Towards an AI narratology: the possibilities of LLM classification for the quantification of abstract narrative concepts in literary studies
C Jimmy Carroll · 2024
While narratology and computational literary criticism share a value for categorisation of formal textual features, the application of computational approaches to narratology has been limited. In this chapter, I argue that the reason for the under-utilisation of computational approaches in narratology has been due to the implicit nature of most narratological concepts, which meant that they could not be effectively operationalised using pre-set textual features. I then outline how Large Language Model (LLM) classifiers offer significant innovations in the realm of text analysis and classification that make them considerably more accurate when identifying elements of narrative that are implicit, rather than explicit, in text. In the second half of the chapter, I demonstrate these increased accuracy rates using results from initial testing of a classifier in the domain of cognitive narratology. This chapter ultimately argues that recent developments in AI, specifically via LLM classifiers, have dramatically expanded the possibilities for the application of computational approaches to narratology.