Page 26 - Encyclopedia of Philosophy of Language

Introduction
to suppose that Davidson simply defines meaning as truth-conditions and leaves it at that. The interest of his account lies in the way he applies the insights of the idealized semantics (of artificial languages) to actual languages. Sentences of actual languages, unlike those of artificial languages, already have appli- cations to objects and states of affairs in the world and this imposes an empirical constraint on the acceptability of the theorems of the semantic theory.
To develop this idea Davidson follows Quine (1960) in postulating a fictional situation of an interpreter attempting to translate a native language totally unknown to him, spoken by people about whom he knows nothing. This is the situation of 'radical interpretation.' The radical interpreter's task is to assign truth-conditions to native sentences without presupposing an understanding of them. This can only be done by observing the linguistic and other behavior of the native people and attempting to correlate sen- tences (utterances) with salient features of the occasions of their utterance. The overall aim is to be in a position to issue reliable formulas of the kind sentence s. is true if and only if ]}. Of course several assumptions are required for this: one is that speakers intend to speak the truth and believe that what they say is true, another is that their sentences actually are true, for the most part. The former is a general assumption of rationality, the latter a 'principle of
charity,' as Davidson describes it. These assumptions play a profound role in Davidson's epistemology and philosophy of mind and have been widely debated (e.g., Grandy 1973; Evans and McDowell 1976). For Davidson, attributing beliefs to speakers and mean- ings to the sentences they utter are closely inter- connected. Although he holds that sentence meaning cannot be reduced to nonsemanticconcepts (like belief or intention) he does suggest that the constraints on radical interpretation provide the best available philo- sophical elucidation of meaning.
One of Davidson's initial principles in developing a theory of meaning is the innocuous sounding dictum that 'the meanings of sentences depend upon the meanings of words.' The principle is important for a number of reasons. If a language is to be learnable, the stock of words in the language—its vocabulary— must be finite; thus there can only be a finite number of axioms specifying the meanings of these basic com- ponents. Also, the principle articulates what many have seen as a fundamental premise in semantics, the compositionality of meaning, which has been more precisely formulated as follows: 'The meaning of a compound expression is a function of the meanings of its parts and of the syntactic rules by which they are combined' (see Compositionality of Meaning). The compositionality principle lies at the foundation of attempts to provide a formal semantics (see Formal Semantics) for natural language and is especially prominent in the work of Richard Montague (see
Montague Grammar). However, it is by no means unproblematic; semantic primitives have to be ident- ified and classified, they have to be assigned meanings, which must not be dependent on the meanings of whole sentences, combinatorial rules have to be defined, and so on. The idealized artificial languages of logic hold out a tempting model for how this might be achieved, and how it might look, but in application even to fragments of natural language the difficulties are formidable.
One issue that has been much debated in the context of Davidson's semantic theory concerns the impli- cations for what a competent speaker knows in know- ing a language. In principle such a speaker knows the meaning of any given sentence from a potentially infinite set which can be generated by the language. As we have seen, Davidson takes it as a constraint on the learnability of a language that a 'theory of meaning,' in his sense, be finitely axiomatized, though that has been challenged by Schiffer (1987). The ques- tion remains as to the explanatory value of a semantic theory of which competent speakers have no con- scious knowledge (the same problem arises for syn- tactic theories (Chomsky 1986)). One suggestion is that the knowledge is tacit (Evans 1985), another that semantic theories are rational reconstructions (Wright 1986).
2.2 Meaning and Communicative Intention
The question of what speakers know who know a language invites a deeper inquiry into the very basis of linguistic meaning. While formal semantic theories—at least those of a broadly compositional nature—attempt to assign meanings to sentences on the basis of the meanings of their component parts, a rather different philosophical enterprise is to try to explain what kind of fact it is, about people's behavior, mental states, the life of a community, etc., that makes it true that a word or expression or sentence has the meaning it does, or indeed has any meaning at all. It is questions of this kind that concerned Wittgenstein and also Quine, the latter developing an essentially skeptical view, on a behavioristic base, to the effect that what facts there are radically underdetermine hypotheses about meaning (see Indeterminacy of Translation).
Perhaps the most systematic attempt to explain meaning in nonsemantic terms comes from the pro- gram of H. P. Grice (1957, 1968, 1969; and others collected in 1989), developed by Schilfer (1972), Bennett (1973, 1976), Strawson (1964), and Searle (1969, 1983). Grice began by distinguishing 'non- natural meaning' from 'natural meaning,' the latter, which he identifies only to set aside, being of the kind Those clouds mean rain. Linguistic meaning, Grice argues, belongs in the genus of nonnatural meaning which includes a wide range of communicative behavior. The bedrock for Grice's account of non-
4