Automatic detection of subtle mental health stigma in online text
Automatic Evaluation of Mental Health Stigma in Online Communication
Computation and Language
Summary
Mental health stigma can be obvious or very subtle, making it hard to spot in online conversations. The authors created a detailed way to label different kinds of stigma in social media and news texts mentioning six mental health conditions. They tested large language models and other tools, finding these models often mistake non-stigmatizing comments for stigma unless given clear instructions. This work helps better identify harmful attitudes toward mental health online.
What this means in practice
- •For social media moderation teams: Identify various forms of mental health stigma in user posts beyond basic hate speech to improve content review policies.
- •For online news editors: Automatically flag subtle mental health stigma in articles and comments to promote more responsible reporting and discussion.
Authors
Naomi Baes, Jemima Kang, Nick Haslam, Chris Groot, Alsa Wu, Luc Raszewski, Yulia Otmakhova
Abstract
Mental health stigma has profoundly harmful impacts but its complexity makes it difficult to evaluate. Stigma may involve explicit derogation, but also subtler forms of blame, fear, paternalistic pity, social distancing, structural exclusion, and discrimination. We introduce a theory-grounded benchmark for automatic evaluation of mental health stigma in online communication, consisting of naturally occurring online news and social media text annotated with a fine-grained taxonomy of stigma across multiple mental health conditions. Our annotation framework comprises a binary stigma-detection task and a multi-level taxonomy covering (i) stigma mode, (ii) domain, and (iii) specific components of certain forms of stigma. We apply this framework to texts mentioning six mental health conditions and evaluate large language models alongside stigma-related classifiers for detecting sentiment, toxicity, and hate speech. Results show that mental health stigma is not well captured by models trained to detect these neighboring constructs, and that LLMs often overpredict stigma unless given explicit operational rules - mirroring the importance of decision rules in human annotation. We release the publicly available part of benchmark, annotations, prototypical exemplars of stigma and code at: https://github.com/jemimakang/mh_stigma.