Using the Structural Topic Model to Explore Learner Satisfaction with LMOOCs

The structural topic model enables the extraction of insights from course reviews to offer actionable feedback for both course instructors and designers, thereby enhancing course quality. The current research collected 70,008 course reviews from 1374 language courses in Udemy and used the collocational network analysis and Structural Topic Model to explore learners’ opinions and satisfaction. Collocational networks of probe words of “teacher,” “course,” and “satisfied” show that learners hold a strong positive sentiment toward teachers and the diverse spectrum of language courses. The topic modeling results show that four negative topics appear more commonly in negative reviews as compared to positive ones. Furthermore, exploration of variations in learner dissatisfaction across different course types and levels showed that expert learners tend to express complaints primarily related to the balance in grammar teaching while beginners show higher sensitivity toward course assignments and assessments. In basic language skills courses, learner dissatisfaction is commonly associated with issues related to practice techniques, grammar teaching, and skill improvement, while in advanced courses there is a tendency for writing skills and professional abilities. Our study contributes to the LMOOC literature by facilitating a better understanding of LMOOC learner satisfaction using rigorous statistical techniques. The analytical framework is adaptable to MOOC course reviews across diverse academic disciplines.