SUNDANESE, INDONESIAN, ENGLISH, AND AI: TRANSLANGUAGING PRACTICES OF EFL STUDENTS IN AI-ASSISTED ACADEMIC WRITING

Abstract

Many Indonesian students of English as a foreign language (EFL) write academically across three languages: a regional first language such as Sundanese, Indonesian as the national language, and English as the target language. Generative artificial intelligence (AI) has now entered this multilingual writing process, raising the question of how students draw on their full linguistic repertoire when they write with AI. This systematic literature review (SLR), conducted following PRISMA guidelines, examined research on AI-assisted academic writing to determine what is known about translanguaging practices of multilingual EFL writers, with particular attention to the Sundanese–Indonesian–English context. From a pool of 99 records, 96 unique records were screened. No study met the core eligibility criterion of examining translanguaging, regional languages, or Sundanese in AI-assisted academic writing. Twenty studies were retained as adjacent evidence across four domains: L2 and EFL writers' AI use (11 studies), Indonesian higher education (3), language and power in AI-assisted scientific writing (2), and multilingual national contexts (4). Synthesis of this adjacent evidence suggests that L2 writers use AI differently from native speakers, that critical AI literacy shapes how they evaluate AI output, and that English dominance in academic publishing frames AI as a tool for linguistic enhancement. Yet the evidence treats writers as monolingual learners of English, overlooking their full repertoires. The review therefore identifies an empty evidence base, proposes a trilingual translanguaging framework for AI-assisted writing, and sets out a research agenda and study design for investigating Sundanese EFL writers.