Integrating Generative AI into Strategy-Based Instruction: Adaptive Scaffolding for Self-Regulated Learning in Secondary EFL Writing

Abstract

While Generative Artificial Intelligence (GenAI) transforms
text production, its potential to cultivate deeper writing
processes and self-regulated learning (SRL) capabilities in
young learners remains largely untapped. Addressing this
gap, this study explored how GenAI-supported writing
instruction influences secondary students' deployment of
SRL strategies. We designed and evaluated a GenAI-powered
adaptive dialogue agent, embedded within a Strategy-Based
Instruction (SBI) framework, to scaffold students’ writing
endeavors. Unlike standard chatbots, this agent utilized a
modular workflow to provide personalized support across the
forethought, performance, and self-reflection phases. Using
a randomized controlled trial, we compared the
GenAI-scaffolded condition (N = 40) to traditional
teacher-led instruction (N = 40) over 8 weeks. The GenAI
condition produced significant gains in writing
performance, alongside deeper, more frequent use of SRL
strategies (specifically text generation strategy). The
findings suggest that engaging secondary students in
adaptive GenAI dialogue acts as a catalyst, transforming
their role from passive recipients of AI text to active
regulators of their writing process. GenAI can function as
a "metacognitive partner" capable of dynamically supporting
personalized writing practice, offering a promising pathway
for cultivating essential writing skills in the AI era.