Exploring L2 Students' Use of GenAI to Generate Model Text as a Feedback Instrument: A Practice and Perception Study

Model texts have been widely recognised as an effective feedback instrument in enhancing L2 writing. Recent advances in generative artificial intelligence (GenAI), capable of producing human-like text in response to user prompts, offer an accessible and efficient alternative for generating customised model texts tailored to L2 learners’ specific goals and proficiency levels. Despite this potential, little empirical research has examined L2 students’ interactions and perceptions of leveraging GenAI for model text creation to support their revision. This study aimed to explore how L2 students perceived and utilised a GenAI tool, Kimi, to produce model text as a form of feedback for an argumentative writing task. Data were collected from 12 mainland Chinese university students’ screencasts, interaction logs with Kimi, stimulated recalls, and semi-structured interviews. The findings revealed a five-stage engagement process: task analysis, initial prompting, output evaluation, iterative refinement, and finalisation. Throughout this process, the students demonstrated their context awareness, genre knowledge, and GenAI knowledge. The students also acknowledged the affordances, limitations, and challenges associated with the autonomous use of GenAI to create model texts. This study contributes original insights into learner agency in feedback-seeking in the age of GenAI, highlighting how students can be positioned in an active role to negotiate appropriate feedback for their writing development. It offers pedagogical implications for fostering the knowledge and skills necessary for optimal and independent use of GenAI in model texts as a feedback instrument practice.