Recognizing intended meaning and speakers' plans

Human conversational participants depend upon the ability of their partners to recognize their intentions, so that those partners may respond appropriately. In such interactional the speaker can encode his intentions that the hearer act in a variety of sentence types. Instead of telling the hearer what to do. the speaker may Just state his goals, and expect a response that meets these goals. This paper presents a new model for recognizing the speaker's Intended meaning in determining a response. We show that this recognition makes use of the speaker's plan, his beliefs about the domain and about the hearer's relevant capacities.