The Impact Of Automatic Speech Recognition On Fluency Of Chinese-English Simultaneous Interpreting

Artificial intelligence has progressed rapidly,bringing new products and technologies impacting the interpretation business.Meanwhile,the pandemic stimulated the development of conference platforms,and online interpretation has become the norm.Online meeting platforms have provided ASR-based captions.This research investigates the impact of live captioning generated by automatic speech recognition(ASR)on C-E simultaneous interpretation(SI)fluency.The research methodology includes both experiments and retrospective interview.The experiment examined the changes in disfluency of interpreters with and without live captions by statistically measuring four indicators-silent pause,filled pause,self-repair,and repetition.Meanwhile,we also studied the effects on fluency in interpreting numerals and terminology.Furthermore,based on the experiment results and the retrospective interviews,the causes of these effects were further explored under the analysis framework of the Effort Model.The experimental results show no significant difference in overall fluency and terminology fluency before and after using live captioning and terminology interpretation.However,there is a significant difference in numeral interpretation.Moreover,its effect was positive.The results of retrospective interviews indicate that live captioning has both negative and positive effects on fluency but no impact on the overall statistical level.Its negative impact is attributable to captions errors,delays,and the way of presentation,which raise the cognitive load of interpreters and increase disfluency.Its positive effect is that visual signals can reduce their cognitive load and decrease disfluency when interpreting numerals and enumerations.Therefore,the overall fluency has mostly stayed the same in aggregation.In conclusion,positive and negative impacts remind interpreters to utilize captions appropriately.Meanwhile,ASR providers should increase the accuracy of live captioning and minimize their delay to meet the needs of interpreters better.