I could teach Lorca but not work on him. He was too problematical and confusing. He was a problem that I couldn't get a handle on. Then I did work on him, with an unpublished paper on the Ode to Whitman, which will now be the core of a chapter once I reconstruct my argument. Then I gave a conference paper on Yerma once.
Then, finally, after many years, I decided that I would work on Lorca, but only through the medium of American poetry and culture. So I was evading the issue. Now I am seeming to confront it head on, but still not analyzing too many Lorca poems. Maybe at some point I will be a good Lorquista.
I want to create Lorca in a certain idealized way. I don't want to be too tied down by other people's preconception, though I seem quite attached to my own. In order to keep my Lorca purified, I have to represent him in a certain way as unrepresentable. The duende becomes an allegory of the unrepresentable itself in chapter 1, but then I go back and read it as a narrative of cultural exceptionalism in chapter 2, deliberately bracketing some of the complexities of the text in order to do so.
I think Paul de Man does make sense: a general theory of literature is often at odds with the results of this theory when applied to literature itself. This is the thesis of Blindness and Insight. He demonstrates this with varying degree of felicity in the essays of the book, but I'm not interested here in his examples but in the more general principle: I might think I'm doing one thing, but really I'm doing something else. Not only that, but that something else should be more interesting than what I think I'm doing.