Mapping Political Negotiation Styles Through AI-Based Language Analysis

Abstract: Political negotiations are often analyzed using surveys, expert ratings, or simplified typologies such as integrative versus distributive styles (Voeth & Herbst, 2015). However, these approaches are subjective, only moderately scalable, and often capture political communication indirectly because they rely on selective perceptions and retrospective evaluations (Mikhaylov et al., 2012). AI-based methods offer new opportunities for systematic, data-driven analyses of political negotiation styles (Ceron et al., 2023). This study develops a replicable approach using AI-generated pairwise similarity judgments to capture negotiation styles from large-scale public communication data (e.g., speeches, press conferences, official statements). For 20 influential political actors (e.g., Merkel, Macron, Trump), stylistic similarities are rated on a 1–10 scale. Multidimensional scaling maps these data into a two-dimensional style space with excellent fit (normalized Raw Stress = 0.032; Tucker = 0.984; DAF = 0.968; S-Stress = 0.058). Ward and k-means clustering identify four stable groups, revealing neighborhoods and isolated profiles.

Keywords: Empirical Study, Negotiation, Artificial Intelligence (AI), Politics, Politicians, Negotiation Styles, Multidimensional Scaling (MDS)