Re: Ensuring Translation Quality

On Tue, Oct 4, 2011 at 9:59 AM, Urmas <davian818@gmail.com> wrote:
> From: "Urmas" <davian818@gmail.com>
> Subject: Ensuring Translation Quality
>
> The following locales have _less than 65%_ of strings translated (site statistics is not correct as it includes fuzzy strings):

The misleading statistics are disappointing, I would note that Pootle
does a much better job of presenting clear statistics on translated,
untranslated, and fuzzy strings.

>
> es-MX

I quite agree with not building an es-MX version, although not for the
reason stated. I performed a merge on es_ES and es_MX when I imported
them into Pootle, fserrador reviewed the merged file and worked to
resolve conflicts between the two PO files.

The exclusion of an es-MX variant is not meant as an insult to the
Mexican localizers who worked to create it, in fact, fserrador
acknowledged that a number of the es_MX strings were improvements over
the previous es_ES entries. It is merely an acknowldgement that the
UI of a word processor does not contain the type of cultural context
that commonly distinguishes national variants of Spanish, nor does
written Spanish embody the so-called yeismo or seseo that so clearly
distinguish spoken forms.

I would note the the Sugar Labs / OLPC project uses a single es_ES
translation at the unanimous request of our Spanish localizers (from
all over the Spanish speaking world) as well as the education
ministries of Peru, Uruguay and Parguay who are purchasing millions of
XO laptops and deploying them in their schools.

> uk-UA
> mg-MG
> ta-IN
> lv-LV
> am-ET
> lt-LT
> he-IL
> jbo
> ja-JP
>

In an open source project, the correct response to low language
coverage levels is to recruit more localizers, you do not accomplish
that by ignoring the work of those you already have.

> Additionally these locales have problems:

> eo : is a conlang and not worthy including (unless proven that "translator" is a "native speaker"*)

I'm not sure why we should substitute your judgement of what makes a
language "worthy" when Esperanto has gained sufficient acceptance to
be included in the official ISO standard list of languages ISO-639
(and variants thereof).

However, for the sake of rebutting your point, I would point out that:

There are in fact "native speakers" of Esperanto.
https://secure.wikimedia.org/wikipedia/en/wiki/Native_Esperanto_speakers

However I doubt that billionaire George Soros will be spending any
time localizing AbiWord and I know of no practical means of requiring
proof of mother tongue in translation efforts. A translation is
judged on it's own merits by co-linguists, not the ethnic or national
origins of the translator.

However, for numbers ranging in the thousands to millions it an
acquired second language.
https://secure.wikimedia.org/wikipedia/en/wiki/Esperanto

Esperanto is indeed a constructed language, which suggests that the
rules for determining the correctness of translations require no
special native knowledge evolved over aeons like 'natural" languages.
In the end, all languages are constructed, some are just newer than
others. I would also note that you've make no similar argument
against Lojban (jbo), which is rather inconsistent on your part.

> be@latin : unless some official resource confirms Latin-based orthography to be used for Belarus language.

A history of the use of a Latin orthography for Belarusian is available here:
https://secure.wikimedia.org/wikipedia/en/wiki/Belarusian_Latin_alphabet

Go to the Gnome L10n team for Belarusian (and scroll down a little)
and you will see that they themselves recognize be@latin as a valid
alternate orthography for Belarusian

http://l10n.gnome.org/teams/be/

> Until these problems are corrected, I reiterate on my suggestion do not include them in future releases.

I reiterate my objections to such a capricious and arbitrary move,
with the exception noted above for es_MX

> ------------------------------------
> * A proof that translator's mental age is greater than 16 would be nice too,,, but that's rather paramount.

The only such measure I can imagine would be an assessment of
emotional maturity based on the author's writings. One of the more
reliable hints of lower mental age would include a propensity to
incite conflict as opposed to seeking clarity and consensus.

cjl
Sugar Labs Translation Team Coordinator
Received on Tue Oct 4 17:39:38 2011