Rethinking Self-Regulated Learning in Online Education: Typologies, Affective Dynamics, and Design Implications from Learner Feedback

Abstract

As self-directed online learning becomes more prevalent, understanding how learners self-regulate in these environments is critical for ensuring engagement, persistence, and success. This study examines how self-regulated learning (SRL) behaviours are expressed in authentic learner feedback and explores their associations with emotional tone and satisfaction. A total of 826 anonymised comments were analysed from participants in asynchronous professional trade finance courses. Drawing on established SRL models, the study combined thematic analysis with natural language processing (TF-IDF vectorisation, KMeans clustering), sentiment analysis (TextBlob), and statistical correlations to identify behavioural patterns. Results showed that SRL behaviours varied widely and often clustered into compound strategies. Help-Seeking and Motivational Beliefs were positively correlated with satisfaction, while Planning, Time Management, and Environmental Structuring were linked to frustration, particularly where platform design failed to support learners’ regulatory intentions. Sentiment analysis confirmed that emotional tone aligned with regulatory challenges or supports. Six learner typologies and five cluster-derived themes were identified, reflecting distinct SRL profiles and systemic pain points such as unclear structure, feedback gaps, and usability issues. These findings affirm that SRL is not merely an internal trait but a contextually shaped, co-regulated process. Implications include the need for emotionally intelligent, scaffolded platforms tailored to diverse learner profiles, and policy shifts toward feedback-rich, adaptive learning environments. While the study is situated in a specific professional learning context, its integrated methods offer a scalable model for analysing authentic SRL behaviours in digital education. This approach demonstrates how learner feedback can serve as a scalable diagnostic lens for SRL-informed design.