Liviu P. Dinu

University of Bucharest, Romania

Linguistic Markers of Depression: Analyzing Social Media Discourse for Mental Health Insights

---

Biography

Liviu P. Dinu is full professor at University of Bucharest, Computer Science Department, director of Computer Science Doctoral School, director of Human Language Technologies Research Center. His main research is in Computational Linguistics and Natural Language Processing (including themes like language similarity, computational historical linguistics, computational mental health analysis, computational stylometry, application in psychology, etc). Solomon Marcus was his PhD supervisor (obtained in 2003), and in 2014 he defended his habilitation thesis (Similarity and Decision Problems in Computational Linguistics). In 2007 he received Grigore C. Moisil Prize, awarded by the Romanian Academy (for 2005). He has published 2 books, 8 chapters in books, over 180 papers in journals and conferences proceedings, has initiated and managed a number of 16 national and international R&D projects and was involved in other 14 R&D projects. He has also initiated in 2020 a master program in Natural Language Processing at University of Bucharest.

Abstract

In recent years, the prevalence of mental health disorders, particularly depression, has garnered considerable attention from researchers in computation linguistics and psychology. This study aims to investigate the linguistic markers of depression by analyzing social media discourse, with a focus on understanding how individuals express symptoms of depression in their online communications. Using data posted on social media, we conducted a comprehensive part-of-speech analysis to identify specific linguistic features that differentiate individuals with depression from their non-depressed counterparts. Our method involves extracting and analyzing discourse from posts made by users identified as depressed and non-depressed by their mention of diagnosis. We focused on the frequency of nouns, adjectives, adverbs, personal pronouns, and verbs as key indicators of psychological states. Statistical comparisons were conducted to evaluate the significance of the differences observed between the two groups. Our findings revealed significant differences in the language used by depressed individuals, highlighting an increased use of the personal pronoun "I" as well as a greater prevalence of past tense verbs—indicative of rumination. These results align with existing psychological literature, thereby reinforcing the notion that social media language can serve as a vital indicator of mental health status. In conclusion, this study provides valuable insights into the linguistic characteristics of depression on social media platforms, underscoring the potential for developing advanced computational models for mental health monitoring. By enhancing our understanding of how individuals with depression communicate online, we can inform both clinical practices and automated tools aimed at early detection and intervention in mental health issues.