Congressional Speech Networks: Their Formation and Importance

Thumbnail Image



Journal Title

Journal ISSN

Volume Title


Research Projects

Organizational Units

Journal Issue


Congressional speeches drive policymaking and political debate at the national level. As a crucial method of communication, these speeches directly reflect legislators’ political agendas and serve as a vehicle by which members of Congress can convey their political priorities and advocate for important issues. Prior to technological advances that made modern natural language processing methods computationally feasible, survey data and analysis of behavior in strategic settings (i.e. legislative roll-call voting) were used to analyze congress members' policy positions and social networks. However, both of these approaches lead to methodological conflicts; new surveys cannot capture past information and represent but a small snapshot of the present, while institutional rules constrain roll-call analysis. Congressional speeches, on the other hand, are less constrained by institutional rules and capture information over a period of time, thus providing a much deeper view into the positions, ideologies, and rhetorical relationships of members of Congress. Previous work applied well known natural language processing methods to determine the level of rhetorical similarity between speeches contained in the Congressional Record from 1981 to 2017, developing relational networks between members of Congress based on these similarities. These networks illuminated the lexical similarities between legislators, the links between rhetorical relationships, partisanship, and legislative effectiveness, and how rhetorical trends developed over three and a half decades. However, more work is needed to illuminate factors that contribute to the similarities in congressional speech that form the basis for network connections. This project extends previous work through the application of exponential random graph models to congressional speech networks, allowing for statistical analysis of the impact of various endogenous structural effects and exogenous covariates on the formation of congressional speech networks over time. Analyzing factors that contribute to the formation of network ties in each Congress provides significant insights into the evolution of relationships over time as well as the underlying dynamics and trends driving legislators’ rhetorical relationships. These insights improve our understanding of social ties in Congress and could even be used to inform strategies that aim to facilitate more effective political discourse and lawmaking in Congress.


Domestic and International Relations (The Ohio State University Denman Undergraduate Research Forum)


congress, speech, networks, language, exponential random graph model