This corpus study aims to explore, both qualitatively and quantitatively, how simultaneous interpreters handle face-threatening, face-enhancing and neutral utterances addressed directly at specific individuals or larger groups participating in plenary debates of the European Parliament. The corpus under analysis comprises all English and Polish material from seven debates held between 2016 and 2020. Eleven hours of original speeches and interpretations were manually annotated for second-person deixis and nominal forms of address. Based on previous research, it was expected that interpreters would tend to reduce directness by omitting and transforming addressative forms, and that face-threatening ones would be prone to mitigation. The results of the analysis disprove these initial hypotheses. While the number of addressatives is indeed reduced in the interpretations, this roughly corresponds to the overall text compression rate. Moreover, face-threatening addressatives are transferred and added by interpreters slightly more often than neutral ones, which, in turn, are omitted more often.
