Aug 2026· International Journal of Educational Technology in Higher Education· Vol 23· 0 citations· 40 references
TL;DR
The findings indicate a dissociation between performance quality and short-term recall in LLM-supported study, which aligns with cognitive-psychology evidence that elaboration improves comprehension while retrieval practice consolidates retention.
Abstract
Large Language Models (LLMs) are rapidly transforming higher education, yet evidence on how interaction patterns affect learning remains limited. This study examines whether explanation-seeking dialogue with an LLM is associated with task quality and immediate recall in a controlled learning session. Twenty-two postgraduate students completed a pre-test, an LLM-assisted neuroeconomics case task, and an immediate post-test. Fine-grained interaction logs captured per-turn telemetry, enabling extraction of Depth (proportion of "why/how/explain" prompts), Volume, and Pacing features. Results showed large immediate pre–post score gains (Cohen’s dz = 2.12, p < .001). In the task-quality regression model, Depth was positively associated with task quality beyond baseline and Volume (β = 6.27, p = .006), indicating that a one-standard-deviation increase in explanation-seeking prompts was associated with approximately six additional marks on a 0–100 scale. Depth was not associated with immediate recall (β = − 0.014, p = .728); instead, gain scores were strongly associated with baseline knowledge, consistent with reduced headroom for improvement among students starting from a higher pre-test score (β = − 0.161, p < .001). The findings indicate a dissociation between performance quality and short-term recall in LLM-supported study. This aligns with cognitive-psychology evidence that elaboration improves comprehension while retrieval practice consolidates retention. Pedagogically, the dissociation suggests that depth-oriented dialogue may need to be paired with deliberate memory-strengthening activities such as self-testing or spaced retrieval if comprehension gains are to translate into recall, although the present design does not test such activities directly. Methodologically, it contributes a replicable, privacy-preserving instrumentation pipeline linking conversational telemetry to learning outcomes. Limitations include single-group design, small sample (n = 22), immediate testing only, and keyword-based depth proxies. Future work should randomise assistance styles, incorporate delayed retention tests, and refine depth measurement through semantic coding.
Large language models (LLMs) are increasingly used as on-demand conversational learning assistants, but they typically do not adapt explanations to a student’s background unless explicitly prompted. We present the Personalized Learning Assistant Interface (PLAI), a web-based prototype that generates explanations from lecture slides, audio transcripts, and a structured student profile through a chat-based interface. We evaluated PLAI in a controlled pilot study with 24 STEM students, comparing profile-based personalization with a baseline condition using the same slide and transcript context. Immediate learning was assessed with a five-item knowledge test, while subjective experience was measured using the User Experience Questionnaire (UEQ) and open-ended feedback. We did not observe clear differences in knowledge-test outcomes, but participants in the personalized condition reported significantly higher UEQ Stimulation. These results suggest that profile-based multimodal prompts may mainly support motivational engagement rather than immediate test performance, while larger and longer-term studies are needed to assess learning effects.
Furkan Ali Yurdakul, Yiman Wu, Maria Torres Vega et al.· Proceedings of Mensch und Co...· 0 citations
This mixed-methods study investigated the integration of the Echo Method with speech-to-text (STT) technology to enhance TOEIC listening instruction. The intervention addresses the washback effects of high-stakes testing—the influence of TOEIC on teaching and learning—by transforming test-focused drills into authentic, learner-centered listening practice. Forty-five Applied English majors at a Taiwanese technological university participated in an 18-week intervention featuring reflective echoing, STT-based self-monitoring, and collaborative transcript analysis. Quantitative results from four diagnostic tests (normalized to a 0–100 scale) showed significant listening gains, with weighted overall means rising from 73.98 (pretest) to 87.87 (posttest) (d = 1.12). Improvements were greatest in sections with lower cognitive load: Part 1 (Picture Descriptions, d = 1.48) and Part 2 (Question-Response, d = 1.52). Regression analysis revealed that STT engagement frequency, rather than class attendance, predicted listening gains (β = .62, p < .001). Qualitative data from journals, interviews, and surveys indicated that STT feedback enhanced metacognitive awareness (73%) and reduced listening anxiety (60%) by externalizing errors in a non-judgmental format. Learners also reported perceived improvements in speaking naturalness, suggesting receptive-to-productive transfer effects. Overall, the findings demonstrate that technology-enhanced articulatory rehearsal can bridge the gap between test-focused drills and communicative competence, transforming high-stakes preparation into a learner-centered experience.
Ching-Jung Yang· LEARN Journal : Language Edu...· 0 citations
Abstract The increasing incorporation of Artificial Intelligence (AI) tools in education urges researchers to examine their efficiency in Foreign Language (FL) learning and teaching. Large Language Models (LLMs) such as ChatGPT are becoming inevitable in promoting English as a Foreign Language (EFL). However, concerns have been raised about overreliance on LLMs, which may hinder learners’ Critical Thinking (CT) abilities. This study explores the use of ChatGPT to develop students’ capacity for critically engaging with LLM-generated content and assessing their abilities in reflective learning, fact-checking, logical reasoning, and bias detection. Using convergent parallel mixed methods, including quantitative rubric-based assessment and qualitative response analysis, students’ interactions with the LLM are evaluated based on Paul and Elder’s (2013) CT nine intellectual standards: clarity, accuracy, precision, relevance, significance, depth, breadth, logic, and fairness. Findings reveal that while advanced students critically reflect on AI responses, lower-proficient learners often adopt them uncritically, highlighting a gap in fact-checking and AI literacy. Many students also struggle to detect inconsistencies or bias, underscoring the need for explicit instruction on AI ethics and limitations. The study concludes that AI is useful for language learning, but its effectiveness depends on learners’ ability to critically evaluate and refine its outputs.
This quasi-experimental study investigated whether purposefully designed multimodal materials boost EFL learner engagement more than text-only materials over a 15-week semester with 122 Chinese undergraduates in four intact classes. The multimodal group received video clips, audio recordings, annotated texts, and interactive quizzes, whereas the control group received identical content through text alone. Pre-post test gains and a 14-item engagement questionnaire served as the dependent measures. The multimodal group outperformed the control group on learning outcomes (d = 1.05, an upper-bound estimate) and across all three engagement dimensions: behavioral (d = 0.84), cognitive (d = 0.74), and emotional (d = 0.50). These findings suggest that even modest technological investments, when guided by deliberate content-modality alignment, can produce practically meaningful improvements in how students engage with and learn from EFL instruction, although the single-instructor design limits the causal inference.
Shuqiong Fang· European Journal of Educatio...· 0 citations
This study explores the use of question-answer videos outside the classroom to improve speaking skills among English 3 EFL students at the National College of Education in Ho Chi Minh City. The study used a quasi-experimental design with 160 students in the treatment group (2023-2024) and 160 students in a historical control group (2022-2023). The treatment group used video-based practice to support their learning. Data were collected through final speaking test scores, questionnaires, and semi-structured interviews with teachers and students. The findings suggest that students who practised with the videos demonstrated greater improvements in grammar and vocabulary, pronunciation, discourse management, and interactive communication than the control group. Participants also reported increased motivation and confidence. Despite challenges related to fast speech, unfamiliar vocabulary, and cognitive overload, the findings suggest that structured video-based practice can support out-of-class speaking development in EFL contexts when tasks are carefully designed.
Hang Le Thi, Thanh Dau Thi, Duong Pham Ngoc Thuy· Proceedings of International...· 0 citations
This article investigates the transformative role of large language models (LLMs), specifically ChatGPT, in contemporary education and their impact on learning fatigue. It examines both the potential benefits of LLMs use as personalized learning, automated content creation, interactive tasks, and real-time feedback from one point of view and the emerging challenges, including increased dependence on digital devices and the risk of digital fatigue - mental, physical, and emotional exhaustion caused by prolonged screen exposure from the other. The results show widespread familiarity with artificial intelligence tools and moderate relationships between AI use and perceived reductions in learning effort. Frequent use of ChatGPT appears to implement improvements in homework, highlighting the need to further develop students’ digital literacy and prompting skills. The study concludes that the responsible, well-regulated integration of LLMs can enhance learning effectiveness while helping to mitigate risks associated with digital fatigue. This research is relevant for educators and institutions seeking to improve student outcomes through the thoughtful adoption of AI-based tools.
Galina Atanasova, Bagryana Ilieva· TEM Journal· 0 citations