A child acquiring language must determine the syntactic reflection of linguistic categories, such as nouns (e.g., dog), noun phrases (e.g., Paul, my dog), adjectives (e.g., yellow), and verbs (e.g., """"""""run""""""""), in the language he or she is acquiring. How do children bootstrap into the system at work in their own language? It has been proposed that children assume a relationship between linguistic categories and semantic categories, for example that nouns tend to refer to kinds of individuals, noun phrases to particular individuals, adjectives to properties, and verbs to actions (Macnamara, 1982; Pinker, 1984). But the reference of nouns and noun phrases overlap. How do children who have yet to learn the syntactic reflection of nouns and noun-phrases induce the correct meaning of a these words? We propose that children make one of two possible assumptions about word meaning early in word learning. One possibility is that children assume that a novel word used in the presence of a novel object refers to that particular object unless given further evidence to the contrary (e.g., a second object of the same kind is referred to with the word). The second possibility is that children assume that a novel word used in the presence of a person or other animal refers to the particular person or animal, but that a word used in the presence of objects of other kinds is a count noun. In the first series of experiments, these two possibilities are tested to determine if either characterizes early word learning. In the second series of experiments, evidence is sought for the latter possibility across different types of noun phrases in children who have already acquired the relevant syntax of English.
