Re: String review



Hi,

Danilo Šegan wrote:
Today at 21:08, Vincent Untz wrote:

The difficult part is, however, to make maintainers use this database
when they add new strings. I don't know how we could do this.

I think it would be simple to make them use it provided we have a good
"similarity" matching algorithm.


Also, this will only make sense doing if we provide such a mechanism
in the first place (or is anyone interested in hand-selecting all the
cases? :).

perhaps something along the lines of building blocks/autotext entries? I guess there might be open source algorithms to borrow from there.


But all such schemes would require either string authors using some certain tool, or some certain database, or that some kind of string (similitude) review can be run on all strings before releasing them for translation.

Regardless how well such guidelines are adhered to, translators *will remain* a crucial part in catching such inconsistencies, as long as string generation doesn't somehow get completely formalized (using a controlled corpora).

In my translating experience, hardly even controlled corpora can stop everything, since there's still someone who's got to control the corpora... (I have translated a lot of stuff made that way and found disturbing inconsistencies).

A good tool for catching these things, would be a reverse consistency checker, that checks terminology/string consistency as viewed *from* the translated (target) language.

That way, e. g. a string or term that turns up as duplicate in the target language, but has two different sources is most likely a terminological inconsistency, either in the translation (collision) or in the source language (like some of Christians examples).

The strings that get translated pass through the semantic-logical filter that is the translators brain, and surprisingly often, what is the essence of the strings more or less always comes out in a uniform way in the target language.

With such a tool, that fact can be put to good use, even upstream.

BR,
Gudmund


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]