Extracting Intersectional Stereotypes from Static and Contextualized Embeddings

Social group-based identities intersect. The meaning of “woman” is modulated by adding social class as in “rich woman” or “poor woman”. How does such intersectionality operate at-scale in everyday language? Which intersections dominate (are most frequent)? What qualities (positivity, competence, warmth) are ascribed to each intersection? Here, we make it possible to address such questions by developing a new stepwise procedure, Flexible Intersectional Stereotype Extraction (FISE), applied to word embeddings (GloVe; BERT) trained on billions of words of Internet text and report on original findings that emerged. First, applying FISE to occupation stereotypes across intersections of gender, race, and class showed alignment with ground-truth data on occupation demographics, providing initial validation. Second, applying FISE to trait adjectives showed strong androcentrism (Men) and ethnocentrism (White) in commanding everyday language (e.g., White Men are associated with 59% of traits; Black Women with 5%). Associated traits also revealed intersectional differences: advantaged intersectional groups, especially intersections involving Rich, were associated with traits that are more common, positive, warm, competent, and dominant. Together, the new empirical insights from FISE illustrate its utility for transparently and efficiently quantifying intersectional stereotypes in existing large text corpora, with the potential to expand the scope of research on intersectionality across time, place, and demographic variation. This project further sets up the infrastructure necessary to pursue new research on the emergent properties of intersectional identities.