Bad example sentence for the entry <ⲕⲉⲗ> & Coptic-English text misalignment

[ouromi]

Hi CDO team,

For the entry for ⲕⲉⲗ (type of fish) [https://coptic-dictionary.org/entry/C12068], the examples are all incorrect:

• 06_Romans_11 has ⲕⲉⲗⲓ knee
• 27_Daniel_08 has ⲕⲉⲗϫ ⲕⲉⲗⲓ bend knee/kneel (and also the English translation does not correspond to the Coptic, which is from Dan 6:10)

BW,
Ambrose

[amir-zeldes]

Thanks for reporting! Based on Crum's entry, it looks like this only comes from Kircher, so basically it's a medieval glossary entry, which means we are unlikely to ever have an example sentence with this word in Coptic Scriptorium. I'll flag it as a 'no examples' entry and these examples will be purged on the next update.

I'll open a separate issue here for the translation alignment for Bohairic Daniel.

[amir-zeldes]

BTW @wjhuah this is a good example for a standard Crum reference that can never have a correct example sentence:

Since K = Kircher is a medieval bilingual glossary, items that only have this source as an attestation in Crum will probably never appear in an attested historical sentence.

[wjhuah]

Many thanks, Amir! So, maybe some of the entries in there might not be substantiated either, and I will be careful! Also, I can relate to this example! Kircher is a bit like the 传抄古文 (e.g., 《汗简》). There are some errors, yet it still offers references for understanding obscure terms.

It seems Sahidic avoids this issue, while Bohairic doesn't. I wonder if the Sahidic data is cleaner?

Could we add a verification step later on to check the translations and lemmas?

[amir-zeldes]

Heh, I guess it's a little like Han Jian in that it's an old dictionary of sorts. But the words in question here are not attested at all in a running text (it's not just an unusual spelling or different 'character'). According to Crum, the only attestation of the word ⲕⲉⲗ in Coptic comes from this bilingual glossary (the Scala Magna).

From a practical perspective, this means that such entries (where "K" is the only source in a Crum entry) will never have an authentic example sentence in Coptic Scriptorium corpora, so it makes sense for the example extraction algorithm to prohibit any sentence, even if the NLP tools say that it contains that word (in this case, because they mis-segmented the word for "knee", ⲕⲉⲗⲓ).

[wjhuah]

Thanks for breaking it down, Amir! I see the difference now:) Excluding these entries is undoubtedly the safest solution during the current procedure.