During my junior and senior years at Washington University in St. Louis, I completed an honors thesis in political science, exploring evidence of language resemblance between congressional speeches and media content in three issues: gun control, abortion, and gay rights. To do so, I used Python to conduct a language analysis, tracking two- and three-word partisan phrases. In the end, I found that the nature of political rhetoric and the complex relationship of the media and members of Congress are both reasons why linguistic resemblance between the media and the language of one party is not a true indicator of media bias.

For more, read my abstract:

In a democracy like our own, the media ideally function to effectively disseminate information on political issues, with language that treats equally the viewpoints of both Democrats and Republicans. An existing literature on media bias seeks to determine how, when, and why the media may, instead, favor the perspective of one party at the expense of the other. In this paper, I explore the underlying rhetorical mechanisms driving the appearance of such media favoritism through a comparative phrase frequency analysis of congressional and media language on three highly partisan social issues: gun control, abortion, and gay rights. Based on the results of the analysis, demonstrating the particular types of partisan language adopted and avoided by the media, I argue that language resemblance between Congress and the media is an insufficient measure for media bias—despite literature that conflates one with the other. By illuminating the nature of both congressional and media language, the phrase frequency analysis serves to shed light on the complex, circular relationship between the two.

If you’re still interested, check out my full paper—but I suggest you grab a strong cup of coffee and get comfortable first.