The effect of AI-powered speech recognition tools on listening comprehension and pronunciation awareness in EFL learners

Abstract

This study investigated whether integrating AI-powered speech recognition (ASR) into regular lessons improved listening comprehension and pronunciation awareness among EFL learners in a private institute. Sixty intermediate, Arabic-L1 female students (aged 18-22) in Baghdad were assigned by intact classes to an experimental group (ASR-mediated practice) or a time-matched control group (teacher-led practice without ASR) over eight weeks. The experimental group completed brief, recurrent ASR cycles embedded in the syllabus. Outcomes were measured with parallel forms of an IELTS-aligned listening test and an adapted Pronunciation Awareness scale. Analyses showed that the ASR group significantly outperformed the control group on post-test listening and demonstrated significantly higher pronunciation awareness. The pattern supported the interpretation that real-time, item-level feedback strengthened noticing and self-monitoring, which in turn aligned with gains in receptive decoding. Pedagogically, two short ASR cycles per week appeared sufficient to yield measurable benefits without increasing total class time. Limitations included the quasi-experimental design, single-site context, and absence of a delayed post-test; future work should examine retention, dosage-response, and targeted supports for suprasegmentals.