Re: [boost] [locale] Review of Boost.Locale library

Subject: Re: [boost] [locale] Review of Boost.Locale library
From: Edward Diener (eldiener_at_[hidden])
Date: 2011-04-17 17:00:50

On 4/17/2011 2:40 PM, Peter Dimov wrote:
> Edward Diener wrote:
>> I do not think you can seriously argue that 'translation' from
>> language X to language Y is more correct if it must go from language X
>> to language E to language Y.
>
> Translation from N source languages to M target languages requires
> O(M*N) resources, whereas translation from N source languages to E to M
> targets requires O(M+N) resources.

The problem is translating between 1 source and multiple targets.

> Similarly, if you have a world with N
> languages, absent a universal second language E, people need to learn N
> languages to be able to communicate, and 2 if there exists an
> agreed-upon E. This is why we are writing in E in this very mailing list.

That's very nice but it does leave out all those who do not know E. Do 
you see that as a practical justification for a programming translation 
system ? I do not.

>
> In theory, it's more correct to translate from X to Y, but in practice,
> it's hard to find people who are simultaneously fluent enough in X
> software terminology and Y software terminology to be able to produce a
> high quality translation.

But it is easier to find someone who is fluent enough in X and E and Y 
to do so ?

>
> And in any event, the fact that the source texts are in E shouldn't
> preclude your translating from X to Y. You just take the translation
> text file for X which is basically a list of (E phrase, X phrase) pairs,
> and translate the X phrases to Y phrases. This requires no E knowledge
> on your part.

If only such a simplistic means of translating between 2 languages 
actually existed. I doubt it, even in the limited use of programming 
phrases. I know in my own area of expertise, literature, it does not 
exist but I will grant that the needs of a computer program may be much 
less linguistic precision. But even a computer program still deals in 
end-users who want to see text that makes sense to them in their own 
language rather than pig-latin type gobbledygook which they will laugh 
at. We are talking about computer users who will pay for a computer 
program in their own language and commercial companies who can not 
deliver poor quality in that regard and hope to be successful.