A toolkit for detecting racist language

The book chapter first clarifies key concepts such as race, racialization, and racism, situating them within a framework of social-scientific theories on racism. It then presents a concrete analytical toolkit for examining how linguistic choices can normalize exclusion and legitimize the differential treatment of groups—such as those referred to as “asylum seekers,” “refugees,” or “foreigners.” Among other things, it is important to pay attention to how people are categorized, whether differences between groups are continually emphasized, who gets to speak, whether certain people are described in derogatory ways, and whether they are systematically associated with what is negative.

“Public debate influences how people are treated. When certain groups are consistently linked to threats, costs, or criminality, it affects both political decisions and people’s everyday lives,” says Kristina Boréus, Professor of Political Science at the Institute for Housing and Urban Research (IBF).

The chapter demonstrates methods for mapping racializing and racist discourse in practice. It also discusses the possibilities and risks of large-scale text analysis. With AI and big data, it is possible to analyze enormous volumes of text and identify patterns that were previously hard to detect, but there are also important methodological challenges associated with such techniques. AI language models trained on large datasets from the internet must also be examined critically: when they are trained on text corpora that contain many racist texts, they too begin to “think” in racist ways.

“We need both better methods for detecting how racism takes shape in language, and more evidence about which counter-strategies actually work in different contexts,” says Kristina Boréus.

In the chapter’s concluding section, Kristina Boréus points to two directions for future research: more integrated analyses where linguistics, social science, and data science intersect—and more knowledge about how racist discourse can be countered.

“The chapter offers an entry point for anyone interested in how racism is expressed through language. The analytical toolkit helps readers see what is often hidden—how what is said, and what is implied, draws the boundaries of who is considered to belong, and on what terms,” Kristina Boréus concludes.