Some notes on how I discovered formal structure in some literary texts, Part 1: Four cases

What is intelligence? How does it work? Those and related questions have been with us for a while, but their salience has been amplified in the wake of ChatGPT, which is given us a demonstration of machine intelligence, whatever that is, that is accessible to anyone with access to the web. Those questions have ceased to be of interest primarily to psychologists, philosophers, and AI researchers. Now “everyone” is interested in them.

Perhaps the single most pressing issue is: While machine intelligence ever reach, or even surpass, human intelligence? The concept of intelligence is so obscure, however, that it’s difficult to produce compelling answers. Earlier in this year I decided to take another crack at the problem, AI, Chess, and Language 1: Two VERY Different Beasts. That post has, in turn, led to a series of posts on chess and language.

More recently, I did a long post centered on analogy, Intelligence, A.I. and analogy: Jaws & Girard, kumquats & MiGs, double-entry bookkeeping & supply and demand, and another in which I talked about wanting an AI that could determine the formal structure of literary texts, What do I personally want from an AI? [as soon as possible, too, NOT in the distant future]. Neither of those posts says anything about chess, but they’re certainly on the language side of the ledger. That’s led to the idea of reflecting on what I’ve had to do to discover formal structures in texts.

I’ve done a lot of this. Which texts should I use?

Kubla Khan

“Kubla Khan” is the most important example. That’s where I started, and it’s the most richly developed example. Furthermore, I’ve published that work in the formal literature and I’ve already written quite a bit (informally) about how I came to do that work.

Here's the articles about the poem:

Articulate Vision: A Structuralist Reading of "Kubla Khan", Language and Style, Vol. 8: 3-29, 1985, https://www.academia.edu/8155602/Articulate_Vision_A_Structuralist_Reading_of_Kubla_Khan_

“Kubla Khan” and the Embodied Mind, PsyArt: A Hyperlink Journal for the Psychological Study of the Arts, Article 030915, November 29, 2003, https://www.academia.edu/8810242/_Kubla_Khan_and_the_Embodied_Mind

These papers have background:

Touchstones • Strange Encounters • Strange Poems • the beginning of an intellectual life, November 2015, https://www.academia.edu/9814276/Touchstones_Strange_Encounters_Strange_Poems_the_beginning_of_an_intellectual_life

Beyond Lévi-Strauss on Myth: Objectification, Computation, and Cognition, Working Paper, February 2015, pp. 20-27, https://www.academia.edu/10541585/Beyond_L%C3%A9vi_Strauss_on_Myth_Objectification_Computation_and_Cognition

I originally published the “Touchstones” piece in a journal edited by one of my teachers in graduate school, the late Art Efron: Paunch 42–43: 4–16, December 1975. It’s about my years at Johns Hopkins and how I came to do a Master’s Thesis on “Kubla Khan.” I’ve linked to a version which I’ve updated with notes that I’ve inserted into the text, notes commenting one what’s happened since then. The Lévi-Strauss piece is generally about the intellectual significance of his work on myth while the section, “Into Lévi-Strauss and Out Through ‘Kubla Khan’” is specifically about how Lévi-Strauss’s structuralism guided my approach to the poem.

Three Shakespeare Plays

This is about three Shakespeare plays, a comedy (Much Ado About Nothing), a tragedy (Othello), and a so-called romance or tragi-comedy (The Winter’s Tale). In each of these plays a man wrongly suspects his beloved of betraying him with another man. Consider the following diagram:

Along the left I've listed five dramatic functions which some character must play. In the columns for each play I've indicated the character that takes these functions. The point of the diagram as the we move from one genre to the next (in this order) one function seems to disappear. Othello has no mentor, but he has a deceiver; and Leontes has neither a mentor nor a deceiver. What I think is going on is that functions are, in effect, being absorbed into the protagonist. At the play's opening, Othello is senior enough in the world that he has no need of a mentor. Leontes is king in his world, so there is no one higher. As for being deceived about his wife, he does that to himself, no external agent required.

There's more to the pattern than that, but that’s enough to give you an idea about what’s going on. The analytic point is that this is a pattern which becomes visible only when you compare texts. That’s something that Lévi-Strauss did in his work on myth.

Here's the article I published about that:

At the Edge of the Modern, or Why is Prospero Shakespeare's Greatest Creation? Journal of Social and Evolutionary Systems, 21 (3): 259-279, 1998, https://www.academia.edu/235334/At_the_Edge_of_the_Modern_or_Why_is_Prospero_Shakespeares_Greatest_Creation

Metropolis

I’m talking about the manga originally published by Osamu Tezuka in 1949 and not Fritz Lang’s movie of the same title. As you know, manga are Japanese comic books, or graphic novels. I’m including this as a case in the first place because it is an example of ring-form construction, which I depict in the following diagram:

Notice the numbering at the left and the descriptive labels at the right. I’m including it in the second place because it is the first time I used to whole-text table to conduct an analysis. The diagram is reduced from that table.

Here’s the article:

Tezuka's Metropolis: A Modern Japanese Fable About Art and the Cosmos, in Uta Klein, Ktaja Mellmann, Steffanie Metzger, eds. Heurisiken der Literaturwissenschaft: Disciplinexterne Perspektiven auf Literatur. mentis Verlag GmbH, 2006, pp. 527-545, https://www.academia.edu/7959634/Tezuka_s_Metropolis_A_Modern_Japanese_Fable_about_Art_and_the_Cosmos

Here's the analytic table:

Descriptive Tables for Tezuka’s Metropolis, A Note on Descriptive Method, Working Paper, 2016, 8 pp., https://www.academia.edu/27370076/Descriptive_Tables_for_Tezukas_Metropolis_A_Note_on_Descriptive_Method

Heart of Darkness

It turns out that Joseph Conrad’s Heart of Darkness is also a ring-form text, or a close variant. Consider the following chart:

Each bar represents a single paragraph in the text. The length of the bar corresponds to the number of words in the text. The bars are arranged in the order they exist in the text, first to last, from left to right.

Notice that there is one paragraph that is noticeably longer than all the others. It has 1502 words. It is the structural center of the text. There’s much more to be said about that paragraph and its place in the text. But this is not the place to do it.

You can find those discussions, and others, in this working paper:

Heart of Darkness: Qualitative and Quantitative Analysis on Several Scales, Version 3, Working Paper, August 9, 2018, 49 pp., https://www.academia.edu/8132174/Heart_of_Darkness_Qualitative_and_Quantitative_Analysis_on_Several_Scales_Version_3

I have two main reasons for including this case as an example of discovery. In the first case, that chart depicts something that is clearly an objective fact of Conrad’s text. It’s not something I conjured out of the mist through interpretive wizardry. I make the same claim about the formal structures in the other three cases. That’s why I’m calling them discoveries; they’re as real as a vein of gold in the Klondike.

And the process by which I discovered that structure in Heart of Darkness was, if anything, more serendipitous than a gold strike. The prospector who gets a gold strike was looking for gold. I wasn’t looking for a ring-form pattern in Conrad’s text. Not only that, but charting paragraph lengths is not something one normally does with literary texts or, as far as I know, any other texts. I did it on a whim. While I did have a formal reason for looking at that particular paragraph, it didn’t have anything to do with ring-composition. While examining that paragraph I noticed that it seemed unusually long. Perhaps it was even the longest one in the text. So I decided to take a look. Out of mere curiosity.

But I’ve said enough about that so far. I’ll say more later. Here’s the point at issue: How do you construct at AI to discover something of a kind that it didn’t know exists? That’s what I did in this case and in “Kubla Khan” as well. The pattern I discovered across those three Shakespeare plays is arguably related to the sort of thing I’d seen in Lévi-Strauss’s work on myth, though quite different in the details. As for the ring-composition in Metropolis, ring-composition is well-enough known, though not much studied by literary critics. I was alerted to it by the late Mary Douglas. 

The importance of visualization

I don't see how I could have done any of that work without visualization. I could have spotted that something was going on in the Shakespeare case, and maybe, just maybe, I could have made an argument without that simple diagram. But the fact is that that's not how I did it. More precisely, I might have imagined the argument I ended up making without drawing that table. But I can't imagine trying to write it up without making the table. The table allows you to see the structural phenomenon at a glance.

In the other three cases diagrams were an essential part in seeing the phenomenon at all. Since I didn't include any diagram from the "Kubla Khan" work in this post, I will here explicitly state that diagramming was essential to my "Kubla Khan" work from the beginning.

All of which is to say that in each case, two distinctly different methods of thought were important. What does that imply about the possibilities of artificial intelligence?