Not enough text
For two years now, I have kept repeating “We don’t have enough text” (meaning in our digital edition).
And it is true. There are a lot of things, especially the big-picturesque ones, that we still cannot look into seriously because we don’t have enough text for that. When I met with my new colleagues from the IMS two weeks ago and they gave me an overview of what could possibly be done with the linguistic methods they are using, I had the opportunity to yet again regret not having more text at hand. If we had all of Boeckh’s correspondence, we could look at his use of the word “Philology” over time and in the different institutional contexts in which he has been writing (as a professor, as an Academy member, as a member of this or that commission, as a friend, etc.). We would be able to see in combination with which adjectives he uses the word over time and according to the context. On the less scholarly, more literary side of our corpus, we could also do some brilliant tracking of ideas and motives – that is, if we had enough letters of different writers and publishers. Somehow, we are still stuck at some middle range between case study and quantitative analysis. If I had only one wish, it would be that we could find a way to reach the next step by the magic of the concept of “intellectuals” – the one that would allow to encompass all of our corpus, to host it under one big umbrella.
But even if I could figure out a coherent protocol on the basis of that premise, there would be some extensive fine-tuning to be done. First, the corpus is in two languages (German and French), so even if we ignore the shorter bits or parentheses in Greek, Latin, Italian, English, etc., there will be a basic linguistic problem in finding a lemma common to all the texts. Within the German corpus, we have the problem of unmodernized spelling, which makes the texts linguistically incoherent, even under the feather of a single author (Boeckh is mostly coherent, but Chamisso is not at all).
So I turned to other corpora, corpora in which spelling and language variation should only marginally bias the analysis. When work on these corpora will have shown me what is possible, it will be easier to make our digital edition evolve towards analyzable text (in mass and in quality). I might even get old transcriptions out of the drawer, those that are nowhere near being editable. But as I have learned, there is a lot you can do with text that is not really presentable or humanly readable, even things that make your eyes shine with excitement and curiosity. Mine, at the very least.