<![CDATA[This study investigates how the Initiation Response Feedback (IRF) pattern, integrated with an AI-based feedback tool, supports public speaking instruction in an English as a Foreign Language (EFL) context. It examines (1) how IRF patterns emerge in classroom interaction, (2) how IRF contributes to feedback effectiveness and students’ public speaking performance, and (3) how an AI tool, AI Feedback Presentation, a Gradio application deployed on Hugging Face Spaces is positioned within the feedback ecosystem and students’ academic literacy. The research adopts a qualitative classroom discourse analysis design. Participants were 19 third semester students enrolled in a Public Speaking course in an English Literature program at a private university in Makassar, Indonesia. Data were collected through participant observation, audio video recordings, verbatim transcripts of four key meetings, and pilot logs from the AI Feedback Presentation application. The four meetings focused on polite (dis)agreement, formal presentation structure and peer feedback, discussions on empathy, plastic use, AI and academic honesty, and modeling of expert speeches. Data were analyzed by segmenting IRF sequences, coding feedback types, and developing themes related to public speaking skills and AI use. Findings show that IRF is consistently employed to scaffold students’ spoken production from short responses to structured, polite opinions and formal presentations. The combination of IRF with recasts, metalinguistic explanations, elaborative feedback, and affective support enhances students’ awareness of politeness, speech organization, and delivery (eye contact, intonation, body language). The AI Feedback Presentation tool provides automatic transcription and simple performance metrics, which function as triggers for reflection rather than grading mechanisms. Students perceive AI as a useful assistant for grammar and idea generation, while also recognizing risks to academic honesty. The study proposes an IRF AI framework in which human and AI feedback are complementary in public speaking instruction.]]>
