Evaluating the semantic categories hypothesis: the case of the count/mass distinction.

It is often claimed that grammatical categories are initially acquired via their semantic properties. In the case of the count/mass distinction, semantic correlations should predispose the child to acquire the countlmass subcategories as a distinction between names for objects vs. substances. This proposal is tested in three experiments. The first two experiments with 3to 5-year-olds employ a word-learning paradigm in which semantic and syntactic cues are either in conflict, in accord or in isolation. Results demonstrate that syntactic cues are clearly the most effective and predominate over semantic cues as a basis for subcategorization. The third experiment with 2to Syear-olds demonstrates that children do not miscategorize nouns whose semantic properties are either inappropriate or indeterminate. Thus, for example, they do not tend to miscategorize a term such as “furniture” which is a mass noun yet denotes a class of objects. These results suggest that the count/mass distinction is not acquired via an objectlsubstance distinction although semantic properties of quantification are probably important for the acquisition process.