From Red Ink to Algorithm:
Reimagining Feedback Cultures with GenAI in EFL Writing
This theoretical paper examines how LLM-powered tools (e.g., ChatGPT, Claude, Gemini) intersect with entrenched feedback cultures in EFL writing classrooms. It critiques the dominant "red-pen" feedback paradigm—rooted in behaviourist models of error correction, compliance, and measurable accuracy—and advocates for dialogic, co-constructed feedback cultures mediated by both human and algorithmic actors. Drawing on sociocultural theory, feedback literacy, and affect-informed pedagogy, the paper introduces the concept of Affective-Epistemic Feedback Habitus: a dispositional framework shaped by repeated exposure to authoritative correction and high-stakes evaluation. This habitus governs how learners emotionally and cognitively interpret feedback, often predisposing them to treat algorithmic suggestions as final verdicts rather than reflective opportunities. While LLMs offer non-judgmental, real-time support that can scaffold autonomy and metacognitive engagement, their use within correction-oriented cultures risks reproducing epistemic dependency and instrumental revision. The paper repositions LLMs as pedagogical interlocutors within culturally responsive, ethically mediated writing instruction.
Keywords: Generative AI; Large Language Models; EFL Writing Feedback; Feedback Cultures; Behaviourism; Learner Agency; Feedback Literacy; Dialogic Pedagogy; Sociocultural Theory; Affective Feedback; Teacher Mediation