Six Degrees of Danish Bacon

by Kieran Healy on May 3, 2011

The current issue of New Left Review has an article by Franco Moretti applying a bit of network analysis to the interactions within some pieces of literature. Here is the interaction network in Hamlet, with a tie being defined by whether the characters speak to one another. (Notice that this means that, e.g., Rosencrantz and Guildenstern do not have a tie, even though they’re in the same scenes.)

The Hamlet network

And here is Hamlet without Hamlet:

Hamlet without Hamlet

I think we can safely say that he is a key figure in the network. Though the Prince may be less crucial than he thinks, as Horatio seems to be pretty well positioned, too. Lots more in the article itself.

{ 43 comments }

1

geo 05.03.11 at 8:23 pm

I think we can safely say that [Hamlet] is a key figure in the network

Egad! Are there any more stunners like this in the article? Should I buy it?

2

Substance McGravitas 05.03.11 at 8:31 pm

When it comes to phenomena of language and style, we can do things that previous generations could only dream of.

Thank god we are past the pencil-and-ruler famines of yesteryear.

3

Gareth Rees 05.03.11 at 8:45 pm

You could make different rhetorical points by drawing the graph in different ways.

A central group of important characters plus hangers-on.

A hierarchy with Claudius at the top.

4

Myles 05.03.11 at 8:52 pm

I daresay, that Lord connexion looks shaky. Has has been any independent verification? He doesn’t seem to very connected; not a chap that mixes well, I see.

5

Jonathan 05.03.11 at 9:10 pm

Here’s an expanded (and free) version of the article:

http://litlab.stanford.edu/?p=337

6

Substance McGravitas 05.03.11 at 9:17 pm

Thanks Jonathan!

7

LFC 05.03.11 at 9:27 pm

That Rosencrantz and Guildenstern do not have a tie may say something about the limits of network analysis.

8

Salient 05.03.11 at 9:33 pm

I think we can safely say that he is a key figure in the network.

No, I think we can safely say that the guy who does Garfield Minus Garfield has some new work cut out for him.

9

Bill Benzon 05.03.11 at 10:15 pm

I’ve been working on Road Runner cartoons. Since Road Runner and Coyote never speak, I guess their graph is just two vertices.

10

Bloix 05.03.11 at 10:27 pm

Rosencranz-and-Guildenstern don’t have a tie because they function as a single character. They don’t need to talk to one another because there is no thought in the head of Rosencranz which is not also in the head of Guildenstern.

11

Strategist 05.03.11 at 11:44 pm

#1 – #10.
Exit Moretti pursued by bears

12

Maurice Meilleur 05.04.11 at 12:19 am

Whenever I encounter Moretti writing about literature, I’m reminded of an example Douglas Hofstadter uses in Gödel, Escher, Bach in his discussion of levels of organization and the reductionism/holism debate: A man who buys an LP of a Bach recording and hangs it on the wall, and invites his friends over to admire it.

What Moretti sees when he charts this stuff may be interesting and all, but does it really contribute to my critical understanding of a novel or a play? Was literature running so low on complexity and interest that we had to find new ways to talk about character and genre (for examples) that don’t require us to just read the damn book already?

13

Kiwanda 05.04.11 at 1:04 am

As the piece says at the end, hopes of applying network theory were abandoned pretty quickly, to do some vamping inspired by the network drawings. But then, attempts at getting some insight by making more elaborate network drawings were unsuccessful. So the upshot of the paper is, pretty explicitly, “We tried to use these fashionable ideas, but no joy.” Well, it’s refreshing to see the publication of negative results.

14

Happy Heyoka 05.04.11 at 3:51 am

Maurice wrote:
What Moretti sees when he charts this stuff may be interesting and all, but does it really contribute to my critical understanding of a novel or a play?
Well, as someone outside the tent I could be naughty and say pretty much the same thing about literary criticism as a whole in regard to actual appreciation of literature.

As someone who uses Directed Graphs reasonably often, I would suggested that the benefit would be in comparing all of the works of one author to another.

The whole point of this kind of tool is to reduce the detail to the barest and most essential relationships – you could overlay a graphs using different coloured edges (the lines) for the dialogue between characters and their appearances in a scene together and gain some higher level understanding of the construction of the play(s).

For example – you could examine character relationships in Shakespeare in comparison to Arthur Miller and Samuel Beckett plays. Using the different coloured edges idea above you could stand back and squint at all the graphs (honest) and get a general overview of the technique of the authors.

It may say absolutely zero about the quality of the work, but may still provide useful insight… or at least that’s how it works in my field.

Of course, you should insert appropriate allegory regarding a man with hammer here.

15

Vance Maverick 05.04.11 at 5:23 am

I don’t quite get the hostility expressed upthread here. These seem like handy visualizations, making manifest some features of the play that, while hardly secrets, aren’t easy to survey without a crib of some sort. Moretti, in the pamphlet version, does essay some strong claims, but inconsistently, and in the end he abandons them. I hold no particular brief for him, but I don’t think anyone here has laid a glove on his actual claims or argument, or the graphs for that matter.

16

anitchang 05.04.11 at 6:33 am

@11 wins the thread

17

anitchang 05.04.11 at 6:44 am

Ok, I venture a guess (without having read the article and taking up suggestions in comment 14): what about producing this kind of graph for different narrations of the same story (say: Ur-Hamlet, Hamlet, “Ophelia” adaptations, all the over 30 version film versions, etc.) … couldn’t that lead to some interesting insights? Not that this couldn’t already have been done without network analysis, but it seems to me that this is something that mostly has value in comparative study and not so much in giving an overview over one single story.

18

rfriel 05.04.11 at 8:21 am

It confuses me when people question Moretti’s methods because the conclusions he produces could be arrived out with more traditional methods. Maybe they could, but why not use all of the tools at one’s disposal? Sometimes, when you’re trying to figure something out, it helps to draw a diagram. I would imagine that this is a claim most people would accept, so why object to Moretti’s pragmatic use of diagrams because we could, in principle, do the same things some other way?

If there’s a valid complaint to be made about Moretti’s methods, it is not that they are unnecessary innovations, but that they are not innovative enough. “Drawing diagrams” is not a revolutionary new methodology; it’s just a common-sensical thinking aid, as should be clear to anyone who’s ever felt the need to whip out a pen and scribble on a pad of paper while thinking about something. If Moretti used statistics or some other sort of quantitative technique to analyze his data, rather than just diagramming it and then coming up with stories about the diagrams, that might be a revolution worth disagreeing with. (I guess his discussion of power-law distributions in the new pamphlet takes him closer to “statistics or some other sort of quantitative technique” than he went in Graphs, Maps, Trees, but I haven’t looked closely enough at the paper to judge that bit.)

19

Bill Benzon 05.04.11 at 9:20 am

Well, as someone outside the tent I could be naughty and say pretty much the same thing about literary criticism as a whole in regard to actual appreciation of literature.

Yep. & I’m someone inside the tent. But also outside it.

Early in my career I became interested in cognitive networks, directed graphs used to represent knowledge structure. So I worked on a Shakspeare sonnet (129) and published an article: “Cognitive Networks and Literary Semantics,” MLN, Vol. 91, pp. 952-982, 1976. Here’s the abstract:

A cognitive network is a type of semantic model developed for simulating natural language on digital computers. A concept is a node in the network while connections between nodes represent relations between concepts. One generates a text by tracing a path through the network and rendering the successive concepts and relations into language according to the appropriate conventions. Elementary concepts are grounded in sensor-motor schemas while abstract concepts are grounded in patterns of network relationship. The semantic structure for Shakespeare’s “Th’ expense of spirit” (Sonnet 129) is given by an abstract pattern for the Fortunate Fall, which is linked to a pattern specifying a fragment of the conceptual basis for faculty psychology.

I went on to write a dissertation on cognitive science and literary theory. & I ran up against a wall. I wouldn’t say it was a dead end, but, well . . . it’s a long story that’s still ongoing. Let’s just say that the intuition I got from the exercise has proved invaluable.

20

maidhc 05.04.11 at 9:32 am

It’s a play. There are some considerations that have been left out.

Some characters may need to be offstage at a certain point because they are changing costumes or other business.

Effectiveness of the performance is paramount. Maybe having Rosencrantz and Guildenstern not talk to each other enhances the overall atmosphere.

It’s sort of an interesting concept in the abstract, for comparative use, but I think it’s a rather blunt instrument.

21

rm 05.04.11 at 1:41 pm

Gertrude doesn’t get to be named? She’s just “Queen”? Or is this an early-80s rock opera version I was unaware of?

22

Frowner 05.04.11 at 1:48 pm

I actually like Moretti’s quantitative/mapping stuff. The part in Atlas of the European Novel about the meaning of location in various early 19th century English novels is pretty neat – it talks about social class and the parts of novel plots that tend to take place in Scotland versus those in the south of England, the way region relates to how morality is depicted and, er, some other stuff I don’t remember because I don’t have the book handy.

Maybe this diagramming of Hamlet isn’t the best example of his recent work, but in the article he talks about the possibility of mapping networks in the huge, complex Qing Dynasty novel The Story of the Stone. (This is one of my favorite novels, so I know it – albeit in translation – fairly well.) Mapping the networks in Story of the Stone would actually be really useful – I’ve read some essays which do this partially in words, and it doesn’t work well because there are about twenty million characters, the novel seems to have been written in multiple versions by several people and it’s partially autobiographical. Mapping would draw out a lot about the structure of the novel that is difficult to see; it would also, I imagine, help sort out some of the social commentary in the novel by making the family and other social connections more visible. If you were to do some different maps that showed relationships over the course of the novel, you could easily get at many of the nuances of the Jia family’s changing fortunes.

The thing is, I’ve read Moretti’s earlier work and found it absolutely incredible. The Way of the World, for me, opened up a new way to understand the political dilemmas of modern radical leftism, for example. There’s quite a lot of “how to live” in Moretti, if you’re into that sort of thing. So I have a bit more faith in Moretti’s smarts than I do in my own, as deferential-to-authority as that may seem. I don’t always get the point of his mapping work, but I tend to assume that further reading and thinking will help me to understand and use it. I know there’s a lot of hostility in reactions to his recent work, but sometimes it sounds like people looking at early modernist art and saying “What’s so important about that? My kid could do that.”

23

roac 05.04.11 at 2:26 pm

you could examine character relationships in Shakespeare in comparison to Arthur Miller and Samuel Beckett plays. Using the different coloured edges idea above you could stand back and squint at all the graphs (honest) and get a general overview of the technique of the authors.

I don’t have access to whatever software is used to construct these graphs, but I can get started on Beckett:

*
Krapp

OK, that was cheap, but it has a point: Graphing the character relationships in Beckett’s plays would certainly tell you something about his technique — but it would be something you already knew.

24

Natilo Paennim 05.04.11 at 4:26 pm

What, no Yorick?

25

roac 05.04.11 at 6:58 pm

I think Yorick is generally classified as a prop not a character.

26

Craig 05.04.11 at 8:00 pm

Um, excuse me?

Rosencrantz: Ho, Guildenstern! Bring in my lord!
(Hamlet, Act IV, Scene 3)

I know it’s the postmodern age and all, but I think that counts as Rosencrantz speaking to Guildenstern. I hope Moretti didn’t highlight that interesting fact.

You really want to have your “Hamlet” down stone cold before you publish results like this…I probably wouldn’t recall if Tamora, say, ever speaks a line to Marcus Andronicus. But Hamlet?

27

Craig 05.04.11 at 8:09 pm

@rm–

Another flaw in the article, re “Queen” versus “Claudius.”

Both characters are identified by name in the stage directions to the Second Quarto edition and the First Folio; in the actual spoken dialogue, Claudius calls the queen “Gertrude,” or some recognizable variant thereof, in every printed edition. No one ever calls the king “Claudius.” So there is no textual basis for calling these characters “Claudius” and “Queen.” “Gertrude” and “King” would make much more sense. Again, you really gotta have your Hamlet _down_ if you want to publish…

28

Satan Mayo 05.04.11 at 8:40 pm

I know it’s the postmodern age and all, but I think that counts as Rosencrantz speaking to Guildenstern. I hope Moretti didn’t highlight that interesting fact.

Maybe “the characters speak to one another” means “each character speaks to the other”, thus doesn’t include purely one-way “exchanges”.

Both characters are identified by name in the stage directions to the Second Quarto edition and the First Folio; in the actual spoken dialogue, Claudius calls the queen “Gertrude,” or some recognizable variant thereof, in every printed edition. No one ever calls the king “Claudius.” So there is no textual basis for calling these characters “Claudius” and “Queen.” “Gertrude” and “King” would make much more sense.

That doesn’t matter.

Again, you really gotta have your Hamlet down if you want to publish…

If you were the editor, that is.

29

mds 05.04.11 at 8:47 pm

Salient @ 8:

No, I think we can safely say that the guy who does Garfield Minus Garfield has some new work cut out for him.

Oooh, I like this idea. The easy way out with Hamlet is to have the ghost be a figment of Hamlet’s imagination. But what if Hamlet were the figment instead? It would cast the murder of Polonius and its consequences in a whole new light.

30

Salient 05.04.11 at 9:05 pm

SPOILERS ALERT

But what if Hamlet were the figment instead?

Then you get Who’s Afraid of Virginia Woolf?

31

belle le triste 05.04.11 at 9:16 pm

“And do you know what I’m going to do now?”
“No,” she said. “What?”
“Buy a copy of ‘Hamlet,'” I said, “and solve that!”

32

geo 05.04.11 at 9:23 pm

HH @14: I could be naughty and say pretty much the same thing about literary criticism as a whole in regard to actual appreciation of literature.

Sorry you feel that way. I’ve found a lot of Empson, Trilling, Leavis, Williams, Pritchett, Bayley, Rahv, Howe, Tate, Jarrell, Davenport, Kenner, Birkerts, Wood, and dozens of other people very useful for helping me appreciate literature.

33

Craig 05.05.11 at 12:00 am

@Satan Mayo:

“Maybe ‘the characters speak to one another’ means ‘each character speaks to the other”, thus doesn’t include purely one-way ‘exchanges’.”

The diagram is broken under this definition, too, as for example the Ghost does not speak to Horatio. The boys are rather miffed at this, you will recall, and are so majestical as to offer it the show of violence.

34

Bill Benzon 05.05.11 at 12:02 pm

I’ve posted a methodological remark about visualization and objectification at my blog:

http://new-savanna.blogspot.com/2011/05/moretti-update-visualization-and.html

35

dsquared 05.05.11 at 12:09 pm

What Moretti sees when he charts this stuff may be interesting and all {…}

But being interesting, “and all” is also a positive contribution to the world; god knows there’s bugger-all on telly these days.

36

Bill Benzon 05.05.11 at 12:37 pm

Interesting is, good, no?

37

LFC 05.05.11 at 12:44 pm

Bloix @10:
Re R&G: yeah; but I couldn’t resist taking a dig at network analysis in general. (Should say that I have not read Moretti and have no view about his work.)

38

aboulian 05.05.11 at 5:16 pm

Literary criticism can be a kind of art. (And I second Mr Scialabba’s list of names.) I couldn’t say that Geoffrey Hill’s criticism wasn’t an addition to his poetic oeuvre, as much as to any tradition of commentary or scholarship.

I sometimes want to ask Ricks, ‘Well which opinions do you believe that you have improved into knowledge?’ Because Ricks, really, is an artist with an idiosyncratically powerful command of English style. And he knows it.

39

peter ramus 05.07.11 at 3:31 pm

Hamlet’s soliloquies are literally off the charts, is all I’m saying.

40

geo 05.07.11 at 5:10 pm

Frowner @22: The Way of the World, for me, opened up a new way to understand the political dilemmas of modern radical leftism, for example. There’s quite a lot of “how to live” in Moretti, if you’re into that sort of thing

This is intriguing. Would you say a bit more?

41

Ray Davis 05.09.11 at 1:40 pm

Frowner@22: This technique (like map-tracing) is well worth teaching students — I’ve made similar charts when working on essays, and I know I’m not alone. But it seems characteristic of Moretti’s recent publications to emphasize technique over results, and characteristic of his most fervent recent readers to prefer things that way.

42

Ray Davis 05.09.11 at 1:51 pm

Probably the most wide-spread application of such diagramming is the Bechdel Test.

43

Ray Davis 05.09.11 at 2:12 pm

Oh hell, that’s what I get for catching up with posts from the bottom rather than the top — the point’s already been made in Holbo’s follow-up.

Comments on this entry are closed.