[sword-devel] Translating the TSK

[Peter von Kaehne]

Hi guys!

I am experimenting at the moment with some new found skills around XML
handling and other Perl stuff.

I have created a Persian TSK module which might be of some interest to
some of you. The process could be used for any other language.

Basically the process is simple:

I pick out the references and translate them (libsword can do this for me)

I then take the keywords and get google to translate them. I take those
translations and check them against the relevant FarOPV verse text. If
the word is in the text, all is fine. If not, I highlight it as worth
checking.

The translation seems to get it absolutely right in about 30%. The
remainder are to 80% "near misses", i.e. perfectly valid synonyms. -
Khodavand instead of Khoda, shoru' instead of ebtedah. And then there
are real corkers "Let [there be light]" "Ejaseh bedehid" ("please allow
me") and some are simply not translated.

I have restricted myself to Gen:1 as I do not want to make myself
unpopular with google translate at this moment - once the process is
improved I will do in a slow way the whole module.

Ways I see forward are : using other online translation sources, testing
each on in turn against the text, and choose it if it works using a
locally installed dictionary (still looking for one) and finally
obviously publication and checking by hand (not something I can see
myself doing personally, but maybe there are volunteers).

What do you think?

Peter