Abstract

In recent years, several explanations have been offered for the critical period in language acquisition, itself, a priori a somewhat surprising phenomenon. Two such explanations are considered here. Both studies use computer simulations, but the factors they model are very different. Hurford (1991) simulates the phylogenetic evolution over hundreds of generations of a species in which the timing of life history traits is under genetic control. The period when an individual is most proficient at language acquisition is just such a life history trait, and is capable of adaptive evolution. Evolutionary simulations lead to a concentration of language acquisition proficiency in the period up to puberty, with a subsequent tailing off. Elman (1993) demonstrates `the advantages of starting small' in neural networks learning mini-languages with many of the complex interacting grammatical factors found in real languages. A neural network which starts mature, with a full adult `working memory' cannot acquire such complex grammatical competence, whereas a net whose attention span is initially limited and then grows with maturation can acquire the appropriate grammar. This explains, in adaptive terms, the existence of a period in which an organism's characteristics, relevant to the language learning task, change, increasing a certain capacity (`working memory') from an immature to an adult value. These accounts are complementary and mutually compatible. An evolutionary account is proposed, in which genetically controlled `working memory' size in relation to life history is the variable operated on by natural selection. This account promises to produce a more detailed explanation of the critical period, which can be related to a wider range of data, including the coincidence with puberty and the involvement of sentence processing in language acquisition The relationships between Elman's `working memory' and the distinct psychological concept of working memory are also explored