This paper explores the questions of how and why children learn names for things. The acquisition of reference is set within a broader discussion of linguistic representation, which defines language as a system functioning in a three-way relationship among the individual language user, the world of experience, and the social community. It is argued that the essential function of human language is to allow indirect (rather than direct) reference, and hypotheses are presented on how the child's motivation, his assumptions about what language is for, and his level of linguistic consciousness determine the role of indirect reference in children's learning of names. In an experiment eliciting indirect reference, four naming strategies were hypothesized: overgeneralization of a real (but incorrect) name, frustration, description, and made-up names. Twelve Ss between the ages of 3;1 and 4;4 were presented strange and novel items in picture and object contexts and asked to name those items once the stimuli were removed. The least mature Ss used mostly over-generalized names. More mature Ss showed some frustration and used both overgeneralization and description. The most mature Ss showed an increased number of descriptions and used made-up names as well. (Author/KM)