Inferring social identities related to mental health and addiction from linguistic style

Abstract: Recent research on the “social cure” and “social prescribing” has shown that membership in social groups may be beneficial for mental health and recovery from addiction. However, important concepts like social identity salience are extremely difficult to assess through self-report. In this talk, I will provide an overview of our recent work on inferring social identity salience from linguistic style using machine learning and natural language processing techniques. I will then provide preliminary evidence that suggests that social identities play an interesting role in perinatal mental health and addiction relapse, respectively.