| Páginas sobre el tema: < [1 2 3 4] > | Poll: Will machine translation ever replace human translation entirely? Autor de la hebra: ProZ.com Staff
|
|---|
Sigurjón Kristjánsson (X)
Reino Unido
Local time: 10:50
inglés al islandés
+ ...
| Machine translations | Oct 9, 2007 |
I personally "SERIOUSLY" doubt, that machine translation, will ever replace human translation. The reason being, that machines, although capable of "artificial" intelligence, even with the use of neural networks, "synthesizers" as well as heuristic programming, emulators and other parafernalia, lack both understanding and insight as well as divine inspiration.
As languages evolve, and new words appear based on new technology, computers will remain "physically and spiritually dead",... See more I personally "SERIOUSLY" doubt, that machine translation, will ever replace human translation. The reason being, that machines, although capable of "artificial" intelligence, even with the use of neural networks, "synthesizers" as well as heuristic programming, emulators and other parafernalia, lack both understanding and insight as well as divine inspiration.
As languages evolve, and new words appear based on new technology, computers will remain "physically and spiritually dead", and will only evolve as far as humans change them.
They can be helpful in medical diagnosis, as offering answers, by including items, that may be rare, based on the probability of the inputed data, but they would never carry out research on mutations of diseases or discover new diseases - Hence when a language evolves and new items need words in a local language, they will not "come up" or "conjour up" a translation.
For example: When sputnik was launched, the whole world spoke of "artificial sattelite" - This through the years got shortened to "sattelite".
In Icelandic, the word "sattelite" is translated as "fylgihnöttur" or "following/accompanying globe/orb"
(e.g. planet or moon as the moon is earths sattelite and the earth is the sun´s sattelite). However, when the word "artificial" ("gervi" in Icelandic) is added to the sentence structure or idiom, we end up with a new translation i.e. "gervi tungl" or "artificial moon". - Regretably, alot of countries, do not persue to maintain the uniqueness of their heritage, by making an effort to translate words into their local language.
I spoke to a man from Finland/Suomi, about 20 years ago, and he told me, that they used the word "Sattelit". I asked him if translating it as "artificial moon" would work in Finnish, without distorting the meaning. He thought about it for 5 seconds, and said: "Yes, it would work".
Another problem, that computers have, is "context specific translations"; E.g. if you have the Icelandic word: "Kenning",
the most likely translation, would by a computer, be the most common used word being the number one selection, i.e. "theory", the second choice would probably be "doctrine", and the third choice would be "metaphor". Unless we fully understand the nature of what we are translating, the result could be catastrophic.
Another problem with translation programs, is grammar. Computers have great difficulty with indefinate and definate articles, and some of the translations I have seen are a disgrace. This also applies to languages like Icelandic and German, where they have four cases, Russian has six and so forth.
We even notice, when people are too sloppy in their own language, and make errors like saying: "She is better than me". - Anyone "knowing" English, will "know", that it is grammatically incorrect. A quick self checking method, is to add a verb, like "to be", at the end of the sentence, giving the following result: "She is better than me am". This should have bells and whistles going off in your head along with flashing red lights and a man waving a red flag. - You therefore immediately know, that the proper way of saying it, is:"She is better than I am", telling you that if you wish to drop the verb, that the sentence should be: "She is better than I".
Moreover, if you look at "modern Greek", you have a "new problem". The changes made over 2 decades ago, resulted in all accents (i.e. ´) etc. to be removed. For a novice to the language, this means, that there is no longer any difference between the feminine definate article and the word "and".
This is a home made problem, which was introduced by the president of the time, who wanted to hide peoples illiteracy by simplifying things, and ended up making things more complicated.
Computers are stupid by nature. I know, for I have learned computer programming, since they day we used punch cards and line numbers. - Any decent computer programmer, will tell you, that computers NEVER do what you want them to, ONLY what you TELL them to do. Hence, if the translation program sucks. It´s the human´s fault.
They will never replace humans. They can speed things up, e.g. if you´ve translated something once, and it appears umpteen times in your work, e.g. "Verily, verily I say unto you..." then it will prompt you to use the same translation again; However you have to decide, whether the grammar needs altering based on the context, as English grammar is very rudimentary, compared to the complex grammar of many other languages.
In closing, computers are machines, they do not think (therefore they are not), and the software is only as good as the programmer. Feed a heuristic program with a lot of bull, and that is what it will come out with.
I do not use translation software, and therefore miss out on alot of business, but the work I do is human, and it won´t be replace by a machine. ▲ Collapse
| | | | | hopefully never but that seems unlikely | Oct 9, 2007 |
I only just finished an evaluation project for a machine translation system.
On the one hand, it's nowhere near the quality of work produced even by terrible human translators.
On the other hand, I am very impressed with how coherent most of the stuff it produces is. Don't get me wrong, it's still mostly incoherent but the bits that aren't seem very fluent indeed.
It has me worried...no doubt about it.
| | | | Williamson 
Reino Unido
Local time: 10:50
flamenco al inglés
+ ...
| Hello, mrs Android: What is your rate? | Oct 9, 2007 |
100 years ago, Jules Verne wrote a couple of fiction books.
Today, a man on the moon is history and the future (moonbase). Some concepts in Star-Trek, such as mobile phones have become part of our daily lives. Teleportation of minute atoms has been achieved. Research on alternate methods of propulsion is no longer science-fiction.
First there was fiction, then science followed by application of science in our daily lives.
In 1980 a pc had a memory of 16 Kb and a hard-disk of... See more 100 years ago, Jules Verne wrote a couple of fiction books.
Today, a man on the moon is history and the future (moonbase). Some concepts in Star-Trek, such as mobile phones have become part of our daily lives. Teleportation of minute atoms has been achieved. Research on alternate methods of propulsion is no longer science-fiction.
First there was fiction, then science followed by application of science in our daily lives.
In 1980 a pc had a memory of 16 Kb and a hard-disk of ? megabyte. In 2007 the average is 4 gig and half a terrabyte. No Trados in 1980, no voice-recognition (what was the Star-Trek movie in which Mr.Spock spoke to a MacIntosh pc?), no MsWindows or Office, no Linux either... All the things we take for granted today simply did not exist. Rember the days when the price of a mobile was about 600 euros. That was only 12 years ago...
Combine M.T., voice-recognition, interpreting skills and put them into a cognitive (female) android, programmed with a translator
http://news.bbc.co.uk/2/hi/science/nature/4714135.stm
If it could interpret all languages at the E.U., it diminish the budget for T&I at international institutions considerably.
A question to Russian translators: how good is ProMt machine translation? It does not translate exactly what is meant, but it transfers meaning. Put it in Promt and rewrite the text. Will that be our future?
[Edited at 2007-10-09 09:52] ▲ Collapse
| | | | | Never, BUT cyber-translators may replace human translators | Oct 9, 2007 |
The question is who will be responsible for wrong translation ? Future Clients are supposed to be humans ?
[Edited at 2007-10-09 10:57]
| | |
|
|
|
| Oh go on Sophie Dzhygir! tell us who the company is! | Oct 9, 2007 |
Oh go on Sophie Dzhygir! tell us who the company is, that does a good job of MT!
Also, anyone familiar with terminology extraction? this could be a way of improving MT. Again there are companies like TEMIS out there doing this - using statistics.
I have tried SDL Phrasefinder (for my desktop) with very little success - the results are promising but the software is dodgy... it might just be my relatively weak pc processor at fault, who knows... anyone have any experience... See more Oh go on Sophie Dzhygir! tell us who the company is, that does a good job of MT!
Also, anyone familiar with terminology extraction? this could be a way of improving MT. Again there are companies like TEMIS out there doing this - using statistics.
I have tried SDL Phrasefinder (for my desktop) with very little success - the results are promising but the software is dodgy... it might just be my relatively weak pc processor at fault, who knows... anyone have any experience of this?
I think a company like Google is going to take MT one step further thanks to its supercomputers and then charge us for the privilege of viewing the material already translated - complete with statistics and references and other quality controls. Then they might buy back our edited versions (they will know a good quality translation from a bad one!) and so keep on improving and monopolising. Then there will be fewer translators (or just faster ones) doing the same amount of work. ▲ Collapse
| | | | | MACHINE TRANSLATOR PROMT IS GOOD | Oct 9, 2007 |
Williamson wrote:
A question to Russian translators: how good is ProMt machine translation? It does not translate exactly what is meant, but it transfers meaning. Put it in Promt and rewrite the text. Will that be our future?
[Edited at 2007-10-09 09:52]
As for Promt - it depends on the owner. I actively use Promt, regularly update my dictionaries, and it saves my physical and mental work. I has never been so dull as to completely rely on its translations. Sometimes it can generate good reading phrases but with the opposite meaning !!! I would like to say that Promt is very useful tool, it shall be further enhanced and sophisticated; I am sure it never replace my work, though our customers sometimes think other way.
The therapy should be as dictated by today's life and practice - leave customers run for a week on the machine translators only (may be in July, or during the Translators day). KIDDING
| | | | Balasubramaniam L.
India
Local time: 16:20
Miembro 2006
inglés al hindi
+ ...
LOCALIZADOR DEL SITIO
I don't think it can ever happen.
Language is not a static thing and is constantly evolving, by the time machines pin down one snapshot of a language, the language would have moved ahead and developed new usages, words, nuances, etc.
Also languages are too complex for machines to replicate, however much memory and processing power they may have. What is required is intelligence and context sensitivity which machines cannot reproduce.
In reply to mediamatrix... See more I don't think it can ever happen.
Language is not a static thing and is constantly evolving, by the time machines pin down one snapshot of a language, the language would have moved ahead and developed new usages, words, nuances, etc.
Also languages are too complex for machines to replicate, however much memory and processing power they may have. What is required is intelligence and context sensitivity which machines cannot reproduce.
In reply to mediamatrix's early post about disappearing languages as a result of English dominance, I would like to point out that languages exist at several levels, and the written form is only one of them. English may dent into the written version of many small languages, but the spoken form which is the real thing will continue as long as a viable number of speakers exist.
Also, in the unlikely event of MT becoming a reality, it is not likely to have a uniform effect on translators of all languages. These will most likely be developed only in the dominant languages (English included, by poetic justice?) and it will be a long long time before they come around to developing one in the less dominant ones. It won't be economical.
So will the meek translators inherit the earth? ▲ Collapse
| | | | Lisa Roberts
Francia
Local time: 11:50
español al inglés
+ ...
Unlikely to ever render the translator completely useless, but I think it is inevitable as time goes on that MT will be capable of performing an increasingly large proportion of the work we do. Scary...
| | |
|
|
|
CMJ_Trans (X)
Local time: 11:50
francés al inglés
+ ...
Many moons ago, I was a member of a working party on machine translation. We had a great time: we got to travel to different countries for meetings at no cost to ourselves; we got to try out the software around at the time and had some great laughs at the rubbish the systems delivered.
Since then, however, things have changed. Several years ago now, an acquaintance was made redundant from a staff job translating for a big French firm because his bosses had decided to opt for machine... See more Many moons ago, I was a member of a working party on machine translation. We had a great time: we got to travel to different countries for meetings at no cost to ourselves; we got to try out the software around at the time and had some great laughs at the rubbish the systems delivered.
Since then, however, things have changed. Several years ago now, an acquaintance was made redundant from a staff job translating for a big French firm because his bosses had decided to opt for machine-translation. They were "kind" to him: they gave him his computer and offered him the job of revising the machine's work. He made a living from that thereafter.
Obviously this big firm had the whole thing sussed out: they produced mainly specifications and invitations to tender. They instructed their engineers to produce texts that were always identical, except for the details that were bound to change. They then bunged the texts through the system and then sent them off to the human being to sort out the mess. And since most engineers don't like writing, they were happy to stick to the rules and simplify the task. Probably still doing it today.
All of which goes to show that for technical and repetitive material, machine translation will see us all out. Not sure I care, given how boring that sort of stuff is but it will mean less bread to go around.
Where machine translation has a long way to go is in the realms of the philosophical, the poetical and the skills with words and turns of phrase. But have no fear, if not perfection, there will be a system that can produce a suitably good workmanlike job.
Frankly, on this site and elsewhere, I have come across some human translators that a good machine should easily be able to give a run for their money...... No names, no packdrill.. But maybe it will root out the ones who should never have been in the job in the first place.
Just a thought... ▲ Collapse
| | | |
I don't believe it ever will as the cost implications for the development and maintenance of such a system are far too high.
Please also remember that computers are digital in nature and can only cope with simple logical ideas.
Also why spend billions to train a computer to do something that a human can do instinctively?
| | | | | I know, I know! | Oct 9, 2007 |
Sophie Dzhygir wrote:
You know Tomás, the fact that you haven't heard of or seen a performant MT system lately doesn't mean there does not exist one. 
Yes, I know. That's why in my last posting I said:
Nevertheless, companies with tons of repetitive/similar items like Microsoft with their Knowledge Base (95% of which is automatic translated) do need and use automatic translation, and these cases will certainly be a success.
| | | | | Let MT take care of the boring stuff! Thank you! :-) | Oct 9, 2007 |
CMJ_Trans wrote:
All of which goes to show that for technical and repetitive material, machine translation will see us all out. Not sure I care, given how boring that sort of stuff is but it will mean less bread to go around.
I remember when translation memory tools were going to take tons of translators our of their jobs.
Personally I think that, as translation of "bulk materials" like user manuals, specs, repetitive technical stuff, etc. shifts to MT in the middle run, companies will be happy to translate into more languages and enter more markets, simply because the bulk of their stuff might get translated at a fraction of the cost. So more and more companies will dare -or will need- to translate more of their trickier stuff (marketing, scientific, agreements, business stuff) and in the end there's going to be plenty of work for good translators.
| | |
|
|
|
| cutting-edge R&D | Oct 9, 2007 |
RobWT wrote:
Please also remember that computers are digital in nature and can only cope with simple logical ideas.
Also why spend billions to train a computer to do something that a human can do instinctively?
Do instinctively? Are you sure you want your translation done by somebody who lets their instincts rule over their brain when doing a translation? I definitely don't. There's way too many translations out there where human translators don't use enough logical thinking when translating and jump to the wildest conclusions.
Simple logical ideas? I definitely wouldn't call the algorithms and rules used to analyze the source language and then produce the target language "simple logical ideas". Of course, they're based on the standard programming logic of true, false, if... then, else, but with the fast computers of today you can crunch tausends of these steps in a split-second. That means even simple linguistic or semantic decisions are based on a wealth of information in form of grammar rules, context information, syntactic information, idomatic information, statistics, etc. etc.
Scientists are actively trying and making progress at reproducing the human neural system as, for instance, combining neurons with chips:
That's because nerve cells and brain waves are not digital systems that simply flip on and off. So the software instructions that drive silicon computers just won't cut it in this realm--and that's fine, since it makes crashes less likely. Brainlike chips will really be brainlike. They will be more creative than the machines on our desks, perhaps even mirroring some of the pluses and minuses of human thinking.
At this stage, Ditto says, it's just too early to tell if neurosilicon computers will have inherent limitations. But he and Sinha are optimistic that biosilicon systems can tackle anything today's hardware can--plus sensory-based computing that only biological ''wetware'' does with ease, such as understanding human language.
http://www.businessweek.com/1999/99_25/b3634137.htm
They have already created some computer/robots that are capable of emotions and of recognizing emotions:
eMo, the emotional robot
Do robots have feelings too? And will they ever recognise how we feel? eMo is an interactive robot developed by Professor Noel Sharkey of Sheffield University. eMo helps us to explore ways in which we might all be communicating with machines in the Future – through our emotions.
Professor Sharkey says “the serious side of trying to develop machines that can express and recognise human emotions is the role they could play in our lives – actually responding to our moods. One day an emotional machine could be fitted to cars to prevent a driver with road rage from doing something stupid.”
http://www.thinktank.ac/explore/futures/futures0.htm
They are working on creating computers that work more like the human brain:
http://www.engadget.com/2006/08/04/spinnaker-project-mimics-human-brain/
In comparison to some of the cutting-edge research in computer technology, "merely" coming up with some improved MT systems (some high-end and research systems are already pretty good, but obviously, you won't find them as free online systems) doesn't seem that difficult...
| | | | | Writers are not machines, instinct is a must! | Oct 10, 2007 |
Heike Behl, Ph.D. wrote:
Do instinctively? Are you sure you want your translation done by somebody who lets their instincts rule over their brain when doing a translation? I definitely don't.
In my opinion, instinct plays a big role in translation, even of technical stuff. Writers are not machines, and even if they do their best, they very often make mistakes which are not always easy to pinpoint. Instinct (and a great deal of experience) will help you notice mistakes and ultimately help the customer improve the source text. If a MT system is able to provide that kind of help to the writers within -let's say- 10 years from now, I will certainly quit translation and work as a farmer once for all!
| | | | Gabriel Csaba
Argentina
Local time: 07:50
inglés al español
+ ...
| Yes, but I won't live to see it | Oct 12, 2007 |
I answered that, but I actually meant I hope I won't live to see it.
I've got mixed feelings, or rather mixed opinions about this.
On one hand, it is true that machines can't handle things like humor, double meanings, nuances, purposeful nonsense, and myriads of other little details that make our work necessary.
But on the other hand, I have to agree with the Verne approach someone used earlier. I grew up in an age in which reality left science fiction b... See more I answered that, but I actually meant I hope I won't live to see it.
I've got mixed feelings, or rather mixed opinions about this.
On one hand, it is true that machines can't handle things like humor, double meanings, nuances, purposeful nonsense, and myriads of other little details that make our work necessary.
But on the other hand, I have to agree with the Verne approach someone used earlier. I grew up in an age in which reality left science fiction behind, and that tought me to avoid concepts like "impossible" or "sure". Technological advances today are only equaled in speed by the lowering of intellectual standards among its users.
There might come a time when only language professionals will care about the details that make their work necessary. ▲ Collapse
| | | | | Páginas sobre el tema: < [1 2 3 4] > | 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 machine translation ever replace human translation entirely? | Wordfast Pro | Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
| | Anycount & Translation Office 3000 | Translation Office 3000
Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
