‘Hope’ is out, ‘fight’ is in: Does tweeting divide Congress, or simply echo its divisions?

USA TODAY analyzed more than 2.8 million tweets posted by members of Congress. The language has become more divided and emotional.

A short decade ago, Congress Twitter was gilded with collegiality, well-wishing and civil debate.

Arizona Republican Sen. John McCain in 2011 thanked Senate Majority Leader Harry Reid, Democrat of Nevada, for his “kind words” following a floor speech. For that year’s state of the union, Rep. John Lewis, Democrat of Georgia, tweeted that he was “very pleased to be sitting with my friend and colleague” Rep. Tom Price, Republican of Georgia.

McCain recommended a book by Sen. Joe Lieberman, Independent of Connecticut. Sen. Tom Harkin, Democrat of Iowa, congratulated Republican Sen. Chuck Grassley in a 2012 tweet for his long tenure in Congress.

Oh, how times have changed.

Although many Americans sense the nation’s lawmakers have become increasingly polarized in the past 10 years, a new USA TODAY language analysis of 2.8 million congressional tweets illustrates just how dramatically this particular form of public discourse has shifted, in tone and in substance, since 2011.

Early congressional adopters of Twitter in both parties commonly used phrases like “read,” “benefits” and “hope.” By 2021, Democrats were more apt to use words such as “victim,” ”condemning” and “firearms.” Republicans were more often saying “communist,” “bureaucrats” and “woke.”

The data show lawmakers segregating into distinct rhetorical clusters linked by party, effectively speaking different languages. And both parties grew more emotional in their choice of words through the implementation of Obamacare, government shutdowns, two impeachments and the overturning of Roe v. Wade.

An unanswered question: Did the decline of common ground on Twitter reflect a fundamental change in attitude among lawmakers, or did the rise of Twitter itself accelerate the breakdown — by encouraging more and more pithy attacks? Either way, experts say the outcome is not healthy.

Libby Hemphill, a professor at the University of Michigan who studies political communication, said the analysis underscores how parties are increasingly talking about different issues, “which means what’s important to one side isn’t important to the other. It means we can’t agree on what we should talk about.”

That has consequences beyond posturing on social media, said Hemphill. “Polarized parties don’t make good legislation. That’s why it matters.”

Equally troubling is that this type of language from politicians “tends to drive mass polarization rather than vice versa,” said Yphtach Lelkes, a political scientist at the University of Pennsylvania. “It makes people see the world as much more polarized and treacherous than it actually is, which then is self-fulfilling.”

USA TODAY used a technique known as agglomerative hierarchical clustering to categorize tweets by their linguistic similarity. The analysis includes all words appearing more than five times in tweets from the official accounts of members of the 112th through 117th Congresses tweeting during those two-year periods. Two Independents who caucus with Democrats are colored blue in cluster visualizations and omitted from further analysis. Twitter text content is courtesy of ProPublica, and news article URLs are from Technical University of Graz, Austria. A list of official congressional Twitter accounts came from Ballard et al.’s 2022 study “Dynamics of Polarizing Rhetoric in Congressional Tweets.” Emotional lexicon is from Simchon et al.’s 2022 study “Troll and divide: the language of online polarization.” Code for this analysis can be found on GitHub.

This analysis was informed by the reporter’s Summer 2022 fellowship at the Complexity Science Hub Vienna.