Quantifying the benefits to students of faking your fluency in English in recorded lectures

Heavily accented speech can hinder learning, particularly for non-English speaking students who struggle with comprehension and engagement. This study explores whether AI-generated voice-modded lectures, mimicking native English speech, improve academic outcomes and engagement in lectures by non-native English-speaking instructors. The study was conducted in an Introductory Statistics large enrolment service unit and compared student performance across original and AI-enhanced lectures. Surveys were used to collect engagement feedback from both domestic and international students. No significant performance difference was observed for international students. However, domestic students showed improved marks when exposed to synthetic lectures, highlighting this intervention's low-cost, scalable impact. This research offers practical insights into using AI voice-modding to improve engagement and equity in large, diverse classes.