Social Sciences

Abstract: Second Language Acquisition (SLA) has long been marked by theoretical fragmentation. Since the 1970s, scholars have explained learner language through constructs such as interlanguage, variability, fossilization, and subsystem interaction. Cognitive models emphasize memory and processing limits; generative approaches highlight innate constraints; sociocultural theories stress mediation and identity; and complexity theory frames SLA as a non-linear system operating at the “edge of chaos.” Each captures a partial truth, yet none provides a unified account of how learners achieve full acquisition of L2. This landscape has led to the portrayal of L2 learning as chaotic, unpredictable, and incomplete. Interlanguage is often described as a transitional grammar shaped by subsystems that interact dynamically but never fully converge. Complexity theory has attempted integration, suggesting acquisition thrives in turbulence, but this remains partial and leaves residues of fragmentation that complicate mastery. This article challenges that predominant view. It argues that interlanguage is not an independent linguistic system but the visible outcome of the learning subsystem a learner follows when acquiring L2. The Law of the Trio offers a paradigm shift: language, thought, and reality are equivalent modalities of existence. Under this universal law, L2 acquisition is not chaotic interplay but a natural developmental process identical to L1, where learners form language by engaging directly with reality. By smoothing out the contradictions of subsystem theories, the Law of the Trio unifies fragmented perspectives into a science of meaning. Empirical evidence strengthens this ontological reframing. Numerous L2 acquirers — Joseph Conrad, Vladimir Nabokov, Chinua Achebe, Ha Jin, Baalu Girma, and Tsegaye Gebremedhin — have not only achieved parity with natives but surpassed them, transforming and advancing their adopted languages. Their contributions demonstrate that L2 acquisition is ordered creativity, not transitional chaos, and that learners are agents of linguistic innovation rather than deficient imitators. By situating SLA within the Law of the Trio, this article moves beyond fragmented subsystem theories and deficit models. It clarifies language learning as the exposition of language’s role as a mirror of being- a process of ontological resonance rather than chaotic struggle. In doing so, it directs SLA research from complexity to simplicity, from chaos to order. It dissolves the residues of fragmentation, validates learner creativity, and reclaims language as a mirror of being — offering a unified science of meaning for the future of SLA research.

Abstract: Background: The public release of generative artificial intelligence (AI) tools in late 2022 has spurred rapid interest in preparing health professionals to understand, use, and critically appraise AI. "AI literacy" has emerged as a focal construct, yet the resulting literature has grown so quickly that its structure, thematic composition, and gaps remain poorly characterized. Objective: To map the recent literature on AI literacy in medical and health professions education, identify its principal research themes and their growth trajectories, and highlight gaps to guide future work.Methods: We conducted a bibliometric scoping review. Records indexed in OpenAlex (2020 to mid-2026) matching a family of AI-literacy phrases were retrieved (n = 12,326 screened) and then filtered by an automated rule to retain those with a strong medical/health-professions signal, yielding 1,141 papers. English-language records with a substantive abstract (n = 957) were grouped into candidate themes using unsupervised text clustering (TF-IDF with latent semantic reduction and k-means), and the themes were then interpreted and characterized by two reviewers. Reporting follows the PRISMA extension for Scoping Reviews (PRISMA-ScR). Coverage of the single source was cross-checked against PubMed. Results: The subfield is overwhelmingly recent: fewer than 30 papers appeared in any year before 2023 (single digits in 2021), rising to 28 in 2023, 86 in 2024, and 452 in 2025, with 553 in the first half of 2026 alone. A compound annual growth rate of roughly 302% across the full years 2023–2025. Eight interpretable themes emerged, spanning governance and regulation; classroom instruction and diagnostic reasoning; knowledge and perception surveys; AI-literacy scales and attitudes; competency frameworks; generative AI tools; nursing education and ethics; and evidence syntheses. The fastest-growing themes were AI-literacy scale/attitude studies (17.7×) and governance and regulation (14.8×). Nursing was the most-represented discipline (345 papers), ahead of medicine (290). The most-cited works remained a small set of 2021–2023 curriculum-readiness surveys and evidence syntheses. Conclusions: AI literacy in health professions education has evolved over two years from asking whether to teach AI to measuring learner readiness and governing institutional adoption. Progress is constrained by a proliferation of non-harmonized measurement instruments, a relative scarcity of validated competency standards, and few studies linking AI-literacy interventions to clinical performance or patient outcomes. Harmonized measurement, consensus competency frameworks, and outcome-anchored evaluations are priorities.

Abstract: Virtual reality (VR) has moved from experimental classroom demonstrations to practical instructional platforms across K-12, higher education, and workforce training. Yet many VR learning environments still rely on static content, fixed difficulty, and limited feedback, which restricts personalization and scalability. Artificial intelligence (AI) offers complementary capabilities—generative content creation, adaptive sequencing, conversational tutoring, multimodal behavior analysis, and learning analytics—that can transform VR from immersive presentation into intelligent instruction. This paper reviews how AI is reshaping VR across the educational pipeline and proposes a conceptual AI-VR educational framework that links intelligent technologies, immersive environments, adaptive learning, real-time analytics, and measurable educational outcomes. The framework explains how AI supports content generation, virtual instructors, personalized learning paths, student behavior analysis, real-time feedback, performance assessment, learning analytics, and continuous improvement. Prior work by the authors on VR-centric behavioral sensing and adaptive thresholding is interpreted as an evolutionary pathway from physiological monitoring toward education-oriented intelligent VR systems. The discussion emphasizes that modern VR becomes substantially more effective when integrated with AI rather than deployed as standalone immersion. Challenges related to cost, teacher readiness, data governance, and equitable access are outlined together with future directions for lightweight analytics, explainable adaptation, and classroom-ready design. The manuscript is intended as a practical, publishable synthesis for educational technology researchers and practitioners seeking a clear roadmap for AI-enhanced immersive learning. Recent syntheses further confirm growing research interest at the intersection of immersive media, generative AI, and learning analytics, positioning AI-enhanced VR as a timely topic for educational technology review and framework development.

Abstract: Hands-on laboratory experiences can strengthen science learning, motivation, and science identity, yet access to molecular biology infrastructure remains limited in socioeconomically vulnerable school contexts. This mixed-methods case study examined learning outcomes and student perceptions following a 16-hour Molecular Biology workshop delivered by CRIDESAT scientists from the University of Atacama to 152 secondary students from eight high schools in the Atacama Region, Chile. The intervention combined theoretical instruction, laboratory activities involving DNA extraction, polymerase chain reaction (PCR), and agarose gel electrophoresis, and discussions on biotechnology applications and bioethics. Students completed a 25-item knowledge assessment and a satisfaction survey comprising seven Likert-scale items and three open-ended questions. The survey demonstrated high internal consistency (Cronbach’s α = 0.88), and qualitative responses were analyzed using thematic analysis. Students achieved high knowledge scores across all content areas, with 85–92% correct responses and mean section scores above 8.5/10. Satisfaction ratings were consistently positive (>4.5/5). Qualitative findings highlighted positive attitudes, perceived learning, and experiences related to “doing science”, including statements suggestive of emerging science identity. These findings suggest that university–school partnerships can expand science capital and provide equitable access to authentic scientific experiences, supporting STEM education in socioeconomically vulnerable educational contexts.

Abstract: Background. Artificial intelligence (AI) literacy has become a rapidly expanding concern in higher education, accelerated by the public release of generative AI tools. The volume and thematic structure of this literature remain systematically unmapped. Objective. To map the scope, growth, thematic structure, disciplinary distribution, and geographic distribution of research on AI literacy in higher education from 2020 to mid-2026. Methods. We queried OpenAlex for works matching an AI-literacy phrase family in the title or abstract (2020–2026) and applied a strong-signal higher-education filter (a higher-education term in the title or ≥2 distinct such terms across the title and abstract). Records were deduplicated, and those with a usable English abstract were embedded using TF-IDF and reduced via truncated SVD, then grouped by k-means (k = 8; one language-artifact cluster excluded, leaving seven substantive themes). Cluster stability was quantified via seed resampling. The review was conducted per PRISMA-ScR. Results. The strong-signal corpus comprised 1,910 works; 89.4% appeared in 2025–2026. Annual output rose from single digits through 2022 to 750 in 2025, with 958 already recorded in the first half of 2026. Clustering of 1,483 abstracted works yielded seven themes, the largest of which were AI-literacy courses and curriculum design (n=342) and institutional integration, policy, and academic integrity. The fastest-growing themes were systematic/scoping reviews and qualitative studies of writing and AI-literacy components (both >18× 2025–26 vs 2023–24). Research was concentrated in the humanities/languages and STEM disciplines and led by the United States, China, and Indonesia. Conclusions. AI literacy research in higher education is expanding rapidly and is organizing around curriculum design, institutional policy, and instrument development. The evidence base is young, survey-heavy, and geographically concentrated, indicating a need for longitudinal and intervention studies.

Abstract: Outcome-based Arabic teacher education requires graduate outcome monitoring, yet program-specific online tracing in Indonesian Islamic universities remains under-documented. This documentary case study analyzes post-enrollment digital student support and alumni engagement—not online instruction—in an Arabic Teacher Education Study Program. The 2025 Self-Evaluation Report (LED; Laporan Evaluasi Diri) was cross-validated against Independent Accreditation Institute for Education (LAMDIK) field assessment minutes and coded for tracer workflow, outcome indicators, and quality-assurance follow-up. Online graduate tracing operates through the Talian platform (talian.uir.ac.id), SMS reminders, alumni networks, and a real-time dashboard (app.uir.ac.id), linked to upstream WhatsApp-based retention monitoring. Across study-year cohorts TS-4 to TS-2 (Tahun Studi), 39 of 42 graduates (92.86%) were traced; 97.5% were employed, self-employed, or studying further; average waiting time to first employment was 4.8 months. Employer surveys flagged relative weaknesses in foreign-language and information-technology domains (58.24% and 69.23% “very good,” respectively). The LED does not report employer sample size, and domain percentages are not independently verified competence measures. Documented institutional responses, planned SMART targets (2026–2027), and unevaluated follow-up are classified separately (Table 11). The study offers a descriptive workflow model for technology-enhanced post-program monitoring in face-to-face teacher education.

Abstract: Medical residency is a demanding period of specialty training, marked by intense clinical exposure, heavy workload, and continuous knowledge development. Identifying how residents prefer to learn may help residency programs refine their educational strategies and better align teaching approaches with the needs of trainees. This study assessed learning-style preferences among medical residents using the Felder–Soloman Index of Learning Styles. A total of 152 residents from clinical and surgical specialties in a Brazil-ian medical residency program were included. In the overall sample, the predominant preferences were sensing (90.79%), visual (73.68%), active (62.50%), and sequential (70.39%). Comparisons between clinical and surgical specialties showed a significant dif-ference only in the understanding domain, with residents from clinical specialties being more frequently classified as sequential than those from surgical specialties (80.0% vs. 63.2%). The limited variation across groups suggests that most residents share a broadly similar learning profile, although some cognitive-processing preferences may differ ac-cording to specialty-related demands. These findings support the use of diversified educa-tional strategies in medical residency, combining visual resources, active learning meth-ods, and structured practical activities adapted to clinical and surgical training contexts.

Abstract: African distance higher education is increasingly shaped by open educational resources (OER) and generative artificial intelligence, yet these agendas are often governed separately. This separation is problematic because OER and AI now meet in resource discovery, translation, adaptation, tutoring, feedback, assessment, accessibility, datafication, and knowledge production. This article develops an OER-AI Equity Framework for African distance higher education. It uses an integrative critical literature review, policy and document synthesis, and conceptual framework development. The synthesis draws on peer-reviewed scholarship, Open Praxis research, UNESCO guidance, African Union digital and AI strategies, connectivity data, accessibility standards, and literature on open pedagogy, digital equity, academic integrity, learner support, epistemic justice, and teacher agency. The analysis shows that equitable integration requires more than low-cost content and tool access. It requires seven interdependent commitments: infrastructural realism; epistemic justice and localisation; open pedagogy and learner agency; accessibility and learner support; assessment redesign and academic integrity; data protection and algorithmic accountability; and teacher agency with institutional policy coherence. The framework contributes a socio-technical, pedagogical, and governance-oriented lens for institutions, instructors, policymakers, quality assurance bodies, libraries, and researchers. It argues that African learners should not be positioned as passive consumers of imported content, platforms, or models, but as capable participants in open and AI-mediated knowledge systems.

Abstract: The rapid expansion of generative artificial intelligence (GenAI) in higher education has intensified debates concerning misinformation, media literacy, academic integrity, higher education governance and regulatory approaches. This study develops a systematic literature review (SLR) following PRISMA 2020 guidelines to examine how recent literature and higher education governance address the relationships between GenAI, misinformation, media literacy, AI literacy and educational governance within university contexts. The review integrated empirical studies, systematic and scoping reviews, institutional documents, university policies and international regulatory frameworks through a comparative thematic synthesis and evidence-based extraction strategy. Findings indicate a persistent tension between the pedagogical opportunities associated with GenAI and the epistemic, ethical and informational risks linked to synthetic content production and informational dependency. The review also shows that media literacy, AI literacy and critical thinking emerge as recurrent educational responses to AI-mediated misinformation. Furthermore, substantial differences were identified across governance frameworks and international regulatory approaches. The conclusions suggest that GenAI governance requires integrated approaches capable of connecting higher education governance, critical pedagogies and evaluative competencies within increasingly AI-supported educational ecosystems.

Abstract: Physical education (PE) is commonly justified through its contributions to physical fitness, motor skill development, sport participation, and health-enhancing physical activity. These aims remain important, but they do not fully capture PE’s potential within the broader landscape of lifestyle behavior and neurobehavioral well-being. This narrative review examines PE as a curriculum-based lifestyle learning environment through which movement experiences may be linked with motivation, stress regulation, emotional regulation, resilience, sleep and recovery awareness, nutrition-related awareness, and lifestyle self-regulation in adolescents and young adults. Using a narrative review design with structured literature mapping, the review integrates evidence and concepts from PE pedagogy, physical literacy, health literacy, Self-Determination Theory, stress and emotion regulation, self-efficacy, resilience, self-regulated learning, sleep, recovery, nutrition-related education, and school and university health-promoting frameworks. The review argues that PE should not be understood only as a setting for activity delivery, fitness testing, or sport-skill instruction. Rather, PE may provide repeated educational opportunities for students to connect embodied movement experiences with everyday lifestyle regulation. The proposed framework does not claim that PE automatically improves mental health or directly produces neurobiological change. Instead, it positions PE as one distinctive educational context in which movement, reflection, social interaction, and health-related learning can be integrated to support neurobehavioral well-being. Future research should examine curriculum mechanisms, teaching strategies, transfer processes, implementation conditions, and measurement indicators for PE-based lifestyle learning.

Abstract: This paper examines the threat to academic integrity posed by students who may use contemporary technology to cheat undetectably in traditional written university examinations. The threat is a cheating pipeline formed by the convergence of three technologies: miniaturised cameras and earbuds; consumer AI smart glasses with near-invisible Heads-Up Displays; and Large Language Models capable of answering university-level questions to a passing standard or higher. The paper argues that, if used, this pipeline can compromise the integrity of closed-book, supervised examinations. Once covert assistance of this kind is feasible, the format can no longer reliably distinguish a candidate's own work from a machine's. It can therefore no longer substantiate the claim to competence that a degree makes on its holder's behalf. This is a structural problem: it concerns what the examination can certify for the cohort as a whole, not only the conduct of those who cheat. The evidence establishes that the threat is already real. Drawing on a narrative review of the academic-integrity literature (1966-2025), UK Ofqual malpractice statistics (2019-2025), and a documentary scan of commercially available AI-assisted devices (May 2026), the paper shows that device use in invigilated examinations is established and increasing in secondary education, with the conditions driving it plausibly extending to higher education, although direct sector-level evidence remains limited. It develops the ethical case that examination design and certification must be reconsidered as a matter of institutional responsibility. It concludes that what a degree certifies in the age of ambient AI cannot be left to detection technology to settle.

Abstract: Generative artificial intelligence (GenAI) is becoming increasingly relevant to foreign language education, as tools such as ChatGPT are used to support teaching, learning, feedback, materials development, and assessment-related activities. However, AI-generated language does not automatically lead to language learning. Its educational value depends on how teachers design prompts that support learner effort, communicative practice, revision, reflection, and responsible classroom use. This article argues that prompt engineering should be understood not merely as a technical skill, but as an extension of foreign language teachers’ pedagogical expertise. It first situates prompt design within key principles of language teaching, and then provides practice-oriented guidance for designing ChatGPT-supported tasks by clarifying learning goals, learner and AI roles, task stages, output limits, feedback boundaries, monitoring procedures, and transparency requirements. The article presents prompting techniques, practical examples, reusable templates, classroom applications, model lessons, and prompt-repair strategies for common foreign language teaching purposes. It also discusses ethical and pedagogical risks and offers safeguards for responsible use. The article concludes that pedagogically informed prompt engineering may help language teachers use ChatGPT as a supportive classroom tool while preserving teacher judgment, learner authorship, communicative participation, and conditions for meaningful language development.

Abstract: This study evaluated the construct validity of the Child Assessment and Education System (CAES) in Hong Kong early intervention services using a cross-sectional Multitrait–Multimethod (MTMM) framework. Reliable CAES scores have been previously established, but validity evidence for score interpretation across methods and objective indicators remained limited. CAES domain scores (Gross Motor, Fine Motor, Social, Cognition, Self-regulation, Language) were compared with theoretically related questionnaire benchmarks from multiple validated instruments administered to 336 children. Objective multi-method cross-validation was further conducted in a randomly selected subsample of 180 children using eye-tracking attention metrics, AI-assisted dyadic interaction coding (MEACI), and salivary cortisol. Results indicated that CAES domains show meaningful validity-relevant structure: convergent and discriminant validity patterns were supported through selective trait–method associations and improved measurement model fit. However, cross-validation with objective indicators was domain-dependent. Among emotion components, CAES Emotion Expression (EE) aligned significantly with salivary cortisol, whereas Emotion Identification (EI) and Emotion Regulation (ER) showed no meaningful correspondence with the eye-tracking metrics used. Overall, CAES demonstrates construct validity for intervention profiling and provides clear directions for refining objective mappings and future longitudinal evaluation.

Abstract: Background: Rwanda’s higher education sector is increasingly expected to support digital transformation, workforce development, flexible learning, and international comparability of qualifications. However, the quality of online and blended programmes cannot be assured by digital platforms alone. It requires coherent alignment among regulation, curriculum design, pedagogy, assessment, accessibility, staffing, infrastructure, learner support, data protection, artificial intelligence governance, and sustainability. Purpose: This article develops the Rwanda Online and Blended Programme Design and Quality Assurance Framework (ROB-PDQAF), a policy-informed framework and practical toolkit for higher learning institutions, programme development teams, internal quality assurance units, and external reviewers. Method: The study applies qualitative document analysis, integrative literature synthesis, standards mapping, and framework construction. Public Rwandan policy and regulatory documents were analysed alongside international and regional quality assurance standards and peer-reviewed literature on online learning, instructional design, assessment, accessibility, open educational resources, data governance, and AI in education. Results: The ROB-PDQAF comprises ten interdependent domains: strategic relevance and regulatory alignment; curriculum coherence and constructive alignment; online and blended course design; digital content, multimedia, OER, and copyright; accessibility, inclusion, and learner support; assessment, feedback, academic integrity, and AI governance; staffing and eLearning support; LMS, infrastructure, data protection, and technical readiness; internal quality assurance and continuous enhancement; and financial planning, partnerships, scalability, and sustainability. Originality: The article contributes a Rwanda-contextual, evidence-based framework that conceptualises online and blended programme quality as an academic ecosystem rather than a technology adoption exercise. Implications: The framework translates policy and research evidence into reviewable domains, evidence requirements, risk indicators, and practical tools for programme design, accreditation, and continuous enhancement.

Abstract: This qualitative case study aims to explore how three Chinese primary school teachers understand, identify and provide for gifted students in ordinary classroom contexts. Drawing on semi-structured interviews with teachers from rural Shandong, an urban public school in Northeast China and an urban school context in Shanghai, the article examines gifted provision as a situated practice shaped by classroom routines, assessment pressure, family resources, school culture and teacher agency. The study uses thematic analysis to compare teachers' accounts of giftedness, identification, differentiation, institutional support and constraints. The findings show that teachers recognize high potential through everyday signs such as rapid comprehension, accurate responses, independence, advanced preparation, wide reading, curiosity, knowledge transfer and subject-specific engagement. However, public recognition of giftedness remains narrowed by examination performance, ranking and visible classroom achievement. Differentiated instruction is valued in principle but is usually enacted informally through questioning, peer sharing, occasional extension, role assignment, after-class suggestions and teacher encouragement rather than through systematic curriculum adaptation. Across the three cases, teachers' professional agency is bound by common pacing, large or mixed-ability classes, workload, safety responsibilities, parental expectations, monitoring, limited training and weak policy guidance. The article argues that the absence of school-based gifted provision is not neutral, because it shifts enrichment opportunities onto families and therefore reproduces uneven access. Building on a comparison with previous case-study work in Madrid, the article recommends a modest, inclusive and sustainable model of gifted support that frames challenge as educational fairness rather than elitism.

Abstract: Artificial intelligence (AI) is transforming educational work, although claims that it will replace teachers often overlook the institutional conditions that make teacher job loss appear more or less likely. This study examined whether current institutional AI risk conditions predict stakeholder expectations of reductions in human teaching positions over the next decade. A purposive sample of 300 education and AI stakeholders rated 11 institutional AI condition indicators and provided a separate 0–10 teacher job-loss expectation score. Overall job-loss expectations were low to moderate, and the institutional AI condition indicators showed good internal consistency. A machine-learning model demonstrated good and stable held-out predictive performance, explaining approximately 60% of the variance in job-loss expectation scores on average. Explainable AI analysis showed that the strongest contribution band consisted of automated feedback, large-scale AI adoption, and standardized content delivery. The findings suggest that stakeholders’ expectations of future teacher job-loss risk are systematically associated with the institutional AI conditions they observe in educational settings. They further indicate that this risk is not an inevitable technological outcome, but is embedded in a broader institutional ecology, with expectations clustering more strongly around direct forms of AI-mediated instructional restructuring than around indirect institutional or governance-related vulnerabilities.

Abstract: This article explores how a 16-year-old Chinese immigrant girl, Mei, narrates science education while studying for her IGCSEs at a British school in Madrid, Spain. Building on Altunbas et al.’s (2024) framework, the study combines self-determination theory with Bourdieu’s concepts of capital, habitus and field. Three semi-structured interviews were analysed narratively, attending to temporal sequencing, turning points, repeated evaluative language, shifts in agency and tensions within Mei’s account. Her story is organised through a marked contrast between her previous school in China and her present British-school field in Madrid. Migration does not erase the earlier achievement habitus, guilt about not studying, fear of losing and an instrumental concern with secure employment continue to shape her relationship with science. At the same time, Mei constructs an emerging autonomous STEM future. A childhood fascination with biology is redirected, for ethical as well as practical reasons, towards mathematics, physics, engineering and renewable energy. Family relatedness strengthens after migration, and her parents support her choices despite limited familiarity with the British system. The analysis shows that high competence, science capital and science aspirations can coexist with insecure relatedness and a legacy of controlled motivation. The combined framework illuminates how migration reconfigures, rather than simply replaces, the dispositions through which an immigrant student understands school science and future possibility.

Abstract: We conceptualized academic success in higher education as a multidimensional outcome shaped by multiple factors. The study examined environmental and sociocultural contexts, institutional support, and relational well-being as predictors of academic success of students in a private higher education institution in South Africa. The study adopted a quantitative cross-sectional survey design. Data from 497 undergraduate students across four campuses in Gauteng Province, were analysed using descriptive statistics, exploratory factor analysis, Pearson correlation analysis, and Partial Least Square Structural Equation Modelling (PLS-SEM). The results showed that relational well-being was the strongest predictor of academic success (β = 0.471, p < .001), demonstrating that relational well-being of students played a significant role in shaping academic success. Similarly, institutional support significantly predicted relational well-being (β = 0.194, p < .001), while sociocultural context also had a significant effect on relational well-being (β = 0.108, p < .05). The study contributes to our understanding of student success by advancing a relationally mediated model of academic performance within the South African higher education context. The findings suggest that institutions of higher learning seeking to improve academic performance must move beyond conventional academic support models with important implications for higher education policy and institutional practices.

Abstract: Grounded in self-determination theory, this study examined the role of psychological distress as a mediator between autonomous learning and academic engagement among university students in China. In this cross-sectional study, survey data from 234 Chinese university students were collected to evaluate autonomous learning, psychological distress including symptoms of depression, anxiety, and stress, and academic engagement. Structural equation modeling indicated that autonomous learning was positively associated with academic engagement and negatively related to psychological distress. Psychological distress, in turn, was negatively associated with academic engagement and mediated the relation between autonomous learning and academic engagement. Additional analyses further revealed significant alternative directionality of effects between psychological distress and autonomous learning, thereby suggesting that the relation may be bidirectional. These findings highlight the importance of fostering autonomous learning and addressing psychological distress to enhance engagement among Chinese students in higher education.

Abstract: Thailand’s Student Loans Fund (SLF) lends on a mortgage-type basis: borrowers repay fixed, time-escalating fractions of principal regardless of realized income. This paper asks how that design distributes financial stress across borrowers and what it implies for fiscal recovery. Using Mincerian age–earnings functions estimated on the 2022 Thai Labor Force Survey and administrative loan parameters, I build a Monte Carlo microsimulation of 400,000 synthetic borrowers and trace each cohort’s repayment burden, scheduled repayment as a share of annual income, over the fifteen-year term. The step-up schedule raises the median bachelor’s-degree borrower’s burden from about 4 percent of income in the first year to 15 percent in the fifteenth, and to roughly 39 percent for the lowest-earning decile; 35 percent of bachelor’s borrowers, and 65 percent of non-completers, breach a 20 percent severe-burden threshold. This affordability-driven stress is concentrated among low earners and non-completers and is steeply regressive. A counterfactual income-contingent loan caps the burden and removes affordability default, and when appropriately parameterized it recovers as much as the mortgage schedule. A utility-based model shows that aggregate collection follows a Laffer curve in repayment stringency, which income contingency removes. The findings reframe SLF default as a product of loan design rather than borrower irresponsibility.