Last Friday, I attended Lisa Marie Rhody’s talk “Reading Against Models: Feminist Text Analysis in Theory & Practice” at UVA’s Scholars’ Lab. Unlike the majority of other attendees, I was unfamiliar with her work. To be completely honest, I knew close to nothing about her scholarship, having only skimmed her article “Why I Dig: Feminist Approaches to Text Analysis” on Debates in the Digital Humanities 2016. This weekend, amid house chores, groceries, and cooking, my mind kept returning to her talk 1.

The reason I liked her presentation is because at the end, I did not only get to know a little bit more about her digital humanities scholarship and pedagogy, I also had an idea about who she is as a person and what she stands for. Lisa Marie Rhody borrowed Sara Ahmed’s moniker “feminist killjoy” to explain her activist approach to digital humanities. A “feminist killjoy,” according to Sara Ahmed, is a person of any gender who is willing to resist and denounce injustices perpetrated by dominant systems, even if that means to be unpleasant. Obviously, that explains the word “against” in the title of her talk. But what are the dominant systems that we should be wary of? And how the study of literature in concert with more technical and statistical approaches to texts can bring them at least to our attention?

Following her “feminist killjoy” approach, Lisa Marie Rhody guided us through her practice and teaching of digital humanities. Her expertise lies in topic modeling, a text mining technique to discover clusters of words that form topics, or patterns, in more or less large textual corpora. In her research, however, she has shown that using algorithms without questioning their assumptions first, only lead to somewhat flattened statistical information that does not unsettle fixed categories. In that regard, it is the idea of distant reading that needs rethinking and repositioning as a metaphor, because algorithms parsing words is not reading in the sense of deciphering. As Johanna Drucker puts it in her article “Why distant reading isn’t”:.

Processing is not reading. It is literal, automatic, and repetitive. Reading is ideational, hermeneutic, generative, and productive. Processing strives for accuracy, reading for leniency or transformation.

That doesn’t take away any of the merits of “distant reading” as a form of not-reading, of course, when applied to huge textual collections in order to detect dormant signals. But we are at a moment in digital humanities where “distant reading” is no longer “one of the hottest trends.”

On the contrary, because of the hot debates in Digital Humanities about “distant reading” we can better understand the stakes of machine learning and AI and how they increasingly affect our lives, our minds, and our identities. Given that all our online interactions are text-based, when our inputs are saved and analyzed by big tech companies, the world that they produce is not some utopian futuristic dreamland, but a simplified construction based on easy assumptions and binary logic. Gender is a complicated question Lisa Marie Rhody reminded us. But research based solely on statistical information from people’s language use on Facebook lead even scientific research to reproduce stereotypical gender representations.

The question then becomes to what extent autofill technologies that finish our sentences, trying to predict our minds, are fed with these types of research and how by succumbing to their convenience, we give up our agency of being ourselves, however messy and complicated. Is autofill a form of distance writing, with distance meaning an unexamined performance of gender roles that have already been predicted to be statistically accurate?

Footnotes:

  1. As a matter of fact, Googling her Saturday morning, not only did I reread more carefully her aforementioned article, but I came across a reference to her work in Martin Paul Eve’s very recent Close Reading with Computers, a book that instantly intrigued me and that I couldn’t help but borrowing from UVA library. Martin Paul Eve departs from macroscopic views of massive literary corpora and turns “distant reading” on its head. He applies a set of similar computational techniques to get as close as possible to the one and only book of his corpus, David Mitchell’s Cloud Atlas. To borrow his words, his method is a “close-textual digital microscopy” and the reason why it resonates with me is because he points his microscopic lens to contemporary literature. But I digress. Let’s go back to Lisa Marie Rhody’s talk. ↩