Contribution of AI to Listening Assessment in SLA: A Systematic Review of Literature

The integration of Artificial Intelligence (AI) into second language (L۲) listening assessment represents a paradigm shift with transformative potential, yet the empirical evidence base remains fragmented and unevaluated. This systematic review synthesizes research from ۲۰۱۳–۲۰۲۳ to examine (۱) what AI technologies are deployed in L۲ listening assessment and for what purposes, and (۲) what empirical evidence exists regarding their validity, reliability, and effectiveness. Following PRISMA guidelines, a systematic search of Scopus, Web of Science, ERIC, and IEEE Xplore yielded ۲۲۷ records, with ۱۲ studies meeting inclusion criteria for final synthesis. Findings reveal a field transitioning rapidly from rule-based automated systems toward generative AI (LLMs, ChatGPT), gamified platforms, and—in nascent forms—multimodal and deliberative assessment systems. Three primary assessment purposes emerge: proficiency testing, formative assessment with personalized feedback, and diagnostic assessment, with formative applications attracting the most innovation. However, the evidence base is critically uneven. While several well-designed studies demonstrate strong validity and pedagogical effectiveness—including randomized controlled trials showing sustained comprehension gains and reduced anxiety—reliability reporting is systematically absent from eight of twelve studies. One study reports a Cronbach's alpha of ۰.۰۲ for AI-generated items, indicating essentially random measurement. Geographic concentration in Asian EFL university contexts further limits generalizability.

کلیدواژه ها:

artificial intelligence ، listening assessment ، second language acquisition ، validity ، reliability ، systematic review