Page 258 - Encyclopedia of Philosophy of Language

Reference
vertebrates have the same truth-conditions in virtue
of the fact that all and only renates are cordates, it by no means follows that the explanation of the two sentences having that truth-condition is the same. So a theory of meaning focused on truth-conditions need not count the two sentences as semantically equivalent.
Of course, this assumption rests on fallible and con- troversial ideas. It is by no means universally agreed that a basic function of language is the representation of the speaker's world, nor that representation is best understood by appeal to truth. It is quite widely argued that an epistemic notion, verification, is the central element of sentence meaning. The logical pos- itivists thought that the meaning of a sentence is deter- mined by its method of verification, that is by the conditions under which it would be rational to accept the sentence into one's belief system, and sophisticated versions of this view are still extant.
2.3 Decompositional and Atomic Theories of Meaning
The ambitious program develops a 'decompositionaF theory of meaning, a theory that explains a sentence's meaning by appeal to the semantics of the words it contains and its structure. The argument for a decompositional theory is simple and strong. Natural languages are productive, that is, they are unbounded. There is no longest sentence of English. We can add to any sentence without turning the meaningful into the meaningless. Of course, as sentences get longer, they become harder to keep track of both in pro- duction and comprehension. So they become less com- municatively salient. But there is a slow fade out of intelligibility, not a line between sense and nonsense. Moreover, languages are systematic. We do not learn to speak a language sentence by sentence. To acquire a new word is to acquire the capacity to speak and understand a range of new sentences; the same is true of mastering a new syntactic construction. In learning a language we learn words one by one together with techniques for combining them into more complex constructions. These facts suggest that a theory of meaning should consist of three parts. One is a theory of word meaning; on this view word meaning is the
most fundamental semantic property. A second is a theory of sentence structure; syntax tells us how words are organized into phrases, and phrases into sentences. The third part is a theory of 'projection rules.' These tell us how the meanings of the words of a constituent and its structure explain that constituent's meaning.
A theory of meaning conforming to this pattern is an 'atomic' theory; a theory which takes the reference of a term to be its key semantic property is 'referential atomism.' Most agree that languages are productive and systematic (though some of those developing 'connectionist' models of human cognitive archi- tecture deny this), but many deny that a theory of meaning should be atomic. Defenders of molecular
theories of meaning, for example, take the sentence to be fundamental. Theorists as different as Davidson, Dummett, and Grice have resisted the inference from productivity to atomism. They do not deny that words have meaning. Instead they take the meaning of a word to be derived from the meaning of sentences in which it appears (though it has proven difficult to give an explicit account of this derivation). Word meaning does not explain sentence meaning but is derived from it. This approach is supported by the fact that the most interesting semantic properties are properties of sentences. The units of linguistic communication and representation seem to be sentences. Even one word utterances like Idiot! are best construed as sentences, not bare words; they seem to have truth values.
It may well be true, as Davidson, Quine, and Dum- mett in their differing ways have argued, that sentence meaning is more evidentially basic than word mean- ing. Davidson, following Quine, has urged that con- fronted with the speakers of an untranslated language, we have some reasonable hopes of determining which sentences they assent to, which sentences they 'hold true,' and that this is the empirical basis of a theory of meaning. One could quarrel with the conception of Quine and others concerning the evidential resources for the semantic theory. But more importantly, the atomists' case is that word meaning is explanatorily fundamental, not that it is evidentially fundamental. There is no doubt that the physical behavior of macro- scopic objects has been, and is, evidentially fun- damental in the development of physical theory, but talk of middle-size objects traveling slowly is not the most fundamental level of explanation in physical theory. Similarly, the evidential primacy of facts about sentences establishes nothing about explanation. A semantic theory which reduces all other semantic notions to word meaning may get its crucial con- firmation from facts about sentences.
2.4 Naturalized Semantics
The final enabling assumption of the ambitious pro- gram is that we should look not just for a semantic theory but for a naturalized semantic theory. The 'naturalist' demand derives from an epistemic idea and a metaphysical idea. The epistemic idea is that philosophical theories enjoy no special status. They differ from the theories of the natural sciences in important ways; for example, they are less precisely formulated and much harder to bring into contact with data. But they share with the sciences their fallible and provisional character. So semantic theory is empirical and provisional. The idea is that semantic theory is a protoscience that will one day jettison its prefix. The further commitment is that semantic theory be physicalist. It is notoriously difficult to for- mulate precisely the requirements of a physicalist ontology, and its bearing on the 'special sciences' (the social sciences, psychology, linguistics, and biology).
236