Digitizing Hervor - Part 2: Approaches & Methods

In the last post, I introduced the Digital Hervararkviða, the labor of love I have undertaken in my quest to come to a better understanding of the ancient Germanic world, it's languages and literature. In this post, I'll talk a bit more about the methods I'm using to create the digital edition of the poem, and why what I'm doing isn't just redundant with previous critical texts of the poem and of Hervarar's saga.

TEI Encoding

First, I'll need to explain a little about TEI encoding. The Text Encoding Initiative (TEI) is "a consortium which collectively develops and maintains a standard for the representation of texts in digital form." Essentially, TEI encoding is a set of standards for taking a text--say the poems of Emily Dickinson or a sixteenth-century Polish fencing manual--and turning them into a robust, searchable XML document which can then be displayed in a number of different formats, inserted into a database in order to perform corpus linguistics analysis, and etc. Basically, once the text is in this standardized XML form, you can do anything with it.

[If you're reading this on a web browser there's a pretty good chance you have an idea of what XML is, but if you don't, you can go and educate yourself here.]

The Menota Project

Now, there are a lot of TEI encoding projects out there (including some which are pretty useful for medievalists and classicists everywhere, like the Perseus project), but the one that is important to this project is the Medieval Nordic Text Archive (Menota). Menota is basically a network of institutions working to do for the Scandinavian languages (mostly Old Icelandic and Old Norwegian at this point, though there are some Old Swedish texts as well) what the Perseus project has done for Greek and Latin. Menota has laid out a process as well as a series of standards for encoding these Old Norse manuscripts at three different levels of text representation:

Facsimile: At this level, the manuscript is transcribed character by character, line by line, retaining all abbreviations and allographic variations found in the original text. This is the closest level of reading possible short of actually handling the manuscript.

Diplomatic: Certain allographic variants are normalized (for instance, the rotunda "long s" is frequently changed to the modern "short" s. Abbreviations are also expanded, and the expansions usually italicized.

Normalized: Spelling and word forms are standardized to conform to grammars and dictionaries for the language in question. For our purposes, this means altering the words to match what you'd find in a dictionary like Zoega's, or a grammar like A New Introduction to Old Norse. This level is useful for newer students of the language who might be confused by the inconsistencies in orthography and even morphology which might be found in an unedited manuscript. It is also the level at which most critical editions tend to be produced. Old English texts, for instance, tend to be normalized to a particular West Saxon literary style, which can often lead to the misleading impression that writing in Anglo-Saxon England conformed to a fairly homogeneous standard.

How normalized spelling is determined for a dead language is a subject for another post, but for now it's enough to know that this...

The beginning of Hervararkviða from G. Turville-Petre's critical edition of Hervara saga.

...is not the same as this:

The same passage from 74r of the Hauksbok manuscript

The Birth of the Digital Hervararkviða 

In a way, that gap between the manuscript and normalized editions is what birthed this entire project. I knew I wanted to do something with the Hervararkviða or with Eddic poetry in general; I knew that pretty much all of the Eddic poetry out there already had at least one good critical edition along with multiple translations (maybe not as many translations as Beowulf, but the Poetic Edda has been done lots of times). It was while chatting with one of my professors at Signum, Prof. Paul Peterson, that I stumbled upon the idea for this project. Paul suggested a TEI text and, knowing I was interested in the Hervararkviða, pointed out that high resolution photographs of the entire Hauksbok manuscript were available online. 

Curious, I began trying to read the manuscript, very quickly realizing I only thought I knew how to read Old Norse. This manuscript was full of strange letters I did not recognize, spelling variations I had never seen, and more abbreviations I had ever seen in a single text (and I work with the US Government, where abbreviations are our stock and trade). I had always known that the normalized text was not quite the text I was encountering in a textbook or reader, but until I actually studied the manuscript for myself I had no appreciation for just how wide that gap could be. 

Bridging that gap, then, for other students of Old Norse, is one of the primary goals of the project. And the Menota Project's encoding standards allow us to do that by encoding each word with readings on all three representation levels.

XML in Action

Let's take the word berserkjanna, the genitive plural of berserkr. This word appears in the prose introduction to the Hervararkviða. In the manuscript, it's actually written as: 

Which in my facsimile I have transcribed as:

I have to render the transcription as an image for this blog post, because otherwise there are characters which won't show up unless you have certain medieval fonts installed. You'll notice that there's a little squiggly line over the b. This is an abbreviation for er. The funky-looking "f" is really a long s. You'll note there's an i here instead of a j, and a bar over the n to indicate a second n has been omitted. I'll talk in greater detail about Old Norse abbreviations in a future post, but it's important to understand that it was fairly standard to use this many abbreviations in a word.

On a diplomatic level, we would render this word: berserkianna. The s has been standardized, and the abbreviations have been expanded and italicized. 

In our normalized edition this word would be rendered (as Turville-Petre in fact does): berserkjanna. Note the i has been changed to a j here, because that's in keeping with the way you'd find the form listed in a grammar or dictionary.

A properly formed Menota XML document allows us to encode all three levels in a single word, like so, with the <me:facs> tag corresponding to the facsimile-level reading, and so on:

<w>
   <choice>
    <me:facs>b<am>&er;</am>&slong;erkian<am>&bar;</am>a</me:facs>
    <me:dipl>b<ex>er</ex>serkian<ex>n</ex>a</me:dipl>
  <me:norm>berserkjanna</me:norm>
   </choice>
</w>

Of course to do this, it means you're encoding every single word three times. Which is what I've done. For the entire poem. Yep, it took me a while. And I'm still not done.

Once it's all in there, though, you can use an XSLT stylesheet (a topic for another post) to render any single reading level. That means that every Menota XML document has the potential to contain readings on all three levels, not to mention a wide variety of information about each word--lexical citation forms, base forms, morphology, syntax, the type of word (noun, proper name, verb, etc.). Although this kind of secondary information wouldn't typically be displayed by your style sheet, it might be accessible via a specialized web page or app. So in the final edition of the Digital Hervararkviða, the student should be able to mouse over a word and see its case/number (for nouns) or tense/mood/number (for verbs), as well as the lexical citation form that they can look up in the dictionary. The student should also be able to toggle between the facs/dip/normalized views at will and compare them to photographs of the text. And of course, since all of this information is being stored in XML, it has the potential to be used in a database for performing corpus-level analytics.

In the next post, I'll talk about the process used for creating the manuscript facsimile. We'll jump into the deep end of the pool of Old Norse paleography, and I'll recount the roadblocks and frustrations I encountered as I attempted to figure out just how people decided how things should be put down on paper more than 600 years ago.