language of doc vs language of entry

mirjam.amsterdam

I always understood "language" to mean the language of the document. Since the upgrade however, the minute I put in Heb or Yid as language, the text direction of the title field or the date, for examples, run from right to left. Even if I put in a Latin-lettered text, running from left to right. The program assumes right to left without checking the fonts I am using. Could this be adjusted?

https://s3.amazonaws.com/zotero.org/images/forums/u4392769/blxzxmxlctz0q9qurns9.png

https://s3.amazonaws.com/zotero.org/images/forums/u4392769/xtdrpv99cekomecqhihw.png

https://s3.amazonaws.com/zotero.org/images/forums/u4392769/v94pkdb6tt7cb5oemxux.png

mirjam.amsterdam

another example: to the left there are dots, the first words of the title is invisible. The alphabet is Latin, running from right to left. Only Zotero decided to have it running from right to left because the language is Yiddish.

https://s3.amazonaws.com/zotero.org/images/forums/u4392769/otxa3hycn4n43sul3oun.png

nickbart

To clear up one misunderstanding, your view, ‘I always understood “language’ to mean the language of the document.’ is not actually correct: In Zotero, the language field specifies what you call the ‘language of entry’ and what is often called the language of the item’s metadata.

So if you put ‘heb’ in the language field, Zotero assumes that the metadata are in Hebrew, and also in Hebrew script – which would seem to explain the effects you're seeing.

I cannot say whether the Zotero developers will consider developing a solution that allows Zotero to account for the script used in the various UI fields. However, it appears that the following approach is already effective with the current version of Zotero (I’m using 7.0.13 on macOS) and may also enhance clarity in any case:

What you should do if the title has been transliterated to Latin script is to use a language tag that expresses this fact.

According to RFC 6497: BCP 47 Extension T - Transformed Content (https://doi.org/10.17487/RFC6497), the formally correct way to express this would seem to be ‘he-Latn-t-he-Hebr’. But at least according to my tests, something as simple (but of course, non-standard and non-interoperable) as ‘heb-t’ would seem to work as well.

I'll leave it to the Zotero developers to look at the potential difficulties if the ‘language’ field is set to ‘heb‘ meaning ‘Hebrew in Hebrew script’, and some fields, e.g., ‘title’ actually do contain ‘Hebrew in Hebrew script’ but other fields, say ‘date’ are supposed to contain content written in Latin script.

mirjam.amsterdam

Thanks for explaining this!