Pages in topic:   [1 2] >
Poll: When will machine translation be infallible?
Thread poster: ProZ.com Staff

ProZ.com Staff
Local time: 09:10
SITE STAFF
Aug 8, 2009

This forum topic is for the discussion of the poll question "When will machine translation be infallible?".

This poll was originally submitted by Transperts

View the poll here

A forum topic will appear each time a new poll is run. For more information, see: http://proz.com/topic/33629


 

Yasutomo Kanazawa  Identity Verified
Local time: 01:10
Never Aug 8, 2009

Machine translations can never convey the subtleties behind the sentences, plus humans must continuously update the new words created and foreign loan words to catch up with the latest glossaries.

 

neilmac  Identity Verified
Spain
Local time: 18:10
Spanish to English
+ ...
Never Aug 8, 2009

... although the temptation is to say "very soon" in the somewhat vain hope that hordes of market-busting competitors will give up in despair and recycle themselves, leaving the world a more open and level playing field for the true professionals who value translation as rather more than just another sordid business angle.

A writeable machine translation program would be nice though, something like a hybrid of Systran and TM but which we could modify as we went along, correcting the worst of its lexical howlers to develop a working tool that avoids derision...


 

raptisi  Identity Verified
Spain
Local time: 19:10
English to Greek
+ ...
Never Aug 8, 2009

MAchine translation will never be infallible, it is made by humans, and nobody can assert that humans are infallible. On the more practical standpoint, the language is something that is alive, it is constantly evolving. Software made for the language of today, will not be applicable in the future.

 

Parrot  Identity Verified
Spain
Local time: 18:10
Member
Spanish to English
+ ...
Are we presuming human translation is infallible? Aug 8, 2009

Written language is codified one (or several) removes from actual living context (i.e., it is never around exactly in the context it is applied to and merely reflects an approximation). Humans can make mistakes on that score. What's the chance a machine can do better, with a human programmer?

 

Sophie Dzhygir  Identity Verified
France
Local time: 18:10
Member (2007)
German to French
+ ...
Never Aug 8, 2009

I think it will never be infaillible (neither are humans, as Parrot said), but in some years (I don't know how many), I think it will do very good job.

 

Simon Bruni  Identity Verified
United Kingdom
Local time: 17:10
Member (2009)
Spanish to English
Never say never but... Aug 8, 2009

A computer programme needs a set of rules on which to base its output and no single set of rules, no matter how big, can cover every possible alternative of a living and evolving language (which are potentially infinite). And this doesn't even take into account extra-textual information that may influence the output (e.g. the client's objectives and the particular readership the text is intended for) - another potentially infinite set of factors.

 

José Henrique Lamensdorf  Identity Verified
Brazil
Local time: 13:10
English to Portuguese
+ ...
After word processors... Aug 8, 2009

... are no longer limited to spelling and grammar checking, but become capable of verifying consistency of ideas, soundness of reasoning, stability of tone, political correctness, targeted audience readability, and other things in written text.

The problem is the cost/benefit of making that happen. The "interested parties" are struggling to bring human translation costs down to 1¢/word. If they succeed, how many gazillion words will have to be translated for less than that to recoup the hefty investment in developing the necessary hardware/software to accomplish it?


 

Miroslav Jeftic  Identity Verified
Local time: 18:10
English to Serbian
+ ...
Hm. Aug 8, 2009

Interesting to see that almost everybody chose never.icon_smile.gif While I agree that the time when a machine will be able to translate a book or a poem isn't near, regarding the translation of instruction manuals, help files, contracts and agreements, etc, I wouldn't bet we aren't going to see that in our life time.

[Edited at 2009-08-08 11:29 GMT]


 

David Russi  Identity Verified
United States
Local time: 10:10
English to Spanish
+ ...
Very little is infallible Aug 8, 2009

especially in the realm of words. Had the question been about about really good, or practically perfect, we probably would see a more nuanced response

 

Tim Drayton  Identity Verified
Cyprus
Local time: 19:10
Turkish to English
+ ...
Other Aug 8, 2009

I opted for 'other'. I really do not know. Obviously as one who earns his living from translation I hope that the output from machine translation will never rival the work of a competent professional. Cotton spinners in 18th century Lancashire were convinced that spinning was too complex ever to be mechanised. They turned out to be wrong. I wonder if my hope will turn out to be equally misplaced.

 

T F F  Identity Verified
Italy
Local time: 18:10
English to Italian
The marketing will do the trick... Aug 8, 2009

José Henrique Lamensdorf wrote:
... are no longer limited to spelling and grammar checking, but become capable of verifying consistency of ideas, soundness of reasoning, stability of tone, political correctness, targeted audience readability, and other things in written text.


Brilliant!...That's why I voted "never"!

Unfortunately, the BIG PLAN is not about raising the quality of Machine Translation, but lowering the Human expectations...

And so: take an "Universal TM" (like Google Translate), add a cutting edge algorithm and you will have a good (poor) recycled translation (still no problems for us at the moment!)
Now, think what a good marketing could do then expecially to Translation Buyers with low budgets!
-> LSPs will have no need to further buy translation from us...
-> We will then go around like zombies keeping on saying "Human translators are irreplaceable, buy our services!"

...Too pessimistic? Maybe, or maybe not...


 
Never????? Aug 8, 2009

Tim Drayton wrote:

I opted for 'other'. I really do not know. Obviously as one who earns his living from translation I hope that the output from machine translation will never rival the work of a competent professional. Cotton spinners in 18th century Lancashire were convinced that spinning was too complex ever to be mechanised. They turned out to be wrong. I wonder if my hope will turn out to be equally misplaced.


Very good point! Our response "never" well could be biased by our own occupational interest as conflict of interest clearly present here.

Just for the sake of it, I would like to add one more episode to Tim's example ........ When typewriter was invented, menfolks said this machine was too sophisticated for women to handle.
As many of us know such time was followed by another era when typist equal woman's occupation. Those were the days when things were not very politically correct.

Another point I wish to make is this.
When cross-training and interdisciplinary practices are more advanted in the future in educational systems, electronical engineers may learn how to write like a human, and thereby become able to invent translation software which can deal with more nuances of human psychology manifested in the arts of languages and words.:cool:

Never say never, as Simon said. The day will come. Let's hope that is not in our lifetime.



[Edited at 2009-08-08 15:20 GMT]


 

Andrew Levine  Identity Verified
United States
Local time: 12:10
Member (2007)
French to English
+ ...
Wrong question Aug 8, 2009

It's beside the point to ask whether machine translation will ever be "infallible." The world already relies on fallible human translators. And of course every machine is fallible in some way, in whatever function it performs.

The more pertinent question is whether (fallible) machine translation will ever be a useful substitute for (fallible) human translators.


 

Emin Arı  Identity Verified
Turkey
Local time: 19:10
Member
English to Turkish
+ ...
main question is "will artificial intelligence be successfull?" Aug 8, 2009

In fact, for the time being, machine translation is a mix of data base and artifical intellgience. As the artifical intelligence is not successfull, most of the job relies on data base. For this reason, we see such an awkward results of machine translation, (should I say, funny results?). If somehow, the scientists will be able to create artifcial intellience, then machine translation will also be infallible. However, the machine will not be machine, it will be a creature with even a soul. that is the dilemma.

 
Pages in topic:   [1 2] >


To report site rules violations or get help, contact a site moderator:

Moderator(s) of this forum
Jared Tabor[Call to this topic]

You can also contact site staff by submitting a support request »

Poll: When will machine translation be infallible?

Advanced search






WordFinder Unlimited
For clarity and excellence

WordFinder is the leading dictionary service that gives you the words you want anywhere, anytime. Access 260+ dictionaries from the world's leading dictionary publishers in virtually any device. Find the right word anywhere, anytime - online or offline.

More info »
SDL MultiTerm 2017
Guarantee a unified, consistent and high-quality translation with terminology software by the industry leaders.

SDL MultiTerm 2017 allows translators to create one central location to store and manage multilingual terminology, and with SDL MultiTerm Extract 2017 you can automatically create term lists from your existing documentation to save time.

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search