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

List:       kde-i18n-doc
Subject:    A different view of translation process
From:       Caslav Ilic <chaslav () sezampro ! yu>
Date:       2003-08-20 14:04:40
[Download RAW message or body]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,


I've recently joined GUI translation within Serbian team and, almost
immediately, something started bothering me. I had some conversation with
my team members, and wrote some code to check my views, but I think I
shouldn't continue further before discussing this with larger public.
There are both social and technical matters to be considered, and I myself
am shaky in both aspects.

Here goes...

- From release point of view, message catalogs are treated the same way as
source code. When there is official release upcoming, both programmers and
translators scramble to polish their job, tie loose ends and fix bugs.
Therefore, translation blindly follows source code, being almost integral
part of it from user's perspective. I think that this neither *must* nor
*should* be the case.

Where is the problem? If translation would be fully satisfactory at every
release, there would be no problem at all. But, in my book, fully
satisfactory means:

- - stable dictionary (from basic terms to constructs)
- - stable style (construction of sentences in appropriate occasions)
- - no spelling or grammar errors
- - no misjudged meanings (original messages can be very misleading)
- - no badly formatted strings in real environment

Translators are usually not linguistically educated and frequently
translate programs which they aren't extensively (or at all) using
themselves, so I believe translations are rarely fully satisfactory in
reality. This is certainly the situation here in Serbia (Southeastern
Europe, Balkans), where things are in a bit of flux (to say the least) on
all mentioned accounts. Heck, I hear that even in France there came a
governmental order not to use word "e-mail" any longer (though, I see that
French team wasn't using it anyway :)

What this means is that, while translators are improving their work within
current scheme of translation process, users of *older* program releases
gain no benefit until they *upgrade* to newer version. One of my team
members said "Who wants new translation, must install new version. Its
that simple." I think this is wrong, both from user's and translator's
viewpoint.

Who are the expected users of translation? I believe that most computer
enthusiasts, aside from us translating the program and small percentage of
others, will use default English version. Translation users will mostly be
people who are using computers for some specific job, especially in
educational and company environments. So, most of the people who will be
upgrading their programs frequently will not use the translation, and most
translation users will not upgrade their programs frequently, not by a
longshot :) It would be far better if users of translation would have the
possibility to use new translation without compromising their system
stability by installing newer versions of translated program.

As for translators, I think they needn't necessarily be as enthusiastic
about new versions as programmers are. For one, I myself simply will not
run CVS snapshot of program I am translating. I would like much more if I
could do environment checks of messages in stable, official release
version, with my own timetable, rather than rushing along programmers in
CVS development. There *might* be more people like me :) And it would
certainly lead to a higher quality translation.

I was very carefull not to mention KDE until now :), since this was all
general stuff. It is obvious that all mentioned problems are further
alleviated when talking about KDE, as it is a major system component.
Users will hesitate even more to install new KDE version (especially
network administrators), and translation is going to be worse due to
enormous amount of work. I am running KDE 3.1.1a and feel no special urge
to upgrade.

Based on all these arguments, I propose that *current CVS translation
should contain all GUI messages from previous official releases, up to
some point in the past*. It should be at least one year of releases,
possibly two -- it would be up to each team to decide how much they can
maintain. Based on the code I wrote so far, increase in maintaining
incorporated translation of all releases from 3.0 to HEAD would be 30%,
and from 3.1 to HEAD only 9%. These percents are ratios of non-duplicate
obsolete messages (not present HEAD source code) to current (HEAD)
messages.

If this would be the case, users should be instructed to use newest
translation, providing they are not running too old release. Since
installation of translation is relatively benign, not likely to mess up
anything, nobody (not even sysadmins) should object to keeping with
up-to-date translation releases. Also, users would be more willing to
report bugs in translation, and even join translation team, knowing they
can immediately see results of their work.

As for technical problems of achieving this, I could explain what I have
encountered (and solved) so far, but I think it would be better if those
interested would ask questions first, in order not to prejudice them.


With regards,
Chusslove Illich
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE/Q3/6MSGXgigGr3ERAh7zAKCOE8zExHsoPcTdUPk4Vkrc8f6OEgCfW/bL
QDbOem/5BK7SlwU8nw8QZsc=
=gPW5
-----END PGP SIGNATURE-----



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