I just discovered Michael Tomasello’s scathing 1995 review of Pinker’s The Language Instinct (1994). Anyone interested in the nativism vs constructivism debate will find this worthwhile. I happen to think Tomasello comes out the winner in this fight. I highly recommend his book The Cultural Origins of Human Cognition (1999). I just started his newer book Constructing a Language: A Usage-Based Theory of Language-acquisition (2003) and I am really impressed so far. He is a phenomenal researcher and his theories are major paradigm busters. Here’s a snippet from his review of Pinker:
Language is not an instinct. In the common understanding of both scientists and laypersons alike an instinct is a behavioral competency, or set of behavioral competencies, that: (a) is relatively stereotyped in its behavioral expression, and (b) would appear in ontogeny even if an individual were raised in isolation from its species-typical set of experiences (e.g.,Birney & Teevan, 1961). But language does not even remotely fit with either of these two criteria. “Language” has several thousand distinct variants in the human species that are fundamentally different from one another, including in their syntactic conventions, and an individual human being can acquire any one particular language only in the context of several years of particular types of linguistic experiences with other human beings. Why, then, has Pinker chosen to apply the term instinct in such a clearly inappropriate manner? The answer is that what Pinker and his fellow Chomskyans mean by the term “language” is not what is normally meant by that term. They do not mean the communicative conventions of the speakers of particular languages such as English, Turkish, or Warlpiri. What they mean is something called Universal Grammar, which is the supposedly species-universal computational structure of language that is, in their view, wholly unlearnable (ironically, the central thesis of Learnability Theory).
In this brief essay I argue three things. First, I argue that although many people bandy about rather loosely the notion of an innate language module, the only theoretically coherent version of such a module that has ever been proposed is that of Generative Grammar-in which resides a priori the theoretically-specific linguistic structures of Universal Grammar. This and only this is Pinker’s “language instinct.” Second, I argue that this view of language and its development, though coherent, is wrong. All of the most important lines of evidence that Pinker’s new book adduces for an innate Universal Grammar are also compatible with a less rigidly nativistic view of language acquisition in which there is a biological foundation for language, just not in the form of specific linguistic structures preformed in the human genome. Finally, I argue that there is an alternative linguistic theory, or group of theories, that should be especially attractive to cognitive developmentalists because they are much more compatible with what is known about development in other domains of human cognition.