Pp. xiii, 304 , Oxford University Press , 2006 , $38.00.
It has taken forty years, but a reaction against Chomsky in linguistics has begun – not to the fruits of his ‘generative grammar’, which is acknowledged and applauded, but to his nominalist theory of language and non-referential epistemology. Devitt wants to pry linguistics away from psychology and disputes the ‘Cartesian speculation’ that distinguished the young rebel of the 1960's in a field previously dominated by the assumptions of British empiricism. The ‘grammar’ of a language is a linguistic reality, not a psychological reality residing in some mysterious ‘pineal gland’ or ‘language faculty’ in the competent speaker's mind. Chomsky thought there were ‘rules’ that underlay linguistic competence in any particular language, which every speaker ‘knows’ at least implicitly, and which it is the ability of the professional linguist to pull out and render explicit by laborious study. Chomsky held that the speaker somehow ‘represents’ these rules in her mind. In analogy with a computer, the mind was ‘programmed’ or ‘hard-wired’ by these rules. Devitt accuses Chomsky here of error by misdirection: there is a psychological reality underlying language, but it is thought, not rules. Or if you like, thought does indeed represent something, but it is not rules. Most competent speakers are in fact ignorant of the rules that a professional linguist can demonstrate structure their language. Clearly, the ‘linguistic intuitions’ that are the ‘voice of competence’, are not generated within the speaker by any such ‘rules’. Following H. P. Grice, Devitt holds that thought is ontologically, explanatorily, and temporally prior to language, and is co-responsible for these ‘linguistic intuitions’ that tell a speaker how to respond in a given situation, together with the brute-causal associations built up by an empirical central processor as is needed to explain any skills acquisition. He writes: ‘(T)he representation of thought is language-like … I argue that the syntactic structure of this representation is likely to be similar to that of the sentence that expresses it in the thinker's language. So, a language is largely psychologically real in a speaker in that its rules are similar to the structure rules of her thought. Linguistic competence should be seen as an ability to translate back and forth between the speaker's ‘Mentalese’ and her natural language … (H)umans are predisposed to learn languages that conform to the rules specified by the (Universal Grammar) because those rules are, largely if not entirely, innate structure rules of thought’ (pp. 11, 13). Devitt is puzzled by Chomsky's claims against linguistic regularities and conventions, against the need for shared and discoverable meaning in communication, and in favour of idiolects rather than common languages as the object of research. By contrast, Devitt argues that it is because of shared conventional meanings in a group that language can play its role of making the thoughts of each member accessible to others; it is therefore the central task of linguistics to explain the nature of these conventional meanings. A revolutionary conclusion might be that the ultimate object of the ‘thoughts’ we have - and thus for the ‘rules’ that structure our language - are ‘things’.