This mixed-methods study investigated Omani EFL university students' engagement, motivation, and perceptions of the Spark digital platform in a blended learning context. Using Self-Determination Theory (SDT), the Technology Acceptance Model (TAM), and the Motivated Strategies for Learning Questionnaire (MSLQ) frameworks, the study analyzed survey data, platform analytics, and focus group discussions from 100 students. The results showed that extrinsic, grade-oriented motivation mainly led to moderate engagement. Although analytics suggested high activity, qualitative findings revealed frustration, redundancy, and time pressure. Students reported moderate levels of satisfaction. The study identifies an engagement-compliance paradox, where observable use masks low intrinsic motivation, and recommends a shift to structured in-class use, targeted curriculum alignment, and AI-powered feedback to achieve meaningful and sustained learner engagement.
This mixed-methods study examined the effectiveness of AI-generated instructional videos (AIVs) on content comprehension, engagement, and reflective practice. Conducted with 117 undergraduate students in an Ecuadorian English-as-a-foreign-language teacher education program, the study compared AIV-supported instruction with traditional materials in two courses: Contrastive Grammar and Research Methods. Quantitative data from tests, surveys, and reflection scores were complemented by qualitative analysis of student reflections and classroom artifacts. Results showed that students using AIVs outperformed the control group in Contrastive Grammar, while gains in Research Methods were positive but not statistically significant. Students reported high perceived usefulness and moderate to high engagement, despite some challenges in video creation. Moreover, watching AIVs helped students clarify and reinforce content, while creating videos fostered deeper understanding, creativity, and digital skills. Reflection quality strongly correlated with academic achievement in Research Methods but not in Contrastive Grammar.
This study examines the relationship between formative assessment (FA) and L2 writing anxiety (WA) by investigating the mediating roles of writing self-concept (WSC) and learning attitudes in Automated Writing Evaluation (AWE)-supported English as Foreign Language (EFL) writing (LAiW). A mixed-methods approach combined covariance-based structural equation modeling (CB-SEM) on survey data from 366 Chinese university students and semi-structured interviews with eight participants. The quantitative findings showed that FA did not exert a direct effect on WA, instead demonstrating a fully mediated relationship. FA significantly predicted improvements in both WSC and LAiW, which subsequently exerted significant negative effects on WA. The qualitative findings triangulated these results, revealing that although AWE enhances metacognitive engagement and immediate linguistic scaffolding, teacher mediation remains essential for higher-order writing concerns and emotional support. The study proposes an affective-cognitive framework for AWE-integrated EFL FA and highlights the need for coordinated AI-teacher collaboration.
This study explores the effectiveness of video captions versus generative artificial intelligence (GAI) mind mapping for enhancing English as a second language (L2) listening skills among 79 EFL undergraduates in Macau. Using TED-Ed videos, the study employed pre-/post-tests and thematic analysis of reflective journals. Results indicated that the GAI mind mapping group significantly outperformed the captions group, demonstrating greater improvements in listening skills, efficacy, and strategy development. In contrast, the captions group experienced cognitive overload and limited engagement. This research offers valuable insights for educators seeking to optimise L2 listening instruction in EMI settings.
This mixed-methods, quasi-experimental pilot study examined free nonpremium AI-conversational apps' impact on Omani English-as-a-foreign-language learners' speaking proficiency and confidence. Over 16 weeks, 140 learners completed speaking assessments and interviews, with experimental groups using AI apps. Results showed significant gains in fluency and pronunciation among advanced learners improved in pronunciation (F = 5.169, p = .026), while lexical and grammatical skills had minimal change. Anxiety decreased from M = 3.70 to M = 3.09, with analysis showing increased confidence and engagement. Intermediate learners reported reduced fear and improved speech, while advanced learners noted enhanced self-regulated learning aligned with the AI-self-regulated speaking development model. Despite limitations in accent recognition and feedback, learners valued AI tools as supplements to instruction. The study concluded that AI apps could support oral proficiency when integrated with instruction and practice.
