Extra! Computational Criticism Breaks with Tradition! Sky Remains Overhead

Andrew Goldstone has decided that perhaps so-called distant reading is not reading at all. The document is relatively short, “The Doxa of Reading” (PDF), and will appear in PMLA for May 2017. Goldstone takes “doxa” from Pierre Bourdieu (whom I’ve not read) who regards it “as what participants in the literary field take for granted.” What critics take for granted is “the assumption that the primary activity of academic literary study is textual interpretation.” Though Goldstone doesn’t say so, this wasn’t always the case, though it has been the case for the last five or six decades.

Bourdieu also allows for orthodoxy and heterodoxy, which stand in opposition to rupture. In Goldstone’s analysis–I refuse to call it a reading–the term “distant reading” functions as a heterodox form of reading and does so, in effect, to ward off the realization that it is not a form of reading at all but a rupture from reading. Goldstone’s analysis is both subtle and complex, just enough so that I’m not sure where he stands. But his penultimate sentence seems clear enough: “If we cease to regard the different versions of distant reading as a singular project [...] we gain a wider sense of the possibilities for scholarship.” I take that to mean in some version distant reading computational criticism really does point beyond “reading” and so to new modes of literary investigation.

In the course of his argument Goldstone considers the visual objects that have become a signal feature of so much work in computational criticism:

Looking back on the work of the Literary Lab in a recent pamphlet, Moretti asserts:

Images come first, in our pamphlets, because–by visualizing empirical findings–they constitute the specific object of study of computational criticism; they are our “text”; the counterpart to what a well-defined excerpt is to close reading. (“Literature, Measured” 3)

As an account of a quantitative methodology, this is a strange statement: visualizations are powerful exploratory tools, but they should be considered provisional summaries of the data of literary history, not primary objects. As a description of argumentative rhetoric, however, Moretti’s analogy between visualization and excerpt is illuminating. It positions the “computational critic” as an expert interpreter of visual texts, a heterodox version of the close reader.

He is correct. Those visual objects are not the “primary objects” of investigation. Yes, they are “exploratory tools”.

It seems to me that they have the character of observations (of the primary object of investigation). Think of the images created in radio astronomy; they make look rather like optical photographs, but they aren’t. The process through which they are constructed is different. The visual objects of computational criticism don’t look like photographs; they look like the charts and graphs that are ubiquitous in so many quantitative disciplines. But they have the same status in the intellectual process as those images of radio astronomy, that of observations.

It is through observations that a phenomenon of interest enters the investigative field. Observations as well have something of a descriptive character. Wouldn’t you know, description has recently become of interest in discussions of critical method and practice, along with “surface reading”.

And with that I want to turn to the aborted structuralist moment in the history of literary criticism. Jonathan Culler published Structuralist Poetics in 1975. In his preface he observed (xiv-xv):

The type of literary study which structuralism helps one to envisage would not be primarily interpretive; it would not offer a method which, when applied to literary works, produced new and hitherto unexpected meanings. Rather than a criticism which discovers or assigns meanings, it would be a poetics which strives to define the conditions of meaning.

Culler himself never followed up on the poetics he proposed, nor did others take him up on it. To be sure, we have narratology, a poetics of the narrative. But it is hardly central to the discipline as it has been practiced over the past half-century or so.

A turn to poetics would have been a rupture from emerging interpretive practices (aka reading). But then those practices themselves constituted a rupture from the practice of a discipline rooted in philology and (traditional) literary history. While interpretive criticism has its roots before World War II, it isn’t until after the war that it became the center of critical activity. As that happened interpretive practice became the focus of scrutiny, scrutiny that led, among other things, to the (in)famous structuralism conference that took place at Johns Hopkins in 1966. But the rupture that actually happened, was not a turn to poetics, but a turn to deconstruction. And if deconstruction was noisy and obstreperous at the time, in retrospect it is clear that it was not a rupture at all, but just a further variation on interpretive reading.

It remains to be seen whether or not computational criticism will flourish as a genuine rupture from interpretative reading, though not so much as a replacement as a supplement. If so, will it find common cause with a new poetics, one based on surface “reading”, description, and perhaps even the newer psychologies (cognitive, neuro-, and evolutionary)? Can we move beyond reading, not only in the analysis of large corpora, but in the analysis of single texts? And can we move beyond a meta language centered on notions of reading and space (close, distant, hidden, surface) to one based on the intellectual operations involved (among others, description, analysis, interpretation, explanation)?

* * * * *

Of course I've addressed these issues endlessly here and in my working papers. For a start, see this post from 3 Quarks Daily, “The Only Game in Town: Digital Criticism Comes of Age”, and in numerous blog entries and working papers (e.g. these working papers on description).