As part of the Magazine Modernisms essay club, I’ve previously written about Franco Moretti’s work with social network analysis. In particular, Moretti’s work creates conversation networks in Hamlet by using lines spoken on stage and directed to another character. Moretti’s article references another recent study in “conversational networks” which was presented at the 2010 ACH/ALLC Digital Humanities conference. The paper “Extracting Social Networks from Literary Fiction” [pdf] traces conversational networks of bilateral conversation between characters. Again, conversation provides the quantifiable exchange between characters to form the nodes and edges of the network. Aditi Muralidharan, of course, does an excellent job of unpacking the article’s premise, methods, and arguments in her post Extracting Social Networks from 19th Century Novels.
What I find useful about this study is that the purpose of the activity was to “test” commonly-held assumptions about location and community in Victorian novels. We assume that Victorian novels set in rural environments include more conversation between closer networks of characters and once characters move to urban surroundings, those conversations become more disparate and therefore relationships begin to break down. Elson, Dames, and McKeown’s analysis demonstrates that this claim is based more on human perception than textual evidence. These are the kinds of results that I think one hopes to find when you work with genre and computational text analysis.
The question for me, however, remains to be how one can extract nodes and vertices from poetic texts. Unlike Victorian novels, poems often have unnamed characters with fluid subjectivities. Conversation is much more oblique, and doesn’t often have textual markers (i.e. quotation marks or italics are not regularized across texts). It seems to me that any quantitative analysis of modern poetry that hopes to do more than simply count words would require extensive mark-up and metadata, by which point the data is so thoroughly manipulated that it becomes difficult to trust the outcome. So, for the moment, network analysis of ekphrastic texts seems most fruitful in those small, hand-drawn models rather than generated through computation.