I recently read in Inside Science that physicists had applied network analytic tools to discover whether three classic myths: Beowulf, the Iliad, and Tain Bo Cuailnge, were based in real life events (Here is the original article). In a project that probably fell to some undergrad RA, the characters from each of the three stories were entered in a database, and links, coded as either hostile or friendly, were assigned between the characters who had relationships.
According to their coding of the data, Beowulf had 74 characters, the Iliad had 716 characters, and Tain Bo Cuailnge had 404 characters. Once the network was constructed for each myth, they calculated different network statistics, like the mean number of people that each person had links with and whether those with many links tended to be connected with others who had many links (called degree assortativity). If networks are highly assortative (meaning that people are more likely to be linked with others who have a similar number of links) it is thought that these networks are more true to real life. They came to the conclusion that the Iliad is the most realistic social network.
This got me thinking about using network analysis to analyse texts, and in particular how to use network analysis to compare the works of different authors. I started wondering if you had a data set containing information on social networks in different novels, if you would be able to determine which novels were written by the same author, or which novels were in the same genre. I also started to think about how the use of longitudinal network analysis seems like a natural fit since the relationships in a fictional work change over time.
Thinking is great, but I really love actually doing something, so I decided to go ahead and start coding up one novel’s social network to start to answer some of the questions that had been floating around in my head. I picked Tales of the City for these reasons:
- It was on my bookshelf
- The characters belong to many overlapping social groups
- It is an easy read
- Also did I mention that I love these books?
I have now been working on this on and off for a couple of weeks. So far I have gone through 274 of the 375 pages, identified 69 characters, and recorded their ties. Unlike the above study I am not keeping track of hostile vs. friendly ties. So for example though DeDe Halcyon-Day is friends with Binky, and enemies with Shugie Sussman (Dede only speaks of Shugie unkindly) both of these relationships are coded the same. Though I think it would be fitting, I am not recording how the relationships change over time as I decided that would be too time consuming. Here is an image of the social network as of page 274:
I made this in R with the SNA package. Nodes for minor characters are in gray while Mary Ann is green, Anna Madrigal is pink, Michael Tolliver is blue, Mona is purple, Edgar Halcyon is red, Beauchamp Day is orange, and Dede Halcyon-Day is yellow.
Nerd fun abounds!!!