Our roundtable today loosely follows several themes within the larger question of what a feminist Text Analysis might look like for disciplines in the Humanities.

We begin with Atilio Barreda’s argument for a more strategic application of transfer learning as a model of a self-consciously feminist textual analysis that would recognize and account for the situated and contingent status of Machine Learning.  

Similarly, Zico Abhi Dey, in his recent examination of Open AI, implies that open-source language models with their shift in scale, and attention to computational costs over efficiency, might provide a feminist alternative to flawed Large Language Models.

Our other panelists, Livia Clarete, Elliot Suhr, and Miaoling Xue, bring a much-needed multi-lingual perspective to current Text Analysis and the applications of Feminist Data principles. 

Clarete examines communication styles in English- and Portuguese-language healthcare systems and asks how feminist notions of care and power-relations might inform our use of linguistic analysis and corpus analysis studies. 

Suhr explores how the biases in data collection, language models, and algorithmic functions can exacerbate disproportions of power in dominant and minoritized languages and suggests that an intersectional feminist framework is essential to unpacking these issues. 

Xue approaches another aspect of Humanistic computing—the construction of the historical past—by looking specifically at what is lost and what can be gained by applying current Western feminist models of digital archival reconstruction when approaching a corpus that differs in space, time, and significantly language. She concludes by considering the implications for representations of women in narrative history and the occluded labor of women in the production of texts, particularly in the so-called “invisible” work of editorial notation and translation.

Taken together, these discussions animate and ground what Sara Ahmed calls “the scene of feminist instruction” which she identifies as “hear[ing] histories in words; . . . reassembl[ing] histories in words . . . attending to the same words across different contexts” (Ahmed 2016) and which could equally be a description of a responsible and informed feminist text analysis itself.

Participants: Atilio Barreda, Bianca Calabresi, Livia Clarete, Zico Abhi Dey, Elliot Suhr, Miaoling Xue