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

List:       kde-i18n-doc
Subject:    Re: Improving KDE with translators
From:       Gérard_Delafond <gerard () delafond ! org>
Date:       2004-08-22 7:04:44
Message-ID: 200408220904.44150.gerard () delafond ! org
[Download RAW message or body]

Le vendredi 20 Août 2004 15:03, Roger Kovacs a écrit :
> I have followed this thread with great interest, learning how other
> countries have implemented quality control in their translating
> projects.  
That was a part of my aim in my initial message.
> It seems like such an ideal world compared to the reality of 
> the challenges facing the smaller emerging nations.  I am fortunate like
> many of you reading this, at home or work with high speed broadband
> connection to the internet.  (I am in the USA.)
> When I started the Tajik Linux translation project a few years ago, it
> was with a key-pal over e-mail, and it just grew.  I e-mailed the po
> files to Tajikistan, had volunteers translate them with text editors.
> They would e-mail them back.  There was no internet connections at the
> time, only a store/forward e-mail system at a non-profit center that
> allowed the citizens to send e-mail with a max quota of 256K per month
> (the telephone connection is unreliable.)  I don't know the Tajik
> language, but I could run the Linux tools, fix the syntax errors in the
> returned .po file and check it in using CVS.  That is how we got
> started.
(...)
Your message is very interesting, and it shows the situation is even worse 
than what I could imagine.
That means I am more right than expected.
As there is a very wide dispersion in the means to translate (some teams are 
very numerous, have begun a long time ago, have an active mailing-list, 
etc.), it is important to communicate to our users on the expected quality of 
the translation.
Some translation are supposed to give the user a full experience, some others 
are only a draft, which can give some help to the user, with no waranty at 
all.

I suggest the criteria to suppose a translation (I mean the whole KDE 
translation, not an application) is good are :
-the number of translators in a team
#there is more in several heads
-the age of the team (beginning of the translation)
#some issues are found after years of usage
-the rate of translation of the GUI
#the more you translate, the more you get experienced
-the rate of translation of the docs and screenshots
#translating docs helps you to understand how the apps work
-the rate of keyboard shortcuts clashes
#this means you tested your translation, and is easy to check even if you 
don't know the language and it improves usability)
-any other item I did not think about

I suggest translations could be classified on a 5 stars scale
-5 stars : perfect = 100% in GUIS, docs, screenshots, and no keyboard 
shortcuts clashes (as possible), more than 10 translators, more than 3 years 
experience
-4 stars : more than 90 % of the GUIs, 70% of the docs, screenshots of all 
translated apps
-3 stars : more than 60 % of GUIs, 50 % of the docs, numerous screenshots
-2 stars : 2 translators at least, more than 1 year, 20% of the GUIS, no docs
-1 star : 1 translator, few GUIS, no doc, new team

Once the criteria defined, the problem is to know who will apply them.
I suggest the teams apply them themselves. I suppose KDE translators are 
honest enough to do it well.
All what KDE core team should do is to publish the criteria for attributing 
stars.

Once the stars attributed, the problem is how to communicate it.
Probably, the best place is where we choose the language (kcontrol).
The user would simply be warned he is choosing a x stars language (with an 
access to the criteria for attributing stars).

I think, with all these criteria, our team (french) could reach 4 stars, like 
about 10 others. Many others could reach these 4 stars easily, making 
screenshots for their docs. Reaching 5 stars would be a sort of graal for 
most of us.

I know this idea of classifying the quality of the translation is not popular 
among translators, is hard to do well, but it would improve KDE's 
credibility.

Thanks for reading

Gerard

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