Analyzing users' interactions with writing feedback and their effects on writing performance

Beata Beigman Klebanov, Jiang Yang, Oren Livne, JIangang Hao

Abstract: With the advance of technology, computer-based tools are increasingly used for writing instruction. However, there is a gap between their practical application and research on how they are used and their effects on writing skills. In the present study, we examined the use of Writing Mentor® (WM), a free Google Docs add-on designed to support academic writing through automated feedback. We used event logs to explore the activities that users-in-the-wild engaged in while revising their submissions. We found that the quality of users’ written products significantly improved from the first submission to the last. Viewing feedback related to the writing being well-edited more frequently and more time spent in WM were significantly associated with a bigger improvement in writing quality. Our findings have implications for the development of writing feedback and the design of AI-assisted tools to support writing.