Text-as-data methods are a broad set of techniques and approaches relying on the automated or semi-automated analysis of text. They have become increasingly prevalent in the social sciences, and are part of a broader trend in which, taken together, the internet and computational social science tools have changed the kinds of questions that social scientists can ask and answer successfully (Golder and Macy, 2014; Lazer and Radford, 2017). Text analysis holds a prominent place in these developments. Texts have always been a primary data source for social scientists. As MonroeandSchrodt (2008, 351) write,“textisarguablythemostpervasive—andcertainlythe most persistent—artifact of political behavior.” In the internet age, texts have become particularly plentiful, and accessible with relative ease. The large amount of text available to researchers, combined with new computational tools, have promoted the development of text-as-data approaches in which texts are analyzed statistically with different degrees of automatization. The promise of the approach is that it can both apply existing theories to new data and uncover new phenomena that previously remained hidden (Evans and Aceves, 2016). As González-Bailón (2017, xviii) writes:“when the right connections are made, much of the data-driven research that is being conducted today speaks directly to long-standing (and unresolved) theoretical discussions.” Text-as-data approaches are becoming mainstream in political science. Typical applications revolve around research question where at least one element is based on political communication theories such as agenda setting, issue definition, or framing (for reviews, see Grimmer and Stewart, 2013; Lucas et al., 2015; Wilkerson and Casas, 2017). From a practical perspective, these approaches allow researchers to conduct more efficiently research they have been doing