26.06.2025

Teaser image to What Words Reveal: Analyzing Language in the Trump–Harris 2024 Debate

What Words Reveal: Analyzing Language in the Trump–Harris 2024 Debate

MCML Research Insight - With Philipp Wicke

In presidential debates, every word counts - not only what candidates say but how they say it. Words are carefully selected to resonate with voters’ values, fears, and hopes. A new insightful study by MCML Junior Member Philipp Wicke and Marianna M. Bolognesi dives deep into the September 10, 2024 debate between Donald Trump and Kamala Harris.

Their central question:

Do Democratic and Republican Candidates Genuinely Differ in Their Linguistic Strategies?


Trump and Harris Word Cloud

The word cloud shows red words for Trump and blue words for Harris, with word size representing frequency, highlighting the most common terms in their responses.

Methodology: Combining AI With Human Expertise

The researchers analyzed the complete debate transcript using sophisticated natural language processing techniques and expert human annotations. They assessed emotional tone, concreteness, specificity, figurative language (like metaphors), and pronoun usage. Large language models, such as GPT-4o, were utilized to detect figurative frames, and these AI-generated results were rigorously validated by human experts, ensuring accuracy and reliability.

What Is Framing?

Framing refers to how authors shape the meaning of their messages by emphasizing certain perspectives or values - often through metaphors or emotionally charged imagery. For example, describing immigration as a “flood” frames it as a natural disaster, influencing how the audience perceives the issue.


«Trump’s figurative frames reveal distinct themes at both content and framing levels… framing the country as facing existential decline, using metaphors like ‘we’re a failing nation,’ ‘our country is being lost,’ and ‘they’re destroying our country.»


Philipp Wicke

MCML Junior Member

Key Linguistic Findings

The analysis unveiled clear rhetorical distinctions aligned with each candidate’s political identity:

  • Framing the Narrative: Harris frequently utilized metaphors emphasizing growth, unity, and recovery, crafting a narrative around national progress and collective strength. Trump, conversely, painted scenarios of decline, threat, and crisis, using vivid, emotionally charged language to depict invasions, societal collapse, and betrayal.
  • Emotion and Subjectivity: Both candidates employed negative emotional language, reflecting the polarized political atmosphere. Trump’s rhetoric was more negative and somewhat more “objective,” framing his opinions as facts to assert authority. Harris used language that was slightly more subjective, indicating empathy and personal conviction.
  • Specificity and Concreteness: Contrary to popular expectations, both Trump and Harris exhibited similar levels of abstract versus concrete language, challenging assumptions about vagueness or excessive policy detail attributed to one side or the other.
  • Pronoun Usage and Identity: Pronouns revealed notable strategies. Harris balanced “we” and “I,” highlighting collective responsibility and inclusive leadership. Trump predominantly used “I” and “me,” reinforcing personal authority and decisiveness. Additionally, Trump never directly mentioned Harris by name, while Harris frequently referred to Trump explicitly, reflecting contrasting debate tactics.

The Bigger Picture

These findings resonate with broader patterns observed in political discourse. Republican language often stresses strength, individualism, and moral urgency, whereas Democrats typically emphasize empathy, collective action, and social justice. This study not only confirms these established ideological divisions but shows how deeply they influence the fundamental structure of candidates’ communication strategies.


Want to Explore More?

The complete study, featuring in-depth analyses, visualizations, and full datasets, is published in the renowned PLOS One journal and freely available. Examine the data or explore the underlying code to discover the strategic language choices that influence political debates. Gain insight into the language of leadership and how it influences political reality or apply the tools to analyze a different debate entirely. While the 2024 debate is now part of the past, the code remains highly relevant and easily adaptable for examining future political showdowns between two candidates.

Top Journal
P. Wicke • M. M. Bolognesi
Red and blue language: Word choices in the Trump and Harris 2024 presidential debate.
PLOS One 20.6. Jun. 2025. DOI GitHub

Share Your Research!


Get in touch with us!

Are you an MCML Junior Member and interested in showcasing your research on our blog?

We’re happy to feature your work—get in touch with us to present your paper.

#blog #research #schuetze

Related

Tiny logo
Link to MCML at KDD 2026

07.08.2026

MCML at KDD 2026

MCML researchers are represented with 5 papers at KDD 2026.

Read more
Link to Benedikt Wiestler: We Want to Build a Time Machine

31.07.2026

Benedikt Wiestler: We Want to Build a Time Machine

MCML PI Benedikt Wiestler explains how AI models help to develop specific strategies in clinical therapy for brain tumor patients.

Read more
Link to Gaps in the Benchmark: Why Medical AI Fails in the Real World

28.07.2026

Gaps in the Benchmark: Why Medical AI Fails in the Real World

MCML researchers show in Nature Health how dynamic red teaming exposes hidden weaknesses in Medical AI beyond static benchmarks.

Read more
Link to Barbara Plank Becomes President of the Association for Computational Linguistics

26.07.2026

Barbara Plank Becomes President of the Association for Computational Linguistics

MCML PI Barbara Plank becomes President of the Association for Computational Linguistics, the world's leading NLP organization.

Read more
Link to Konstantin Riedl Receives 2026 SIAM Student Paper Prize

25.07.2026

Konstantin Riedl Receives 2026 SIAM Student Paper Prize

Former MCML Junior Member Konstantin Riedl receives the 2026 SIAM Student Paper Prize for research on nonconvex optimization.

Read more
Back to Top