In the other course I’m taking this term– Paul Kameen’s core course in rhetoric which is focused on referential language– we’re currently reading through some essays and works from poets in the 1970s and 80s, namely from the L=A=N=G=U=A=G=E movement. It was/is a movement that foregrounds play, collage (material collage, literal cutting and pasting, as well as later digitized versions and other means of appropriating texts from other sources), strangeness, and a deliberate disruption and intervention into the typical modes of close reading. Above all, the poets from this movement were invested in troubling the relationship between reader and writer, in calling to attention the reader as an equal if not majority maker of a text. Such works resist “a reading”. In Lyn Hejinian’s exceptional essay “The Rejection of Closure,” she writes:
It is not hard to discover devices–structural devices–that may serve to ‘open’ a poetic text, depending on other elements in the work and by all means on the intention of the writer… the ‘open text,’ by definition, is open to the world and particularly to the reader. It invites participation, rejects the authority of the writer over the reader and thus, by analogy, the authority implicit in other (social, economic, cultural) hierarchies. It speaks for writing that is generative rather than directive. The write relinquishes total control and challenges authority as a principle and control as a motive. The ‘open text’ often emphasizes or foregrounds process, either the process of the original composition or of the subsequent compositions by readers, and thus resists the cultural tendencies that seek to identify and fix material, turn it into a product; that is, it resists reduction (88).
Reading Hejinian and then also Ramsay and Underwood and Sellers at the same time has me wondering, if algorithmic criticism serves to open up new critical readings across literary texts, what is the role of the “reader” of its outputs? On the one hand, the reader is maker. The reader/coder designs (or appropriates) an algorithm with which to sort through a mass of textual data, and determines what that algorithm is meant to do. It re-reads the data to assign and sort through tokens, making sure the human sense-making system of word combinations are represented by the data (e.g. “high school” becomes a bunch of “high” students at “school” if we’re not careful to assemble the singular term). The algorithm designer reads the world as much as she reads the data, but the design of the thing makes strange before it makes sense. The reader still has to make “value” of the repetitions. To ask what it means.
This version of reading strikes me as quite similar to the request Hejinian is putting on the reader to bring themselves to the text. Her process might be idiosyncratic and not automated in the same sense that a machine algorithm is, but her texts could read like a computer-generated recombinatory text. Hers is not a poetic that is based off of “natural language”. Take this moment from My Life:
The continent is greater than the continent. A river nets the peninsula. The garden rooster goes through the goldenrod. I watched a robin worming its way on the ridge, time on the uneven light ledge. There as in that’s their truck there. Where it rested in the weather where it rusted. As one would say, my friends, meaning no possession, and don’t harm my trees.
Hejinian verbs the noun “nets,” and describes the robin’s search for food as “worming,” and gives “time” a physical location (read, perhaps, the passing of the light by the shadow of the ledge tells us the time… but that is too many words and perhaps not what is “meant”). The play with “there” and “their” in sonic spheres. The weather as something that one can be “in” apart from what happens to us. This is not natural language in the same sense that other texts are, even the poetic works of the time period Underwood and Sellers are looking at. If we determined the frequency of light to be quite common it is still “unreadable” as a figure or trope, as only context determines these things and even then the context doesn’t save you with the Hejinian poem. The language would probably flag as not latinate and more common. There’s plenty of idiom if you could have the code look for such things, but that can in no way signal a desire for “universal experience”.
In short, while perhaps in the window that Underwood and Sellers are analyzing some kinds of “closure” is possible, but feels to me to be too limited in terms of their conclusions. In the sense of most of these algorithmic criticism machines, they seem to me to be mechanisms for opening up the text to the level that Hejinian hopes her works have from the outset, but then the critic’s aim is still to close them down again. Perhaps Hejinian would say that these are new technologies at all, and that it’s actually a responsibility of the writer (or algorithm designer and executer) to leave a text open to such participation from the reader. Sense-making is not unlocked by the texts themselves but by the interaction between a text or texts and its reader. This is not a new concept. Of course we all participate in texts as readers, but as with so much we’re thinking about in this class, it’s a matter of scale. The Hejinian poem demands much more of us. These algorithmic critical programs demand much more of us. How, then, do we amplify the openness and work with the results of these programs without succumbing to reading the output through critical modes that “seek to identify and fix material, turn it into a product”?