Assembling the People, Defining the Enemy: A Discursive Cartography of Mexican Populism Using Word Embeddings

• In Event: Political Sociology Section Roundtables
  In Refereed Roundtable (60 minute): Measuring Politics (Table 5)

Abstract

This paper examines how populism constructs its antagonists not as a fixed social category but as a shifting set of actors linked through variable relational contrasts. Using the case of Andrés Manuel López Obrador (AMLO) in Mexico, it analyzes a prominent instance of left-wing populism and shows that its antagonisms extend well beyond economic elites. Drawing on a corpus of more than 10 million words from AMLO’s presidential press conferences, the study uses BETO transformer embeddings and semantic axis analysis to map how opposition is assembled across four conceptual dimensions: morality, wealth, power, and quantity. The results show that actors become “enemies of the people” when they occupy similar positions at the extreme ends of these axes, even when they differ in their social roles or institutional locations. Criminal organizations, usurping political actors, exclusionary publics, and ostentatious consumers emerge as antagonists alongside traditional elites because they share comparable moral and relational profiles in the semantic space of AMLO’s discourse. The paper contributes by (1) demonstrating that left-wing populist antagonism is a variable configuration of positions rather than a stable people-versus-elite divide; (2) clarifying how left-wing populism incorporates multiple kinds of enemies through semantic linkages rather than categorical definitions or class positions; and (3) developing a computational hermeneutic approach that models the geometry of political discourse using transformer-based embeddings.

Author

• Victor Patricio Verde Neri, University of California-San Diego