Confusing automated translators.

Borcis borcis at users.ch
Tue Jul 15 15:33:47 EDT 2003


Francois Pinard wrote:
> 
> Automated translators which ignore punctuation are pretty fragile, too.
> Here is a case where the exact same words are used, besides punctuation.
> I read in one of Audouard's books ("De la connerie et des cons", if
> I remember correctly), about Montcalm, a French Canadian military of
> old times.  The history reports that he said:
> 
>    "Messieurs les Anglais, tirez les premiers!"
> 
> but Audouard wrote that he fears the correct writing should have been
> something like:
> 
>    "Messieurs!  Les Anglais...  Tirez les premiers!"
> 
> P.S. - I confess I would have more difficulty relating this one to Python.
> The word categorisation disambiguation program that I intend to rewrite
> in Python, one of these days, would (correctly) yield the same results
> for both sentences, so Python-wise for me, this is a non-issue! :-)

Reminds me of sitting, a dozen years ago, through the exposition, by an
amator, of the theory of word stemming he had spent years, in his
corner, to polish; applied to machine translation - that was the
research activity of the institute I happened to work for at the time.

Someone had invited a local University's professor of linguistics,
specialized in morphology. That person apparently felt he needed to
debunk the amator's proposal, and started pedantically expounding :
"Your proposal to decompose prefix-stem-suffix to translate them
separately and recompose the result won't work if the stems are
not etymologically related (somehow the topic was translation between
closely related languages). Take for instance the pair <faire>, <do>..."

-- Of course it's possible : "faisable", "doable" !

Cheers, Boris Borcic
--
python >>> filter(lambda W : W not in "ILLITERATE","BULLSHIT")






More information about the Python-list mailing list