From Complex Emotion Words to Insomnia and Mental Health: A Corpus-based Analysis of the Online Psychological Consultation Discourse about Insomnia Problems in Chinese

This paper reports the preliminary findings of a cross-disciplinary research on emotions, insomnia and mental health with joint efforts from linguistic, computer science, and medical researchers. We take a computational linguistic approach to analyze a corpus of over 400 posts crawled from online psychological consultation platforms in China that complaint about insomnia problems, with annotations for the psychiatric conditions of post contributors made by professional psychiatrists. Based on results of our automatic analysis of six basic emotions in the posts and their intensity levels, logistic regression and artificial neural network models were successfully performed to predict the mental health conditions of post contributors. Relevance analysis were further run to explore the associations among emotion types and between emotions and the psychiatric disorders of depression and anxiety. Depression and anxiety were found to have divergent association patterns with the basic emotion types. Overall, the findings support the previous hypothesis that emotions modulate insomnia and psychiatric disorders, and have important linguistic and clinical implications.