Page 170 - Encyclopedia of Philosophy of Language

Truth and Meaning
native talk solely on the basis of his observations of native assent and dissent to stimuli. Quine's claim is that the behavioral evidence, however extended, underdetermines the translation of individual sentences, and, since this evidence is all there is to attributions of meaning, their translation is radically indeterminate. Nonetheless, relative to one of the empirically acceptable schemes of translation of the native language, translation of single sentences can be achieved.
Quine's conclusion seems no less problematic than the reductionist one that he sought to avoid. For the indeterminacy thesis cannot be restricted to alien languages: it must begin at home, if at all. Yet, one cannot easily acknowledge that the meanings of one's own utterances, including those in which the inde- terminacy thesis itself is formulated, are radically inde- terminate. To avoid this conclusion, Dummett has argued that Quine's semantic holism should be rejected in favor of a 'molecular' conception of mean- ing according to which meaning is given by a deter- minate specification of the kinds of evidence relevant to the assertion or denial of sentences of various types. This position implies that one can give a nonholistic account of the evidence which warrants the assertion of sentences containing theoretical terms; not sur- prisingly, this denial of the Duhem-Quine thesis is much contested. But a more pressing question is whether Dummett avoids the reductionism that Quine's holism enables him to escape.
Dummett's answer, which he connects with Wittgenstein's thesis that meaning is use, is that the link between meaning and evidence concerns the 'assertibility-conditions' of sentences rather than their truth-conditions (Dummett 1978). One side of this is straightforward: assuming that there is an intrinsic link between meaning and truth-conditions (a point which is revisited below), the characterization of the existence of some type of evidence as the truth- condition of a sentence implies that the sentence merely means that there is evidence of that kind. What is less clear is how it helps to describe the evidence as the sentence's assertibility-condition. The explanation suggested by Dummett's self-description as 'anti- realist' is that meaning is fundamentally an epistemic matter, so that an account of the meaning of a sen- tence is based upon an account of the types of evidence which warrant its assertion or denial. As long as it is then held that such an account is not an account of its truth-condition, the reductionist conclusion is avoided. But there still remains the question of how its truth-condition is determined. One alternative (advocated more clearly by Peacocke (1986) than by Dummett himself) is that these types of evidence suffice to identify a distinct possible state of affairs as the sentence's truth-condition. Another, more radical, alternative is just to deny that an account of meaning should yield an account of truth-conditions. On this
view meaning, as traditionally conceived, is a 'realist' illusion.
This radical alternative is one which Saul Kripke has suggested should be embraced anyway in the light of Wittgenstein's 'rule-following' argument (Kripke 1982). As reconstructed by Kripke, this argument starts from the thesis that there is nothing in a speak- er's use of language, in either past practice or linguistic dispositions, which determines how words used by that speaker should be applied to future cases: even if there is a natural rule of projection ('green' has previously been applied only to green things, so on the next occasion it will be correct to describe some- thing as 'green' just in case it is green), there is nothing in principle to rule out a deviant rule ('green' has previously been applied only to green things, but on the next occasion it will be correct to describe a thing as 'green' just in case it is blue). Yet, the argument continues, the traditional concept of meaning implies that the assignment to a sentence (e.g., 'grass is green') of a definite meaning (that grass is green) prescribes in advance under what conditions the sentence is true (namely, when grass is green). Hence, Kripke con- cludes, since this traditional concept cannot be grounded in the use of language by speakers, it should be rejected as illusory, and instead one should embrace the antirealist account which limits itself to an external description of the practices of speakers, including their practices of correcting each other.
It will be obvious that this 'skeptical' conclusion is even more paradoxical than that of Quine, which it extends. In Wittgenstein's writings, the rule-following considerations are associated with his criticisms of any Lockean approach which attempts to ground meaning on introspective psychology; hence Wittgenstein emphasizes the importance of 'blind' rule-following, assertion unguided by introspective directions (Witt- genstein 19S3). One can accept this, but question Kripke's extension of the argument in one of two ways, depending on one's attitude to the relation between psychology and semantics. Those who hold that semantic facts are irreducible to psychological ones welcome Kripke's negative argument, but hold that it does not follow that meaning is an abstract illusion; instead, they hold, people's lives—their reasonings and actions—are permeated by an involve- ment with meanings which does not need to be grounded in anything else. By contrast, those who accept the reducibility of semantics to psychology hold that once a description of the use of language includes both its social context and the underlying causal facts concerning human psychology, it is poss- ible to understand how meaning does not transcend physical fact.
4. Truth-conditionsandMeaning
Donald Davidson's position exemplifies the first alter- native (Davidson 1984). Like Quine and Dummett,
148