From Simple Input to Complex Grammar

Abstract

The paradox of how children regularly learn highly complex natural languages upon limited exposure to simple data lies at the center of any study of language acquisition. This book explores a new and important hypothesis for how young children might be able to learn a language from very simple sentences.The Bracketed Input Hypothesis suggests that children may receive information from their input about how the words of sentences group together to form the basic phrases from which sentences are constructed. Such information, in conjunction with the child's native and acquired language learning abilities, Morgan argues, forms part of the necessary basis for language acquisition.From Simple Input to Complex Grammar first reviews the empirical and mathematical literature on language learnability, particularly Kenneth Wexler and Peter Culicover's influential work (Formal Principles of Language Acquisition, MIT Press 1980) which provides a basis for Morgan's hypothesis.The book shows that, assuming information about phrasal structure to be available in input, one can prove that the child can learn a natural language from extremely simple sentences, sentences of the type that children actually hear. This learnability proof, by virtue of requiring both less complex input and fewer assumptions concerning the child's native endowment, represents a significant advance beyond previous results.Morgan's own work on the nature of mothers' speech to children is then discussed, indicating that requisite phrase bracketing information is present in such speech. Moreover, his study of children's representation of speech suggests that children are indeed sensitive to such information.James L. Morgan is Assistant Professor, Institute of Child Development, University of Minnesota. From Simple Input to Complex Grammar is included in the series Learning; Development, and Conceptual Change, edited by Leila Gleitman, Susan Carey, Elissa Newport, and Elizabeth Spelke. A Bradford Book.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 270 citations exist upstream.