Computational literary scholars have used many different data sources as proxies for measuring the popularity and/or prestige of individual authors and works of literature. However, many of these data sources apply to relatively limited contexts, such as professional literary study (e.g., The MLA International Bibliography, W. W. Norton & Co.’s anthologies of literature), or communities of habitual readers (e.g., Goodreads). What kinds of literary knowledge and valuation exist in US culture more broadly?
Jeopardy! frequently asks contestants questions about literature, and the relative presence or absence of authors and texts from Jeopardy! as compared to other data sources mentioned above says something about the extension of literary study’s disciplinary knowledge into the more general realm of trivia. This is not to say that Jeopardy! is some how representative of US culture more broadly, but rather that as a popular television show, it reflects aspects of literary knowledge not well captured by preexisting approaches.
By using large language models to classify the 405,874 clues asked over the seasons that Alex Trebek hosted (1984-2020) to identify those that pertain to literature and literary history, this project seeks to identify the pieces of literary knowledge most frequently cited on Jeopardy!
However, the project goes further than mere frequency: Because of the structure of the game and the data set, this process will also reveal how individual pieces of literary knowledge are priced. Questions asked in earlier rounds and at lower dollar values are presumed to be easier (i.e., more common knowledge) than questions asked at higher dollar values and in later rounds. I compare and contrast Jeopardy!‘s pricing of literary knowledge from other forms of consecration alluded to above to see how Jeopardy!‘s metrics map on to theories of and data about literary canonicity.