The way a person uses language is a direct route to understanding the way they think. Today, researchers don’t just bring people into the lab to study this connection -- the online world gives us access to what social scientist David Lazer and colleagues have called “digital traces that can be compiled into comprehensive pictures of both individual and group behavior, with the potential to transform our understanding of our lives, organizations, and societies.” At the University of Maryland, Dr. Philip Resnik is developing new computational methods for analyzing the digital traces of language use and online behavior, with two goals: to advance our scientific understanding of how language reflects and influences our perceptions, and to develop practical technology, using the science to help solve pressing societal problems.

Mental health issues are a case in point. In the U.S., the human cost of depression-related conditions is staggering: some 25 million American adults will have an episode of major depression this year, suicide is the third leading cause of death for people 10–24 years old, and 18.5% of returning military service members meet criteria for PTSD or depression. Worse, 89.3 million Americans cannot even obtain ready access to a clinician who is qualified to perform a psychological evaluation, and standard screening and diagnostic methods are often limited by the motivation of many individuals to underreport psychiatric symptoms and lack of the self-awareness to report accurately. What if we were able to make early detection and monitoring of mental health problems easier and dramatically more accessible, by analyzing the day-to-day language that people use in their online interactions?

The research agenda goes beyond mental health, though. Language behavior is also a powerful signal of other mental states such as political ideology -- and, conversely, language in traditional and social media can powerfully influence people’s perceptions of political and policy issues. The potential of online language use and behavior to help us understand and even improve the political process is largely untapped. Using the new data and computational methods available to us, how can we identify subtle “spin” in political language and better understand the influence of framing on people’s perceptions and decision-making? How can we track the rise and fall of different “angles” on public issues, or even identify signs that someone’s extreme perceptions may be leading them in the direction of violent action?

Current areas of emphasis in this research include the following:

• Mental health. In 2014, Dr. Resnik launched the first research event designed to bring together computational linguists and clinical psychologists -- the Association for Computational Linguistics Workshop on Computational Linguistics and Clinical Psychology, clpsych.org -- and he is working with psychologists to develop new computational models of language and mental state that take advantage of his decades of experience combining linguistic insight, subject matter expertise, and “big data” analysis.
• Political framing. This question, too, is about the connection between language and mental state. Dr. Resnik is working with political scientists to apply the same key ideas to the analysis of language in the legislative process, traditional media, and social media, in order to shed light on the way that language is used (and sometimes abused) in affecting our decision-making, and to provide new methods for public opinion research.
• New computational tools. Dr. Resnik introduced a new, mobile technology for measuring people’s responses to the language they hear with unprecedented scale and accuracy. The mobile platform helped a team of WJLA-ABC7 and Politico journalists create award-winning coverage of the Obama/Romney debates in 2012, collected reactions to Super Bowl ads for a top consumer-media research firm, and guided Washington Post fact-checking during a recent gubernatorial race. (It is now being spun out as a startup.) His next phase of technology development is focused on methods for identifying and monitoring mental health issues, with an emphasis on collaboration with clinicians to create accurate techniques and address privacy issues responsibly.