Understanding Emotions in Online Learning: Using Emotional Design and Emotional Measurement to Unpack Complex Emotions During Collaborative Learning

Abstract

Many educational researchers explore the role of emotion in learning and there are many new affordances for emotional measurement. Just as there are many options for emotional measurement there are many theories of emotion. When it comes to the measure of sentiment analysis recent findings suggest it is beneficial to online and blended learning research. The sentiment analysis technologies used for educational research are general purpose technologies suggesting that creating a measure designed for the context of learning would improve the alignment between the measure and context. In addition to aligning measure with the context, there is a need to consider how sentiment analysis relates to emotion theory to determine an appropriate method to evaluate the accuracy of sentiment analysis. In this PhD thesis I adopt the Constructed Theory of Emotion, which considers emotion as a collective intentionality indicating that consensus on emotion is the best approach toward examining accuracy. From this perspective I create a sentiment analysis measure in the context of learning to contribute to emotional learning analytics the emerging sub-field of learning analytics. The field of learning analytics acknowledges that design and measurement are intertwined. I adopt a design-based research approach by designing supports for emotional communication and examining how such a design impacts the accuracy of sentiment analysis. I then examine correlation analysis with other established measures of emotion. The results contribute to the field of emotional learning analytics by: • demonstrating promise for generating a classifier based on student perception • demonstrating benefits of supporting emotion expression in text for students • demonstrating that students’ emotion expression in text does not appear to align with their internal emotional experiences These findings provide opportunities for further research and suggest caution should be used when interpreting sentiment analysis results in the context of learning.