Students’ perceptions of the use of artificial intelligence-driven tools for enhancing EFL pronunciation

Abstract

The accessibility of artificial intelligence applications has made autonomous pronunciation practice possible outside any formal instructional framework. This descriptive, cross-sectional, non-experimental quantitative study examined EFL university students’ perceptions of their current use of AI-driven pronunciation tools. A branched questionnaire organised into sixteen dimensions—among them professional relevance, self-efficacy, accent-related anxiety, autonomy, attitudes towards technology, perceived improvement, feedback quality, phonological awareness, intention to continue, recognition failures, and teacher mediation—was administered on a five-point Likert scale to 49 third-semester students enrolled in an English teaching programme and selected through convenience sampling. Adoption reached 87.8%. Although 77.6% regarded pronunciation as decisive for their professional future, merely 38.8% expressed confidence in their own performance. Perceived improvement proved moderate (53.5%) and declined from segmental accuracy (72.1%) to intelligibility (58.1%). Phonological awareness drew the highest agreement (79.1%), exceeding both perceived improvement and feedback quality (67.4%). Among non-users, 83.3% would adopt these tools under teacher guidance, their barriers proving informational rather than attitudinal. The findings describe perceptions rather than measured performance and indicate that adoption is widespread but unguided, with teacher mediation emerging as the decisive factor.