What Is Being Argued (WIBA)? An Application to Legislative Deliberation in the U.S. Congress

📅 2024-07-08
🏛️ arXiv.org
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This study investigates the evolutionary mechanisms underlying deliberative dynamics in political debate, specifically examining how partisan polarization and information manipulation degrade deliberative quality. Methodologically, it constructs an extended WIBA (Weighted Ideological Balance Analysis) framework applied to U.S. congressional hearing transcripts from 2005–2023, integrating pretrained language models for argument extraction, dependency parsing, and semantic role labeling, alongside novel interpretable metrics—such as deliberativeness and contentiousness—and interactive visualizations. The analysis systematically identifies structural partisan asymmetries in argumentative dominance, agenda-setting power, and rebuttal intensity within “hot-button” issues—a first-of-its-kind finding. It empirically confirms institutional information filtering and party-driven discursive control at the committee level. These results establish a reproducible methodological paradigm and empirical benchmark for computational political science and digital governance research.

Technology Category

Application Category

📝 Abstract
How can we utilize state-of-the-art NLP tools to better understand legislative deliberation? Committee hearings are a core feature of any legislature, and they offer an institutional setting which promotes the exchange of arguments and reasoning that directly impact and shape legislation. We apply What Is Being Argued (WIBA), which is an argument extraction and analysis framework that we previously developed, to U.S. Congressional committee hearings from 2005 to 2023 (109th to 117th Congresses). Then, we further expand WIBA by introducing new ways to quantify various dynamics of democratic deliberation. Specifically, these extensions present a variety of summary statistics capturing how deliberative or controversial a discourse was, as well as useful visualizations to the WIBA output that aid analyzing arguments made during the legislative deliberation. Our application reveals potential biases in the committee system, and how political parties control the flow of information in 'hot topic' hearings.
Problem

Research questions and friction points this paper is trying to address.

Analyze discourse dynamics in legislative and social media debates
Identify arguments and deliberation intensity from raw texts
Model discourse evolution over time in diverse platforms
Innovation

Methods, ideas, or system contributions that make the work stand out.

Data-driven argument-centric framework DALiSM
Computational argumentation for discourse analysis
Interactive dashboard for dynamic discourse visualization
A
Arman Irani
UC Riverside
J
Ju Yeon Park
The Ohio State University
K
Kevin E. Esterling
UC Riverside
M
Michalis Faloutsos
UC Riverside