(tlg2045.tlg001.perseus-grc2.xml) Large number of corrections to Nonnus Dionysiaca (stoa2045)

[polycrates]

I've been working on metrical tagging for Nonnus, and I've ended up with changes to about 2800 lines (consulting Rouse's Loeb, the source text). What would be an efficient way to share these? I'm using a diff program on plain text versions, so I could share the report, but the whole thing is about 18M. For now I'll post it on my site, but let me know if there's a better way:

https://hypotactic.com/nonnusCorrections.html

Notes:

• For the sake of my own text, I've changed one aspect of Rouse's text: he adds enclitic accents to paroxytone words, e.g. δέξό με. I'm not aware of any modern editor who would accept this, and since both my texts and Perseus' texts are used by students, this seems like something that should not be retained.
• I've changed a very few words that are clearly typos in the Loeb, often revealed as such by the translation (e.g. χέρμα for θέρμα, translated as 'warm'), or by metrical impossibility.
• I made a few fixes to capitalization to accord with the printed Loeb, but ended up putting that job off for a later day.
• There are interesting overlaps between odd errors in the Perseus text and in the online Loeb, though the latter has been much more thoroughly proofed. Who digitized this first?

[lcerrato]

@polycrates
I believe this text was digitized via a Google grant which would mean that there was an initial pass done via rough OCR with some later automated cleanup. The extent of issues corrected by SEDES et al would put it in that particular batch of work.

Some answers:
We wouldn't normally regularize something versus the print, but if this is done, it needs to be tagged as such. A regularized edition (fixing the enclitic accents) might be an entirely new edition, depending on the scope of the change.
Anything that deviates from print should also be marked as such. We would typically add a <choice> tag and/or a note. (A typo is of course fine as a silent fix.)
There was rarely capitalization captured in this batch, although I know SEDES made many corrections to fix proper nouns that were lowercase.
There is no relation between the online Loeb and this work, as far as I know. There could be a common derivation from an Internet Archive OCR version, but I doubt it.

All in all, the best way to integrate changes would be to compare XML to XML. Otherwise, I'd have to check line by line to determine what types of changes we have (typo, variant, stylistic, etc.). Probably not easily managed in a work of this size, but am open to suggestions on that!

[polycrates]

Double accute accents from enclitics are the only thing I changed against the editor's choice. Those should be easy to ignore if you want to keep things simple. Everything else is either a typo in the Loeb itself (rare) or an error in digitization (not rare). When you say typos are ok as silent fixes, do you mean the latter or the former?

[lcerrato]

An error in the digitization or transcription where the print reads differently is a silent correction. A correction that differs from the print would ideally be marked as such.