computer-assisted-language-learning
Filtering by topic computer-assisted-language-learning(3)Clear all filters
- PaperarXiv — Language & NLP (cs.CL)4 Jun 2026
Domain-Aware Mispronunciation Detection and Diagnosis Using Language-Specific Statistical Graphs
Huu Tuong Tu, Hanh Nguyen, Thien Van Luong, Nguyen Tien Cuong et al.
Proposes a method for constructing statistical graphs to learn phoneme confusion patterns, and introduces a language-specific strategy to capture pronunciation differences across native language backgrounds. Achieves an F1-score of 59.52% on the L2-ARCTIC benchmark, outperforming several competitive baselines.
Original abstract
Mispronunciation Detection and Diagnosis (MDD) has gained increasing importance in computer-assisted language learning and speech technology in recent years. In this paper, we propose a method for constructing statistical graphs that enable models to learn phoneme confusion patterns represented as directed graphs. Furthermore, we introduce a language-specific strategy to capture systematic pronunciation differences across various native language (L1) backgrounds. The effectiveness of our approach is demonstrated through extensive experiments on the L2-ARCTIC benchmark, where it achieves an F1-score of 59.52%, outperforming several competitive baselines.
- PaperLanguage Learning & Technology1 Jan 2026
Pride and shame in CALL: Links to appraisals, engagement, and performance
Kaiqi Shao, Elmakki Amiri, Gulsah Kutuk
This study examines pride and shame as achievement emotions in a Computer-Assisted Language Learning (CALL) setting, guided by control-value theory. Survey data from 652 Chinese university students in a MOOC showed that control and value appraisals positively predicted pride but negatively predicted shame, and that pride and shame differentially predicted cognitive, emotional, and behavioral engagement. Emotional and behavioral engagement, but not cognitive engagement, predicted performance, with pride and shame acting as mediators between appraisals and engagement outcomes.
Original abstract
Guided by the control-value theory of achievement emotions, this study examines the relationships among two understudied foreign language emotions, namely pride and shame, control-value appraisals, engagement, and performance in a Computer-Assisted Language Learning (CALL) setting. A total of 652 Chinese university students from a massive open online course (MOOC) participated in the study. Structural equation modeling (SEM) results showed that control and value appraisals positively predicted pride but negatively predicted shame. Pride positively predicted each of the three dimensions of engagement (i.e., cognitive, emotional, and behavioral) while shame negatively predicted these dimensions, except for cognitive engagement. Emotional and behavioral engagement, but not cognitive engagement, positively predicted performance. Pride and shame mediated the relationship between control-value appraisals and emotional and behavioral engagement, which, in turn, mediated the relationship between pride or shame and performance. Cognitive engagement consistently showed no significant pathways. Overall, pride and shame, along with emotional and behavioral engagement rather than cognitive engagement serially mediated the relationship between control-value appraisals and performance. We discuss the implications for language teachers and highlight the importance of addressing pride and shame, alongside their appraisal antecedents and learning outcomes in CALL.
- PaperReCALL11 Nov 2025
Investigating the status of mixed-methods research in CALL published research articles
Saleh Arizavi, Yazdan Choubsaz
Analyzed 204 mixed-methods research articles in CALL, revealing that triangulation and complementarity are the primary purposes. Core designs are more common than complex ones, with moderate random sampling and parametric tests frequently used. The study offers implications for CALL stakeholders and authors.
Original abstract
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.