Sequencing affective preparation before AI-mediated instruction: a quasi-experimental study of L2 writing outcomes and sustainability among multilingual undergraduate learners

This mixed-methods quasi-experimental study examined whether sequencing affective preparation before AI-supported instruction enhanced second language (L2) writing outcomes and promoted sustained learning among multilingual learners in an underserved educational context. Sixty low-proficiency undergraduate EFL learners from diverse linguistic backgrounds (Telugu, Tamil, Malayalam, Hindi) at a South Indian university participated in an eight-week classroom intervention, with an experimental group receiving structured affective preparation followed by guided AI-supported writing instruction, and a control group receiving traditional process-oriented writing instruction. Quantitative analyses revealed that the experimental group achieved significantly greater gains in writing performance, writing self-efficacy, and metacognitive awareness than the control group did. Importantly, these gains were largely retained at a delayed post-test administered five weeks after the intervention without AI support, providing evidence consistent with skill internalization rather than tool dependency within the five-week follow-up window of the present study. Qualitative findings further demonstrated that affective preparation reduced writing anxiety and fostered psychological safety, enabling learners to engage with AI tools productively. Framed as an instructional application of the Emotion- Cognition-Metacognition (ECM) model grounded in sociocultural theory, the findings suggest that affective readiness is a critical pedagogical condition for maximizing the effectiveness and sustainability of AI-mediated L2 writing instruction. These results have implications for educational equity in digital learning environments, suggesting that thoughtful instructional sequencing can make AI tools more accessible and beneficial for anxious, low-confidence learners who might otherwise be marginalized in technology-rich classrooms.