(Assistance provided by Claude AI)
This research paper tackles a compelling question: Do American presidents actually speak differently from each other, and if so, how can we measure those differences scientifically? The researchers from the University of Chicago developed sophisticated new tools to analyze presidential speech and discovered some striking patterns, particularly regarding Donald Trump’s communication style.
Zhou K, Meitus AA, Chase M, Wang G, Mykland A, Howell W, Tan C. Quantifying the uniqueness and divisiveness of presidential discourse. PNAS Nexus. 2024 Oct 7;3(10):pgae431. doi: 10.1093/pnasnexus/pgae431. PMID: 39440021; PMCID: PMC11495212.
Understanding the Research Question
Think about how you might immediately recognize a friend’s voice on the phone, even before they say who they are. Similarly, you might notice that some politicians have distinctive ways of speaking – perhaps they use certain phrases repeatedly, or they structure their sentences in particular ways. This study wanted to move beyond those casual observations to measure presidential speech differences with mathematical precision.
The researchers faced an interesting challenge: previous studies of presidential communication relied mainly on simple word counts or basic sentiment analysis. But human language is far more complex than that. The way we arrange words, the context we provide, and the flow of our sentences all contribute to our unique “linguistic fingerprint.” To capture this complexity, the team needed more sophisticated tools.
Three Innovative Measurement Approaches
The researchers developed three complementary methods to analyze presidential speech, each capturing different aspects of communication style.
Large Language Model Analysis: Measuring “Uniqueness”
The first approach leverages artificial intelligence models similar to ChatGPT to measure how “predictable” or “unique” each president’s speech patterns are. Here’s how this works: imagine you’re playing a word-guessing game where you try to predict the next word in a sentence. If someone always speaks in very predictable patterns, you’d guess correctly more often. But if they have a unique speaking style, their word choices would surprise you more frequently.
The researchers trained AI models on presidential speeches and then tested how well these models could predict what each president would say next. When a president’s actual words were much harder to predict than expected, this indicated a more “unique” speaking style. This approach considers not just individual words, but the entire context and flow of speech.
Divisive Language Lexicon: Quantifying Antagonistic Speech
The second tool addresses something that many people sense intuitively but that hadn’t been systematically studied: some political figures use more antagonistic, divisive language than others. The researchers created the first scientific lexicon (or dictionary) specifically designed to identify “divisive” political speech.
They define divisive language as words explicitly intended to “impugn and delegitimize” a target – going beyond mere disagreement to language designed to attack someone’s intelligence, integrity, or intentions. Words like “corrupt,” “stupid,” “racist,” or “ridiculous” fall into this category. Four independent researchers reviewed hundreds of potential words to create a final list of 178 terms that consistently qualified as divisive in political contexts.
This lexicon differs from other measures you might have heard about. It’s not just about “negative” language or “impolite” speech, but specifically about language designed to marginalize and distance the speaker from their target for political purposes.
Opponent Reference Analysis: How Politicians Talk About Their Rivals
The third approach examines how candidates specifically talk about their political opponents. During presidential debates, for instance, candidates frequently reference each other or the opposing party. The researchers developed methods to isolate these opponent-focused statements and analyze what kinds of descriptive words each candidate uses when discussing their rivals.
They employed a technique called “Fightin’ Words” analysis to identify which adjectives each candidate uses most distinctively when talking about opponents, then measured how much overlap exists between different candidates’ opponent-describing vocabulary.
The Data: A Comprehensive Collection
To ensure robust findings, the researchers analyzed three distinct types of presidential communication spanning several decades:
Presidential debates from 1960 to 2020 provided a controlled setting where candidates respond to similar questions and directly interact with opponents. State of the Union addresses from 1961 to 2022 offered examples of formal, official presidential communication. Campaign speeches from 2008 to 2020 captured the more informal, rally-style communication that candidates use when speaking directly to supporters.
This multi-format approach was crucial because it allowed the researchers to determine whether any observed differences were specific to certain types of speaking situations or represented consistent patterns across all contexts.
Key Findings: Trump as a Statistical Outlier
The results revealed a clear and consistent pattern across all three measurement approaches and all types of speeches: Donald Trump’s communication style differs dramatically from every other modern president and major party presidential nominee.
Uniqueness Scores
When measuring predictability using the large language model approach, Trump consistently scored as the most “unique” speaker. This means his speech patterns were the most difficult for AI models to predict based on patterns from other presidents. Notably, the difference between Trump and other Republican presidents was actually larger than the typical difference between Republican and Democratic presidents.
This finding is particularly striking because it suggests Trump’s distinctiveness isn’t simply about partisan differences – he speaks differently even from presidents of his own party.
Divisive Language Usage
Trump used words from the divisive language lexicon more frequently than any other president across all types of speeches. In campaign speeches, his most commonly used divisive words included “crazy” (used 135 times), “corrupt” (111 times), and “stupid” (69 times). Even in more formal settings like State of the Union addresses, where divisive language is generally rare, Trump used such terms more frequently than his predecessors.
The researchers found that sentences containing divisive words also tended to be more “unique” according to their AI analysis, suggesting that antagonistic language contributes to, but doesn’t fully explain, Trump’s distinctive communication patterns.
Opponent References
Trump mentioned his political opponents more frequently than other candidates, and when he did so, he used distinctively different descriptive language. While other candidates might use words like “wrong” or “misguided” when criticizing opponents, Trump’s opponent-focused vocabulary included terms like “disgraceful,” “radical,” and “super” – language that overlapped less with other candidates’ opponent-describing words.
Important Patterns and Robustness
Several additional findings strengthen confidence in these results:
The researchers confirmed their findings using multiple different AI models, not just one, and the patterns held consistently. Trump’s uniqueness wasn’t simply due to using shorter sentences (though he does), because the patterns persisted even when comparing sentences of similar lengths. The differences weren’t artifacts of changing communication norms over time – Trump appeared just as distinctive compared to contemporary candidates as to those from earlier decades.
Perhaps most importantly, Trump’s distinctive patterns appeared across all three types of communication contexts, suggesting these represent fundamental aspects of his communication style rather than situation-specific adaptations.
Understanding the Broader Implications
These findings contribute to our understanding of several important questions about modern American politics. The research provides empirical support for widespread intuitions about Trump’s distinctive communication style, but quantifies these differences in ways that allow for more precise analysis.
The study also offers insights into populist political communication more generally. The researchers suggest that Trump’s combination of unique speech patterns and divisive language may contribute to his appeal as a political outsider who “tells it like it is” – even when this involves breaking conventional norms of political discourse.
From a methodological standpoint, the tools developed in this research could be applied to analyze other political figures, both in the United States and internationally. The techniques for measuring speech uniqueness and divisiveness could help researchers study populist movements, authoritarian rhetoric, or changes in political communication norms over time.
Limitations and Future Directions
The researchers acknowledge several important limitations in their work. Their campaign speech data focuses primarily on recent elections due to availability constraints, and they don’t include social media communication (like Trump’s extensive use of Twitter) because comparable data doesn’t exist for earlier presidents.
The study also raises questions about the relationship between distinctive communication styles and political effectiveness. While they demonstrate that Trump’s speech is measurably different, they don’t directly examine whether these differences contribute to his political success or appeal to specific voter groups.
Future research might explore whether other political figures adopt similar communication strategies, whether Trump’s style influences other politicians (a “contagion effect”), or how these speech patterns relate to broader measures of political polarization and democratic norms.
Conclusion: Quantifying Intuition
This research represents a sophisticated attempt to quantify something many people sense intuitively about political communication. By developing new computational tools and applying them systematically to decades of presidential speech, the researchers provide strong empirical evidence that Donald Trump’s political rhetoric represents a significant departure from established patterns of presidential communication.
The study’s strength lies not just in its findings about one particular figure, but in its methodological innovations that could reshape how we study political communication more broadly. In an era of increasing concern about political polarization and the erosion of democratic norms, having precise tools to measure and track changes in political discourse becomes increasingly valuable for both scholars and citizens trying to understand our changing political landscape.