Bloomington Leader

Two researchers from Indiana University Bloomington have received a $2.8 million grant from the National Institute of Mental Health to explore the connection between social media use and youth mental health. The five-year study aims to understand how language used on social media can impact adolescents' mental well-being.

Lorenzo Lorenzo-Luaces, an associate professor in the Department of Psychological and Brain Sciences, and Johan Bollen, a professor in the Luddy School of Informatics, Computing and Engineering, will lead this research. They plan to investigate whether the language young people use online can predict internalized disorders such as depression and anxiety.

“Over the past couple of years, there has been focus on social media impacting the mental health of our adolescents and young adults, but if social media does impact mental health, it is unclear why,” Lorenzo-Luaces said. “Since so much communication on social media relies on language, we wanted to put a spotlight on language use to see if it can predict how adolescents feel when using social media.”

The U.S. surgeon general's advisory report last year highlighted potential risks associated with social media for young people. With nearly 95% of teens using these platforms, understanding its effects is crucial.

The Study of Online Cohorts for Internalizing Symptoms and Language (SOCIAL) will track 1,000 individuals aged 13-20 through survey diaries and machine learning techniques. The aim is to identify harmful usage patterns and those most affected by them.

This initiative builds upon previous work by Lorenzo-Luaces and Bollen that linked depression with negative language on social media in 2021. Their collaborative efforts blend psychology with artificial intelligence to model online interactions.

“IU has always been a very interdisciplinary place; particularly, Luddy and the College have a strong history of collaborating,” Lorenzo-Luaces stated. “These kinds of projects we’ve done stretch both of us outside of our comfort zone, and the knowledge that gets produced is more than either one of us could have done alone.”

Participants will be recruited via social media platforms where they will complete daily questionnaires about their mental health experiences. The collected data will help develop models using natural language processing to determine if negative language predicts mental health issues.

“With artificial intelligence, you can develop natural language processing algorithms to examine the language of social media posts and look for markers that are indicative of mental health disorders,” Bollen explained. “That’s technically possible, but here we take a theory-driven perspective. We want to understand the very cognitive processes and their dynamics that lead to internalizing disorders.”

Bollen emphasized that combining insights from social media with natural language processing offers comprehensive insights into how depression develops and potential intervention strategies.

“Since we can see who’s connected to whom and what they’re saying and responding to, we can see if this kind of language propagates a little more easily on social media,” he added. “This could explain why social media in some cases is indeed associated with higher levels of depression, anxiety and other internalized disorders.”