These two projects use machine learning technology to examine gender bias in language. The first project is a text classification model fine-tuned from the anti-trans legislation that is currently proliferating across the US. From this legislation, which I scraped and processed from government records, I extract definitions for gender and related terms, with the goal of scoring language about gender on a scale from affirming to non-affirming. The second project is a text generation model, the "homo-generator," fine-tuned on the Homosaurus vocabulary.
Both projects are still in development, and you may learn more about them on Github.
This online workshop series teaches Python programming from an ethical and feminist approach to working with text data. It begins with foundations of programming and proceeds to data gathering (with web scraping), and text analysis. Future units are currently being developed on text generation with machine learning.
My dissertation, "Since No Expressions Do": Queer Tools for Studying Literature explores how digital methods and tools for studying text engage with queer literature via Queer Studies frameworks. I critique digital methods and tools by exploring how computation, which disambiguates and fixes data for electronic processing, might be used to analyze the complexity of queerness expressed in textual style, form, and voice. Download the dissertation here.
Below are two small digital projects that demonstrate in practice how one might digital tools to work within Queer Studies frameworks.
This project draws connections between programming logics and gender theory to propose a text analysis methodology that iterates through distant and close reading. Using Virginia Woolf's novel, Orlando: A Biography (1928), as a test case, I demonstrate how this method of text analysis leads from a binary understanding of gender into a plurality of gender significations in the novel, suggesting how language and gender are mutually constructed.
See the digital component of this project on the "Queer Distant Reading" Github repository.
This project uses the Text Encoding Initiative (TEI) standard, an electronic editing tool, to encode the homoerotic elements that Oscar Wilde edited while composing his novel, The Picture of Dorian Gray. (1890). It explores the mutually reinforcing nature of TEI's hierarchical structure and of dominance structures in archival data and practices. See a customized rendering of the manuscript's first chapter and access the XML/TEI files .