FoDiRa-Project: Regionalization of racist and discriminatory discourses on the social web

Racism is a social practice that (re)produces the exclusion and devaluation of certain socially constructed collectives (Fields & Fields 2012) and is often anchored in the center of society (Zick et al. 2016). Forms of racism differ between social groups and can therefore sometimes also differ significantly from region to region.

Until now, there has been little research that focuses on the small-scale regionalization of racism in the German-language social web. The subproject presented here aims to quantify various forms of racism regionally and to explain regional differences through subsequent analyses with the addition of further contextual indicators.

To this end, the project team is constructing a regional racism monitor that draws on text data obtained from social media platforms and websites of regional newspapers. Both data sources are “digital trace data”, which depict traces of real human behavior in digital space (Lazer et al. 2009). Social media data can be assigned to regional entities via geotagging or information from corresponding user profiles and subsequent geocoding, while newspapers can be geographically located via their regional reference.

Overall, the dissemination of racist content on social media platforms, but also on other digital media such as online news portals, has become increasingly important in the international research landscape in recent years (cf. Sigurbergsson & Derczynski 2019; for contributions on Germany, cf. Jaki, Sylvia, and Tom de Smedt 2018, Darius & Stephany 2019). At the same time, the spread of racist and discriminatory discourses can be understood as a warning indicator for processes of political polarization and radicalization. With an analysis of racism online, a comprehensive picture of racism in Germany can therefore be drawn, as racism is also carried from many parts of society to the internet and especially the social web.

The focus of the project is on revealing latent structures, which is realized by using methods from the field of unsupervised learning (word embedding models, various clustering procedures). These make it possible to measure and differentiate different racisms within a large amount of data (e.g., social media data from a specific region). Subsequent analyses will employ well-proven and frequently used social science statistical models (generalized linear models, multi-level models), to examine whether and to what extent regional differences between racisms in the network can be explained.

Research questions

  • What forms of racist stereotypes can be found regionally on the social web?
  • Can regional variation explain the extent of certain racisms on the web through context effects (inequality, demographic differences, etc.)?
  • What would a racism monitor based on social media platforms and regional newspaper articles look like?

Scientists involved in the project

Project management

Staff members

Student staff

  • Sophia Heuer
  • Daniela Wolf

Contact

Stefan Knauff 
Research assistant in the field of empirical social research with a focus on quantitative methods at the University of Bielefeld.

stefan.knauff(at)uni-bielefeld.de