Mainstreaming: Using Natural Language Processing to Assess Linguistic Patterns in Think Tanks and Extremist News Sites

This paper investigates a worrying trend: while Russian disinformation was once relegated to fringe outlets like RT and bot farms, current networks now leverage American and Canadian think tanks and pundits to spread narratives that undermine trust in American democracy and global leadership. By using domestic actors, disinformation becomes harder to identify in U.S. media. Using NLTK, I examine the linguistic features of disinformation organizations. I find that the Center for the National Interest—a think tank with alleged Russian ties—shares nearly indistinguishable policy language and linguistic features with mainstream think tanks. Similarly, GlobalResearch, a leading Canadian-Russian disinformation site, exhibits linguistic similarities to InfoWars. These findings reflect key trends: a convergence of sources reusing content and a flattening of the news hierarchy, where previously “reputable” outlets no longer stand clearly apart from fringe voices.