Re: [Translation-i18n] Proposal for declinations in gettext

[Danilo Segan <dsegan gmx net>]

Deniz Akkus Kanca wrote:

>If the word takes a pronoun, such as my, you add the suffix for the pronoun 
>(genitive) and keep the declinations at the end: 
>
>evim - my house
>evimde - at my house
>evimden - from my house
>evime - to my house
>evimi - objectify as in I am painting my house - evimi boyuyorum
>  
>
My experience in looking at and translating strings is that composition 
is usually done as:
"Painting %s", where "%s" would take the form of "my house". In this and 
similar cases, I guess declinations would even help you.

Instead of "evim boyuyorum", you'd be able to get "evimi boyoyorum" (ok, 
some part of "boyoyorum" will probably change, because it's not realted 
with "I" in my example, but you get the point).

Of course, I realize that it would not work for all cases. As a matter 
of fact, it wouldn't solve all problems one might have with my (Serbian) 
language (verbs changing form depending on gender, who is doing the job 
[I, you, (s)he/it, we, you, they]). But, the proposal was not made to 
solve that problems, rather only a small set of problems caused with 
composition which uses the form:
"verb %s", and "%s" is any "noun-form" (adjectives, nouns, pronouns,... 
eg. "my wonderful house").

In your examples, regarding Serbian language, the word "my" would also 
go through declinations, genders, etc, so if the composition is used as 
you suggested, there would be no way to translate it correctly either.

>>Your feature would encourage the programmers to keep constructing the sort
>>of strings that you present in the beginning. My point is that this will
>>make the programs impossible to translate for many target languages.
>>    
>>
>
>Exactly. The minimum translatable unit is a stand-alone clause or sentence. 
>This principle works for all languages. I despair at trying to explain 
>anything else to mono-lingual developers. 
>  
>
The last thing I wanted to do is to "encourage" programmers to use bad 
strings. But, my small experience with the kinds of strings programmers 
do produce is that they are usually not that bad, just "somewhat" bad, 
and consist of separating verbs and nouns. In many languages, as I have 
seen on this list, these cases can be solved using the 
declinations-mechanism.

As I've pointed out above, even though I am "mono-lingual" (not counting 
the English, and long-unused Deutsch), this solution is not "The 
Solution for all problems of Translation to My Language", but rather a 
simple solution for problems that exist, and are pretty annoying.

I don't know how any of you have translated the Metacity strings of the 
form:
"Switch to %s", where %s is "Workspace %d" (which is also used as 
standalone for naming the workspaces),
but the solutions here are (because we can have an arbitrary number of 
workspaces) either what I proposed, or context markers that Miloslav 
Trmac mentioned:

"[Switch to ]workspace %d"
"Workspace %d" # for nominative
"Switch to %s" # here, only the first string would be used, so no problems

In my "solution", it would become:

"Workspace %d" # and translation includes all the needed declinations
"Switch to %<identifier-of-particular-word-form>s"

I am not certain what is better, but any of these would work. The second 
is easier on the programmers, but harder on translators, while the other 
is vice-versa. Since I have found out that translators *are* willing to 
work harder to get the correct translation for their beloved language 
:-), I prefer the second solution (it would also make some 
translator-unfriendly programs a bit more friendly without (big) 
intervention of the programmer).


Cheers,
Danilo