Enhancing numerical accuracy in simultaneous interpreting: a comparative study of human and AI-based support
Simultaneous interpreting (SI) of numerical elements is a complex task and numbers are widely acknowledged as a problem trigger and a source of errors for interpreters. However, recent advances in Automatic Speech Recognition (ASR) and Artificial Intelligence (AI) integrated into Computer-Assisted Interpreting (CAI) tools have shown potential in enhancing accuracy during SI of numerical elements, as well as of other parts of speech, such as specialised terminology. Although many recent studies are focusing on this area, research comparing the impact of CAI tool support during SI against the support provided by human boothmates is scarce. This paper aims to address this gap within this emerging field of research by comparing the performances of interpreting students using the ASR-powered CAI tool InterpretBank with those who worked with the support of a human boothmate while interpreting a source text (ST) with a high density of numbers. The pilot study assesses the accuracy rate of numerical renditions, intelligibility of audio extracts, and interpreters’ self-perception/satisfaction with the CAI tool. The findings show that CAI tools can improve interpretation quality, and the digital boothmate provides equivalent support to a human boothmate in terms of accuracy of numerical rendition, while considering latency limitations.
Claudio Russello, Matilde Carbutto, "Enhancing numerical accuracy in simultaneous interpreting: a comparative study of human and AI-based support" in: "The Interpreters' Newsletter n. 28 - 2023", EUT Edizioni Università di Trieste, Trieste, 2023, pp.