Scholarlink

<p>As artificial intelligence (AI) tools such as ChatGPT become increasingly integrated into language classrooms, English instructors are exploring effective ways to incorporate them into writing instruction. This study examines two approaches to AI-supported multimodal writing in a first-year university composition course in South Korea: (1) instructor-provided AI-generated images and (2) learner-generated AI images. The study investigates how these approaches influenced learners’ writing development, engagement, and perceptions of AI-assisted writing. Data were collected through image-based writing tasks, surveys, semi-structured interviews, and pre- and post-writing assessments. The findings revealed no significant differences in overall writing performance between the two groups; however, important differences emerged in how learners engaged with visual support. Instructor-provided images reduced cognitive load and supported idea organization, whereas learner-generated images promoted greater autonomy, active engagement, and deeper cognitive processing through iterative prompt refinement. These findings offer practical implications for language instructors seeking to integrate AI-generated images into multimodal writing instruction while balancing pedagogical scaffolding and learner autonomy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

GenAI Image Integration in EFL Writing: Instructor-Led and Learner-Led Approaches

  • 期刊论文

  • 发表日期：

    2026年7月14日

查看全文
  • SoHee Kim,
  • Sandra Morales

摘要

As artificial intelligence (AI) tools such as ChatGPT become increasingly integrated into language classrooms, English instructors are exploring effective ways to incorporate them into writing instruction. This study examines two approaches to AI-supported multimodal writing in a first-year university composition course in South Korea: (1) instructor-provided AI-generated images and (2) learner-generated AI images. The study investigates how these approaches influenced learners’ writing development, engagement, and perceptions of AI-assisted writing. Data were collected through image-based writing tasks, surveys, semi-structured interviews, and pre- and post-writing assessments. The findings revealed no significant differences in overall writing performance between the two groups; however, important differences emerged in how learners engaged with visual support. Instructor-provided images reduced cognitive load and supported idea organization, whereas learner-generated images promoted greater autonomy, active engagement, and deeper cognitive processing through iterative prompt refinement. These findings offer practical implications for language instructors seeking to integrate AI-generated images into multimodal writing instruction while balancing pedagogical scaffolding and learner autonomy.