This study investigates the effectiveness of Animal Crossing: New Horizons as a game-based learning tool for English vocabulary acquisition among elementary-level EFL learners in Taiwan, with particular attention to gender differences. While previous research on game-based language learning has largely focused on purpose-built educational games or reported ungendered outcomes, little is known about how commercial entertainment games support vocabulary learning across genders in formal classroom settings. To address this gap, a quasi-experimental mixed-methods design was employed with 130 sixth-grade students (aged 11–12). Participants completed vocabulary pre- and post-tests, self-report questionnaires, and guided gameplay activities supported by structured checklists and teacher-led discussions. Quantitative data were analyzed using paired- and independent-samples t -tests, while qualitative data were examined using qualitative content analysis of learner reflections. Results showed that gameplay significantly enhanced vocabulary acquisition for both genders; however, female learners demonstrated substantially larger and more consistent gains than male learners. In contrast, male learners reported higher perceived learning and enjoyment despite showing smaller measurable improvements. These findings suggest that pedagogical scaffolding plays a critical role in moderating gender-related motivational tendencies in game-based learning contexts. The study contributes empirical evidence that commercial narrative-based games, when systematically scaffolded, can function as effective and inclusive tools for situated vocabulary learning in elementary EFL classrooms.
A central goal of foundational dietary education is to enhance nursing students’ learning achievement, self-efficacy, and critical thinking awareness. However, providing personalized guidance in conventional instructional settings remains challenging. To address these limitations, the present study introduces a structured metaverse-based self-regulated learning framework designed to promote personalized, immersive, and contextually rich learning experiences in dietary education. A quasi-experimental study was conducted with 44 undergraduate nursing students enrolled in an introductory course on cancer-preventive dietary practices. Participants were assigned to either a metaverse-based self-regulated learning condition (experimental group) or a conventional 2D video-based instruction condition (control group). Findings indicate that students in the metaverse-based condition demonstrated significantly higher levels of learning achievement, self-efficacy, and critical thinking awareness than peers in the conventional instruction group. Exploratory behavioral analyses of the 22 experimental-group participants, categorized by a median split, suggested that learners with higher critical thinking awareness exhibited more structured, reflective, and proactive learning patterns, whereas those with lower awareness relied more on feedback-dependent strategies. Overall, metaverse-based self-regulated learning enhanced learning achievement, self-efficacy, and critical thinking awareness, while providing preliminary insights into learners’ behavioral patterns in nursing education.
As artificial intelligence (AI) becomes increasingly embedded in higher education, understanding how students perceive, interpret, and respond to AI-generated feedback is critical for designing pedagogically meaningful learning environments. This study investigates students’ experiences with TaskInsighter, a GenAI-supported feedback chatbot that combines a conversational interface, generative feedback, Socratic questioning, formative scoring, and code-specific prompts in a university-level Python programming course. Grounded in Self-Regulated Learning (SRL) theory and Sensemaking Theory, we conducted a qualitative analysis of 38 open-ended student reflections and 8 in-depth semi-structured interviews. Notably, the interviewed students had engaged with the chatbot continuously over a 16-week semester, allowing the study to examine how learners described their evolving interpretations, feedback engagement, and regulatory behaviors over time. The findings suggest that students perceived the chatbot as supporting metacognitive regulation, explanation-oriented learning, motivational engagement, and strategic learning behaviors, while also reporting usability challenges related to system constraints and feedback clarity. Rather than accepting AI feedback passively, students actively evaluated the chatbot’s responses and adjusted their learning strategies in response to feedback. These insights contribute to understanding how SRL operates in AI-mediated learning contexts and suggest implications for designing AI learning tools and instructional strategies to support autonomy, reflection, and student agency.
Role-playing activity has shown promise for language learning, yet many constraints of traditional pedagogical agents (e.g., teachers, peers) limit its practical implementation and full potential. While generative AI (GenAI) emerged as a promising additional conversational agent to enhance students’ educational experiences in such activities, existing research has mostly focused on speaking and writing, leaving its pedagogical effects on grammar acquisition and its strengths and weaknesses uncertain. The study compares the pedagogical effects and educational experiences of GenAI, teacher, and peer agents in grammar-focused role-playing activities. A total of 155 undergraduates participated in a 12-week within-class randomised experimental design. Results revealed distinct pedagogical profiles: GenAI and teacher agents produced equivalent overall learning gains, significantly outperforming peers; however, GenAI demonstrated superior effectiveness in the use dimension and cognitive presence, teachers excelled in the meaning dimension and teaching presence, while peers showed the highest social presence. These findings support that teachers, GenAI, and peers play complementary roles. We proposed a triadic agent framework for grammar-focused role-playing, offering practical and actionable insights. Theoretical implications, limitations, and future research were finally discussed.
While generative artificial intelligence (GenAI) is increasingly integrated into higher education academic writing, it introduces profound epistemic challenges such as AI hallucinations, including fabricated citations and data. Current literature predominantly focuses on macro-level technology acceptance, leaving students’ micro-level subjective perspectives relatively underexplored. This study investigates the subjective perspectives and evaluative orientations of university students when navigating AI hallucinations during human–AI collaborative writing. Employing Q methodology structurally anchored in the Cognitive Process Theory of Writing, 45 humanities and social science students prioritized a refined set of 40 statements. Quantitative factor analysis complemented by qualitative post-sort interviews identified five factor-defined perspectives: Autonomous Directors, Pragmatic Operators, Epistemic Sentinels, Conflicted Compromisers, and Creative Opportunists. These perspectives reflect shared configurations of epistemic trust, authorship control, and verification responsibility, which are interpreted in relation to cognitive agency rather than treated as separately measured constructs or fixed categories of students. The findings offer empirical material for formulating tentative, data-informed pedagogical hypotheses concerning different patterns of epistemic and cognitive support in AI-assisted academic writing; these hypotheses require validation through future intervention-based research.
Grounded in experiential learning theory, this study proposed a virtual reality-facilitated creative thinking mapping (VR-CTM) approach, designed to guide learning through immersive experiences. To evaluate its effectiveness, a quasi-experimental design was implemented to compare the VR-CTM approach with a conventional creative thinking mapping (C-CTM) approach. Two classes of students enrolled in elective university performing arts courses were recruited. One class (N = 45) was the experimental group adopting the VR-CTM approach; the other class (N = 40) was the control group adopting the C-CTM approach. Results indicated that students adopting the VR-CTM approach demonstrated significantly better performance skills than those adopting the C-CTM approach. Furthermore, for learners with higher initial levels of choreographic creative expression, students in the VR-CTM group demonstrated better post-intervention performance than those in the C-CTM group. For learners with lower initial collective self-efficacy, students in the VR-CTM group demonstrated greater post-intervention self-efficacy than those in the C-CTM group. Finally, the creative thinking maps produced by the VR-CTM group contained more specific and comprehensive creative planning than those produced by the C-CTM group.
The emergence of the metaverse marks a transformative shift in digital education, offering immersive, collaborative and personalised learning experiences. However, its convergence with artificial intelligence (AI) raises notable concerns related to data privacy, learner security and ethical transparency. Whilst various studies have explored these domains independently, comprehensive understanding regarding the intersection of security and privacy-preserving AI methods within educational metaverse environments is still lacking. Aiming to address this gap, this paper presents a comprehensive systematic review conducted in accordance with the PRISMA protocol, synthesising current research on the integration of AI with security and privacy mechanisms in metaverse-based learning systems. A total of 60 research studies published between 2020 and 2025 were examined in the present systematic review. The review focuses on key educational domains, including language learning, STEM education and health education, and critically examines the role of AI techniques such as federated learning, differential privacy, blockchain and machine learning to enhance trust and data governance in immersive environments. Based on this synthesis, we introduce the Secure and Ethical AI Framework (SEAF), a conceptual model designed to guide the development of secure, privacy-aware and pedagogically aligned metaverse learning environments. SEAF integrates ethical, technical and educational considerations to support trustworthy AI-driven learning. This study contributes to the field of educational computing by deepening theoretical insights and offering a practical framework for implementing AI-driven metaverse systems that prioritise ethical learning and digital trust. Additionally, the study identifies key research gaps and outlines a future research agenda to support interdisciplinary collaboration at the intersection of AI, privacy and educational innovation.
Generative AI (GenAI) has rapidly emerged as a promising tool for supporting self-regulated learning (SRL). However, little is known about how its mechanisms compare with, or extend beyond, earlier artificial intelligence (AI) approaches. Using Winne and Hadwin’s COPES (Conditions, Operations, Products, Evaluations and Standards) architecture as an organizing framework, this comparative systematic review synthesized 70 articles from 2015 to 2025 to examine how AI and GenAI were technologically and pedagogically implemented to scaffold SRL. Technically, SRL systems predominantly employed knowledge-based systems, machine learning, natural language processing, and, more recently, customized large language models, with limited integration of external knowledge bases. Pedagogically, interventions in the reviewed studies concentrated on Operations and Evaluations during task enactment, with less attention to task definition, goal setting, and adaptation. In terms of operational distinctions between AI and GenAI, AI tended to instantiate analytics-driven scaffolding (data-based adaptivity, progress monitoring, and standards setting). In contrast, GenAI more often enabled dialogic, context-adaptive cognitive scaffolding and content generation. Building on these findings, this review proposes a framework for comprehensive SRL scaffolding.
Cognitive load is fundamental to learning success, but most mobile learning studies address it only at a surface level. This approach limits understanding of how instructional designs manage complexity, reduce mental burdens, and create space for deeper processing. This paper addresses this gap by reviewing 53 studies published between 2016 and 2025. Results include the identification of prevalent strategies such as autonomy, scaffolding, and practical scenarios. These practices’ associations with intrinsic and extraneous load were examined using epistemic network analysis, while germane-aligned strategies were thematically analyzed for cognitive purposes and challenges. Findings revealed that intrinsic load is primarily addressed through the structuring and sequencing of task complexity, while extraneous load emerges from how multiple instructional features are integrated within learning designs. Although germane load was excluded from most measurements, the reviewed studies consistently embedded designs aimed at promoting deeper learning. This reflects germane processing as a guiding principle for instructional design rather than a directly measured outcome. Learning achievement in mobile learning should be interpreted in relation to instructional coherence that supports germane processing and manages cognitive burdens rather than technology use alone. Alignment across technology, pedagogy, and learner agency emerges as a key design consideration that requires early decisions.
Grounded in activity theory, this study conducted a systematic review and meta-analysis, following PRISMA guidelines, to evaluate the impact of generative AI-powered conversational agents on students’ learning outcomes in both cognitive and non-cognitive skills. Twenty-seven empirical studies published between 2022 and 2025 were analyzed. The results revealed that generative AI-powered conversational agents had moderately positive effects on both cognitive ( g = 0.462, p < .001) and non-cognitive skills ( g = 0.519, p < .001). Moderator analyses identified several contextual variables—educational levels, knowledge types, intervention durations, agent roles, and instructional modes—as significant or marginally significant. These findings offer practical implications for educators, developers, and researchers, aiming to enhance the effectiveness of generative AI-powered conversational agents in improving learning outcomes in educational settings.
This editorial presents a structured synthesis of the thematic development of the Journal of Educational Computing Research (JECR) from 1987 to 2024, drawing on a computational analysis of 1,597 articles. Applying Latent Dirichlet Allocation (LDA), the analysis identifies thirteen thematic structures that reveal both the enduring commitments and the emerging orientations of the journal across nearly four decades. Themes such as problem-solving through programming and self-efficacy in programming education have constituted persistent research agendas since the journal’s earliest volumes, while computational thinking and game-based learning have gained substantial momentum since 2015. These patterns suggest that JECR’s intellectual identity is best understood not as a sequence of discrete topical shifts, but as a sustained process in which a bounded set of foundational problems is repeatedly re-articulated under changing technological and epistemic conditions. The editorial situates this trajectory in relation to contemporary developments in artificial intelligence and educational technology, arguing that the journal’s historical archive offers an indispensable conceptual framework for interpreting present-day transformations.
Generative AI (GenAI) is now common in university project work, yet previous studies often examine students’ trust, creativity, overload, or engagement separately. This leaves a key gap: how students regulate GenAI across a full project workflow. This exploratory study addresses that gap by examining a four-process “ AI-mediated Self-Regulated Learning ” ( AI-SRL ) cycle: (1) evaluating and selecting GenAI suggestions, (2) experiencing shifts in creative agency, (3) managing overload through filtering and summarizing, and (4) monitoring time and energy to stop or continue working with GenAI. We conducted a two-course basic qualitative design study with 97 undergraduate and graduate students. Data came from an open-ended questionnaire aligned to the four processes. We used inductive content analysis with a shared codebook, reliability checks, and cross-level comparisons. Findings show that students use combinations of strategies across the AI-SRL cycle. They exercise agency through goal alignment, revision, and verification, with graduates reporting stronger cross-checking and source-based justification. Creativity was described as a conditional outcome: it increased when GenAI widened ideas but declined when it replaced personal exploration. Overload was managed through targeted prompts and structured outputs, again more common among graduates. Most students did not lose track of time; they used clear stopping cues such as fatigue, repetition, or satisfaction. Together, results reveal two distinct metacognitive regulation styles: “ Exploratory-Simplification ” and “ Systematic-Methodical ”.
This umbrella review provides a thorough synthesis of evidence on the impact of virtual reality on three domains of learning (cognitive, affective, and psychomotor), informed by Bloom’s Taxonomy. Content analysis was applied to 93 articles comprising 70 systematic reviews, 17 systematic reviews and meta-analyses, and 6 meta-analyses, extracted from three databases (i.e., Web of Science, Scopus, and IEEE Xplore). The results indicated that affective learning in VR was most widely reported, followed by cognitive and psychomotor domains of learning. In the cognitive domain, VR showed more consistent benefits, especially for knowledge acquisition and higher-order cognitive skills (e.g., problem-solving, intellectual skills), although a few meta-analyses reported non-significant effects. Across the affective domain, findings were mixed but generally suggested VR could improve self-efficacy, self-confidence, and satisfaction, while evidence for various affective outcomes remained highly inconsistent. For the psychomotor domain, VR often enhanced motor, procedural, and safety-related skills, but results specifically for complex clinical skills were inconclusive. Future research should advance theory through exploring why and how VR influences learning while strengthening research designs and standardized measurements to determine under what conditions VR produces reliable improvements.
Despite growing calls to challenge epistemic inequalities within business education, there remains a lack of systematic and scalable approaches for identifying coloniality within routine curricular artefacts. This study addresses this gap by asking: How can a reproducible AI-assisted workflow be developed to identify and audit colonial markers within MBA PowerPoint lecture slides? This ‘Systems and Tools’ article adopts a sequential three-phase research design. First, a Colonial Markers Framework was developed through iterative expert review, AI-assisted analysis, and comparative thematic synthesis of MBA teaching materials from two programmes within a UK university. This process resulted in seven colonial markers relating to language and terminology, inclusivity and representation, capitalist and economic focus, Eurocentrism and Western dominance, epistemological dominance and knowledge production, historical context and oversight, and case study and example bias. Second, the framework was implemented within an exploratory Langflow-based workflow to assess the feasibility of automated colonial marker detection. While the approach demonstrated promising alignment with expert interpretations, technical limitations relating to document processing, citation accuracy, and workflow stability constrained its effectiveness. Third, insights from this exploratory phase informed the development of a more robust LangChain-Python workflow incorporating optical character recognition, automated reporting, structured data generation, and enhanced reproducibility. Findings demonstrate substantial convergence between AI-generated and expert-identified colonial markers, suggesting that AI can support the systematic auditing of teaching materials when guided by a human-developed analytical framework. Exploratory evaluation with MBA programme leads further indicated that the resulting outputs were interpretable, accessible, and useful as prompts for reflective curriculum review.
This editorial reflects on the past year of the Journal of Educational Computing Research, highlighting record impact factor and submission growth alongside initiatives to strengthen transparency, ethical reporting, and methodological rigour. Key developments include the introduction of a mandatory ethical statement, expansion of the editorial board, the introduction of the RAISE framework for AI-focused research, and evolving attention to future research needs. Submission patterns, editorial priorities, and future objectives are discussed, highlighting the journal’s commitment to stewardship, scholarly standards, and responsible, innovative research.
The COVID-19 pandemic catalyzed a dramatic reshaping of the educational technology landscape, creating an urgent need to analyze the field’s new trajectories and inform future directions. This study addresses that need by uniquely combining bibliometric analysis with machine learning forecasting to map and predict research trends using a comprehensive dataset of 9,630 articles from 20 high-impact journals (2020–2024). Our analysis reveals three significant findings: (1) research is heavily concentrated on scalable technologies like MOOCs and AI, while critical gaps persist in equity, open educational resources, and micro-learning; (2) methodological sophistication is increasing, with a rise in mixed-methods and meta-analytic approaches; and (3) nascent interdisciplinary connections with fields like healthcare offer promising new research frontiers. Forecasting models predict exponential growth in AI and computational thinking research through 2030, while pandemic-specific topics are expected to decline. These results provide a strategic roadmap for researchers, policymakers, and practitioners to prioritize underexplored areas, foster impactful interdisciplinary collaborations, and critically address the ethical implications of emerging technologies to build a more equitable and effective educational future.
Computational thinking (CT) is crucial for enhancing students’ complex problem-solving abilities in the intelligent era. The emergence of generative artificial intelligence (GenAI) is profoundly transforming the global educational landscape and demonstrating significant potential for promoting personalized learning. However, the literature offers varied results on the effectiveness of using GenAI to cultivate students’ CT. This study comprehensively investigated the effects of GenAI on students’ CT and the role of moderating factors, integrating 45 effect sizes from 25 empirical studies published between 2022 and 2025. A theoretical framework of factors influencing students’ CT was proposed based on activity theory, and the moderating factors included educational level, region, intervention duration, teaching mode, interaction mode, role setting, and feedback type. The results indicated that GenAI had a significant overall positive effect on students’ CT development. Specifically, the largest effect size was computational practice, followed by computational concept and computational perspective. Furthermore, the analysis revealed that region, teaching mode, and interaction mode had significant moderating effects. Based on these results, this study offers targeted implications across the dimensions of theoretical foundation, educational practice, and technological development, providing empirical evidence for implementing GenAI teaching and developing GenAI tools to cultivate students’ CT.
Generative Artificial Intelligence (GenAI), exemplified by models such as DeepSeek and ChatGPT, is rapidly reshaping education by fostering new pedagogical approaches, including personalized learning, adaptive feedback, and multi-modal instruction. This pedagogical transformation has led to a growing number of review studies examining the applications of GenAI applications across diverse educational contexts. Existing reviews tend to concentrate on various dimensions, such as educational levels, subject domains, or particular GenAI tools and their applications to support teaching and learning. However, to the best of our knowledge, no meta-review has yet been conducted to systematically examine and consolidate the findings of existing review studies on GenAI in education. To address this gap, the present study conducts a systematic meta-review of 35 published reviews, guided by PRISMA protocol. The analysis is structured around three key dimensions: methodological characteristics, thematic focus, and existing issues. Results revealed both advances and inconsistencies in methodological characteristics, including variation in database selection, search strategy transparency, and quality appraisal. The thematic focus shows diverse applications of GenAI across educational levels and disciplines, yet lacks theoretical grounding and comprehensive evaluation of learning outcomes. Furthermore, although the reviews acknowledge GenAI’s potential benefits, few offer concrete strategies to mitigate identified risks such as bias, over-reliance, or ethical concerns. This meta-review provides an integrated overview of the current evidence base and identifies directions for future research to support more rigorous, equitable, and pedagogically sound implementation of GenAI in education.
Research on AI in education has expanded rapidly, resulting in numerous first-order meta-analyses, yet these studies are limited by inconsistent findings, heterogeneous methods, variable quality, and overlapping primary studies, requiring a second-order meta-analysis to obtain a more reliable estimate of AI’s effects. This study examined the overall effect of AI applications on student outcomes, including academic achievement and higher-order thinking skills. We included first-order meta-analyses published between 2020 and 2025, covering primary studies published from 1993 to 2024. Nineteen meta-analyses obtained from electronic databases, involving a total of 58,702 participants, were analyzed. Using a random-effects model, we found a statistically significant moderate mean effect size (ES = .67, 95% CI [.55–.78]), indicating that AI technologies meaningfully contribute to student learning. The moderator analysis revealed that the moderators influencing the variability of effects of AI on student outcomes include education level, education field, and publication bias status. Effect sizes were robust across AI types and learning outcome types. Meta-regression showed that sample size and publication year did not predict effect sizes, whereas the number of primary studies did. These findings highlight the need for informed AI integration, strengthened pedagogical and institutional capacity, and evidence-based strategies to ensure meaningful improvements in student learning.
AI chatbots have emerged as innovative educational tools and drawn increasing attention from educators and researchers in programming education. Although previous research has highlighted potentials of applying AI chatbots in programming education, there is a lack of empirical evidence to understand the overall effects of using AI chatbots in programming learning as well as the critical factors that influence the effects. To fill this gap, this study conducted a meta-analysis of 32 empirical studies published between 2015 and 2025 to examine the overall effect size of applying AI chatbots on programming learning performance and identify significant moderators. The results indicated a small-to-medium effect on posttest performance ( g+ = 0.538, 95% CI [.202, .873], p < .01) and a medium-to-large effect on practice performance ( g+ = 0.650, 95% CI [.330, .970], p < .001), based on robust variance estimation models. Moderator analyses revealed that research design and AI chatbot-to-student ratio significantly influenced posttest performance. Specifically, true experimental designs demonstrated significantly larger effects than quasi-experimental designs, and a 1:1 chatbot-student ratio was substantially more effective than a 1:N ratio. These findings underscore the potential of AI chatbots in programming education and offer practical insights for optimizing their integration into instructional design.
