Re: Ensuring Translation Quality

Op Vr, 2011-09-23 om 11:06 +0700 skryf Urmas:
> From: "Kathiravelu Pradeeban" <kk.pradeeban@gmail.com>
> Subject: Re: Ensuring Translation Quality
>
> >> 1) Remove .strings files for languages with translation less then
> 90% complete. We should preserve po-files in case somebody will want
> to improve them, obviously.
> >
> > That doesn't make any sense to do so. Many words are not translated
> > yet to the local languages. So we need time...
>
> They have more than enough time to update them. In fact, we have a
> handful of people who care to update their translation, but other do
> not. The maintanability and reasonable completeness is mandatory in
> other community projects like Gnome or KDE.

To my best knowledge incomplete localisations are released and welcomed
in at least GNOME where I am active. I was recently involved in an
effort to make it easier to make (better) incomplete localisations of
GNOME, so I am reasonably confident that it is the case:
http://translate.org.za/blogs/friedel/en/content/better-lies-about-gnome-localisation

I also feel that we can make reasonable arguments for releasing
incomplete localisations, as we have seen in this thread so far.

> The undeniable fact is that most translations are abandoned, have a
> little hope that someone will take interest in them again, and should
> not be included as their quality hurts Abiword image.

Depending on your definition of "abandoned" they might be. The truth is
that the world of FOSS is huge, and many of us are part of small groups
trying to localise as much of the FOSS software as our time allows. That
means some projects only receive attention on a few occasions rather
than daily/weekly or even monthly. Depending on the amount of change in
the source text of the software, it might or might not make a big
difference in how users experience the incomplete translations.

> >> 2) Remove Esperanto [eo] translation entirely due to its low
> quality, and considering the fact that Esperanto is toy, artificial
> language, and quality translation can not be assured without
> contributors' voluntarism.
> >
> > ...All the languages came from humans. So did
> > Esperanto. Calling Esperanto a toy language is an offense.
>
> From _people_, not from ophtalmologists of XIX century. I understand
> that a little part of several thousands of humankind is fond of
> conlangs, and delights playing with funny letters and other puerile
> stuff, but translations are to be made by people actually knowing the
> language. Quenya, Esperanto, Interlingua, Na'vi or Klingon have no
> native speakers, so translations can only be done in the form of
> vision each translator have on that conlang, without any guidance
> possible.

I also sometimes like to make jokes about Esperanto, but I don't know if
you meant this as a joke. As far as I know, Esperanto grammar is quite
well established, and there are native speakers, like George Soros,
who's Open Society Institute has been very kind to several FOSS projects
and groups.

In fact, my own language, Afrikaans, was only formally described in the
twentieth century, compared to the first Esperanto grammar book from
1887 (according to Wikipedia, anyway).

> Because we should avoid translations made by non-natives as we cannot
> judge their veritability, machine translations and translations into
> conlangs are not acceptable.

I think you already have a problem: you have no idea which of the
current translations were done by native speakers or with the help of
machine translation. I don't know what the process is for writing source
text, but we might have an even greater problem: some of the English
source text might have been written by non-native speakers of English.

So either we introduce an acceptance test (won't happen, right?) or we
rather try to focus on aspects of quality that are easier to
measure/fix. Tools like pofilter and poconflicts might provide a
starting point. Doing something like adding comments to some ambiguous
strings in the English text might help to avoid translation mistakes.

> > The localization of Esperanto is of very high quality...
>
> It is not. It is full of inconsistencies which are result of the
> language situation described above.

I haven't looked at the translation at all. If it is indeed of low
quality, I would propose that it is addressed like we do with any bug or
other shortcoming in the software: through a friendly request to the
previous translators, or in their absence, maybe reaching out to other
related communities where we can maybe find interested people. Maybe the
Esperanto group at GNOME would be interested, for example.

If speakers of Esperanto can confirm that it is indeed harmful to the
project and we can't improve it, then removing it from the build might
be the appropriate thing to do.

Keep well
Friedel

--
Recently on my blog:
http://translate.org.za/blogs/friedel/en/content/virtaal-070-released