I just had an abstract accepted to present at the MLA 2016 (Austin, TX) for Société Rencesvals American Canadian Branch division panel “Digital Humanities and the Romance Epic: A New Approach?” organized by Paula Leverage. As my first post here suggested, I am interested in ways of visualizing as many texts as we can from the corpus of medieval French. My MLA proposal is entitled “Visualizing Romance Epic Space-Time” and aims to examine variance of place name usage over time in romance epics (a.k.a. chansons de geste).
The chansons de geste seemed like a perfect use case for digital analysis. The corpus is vast and the genre is highly repetitive (just think of Rychner’s 1955 landmark La chanson de geste: essai sur l’art épique des jongleurs).
I come back to the question of scale: just how many romance epic texts are there? Rychner only looked at nine. The Wikipedia article on the chanson de geste claims there are “over a hundred” (accessed 8 April 2015). Working with the entry on the “chanson de geste” at ARLIMA, and supplementing it with searches at IA, HathiTrust (my institution became a partner!), La Vie en Proses, along with ProQuest for more obscure titles, I have built a working handlist. It seems that we can say with certainty that there are at least 185 unique texts. There are, of course, so many versions for some texts: different meter, verse, prose, differing lengths, etc. Using my rule of thumb of 60-70 place names per text on average, this means there are potentially 11000-13000 spatial data points for this genre. The corpus will surely become clearer with more research.
I have some existing data from the VMP project from epic-inflected texts, so I decided to try a map. The example below contains some 1800 place names (96% geocoded) from 22 different texts from across the 12-15th centuries. (You can hover over the glyphs and see the place, the text and the supposed composition range).
I have used a choropleth map, with a multicolor ramp, to show progression over time. Dark and light green glyphs indicate that the place name occurs in a text dated closer to 1200, whereas salmon and red glyphs point to a place name showing up around 1450. A note of caution: some glyphs may actually be overlapping with others, masking persistence of the place name.
What I find most interesting about this initial exploration of the data is the flip-flop of the axes of the chanson de geste. The earlier period tends to align on a NE-SW axis of France-Iberia, whereas the later period follows a NW-SE axis. The most obvious guess for why we find such a large scale flip is perhaps a shift from the Charlemagne material and toward localizing epic form for other political (Mediterranean), military encounters. Just a guess.
- How much spatial data can be extracted from the medieval French corpus?
- Deep, Thick, Playful Mapping: a Spatial/GeoHumanities Reading List for Beginners