Translators need not fear for their jobs just yet!!  None of the
programs i have seen so far are tremendously exiting in the execution of
their appointed tasks (and, i might add, the languages they claim to
know something about!!).  Some of the most frequent drawbacks seem to
be:
 
        1)  They have little or no concept of idioms or commonly used
allegories.  Hence, "China syndrome" (complete meltdown of a nuclear
power plant) may become "oriental disease".
        2)  They often take no account of context when deciding which
roughly
equiviant word or phrase to use.  Thus, people get to carry suitcases or
luggage under their eyes, instead of mere bags.  This is a serious
drawback, not easily corrected!
        3)  They have no idea which of two or more synonyms will sound
more
fitting in a given, often culturally-determined context.  "The poor
little boy" can come out looking like "the impoverished small
male-child".  This is closely related to #2.  To take context into
account, a vast amount of knowledge is nessessary, much of which we take
for granted.  Truly effective translation programs will have to be
gigantic, and will take many years yet to develop.
        4) They have no clues at all (well, most of them don't) as to
whether
or not their linguistic constructions actually make logical sense.  For
examples of this, take a peek at the translation selections below, from
Mr. Nisters' site.
        5)They are usually not programmed to translate technical or
specialized
terms and phrases.  Try using one to translate legal or mediacal jargon,
for example!!
        6)The one used by Alta Vista at least, seems to have great
difficulty
with compound words and other specialized constructions.
        7) Most programs seem about on par with 2-year-olds and teens
when it
comes to such refinements as grammer.
        8) Dialects??  Forget it!!  Try shoving a Robert Burns poem
through a
mechanical translator and see what you get back!!  The problem of idioms
and which words make sense or sound better in any given context, is
greatly compounded by the myriad dailectical variations of most
languages in use today.
Even to include a *basic* grasp of the more frequent English dialects in
a translation program would be a mammoth task, not likely to be
accomplished any time soon!!  Then, there is the problem of recognizing
which dialect or combination thereof is being used: very few relatively
"pure" dialects survive these days.
        9)  Every language has a certain "flow" to it when used in
everyday
conversations, and a somewhat different feel when used in literary,
academic, legal, or other specialized settings.  This sort of concept is
next to impossible to incorporate into a machine-based translation
system.  Thus, computerized translations come out sounding "awkward" at
best, and completely silly at worst, even when they actually manage to
make a bit of purely logical sense.
        10)  Many words and concepts simply do NOT translate well from
one language
to another, since they represent cultural contexts (ways of thinking and
seeing the world: ("umwelt" in German- itself a concept with no exact
English equivilant!)).  For example, the way we treat time as something
divided into discreet, indefinitely-subdividable units, is totally
foreign to the cultures (hence languages) of many native peoples.  To
try to translate or explain the concept of a "nano-second" to a Hopi
indian or most varieties of Australian aboriginal,  for example, would
be a task no machine could even HOPE to achieve, let alone tackling the
many French idioms and words which just don't make much sense to the
average New-worlder!!
 
        So, at the present state of the art, the best a machine can do
in the
translation field is to give a rather rough idea of what the passage in
question is about.  Human translators need not consider Unemployment
Insusurance as a career option anytime soon!
 
To illustrate a few of these points, i offer a few selections from
Helmut Nister's site, as "translated" by the babel-fish:
 
>                                           Hobby - Malakologe
>                                           Interests: Land and Suesswassermollusken of Europe and
>                                           Mittelmeerconchylien
>                                           Assembly areas: Land and Suesswassermollusken of Europe and Mittelmeermollusken.
>                                           Exchange contact with institutes and private collecting tanks abroad the in and. *See #2,4 and 7 above *
 
<Together with his nut/mother, Mrs. Dr. Irmgard
Nisters,honorary
coworker of the natural sciences of the Tiroler federal
statemuseum
Ferdinandeum and in this regard responsibly for the
there
Molluskensammlung revised, cares for and creates school collections
again.>
 
>                                         The Molluskenforschung in Tirol began 1849
>                                         in the years 1844 - by the brothers Pellegrino
>                                         and Giuseppe Strobel, whereby particularly
>                                         Giuseppe processed the Inn valley from
>                                         Hochfinstermuenz and its Nebentaeler (rocker
>                                         valley, Halltal, oh valley, Zillertal).
 
>                                         (single dump one for schools and private
>                                                    purposes wrote report over Tirol,
>                                          to the full extent with me to be requested can.)
 
        I rest my case, and will apologise to Mrs. Nisters on behalf of
AltaVista.
 
 
-Ross M.