On the study of first language acquisition (Q1899546)

Human languages are immensely complicated systems for pairing sounds with meanings, yet most infants can acquire any one of them from a few years of sustained, casual contact. However difficult it is to give sharp expression to this phenomenon, few students of language have failed to be impressed by the child's achievement. It is thus not surprising that many questions about language acquisition have been posed and investigated, including the following questions: (a) From among the vast class of logically possible languages, which can in fact be assimilated by children on the basis of the kind of linguistic experience typically afforded the young? That is, which languages are ``natural''? (b) What is the mental mechanism upon which language acquisition rests? Question (a) requests an illuminating characterization of the class of learnable languages, for example, by formally specifying a class of grammars rich enough to represent any such language but none others. Question (b) requests information about the program embodied by the child's nervous system, the program allowing children to arrive at a stable and accurate representation of the ambient language in case the latter is of the human sort. Both questions presuppose familiar idealizations of linguistic competence that have been discussed many times. Rather than reexamine these idealizations, we intend to rely upon them for the purposes of projecting (a,b) into a recursion-theoretic framework. The framework will allow us to represent developmental psycholinguistics as a process of theory-discovery within a formally defined paradigm of empirical inquiry. It will then be possible to consider in mathematical terms the prospects for success in this enterprise. The ensuing discussion proceeds as follows. The next section reviews the definition of ``language identification from text'', due to \textit{E. M. Gold} [Inform. and Control 10, 447-474 (1967; Zbl 0259.68032)]. These definitions constitute our model of the child's discovery problem, which is to acquire a grammar for the ambient language. Section 3 formalizes the discovery problem faced by the developmental psycholinguist, namely, to understand how the child solves his own discovery problem. Section 4 is devoted to theorems bearing on the prospects of successful linguistic inquiry. It is shown that even strong assumptions about the child's learning mechanism do not guarantee that linguists can discover even an approximate description of the child's competence. Discussion of our results occupies Section 5. Proofs of theorems and propositions are relegated to the Appendix.