26

p For a long time it was considered in linguistics that an exclusive orientation to the systematic aspect of natural language, a rejection of any references of a semantic order (references that signified, it was presupposed, a dangerous departure from the system of language into vague, extralinguistic realities difficult to define) was the sole correct scien- 27 tific strategy (it was most consistently defended in structural formal linguistics by de Saussure, Bloomfield, Harris, Fries, and Chomsky in his early works). Doubt in the reliability of semantic studies (though even possibly with recognition of their need) mainly concerned one point, namely whether it was possible at all to indicate the semantic structure of a linguistic expression as objectively, rigorously, and exactly, for example, as its syntactic and phonological structure. But the incompleteness of linguistic description (when the semantic aspects remained outside the structures ascribed to expressions) and its inadequacy (when one and the same structure was ascribed intuitively to linguistic expressions that were different from the semantic standpoint) inevitably prompted researchers to look for ways of realising the programme mapped out in his time by Carnap, and inevitably led to rehabilitation of semantic studies.

p The first steps towards realising this programme were made in the early works of Noam Chomsky, the founder of the transformational-generative theory of natural language. As an example of the difficulties entailed in partial rehabilitation of semantics in formal linguistic models of language, one can point to the interpretation of the concept ’grammatical’ in regard to the sentence ’Colourless green ideas sleep furiously’, which later became standard. In the early work of Chomsky and others the question was put as follows: should this expression at least be considered grammatically correct even if it is nonsense? If it was not the one or the other, in what way then did it differ from the sentence ’Interesting new ideas seldom arise’?^^18^^ The opinions of logicians, philosophers, and linguists were not unequivocal on this point. The linguist Roman Jakobson, for instance, thought such expressions grammatically correct but semantically false. In the opinion of the logician and linguist Paul Ziff such sentences were not false, but were not meaningless.^^19^^ The logician Curry considered them to be on a ’different level of grammaticalness’,^^20^^ with which the logician and philosopher Hilary Putnam agreed, though he did not find them logical,^^21^^ while the logician Willard Quine would class them as trivially false propositions.^^22^^

p The evolution of the views of Chomsky himself is re- 28 arkable in the matter of definition of the place of semantics in the investigation of natural language. Thus, in his Syntactic Structures, the sentence cited above was considered grammatically correct because he supposed at that time that semantic considerations had no relation to linguistic analysis.^^23^^ Later, and especially in Aspects of the Theory of Syntax, in which he developed a ’standard theory’ of transformational-generative analysis (see Ch. II, 1), the sentence considered was already qualified as infringing the norms of grammar.^^24^^ The point is that his conception of grammar then also included a semantic component. Although Chomsky never undertook to build a semantic theory of natural language proper on the basis of transformational-generative grammar (that was done later by Jerrold Katz on the basis of Chomsky’s theory^^25^^, he no longer identified the ’grammaticalness’ of the sentence with its syntactic correctness.

p In subsequent studies that represent the period of ’ extended standard theory’, the realia of natural language were related more and more broadly to the semantic aspect, including certain characteristically ’surface forms’ or ’ surface structures’ of linguistic expressions (see Ch. II, § I),^^26^^ But having recognised that construction of a semantic theory adequate to the intuition of a speaker of natural language was beyond the possibilities of linguistic theory, since it called for full allowance for extralinguistic knowledge or the speaker’s knowledge of the world, Chomsky formulated a ’revised extended standard theory’, according to which formally described structures could be given a full semantic interpretation only when they were tied up with other non-linguistic cognitive structures, including the beliefs, intentions, etc., of speakers of natural language.^^27^^

p The construction of this theory can be characterised simultaneously as a search for a limit of the semantic claims of linguistic theory and as recognition of the significance of the factor of extralinguistic knowledge for building an adequate semantic theory.

p The problem of the meaning of expressions of natural language had already been treated long back in both the Soviet and the Western logico-philosophical literature as a 29 problem of accepting one of the possible answers to the question ’what is meaning?’ A very general conception of meaning (in the sense of determinacy) was associated with each of the proposed answers.

p The problem of meaning was discussed particularly intensively in connection with the exceptional role ascribed to natural language as an object of philosophical analysis in neopositivist and linguistic philosophy. One must note in that connection such different methodological consequences of philosophical interest in the analysis of natural language as the transition from its supposed structure to the structure of reality, to the structure of the brain, consciousness, culture, and history, and in general to the structure of ’human phenomena’. If these theoretical extrapolations were not always formulated as philosophical doctrines, they functioned at least as specific conceptions of the analysis of these phenomena, and special trends of investigation subsequently took shape on their basis.

p In that connection it must also be noted that the problem of meaning (by analysis of which the main trends of modern logico-philosophical studies of natural language have been determined) is not discussed just in analytical philosophy. The studies carried out in the context of French structuralism, for instance, deserve particular attention. One should indicate first of all the work of the French semiotician A. J. Greimas and his school, in which the problem of meaning is examined in a broard socio-cultural aspect —as a problem of the discovery of the semantic structure of various systems of signs and symbols and their interaction.^^28^^ These studies call for special consideration.

p Before going into the characteristics of the separate conceptions of meaning, let me state a few other considerations of a methodological order about the tasks of formal semantic analysis of natural language put forward in them.

p Formalisation of the concept ’meaningful expression of natural language’ in the descriptive and explanatory aspects is usually indicated as the final guideline of formal analysis of natural language. It is a matter of constructing a decision procedure for understanding the concept ’well-formed expression of natural language’ that will serve not only 30 as a description of given expressions but also for predicting any new correctly constructed ones both syntactically and semantically. What is intended is a pointer to a definite procedure that will make it possible to determine the correctness of the construction of a linguistic expression in a finite number of steps. The point concerns bringing out the logical or semantic structure of linguistic expressions from which their semantic properties and relations can be determined. Such a formalisation of natural language is a necessary precondition of its automatic, machine processing, pursuant of such extremely important aims in practice as automatic search for information, machine translation, the creation of systems of ’artificial intellect’, and so on.

p As has been shown above, it is not enough for this to point out the rules of the syntactic combinability of the lexical units (as a definition of the concept of ’a syntactically ; well-formed expression of natural language’), which represent (in terms of traditional grammar) the categories of noun (substantive), verb, adjective, adverb, etc. (Thus the sentence ’The new virtue is bald’ does not differ syntactically from ’The new shirt is dirty’). (1) Expressions of natural language that have one and the same syntactical or, possibly, even phonetic and lexical structure, may differ in meaning, and vice versa. (2) One of them may be perceived as meaningful and the other as meaningless. (3) Expressions meaningful in themselves may not form a meaningful text, especially in view of the possible ambiguity of its components.

p At the same time (and this is no less important) the point concerns the theoretical reconstruction of knowledge, along with a formal description of the language, that pursues certain practical aims, a reconstruction through which the speaker of the language correlates acoustic or visual signals (as expressions of the language) with information coded in them (as the meaning of these expressions). This refers to the bringing out of the knowledge by which a person understands expressions and so determines their meaningfulness, ’ascribing’ some meaning or other to a word, sentence, or text, establishing some semantic link or other between them, and so noting that some of them describe the 31 world as it actually is, or will be, could be or, on the contrary, could not be, that the truth of some expressions follows from, or only presupposes, the truth of others, that some one expression is suitable or, on the contrary, unsuitable for performing a certain speech act (statement, question, order, vow, promise, etc.), that there are such- and-such semantic relations between expressions, of the same or different natural languages and so on.

p The task of semantic theory consequently includes establishing those principles and rules that would explain the capacity of a speaker of a language to understand its expressions. In that connection a precept is accepted as the initial methodological premiss that both the natural language and the knowledge are amenable to rational analysis and can ultimately be characterised by clearly formulated principles.

p In other words, the task of semantic theory includes explication of what is called the ’linguistic intuition’ of a speaker of natural language, by which is understood his unconscious or not quite conscious knowledge of the principles in relation to which the researcher builds a theoretical model in the form of a system of clearly formulated statements. When it is said, for example, that a speaker of natural language has an intuitive understanding of the grammatical correctness of linguistic expressions, one has in mind (as Quine has noted) that he usually distinguishes a grammatically correctly constructed expression from one incorrectly constructed in that respect ’by intuitive judgment.’^^29^^ In that case criteria of the grammaticalness of expressions must be explicitly formulated in the theory that models this capacity, in the shape of rules that generate only correctly constructed expressions of the language. These rules are a certain formal analogue of the ’mechanism’ postulated in the speaker’s natural language that accepts its correctly constructed expressions and rejects incorrectly constructed ones.

In view of the foregoing it is a matter of explication of linguistic intuition, and of looking for criteria of the adequacy of the theory in accordance with it, or the possibility of verifying it empirically, It is thus a matter of a special 32 situation in which linguistic intuition is not only the object of explication but is itself the criterion of its adequacy, which in turn depends essentially on the conception of meaning that is proper to a concrete semantic theory.