Dynamical systems, literary theory, and the computational modelling of narrative

Abstract

Narrative can be conceptualised as a complex system, with pragmatic benefits for our description of interpretative and affective processes involved in reading. In this article, I show how systemic thinking has shaped many literary theories, and I offer some suggestions about the modelling of narrative as a complex system. Considering the relations between text, reader, and reading situation as constitutive of narrative requires a way of looking at stories keeping in mind that the reader's experiential background, their cognitive-affective states, and the situational context all play a crucial role in the emergence of what we call a narrative. Here I suggest that Transformer-based language models can be considered artificial systems that simulate a (disembodied) reading process and make it possible to model narrative as a complex system, including the encoded experiential background of the (simulated) reader.