Integrating automatic speech recognition into Chinese–Portuguese simultaneous interpreting

Abstract

The rapid development of automatic speech recognition (ASR) has created new opportunities for technology-enhanced simultaneous interpreting, yet empirical research has focused largely on English-paired combinations. This study examines the integration of real-time automatic transcription in Chinese–Portuguese simultaneous interpreting, investigating its effects on interpreter performance and on interpreters’ interaction with and perceptions of the technology. Ten first-year MA students completed simultaneous interpreting tasks with and without ASR-generated transcription in a mixed-methods design. Performance was assessed across four dimensions (information completeness, fluency, target language quality, holistic performance), with ear-voice span measured to capture temporal processing patterns. Qualitative data were collected through questionnaires and retrospective interviews. Results revealed that transcription support significantly improved information completeness but yielded no significant effects on fluency, target language quality, or holistic performance. Participants reported allocating, on average, 60% of their attention to the visual display, with ear-voice span nearly doubling when transcription was available. Perceptions of the tool were generally positive, though concerns were raised regarding cognitive fatigue and error propagation. Participants preferred continuous transcription with highlighted problem triggers, partly diverging from findings in other language pairs. Overall, the results show that ASR-generated transcription offers selective benefits while introducing risks of visual over-reliance, underscoring the need for targeted training and customizable, user-centered interface design in technology-supported simultaneous interpreting.