Analisis Kesalahan dalam Karangan Pembelajar Bahasa Jepang Penutur Indonesia dalam konteks pembelajaran yang berpotensi melibatkan AI

Abstract

This study aims to identify the characteristics of errors that commonly occur in the written compositions of intermediate-level Japanese language learners at the tertiary level in Indonesia. The participants were third-year students (fifth semester) enrolled in the Japanese Literature Program, Faculty of Cultural Sciences, Universitas Hasanuddin, who took the Japanese Composition course during the odd semester of the 2025/2026 academic year. The data analyzed consisted of in-class writing assignments (Compositions 3–5) and final examination compositions. Compositions 1 and 2 were excluded from the analysis because, at that stage, students had not yet received instruction on the use of genkōyōshi and paragraph structure. Compositions 3–5 were produced through a staged instructional process that included teacher guidance on writing, draft preparation, and group-based revision through peer response activities. In contrast, the final examination compositions were written under time constraints, with a 50-minute limit and without access to dictionaries or other reference materials. In this study, the analysis focused exclusively on errors that appeared repeatedly and were shared by multiple learners, while individual or incidental errors were excluded from consideration. The findings indicate that errors related to basic grammatical aspects, such as particle usage and verb conjugation, were relatively limited. By contrast, errors associated with the use of genkōyōshi, paragraph organization, stylistic choices, and the mixing of written and spoken language were consistently observed. This article further discusses the underlying factors contributing to these errors by considering instructional practices, differences in writing conditions between assignments and examinations, and the potential influence of artificial intelligence technologies in the writing process,  which is interpreted as a contextual consideration based on the learning environment rather than as an empirically verified finding of this study.