The Paradox of the Text and Computational Criticism

I've been thinking about the text, especially the standard spatial metaphors literary critics use in talking about them, so I'm bumping this to the top of the queue. I'm working on another post on the topic.

Computational criticism, in Franco Moretti’s well-known terminology, is said to be “distant” reading. And there is an obvious sense in which it consistently deals with the text while so-called “close” reading often elides the text. It all depends on what we mean by “the text.”

There are situations where we mean the physical signs on the page, or even the pages themselves. But these situations are generally circumscribed. First of all there is old school textual criticism, concerned with the editing and preparation of texts. Then there are theoretical discussions about signs. More recently we have the history of the book. Practical criticism may deal directly with the signs on the page, or with the sounds of words (as in working out the rhyme scheme of a poem), but more often than not “looks through” the signs to their (various and possible) meanings.

Most importantly, in discussions of meaning and interpretation, the text is in fact a ghostly object – not the signifiers on the page, much less the pages themselves – which we apprehend through vague spatial schematization. In this schematization the text has an inside, an outside, and a surface. A major discussion has taken place around the question of whether or not the “meaning” of the text is determined only by items inside the text or items outside the text (such as authors and social context). Moreover the interior of the text is often presumed to have some kind of structure such that some meanings can be hidden within it. In contrast, other meanings are said to be on the surface of the text.

The general nature of the space that is articulated through these terms – inside, outside, surface, hidden – is unspecified. It is thus not clear where one must be in this space, or even how one can be in it, to render a so-called close reading. That a reading is close is generally indicated by extensive quotation of the text. Not only is such quotation lacking in so-called distant reading, but individual texts may not even be distinguished from one another.

And yet where the process of distant reading is mediated by digital computation, it is the text itself, the text understood as a collection of signifiers, that enters into the process. That is what I mean by the paradox of the text: computational criticism takes the text into account much more directly than does so-called close reading. The spatial schematization that guides our thinking about close reading is replaced by computational processes that operate directly with and on digital representations of physical signs (and the process by which those digital representations are created is a transparent one).

The critic, of course, supervises this digital processing. And the critic must select and interpret the visualizations of that processing. But that’s quite different from interpreting quoted passages. And the process of creating those visualizations is open to examination in a way that the selection of passages to quote is not. In “close” reading the process by which we get from the signs to meanings is opaque. By contrast, in “distant” reading that process is, at least in some respect, explicit and open to examination.

I note, finally, that this post is subject to deconstructive critique, but that critique will not erase the point that signs play a different role in computational criticism than they place in the many varieties of close and not-so-close reading.

* * * * *

Also see this older post, Commensurability, Meaning, and Digital Criticism.