Abstract
Recent debate about the preservation of literary works that present outdated or disfavored views, has shown that literary representation is a highly controversial issue. Books influence to a significant extent how we perceive reality, ourselves, and others. Nevertheless, we are still missing a comprehensive overview of representation in Dutch-language children's literature. My research aims to fill that gap by answering the question which biases regarding age, gender, ethnicity, and social class can be detected in illustrations to Dutch children's literature from 1800 to 1940. I will pay particular attention to the intersectional nature of these biases, the interplay between visual and textual elements in its construction and how it is given shape at a surface level. For this research, I propose a computational, data-driven method to analyze bias in a corpus of 3000 illustrated children's books. Certain machine learning models are proven to implicitly model social biases, an issue which is rightfully criticized for its undesirable – and at times unethical - effects in downstream applications. This project will turn this deficiency to our advantage, using it as a unique research possibility. I will employ a combination of distant viewing and reading methods to uncover patterns of biased representation in the dataset. The results of this analysis will then serve as a touchstone for more contextualized close readings.
Researcher(s)
Research team(s)
Project type(s)