Developing Digital Infrastructure and Automated Text-Analysis Methods for Identifying Representative Claims in Political Texts: First Findings from Speeches and Social Media Data

Representative claims are important yet understudied parts of political representation, including the ways political actors construct constituency identities, shape public discourse, and legitimize policies. However, established methods for analyzing these claims are labor-intensive and context-specific, limiting their scalability and applicability to large corpora of text. While qualitative studies of representative claims provide us with the necessary theoretical foundations, quantitative approaches to study, for example claim-making patterns and predictors, are still limited. In this presentation, we introduce the early findings of the project "Developing Digital Infrastructure and Automated Text-Analysis Methods for the Identification of Representative Claims." Leveraging advanced Natural Language Processing (NLP) techniques, including Bidirectional Encoder Representations from Transformers (BERT) models, we aim to systematically identify and analyze representative claims across diverse text types and languages. Our initial results focus on two key text corpora: social media posts and political speeches. These corpora provide rich, high-variance datasets that showcase the potential of semi-automated approaches to uncover implicit and explicit representative claims. We present the methodological framework, including the preprocessing pipeline and model training, along with an evaluation of performance metrics. Our initial findings show the capacity of automated methods to capture the performative aspects of representation, offering insights into claim-making strategies and constituency construction, and invite the audience for feedback as we aim for user friendliness and applicability in political science research. By developing scalable tools and workflows, we aim to enhance the understanding of representative dynamics in political discourse, with implications for comparative research across time, political systems, and languages. The presentation concludes with a discussion of future research directions, including validation strategies, multilingual model development, and applications in policy analysis.