Riva Quiroga—Outlier 2021—Beyond Word Clouds: Visualizing Linguistic Patterns of Political Speeches

Beyond Word Clouds: Visualizing the Linguistic Patterns of Political Speeches This talk presents a project about visualizing 188 years of presidential addresses to the congress in Chile. It aims to analyze these speeches not as isolated events, but as pieces of a major sequence that can help us understand history and change. Along with showing the resulting data visualizations, the talk will discuss some aspects related to the analysis of textual data, specifically, the importance of not concentrating only on word frequencies, but also on their distribution, dispersion, and absence. When exploring these speeches using word clouds (probably the most extended type textual data visualization), they look strikingly the same, like if every year presidents talk about the same topics. But when experimenting with other ways of visualizing textual data, patterns emerge. Even punctuation, a type of data that is usually discarded in analyses focused on content, can be relevant to comprehend how speeches can help us understand historical events. The talk will also address the whole process implied in working with historical textual data (from digitization to visualization) and discuss how to approach its interpretation in a way that considers the human communication processes involved in their production.