Productive Friction and Learner Engagement in Genai Supported EFL Vocabulary Learning

Abstract

This article explains how generative artificial intelligence (GenAI) prompts can support durable vocabulary learning while preserving learner engagement and autonomy in English as a foreign language (EFL) education. A purposive conceptual synthesis integrates vocabulary research, retrieval practice, task involvement, cognitive load, self regulation, and AI literacy. The analysis distinguishes established findings from proposed relationships and develops the AI Mediated Cognitive Engagement Model (ACEM). ACEM locates potential learning benefits in the sequencing of an initial attempt, contingent assistance, evaluative feedback, contextual reuse, and later retrieval. Productive friction is defined prospectively through learning relevant demands, accessible support, and opportunities for error repair; its effectiveness requires independent testing. Four propositions address retrieval, proficiency dependent support, evaluative reuse, and learner control. The synthesis is selective and reports no intervention data. The proposed mechanisms require comparisons that separate prompt sequence effects from exposure, time, feedback quality, and technological novelty. Teachers can use a graduated prompt sequence, verified lexical resources, and equivalent offline tasks to protect participation, reduce unnecessary interface demands, and assess learning after AI assistance is withdrawn. The framework connects prompt design with observable learner activity and delayed lexical outcomes, while treating emotional engagement, teacher judgement, and equitable access as conditions of implementation.