Middle Distance Reading

In on a Sunday to, among other things, layer some more data into the Shakespeare database that runs the intra-corpus inter-textual association “game” that I’m calling Ariel for the time being. I had expected a process I started Friday evening to be completed by now, but it isn’t. I probably should have run it on the cluster.

In the database’s previous iteration, I found line-to-line links through co-citations in the JSTOR Shakespeare match data set. But when I did that, I filtered out a bunch of those links on the front end because they wouldn’t be relevant to the way that the test version of the game was going to look and run. Slim down the database, save myself some time, and speed up the game easily.

But now, a programmer colleague of mine is going to assist, by building a much more robust interface. This will allow us to make use of much more data, so I’ve decided to stop filtering out co-citational links out. It’s already much, much larger (which is why it’s taking so long to build even with Condor’s help!). Again, I probably should have run it on the cluster, and probably will if I can carve out time to restructure the job in the next couple weeks.

I’m excited about building this out, as a model project for what another colleague and I have been privately calling “middle distance reading.” A couple articles use variants of the term to describe 1) the computer-aided qualitative methods of anthropologists and social scientists, or 2) the method of productively abstracting from close readings to interface with context. The approach we’re trying here goes in a somewhat different direction, but I suspect it’s motivated by similar concerns.

The idea, essentially, is to re-center reading and to sideline the vestigial positivism of what we call the computational humanities, while experimenting with the potentialities of computational processes for this act of reading. The motivating humanistic objection being: too many graphs and statistics, not enough reading. The motivating computational objection being: these data sets are richer than we’re giving them credit for, and can be leveraged in some very useful (and–which is a precondition–interesting) ways for humanists.