The effects of different voice modalities on EFL students' listening anxiety and comprehension

This study investigates the influence of different voice modalities such as AI-generated voices (AGVs), live teacher voice (LTV), and recorded human voice (RHV) on the language learning processes of English as a Foreign Language (EFL) learners, with a focus on listening anxiety, comprehension skills, and learner perceptions. Utilizing a mixed-methods approach, the research examines how different voice modalities affect EFL learners' anxiety levels and listening test scores. Additionally, the study explores learners' perceptions of different voice modalities in their learning experiences, identifying specific attributes that are considered beneficial or detrimental. The participants of the study were comprised of 141 EFL students from a state university's preparatory school. The participants were randomly assigned to three groups, with one control group and two experimental groups. The quantitative data was collected using a modified version of the Foreign Language Listening Anxiety Scale (FLLAS) and the listening section of a Preliminary English Test (PET) from Cambridge. The qualitative data was collected through interviews with 21 participants, using semi-structured interview questions developed by the researcher. Findings suggest that voice modalities have varying effects on listening anxiety, comprehension, and learner perceptions. The AGV group had the most significant reduction in listening anxiety, while the LTV group performed significantly better on the listening test. On the other hand, the RHV group yielded the highest anxiety scores and lowest listening scores. Additionally, the results indicated a significant negative correlation between listening anxiety scores and comprehension scores. Finally, learners in each group demonstrated varying attitudes towards different voice modalities. Keywords: Listening anxiety, AI-generated voice, English as a Foreign Language