| Pages in topic: [1 2 3] > | Will translation software put us out of business in 10 years time? 投稿者: Maurice Koopman
|
|---|
Dear collegues,
How likely is it in your opinion that say in 10 years time translation software has advanced to such an extent that companies can translate their own manuals? Have you ever thought of this possibility? Are in your opinion specific areas of expertise more suitable to automatic translation than others? I am curious what you think.
Thanks and best regards,
Maurice
[Edited at 2013-08-02 07:25 GMT]
| | | |
I think this paragraph from Wikipedia answers your question:
"The history of machine translation generally starts in the 1950s, although work can be found from earlier periods. The Georgetown experiment in 1954 involved fully automatic translation of more than sixty Russian sentences into English. The experiment was a great success and ushered in an era of significant funding for machine translation research in the United States. The authors claimed that within three or five years, ... See more I think this paragraph from Wikipedia answers your question:
"The history of machine translation generally starts in the 1950s, although work can be found from earlier periods. The Georgetown experiment in 1954 involved fully automatic translation of more than sixty Russian sentences into English. The experiment was a great success and ushered in an era of significant funding for machine translation research in the United States. The authors claimed that within three or five years, machine translation would be a solved problem.[1] In the Soviet Union, similar experiments were performed shortly after."
In my opinion we as a profession should really stop discussing this topic because by bringing it up over and over again we're damaging our own credibility. ▲ Collapse
| | | | | I think differently | Aug 1, 2013 |
Andy Lemminger wrote:
In my opinion we as a profession should really stop discussing this topic because by bringing it up over and over again we're damaging our own credibility.
My expectation is that yes, machine translation should put out of business all those extremely cheap translators whose delivery is lower quality than MT.
Every time a prospect offers me anything below US 5¢/word, and tells me that it's all they can afford, I advise them to save that money, and use machine translation instead. The flaws will be different, however the overall quality will be about the same.
Since MT is free, online, and immediate, they can get results quickly. If it serves the purpose, they won't have wasted any money. And if that output is not adequate, there will be more money left in the budget to hire someone competent to do it.
Now and then I am contacted by some desperate prospect who literally wasted money on cheap translation when high quality was necessary. They want to know how much I'd charge to "salvage" whatever I can there and fix it. In all cases so far, there was so little worth salvaging, that it would be better to redo everything from scratch... at my normal translation rate. Though I felt sorry for the money they had wasted on rock-bottom-cheap unskilled human translation, it was not my fault, especially if I had the chance to forewarn them.
The above explains how I know what cheap translation looks like. I've seen quite a few.
Of course, there are several shades between 5¢ and my actual rate, as well as rates above mine. It's a free market, so the client may choose. However my general guess is that if I devised objective parameters to evaluate a translation, human translation at 5¢/word would get about the same score as the current stage of Google Translate, at least for my pair, EN-US/PT-BR. As always, your mileage may vary.
Will MT develop further? I doubt it. Lately it has approached so asymptotically its current level, that only a major breakthrough could really make a difference.
For the time being, I'm spreading the word that low quality translation can be obtained for free from MT sites. If they want quality, there are many thousands of competent professional translators worldwide. Merely "cheap" translation is no longer an intelligent choice.
| | | | | Not 10 years, but maybe 100 | Aug 1, 2013 |
I think José's comments are pretty lucid - translation technology does not look like it's going anywhere spectacular in the near future.
However, on an armchair philosophical level, I personally don't believe that language (or even the brain itself) is some impenetrable mystery that computers will never be able to crack. One day, computers will be good enough for all commercial and technical purposes. Some day after that, they will be able to appreciate and process linguistic subtl... See more I think José's comments are pretty lucid - translation technology does not look like it's going anywhere spectacular in the near future.
However, on an armchair philosophical level, I personally don't believe that language (or even the brain itself) is some impenetrable mystery that computers will never be able to crack. One day, computers will be good enough for all commercial and technical purposes. Some day after that, they will be able to appreciate and process linguistic subtleties that any person would need weeks with a library of dictionaries and some inspired insight to pick up (they will only be able to do this because of the ingenious scientists and engineers who will have done the necessary groundwork, of course!). These two days will be nails in the industry's coffin, and the second one will drive the profession of translator to become fully obsolete.
Still, I can only guess at the timescale needed for advances of such scope. But it will most certainly be counted in decades, if not centuries. In the mean time and the near future, I think that translation will remain a very real and enduring market (maybe even expanding as globalisation is consolidated), and high-level translators will not be threatened. There are some types of document that need to be nearly perfect - the maybe 70-80% accuracy rate that can plausibly be achieved within the next few years by current technology simply isn't good enough, and we haven't seen anything that comes close to offering an aesthetically pleasing style and sentence flow by machines, which currently operate statistically - not by simulating the emotional and intellectual dialogue that creative writers engage in when producing beautiful prose (a totally separate but equally important shortcoming).
Of course, other sectors of the industry will probably shift and groan. As José said, low-quality translators become useless, because they can't beat the machines. Removing the whole lower stratum of the professional community will necessarily have enormous upward economic repercussions on the rest... but it's pretty futile to try to qualify these any further. Everybody knows how bad people in general are at predicting economic trends.
In summary, hold on to your hat for an interesting ride, but don't go looking for another way to earn your keep if you enjoy translation and you know you're good at it. Even if you are a less competent translator (and are self-critical enough to realise it), you will realistically be able to find plenty of work. Unfortunately, this work will be on the back of client ignorance, rather than any intrinsic economic value. If you can live with that, fine (for you). Otherwise, you could always look for something else. ▲ Collapse
| | |
|
|
|
564354352 (X) 
デンマーク
Local time: 21:42
デンマーク語 から 英語
+ ...
| No, but a new sideline in translation may well emerge | Aug 2, 2013 |
Considering the slow development within MT, I am not the least bit worried.
Even if you just take a quick look at the many KudoZ questions asked here, you get the sense that language is a very complex affair that needs a human mind to distinguish between subtle differences.
However, I don't think there can be any doubt that MT is most definitely here to stay and that it will keep getting better. I also believe that in some cases, it IS possible to feed standard texts th... See more Considering the slow development within MT, I am not the least bit worried.
Even if you just take a quick look at the many KudoZ questions asked here, you get the sense that language is a very complex affair that needs a human mind to distinguish between subtle differences.
However, I don't think there can be any doubt that MT is most definitely here to stay and that it will keep getting better. I also believe that in some cases, it IS possible to feed standard texts through excellent MT databases, based on huge amounts of accumulated texts, and come out with decent results. This would apply, for instance, to standard forms into which only names or numbers have been added. But just think of it: would a huge corporation feed its annual accounts through a machine for translation and trust the result to be 100 % accurate. Wouldn't they always want a human to run through it for quality assurance? Even a comma out of place can be pretty disastrous.
So, I think that post-editing of machine-translated texts will become more common. In my view, this is a skill entirely different from actual translation, and it might be worth translation programmes to include this in their curricula (I would presume that some have already started doing this). As I imagine that most translators would rather translate than fiddle around with correcting the translations of others, people who set out for a full-time career as translators may not be at all interested in post-editing MT work. But some people may actually find some joy in nitpicking at substandard translations and turning them into useful, localised texts. The market for such people will only keep increasing. ▲ Collapse
| | | | | When a computer can translate, it can be prime minister | Aug 2, 2013 |
That's not a joke.
All the attempts to get computers to translate thus far have failed because the computers don't understand text. They try to play games with the words, and those games will always fail at some point, because language, pace Wittgenstein, is not a game.
A computer that can understand a text will be able to translate it. And that might happen an awful lot sooner than we imagine. But when it does, it's not just translators that will be out of a job. All t... See more That's not a joke.
All the attempts to get computers to translate thus far have failed because the computers don't understand text. They try to play games with the words, and those games will always fail at some point, because language, pace Wittgenstein, is not a game.
A computer that can understand a text will be able to translate it. And that might happen an awful lot sooner than we imagine. But when it does, it's not just translators that will be out of a job. All the philosophers in the world? Fired. Computer can read Plato better than you, find all your mistakes and write a more interesting thesis. All the town planners? Fired. Computer has read every book on engineering and planning ever written, and can calculate the best solution for your town much better than you can. All the politicians...? You get my point.
At some time in the future, computers will be able to do our job. But when that point comes, all of humanity, with the exception of plumbers, will be out of a job, and we'll just have to join in the hunt for the next thing to do. ▲ Collapse
| | | | | MT will be perfected when computers can think like humans.... | Aug 2, 2013 |
Joshua Pepper wrote:
However, on an armchair philosophical level, I personally don't believe that language (or even the brain itself) is some impenetrable mystery that computers will never be able to crack.
...And when that happens, we'll have bigger fish to fry than worrying about them doing a bit of translation.
| | | | | In the meantime | Aug 2, 2013 |
A friend just finished proofreading an EN-FR translation of a presentation in which slide (as in PPT slide) was variably translated as
the slide in the kids' playground
the act of sliding down the slide in the kids' playground
a type of hairgrip
No kidding!
| | |
|
|
|
| I fully agree with José | Aug 2, 2013 |
José Henrique Lamensdorf wrote:
For the time being, I'm spreading the word that low quality translation can be obtained for free from MT sites. If they want quality, there are many thousands of competent professional translators worldwide. Merely "cheap" translation is no longer an intelligent choice.
I am spreading the word as well. On a regular basis. Last time I did so was an hour ago to finally terminate some extensive email exchange with a tenacious English agency who insisted on the following:
"Although your rate seems a bit too high for us – we are still interested to have you in our Team of linguists.
Please consider if you could agree to 40%-50% discount.
This would allow you to be more competitive and would ease our way to offer you translation/proofreading tasks."
The larger the split between MT and real, top-notch translation, the better for professional translators. It is the overlap between both sectors that is killing our rates.
| | | | | No-- it ill never happen. | Aug 2, 2013 |
Actually there will be more work for translators when people finally wake up from the semi-comatised state of believing that computers can replace human beings in all the branches of life, including their unconditional belief in MT and computer aided translation Yes, there ill be computer aided translation but much less mechanical than it is now.
| | | |
Nicole Schnell wrote:
The larger the split between MT and real, top-notch translation, the better for professional translators. It is the overlap between both sectors that is killing our rates.
| | | | | Don't go by the quality of GT | Aug 2, 2013 |
Gitte Hovedskov, MCIL wrote:
I don't think there can be any doubt that MT is most definitely here to stay and that it will keep getting better. I also believe that in some cases, it IS possible to feed standard texts through excellent MT databases, based on huge amounts of accumulated texts, and come out with decent results.
GT has the volume, but the quality is poor to atrocious, as we all know only too well. It's a free product, one that we all know about, simply because it's free, but it isn't the only one and it's far, far from the best one.
From what I've heard, other MT engines of the very-near future will not only build databases of very specific content but will also be capable of using a whole raft of client-designated terminology - at a price. I suspect these will soon be capable of, for example, updating manuals with 100% consistent client-approved terminology, that only need checking for relatively minor glitches of grammar. I think that area of our business might be affected quite soon. Some technical translators (who refuse to do PE) may find their market shrinking somewhat. However, there are a relatively small number of real specialists, so I'm sure there will be enough one-off and non-MT-friendly jobs for them, and in fact they may actually represent a cheaper option than the equivalent MT+PE . I think the ones who might be challenged by developments are the translators who 'specialise' in a wide range of technical texts such as manuals, without having a background as an engineer, scientist...
I think the areas where MT will probably not make any great headway (at least until Phil's prophecy comes true) are those where it isn't the translation of the words so much as the transcreation of the message behind the words that's needed. Websites, brochures, tourist guides, advertisements, magazine articles... These are areas that often don't have 'difficult' terminology but where the translation of a single sentence by a specialist translator would no doubt bear very little resemblance to the source after literal back-translation. It's also an area where repetitions are often avoided even if they're grammatically and terminologically correct. I can't see an MT engine ever getting the hang of the whole range of bilingual idioms, let alone word play.
I think that post-editing of machine-translated texts will become more common.
I agree. But I think most of the work will be post-editing of high-quality tailored MT, with reliable terminology usage, so that subject becomes less important. And it will be well paid and for good linguists only. If you've spent a mint on a top MT engine, why would you then give its output to someone who has no idea of the finer points of translation and writing, and will probably 'correct' it to show?
OTOH, I see GT output post-editing remaining a sweat-shop activity that we need to distance ourselves from.
(BTW, Phil, what makes you think we're going to have free will? I think They will determine what we do, if in fact they want to keep us around. I suppose we might make fascinating pets. )
| | |
|
|
|
| Butlerian Jihad | Aug 2, 2013 |
I believe that MT will not put good translators out of business, save for good translators with bad marketing (or lacking business skills, which is no shame but a hindrance... in business).
MT will, however, put bad translators out of business. I have already seen human translation worse than machine translation. From a client's point of view, there's really little reason to pay such a translator over such a free-to-use machine.
MT might possibly put out of business the... See more I believe that MT will not put good translators out of business, save for good translators with bad marketing (or lacking business skills, which is no shame but a hindrance... in business).
MT will, however, put bad translators out of business. I have already seen human translation worse than machine translation. From a client's point of view, there's really little reason to pay such a translator over such a free-to-use machine.
MT might possibly put out of business the translators who use it, rely on it, and then claim they are returning human translation.
MT might put a dent in the business of good translators who don't have the most open-handed or translation-aware clients, but the effect should not be overwhelming.
On the other hand, MT complicates the life of translators and will continue to do so, and more and more in fact, by frequently being employed in the creation of those source texts which are translations from yet another language.
In the above connection, if PMs in agencies based in English-speaking countries – but many PMs in many countries have at least the kind of knowledge of English to recognise senseless gibberish when they see it (an FCE holder should be able to recognise the absence of grammar) – look at texts and refuse to accept machine translations as sources to be translated into different languages, then the problem shouldn't get out of hand... Except that bloody murder is performed on English grammar and syntax every day throughout the world (with no small fault of education methods that value clumsy but working communication over correctness with the aid of growingly lax description-oriented linguistics), but anyway.
Other languages, by contrast, are likely to have it worse since the PMs will more often not speak them, while those languages may also be more resistant on their own to machine translated source encroachment.
What translation software can possibly put us largely out of business or otherwise seriously complicate our lives is CAT software.
In the past years, we used to charge per page (or even per word), and the existence of repetitions was understood to be a fact of life that perhaps sometimes justified a discount when it was really, really visible in a large amount. There were certainly no CAT grids to pick up on every single '75% similarity', effectively lowering our rates 5 to 15% (after we were made to pay for the software).
Anybody translating from English into a language that has inflection knows a thing or two about even 100% matches.
With that penny-pinching bean-counting maths, CATs cheapen our profession. Translation should be about better or worse understandings of the source, better or worse expression in the target language, better or worse equivalence, not about the tiniest bit of maths doable on the rates while often not being worth the financial value of the time spent thinking about it.
Also, translators are putting themselves out of business or making their business worse due to accepting the trends or passively or even actively furthering them. Accepting re-writes without a good reason, accepting corrections from less than competent clients or client staff, accepting to perform secretarial and concierge duties, the list is long. The worse is probably the willingness to stand correct by everybody and his dog*.
* 'My dog grew up in your country and he wagged his tail the wrong way when being read your translation. Are you sure everything is correct there? Please change some words.' ▲ Collapse
| | | | | It will not happen, Sheila -- I can assure you. | Aug 2, 2013 |
If you look at the MT within the Germanic languages, you may still have some hopes. If you look at the Slavic languages, not to mention the Baltic languages, you arms just drop in dismay. I don't think it will ever be possible that computers will be able to handle complex translation -- perhaps when they will also learn how to cook, and write interesting books.
I also agree with Lukasz about the repetition part -- it is a total nonsense. I agree that if it is a part of the text -... See more If you look at the MT within the Germanic languages, you may still have some hopes. If you look at the Slavic languages, not to mention the Baltic languages, you arms just drop in dismay. I don't think it will ever be possible that computers will be able to handle complex translation -- perhaps when they will also learn how to cook, and write interesting books.
I also agree with Lukasz about the repetition part -- it is a total nonsense. I agree that if it is a part of the text -- boilerplate-like-- with exactly the same wording, it can be counted as the text repeated. Many linguistically unaware companies, however, would like to count the definite articles as repetitions, or the word all, or any other word almost. This can serve as a proof of their total ignorance of any language related problems.
And then the fuzzy matches -- this is even more ridiculous. Why would a good translator need any fuzzy matches -- to confuse him or her even more? Fuzzy is fuzzy -- wrong.
[Edited at 2013-08-02 11:24 GMT] ▲ Collapse
| | | |
[Edited at 2013-08-02 11:28 GMT]
| | | | | Pages in topic: [1 2 3] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Will translation software put us out of business in 10 years time? | TM-Town | Manage your TMs and Terms ... and boost your translation business
Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.
More info » |
| | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
