I enjoyed Miriam Posner’s keynote address at the Keystone Digital Humanities Conference. It was far from the only talk last week that was animated by a desire for justice and compassion, and it was good to see that desire given such prominence by the organizers and applauded by the attendees.
As a linguist I also welcomed Posner’s focus on categorization and language diversity. I was trained as a syntactician, but over the past several years I have paid more and more attention to semantics, and categorization in particular. Building on the work of Ludwig Wittgenstein, Eleanor Rosch, George Lakoff and Deborah Cameron, I have come to see categorization as a touchpoint for social justice.
I should note that for me, categorization is not just where I can advocate for others with less power. As a transgender person, the power to categorize myself, my feelings, my beliefs and my actions is denied to me on a daily basis. The main reason that I study categorization is to regain that power for myself and others.
Much as I share Posner’s passion for justice, her talk raised some concerns in my mind. First, digital humanities cannot bear the entire burden of social justice, and even language as a whole cannot. Second, categories are slippery and flexible, which is a great strength of humanity but also a great weakness. Third, there are limits to how much we can trust anyone, no matter how high or low they are in the hierarchies of power. These concerns are not insurmountable barriers to a radical approach to categorization, but keeping them in mind will help us to be more effective fighters for social justice.
I plan to address the issues of the burden and trust in future posts, and in this post focus on the slipperiness of categories. As I understood it, Posner drew a distinction between the data models used by digital humanists (among many others) to categorize the world, and the lived experience of the people who created and consume the data.
There is often conflict between the categories used in the model and in the experience, and there is often a power imbalance between the digital humanist and the humans whose data is being modeled. Digital humanists may be perpetuating injustice by imposing their data models on the lived experiences of others. Posner gave examples of binary gender forms, database fields for racial classification, and maps of places. She contrasted these models imposed from above with examples where humanists had contested those models, aiming to replace them with models closer to the lived experience of people with less power who had a stake in the categorization.
The problem is that our lived experience is also a data model. As George Lakoff and other cognitive scientists have shown, the categories that humans use to describe and interpret our experience are themselves conventions that are collectively negotiated and then imposed on all members of the language community, with penalties for non-compliance. They are just as distinct from reality as the fields in a SQL table, and they shape our perceptions of reality in the same ways.
Whether or not they are encoded in HTML forms, categories are always contested, and the degree to which they are contested is a function of what is at stake to be gained or lost from them. In my experience, categorizing people, whether by race, gender, nationality, religion or other criteria, is the most fraught, because these categories are often used as proxies for other factors, to grant or deny us access to valuable resources. The second most fraught is the categorization of place, because places contain resources and are often proxies for categorizing people. After that, food seems to be the most fraught; if you doubt me, ask your Facebook friends for the most (or least) authentic Mexican or Italian restaurant.
And yet, as Wittgenstein and Rosch have observed, it is normal for categories to have multiple, slightly different, meanings and memberships, not just in the same language community but in the same individual. In my observations it is possible for the same person to use two different senses of a word – a category with two different but overlapping memberships – on the same page, in the same paragraph, in the same sentence.
Responding to Posner’s talk, Matt Lincoln posted a recipe for using the Resource Description Framework (RDF) to describe overlapping, contrasting systems of categorization. I think that is an excellent start, particularly because he places the data models of lived experience on the same level as those imagined by the researchers. My word of caution would be to keep in mind that there is not one singular data model for the lived experience of a community, or even for an individual. As Whitman said, we contain multitudes. Each member of those contradicting multitudes has its own data model, and we should thus be prepared to give it its own entry in the RDF.
I and others brought up similar issues in the question and answer period, and in the reception after Posner’s keynote, and I very much appreciate her taking the time to discuss them. As I remember it, she acknowledged the challenges that I raised, and I look forward to us all working together to build a humane, compassionate humanities, whether digital or not. I will discuss the challenges of bearing the burden and of trusting the community in later posts.