utilizing chatgpt to enhance iranian efl learners’ pronunciation: a focus on phonological processes

Recognizing that conventional iranian efl pedagogy often marginalizes pronunciation, this study investigates the effectiveness of chatgpt, an artificial intelligence tool, in improving the pronunciation of iranian efl learners by focusing on key english phonological processes. employing a quasi-experimental non-equivalent groups pre-test–post-test control group design, 40 lower-intermediate male efl learners (aged 17–21) from pardisan language institute in tabriz, iran, were selected through convenience sampling and homogenized using the oxford placement test. participants were divided into an experimental group, which received chatgpt‐mediated pronunciation practice with optimized prompt designs that explicitly highlighted persian‐specific transfer issues, and a control group, which followed conventional teacher-led instruction. both groups received instruction on phonological processes over five weeks. data were collected using teacher-made pronunciation tests, consisting of word reading, sentence reading, and spontaneous speech tasks. pronunciation performance was evaluated using acoustic analyses (e.g., spectrographic measures of flapping accuracy) and intelligibility ratings by two blind raters. results from independent samples t-tests demonstrated that the experimental group achieved significantly greater improvements in both segmental and suprasegmental accuracy compared to the control group. notably, chatgpt reduced misclassification rates for connected‐speech rules compared to the rule‐based tutor. these findings suggest that ai-powered tools like chatgpt can serve as effective supplements to traditional instruction by offering interactive and linguistically informed pronunciation support in efl contexts.