Forums

Full Version: Has the threat become more fearsome?
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Buen morning, Yves!

Your presentation at the Law School in 2006 was really very interesting.

Could you share with us your explanation on why we still defeat the machine at translating? (the example of the long mathematical calculation which is solved faster [for the time being] by a human being because of the way in which we approach it)

With news like the one below (Reuters - March 29), is our end nearer?

Thank you!

Worried Au Sad


MOUNTAIN VIEW, California (Reuters) - In Google Inc.'s (GOOG.O) vision of the future, people will be able to translate documents instantly into the world's main languages, with machine logic, not expert linguists, leading the way.

Google's approach, called statistical machine translation, differs from past efforts in that it forgoes language experts who program grammatical rules and dictionaries into computers.

Instead, they feed documents humans have already translated into two languages and then rely on computers to discern patterns for future translations.

http://www.nytimes.com/reuters/technolog...ref=slogin
This is yet another enthusiastic newsbite for the general public. Example-based MT - I guess that's what they're talking about - is even worse than regular MT. But people can dream. I do applaud Google's effort, though.

Decent MT is not an impossibility, but it is certainly not around the corner. Even myself, I predicted that MT would become much better by the end of the decade, but I now revise my position. Add a second decade for MT to improve. By 'improving' I mean, translate more or less correctly 3 sentences out of ten, when it's one or two now.

Business will remain basically unchanged for at least a decade, this is my intimate conviction. Good news for all... who are only a decade away from retirement :-)))))

- yves

Aurora Humarán Wrote:
Buen morning, Yves!

Your presentation at the Law School in 2006 was really very interesting.

Could you share with us your explanation on why we still defeat the machine at translating? (the example of the long mathematical calculation which is solved faster [for the time being] by a human being because of the way in which we approach it)

With news like the one below (Reuters - March 29), is our end nearer?

Thank you!

Worried Au Sad


MOUNTAIN VIEW, California (Reuters) - In Google Inc.'s (GOOG.O) vision of the future, people will be able to translate documents instantly into the world's main languages, with machine logic, not expert linguists, leading the way.

Google's approach, called statistical machine translation, differs from past efforts in that it forgoes language experts who program grammatical rules and dictionaries into computers.

Instead, they feed documents humans have already translated into two languages and then rely on computers to discern patterns for future translations.

http://www.nytimes.com/reuters/technolog...ref=slogin

Yves Champollion Wrote:
Decent MT is not an impossibility, but it is certainly not around the corner. Even myself, I predicted that MT would become much better by the end of the decade, but I now revise my position. Add a second decade for MT to improve. By 'improving' I mean, translate more or less correctly 3 sentences out of ten, when it's one or two now.


This if you are considering User's Manuals, general stuff, Yves. What about... Literature? The translation of a manual, of a birth certificate have to do with 'practical' needs. (If I wrote a paragraph of Little Red Riding Hood in the middle of most of the legal stuff I have been translating for 25 years... nobody would realize! These translations are just (just!) instruments. A certain officer reads the name in bold in the first lines and a couple of other details, then double checks the translation formula and ... that's it!

What about books?... They are bought to enjoy the language also, for the pleasure of 'savoring' each and every word. They are the means and the end, if you follow my comparison.

Which machine will be able to get all the nuances of languages? May be one which is so perfect as to be imperfect at the same time (as we are) to be able to perceive and transmit second, third, fourth meanings?

Thanks!

Au

> Which machine will be able to get all the nuances of languages? May be one which is so perfect as to be imperfect at the same time (as we are) to be able to perceive and transmit second, third, fourth meanings?

IMHO, never. Unless machines learn sentiments, joy and sorrow, etc. which I doubt very much, they'll only learn to imitate them, notto feel them. So, long live the literary translator!
yves

Aurora Humarán Wrote:

Yves Champollion Wrote:
Decent MT is not an impossibility, but it is certainly not around the corner. Even myself, I predicted that MT would become much better by the end of the decade, but I now revise my position. Add a second decade for MT to improve. By 'improving' I mean, translate more or less correctly 3 sentences out of ten, when it's one or two now.


This if you are considering User's Manuals, general stuff, Yves. What about... Literature? The translation of a manual, of a birth certificate have to do with 'practical' needs. (If I wrote a paragraph of Little Red Riding Hood in the middle of most of the legal stuff I have been translating for 25 years... nobody would realize! These translations are just (just!) instruments. A certain officer reads the name in bold in the first lines and a couple of other details, then double checks the translation formula and ... that's it!

What about books?... They are bought to enjoy the language also, for the pleasure of 'savoring' each and every word. They are the means and the end, if you follow my comparison.

Which machine will be able to get all the nuances of languages? May be one which is so perfect as to be imperfect at the same time (as we are) to be able to perceive and transmit second, third, fourth meanings?

Thanks!

Au

Reference URL's