This thesis advances the hypothesis that the child's theory of mind and metalinguistic awareness are both based on a general understanding of representation. A priori considerations lead to a definition of metalinguistic awareness as representation of language as a representational medium. Since no existing tasks tap this competence reliably and validly, three novel tasks based on the understanding of synonymy in naming situations were developed. Experiments 1 and 2 examined preschoolers' ability to produce synonyms. This associated highly with their false belief understanding (r = .73, p<.OOI and r = .64, p<.OOl, respectively) and persisted beyond a common association with verbal mental age and general production difficulties. The danger remained, however, of success through some associative strategy or failure through word finding difficulties. To avoid these possibilities, in Experiments 3 and 4 children judged the synonym production of a puppet. With these sources of error removed, association was even higher (r =.76, p<.OOI, r = .84, p<.OOl, respectively) beyond a common association with age or verbal mental age. Experiment 9 examined the ability of autistic children on a version of the judgement task to see whether their understanding of mental arid. non-mental representation was also related. Results were suggestive of a relationship, but inconclusive. The possibility remains that normal children may represent form in a nonrepresentational way. Experiment 5 and 6 showed that although even very young children could recall synonyms verbatim, most preschool children deny that one of the synonyms applies. I argued that children assume that categories, not words, are mutually exclusive. Experiment 7 showed a similar rejection effect for hierarchical terms. In Experiment 8, more metalinguistic terminology aided only younger children to accept both words, consistent with the assumption that the use of two "is a" phrases prompts children to employ their category mutual exclusivity assumption. Finally, the synonym judgement task was modified for use with autistic children to test the theory that autistic children have general difficulties understanding representation. Results were inconclusive, although they suggest that autistic children have similar difficulties with the false belief and synonym tasks. The overall conclusions are-that metalinguistic awareness and theory of mind have a common basis in representational understanding, but that prior to this children can employ the form of language to make judgements about category membership.