Crossing the Gender Binary: Modelling Gendered Reference in Science Fiction and General Fiction

Title: Crossing the Gender Binary: Modelling Gendered Reference in Science Fiction and General Fiction Source: arxiv RoBERTa-based substitution analysis reveals that science fiction exhibits higher rates of gender-neutral term displacement compared to general fiction corpora. When masking gendered pronouns, the model predicts masculine substitutes twice as often as feminine ones for masked female subjects, though science fiction shows statistically greater linguistic fluidity across the binary.