Automatic Speech Recognition in conference interpreting: an exploratory study on consecutive interpreting assisted by Sight-Terp

This article aims to observe potential interpreter-machine approaches in Consecutive Interpreting (CI) assisted by Automatic Speech Recognition (ASR). Sight-Terp was used in the study, a CAI tool for CI equipped with ASR and Machine Translation (MT). Six student-interpreters at the end of their second year of the master's degree in Conference Interpreting at SSLMIT, Trieste were asked to interpret one speech in the traditional consecutive manner and one using Sight-Terp in a randomised order. The results show that a) the transcript produced by Sight-Terp does not always enable the interpreters to concentrate more on speech listening and comprehension, b) the duration of the interpreters' renditions tends to increase, owing to an increased dependency on the written text and to additional cognitive efforts involved in the rendition phase, and c) ASR can be a useful resource for the retrieval of names and figures (although there may be errors in the transcript), but it presents limitations in the transcription of spontaneous spoken language. In conclusion, despite the significant advances in ASR technologies, more research is needed to further explore the advantages of ASR-assisted consecutive interpreting and strategic approaches.