Explorations in semantic theory¹

Introduction

In its current surge of rejuvenation, linguistics faces opportunities long unavailable for reintegrating semantics into the range of its legitimate interests. That sounds associated with meanings are the proper objects of linguistic study has never been denied. But unlike sounds themselves, the meanings with which they are somehow paired are not physically manifest in an utterance or its graphic rendition. And so, when squeamishness about ‘ mental ’ data prevailed, particularly in America, the only official role left for the informant was that of an emitter of uninterpreted texts. Semantic material - whether it was imagined to reside in the situational stimulus, or in the speaker’s brain, or in another speaker’s overt response - was, in any case, inaccessible to observation: it was, in fact, as elusive in the case of living languages as of dead ones. Lexicography carried on in paradisiac innocence without questioning its own theoretical foundations; but for critical linguistics, no theory of meaning was on hand for semantic statements to conform with, and no procedures were in sight for testing semantic claims against finite, surveyable bodies of evidence. As for lay opinions about variant forms - what Bloomfield (1944) dubbed ‘tertiary responses’ -these were read out of linguistics altogether. ‘The linguist’s gospel’, it was said (Allen 1957: 14), ‘ comprises every word that proceeds from his informant’s mouth - which cannot, by definition, be wrong; but. . .as a matter of principle, whatever the informant volunteers about his language (as opposed to in it) must be assumed to be wrong.’

Today many linguists are breaking out of these self-imposed restrictions on the scope of their science. As if fed up with the positivism of the past century, linguists are trying out a bolder stance of much further-ranging accountability. The unedited finite corpus of physical events has lost its paralyzing hold. A concern with informant evaluations of occurring and non-occurring expressions has revolutionized syntax and has opened new perspectives in phonology. ‘Tertiary responses’ have yielded to systematic sociolinguistic analysis (Labov 1964, 1965), which is liquidating the homogeneity of dialects and the unobservable nature of sound change. The constructs which linguistics is developing in order to deal with language-users’ intuitions and attitudes are already more abstract than the constructs of conventional structuralism. When compared with the ‘underlying forms’ and ‘variables’ which are the new stock-in-trade of description and dialectology, the conceptual apparatus required for semantics no longer stands out by any glaring degree of non-objectivity.

An aroused curiosity about universals, too (cf. Greenberg 1963), presages a new deal for semantics. For decades every linguistic generalization was hedged with qualifications about the infinite variety of language; the appropriate policy with regard to the definition of ‘ language * was to reduce it to the bare bones of double articulation and arbitrariness. Today linguists are resuming their search for a far richer characterization of the notion ‘ human language ’, and it is apparent that in such a characterization, a detailed statement of the semiotic submechanisms of language will occupy a prominent place.

The fresh opportunities for semantics are, of course, matched by unprecedented requirements regarding the nature of semantic research. Semantics, too, must rise to the Chomskyan challenge of generativeness - the ideal, that is, of fully explicit and literally applicable descriptions. If a semantic theory of a language is to be held accountable for the intuitions of language users as well as for their manifest output, the range of skills for which the theory is responsible must be formulated with great care, and the nature of confirming and disconfirming evidence for theoretical claims in semantics must be determined in advance. Moreover, if semantic theory is to furnish a procedure for evaluating alternative descriptive statements, it must assure the comparability of such statements by specifying the exact form in which they are made.

In several earlier publications. I dealt directly or indirectly with the question of the form of semantic statements as it relates to lexicography. But lexicographic considerations are not the whole story: a full-fledged semantic theory must guarantee that descriptive statements will be compatible with the description of the grammar of a language in all its depth. While the above-mentioned publications were not oblivious to this question, they did not face it in its full complexity. It is the specific purpose of the present paper to explore a semantic theory which might fit into a comprehensive and highly explicit theory of linguistic structure.²

A recent attempt to achieve this goal was made by Katz and Fodor (1963). The immediate impact of their work testifies to its importance: it was quickly incorporated into an integrated theory of linguistic descriptions (Katz and Postal 1964) and became a major stimulus for fundamental revisions in transformational syntactic theory (Katz and Postal 1964; Chomsky 1965).³ In a number of ways, however, the proposals of Katz and Fodor (hereinafter KF) are unsatisfactory. Since an analysis of these inadequacies is a prerequisite to the development of alternative proposals,⁴ the first portion of this paper is devoted to a critical discussion of KF [ch. 2]. The next part [ch. 3] develops, in outline, a semantic theory which would contribute to a more satisfying conception of linguistics as a whole. The concluding remarks [ch. 4] compare the two approaches.

¹ This paper comprises the first two sections of a paper of the same title in T. A. Sebeok (ed.), Current Trendsin Linguistics(The Hague, 1966), vol. in. Page, section and chapter numbers given in square brackets after references to later sections of the paper refer to the Sebeok volume.

² In preparing this paper I have profited greatly from discussions with Erica C. Garcia. A number of mistakes were caught by Edward H. Bendix and William Labov; both of them contributed many useful suggestions for improvement - some, unfortunately, too far-reaching to be incorporated in the present version. The research on which the article is based was supported in part by Public Health Service grant MH 05743 from the National Institute of Mental Health.

³ My indebtedness to Chomsky, which is evident throughout the article, covers far more, of course, than his loan of a pre-publication version of his 1965 monograph. Since his Aspects of the Theory of Syntax was not yet in print when the present paper went to press, page references had to be dispensed with; only the most important ones were added in the proofs.

⁴ Space limitations prohibit more than passing references to the older literature.