Learning proper and common names in inferential versus ostensive contexts

A single, indirect exposure to a novel word provides information that could be used to make a fast mapping between the word and its referent, but it is not known how well this initial mapping specifies the function of the new word. The four studies reported here compare preschoolers' (N = 64) fast mapping of new proper and common names following an indirect exposure requiring inference with their learning of new names following ostension. In Study 1, 3-year-olds were shown an animate-inanimate pair of objects and asked to select, for example, Dax, a dax, or one. Children spontaneously selected an animate over an inanimate object as the referent for a novel proper name, but had no animacy preference in common name or baseline conditions. Next, the children were asked to perform actions on, for example, Dax or a dax, when presented with an array of three objects: the one they had just selected, another member of like kind, and a distracter. An indirectly learned proper name was treated as a marker for the originally selected object only, whereas a new common name was generalized to include the other category member. Study 2 showed that mappings made by inference were as robust as those made by ostension. Studies 3 and 4 demonstrated that even 2-year-olds can learn as much about the function of a new word from an indirect exposure as from ostension.