e n t u r y , as can be concluded f r o m Leibniz' philosophy. 14. I read t h e reprinted 1972 edition. Since the original t e x t of 1947 was k e p t u n t o u c h e d in the reedition, I shall refer to Piaget and Inhelder (1947). T h e pages referred to will be f o u n d in the 1972 edition. 15. In this sense it is opposed to t h e aprioristic constitution system of G o o d m a n . T h e rules of constitution, as well as t h e distinguished levels must be psychologically real. It is evident that t h e y can at least partly be revealed through empirical research (e.g. studies on concept f o r m a t i o n ) . For this reason, the logical term 'analytical' will not be used to designate or characterize t h e relationship between features of d i f f e r e n t levels. Rather, t h e psychologically, or anthopologically, colored t e r m 'constitutional' will be used. 16. One of t h e main tasks Piaget and Inhelder ( 1 9 4 7 ) engaged in with this b o o k is presented as follows: Ί1 s'agit done de verifier de pres si vraiment la construction de l'espace representatif repasse par les memes phases, avec un decalage de quelques annees, q u e Celles de l'espace perceptif . . .' (p. 60). 17. Piaget and Inhelder (1947) originally agreed o n t h e relation of 'neighborhood (voisinage p. 15) as t h e f u n d a m e n t a l topological notion. A f t e r multiple discussions, Piaget ( 1 9 6 4 ) seems inclined t o accept a second primitive relation, namely a n o t i o n of 'betwixtness': 'On peut penser qu'il existe une relation seriale 'entre' qui suppose une construction operatoire de l'ordre' (p. 35). More detailed observations are required to decide w h e t h e r or not and in what f o r m the relation can be called primitive. This would be

170

18.

19.

20.

21.

22.

Rik

Pinxten

a psychological peculiarity. In mathematical topology, of course, the mere choice of one basic notion is sufficient. 'Une operation est une action susceptible de revenir ä son point de depart et de se composer avec d'autres selon ce double mode direct et inverse' (p. 50). I do agree that notions such as 'mental operation' and 'mental structure' raise difficult but important questions. On the other hand, I feel incapable of critisizing or amending these concepts here. Consequently, I shall use them in the sense I gather Piaget meant them to be read. Still, I must recognize one other main influence in this respect. The approach that must be held responsible for this influence might as well give hints for a suitable reformulation of these concepts. An ad rem example of this approach, and a tentative delineation of the influence, is Simon's (1967) restatement of the notions of 'cognitive structure' and 'cognitive process' according to the information theory. He heavily drew on the GPS language for this purpose, with the consequence that at least an unambiguous formulation was served. The problem now remains, I think of adapting Simon's concepts in order to give them more extended contents. The acquisition of a point of view in fact implies, as was convincingly explained by Piaget and Inhelder (Ch. 6, par. 3) the acquisition of ALL possible points of view' In sum, the perceived things that are external to us are put in a necessary and predominant relation towards the observer that faces us. An excellent example of this is the genesis of the correlation between the different sizes of the shadow of the same object relative to the position of the light source: The resulting operation consists of 'emboiter les petites sections de l'objet dans les plus grandes, en comprenant que ce sont les grandes sections qui masquent la lumiere, et expliquent ainsi la projection de l'ombre portee . . .' (Piaget and Inhelder 1947: 241). The correct representation of sections depends on the acquisition and suitable application of the elaborate coordination system of perspectives — of the sufficient differentiation between perspectives (cf. under 2). This is shown quite elegantly and convincingly by Piaget and Inhelder (1947: 310-313). These dimensions constitutionally and genetically appear late in the spatial constitution system. In fact, there is a major difference between the conceptual attainment and use of these dimensions (about ten years), and their perceptual use. The baby initially looks at the world f r o m a horizontal position, later (when able to sit up) from a vertical point of view. But, 'De meme, de ce qu'il sait se tenir droit ou se coucher, il tirera simplement la conscience toute pratique de ces deux postures, et d'abord rien de plus' (Piaget and Inhelder: 1947: 439).

23. Linguistic elements are probably bound to have no influence on it because the topological notions are developed mainly in a nonlinguistic period (from birth until two years), and are already completely present in a nonsymbolized form in this sensori-motor period (Sauvy and Sauvy 1972: ch. 1; Piaget and Inhelder 1947: ch. 1, par. 2). 24. Their role would be comparable to that of say Galilei or Newton with respect to European thought. Their theories and methods are not totally

Epistemic

25.

26.

27.

28.

29. 30. 31.

32. 33.

34.

35.

36. 37.

universals

171

integrated in common sense, but some general ideas about universe, gravitation, light, etc., are so far grooved in Western thought and technology (through education mainly), that they can be said to be common knowledge of all Western people by now. It is evident that I shall omit all non denotative elements. This will certainly impoverish the concept. It must be remembered, though, that Piaget's spatial characteristics were to be searched for; their meaning must not be enlarged or diminished at random, following the instance of comparison. Since it is claimed that Piaget's model is only a 'minimum program of universality', complete overlapping of its features with the elements to compare them would be an unfair requirement. It is a delight to notice here that Piaget's system holds for this concept, since distance can be conceived directly out of the topological notion of betweenness, together with the general (Euclidean) conservation operation. The lacking Piagetian elements in Chinese representation do not oppose the appearance of the distance notion. The tribes are: Bambara, Mandigue, Dogo, Boro, Minianka, Samogo, Mossi, Koule, Forgerons, and Trouveres. According to the sociological analysis of Agblemagnon (1969), they are to be found in Liberia, Ivory Coast, Dahomey, Sierra Leone, Guinea, and Nigeria. Much of what follows is the result of long and enthusiastic discussions with the Africanist, Prof. Dr. Van Spaandonk. Without his kind advices, my work at this topic would certainly be unable to stand criticism at all. I am most grateful for his patient and competent support. Of course, the faults in the analysis remain completely on my account. Information received from Prof. Van Spaandonck. Information received from Prof. Van Spaandonck. I could only read M. Griaule (1952) and J. Jahn (1964), but I found many references to similarly synthetic work of the same authors in other works on African culture. Primarily, Prof. Van Spaandonck's reservation kept me from joining Griaule's position. It is clear that the central part, the Eve culture of Agblemagnon, is deliberately left out. Since it could not be analyzed, it is, obviously, not represented in the conclusions. It must be said in advance that at least one source of information, namely Calame-Griaule (1965), can only bear indirectly relevant data on space. Still, it is an unusually rich ethnological report on the central role of language in the Dogo-culture. The work by Cole, et al. (1971) did not intend to be a systematic analysis of space. Neverthelesss, in some experiments (e.g. those on pp. 160—170) the material is organized following spatial criteria. The two features are taken together because they imply each other They are complementary. In order to reduce the length of the explanation, they are treated as one construct. Some Kpelle claimed that 'planted things' belong to 'town things'. This is of no importance since analysis would bear the same results. The male sex defines a free, but very important, part of the village space.

172

Rik

Pinxten

38. Compare this statement with Piaget's definition of projective line. It is a pity the authors were not able to make this comparison in order to formulate the Kpelle concept in a more precise and explicit way. 39. It must be clear that the purpose of this section on African spatial representation is, in fact, an attempt to define these axioms. 40. I gather the North American Indian studies have reached such a degree of sophistication, that an untrained approach to the field would seem plainly ridiculous. Nevertheless, I insist on claiming that a similar analysis (like, the one attempted here) ought to be made on North American Indians as well. 41. This measure (17") is based on the precise direction of the sundown from where the sun reaches its zenith during midday (Tropic of Capricorn). 42. One should pay attention to the fact that no metrical notion of parallelism (or of angle) is tested, only the presence of the conservation of parallels. Another piece of information is useful: there appear to be two suns in this picture. The first and most remote one is the real sun, not to be seen by man. The second one, neighboring the moon and the sun people can see, is directed (as the whole set of elements actually is considered to be) by the real Sun. It is, however, the creator-representative that gives light and life to the earth, with which the people are directly concerned. The moon was originally at the same level as the real Sun, but after a quarrel out of jealousy, it was banned to the regions of the representative sun and deprived of its light (myth, pp. 2 4 - 3 7 ) . 43. A most peculiar aspect of Brunner's (1957) article is his a priori distinction between 'qualitatively different' sorts of space — mythical and everyday — which are said to exist synchronically. He claims: 'So stehen Mythos und Arbeitswelt in eigenartiger Weise nebeneinander' (p. 619). Since he is the only author to reveal such inconsistency in 'primitive' spatial representation, I shall take it for granted the Egyptian space might be specific in this matter. This will not disturb my analysis, however, since the abstractions Egyptians are capable of must show in one or the other of both notions of space. 44. Vernant happens to be a distinguished writer on the subject of ancient Greek civilization. Moreover, the book I read contains a whole chapter (1965: ch. 3) where a thorough analysis is made of the problem that concerns me here: space. 45. The idea as well of the actual conception of Scheme II came to us from an article by the group on Eluard, sent very kindly by Prof. Minguet. They analyzed a poem of Eluard and integrated the whole of their data in similar, but more complex, frame. 46. In the article referred to, a distinction is made between the synchronic and the diachronic intent of the Whorfian hypothesis. In case of universals, both are involved in equal way, I suppose. Moreover, a distinction was made between psychological, sociopsychological, and anthropological (= macrosocial) levels. I treat the anthropological level. It goes without question that the results of such an analysis should incorporate concepts and regularities that prove valid for psychological and sociopsychological phenomena as well. Here again, empirical research must be set up. 47. Cf. t h e foregoing footnote: determining influence can only be spoken of, in

Epistemic

universah

173

my opinion, if in fact language would prove to be the only universal f a c t o r . Since, apart f r o m language universals, epistemic universals seem t o exist as well, mutual or unilateral conditioning can still occur. Piaget ( 1 9 5 4 ) would certainly go much f u r t h e r and claim a determination f r o m the cognitive side. 48. In this article stress is laid on t h e sole epistemic value of space. This must be seen in the research f r a m e of a general (cognitive) a n t h r o p o l o g y , t h e search for the (or an) elaborated cognitive m a p that constitutes a subsystem of a cybernetic model of m a n (Vermeersch 1967). T h e m o d e l was proposed by Vermeersch, its actual working out, the filling in of t h e f r a m e , is still t o be done. I would like to express m y gratitude to N. Vandendriessche, who was kind enough to do most of the t y p e w o r k for me. A special word of g r a t i t u d e goes to F. Winter who was so kind as t o correct my English, in order to m a k e the text readable. REFERENCES Agblemagnon, F. N'Sougan (1969), Sociologie des societies orales d'Afrique Noire. The Hague — Paris, Mouton (Le Monde d ' O u t r e Mer, Passe et present, Etudes 25). Allport, F, H. (1955), Theories of perception and the concept of structure. New York, J o h n Wiley. Brunner, Η. (1957), ' Z u m Raumbegriff der Ägypter', Studium generale, 10: 612-620. Calame-Griaule, G. (1965), Ethnologie et langage. Paris, Gallimard. Campbell, D. T. (1964), 'Distinguishing differences of perception f r o m failures of communication in cross-cultural studies', in F.S.C. N o r t r o p and X. Livingston (eds.), Cross-cultural understanding, pp. 308—336. Casagrande, J. B. (1963), 'Language universals', i n j . Η. Greenberg (ed.), Universals in language, pp. 279—298. Chomsky, N. (1970), Taal en mens. Deventer, Van L o g h u m Slaterus. D u t c h translation of Language and mind, 1968. Cole, M., Gay, J., Glick, J. Α., and Sharp, D. W. (1971), The cultural context of learning and thinking. An exploration of experimental anthropology. London, Methuen. Diamond, S., ed. (1964), Primitive views of the world. New York, Columbia University Press. Pb. Frey, L. (1967), 'Langages logiques et processus intellectuels', in Modeies et la formalisation du comportement. Paris, CNRS. Goodman, N. (1968), The structure of appearance, 2nd ed. Indianapolis, In., Bobbs Merrill C o m p . Granet, M. (1934), La pensee chinoise. Paris, Albin Michel. Greenberg, J. H., ed. (1963), Universals of language. Cambridge, Mass., MIT Press. — (1966), Language universals. With special references to feature hierarchies. The Hague — Paris, M o u t o n (Janua linguarum, Series m i n o r 59). Griaule, M. (1952), 'Connaissance de l ' h o m m e noir', in La connaissance de I'

174

Rik

Pinxten

homme au XXe siecle, pp. 11—24. Neuchatel. Grünbaum, A. (1964), Philosophical problems of space and time. London, Routledge and Kegan Paul. Jahn, J, (1964), 'Value conceptions in sub-Saharan Africa', in F. S. C. Nortrop and X. Livingston (eds.), Cross-cultural understanding, pp. 55—69. Lenneberg, Eric H. (1953), 'Cognition and ethnolinguistics', Language 29: 463-471. — (1961, 'Color naming, color recognition, color discrimination: A reappraisal', in Perceptual and motor skills. — (1967), Biological foundations of language. New York, John Wiley. Leon-Portilla, M. (1964), 'Philosophy in the cultures of ancient Mexico', in F. S. C. Nortrop and X. Livingston (eds.), Cross-cultural understanding, pp. 3 5 - 5 4 . Miller, G. Α., Galantere, Ε., and Pribam, Κ. (1960), Plans and structure of behavior. New York, Holt Rinehart and Winston Inc. Needham, J. (1956), Science and civilization in China. London, Cambridge University Press (History of scientific thought 2). Niyekawa-Howard, A. M. (1972), 'The current status of the linguistic relativity hypothesis', Hawaii working papers, pp. 1—30. Nortrop, F. S. C. (1964), 'Toward a deductively formulated and operationally verifiable comparative cultural andthropology', in F. S. C. Nortrop and X. Livingston (eds.), Cross-national understanding, pp. 194—222. Piaget, J. (1954), 'Language and thought f r o m the genetic point of view', in P. Adams, Language and thinking, pp. 170—180. Harmondsworth, Middlesex, Penguin. — (1964), 'Les travaux de l'annee 1 9 6 0 - 1 9 6 1 et le V I e Symposium ( 1 9 - 2 4 juin 1961) du Centre International d'Epistemologie Genetique, in L'Epistemologie de l'espace, pp. 1—40. Paris, PUF. Piaget, J., and Inhelder, B. (1947), La representation de I' espace che ζ I' enfant. Paris, PUF. 2nd ed.: 1972. Piaget, J., Inhelder, B., and Szeminska, A. (4948), Lageometrie spontanee de Γ enfant. Paris, PUF. Piaget, J., et al. (1964), L' Epistemologie de I' espace. Paris, PUF (Etudes d' epistemologie genetique 18). Pinxten, R. (1972), 'Reformulating the Whorfian hypotheses. Making them instruments in the study of problems in the field of general anthropology and behavioral sciences', Communication and cognition (Ghent), 5(1): 25-41. Reichel-Dolmatoff, G. (1971), Amazonian cosmos, Parts 2 and 3. Chicago, University of Chicago Press. Robins, R. H. (1973), 'The current relevance of the Sapir-Whorf hypothesis', Communication and cognition (Ghent), 6. Sauvy, J. and S. (1972), L' enfant a la decouverte de l'espace. Paris, Casterman. Schaff, A. (1966), Langage et connaissance. Paris, Anthropos. Segall, Ν. H., Campbell, D. T., and Herskovits, M. J. (1970), 'Some psychological theory and predictions of cultural differences', in Proshansky, Ittelson and Rivlin (eds.), Environmental psychology, pp. 153—165. Sejourne, L. (1966), La pensee des anciens Mexicains. Paris, Maspero.

Epistemic

universals

175

Simon, H. A. (1967), 'The use of information processing languages in psychology', in Les modeles et la formalisation du comportement. Paris, CNRS. Vandamme, F. (1968), 'Is transformational grammar a contribution to the theory of innate ideas?', Philosophica Gandensia (Ghent), pp. 93—108. Vermeersch, E. (1967), Epistemologische inleiding tot een wetenschap van de mens. Brugge, De Tempel. Vernant, J. P. (1969), 'Mythe et pensee chez les Grecs', in Etudes de psychologie historique, 2nd ed. Paris, Maspero. Whorf, B. L. (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J. B. Carroll. Cambridge, Mass., MIT Press.

PAULA BURGHGRAEVE

8

A Critical Analysis of Schaff's Views in Connection with the Relation between Language and Thought

l . T H E EQUATION O F THINKING T O HUMAN THINKING

As a starting point of the chapter concerning the relation between language and thought, Schaff (1966: 101—150) chose one of the possible definitions one can attach to the notion of thinking. By definition he equates thinking to human thinking. To clarify the relation between language and thought he believes it necessary to hold to a differentiating notion of thought. With a differentiating notion of thought he means that we can clearly discriminate between the behaviour of human beings and that of other living organisms. The question whether, in connection with human thinking, we are concerned with a process of pure thought followed by secondary verbalisation of this thinking or with a unique thought process realised in a given language is important {Ibid.: 101—102). This is the main problem to Schaff, and it can be translated in the two following questions: 1. Can language exist without thinking? 2. Can thinking exist without language? Schaff states that no author will hold that the use of language can exist without thinking because this use of language presupposes that the meaning of the words must be understood in a given language by the native speakers of that language. The reversed question — can thinking exist without language? — leads, as he himself remarks, to a controversy which he would answer negatively. In Schaffs view this means that if we speak of thinking this implies the use of language by the thinking organism. Schaff says that we can look upon this statement as the central hypothesis of his views concerning the language-thought relation. Roughly speaking the following relations are expressed:

Paula Brughgraeve

178

That individual X uses speech implies that individual X thinks. This relation is put up on the definition given to a linguistic system and consequently cannot be doubted. Schaff puts forward as a hypothesis a second relation, i.e. the inverse of the foregoing: That individual X thinks implies that individual X uses language. Logically this would lead to an equivalence relation·. L(X) * Τ (X) Such a relation would suggest that language and thought are interchangeable. However, Schaff himself warns against too literal an interpretation of this equivalent, relation and states that with this given monism he never means to establish an identity between language and thought. They form an organic unit which does not imply that its elements are interchangeable. Thinking also consists of other elements but only the mastering of language together with those elements allows us to speak of thinking (Ibid.: 13 3—134). There cannot be any doubt that Schaff detects a certain characteristic in the language that moves him to draw such a sharp line between the mental activity of man and that of all other animals. Schaff discusses this characteristic in the course of his expose whose purpose is to strenghten the proposed hypothesis. He wants to find proof in empirical material. Consequently he makes use of a number of empirical sciences. The empirical facts alone cannot justify a decision in favour of the hypothesis. Schaff makes the remark that this material can only be used profitably when we have an adequate definition of the notion of thinking, but he does not attempt to put forward such a definition. In the next paragraphs we will discuss not only the argumentation in full scope but also the denomination of the characteristic that in Schaffs view pins the use of language down to a level where thinking is included. As a conclusion regarding Schaffs view we think it important to stress his reserved position; although he poses the monism of language and thinking he is fully aware that a total breech with former phylogenetic levels is in sharp contrast with what has been discovered in biological sciences.

The relation between language and thought

179

2. ARGUMENTATION IN FAVOUR OF THE UNITY BETWEEN L A N G U A G E A N D THINKING

Schaff suggests two ways to solve the problem of the relation between language and thought. The first method, purely an analysis of concepts, is rejected as speculative. An analysis of the material that the exact sciences offer him remains as the second method. He thinks that the empirical facts have to be sought in genetic psychology, particularly in the period of the acquaintance of the child with a given language. In connection with this item we must attach importance to the pathological cases, for example the deaf, and the education of such cases. Further, Schaff refers to the study of aphasia. Neither pathology nor the study of aphasia takes a particular interest in the question put forward by Schaff. Only a few authors will mention the relation between language and thought. It is also true that these disciplines are not on a level where the formulation of explanatory models for the acquired facts is consistent. On the contrary, Schaff is placed before the most differentiating explanations we can imagine. It is from those scientific disciplines that Schaff must extract the appropriate material to give his hypothesis a higher probability. In the case of genetic psychology Schaff is particularly interested in the work of Vigotsky and of the Polish school. Vigotsky starts his remarks about the language-thought relation with the description of thinking as 'an orientation in the world'. From this definition, thinking exists before the use of language. But the important thing according to Schaff is the fact that Vigotsky states that the separate courses of language and thought on certain levels of phylogenesis and ontogenesis merges into a single unit. This is for Schaff a reason to put forward his hypothesis and to disagree with the definition stating that thinking has to be 'an orientation in the world'. Only with the acquisition of language are we allowed to use thinking as the characteristic of the behaviour we observe. In the Polish school Schaff finds a second argument to use in favour of his hypothesis. He refers to the idea that a child builds up its world of objects through its acquintance with the appropriate names for the objects. For the study of deaf children and children reared in nature he can give few results if any. The first sort of investigation of the Polish school is only interested in educational results; the second does not possess systematically collected observations on a scientific basis. In Schaffs view it is noteworthy that

180

Paula Brughgraeve

children lacking speech and not having learned another system of transmission of signs are condemned to a lasting mental infirmity. At the same time those children possess, on a potential level, enough capacity to become normally or even highly intelligent persons. (How would it be possible for Schaff to detect intelligence — which involves evaluating the ability to think — in individuals without a system of signs equal to language, if, as he says, one can only speak of thinking when language is involved?) Schaff states that every study about infirmities of speech refers to the importance of the part language plays in the development of the capacity of human thinking. He stresses the specific possibility created by the use of language, i.e. the capacity to generalize or to abstract. It is this characteristic that creates the situation where a representation of objects is not at all necessary. Those arguments from genetic psychology, in their scarcity, are useful to Schaff and favour his idea that pure thinking does not exist. For him this means that a thinking organism necessarily makes use of a language. He infers from the genetic material that the hypothesis he suggested at the beginning of the chapter 'Langage et pensee': 'Thinking implies language, becomes more probable'. We, on the contrary, believe that Schaff makes a capital mistake in deduction because he generalises whereas he is not allowed to do so. He carefully investigated the material in genetic psychology; he found a clear connection between the use of language and a sudden explosion of the possibility to think. An individual's specific exteriorisation of the capacity to think seems to be directly involved in the use of a language. Schaff concluded that only in the presence of language can the potential possibilities of thinking be realised. A pure thinking that makes no use of language can not exist. This generalisation is not acceptable because Schaff immediately assumes that thinking can use only language as a means for its exteriorisation. In other words, from the fact that he finds indications to presume that pure thinking (thinking not having any instrument or tool tc exteriorise) does not exist, he concludes that only language can be such a tool. Nowhere, however, was he able to produce the arguments in favour of such a decision. Instead of putting forward his immediate conclusion that thinking without language can not possibly exist, we would argue that it might be possible that other instruments of realising thought processes exist. Those instruments or tools can be more, or less, efficient than language, but their presence gives to the possessor the possi-

The relation between language and thought

181

bility of realising certain forms of thinking. We can easily follow Schaff when he speaks of the impossibility of indicating one pure thought process. On the contrary, we cannot accept the conclusion that thinking without language can not exist. In our view the collected material does not allow for such a conclusion. The question whether thought without language can or cannot exist can be answered after investigation only. The investigation must give us indications of how to answer the question concerning whether language is the only tool available to realise thinking processes. In the beginning of this section we mentioned that Schaff uses the material he finds on the study of aphasia. The same difficulties about scarcity of material and lack of interest for the relation between language and thought exist; however, to complicate matters, he must confess that the discipline in question is dominated by controversies about whether there is one or several types of aphasia, and whether a part or the whole of the brain is affected. The use of results and ideas from the study of aphasia does not seem to us a strong basis to prove a hypothesis. This criticism, however, will not withhold us from accepting the material and we propose to examine it without referring to the doubtful context from which it is taken. It is quite understandable that Schaff chose those authors who in one way or another could give him arguments in favour of his ideas. In particular he refers to K. Goldstein (1948) who defends the idea that as a result of aphasia a total change of the personality appears in an individual and he acts differently toward the outside world. Goldstein tries to explain this fact by the introduction of two attitudes the individual is normally capable of adopting with regard to the world around him. First, the concrete attitude connected with a concrete use of language. Second, a categorial attitude connected with a categorial use of that language. With the help of those two concepts Goldstein tries to clarify different forms of aphasia. He refers to the fact that in case of aphasia the categorial attitude and the connected categorial use of language disappear. Schaff assures us that a transposition of the different attitudes in the language finds an argumentation in the study of aphasia. Important to S c h a f f s proposals is the idea that the transition of categorial use of language to concrete use of language is the result of losing the ability to understand the meaning of the words (1966: 120—123). Schaff

182

Paula Brughgraeve

argues that from the moment the categorial use of language disappears in the individual, we are confronted with an absence of the possibility to generalise or to abstract. In other words, generalisation or abstraction is for Schaff exclusively the result of the possession of a linguistic system.

3. THE CENTRAL CHARACTERISTIC OF THINKING: THE CATEGORIAL USE OF LANGUAGE

Only after he has enumerated the empirical facts, does it become clear that it is the categorial attitude connected with the categorial use of language which must be involved to enable us to use the notion of thinking. The idea arises in the discussion of Goldstein's conceptions about different cases of aphasia. In the foregoing section we referred to the idea that the disappearing of the categorial use of language entails a disappearing of the categorial attitude, which means that the individual loses his ability to generalise or to abstract. We tried to find in the text some explanation or definition of what is meant with 'possibility to generalise or to abstract'. Quite strikingly, those notions are made understandable through a reference to categorial use of language, some examples of which are given (Schaff 1966 : 121, 127); and reversed, categorial use of language finds its explanation in an individual's ability to generalise or to abstract. Schaff seems to be satisfied and he proposes the following reasoning: the notion of thinking must be equated to human thinking because humans master a linguistic system which has as its most important characteristic the categorial use of language connected with a categorial attitude whose existence or non existence justifies the use of the notion 'thinking'. In favour of his hypothesis that 'thinking implies the use of language', Schaff uses arguments of two kinds: 1. Those that give us evidence for believing that pure thinking or thinking without tools to realise it can not exist; 2. Arguments that seem to give evidence for the disappearrance of the ability to generalise from the moment the categorial use of language no longer exists. In connection with the first sort of arguments we referred to the fact that Schaff immediately excludes the possibility of the existence of instruments other than language that could be capable of shaping thought.

The relation between language and thought

183

In connection with the second argument we referred to the fact that nowhere in the text does Schaff try to give an acceptable definition of the concepts 'generalisation or abstraction'. Categorial use of language is described by the possibility of generalising which itself is described through a reference to the categorial use of language. In that case it is indeed easy to state that generalisation is strictly connected with the mastering of language. It seems clear to us that in the given situation a definition of the 'possibility to generalise' must be found. Empirical investigation is necessary which does' not exclude the results of the study of the behaviour of living organisms other than humans. The same can be said about the question dealing with the problem of the instruments that help in the realisation of thought.

4 . A N HEURISTIC APPROACH TO THE NOTION OF THINKING

Schaff well knows that other definitions of thinking can be put forward. He mentions, for example, as a possible definition: 'orientation in the world'. He does not try to deepen it, and it is not necessary to his purpose because it is the starting point of Vigotsky's ideas about thinking. As we tried to show in earlier paragraphs, those ideas procure Schaff the arguments he needed to strengthen his own position. Meanwhile we discovered that the conclusions'drawn from the arguments were too strong. This gives us a right to doubt the definition of thinking proposed by Schaff. We might as well examine the other descriptions we found in the text. We shall pay special attention to a definition mentioned, but rejected, by Schaff. Schaff gives the following alternative description of thought: 'Un proces de pensee se produit ä chaque fois qu'un etre vivant resoud un Probleme' (1966: 126). Immediately Schaff mentions that this description simply transfers the problem. He says that we have now to ask ourselves what is the' meaning of 'solving a problem'. To defend his own position and ideas he gives a series of totally different examples which can be understood as practical situations in which different animals could be said to solve problems (Ibid.: 126—127). He refers to the absurdity of equating completely different species by saying they all think. He also makes an analogy with the use of the word 'language' to discriminate different systems of communication used by man, animals, etc, The use of the word 'thinking' in such different cases

184

Paula Brughgraeve

causes the same confusion. Schaff does not even attempt t o give a clear description of the meaning of 'solving a problem'. He merely states that the difficulty gives him the right t o exclude the proposal. Our problems, however, are n o t solved. We do not know what the meaning of generalisation is or if generalisation necessarily goes together with the use of language. We still are n o t convinced that language is the only instrument of realising thinking. Consequently, we shall try to clarify the proposed definition and for this purpose we shall use general information about thinking which we find in the study of problemsolving. As a rule in problem solving, a problem is presented as a conflict situation where an organism wants to reach a specific goal or wants t o find himself in a specific situation different f r o m the one he is in. The process of going f r o m the initial unwanted situation t o the final desired situation is seen as the solving of-a problem. This requires the application of means available to the organism which help to shorten and finally t o overcome the distance between initial and final situations. We can n o w describe a problem situation and the solving of the problem by the following five elements: 1. The initial situation is different than the situation wanted by the organism; 2. the final situation is the situation wanted by the organism; 3. there are intermediate situations that shorten the distance between the initial and final situations; 4. there are operators applicable to the different situations transforming them into the final, wanted situation; 5. t o decide whether it is confronted with a problem situation the organism must have the possibility of comparing the initial and intermediate situations with the final situation t o see whether they are alike or not. The last item means t h a t the organism must be able to state an identity relation to enable it to go f r o m the initial to the final situation. We can ask ourselves what this means. Undoubtedly this means that the organism must be able to discriminate between states if this seems necessary for the reaching of its goal. Similarly this means that the organism must be able to generalise states to do so. In other words, the identity relation requires f r o m the organism the possibility to classify. Without this possibility the organism cannot detect a difference or a distance between initial and final states, which means that the organism cannot even find itself in a con-

The relation between language and

thought

185

flicting situation and cannot be confronted with problems it has to solve. We can conclude that the existence of the categorial attitude, being the possibility to generalise, is absolutely necessary in an investigation of thinking started from another definition. This result is similar to the one Schaff found. Nothing however is said about the necessity of the use of language. The question whether or not this is the case must be solved by an investigation that tries to find out if the possibility to generalise exists for organisms not having the use of language. We shall consider that topic in the next section. Here we only stress the fact that as a result of approaching thinking from a different starting point we once more find the importance of the possibility to generalise or to abstract. In heuristics we are concerned with the methods we possess to solve problems. An approach generally used as a starting point in the investigation of the several methods consists in trying out all possible moves at random until the goal is attained. This is an algorithmic approach to a problem where the organism uses trial and error. When a problem has to be solved it is rather unusual that pure algorithm is used. The thinking organism will try to profit from all the relevant information it possesses in order to select, from the field of possible operators applicable to the present state, only those giving the greatest chance to succeed. The thinking organism makes use of heuristic information and does not submit itself completely to trial and error. Eventually we find in the literature about heuristic methods another quite important aspect of the question of under which circumstances we can speak of thinking. The aspect under discussion is clarified in the following: 'We assume, of course, that such a search can be carried out conceptually, say with a computer, rather than by using an actual 15-puzzle' (Nilsson 1971: 4). It seems clear that in the context of an investigation of heuristic methods it is presupposed that the organism under investigation possesses means that enable it to select only those paths with the greatest probability of leading to the goal before turning the process of thinking into acting. This gives the organism the possibility of acting at the moment he feels that the problem is solved, which means at the moment when the problem does not exist anymore. In an unpublished paper by L. Apostel (no date) all these characteristics are put together within a new definition of thinking: 'Interiorized experiments on models, using trial and error, not totally construed at random, but using certain systematically chosen

186

Paula Brughgraeve

heuristics or search strategies'. Starting from this approach we think we can put forward more explicit questions than was the case from Schaffs view. Three important aspects, which we found in the discussion of how to solve problems, must be present in the behaviour of an organism to justify the use of the notion 'thinking'. It is interesting to see if those aspects appear in the behaviour of organisms not having a speech mechanism. The questions we ask ourselves are: 1. In organisms which do not use language. Can we find indications of the existence of the possibility to discriminate and to generalize? 2. Is the typical process involved in a nonheuristic course of solving problems — trial and error — the only method possessed by organisms which do not use language? 3. The process going from the initial state to the final state must have the possibility of taking place on a conceptual level. This means that without a repeated concrete testing of all the possible moves an efficient solution must be actualised in an adequate behaviour of an organism not using language.

5 . E X P E R I M E N T S WITH A N I M A L S

The three aspects put forward in the former paragraph plead, at first sight, in favour of an equation of thinking to human thinking. The conceptual level clearly exists together with a use of language. With the elimination of the use of language it becomes quite difficult to see how an organism could possibly formulate a solution of a problem without a constant use of actual behaviour. In this view one gets a strong impression that trial and error must preeminently be the procedure for problem solving used by an organism that cannot master language. Nevertheless, in experiments done with several nonhuman species, we find indications that a connection between conceptual level — possibility to generalise — and the use of heuristic information by means of language is not as evident as one would guess. For each of the three elements we mentioned above we shall refer to experiments which make Schaff's ideas more doubtful.

The relation between language and 5.1 Generalisation and

thought

187

discrimination

When working with animals, a clear differentiation must be made between real generalisation on the basis of the possibility to differentiate, and generalisation on the basis of a lacking of this possibility. Only generalisations of the first sort are interesting for us. Various experiments made, for example, by Lashley (1934, 1938), Smith, etc., show that identification and the capacity of identifying distinguishes the different species. For example, a chimpanzee is in most cases able to identify a triangle in spite of changing colour effects. A certain degree of generalisation appears, but the appearance is only the result of a learning situation. Such a generalisation would not be possible for a rat. The experiment enables us to state that different species have specific possibilities of generalisation that have an upper limit and only appear after a learning situation. It is quite important that without a name connected to the discriminated object, certain species can realise identifications that require an act of generalisation. Animals must have at their disposal certain concepts without having a la guage. Material about animals learning to discriminate and to generalise has been systematicaly worked out by O. Hebb (1949). The author particularly gives his attention to the possibilities of learning by animals — by organisms who have not mastered the use of language.

5.2 Trial and error To avoid any reference to intelligence in other than the human species, learning on the basis of trial and error has for a long time been put forward as the only way of acquiring knowledge for nonhuman species. A reaction against this idea was launched on the basis of experiments that proved the mechanism of trial and error as insufficient to explain the behaviour of animals in problem situations. Trial and error implies a progressive increase of efficient behaviour until a maximum of positive results is reached. A whole series of experiments showed a different course than we might expect from learning by trial and error. After a period of fruitless search with different orientations leading to no result, there would be a sudden change, creating a total breach with former efforts, and an efficient behaviour would be produced. In literature on the subject, the sudden appearance of the appro-

188

Paula Brughgraeve

priate behaviour is explained through the notion of insight. We do not want stick to this particular term, knowing that its meaning is as little clearly defined as the meaning of thinking. The interesting fact for us is that within the organism a change seems to have taken place that produced the sudden aappearance of the appropriate behaviour. We are unable to define the nature of the change at this moment. Other experiments have shown that even animals work with hypotheses. We only mention here Krechevsky's experiments with rats (1971). Hungry rats were given the choice of opening one of two given doors, one to the left and one to the right, one white and one black. The colour of the doors was changed. Food was to be found behind one of the given doors, the other was kept closed. Krechevsky found that rats did not act at random. The rats tested a systematic series of strategies until one of these attempts solved the problem. Psychologists interpret this behaviour by saying that the rats have a series of hypotheses. Learning only with the help of trial and error did not show up in these experiments. A pure trial and error process appeared to be a less frequent and less evident behaviour than one would think. We see rather, a behaviour that tries to procure a maximal efficiency at minimal cost. 5.3. Conceptual level process It is not easy to find evidence for the existence of a conceptual level process without the use of words. The difficulty arises from the fact that an organism lacking an intersubjective communication system can not give a report of its mental activity. We can not find evidence for the mental activity of those organisms other than the material discussed above. We already mentioned in 5.2 the sudden changes in the behaviour of an animal connected with an appropriate solution of a problem. Every one knows Köhler's experiments with chimpanzees. The experiments clearly show the appearance of 'insight'. The fact that the sudden change cannot be attributed to a progressive upbuilding of the solution through concrete behaviour is important. On the contrary, the chimpanzees stopped doing things, and it was after the period of inactivity that the efficient behaviour is suddenly actualised. The only thing we can be sure of is that in the animal something has taken place that caused the finding of an efficient solution. If for such processes we ask for an irrefutable proof we must confine ourselves to the experiments that give an indication for

The relation between language and thought

189

the existence of generalisation without the use of language. We mentioned some of them in 5.1. An investigation to deepen the results is necessary. This would give us at last the possibility of deciding in favour of or against Schaff's hypothesis. In any case as we have seen we can not immediately exclude the possibility of the existence of an interiorised experiment in those organisms. It is important to ask if we can explain any form of behaviour we find in the organisms not mastering language without referring to the three aspects, or analogues of those aspects, which we mentioned as characteristics of thinking. Schaff himself, at the end of his expose, can not deny the fact that an investigation of the behaviour of other organisms may shed more light on the problem. If he does not do so it is because he is sure that mastering a language implies mastering generalisation or abstraction and that lacking a language implies lacking generalisation or abstraction. For him, having the use of language creates a qualitative difference on the basis of generalisation that clearly discriminates between the users and nonusers of language. A superficial glance at the experiments with animals shows that the exclusive characteristic 'generalisation' can be traced even in the absence of language. Tnis gives us a right to conclude that Schaff's hypothesis can only be retained if he defines generalisation as the labeling of objects, relations, etc., with words. He can only put forward such a definition if he has enough arguments to defend it. This means that he must find arguments which would make us believe that, for example, the chimpanzee, who recognised a triangle without being disturbed by changing colours and backgrounds, did not generalise because he could use the word 'triangle'.

6 . THE ROLE OF LANGUAGE IN A THOUGHT PROCESS

We mentioned the fact that Schaff boldly concludes that thinking without language cannot exist. He reached this conclusion through arguments that give evidence for the fact that pure thinking does not exist. We tried to weaken his conclusion by stating that thinking is possible if the organism possesses an instrument to realize his thinking. We did not say that the instrument has to be a language: it could be some other means. From the material he found, Schaff also concludes that language is responsible for the rising of the categorial attitude connected with the categorial use of language which enables

190

Paula Brughgraeve

the use of generalisation. A definition of generalisation is not given. We tried to work out another approach of thinking mentioned, but not accepted, by Schaff. This approach procured the three aspects or characteristics of the process that have to be present if we want to talk about thinking. It also showed that the possibility of generalisation is necessary for thinking. Nowhere, however, is the mastering of language regarded as indispensable. Experiments with animals procured the material to presuppose that the different aspects (generalisation included) could be traced in organisms who have not mastered language. Explaining the whole behaviour of animals without referring to generalisation seems impossible. It seems quite clear to us that more investigation is necessary to define 'generalisation' or 'abstraction'. From this viewpoint the central item of investigation is the forming of concepts. We did not suggest anything about the possible existence of instruments for thinking other than language. The answer to this question hangs together with the answer to the question about whether or not we have the right to speak about concept formation in organisms not using language. Let us suppose we have the necessary answer to the question about concept forming that asserts the existence of generalisation without the use of language. This would be enough to presuppose the existence of instruments having the same use of language. We mean for example — and this has to be understood as pure speculation — that emotional differentiation connected with impressions of the several senses could create total images we could describe as concepts without labelling with the help of words. Whether this is the case dr not we are unable to state at this stage. To state here and now the exact role of language would come down to drawing conclusions that go further than the available material offers us. Language plays an important part in the realisation of human thinking. Thus far we agree. At the other hand we mean to have the right to doubt strongly the implications of S c h a f f s hypothesis that thinking only appears at the human level and human beings only think on the basis of the use of language. This means that other human thinking does not exist. To put an end to the confusion about the meaning of the notion 'thinking', we could come to an agreement and speak of thinking when language is involved, but this agreement certainly does not follow from Schaff's arguments. This means that we have to introduce a covering notion to indicate all behaviour having certain

The relation between

language and

thought

191

characteristics, for example those we found in paragraph 4, but not using the specific instrument language.

REFERENCES Apostel, L. (no date), Logika als theorie van de efficiente heuristiek. Gent (Unpublished paper). Goldstein, K. (1948), Language and language disturbances. New York. Hebb, D. O. (1949), The organization of behavior. A neuro-psychological theory. New York, John Wiley. Krechevsky, I. (1971), ' "Hypothesis" versus "chance" in the pre-solution period in sensory discrimination learning', Univ. Calif. Publ. Psychol., 6: 27—44. Lashley, K. S. (1934), 'The mechanism of vision. The projection of t h e retina upon the cerebral cortex of the rat', J. Comp. Neurol., 60: 57—59. — (1938), 'An examination of the "continuity t h e o r y " as applied t o discrimination learning',/. Gen. Psychol., 26: 241—265. Nilson, N . J . (1971), Problem-solving methods in artificial intelligence. New York, McGraw-Hill. Schaff, A. (1966), Langage et connaissance. Paris, Ed. Anthropos. Original edition in German: Sprache und Erkenntnis. Wien — Frankfurt — Zürich, Europa Verlag, 1964.

PAUL VAN GEERT

9

Language and Meaningful Intuition of Reality A General Contribution to the Problem of Linguistic Relativity 1. INTRODUCTION: THE PROBLEM OF THE ELEPHANT

In making an appeal to a series of more or less scientific deductions, I shall start from a rather well-known parable, in order to clarify the concept of meaningful intuition of reality (MIR) and to determine its position towards the Whorfian hypotheses of linguistic relativism. This parable is called: 'The Elephant and the Blind Men'. As the parable goes, on a good day, some blind men struck upon an elephant, which they directly started to examine. One of them made a thorough investigation of the trunk, another one of a leg, and still another one of the tail. They had rather few contacts with each other — by reason of some terminological divergencies, among others — and this explains the fact that for a long time legs, trunk, tail, and so on were considered to be rather independent entities. But one of them formulated an hypothesis which he called 'trunk relativism'. He had discovered that the characteristics of the trunk were conditional for the characteristics of the legs (big trunk, big legs; wrinkled trunk, wrinkled legs; and so on), and when there was no trunk there could not be any legs at all. All this disproved the old conception of trunk-leg independency. We shall use this — disrespectful — history as a working parable. First of all, it seems rather clear that the contribution of the trunk relativist constitutes an important step forward. The trunk relativist has indirectly shown the fundamental unity of the described phenomena. This, however, does not improve the fact that his problem, f r o m the viewpoint of someone who can see, is a false problem: it is an illusive problem to look for causal or conditional connections between trunk and legs of an elephant. These correlations only

194

Paul Van Geert

depend on the fact that trunks, legs, and so on are components of one and the same elephant. To the blind men, 'elephant' is a postulate that summarizes trunk, legs, and so on in a sense-making entity, wherein trunk and legs get a functional significance. 'Elephant' is the primary level of description, while trunk, legs and tail form the secondary level of description. The problem of the interactions on this level necessarily must be faced against the primary connection 'elephant'. On the primary level of description, 'elephant', new problems can be posed, e.g. the construction of trunk and legs in function of the relation between the elephant and its environment. Starting from this parable, the Sapir-Whorf hypothesis may be regarded as an important step forward on the one hand; on the other hand, however, it is a fictive problem. Against the background of the general human behavioral situation, language, meaning, thinking, and so on are concepts that stand on a secondary level of description: they are related in the same way as trunk and legs of the same elephant. The only problem now is to find an 'elephant', i.e. a generalizing postulate that could serve as our primary level of description, towards which the relation between perception, cognition, and communication could be put in a new light.

2 . MEANINGFUL INTUITION OF REALITY: GENERAL DEFINITION

Language (as a system of meaning), perception, thinking, and so on are diverse forms of so-called epistemic behavior that can be considered, more or less autonomous, 'mechanistic', and logical structures (or, they can be described as such). On the primary level of description of this epistemic behavior we can not do without the fact that man is a conscious being. (Sorry for this perhaps too phenomenological analysis; however, the reader should be aware of the fact that even in empirical science making, in the process of hypotheses formation, not a single method is forbidden. In reality, the formation of the MIR concept happened in an inverse way: First I executed two preparatory experiments starting from some confused notions at the back of my mind, then I came to the phenomenological interpretation, and after all to a trial for formalization.)

Language and meaningful

intuition

of reality

195

In his direct confrontation with reality, man's own consciousness is involved, this means that in the conscious confrontation with reality, the bound between 'me' and 'environment' does not exist, but this does not alter the fact that one can make oneself conscious of such a bound (reflection). The fact of being present one self, as subject, in the situation one perceives, is a phenomenon of primary importance. The way on which one's own existence or presence is fused with his perceived environment is called the meaningful intuition of reality. Let us clarify this by means of an example coming from physics. The subject and his environment can be compared with two unlike electric poles. In this situation the content of the subject's consciousness is the magnetic field between the two poles, subject and environment. (Consciousness may elucidate the existence of subject as well as environment, starting from the characteristics of the magnetic field.) The concept of meaningful intuition of reality rests on some preliminary notions: — every experience is an experience of reality, which does not imply, however, that every experienced reality belongs to the field of accepted reality (e.g. dreams, visions . . .); — there is no such thing as pure experience of reality. If there is such, it is an uncommon thing, the result of continued mental training or of specific situations that cause states of profound alienation and/or raise a negative effect; — every experience is meaningful and structured. Nevertheless, this meaningfulness and structure are merely nonpropositional. (All these characteristics are summarized in the term intuition used in an uncommon way.) The concept of meaningful intuition of reality is used as the 'elephant' here: It is the summarizing concept to denote the general epistemic relation of a human being towards his environment. This epistemic relation is traditionally divised in perception, recognition, understanding, talking about, thinking about, knowing, . . . These phenomena are related to MIR in the same way as trunk, legs, and tail are related to the elephant. This means that none of these phenomena can be related to other ones in a causal or conditional way (the illusion of trunk relativism). We adopt the idea that a specific MIR (the MIR of a specific situation) can be described in a way that is highly analogous to the description of language in T.G.G. This means among others:

Paul Van Geert

196

— that a given MIR can be regarded as a surface structure whose characteristics can only be understood when confronted with the underlying deep structure. (This deep structure is a merely theoretical notion, necessary to understand the relation between a MIR and expressions or behavior based hereon.) — the deep structure is a description of a MIR in the form of a transformational entirety. The way in which such a 'MIR-marker' is constructed differs fundamentally from the way in which a 'phrase marker' is constructed; (see further.) Hitherto, two things appear to be fundamental: a) the insection of the MIR as 'elephant', i.e. a notion to summarize or to describe at primary level all forms and sides· of epistemic activity as it is given in the direct conscious contact of a subject with his environment; b) the transformational and structural approximation, which permits an answer to the trivial looking question why one says, 'John eats an apple', when he sees John eating an apple.

3.

P R E P A R A T O R Y EXPERIMENTS

The empirical material on which this theoretical frame is based may not be regarded as a proof of it, but as background against which it becomes significant. (These were parts of a more general investigation on the linguistic capacity of nursery and primary school working class children.)

3.1 Enchanted world game The subject receives the instruction to tell a coherent story about two, three, or f c u r drawings, coming from an enchanted world in which nothing is impossible. What cards will be presented together is due to random choice. The whole problem is to formulate a sense-making story about non-coherent, non-existant and never perceived things and situations. After all, it is remarkable that the complete coherence and structure of the story is already given in the perception of it. (It would sound rather strange to say that 'the perception of the story already would be given in the structuring of it', but it would not be wrong after all.)

Language and meaningful intuition of reality

197

An analysis of the stories shows, among other things, the impossibility of stating that either perception or language played a conditional role in the structuring of the story. (The experiment was initially developed to show the influence of language on the perception of unexpected phenomena. The original hypothesis was linguistic-relativistic.) This fact originated the hypothesis that language (concept structure), perception, and meaning in the approximation of reality must be seen as an entirety where afterwards the parable of the elephant made it possible to elucidate this connection. An analysis of the stories would also show that convergently thinking subjects, having little creativity, confirmed most strongly the hypotheses of learning relativism: These subjects relate as having seen what they suppose they are forced to see (what has been taught that reality 'really is').

198

Faul Van Geert

Such a teaching relativism could easily be interpreted — wrongly, however — to be linguistic relativism, because much of what one has learned precipitates in one's own language. Let us illustrate this by giving an example of t w o stories about the same drawing. This drawing (see below) forms part of a series of three: this one, the second, shows a lady putting a strangely composed kettle on a gasstove. A: This is a watch. This man is busy winding it. There is a bird on t o p of the watch. There is a glass of beer on the cupboard, it belongs to the man, for he is thirsty. B: This is a machine to make beer. It works with a hand. The man puts it on . . . rrr . . . there comes a hand that gives him the beer. It is important to remark that there was no perceivable span of time between the presentation of the drawing and the telling of the story. The essential element of b o t h stories remains in other versions of it, although in different order. This suggests that the ambiguity of the drawing does not constitute a problem that has to be solved by means of a linear process of deduction, dominated by the structure of the subject. We might better see it as a hierarchical process of explicitation (see par. 4.2) of a base scheme in which the constituting elements of the perception and interpretation of the situation are immediately given.

3.2 Choosing

alternatives

The experimental set contains t w o series of drawings. Fifteen drawings represent simple situations (a boy eats an apple, boy and girl looking to each other, and so on). Fifteen series of drawings represent the elements of the drawings of simple situations but in a splitted f o r m (drawing of boy and drawing of apple). In the first part of the experiment, the experimenter shows one card and asks the subject to tell what he sees on it. The sentence(s) is (are) recorded and eventually written down. T h e sentence(s) is (are) called the 'referential description' or 'referential sentence'. This first part is normally followed by another experiment or pause, in order to avoid too explicit influence of the referential descriptions on the second part of the experiment. The reason for this requirement is that we want to study the predicting value of the

Language and meaningful intuition of reality

199

referential description with respect to the choice of behavior in the second part of the experiment. When the referential description has a predicting value with respect to the choice behavior, in spite of the lapse of time between the two parts of the experiment, and this predicting value can be found via the transformational system of the semantic matrices (see further), then we have at our disposal a certain proof of a) the fact that the referential description represents the way in which the described situation is perceived; b) that the system of semantic matrix transformation is a valuable instrument for the description of meaning-producing perception (MIR). In the second part, the experimenter shows two "emantic 'halves' of the initial drawing. For example, when the initial drawing shows a dog running after a cat, the two halves show a running dog and a running cat. When the initial drawing shows a boy eating an apple, the two halves show a boy and an apple. It all depends on the sort of situation whether the referential meaning of the verb can be kept in the separate 'half' drawings. The experimenter reads a sentence, e.g., 'The boy eats an apple'. The child has to show the picture that matches the best with the sentence. To every pair of pictures, there are six or seven sentences that are more or less synonymous but vary in grammatical structure. There are three types of situations. Type a: This is a green tree

Fig. 2

200

Paul Van Geert

Type b: The boy eats an apple

Fig. 3 Subtype: Mama knits a

sweater

Type c: The man is sitting in an air

balloon

Fig. 4 Subtype: The boy looks at the girl For detailed results and interpretation see appendix. Neither the syntactic structure or the logico semantic content of the stimulus sentences is a good criterion to predict the choice pattern. The choice pattern depends on the way in which the referential MIR (i.e. the meaningful intuition of the situation, leading to the referential sentence of the subject) is structured, and on the way in which this structure of the MIR, that, following the principles of the

Language and meaningful intuition of reality

201

semantic matrices, should underlie the stimulus sentence. We shall give an example of this in the text on the semantic matrices.

4 . THE STRUCTURE OF MIR: SEMANTIC MATRICES

4.1 General notions The MIR structure is nonpropositional, but it supplies the non propositional base on which propositions may be generated. If we would describe the MIR structure by means of propositional forms (linguistic forms), we would reduce it a priori to the possibilities of our language. This would lead us automatically to linguistic relativism, because we ourselves have adopted the structure of the MIR to the structure of language. Nevertheless, we are not able to say anything on MIR structure but in propositional terms. So we face the dilemma either to say nothing at all, or else to say something that can not choose but be wrong or violates our specific intentions. Consequently, we shall have to strike the happy medium: we shall try to develop a propositional system that has a maximum of translatibility and a minimum of formal restrictions. We can find this system in a kind of 'logical language'. Of course the 'words' of this language shall be very abstract; they will not be defined sharply and will need to be understood with the aid of intuition. In order to construct this logical language, we shall have to take over notational forms and concepts belonging to traditional logic, without accepting the axiomatical system that goes with it. This structure underlies a system of semantical matrices, which represent techniques of analyzing the non differentiated basis of the meaningful intuition of reality. 4.2 Semantic

matrices

Some old painters used a sort of 'finder' to be able to represent their subjects as realistically as possible. This finder looked like a small square tennis racket, with threads forming a network of small squares. On his paper the painter had the same network, with the same proportions. He only needed to transfer square after square in order to receive a true copy of his subject. The use of such a finder allowed

202

Paul Van Geert

the painter to perceive the details of his subject in their right mutual relationships — the painter could structure his subjects. The semantic matrices highly resemble these old painters's finders. These also are simple networks that allow us to structure the incoming information; they reveal in a direct way the mutual relationships in a given perception. Nevertheless there remain some important differences: 1. The painter's finder only clarified connections that already existed in reality. The semantic matrix creates relations that do not a priori exist in a given perception. 2. The painter can only use one finder (the interference of another one would disturb his image of his subject), whereas a human being can intermingle different matrices. 3. The relations between the squares of the finder are spatial, e.g. square A is right up, square C is left down, and so on. The relations between the squares of the matrices are of a totally different nature: The squares mutually fulfill intuitive functions, dependent on the used matrix. These matrices are intuitive and non propositional: they are not explicitable without essential deformation. (By 'explicitable' we mean able to be stated explicitly, by means of an explicit statement or proposition.) Principally we can only denote them by means of an abstract notation, e.g. a letter. Because we suppose that the matrices are more or less non ambiguously related to general propositional structures, we can clarify the intuitive meaning of the matrices by showing how certain matrix structures are transformed into specific propositional structures. Let us first say in which way the matrices are related to the meaningful intuition of reality. The information that reaches us from the world outside exists only unconsciously. This information is structured by means of our basematrices, and thus structured, it becomes a meaningful intuition of reality. This MIR can be transformed via P-forming rules, into a series of propositions or into a series of quasi, merely internal, propositional fragments. The following hypothesis concerning the relation between language and semantic matrices are considered to be important. 1. The complete stock of semantic matrices is universal, in the sense that individual fluctuations in it are not culture bound nor race bound. 2. Language is a system to transform the MIR into an expression

Language and meaningful intuition of reality

203

(proposition) by which another individual (through inverse transformation) is able to construct a MIR, relatively equal to the speaker's MIR, so that they arrive at a kind of common consciousness. 3. Principally, there are no limits in combining the matrices, so that an infinite number of differently structured MIR's arises. Yet, there is no language that is capable of transforming all possible MIR structures in an equally elegant way. (Not only language plays the role of inhibitor. Also, there is a specific limit to what is cosmologically allowed.) 'Difficult' MIR structures may only be expressed by means of uncommon propositional structures. But, with the increase of propositional uncommonness, the semantic ambiguity increases. In other words its comprehensibility decreases. This explains the difficulty of communicating — of making common — MIR's which are linguistically difficult to structure, i.e. difficult to communicate in fact do not exist. 4. MIR structures that are not communicable in one language may be easily communicable in another language. So it looks as if language creates the intuition of reality, whereas the only thing language is capable of, is inhibiting or facilitating the communication of certain MIR's. Language does not have particular influence on the existence of an image of reality, only on its explicitability. 5. If language only illuminates the existence of certain MIR's by making them communicable, it may happen that the speech community indulges a process of repression. (This throws a different light on the validity of the Sapir-Whorf hypothesis: It states that the image of a certain speech community a 'stranger' may have is partly determined by the speech community's language structure. This image contains the speech community's way of life, its culture, its methods of facing reality, as is interpreted by the 'stranger'.) The community makes the propositional structure of communicable MIR's more stable, more elaborate, and supply it with an official or 'holy' status. A process of repression is evolved against the non- or difficult communicable MIR's. They are sometimes declared taboo and finally their conscious form of existence disappears. (Here it needs to be stated that repression of difficult MIR's is not the only way in which an official image of reality is stated. One of the global ways to do this is to choose a favorite idiom or medium of communication.) Language is adopted and elaborated to be able to express more and more adequately this official image of reality. The unofficial remainder of meaningful intuitions of reality may find its communication channels in art, dreams, or 'mental illness'.

204

Paul Van Geert

6. Principally, education is nothing else than a reduction of the total treasure of possible MIR's children may have. Education reduces it to the officially accepted ones and gives t h e m a p r e p o sitional superstructure. In this sense, education is n o t an enrichment, b u t an impoverishment. 4.3 Nature and function

of semantic

matrices

When a specific MIR becomes conscious, it is completely structured — a priori. T h e structural process we talk a b o u t is merely a transformational description of it. We shall t r y t o clarify the nature and f u n c t i o n of semantic matrices by analyzing t h e process of intuition structuring. 1. We start with a global perception 'P' (or perception, m e m o r y , representation, . . .). It is functionally equivalent t o Chomsky's n o t i o n of 'S'. 2. In P, we define a certain ' p o i n t of reference' labeled X of REF(X): Ρ -»· REF(X) 3. X leads t o the fixation of its c o m p l e m e n t : REF(X) REF(X)

COMP(Y) + COMP(Y) = Ρ

This c o m p l e m e n t is n o t a logical c o m p l e m e n t , but an intuitional one: Y is the c o m p l e m e n t of X, seen in the c o n t e x t of P. Principally, Ρ does have an infinite n u m b e r of possible 'attributes', b u t as a result of the way R E F and COMP mutually define each other, their n u m b e r is rigidly reduced. (On this level of analysis, the word ' a t t r i b u t e ' is t o o specific and restricted. ' A t t r i b u t e ' m a y stand f o r ' a t t r i b u t e ' , ' c o m p o n e n t ' , 'dimension', 'part', etc.) In other words, t h e 'attributes' t h a t thus define Ρ depend on the way in which Y is t h e c o m p l e m e n t of X. T h e interaction of REF and COMP might be illustrated by what we call 'background and f o r e g r o u n d ' . T h e choice of REF and COMP is an application of the rule DOM/DEP, which states t h a t in every meaningful intuition of reality, t h e r e is some d o m i n a n t (being the 'subject' of the intuition, 'the point of reference') on which a particular d e p e n d e n t depends. Viewed on a propositional level, the DOM is in fact the subject of the proposition, whereas the d e p e n d e n t essentially represents the reason w h y the proposition a b o u t the DOM (subject) has been formulated.

Language and meaningful intuition of reality

205

DOM/DEP seems to show a certain correlation with the notions NP-VP. However, NP-VP adds the value of syntactical category to the merely intuitive DEP/DOM relationship. Here it needs to be said that these DEP/DOM relationships can be reordered, if only it is done consistently. This interchangeability shows in fact that these intuitive relationships are not necessarily connected with fixed conceptual or perceptual features. The meaning of DOM, for example, depends on its connection with DEP and vice versa. (Here we need to warn against interpreting the matrices as if they would represent 'dialectical' connections. Additionally it needs to be said that the names we have given the matrices are considered to be the closest approximinations of the intuitive notions they represent. If other names would fit better, these need to be changed.) 4. Furthermore, there is the rule of 'approximation of complements', or, logically said, the rule of (intuitive) identity and (intuitive) difference ( I D / D I F ) that bears to the notion of 'boundary'. Namely, when two things are complimentary, they touch each other in a particular way, they have a sort of 'borderline' (boundary) through which the one element differs from the other. As to X, this point of contact is called X y ; as to Y, this point of contact is called Υχ. (Χ

, Υ χ ) - Β, or boundary between X and Y

Being written as a logical set, we have: 1. 2. 3 4. 5.

X=X Y=Y .ΧΦΥ+Χ ΥΦΧ •* (Χ Y x ) = Β, or boundary

Β represents the way in which (the place in which) differentiation between Xand Y can be compared (comparability of differentiation). If X differs from Y, this must be in a certain feature or characteristic ('place', 'point') Xy, and if Y differs (from X, it must be in Yx, and the set (X y , F x ) constitutes the boundary between X and Y. 5. We can now insert the following two matrices. (There certainly are more, but we did not clarify them till now.) a) CON/TRANS indicates that something can be seen as the transitional aspect. Something else as the coherent conservational or constant aspect of a meaningful intuition of reality I. This coherence

206

Paul Van Geert

between CON/TRANS presents a way in which this I is intuitively structured and explicitly stated. b) M-mod indicates that a fragment of a particular MIR is defined by means of two classes of aspects, of which one class assumes a relation of dependency on the other. This dependent subclass is called the modification {mod) of the dominant M. 6. Further application of the semantic matrices leads to a fully structured intuition on which can be based the generation of an expression that bears a proposition concerning the 'content' of this intuition. As we have already stated, not every intuitional structure automatically leads to a proposition or set of propositions because the specific structure of a certain language only allows the transformation of a limited set of all possible MIR's. Because this paper does not allow us to draw up an extensive description of the logical structure of the matrices and their applicational rules we shall try to clarify them giving some examples of the way in which a MIR structurally can be described. 4.4 An explicitation matrices

of the structuring

of MIR by means of

semantic

Up to now we have illustrated in what way the structure of intuition can be described. The reason for this paradoxical attempt to describe something what is a priori not describable is that we would like to answer the question about what connection there is between the intuition of reality and language. In other words, why does a subject in a certain situation say what he says. Why does someone say '. . . that's a boy that eats an apple . . .' when he sees a boy that eats an apple? As we have already stated — keeping in mind the specific situation in which the proposition is formed — the reason for this is twofold: 1) the structure of the propositions depends on the expressional attitudes of the subject (which we shall deal with later in this paper); 2) the structure of the propositions depends on the progressive structural explicitation of a specific intuition, so that it can be transformed into a specific application of the P-forming rules. We shall try to give some examples of this process of structural explicitation of intuition. First of all it needs to be said that we only deal with language as a representational act of symbolization. Because language forms a system on its own, it may appear as a mere

Language and meaningful intuition of reality

207

sign-creating system, forming propositions according to the laws of language and logic and not according to the structure of an intuition. First example We shall try to analyze an intuition we may describe thus: a subject has a MIR that is composed of something he might call 'John', busy doing something, and another thing 'apple', playing a decisive role towards the nature of the activity. We stress the fact that in this description we already have inserted mutual relationships that do not analogically exist in the intuition we try to describe. But as we have already said, this is unavoidable. The progressive structural explicitation of this intuition looks like this: explicitive form

used rule

explication by means of 'splitting up'

Dom/Dep

ID/DIF explicitation by means of 'joining together'

ID(X x )

DIF(Xy)

ID(YV)

DIF(Y X )

This joining together can be done in two ways: 1) by following the rule of 'approximation of complements': We already have mentioned this rule but we did not mention its useful counterpart (see 2). This rule is connected with the use of the Con-Trans matrix. According to this rule, the two DIF elements are joined in order to specify the relation between the dom/dep groups; 2) following the rule of 'complementarity of approximation'. This rule, used with the M-mod-matrix, joins the two ID elements in order to make explicit the relation between dep and dom. Let us first apply rule (1): This structure is the basematrix of the 'normal' sentence, e.g. 'John eats an Apple'. It is an easily transformable base matrix because it is highly similar to linguistic base structures as, for example, 'terminalmedial-terminal' or 'NP — (V-VP).' In the sentence 'John eats an apple' the linguistic dominant — i.e. the linguistic item that, for

208

Paul Van Geert

DIF(Xy)

ID(Xx)

ID

I

:

(DIF

1

-

DIF(Y X )

ID J

DIF)

I

Dom

ID(Y y )

Dep

specified with a con-trans matrix:

CON

-

( T R A N S - TRANS)

CON

-

(TRANS) Dom

-

CON

-

CON

Dep

example, determines the verbal conjugation — coincides with the intuitive dominant (Dom-Con [John]). In the sentence 'the apple has been eaten by John', the linguistic dominant does not coincide with the intuitive dominant. This causes a reduction of the sentence's semantic tension, which sometimes may be useful in relation to former or future propositions. Second

example

We shall try to analyze the intuition that finally may lead to the proposition, for example 'red apple' explication by means of 'splitting up'

REF(X) j^ ' ID(X x )

DIF(X y )

COMP(Y) ID(Yy)

DIF(Y X )

On this follows an explicitation by means of 'joining together' according to the rule 'complementarity of approximation' (in relation to M-mod). This is easily transformable into the linguistic form of modification or specification. The linguistic process of transformation of a Λί-mod group differs from the process of transformation of the Con-Trans

Language and meaningful

intuition

DIF(Xy)

of reality

209

( ID(X x ), ID(Y y ) )

DIF(Y X )

M(Xx,Yy)

mod(Y x )

application of of the M-mod matrix: mod(Xy) Dom

Dep

group. In the Con-Trans group the elements of the intuition 'remain' or have a substantial counterpart in the proposition, the relations being expressed by means of specific syntactic forms. However, the linguistic transformation of the M-mod group executes an elimination of the dep(mod) element, and supplies the Dom(mod) with a specific intonation or accent. This intuitional structure may lead to propositions as, for example, 'the red apple', 'an apple, a red one', and so on.

Third and fourth

examples

The final structural explicitation of examples 1 and 2 may be taken as final states directly transformed into a proposition. On the other hand, their explicitation may continue. The final state of the first example can be made more explicit by means of the matrix M-mod. In this kind of structure, sentences as 'that's John, who eats an apple. . .' may be generated, wherein 'who eats an apple' is a modification of specification of 'John'. The accent a mod group normally bears is substituted by a specific VP structure. The way an initial Ρ is made explicit is not a matter of accident, but it depends, for example, on the possibilities of conceptualization of the intuition, on its intuitive structure, on the specific relationships between REF and COMP, and so on. When the final state of the second example is extended or specified by means of Con-Trans a structure occurs on which sentences of the type 'John walks' may be based. These examples are merely introductory. Many problems of formal and theoretical nature still remain. (For further illustrations, see appendix B.)

210

Paul Van Geert

APPENDIX

A. Results of experiment

2 choosing

alternatives

Every pair of drawings was accompanied by seven or eight sentences that were grammatical or slight semantic variations of the first one of the series. For most sentences, only the first choice, the choice based on the first sentence of each series, was relevant. The reason for this is that the children did not understand the communicative sense of the second and following sentences of each series ('Why. does that man say always the same thing?')· The repitition of sentences describing one basic situation alienates them from their direct and intuitive meaning. Series of sentences that all lead to relevant choices are specific ones; we shall try to analyze in the second part of this appendix. Results stimulus sentece s:

number of choices:

S, 52 53 54

green:7 boy: 8 banana: 4 ( + 2 ? ) dog: 14

This The This The

is a green tree boys eats an apple is a red banana dog runs after the cat

tree: 7 apple: 6 (+1?) red: 8 cat: 0

Note on S 4 : All sentences show an almost unique choice of the grammatical subject, also for the passive structures. The reason for this is that children of this age do not 'understand' passives: they interpret them as if they were active constructions, having the same word order.

Ss 56 57 58

The The The The The

sugar is in the box boy drinks petrol bird flies to the cat cat is beneath the bird boy looks to the girl

sugar: 3 (+1?) boy: 7 bird: 9 (+1) bird: 1 boy: 9

box: 1(+1?) petrol: 6(+l?) cat: 1 cat: 0 (+2) girl: 0

Note on S 8 : All sentences of this series gave extremely significant choices (as for

S 4 )· S9 S 10 Sn

The butcher cuts the meat with a knife Mama knits a sweater The mouse gnaws at the cheese

butcher: 1 knife: 5 mama: 3 cheese: 2

meat: 3 sweater: 9 mouse: 8

Language and meaningful intuition of reality

211

Note on S n : This sentence illustrates that the more the predicate is a semantic inplication of the subject, the less the direct object is found to be an important part of the sentence (or the less the predicate is found to be so).

512 The boy draws a picture 5 1 3 The woodcutter cuts down a tree Sj4 The man is sitting in an airballoon The car rides through the tunnel

boy: 3

picture: 7

woodcut.: 6

tree: 4(+l)

man.· 4

balloon:6(+2?)

car: 4

tunnel: 6(+l?)

The experiment was initially executed with thirty subjects. The first subjects had a less standardized version of the experiment (nonstandardized intonation and accent), less situations, other drawings, and so on. Their results played a role in the formation of the hypotheses, but those results cannot be considered for inclusion in this text. B. An analysis of some results of experiment 2 in terms of semantic The analysis of these two stimulus sentences serves as an illustration S 10 Mama knits a sweater (Related sentence S 12 Theboy draws a picture The results were more significant when the following sentence was presented: The boy draws a house)

mama: 3

sweater: 9

The grammatical structure of this sentence is the same as the one of 'the boy eats an apple', although there is quite a different choice pattern. This difference can be described when analyzing the child's referential sentence and the presented stimulus sentence in terms of semantic matrix structure. Referential sentence:

Si 0

Mama knits is knitting

Stimulus sentence:

Sn

Mama knits a sweater

MIR structure of referential sentence:

212

Paul Van Geert

REF(K)

/

Dom

COMP(L)

Dep

\

/

\ DIF (CON - TRANS)

(mother) -

(knit) becomes

'mama (knits - Δ)'

I 'mama knits' (referential sentence) Ps

(M-mod) MOD

'sweater'

'a sweater'

Δ = non differentiated or conceptual class. P„ = the subject structured the meaning of the stimulus sentence. * a sweater' : complement of 'sweater' in its conceptual class.

In this sentence, 'sweater' played the role of a mod element, which is the psychologically dominant part of a M-mod structure. Sweater is the explicitation of a modificational class that is open (not explicit)

Language and meaningful

intuition

of

reality

REF

ID

COMP

DIF

MOD

213

ID

(M)

DIF

MOD

(M-mod)

(con-trans)

C 'running'

—^

A (cat & dog)

-'running 1

(cat & dog)

running

The latter is represented as follows: / REF

\ COMP

'dog'

'cat'

Dom — Dep The relation between 'dog' and 'cat' is made explicit by means of a spatial relationship 'after'. I I Dom (dog) Dep (cat) after

I

('dog' = f(after) 'cat')

'the dog runs after the cat'

4. The dog runs after the cat Δ (cat, dog) : conceptual class containing 'cat' and 'dog'

dog : 14 cat : 0

214

Paul Van Geert

in PR. The explicitation thus generates special attention, that is expressed in the choice pattern. To get this sentence, the Δ (cat, dog) in which all the concepts have an equal 'attention' value, must be transformed in a REF (dog)/ COMP (cat) structure, in which 'dog' becomes the dominant. This operation draws special attention to 'dog', which explains the choice of pattern. The analysis of this two stimulus sentences serves as an illustration of the way in which choice patterns may be understood by means of the semantic matrices. One should remark on the basic importance of the referential sentence.

PART THREE

Epistemological and Methodological Aspects of Relativism

HELMUT GIPPER

10

Is there a Linguistic Relativity Principle?

It was during the fifties that I, by chance, became acquainted with B. L. Whorf's Four Articles on metalinguistics (1952). I read these papers with growing interest, struck immediately by the resemblance of Whorf's leading ideas to those of Wilhelm von Humboldt whose influence on my work had been decisive from the beginning of my studies. At the same time, however, I found some of Whorf's assertions regarding the conception of space and time in the Hopi language and his generalized Standard Average European (SAE) puzzling. Whorf's thought provoking ideas were remarkable. But could they be regarded as valid? I felt it was my task to pursue some of these problems. Since that time I have been concerned with the verification of Whorf's theses, with his theoretical and methodological premises, with his notions concerning the Hopi language, and, last but not least, with his conception of the linguistic relativity principle. Isolated in Germany with no access to the Indian world or to authentic material, it was difficult for me to contribute to these problems. The only possibility was a careful study of Whorf's writings in order to discover whether his argument was in itself consistent and free of contradictions. I set about reanalyzing all of Whorf's data, comparing them with one another, and found that, at least in the form they had been presented, some of his statements were inconsistent, misleading, and scientifically scarcely tenable. These results were published in my book Bausteine zur Sprachinhaltsforschung. Neuere Sprachbetrachtung im Austausch mit Geistes- und Naturwissenschaft (1963). In Germany my preoccupation with Whorf was regarded with reserve and scepticism by my colleagues. My lack of competence as

218

Helmut Gipper

a non-Americanist was self-evident. I was, however, much encouraged by Prof. Charles F. Voegelin of Indiana University, one of the few experts in Hopi language, who wrote to me, remarking kindly, Ί think you have done a first class job in explicating Whorf. This was more than I had expected, and emboldened by this generous tribute, I determined to seize the fist opportunity to visit the Hopis, to study there more intimately the problem involved. In 1967 and 1969 I was invited as Visiting Professor to the University of New Hampshire, USA, and taking advantage of my visits I journeyed on each occasion to the Hopi reservation, on the second occasion aided by a grant from the Deutsche Forschungsgemeinschaft. At that time I was accompanied by my wife who helped me greatly in my investigations. Although forewarned by American experts about the difficulty of inquiries among the Hopi, I did obtain much valuable information. I was obliged to carry out my interviews in English and make use of the help of bilingual Hopis. I made tape recordings and obtained further material from the Bureau of Indian Affairs in Flagstaff, Arizona. On returning to Germany I analyzed my interviews, aided by one of my students, Mrs. Andrea Stahlschmidt. Before publishing the results, I had the opportunity of bringing to Bonn, with the benignant help of Ministerialrat Dr. Malotki, a Hopi informant, Mr. James Kootshongsi from Hotevilla, Hopiland. He read my material, checking it and making very useful corrections. I was thus able to publish tenable results in book Gibt es ein sprachliches Relativitätsprinzip? Untersuchungen zur Sapir-Whorf-Hypothese (1972). On the following pages I present a brief resume of these researches. However, the results I wish to communicate and the corrections to Whorf I propose do not represent the primary purpose of my book. Its purpose is to elucidate the important problem of the interrelations between thought and language, language and culture, and language and society. I have gathered together all published comments on Whorf and this central question. I have discussed and compared the most important contributions from the standpoint of Neo-Humboldtian linguistics, with which I am familiar. The most balanced analysis of the problem is that of the Polish philosopher Adam Schaff in his book Sprache und Erkenntnis (1964). Schaff has presented a measured, painstaking analysis of several of the points in discussion. As a declared Marxist he cham-

Is there a linguistic relativity

principle?

219

pions the well-known Widerspiegelungstheorie, presenting, however, a flexible interpretation. He rejects a too rigid interpretation of the reflection of reality in language, and admits that, besides the objective pole of given reality, there is also the subjective pole of cognition. The mind does not merely perceive the world through its senses but conceives it through the categories of thought fashioned predominantly by the mother tongue. This approach suggests a possible reconciliation between the materialistic viewpoint and Humboldt's more idealistic procedure, a dialectical synthesis of the two approaches. This is a problem which I have tried to resolve, and, from the standpoint of my book, my critique of Schaff represents a major theme of my argument. I have further tried to place the 'linguistic world view' thesis on more solid foundations by reviewing concurrent concepts. Without repeating every stage of the argument, it seems advisable to make the following distinctions, in order to avoid misunderstandings and provocation of fruitless discussions. To explore the fundamentals of the linguistic relativity principle, we must distinguish among three concepts, which are clearly delineated in the German philosophical tradition: (1) Weltbild, (2) Weltanschauung, and (3) sprachliche Weltansicht (Humboldt) or sprachliches Weltbild (Weisgerber). 1. By Weltbild is meant the scientific view of the interconnections of the world or cosmos, i.e. a cosmography. Thus we speak of a Ptolemean Weltbild, of a Galilean or a Einsteinian Weltbild. In this sense Weltbild is the product of scientific thought of an era. 2. Weltanschauung refers to the structure of human belief concerning the nature of the world and human affairs. It is an ideological concept with religious or political implications. We speak thus of a Christian, a Marxist, an idealistic, or materialistic Weltanschauung. 3. With the term sprachliche Weltansicht or sprachliches Weltbild we imply that our experience of the world is construed not only by our senses but also by the categories of the language we have learned. The vocabulary we use construes a specific view of things, of facts, and of events. Sprachliches Weltbild is an a priori condition making specific thought and speech possible. Following G. Frege we may say that sprachliches Weltbild has to do with the 'Art des Gegebenseins von Welt in den Kategorien und semantischen Gliederungen einer natürlichen Sprache' (the nature of the representation of the world in the categories and semantic structures of a natural language). The concept is devoid of ideological speculation; it points to the condition of human existence.

220

Helmut Gipper

There are, of course, relations between (scientific) Weltbild, (ideological) Weltanschauung, and (linguistic) Weltansicht or Weltbild. The first two are impossible without the mediation of the last concept. They are founded and expressed by means of a particular language and hence are colored by its semantic categories and structures. We must, therefore, at the same time keep in mind that different, even contradictory, views of the world may be expressed in the same language. A competent speaker of German or English may be an adherent of different Weltbilder and Weltanschauungen: He may become a partisan of Kopernicus or Einstein, of Christ or Marx. Nevertheless, whatever he may think or believe, he owes to the means his language places at his disposal. This undeniable fact must be accounted for in our discussion of the linguistic relativity principle. If we hold these distinctions in mind and if we take care not to loose sight of them in the translation of relevant contexts from one language to another we shall avoid tedious, unproductive discussions. We might add that by comparison with other languages and discovering the relationships between words and things we can gain insight into the specific world view of our own language. We are not hopelessly bound by the limits of our mother tongue. I shall now turn to the questions concerning Hopi to offer some provisional answers.

THE CONCEPTION OF TIME A N D SPACE IN HOPI SOME NECESSARY CORRECTIONS TO THE VIEWS OF B. L. WHORF

An exhaustive analysis of Whorf's own controversial papers on the conceptions of time and space in Hopi — including his grammatical sketch of the Hopi language, Toreva dialect — reveals that such provoking statements of his, as 'The Hopi language contains no reference to "time", either explicit or implicit' (Whorf 1950: 57), are misleading, and even false. The Hopi language does indeed contain a considerable number of expressions referring to space and time. This statement can partly be supported by Whorf's own material (Gipper 1963: 297—366). Since Whorf fails to offer a synopsis of the relevant expressions in Hopi, it is impossible to verify his interpretations of the specific Hopi world view. After a careful study of the linguistic data during the last few years I am able to support some of Whorf's theses but I must at the same time refute some of his particular statements. I propose the following corrections to the conception of

Is there a linguistic relativity

principle?

221

time and space in the Hopi language, using, in addition to my own observations, material from the following publications (listed chronologically): 1936 Alexander M. Stephen, Hopi journal. 1944 E. A. Kennard and A. Yava, Field mouse goes to war [Tusan homichi tuwvöta]. Coyota tales. NASEC. 1957 C. F. and F. M. Voegelin, Hopi domains. 1960 C. F. and F. M. Voegelin, 'Selection in Hopi ethics', Linguistics and translation. 1963 E. A. Kennard, Linguistic acculturation in Hopi. 1967 Robert A. Black, Hopi grievance chants. A mechanism of social control. 1967 C. F. and F. M. Voegelin, Passive transformation from nontransitive bases in Hopi. 1. A linguistic analysis of the Hopi expressions in question shows that they can be interpreted by Indo-European grammatical categories. There are nouns, adjectives, verbs, and adverbs or particles similar to adverbs (Whorf's locators and temporals). The Hopi language even shows a striking tendency to verbalize the nonverbal categories. Verbalization is effected by adding special suffixes to nouns, adjectives, etc. 2. There is evidence for time intervals which can be said to belong, contrary to Whorf's view, t o the category of nouns, and at least some of these nouns can be pluralized. Pluralization is o f t e n made by reduplication of the first morpheme of the word in question. 3. These nouns for time intervals can also occur in a grammatical or syntactic function which corresponds to that of a*subject (in the nominative) in Indo-European sentences. This Whorf denied. 4. Contrary to Whorf's opinion, expressions for spatial relations are used metaphorically in a temporal sense, in other words, there are space-time metaphors as in Indo-European languages. 5. Whorf's statement, that in Hopi time intervals cannot be counted in the same manner as are material objects but are used in the singular combined with the ordinal number, is questionable. The problem seems to be a question of interpretation. In order to find a satisfactory answer to this question we have to consider first the numeral system in Hopi. There are simple numerals up to ten on First and Second Mesa, and up to twenty on Third Mesa. Higher numbers can be formed by combination (multiplicatives) with tens and twenties, at least theoretically, but this method seems to have

222

Helmut

Gipper

been largely replaced nowadays by the English counting system. So on Third Mesa the number seventy-five may be expressed by saying three times twenty (plus) ten (plus) five, but certainly it is not usual t o do so. Besides these cardinal numbers there is another set of numerals with the element — sikis (respectively — is, -s) which means literally '-times', as in English four times. Thus we have expressions corresponding to English once, twice, three times, etc. This meaning is confirmed by sentences as 'The chief smoked four times', etc. The most important intervals in Hopi life are years, months, and days. (I have not found expressions for minor intervals corresponding to English hour, minute, second.) Time intervals, above all days, are counted, especially in connection with the ceremonies. The most important intervals are four, six and twenty days. To count these intervals the second system is often used, e.g. the numeral with the -sikis element plus the (noun) interval in the singular. In relevant publications we often find in this connection the English translation 'first, second, third (day)', etc., that is the English ordinal number with the noun in the singular. Is this translation justified? And is it correct to conclude, as Whorf does, that the intervals, strictly speaking, are not counted at all? Yes and no. In fact, the meaning of the two translations 'fourtimesday' and 'the fourth day' may be nearly identical in many cases. Nevetheless it is hypothetical to interpret, as Whorf does, that time intervals are not counted: It is the same daylight which returns, only a little older. My informants did« not even understand this explanation. Some of them gave me combinations of a cardinal numeral plus noun in the singular, e.g. civot tala 'five day(s)'. And we also find examples, where the -sikis form is combined with the plural, e.g. paish'tala toto'kya — 'third day sleeps' (Hopi Journal 1936: 1273). It is worth mentioning in this connection that in German we have also cases in which weights and measures are counted with cardinal numbers plus singular. So we have fünf Pfund — 'five pound', fünf Fuß — 'five f o o t ' {fünf Füße meaning five different members of the body). Nobody would say that these weights and measures are not regularly counted. I think we can assert that time intervals are counted even if the singular be used. Another possible way of counting time intervals, which seems to be the most usual one, is to use the cardinal number combined with the element -tok which originally meant 'night' (toki), the final vowel

Is there a linguistic relativity

principle?

223

being lost. So we have lötog — 'two-night', paytok — 'three-night', etc. It is possible that an original reduplication of the morpheme meaning 'night' has been lost in these patterns. The element -tok seems to have been reduced to a mere grammatical component. We know that the counting of nights is important in Hopi ceremonies. Nevertheless none of my informants was aware of the fact that there is the element 'night' in these constructions. The old meaning of the element has been completely lost. Last but not least there is — as my informant Mr. Kootshongsi assured me — even the possibility of using the cardinal number for one plus singular, two plus dual, and three plus plural, etc., e.g. stika mtiyaw(u) — 'one month', löyöm müyawt — 'two months', päyom mümiiyawt — 'three months' (with reduplication!). 6. There are time expressions for the different phases of a day from dawn to dusk and there are also expressions corresponding to today, yesterday, tomorrow, etc. 7. There are grammatical means in Hopi to express present, past, and future, though the Hopi thinking seems to be governed by a bipartition of time ('present + past' and 'future') instead of the tripartition of time ('past' — 'present' — 'future') we are accustomed to. In the verb there are forms for actual events which can also be used for past events, often combined with special particles denoting past. Thus we often find in stories and tales the particle yaw (Whorf's 'quotative modality') which is generally translated by 'it is said'. This translation is completely misleading since yaw is not a form of a verb 'to say' but a particle having the approximate meaning of 'then', 'after that'. If this particle is used we know that the reported event is in the past. Even when direct speech is used in a story in which a person is speaking of something to happen in the future the particle yaw is used. It reminds the hearer or the reader that he is still listening, to a story related in the past. (Other particles, which are able 'to make events more past', as some of my informants put it, are pai, ep.) 'Future' in the broadest sense — including everything which has not yet happened, which is desired, wished, wanted, planned, etc. — is expressed by the suffix -ni which can be combined with different word classes. There are still other particles to place an event in the present, past, or future. This corresponds at least partly with Whorf's dispersed data, although he denies a temporal interpretation of these linguistic forms.

224

Helmut

Gipper

8. The Hopi verb is a most interesting and complex phenomenon containing many temporal elements. There are several 'aspects' (I would prefer the German term Aktionsarten in order to avoid confusion with the well-known aspects of the Slavonic languages) to express durative, progressional, continuative, ingressive, and iterative events. All of them refer directly or indirectly to time, a fact Whorf failed to acknowledge. 9. There are also interesting possibilities of expressing, by the means of certain suffixes, the time relation between two and more events in complex sentences with different clauses. Since Mrs. Stahlschmidt is preparing a dissertation containing a complete analysis of these complex details, I omit the problem here. 10. There seems to exist — or at least to have existed — a special word for 'time', which, however, Whorf — and my informants — denied. The word is sbato — 'time' (Hopi Journal 1936: 1286), which occurs in the expression nono'bshato — 'food time'. Whorf himself mentions sa 'to, n. (noun) (Hopi Journal 1936: 1288). Mrs. Stahlschmidt suggests that this element is contained in hi(n)sato, that is, in combination with the indefinite or interrogative pronoun. The word bisato is common in Hopi. 11. I was not able to verify Whorf's statement that the Hopi world view is dominated by the distinction of 'two aspects of one reality', the 'objective or manifested' and the 'subjective or manifesting (or unmanifested)' (Whorf 1939) or that this dualism is expressed by use of special linguistic forms. We may thus conclude that although Whorf's exaggerated statements require some correction, there remains evidence for the fact that Hopitime is, in various aspects, different from ours. As to the conception of space, the difference between Hopi views and ours is far less evident. There are many expressions for spatial relations and for dimensions of things. They are in part comparable to the spatial prepositions and adverbs in Indo-European languages. In Hopi grammar cases of the locator type (Whorf 1946: 166), Locative, Allative, Illative, and Ablative, are important in connection with the use of spatial expressions. Some of the spatial expressions such as 'on top o f and 'down-below' are frequently used because the geographical location of the mesas requires such possibilities. Placenames can be used in the nominative. Whorf also denied this. The four cardinal directions are very important. We have to add to these the two vertical directions (up to the sky and down to the hadir), These six dimensions are often referred to during ceremonies.

Is there a linguistic relativity

principle?

225

Whorf seems to be right in stating that there are no special words for the different rooms of a Hopi house (such as kitchen, bedroom, etc.)· But the word kiva is certainly the expression for the underground ceremonial chamber. To date it has not been possible for me to verify Whorf s statement concerning the operational character of time-space expressions (an event which has happened far away is said to have happened in the past since the report of it can only come when the event is past). Following on all these corrections, what remains of Whorf's statements? Can we assert that the Hopi conception of time differs radically from that of SAE (Standard Average European)? Our analysis seems to imply a negative answer, but in reality the problem is much more complicated. First, we must not forget that fundamental conceptions of human life such as space and time are not exclusively bound to language but are dependent — as language itself — on the whole 'context of culture' of the given society. Language can not express all the details; language is always abstractive and the meaning of words has to be supported by the thing meant. We must take into account the whole of Hopi life, the beliefs of the people as well as their remarkable geographical location. Let us, therefore, consider the Hopi cycle of the ceremonial year, the Hopi road of life, the world view of the people reflected in the events during the year and during the life of the individual. Here we find evidence for the predominance of a cyclic conception of time combined with a dual conception of human life on earth and the 'life of the dead' in the underworld. In Hopi thinking this, cyclic principle is characteristic. We may compare the cycle of Hopi life to a wheel which turns eternally on one spot without progressing. As the Hopi have no calendar, years are not counted continuously. What people experience seems to be the steady repetition of the same sequence of events, of seasons, of seedtime, harvest, winter, spring, and so on. They live in time, but not apart from it, they are bound up in time but are not neutral observers of objective physical time. They do not live by the clock as we do. The course of the sun is the measure of Hopi time. The point of sunrise is observed precisely and constantly. The 'Crier Chief announces what has to be done when the sun has reached certain points. The names of the months reveal that the cycle on the surface of the earth finds its repetition in the underworld. The Hopi are still completely integrated in their environment.

226

Helmut

Gipper

They have not yet reached that detachment from events which is p n e of the outstanding prerequisites for our Western concept of physical time. Their conception of time corresponds to that of old peasant societies. I should like to add here a personal experience. When I first entered the house of Chief Joe (more precisely of his sister Elsie) in Shipaulovi (Second Mesa), I was much surprised to see an old alarmclock. But I soon discovered that this alarmclock was not going and that it was not needed. I was told that this clock had been used t o have the school children ready for the schoolbus which took them to the American Day School. Watches and clocks are even today no real ingredients of Hopi life. There is, a s E.T. Hall put it, a 'silent language' of time. This fact became evident to me on the occasion of the famous snakedances. Nobody could tell me at what time these dances would take place. Time must be 'ripe', they would say. Before these dances begin, people wait for hours until all preparations in the kivas are completed. Time is important on these occasions, but it is not our time; it is rather the duration of certain ritual events relevant to Hopi life. Nevertheless we are able to understand what happens. There are, as already mentioned above, traits in the Hopi thinking and behavior which are typical for all old peasant cultures, and so we can find even today in Europe similar phenomena. All people, who live a simple peasant life, who depend on the sun and the rain as the Hopis do, are supposed to have a more cyclic time experience rather than a linear concept of time, which is characteristic for members of modern Western civilizations. An additional remark may be useful. In a 'primitive' or original culture like that of the Hopis a scientific Weltbild has not yet been developed. The Hopi cosmography — if there is any — closely linked with the Hopi Weltanschauung, that is their religious beliefs. Hopi thinking has not yet reached a critical distance towards language. The confidence in the 'truth values', so to speak, of the mother tongue is therefore greater than in our societies. Thus the Hopi language can be said to be an authentic key to the understanding of the Hopi Weltanschauung. In other words, the Sprachliche Weltansicht of the Hopi language is a more adequate expression of Hopi thought than this is nowadays the case with the Indo-European languages. Therefore a careful study of the Hopi language and that of the other Indian languages is important for the problem of the linguistic relativity principle.

Is there a linguistic relativity

principle?

227

For further details I refer to my book Gibt es ein sprachliches Relativitätsprinzip?, which contains more Hopi material including special word lists for the expressions of time and space. The interpretation I had to offer takes account of the whole cultural context of Hopi life including language. Language is understood as an obligatory medium of thought which has been formed by the speaking community during the centuries and which, as a soziales Objektgebilde in the sense of the German sociologist Alfred Vierkandt, shapes and influences the behavior of the speaker. Since this interrelationship is dialectical in nature, linguistic research cannot be undertaken in isolation from the speakers and the world they refer to. Human thought is, to be sure, relative to the possibilities of the languages in which it is expressed, but it is not determined by language. Each natural language represents an open system and therefore it is open to be changed by the speakers. There are universal traits in each natural language because they all depend on general conditions of human existence and human life. But the universals are either fundamental, e.g. they refer to existentials of human life, or are situated on a high level of abstraction attainable only by scientific reflection. In, however, the wide middle zone in which human life and behavior takes place, we find differences which characterize cultures and civilizations. It is highly important and relevant to discover these differences. Therefore we have to investigate the linguistic world views of the given languages in order to find a key to better understanding among the peoples of this world.

REFERENCES Gipper, Helmut (1963), Bausteine zur Sprachinhaltsforschung. Neuere Sprachbetrachtung im Austausch mit Geistes- und Naturwissenschaft. Düsseldorf, Schwann. 2nd ed.: 1969. — (1972), Gibt es ein sprachliches Relativitätsprinzip? Untersuchungen zur Sapir-Whorf-Hypothese. Frankfurt a.M., S. Fischer (Series 'Conditio humana'). Contains a bibliography of all relevant publications. Hall, Edward T. (1959), The silent language. Greenwich, Conn., Fawcett Publications. 9th ed.: 1967. Schaff, Adam (1964), Sprache und Erkenntnis. Wien — Frankfurt — Zürich, Europa Verlag.

228

Helmut

Gipper

Whorf, Benjamin Lee (1939), 'The relation of habitual thought and behavior to language', in Whorf (1956). — (1946), 'The Hopi language; Toreva dialect', in H. Hoijer (ed.), Linguistic structures of native America. New York, Viking Press. — (1950), 'An American-Indian model of the universe', in Whorf (1956). — (1952), Four articles on metalinguistics. Washington. — (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J. B. Carroll. Cambridge, Mass., MIT Press. See also the list of publications in the text, on p. 221.

ETIENNE VERMEERSCH

11

Epistemologie et Hypotheses de Whorf

Dans cet expose je voudrais proposer quelques prises de position epistemologiques et introduire quelques notions qui me semblent utiles dans la discussion des themes de ce colloque. Ma conception philosophique generale est apparentee a celle de FEmpirisme Logique sans que je puisse pour autant m'associer completement a Tun ou l'autre des philosophes de cette tendance. D'autre part je suis convaincu que le discours philosophique — et scientifique en general — peut etre enrichi d'une fa£on considerable si on tient compte de points de vue empruntes ä la Cybernetique ou a une theorie generale des systemes. Je diviserai mon expose en trois parties; d'abord je rappellerai brievement quelques theses generates suggerees par l'actuelle philosophic de la science qui tendent a eliminer un realisme naif et un relativisme absolu en ce qui concerne notre connaissance de l'univers. Ensuite, je tächerai d'esquisser une minitheorie sur la fa$on dont l'individu construit sa vision du monde et pour finir j'appliquerai ces conceptions aux questions qui se posent par rapport aux hypotheses Sapir-Whorf. I. Dans une certaine conception naive on se represente les sciences naturelles composees, d'une part d'un ensemble de concepts, de lois et de theories, — c'est-ä-dire d'un cadre de reference — et d'autre part d'un ensemble de donnees, de faits, fournis par l'observation et l'experimentation. On s'imagine alors que le cadre de reference constitue une sorte de schema qui classifie et organise l'ensemble des faits bruts. Le realisme naif va meme aussi loin de penser que cette organisation existe deja dans l'ensemble des faits et que c'est la täche de la science de la trouver et de l'expliciter. En realite la structure de la science est beaucoup plus complexe.

230

Etienne

Vermeersch

On sait maintenant q u ' u n grand n o m b r e des soi-disant faits ne peuvent etre consus que grace ä l'existence prelable d'une theorie. Par example, pour pouvoir considerer la temperature d'un gaz comme un fait d'observation, il faut supposer une theorie concernant le fonctionnement de l'instrument de mesure, le thermometre. Ce n'est q u ' a partir du m o m e n t oil on accepte cette theorie comme valable q u ' o n peut tenir la temperature donnee p o u r un fait et l'utiliser pour verifier ou falsifier une autre theorie. Cette interdependance des faits et des theories existe non seulement au niveau de la verification, mais egalement a celui de l'identification. Par example l'identification des unites de temps telles que la minute, la seconde, n'est possible que si on dispose d'une horloge; mais pour interpreter le fonctionnement de celle-ci on a besoin d'une theorie de la pendule, ou, en fin de compte d'une theorie de la rotation du globe terrestre. Si, dan ce m e m e ordre d'idees, on tient compte de la these Popperienne tres plausible, a savoir q u ' u n e verification definitive et absolue d'une theorie ne peut etre atteinte, il y a lieu de conclure que m e m e dans les sciences naturelles on doit accepter une certaine relativite, voire une relativite certaine. Pourtant, nul ne peut nier qu'un grand nombre de ces theories nous inspirent beaucoup de confiance et sont de ce fait considerees comme 'objectives'. Cette impression de certitude, de fiabilite, est t o u t a fait justifiee: ces sciences nous permenttent une adaptation t o u j ours meilleure au monde qui nous entoure et les actions que nous entreprenons en nous fiant ä elles re$oivent generalement un 'renforcement'. On peut done conclure qu'un absolutisme concernant la connaissance scientifique est exclu, mais qu'un relativisme total est egalement contraire au bon sens; il n'y a pas un cadre de reference qui s'impose, mais cela n'implique pas que tous les cadres de reference se valent. Comme Μ Gipper l'a justement souligne dans l'introduction de son livre sur l'hypothese Sapir-Whorf (1972), un principe de relativite n'implique pas forcement un relativisme complet. L'erreur de la conception naive de la science ne consiste pas dans une tres large croyance a ses resultats mais dans une analyse inadequate de la situation, analyse qui tend a accentuer l'importance des donnees empiriques et a reduire le role tout aussi indispensable des concepts et des theories crees par l'homme de science et par la c o m m u n a u t e scientifique. Pour nous rapprocher deja un peu plus des theses de Whorf, je tiens a rappeler que des schemas conceptuels, des paradigmes comme Kuhn les appelle, exercent une influence indeniable sur la pensee des

Epistemologie et hypotheses de Whorf

231

chercheurs. On peut prouver historiquement que bon nombre d'hommes de science ont du surmonter des inhibitions enormes avant d'etre capables de suivre des raisonnements revolutionnaires et meme d'observer des donnees qui ne trouvaient pas de place dans leur paradigme. II est done plausible a priori que la structure grammaticale et lexicologique des langues naturelles puisse egalement avoir une influence analogue. D'autre part on remarque que les schemas existants n'ont jamais empeche radicalement la decouverte de faits nouveaux et la construction de paradigmes novateurs. Cela pourrait etre attribue a un mecanisme d'adaptation propre au systeme que constitue l'ensemble des chercheurs; mais il est peu probable que des cultures meme tres primitives soient denuees de mecanismes d'adaptation. Au niveau des sciences sociales l'importance de cette relativite devient encore plus manifeste. Le degre de certitude et d'utilite pratique de ces sciences est generalement moins grand: sur beaucoup de problemes il y a des divergences d'idees et du point de vue theorique et methodologique il existe souvent plusieurs ecoles. La situation se complique du fait que dans beaucoup de domaines les buts de la recherche ne sont pas seulement determines par la curiosite scientifique, mais aussi par des valeurs humaines individuelles, sociales ou politiques. Partout oil cela est le cas, l'apport subjectif de l'homme de science dans la construction des theories et dans l'elaboration des concepts sera beaucoup plus marque que dans les sciences naturelles. Aussi longtemps qu'il n'y a pas une theorie qui l'emporte sur toutes les autres, un relativisme au sens propre du terme ne peut etre evite. Ii y a pourtant deux restrictions: (1) ce relativisme n'est pas forcement definitif car il est possible que nous parvenions un jour a des solutions convaincantes; (2) ceux qui travaillent dans un paradigme donne peuvent etre influences par les theories des autres. La rigidite des cadres de reference est neanmoins plus grande que dans les sciences naturelles parce que les faits qui pourraient falsifier une theorie sont plus contestables et surtout parce que les schemas conceptuels ne dependent pas seulement d'un systeme purement 'theorique' (speculatif), mais peuvent avoir des relations tres intimes avec des valeurs et des prises de position ideologiques. Le degre d'engagement, de commitment, vis-a-vis de ces idees-la est en regle generale plus grand que vis-a-vis de theses purement scientifiques; le conflit interne, la cognitive dissonance dans le langage de Festinger, sera d'autant plus important. Par analogie on peut supposer que l'influence des categories de la

232

Etienne

Vermeerscb

langue egalement sera d'autant plus grande qu'il s'agit d'idees liees aux jugements ae valeur. C'est done a juste titre que M. Gipper met l'accent sur la difference importante entre une conception du monde, une Weltanschauung oil les aspects ideologiques sont importants, et l'image du monde, le Weltbild, oü il s'agit plutot d'une structuration des donnees empiriques. Les deux peuvent subir des influences de la langue mais les mecanismes qui pourraient les changer sont de nature differente. Il semble d'ailleurs que les differences de Weltanschauung entre diverses cultures sont plus importantes et qu'un vrai relativisme se defend mieux dans ce cas-ci que dans celui des Weltbild. L'accent que j'ai mis jusqu'ici sur l'interdependance inevitable des donnees et du cadre de reference dans les sciences naturelles et sociales, reste valable pour toute forme de pensee et d'acquisition de connaissance, et presente done un interet epistemologique general. II. Tout comme un grand nombre d'organismes vivants et comme certaines machines, l'homme est compose de deux systemes qui sont en interaction constante. II y a ce que j'appelle un EM-systeme qui Γεςοϊΐ et transforme de l'energie et de la matiere et un systeme informationnel ou I-systeme qui est capable de recevoir, de transformer, de conserver et d'emettre l'information. Chez l'homme cet I-systeme correspond plus ou moins au systeme nerveux. Quand on etudie de plus pres les I-systemes on constate que les donnees qu'ils re^oivent se presentent en premiere analyse sous une forme materielle ou energetique: des ondes lumineuses, des ondes sonores, des objets qu'on peut täter, etc. II est pourtant evident que ces inputs materiels ou energetiques ne peuvent en tant que tels avoir aucun interet pour ces systemes. Hormis les plantes, les organismes vivants n ' o n t aucun besoin biologique de lumiere. Si l'input lumineux est neanmoins d'une importance capitale pour la plupart d'entre eux, cela est du a leur capacite de reagir d'une fa£on specifique sur des etats differents de ce flux energetique. Dans les milieux sans lumiere, les memes fonctions sont accomplies par d'autres flux energetiques (tels les ondes sonores) qui presentent des possibilites analogues de detection et de reaction specifiques. II en resulte que la caracteristique essentielle des I-systemes ne reside pas dans la captation d'une forme d'energie particuliere, mais plutot dans la capacite de detecter differents ensembles d'un sub strat energetique (ou materiel) quelconque et d'y associer une reaction specifique. (Un chauffeur, par exemples, reagit de la meme fa^on chaque fois

Epistemologie

et hypotheses

de Whorf

233

qu'il se trouve devant un feu rouge, et d'une autre fa^on devant un feu vert.) Les donnees qui sont reconnues par un I-systeme dans le chaos d'influences energetiques et materielles qui accablent ses sens, je les appelle des formes. Chez les systemes les plus interessante, les organismes vivants et les machines complexes, la reconnaissance des formes n'est possible que si l'l-systeme dispose de procedes d'analyse de l'input. Meme un cerveau electronique tres puissant auquel on aurait branche un micro ou une camera de television serait incapable de reagir d'une fa^on significative s'il ne possedait pas des programmes de pattern recognition. On sait que ces programmes sont d'une etonnante complexite meme quand il s'agit de la reconnaissance de configurations tres simples. Les formes ne sont done pas donnees en tant que telles, mais sont le resultat d'operations de la part de l'l-systeme sur le flux energetique au materiel qui se presente a l'entree. Quels que soient, dans chaque cas particulier, les procedes utilises il en resulte toujours deux performances fondamentales: (1) un certain nombre d'etats du substrat sont identifies entr'eux, e'est-a-dire, qu'ils provoquent une reaction identique, interne ou externe, chez l'l-systeme, et (2) l'ensemble ainsi constitue est discerne clairement d'autres ensembles d'etats (qui causent une reaction differente). La definition de base est done celle-ci: une forme est un ensemble d'etats d'un substrat materiel ou energetique qui sont identifies entre eux et distingues d'autres ensembles d'etats. Cette identification et cette discrimination sont effectuees par un I-systeme quand il realise une correspondance bi-univoque entre ces ensembles d'etats du substrat, et des outputs ou etats exterieurs du systeme. On remarquera que ce concept de forme a une extension beaucoup plus large que le concept ordinaire; il ne comprend pas seulement des configurations ou Gestalt, mais egalement des couleurs et des sons, pour autant qu'ils soient identifies, des phonemes, des lettres ecrites,· des objets, et egalement des combinaisons ou des ensembles de formes, quand ceux-ci sont identifies comme tels. II m'est impossible d'exposer ici d'une faijon detaillee les proprietes et les diverses applications possibles de cette notion de forme (pour plus de details, voir Vermeersch 1965 et 1967); j'ajoute seulement deux remarques importantes. (1) Les I-systemes complexes, tels que l'homme, peuvent changer leurs criteres de discernement et il peut done y avoir differents niveaux de formes. On peut, par exemple analyser un texte ecrit au niveau des formes elementaires, des lettres, puis au niveau des mots, des phrases entieres etc. (2) II faut se rendre compte que,

234

Etienne

Vermeersch

quand on parle ici de discrimination, il s'agit bien de ce qu'on appelle en psychologie la discrimination absolue, c'est-a-dire celle qui donne lieu a une identification, une reconnaissance de la forme, en dehors d'elements de comparaison. Celle-ci est a distinguer de la discrimination relative ou comparative qui est le resultat d'une comparaison entre deux donnees presentees en meme temps. On sait que la discrimination comparative nous permet de faire des milliers de distinctions dans une dimension visuelle ou auditive, tandis qu'avec la discrimination absolue on peut identifier une douzaine de formes dans un continuum donne. A part les daltoniens tout le monde peut distinguer des centaines de couleurs differentes quand on les voit dans un album comme celui de Munsell, mais quand il s'agit de reconnaitre d'une fa£on certaine une couleur particuliere, peu de gens seront capables d'en identifier une vingtaine. (Pour un aper^u et une evaluation des resultats experimentaux de Pollack, Miller, et autre;, en ce qui concerne la discrimination absolue et 1'importance de la multidimensionalite des stimuli, voir, par exemple, Garner 1962.) Cette remarque est d'autant plus interessante qu'on peut supposer, pour des raisons neurologiques, que les hommes ont dans une large mesure les memes capacites de discrimination comparative, tandis qu'on trouve des ecarts considerables en ce qui concerne la discrimination absolue. Je n'ai qu'a rappeler une fois de plus, l'exemple des quantites de 'formes' de neige identifiees par les Esquimaux. Peut-etre n'est-il pas superflu de remarquer que chaque fois qu'on refere a des influences de la langue sur la faculte discriminatoire, il s'agit toujours de cette discrimination absolue! Quand on applique maintenant ces notions au probleme de l'acquisition de la connaissance par l'homme, il est etonnant de constater (a) que cet I-systeme identifie dans le monde qui l'entoure un nombre incroyable de formes, et (b) que cette reconnaissance de formes n'est pas due a un programme de pattern recognition inne puisqu'on constate de grandes differences entre les individus en fonction de leur formation particuliere, et chez le meme individu une etonnante plasticite a changer de criteres.(Je suis incapable pour le moment de distinguer les phonemes arabes, mais je sais tres bien qu'apres quelques semaines ou quelques mois d'exercice cela ne me poserait plus aucun probleme.) Meme si on imagine, ce qui est parfaitement plausible, que notre representation interne du monde peut etre con£ue comme un ensemble de formes internes, encodees dans des circuits neurologiques et presentant une isomorphie avec l'ensemble des formes du

Epistemologie

et hypotheses

de Whorf

235

monde exterieur, il n'en resulte pas un realisme naif. En effet, les formes exterieures et interieures ne peuvent etre que le resultat d'operations sur les inputs materiels et energetiques et ne peuvent done exister en tant que formes qu'a partir de l'interaction entre des donnees externes et des programmes internes. Ici, comme tout a l'heure quand nous parlions des cadres de reference des sciences, on serait tente de conclure a un relativisme cognitif. II faut pourtant se rendre compte que si un I-systeme introduit des identifications dans un chaos de stimuli, il ne pourrait faire cela d'une fa£on systematique si les donnees ne presentaient pas quelque ressemblance entre elles et quelque difference avec d'autres donnees; il y a done toujours un fundamentum in re. En plus, les operations d'identification introduites chez un individu particulier, ou dans une culture ne pourraient pas subsister si elles n'avaient pas une certaine survival value, e'est-a-dire, si eile n'aidaient pas le systeme dans son adaptation au milieu. Finalement, il est clair que s'il peut y avoir un grand nombre de phenomenes oil l'identification est tres arbitraire (par exemple, les constellations des etoiles, comme Orion, La Grande Ourse, etc.) il y en a d'autres qui s'imposent d'une facon imperative, ou bien pour des raisons fonctionnelles (par exemple des 'formes' de danger ou de nourriture), ou bien pour des raisons intrinseques par exemples lorsqu'elles representent un caractere d'isomorphie ou d'invariance si elles sont soumises a des transformations diverses (tels que des objets a forme geometrique simple qu'on peut soumettre a des translations, des rotations, etc., ou des objets commes les animaux d'une meme espece qui frappent par leur etonnante isomorphic). III. On peut se demander qu'elle est la portee de ces reflexions epistemologiques pour les hypotheses Sapir-Whorf. Vu les limites de temps, je dois me borner ä proposer quelques suggestions, sans en donner aucune preuve. Acceptons d'abord le postulat suivant. L'I-systeme de l'homme est organise de telle fa^on qu'il permet au nouveau-ne de s'adapter ä des milieux les plus divers et d'acquerir une plasticite dans le choix de methodes de manipulation d'information. On peut supposer d'autre part qu'il y a un principe d'economie dans l'organisation et dans le mode de structuration du cerveau, en ce sens que le reservoir des formes cherchera un optimum entre la vitesse d'acces et le nombre de formes conservees. Suivant le critere de plasticite, le degre d'organisation innee sera minimal. II en resulte le principe methodologique

236

Etienne

Vermeersch

suivant: des qu'une forme, une structure ou une operation peut etre construite a partir de formes ou de procedes plus simples — et dont on sait par ailleurs qu'elles sont dejä presentes — on devra preferer l'hypothese de la construction a celle de l'inneite. En ce qui concerne le probleme des universals, il y en a certainement quelques-uns d'ordre neurologique: il existe pour les differents sens un repertoire de stimuli possibles qui correspond au nombre de discriminations comparatives; ensuite il y a quelques limites universelles imposees entre autres par le memory span, le temps de reaction, les capacites de discrimination absolue dans une dimension donnee et Celles, beaucoup plus grandes, qui resultent des combinaisons simultanees de dimensions. D'autre part il y a les possibilites motrices du corps humain et en particulier le fait qu'on peut apprendre tres vite a effectuer des operations simples comme la translation et la rotation. Finalement tout homme a la faculte d'identifier des formes et d'en acquerir une representation interne, y compris une representation interne des operations, de leurs inverses, et de quelques compositions d'operations. Je propose maintenant les hypotheses suivantes. (1) Les possibilites de discrimination absolue dans une dimension sensorielle donnee — qui sont de l'ordre de 7 a peu pres pour tous les hommes — augmentees par les moyens de discrimination simultanee dans plusieurs dimensions, suffisent a garantir la formation d'un nombre considerable de formes elementaires visuelles et auditives. (2) Au moyen de quelques operations simples, effectuees sur ces formes elementaires on peut elaborer des formes internes complexes. (3) La representation interne des mouvements de l'organisme dans son milieu permet la construction d'une image interieure de 1 'Umwelt (c'est-ä-dire, un ensemble de formes interieures isomorphe avec l'ensemble des formes perijues). (4) L'identification de phonemes entendus dans le milieu social resulte de la capacite de discrimination absolue qui est basee sur la multidimensionnalite des distinctive features, et qui est aidee par un mecanisme de feedback entre l'oui'e et l'organe de la parole. A partir des phonemes on peut construire des formes ou des structures plus complexes au moyen des operations dont j'ai deja parle. Si ces theses sont vraies, il s'ensuit que tous les hommes identifieront dans leur entourage un certain nombre de formes qui s'imposent, et puisque leurs mouvements corporels sont quasi identiques ils construiront un schema de base d'une image du monde qui sera identique et done universel. On peut accepter egalement

Epistemologie et hypotheses de Whorf

237

l'existence d'un systeme elementaire d'operations mentales qui s'est forme a partir d'un mecanisme de feedback entre les operations externes et leurs representations internes — ce que Piaget appellerait l'intelligence sensorimotrice. Ce serait une deuxieme structure universelle. II semble done qu'il n'est pas necessaire d'accepter l'hypothese des linguistic universals proprement dits. D'un point de vue d'economie scientifique il faut d'abord tächer de construire les transformations grammaticales a partir de ces operations plus simples et n'avoir recours au nativisme linguistique que si cette construction s'avere tout a fait impossible. Pour evaluer l'hypothese de Whorf il faut de nouveau tenir compte des criteres d'economie et d'adaptation. Il est inconcevable que n'importe quelle langue puisse avoir elabore des structures qui soient en opposition avec le schema d'image du monde dont j'ai parle puisqu'il s'agit Iä de formes et de complexes de formes qui s'imposent. J'admets d'ailleurs que l'acception de l'existence d'une telle image elementaire ne suffit pas a en donner une description exhaustive, ce qui serait une täche extremement difficile. A ce schema de base doivent s'ajouter les formes dont l'identification est d'une necessite imperative dans un milieu geographique, social et culturel donne: s'il y a des scorpions dangereux il faudra les identifier; s'il y a un tabou de l'oncle paternel ou une importante structure geometrique du village ou des techniques de peche ou de chasse, il faudra en construire les formes interieures. Selon le critere d'economie il est evidemment tres probable que des formes linguistiques repondant a toutes ces necessites existeront dans les communautes concernees puisque l'apprentissage des formes est facilite par la preexistence de leur equivalent linguistiqufc. Si done on peut dire que la vision du monde d'un individu est en quelque sorte fa9onnee par sa langue, cela n'est le cas que parce que cette langue a ete adaptee aux necessites de ce milieu. II est toutefois possible qu'un certain nombre de formes et de structures de la langue ne soient pas particulierement favorables ou defavorables dans un milieu donne: elles peuvent avoir subsiste bien que n'ayant plus de fonction utile ou elles peuvent resulter du jeu des lois strictement linguistiques. Dans ce cas-la, etant donne le role indeniable de la langue dans l'apprentissage, il est normal que ces structures et ces formes soient egalement assimilees. Elles contribuent alors a la formation de la vision du monde et dirigent la pensee dans une direction plutöt que dans une autre. Formulee de cette

238

Etienne

Vermeersch

fa9on, je considere l'hypothese de Whorf comme un truisme. En effet, hormis les universaux dont j'ai parle, et les quelques discriminations particulieres que chacun peut elaborer dans son contact avec le monde, on se demande de quelle fa^on un individu pourrait bien penser si ce n'est en appliquant les seules formes et structures qui lui sont donnees en surplus, notamment Celles de sa langue. La seule question qui se pose alors a mon avis, est celle-ci: quelle est l'importance quantitive et qualitative de ces structures et de ces formes que nous pourrions appeler neutres, c'est-adire, celles qui ne sont pas particulierement eufonctionnelles ou dysfonctionelles et quel est leur degre de rigidite,.en d'autres termes dans quelle mesure peuvent-elles empecher la formation de structures nouvelles. II s'agit la de question purement empiriques qu'un philosophe aimerait laisser resoudre par d'autres plus competents que lui.

SUMMARY In this paper the Sapir-Whorf hypothesis is looked at from a philosopher's point of view, some epistemological notions are introduced and some basic hypotheses concerning the nature of man are proposed. The first section is meant to eliminate a naive realism as well as a radical relativism concerning our knowledge in the natural and the social sciences. It is emphasized that conceptual schemes or 'paradigms'guide the scientist even in the identification of facts', but that, on the other hand, the necessity of adaptation to new situations often leads to a breaking of the existing schemes. It seems plausible that the conceptual structures of natural languages may cause analogous constraints, but also that an extreme rigidity is unlikely there. In the second section a few concepts are introduced which may help to elucidate the way man functions as an informationmanipulating system (I-system). Every acquisition of information or of knowledge is considered to be the result of an operation of discrimination and identification performed on a set of states of an energetic or material substratum. The result of this operation is called a form'. The point is that these forms, which constitute the substance of knowledge, cannot exist without the interaction of the energetic (or material) input and some manipulating programs with the subject. Just as in the first section, a moderate relativism seems indicated.

Epistemologie et hypotheses de Whorf

239

In the third section these views are applied to the problem of linguistic universals and the hypothesis of Whorf The thesis is put forward that, applying Occam's razor, one should not introduce more universals than necessary. On neuropsychological grounds we may assume that a minimal reservoir of 'forms' and some basic operations on forms will be constructed by a human organism in his interaction with the environment; this will lead to a universal elementary picture of the world. Linguistic transformations should be considered as constructed out of the basic operations, whereas the elaborate world picture of the adult may depend largely on the form discriminations suggested by his language. In this moderate sense the Whorf hypothesis is a truism. One should not wonder that all human languages have much in common since all cultures have to cope with a large amount of analogous problems. It remains an open question, however, to what extent the linguistic schemes are constraints or are open to adjustment.

REFERENCES Garner, W. R. (1962), Uncertainty and structure as psychological concepts. London. Gipper, H. (1972), Gibt es ein sprachliches Relativitätsprinzip? Untersuchungen zur Sapir-Whorf-Hypothese. Frankfurt a.M., S. Fischer (Series 'Conditio humana'). Vermeersch, E. (1965), 'Some remarks on the analysis of the culture concept', Studia philosophica gandensia, 161—213. — (1967), Epistemologische inleiding tot een wetenschap van de mens. Brugge, De Tempel. Whorf, Β. L. (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J. B. Carroll. Cambridge, Mass., MIT Press.

MAX BELUFFI

12

La Verification Psycholinguistique de la Psychiatrie et l'Anthropoanalyse

Le langage joue un röle essentiel dans trois secteurs de la psychologie: (1) dans la Pathogenese originaire des phenomenes psychotiques et dans la constitution de leur propre Weltlichkeit, (2) dans les differentes methodes scientifiques adoptees pour la designation de la phenomenologie des psychoses; et (3) dans notre fa£on de penetrer l'univers semantique-ontologi que des psychoses, avec nos differents dessins psychotherapeutiques. Compte tenu de ces differentes perspectives, il est indiscutable qu'une consideration nouvelle de la Psychopathologie, en son ensemble, dans un sens veritablement linguistique et communicationel, devrait pouvoir nous permettre, en futur, beaucoup de verifications interessantes. Le poids du facteur communicationel, par exemple, nous apparait preeminent dans la formation de la Stimmung psychotique primaire. Les observations de G. Bateson et collegues (a propos de l'ainsi-dit double bind) temoignent du fait que l'ambiguite et la contradiction des messages (envoyes a un enfant par ses parents) constituent les elements suffisants pour declancher des developpements psychotiques assez graves. Dans les etudes de ce genre, toutefois, n'a pas ete souligne suffisamment un trait assez particulier, c'est-a-dire le fait que la contradiction des messages ne suffit pas, par eile meme, a mettre en route une reaction emotionelle cronique de valeur destructrice. Cette reaction apparait seulement apres qu'un veritable 'monde' (Weltlichkeit) parvient a definir sa structure, du point de vue existenciel, sur les traits basilaires de la contradiction et de l'ambivalence vecue. Pour passer d'une ambivalence semantique (ou logonomique) a une anfibologie existencielle (ou cosmonomique) une condition speciale doit se realiser: c'est le passage du sentir (psychologique) ambigu et

242

Max Beluffi

conflictuel du nevrotique compense, au vecu (anthropologique), dereistique du schizophrene decompense. Dans le premier $as l'ambiguite ne parvient proprement jamais a traduire les resonances semantiques conflictuelles dans le decor d'un monde lointain et perdu; dans le deuxieme, au contraire, l'intention figee cristallise le monde de l'autisme (du manierisme, de l'exaltation, etc.) dans son Ausrüstung paradoxale et barocque. On comprend aisement, done, quel serait le travail qu'il faudrait programmer — justement sur la base des hypotheses de Sapir et de Whorf — pour bien saisir, au sein meme de cette metarmophose catastrophique, le rapport special qui vient se composer entre 'message' et 'vecu', entre 'communication' et 'cognition', entre 'Geschehnis' et 'Erlebnis', entre 'evenement' et 'resonnance'. Une telle verification pourrait nous permettre de pouvoir evaluer les structures elementaires de cette metamorphose, e'est-a-dire le profil exact des Weltlichkeiten qui se succedent et qui se dessinent (entre le premier et le dernier degre de cette regression destructrice) au fur et ä mesure que la Spaltung psychotique avance. A ce propos il est indispensable de rappeler ici les suggestions qui decoulent de l'oeuvre de Κ. Conrad (1958). Le clinicien de Göttingen distinguait soigneusement, pendant le proces schizophrenique initial, les phases successives du 'trema', de Γ 'apophanie' et de Γ 'apokalupsis'. On peut dire aujourd'hui qu'une analyse psycholinguistique attentive de ces Gestalten intermediates pourrait nous offrir un materiel d'importance fondamentale pour reussir a classifier, finalement, les correspondances semantic-gestaltiques des differentes etapes de la schizogenese psychotique. Une repartition soigneuse de ces 'reperta' sur des schemas analytiques (du type des 'Spracbpathogrammen' proposes par H. Flegel (1965) bien dessines nous fournirait en outre un panorama analytique tres important, aussi et surtout, sous la perspective de l'intervention reeducatrice precoce. La täche que la Psychopathologie du futur devra affronter, done, avec une conscience assez plus avertie, par rapport a l'actuelle, e'est justement celle de reussir a comprendre quel est le rapport exact qui passe entre 'langage' et 'cosmogenese' des psychoses et bien evaluer surtout pourquoi certains styles reactifs (nevrotiques et psychotiques) repondent mieux ä certains types de langage et quelle est exactement la typologie linguistique de leur genese et de leur reductibilite. Un autre secteur qui n'a pas ete elucide suffisamment, jusqu'-

La verification psycbolinguistique

de la psychiatrie

243

ajourd'hui, dans cet ordre d'etudes, par les linguistes, est celui de la semiologie clinique, en psychiatrie. Comme toute autre semiologie specialisee, la semiologie psychiatrique a ete toujours consideree au niveau d'un langage veritablement scientifique qui n'exigeait pourtant, aucune verification prealable par rapport ä ses qualites constitutives possibles. Mais il est temps de remonter aux sources et bien saisir le procede historique et linguistique collectif a travers lequel ce langage trouva les conditions necessaires pour sa constitution. II est temps de verifier, enfin, a travers quelles interferences (entre la Weltlichkeit psychotique dominante pendant une certaine periode historique et la Weltlichkeit correspondante des psychiatres-nosologistes) se constitua ce systeme communicationnel metalinguistique qui se qualifie, desormais, comme nosographie psychiatrique. II est indiscutable, sans doute, que ce meme metalangage donne naissance, a son tour, a une autre Weltlichkeit assez particuliere qui poursuit son devenir phenomenologique independamment de la vie reelle des 'malades' (a la description fidele et immediate desquels eile semblerait destinee), d'une part, et de la vie et de revolution de la veritable science psychopathologique (anthropologiquement consideree), de l'autre. Ce monde symbolique artificiel, domine par une semantique capricieuse (et frequemment absurde) represente, en fin des comptes, la structure derisoire ultime du destin de nos 'malades' mentaux et des souffrants psychologiques de tout genre. Nous nous trouvons a faire, ici, done, avec le monde de la denomination discriminatoire, avec l'univers de la classification stygmatisatrice, avec le regne de l'histoire naturelle de la folie (Η. Ey). C'est un paysage linneen qui n'a rien a faire avec l'homme souffrant reel ni par rapport a la traduction fidele de ses messages desesperes, ni envers ses simples demandes d'aide. C'est le monde 'apotropai'que' des signes conventionnels — neutres et neutralisants - avec le pouvoir duquel les differentes societes historiques se sont toujours defendues vis a vis de la folie. Mais, quelle est la fonction ulterieure de ce monde intermediaire de la denomination scientifique-apotropai'que? Je pense que la fonction de cette structure d'intermediation denominatoire est veritablement une fonction 'cosmophylactique' qui permet aux differentes societes historiques de perfectionner leurs systemes juridiques et sociaux de defence contre le pouvoir dissolvant de la folie. Face ä l'emergence continuelle de la 'deraison' (comme dirait M. Foucault), ces systemes taxonomiques exercent une veritable fonction de

244

Max

Beluffi

demonisation de la Nature. lis permettent en meme temps de cultiver l'illusion d'avoir maitrise, dans les mailies de leur phonetique, l'essence meme du 'mal' que nous identifions avec les extravagances de la raison et les egarements de Tesprit. II serait done interessant que la Weltlicbkeit meme de cet univers intermediate (de la taxonomie psychiatrique) puisse devenir l'objet d'une analyse psycholinguistique revelatrice. Nous pourrions ainsi demonter, peut-etre, beaucoup des plus douloureuses superstructures institutionnelles du secteur. L'univers de la taxonomie officielle, en effet, perpetue le 'mythe de la maladie mentale', soutient la dangereuse metaphore de Γ 'hygiene mentale', renforce les innombrables malentendus qui peuplent la psychiatrie medico-legale, continue a justifier des systemes d'intervention judiciaires qui ne possedent aucune conscience de l'enorme faiblesse de leurs racines conceptuelles. II est facile de comprendre, par ces simples indices, quel est le besoin qui domine la phase acteulle de notre discipline: la demistification nosographique en tant que revelation d'une 'Weltlichkeit' factice et petrifiante a qui une telle taxonomie nous condamnent tous: medecins, malades, institutions. Le probleme, a cet egard, n'est pas different de celui qui se presente aujourd'hui en psychanalyse et en psychologie. La seule difference consiste dans le fait que le monde des taxonomies de ces disciplines n'a pas encore influence, d'une maniere appreciable, ni les lois ni les reglements de nos societes. Mais nos moeurs ont ete dejä influences par les Weltanschauungen que ces psychologies et metapsychologies soutiennent et propagandent. Quelle pourrait etre notre destinee collective, on se demande, le jour oil, par exemple, la taxonomie psychanalytique (et sa dependante cosmonie) parviendrait a inspirer, par le systeme de son hermeneutik0 dogmatique, l'elaboration de nos lois? Les consequences pourraient etre assez plus lourdes de celles qui derivent de l'influence operee par la psychiatrie sur la pensee juridique actuelle. La psychiatrie, en effet, ne possede pas une Ideologie originale: eile la derive simplement des ideologies sociales dominantes dans les differentes periodes historiques. La psychiatrie ne possede non plus une Welanschauung, sa propre et bien definie: elle la derive des differentes psychologies auxquelles les differents psychiatres adherent de temps en temps. Ainsi la psychiatrie — avec sa taxonomie idiosyncrasique et traduisante une dangereuse Weltanschauungslosigheit specifique — ne parvient qu'a influencer un

La verification psycholinguistique

de la psychiatrie

245

secteur social et juridique assez restreint. II s'agit du secteur de l'alienation clinique et de celui (assez plus dangereux en rapport a son ambiquite d'inspiration mais encore assez limite) de l'ainsidite 'hygiene mentale'. La psychanalyse, au contraire, en tant que systeme interpretatif de l'essence humaine, (la psychanalyse nourrit des anciennes nostalgies anthropologiques [G. Roheim] et cosmologiques [W. Reich]) ne se borne pas au simple secteur de la clinique ni a la sphere medicol e g a l et juridique plus restreinte: eile vise directement ä l'Homme, eile envisage comme son point de mire electif la societe organisee et les differentes cultures humaines dans leur ensemble. La psychanalyse possede, a cet egard, des programmes missionaires bien precis qu'elle nous fait entrevoir a travers le slang mythomorphe de ses differents dialectes et par l'intermediaire des ceremonies initiatiques organisees d'une fa^on si solemnelle au sein de ses confreries. Les 'therapeias' analytiques sont bien connues par la foule des cathecumenes qui doit assurer la 'catholicisation' (au sens etymologique du mot) de 1 'Es (Gr.: Therapeia, dans le sens de cultus, de servitium, de cura). Quelles sont les operations qu'on s'attend de voir accomplies par ces nouveaux missionaires? II s'agit surtout de la translation de la Weltanschauung et de Yhermeneutike psychanalitique dans les differentes Weltlichkeiten propres aux plus disparates groupes humains de notre temps. Or, quelle a ete la position des chercheurs a inspiration humanistique pendant cette periode si complexe et contradictoire? La psychologie humanistique a dü affronter, pendant un demisiecle, au moins, des resistances enormes qui ne cessent pas d'operer, encore aujourd'hui, a l'echelle mondiale. Des hommes tels que Buber, Cohen, Dorsey, Fromm, Goldstein, R. Hartman, Huizinga, Huxley, Korzybski, May, Maslow, Munford, Murphy, Polanyi, Rogers, Sorokin, Tillich, etc., ont deploye des efforts Continus dans cette entreprise. La merae chose pourrait etre repetee pour la psychiatrie humanistique et pour l'anthropophenomenologie. Des savants tels que Gruche, Berze, Boss, Minkowski, Zutt, Franki, Kuhn, Rumke, V. Weizsaecker, Wyrsch, E. Straus (et combien d'autres encore) ont conduit, pendant des dizaines d'annees, une Campagne culturelle opiniätre qui seulement des nos jours montre quelques timides possibilites de penetration dans les entourages cliniques les plus avertis. Du cote des responsabilites on peut dire seulement que les

246

Max

Beluffi

hommes de science a orientation humanistique ont lache prise, inexplicablement, a partir environ de la fin de 1800, juste dans un domaine oil, pendant des siecles, ils avaient tenu chaire avec beaucoup de prestige. Harcelees par le positivisme (par rempirisme, le pragmatisme, le behaviorisme, le reflexologie pawlovienne et bien d'autres 'ismes' encore) les 'humaniora' ont ete carrement chassees du domaine des ainsi-dites 'sciences' psychologiques et psychopathologiques. Mais ä quel prix? La crise actuelle des psycho-sciences nous renseigne a suffisance sur le proces de tarissement morale qui ravage dans ces disciplines. Les scientisme de maniere les a menees dans une impasse dont elles cultivent seulement l'espoir de parvenir a se tirer, un jour ou l'autre.apres une periode ulterieure de penitence, de desorientation et de sterilite. Si on considere, a son tour, la Daseinsanalyse on doit, par contre, nous demander: pourquoi cette doctrine, avec son puissant arriere plan humanistique, n'a jamais evolue comme methode specifique de travail veritablement idiographique et ' c u r a t i f ? En effet, il est indiscutable que la Daseinsanalyse a envisage, par exemple, avec une attention particuliere, et des le debut meme de son projet de travail, le problem e du langage humain. Je dirais carrement que les preoccupations gnoseologiques visant les grandes inconnues du langage ont inspire, des son origine, toute la pensee clinique et philosophique de L. Binswanger. A ce propos j'aimerais d'attirer votre attention sur deux essais du philosophe-psychiatre: un premier de 1926 ('Zum Problem von Sprache und Denken') et un deuxieme de 1946 ('Über Sprache und Denken'). Les deux essais representent d'une fa^on tres fidele revolution j'oserais presque dire 'psycholinguistique' de L. Binswanger. En partant d'une position enrichie d'etincelantes references classiques, L. Binswanger montrait tout de suite, en 1926, une conscience nette des allees qu'il fallait parcourir, dans le contexte des futures recherches, pour faire epanouir, dans ce secteur de travail, des hypotheses plus fertiles. Ii s'agissait de trouver la juste synthese parmi une foule de suggestions qui remontaient jusqu'a V. Humboldt et ä Herder et qui avaient ete multiplies en cascade a travers la meditation de penseurs d'une envergure exceptionelle tels que V. Monakow, Bühler, Gomperz (avec sa Weltanschauungslehre) et finalement Cassirer. 'Dass die Sprache "in einem Brennpunkt des geistigen Seins" steht, wie Ernst Cassirer sagt, und nicht nur des geistigen Seins, haben wir

La verification psycholinguistique

de la psychiatrie

247

gesehen. Vor Einsitigkeiten suchten wir uns zu bewahren. Weder Hessen wir die Sprache in einer universellen Grammatik oder Logik aufgehen, was nich einmal bei Husserl der Fall ist, noch in einer allgemeinen Ästhetik, wozu Vossler mit Benedetto Croce neigt, noch erkennen wir ihr lediglich eine praktische und empirische Wirchlichkeit zu, wie es bei den schweitzerischen Sprachforschern de Saussure und Bally, dem franzosen Meillet, aber vor allem auch bei der Wundtschen Schule der Fall ist, noch auch glauben wir, ihr und ihren Störungen von der Seite der Neurologie allein oder der Biologie oder der Psychologie allein beikommen zu können' (Binswanger 1955: 345). A cette epoque L. Binswanger etait bien convaincu qu'on ne peut pas affronter les problemes ultimes du langage, meme et surtout pour un chercheur d'orientation biologique (tel qu'il l'etait) en dehors de la charpente philosophique indispensable. 'Sollaus einer solchen "Vielseitigkeit" der Betrachtung kein blosses "rapsodisches" Konglomerat oder Chaos, sondern ein wissenschaftlicher Kosmos werden, wobei wir "Spiel in Arbeit, Meinung in Gewissheit" zu verwandeln haben (Kant) so müssen wir uns, um mit Husserl zu reden, "hier wie sonst der grossen Kantischen Einsicht fügen, uns ganz mit ihrem philosophischen Sinne erfüllen: dass es nicht eine Vermehrung, sondern Verunderstaltung der Wissenschaften sei, wenn man ihre Grenzen ineinanderlaufen lasse".' Um dies zu vermeiden, müssen wir uns ein weiteren philosophischen Einsicht fügen, von deren Sinn schon Sokrates und Plato erfüllt waren, des Einsicht, dass wir uns stets Rechenschaft abzulegen haben, λόγοι; δούναι' von den Voraussetzungen unseres wissenschaftlichen Tuns und von dem Sinn und den Prinzipien der Wissenschaft selbst' (Ibid.: 345). L. Binswanger evaluait correctement, alors, les relations qui passent entre le langage du schizophrene par exemple, et celui du reveur ou de l'aphasique. Ii envisageait bien nettement la racine de ces langages particuliers dans les correspondantes typologies de 'prise' de la conscience sur des Weltlichkeiten bien distinctes. Et ces Weltlichkeiten etaient et sont distinctes et meme opposees, par rapport aux disparates idiosyncrasies des differentes 'corporeites', vecues par chaque malade dans chaque situation momentanee particuliere. Une telle 'corporate vecue' resulte determinee, a la fois, par les facteurs affectifs et emotionnels qui regissent le rapport ou de participation ou bien d'exclusion entre l'individu et son milieu physique et moral. Pendant cette premiere etape de sa meditation 'psycholinguistique'

248

Max

Beluffi

L. Binswanger analyse une literature tres riche. En elaborant les idees de Gelb, de Goldstein, de Eliasberg, de Feuchtwanger, de Boumann et Grünbaum (sur le versant neuropsychologique) et Celles de Husserl, Vossler, Steinthfil, Paul, etc. (sur le versant plus proprement philosophique et linguistique) il nous introduit a revaluation d'une serie de dimensions nouvelles pour son epoque. C'est la dimension de la situation (Heilbronner), c'est l'importance de lapersonnalite (Pos), c'est le role de la structure conversationnelle (Vossler), c'est le poids du facteur express/'(Buytendijk und Plessner). 'Und wir finden uns als moderne Psychopathologen und Neurologen wieder in einer sehr vertrauten geistigen Sphaere, wenn wir sehen, wie das lebendige "Sprachbewusstsein" jetzt auf sein möglichen "Stadien" untersucht wird mit der klaren Einsicht, "dass hinter jedem niedergeschriebenen oder eingemeisselten Wort als Urheberin die lebendige Person steht" (Pos), und wie wieder das lebendige und konkrete Sprechen der Person, "das verständnisvoll angehorte, beantwortete und seiner Wirchlichkeit versicherte Sprechen" oder das Gespräch es ist, das neben dem vereinzelten Sprechakt und der abstrakten Sprache wieder zur Geltung kommt (Vossler). Dasselbe gilt von den neuesten Anschauungen über die Beziehungen zwischen Sprache, einerseits, Handlung und mimischen Ausdruck, andererseits. "Zwischen reinen Ausdruck als untrennbarer Einheit von erscheinender Leibgestalt und Sinngehalt und zwischen reiner Handlung halt sprachliches Bedeuten die Mitte" [ . . . ] "Sprachliches Bedeuten ist eine zielgerichtete Bewegung, die sich weder im Ende, noch in der Bewegung, sondern in der Einheit der Bewegung erfüllt". (Buytendijk und Plessner). Auch darin zeigt sich die Mittelstellung der Sprache zwischen "Ausdruck" und Handlung' {Ibid.: 339). Ii n'echappait pas a L. Binswanger, des le debut de sa meditation sur ces problemes, l'importance d'une antinomie centrale, dans ces etudes, une antinomie qui pendant ces dernieres saisons de l'histoire de la linguistique, a provoque tant de polemiques fausses et inutiles: l'antinomie entre psychologie et grammatologie. 'Es braucht nunmehr kaum wiederholt zu werden, dass diese Lehre Humboldts und Husserls von den grammatischen Formen als einem Korrelat der geistigen Bedeutungen durchhaus nicht in Widerspruch steht zu der Lehre von der Beziehungen zwischen grammatischer und psychologischer Kategorie, wie wir sie aus der Sprachgeschichte kennen. Wir sahen oben dass Husserl nichts einzuwenden hatte gegen die Auffassung Herman Pauls, die Vossler überall weiter ausgearbeitet hat: "Die grammatische Kategorie ist gewissermassen eine Erstarrung

La verification psycholinguistique

de la psycbiatrie

249

der Psychologischen. Sie bindet sich an eine feste Tradition. Die psychologische dagegen bleibt immer etwas Freies, lebendig Wirkendes, das sich nach individueller Auffassung mannigfach und wechselnd gestalten kann" (nach H. Paul)' (Ibid.: 343). Mais c'est seulement avec l'essai de 1946 que L. Binswanger s'approche decidement aux themes plus propres que Sapir et Whorf ont eux aussi envisage. Dans le chapitre 'Sprache und Welt' on trouve les considerations les plus significatives pour notre recherche. L. Binswanger y souligne le role createur du langage par rapport a la constitution du 'monde' et il y envisage l'importance de la participation commune de plusieurs 'conversateurs' pour la Gestaltung d'un 'monde commun', veritablement humain, dans lequel l'intention significative puisse se realiser un tant que veritable koine. 'Wenn Sie mich fragen, wo ein Sprechen als Intonation einer Abfolge von artikulierten Lauten und gänzlich isoliert vom Denken in der Welt vorkomme, so antworte ich: nur beim Papagei. Und wenn Sie mich fragen, "wo ein Denken rein als Spontaneität des Geistes und ganzlich isoliert vom allem, was Sprache heisst, in der Welt vorkomme", so antworte ich: nirgends. Wo wir von einem reinen, d.h. nicht-sprachbezogenen Denken reden, reden wir von einem ausseroder überweltlichen Wesen, von Gott; denn wo Sprache ist, ist Welt (soulignement a moi). Es ist daher ein unausschopfbar tiefer Gedanke des Johannes-Evangeliums, dass es sich die Welt enstahden denkt durch das Wort Gottes, durch das Verbum divinum, den göttlichen Logos. Wenn Gott spricht, ist er nicht mehr " G o t t ohne Welt", sondern Weltschöpfer. Das Wort is hier nicht nur vergänglicher Laut und nicht nur bleibender Gedanke, sondern, wie Faust mit Recht übersetzt, Tat, einmalige Schöpfung. Fragen Sie mich schliesslich, wo das Sprechen in der Welt vorkommt als "sinnbelebter Ausdruck", als sprachliche Verlautbarung und Mitteilung von Gedanken, als Ansprechen und Antworten im Sprechen über etwas, mit einem Wort: als Gespräch, so antworte ich: nur beim Menschen. So steht das* menschliche Gespräch zwischen dem blossen Nachplappern von artikulierten Lauten und dem Sprechen als kosmogonischer weltschöpferischer Tat' (Ibid.: 346). Mais, qu'est-ce que c'est exactement ce 'monde' et quelle est veritablement la fa9on par qui il se constitue? L. Binswanger affirme qu'il s'agit surtout d'evaluer les 'modalites' constitutives d'une teile Weltlichkeit, et la täche la plus veritable du linguiste-philosophe est, sous ce point de vue, celle d'une analyse stylistique fonciere.

250

Max

Beluffi

'Gleichwahl bleibt auch für die mensliche Sprache der Satz zu Recht bestehen: Wo Sprache ist, ist Welt, und gleichwohl ist auch die mensliche Sprache in einem seit Nietzsche und Heidegger genau bestimmbaren Sinne des Wortes weltschöpferisch. Um diesen Sinn von Welt zu verstehen, müssen Sie sich klarmachen, dass das Seiende uns nicht unmittelbar zugänglich ist, sondern immer nur als etwas, als dieses oder jedes, m.a.w., dass es uns nur zugänglich is je nach der Art und Weise, wie wir es verstehen, wie es uns anspricht, und wie wir es aussprechen, wie wir mit ihm umgehen, wie es uns anmutet, überfällt, stimmt. In all diesen Arten und Weisen, wie uns das Seiende zugänglich, bedeutsam oder offenbar wird, konstituiert es sich für uns erst als Welt. Welt bedeutet hier also nicht mehr Universum im Sinne der Gesamtheit des Seiende als solcher, sondern Kosmos im Sinne der Art und Weise, wie das Seiende im Ganzen sich dem Menschen offenbart, ihm zugänglich oder bedeutsam wird, und zwar im alltaglichen Umgang mit ihm in Arbeit und Handwerk, im Bedrücktsein von ihm in Angst und Schrecken, im Gehobensein von ihm in Freude und Begeisterung, in seinem Genuss, in seiner künstlerischen Gestaltung oder religiösen Verehrung, in Liebe und Freundschaft, endlich in seiner distanzierten, ruhigen Betrachtung, Vergleichung und begrifflichen Durchdringung, mit einem Wort: in seinem Denken' (Ibid.: 347). Par tout ce qui a ete dit jusqu'ici et a travers la prose meme de L. Binswanger on peut comprendre aisement quelle est la clef de voüte de la constitution du 'monde' interhumain: il s'agit essentiellement du Gespräch. Un Gespräch vivant (lebendige) et bien articule sur le plan de la dialectique interpersonnelle: 'Der Mensch ist nur "Mensch" im Miteinander-Sprechen, in der Verständigung von Ich und Du auf dem Grunde einer gemeinsamen sprachlichen Welt oder, wie wir mit Heidegger sagen, eines gemeinsamen sprachlichen Weltentwurfs. Sprache is kein blosses "Austauschmittel" sondern Miteinandersein in einer Welt möglicher Verständigung. [. . .] Damit hängt unmittelbar zusammen die Auffassung des Gegenstandes als einer Bedeutungseinheit, über die die Verständigung geht. Verständigung und Gegenstand lassen sich nicht trennen. [. . .] Wo kein gemeinsames Meinen über etwas, also kein gemeinsamer Gegenstand, möglich ist, wie z.B. bei der schizophrenen Psychose, da ist auch keine Verständigung möglich und umgekehrt. [. . .] Um sich verständigen und erst recht um sich verstehen zu können, muss man schon in einer Welt gemeinsamen Meinen sein. Zwar ist Mensch-sein an uns für sich schon ein Sein mit andern in einer irgendwie geglieder-

La verification psycholinguistique

de la psychiatrie

251

te Welt, einer Welt irgendwelcher Bedeutsamkeiten [. . .] Aber erst im Denken schlägt die Welt als eine Welt von Bedeutsamkeiten um in eine Welt von Bedeutungen schlägt die Welt des blosses Umgangs und Verkehrs, des blossen Sympathie und der Andacht um in die Welt der gedanklichen Kommunikation, der Verständigung durch Sprache, mit einem Wort, in die Welt des Logos' (Ibid.: 350—352). Tels etant les caracteres du 'Welt' qui, selon L. Binswanger se constitue a travers la mediation merveilleuse du langage, on peut se demander, a ce point, quelles puissent etre-sous une perspective pratique-les deductions qu'on en peut tirer. Et tout evidemment et avant tout, dans le secteur specifique de la psychotherapie. En effet, il est certain qu'on peut songer a constituer un verkable 'thesaurus' daseins-analytique a travers la_ definition des 'tropes' les plus caracteristiques du langage meme de L. Binswanger et tels qu'il les a employes, par exemple, dans son hermeneutique phraseologique de la serie radicale du 'nehmen' (1962). Ces tropes pourraient etre ordonnes dans un systeme de categories — deja esquisse, en ligne generale, par L. Binswanger lui meme — jusqu'au point de nous permettre la constitution d'un veritable systeme ou structure de triage des textes qu'on entend analyser du point de vue daseinsanalytique. Par des approaches successives il serait possible de constituer, ensuite, un veritable systeme d'analyse structurelle-stylistique süffisante a sonder les implications intentionnelles qui se cachent dans les langages des ainsidit malades et de tous ces cas qu'on envisage d'explorer, en generale, sous une perspective daseins-psycholinguistique. J'ai eu l'occasion de commencer un certain travail de ce genre pendant les annees 1966—1968 dans le cadre d'une recherche sur l'anthropoanalyse de certains groupes de psychotiques äuxquels je presentais en vision, avant le Gespräch de groupe, certains films d'art. Les themes les plus importants qui emergeaient de l'analyse de ces films constituaient, ensuite, les references fondamentales sur lesquelles la conversation de groupe se developpait. Les textes de ces conversations etaient enregistres sur bande magnetique; dans un moment successif on le transcrivait a la machine. De cette faijon on pouvait rassembler un veritable corpus linguistique se referant a un complexe de 'malades' bien determine, ä des conditions experimentales assez bien definies et ä des styles de representation du monde extremement caracteristiques. Les textes d'une douzaine de ces seances de groupe ont ete rassembles et commentes dans un livre que j'ai publie et dans lequel

252

Max

Beluffi

j'ai resume les premieres deductions d'une recherche tres complexe qui demande encore des annees de travail (Beluffi 1969). Je n'ai pas en programme de vous entretenir ici sur les details d'une telle experience. Je me bornerai a souligner que, du point de vue psycholinguistique, cette experimentation m'a fourni beaucoup de suggestions. Je suis convaincu, par exemple, de l'interet qu'il y aurait de pouvoir constituer un thesaurier daseins-analytique compare (dans les differents langages europeens) pour parvenir, ainsi, ä des confrontations interlinguistiques entre les veritables formes-langage des Weltlichkeiten anthropo-analytiques propres a chaque Iangue. Get instrument" (et les autres analogues qu'on pourrait multiplier dans ce domaine) devrait nous fournir les categories phenomenolinguistiques indispensables a transformer la Daseins-analyse de ce qu'elle est actuellement (c'est a dire un simple projet anthropophilosophique general) en une methode psychotherapeutique suffisamment bien definie dans ses contours. L'analyse s'ouvre, a ce point, sur l'essence meme de la psychotherapie et sur ses rapports specifiques avec le langage, en general, et la psycholinguistique, en particulier. En effet la psychotherapie a ete consideree pendant de longues annees comme une activite d'ordre suggestif (Charcot, Breuer, Bernheim, le premier Freud, etc.) et seulement ensuite comme une activite 'analytique' (le Freud classique et les differentes ecoles de la Tiefenpsych ologie). Mais qu'est-ce que signifie 'analytique', ou 'individuel', ou 'profond'P Les chercheurs de l'ere proto-psychotherapeutique etaient convaincu que les 'maladies psychiques' (nevroses et psychoses) pouvaient etre soignees a travers la resolution des conflits element a r e s qui les soutenaient. Ces conflits, a leur tour, etant centres sur des representations symboliques tres particulieres, la täche que ces chercheurs envisageaient c'etait celle de resoudre ces concretions semantiques abnormes dans leurs hypothetiques essences significatives 'originaires' de fa9on telle que la charge pathogenique respective pouvait en etre eliminee. Tout ce travail fut toujours conduit sous 1'inspiration d'une linguistique assez rudimentaire et d'une rhetorique primitive, non bien conscientes — soit l'une que l'autre — de leur propre grandes possibilites justement dans ce champ de travail. La linguistique freudienne et junghienne, en particulier, quoique bien dirigees a la recherche d'horizons hermeneutiques assez precis, n'avangaient pas tres en

La verification psych ο Unguis tique de la psychiatrie

253

profondeur sur le terrain que les linguistes de ces memes epoques historiques (et les philosophes du langage) contribuaient a eclaircir avec les contribution, meme psychologiques, si originales. C'est seulement avec l'avent de la Daseinsanalyse (et de l'analitique existencielle heiddeggerienne a qui cette discipline s'inspire) que le langage psychotherapeutique s'enrichit de references originelles fondees sur une nouvelle conception du monde en general et du monde de la participation psychotherapeutique en particulier. II s'agit d'une espece d'anabasis aux sources memes de la significativite existencielle du langage (comment oublier la Ιεςοη d'un R. M. Rilke, a ce propos?) qui permet a la linguistique daseinsanalytique de devenir une veritable ontoanalyse, avec toutes les implications 'apokalyptiques' qu'une telle orientation comporte. La Daseinsanalyse introduit en psychotherapie la dimension evaluative et communicative de la Weltlichkeit et celles des 'univers significatifs'; eile provoque dans le langage psychotherapeutique un veritable bouleversement en rappelant au psychotherapeute que la 'nevrose' ou la 'psychose' ne representent pas des simples aggregations symboliques; eile avertit le psychotherapeute que les 'univers significatifs' de ces 'maladies' doivent etre approches et traites avec un outillage anthropologique (au sens philosophique) et non plus simplement psychologique. Dans cette perspective de travail on commence a comprendre mieux l'universalite meme du langage dont les frontieres ne se bornent plus a la simple dimension du langage verbal conventionnel. La psychotherapie cesse de representer une pale technique dialogique (ou polidialogique) en devenant a nouveau une situation humaine polivalente soit du point de vue ontologique que linguistique. En cessant d'explorer des champs significatifs aprioriques, restreints et artificiels, eile dirige son projet sur la mediation communicationelle de rencontres humaines bien plus concretes et eclaircissantes que celles du 'divan'. On entrevoit done, desormais, une veritable linguistique psychotherapeutique generale, qui, en derivant ses categories interpretatives et ordinatives des differents univers expressifs qu'elle considere (verbal et preverbal, pictorial et plastique, gestique et proxemique, ludique et dramatique, poetique et satyrique, parle et ecrit, presentiel et documental, etc.) se trouve en condition de realiser un projet de travail completement nouveau dans le domaine des recherches fondatives sur la psychotherapie. Un tel programme exige une collaboration interdisciplinaire

254

Max Beluffi

considerable et une grande flexibilite spirituelle de la part de tous ceux qui voudront s'y interesser serieusement. En effet il faut parvenir a considerer l'homme (malade) en tant que sujet, et non plus comme objet, de nos differentes psychotherapies, en lui restituant sa position de veritable protagoniste de l'acte psychotherapeutique. En somme, il faut fonder le projet de cette linguistique psychotherapeutique generale sur des piliers d'un style nouveau, puisque le style est l'element imponderable mais determinant de la vitalite de n'importe quel projet humain. 'Denn wo Stil ist, da ist Liebe zur Form-, Stil der Persönlichkeit ist Liebe zur Einheit von Herz, Seele un Vernunft, von Ich und Du, von Ich und Gott, von Ich und Welt, von Zeit und Ewigkeit. Dass diese Ziel aller Bildung estrebt werden kann, verdanken wir der Sprache des Volks, dem Sprechen der Dichter und Denker und dem über die Jahrtausende hinweg fragenden und antwortenden Gespräch der Menschheit mit sich selbst' (Binswanger 1955: 362). Tels sont les presupposes principaux de notre orientation actuelle de travail: l'experience future pourra nous deceler progressivement les limites et les avantages d'une teile perspective.

SUMMARY

The author treats some problems in psychopathology. It is claimed that the impact of communication is generally poorly studied and seriously underrated in the analysis of psychopathological syndroms as well as in the different therapies proposed until now. In psychiatry, the diagnostic problem that will have to be confronted is primarily the establishment of the exact relationship between 'language' and 'cosmogenesis' of a psychosis. It is important to keep in mind that 'cosmogenesis' cannot be equated to the 'cognitive semantic map' of an individual; rather, it stands for the 'existential' world view, the life world (both synchronic and diachronic) of the mentally suffering subject. The actual description of this relationship can not be appropriately provided by contemporary scientific models, or through psychoanalysis. On the other hand, the long developed humanistic psychiatry can serve as a basis to confront the task. Especially L. Binswangens refreshing work would be relevant here, provided it be extended to become truely general psychotherapeutic linguistics.

La verification

psycholinguistique

de la

psychiatrie

255

REFERENCES Beluffi, Μ. (1969), Cinema d'arte; alienazione e psicoterapia. Bologna, IlMulino. Binswanger, L. (1955), 'Zum Problem von Sprache und Denken' [1926], and 'Über Sprache und Denken' [1946], in Ausgewählte Vorträge und Afsätze, Band IL. Bern, Francke. Conrad, K. (1958), Die beginnende Schizophrenie. Versuch einer Gestalt-analyse des Wahns. Stuttgart, G. Thieme-Verlag. Flegel, H. (1965), Schizophasie in linguistischer Deutung. Berlin, SpringerVerlag.

ERVIN LASZLO

13

The Reduction of Whorfian Relativity through a General-Systems Language

INTRODUCTION

We shall speak to the issue posed by Whorf in the following statement: 'Every language is a vast pattern-system, different from others, in which are culturally ordained the forms and categories by which the personality not only communicates, but also analyses nature, notices or neglects types of relationships and phenomena, channels his reasoning, and builds the home of his consciousness' (Whorf 1956: 252). We shall take 'language' to include both natural and scientific languages, possibly coexisting within the same culture. We shall not address the question of the origin of these languages but restrict ourself to a consideration of their synchronous relations. We shall also not speak to the problem of coexistence of these languages on the level of entire societies or cultures, but deal only with their presence within the cognitive maps of individuals, and groups of analogously cognizing individuals. Hence our paper will address the problem of epistemological relativism induced by different but coexisting natural and scientific languages within individuals, and tight epistemic communicates in a synchronic analysis. More specifically, We shall examine (1) the fragmentation of the world picture of average individuals and communities of such individuals through the use of the natural language typical of contemporary Western cultures; (2) the fragmentation of the reconstructed world picture of individuals and communities using Western scientific languages without integration among them; and (3) the potentials of a general systems language to reunite the contemporary Western world picture in a metascientific reconstruction.

258

Ervin Laszlo

1. THE FRAGMENTATION OF THE COMMON-SENSE WORLD PICTURE

The evolution o f common-sense logic and ordinary language presents each user with a framework for viewing the empirical world. This framework is functional for the purpose of handling information in the contexts of interpersonal behavior, but is not necessarily true to the nature of the empirical world. The world picture shaped by experience in the framework o f ordinary language contains terms for objects and events which are assumed to have counterparts in the world, existing independently of being named and experienced. For example, the term 'tree' is assumed to refer to trees which are empirical realities beyond the confines o f language and cognition. However, the range o f objects and events assumed to have such reference is narrower than the range o f existential terms in the languages of the arts, sciences, and religions. For many persons in contemporary Western societies, the empirical world consists primarily o f the persons with whom the language user is in personal contact, and of the class of persons he assumes to exist elsewhere in the world. Secondarily, independent existence is assumed with regard to the things which form the background o f interpersonal relationships, things such as animals, plants, and the inanimate natural objects and artifacts o f the environment. But the microcomponents which constitute persons, as well as animate and inanimate things, are usually conceived as abstractions, populating the minds of theoretical scientists with little if any reality apart from their theories and formulas. The social and ecological entities formed by the interactions o f many living things appear likewise abstract and pale in comparison to the concreteness o f the everyday world. In the context o f everyday common sense, ordinary language confers reality on the persons and things which one can touch and which have a noticeable influence on one's purposes and projects. Because o f their subject-predicate structure, Indo-European languages confer substantiality on the reified things o f everyday life. Predicates refer to real, material things and persons clothed with the predicated properties. Greenness refers to the substance of grass; tallness, handsomeness, and similar predicates to the substance that makes up persons — the flesh and bone. In order to account for mental attributes and traits of character, the logic of common sense Western languages a 'mental substance' — a mind or personality of which one can say that it is good or bad, likeable or detestable.

The reduction of Whorfian

relativity

259

Consequently common sense combined with the structure of ordinary language structures a world picture for Western man which has in its sharpest focus oneself and those around one; in lesser focus are the animate and inanimate things of the immediate environment; and then the picture fades toward abstraction as one goes down the scale to microphenomena and up to macroentities. The reified entities appear as substances clothed with qualities — physical as well as (in the case of persons) mental. Such a world is relatively narrow: It is implicitly dualist, and it is anthropocentric. It is the world of childhood when the elements of fantasy are dispelled and one is left with 'stark reality'. The sciences as well as the arts, and religion, expand this anthropocentric bubble built by common-sense logic and ordinary language. The language of art (if indeed it uses a language) expands the bubble through imagination. The language of religion expands it through faith. Science expands it through a suspension of common sense and ordinary language, and a reliance on its own method of hypothesis formulation and validation. But already life in a complex social community expands the bubble by bringing into the sphere of immediate concern various multiperson entities and man-machine systems. Businesses and corporations emerge into the realm of reality; one speaks of them as entities subsisting independently of the persons forming them. Legal and political doctrines reinforce this conceptualization by endowing formally constituted multiperson entities with rights and responsibilities. Group identification has provided a particularly sharp focus for multiperson entities. At the dawn of civilization tribes and clans, later communities, city states, small and large states, and presently even bloc alliances among states provide a vehicle for reifying entities beyond the level of persons. In a complex and interdependent world, the bubble of common sense and its language expands upwards, as persons fit themselves through specialized training and role functions to enduring social, economic, and political systems. The subject-predicate structure of our languages conditions our perception of these entities and masks the fact that we cannot common sensically assume that they have a substance of their own. Hence we can speak confidently, if inconsistently, of teams, corporations, social, political and cultural organizations, and entire nation states as being good or bad, efficient, or malicious, or agressive. If questioned, common sense would dictate that we clarify our language and point out that we mean the

260

Ervin Laszlo

character o f the individuals making up the predicated entities, but in fact we do not mean the individuals, for individuals can be transient whereas the character of the entities may be relatively invariant. Thus the character of multiperson entities 'inheres' in some unperceived substance, much as psychological characters inhere in minds and personalities. The logic o f common sense and the structure o f ordinary language forces such conceptualizations on those who would use them indiscriminately.

2. THE FRAGMENTATION OF THE SCIENTIFIC-WORLD PICTURE

The Worf-Sapir hypothesis can be extended to the logic and language o f the sciences. Scientists tend t o perceive the world in terms relevant t o their scientific specialties. Since different specialties use different concepts and modes of structuring their empirical data, scientists in different disciplines tend to have differing organizations o f empirical reality. The differences occasioned by adherence to specific scientific theories take the form o f partial qualifications of the everyday world picture. No scientific theory meshes fully with common sense and ordinary language. Bertrand Russell pointed out that common sense, if true, leads to physics, and physics, if true, shows that common sense is false. Hence common sense, if true is false; therefore it is false. What Russell said of physics can also be said of other natural and social sciences. Inasmuch as they go beyond the description and classification o f observations and attempt an explanation o f their data, they make use o f theoretical frameworks o f which the axioms as well as the rules o f deduction are constructed in reference to the methods and criteria o f the sciences, not o f common sense. Thus theoretical sciences do not hesitate to postulate unobserved entities provided thereby consistency can be introduced into the explanation o f observations, as well as parsimony, elegance, extensibility to other phenomena, predictive power, and so on. The consequence is that the processes of reification become autonomous in science; and the reified world o f the scientist differs from the common-sense world. Insofar as a scientist does not fully divorce his professional and his private spheres of cognition but demands some degree o f consistency between them, he will begin to qualify his everyday logic and language in the light o f his theories. The modification

is likely to vary with the kind of specialty

The reduction of Whorfian

relativity

261

embraced by the scientist. The almost monolithic unity of modern science, induced by the paradigmatic role of Newtonian physics, was shattered with the advent of relativity. Physics as the paradigm to be emulated was increasingly questioned by the life sciences when it became apparent that the equations of quantum mechanics are incapable of providing an explanation of living phenomena. Also, the social sciences came into their own, and the sweeping reductionism, expressed in the last century in Fechner's concept of a 'psychobiological-physics' as the perfect science, came to be replaced by relatively independent theories, building conceptual universes as best suited to account for the phenomena of their particular concern. Although contemporary science is pluralistic it is nevertheless different from the logic of common sense and ordinary language. It rejects the subject-predicate structure of ordinary Indo-European languages as a mode of conceptualizing reality. What a thing is, independently of its attributes, cannot be answered in science; nor does it make sense to explain observations as attributes, since they would then resemble the grin of the Cheshire cat. Instead, science conceptualizes by building models of the structure and function of diverse entities and of their interactions. These models answer questions concerning how the entities are organized and how they function when conditions are specified. They are incommensurate with the entities reified in ordinary discourse as having qualities inhering in substances. Hence scientists tend to modify their everyday construction of the world in accordance with the language they use qua scientists. A nuclear physicist is likely to modify his common-sense world picture to accomodate the existence of elementary particles as the constituents of all observed things. Introspectionist psychologists will expand the mind concept of common sense to a rich internal structure consisting of drives, tensions, emotions, fixations, and similar structural elements, whereas psychologists of the classical behaviorist school will look on human beings as complex reactive systems. Life scientists recognize unicellular organisms, genes, and species, as furnitures of reality, but only ecologists among them tend to affirm the reality of ecosystems. Social scientists recognize the entities on the level of their investigation: Microsociologists and social psychologists make entities of groups and other primary units of social, economic, and political interaction, and macrosociologists and international relations theorists recognize nation states, supranational blocs, transnational organizations, and universal actors, such as the

262

Ervin

Laszlo

United Nations. Since scientists participate in the everyday world and use ordinary language, their specialties condition their everyday world outlook but do not totally replace them with the perspective of their discipline. Scientists structure their perceived world to include persons, other animate things and the observed inanimate things in their milieu, but also include in their world picture the principal units of their professional investigations. For natural scientists other than astronomers and ecologists, these units tend to lie below the order of magnitude of the sense perceived world, for atronomers, social scientists, and ecologists, above the perceived magnitudes. Their respective world pictures are extended in these directions beyond the confines imposed by common sense logic and the structure of everyday language.

3. THE REUNIFICATION OF SCIENTIFIC-WORLD PICTURES THROUGH GENERAL-SYSTEM THEORY

The great cultures of the past have not offered a fragmented world outlook but conserved some element of consistency in the way they structured experience for their members. Classical Hellenic culture, for example, saw man as part of the plan of creation, or of the evolution of the cosmos from chaos to order, and regarded human institutions as partaking in the universal scheme. There were gods, men, animals, and other living things, as well as inanimate nature, but these were all products of a creative demiurge or of the crystallization of order through the separation and conflict of opposites. Classical Oriental thought likewise distinguished various levels of reality but did not isolate or oppose them. Man was to fit himself into nature and seek union with the cosmos through illumination. The Middle Ages also produced a harmonious and unified world picture by ascribing the reason for the existence of all things to God. Man alone had a divine soul, but the whole world was God's creation and man was to seek his salvation through fitting himself into the earthly order without arrogance or agression. The rise of modern science with its technological spin-offs, and the contemporaneous rise of a Protestant ethic of hard work for the good of man equated with the glory of God, splintered the harmony of the Medieval world and placed man above nature as its master. It gave free reign to the development of science, only to find that the mechanistic world view of Newtonian physics, and the transcenden-

The reduction of Whorfian

relativity

263

ral spiritual outlook of classical religion, provoked a fission which philosophers, even of the stature of Hume and Kant, were unable to mend. The result was the separation of a common-sense perceptual world based on sole reliance on ordinary language: There came into being the worlds of the specialized sciences, and the distinct worlds of the arts and religions. There are currents taking shape today which seek to overcome the multiple splintering of the empirical world pictures through the creation of an internally consistent general metalanguage to give structure to perception. One of the most powerful elements in these currents is general systems theory. General systems theory adopts the emphasis on structure and function in the sciences, but extends it to all sciences and, through systems philosophy, to the humanities as well. The result is a new structuring of the empirical evidence into levels of organization. Two general level structures are distinguished. One level includes the entities of physics and astronomy: Space-time field, elementary particles, atoms, cosmic" dust, and interstellar matter in micro aggregations (meteorites, etc.), stellar objects of all kinds, satellitic systems, star families, stellar clusters, galaxies, galaxy clusters, and the metagalaxy as the totality of astronomical objects. The other level comprises a level structure evolving on the surface of the earth and assumed to be paralleled in some respects or other suitable planetary surfaces. This structure can be conceptualized as starting with some varieties of stable atoms, molecular systems, crystals, macromolecular compounds, cells, multicellular organisms in speciesspecific populations, and interspecific ecosystems. It includes man and mankind. Abstracting here from the astronomical hierarchy of stars and galaxies, we can introduce order into our knowledge of the terrestrial scheme of things by observing that some general systems laws hold true on all levels of its organization. All modules can be conceptualized as structures maintaining themselves amidst change, and adapting to the patterns of change by regulating whatever exchanges they have with their environment. Thus the theory of open dynamic systems will apply, and with it the concepts recently developed in cybernetics, information theory, and related fields. Recurrent factors can be observed, applying to the microhierarchy as a whole. The higher we penetrate on the scale of organization, the more openness we find in the modules. Stable atoms are closed unless disturbed by radiation exceeding their bonding forces. By contrast multicellular organisms and their ecosystems are continuously open and depend

264

Ervin Laszlo / o = H metagalaxy (^astronomical universe) \

»

Ρ Α - η ι galaxy aggregations galaxy stellar clusters stars,planets

atoms of the e l e m e n t s (H 1 - L r 1 0 3 ) e l e c t r o n s , nuclei, protons radiation quanta, / fundamental energy condensations (quarks) / s p a c e - t i m e energy and field continua

— H5(=h6)

v-H 7 . -Hi Η = macrohierarchy h* = m i c r o h i e r a r c h y ( t e r r e s t r i a l ) h n = microhierarchy (other) ' U p p e r levels a r e hypothetical.

on -their environmental transactions for their persistence. We find that the lower level modules are strongly bonded (e.g. nuclear and electronic forces bond the structure of atoms) whereas the higher modules make use of progressively weaker bonding forces (organisms have chemical bonds, populations and ecosystems bonds based on niche structure and symbiotic behavior coded by genes, and social systems bonds based on acquired and retransmitted values, mores, laws, and so on). The higher modules depend on self-regulation for

The reduction of Whorfian

relativity

265

their persistence rather than the force of the bonds. Theirs is not a structural form of stability, but a cybernetic form of ultrastability — the capacity to respond to the environment, and cope with it through adaptive changes in the systems themselves. Further, as we ascend the scale we find that the abundance of the modules decreases: There are more atoms than molecules, more molecules than cells, and so on. Indeed, the level structure forms a branching Chinese-box hierarchy, with many lower-level modules inside every higher level module (e.g. many molecules constitute cells, many cells ' multicellular organisms, and many organisms populations and ecosystems). But the decreased quantitative abundance of the systems is balanced by an increased variety of species and a correlated diversity in their manifest structural and functional features. (There are but eighty-two stable forms of atoms, but many thousands of molecules, untold millions of organic species [the roughly one billion species assumed to have existed on earth — of which 99 percent are now extinct — are but a fraction of all the species which could arise by combinations of genetic codons based on nucleic acids], and the variety of possible ecological and social forms is probably beyond human imagination.) Some species of systems may have highly sophisticated selfregulative functions: these can have been naturally selected in the course of phylogenetic evolution. In such species we find an enhanced capacity to adapt organism and environment by manipulating the latter to the needs of the former. Cooperative patterns can develop among members of local populations, selecting for a flexible mode of intercommunication. Such patterns of communication can facilitate the development of operations, of the imaginative recreation of perceived objects and events, and reflections on ultimate purposes. In sum, culture may evolve out of a reliance on a high level of self-regulation in a population of organisms. Therewith we may witness the emergence of justice, morality, beauty, knowledge, and so on. These phenomena are not strangers in the natural world, but may be accounted for in thee context of the evolution of a particular species of complex open systems. The living and the nonliving, as well as the individual and the cultural, are united through the general systems conception of persistence and development, which provides a 'vertical' linkage among the sciences specializing on the diverse levels. A statement of the specialized languages of the sciences and the humanities in the general systems metalanguage can bring about a reunification of the world pictures fragmented by adherence to one's

266

Ervin Laszlo

theoretical paradigm. Hereby the effect of Whorfian relativity within contemporary Western culture can be mitigated, and better conditions provided for communication and understanding among scientists and humanists.

4. CURRENT EFFECTS OF THE REDUCTION OF WHORFIAN RELATIONS THROUGH GENERAL-SYSTEMS LANGUAGE

The reunification of the perceived worlds of contemporary scholars through the adoption of a general systems language already bears some remarkable fruits. It should be noted, first of all, that the person who masters the general systems paradigm can readily adapt to different sciences: he can translate their theoretical languages into his own systems metalanguage and offer new insights by relating one theoretical language to another. I have had occasion to work with biologists in the past, and am currently engaged in a research project with specialists in international relations theory. I experienced little difficulty in moving back and forth among these fields, and communicating with specialized workers. Moreover insights gained in one field have repeatedly proved to have value when applied to others, when the application was not a direct reduction of one theory to another but a translation into a systems model which could be specified to simulate behavior on different levels of observation. Secondy, we should note that interdisciplinary communication occurs these days at a high level of competence through the meetings organized by the Society for General Systems Research. The Society has members coming from a wide variety of disciplines, including biophysics, international relations, computer sciences, history, philosophy, and linguistics. At the meetings of the Society the members gather to exchange information and comment on each other's work. The level of communication at these meetings (some of which I have organized) is as high or higher than at comparable meetings within the boundaries of a single discipline. It is not that the biologist acquires an instant knowledge of economics or the computer expert a knowledge of international relations. Rather, the language of the analysis frames a set of concepts that are familiar to all, and make possible the assessment of phenomena on fields different from one's own narrow domain. General systems people share a language which restructures their world pictures along isomorphic lines.

The reduction

ofWborfian

relativity

267

The communication at general systems meetings may be contrasted with that typical of interdisciplinary meetings. I participated in several such meetings as well, of which one recent experience stands out particularly clearly. The meeting was called to discuss the nature and goals of human purpose from various perspective — biological, cultural, epistemological, social, and theological. The distinguished participants included two physicists, three biologists, a neurophysiologist, an anthropologist, a sociologist, two philosophers, and two theologians. Most participants had been acquainted with one another from previous contacts, and all have read each others' published works. The entente was excellent and cordiality high. But the sessions, which lasted for six days in an idyllic secluded spot of the New York Catskills Mountains, bogged down through mutual incomprehension. The physicists failed to see what the theologians were driving at; the three biologists, of whom one was a specialist on termites, another on genetics, and the third on population dynamics and evolution, disagreed among themselves; few conceded that the philosophers had any real competence to speak on the subject, and no one understood the dualist ideas put forward by the neurophysiologist. It so happened that three of those present — a theologian, a biologist, and a philosopher — shared a general systems orientation. They had no difficulty in communicating, but were unable to convince the others of the cogency of their viewpoint. (That something of value might still come out of this meeting is possible: All participants agreed to exchange papers and revise them in accordance with feedback from the rest. Given the close personal entente among the group, there are powerful motivations t o bridge professional differences.) Let me take as my final example a meeting uniting general systems theorists who differed not only in disciplinary, but also in culturalideological, background. The occasion was the Thirteenth World «Congress of the History of Science in Moscow where a section was devoted to the history and concepts of general systems theory. Present at the sessions were a large number of Russians, an almost equally large number of Americans, and a few Europeans. Papers were given in Russian, English, French, and German. A condensed summary was provided of Russian papers in English, and of the rest in Russian. Despite obvious linguistic difficulties of comprehension, the comments were incisive and the discussion live. Moreover the sessions were prolonged into informal discussion groups and dinnertable exchanges, with a level of communication as high as that of any

268

Ervin Laszlo

disciplinary meeting in one country. Friendships were formed between Soviet philosophers and American mathematical psychologists, American sociologists and Soviet historians of science, etc., which would not have been likely to form in the absence of the general systems metalanguage, even among members of any one nation. Apart from an unspoken agreement not to raise specifically ideological questions, there were no areas in the respective world pictures which would have been too differently structured to permit real dialogue. Contacts established at that meeting issued in a number of publications of Soviet researchers in America, and of Americans in the Soviet Union, Poland, and other East European countries. The above examples were offered to illustrate the thesis of this study: that the effects of Whorfian relativity include the fragmentation of the world pictures among scientists and humanists belonging to different specialties, and that such fragmentation can be reduced through the use of a general systems metalanguage which assigns reality to hierarchic sequences of dynamic open systems from atoms to global sociocultural systems, and thus effects a vertical unification of the disciplines without reducing one to another. The general systems theorist perceives the world as constituted of systems within systems, in complex hierarchical structures with multiple feedback paths, and can select for detailed attention systems and relationships on different levels without distorting his general framework and isolating himself from investigators interested in other levels and relationships. The consistency introduced into our ordering of the knowledge systems of the empirical world through the use of this metalanguage merits further study and refinement.

REFERENCE Whorf, B. L. ( 1 9 5 6 ) , 'Language, mind and reality', in Language, thought and reality. Selected writings of Benjamin Lee Whorf, edited and introduced by J . B. Carroll. Cambridge, Mass., MIT Press.

MARC DE MEY

14

Incommensurability of Theories and Untranslatability of Languages

For more than a decade, epistemological relativism has gained some popularity among philosophers of science in a form which is remarkably comparable to some of Whorf's formulations of his well-known hypothesis. The recent relativistic movement is associated with the names of Hanson, Kuhn, Toulmin, Feyerabend, and others. If one substitutes the term 'theory' or better, 'paradigm' for 'language' in some Whorfian statements, one gets almost Kuhnian propositions. Like natural language, theories are considered never to be ontologically neutral. Theories are phrased within some 'conceptual scheme', some picture of the world, sometimes called 'metaphysics'. Like natural languages, this picture of the world has influence on perception. In Whorf's favorite expression: 'All observers are not led by the same physical evidence to the same picture of the universe, unless their linguistic backgrounds are similar, or can in some way be calibrated (Carroll 1956: 214). One would arrive at the position of Kuhn by substituting 'paradigm' for 'linguistic background'. With respect to the calibration of those linguistic backgrounds, Whorf discussed the possibility of ease of translation as an indicator for a shared 'picture of the world' and he believed in the essential untranslatability of some languages into some others, e.g. Hopi into English. Kuhn's relativism is expressed in what he calls the 'incommensurability' of theories. Apparently competing theories developed within different frameworks or different paradigms can not be compared with each other in a meaningful way. Empiricists have found this latter view to be particularly irritating. Classical empiricism somehow believes in the possibility o f evaluating theories as isolated conceptual systems having some obvious link with 'reality' through perception. Through perception any theory

270

Marc De Mey

can be verified or falsified by anyone who cares to look or to listen. Sophisticated forms of empiricism, like logical positivism in its more mature stages, no longer adhere to the view that absolute judgment of theories taken in isolation can yield a rational reconstruction of scientific development. Science does not develop by accumulation or rejection of isolated theories presented as candidates for the inventory of eternal truths. The evaluation of theories is recognized as a rather complex and subtle decision process in which several criteria may play a role, among them the context of other theories (among the other criteria are included internal consistency, simplicity, empirical support, predictive power, etc.). Although it is no longer presupposed that a theory can be evaluated on its own merits, it is commonly accepted that it is possible to compare theories with each other in order to arrive at some classification according to some values so that one can choose the 'best one' on the basis of a rational decision. This relative judgment of theories is essential for logical positivists since they want a selective principle to distinguish the scientific from the nonscientific and the more scientific from the less scientific. The empiricist tradition also emphasizes accumulation of knowledge over time and looks quite rightly for the mechanism of progressivity in science. So, if theories follow upon each other, the latter ones should be better ones. If one gives up the comparability of theories, then the absence of criteria for absolute judgement and the renouncing of relative judgment leads to complete anarchism with respect to theories. It appears as if there are no longer criteria either for distinguishing the better theories from those which are inferior or for distinguishing science from what is not science or nonsense. If, as Feyerabend proposes 'everything goes' then everything goes, astrology and parapsychology as well as biochemistry and solid state physics. The incommensurability relativists can make two remarks with respect to relative judgment of theories. The first one is concerned with the weighting of the criteria involved in the comparison. How does one determine the relative importance of criteria like predictive power, internal consistency, coordinatability with accepted theories in related fields, etc.? If predictive power should be judged more important than consistency, what do we do when we arrive at a choice between a theory which is very consistent and predicts reasonably well and a theory which is reasonably consistent and predicts very well? If one studies scientific activity in several disciplines, one

Untranslatability

of languages

271

discovers that disciplines differ with respect to the weights they assign to those criteria. If one studies their history, one discovers that the weights also vary during the 'life cycle' of a discipline (cf. Radnitzky 1973). The second remark a relativist can make with respect to comparability of theories is that one cannot know if two theories deal with the same thing. To save his face an author of an inferior theory can always find a way out of an embarassing confrontation with a superior theory by saying that the two theories do not deal with the same things, that he and his colleague study different things. This cheap way of avoiding conflict between competing theories is an inescapable experience with conflicting paradigms: they do not deal with the same phenomena. A scientific tradition or paradigm defines its own phenomena, its own problems, its own facts. What is a fact in one paradigm is not necessarily a fact in another one, notwithstanding whatever empirical support it may have. To illusstrate this point we can refer to Wason's recent history of the development of negation experiments in his own work and in psycholinguistics in general (Wason 1972). In a period of one decade we can witness the rise of a highly attractive and ingenious way of experimenting with single sentences comprising simple and complex negations, and the rejection of this approach on the basis of the discovery of the relevance of preconceptions and expectations previously derived from the verbal, nonverbal or extralinguistic context in verification tasks. Wason made many measurements and observations. Probably, if we should care to repeat his experiments, we would obtain comparable data. We would discover that in verification tasks negative sentences take longer than affirmatives, and that false negatives are faster than true ones. However, the point is, I guess, that Wason would not care so much anymore. Meanwhile for him, it became obvious through his own work and from the work of others that studying isolated sentences in a limited artificial context does not yield a good picture of what occurs when people understand natural language. If one wants to understand natural language, the study of isolated sentences seems to be the wrong segmentation of the area under investigation. One should deal with sets of related sentences and take into account the shifts of attention and preconceptions induced from sentence to sentence. This type of argument is an important point in Kuhn's theory. Science does not make progress by becoming more scientific. At any time scientists take into account a number of empirical facts in a

272

Marc De Mey

rational way. It is not that the older scientists were more irrational t h a n the present generation of scientists: they sought for empirical evidence as much as we do. Neither is there any progressivity in science in the sense that the stock of empirical findings grows larger and larger. In effect, this stock of findings is growing but an important part of that growing stock also quickly becomes obsolete. With respect to our example we can say in a Kuhnian way that, although the facts about negative sentences remain empirically valid, they no longer mean anything for a theory pretending to account for the processes which intervene in understanding natural language. They are facts in the sense that they are operationally defined and that they can be observed by anyone who follows the methodological prescriptions. They cease to be meaningful in the sense that they no longer belong to what De Groot (1961) calls a nomological network of concepts applied to a well-defined set of problems and phenomena. (Obviously, certain traditions may prefer to adhere to a view in which processing of isolated sentences is considered to be relevant for an analysis of mental processes as one can witness in several stage-oriented models.) In psychology this could be compared to the loss of construct validity. Relativists like Kuhn and Hanson emphasize the importance of meaning in terms of conceptual networks. Terms and concepts derive their meaning f r o m their multiple relationships with other terms: they are nodes in a complex network. Scientific revolutions, changes of paradigms, involve radical reorganizations of the system of interrelated concepts in a given discipline. To understand theories or statements produced within such a framework, one has to acquire the semantic network. It is not sufficient to have clear empirical references in order to have a meaningful statement for everybody. Meaning is restricted to those who are conversant with the conceptual framework. Here the parallel with translation is obvious. Problems of the indeterminacy of translation may arise f r o m the absence of knowledge of this semantic network. An interpretation of Wittgenstein's famous dictum that 'Understanding a sentence means understanding a language' points to the conceptual system represented by the structure in the lexicon of a language. Words and sentences n o t only apply to objects and states of affairs; they f o r m part of a system and they derive their meaning as much f r o m their relationships with other words as f r o m their reference to extra linguistic phenomena. (This interpretation of meaning has been recently expanded and operationalized in the development of certain 'question answering

Untranslatability

of Languages

273

systems' like Ross Quillian's [1967] and the psychology of long term memory [cf. Frijda 1972.].) The controversy with respect to this problem of uncommensurability centers around the notions of progress and criteria for ultimate solidity. If one maintains that, with respect to translation, one can always fall back upon a set of basic categories in terms of which concepts in the languages involved can be reconstructed, perfect translation can be achieved. It appears to me that in the tradition of logical positivists, the belief still prevails that it is possible to develop a universal language which could play that role. Logical positivism is an essentialistic philosophy for which in the last resort the only real things are somewhat like Machian 'sensations'. The ultimate units of knowledge have to be immutable units of experience and these units, no matter how elementary they are, become the universal categories to which every statement in whatever language should be able to be connected. Relativists give up the idea of ultimate solidity in terms of immutable units. However this does not mean that they also give up the notion of progress. Two points should stand out. First, the Kuhnian and Hansonian relativist is an empiricist. He believes it is necessary to interact with the world in order to achieve knowledge. What he does not accept is that this interaction should be confined to what is possible in terms of essentialistic ultimate units of experience which are established once and for ever. Second, although he also gives up the idea of an ultimate reality, the Kuhnian relativist believes in some kind of scientific progress. Yet the radical break which Kuhn places between successive paradigms makes it difficult for him to explain that. However, the solution proposed by Blachowicz (1971) is a solution which is as instructive to logical positivists as to Kuhnian relativists. It is a solution which has already been suggested by Laszlo (1976) and which makes an appeal to General Systems Theory in one of its most important concepts: the concept of hierarchical organization. Up to now, we have been talking about theories and paradigms. The point is that we get in trouble because we confine ourselves to only one level of analysis. Science as such, and scientific disciplines, do however form systems consisting of many subdisciplines containing schools in which one finds several theories and hypotheses constructed out of statements which in turn consist of observations and preconceptions. Science is a multilevel phenomenon and seien-

274

Marc De Mey

tific progress is not restricted to one particular level. Now, this principle of hierarchical construction is important f o r both empirical feedback and the construction of new conceptual frameworks. With respect to empirical data, there is no privileged level at which empirical evidence enters into science; rather, it is relevant on all levels — on t h e level of schools and theories as well as on the level of concepts and their operational definitions. If a paradigm is discarded, it is because of some empirical evidence — it does not lead to the solution of a sufficiently large number of problems, it does not lead to the accuracy of prediction which is wanted, etc. These are empirical observations, although they are n o t stated in a physicalistic language. With respect to the construction of new conceptual frameworks, paradigms or traditions or pictures of the world or whatever they may called, do not come into being all at once. They are assembled f r o m components the solidity of which is n o t at stake at the time of construction of the new paradigm. In terms of the semantic network, not all concepts are thrown away, only some links are cut and some new ones introduced. Kuhn refers to biological evolution as being a good metaphor for scientific development. Blachowicz has quite rightly indicated that Kuhn's theory does not live up to his own metaphor. He should take into account the hierarchical structure of biological systems and their possibility to conserve earlier achievements while experimenting with new combinations. The metaphor of biological evolution is not new, either for disciplines in science or for languages. (A very impressive and elaborate application of evolutionary concepts on scientific development is Toulmin [1972]; see also the critical review by Cohen [1973].) It leads to the view that languages or disciplines are somehow like biological species, bounded to some more or less specific ecological setting. It is a stimulating idea which, among other things implies that disciplines and languages adapt to each other as much as t o the world they are supposed to describe. For each species, the environment includes the other species. The development of knowledge is the result of a process of interaction with the world leading to a more adaptive relationship. Interaction with the world does not only involve observation and experimentation with physical objects b u t also interaction with fellow men. The contribution of the relativistic trend in today's philosophy of science consists in its emphasis on the fact that scientists submit to the reality and the feedback of their interaction with the scientific community as well as

Untranslatability

of Languages

275

to the reality and the feedback they get from their interaction with the physical world. Communication and 'empirical' research are both essential to scientific development.

NOTE

This paper is a slightly modified version of 'Absolute and relative judgment of theories', Communication and cognition, 6 (1973).

REFERENCES Blachowicz, J. A. (1971), 'Systems theory and evolutionary models of the development of science', Philosophy of science, 38: 178—199. Carroll, J. B., ed. (1956), Language, thought and reality. Selected writings of Benjamin Lee Whorf. Cambridge, Mass., MIT Press. Cohen, L. J. (1973, . . ., The British journal for the philosophy of science, 24: 21-61.

Groot, A. D. de (1961), Methodologie. Grondslagen van onderzoek en denken in degedragswetenschappen. The Hague, Mouton. 8th rev. ed.: 1975. English revised edition: Methodology. Foundations of inference and research in the behavioral sciences. The Hague, Mouton (Psychological studies, 6), 1969. Frijda, N. (1972), 'The simulation of human memory', Psychological bulletin. Laszlo, E. (1976), 'The reduction of Whorfian relatively through a general systems language', Article 13 in this volume. Quillian, R. (1967), 'World concepts. A theory and simulation of some basic semantic capabilities', Behavioral science, 12: 410—430. Radnitzky, F. (1973), 'Towards a theory of traditions in science', Communication and cognition (Ghent), 6. Toulmin, X. (1972), Human understanding..Oxford, Clarendon Press. Wason, P. C. (1972), 'In real life negatives are false', Logique et analyse, 15: 16-38.

KAREL BOULLART

15

A l'Occasion de l'Hypothese Sapir-Whorf: L'lncompatibilite des systemes Remarques generates

I. Considerant l'hypothese Whorfienne, il est necessaire de faire une distinction nette entre la version strictement deterministe et la version plus large qui postule seulement une influence plus ou moins prononcee du langage sur la culture humaine (Penn 1972: 28—32). En outre, quand on etudie le relativisme implique, il faut se demander si celui-ci s'applique a la langue comme moyen de communication, a la syntaxe, ou bien a l'information communiquee, a la s6mantique (cf. Schaff, 1976). Partant de l'idee qu'il n'est pas vrai que n'importe quel contenu peut etre formule dans n'importe quelle forme, il pourrait — dans certains cas-suffire d'etablir un relativisme semantique pour necessiter des divergences syntactiques significatives et vice versa; cele ne signifie pas qu'il soit impossible que le systeme de communication de n'importe quelle langue ne puisse etre elargi infiniment, de telle fa^on que n'importe quel contenu pourrait etre exprime dans cette langue, ce qui impliquerait que le relativisme syntactique et semantique soit seulement temporel. Ce qui, d'un autre cote, est probablement impossible, c'est la possibilite a un moment donne de l'histoire d'exprimer la totalite des informations semantiques possibles dans un seul et meme systeme de communication completement elabore. Autrement dit, la langue, dans son aspect syntactique et semantique, comme la culture humaine qu'elle exprime et qu'elle fige d'une certaine maniere, ne peuvent etre considerees comme des systemes proprement dits. Ce sont des systemes qui realisent seulement une partie restreinte des fins pour lesquelles elles ont ete con