Speaker
Description
Abstract
Anyone who has taught an English-Medium Instruction (EMI) class knows the quiet, common problem of it: finding or building the reading material students will actually be able to use. Content lecturers, who are rarely trained in language pedagogy, are expected to either hand students a native-speaker textbook that is too dense or spend their own limited time rewriting it by hand. Generative AI (GenAI) has recently entered this gap, promising to adapt, level, and multiply reading texts far faster than any one teacher could manage alone. This review pulls together what the current empirical and conceptual literature says about that promise, specifically for reading-material preparation in EMI and the closely related fields of English for Academic Purposes (EAP) and English as a Foreign Language (EFL). It looks at the theory behind AI-assisted scaffolding, at what AI is actually being used for in practices like generating and adapting texts, differentiating them by proficiency, producing vocabulary-controlled extensive-reading material. Also, it can be used for cutting down teacher prep time, and be at the risks that keep showing up across studies: hallucinated content, loss of academic fidelity, and uneven access. The review is based in the Vietnamese EMI context, where reading comprehension is one of the most frequently reported student struggles, and argues that the interesting question is no longer whether AI can produce a text, but whether it can do so without quietly hollowing out the disciplinary and linguistic substance that EMI reading is meant to teach. It closes with the clearest gaps in the field, and how little of this research has actually happened inside EMI content classrooms, as opposed to EAP or general EFL settings. It also suggests where Vietnam-based work could genuinely add something new.
Keywords: English-medium instruction, generative AI, reading materials, EAP, text adaptation, material design
