Legislative committees in the US Congress play an important role in holding policymakers to account in oversight hearings. While most empirical studies to date which look at how these committees deliberate have focused on the words and arguments spoken, very little work has been done on the effect of nonverbal communication in accountability hearings. Because of the COVID-19 pandemic, the use of Zoom and other visually remote ways of communicating have made the importance of nonverbal communication even more important.
This project has examined the hearings on economic policy by the House Financial Services Committee and the Senate Banking Committee, both immediately prior to the lockdown conditions brought about from COVID-19 and then those conducted remotely in the months following these restrictions. Through the lense of natural language processing, we have sought to unlock the changes the effectiveness of accountability hearings in the changing context of remote communication.
Methods
First, my research began with an indepth literature review, analysing the impacts of a society wide shift to zoom. Everyone has shifted most all aspects of their life online, and all of us have experienced the effects of this – be it a reluctance to engage in yet another Zoom meeting, or pining after a watercooler conversation. We first began by attempting to understand the academic work specifically characterising the impact of Zoom in non-political elements of life, looking at the impact in business meetings of small lags causing cross talk and interuptions, and venturing even into scientific studies discussing why an infinitesimal delay can prevent the natural flow of conversation.
After understanding the broad context into which our work would fall, I next began performing the specific analytical work. Here, I took the transcripts from the congressional hearings and edited them in a manner such that they could be fed into different software analysis packages. This step took the majority of my time: data cleaning and reconciling differences between different transcript formats and different software preferences. Through collecting this data, I was able to use natural language processing software to analyse the level of deliberative accountability.
Personal Reflections
This internship has been a fantastic opportunity in so many ways. In a direct sense, this project has been fantastic for giving me the chance to dive into topics I find fascinating: I have taught myself how to perform sentiment analysis on R, or use other non-standard software packages; I have had the opportunity to dive into policy debates, and gain exposure to politics in its most raw sense – allowing me to engage in a depth I’d never previously considered possible.
Beyond these direct skills, this project has been a peerless opportunity for allowing me the chance to develop my character holistically.
Professor Schonhardt-Bailey trusted me with great independence and I benefited from an academic freedom and feeling of responsibility on offer in no other part of my LSE education. Working with a world-leading academic has given me exposure, and some experience, of how research is performed at the highest levels.
The value of this experience can not be overstated. I have found my interests shaped and developed through this position, where such independent study gave the me the chance to learn so much about myself. In many a job interview I have discussed facing challenges in this project, or how this project developed in me an ability to deal with responsibility. Without this internship, I would have had very little to say in these interviews – and I believe the internship is wholly responsible for my success therein.