John M. Gayed

Tokyo Institute of Technology


Currently researching the potential to use an AI-based writing assistant for second language learners at Cross lab. Little research has been done on how these systems affect L2 writing output and the researcher believes these systems will be as prevalent as spell-checking/grammar checking systems that were first developed more than thirty years ago. He plans to develop these tools to assist English Language Learners (ELLs) overcome the various cognitive barriers they face when they attempt to produce written text in English. The researcher is developing the AI system based on Open AI’s GPT-2 language model. The expected outcome of the research is that AI-based writing assistants can improve students' writing fluency.


Paper Presentation Exploring the impact of Grammarly on EFL students’ writing more

Sun, Jun 6, 16:00-16:30 Asia/Tokyo

Although the topic of automated corrective feedback has been studied extensively in CALL literature, as Frankenberg-Garcia (2019) notes, empirical research on predictive text and intelligent writing assistants is lacking. Predictive text in particular seems to be a potentially useful tool for L2 learners, as the technology may help students improve different aspects of their writing in the target language. Thus, this presentation details a small-scale study that addresses this gap in the literature by examining the impact of Grammarly, an intelligent writing assistant that incorporates predictive text technology, on the writing quality of Japanese L2 English students. Specifically, the following research question is addressed: Does Grammarly have a significant effect on the lexical complexity, syntactic complexity, and grammatical accuracy of EFL students’ writing? The study utilized a counterbalanced research design to assess the impact of Grammarly on L2 students’ writing. A total of 31 students at a Japanese university participated in the 8-week study, with the learners completing weekly guided freewriting tasks on their smartphones under two conditions: writing with the assistance of Grammarly (treatment) and writing without any aids (control). The results of the study will be presented and implications for language learning and research will also be discussed.

John M. Gayed Gilbert Dizon