Which Voice Leads Learning? A Qualitative Comparison of AI-Generated Input and Teacher Talk in Listening Instruction

Abstract

This study explores the comparative influence of artificial intelligence (AI)-generated input and teacher talk as listening materials for first-year university students learning English as a foreign language (EFL). As AI technologies become more prevalent in education, tools such as AI-generated speech offer new modes of listening input, while teacher talk continues to serve as an authentic and interactive source of language exposure. Employing a qualitative research design, this study involves classroom observations, semi-structured interviews, and reflective journals to examine students’ comprehension strategies, engagement, perceptions, benefits, and challenges with both types of listening input. The participants consist of university EFL students. The data will be analysed thematically to uncover patterns in learner responses of perceived benefits, challenges, comprehension strategies, and engagement associated with each input type. Findings indicate that learners perceived AI-generated voices as clear, consistent, and supportive of self-paced listening practice, facilitating vocabulary recognition and reducing listening anxiety. However, AI input was also described as limited in interaction and emotional presence. Teacher talk, in contrast, was experienced as more authentic and socially meaningful, providing real-time clarification, contextual scaffolding, and affective support, although natural speech variation sometimes increased cognitive demand. In terms of comprehension and engagement, AI-generated input promoted focused individual effort and autonomous learning, while teacher talk encouraged active participation, interaction, and sustained motivation. Overall, the findings suggest that AI-generated voices and teacher talk function as complementary listening sources, highlighting the importance of balanced integration to support effective EFL listening instruction.