Vietnamese should have two plural forms

Thank you for your report!

In order to learn more about that topic, I read the following GNU help page:
https://www.gnu.org/software/gettext/manual/html_node/Plural-forms.html

There Vietnamese is especially listed as a one plural form language.

I also checked a Gnome and a Debian package, as these are wide-spread open source projects, and both also use the single plural form:
https://sources.debian.org/src/grub2/2.06-2/po/vi.po/
https://gitlab.gnome.org/GNOME/gnome-shell/-/blob/main/po/vi.po

I have seen your report at the unicode issue tracker ( https://unicode-org.atlassian.net/browse/CLDR-14273 ) and the pull request over at GitHub ( https://github.com/unicode-org/cldr/pull/1774 ) which has not been merged yet.

As I do not speak Vietnamese, I would appreciate if some more native speakers/translators would speak up here to get some "extra confirmation" :-)

I will also reach out to some fellow engineers - maybe somebody knows someone from Vietnam.

Thanks for your patience!

While I agree that Vietnamese should have two plural forms, treating n = 0 as plural is a workaround at best, since

> No demonstrative pronoun in Vietnamese can refer to zero of something.

The given example is not really convincing since the zero sentence shouldn't exist at all as the button should be disabled when there is no update and the description should be reflective of that. I would like to take a look at the .po file or equivalent to provide a more informed opinion.

Regarding the Unicode ticket, I think it should be emphasized that excessive use of pronouns and counting markers sounds unnatural in many cases. Back to the example, the second sentence should have just been for all cases (notice capitalization)

> Để cài đặt, bấm Cài đặt cập nhật.

In other words, the plural rule should be change to facilitate translation, but doing that by itself does not magically improve the translation quality. Translators should be encouraged to come up with good sentences instead of mapping words literally.

[Expired for Launchpad itself because there has been no activity for 60 days.]

> I also checked a Gnome and a Debian package, as these are wide-spread open source projects, and both also use the single plural form:

Practically everyone other than CLDR and MediaWiki originally based their plural rules on gettext. As far as I can tell, this whole misunderstanding stems from a single commit in 2005:

<https://git.savannah.gnu.org/gitweb/?p=gettext.git;a=commit;h=d78c911e16a5d2a15febfa1112c88612a9bb0a0b>

I’ll reach out to Clytie for confirmation.

> I have seen your report at the unicode issue tracker ( https://unicode-org.atlassian.net/browse/CLDR-14273 ) and the pull request over at GitHub ( https://github.com/unicode-org/cldr/pull/1774 ) which has not been merged yet.

CLDR’s maintainers agree with my proposal and consider the proposed change to be a release blocker. However, they also consider it to be blocked by a new feature that would explicitly indicate whether a noun phrase (“1 meter”, “3 meters”) needs to vary by number, or whether grammatical number only affects other parts of the sentence, like demonstratives in Vietnamese. <https://unicode-org.atlassian.net/browse/CLDR-15634> This feature would reduce some redundancy for software that uses just CLDR’s formatting functionality, as opposed to reusing the plural rules for translation in general. I believe the question for Launchpad is more straightforward: as far as I know, it doesn’t come with built-in measurement or currency formatting functionality.

> The given example is not really convincing since the zero sentence shouldn't exist at all as the button should be disabled when there is no update and the description should be reflective of that.

That’s fair, it is a pretty contrived example. It’s hard to come up with an uncontrived example, which is probably a point in favor of going ahead with the proposed change.

The zero behavior is something of a moot point for CLDR, because the project advises implementations to let translators optionally customize the zero message even if the language has an n=1 plural rule. MediaWiki/Translatewiki.net also offers this flexibility, but I don’t view it as a showstopper.

> Regarding the Unicode ticket, I think it should be emphasized that excessive use of pronouns and counting markers sounds unnatural in many cases.

Yes, as a translator, I take every opportunity to omit pronouns and counting markers. But sometimes a UI design gives me no choice: when a user is about to delete multiple files, triggering a confirmation dialog, I want the user to realize, as in English, that clicking the confirmation button will have the destructive effect of deleting multiple files:

* “Xóa” would lose this emphasis.
* “Xóa tập tin” would be potentially misleading.
* “Xóa các tập tin” would be ideal.
* “Xóa (các) tập tin” would be stilted (and alarmist when deleting only one file).

In the zero case, which would disable this button, any of these translations would be plausible.

> Translators should be encouraged to come up with good sentences instead of mapping words literally.

Totally agree, but this presupposes that software structures messages flexibly and allows for sufficient plural forms. One step at a time. It’s enough work as ...

> I’ll reach out to Clytie for confirmation.

Unfortunately, this won’t be possible: <https://mail.gnome.org/archives/gnome-i18n/2015-February/msg00000.html> :-(

It does seem that Phong at least agrees with me that there needs to be a second plural form, and as I said, the CLDR maintainers agree as well. If needed, I can reach out to other translators, such as on the Mozilla and MediaWiki projects, for confirmation.

I understand that Launchpad would like to remain compatible with gettext, so I’ve opened a parallel bug report there: <https://savannah.gnu.org/bugs/index.php?62876>.