Re: Latin Belarusian

[Ihar Hrachyshka]

On 10/27/07, Bruno Haible <address@hidden> wrote:
>
> Hello Ihar,
>
> I have heard that you are setting up a translation team for Belarusian
> with
> Latin script. I also see in [1] that there is a standard way to
> transliterate
> Belarusian text from Cyrillic to Latin script.


Yes, there is a half-workable script that can do some conversion work
automatically though there are some circumstances that prevent fully
automized conversion. F.e., Cyrillic variant of Belarusian language adapts
all foreign words phonetically, and that's not the universal case for Latin
variant (f.e. such words as "interview", "widget", "tag", "debug" are not
adapted in Latin subscript as in Cyrillic). There are also some problems
with po-messages where a word is divided by shortcut symbol (like
"Raści_ahni") that will be converted incorrectly (like "Расьці_агні", not
"Расьц_ягні") though this kind of problem can be resolved programmatically.
And the last and - maybe - the largest problem is that Latin and Cyrillic
communities have different views on some fundamental l10n questions (f.e.
what kind of form to use when translating commands like "Close" -
"Зачыніць/Закрыць" or "Зачыні/Закрый" (infinitive or imperative), different
terminology and so on.

That's why it would be great to distribute conversion script with
gettext-tools package though I'm absolutely sure that we shouldn't use it
for automatic conversion. There are Ruby scripts in attachement that are
capable to do the conversion (into both sides). Should it be rewritten into
any other language to be distributable with main gettext package? Then what
language is it?

Is it true that such a transliteration can be done automatically? If yes,
> then we can implement this transliteration as a filter for the 'msgfilter'
> program. For Serbian, such a conversion is indeed possible to do
> automatically,
> and it's implemented in GNU gettext already:
>
>     $ recode-sr-latin --help
>     Usage: recode-sr-latin [OPTION]
>
>     Recode Serbian text from Cyrillic to Latin script.
>     The input text is read from standard input.  The converted text is
> output to
>     standard output.
>
>     Informative output:
>       -h, --help                  display this help and exit
>       -V, --version               output version information and exit
>
>     Report bugs to <address@hidden>.
>
>     $ echo 'Данило Шеган' | recode-sr-latin
>     Danilo Šegan


Yes, there is a script but we shouldn't automatize its work. It's not
capable to resolve all our l10n issues.

msgfilter can use this program to recode an entire PO file at once.
>
> Do you think such a converter would be possible and useful for Belarusian
> as well?


Yes, it would be useful to be reachable with one movement for many
Belarusian localizers but nor more, neither less.

Bruno
>
>
> [1] http://en.wikipedia.org/wiki/Romanization_of_Belarusian
>
>

lacinka-pack.tar.gz
Description: GNU Zip compressed data