Racism is communicated and reinforced through language, but it is also challenged. The analysis of text data is therefore central to understanding how social inequalities arise and are reproduced. At the same time, measuring racism presents a particular challenge: as a socially contested concept, racism is shaped by social power relations, including within academia. Against this backdrop, the dissertation project asks: How has, does, and can racism be measured in text data using digital methods? How do biases in knowledge production as well as in AI models affect such measurements? How can these methods be improved for future research? The dissertation brings together five studies that explore the opportunities and challenges of computational approaches to measuring racism. One application study examines the presence and contexts of racist language in media reporting over the past twenty years.
Overall, the project provides a critical yet constructive contribution to how computational methods can be used responsibly in the social scientific study of racism.