Re: GLADE, Linux Kernel Modules etc,

[Larry Elmore <ljelmore_@_comcast_._net>]

sk wrote:
> Georg Bauhaus <sb463ba@l1-hrz.uni-duisburg.de>:
> 
>  > Traduttore, traditore in Watzlawick's classic notwithstanding
> I do not have a classic education so I have no understanding of
> the reference.
> 
> However, and since this is a "comp.lang.*" usenet forum ...
> 
>  > ... if you know French (or Latin) ...
> 
> ... and my previous comments concerning "google" translations.
> 
> I had, a long time ago, a thought : if I studied Latin, which
> I believe has a more rigid structure than English, French,
> German and Spanish, could a reasonable human language
> translator be developed, better than those which I have tried
> on the internet ?
> 
> The idea being that any of the above 4 languages would be
> translated to Latin and then translated into the desired
> language ?
> 
> English  >--+               +--> English
>             |               |
> French   >--+               +--> French
>             |---> Latin --->|
> German   >--+               +--> German
>             |               |
> Spanish  >--+               +--> Spanish
> 
> Does Latin have a more "logical" rigid structure ? And if so,
> would it be a reasonable middle language for a more universal
> translator ?

No, I fear it wouldn't. Latin is a good deal more 'regular' in some ways 
than English (or any of the other Germanic languages), but there's still 
a lot of irregular parts. While it has more structure in some ways, this 
allows greater flexibility in other ways. Word order is a lot freer in 
Latin than in English and serves more as a hint on emphasis than core 
meaning. Latin is very much a human languages and has all the foibles of 
the others, just expressed in different ways.

Perhaps a language like lojban might work better as an intermediate 
language, but I have real doubts about _consistently_ _good_ machine 
translation becoming possible before we have some form of AI.

--Larry