[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference thebay::joyoflex

Title:The Joy of Lex
Notice:A Notes File even your grammar could love
Moderator:THEBAY::SYSTEM
Created:Fri Feb 28 1986
Last Modified:Mon Jun 02 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:1192
Total number of notes:42769

1135.0. "Help, Franco-Deutscho-Espano-Italo-phones in our web..." by LJSRV2::KALIKOW (Technology Hunter/Gatherer) Mon Feb 13 1995 18:39

                    (Crossposted in ::DIGITAL and ::VALBONNE)
    
    As part of my work as a technocatalyst in Rose Ann Giordano's Internet
    Business Group (part of Bill Strecker's Advanced Technology Group),
    I've been doing some informal evaluation of the mechanical translation
    engine produced by an outfit called Logos Corporation.  
    
    Without saying much more than "they do as complete a job as I've heard
    of in this area," and "the raw translations their algorithms produce
    are sufficient for many purposes, but for the Real Thing, the Logos
    engine should be thought of as a Power Tool which aids a human
    translator in producing a polished result in far less time than by
    creating a translation _de novo,_" I would like to ask you, my esteemed
    THEBAY::JOYOFLEX and HUMANE::DIGITAL brethren, to check out the
    translations of a "guinea-pig" article chosen at semirandom from our
    Corporate Web-Server.
    
    These translations are into French, German, Spanish, and Italian.
    
    The words that were not known to the various "jargon dictionaries" that
    were in use are set off typographically from the bulk of the text (and
    obviously no tuning has been done for the DIGITAL HW/SW/PR
    environment).  It appears to me that if some minimal tuning were to be
    done, then the number of translated tokens would be around 99%...
    
    But of course this doesn't take into account possible mistranslations. 
    These are most amusing around the names (Gail Grant, Dan Cross, Al
    Gore, etc., have some marvelous things done to them) but I'm sure it
    extends to other areas as well.
    
    I would really appreciate wiser, more multilingual heads than mine
    taking a look at these raw translations.  Any and all comments are
    welcome.  I'm particularly interested in generalities like "Would even
    the RAW translations (suitably augmented by obvious improvements in a
    jargon dictionary) be of use?  How good are the raw translations?  How
    much would you pay...  What kind of polishing would be needed...  How
    expensive is that likely to be...  Where are the obvious errors of
    sense...  Are there any regionalisms present or absent in the
    translations...  Things I can't even imagine but YOU can, by way of
    critiques...????  Who's their competition, are they any better (can you
    give me some proof)? ...
    
    Unfortunately, in order to view these translations, you'll need a Web
    Browser -- they use HTML markup to achieve the accented characters.
    
    So, if you will, please fire up your browser (even LYNX will work OK)
    and point it at 
    
       http://www.ljo.dec.com/IBG/people/kalikow/woad/logos-intro.html
    
    And then, fire away...  Feel free to post your comments as followups
    here, or via EMail to me at DRDAN::KALIKOW or kalikow@ljo.dec.com
    
    Please be aware of the fact that these files are being worked on from
    time to time, and they may change from visit to visit...
    
              Also please keep in mind that all this information
                        is DIGITAL Internal Use Only.
    
    Thanks for any time you can spend critiquing them!!
    
    Dan
    
    PS -- If the collective opinion is that these guys do an OK job, I'd be
    interested in receiving possible similar-in-quantity-and-subject-matter
    German articles, for use as Guinea Pigs for the LOGOS engine in its
    German-to-English, -French, -Italian,... incarnation.  If that issue
    arises, I'll post a followup notice saying that nominations are open &
    submissions are welcome.
                                  
T.RTitleUserPersonal
Name
DateLines
1135.1JRDV04::DIAMONDsegmentation fault (california dumped)Tue Feb 14 1995 01:2410
    >Unfortunately, in order to view these translations, you'll need a Web
    >Browser -- they use HTML markup to achieve the accented characters.
    
    Oh great, this company can't even produce a copy of mosaic that
    retains accented characters in the HTML markup.  (Yup, I see words
    like    E  s  p  a  square-box  l    just like in DECterm windows.)
    
    Not your fault of course.
    
    -- Norman Diamond
1135.248360::MAILLARDDenis MAILLARDTue Feb 14 1995 06:0016
    Re .0: Dan, I've had a look at the French translation. I'd say it is
    practically unusable. However, under closer examination, one can see
    that most (but by far not all) of the problems come from the sentences
    where proper names and technical terms appear. All proper names that
    can be given any substantive meaning have generated surrealist
    translations. A good number of the technical terms have incorrect
    translations as well. If you look at the English text, you'll realise
    that this already makes most of the translation unusable. When you take
    all this into account, you still have another problem: English
    idiotisms give erroneous translations, and places where French
    idiotisms should take place look very queer to a French reader. With a
    text such as the one you have used, the general result is only
    understandable if you can compare it word for word with the English
    original (i.e. totally useless), but it will at least provide you a
    good laugh...
    			Denis.
1135.3HLDE01::SOEMBA::RIKMostly HarmlessTue Feb 14 1995 06:247
Having looked at the German version I have to agree with Dennis.

Some errors are mindboggling as well, in the sense of 'how did the program
decide to come up with _that_'.

                                                        
                                                  - Rik -
1135.4BBRDGE::LOVELLWed Feb 15 1995 03:4611
    I made a conscious effort to read the French prior to the English and
    despite the fact that I thought it was grammatically pretty sound, I
    had great difficulty in general comprehension until I referred to 
    the English original.
    
    Far be it for me to poke fun at Denis here whose English is vastly
    superior to my French but I had to chuckle over the "idiotisms" comment.  
    Maybe this whole exercise should be regarded as an idiotism.
    
    /Chris.
    
1135.548360::MAILLARDDenis MAILLARDThu Feb 16 1995 09:4212
    Re .4: Sorry, Chris, I incorrectly assumed the French word "idiotisme"
    also existed under the same form in English. A check in a dictionnary
    just taught me that English uses "idiomatic expression" instead. But
    the word sure gives a good feeling of what the French translation looks
    like. On further reading, I saw that there were also serious
    mistranslations that were not related to the two main causes I noted in
    my previous reply. It probably involves already quite a hell of a job
    to realise a translator able to come out with such a translation, but
    it still is not able to do the trick. I seriously doubt that even a non
    technical text without any misleading proper names would even approach
    something usable by an average reader.
    			Denis.
1135.6No hay nada que hacer con la version CastellanaTAVIS::JUANSun Feb 19 1995 10:2743
	Hi Dan!

	First of all I'd like to give you some short info about myself. I was
	born and rised in Argentina: I am a native Spanish speaker.
	For the last 12 years I am located in Israel - and this is only 25%
	of my life. 

	I read the raw Spanish version of the article on the INTERNET. I know
	the subject and was able to decipher the text. The translated words
	were standard Spanish. The syntax... well, there was no recognizable 
	Spanish syntax at all.

	In order to translate the article in question I'd need to go back to
	the English original. The raw translation is almost useless. It's 
	quality is that of any WASP high school graduate, with 2 years Spanish
	and a good Eglish-Spanish Dictionnary.

	It was very funny to see the VP of the USA being called Al "the Knife".

	I didn't see a problem with regionalisms or Jargon. The text was so
	un-natural that it was impossible to see if they were using a regional
	expression or jargon.

	Regards,

	Juan-Carlos Kiel
==============================================================================
>    
>    welcome.  I'm particularly interested in generalities like "Would even
>    the RAW translations (suitably augmented by obvious improvements in a
>    jargon dictionary) be of use?  

	I would not use the RAW translations for any use - but as making fun!


>    How good are the raw translations?  How
>    much would you pay...  

	I would suggest my company NOT to pay for such a translator, unless 
	other, better, translations would be brought as justification. Maybe 
	there are other kinds of text that suit better for their translation.

	JCK
1135.7vicino, ma senza cigarroFORTY2::KNOWLESMon Mar 13 1995 09:1021
    As a one-time interpreter from French and Spanish, I've sent Dan my
    detailed comments by mail. While I agree that the translations are
    laughable (I came across the French word `idiotisme' years ago and
    didn't bother looking it up in a dictionary because the context made
    its apparent meaning so appropriate, and I put it down as just another
    of those words that meant something clear but didn't translate well -
    I'm disappointed to know it means `idiomatic expression') I'd find them
    useful if I had to do a translation (in which case I'd have the English
    original alongside and - as others have said would be necessary - refer
    to that). If the service is going to be automatic, isn't the fact that
    these translations are raw irrelevant? Anything in an automatic system
    would have to be raw, therefore not very useful.
    
    I've never heard of Logos. Are they a big name? I read a newspaper
    article quite recently that said the biggest had just swallowed up the
    next biggest. The swallower or the swallowee was Global-something -
    Globalnet, perhaps (a PC package). But I doubt if `the best around' -
    even if the biggest were the best - could be useable; I've never seen
    an automatic translation that was of use to anyone but a translator.
    
    b