Structured Post-Lecture AI Conversation as Linguistic Scaffolding in English Medium Instruction: Vocabulary Retention and Learner Perceptions
CALICO Journal · 01.11.2026 · 43 (3) · 453-482
DOI: 10.3138/calico-2025-0052 ↗Kurz gefasst
Fachartikel in der Zeitschrift CALICO Journal, erschienen am 01.11.2026. Verfasst von Yuka Akiyama und 1 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben.
Abstract (Originalsprache)
Students in English medium instruction (EMI) often struggle to develop vocabulary knowledge despite sustained English exposure, as courses prioritize content over language. This study examines whether structured AI-mediated oral conversations, designed as post-lecture linguistic scaffolding, can support vocabulary retention in EMI. A quasi-experimental, multi-cohort design compared 3 groups of Japanese science and engineering undergraduates enrolled in an EMI course: Summary Writing ( n = 10), Structured AI Conversation (AI-mediated dialogue with target vocabulary embedded in the prompts; n = 19), and Explicit + AI (teacher-led vocabulary instruction followed by open-ended AI-mediated conversation; n = 15). Receptive knowledge of 30 discipline-relevant target words was assessed through meaning recall and recognition tests administered at pre-lecture test, post-lecture test, and delayed post-test. In addition, learner perceptions were examined through surveys and interviews with the Structured AI Conversation group. While all 3 groups showed immediate gains following the lectures, the Structured AI Conversation group demonstrated greater retention of the learned words than the Summary group, whereas its performance did not differ significantly from that of the Explicit + AI group. Qualitative findings indicated that students perceived the structured AI conversation task as beneficial for reinforcing content, supporting language practice, and building confidence, though perceptions appeared to vary by proficiency, with lower-proficiency learners emphasizing supportive aspects and higher-proficiency learners focusing on limitations such as conversational unnaturalness and feedback inconsistency. Findings suggest that vocabulary scaffolding can be embedded into EMI through AI-mediated interaction; however, effective implementation requires careful task design that addresses learner needs across proficiency levels.
Zitieren
APA
Akiyama, Y., & Uchihara, T. (2026). Structured Post-Lecture AI Conversation as Linguistic Scaffolding in English Medium Instruction: Vocabulary Retention and Learner Perceptions. CALICO Journal, 43(3), 453-482. https://doi.org/10.3138/calico-2025-0052
Vancouver
Akiyama Y, Uchihara T. Structured Post-Lecture AI Conversation as Linguistic Scaffolding in English Medium Instruction: Vocabulary Retention and Learner Perceptions. CALICO Journal. 2026;43(3):453-482. doi:10.3138/calico-2025-0052
BibTeX
@article{Akiyama2026,
author = {Akiyama, Yuka and Uchihara, Takumi},
title = {Structured Post-Lecture AI Conversation as Linguistic Scaffolding in English Medium Instruction: Vocabulary Retention and Learner Perceptions},
journal = {CALICO Journal},
year = {2026},
volume = {43},
number = {3},
pages = {453--482},
doi = {10.3138/calico-2025-0052}
}
EUVIMED