Introducing discourse-driven text mining: A novel method to critically analyse discourses on Twitter

This chapter responds to the numerous calls for new analytical methods especially designed for the study of digital communication. Digital media research has indeed highlighted the unsuitability of traditional analytical paradigms for the analysis of the complex relationships between discourse and digital practices. Informed by digital media theory and critical discourse theory, the chapter introduces a new analytical method: discourse-driven text mining. This method expands on discourse-driven topic modelling by merging critical discourse analysis and text mining. The chapter investigates social media discourses of social inclusion produced in Italy and tests the methodology on a corpus of Italian tweets from 2010 to the first half of 2020. The results proved the method effective in obtaining a ‘zoom-out’ perspective on social inclusion discourse studies in which works have been mostly qualitative in nature or limited to specific use cases. The combined analysis of topic modelling, bigrams, and discourse showed that two public macro-conceptualisations of social inclusion can be observed: 1) at the European level, social inclusion is predominantly associated with disability and 2) at the national level, social inclusion is associated with employability. The results also revealed that other pressing issues or vulnerable categories are missing.