The Tone of Awareness: Topic, Sentiment, and Toxicity Maps During Mental Health Month on TikTok

Problem — This paper addresses the gap in understanding how mental health content is framed by creators and perceived by audiences on TikTok, particularly during Mental Health Awareness Month. Despite the platform’s growing influence on mental health discourse, there is limited empirical analysis of the emotional tone and toxicity of such content. The work is a preprint and has not undergone peer review.

Method — The authors collected a dataset comprising 28,341 TikTok videos and 80,130 comments from May 2023 and 2024 using the TikTok Research API. They operationalized “tone” through sentiment and toxicity measures, employing BERTopic for topic extraction and XLM-T for sentiment analysis, alongside Detoxify for toxicity assessment. The analysis was conducted separately for video transcriptions and comments, allowing for a nuanced understanding of content production versus audience reception. The study identifies recurring themes across years, including clinical conditions, emotional disclosure, self-care, and campaign-oriented content.

Results — The findings reveal a stable set of themes with significant engagement skewed towards a small subset of topics. Sentiment analysis indicated that videos on emotionally charged topics often exhibited negative sentiment, while comments showed a shift towards mixed or positive sentiment, particularly in discussions around suicide prevention. The median toxicity level was low across the dataset, but there were notable outliers in comments, especially for topics like “Duet,” “Suicide Prevention,” and “Psychisch.” The authors provide a detailed topic-level decomposition of sentiment and toxicity, highlighting the differences in emotional framing between video creators and audience responses.

Limitations — The authors acknowledge that their analysis is limited to content from a specific time frame (Mental Health Awareness Month) and may not generalize to other periods or contexts. Additionally, the reliance on automated sentiment and toxicity measures may overlook nuanced human interpretations of language. The study does not explore the potential impact of creator demographics or the influence of TikTok’s algorithm on content visibility and engagement.

Why it matters — This research contributes to the understanding of mental health discourse on social media platforms, particularly in how creators frame sensitive topics and how audiences engage with this content. The insights gained can inform future studies on the impact of social media on mental health, as well as guide content creators and mental health advocates in crafting messages that resonate positively with audiences. The findings are particularly relevant for ongoing discussions about the role of social media in mental health awareness, as published in arXiv cs.CL.