> The two fields do different things.

I know that. My point is: Generate **both** `langid = {japanese}` and `language = {japanese}` in the bibtex file from the **single** `language = japanese` field of Zotero.

Please report BBT issues on https://github.com/retorquere/zotero-better-bibtex/issues

According to the biblatex manual, there is no field language, just langid/langidopts. language a bibtex field (or more specifically, a babelbib field). If you have a biblatex style that needs language, I can take a look at that.

For the short term, you can put this in your preamble: \DeclareSourcemap{
\maps[datatype=bibtex]{
\map{
\step[fieldsource=langid,fieldtarget=language]
}
}
}

Looks like you're right, apologies. Please open an issue on github.

Turns out BBT already has this, I just forgot:

https://retorque.re/zotero-better-bibtex/installation/preferences/export/index.html#export-language-as

Thanks!!!! I confirm that "export-language-as"="both" solves my problem: the generated reference list now has "[in Japanese]".

Now, would there be any harm if "export-language-as" were set to "both" by default?

Actually, that was the reason why I posted here instead of posting to the BBT github: in order for other users to learn from the discussion.

Given they're both in the biblatex manual I've just made them fixed for biblatex and only configurable for bibtex. Will be in the next release.

Which can be done using a tex.language entry in extra -- or by setting BBT's language-field handling to the non-default "both".

If the original writer of the metadata means merely to indicate the language the article is written in and if Zotero restricts its meaning to just formatting instructions, it's Zotero which has to be fixed.

I mean -- BibLaTeX has two language fields. Zotero has one. To the extent that two fields with different functionality are needed, yes, Zotero needs to get fixed. This is a bit of a tricky issue because Zotero's general approach to metadata is a bit different from Bib(La)TeX -- in BibLaTeX the general approach (though not without exceptions) is to only include things to be cited in the metadata. Zotero's approach is to generally prefer completeness.
That's a challenge for something like "Language" which, typically, you only want displayed in some very select cases (you certainly wouldn't want every English publication to say "in English", say).
This is also the reason this would 'break' biblatex styles -- they'd suddenly be displaying the language of pieces when, in most cases, that's not actually desirable.

This rarely comes up in the Zotero context, so it's not, I think, a high priority to address, to be honest.

I guess I don't understand the question? As I say above, Zotero is designed to always prefer complete data, so including an en tag is strictly better than not. It's just that this means mapping Zotero's language field to the biblatex field of the same name is problematic -- I'm going to assume this latter part is irrelevant for your users.

@nickbart
> Even if BBT continues to export Zotero’s language fields to biblatex’s langid fields, and never to language, which I maintain is the only sensible way,

Please define "sensible". My question has been: What *practical harm* would result if BBT exports Zotero's `language` both to biblatex's `langid` and `language`? If there is no practical harm, then do it, until Zotero is "fixed". Zotero is just a tool. What matters is the result in practice.

First of all, this is the workflow if you don't use Zotero:

1. Download the bib database from the publisher's website.
2. Use it with biblatex.
3. The reference list includes the wording "[in Japanese]".

If you use Zotero and BBT, then the reference list doesn't include "[in Japanese]". Obviously, it's Zotero+BBT which needs to be fixed.

**Zotero is just a tool** to help the workflow. It's **not** a standard. That means, it's Zotero who is breaking the workflow.

What Zotero *should* do is to have two language-related fields: one to indicate the language the article is written in and the other to indicate how the title should be formatted.

Until that happens, BBT *should* export `language` both to biblatex's `language` and `langid`, not to break the workflow. This is a temporary workaround.

---
1. There are too many metadata standards and publishers' attitudes toward standards are generally lax.

2. The Zotero developers don't have the power to convince the publishers to use Zotero's strict standard.

3. Therefore, Zotero or BBT has to adapt to this word's chaotic reality.

My question has been: What *practical harm* would result if BBT exports Zotero's `language` both to biblatex's `langid` and `language`? If there is no practical harm, then do it, until Zotero is "fixed". Zotero is just a tool. What matters is the result in practice.

It seems to me @nickbart has described this at length above. It would cause output to appear in the rendered bibliography of which the user has not expressed intent it should be there.

BBT, for better or for worse, already allows what you want, it's just not the default behavior.

If you use Zotero and BBT, then the reference list doesn't include "[in Japanese]".

what you are arguing is that import should make better choices. That is something we can talk about.

Obviously, it's Zotero+BBT which needs to be fixed ... Until that happens, BBT *should* export `language` both to biblatex's `language` and `langid`, not to break the workflow.

You talk about "obviously" and "the workflow" as if your view on this is a universal standard. Given that BBT has existed for 11 years, and this is just now being debated, it doesn't look like your opinion is widely shared.