The input–output relationship in first language acquisition

Abstract

This study provides an account of the distributional information and the production rates in a particularly rich corpus of German child and adult language. Three structural domains are analysed: the parts-of-speech distribution for a coded corpus of circa one million words as well as the internal constituency of 300,000 noun phrases and almost 200,000 verb phrases. In all three domains, the distribution over time in the adult input is extremely homogenous. The child shows a steady approximation towards the adult distribution. It is argues that two notions of acquisition have to be distinguished: acquisition in terms of the availability of a given structure, for example in terms of first occurrence of a structure or according to various criteria of productivity, and acquisition in terms of full communicative competence, i.e., using structures in the way adults use them (cf. Slobin, 1991, 1997). The data presented here show that the child acquires not only the structural options of German but also highly conventionalised ways of encoding concepts. The amount of information about the structure and conventions of German that is available in the input has the potential of making innate stipulations unnecessary. Instead, the data support usage-based and probabilistic theories of language and language processing.