Tomer Chachamu wrote:
On Wed, 23 Feb 2005 18:32:59 +0200, Andy Rabagliati
<andyr(a)wizzy.com> wrote:
If we plug away at the en: wikipedia, adding
African-related articles,
maybe a couple of years down the line we will be able to automatically
translate to Xhosa. I think that is a /much/ better use of everyone's
time than writing a Xhosa wikipedia.
I think not. As your KDE example showed, translation is far more
complicated than simple dictionary lookups. Xhosa is only spoken by
7.9m speakers (mainly in South Africa) and with so few speakers (and
so little commercial possibility for making a translator) I would
imagine that Xhosa may never be able to be translated.
Actually, I just read that it is similar to Zulu, which probably adds
some more "speakers". Still, I think development of a machine
translation program is unlikely.
http://en.wikipedia.org/wiki/Machine_translation
http://en.wikipedia.org/wiki/Xhosa_language
I fully agree with the fact that translation is far more compilcated
than many think - babelfish does not at all do the trick ;-)
Never ever a machine will be able to translate correctly - not now, not
tomorrow, not in 100 years - it's a dream many have - sorry, I must
destroy it. A spoken and "living" language changes style and partly even
writing and grammar at least every 20 years. It's an evolutionary
process and a machine translator will only be as good as the human who
fed it. It can give raw translations, but they will never be perfect.
Often the same sentence used in another context needs a different
translation as it has a different meaning. So whatever anyone thinks or
belives: you will always need human beings to transmit a concept from
one language to the other.
My 2 cts.
Sabine
(professional translator - and not even a bit worrying about computers
taking over high qualified work :-) go ahead programming machine
translation software - thank you! - you help me to show that humans are
the better translators and are woth every cent a customer pays :-)