What are the basic mechanisms that make language possible? Is the kind of grammar that a child learns completely arbitrary, or is it constrained by biology? In this section we address these questions by describing Noam Chomsky's influential arguments about language learning.

In the 1950's, a major debate raged between American psychologists who believed that language could be explained by the patterns of reinforcement that children received for their verbal behavior and psychologists who believed that behaviorist accounts could never satisfactorily explain language. B. F. Skinner and Noam Chomsky were the primary representatives of the two camps. Chomsky's scathing review of Skinner's book Verbal Behavior is commonly seen as the beginning of the "cognitive revolution" in psychology. In that review and in later works, Chomsky argued that several properties of language--primarily its "deep structure" and the fact that it can be used to generate an infinite number of different behaviors (sentences)- -made it impossible for the child to learn according to the rules of behaviorism. The fact that children did learn to speak, and with apparent ease, was a strong argument that 1) children were genetically endowed with biological structures that aided language acquisition and 2) behaviorism was wrong.

The "deep structure" of language is its basic grammatical categories and the ways in which they interact in particular sentences. Chomsky's argument against behaviorism was based on the fact that deep structure, while crucial to the correct use of language, is not directly perceivable. Chomsky called this the "poverty of the stimulus," by which he meant not that children receive very little linguistic input, which is untrue, but that no matter how much language children heard they could not learn it according to behavioristic rules. Even if children were to receive perfect feedback on their use of grammar--which is far from the actual case--it would be logically impossible for them to extract language's deep structure unless they were equipped with some innate constraints on what that structure could be.

As a solution to the problem of the poverty of the stimulus, Chomsky suggested that each human child is equipped with a "language acquisition device" (LAD) that drives the child to acquire language and constrains the kinds of grammars that the child can learn. Evidence for something like a LAD comes from several sources: 1) children who are raised without a complete language develop a grammatical, full language of their own (such as Creole languages); 2) some family lines appear to have genetic impairments in the acquisition of grammar although the rest of their mental faculties are intact; 3) children with Williams syndrome are skilled at language despite often being severely mentally retarded; and 4) across cultures, languages seem to share a wide range of basic grammatical features.

This evidence is convincing but not conclusive. The original claim that there was a dedicated module for language acquisition has been tempered somewhat in recent years, for several reasons. First, closer studies of people with specific language impairments or specific language enhancements (such as in Williams syndrome) have shown that other aspects of cognition are invariably affected as well. Language is part of a large system of cognitive abilities, and it is likely that many of the aspects of the mind that make language possible also make non-linguistic aspects of thought possible as well. Second, children receive a wide variety of linguistic and non-linguistic input in their years of language acquisition that make the task of acquisition, although still formidable, less difficult than it may seem at first. The existence of some complex and powerful learning mechanisms is obviously necessary, but their specificity to language is not clear.