Parsing the Language of Legislation: A New Pipeline for Rhetorical Discourse Analysis of Committee Hearings
arXiv:2611.05381 [cs.CL] · Submitted 30 July 2026 · 10 pages, 4 figures
Authors: Isaac Rudnick, Nipun Das, Foaad Khosmood
Overview
Legislative hearings are distinct communication events with their own human-recognizable structure and logic, yet they have largely been analyzed as isolated phenomena divorced from their transcripts. Understanding hearings at a deeper level is a major step toward analyzing the work and power relationships within democratic bodies. In this work, we present a novel end-to-end pipeline for parsing rhetorical structures from legislative committee hearings, with the goal of improving legislative discourse analysis and democratic accountability.
Approach
We propose a rhetorical structure grammar and find that 55.21% of California state bill discussions in our dataset can be parsed with this grammar. We then evaluate our pipeline using two utterance section classifiers on a random subset of over 2,000 annotated utterances corresponding to 100 hearings.
Results
We find that a logistic regression-based classifier with grammar-based output masking and utterance tag smoothing outperforms more complex models, achieving an F1 score of 52.8% against our ground truth discourse section labels.
Implications for 2026 and Beyond
As legislative bodies increasingly publish full transcripts and streamed hearing archives—and as AI-assisted policy analysis becomes standard practice—structured, computationally tractable representations of democratic deliberation are in growing demand. This work offers a foundation for automated discourse analysis that can support transparency, accountability, and downstream civic technology.
Metadata
- Subjects: Computation and Language (cs.CL)
- Cite as: arXiv:2611.05381 [cs.CL]
- DOI: https://doi.org/10.48550/arXiv.2611.05381
- Comments: 10 pages, 4 figures
Submission History
- [v1] Thu, 30 Jul 2026, 00:13:23 UTC (1,193 KB)
via ArXiv CL+LG