A data science approach to 138 years of congressional speeches

• Main Authors: Ethan C. Tucker, Colton J. Capps, Lior Shamir
• Format: Article
• Language: English
• Published: Elsevier 2020-08-01
• Series: Heliyon
• Subjects: Speeches; Text analysis; Data science; Politics; Feature extraction; Data analytics
• Online Access: http://www.sciencedirect.com/science/article/pii/S2405844020312615

The availability of automatic data analysis tools and large databases have enabled new ways of studying language and communication that were not possible in the pre-information era. Here we apply a quantitative analysis to a large dataset of USA congressional speeches made over a period of 138 years. The analysis reveals that the readability index of congressional speeches increased consistently until the 96th congress, and then started to decline. Congressional speeches have also become more positive over time, and in general express more sentiments compared to speeches made in the 19th century or early 20th century. The analysis also shows statistically significant differences between Democratic and Republican congressional speeches.