Given that I'm a text-mining skeptic, it is only fitting that I hope to pursue a small project with Aditi Muralidharan next semester.
For any humanistic question really worth asking, text-mining can never provide an adequate answer. But it can provide supplementary evidence, or provide part of an answer, or point one toward a way of reading. Aditi is working on an interface that will reduce the up-front cost (in time, etc.) of doing text-mining, which, to my mind, will make it more natural for scholars to use quantitative evidence without feeling pressure (due to massive time-investment) to make it the centerpiece of the argument.
I'm still persuaded that text-mining, or any operation that wrings data out of discourse, is an incitement to automatic writing, a way of forcing the body of the text to reveal an unconscious that it didn't know to keep secret. There's something unsporting about it--and something naïvely idealistic, too. The hidden is accorded special powers, its occultism its epistemic guarantee. (After all, what would be the point of using a computer to do something that could easily be done by hand? The whole point is that we're not experiencing it, not actually reading it.) Despite its apparent superficiality, text-mining is a hypnosis to close reading's talking cure.
If we must make our texts hysterical, then, what better questions to ask them than questions about gender circa 1900?
That's what we're going to do.