[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-i18n-doc
Subject:    Re: broken links in the sq branch
From:       Sveinn í Felli <sveinki () nett ! is>
Date:       2011-05-02 9:42:58
Message-ID: 20110502094417.7974.qmail () ktown ! kde ! org
[Download RAW message or body]

, skrifaði Mike Dupont:
> This all makes good sense,
> but let me tell you about what i have experienced, there are many
> young people in Kosovo and Albania,
> they are not linux gurus, they are not computer gurus, but they might
> be willing to help and get interested if it is easy for them to get
> started.
> There are 2-3 hard core translators who have done 99% of the work,
> Besnik and Blendi to name two, and besides that there has been not so
> much effort.
> How can we get people interested? how can we get them started? I think
> a facebok application might be a good way to find new people.
>
> mike

Surely it's no harm if people are able to quickly submit 
_suggestions_ in an integrated environment ("Translate this 
application") or via an web interface. It could even be 
amusing for the persons that _must_ then review those 
phrases ;-)

None of the web interfaces in actual use are really that 
good, and some are plain pieces of junk. My experience is 
that a well thought implementation of Pootle has many 
benefits without much of the rubbish; it has role 
permissions, terminology, language-wide submission via zip 
archives, and some more. But like most web interfaces it's 
slow, breaks often and has sloppy search functions.

While translating locally with specialised tools 
(translation memory, good search/replace, etc.) is much 
faster, more consistent and not that difficult to learn. 
What may be complicated is to set up the update/commit 
processes.

Albert has mentioned the one-off translation visitors (short 
commitment lifespan) seconded by Yuri's statistics. My 
experience is the same, even if the stats may be inaccurate 
due to a few persons committing files for others.

I think there's a dilemma here, good translators may be 
language-geeks but not computer-savy, and computer-geeks may 
be not that fluent in their own language. Best is to have 
both kinds, working together in a team.
It has to be a compromise between quality and quantity - 
otherwise we end up with situations like Ubuntu (quantity of 
rubbish) and Debian (few quality strings). No offence 
intended...

This all boils down to the problem of making and maintaining 
a team, working together, delegating tasks, renewing 
participation etc. etc...

Best regards,
Sveinn í Felli

> On Sun, May 1, 2011 at 10:08 PM, Yuri Chornoivan<yurchor@ukr.net>  wrote:
>> написане Sun, 01 May 2011 22:20:29 +0300, Mike Dupont
>> <jamesmikedupont@googlemail.com>:
>>
>>> asking people to help in a safe manner is a good way to engage people.
>>> even if we have to discard some users input, it would be good to get
>>> people involved.
>>> If you think about the social network idea it could also be used to
>>> also rate users input and weight them, give people a way to build up
>>> positive or negative feedback
>>> I still think that it could be used to get new people interested and
>>> in the long term produce good results.
>>> mike
>>
>> The previous statements can be truth iff we suppose that the picture is
>> static: no new applications, no new messages, just new people fixing the old
>> mistakes and adding new translations.
>>
>> In fact, e.g. Ukrainian team in LP has the following results for
>> untranslated:
>>
>> 9.10: 92979 (-~20000 for OO.org)
>> 10.04: 73165 (without OO.org)
>> 10.10: 77237
>> 11.04: 81519
>>
>> The last known data (11.04) shows that 6.85% of messages were translated on
>> LP.
>>
>> Thus, the online tool cannot give the number of translations.
>>
>> Can it give the number of translators? Consider this statistics (team has
>> 302 members now):
>>
>> https://translations.launchpad.net/+languages/uk
>>
>> Name                         Karma
>>
>> Yuri Chornoivan              13512
>> Sergiy Gavrylov              7009
>> Fedik                        3741
>> andygol (Russian)            1744
>> lokster (Bulgarian)          1449
>> Dirk Stöcker (maintainer)    1302
>> etc.
>>
>> The similar statistics (only the persons will change) can be found on
>> translatewiki.net.
>>
>> Conclusion: like in wikipedia, 90% (or more likely 95%) of contributions are
>> made by 5% of contributors. The way of contribution does not matter.
>>
>> Can the online tool give the better quality? Ask the users of Ubuntu or
>> MeeGo 1.1 translations.
>>
>> So why should I maintain the online translation tool if it gives me 10% (or
>> likely 5%) of contributions (most of them are worse than Google Translate)?
>>
>> This open online scheme fails (at least does not win) for Ubuntu, Drupal,
>> Django. Now Ubuntu team members are trying to close the access to their
>> "open" translations with the rules that are far more strictly than KDE ones.
>> Almost any "open" translation with 5-6 accident contributors is garbage. You
>> have to register, you have to register translation group, you have to
>> acquire membership in the other translation group, etc., and then you can
>> translate.
>>
>>>
>>> On Sun, May 1, 2011 at 9:14 PM, Vít Pelčák<vit@pelcak.org>  wrote:
>>>>>>
>>>>>> Just a joke:
>>>>>>
>>>>>> And if you give them the bananas, it is likely you will have the monkey
>>>>>> translation. ;)
>>>>>
>>>>> Sure but a monkey translation is better than no translation, no?
>>>>
>>>> Not at all. It is better to not to translate at all than confuse user.
>>>>
>>>> He, who doesn't understand english will get help neither from
>>>> incorrectly translated nor untranslated message.
>>>>
>>>> He, who does will in same case get help from only from untranslated
>>>> message.
>>>>
>>>> Frankly, I think we should also support people learning English rather
>>>> than their ignorance. But that's just personal oppinion.
>>>>
>>>> Regards
>>>> Vit Pelcak
>>>>
>>>
>>>
>>
>
>
>


[prev in list] [next in list] [prev in thread] [next in thread]