In honor of the New Chaucer Society conference going on in Reykjavik right now, I’m going to bring back network graphs to see how Langland and Chaucer might network together.
Ok, well, if you’ve been reading this blog at all (which maybe you haven’t, and that’s ok–here’s what you’ve been missing), you already know that they really don’t. Nevertheless, let’s put some Piers and Chaucer network graphs side by side anyway to compare.
In an earlier blog, we looked at the Piers Plowman corpus as one big networked graph. As mentioned in several previous posts, this kind of network graph displays relationships of material co-occurrence. That is, every line on the graph (an undirected “edge”) connects two nodes. The nodes it connects are two works (or in some cases, categories of works that needn’t be listed individually), and these works (as nodes) are connected because they occur side by side in the same manuscript.
If you see a large group of nodes all connected to each other, then it’s a safe bet you’re looking at a single collection–a compilation manuscript that contains all of these nodes. On this graph, the big cluster is the Vernon, which is mentioned both in the blog where this was originally presented, and in the Vernon special edition.
The second major way that information is conveyed on this graph is through the size and thickness of the nodes and edges (respectively). The larger a node is, the more often it occurs in a corpus. The same goes for edges: the thicker they are, the more often this co-occurrence (the two works together in one MS) happens.
So, to get back to the Chaucer part (which I know you were all waiting for), let’s compare the Piers graph with the one and only work of Chaucer’s that it ever appears alongside (and even then only in the anomalous, conflated and “rejected” manuscript Hm114):
TROILUS AND CRISEYDE
What we see in this data visualization network (DVN) is actually relatively simple. Troilus is represented by the larger center node (the focus node of each graph will always be red), and around it you can see its co-occurrence with a modest number of works largely grouped into four moderately large collections, and a few smaller collections. Certainly some Troilus MSS contain only T & C, which makes the node grow much larger than everything it’s connected to.
Note also, that it is connected to three nodes representing some of Chaucer’s other major works. In all the graphs the 6 long works (if present) will be highlighted:
- The Canterbury Tales
- Troilus and Criseyde
- The Parlement of Foules
- The Legend of Good Women
- The House of Fame
- The Book of the Duchess
For simplicity’s sake, I have only made DVNs for these six works (and a master), and **I have only included the other MIDDLE ENGLISH verse works** in the MSS (for two reasons 1. I’m not an expert in the Chaucer MSS, and without researching each MS I can’t tell you what else is in each and 2. I used the DIMEV for a quick and dirty breakdown, so if it wasn’t there, it isn’t here).
In addition to highlighting other major works of Chaucer in yellow, I have also made sure to locate Piers Plowman (in purple) and Gower’s Confessio Amantis (in green–done for the delight of Jonathan Hsy and to allow us to see all these Ricardian poets in relation to each other).
So, ok, Troilus has a pretty small corpus, and it’s network is a bit smaller (perhaps even proportionally so) than the Piers network, so let’s compare something that’s witnessed a little more similarly. There are 65 MSS of the Canterbury Tales that were “intended” to be complete, which is much closer to the 52 of Piers than the 20-something of Troilus.
THE CANTERBURY TALES
This looks more similar to the graphs we are used to with Piers. There are several sizable collections, one of which dominates the view (CUL Gg.4.27, which seems to have collected the most Chaucer poems in one MS and will thus appear on multiple graphs). Note for a moment here that we have a couple other colorations on this graph, mostly because I am absolutely fascinated by who the Canterbury Tales seems to circulate with.
In the case of the CT, not only does it not circulate with two of the three other “Ricardian Poets” (Gawain-poet or Langland), but it very rarely circulates with Gower (twice) and doesn’t even circulate very often with many other works from Chaucer himself.
And while one may be tempted to say, well maybe that’s because it was so long, just the sheer number of other nodes on the graph should force us to think otherwise!
Instead, we notice that the Canterbury Tales circulate primarily with Lydgate, and also occasionally with Hoccleve. So, the CT seem to have a distinctly 15th-century flavor about them (or at least the MSS they’re collected in). While there is a lot of Lydgate throughout the corpus, it seems especially prevalent here.
But buckle up. Here comes MUCH MORE Lydgate.
THE PARLEMENT OF FOULES
Yeah. Just…yeah. I mean, bloody hell. That graph is such a mess it almost doesn’t tell us anything. I mean, in some places there are so many nodes that titles of works they represent are completely illegible. But, I think it’s important to point out that we can clearly see that the Parlement circulates with a lot more works in far fewer manuscripts. Moreover, of the other major Chaucer works, it least frequently occurs with The Canterbury Tales.
Second, I will just point out that the shorter of his major works do tend to circulate more frequently in collections, and the compilations they circulate in tend to be larger collections than any of those his two longest compilations circulate in. This, however is not an inverse proportion. I.e., it is not the case that as the poems get shorter, the compilations get bigger and more frequent. The Parlement of Foules has, by far, the largest corpus of other Middle English verse accompanying Chaucer’s poem (and if you can find it, the Confessio is in there twice).
In a close second, though, is
THE LEGEND OF GOOD WOMEN
The Legend of Good Women DVN clearly shows a few large clusters, and then a few smaller clusters that dominate an even smaller codicological landscape (with fewer manuscript testimonies). Interestingly, though, LGW is the only one of the six major works to circulate with all of the other five large works.
The last two of these, The Book of the Duchess and The Hous of Fame, have considerably lower circulation (in only four and three manuscript witnesses brought up by the DIMEV). In fact, their circulation is so infrequent, and it always occurs in conjunction with at least one of the other major works (regardless of which work you’re talking about), that their graphs (DVNs) are entirely subsumed by the larger graphs above.
Just to get a sense of what they circulated with in particular, though, let’s have a look.
THE BOOK OF THE DUCHESS
In terms of the number of nodes, the BoD is giving the other shorter works a run for their money. You can clearly see from the butterfly shape of this graph, though, that it’s mostly because of the poem’s place in only two major compilations.
THE HOUS OF FAME
In contrast, the HoF DVN clearly displays three distinct compilations–one of which is very large, which means HoF (like BoD) absolutely never circulates by itself. Three MSS, three compilations, three clusters.
So, just in case that wasn’t enough information for you, let’s pull in one last network graph of THE ENTIRE CHAUCER CORPUS
This is a pretty neat graph in that it allows you to see just how many co-occurrences there really are in a absolutely bloody HUGE corpus of manuscripts containing just complete (or near complete) versions of his major works. If we started to add in fragments, or MSS with just individual tales (or a handful of tales) it would get even crazier!
To know anything more specific, though, we have to start slicing the data down into more manageable chunks, like I’ve done here by splitting up the corpus according to which work the network centered around. By doing so, I’m essentially selecting which part of the overall corpus to focus on, which will bring certain relationships into focus by allowing other things to fall out of focus. If you want to learn more about why we have to do this kind of slicing, and then superimpose all the slices to draw a conclusion, be sure to check out this explanation.
So, the big conclusion?
Clearly, Chaucer(‘s works) were a lot more well-travelled, were in relation to a lot more works, and occurred with a really different kind of corpus than Piers did.