Abstract

Genre is a classic problem in literary studies: How do critics and readers define genres? What genres exist? Which books belong to each one? This talk describes an unsupervised learning approach to mapping genre memberships in two large corpora of American novels published in the twentieth century. It offers a model of genre that incorporates topical, stylistic, geographic, and extra-textual features. The model and method can be used both to locate existing genres (detective stories, regional writing, science fiction, etc.) and to identify new, genre-like clusters of related texts. The talk concludes with a discussion of the generic properties of so-called "literary fiction," a high-prestige subset of contemporary literary production.

About the speaker

Matthew Wilkens, Associate Professor

University of Notre Dame, Dept. of English