Using Riveter to map gendered power dynamics in Hades/Persephone fan fiction

Abstract

I examine the affordances and limitations of Riveter, an NLP-pipeline, for analyzing dynamics of power and agency in fan fiction. Riveter detects entities in a text, links together references to the same entity, parses grammatical relationship between entities, and assigns entities lexicon-based scores reflecting their levels of power and agency. I take the gendered dynamics of the Hades/Persephone relationship from Greek mythology fandom as a case study, using a corpus of fan fiction from Archive of Our Own. I also use Riveter to assign aggregate power and agency scores to all entities referred to with the same gendered pronouns in fan fiction. I evaluate Riveter's performance by manually annotating a single fan fiction text. The tool overlooks some of the implicit ways characters negotiate power and misses many of the ways power and agency are represented, reconfigured, and negotiated. However, because of its ease and transparency of use and its capacity to examine textual relations at a large scale, Riveter is useful for mapping (gendered) character dynamics in fan fiction, especially when combined with more fine-grained methods such as selected close readings.