I think that Trevor already indicated several relevant aspects, so I want to add just a few remarks. First of all, I think it could be helpful to have a look at the extensive literature on translation problems and translation errors in human translation. The topic of errors and their evaluation is at the heart of translation studies and practical translation. Foreign language learning is, of course, related, but translation presents some specific problems (e. g. problems of an extralinguistic nature, terminology and the like) that may be less relevant in language learning. Moreover, to me it seems that research in foreign language acquisition generally is concerned more with errors in language production, but in the case of translating texts from ancient languages, the situation would be the neat opposite. So literature on problems in the more specific field of foreign language reading (especially concerning the understanding of complex syntax and grammatical structures which probably is an important aspect of teaching Greek and Latin) might also help. I am also aware that error typologies have been researched in MT as well. I am not an expert there, but I know of two papers: Vilar et al. 2006, Error Analysis of Statistical Machine Translation Output Ghoniem et al. 2011, An Analysis and Evaluation of English Arabic Statistical Machine Translation of Terminology-Rich Text, http://tia2011.crim.fr/Proceedings/pdf/TIA19.pdf Big language service providers and translation departments usually also have their own error typology and grading system. As for the corpus, it may be helpful to contact the local philology department, as I don't think that translation from ancient languages is done a lot by professional translators. However, in philology, translations are produced to support language learning and test the students' linguistic knowledge (a practice that I find really awkward if applied to modern languages). Therefore the errors marked by the teacher and their evaluation may differ from error typologies in professional translation and the errors in translations from Greek/Latin(/Hebrew?/others?) can certainly only partially be compared to the problems encountered in modern text (see above). So depending on the aim of your application it may be necessary to elaborate on that. Hope this was not confusing.
Best regards, Anne-Kathrin Schumann PhD student University of Vienna
-------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 4521 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20120112/6a8c1aac/attachment.txt>