---
title: "Computational reading as 'reading against the reader'"
layout: post
---
I'm probably not the first to think these thoughts, but I thought I would write them anyway as they are fresh in my mind. When dealing with computational reading methods, it is easy to encounter an aesthetic/teleological opposition to stylometry from some quarters. Indeed, among the most common questions that are asked by non-stylometrists about its processes are: “so what?”; “why should I care?”; and “what does this actually tell us that we didn’t already know?” In other words, when confronted with mathematical and computational processes for studying texts, the frequent response is to ask what it tells us about a work. The obvious retort is that it tells us neither more nor less than any other study of an aesthetic object; a work of literature. Indeed, the study of aesthetics is answerable to nothing except itself at some point in the chain; it is a human pursuit to understand how literary works achieve their affects and sometimes effects.

Yet, [as Ted Underwood put it to me in a statement that has haunted my thinking ever since](https://twitter.com/Ted_Underwood/status/756135378742943744), this challenge of purpose and teleology can be “understood as an aesthetic problem”. For literary criticism traditionally makes “fragments of individual experience work to illuminate a big picture” while stylometry takes unexperienced quantitative data to do the same, which feels like an “aesthetic loss”. In other words, there is something not-like-reading about stylometry and computational "reading" that disconcerts people outside of its practices, compounded by a fear held by many that the fund-able future of humanities research might compel them into this space against their wishes. Indeed, “distant reading”/computational "reading" is actually a non-consumptive use, to use the phrase from American copyright law. It is not actually a form of reading; it is a set of utilitarian techniques for evaluating large-scale corpora. At the same time, though, the common curse uttered by academics working on fiction is that they have “lost the ability to read for pleasure”. Indeed, traditional literary criticism coerces texts into new narrative forms conducive to argument, its practitioners reading to seek case studies suited for exegetic purpose. But we still call this reading.

I wonder, too, whether there is an aesthetic antagonism to literary criticism,as it has existed since the New Criticism and Poststructuralist anti-intentional criticism, in computational reading practices. Of course, there have always been archival, biographical, and other more seemingly material literary-critical practices. But one of the legacies of the New Criticism was to turn power to readers, away from authors. The poststructuralist "Death of the Author" extensions of such New Critical anti-intentionalist practices -- even if they might not have wished them to be billed as such as "extension" -- only strengthened such readerly-centric approaches. It was empowering as a reader to be told that there was nothing outside of the text and that readers could interpret on this basis without a master-author figure undermining such readings.

The features that can be discerned through stylometry and other computational approaches are, though, a _disempowerment_ of the reader to some extent. Most readers are not likely to notice statistically significant deviations in part-of-speech usage, nor differences in the most-frequently used words within a text. In a way, then, stylometry seems to bring back an authorial subconscious and to read this in a way that counters the aesthetic sense of actual, human reading. It is a type of "reading against the reader" as other paradigms were "reading against the author". The challenge, then, is to connect such findings with the aesthetic experience; to argue why the measurement of lingustic stylings matters by showing how it connects with the experience of reading.

Now the other weird thing that strikes me here, though, is that this "reading against the reader" is still a facet of much traditional literary criticism. The best lit. crit. shocks the reader into a previously unknown and fresh viewport. The best work forces us to see texts in new lights, to bring the shock of the new to the familiar and to critically deform and reform those literary pieces that we thought we knew so well. And this is also a type of "reading against the reader", for it shows how shallow my own readings were whenever I feel that satisfying jolt of what was previously unseen. It appears to me, though, that the shocks of the new of stylometry do not bring this satisfaction, for the reasons that Ted already laid out: there is nothing with which I can connect them in my experience of reading. And so the next part of my thinking is to weave a set of arguments that reconnect the linguistic to the hermeneutic, that bring computational "reading against the reader" back within acceptable bounds once more.