Rhetoric and Research: The Political Language of Autocrats and Its Impact on Science Policy
P9-S239-3
Presented by: Sukayna Younger-Khan
Does the science rhetoric of political leaders influence science policy in non-democracies? This paper investigates how autocrats use science rhetoric in speeches and how it affects science policy outcomes during their tenure. While research has focused on the politicization of scientific issues like climate change and vaccination in liberal democracies, this work examines the oft-neglected intersection of political communication and policy outcomes in non-democracies. It explores how politicized science discourse impacts funding for higher education and research and development, analyzing data from 55 leaders across 33 countries from 1990 to 2023. The paper makes three key contributions. First, it develops a conceptual framework for science rhetoric, distinguishing between normative and institutional aspects of political language. Second, it introduces a novel method using natural language processing to detect these dimensions of rhetoric through contextual semantic similarity. Third, it assesses the relationship between rhetoric and policy outcomes in regression models. Early results show significant variation in normative and institutional rhetoric among authoritarian leaders both within and across countries. Autocrats using more institutional rhetoric tend to allocate more funding to higher education and R&D, in contrast to those relying on normative rhetoric. Institutional constraints mediate the relationship between rhetoric and policy outcomes, with more constrained leaders showing stronger alignment between rhetoric and action. These findings highlight the stakes of scientific discourse as a battleground for both open inquiry and political manipulation, with implications for global challenges like climate change and public health.
Keywords: Political Communication, Politicization of Science, Science and Technology Policy, Natural Language Processing, Large Language Models