Towards a Treasure Legend Corpus

Thanks to the organizers of this years meeting of the International Society for the Study of Contemporary Legend, which thankfully has the acronym ISCLR, I have a reason to explore text analysis on objects to which I am more accustomed. And so, for a moment, on the one hand, the work on the intellectual history of folklore studies using topic modeling and other forms of algorithmic / quantitative inquiry is paused. On the other, the necessary revising of the boat book manuscript is not yet cranked up, as it will be in two weeks, when the semester ends and my editor starts counting down the days to my self-imposed deadline of May 31. (Okay, Craig, that’s really going to be June 31, just so you know.)

This study of legendary is something of a bridge. It’s addressing a topic that I thought would be a part of the book, an examination of traditional folklore materials for an understanding of how people in south Louisiana understand the landscape, but using methods that I have learned since starting that project and that really point to the next chapter in my research and the next phase of my career, wherein I plan to focus on narrative studies, especially computational / quantitative / whatever forms of narrative studies. (And I’m especially thinking about looking into doing some work with science fiction, because it could be a lot of fun, and I think some of those authors would be more amenable to such things.)

I am, in particular, interested in legends about treasure, since they would seem to focus our attention on where a bounty might be collected from the landscape. I am drawn to these tales because in collecting a few narratives of my own, and then looking at published legends, I realized that a number of them located the treasure at the intersection of land and water, which seemed a really compelling point, given that I was (am) writing a book about boats that go on land and water, which is really about trying to understand how people think about the Louisiana landscape from a practical point of view.

Thanks to the generosity of Maida Owens, who copied to a thumb drive her collection of texts from the _Swapping Stories_ books, I had a nice collection of legends with which to begin. Ten texts, numbers 157 to 165, to be precise.

(Please note that I cannot, at any time, make these texts available. I was able to get this copy from Maida, with the generous permission of Craig Gill to make the transaction, with the proviso that I would not make the texts available — I can run any kind of analysis you might imagine on them and report those results, but not the texts themselves.)

I picked up four more legends from Barry Ancelet’s _Cajun and Creole Folktales_. And, following some suggestions based on what Ancelet chose to include in his work, I have about two or so tales (narratives) to contribute from my own work.

That makes 16 folkloric legends.

Where do those other texts come from? Well, it turns out that there is a fairly diverse community of treasure seekers and thus also an interesting collection of websites serving the treasure-hunting market. Most of the stories I have chosen to include were found in posts in treasure-hunting forums, like the ones hosted at [TreasureNet][]. A few of the stories were seemingly front page items from such sites. That is, these stories had more of the authorial framework of “I wrote this.” I would not have considered them at all, but so much of what appeared with a byline looked exactly like the kinds of things that were found in the forums: which themselves were, in most cases, coped from elsewhere. In a number of instances, as a matter of fact, the material in the forums appeared to be things transcribed out of old newspapers or locally-published books that were no longer otherwise accessible.

And so 16 of those texts.

Overall, the current collection stands at 32. Not a huge number, but it’s a start.

What am I going to do next? A couple of things.

On the computational side, I want to:

* run word frequency analysis on both the individual texts as well as comparatively across the two groups. (I want to call them something like “folklore” and “digital” but that’s a terrible distinction.)
* perhaps do some form of PCA, to see if I can’t find some clusters (I think this is too small a collection for topic modeling, but I might give it a try).
* use something like Wordij to compare word collocation across the whole collection as well between the two groups.

On the narratological side, I want to:

* break all the texts into component clauses to see if time and space are managed in a similar fashion and to compare the amount of description, narration, and exposition contained in each text and across the collection *and* the two groups.
* see if there is any kind of reliable morphology.

I’ve given myself until the end of the semester to do what work I can. Once the semester ends, the race is on to get the boat book wrapped up and off its publisher, the University Press of Mississippi.


Leave a Reply