I wonder how many of the translations from translatewiki that shows such an outstandig quality as http://translatewiki.net/wiki/MediaWiki:Expensive-parserfunction-warning/da does.
The attempt at translating the original text, "Warning: This page contains too many expensive parser function calls." to danish is actually worse than what Googles translator suggests. An attempt to translate back to english from the munged attempt at danish is something like "Warning: This page also contains many costly analyse the function phone call."
I think that the basic assumption that chinese horde translation is viable should be re-evaluated.
So I thought of a very short answer in this mail, and I decided not to write it.
Then by some associative browsing, looking into an error reported using the excellent ForeignAPIRepo functionality, I came across Tim Shell's user page on the English Language Wikipedia[1]. I went on and did a few other things, and decided it might be an interesting read. When I was done reading, I decided to reply to your e-mail anyway. All credits for the reply should go to Tim Shell. Flames can be addressed to me.
MediaWiki localisation moves asymptotically towards perfection. At any given moment, there is stuff we don't have, factual errors, etc. As MediaWiki localisation grows, and moves closer to perfection, errors and shortcomings grow smaller. MediaWiki localisation is thus a process, rather than an end state. Criticizing MediaWiki localisation for errors and shortcomings that exist now misses this point entirely.
What is preferable: An authoritative MediaWiki localisation that is 99.9% accurate, but which costs $300,000 a year to maintain? Or a MediaWiki localisation that is 95% accurate, but which costs maybe $5,000 a year to maintain? And that, on this much smaller budget, will grow more accurate every year? And that, because it does not demand perfect accuracy, is able to cover a much broader range of topics? And that, on top of this, is free?
Cheers! Siebrand
[1] http://en.wikipedia.org/wiki/User:TimShell
-----Oorspronkelijk bericht----- Van: translators-l-bounces@lists.wikimedia.org [mailto:translators-l-bounces@lists.wikimedia.org] Namens Anders Wegge Jakobsen Verzonden: zaterdag 2 augustus 2008 15:14 Aan: translators-l@lists.wikimedia.org Onderwerp: [Translators-l] many costly analyse the function phone call.
I wonder how many of the translations from translatewiki that shows such an outstandig quality as http://translatewiki.net/wiki/MediaWiki:Expensive-parserfunction-warning/da does.
The attempt at translating the original text, "Warning: This page contains too many expensive parser function calls." to danish is actually worse than what Googles translator suggests. An attempt to translate back to english from the munged attempt at danish is something like "Warning: This page also contains many costly analyse the function phone call."
I think that the basic assumption that chinese horde translation is viable should be re-evaluated.
-- // Wegge http://blog.wegge.dk - Her hænger jeg også ud. http://geowiki.wegge.dk/wiki/Forside - Alt om geocaching. Bruger du den gratis spamfighther ser jeg kun dine indlæg *EN* gang.
-- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
_______________________________________________ Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
"Siebrand Mazeland" s.mazeland@xs4all.nl writes:
What is preferable: An authoritative MediaWiki localisation that is 99.9% accurate, but which costs $300,000 a year to maintain? Or a MediaWiki localisation that is 95% accurate, but which costs maybe $5,000 a year to maintain? And that, on this much smaller budget, will grow more accurate every year? And that, because it does not demand perfect accuracy, is able to cover a much broader range of topics? And that, on top of this, is free?
You are missing the point entirely.
Around new year, you told the world how superior the translatewiki process was, and implicit how bad a translation was the result of individual translators. This is the reason I have no intentions whatsoever of doing anything else than proving how wrong that statement was. My translation work until that point was, and still is, fee, so I fail to understand the point of your different dollar amounts.
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated. There are many more similar examples, and as you can see, this particular instance is more than two months old.
2008/8/2 Anders Wegge Jakobsen wegge@wegge.dk:
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated.
Perhaps it would be possible to add basic validation functionality (not necessarily something as sophisticated as FlaggedRevs) to TranslateWiki? Then the quality of a translation could be ranked by the number of people who have looked at and validated it.
On Sun, Aug 3, 2008 at 1:54 AM, Erik Moeller erik@wikimedia.org wrote:
2008/8/2 Anders Wegge Jakobsen wegge@wegge.dk:
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated.
Perhaps it would be possible to add basic validation functionality (not necessarily something as sophisticated as FlaggedRevs) to TranslateWiki?
If it is, it is more than great I think. See also the thread about http://jp.librarything.com where they provide the registered users the way to evaluate the current version, not only the opportunity to submit the alternative.
If the entire site has a feature to recommend an alternative to system messages (since it is read-only for most visitors anyway), like Google Translator gives its visitors, I think it better from the point of proofreading, but not sure it is balanced with the other aspect & workload.
Then the quality of a translation could be ranked by the number of people who have looked at and validated it.
-- Erik Möller Deputy Director, Wikimedia Foundation
Support Free Knowledge: http://wikimediafoundation.org/wiki/Donate
Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
From personal experience I would say that its quite hard for 1) as a translator to guess at points where does a given sentence go, and what do the numerous "$1" become in use 2) as a visitor to a site using the translation to report or just know how and what to do to make the translation better. I think a solution like that of Facebook might be good: if you enable it in the settings you would have a link in the corner that would say "translate" you could click on anything on that given page and either provide the translation for it, or review the translated version and the attached comments (which then would have to be sent to TranslateWiki somehow). Otherwise its quite tiresome for the end-user to go and find the actual sentence again on TWiki (as the search doesn't search in the MediaWiki namespace by default), and than make the change (or be turned away by fear of disrupting something, after seeing the dollar signs for the included sub-messages).
Regards, Bence Damokos
On Sat, Aug 2, 2008 at 7:05 PM, Aphaia aphaia@gmail.com wrote:
On Sun, Aug 3, 2008 at 1:54 AM, Erik Moeller erik@wikimedia.org wrote:
2008/8/2 Anders Wegge Jakobsen wegge@wegge.dk:
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated.
Perhaps it would be possible to add basic validation functionality (not necessarily something as sophisticated as FlaggedRevs) to TranslateWiki?
If it is, it is more than great I think. See also the thread about http://jp.librarything.com where they provide the registered users the way to evaluate the current version, not only the opportunity to submit the alternative.
If the entire site has a feature to recommend an alternative to system messages (since it is read-only for most visitors anyway), like Google Translator gives its visitors, I think it better from the point of proofreading, but not sure it is balanced with the other aspect & workload.
Then the quality of a translation could be ranked by the number of people who have looked at and validated it.
-- Erik Möller Deputy Director, Wikimedia Foundation
Support Free Knowledge: http://wikimediafoundation.org/wiki/Donate
Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
-- KIZU Naoko http://d.hatena.ne.jp/Britty (in Japanese) Quote of the Day (English): http://en.wikiquote.org/wiki/WQ:QOTD
Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
Having done contextual inquiries with a number of professional translators, and having done a bit of translation myself, I can attest that it is very hard to translate sentences out of context.
So being able to translate a sentence in context, for example by clicking on it in the actual UI where it appears, is bound to be superior.
Alain
From: translators-l-bounces@lists.wikimedia.org [mailto:translators-l-bounces@lists.wikimedia.org] On Behalf Of Bence Damokos Sent: August 2, 2008 1:13 PM To: Wikimedia Translators Subject: Re: [Translators-l] many costly analyse the function phone call.
From personal experience I would say that its quite hard for 1) as a translator to guess at points where does a given sentence go, and what do the numerous "$1" become in use 2) as a visitor to a site using the translation to report or just know how and what to do to make the translation better. I think a solution like that of Facebook might be good: if you enable it in the settings you would have a link in the corner that would say "translate" you could click on anything on that given page and either provide the translation for it, or review the translated version and the attached comments (which then would have to be sent to TranslateWiki somehow). Otherwise its quite tiresome for the end-user to go and find the actual sentence again on TWiki (as the search doesn't search in the MediaWiki namespace by default), and than make the change (or be turned away by fear of disrupting something, after seeing the dollar signs for the included sub-messages).
Regards, Bence Damokos
On Sat, Aug 2, 2008 at 7:05 PM, Aphaia aphaia@gmail.com wrote:
On Sun, Aug 3, 2008 at 1:54 AM, Erik Moeller erik@wikimedia.org wrote:
2008/8/2 Anders Wegge Jakobsen wegge@wegge.dk:
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated.
Perhaps it would be possible to add basic validation functionality (not necessarily something as sophisticated as FlaggedRevs) to TranslateWiki?
If it is, it is more than great I think. See also the thread about http://jp.librarything.com where they provide the registered users the way to evaluate the current version, not only the opportunity to submit the alternative.
If the entire site has a feature to recommend an alternative to system messages (since it is read-only for most visitors anyway), like Google Translator gives its visitors, I think it better from the point of proofreading, but not sure it is balanced with the other aspect & workload.
Then the quality of a translation could be ranked by the number of people who have looked at and validated it.
-- Erik Möller Deputy Director, Wikimedia Foundation
Support Free Knowledge: http://wikimediafoundation.org/wiki/Donate
Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
-- KIZU Naoko http://d.hatena.ne.jp/Britty (in Japanese) Quote of the Day (English): http://en.wikiquote.org/wiki/WQ:QOTD
_______________________________________________ Translators-l mailing list Translators-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/translators-l
On Sat, Aug 2, 2008 at 7:47 PM, Anders Wegge Jakobsen wegge@wegge.dkwrote:
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated.
Naive thought.. why? I *do* proofread the translation. these errors you mention would be sorted out eventually.
"Mohamed Magdy" mohamed.m.k@gmail.com writes:
On Sat, Aug 2, 2008 at 7:47 PM, Anders Wegge Jakobsen <[1]wegge@wegge.dk> wrote:
I will continue pointing out those hilarious examples of worse-than-none translations that ensues from the naive thought that anyone will ever proofread a translation, when it has first been marked as translated.
Naive thought.. why? I *do* proofread the translation. these errors you mention would be sorted out eventually.
Well, the practical example show that it does not happen. Highly visible messages are translated locally on dawiki, but the fixes stay there. The more obscure errors goes unnoticed unless I feel like looking for confirmation of my prediction that it's just not possible to get a decent danish translation by way of the chinese horde.
Anders Wegge Jakobsen, Aug 2, 2008 19.19:
Highly visible messages are translated locally on dawiki
Very bad. Actually, I see that there are not so many MediaWiki edits[1], and that the most active translator, Kaare, has just discovered Betawiki[2], so I hope that he will correct and update messages here.
Nemo
[1] http://toolserver.org/~vvv/adminstats.php?wiki=dawiki_p&tlimit=7884000
[2] http://translatewiki.net/w/i.php?title=Betawiki:Translator&diff=prev&...
translators-l@lists.wikimedia.org