New Study Reveals Online Language Indicators of Self-Harm Risk

A recent study conducted by researchers at The University of Texas at Dallas and the University of Liverpool has unveiled significant insights into how online language choices can serve as early indicators of self-harm risk. Published in the journal *npj Mental Health Research* on July 25, 2025, the study analyzed over 66,000 posts from nearly 1,000 Reddit users diagnosed with borderline personality disorder (BPD), focusing on how specific language patterns correlate with self-harming behaviors.

The research, led by Dr. Ryan L. Boyd, Assistant Professor of Psychology in the School of Behavioral and Brain Sciences at The University of Texas at Dallas, and Dr. Charlotte Entwistle from the University of Liverpool, aimed to explore the dynamics surrounding self-harm in natural online settings. The findings revealed that posts indicating decreased social connectedness and heightened negative emotions could predict self-harm behaviors, including suicidal thoughts, as much as several weeks before they occurred.

Dr. Boyd emphasized that the study highlights the potential dangers associated with online support communities. “Our research underscores not only early linguistic predictors of self-harm but also how online interactions may unintentionally reinforce harmful thoughts and behaviors,” said Dr. Boyd, who serves as the senior author of the paper.

Dr. Entwistle further elaborated on the study's distinctive focus, noting that it addresses both nonsuicidal self-injury and suicidality simultaneously, specifically in individuals with BPD. “This population is particularly vulnerable, with alarmingly high suicide rates,” she stated.

The researchers utilized natural language processing, a form of artificial intelligence, to dissect the language used in Reddit posts. They found that negative and hostile language was associated with a higher risk of self-harm, indicating a possible social contagion effect within these online communities. “Posts that contained more negative emotions—like anger, sadness, and anxiety—received higher engagement, suggesting a troubling reinforcement of harmful behaviors,” added Dr. Boyd.

Contrastingly, Dr. Entwistle referenced previous studies that found general online communities to be less supportive of negative posts, indicating that the dynamics observed in the BPD subreddit may be unique to this specific group. This raises questions about the effectiveness of informal online support in preventing self-harm and suicidal ideation.

“The engagement metrics we observed suggest that negative posts attract more attention, which can lead users to focus on self-harm to receive the same level of care from the community,” Dr. Boyd explained. “While online communities can provide essential support, they can also contribute to a downward spiral of harmful thoughts.”

The implications of this research extend beyond social media platforms, influencing clinical practices as well. The findings suggest that emotional and social factors are critical triggers for suicidal thoughts and self-harm behaviors, presenting opportunities for targeted interventions. Dr. Entwistle remarked, “Our study lays the groundwork for more advanced predictive models that could aid in early intervention for high-risk individuals.”

Supported by grants from the National Institute on Alcohol Abuse and Alcoholism and the National Institute of Mental Health, this research underscores the necessity for thoughtful moderation of online support communities. As Dr. Boyd concluded, “We may need to rethink how social media frames engagement, particularly around distressing content, to mitigate unintentional harm.”

This study not only sheds light on the complexities of mental health discourse within online communities but also calls for a re-evaluation of the role these platforms play in the lives of individuals struggling with self-harm and mental health issues.