柴田 直哉

This exploratory case study investigates Japanese English-as-a-foreign-language university students’ beliefs about text and suggestions produced by generative artificial intelligence (GenAI). Qualitative data were collected from questionnaire responses from 83 upper-beginner and lower-intermediate students and semi-structured interviews with 46 volunteers at a private university in central Japan. Results showed that a substantial proportion of students had a strong belief that GenAI output would be appropriate and trustworthy for their writing development due to their limited knowledge, low writing self-efficacy and perceived native-like writing norms. However, some comments highlighted the dilemma that learners understand the overuse of GenAI functions should be avoided, but they still rely on its accessibility because they believe GenAI technology produces linguistically ideal writing models. A few reported that they doubted its linguistic reliability and appropriateness even though their writing self-efficacy was low. Based on these findings, these learner belief patterns are conceptualised as generative artificial intelligence writerism (GenAI-Writerism). Although GenAI-Writerism is not necessarily negative, strong GenAI-Writerism can foster overreliance on GenAI technology and hinder students’ writing skill development. To avoid these risks, the importance of critical artificial intelligence literacy education is discussed. Finally, future research foci on GenAI-Writerism are proposed whilst considering research limitations.

Theme-based instruction (TBI) is one of the models of content-based instruction, in which learners study topics that are interesting and relevant to them (Brinton & Snow, 2017; Snow & Brinton, 2023). Although this approach is well-known in many English-as-a-second/foreign-language (ES/FL) educational contexts (Brown & Lee, 2015; Curtis, 2017; Yugandhar, 2016), research on the effects of TBI on students’ writing performance is still scarce (Ostovar-Namaghi & Nakhaee, 2019). To address this research issue, the authors conducted a one-shot model of a quantitative study with 86 first-year university students in Japan over a 15-week semester. The course covered four thematic units. The authors collected students’ perceptions of each theme before learning the unit content and their pre-/post-timed writing papers about each theme. The Wilcoxon signed-rank tests were conducted to analyze their writing samples from two perspectives (the number of tokens and syntactic complexity). Regardless of assigned themes, statistical test results revealed that the number of word tokens increased significantly, whereas syntactic complexity did not. Based on these findings, TBI does not appear to be a practical approach for facilitating learners' writing abilities to use grammatically complex sentences. Hence, other teaching approaches need to be explored and integrated with TBI when syntactic complexity development is a teaching and learning objective in writing courses.