From Feedback to Feedforward: AI-Enhanced Formative Assessment for Self-Evaluation in EFL Non-English Majors’ Listening–Speaking Development

Research on formative assessment in English as a Foreign Language (EFL) contexts has emphasized its potential to enhance learner autonomy, reflection, and sustained improvement; however, limited research has explored how Artificial Intelligence (AI) can transform formative feedback into feedforward practices that strengthen learners’ self-evaluation and skill development, particularly in listening and speaking. This study examined the effects of AI-enhanced formative assessment on learners’ self-evaluation and their listening–speaking performance. Adopting a mixed-methods quasi-experimental design, the study involved 72 non-English majors enrolled in an “English for Communication” course at a private university in West Java, Indonesia. Participants were assigned to an experimental group (n = 36), which received AI-supported formative feedback, and a control group (n = 36), which received conventional feedback, over an eight-week intervention. Quantitative data were collected through pre- and post-tests of listening and speaking, while qualitative data were obtained from a closed-ended questionnaire incorporating AI feedback log items and semi-structured interviews with six selected participants. The results indicated that the AI-supported group achieved significantly greater gains in listening comprehension, pronunciation accuracy, fluency, and discourse organization than the control group. Qualitative findings further revealed enhanced self-monitoring, reflective awareness, and feedforward-oriented learning behaviors among learners using AI feedback. However, some participants reported challenges in interpreting nuanced AI-generated comments and translating them into actionable improvement strategies. Overall, the findings demonstrate that AI-enhanced formative assessment can effectively support self-evaluation and listening–speaking development, provided that pedagogical guidance is integrated to facilitate meaningful use of automated feedback.