As the US transitions into post-election fallout, a recent study from Carnegie Mellon University might demonstrate how language has defined the nation’s dividing lines.

Researchers at Carnegie Mellon University began a study in January 2014 that has observed the correlation between political leanings and language… language that might even need translation. Publishing the findings last month, they discovered a consistent pattern of differing language across the political spectrum.

Applying the hypothesis that conservative and liberal Americans speak two distinct languages, the researchers collected data from the YouTube channel comment sections of CNN, Fox News, MSNBS, and One America News Network (OANN). They used that data to compile a lexicon of words and phrases with strong connections, albeit polar opposites.

The derogatory ways people might refer to those with opposing political views (trumptards, snowflakes) was consistent, as were references to individuals (chump, trump), (pelosi, pelousy), (obama, obummer), (barr, weasel), (biden, creep). The researchers also discovered opposing phrases like “Republicans/Democrats are the greatest threat to America…are traitors…are patriots.”

“We do not intend these categories to be formal or exhaustive, but rather to be illustrative of the types of misaligned pairs we encountered,” the study explains. Some of the pairs map to the same entity, such as the terms “liberals” and “libtards,” while the rest map to completely different entities and beliefs.

With 10.6 million subscribers and almost 100,000 videos, CNN provided the most voluminous set of data, with Fox News just shy of 6.1 million subscribers and 65,000+ videos. MSNBC follows with about 3.5 million, and OANN, which largely aligns with President Trump, has 800,000+ subscribers.

While the study provides some interesting connections between conservative and liberal language, the researchers also acknowledge some drawbacks to using an internet platform.

“It is not possible to unambiguously identify if a YouTube user prefers CNN over Fox News or not,” they explain. “We assign a user to CNN or Fox News using a simple filter. We acknowledge that our filter makes certain assumptions that may not hold in the wild. It is possible that a user only comments on a video if she does not agree with its content.”

Whether these commenters are left or right, boomer or millennial, serious or trolling, this study at least illuminates the language that the commenters have determined appropriate for the context.

To learn more about the stark contrasts between the American political lexicons, MultiLingual‘s most recent issue features a deep analysis of the language propagated among both ideologies, and considers this Carnegie Mellon study as well as other research.