There seems to be a consensus in populism studies around the ideational approach although there are still discrepancies on the conceptual genus of populism (thin-ideology, discourse, communication style). Notwithstanding this consensus, multiple voices warn about an excess of description (quantifying and comparing populist ideas or discourse) and, more generally, about an analytical deadlock in the field of populism studies. As observed by De Cleen and Glynos (2021), the consistency and comparability of the concept of populist has been achieved, at least in part, by abandoning complex and context-bounded explorations of populist politics.

This study aims to be a contribution to remedy this analytical impasse. First, it identifies the central problem in many studies under the ideational approach rubric. This paper argues that populism is generally conceived as a “thing” or “attribute” that can be possessed by political and social actors and endowed with a more or less permanent essence. This explains the way in which populist attitudes are studied as an attribute distributed among voters and a predictor of support for populist parties. The main argument contends that a more dynamic and interactive analytical framework is necessary to show the contingent and fragile nature of populist discourses and complement the current research agenda in populism studies.

In this last vein, I analyze two cases to scrutinize how the populist discourse varies across time for the same populist actor and to what extent this responds to changing socioeconomic and political contexts. I collect manifestos and campaign speeches of Podemos in Spain and Alternative für Deutschland in Germany between 2013 and 2017, to explore both left-wing and right-wing subtypes of populist parties. I combine a thick description of the cases with a corpus assisted discourse analysis of people-centrism and anti-elitism, the two core elements that allow us to measure change in populist discourses. The study shows that populism mutable nature can be related to two main factors: social protest and actor configuration. I argue that these two dimensions should be explored in their interactions with populist discourse or populist ideas to accurately capture populist politics. Finally, we draw several theoretical implications for future research on populism and adjacent topics.