“It’s the market, stupid”. Computational approaches to digital neoliberalism: a case study

The research reported in this paper attempts an empirical analysis of large textual data with innovative computational tools that process and query texts within a highly debated field: neoliberalism and the financialization and marketisation of public life. We built a corpus of neoliberal texts comprising of articles and blog entries derived from the websites of well acknowledged think tanks of neoliberal orientation which allowed for a theory- driven detection of ideological and linguistic markers as well as a data- driven approach of feature extraction and domain specific concepts. In our approach we operationalized linguistic and discursive devices of the market ideology and its persuasive techniques such as metaphors, repetitions, and the detection of presuppositional triggers that signal assumptions and therefore commonsensical establishments. We experimented with distributional semantics to extract conceptual dictionaries and “seed” words. By implementing features and techniques from discourse analysis, rhetoric devices, corpus linguistics, results from previous qualitative research and the output of our machine learning algorithms we built a model that can effectively detect ideological assumptions and enabled the foundation of a novel pipeline for computer assisted research in political analysis and ideology.