Proposal talk:Interlanguage reunification

Implementation ideas

Great idea! Here are some thoughts. For one wikipedian to harmonize two pages requires three skills: language A (let's say French), language B (let's say English), and subject matter expertise/interest. An unlikely combination. So, break down the task:

Offer a machine translation (e.g. Google Translate) of say http://fr.wikipedia.org/wiki/Minitel to Editor 1 who reads French, writes English natively, but is not a subject matter expert on telecommunications technology.
Editor 1 touches up the machine translation (like Google Translate which has a "Contribute a better translation" option today) into good English.
Editor 2, who is a subject matter expert on telecomm and a native writer in English but doesn't speak French, draws on the translated material to improve the corresponding article in English (http://en.wikipedia.org/wiki/Minitel).

The process could work into multiple directions:

Editor 2 might draw on the best contributions, translated to English from not only French, but also German (From Editor 3), Dutch (Editor 4), etc.

Editor 5 (a French-speaking technologist who doesn't speak English) might draw good bits from the English Minitel page back to the French one via Editor 6, who reads English, writes French natively but is not a technologist. Woz2 00:48, 14 August 2009 (UTC)[reply]

Yes, some use might be made of machine translation (as indeed it could be now, for translating articles between languages), although that possibility isn't the main thing that motivated this proposal. It might also be an idea to offer readers a machine translation of a foreign-languge article (with strong caveats as to accuracy) if their own language doesn't have an article on the subject yet. But especially the tabular data should be available in all lanugages, since that can be auto-translated successfully.--Kotniski 07:23, 14 August 2009 (UTC)[reply]

Thanks. Rather than imposing bureaucracy, we should offer tools to editors. You can use machine translation now, but it isn't very convenient, nor does it have the polish that a human editor could offer. Woz2 02:05, 15 August 2009 (UTC)[reply]

See Proposal:Integration_across_languages_and_WMF_projects, which links to a technical proposal along these lines

Fantastic idea

I edit in more than one language and think this is a great idea. I agree that it would increase productivity. On the other hand, I think it's probably impossible to implement (bureaucratically, not technically); I imagine a lot of the language versions will not like this and the policy concessions that would go with it. To get it done would require some good old-fashioned dictatorship. Eb.eric 01:09, 14 August 2009 (UTC)[reply]

You may be right, unfortunately, but even if the full proposal isn't implementable, we could still go some way down the road.--Kotniski 07:23, 14 August 2009 (UTC)[reply]

This is EXACTLY how I feel about the situation. We should be one project. What is happening at present isn't going to work in the long run because of the uneveness. Some wikipedias will always be better than others and given that one of our core goals is the sum of knowledge to anybody in their own language we have an obligation to try to achieve it. We seriously need to do something about the translation on here if we are serious about other languages gaining access to knowledge and vice versa. I don't know though how articles can be exactyl translated and shared. Dr. Blofeld ^{White cat} 09:33, 14 August 2009 (UTC)[reply]

What's feasible, and what isn't

Short term unification is not feasible: too much work for too less gain.

Why?

Variant A: To write a perfect article in one metalanguage we should have metalanguage, and even though English is kind of todays lingua franca, it's not omnipresent.
Variant B: Maintenance of articles in all languages on one project, with or without (some kind) machine translation is a mess, not so much because of aforementioned machine translation, but because of need for admins for all languages on that one superproject. If you could there edit article only in one language at a time, project could be divided to enable 200+ groups of language admins so everybody would check it's language, but either way, it would be mess :D

What's feasible (short term)?

Some kind of commons for statistics data, so census data of Paris wouldn't be changed separately 266 times on 266 wikis, but once every 10 years (when census data become public) e.g. on commonsource.wikimedia.org, from where all such global data could be transcluded.

Where to look (long term)?

What I described above as Variants A & B really is a future (A or B or both), but I don't see that either variant will be realized sooner than 2020 :-) SpeedyGonsales 01:44, 14 August 2009 (UTC)[reply]

Well, there's no deadline;) But I certainly agree about the commons for statistics data (and for interwikilinks, to avoid all those millions of bot edits, although I've heard that might already be in the pipeline).--Kotniski 07:23, 14 August 2009 (UTC)[reply]

Good idea but please not at cost of independence

This Idea has some good benefit to hold but please let us work out some thing else.At local levels present system offers some good flexibility of independent decision making and flexibility, especially wikipedia's with active but smaller communities may still prefer to have independence.

Rather Proposal:Change languages quickly also needs consideration.

I would also suggest that Proposal:Integration across languages and WMF projects is another less ambitious proposal that would take us a little way down the road toward this eventual goal. Dupuy 16:42, 14 August 2009 (UTC)[reply]

I've made a more modest proposal along these lines; Proposal:Interwiki links for redlinks. Joey the Mango 10:43, 15 August 2009 (UTC)[reply]

History Sections of translated pages do offer a good corpus for ideal machine translations.Some repository of most used sentences also can be generated for faster translations.

We need some refined proposal on translation side

Mahitgar 10:39, 14 August 2009 (UTC)[reply]

some problems

In my understanding, you want all Wikipedia to be conbined into one wiki. But what about page titles? All-in-one title, such as United States of America/United States of America (Fr)? Or different titles?
- Also, one article in English may be connected to 2 articles in one other language, but that's not the case of duplication. I have an example of en:MTR with zh:香港地铁 & zh:港铁.
And if Machine Tranlsation is considered, there will be a Big problem of the quality of some articles. In Chinese Wikipedia, Machine-translated articles will be speedy deleted.
- Some languages don't have machine translators, such as za:.
--Hat600 06:16, 15 August 2009 (UTC)[reply]

The same title will not be used across languages; the titles will not change from what they are now. The articles will be simply be grouped, so for example from the English page, you click French and see the French page, just as it is now. They will probably all have some sort of identifier, like an article ID number, that links them, but that would be all back-end. The end-user won't see very many changes from how things are currently, in this respect.

As for the 2:1 problem: these are the sorts of things that will have to change. In a situation like this, the English Wiki could decide to create an English version of both the zh articles, or one of the English versions could just be a link to the other English article. Things like this need to be worked out by the community.

Finally, your concerns about machine translation is misplaced, because this proposal doesn't suggest any increased use of machine translation, or any changes to the articles themselves. This proposal is to bring all language Wikipedias under one project, thus simplifying the currently very inefficient interwiki link system, and allowing efficient sharing of table data, images, etc. The proposal author can correct me if I'm wrong. Eb.eric 14:21, 15 August 2009 (UTC)[reply]

What is wiki

Sorry to be a skeptic, but it sounds very far-fetched for

the sets of concepts are different in different languages; there is no one-to-one map between concepts and words
Wiki is a bottom-up system, and this is a top-down proposal; so it is hard to get community support

Articles are about concepts. They can be called whatever. Eb.eric 03:38, 16 August 2009 (UTC)[reply]

Yes but concepts are language dependent. You cannot truely isolate concepts from the base language. Hillgentleman 16:28, 18 August 2009 (UTC)[reply]

No!

It might work for a few related language wikipedias, such as east nordic speakers, Swedish, Danish and Norwegian, but let's imagine every administrator on the wikipedia learning every major language, such as Jimbo, learning English, German, Arabic, Hindi, Mandarin and Japanese. Surely he is capable, but does he have the time? And walking around and trying to see if some imagined vandalism really is vandalism, just the text in question happens to be in Korean in a Korean page and the administrator is a Tamil? The reunification will for certain make the Wikipedia servers crash in a millisecond, the next millisecond incinerating the buildings where they happen to reside, the third millisecond the local town will burn to ashes, up to including all inhabitants clothes, inhabitants that the fourth millisecond will be wondering why they're standing there naked knee-high in ashes. Rursus 13:02, 18 August 2009 (UTC)[reply]

A bunch of good ideas

I can't even say I understand the concept of this proposal completely, but can at least heartily agree that it would be great towards WPs' objectives to have articles translated faster. As a translator, I am thoroughly disappointed with machine translations (interpreting is the first step in translation; machine renditions I have seen are so garbled up that really it is easier not to use them) but others may have different opinions. Whatever device we use, we ought to act fast on this subject. Also: I humbly suggest there are a bunch of good ideas on this talk page; sounds to me as though they deserve to be retrieved somehow. Say, people here got together and rewrote this proposal, filter out impracticalities... you get the gist. Thamus 01:39, 20 August 2009 (UTC)[reply]

There will be serious communication problems

If all languages are united in one wiki, then many administrative issues will have to be discussed between users of different languages. They will have to choose a language which they'll use for discussions. Probably this will be English most of the time. Now this would cause serious communication problems for those Wikipedians who don't speak enough English for discussing in English. Additionally, it will be highly discriminating against non-English-speakers.

Eb-eric wrote above that two articles in one language corresponding to one language in another should no longer be possible when this one-wiki system will be established. Now this would cause even more serious problems: How one divides knowledge into articles with their respective titles isn't and can't be language independent. Just visit the article "vintage car" on the English Wikipedia, and then click on the German link. As you will see, the German article uses the German pseudo-anglicism "Oldtimer" for its title; no problem so far. But as you can see from the articles, an "oldtimer" in German is any car older than 25 or 30 years, while a "vintage car" in English is a car from the period 1919-1930. How would you solve that in a one-wiki system? Should every language have an article on every concept used in any language whatsoever? This would cause a complete overblow. Or should one language be chosen as the language used to define how to group the knowledge into articles, while other articles just translate this knowledge? That would be a highly discriminating asymmetry, that would additionally make it impossible for Wikipedias in languages other than the central language to represent the knowldege in a way that is sensible for native speakers of those language. Or should a case-by-case evaluation be used to decide how to group the knowledge into articles? This would cause endless discussions (which as noted in the first paragraph would include communication problems and linguistic discrimination).

So I strongly oppose this proposal. Marcos 23:20, 24 August 2009 (UTC)[reply]

Simpler idea

I have an idea that is much simpler, and that would avoid lots more of these problems: just teach every human on the earth to use English at a professional level. All interlanguage problems solved! - Brya 05:49, 27 August 2009 (UTC)[reply]

Fantastic idea, but I think we can simplify the task a bit if we replace English by Esperanto in your proposal :-) Marcos 10:30, 28 August 2009 (UTC)[reply]

Well, maybe, but not if we go by the number of existing English speakers (versus existing Esperanto speakers) and especially by the existing amount of original written material in English (versus existing amount of original written material in Esperanto). From a quantitative perspective English wins by a long stretch. - Brya 18:31, 28 August 2009 (UTC)[reply]

Well, less than 5% of the world popoulation speak English natively (but some of them not at a professional level). If you add foreign language speakers that use English at a professional level, you're still certainly below 10% of the world population. So Esperanto just needs to be 10% easier than English to make the task of making all people in the world speak a single language at profession level more easily attainable for Esperanto than for English. Actually Esperanto is much more than 10% easier than English!

At any rate, at least the "Fantastic idea" part of my previous comment was actually a joke. I think your idea is a very silly idea; I actually thought you were joking, so I responded with a joke. But now it seems you are beeing serious, so I respond seriously:

Teaching everyone to use English at a professional level is not only completely out of the scope of what the Wikimedia Foundation could possibly attain, it is also something that shouldn't be done if it was attainable: Teaching everyone English necessarily implies that everyone learns more about the culture of English speakers than about other cultures; this would increase the imbalance of cultures in the world even more, and would contribute severely to the loss of cultural diversity; it would also put speakers of English in an economical advantage for the (necessarily very long) period during which English would have to be massively taught to non-native speakers, which is very unjust to speakers of other language, who would be in an economical dis-advanced position (one reason for this economical inequality is that English speakers could make money by teaching English, while the others would have to pay them; another reason is linguistic tourism to English speaking countries, which brings a lot of money to them).

In the case of Esperanto, these inequality, injustice and cultural hegemony issues aren't problematic, so for Esperanto it could actually be desirable that everybody in the world learns it to a professional level. But this would still be completely out of the scope of what the Wikimedia Foundation could attain (even though it would be more easily attainable than for the case of English). Marcos 22:45, 28 August 2009 (UTC)[reply]

Well first of all, this plan is obviously unrealistic, just as unrealistic as the proposal, but simpler in concept and more worthwhile in its effect. A further-reaching plan would be to teach every human on the planet telepathy; however, this would be even more outside of the scope of the Wikimedia as the result would be that nobody would read anything in the Wikimedia projects anymore.

Having said that, the number of people who speak a language natively is pretty much irrelevant, and I certainly did not introduce that into this discussion. If that were a criterium, everbody should learn Russian or Chinese. However, English is very widely spoken, when one considers the use as a second or third language. Perhaps more importantly, much of the primary literature on very many topics is in English.

It is true that a language is linked to culture; the very concepts that can be expressed depend on the language used. But here too, English has the advantage, as English is not only just one language (after 1006 it became two different langauages, forcefully fitted into the one jacket) but has been used in so many cultures and has borrowed words and concepts form so many other languages that it has a distinct advantage. Esperanto may have been a good idea, but it just did not make it. - Brya 08:06, 29 August 2009 (UTC)[reply]

Impact?

Some proposals will have massive impact on end-users, including non-editors. Some will have minimal impact. What will be the impact of this proposal on our end-users? -- Philippe 00:11, 3 September 2009 (UTC)[reply]

as simple way to start: wiktionary

The simplest way to start might be Wiktionary, where there is already emphasis on providing translations, where much of each article is formulaic, and where they are no long blocks of text. DGG 18:43, 11 October 2009 (UTC)[reply]