Le langage est-il un instinct? Une critique du nativisme de Chomsky à Pinker

International audienceIn this paper, I review the main arguments which purportedly establish that the language faculty is innate. In its modern guise, this linguistic nativism goes back to Chomsky, whose views on the matter I try to set in their intellectual context. I turn next to his famous stimulus poverty argument, that is, to his claim that what a child knows or comes to learn about her language far outstrips her linguistic experience. Chomsky’s arguments, I claim, hold up only insofar as one accepts his commitment to a formal (or configurational) view of grammar, and his downplaying of semantic clues. Thereafter, I discuss the evidence gathered by Pinker in his Language Instinct. I devote special attention to the various cases of language “invention” by children who (so is claimed) have been deprived of a normal linguistic input. In particular, I review the evidence which pertains to the creation of sign languages and creoles. It is pointed out that Pinker does not simply reiterate Chomsky’s theory of language acquisition. He has his own views on acquisition, which are elaborated in other works (notably in Language Learnability and Cognition). These views, I argue, run afoul of some of the arguments put forth in The Language Instinct. They entail a significant departure from Chomsky’s ideas and could be taken as supporting a non modular view of language acquisition. Another type of nativist argument concerns the notion of critical period, which would reflect the maturational process of the language faculty. Contrary to what Pinker says, there is no strong evidence that the critical period corresponds to the maturation of a domain-specific faculty. Last, I review the data from pathological dissociations of language and cognition (Specific Language Impairments and Williams Syndrome). The facts are more complex than Pinker claims, and his conclusion that language is dissociated from cognition because it is innately specific appears to be dubious. Thus, the hereditary component of SLI may not be linked to a gene specifically coding for “grammatical” abilities. In conclusion, nativists’ positions go far beyond anything warranted by the data to hand, and their own presentation of the data is heavily skewed in favor of these positions