Central to the study’s findings is the identification of what the authors term the “conflict entrepreneur” — a political figure who strategically employs personal attacks focused on the character, morality, or intellect of rivals rather than engaging on policy grounds. By examining the 118th United States Congress, convened from January 2023 to January 2025, the researchers analyzed an extensive database comprising over 2.2 million public statements, including floor speeches, press releases, newsletters, and social media communications. Leveraging advanced natural language processing via large language models, the team distinguished between legitimate policy critique and inflammatory personal invective, allowing for unprecedented granularity in understanding patterns of congressional rhetoric.

Intriguingly, the research reveals a pronounced asymmetry in the use of derogatory language between political parties. While both Democrats and Republicans engage in personal attacks to varying degrees, Republicans exhibit this behavior approximately 2.7 times more frequently. Furthermore, the incidence of such attacks is more prevalent in the House of Representatives as opposed to the Senate, with a 1.3-fold increase. These disparities point to systemic differences not only in partisan culture but also institutional norms that shape congressional interactions.

Perhaps the most striking insight from the study is the pronounced disjunction between conflict-driven communication styles and traditional markers of political success, such as fundraising prowess, electoral margins, legislative achievements, or personal wealth accumulation. Politicians dedicating a mere 5% of their discourse to personal attacks garner media attention on cable news networks comparable to peers whose discourse is overwhelmingly substantive. This disproportionate visibility is starkly evident in social media metrics: posts laced with insults enjoy nearly two and a half times more shares than those focusing on detailed policy discussions.

Nevertheless, this aggressive rhetorical style exacts tangible costs within the legislative domain. Lawmakers who habitually engage in personal invective tend to participate less in co-sponsoring legislation and receive fewer appointments to influential standing committees. This detachment from policymaking suggests that the pursuit of media attention may undermine traditional legislative roles and responsibilities, representing a strategic diversion from governance towards celebrity via conflict.

One of the study’s more unexpected findings challenges conventional wisdom linking incivility to the political leanings of legislators’ constituencies. The researchers found no statistically significant correlation between the extent of personal attacks and the level of partisan animus or polarization within electoral districts. In fact, many of the most vociferous conflict entrepreneurs represent electorates characterized by relative moderation. This decoupling implies that for a subset of political actors, objectives extend beyond traditional goals of reelection, legislative influence, or institutional power, gravitating instead towards maximizing media exposure.

The implications of these findings extend far beyond academic interest, illuminating the calibration of incentives in America’s burgeoning media ecosystem. The study underscores how the prevailing media attention economy rewards spectacle and conflict at the expense of deliberative policy debate, distorting political incentives toward incivility. Cable news and social platforms magnify the visibility of confrontational rhetoric, effectively incentivizing politicians to adopt aggressive communication tactics as a shortcut to national relevance.

Jacob and his colleagues caution that this burgeoning political communication strategy risks eroding foundational democratic norms by severing the link between visibility and accountability. As conflict entrepreneurs dominate airwaves and social feeds without corresponding legislative achievements or constituency representation, the integrity of democratic discourse faces unprecedented strain. The ascendancy of personal attacks threatens to normalize antagonism and reduce politics to entertainment spectacle.

The study’s sobering conclusion offers a dual mandate: political party leadership and media gatekeepers possess crucial agency in reshaping incentive structures. Leadership must champion policy-oriented discourse and curtail the tacit endorsement of incivility, while media organizations are called to critically evaluate what counts as newsworthy, prioritizing substantive governance over provocative personality clashes. Such corrective actions are essential to safeguarding the stability and health of American democracy in an era increasingly dominated by polarization and media saturation.

This research, conducted through the Polarization Research Lab, collaborates with scholars from Dartmouth College and the University of Pennsylvania, collectively advancing our understanding of political polarization and rhetorical strategies in contemporary governance. Their innovative integration of computational linguistic tools with large-scale political data establishes a new frontier in political science, blending empirical rigor with relevant real-world implications.

Ultimately, this study signals a critical inflection point in American political communication. The choices political actors make today, amplified by media platforms hungry for conflict, will shape democratic norms and citizen engagement for years to come. As Marc Jacob summarizes, the path forward demands deliberate cultural and institutional reforms that restore respect for policy debate as the cornerstone of representative democracy, resisting the alluring but corrosive temptation of personal attacks for fleeting attention.

Subject of Research: People

Article Title: Entrepreneurs of conflict: A descriptive analysis of when and how political elites use divisive rhetoric

News Publication Date: 17-Mar-2026

Web References:

• PNAS Nexus Article

References:
Jacob, M., Lelkes, Y., & Westwood, S. J. (2026). Entrepreneurs of conflict: A descriptive analysis of when and how political elites use divisive rhetoric. PNAS Nexus. https://doi.org/10.1093/pnasnexus/pgag038

Image Credits: University of Notre Dame

Keywords: Democracy, Political process, Sociopolitical systems, Mass media, News media, Social media, Rhetoric, Verbal communication

At the heart of this research lies the concept of affective alignment, a nuanced psychological phenomenon where individuals’ emotional responses and attitudes toward political subjects synchronize more deeply than their overall partisan identities might imply. Traditional political psychology often emphasizes partisan identity—the labels people choose, claim, or inherit (Democrat, Republican, Independent). However, this study challenges such conventional wisdom by applying natural language analysis to vast corpora of individuals’ expressed political opinions, demonstrating that emotional and cognitive congruence on issues is a driving force behind partisanship rather than the party label itself.

The research team deployed advanced computational techniques to analyze extensive bodies of text drawn from political speeches, social media posts, and survey responses. By utilizing sentiment analysis, topic modeling, and semantic vector representations, they could quantify the emotional valence and thematic coherence in people’s political communications. These methods enabled the researchers to move beyond surface-level categorizations of party identification and delve into the underlying psychological architecture of how political attitudes are structured. The results revealed that affective patterns around particular policy areas were more predictive of partisan divide than self-reported political identities.

This paradigm shift has profound methodological implications. It suggests that researchers must reconsider how political partisanship is operationalized in empirical studies. Rather than relying heavily on traditional self-report questionnaires that focus on party membership or identity labels, the incorporation of natural language processing offers a richer, more dimensional insight into political affect and cognition. This approach vividly illustrates that individuals can display higher emotional and evaluative overlap on political issues even when their official partisan identities differ or are modestly committed.

Further, the findings underscore why American political polarization has become increasingly intractable. Emotional alignment on divisive issues—such as immigration, gun control, or healthcare—is often deeply entrenched and resistant to change because it is tied to affective motivations rather than mere intellectual agreement. This study highlights how affective polarization reinforces political echo chambers, confirming the role of emotions in shaping and sustaining political tribalism. The stronger collective feelings toward politically charged topics overshadow simple nominal party affiliation and create fertile ground for division and conflict.

The use of natural language processing in this context marks a novel fusion of technology and political science. By examining the emotional and semantic content of political language at scale, the researchers demonstrate how computational tools can dissect complex human attitudes in ways unattainable by traditional survey methods alone. Their methodology involved aggregating millions of political utterances, mapping them onto multidimensional affective spaces, and statistically analyzing the degree of alignment within and across partisan groups. This kind of fine-grained linguistic analysis provides new evidence that personal emotions and issue-based evaluations are pivotal in shaping contemporary political identities.

Interestingly, the research also examined the phenomenon of cross-cutting alignments—instances where individuals who identify with one partisan group nonetheless express affective consistencies that overlap with ideological opponents on specific issues. This highlights the fluid, dynamic nature of political identity and suggests that issue-based affective congruence might bridge divides in certain contexts. Such findings could inform future political strategies and conflict resolution approaches, emphasizing shared emotional resonances rather than focusing solely on partisan differences.

Moreover, this study challenges the simplistic notion that partisanship is a stable, categorical identity. Instead, it supports a more dynamic model whereby political affiliations are context-dependent, malleable, and deeply entwined with emotional investments in political topics. This insight holds the promise of developing interventions to reduce polarization by targeting emotional undercurrents and perceived affective divergences on key substantive issues rather than attempting to shift ideological identities in bulk.

From a broader perspective, the study’s implications stretch beyond the academic sphere. Understanding affective alignment’s predominance offers a lens for media organizations, political campaigners, and social platforms to better comprehend audience segmentation and communication strategies. It suggests that political messaging that resonates affectively on specific issues may have far greater mobilizing power than broad appeals to partisan identity. This insight could revolutionize how political engagement and discourse are shaped in an era dominated by emotionally charged narratives and social media dynamics.

Additionally, the findings have repercussions for democratic governance. As political affective divisions deepen, policymakers must grapple with the challenge of crafting solutions that acknowledge and address the emotional attachments constituents have to contentious issues. Recognizing that emotions and affective alignments are at the core of partisanship may enable more empathetic and effective policymaking, promoting dialogue that bridges emotional divides and fosters more cooperative political environments.

Technically, the multidisciplinary approach of the study serves as a model of integrating psychological theory with computational linguistics. The team’s application of sentiment analysis tools, such as BERT-based sentiment classifiers fine-tuned on political discourse, coupled with dimensionality reduction techniques like t-SNE projection of semantic embeddings, allowed for unprecedented insight into the emotional contours of political speech. This methodological framework could be adapted to study affective alignments in other domains, such as social movements, international relations, or corporate communications.

Furthermore, these results challenge political theorists to rethink classical assumptions regarding partisan identity coherence. Traditional identity theories often assume that individuals derive stable self-concepts from party membership. This research indicates that affective consistency around issues may be a more salient marker of political identity, suggesting a layered, interactional construction of political selves. Such a framework aligns with emergent theories of identity fluidity and intersectionality but within the highly charged arena of electoral politics.

Looking ahead, the study opens multiple research avenues. Future work may explore the temporal dynamics of affective alignment—how affective patterns evolve over election cycles, in response to political events, or under the influence of media framing. Longitudinal studies could help identify whether affective alignment is a cause or consequence of partisan sorting or polarization. Moreover, expanding the natural language corpus to encompass international political contexts would test the generalizability of these affective alignment patterns beyond the American political landscape.

This pioneering work by Rim and colleagues contributes to a growing corpus of research emphasizing the centrality of affect in political behavior. By demonstrating that political partisanship is more accurately described in terms of affective alignment on policy issues rather than simplistic partisan identification, the study inspires a reexamination of political mobilization, polarization, and identity formation. It invites scholars, practitioners, and citizens alike to appreciate the complex emotional tapestries that underlie political division and solidarity.

Ultimately, the synthesis of computational methods and psychological insight embodied in this research marks an important step in decoding the emotional machinery of partisanship. It provides robust empirical evidence that political discourse is not just about facts, ideology, or group membership, but deeply about emotional convergence and alignment around the issues that matter most. As societies navigate the challenges of polarization, this insight could help foster more informed, compassionate, and effective approaches to democratic dialogue and political engagement.

Subject of Research: The study investigates the affective underpinnings of political partisanship, focusing on how emotional alignment over political issues drives partisan behavior more than formal partisan identities.

Article Title: Natural language reveals that political partisans are more affectively aligned over political issues than partisan identities.

Article References:
Rim, N., Jackson, J.C., Berman, M.G. et al. Natural language reveals that political partisans are more affectively aligned over political issues than partisan identities. Commun Psychol (2026). https://doi.org/10.1038/s44271-026-00430-x

Image Credits: AI Generated