Drawing on Sperber and Wilson's relevance theory and Levinas's theory of conversation, this paper attempts to elucidate the process of hypothesis formation and selection in utterance interpretation. The principal conclusions suggest: 1. that psychological processes other than intention recognition accelerate or impair the derivation of assumptions and that they should be considered as falling within the study of intention recovery; 2. that the process of discourse interpretation is doubly dynamic, as created in-between the interlocutors.