Monday, January 16, 2006
Distant Reading Minds
Full disclosures first: Moretti is a friend, and was my advisor for several years during my grad school days at Columbia in the early nineties. I was there at the beginning of this project; I can still remember the slightly baffled silence the followed his announcement—over dinner with a half dozen grad students—that the future of literary criticism was going to lie in mapmaking. I spent an insane number of hours generating some of the first maps for him over the next year, using some now-obsolete cartography software that was, to say the least, not optimized for mapping narratives. So perhaps I’m biased by my fondness for Moretti—or by the desire not to think of all those hours generating maps as wasted ones—but I really do think that the two books that eventually emerged out of this research (Atlas Of The European Novel and Maps, Graphs, and Trees) constitute a welcome and significant turning point in recent literary criticism. Just the density of ideas in Maps, Graphs, and Trees alone is noteworthy; most scholars would have spun the analysis of genre cycles into an entire book, but Moretti gives you ten pages on it, and marches on to the next case study. It’s exhilarating stuff, and I wouldn’t have it any other way, but it does feel like we’re going to be spending the next ten years unpacking some of these examples.
When Jonathan asked me to contribute to this discussion, he suggested one of the topics I might address would be the connection between my work in Everything Bad Is Good For You, and Moretti’s model in Maps, Graphs, and Trees. As it happened, Moretti and I had just been exchanging a series of email messages remarking on how our thinking had developed along parallel lines since we both left Columbia. I talk about this a little in the Appendix to Everything Bad: the shift from a symbolic model to a systemic model. Most criticism ultimately involves, in one fashion or another, decoding the literary work to find the hidden meaning that lurks behind it. What form that meaning takes depends on the kind of critic you are: the genius of the author, the zeitgeist, the class struggle, the unconscious, the subaltern, différance. But the systemic approach to culture isn’t looking for hidden meaning in the same way. This is how I described it in Everything Bad:
My argument for the existence of the sleeper effect comes out of an assumption that the landscape of popular culture involves the clash of competing forces: the neurological appetites of the brain, the economics of the culture industry, changing technological platforms. The specific ways in which those forces collide plays a determining role in the type of popular culture we ultimately consume. The work of the critic, in this instance, is to diagram those forces, not decode them. Sometimes, for the sake of argument, I find it helpful to imagine culture as a kind of man-made weather system. Float a mass of warm, humid air over cold ocean water, and you’ll create an environment in which fog will thrive. The fog doesn’t appear because it somehow symbolically re-enacts the clash of warm air and cool water. Fog arrives instead as an emergent effect of that particular system and its internal dynamics. The same goes for culture.…
Diagramming not decoding: you can hear the Moretti disciple loud and clear in that language. (You can also hear a little Manuel DeLanda, for what it’s worth.) But the passage also points to an area that I hope Moretti will turn to next. You can’t analyze the literary system purely from the bird’s-eye-view of distant reading. You need to zoom in as much as you need to zoom out: all the way to the human brain itself. Consider this passage from Maps, Graphs, and Trees:
Everybody, from the first readers onwards, had noticed the country walks of Our Village; but no one had ever reflected on the circular pattern they project on the English countryside, because no one—in the absence of a map of the book—had ever managed to actually see it.
Those readers might not have seen that circular pattern, but on some level, they must have perceived it; it must have shaped their experience of Our Village in some meaningful way. Otherwise, it’s just a random pattern, formalist trivia, like discovering there are the same exact number of words beginning with “c” and “t” in the novel. The same goes for Conan Doyle and his rivals: the reading audience isn’t consciously analyzing the formal properties of a stack of detective stories, and saying: “You know, I really like the ones with decodable clues.” But nonetheless they’re somehow processing those formal innovations, and building a sensibility around them, a preference for one configuration over the others.
So my question for Moretti is: how does this happen? How is the reader influenced by formal properties without being fully conscious of the influence? Maps, Graphs, and Trees is silent on the question—understandably so, since he’s got his hands full persuading us of the birds-eye-perspective. But a systemic theory has to work at all the relevant scales. All the science and empiricism of distant reading disappears when you get down to the level of the reader’s mind: we’re left with watered-down Freud: somehow the reader unconsciously digests the form, emphasis on “somehow.” (I tried my hand at this approach for a few pages in Everything Bad—explaining how the formal architecture of reward and exploration in gaming connects to the brain’s so-called “seeking circuitry.") In filling in the gap, I suspect that the cognitive sciences will be more relevant than evolutionary psychology, despite the fact that the Darwinian approach to literature has been attracting all the buzz lately. I’m less interested in where the brain’s capacity for apprehending formal variation came from, and more interested in how it works. No doubt some these cognitive tools for apprehending literary form will turn out to be spandrels, or even exclusively the product of cultural training. That won’t make those tools any less interesting. But we need to be able to talk with some rigor about how those semi-conscious assessments take hold in the brain—the same rigor that Moretti has applied to the system of literature itself. Right now all we have is superstition…
Steven Johnson is the author of four books on science, technology, and culture, mostly recently Mind Wide Open and Everything Bad Is Good For You. He is a Distinguished Writer In Residence at NYU’s Department of Journalism. He can be found on the web at www.stevenberlinjohnson.com.
"The future of literary criticism was going to lie in mapmaking”.
It’s his man-of-destiny shit that I don’t like. I suppose that that was just your tongue in cheek paraphrase, but the guy is far too insistent on the importance of what he does, and seems to be stancing himself for remethodologizing the MLA and starting another trend in the grad schools.
That definitely was a tongue-in-cheek paraphrase—perhaps I should phrase it differently. What he actually said was that there was a big hole in literary criticism: there were no atlases. It’s not that he’s trying to replace a lot of what’s been done already; it’s that he’s trying to fill a blank spot that he been largely ignored for too long…
Yeah, John, a “man-of-destiny” vibe does not exactly emanate.
correction to my last post: “that HAD been largely ignored” not “he been largely ignored”—for those of you keeping score at home.
for what it’s worth, I like that Moretti’s got a Big Idea here, and he’s willing to make to the case for it—and willing to do it in such concise, readable prose. And he’s awfully engaging for a man-of-destiny; somehow I have a hard time imagining Baudrillard dropping by one of these Valve discussions and defending his work…
If this were 30 years ago and you said, “go cognitive young man!” I would gladly have done so. In fact, that’s what I did 30 years ago and have been doing it ever since, not to mention neuro and even a bit of evo. As a result of all that, I’m not sure you can get there from here.
Yes, if we want to know WHY people prefer detective stories with decodable clues we need to know what’s going on inside their heads. And so it is for a whole lot of things. Where we actually are is that we’ve got a good hypothesis about why the lines of poems are roughly 3 o4 4 seconds long, plus or minus. Actually explaining this stuff is going to be hard.
A couple of years ago Meir Sternberg published two very long articles in Poetics Today reviewing a large multi-disciplinary literature on narrative, including various kinds of cognitive work. It was excruciating to read because he pretty much said all this work on narrative hasn’t yet produced much in the way of new insight. It was excruciating because, on the one hand I thought his POV was just wrong-headed, but I pretty much agreed with him on many/most of his substantive claims.
So I think we’ve got a long way to go on the explanation side, a very long way. But then, however much he is interested in explanation, Moretti doesn’t get around to it in GMT. What he does is describe phenomena, some of them, I think new. And that’s one thing we can do: describe stuff. And I think we need to do a lot of that at the level of individual texts, and in detail. I’ve done enough of that—and published the results—and read enough by others that I’m confident there’s a lot there to be done.
I keep reminding myself that Darwin’s work was built on generations of painstaking naturalistic description of plants and animals. And I rather suspect it took a long time for people to learn just how to do the descriptions, just what features were important and which were not, and how to organize and compare them. We need to do that for literature. We’ve got fragments of such descriptions all over the place. But we need to be more systematic and rigorous.
And doing that, it seems to me, means that some of us need to set aside the business of decoding texts, of dredging them for meaning. That’s a different business.
Basically, if we’re scientists of literature, then explanations of literature should ultimately replace interpretations. I’ve just always thought that the main reasons we are interested in literature have little to do with our desire to find explanations of it.
As I said earlier in a garbled way, I’m sure maps and graphs and trees of literary forms have the same kind of validity that maps and graphs and trees of shoe styles and designs would have.
And I don’t object to the very thought of doing that, but there seems to be a mood that this is an important breakthrough, and that now at last literary criticism has done [something big and important], and this mood has a positivist and Braudelian tinge, and I also sniff a new methodologic trend coming down the road to vacuum up the graduate students of the future and vote as a bloc on methodology at the MLA.
"Positivist” is one of the most-abused words in the language. I don’t think it’s a tenable position, mind you, but people frequently denounce it with the same venom they would use for “racist” or “Comte.”
I don’t see what’s wrong with a new methodologic trend, even if it’s just a another fashion to replace the current ones. Why deny people their fun? I’m really not sure what “vote as a bloc on methodology at the MLA” means. I don’t know much about the MLA, but surely they don’t have party-line votes on permissible methodology or something.
My only concern about the methodological trend, if an actual one should appear, is that as I mentioned previously I don’t see what kind of publication a grad student can get out of doing data collection. In that sense I think that this work may be bad for grad students, even if it is arguably good for the field as a whole.
Positivism has been one of my favorite objects of abuse for about 35 years. I’m glad that it’s becoming a trend.
I have no understanding of the mechanics of the MLA, but I’ve been listening to grad students explaining what methodologies are expected of them for most of those 35 years. And I’ve also heard of shifts from time to time, when people learned that certain kinds of things no longer were to be done any more.
Someone tossed out the pluralism option just now, and Moretti knocked it down. So I guess it’s going to be interpretation vs. maps. And that makes sense, if he’s planning on carving out market share versus other methodologies.
That’s the way the system works, so I suppose this ins’t especially about Moretti. To me he’s playing the game a little too enthusiastically, though.
As fas as the MLA membership is concerned, positivism has NEVER been a good thing, not now, not 35 years ago, not 50 years ago, not since the MLA’s founding before positivism happened in the early 20th century. As for market share, do you think that’s all that’s involved, or do you think there might be deep intellectual convictions held in good faith—on all sides?
I’ll keep on.
There’s no particular reason why someone doing Moretti’s kind of thing couldn’t also be doing the various other sorts of things that other people in literature do. Moretti was given a chance to say “let a hundred flowers bloom”, and he turned it down. For him there has to be a decision between interpretation and maps.
In my experience this kind of talk comes from academic-bureaucratic faction fights with something at stake, and ends up with a methodology to be imposed on graduate students.
When he talks about “explanation” that raises another red flag for me. He even asked, at one point, for an explanation of why certain things had happened differently in Japan and in England during the eighteenth century, as though he was trying to discover the Law of Fiction and the Japanese data had to be accounted for. But Japan and England were not in communication at all at that time, and their societies were not only very different but had no common tradition at all, and you’d hardly expect anything to be the same in the two places.
Not quite on topic, but I thought those involved in these discussions might be interested in this article about the application of various algorithms to the google index in order to answer historical questions of fact.
Insincerity isn’t the issue. (Is my love less sincere if I’m rewarded by the presence of my beloved and if I haven’t sought out and compared all other possible beloveds?) But Moretti’s and journalism’s lack of interest in the hard-working materialists who don’t make such sweeping claims bodes ill.
As Bob O’Hara summarized in passing, contingent history cannot be made isomorphic with verifiable lab science. Some of this recent rhetoric seems intent on remaking the mistakes of such historical system-builders as Brooks Adams and Oswald Spengler. A scientised literary history will likely contribute to fewer deaths than scientised general history did, but it seems even more confused: the aesthetic realm is defined by particularity, atemporality, and variation of taste.
Regarding “rigor”: If someone vigorously misapplies techniques to rigidly-held arbitrary premises, are the results “rigorous”? They appear so; they even feel so. Still, there’s no reason to believe they’re true.
An unconsciously developing preference for visible decodable clues “sounds reasonable.” But Moretti’s clue tree doesn’t prove it: the tree never reaches the supposed end goal of evolution; chronological ordering by height is implied but not documented; no evidence is shown that Doyle’s rivals at each “level of development” were as popular as Doyle’s stories at the same level. The parsimony principle suggests instead that the popularity of the “Sherlock Holmes stories” was due to Sherlock Holmes, with visible clues a side-effect of building narratives around such a character: the magician is more effective when audience members inspect the box. What defines a MOR-generic detective novel isn’t the quality of its puzzle but the presence of a trademark detective.
But then we’re back to personal names, which aren’t nearly as scientistic as abstractions.
Sullen bastard that I am, I’m not even cheered up by Steven Berlin Johnson’s call for equally “rigorous” use of brain studies.
We’ve heard what professional philosophers think of the English department’s use of philosophy, what mathematicians think of English department Gödel and chaos, and what physicists think of English department entropy and quantum theory. And we’ve heard some of what professional biologists and cognitive scientists think of “Evolutionary Psychology." Popularized metaphorically applied science is not science. But it is what consumers want, and it’s lucrative.
Since 1993, I’ve been fascinated by application of the cognitive sciences to aesthetics. Since 1995 or so, people who find this out have asked me what I thought of some best-selling book or newspaper article. And I’ve always had to say that I prefer reading in the primary disciplines. Outside them, we seem to get Just So stories which justify a conservative canon, New Age vapidity, or pop culture tautology. ("People really do enjoy what they seem to enjoy. EEGs prove it!") Working scientists who venture into criticism fare little better. Vilayanur S. Ramachandran’s defense of this Chola bronze sounds like a Jungian art appreciation class. If he’s describing biologically determined reactions, why does his audience have to be won over? I’d say it’s because aesthetics is too complex and socially-personally contingent to support the sort of argument he’s making.
Unless we burn art, it outlives our theories about art. I believe the cognitive sciences, cross-cultural history, and philosophy can all shed light on why that is. But where’s the percentage? What sells is a familiar story with a twist. 100 newspapers and NPR can’t be wrong.
Those readers might not have seen that circular pattern, but on some level, they must have perceived it; it must have shaped their experience of Our Village in some meaningful way. Otherwise, it’s just a random pattern, formalist trivia, like discovering there are the same exact number of words beginning with “c” and “t” in the novel.
I think this precisely is what needs to be proven. How do we know that anyone perceived it or even cared? “Otherwise, it’s just a random pattern.” Why couldn’t it be? It looks like the null hypothesis has a lot going for it. It doesn’t seem to be that much of a stretch to think that a four-volume collection of village stories called My Village would be centered on a particular village, and that the stories concerning things taking place farther away could be arranged in a “roughly concentric” fashion around it.
Anyway, thank you, Valvistas, for this very interesting event and for making the documents available on your site.
In 1957 Chomsky published Syntactic Structures. By the mid-1960s (I believe) psychologists had figured out how to test his grammatical theories and empirical papers were appearing in the literature. Thus was psycholinguistics born. It happened quickly because there was an “affinity” between Chomsky’s formal linguistics and the research methodologies available to psychologists.
As far as I can tell the “distance” between literary criticism and the cognitive sciences is much greater. I don’t expect to see deep empirical results anytime soon. And when that begins to happen, I don’t expect those results to take the form of experimental confirmation of this or that interpretation for this or that text.
* * * * * *
Some years ago I read an article in PMLA about patterns in literary texts. The article asserted that the “Author’s Prologue” to the Collected Poems of Dylan Thomas had a rhyme pattern that was the same forward as it was backward. That poem is 100+ lines over five pages. While I had no reason to doubt the article, the notion was so astounding that I checked for myself. Here’s how the poem opens:
This day winding down now
At God speeded summer’s end
In the torrent salmon sun
Molton and mountainous to stream
Over the wound asleep
Sheep white hollow farms
To Wales in my arms,
Hoo, there, in castle keep,
You kin singsong owls, who moonbeam
And the end:
My ark sings in the sun
At God speeded summers end
And the flood flowers now.
Was Thomas explicitly conscious of constructing this elaborate yet simple pattern? Unless it is pointed out, how likely will is it that a reader notice it? And when the reader notices it, what then?
Come to think of it, those opening and final lines are a pretty strong clue. Not only do they rhyme—inversely—but the rhyming words are the same. And then note the repeititon of the second line as the penultimate line. It’s as though Thomas is inviting you to pay close attention to what’s going on at the ends.
>But Moretti’s clue tree doesn’t prove it: the tree never reaches the supposed end goal of evolution; chronological ordering by height is implied but not documented
Moretti’s next figure in the chapter makes it clear that there is no strict chronological ordering by height in the earlier figure. Things are complicated. Part of what I liked about the book is that Moretti is willing to follow the data where it takes him.