Mapping Eileen Chang’s Novels with a Computational Analysis of Themes and Emotions

In the domain of digital humanities, the integration of computational methods with literary studies has opened new approaches to understanding and interpreting literature. This study applies Latent Dirichlet Allocation (LDA) topic modelling and sentiment analysis to eight English novels by Chinese-American author Eileen Chang. T he research uncovers seven main themes, reflecting Chang’s focus on women’s lives, cultural intersections, and societal changes in 1940s Shanghai and Hong Kong. The sentiment analysis of Chang's novels revealed Joy as the dominant positive emotion (19.43%) and Fear and Surprise as the most prominent negative emotion (17.21% each), challenging traditional interpretations of her work as predominantly melancholic. The study demonstrates both the potential and limitations of computational methods in literary analysis. While Natural Language Processing (NLP) techniques provide quantitative support and new perspectives, they sometimes fall short in capturing nuanced literary devices like irony and metaphor. Th e research advocates the convergence of the traditional literary approach and the burgeoning field of digital humanities to provide a comprehensive perspective. Keywords: NLP; Eileen Chang; topic modelling; sentiment analysis