The social-pragmatic theory of language acquisition proposes that children only become efficient at learning the meanings of words once they acquire the ability to understand the intentions of other agents, in particular the intention to communicate (Akhtar & Tomasello, 2000). In this paper we present a neural network model of word learning which proposes a possible mechanism for the dependence of efficient word learning on the concept of a communicative action. The key idea in the model is that communicative actions are initially identified by a child as cues which predict a reliable mapping between concepts and words in an otherwise noisy training environment. Once the significance of communicative actions has been learned, the child can selectively engage an associative word-learning mechanism when communicative actions are recognized. (Contains 1 table, 9 figures, and 10 footnotes.)