This study developed and evaluated an online English speaking training approach that integrates corpora and artificial intelligence (AI) tools. The training integrated a self-developed spoken corpus, generative AI tools, and text-to-speech AI tools. Pre- and post-test results identified improvements in participants’ speaking performances. Participants attempted to use more positive linguistic features (e.g. producing complex sentences more frequently) and avoid using negative linguistic features (e.g. reducing the number of vowel errors) after receiving the training. Participants showed positive attitudes towards this corpus-based and AI-integrated English oral ability learning approach and affirmed the importance of integrating both tools. The corpus helped raise participants’ awareness of features that influence speaking performance and offered prompt engineering and feedback-checking functions, while the generative AI tools provided useful feedback and tailor-made sample responses. Additionally, text-to-speech AI tools offered learners with tailor-made native speaker samples for imitation and helped learners learn pausing. Results also revealed that this approach helped create an interactive oral ability learning environment, and the combination of corpora and AI tools provided more accurate feedback for each subskill of speaking.

This study examines the status of mixed-methods research (MMR) in computer-assisted language learning (CALL). A total of 204 studies employing MMR were analyzed. Manual coding was carried out to reveal MMR purposes, designs, features, and rhetorical justifications. Findings indicate CALL authors mostly adopt MMR for triangulation and complementarity purposes. Core designs are more favored in CALL MMR research articles, compared to complex designs. Moderate size random sampling prevails in the data, where data sources are sequentially collected and analyzed using parametric tests. Symptomatic argumentative schemes are found to be the most common justification of MMR. Based on the findings, it is evident that most CALL researchers employ conventional MMR designs. The study concludes with implications for CALL stakeholders and authors.

Limited research has examined the quality of language MOOCs and no existing instrument has been developed to gauge learners’ evaluation of LMOOC quality. This study develops an LMOOC Quality Evaluation Scale (LQES) and validates it in the Chinese context, which has the largest number of LMOOC learners in the world. The data were collected from 2,315 LMOOC learners in China using a mixed-method approach. Development and validation of the scale involved (1) generation of an initial item pool based on a semi-structured interview and literature review, (2) refinement of scale items through consultation of LMOOC experts and a focus group interview, (3) exploration of the factor structure of the scale using exploratory factor analysis, and (4) validation and confirmation of the final scale using confirmatory factor analysis. A four-factor model, comprising Instructional Design, L2 Teachers’ Competence, Teaching Implementation, and Technical Support, emerged and was validated. The 26-item LQES provides an original and comprehensive framework for understanding the complexities of LMOOC quality. This study highlights the critical factors underpinning the evaluation of LMOOC quality and paves the way for further refining of the instrument in future research.

This study investigated an 18‑week teacher education model grounded in technological pedagogical content knowledge (TPACK). Known as CATERR (comprehending, analyzing, teaching, evaluating, reflecting, and refining), this teacher education model cultivated the computer-assisted language learning (CALL) competencies of 43 content and language integrated learning (CLIL) preservice teachers (PSTs) from Taiwan. The model promotes peer coaching, where participants collaborate, reflect, and refine their teaching over three rounds. The study utilized a multi-method case study and triangulated the quantitative and qualitative data. Quantitative data refers to the TPACK-CLIL questionnaire administered before and after the teacher education model. Qualitative data included lesson plans, self-analysis, teaching demonstration videos, revised lesson plans, classroom discussion records, peer evaluations, and reflection notes. Data analysis involved paired-samples t-tests and descriptive statistics for the coding framework, thematic analysis for qualitative data, and a repeated measures ANOVA to compare three total scores across three rounds using scoring rubrics. Results showed that the CATERR teacher education model enhanced CLIL PSTs’ self-perceived and observed CALL competencies. Specifically, as “digital native” PSTs with high levels of technological knowledge (TK), they successfully transferred their TK into TPACK by adding pedagogical values and contextualizing the ICT tools in their CLIL lessons. Meanwhile, their ability to use ICT tools to facilitate interaction and students’ autonomous learning substantially improved. The theoretical and pedagogical implications for CALL teacher education research and practice are discussed.

This study investigated the impact of familiar versus unfamiliar environments on mobile-assisted language learning (MALL) task writing performance, English as a foreign language (EFL) writing proficiency, and learner perceptions. Fifty undergraduate students were divided into an experimental group and a control group. Both groups engaged in EFL learning in the classroom and later completed writing tasks in different learning environments outside the classroom: the experimental group in familiar environments and the control group in unfamiliar ones. Using a mobile learning system on tablet PCs, students completed five writing tasks describing resources in their environments, such as objects, people, situations, and scenarios. We assessed MALL task writing performance based on factors including the amount of writing, content quality, organization, creativity, grammar, and vocabulary, and compared results between the two groups. EFL writing proficiency was evaluated through a post-test directly related to the MALL tasks, and student perceptions of the MALL experience were measured through a survey. The results indicated that the experimental group outperformed the control group in both writing tasks and the post-test. Furthermore, the experimental group reported more positive perceptions of their MALL experience, reflected in higher emotional engagement and cognitive involvement. Based on these findings, we offer both theoretical insights into the role of familiar environments in facilitating language learning and practical suggestions for EFL teachers and researchers to incorporate real-world, contextually rich environments in MALL activities.

Speaking is often challenging for language learners to develop due to factors such as anxiety and limited practice opportunities. Dialogue-based computer-assisted language learning (CALL) systems have the potential to address these challenges. While there is evidence of their usefulness in second language (L2) learning, the effectiveness of these systems on speaking development remains unclear. The present meta-analysis attempts to provide a comprehensive overview of the effect of dialogue-based CALL in facilitating L2 speaking development. After an extensive literature search, we identified 16 studies encompassing 89 effect sizes. Through a three-level meta-analysis, we calculated the overall effect size and investigated the potential moderating effect of 13 variables spanning study context, study design and treatment, and measures. Results indicated a moderate overall effect size (g = .61) of dialogue systems on L2 learners’ speaking development. Notably, three moderators were found to have significant effects: type of system, system meaning constraint, and system modality. No significant moderating effect was identified for education stage, L2 proficiency, learning location, corrective feedback, length of intervention, type of interaction, measure, and key assessment component. These findings suggest directions for future research, including the role of corrective feedback in dialogue-based CALL, the effectiveness of such systems across proficiency levels, and their potential in diverse learning contexts with the integration of generative artificial intelligence.

Digital oratory skills have become essential for academic and professional success in today’s digital world, making it imperative to integrate digital oratory training into public speaking pedagogy. This study examined second language (L2) speakers’ public speaking anxiety and nonverbal speech performance in the context of digital oratory. Participants were 40 English as a second language students enrolled in a public speaking course at a Hong Kong university. Each student recorded and uploaded an 8-minute speech to a digital learning platform. They also completed a questionnaire measuring digital oratory anxiety and participated in semi-structured interviews sharing their perceptions of digital presentation. Nonverbal speech performance was assessed, and correlations with digital oratory anxiety were analyzed. The results showed that cognitive and physiological factors had a greater influence on digital oratory anxiety than behavioral and technical factors. Although no significant correlations were found between digital oratory anxiety and nonverbal speech performances, the technical factor had the least impact on L2 students’ anxiety, leading to positive outcomes regarding the technical quality of the speech videos. Comparatively, eye contact and gestures attained much lower mean scores than voice control and facial expressions. Interview results further elucidated the benefits and challenges students experienced during digital presentations. Pedagogically, the findings highlight the need of a holistic approach considering cognitive, physiological, behavioral, and technical factors to address L2 learners’ digital oratory anxiety. Given its affordability and accessibility, digital oratory can be effectively integrated into instruction to develop L2 students’ multimodal communication and nonverbal delivery skills.

Consulting dictionaries during writing requires time and cognitive resources. ColloCaid, a writing assistance prototype freely available online, was designed to minimize the cognitive strain on writers by embedding a collocation database within the writing environment. Usability surveys have shown ColloCaid can indeed help. In this study, we go beyond user perceptions. Using authentic excerpts of student academic writing by 27 advanced L2 English speakers, we analysed (1) the lexical coverage of the tool, (2) the collocation changes prompted by the tool, (3) the reasons behind decisions to revise collocations, (4) the effect of revisions prompted by ColloCaid, and (5) the participants’ perceptions of using the tool to revise authentic writing assignments. Our findings indicate that ColloCaid offered good academic collocation coverage, that the participants tended to accept its collocation prompts with discernment, and that the revisions made resulted in more fluent texts overall.

Learners’ confidence in using a second language (L2 self-efficacy) and their L2 grit are key psychological factors in developing intercultural competence (ICC). As English as a foreign language (EFL) students increasingly encounter diverse cultures though informal digital learning of English (IDLE), this study examines whether IDLE serves as a pathway connecting these psychological traits to ICC. Grounded in the broaden-and-build theory, this explanatory mixed-methods research investigates how L2 self-efficacy and grit contribute to ICC through IDLE among 416 Chinese EFL students. Structural equation modeling revealed that higher L2 self-efficacy fosters greater L2 grit, which in turn promotes more frequent engagement in both receptive IDLE activities (e.g. watching English media) and productive ones (e.g. participating in online conversations). This increased engagement was positively linked to higher levels of ICC. Qualitative findings further illuminated the mechanisms behind this process, illustrating how psychological strengths support meaningful digital encounters across cultures. The findings offer pedagogical insights: by cultivating students’ self-efficacy and grit, educators can encourage deeper engagement in IDLE, thereby equipping learners with effective and culturally sensitive communication in an increasingly interconnected digital world.

This study synthesized 65 (quasi-)experimental studies published between 2010 and 2024 that examined the use of mobile applications to develop language learners’ vocabulary learning. Bayesian meta-analysis was adopted to assess (1) overall effect size; (2) subgroup analyses (i.e. education level, vocabulary knowledge, aspects of vocabulary learning, learning environment, sample size, mobile application type, gender, and cultural background); and (3) publication bias. A large effect size of 1.28 was found for the overall effectiveness of using mobile applications for vocabulary learning when we restricted the studies to long-term treatment duration of 10 weeks or above. Each moderator was analyzed and discussed, and implications for language teaching and research were provided.

The emergence of large language models, exemplified by ChatGPT, has garnered growing attention for their potential to generate feedback in second language writing, particularly automated written corrective feedback (AWCF). In this study, we examined how prompt design – a generic prompt and two domain-specific prompts (zero-shot and one-shot) enriched with comprehensive domain knowledge about written corrective feedback (WCF) – influences ChatGPT’s ability to provide AWCF. The accuracy and coverage of ChatGPT’s feedback across these three prompts were benchmarked against Grammarly, a widely used traditional automated writing evaluation (AWE) tool. We find that ChatGPT’s ability in flagging language errors grew considerably with prompt sophistication driven by the integration of domain-specific knowledge and examples. While the generic prompt resulted in substantially lower performance than Grammarly, the zero-shot prompt achieved comparable results to it and the one-shot prompt surpassed it considerably in error detection. Notably, the most pronounced improvement in ChatGPT’s performance was observed in its detection of frequent error categories, including those of word choice or expression, direct translation, sentence structure and pronoun. Nonetheless, even with the most sophisticated prompt, ChatGPT still displayed certain limitations when compared to Grammarly. Our study has both theoretical and practical implications. Theoretically, it lends empirical evidence to Knoth et al.’s (2024) proposition to separate domain-specific AI literacy from generic AI literacy. Practically, it sheds light on the pedagogical application and technical development of AWE systems.

The adoption of corpus technology in school classroom settings remains limited, largely due to insufficient technological pedagogical content knowledge (TPACK) training for pedagogical corpus use. To address this gap, we investigated how teacher education in corpus-based language pedagogy (CBLP), a subdomain of TPACK for corpus technology tailored to language teachers, influenced student TESOL teachers’ self-efficacy for independent language learning and teaching. Employing a mixed-methods approach, including a CBLP training intervention (n = 120), survey data (n = 96), and interviews (n = 8) with student teachers at a university in Hong Kong SAR, China, the research validates a theoretical model through confirmatory factor analysis and structural equation modelling. Results demonstrate that corpus literacy (CL) is foundational for effective CBLP implementation and development of independent learning self-efficacy, which in turn fosters innovative, resource-rich instructional strategies. CBLP also enhances teachers’ self-efficacy for student engagement, fostering more interactive and motivating classrooms. These findings emphasise the value of embedding CL and CBLP within TESOL teacher-education programmes to prepare future language teachers for self-efficacy within dynamic, technology-enhanced classrooms.

Efforts to integrate intelligent chatbots into academic courses, particularly for language learning, have been gaining popularity. However, the impact of chatbot-supported collaborative learning (CL) on student engagement and English speaking skills is under-researched. This study explored the impact of utilizing intelligent chatbot–supported CL on student engagement and speaking skills of English as a foreign language (EFL) learners. It investigated how chatbot-supported CL influences student engagement and speaking skills. The experimental group was taught using chatbot-supported CL, while the control group followed conventional CL. A total of 75 first-year undergraduate students participated, with 39 students in the experimental group and 36 in the control group. Data were collected through a 14-item engagement questionnaire, a speaking test based on the IELTS speaking evaluation rubric for both groups, and a 5-item CL questionnaire administered solely to the experimental group. The data were analyzed using repeated measures analysis of variance (RM-ANOVA) and linear regression analysis. The RM-ANOVA results showed that chatbot-supported CL positively affected student engagement and speaking skills. The linear regression analysis further indicated that CL supported by intelligent chatbots influenced student engagement, which in turn significantly impacted speaking skills. The findings suggested that active engagement in CL speaking classes is crucial for improving EFL speaking skills and that intelligent chatbots can be valuable and effective tools for promoting such engagement.

Feedback is integral to second language (L2) writing instruction. However, large class sizes and limited teacher time often challenge the delivery of personalized feedback, prompting interest in AI-powered solutions such as ChatGPT (Escalante et al., 2023; Huete-García & Tarp, 2024; Steiss et al., 2024; Yoon et al., 2023; Zhang, 2024). This study evaluates a task-customized GPT model, “Belinda,” trained to assess A1-level Spanish learners’ writing and provide feedback. Two research questions guided the investigation: (1) Can Belinda accurately score beginner Spanish writing using a provided rubric? (2) Can Belinda deliver constructive qualitative feedback? Human and GPT-generated scores were compared for inter- and intrarater reliability, and qualitative analyses categorized the feedback for usability in the classroom. Results revealed moderate alignment between Belinda’s scores and human raters, though reliability of the GPT fell short of calibration benchmarks. Feedback quality varied, with Belinda often providing vague, incomplete, or inaccurate suggestions. Despite iterative training, the GPT struggled to balance error correction with encouragement, a critical need for novice learners. Additionally, inconsistencies in identical GPT versions raised concerns about reliability. While Belinda showed potential in automating feedback, its limitations in accuracy, contextual understanding, and positivity suggest it is not yet a viable substitute for human evaluation by itself. These findings emphasize the challenges of integrating AI into L2 instruction and call for the need for extensive datasets, robust training, and human–AI collaboration to achieve pedagogically sound outcomes. Future research should explore hybrid feedback models and scalable solutions to enhance AI’s role in language education without compromising learner progress or confidence.

This scoping review directs attention to artificial intelligence–mediated informal language learning (AI-ILL), defined as autonomous, self-directed, out-of-class second and foreign language (L2) learning practices involving AI tools. Through analysis of 65 empirical studies published up to mid-April 2025, it maps the landscape of this emerging field and identifies the key antecedents and outcomes. Findings revealed a nascent field characterized by exponential growth following ChatGPT’s release, geographical concentration in East Asia, methodological dominance of cross-sectional designs, and limited theoretical foundations. Analysis also demonstrated that learners’ AI-mediated informal learning practices are influenced by cognitive, affective, and sociocontextual factors, while producing significant benefits across linguistic, affective, and cognitive dimensions, particularly enhanced speaking proficiency and reduced communication anxiety. This review situates AI-ILL as an evolving subfield within intelligent CALL and suggests important directions for future research to understand the potential of constantly emerging AI technologies in supporting autonomous L2 development beyond the classroom.

Generative artificial intelligence (GenAI) has been heralded by some as a transformational force in education. It is argued to have the potential to reduce inequality and democratize the learning experience, particularly in the Global South. Others warn of the dangers of techno-solutionism, dehumanization of learners, and a widening digital divide. The reality, as so often, may be more complicated than this juxtaposition suggests. In our study, we investigated the ways in which GenAI can contribute to independent language learning in the context of Pakistan. We were particularly interested in the roles of five variables that have been shown to be particularly salient in this and similar contexts: learners’ Generative Artificial Intelligence-mediated Informal Digital Learning of English (GenAI-IDLE) participation, AI Literacy, Foreign Language Enjoyment (FLE) and Foreign Language Boredom (FLB), and their second language Willingness to Communicate (L2 WTC). Employing a structural equation modelling approach, we surveyed 359 Pakistani English as a foreign language (EFL) learners to investigate their interrelationships between variables. The results demonstrate that EFL learners’ GenAI-IDLE activity directly and positively influences AI literacy and FLE. Students’ AI literacy and FLE play a chain-mediating role in the relationship between GenAI-IDLE participation and L2 WTC. However, FLB lacks predictive power over L2 WTC. We discuss the implications of these results for language learning, in particular in low-resource contexts.