To learn a new word, a child must create two distinct mental representations: a lexical representation, corresponding to the sound form of the word, and a semantic representation, corresponding to the meaning of the word. For the most part, acquisition of lexical and semantic representations has been studied independently. Consequently, most current word learning models focus on the acquisition of only one type of representation. Moreover, when learning a new word, these newly created lexical and semantic representations must be integrated or linked with existing phonological (individual sounds), lexical, and semantic representations. Thus, similarity between the novel word and existing representations may influence word learning, and this has not been fully captured by current models. Finally, the majority of word learning research has focused on factors affecting rate of simple associative learning, referred to here as initial mapping. Few studies have considered factors affecting the quality of the newly learned lexical and semantic representation, often referred to as extended mapping. Thus, models of word learning capture associative learning but not continuing refinement of representations.