ABSTRACT According to a currently influential view, children begin language acquisition by mapping linguistic forms onto concepts previously established through nonlinguistic cognitive development. The project will evaluate this hypothesis by comparing the early semantic categorization of topological spatial actions (e.g., PUT IN/ON) by children learning English, Korean, and Dutch. These languages classify spatial actions in strikingly different ways for purposes of linguistic expression. However, to the extent that children rely on nonlinguistic concepts rather than analysis of the linguistic input for the meanings of words, their spatial semantic categories should be similar, or at least should not differ systematically according to language. Two cross-sectional studies will be carried out with 80 children in each language (age range 1;0-3;6). In the production experiment, children are encouraged to tell the experimenter how to spatially arrange objects. The comprehension study assesses sensitivity to language-specific semantic distinctions with the "preferential looking paradigm". The data will be analyzed (1) to determine how early children learning each language develop language-specific principles of categorizing spatial actions, and (2) to aassess, through analysis of correct and incorrect patterns of generalizations, the relative difficulty for children of alternative principles of spatial classification. Our conclusions should enable us to characterize more precisely than is now possible the nature of the interaction in early language acquisition between nonlinguistic categorization predispositions and children's sensitivity to the semantic categories of the input language.
