Exercise 5: The Dragmaticon and Voyant Tools

In this exercise, I was tasked to choose some texts related to the final project and analyze them using Voyant Tools. In this exercise, I found many limitations, but also a few interesting possibilities. The current project may not be especially suited to Voyant Tools’ particular methods of analysis, because of the language and the paucity of data. However, that is as much a problem with the data as it is with the tools. The tools are very robust and impressive, to be sure, but there’s (as of yet) no accounting for the peculiarities of works like this.

The Analysis

After some difficultly, I managed to extract the transcription of the Dragmaticon from T-Pen, but was unable to analyze it in Voyant Tools for an unknown reason (most likely the format of the data). Ben reformatted it, and I finally had the Dragmaticon in Voyant tools. Well, not really. I had a transcription of the text, which amounts to seven individual pages, separated throughout the text. The transcription itself is made up of our best guesses at the obscure hand and tricky abbreviations of the original. I also had to clean up the data by removing any brackets around abbreviations, and by reuiniting words split by page-wrapping. After the application of a few stop words (cum de enim esse est et hoc in que supra ut), I found that the most common words were substantia (9); dicitur (8); dux (8); philosophus (8); and sol (8). This shows us a few things we already know; the book is a dialogue between a duke and philosopher, and concerns the movement of celestial bodies. While I must confess I didn’t look at all the visualization tools Voyant had on offer, it didn’t seem like I could get much further from there. The phrases are always interesting to see, though; the phrase “si terra plana esset” occurs twice, and reveals how that page in the book concerns William’s proof that the earth is round.

I then decided to take a look at the first book of the Philosophia Mundi, William’s controversial work which was revised to make the Dragmaticon. I retrieved the transcription from here. After another cleaning of the data, we can see that (36); elementa (33); aqua (32); sapientia (22); and spiritus (22) are the most present words. This piqued my interest in the Philosophia Mundi, as I got the impression from these content words that the work addresses similar content to the Dragmaticon, but in terms of elements, wisdom, and spirit–that said, I’d have to do actual close reading to figure out how those words were being used. I wonder, also, why the choice was made to make the Dragmaticon a dialogue when the Philosophia is not. It’s also interesting that the phrase “divina voluntatem,” or divine will, appears 4 times in the text, especially in the context of its supposedly heretical contents.

It was somewhere around this point that I realized how the tools might be unsuited for the exercise. The particular algorithm Voyant uses seems to be very keyed towards English (texttexture as well), and for good reason–most times in English, a word is a word is a word. However, Latin’s changing word endings renders this approach much more difficult. For instance, the tools found 36 instances of “terra,” but also 10 of “terram,” and counted them as different words. Also, to reiterate a prior point, we don’t have very much of the Dragmaticon transcribed, and what we do have is selected from different areas of the book (sadly, Stanford hasn’t digitized the whole text yet.)

These visualizations were cool to look at, and I will definitely keep exploring them, but they don’t do a great job representing the text. They can tell you major themes of the books, or common phrases that might give something away. However, something is lost here; the binding of the Dragmaticon, the illustrations, the abbreviations. Still, I think Voyant Tools are nifty, and I hope to see more specific development of digital humanities tools in future, hopefully to the point where we can teach machines how to better read Latin.


08 March 2017