Designing ChatGPT-Mediated Feedback Activities in EFL Writing: A Design-Based Study of the Dialogic Feedback Triangle

• In Event: AERA Poster Session Thursday 7:45am
  In Poster Session: AI, Learning, and Innovation: Emerging Research in Education

Abstract

ChatGPT has shown potential as feedback providers and interactive partners, while little research has explored how to design and implement effective ChatGPT-mediated dialogic feedback. This study adopted a design-based approach to iteratively develop an optimal design for ChatGPT-mediated dialogic feedback, drawing on a dialogic feedback triangle as conceptual framework. Across three cycles with four EFL learners, data from semi-structured observations and interviews informed the design process. Seven design elements emerged, organised into a three-dimensional feedback triangle: cognitive (disciplinary knowledge, cue-consciousness, self-evaluation), socio-affective (adaptive roles of ChatGPT, learners’ critical attitudes), and structural (mobilisation of learners’ tool repertoires, feedback structures based on specific formats and standards). These elements underpin seven practical implications for educators and technology designers to implement effective GenAI-mediated feedback.

Authors

• YUNAN ZHANG, University of Hong Kong

• Yongcan Liu, University of Cambridge