| Pages in topic: < [1 2] | Poll: Will the Google Translation Center affect your business? Thread poster: ProZ.com Staff
|
|---|
Williamson 
United Kingdom
Local time: 08:17
Flemish to English
+ ...
| In the long run: Yes | Aug 16, 2008 |
Not on the short term, but in the long run. Whereas now some call giving discounts for Trados "Professional", we will be forced to give discounts for Google Translation Centre.
Don't forget that translation is a cost-factor, which companies will try to eliminate.
If a text is translated to a satisfactory level (understandable enough) to be rewritten by a management assistant (secretary) or an engineer or both, who are already on the pay-roll, why pay a freelance translator?... See more Not on the short term, but in the long run. Whereas now some call giving discounts for Trados "Professional", we will be forced to give discounts for Google Translation Centre.
Don't forget that translation is a cost-factor, which companies will try to eliminate.
If a text is translated to a satisfactory level (understandable enough) to be rewritten by a management assistant (secretary) or an engineer or both, who are already on the pay-roll, why pay a freelance translator?
What will the machine be able to churn out in 5 years from now?
What about the role of the translator:
Definition of the task: a person who rewrites and adapts texts produced by M.T. Don't forget your discounts. After all, we already have a basic version and an idea what the text is about.
[Edited at 2008-08-16 13:04] ▲ Collapse
| | | | Heidi C
Local time: 03:17
English to Spanish
+ ...
| Translator should define what his or her role is... | Aug 16, 2008 |
Williamson wrote:
What about the role of the translator:
Definition of the task: a person who rewrites and adapts texts produced by M.T. Don't forget your discounts. After all, we already have a basic version and an idea what the text is about.
[Edited at 2008-08-16 13:04]
This will be the role of the translator who accepts this definition. (Just like translators who accept 0.002 fees plus discounts)
Translators who do not remain passive and accept the roles assigned to them, will follow this fate.
Quality translators who deliver a quality product and do not accept this role will still have work. Because the need for quality translated texts will always exist, and a good editing or adaptation job could cost the same (or more) than getting a translation right the first time!
Another example of the self-fulfilling prophesy...
| | | | | Agree - we must know what role we wish to play | Aug 16, 2008 |
Heidi C wrote:
Williamson wrote:
What about the role of the translator:
Definition of the task: a person who rewrites and adapts texts produced by M.T. Don't forget your discounts. After all, we already have a basic version and an idea what the text is about.
[Edited at 2008-08-16 13:04]
This will be the role of the translator who accepts this definition. (Just like translators who accept 0.002 fees plus discounts)
Translators who do not remain passive and accept the roles assigned to them, will follow this fate.
Quality translators who deliver a quality product and do not accept this role will still have work. Because the need for quality translated texts will always exist, and a good editing or adaptation job could cost the same (or more) than getting a translation right the first time!
Another example of the self-fulfilling prophesy...
I agree that a 'translator' has to define her / his role. From what I have read, I don't think most of the colleagues who are members of Proz.Com would accept a demeaning role. As some colleagues have already pointed out, many of us are not at this time performing roles which can, in the near future, be acceptably carried out by automated means, nor are they going to be accepting the task of merely 'fixing' MTs to make them more acceptable.
In addition to specialised language skills, many colleagues bring to the table skills born of years of experience in other disciplines (law / medicine / exact sciences / biology / finance and the like), besides being HUMAN beings of integrity and discretion. Hard to see how they can in fact be replaced so easily by automation, however sophisticated, unless they allow themselves to!
| | | | Nigel Greenwood (X)
Spain
Local time: 09:17
Spanish to English
+ ...
| I totally agree | Aug 17, 2008 |
Venkatesh Sundaram wrote:
Heidi C wrote:
Williamson wrote:
What about the role of the translator:
Definition of the task: a person who rewrites and adapts texts produced by M.T. Don't forget your discounts. After all, we already have a basic version and an idea what the text is about.
[Edited at 2008-08-16 13:04]
This will be the role of the translator who accepts this definition. (Just like translators who accept 0.002 fees plus discounts)
Translators who do not remain passive and accept the roles assigned to them, will follow this fate.
Quality translators who deliver a quality product and do not accept this role will still have work. Because the need for quality translated texts will always exist, and a good editing or adaptation job could cost the same (or more) than getting a translation right the first time!
Another example of the self-fulfilling prophesy...
I agree that a 'translator' has to define her / his role. From what I have read, I don't think most of the colleagues who are members of Proz.Com would accept a demeaning role. As some colleagues have already pointed out, many of us are not at this time performing roles which can, in the near future, be acceptably carried out by automated means, nor are they going to be accepting the task of merely 'fixing' MTs to make them more acceptable.
In addition to specialised language skills, many colleagues bring to the table skills born of years of experience in other disciplines (law / medicine / exact sciences / biology / finance and the like), besides being HUMAN beings of integrity and discretion. Hard to see how they can in fact be replaced so easily by automation, however sophisticated, unless they allow themselves to!
Don't forget the role we have in DTP, formatting tables, deciphering badly written original texts, etc.
| | |
|
|
|
XX789 (X)
Netherlands
Local time: 09:17
English to Dutch
+ ...
| It's not going to happen | Aug 17, 2008 |
Example-based machine translation is never, ever going to take over this world. The Oxford English dictionary features 171,476 words. Let's round that off to 170,000.
Theoretically, there are therefore 170,000 possibilities for a one-word sentence.
That means there are 170,000 x 170,000 = 28,900,000,000 possibilities for a two-word sentence.
That is 170,000 x 170,000 x 170,000 = 4,913E+15 possibilities for a three-word sentence.
This relative... See more Example-based machine translation is never, ever going to take over this world. The Oxford English dictionary features 171,476 words. Let's round that off to 170,000.
Theoretically, there are therefore 170,000 possibilities for a one-word sentence.
That means there are 170,000 x 170,000 = 28,900,000,000 possibilities for a two-word sentence.
That is 170,000 x 170,000 x 170,000 = 4,913E+15 possibilities for a three-word sentence.
This relatively short sentence contains six words. That means 2,41376E+31 possibilities. That's about 24,137,600,000,000,000,000,000,000,000,000 possibilities. Average sentences are much longer than that.
But the situation is worse. We're not only talking language combinations like English to Spanish. No, there's also English to Russian. Russian to English. Chinese to Japanese. Japanese to Dutch. Dutch to French. French to German. French to Russian. And on, and on, and on...
Example-based machine translation will book some small successes in phrases that are used relatively often, like "Welcome to our website", but is absolutely useless for phrases that are used less often, like "Welcome to my forum thread".
And we can forget about translations that involve creativity. In my field of game translations, texts are often infested with puns.
Why couldn't the little pony talk? Because it was a little hoarse.
Waarom plant een pony zich zo snel voort? Omdat-ie graag paard.
Coming up with translations like that involves a lot of creativity. Creativity that machines will never have - it's impossible to program it.
As a translator there's no need to worry. As Proz.com there is. It can't be a coincidence that only a few days after Google's announcement, Proz.com came up with the free Try Out Our Site offer for clients.
Some clients won't care whether their texts are used for a public memory. Some clients will. And some clients will be too lazy to even check. But for you as a translator, work is work. A 1000 euro job coming in via Proz.com is worth as much as a 1000 euro job coming in via Google. I don't care, as long as I get paid
But no, I'm not worried that the translations I deliver will eventually put me out of business. 24,137,600,000,000,000,000,000,000,000,000 possibilities for 6-word sentences. Assuming that all 16 million inhabitants of Holland helped building up Google's translation memory, each inhabitant still needs to translate 1,508,600,000,000,000,000,000,000 words. If each of them does 2,000 words a day, that will take them 754,299,000,000,000,000,000 days = 2,066,570,000,000,000,000 years.
And that is just for 6-word sentences. For one of the several thousand language combinations.
We're not going to live to see that day, people.
[Edited at 2008-08-17 15:57] ▲ Collapse
| | | | | Think sub-segments and terminology - not full sentences. | Aug 17, 2008 |
You make some excellent points Loek, but there is a case for factoring in repetitive chunks and terminology. I believe Google's power lies in sub-segment analysis, so that, for example, your own set phrases:
"never, ever"
"take over [the] world"
"round that off"
"two-word"
"And on, and on"
"live to see [that] day"
"put [me] out of business"
and specialised terminology:
"Example-based machine translation"
"language c... See more You make some excellent points Loek, but there is a case for factoring in repetitive chunks and terminology. I believe Google's power lies in sub-segment analysis, so that, for example, your own set phrases:
"never, ever"
"take over [the] world"
"round that off"
"two-word"
"And on, and on"
"live to see [that] day"
"put [me] out of business"
and specialised terminology:
"Example-based machine translation"
"language combinations"
...could easily become part of a Machine Translation engine.
I can see Google taking over first 1%, then 5%, then more and more of the research effort, and coming up with a range of brilliant suggestions (such as your own Pony joke) eventually. There aren't as many clichés, set phrases and jokes as the total possible numbers of phrases you suggest...! In fact these 'chunks' are quite common; what CAT tools are pretty poor at, at the moment, is allowing us to extract these from the randomly long sentences that pad them out and make each author 'unique' (...pad out? - cliché!)
I am already working with some MT software and even my manual database is worth compiling in terms of return on time investment, never mind a massive automated system like Google's. I think it could become a huge phenomenon and if it increases the output of current translators, new translators might find it harder to get work (unless of course more work becomes available). Tragic if you like looking words up in dozens of dictionaries and on-line sources, rooting through dodgy old bookmarks, or being told 'access denied' or even finding nothing! But good if you prefer making things plain and readable for people with a brilliant, organic and growing knowledge base at your disposal.
Simon
www.italiantranslator.org.uk ▲ Collapse
| | | | Wil Hardman (X)
United Kingdom
Local time: 08:17
Spanish to English
+ ...
| The human element will never go away. Quality will always cost money. | Aug 17, 2008 |
The only way I can see it affecting our buisiness is that perhaps more people may turn to it, as it will produce better results than machine translation. However they will still be misguided to do so and will never get a top quality translation.
It may also turn out to be a useful tool for translators or people who know what they are doing.
Many people resent paying for translation, they don't appreciate the effort and skill that producing a good translation requires. T... See more The only way I can see it affecting our buisiness is that perhaps more people may turn to it, as it will produce better results than machine translation. However they will still be misguided to do so and will never get a top quality translation.
It may also turn out to be a useful tool for translators or people who know what they are doing.
Many people resent paying for translation, they don't appreciate the effort and skill that producing a good translation requires. The Goolgle Center just seems like another way to get around paying. However, without the input of a human expert, good results will never be achieved, for the reasons that Loek has exposed (too many possibilities).
The concept of a global tranlsation memory, to me seems unworkable- it would spiral out of control. When translation agencies provide me with a TM, even a small one, there are always inconsistencies and errors. A huge TM where all translations are inputted would be a minefield of errors and completely unreliable.
The articles I've read talk about a review process, where translations in a global tranlsation memory are reviewed by volunteers, however the best translators are never going to participate in this free of charge. Kudoz works because it raises a translator's profile, and this is also suggested as motivation for translators to review translations in the Google translation Center! An experienced translator would have to be insane to waste their time reviewing chunks of text to raise their profile when they can do it on kudoz by just translating a term. Not to mention that they would be contributing to the Center's ultimate goal of putting translators out of work.
Kudoz works because it is restricted to translating single terms, anything beyond that the possibilities become too vast and it would no longer be a useful reference. However, not even Kudoz is 100% reliable as a reference and a knowledge of translation and the languages in question is needed to pick out which terms are applicable.
You will never eliminate the human element from translation. Proz.com works because it places the emphasis on this human element and on trying to help translators earn a living, whereas the Google Translation Center seems to be trying to bypass the human element and reducing the input needed from translators. ▲ Collapse
| | | | XX789 (X)
Netherlands
Local time: 09:17
English to Dutch
+ ...
| There definitely is | Aug 17, 2008 |
but there is a case for factoring in repetitive chunks and terminology.
There definitely is. And these will deflate the above numbers a bit. The number will be deflated even more because in practise, not all words will be combined with all words. The chances that a two-word term like "follicle skates" will ever be used is close to nil for example (except in this example that is)
But even then I think the numbers are high enough to make sure that we won't live to see the day. Especially because in many languages, segments depend on other segments. Think cases for example (der die das, anyone)? Or think politeness levels, of which Japanese has about 6 or so. Or think context. Like the Japanese "iku", which can mean both come and go, depending entirely on the context Japanese even has words which can mean yes and no at the same time.
But let's not go there. The problem is, that besides the enormous mathematical obstacles described above, language is more than just words. There's a whole load of invisible information between the lines that machines will never ever be able to grasp. Machines are just really bad at stuff like this.
| | |
|
|
|
| What sort of material could possibly get translated this way? | Aug 17, 2008 |
As I'd feel obliged to point out to a customer approaching me via the project site that his text would get included in a publicly available database, which customer group is likely to forego confidentiality, in fact?
To me, the most probable answer is - only those who intend to put the material into the public domain anyway, e.g. website owners, leaflet publishers etc. Having signed several (dozen) NDAs during your career, can you see any other group of potential users on the customer side... See more As I'd feel obliged to point out to a customer approaching me via the project site that his text would get included in a publicly available database, which customer group is likely to forego confidentiality, in fact?
To me, the most probable answer is - only those who intend to put the material into the public domain anyway, e.g. website owners, leaflet publishers etc. Having signed several (dozen) NDAs during your career, can you see any other group of potential users on the customer side?
And even website owners specify copyright on home pages...
[Edited at 2008-08-17 18:52] ▲ Collapse
| | | | Rob Edwards
Local time: 08:17
Member (2008)
German to English
| Thanks Loek. | Aug 17, 2008 |
You have done a great job of expressing what i have been trying to tell people about MT for years.
Loek van Kooten wrote:
Example-based machine translation is never, ever going to take over this world. The Oxford English dictionary features 171,476 words. Let's round that off to 170,000.
Theoretically, there are therefore 170,000 possibilities for a one-word sentence.
That means there are 170,000 x 170,000 = 28,900,000,000 possibilities for a two-word sentence.
That is 170,000 x 170,000 x 170,000 = 4,913E+15 possibilities for a three-word sentence.
This relatively short sentence contains six words. That means 2,41376E+31 possibilities. That's about 24,137,600,000,000,000,000,000,000,000,000 possibilities. Average sentences are much longer than that.
But the situation is worse. We're not only talking language combinations like English to Spanish. No, there's also English to Russian. Russian to English. Chinese to Japanese. Japanese to Dutch. Dutch to French. French to German. French to Russian. And on, and on, and on...
Example-based machine translation will book some small successes in phrases that are used relatively often, like "Welcome to our website", but is absolutely useless for phrases that are used less often, like "Welcome to my forum thread".
And we can forget about translations that involve creativity. In my field of game translations, texts are often infested with puns.
Why couldn't the little pony talk? Because it was a little hoarse.
Waarom plant een pony zich zo snel voort? Omdat-ie graag paard.
Coming up with translations like that involves a lot of creativity. Creativity that machines will never have - it's impossible to program it.
As a translator there's no need to worry. As Proz.com there is. It can't be a coincidence that only a few days after Google's announcement, Proz.com came up with the free Try Out Our Site offer for clients.
Some clients won't care whether their texts are used for a public memory. Some clients will. And some clients will be too lazy to even check. But for you as a translator, work is work. A 1000 euro job coming in via Proz.com is worth as much as a 1000 euro job coming in via Google. I don't care, as long as I get paid
But no, I'm not worried that the translations I deliver will eventually put me out of business. 24,137,600,000,000,000,000,000,000,000,000 possibilities for 6-word sentences. Assuming that all 16 million inhabitants of Holland helped building up Google's translation memory, each inhabitant still needs to translate 1,508,600,000,000,000,000,000,000 words. If each of them does 2,000 words a day, that will take them 754,299,000,000,000,000,000 days = 2,066,570,000,000,000,000 years.
And that is just for 6-word sentences. For one of the several thousand language combinations.
We're not going to live to see that day, people. [Edited at 2008-08-17 15:57]
| | | | hakim_sm
Sweden
Local time: 09:17
Swedish to Arabic
+ ...
| human will never substitute the machin. | Aug 18, 2008 |
is google the first wich comes with this service ? i dont think so
| | |
|
|
|
| Other threats too - but response of colleagues is reassuring | Aug 18, 2008 |
Williamson wrote:
Not on the short term, but in the long run. Whereas now some call giving discounts for Trados "Professional", we will be forced to give discounts for Google Translation Centre.
Don't forget that translation is a cost-factor, which companies will try to eliminate.
If a text is translated to a satisfactory level (understandable enough) to be rewritten by a management assistant (secretary) or an engineer or both, who are already on the pay-roll, why pay a freelance translator?
What will the machine be able to churn out in 5 years from now?
What about the role of the translator:
Definition of the task: a person who rewrites and adapts texts produced by M.T. Don't forget your discounts. After all, we already have a basic version and an idea what the text is about.
[Edited at 2008-08-16 13:04]
Going a little beyond the discussion on Google in which so many colleagues have participated, I think we all do realise that there are perhaps other treats which are perhaps more severe than Google at this time. Only this morning, I've received a mail asking me whether I would be interested in working for a 'large project' at about HALF THE RATE which I have quoted on my Proz.Com page (the potential outsourcer used Proz.Com to get to know about me in the first place). If they've contacted a large number of colleagues, and sufficient numbers of them agree to the low rates they've offered, then its perhaps only a matter of time before the profession becomes unviable for the people like me! I'm sure this is a threat several other colleagues deal with every day too - but that perhaps has already been the topic of (numerous) other discussion(s)! Nevertheless, its' reassuring to note that so many colleagues are so sure of their credentials, experience and what they are offering to their clients that they fear no machine, Google - run or otherwise!
| | | | | Positively for sure | Aug 25, 2008 |
Google Translation Center positively affects my bussines for sure.No one can deny the massive and great Efforts exerted by Google team in assisting translators nowadays.
| | | | | Will it affect my business? | Sep 9, 2008 |
As a disruptive technology under Convergence conditions yes...
Creative Commons License
http://en.wikipedia.org/wiki/Disruptive_technology
http://en.wikipedia.org/wiki/Technological_convergence
The machine translation function is... See more As a disruptive technology under Convergence conditions yes...
Creative Commons License
http://en.wikipedia.org/wiki/Disruptive_technology
http://en.wikipedia.org/wiki/Technological_convergence
The machine translation function is only one function among many, like translation memory,
collaborative structure and collaborative processes, and Creative Commons License, like
the following links (meaning such not in violation to ProZ site rule 3, moderators claiming
otherwise are false):
Creative Commons License like Wikipedia
http://collaborative-translation.ning.com/group/googletranslationcenter
Web 1.0, Gutenberg-like, and hierarchy-like processes are becoming obsolete, even if you
deploy a collaborative wiki now already. There is no need to just wait for Google Translation
Center. Web 2.0 and Wikis are now.
In conjunction with the above Creative Commons License:
http://collaborative-translation.ning.com/forum/topic/show?id=2237585:Topic:341
Even the Translation Grid (the next big Internet thing, derived from Grid Computing in order
to collaboratively investigate Black Hole data in CERN) is being tested...
University Link
http://langrid.org/association/indexe.html
http://langrid.nict.go.jp/en/index.html
http://en.wikipedia.org/wiki/Grid_computing
All links are public domain information.
[Edited at 2008-09-09 13:30] ▲ Collapse
| | | | | Pages in topic: < [1 2] | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: Will the Google Translation Center affect your business? | Trados Business Manager Lite | Create customer quotes and invoices from within Trados Studio
Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.
More info » |
| | CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
