Tame the beast of ChatGPT: developing design principles to strategically integrate ChatGPT into EFL writing through an activity theory perspective
Yunan Zhang, Yongcan Liu
Abstract
Despite the growing interest in the potential of ChatGPT in writing, empirical research on ChatGPT-assisted English as a Foreign Language (EFL) writing strategies is scarce. This study addresses the central question of how ChatGPT can be strategically integrated into writing, which has been conceptualised as the resolution of contradictions through the lens of Activity Theory (AT). Employing a Design-Based Research (DBR) approach, we have developed an intervention with design principles to guide ChatGPT-assisted writing through three iterative cycles of implementation and revision. In each round, screen recordings of four Chinese EFL writers’ ChatGPT-assisted writing, semi-structured interviews on the intervention and AT framework, and stimulated-recall interviews on contradictions and solution strategies were collected and analysed using Reflexive Thematic Analysis (RTA). Four contradictions manifested themselves, wherein the efforts to solve contradictions resulted in the iterative development of the intervention, AT framework, and design principles. Underpinned by the AT framework, four intersected design principles, each consisting of two to three strategies, have emerged from the process of resolving ChatGPT-related contradictions in the AT framework. The four design principles with nine strategies could guide the strategical integration of ChatGPT into EFL writing and have implications for self-regulated learning, AI literacy, and EFL writing instructions.