Page MenuHomePhabricator

Update Levantine Arabic language
Closed, ResolvedPublic

Description

Hi,
I created a pull request to add ajp (South Levantine Arabic) to the list of languages: https://github.com/wikimedia/language-data/pull/165

I did this based on advice from the Moroccan community (ary, another language from the Arabic macrolanguage family) who created a similar task in November 2019: T237672

There's an ongoing discussion on the English Wikipedia about the problem of font and directionality for Levantine Arabic.

Please let me know if there's anything else to do.

Event Timeline

Hi, I think this does not need a ticket here but would rather be in scope for https://github.com/wikimedia/language-data/issues

I created a pull request on GitHub: https://github.com/wikimedia/language-data/pull/165

But I think this will only solve the problem of directionality, not the font issue (text tagged as ajp is hard to read on Windows).

Noting that the PR in that repo might be useful, but probably won't have the desired effect. MediaWiki itself doesn't use that, but ContentTranslation does

Depending on your desired result, you may need patches to UniversalLanguageSelector, and also MediaWiki core itself to add the language code...

@Reedy the desired result is to have ajp setup as "right to left" by default.

(and maybe as well to decide what its font would be, but the first issue is directionality)

Is there anything else to do regarding directionality?

I also opened a ticket there to create a new ajp locale: https://unicode-org.atlassian.net/browse/CLDR-14884

I think a language file in MW core to set that is one step then…

Thanks. How do I add/edit a language file in MediaWiki core?

Thanks. How do I add/edit a language file in MediaWiki core?

You need to create a patch like rMW3d007ba9f6ed: Add the Carpathian Romani (rmc) language that was created for T284044: Add Carpathian Romani (rmc) to Names.php.

Add ajp to Names.php. Add a MessagesAjp.php, setting $rtl = true; along with any fallback information that may be relevant.

Amire80 subscribed.

Thanks. How do I add/edit a language file in MediaWiki core?

You need to create a patch like rMW3d007ba9f6ed: Add the Carpathian Romani (rmc) language that was created for T284044: Add Carpathian Romani (rmc) to Names.php.

Add ajp to Names.php. Add a MessagesAjp.php, setting $rtl = true; along with any fallback information that may be relevant.

No, don't do it. This is done semi-automatically after there's a certain level of localization, and at the moment there's no localization to this language at all. By "localization" here I mean the translation of user interface strings (messages) into this language. Does anyone want to do it? I can set it up, if someone actually plans to do it.

@Reedy the desired result is to have ajp setup as "right to left" by default.

Right to left where, for example? There are a lot of possible contexts. If you can give examples, I'll be happy to help. language-data is appropriate for some of them, but not all. (Sorry, it's more complicated than it should be...)

Hi,

Our first goal is to create an ajp Wikipedia (today in the Incubator), so we need:

  1. Right-to-left directionality on the Incubator
  2. To translate the interface at some point

I've already posted messages on the Incubator. No answer so far: MediaWiki talk:Common.css ​ and Incubator:Community Portal.

I opened this ticket because those who created the Moroccan Arabic Wikipedia (ary) said it would solve the problem. (discussion here).

Then we have a font issue: Windows users don't like the font used for ajp (hard to read according to them), either on the Incubator or on the English Wikipedia when we apply the language tag ajp to some text (for instance using the models {{lang}} or {{wikt-lang}}). There's a whole discussion with examples here. What is weird is that, according to these users (I don't have Windows so I can't judge), ary (Moroccan) and arq (Algerian) look good.

Thanks for your help! (and yes, it's extremely complex, I thought it would just be one line in a CSS file...)

Amire80 added a project: translatewiki.net.

Oh, I can help with localization and direction. Some questions to clarify:

  1. Do you want to use Standard Arabic as the fallback if a message is not translated?
  2. I see that you suggested شامي as the name of the language. I know a tiny bit of Arabic and it looks legitimate to me, but just to be sure: is it used anywhere else? Are there any other websites or books that use this name as the title of the language?

I'm less sure about the font. I'll try to see what I can do, but I cannot promise anything. In any case, it should be a separate Phabricator ticket.

Our first goal is to create an ajp Wikipedia (today in the Incubator), so we need:

Wikipedia South Levantine Arabic was rejected by the language committee in accordance with the Language proposal policy.

Our first goal is to create an ajp Wikipedia (today in the Incubator), so we need:

Wikipedia South Levantine Arabic was rejected by the language committee in accordance with the Language proposal policy.

Yes, but that happened very long ago and on rather technical grounds, so it can be reconsidered.

Hi,

@Amire80: yes MSA as a fallback would be perfect.

Where do you see that I suggested شامي (Shami) as the name of the language? Do you mean on the main page of the incubator? Yes I did. I also think the name is legitimate. However, most speakers instinctively use the name "العامية" (Amiya, dialect/colloquial) to describe their language, to oppose it to MSA/fusHa. Or they use the name of their country (Lebanese, Palestinian, Jordanian, etc.). Or, if they want to distinguish it from other dialects, they call it "Shami" (and by doing so they implicitly don't distinguish ajp, South Levantine, from apc, North Levantine). This is confirmed by Ethnologue. According to Ethnologue, the alternate names of "South Levantine Arabic" (ajp) are "Levantine Arabic" (simply, no mention of "South") or "Palestinian-Jordanian" (the two countries) or أردني‎ (Urduni, Jordanian). And for North Levantine Arabic they say that the alternate names are "Levantine Arabic" (simply, no mention of "North"), "Lebanese-Syrian Arabic", "Syro-Lebanese Arabic", or شامي‎ (Shami, Syrian). Linguasphere calls Levantine "syro-palestinian". Another source using Shami: Kwaik, Kathrein; Saad, Motaz. "Shami: A Corpus of Levantine Arabic Dialects"., https://aclanthology.org/L18-1576/

Ok please let me know for the font, I'll be happy to open a new ticket if necessary!

@alaa: yes, the proposal was rejected in 2008 because "As this proposal has been proposed by an anonymous coward, this proposal will not be considered without people indicating that it is them who propose and indicate their willingness to contribute to this project." (quote from GerardM, who took the final decision).

Another answer to the proposal was "Moderately skeptical: This request is much like Egyptian Arabic and Moroccan Arabic". Since, both the Egyptian and Moroccan Wikipedia editions have been created. (And the Algerian may be the next one)

So there is no doubt that IF there are enough active contributors on the ajp Wikipedia Incubator this Wikipedia will be created. However, without the correct directionality and with the font issue, it's sure that no one will ever contribute.

Also, please note, as I said previously, that this issue is not only related to the potential ajp Wikipedia but also to issues on the English Wikipedia, on Wiktionary, and on Wikibooks, where characters tagged as ajp are not readable on Windows as you can see on this image: https://en.wikipedia.org/wiki/File:Scripts_W10_Edge(Chrome).JPG

@Amire80 https://en.wiktionary.org/wiki/MediaWiki:Common.css#L-783 Would it be possible to replicate this CSS on all Wikimedia projects? It seems to solve the font issue for all languages written with the Arabic script.

Hi,

@Amire80: yes MSA as a fallback would be perfect.

Where do you see that I suggested شامي (Shami) as the name of the language?

Here: https://github.com/wikimedia/language-data/pull/165

I'll take a look at the sources you suggested, thanks a lot.

@Amire80 https://en.wiktionary.org/wiki/MediaWiki:Common.css#L-783 Would it be possible to replicate this CSS on all Wikimedia projects? It seems to solve the font issue for all languages written with the Arabic script.

Conceivable, but not trivial :)

I'll take a look.

Hi,

@Amire80: yes MSA as a fallback would be perfect.

Where do you see that I suggested شامي (Shami) as the name of the language?

Here: https://github.com/wikimedia/language-data/pull/165

I'll take a look at the sources you suggested, thanks a lot.

Indeed! (I opened so many conversations and tickets, on different platforms, about this issue, that I start forgetting them...)

Change 703921 had a related patch set uploaded (by Amire80; author: Amire80):

[translatewiki@master] Add Levantine Arabic

https://gerrit.wikimedia.org/r/703921

Change 703921 merged by jenkins-bot:

[translatewiki@master] Add Levantine Arabic

https://gerrit.wikimedia.org/r/703921

Hi,

Our first goal is to create an ajp Wikipedia (today in the Incubator), so we need:

  1. Right-to-left directionality on the Incubator
  2. To translate the interface at some point

Translating the interface is now possible. Start here:
https://translatewiki.net/wiki/Special:Translate?filter=%21translated&action=translate&language=ajp&group=core-0-mostused

You appear to have a translatewiki account. I added a list of useful links to translatable projects to your user page. When you complete the "Most used" messages, the RTL direction will sort itself out, too. (For technical reasons it's a bit complicated to do now.)

Thank you so much!

I think I have a translatewiki but of course I forgot my password long ago and I don't remember the email I used either... Do you know if the username is case sensitive? (a455bcd9 vs A455bcd9)

[Update: problem solved]

Thank you so much!

I think I have a translatewiki but of course I forgot my password long ago and I don't remember the email I used either... Do you know if the username is case sensitive? (a455bcd9 vs A455bcd9)

[Update: problem solved]

I think it doesn't matter :)

If you have any trouble resetting the password, let me know and I'll try to help.

Thanks, I solved the problem and I now have access. Thanks a lot also for the resources on my user page!

Regarding the font issue: do you have any idea?

Thanks, I solved the problem and I now have access. Thanks a lot also for the resources on my user page!

Regarding the font issue: do you have any idea?

Can you please create a separate task about it, with a few examples of the problem?

Thanks, I solved the problem and I now have access. Thanks a lot also for the resources on my user page!

Regarding the font issue: do you have any idea?

Can you please create a separate task about it, with a few examples of the problem?

I created T286417

Does the PR still make sense by the way? https://github.com/wikimedia/language-data/pull/165

Thanks, I solved the problem and I now have access. Thanks a lot also for the resources on my user page!

Regarding the font issue: do you have any idea?

Can you please create a separate task about it, with a few examples of the problem?

I created T286417

Does the PR still make sense by the way? https://github.com/wikimedia/language-data/pull/165

Kind of, but needs a few fixes. I submitted another pull request based on it. No more action needed :)