In this paper,ｗ e examine the first step of conference interpretation training at ESIT: message comprehension and restitution without note taking, in other words, the acquisition of the basic interpretation process. We first present some findings from the recording of two Japanese-French interpretation classes in the first term. The second part is devoted to theoretical analysis based on the Interpretative Theory of Translation (TIT), to which some recent findings in cognitive science have been integrated (Ito-Bergerot, 2005a). In the last part, we propose an Interpreter’s Speech Comprehension Model. This model derives from the text comprehension model proposed by Ericsson and Kintsch (1995), which can be compared to Lederer's “unité de sens”. We also refer to the comprehension model of Gernsbacher (1990), who stresses the importance of suppressing irrelevant information during story understanding. In our model, the interpreter listening to a speech in X language, in order to translate it into Y language, constructs a situation model, which becomes a multi-layered complex structure of mental representations as the story develops. These mental representations have, under certain conditions, links with either the Y language system or the X language system, corresponding to the interpreter’s awareness of specific words in the Y language that come to mind while listening to the speech, or the recall of an expression used by the speaker in the X language. This model can be used to describe students’ errors, as well as experts’ skillful processing.
