Community
June 28, 2010
TEI, one letter at a time
The project I was working on at DEMiC was a body of correspondence between various figures active in the middle years of the century within Canadian modernism: not only authors but editors, publishers and other figures who participated in the dissemination of modernist aesthetics and artefacts. The letters form part of a volume which currently has the working title Enduring Traces: Correspondence from Canadian Modernism’s Archives, and which I’m still gathering the archival material for. (So if anyone comes across some intriguing correspondence tucked away in an archive somewhere that deserves to come to light, I would be glad to hear about it …)
Below is a little section from one of the letters I was working on, which is from Earle Birney to Alan Crawley:
And here is what it looked like once I’d tweaked the style sheet to work with the letter format:
[If the size is too small to see, click on the images to view them at full size. I haven’t posted them as large images in this post because they crash into the menu at the right.]
This is not a particularly complex document to encode, but all the same, a number of issues came up during the coding, including the following:
– How do you code the date of a letter when its year of composition is unclear, given that TEI insists that you supply at least a year? (In this case it isn’t difficult to figure out what year the letter is written from other contextual evidence, but that isn’t always the case. In a print edition you could presumably just put [1945?] or something similar, and move on, but TEI demands a year.)
– Would readers prefer a text in which wasnt and wont are silently corrected to wasn’t and won’t? Should I bother giving readers the option of toggling errors such as these on and off? (I rather like the way Birney wrote it, but if the original is supplied then it might cause difficulties when text mining down the line, when wasnt and wasn’t might be treated as different words.)
– If some authors italicise or underline titles of books and journals and others don’t, how should this be standardised across different authors? What do readers of an edition value more: the ability to see at a glance which publications are under discussion, or an individual author’s bibliographic habits?
– The xxxx that is used to cross something out after the word moods a few lines from the bottom was a problem that neither Julia nor Martin could solve without the use of XLST. I wanted to be able to include the word that was obliterated in the XML, so that it could if necessary be retrieved, but TEI could not manage it. The compromise was to represent the xxxx as it appears in Birney’s letter, but the way I’ve done it means that the bit of information about what exactly was crossed out has been lost.
As you can see, these are questions which are not just about the mechanics of digital editing but also about the theory of textual editing. I have the feeling historians rather than lit scholars are the ones to ask about conventions for editing documents such as the ones I’m working on … but then they will also see the significance of these letters in quite a different way, so perhaps I should not be seeking their advice …?
Looking at the XML now, I see so many things still to code – there’s not yet an entry in the personography for Jean Crawley; the ‘two letters’ Birney refers to should have a reference, as should the issue number of Q’s Q under discussion, etc etc (and these are quite straightforward and easily tracked-down things, compared to the palimpsest of allusions in the excerpt Vanessa quotes from By Grand Central Station, or the question of how to evade the hierarchical nature of TEI that Bart poses). Much more to do, and this is just one section of one letter!