Re: Ensuring Translation Quality

On Thu, Sep 22, 2011 at 11:47 AM, F Wolff <friedel@translate.org.za> wrote:
>
>
> Op Do, 2011-09-22 om 18:52 +0700 skryf Urmas:
>> To improve the translation quality I propose:
>>
>> 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.
>
> Hi
>
> I'm strongly against this idea. When several of us translated Abiword as
> part of ANLoc, we actively worked on reducing the number of strings to
> the most visible so that we create a reasonably good coverage of the UI
> without needing a very long time to do it.
>
> I use incompletely translated software everyday of my life, and it is
> still great to have most of it in my language. Some strings might just
> be incomplete because a volunteer didn't manage to give it attention for
> a while, or because the English is unclear and it is better to leave it
> out until confirmed.
>
> Please leave incomplete translations. We won't remove ODF support if
> there is a small gap in the implementation, right?
>
> Friedel

I strongly agree with Friedel, your proposal would exclude about 25
languages from the release, including a number that are very important
to OLPC (which redistributes full AbiWord in the Gnome boot of builds.
 I would however agree to dropping es_MX in favor of the unified
es_ES work that was done mostly by fserrador and that had been
discussed on this list.

I think one way to improve coverage is to publish what you have and if
the user is not satisfied with the level of coverage, then they may
very well be motivated to contribute to the L10n effort (to scratch
their own itch). Eliminating the partial PO files at nearly any level
above about 5% to 10% would be too high a bar, but even in that case,
there are exceptions (i.e. en_IE and en_AU).

AbiWord compares very favorably to other "competitive" FOSS word
processing packages in it's language coverage and I think it can be a
real source of competitive advantage for "mindshare" and user adoption
and attracting contributors.

I do think that known issues with quality of L10n should be addressed,
but I do not think of low-levels of coverage as a bug, but an
opportunity to recruit localizers. Big distro projects may worry a
lot about "supported languages" and cut-off levels for coverage, but I
do not think that AbiWord's reputation will suffer for trying to reach
out in as many languages as it can, even if in some cases, that
attempt reflects a currently lower-than-optimal coverage level.

On a L10n community note, I think it would be discouraging to
beginning teams not to see their hard work (as partial as it may be)
reflected in current builds. Setting such a high bar for their work
to be published is more likely to cause them to go and work elsewhere
than it is to spur them to re-double their efforts to achieve that
level of coverage.

cjl

According to http://abisource.com/contribute/translate/

The only languages below 50% coverage are en_IE and en_AU and there
are obvious reasons that they might only chose to translate a small
number of strings and accept the lang-en default strings to display.

Language Status
en-US 99.00%
eo 98.00%
gl 98.00%
da-DK 98.00%
ff 98.00%
sk-SK 98.00%
de-DE 98.00%
kk-KZ 98.00%
ar 98.00%
es-ES 98.00%
sl-SI 98.00%
br-FR 98.00%
nl-NL 98.00%
ru-RU 98.00%
ro-RO 98.00%
wo-SN 97.00%
zh-CN 97.00%
ca-ES 97.00%
ms-MY 97.00%
mnk-SN 97.00%
fr-FR 97.00%
ast-ES 97.00%
eu-ES 97.00%
bg-BG 96.00%
en-GB 96.00%
sq-AL 96.00%
cy-GB 96.00%
aym-BO 96.00%
qul-BO 96.00%
quh-BO 96.00%
en-CA 96.00%
be@latin 96.00%
et 96.00%
it-IT 96.00%
ps 96.00%
hr-HR 96.00%
ga-IE 96.00%
ayc-BO 96.00%
ne-NP 96.00%
pt-BR 95.00%
pl-PL 95.00%
cs-CZ 95.00%
el-GR 95.00%
yi 94.00%
mk-MK 94.00%
zh-TW 94.00%
hu-HU 93.00%
ku 93.00%
nb-NO 92.00%
nn-NO 90.00%

pt-PT 89.00%
sv-SE 89.00%
sr@latin 87.00%
sr 87.00%
sc-IT 85.00%
fi-FI 83.00%
tr-TR 81.00%
ko 81.00%
af-ZA 80.00%
lt-LT 79.00%
id-ID 75.00%
be-BY 74.00%
am-ET 74.00%
vi-VN 73.00%
ja-JP 70.00%
de-CH 70.00%
zh-HK 67.00%
lv-LV 66.00%
uk-UA 65.00%
mg-MG 62.00%
he-IL 61.00%
es-MX 61.00%
jbo 53.00%
ta-IN 53.00%
en-IE 26.00%
en-AU 3.00%
Received on Thu Sep 22 18:46:04 2011