Work Product

This past semester I taught a grad seminar on digital humanities, one with more technical content than has been the case in my previous (undergrad) DH classes. On the whole, it went shockingly well; my students came in with very little background in either programming or statistical or quantitative analysis, and they left with enough of each of those things to do genuinely interesting work on their final projects. More importantly, they now know enough to go much further in the future, which many of them are already promising/threatening to do. I’m very pleased.

A few thoughts on specific aspects of the class and the syllabus. (NB. I posted an initial syllabus back in September, but it had some holes toward the end of the semester; the final, complete version (PDF) is now available.)

I was especially pleased with the response to the weekly problem sets, which were difficult and time consuming. The students ended up mostly working together in study groups, often meeting on campus over the weekend to finish the exercises for Monday’s seminar. This was exactly what I was hoping would happen. There’s no getting around the fact that programming, like language learning, requires hours and days of hands-on practice. Problem sets are an odd form in the humanities and I know there were students who thought the exercises consumed too much time or required more groping for answers than they would have liked, but I think this part of the course worked exactly as designed. I’m glad everyone was willing and able to struggle productively with them. On a semi-related front, grad school can be an isolating experience; one of the very few things I missed about chemistry when I moved to an English PhD program was the camaraderie and feedback I got from group work in the sciences. The problem sets were an opportunity to bring more collaborative structure to an English PhD program.

The biggest problem we faced was lack of time. This is a course that wants to be an intro to programming, an intro to statistics, a survey of recent work in DH (broadly defined), a theoretical treatment of digital media, a chance to think about the future of the discipline, and a grad-level seminar on nineteenth-century American fiction (nineteenth century rather than twentieth due to corpus constraints). We spent two weeks at the beginning of the semester on media studies (McLuhan, Galloway); that was fun, but it was ultimately to the side of our main concerns. The time could have been more profitably spent on an extra week of intro programming concepts and another week later in the semester on advanced computational topics.

The exercises from The Programming Historian were invaluable and I recommend them to anyone teaching a similar course. But/and I’d add a week of more introductory proper CS concepts (branching, looping, variables, return values, etc.) before beginning them. This is true even though the PH exercises really do start from “Hello, world”; they then ramp up by way of very concrete examples, which the students sometimes found difficult to separate from the concepts those examples were meant to convey.

I’d spend two weeks (rather than one) on mapping and GIS, which was popular and useful for several final projects.

I’d also spend more time on visualization in general, maybe assigning Tufte’s book and all of Yau’s (rather than just one chapter). An additional merit of Yau’s book is that it would provide some intro to R, which the students said they’d like.

I’d assign all of Jockers’ Macroanalysis once it’s available. Many thanks to Matt for sharing a handful of draft chapters with us.

Not sure it’s worth reviewing in depth the merits of individual articles and chapters (of which we read many); by the time I teach this course again in a year or two, most of those will probably drop off in favor of newer results. This is both the joy and the frustration of teaching DH at the moment.