This paper expands current dictionary approaches to measure populism as the latter often carry a ‘broad brush’ and are prone to conflating populism with empirically related concepts, e.g. nationalism. I argue that established theoretical assumptions, i.e. the moralizing notion of populism and the construction of the people as morally superior, should be considered more thoroughly when measuring populist discourse. Using different dictionaries for each core feature, i.e. people-centrism, anti-elitism, and popular sovereignty captures populism in a more appropriate manner. Moreover, I use dictionaries for explicitly non-populist discourse as counter-measurement, e.g. technocratic language. My dictionaries are based on theoretical considerations and an actor-based qualitative assessment. Comparing them to other automated and qualitative approaches of measuring populism, I show their validity using – among others - a text corpus of 207,700 speeches given in the European Parliament from 1999 to 2014. Identifying the key features of populism separately enhances conceptual clarity and allows for identifying clusters of parties based on how they use or combine those features differently in a comparative manner across time and countries.
