Craig Dworkin's Parse (Atelos, 2008).
From the note at the back of the book:
Parse is a translation of Edwin A. Abbott's How To Parse: An Attempt to Apply the Principles of Scholarship to English Grammar. First published in 1874, the book played a leading role in the pedagogic debate over whether English should be analyzed as if it were Latin, and thousands of copies were printed as textbooks in the last quarter of the 19th century.
When I first came across the book, I was reminded of a confession by Gertrude Stein (another product of 1874): "I really do not know that anything has ever been more exciting than diagramming sentences." And so, of course, I parsed Abbott's book into its own idiosyncratic system of analysis.
What this means is that the bulk of the book consists of Abbott's sentences converted into descriptions of their own grammatical structure. Almost all the original content has been replaced by self-reflexive language like this:
Cardinal Roman Numeral period Preposition Noun parenthesis cardinal arabic numeral parenthesis colon dash
(I believe the capitalized words correspond to those that are capitalized in Abbott's text, but I'm not sure.)
At nearly 300 pages, even the most diehard conceptualist might balk at the prospect of actually reading Parse front to back, and in fact Kenny Goldsmith has used it as an example of conceptual texts that don't actually need to be read: the idea is enough. As appealing as I find this notion in many ways, I don't ultimately find it fully adequate to an assessment of Dworkin's work (or Goldsmith's, for that matter). What I think books like this do is ask us to reconsider what it means to read, and to find ways of actualizing new reading practices. This might mean something as simple as flipping around here and there throughout the book rather than reading straight through. It might mean reading one or more sections in an intense state of attention, and generalizing outward from such readings to a larger engagement with the total work. It might mean submitting for extended periods of time to the monotony of the governing structure, so that when there is some kind of variation in the pattern, it takes on an added value of surprise, as when Dworkin occasionally retains an entire phrase or sentence from Abbott without "translating" it, often creating the effect of editorial comment ("plural first person subjective case pronoun used in bad faith to suggest a camaraderie with the reader auxiliary verb adverb" etc.).
The truth is that the more one looks into Parse, the more one discovers in it--not "depths" in the familiar sense, necessarily, but details and complexities that activate underused and underappreciated areas of the intelligence. Highly recommended.