Conference Paper Published

Does AI-INT Training Transfer Across Interpreting Modes? Evidence from Eye-Tracking

Li, J., & Weng, Y. (2026). Does AI-INT Training Transfer Across Interpreting Modes? Evidence from Eye-Tracking. In Proceedings of the 3rd International Conference on New Trends in Translation and Interpreting Technology, 189-192.

URL: https://nettt-conference.com/2026/wp-content/uploads/2026/06/NeTTIT-2026-Proceedings.pdf

Abstract

This paper investigates two related questions: whether current AI-assisted interpreting (AI-INT) training helps students process AI-generated transcripts more efficiently, and whether such training also benefits conventional text-based interpreting tasks. Forty-one postgraduate interpreting students (22 AI-INT-trained, 19 untrained) completed three types of tasks: AI-assisted simultaneous interpreting (AISI), simultaneous interpreting with text (SIT), and sight translation (ST), yielding 454 valid eye-tracking trials. Fixation rate (fixations/min) and mean fixation duration (MFD) were used to gauge reading ease and efficiency. Linear mixed-effects models revealed a significant Training × Mode interaction for fixation rate: compared with untrained interpreters, AIINT-trained interpreters produced significantly fewer fixations in ST and slightly fewer in SIT, but slightly more in AISI. MFD did not differ significantly between groups in any mode. These findings suggest that current AI-INT training may enhance fixation efficiency in conventional static-text tasks but does not transfer to AISI.