Translation has recently been analysed in the terms of modern cognitive–pragmatic theory (relevance theory) as an interlingual interpretive use of language (Gutt, 1991/2000). But Gutt's account primarily addresses the principles and processes of text or written translation, where the displacement in time and place between the original communicator, the translator and her readers requires the translator to reconstruct the original informative intention, project the original and target addressees’ cognitive environment, and craft a stimulus according to the degree of interpretive resemblance sought. By contrast, oral translation, in particular simultaneous interpreting (SI), is performed in live situations in which the interpreter shares most of the manifest cognitive environment with the participants and is thus better able to project and control the contexts in which her addressees will process her utterances. Since the condition of simultaneity severely constrains the simultaneous interpreter's choice of stimulus, she relies heavily on this access to immediate context and her audience's inferential abilities. Text translators need time to project context and choose their stimuli, while in SI, access to live contexts compensates for temporal constraints. The paper concludes with a discussion on prospects for exploring patterns and possible biases in interlingual text and oral communication on this basis.