Abstract

Does successful literature have certain properties? This question is highly contested in research, between the position that reader appreciation is subjective and context-dependent, and a work-intrinsic position, i.e., that readers react to work-internal features that are relatively stable across time and space.  Over the last three years, the Fabula-NET project has focused on exploring both extrinsic and intrinsic features to gauge the reception of narrative texts. In this talk, we will be presenting on the evolution of the project, starting with our initial focus on a balance between unpredictability and predictability in the reading experience, which we measured via the dynamics of texts’ sentiment arcs.

While sentiment arcs are relatively coarse representations – a simplification of diverse narrative emotions – our findings indicate that in simpler domains, such as in fairy tales, there is a linear relationship between sentiment arcs' self-similarity and their reception. In contrast, more complex and diverse texts, like Nobel prize-winning literature, exhibit a "sweet spot" between predictability and unpredictability. While points of balance vary depending on the text type and reception dimension or quality proxy chosen, controlled experiments reveal significant relationships between sentiment arcs' self-similarity and their appreciation across different audiences. 
Moreover, our research also underscores that the text which sentiment arcs are conveyed through affects novels’ reception. We show that combining deep narrative and sentiment features with surface-level stylistic features – such as how ‘readable’ a text is – enhances the prediction of a book's reception, with semantic categories providing additional insights.

So while the coherence of sentiment arcs varies among different appraisal groups, the interplay between stylistic and semantic levels in narrative novels significantly impacts their reception. We observe that different quality proxies tend to strike a balance of affective and linguistic unpredictability, reflecting, perhaps, a certain literary way of optimizing communication strategies with readers.

About the speakers

Pascale Feldkamp Moreira, Research Assistant, Center for Humanities Computing, Aarhus University

Yuri Bizzoni, Postdoc, Center for Humanities Computing, Aarhus University

Free of charge - All are welcome