For reference, these are the languages which were prioritized for the elections. Additional translation support was utilized to help with the 48-72 hour deadlines:

1. العربية (ar) 2. বাংলা (bn) 3. Deutsch (de) 4. español (es) 5. فارسی (fa) 6. français (fr) 7. हिन्दी (hi) 8. Bahasa Indonesia (id) 9. italiano (it) 10. 日本語 (ja) 11. Nederlands (nl) 12. português do Brasil (pt-br) 13. русский (ru) 14. Kiswahili (sw) 15. Türkçe (tr) 16. Tiếng Việt (vi) 17. 中文 (zh)

-greg (User:Varnent)

On Wed, May 27, 2015 at 6:44 PM, Richard Symonds < richard.symonds@wikimedia.org.uk> wrote:

Milos,

The formatting has been stripped from the email because the list software doesn't like HTML, so the table at the bottom is illegible. Is it available elsewhere?

Richard Symonds Wikimedia UK 0207 065 0992

Wikimedia UK is a Company Limited by Guarantee registered in England and Wales, Registered No. 6741827. Registered Charity No.1144513. Registered Office 4th Floor, Development House, 56-64 Leonard Street, London EC2A 4LT. United Kingdom. Wikimedia UK is the UK chapter of a global Wikimedia movement. The Wikimedia projects are run by the Wikimedia Foundation (who operate Wikipedia, amongst other projects).

*Wikimedia UK is an independent non-profit charity with no legal control over Wikipedia nor responsibility for its contents.*

On 27 May 2015 at 22:04, Milos Rancic millosh@gmail.com wrote:

...

Below is the list of the languages sorted by the number of L2 speakers (more than one million of them).

L2 speakers appear in two occasions:

• First and important to us is about languages used for wider

communication. For example, French is L2 among educated people of West Africa.

• The second type is related to the native languages in not so good

position (either dying or reviving). For example, English is L1 language

of

...

the most of Native Americans, as well as Russian is L1 language of the

most

...

of ethnicities of former Soviet Union, while their own languages are L2 ones. (They are important in other cases, but not for this purpose.)

I omitted English (there is no sense, as we are communicating in English and English is default for all the localization) and few spoken languages (our content is [mostly] written).

I also removed some languages which belong to the second category (Irish Gaelic and Scots, for example), but it could be the case that some of the languages from the list belong to that category, as well (though I am pretty sure they don't).

There are languages inside of this list with well developed Wikimedia projects and without particular need to promote work on Wikimedia

projects

...

among them: French, Spanish and German are the examples. There is no Russian inside of the list, as it's usually L1 language, as mentioned above, but it belongs to the category of the languages with well

developed

...

Wikimedia projects.

There are also languages spoken in countries with low level of internet access and issues much more important than writing an encyclopedia, like Congo Swahili is. Those are the areas not yet ready even for the projects like OLPC is and we don't have a lot to do there.

But there are a number of languages in between with active chapter(s) or user group(s) inside of relevant countries. Those languages should be the priority in promotion collaboration.

They are: Arabic (Arabic user groups), Indonesian (WM ID), Hindi (WM IN), Urdu (Pakistani user group), Thai (Thailand UG), Bengali (WM BD), Zulu

(WM

...

ZA), Hausa (West African user groups), Xhosa (WM ZA), Afrikaans (WM ZA), Kannada (WM IN), Telugu (WM IN), Tsonga (WM ZA), Malay (WM ID and

Malaysian

...

Wikimedians), Marathi (WM IN).

The priorities for those languages should include (but likely not limited to):

• Translation of MediaWiki messages should be 100%.
• Those languages should be priorities for every document which should be

translated. For example, ongoing Board elections; but also various Meta pages.

• We should have the pool of literate people in those languages for

various

...

purposes, not just for translation. For example, if we want to create projects in languages of Pakistan, we should have a number of literate

Urdu

...

speakers, willing to help newcomers speaking Urdu as their L2 language.

Will be back with other languages-related data :)

LanguageCodeL1 speakersL2 speakersStandard Arabicarb206,000,000246,000,000Mandarin

Chinesecmn847,808,270178,000,000Indonesianind23,200,480140,000,000Hindihin

...

260,333,620120,000,000Spanishspa398,931,84096,990,000Urduurd64,035,800

94,000,000Frenchfra75,916,15087,000,000Thaitha20,396,93040,000,000Bengaliben

...

189,261,20019,200,000Zuluzul11,969,10015,700,000Hausahau25,109,00015,000,000

...

Xhosaxho8,177,30011,000,000Afrikaansafr7,096,81010,300,000Bamanankanbam 4,072,04010,000,000Burmesemya32,035,30010,000,000Congo Swahiliswc1,000 9,100,000Northern Sothonso4,631,0009,100,000Kannadakan37,739,0409,000,000 Germandeu78,093,9808,000,000Tamiltam68,776,4608,000,000Juladyu2,550,000 7,000,000Lingalalin2,141,3007,000,000Koongokng5,016,5005,000,000Telugutel 74,049,0005,000,000Ibibioibb1,500,0004,500,000Tok

Pisintpi122,0004,000,000

...

Kriokri495,6004,000,000Amharicamh21,811,6004,000,000Bangalabxg~0 3,500,000Tsongatso4,009,0003,400,000Malayzlm15,848,5003,000,000Marathimar

71,780,6603,000,000Sinhalasin15,613,9802,000,000Efikefi405,2602,000,000Duala

...

dua87,7002,000,000Yorubayor19,380,8002,000,000Shonasna10,741,7001,800,000 Vendaven1,294,0001,700,000Sangosag404,0001,600,000Manado Malayxmm850,000 1,500,000Sylhetisyl10,300,0001,500,000Ambonese Malayabs245,0201,400,000 Ndebelenbl1,090,0001,400,000Rakhinerki1,000,0001,020,000Gandalug4,130,000 1,000,000Akanaka8,314,6001,000,000Khmerkhm14,224,5001,000,000 _______________________________________________ Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe

---

Wikimedia-l mailing list, guidelines at: https://meta.wikimedia.org/wiki/Mailing_lists/Guidelines Wikimedia-l@lists.wikimedia.org Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/wikimedia-l, mailto:wikimedia-l-request@lists.wikimedia.org?subject=unsubscribe

Richard, the spreadsheet could be found here [1]. I compiled the list again and it assumes some more languages (Swahili, for example), which didn't have numbers in Ethnologue's data.

On Thu, May 28, 2015 at 7:04 AM, Asaf Bartov abartov@wikimedia.org wrote:

For some of these languages, I don't see that this makes sense, in terms of investment versus impact, or in terms of putting the cart before the horse. Tsonga, for example, seems to have precisely one active editor[1] -- no doubt, our colleague Dumisani Ndubane (CCed as a courtesy). While it is indeed his native language, it does not seem like a good investment of effort to translate hundreds of pages and thousands of strings into Tsonga when he (and, with overwhelming likelihood, any other literate speaker of Tsonga) is also fluent (and educated in) English. Xhosa, Hausa and Zulu are in the same class[2][3][4].

It's not just about English. I've realized that there is the third group of the languages, which include the fact that speakers of various languages are in fact fluent in the other languages of the same area. Thus, those numbers could mislead. But, this is the initial data, for further analysis.

... Again, there already exists a community of dedicated contributors to the Urdu Wikipedia[5] (apparently more from India than from Pakistan, no doubt partially due to script issues[6]). Some of you, particularly in the last year, have been energetically mentoring newcomers and doing outreach activities. Our colleagues Nisar Ahmad Syed and Muzammiluddin Syed (CCed) are two such volunteers. Now, what, precisely, are you suggesting?

Members of our community native in a language widely as L2 (or similar) have wider responsibility for keeping translations up to date. According to the present situation, their responsibility to [usually] minority languages is even greater than the responsibility toward their own language, as it's matter of would that language have or not Wikimedia project(s).

For example, if we get a community interested and being capable to have a number of Wikimedia projects, but not that fluent in English, we need to cover in their L2 all important MediaWiki messages and keep them up to date.

There are two problems with which a group of enthusiasts in such linguistic situation: * They have to translate ~500 messages and it's not a time consuming problem. However, even Hindi and Arabic don't have 100% of translated messages from that group [2] (likely one message is the issue). If we count on, let's say, dozens of volunteers educated in Hindi or Arabic but far from being fluent in English, that could be significant problem. If they want to create the next project, let's say Wiktionary to gather lexical data -- and I think the second two message groups have to be translated for that (not quite sure, will have to take a look into the LangCom list archives) --, they have even bigger problem. If they want to create the third one, let's say Wikibooks, to create educational textbooks, they are likely in front of the problem which they won't be able to solve. * MediaWiki is developing and messages are changing. While it doesn't matter a lot for the main language to have 99% and not 100% of translated most used messages, the new one won't get a project if it's not 100%. (The situation as it is; I don't like it, but I can't change it.)

It turns out that the threshold for very small languages is much higher than for the big ones. And we have to find a way to help them. L2 languages are one of the good starting points.

Further... We don't have statistics related to the translation of the articles which define Wikipedia, Wikimedia and other projects. And that has to be translated by Wikimedians who are well into the matter and understand fully the meaning of the content. (I don't even want to open the question of the fact that we don't have those defining documents on Meta, but on English language projects; more than a decade of Wikimedia Foundation existence.) So, even we use customary rules "translate it from English Wikipedia", we are pretty incapable of following the translation paths. And that's quite important to us. I mean, important so much that it defines who we are and what we expect from newcomers.

And I didn't say anything about current events, of which the most important ones should be transferred to as many as possible Wikimedians, including those who don't speak English. (It isn't realistic to expect transfer of information to a Wikimedian who speaks just his or her native language, while Wikipedia in that language has few dozens of active editors or less.)

Now, what I suggest...

First of all, I'd like to see the most important content translated into the languages used for wider communication. Which methods would be used, it's different question and while I will suggest some of them, I really don't mind if a different approach or approaches would be used.

If we want to do that, we should define "the starter kit": What's the most important to be translated? How often do those document change? How many translators do we need to keep the messages and documents up to date? How can we increase the number of translators? What could we use to improve response for calls for translation? What's the minimum capacity one community should have to maintain translated various segments?

Thinking generally, it's probably good idea for the beginning that WMF promote among chapters and user groups translation editathons according to the recommendations created based on the starter kit: "Don't you area have those languages spoken? Would you like to organize annual translation editathons?"

But, as I said, the last paragraph is particular methodology. If it could work better by using something else, I am for it.

[1] https://docs.google.com/spreadsheets/d/1dbL-aJAStMuGlnxq5Md0s0zCmX--35JzRKcX... [2] https://translatewiki.net/wiki/Translating:Group_statistics

On Jun 2, 2015 00:39, "Benjamin Lees" emufarmers@gmail.com wrote:

On Mon, Jun 1, 2015 at 11:53 AM, Milos Rancic millosh@gmail.com wrote:

...

• MediaWiki is developing and messages are changing. While it doesn't

matter a lot for the main language to have 99% and not 100% of translated most used messages, the new one won't get a project if it's not 100%. (The situation as it is; I don't like it, but I can't change it.)

"Won't get a project"? Are you saying that new project language editions are only approved if the MediaWiki messages for that language are all translated already? (Maybe I'm misunderstanding.)

Writing from the phone, so can't give links... Search for Language proposal policy on Meta. That's theory. I described above how it translates into the practice.

It makes sense up to certain point. MediaWiki interface should be in the native language. The condition for the first project isn't hard. It's about 500 messages.

In relation to the second and third project I think there are much more sensible work to be done than translating various obscure places of MW interface. (Few years ago few of us, after a lot of arguing removed translation of CheckUser interface as a requirement for the third project, likely Wikibooks or Wikisource.)

It would be useful for the sake of future arguments to have data how often people access to particular messages.

On Jun 2, 2015 02:08, "MZMcBride" z@mzmcbride.com wrote:

Milos Rancic wrote:

...

On Jun 2, 2015 00:39, "Benjamin Lees" emufarmers@gmail.com wrote:

...

"Won't get a project"? Are you saying that new project language editions are only approved if the MediaWiki messages for that language are all translated already? (Maybe I'm misunderstanding.)

[...]

It would be useful for the sake of future arguments to have data how

often

...

people access to particular messages.

Directly related: https://phabricator.wikimedia.org/T65416#1042471. Though upon re-reading it just now, the specific wording used at https://meta.wikimedia.org/wiki/Language_proposal_policy is actually softer than I thought ("it is recommended" instead of a hard requirement).

Will read Phabricator discussion in the morning...

Regarding LPP wording, as I mentioned above, it's theory. Practice is pretty hard and was even harder in the past. I remember Robin and I were waging hard battles for every set we wanted to remove from requirements.

I am sure that's documented somewhere, but I forgot where. It should be somewhere on Meta.