Teaching Literary History through Computational Analysis

Understanding literary history is a fundamental component of the literary studies curriculum (Summit). Traditionally, students have been introduced to the discipline from a historical perspective, often following a linear temporal progression that outlines literary periods, or simply using centuries and national or linguistic boundaries as markers. In many academic settings, such as in Germany, students of literary studies often also pursue teacher training. Consequently, they learn about literary history as future mediators of a constructed (national) literary tradition. However, as many critics have pointed out in recent decades, doing literary history poses certain challenges: Which texts are read and which are forgotten? Who becomes part of the canon, and how does this intersect with power relations? What kind of literary history are we teaching if we only address a very small part of the literary archive?

Outside of the increasingly popular but not yet widespread digital humanities programs, literary history is generally taught through the lens of hermeneutics, or, less frequently, social history. Complementing these, digital humanities and computational methods can offer promising solutions to questions about largescale phenomena over long periods of time. By working quantitatively with large data sets—some call them capta to emphasize their constructed nature (Drucker)—computational literary studies strive to ”explain or provide general laws of literature, and even of history and culture” (Bode, 14). 

While scalability in data analysis offers broader perspectives, it does not eliminate the need for narrativization. Any literary history inherently synthesizes information and reduces complexity. Expanding the data set accentuates potential issues related to data selection (Herrmann and Lauer). So the process of building a corpus doesn’t eliminate the challenges of hermeneutic or social-historical literary history. Rather, the practices of quantitative research tend to expose problems around canonicity, periodization, and intersectionality, by making the selection process transparent. The same is true of diagrammatic visualizations, which play a crucial role in quantitative analysis in explaining relationships and developments within literary history by providing visual frameworks for concepts of time and change, causality and continuity (BÅNorner et al.). Therefore, a key advantage of computational literary studies is its ability to  identify and address selection processes and blind spots (Herrmann et al.), while critically engaging with issues like availability, selection, bias, and canonicity in the narrativization of literary history (Underwood).

We discuss these questions using an interactive Open Educational Resource (OER) in data literary studies as a case study. This OER introduces concepts such as modeling, operationalization, corpus building, and various measures of quantitative analysis to students and teachers of literature who are not yet familiar with digital humanities methods. Through this example, we highlight how fostering data literacy in students equips them with the skills to critically engage with narrative constructions of literary history.