Network Analysis

Another Research Question

Network analysis was not part of the original research question for this project. Our interest was piqued by a wonderful presentation by Dr. Elisa Beshero-Bondar on network analysis and her applications of it to the Digital Humanities project Spectacular Intersections of Place in Southey’s Thalaba the Destroyer. Much like the duality of Southey's Thalaba text lent itself to an edged network analysis, we saw an opportunity in the separation between Muggle and wizarding worlds in Harry Potter. As linguists and coders who had spent far too much time poring over three editions of the first HP novel, the research question for our network analysis formed along linguistic boundaries. All three languages (French, Russian, and Swedish) that we were already comparing syntactically had a register distinction - that is to say, speakers choose between informal and formal pronouns and grammatical forms based on their familiarity with the addressee, or the situation at hand. The registers used vary based on several sociolinguistic factors that are fluid within the languages themselves, so their usage would be expected to be theoretically similar, but not realized in exactly the same ways (2). Our research question thus became: How do the characters of Harry Potter address each other?

Graphical Representations of Networks

The network images were generated with the program Cytoscape, with much help from Dr. Beshero-Bondar's Cytoscape tutorial. First, the data was collected from the first chapter of Harry Potter in each language. In this chapter, the characters represented are Albus Dumbledore, Minerva McGonagall, Rubeus Hagrid, Vernon Dursley, Petunia Dursley, and one unnamed wizard. Russian was done first as a test network, then French and Swedish were added. The following attributes had to be identified for each instance of personal pronouns or direct address: speaker, addressee, and level of register. These data were put into a .txt document, which was then imported into Cytoscape as a table. The nodes in the network are speakers and addressees, while the edges represent the level of register. The margin of error comes from an unfamiliarity with the languages other than Russian by Taylor, who produced the networks. Relying on the provided list of formal and informal pronouns in French and Swedish, and using best judgment for direct address, resulted in a slightly smaller sample size in both French and Swedish. However, the quality of the nodes represented is the same across languages.

Room For Future Study

Our project as a whole only took into account chapters 1, 8, and 17 of Harry Potter and the Sorcerer's Stone. The network analysis only represents the first chapter of that sample. Character interaction is thus relatively limited; a more complete analysis would include other chapters to enhance the existing networks. For example, Hagrid speaks in direct address to Professors Dumbledore and McGonagall in chapter 1, but in a later chapter, his register uses second person pronouns. Thaicharoen writes about the challenges of translating Hagrid's idiolectic use of "yeh" and "yer" for "you" and "your" into Thai, implying a register that isn't easily translatable (4). A larger project more focused on network analysis might try to represent these idiosyncrasies as well as traditional register distinctions.


1. Beshero-Bondar, Elisa. "Spectacular Intersections of Place in Southey’s Thalaba the Destroyer."

2. Gołąbek, Rafał. "Culture as a factor determining tu-vous usage."

3. Beshero-Bondar, Elisa. "An Introduction to Network Analysis and Cytoscape for XML Coders."

4. Khanittha Thaicharoen. "Translation quality: A study of Harry Potter: The Philosopher's Stone." Master's thesis submitted at Srinakharinwirot University, 2007.