Mental health is becoming a growing issue in our society, and the problem may be showing up in our language and literature.
In a study published in PNAS, “Historical language records reveal a surge of cognitive distortions in recent decades,” researchers from the Luddy School of Informatics, Computing, and Engineering and colleagues at IU’s Department of Psychological and Brain Sciences analyzed millions of books published over the past 125 years. They discovered that, since the 1980s there has been a surge of language indicative of “cognitive distortions,” thinking patterns usually seen in internalizing orders such a depression and anxiety.
“It has long been known that depression is associated with cognitive distortions, wherein people think about themselves, others, and the world in absolutist, inaccurate, and negative ways,” said Johan Bollen, a professor at the Luddy School and one of the lead researchers on the project. “For instance, someone might think to themselves, ‘Nobody loves me because I will always be a total loser.’ Those kinds of thoughts are associated with negative changes in mood, behavior, and motivation, which is broadly why they are targeted by cognitive behavioral therapy.”
Bollen and his group previously discovered that the online language of people with depression contains significantly more makers of cognitive distortions than a random sample, a difference which can be automatically detected via computer algorithms. They decided to investigate whether that fact held for society at large.
“Our analysis of the language used in a collection of more than 14 million books published from 1855 to 2019 in the United States, and German-, and Spanish-speaking countries, reveals a worrisome pattern,” Bollen said. “We see a pronounced ‘hockey stick’ pattern in which the use of cognitive distortion expressions surged well above historical levels in recent decades, including those of the great depression, and World War I and II, after declining or stable levels for most of the 20th century. Our results point to the possibility that socioeconomic changes, new technology, and social media have involved societal changes that lead to higher levels of expression of cognitive distortions.”
The research suggests that it may be possible to detect whether people individually—and society as a whole—are undergoing changes in their mental health by looking for subtle changes in the structure of the language being used.
“Early detection of such patterns may allow more effective prevention efforts from the perspective of public health, to promote societal well-being at the level of culture and language, as well as efforts to promote individual mental health” Bollen said.
This analysis is part of a large interdisciplinary research program that explores the development of mental health disorders from large-scale data using machine learning and AI. The group plans to expand this analysis to other sources of language, other languages, and cultures while continuing to develop models to study individual and societal mental health trajectories.
“Using algorithms and machine learning to identify changes in language patterns showcases the way technology can be used to draw out insights that otherwise would have been impossible to identify,” said Kay Connelly, associate dean for research at the Luddy School. “This project and the methods being developed hold so much potential for helping change public policy going forward.”