A longitudinal study on prediction in simultaneous interpreting


Thematic Section: The intersection between interpreting and the language hierarchy

simultaneous interpreting, expertise, cognition, language proficiency, working memory

Rhona Amos, University of Geneva
Martin Pickering, University of Edinburgh
Kilian Seeber, University of Geneva

Experience in simultaneous interpreting may confer verbal and executive advantages upon simultaneous interpreters, and longitudinal studies have detected neural differences in interpreting trainees before and after training. This suggests that interpreting training may improve regions of the brain associated with language processing.
Prediction routinely takes place in language comprehension, and has long been considered a skill or strategy in simultaneous interpreting. We hypothesised that training might lead to an increase in predictive processing during an interpreting task.
Two cohorts of a total of 23 interpreting students participated in our study from March 2019 to December 2020. Students heard, and simultaneously interpreted into their most dominant language, an English sentence containing a highly predictable word, for example, “mouth”, in the sentence, “The dentist asked the man to open his mouth a little wider”. In a visual-world design based on Ito et al. (2018), students’ eyes were tracked as they viewed a visual scene, which appeared 1000ms before the onset of the predictable word. It contained four objects, one of which was either the predictable word (mouth) or an unrelated word (bone). Students participated in the study before and after two semesters of simultaneous interpreting training. Two versions of the study were used, with different stimuli. Their order was counterbalanced between participants.
We modelled the time-course of fixations to the critical object in the target condition as compared to the baseline condition. Prior to training, students begin to predict 550ms before the onset of the critical word. Preliminary results from the first cohort suggest that predictive behaviour is not greatly influenced by training in simultaneous interpreting.
We discuss the implications of these findings for theories of prediction in psycholinguistics and interpreting.