Analysis of Students’ Emotion from a Text Corpus

Abstract

Emotions play an important role in e-learning environments. Previous studies have investigated the prediction of learners’ emotions using various features such as acoustic-prosodic features, mouse movements, facial features, and body postures. In addition to these features, linguistic features are also useful for identifying learners’ emotions especially in text-based e-learning systems. Therefore, this study attempts to analyze the linguistic features for different types of emotions. To accomplish this goal, we first collect a text corpus of emotion sentences from student-teacher dialogs in mathematics learning. Each sentence is then annotated to provide analysis results such as the linguistic features, proportions, annotator agreements, and annotation accuracy for different emotions.