On Being Translucent

Submitted by Sarah Stanley on Fri, 12/12/2014 - 21:26

For my work on the Early Caribbean Digital Archive, I've helped to create some interesting elements, meant to tag various phrase-level and division-level textual elements. The ones that my fellow encoders and I were most interested in were a subset of elements meant to describe commodities, food, flora, and fauna. Our original intent was to use these tags to track the circulation of commodities in various stages of production, sale, and use. We based our markup practices in Marx's theories of the commodity, specifically tagging items when they are discussed without reference to their use-value.

We hoped that, using our placeography, we would be able to show various relationships between commodities and place. We had an interpretive group (<interpGrp>) specifically devoted to production stages, so that we could differentiate between when commodities were being produced or sold in given places. The markup itself was very complex, and the encoding team ended up postponing the encoding of some data until later in the project. For example, we decided to hold off on explicitly associating each commodity with a place, by using a pointer to the placeography. Since we still weren't certain about how we wanted to use the data we were collecting, we had only recorded production stage (where applicable) and other basic descriptors of the commodity.

Since our markup team has been treating our use of the TEI as primarily exploratory (rather than descriptive or procedural), our commodity encodings have varied from document to document. Usually, our understanding of the commodity in our discrete texts came from spending a lot of time reading each text. This is why I decided to only focus on James Grainger's The Sugar Cane. It's the only text that I've gotten to work with in-depth.

I decided that the best way to show the relationships between commodity and place would be to create a bi-modal network graph. I could have two types of nodes (one for commodity and one for place), and each edge would have an attribute, with a value that specified its production stage. Since we never found a way to explicitly link each element to a place, I had to read through the text and hand-create an edge table (in another universe, creating both my nodes and edges tables could have been created through basic xslt).

After many days of creating tables, I finally got to the end of my document. Upon importing my data into Gephi, and messing around with the visualization for a bit, I was left with this:

Image removed. 

Unfortunately, there wasn't nearly enough data to make an awesome opaque visualization. It was all so... translucent. I've recently come to the conclusion that a visualization that allows for you to clearly see node labels is probably not a good visualization. I mean, what even are "cane rats" and why are they there?? The most I could really do was force the edges for production, transport and sale/consumption to display in different colors (blue, red, and green, respectively). It was sort of neat seeing how certain geographic areas very roughly clustered together, but that isn't really a finding.

So I went back to the text to see what else I could do (I mean, I've spent a lot of time collecting <commodity> data. There had to be something). And as I was thinking about the process of data collection, I remembered something interesting. If you look back at the graph, you can see (very faintly) that there are tiny blue lines (production edges) coming out of England, the United States¹, Scotland, and Newfoundland. These edges are all made up of commodity/place relationships that I found at the end of the poem (specifically, Book IV). I would have assumed that Europe and the United States would have largely been consuming products produced in the Caribbean, but in this portion of the text, slaves in the Caribbean were meant to consume products produced by paid, (presumably) white laborers.

This got me thinking about my post on the ECDA's use of <distinct>. In it, I make the claim that the text is anxious about linguistic influence and contamination resulting from European interactions with the Caribbean colonies. Looking at the distribution of commodity/place relations across time, I would argue that the text attempts to move from a model in which the colonized influences the colonizer (e.g. the sale of Caribbean sugar in England), to one in which the colonizer influences the colonized (e.g. slaves are clothed in linens from England).

Now that I've begun to look at the relationship and commodity in a different light, I have decided to play around a bit with Neatline as a tool for visualizing my data. It's a perfect way to think about the commodities and places I've already collected, as it allows show information across time. You can find the beginnings of my Neatline map here. For the time being, I've needed to use the "year" as my base unit on SIMILE Timeline, but each number is really meant to represent a page. I'm currently trying to work out how to deal with commodities that are mentioned frequently throughout the text (i.e. sugar cane). The tag function seems to be helpful for clustering commodities that are referenced in different parts of the text, but I'm still figuring out how to make those tags useful. My hope is that this Neatline exhibit will better represent the function of commodities in the text by showing how it deals with commodities over time.


¹ The Sugar Cane was published a few years before "The United States" even became an entity, but I lumped them together, since having separate nodes for Pennsylvania, Virginia, and one of the Carolinas was probably less-than-helpful.