LANGUAGE LEARNABILITY AND LANGUAGE DEVELOPMENT

提出了一个多方面的研究项目，以检验和完善的理论

A multifaceted research project is proposed to test and refine a theory of how children use the sentences addressed to them to learn the syntax of their mother tongue. The theory describes an acquisition mechanism that takes sentences plus context-induced semantic representations as input, and yields as output a set of grammatical rules defining the language's phrase structure, affixation, control and complementation, long-distance binding, and lexical productivity. The acquisition machanism, whose output rules are in the form described by Bresnan's Lexical Functional Grammar, first uses certain types of semantic information as evidence for the arrangement of specific syntactic primitives in the input, and then analyzes remaining unknown elements in terms of their ditribution within partially known structures. The theory is tested and refined in five ways: 1) examining the structure of languages other than English to see of the existing mechanism can acquire them; 2) accounting for existing data on children's language by considering them as intermediate stages of the operation of the learning mechanisms; 3) gathering new data on children's language to address questions raised by the theory, especially in studies that expose children to novel forms and then test their comprehension, production, and judgment of releted forms; 4) examining the syntactic and semantic properties of parental speech to children, to see if the model's reliance on syntax-semantics correspondences in early parental speech is well-motivated; and 5) simulating the mechanisms as a computer program, to test the robustness of the entire set of acquisition pocedures acting together.