TJ Blog
Wednesday, July 28, 2004
The Internet and the Translation Profession
It's a commonplace to state that the Internet has profoundly changed our profession. It has erased distances and national borders, and it's providing us with research tools of which we had never dreamed before.
Depending on where you live and how comfortable you feel with the new, ever-changing technologies, you may consider the Internet a curse or a blessing. How did it change your way of doing translations and doing business, and what would be your advice to your colleagues on facing the challenges of this Brave New Cyberworld of ours?
Depending on where you live and how comfortable you feel with the new, ever-changing technologies, you may consider the Internet a curse or a blessing. How did it change your way of doing translations and doing business, and what would be your advice to your colleagues on facing the challenges of this Brave New Cyberworld of ours?
posted by Gabe Bokor at 6:52 PM
Table of Contents
Entering the Translation Market
Translation Industry Survey
Communication is Power
Translators and the Recession
Quality
Contracts
Edit or Not to Edit
Specialist or Generalist
Certification
Machine Translation
The Goalie
Liability Insurance for Translators
Translator Licensing
Brave New World of Translation
Globalization
Professor Camayd-Freixas and the Postville, Iowa Raid
Who Pays for Technology?
Test Translations
Globalization
Good Translations from Poor Originals?
Simple comme bonjour
How to Predict Translator Performance
Holidays and Missed Calls
Late Payments and Non-Payments
On-Line Job Mediation Services: Are They Worth It?
The Role of Translation "Agencies"
Native Language
The Internet and the Translation Profession
43 Comments:
The Internet has become our "daily bread": we use it to advertise our services, look for work and to do research.
With the Internet, however, we have also seen a tremendous rise in the number of translators - or "translators" I should say. Anyone with an Internet connection and a smattering of foreign-language skills thinks he can hang out his shingle and call himself a translator.
Internet portals such as Translatorscafe or ProZ have done a lot of good for the profession, but also a lot of harm. Anyone can create a profile and "pad" it with "credentials" and "qualifications".
Recently I came across one "translator", who maintains profiles on both TranslatorsCafe and ProZ, and he states that he is a member of ATA. The only problem is that no one at ATA knows him, and he is not a member. But his profile on ProZ shows him as ID verified, thus giving him more credibility, even though his CV is filled with lies and fraudulent claims.
Or how about that Portuguese woman on one of these sites who claims to be a native English speaker, but can't write a single straight sentence in English?
Clients need to be extra careful these days. In fact, I would say, it's clients that benefit and, particularly, suffer the most in this new virtual translation market, because for them it's especially hard to distinguish truth from "fiction". I always advise them to check and double-check first before hiring a translator found somewhere on the Net. Too many among us, unfortunately, make fraudulent claims about their qualifications, and that gives us all a bad name (and could explain some of the insultingly low rate offers we have seen over the last few years).
Werner George Patels
www.translations-canada.net/wpatelsThe German-English Translation Blog
@Blind Tangerine:
Yes, but the Indian translator you think of speaks "Indian English", which is completely useless if the translation is intended for an English-speaking target group in Britain, the US or Canada. And they are not native English speakers and should therefore not translate into English.
In short, I would not consider him/her competition.
Werner George Patels
www.translations-canada.net/wpatelsThe German-English Translation Blog
It is a misperception to assume there are not many qualified Indians who have an excellent command of English. If you are translating from Japanese to English as I do, they are very much "the competition." Even if the work is not always as high quality, they still get many jobs and at rates that drag other rates down. But that is OK, MT will be largely replacing Indians and everyone else within a decade. Those who laugh at this haven't seen what they are creating at John Hopkins and MIT. Human translation will be largely gone by 2015, with some editing jobs opening up...
Yamishogun, Japan
Machine Translation does not work - and never will. Translation involves intuition, personal life experience, nuances, etc. - all things that a machine can never be programmed to do. And how could they? They don't even know how all these things work in the human brain.
Translation will always be a human activity. I said elsewhere not too long ago: there is more of a chance that we will be able to beam from point A to point B by the 24th century than have a working Universal Translator that can even recognize hitherto unknown languages.
Werner George Patels
Language is one of the most human activities – it is unpredictable and can never really be fitted into a strict regime of rules and regulations (and don’t be fooled by all those grammar books you find out there: even though some grammar books fill several hundred pages, native speakers use their language instinctively and intuitively).
In fact, very little is known about the processes in the human brain, and language and how it works in humans is just one of the many mysteries that still baffle scientists.
Translation, a process that involves the transfer from one language to another, is therefore even more complex than the “simple” use of everyday language. In order to translate well, you need to be able to “read between the lines”, which computers simply cannot do (and NEVER will). Just as much of an author’s life experience and personal feelings go into writing a novel or poem, each translation is a reflection of the translator’s experience, “world knowledge”, intuitive capabilities and creativity. If you think that a machine could ever reproduce that, you are sorely mistaken (and probably live in an ivory-tower sort of world).
Linguists at several world-renowned institutions such as MIT have been working hard to make machine translation work. The problem is that the minds behind these projects are linguists or theorists, but not practising and experienced translators.
Look at departments of linguistics or philology. Most curricula include some form of “translation exercise”, but these are exercises in what I call “grammatical translations”: they serve to test a student’s knowledge of grammar by having him or her translate relatively standard sentences into a foreign language.
For example: at the Department of English literature and language at the University of Vienna, students are required to translate German sentences into English to test their knowledge of English grammar and syntax, such as the proper use of “conditional clauses”.
Wenn ich reich wäre, würde ich mir ein Haus in Hollywood kaufen. (If I were rich, I would buy a house in Hollywood.)
These exercises are not about translation proper, that is, the transfer of meaning and information from one language to another. The important thing is to “translate” syntactical structures correctly.
Most sophisticated machine translation programs can master such simple sentences. In other words, the grammar and syntax in the target language will usually be correct, but that does not mean that the information is conveyed correctly.
Translation requires that you understand the source text. Without understanding, there is no way the translator can convey the meaning of the source text properly. Understanding is a human skill; a computer does not understand. A computer does not even “see” individual words: to the program this is all just code.
Let’s do an experiment: I took the above paragraph and had it translated by Google machine translation into German:
Übersetzung erfordert, daß Sie den Ausgangstext verstehen. Ohne zu verstehen gibt es keine Weise, die der Übersetzer die Bedeutung des Ausgangstextes richtig übermitteln kann. Verständnis ist eine menschliche Fähigkeit; ein Computer versteht nicht. Ein Computer glättet nicht?see? einzelne Wörter: zum Programm ist dieses aller gerade Code.
Now let us translate this back to English, again using Google machine translation:
Translation requires that you understand the output text. Without to understand there is no way, which the translator the meaning of the output text can convey correctly. Understanding is a human ability; a computer does not understand. A computer smoothes nicht?see? individual words: to the program is this all straight code.
While some of it is sort of intelligible, you can clearly see that the machine translation would require heavy-duty editing and/or re-translation. Also keep in mind that this passage consisted of relatively easy and short sentences. Imagine what would happen if I had used a long-winded paragraph from a contract.
Some companies believe they can cut costs by translating their business documents, manuals, contracts, etc. using machine translation. They know about the quality of machine translation and figure they can always get a human translator to “smooth out” any of the shortcomings. Well, that may be true, but it won’t cost them any less.
First, they have to invest a considerable amount of money in a high-level machine translation program. Second, human translators don’t edit machine translations for a pittance. Editing a bad translation is invariably more difficult than translating the whole document from scratch. It is also more likely that the editor will overlook some mistakes. As a result, the fee for editing a machine translation is at least 75% of what the translator would have charged for the whole translation. In addition, if the machine translation is really bad (= requiring at least 50% re-write/re-translation), the translator will charge the full translation rate and re-do the translation from scratch.
I do not doubt that some translators will lose business because of machine translation, but it will only affect cheap unqualified wannabe translators. If you are a professional translator (= with a degree in translation and a native speaker of the target language), you will not suffer at all. Cheap wannabes charge only a few pennies a word, and the quality of their “translations” is no better or worse than the quality of machine translation anyway. So, companies can easily bypass those “cheapies” by relying on machine translation. A real professional will then edit or re-translate the document for 75% or 100% of the regular translation rate.
For us professionals, therefore, nothing much will change: even now we are often asked to edit those horrible translations done by “cheapies” (= anyone working for less than 10 cents a word). In future, those horrible translations will not come from wannabes, but from machines. Big deal!
Werner George Patels
German-English Translator Blog
MT will replace human translators when novelists are put out of work by programs that can write literature.
I may as well take these comments one by one because some serious fallacies need to be addressed:
1) Machine Translation does not work - and never will. Translation involves intuition, personal life experience, nuances, etc. - all things that a machine can never be programmed to do. And how could they? They don't even know how all these things work in the human brain.
With increasingly fast computers , translations are probabalistically simulated. The working of the human brain is not needed. Look at even simple bablefish translation. Many sentences aer perfect. How can the computer do that withuot 'thinking?'
2. In order to translate well, you need to be able to “read between the lines”, which computers simply cannot do (and NEVER will).
No, they don't have to read betwen the lines. To say a computer can never do that is also a gross understimation of the potential of computers.
3. If you think that a machine could ever reproduce that, you are sorely mistaken (and probably live in an ivory-tower sort of world).
Mt fellow translator friends often convey how much "art" is in their work, but for over 95% of what we do, this simply not the case. A computer will be able to simulate this.
4. Linguists at several world-renowned institutions such as MIT have been working hard to make machine translation work. The problem is that the minds behind these projects are linguists or theorists, but not practising and experienced translators.
Because translators would have no contribution to make.
What contribution could we make? Just point out over and over "you have to read between the lines?"
5. Translation requires that you understand the source text. Without understanding, there is no way the translator can convey the meaning of the source text properly. Understanding is a human skill; a computer does not understand.
First, as I have said, it is NOT necessary to "understand" to have a perfect tranlation. Is the translation at Bablefish 'understanding?" Of course not. One can argue that this is a simple example, but similar processes are being expanded greatly.
Second, it is presumptious to say computers will never understand. But for this argument, I agree they won't.
6. While some of it (Google translation) is sort of intelligible, you can clearly see that the machine translation would require heavy-duty editing and/or re-translation. Also keep in mind that this passage consisted of relatively easy and short sentences. Imagine what would happen if I had used a long-winded paragraph from a contract.
You are using a free MT program, yet even this could not have been reproduced 10 years ago. It strikes me how translators laugh at these translations when those working in the 1970s would have been spooked by the quality, the speed and the cheapness (free).
7. They know about the quality of machine translation and figure they can always get a human translator to “smooth out” any of the shortcomings. Well, that may be true, but it won’t cost them any less.
This is true in 2004. I am talking about 2009 and 2014.
As MT gets better, there is less smoothing out.
8. First, they have to invest a considerable amount of money in a high-level machine translation program.
The costs continue to come down sharply. These companies are payign thousands to translators where this adds up over time. The cost savings will be enormous.
9. I do not doubt that some translators will lose business because of machine translation, but it will only affect cheap unqualified wannabe translators. If you are a professional translator (= with a degree in translation and a native speaker of the target language), you will not suffer at all.
Basic economics (which sadly very few translators understand) dictate supply and demand will determine fees. As MT gets better, it will lower the supply of work available to the better translators. And MT will hit anything technical first. If you are in the Jap/Eng line of work as I am, this is a huge percent of what needs transling. So more translators will be fighting over a smaller pie,and wages drop. Period.
10. For us professionals, therefore, nothing much will change:
Except that if you are translatsing European languages, you will be out of a job within 5-10 years. Seems like a big change to me. Japanese, Chinese and a few others will be harder to crack, but wages will be pushed down here as well. Those who translate novels will be working, but they will have to compete with many translators who no longer get tech work.
It also strikes me as odd that for all the (nervous) laughter MT geberates among translators, the topic sure does appear in the Translation Journal quite often.
From the NY Times (2003):
"Big Leap in Machine Translation"
...N-grams (with "N" representing the number of terms in a given phrase), are the basic building blocks of STASTICAL machine translation, or MT.
Although in one sense it was more economical, this kind of machine translation was also much more complex, requiring powerful computers and software that DID NOT EXIST for most of the 1990's. But a workshop at Johns Hopkins University in Baltimore changed all that in the summer of 1999. ...
...Today, researchers are racing to improve the quality and accuracy of the translations. The final translations generally give an average reader a solid understanding of the original meaning but are far from grammatically correct. While not perfect, statistics-based technology is also allowing scientists to crack scores of languages in a FRACTION of the time, and at a FRACTION of the cost, that traditional methods involved....
... "All our techniques require is having texts in two languages. For example, the Klingon Language Institute translated 'Hamlet' and the Bible into Klingon, and our programs can automatically learn a basic Klingon-English MT system from that."
Yarowsky said he hoped to have working translation systems for as many as 100 languages within five years. Although the grammatical structures of languages like Chinese and Arabic make them hard to analyze statistically, he said, it will only be a matter of time before such hurdles are overcome.
[5 years? Maybe, maybe not. But the writing is on the wall. It is important that translators, especially those who translate European languages understand what is coming this decade. -- Yamishogun]
Translation does not involve "understanding"? Gimme a break. Translation is ALL ABOUT UNDERSTANDING. Without understanding the information, you cannot convey it. Computers don't understand, therefore they cannot translate properly.
You are talking about "transliterations", and that's what computers do. Yes, occasionally they will get it right, because some transliterations "happen" to be correct translations, but this is the exception, not the norm. Just as wannabe translators sometimes get it right. That does not mean that they are good translators, and the same is true of MT.
I am not talking about lists of machine parts - something like that could be handled by a computer. But whenever you have "narrative", you need a human professional translator to get it right.
Translation = gleaning of meaning - words don't matter at all! But the computer only sees words (or, rather, code).
Please don't repeat that nonsense (translation does not involve understanding) again, unless you want to be seen as a "transliterator" instead of a translator.
Werner George Patels
The German-English Translator
And, please, get off that high horse of yours. European languages are not "inferior" to Asian languages. Computers can't crack any of these.
How do you suppose a computer can achieve this "reading between the lines"? How can a computer "see" or "understand" some hidden reference to something entirely else? Only the human brain and experience can do that, and we perform this "task" in our translation routine several times a day every day. To say that computers don't need that reveals your poor understanding of what translation is all about (many engineers that stumble into technical translation have no clue and are therefore bad translators).
Even the biggest and most sophisticated computer cannot achieve what the human brain does - not as far as emotions, intuition and language are concerned.
Werner George Patels
Do you honestly think that "tech work" is all translators ever do? How conceited!
Technical translations don't even make up 1% of my work: I handle highly complex and convoluted legal contracts (good luck trying to "translate" that with a computer), marketing texts, business-related texts, etc. - all of which is not "tech work" at all and could not be "deciphered" by even the most expensive computer on earth.
Werner George Patels
If anything, my learned friend in Japan, it's Japanese that would bite the dust first, given its simplistic and almost mechanical grammar and syntax (I studied Japanese at university for over two years).
Japanese sentences are very "simple", compared to European languages, and if a computer were ever to replace any language, it would be Japanese, and not European languages.
For example: as regards European languages such as English, you can study grammar, idioms, vocab until you're blue in the face, but you'll never pass for a native - European languages, and especially English, are highly intuitive languages.
Japanese, however, is a different story. A "gaijin" can learn it and learn it well - to the point that you can write like a native.
Werner George Patels
Dr. James McCabe has a regular column on Business English in the German version of Business Week ("WirtschaftsWoche"). In his most recent column he writes:
"Translating your critical findings directly into English may not only hinder progress but endanger your business relationships."
Essentially he says that a direct translation (even if it is grammatically, etc. correct) does not work; you need to customize and adapt the translation with respect to your target audience and the target culture. A computer cannot do that.
True, with many technical translations (e.g., lists of machine parts) that isn't an issue, but when you translate even the simplest of annual reports, you'll have to consider the target audience and culture. Again, a computer cannot do that.
By the way, speaking of technical translations: I have been in business since 1987, but I noticed a drastic drop-off in the number of technical translations over the last five years or so. Even though I specialize in commercial, legal and financial translations, I would also handle technical translations. I think there is not that much demand for purely technical translations anymore, because
a) these texts are written by engineers from scratch in various languages
b) many companies in that field have their own technical writers (for most major languages) - and technical writing has become a fully recognized profession
c) most technical translations involve very "mechanical" texts such as parts lists, and that CAN, in fact, be handled by MT (which is then reviewed and edited by in-house technical writers/translators).
Werner George Patels
Again, for clarity, I will respond point by point.
1.
>Translation does not involve "understanding"? Gimme a >break. Translation is ALL ABOUT UNDERSTANDING. Without >understanding the information, you cannot convey it. >Computers don't understand, therefore they cannot >translate properly.
I used Babblefish to translate your first sentence.
Translation does not involve "understanding"?
翻訳は理解することを含まないか
Looks pretty good to me. This is a _free_ program
written a few years back. My point is that there are now sentences that can be translated flawlessly, and so yuor insistance that translation requires "understanding" is clearly erroneous.
What you may intend to say is that "at the present time, complex translation requires "understanding." As computers become more powerful combining statistical and other approaches, complex sentences will also be translated as well as the best humans. It is also human conceit (well, hey, Im cheering on the humans!) to assume even the best translators have 100% accuracy.
2. You are talking about "transliterations", and that's >what computers do. Yes, occasionally they will get it >right, because some transliterations "happen" to be ?>correct translations, but this is the exception, not >the norm.
It is not the occassional random sentence that Babblefish is getting correct. and this is a FREE program written years ago.
3>I am not talking about lists of machine parts - >something like that could be handled by a computer. >But whenever you have "narrative", you need a human >professional translator to get it right.
You are talking about two extremes. Obviously MT could handle lists of tech terms years ago. And no, MT will not be doing poetry. But technical tranlsation which still can't be done well with MT is in striking distance. Using a stastical method, there is no reason that pulp novels like a Tom Clancey mystery couldnt be translated with MT in the future (7-15 year range). It may require an editor, but so do translators. It is going to stun people how good the translations get, even with no "understanding."
4. >Translation = gleaning of meaning - words don't >matter at all! But the computer only sees words (or, >rather, code). Please don't repeat that nonsense
>(translation does not involve understanding) again, >unless you want to be seen as a "transliterator" >instead of a translator.
If you choose to include "gleaning of meaning" in your definition, then of course I can't argue against it. But most people would accept "translation" as the written near equivalent of another language. Just as if one says "love is an emotion between a man and a woman" I'd be at a loss to explain what homosexuals feel. The definition precludes it.
5. >And, please, get off that high horse of yours. >European languages are not "inferior" to Asian >languages. Computers can't crack any of these.
You put "inferior" in quotes, although of course I never implied any such thing. I'm simply repeating what the John Hopkins researcher stated: We expect to crack 100 languages in 5 years, and while Chinese and Arabic pose more of a problem, those can be beaten. What he didnt mention was the difficulty in going from Japanese into English due to specificity problems. This could b ebeat by simply requiring Japanese tech writers to be explicit where they might not be in the vernacular.
6. >How do you suppose a computer can achieve >this "reading between the lines"? How can a >computer "see" or "understand" some hidden reference >to something entirely else? Only the human brain and >experience can do that, and we perform this "task" in >our translation routine several times a day every day. >To say that computers don't need that reveals your >poor understanding of what translation is all about
>(many engineers that stumble into technical >translation have no clue and are therefore bad >translators).
It is simply arrogance for one translator to tell another he has "poor understanding of what translation is all about. And for the record, several technical translators I know also translate non technical Japanese very well. But you hit on an important point Ihave made: I know an excellent translator who jokingly boasted to me "I knew nothing about banking, but I still did that banking translation ....pretty well." OK, so did he have "understanding?" Probably with narrow sentence structure, and maybe for the job that was sufficient.
7. >Even the biggest and most sophisticated computer >cannot achieve what the human brain does - not as far >as emotions, intuition and language are concerned.
In 2020, the machine on your desk will have the capacity of a human brain. What was the capacity of your computer when you started translating years back? The speed was a blazign 4 MHz, right? The 2020 computer wont have emotions, nor will it think, yet it is not at all inconceivable that thinking machines will exist in 25 - 50 years. But I am always assuming this wont ever happen.
8 >Do you honestly think that "tech work" is all >translators ever do? How conceited! Technical >translations don't even make up 1% of my work: I >handle highly complex and convoluted legal contracts
>(good luck trying to "translate" that with a >computer), marketing texts, business-related texts, >etc. - all of which is not "tech work" at all and >could not be "deciphered" by even the most expensive >computer on earth.
Let's see, you have done tech work, leagal work and business texts. Do you have degrees in law, science and business? Can you honestly say you have "understood" everything you have done?
Technical translations comprise of the overwhelming majority of work from Japanese into English. Consequently, there is a multi-billion dollar incentive to improve MT. What you clearly fail to understand is that this is not "all or nothing" as more sophisticated MT will undoubtedly take away work from translators and this will effect _every level_ of translator as it already has in my language pair. If you know enough to translate business documents, surely you must no basic supply and demand to figure this out.
and yes, legal translations will be tougher to crack, althuogh it isnt unreasonable to assume that lawyers would be forced to use certain language much as the US pattent office forces the use of certain language. But if MT is takign away many other jobs than translators will be fighting for the smaller legal pie.
Anonymous said...
9>If anything, my learned friend in Japan, it's >Japanese that would bite the dust first, given its >simplistic and almost mechanical grammar and syntax (I >studied Japanese at university for over two years).
Two years is not enough to get past the most basic grammar.
>Japanese sentences are very "simple", compared to >European languages, and if a computer were ever to >replace any language, it would be Japanese, and not >European languages.
For the reason I mentioned earlier (specificity) translating _into_ Japanese would be doable, but translating into English is far , far harder. Typing the above with respect to the "simplicity" of a Japanese sentence only serves to embarrass yourself.
>Japanese, however, is a different story. A "gaijin" >can learn it and learn it well - to the point that you >can write like a native.
No offence, but you are absolutely ignorant on this score. You knew nothing about the level of some Indians speaking English, so I should not have expected much.
I don't think you have an accredited degree in translation, do you? So who's the ignorant one here?
I did study all those disciplines as part of my degree program, and, yes, I do understand everything I translate; otherwise, I would not handle those jobs. Being well-read in a number of disciplines is more than enough by way of specialized knowledge. Remember: the translator does not have to be able to perform the actual brain surgery; we "only" have to translate ;-)
Your friend must have understood the banking text, because if he didn't, his client would have thrown the translation back in his face.
Translation is all about meaning and understanding - this is the fundamental principle of all translation theories. Who are you to discard that?
My Japanese studies were very intensive. Trust me, in two years we covered what you did in 10 years, and I am not exaggerating.
Babelfish is a "toy". Of course, it will get the occasional (simple) sentence right. I never disputed that fact.
A computer that can do everything the human brain can? By 2020? No way! No one knows how the brain really works (apart from assigning various human tasks to different areas of the brain and measuring brain waves and such - but that still does not tell us HOW and WHAT and WHY). And I'll bet they won't have those answers by 2120 either (or 2220, 2320, ...).
If you call yourself a translator, you should actually be ashamed of yourself. You come across as a member of our profession that does not have much confidence in what he does for a living (nor much respect, for that matter). You seem to be hellbent on reducing translation to a mechanical process that can be reproduced with relative ease (while also denigrating humans and what they stand for). I suppose all this stems from major frustrations at work. I guess you are not really a "happy translator", eh? This is also why you don't reveal your identity, because you don't want any of your clients (or potential clients) to see your frustrated rants.
I pity you, because with an attitude like that, you'll be the first to fall by the wayside - MT or no MT.
Werner George Patels
Regarding Indians: they are not native speakers of English and therefore should not translate into English - or French, German, Italian, etc., as they frequently claim they can.
They might serve a purpose when it comes to outsourcing programming jobs, but even something as straightforward as medical transcription is not outsourced to India anymore. The Economist reported on that: the quality was so bad that the companies in the US decided to repatriate their medical transcription services. If Indians cannot even handle transcriptions, how can they be relied on for translation services into non-Indian languages?
You, sir, still have a lot to learn about languages and translation. How about going back to school and enrolling in a quality T&I degree program?
Werner George Patels
There are many Indians who have "near native" English fluency, and two of them are friends with excellent Japanese who translate now in India. Who are you to say that they should not be translating Japanese into English? They are inexpensive AND the companies they work for are pleased with their work. There are many native English speakers translating Japanese into English who are not nearly as good as they are. This is a common translator's conceit: "only the very elite translators should be working." But in a free market, this is rightly ignored.
--------------
If Indians cannot even handle transcriptions, how can they be relied on for translation services into non->
Indian languages?
--------------
It isnt a problem with "Indians" per se but the ones employed who could not do quality transcriptions. Those with excellent English do other work that is more highly valued, but there is an English explosion in India (and China) and those jobs will be going back.
---------
You, sir, still have a lot to learn about languages and translation. How about going back to school and enrolling in a quality T&I degree program?
---------
You, who just made utterly laughable "insights" about the ease of the Japanese language, are telling _me_ I know little about languages? Why don't you ask my clients if I need to learn "a lot about translation." Maybe you could take your 2 years of Japanese training and start translating. That should be amusing. Why not get a degree in translation? Because one is not needed to do high quality work. I have degrees in physics and economics. Those help me quite a bit.
But why be so upset? Why shoot the messenger who says human translators are in serious trouble? Shouldn't you rather be thanking me for informing about the future of this profession? You are most welcome.
Yamishogun
-----------
My Japanese studies were very intensive. Trust me, in two years we covered what you did in 10 years, and I am not exaggerating.
-----------
Nothing surprising about this Euro attitude in learning Japanese. Many of my European friends said the same thing -- until they went to Japan. Now they are a bit more humble. In my ten years, I had two years of nothing but Japaense, I lived with a Japanese family for two years, and I have lived in the country for 8 years total. Your two years somehow outstrips this?
---------------------
A computer that can do everything the human brain can? By 2020? No way!
---------------------
Read my sentence again. I said the capacity and in no way said "everything the human brain can."
------------
If you call yourself a translator, you should actually be ashamed of yourself.
------------
ah, more moralizing from my German friend. I call myself a translator because I translate for money. No, I don't do poetry. Is that the requisite quality? Are you more of a translator than I am? My clients are happy, your clients are happy. What shoud I be ashamed of?
------------------
You come across as a member of our profession that does not have much confidence in what he does for a living (nor much respect, for that matter).
------------------
I have plenty of confidence, and I seem to have more respect than you do for the profession as I don't tell which people should enter and which should not.
---------------------------------
You seem to be hellbent on reducing translation to a mechanical process that can be reproduced with relative ease (while also denigrating humans and what they stand for).
----------------------------------
I am simply saying what is going to be achieved in the near future. You are the one who is so emotional about this, even going so far as to say "I denigrate humans and what they stand for" Isnt that a wee bit over the top? How can I take you seriously with comments like this?
-------------------------
I suppose all this stems from major frustrations at work. I guess you are not really a "happy translator", eh? This is also why you don't reveal your identity, because you don't want any of your clients (or potential clients) to see your frustrated rants.
---------------------------
I am a very happy translator, but I also know that the ability to earn a living as a translator in a few years will be quite limitted. I have other skills , and that work should keep me happy as well. Heck, maybe I will do a little editing on the side, but I doubt it.
The laughing at the dawn of truly functioning MT has become a religion among many translators, and this is sad. As MT becomes much more powerful in the coming 5-10 years, you can scream as loudly as you wish about how "they cant truly translate" and "they can't think" but companies will collectively nod, and continue using computers to do most of the translations people currently do.
By the way, an editorial in this very journal a year or so ago said the same thing I am saying. Maybe you could look that up.
Did you notice I haven't mentioned anything about 1. your ability to do your job 2. your outlook on humanity 3. your general emotional state 4. your pet goldfish. This can all be done without the childish commentary.
Yamishogun
For someone who makes so many spelling mistakes (not to mention others), I am sure your chances of making good money must be "limitted".
Just for the record: I am not German, I am Canadian.
As I just wrote in the other thread on this blog, both ATA and ITI (in the UK) clearly state: "Professional translators work into their native language." Having "near native" command does not cut it. You're either a native speaker or you're not. If you are not, stick to translating into your native language.
According to ATA, ITI and others, a translator who flouts this "basic rule is likely to be ignorant of other important quality issues as well." (Source: Translation - Getting it right. A guide to buying translations. 2003).
Werner George Patels
Your rant was nothing but utterly childish. I will not tell you what to believe. But given what you do believe, it is quite clear that you have never mastered the fundamentals of professional translation. Someone like you will be among the first to be replaced by MT (even if it's as crappy as Google translation, because even that little "toy" produces a better translation, apparently).
Werner George Patels
You mentioned an article in the TJ. Well, perhaps, you should have read this one more closely:
http://accurapid.com/journal/29computers.htm
Here it says, for example:
"Given the complexity of the phenomena that underlie the work of a human translator, it would be absurd to claim that a machine could produce a target text of the same quality as that of a human being."
Well, apparently you have been quite "absurd" in claiming that many of us will be out of business in 5 to 10 years.
The authors of the article clearly explain that even after 2007 (or whatever date you'd like to place on it) no MT can ever do without the input of a human translator (if for no other reason than the fact that the "knowledge of socio-cultural aspects, that is, of the customs and conventions of the source and target cultures" cannot be translated into code and fed into the computer).
They also explain that MT (if ever) will only be an aid at the first stage of the translation process (= producing a rough draft), after which a human translator will take over. That's why I said in the very beginning that I would simply charge 75% of my regular translation rate for editing/revising machine translations, or 100% if more than 50% has to be re-written. But, of course, it will never come to that, because most companies are smarter than you think.
You see, they cannot be sure of the end product and don't know whether the machine translation will have to be re-translated from scratch at the full rate. So, they will simply give it to a professional human translator (one with a degree and all other necessary credentials and experience) and have it done right the first time round.
And this is happening even now: some companies have "cheapies" (= unqualified translators working for peanuts) do the rough draft and then hand it to a professional for editing and any other "repairs". In future, these types of companies will surely use MT for the first stage, no doubt, which means that all the cheapies will be out of work - COMPLETELY.
Then, there are companies, and these are the ones I work with, that are very demanding and careful in selecting a suitable translator: they demand a native speaker of the target language who has a degree in translation, is accredited and has plenty of experience. And these clients will NOT switch to MT. In fact, these companies are only too happy to pay a premium for quality HUMAN translation. They are not cutting corners now by assigning translations to wannabes and cheapies, so why should they change their behaviour later on?
Werner George Patels
For someone who makes so many spelling mistakes (not to mention others), I am sure your chances of making good money must be "limitted".
----------------------------
I type ectremely quickly typing something like this with no importance. If my spelling offends you, you could always run it through a spellchecker. (amazing what computers can do these days...) Money is pretty good in Jap/Eng translating. Thanks for asking.
--------------------
Just for the record: I am not German, I am Canadian.
---------------------
When I saw you mention how your Japanese is 2 years ran ccircles around my 10 years, I assumed you must have studied in Germany. Wrong assumption.Canadains can be as misguided about their lnguistic skills, so a faulty assumption on my part.
------------------------
As I just wrote in the other thread on this blog, both ATA and ITI (in the UK) clearly state: "Professional translators work into their native language." Having "near native" command does not cut it.
-------------------------
Well of course these organizations would say this! That is the ideal, but companies that want quality yet do not want to pay premium prices do very well with non native translators. The Indians are happy and the clients are happy. Who aer you to interfere?
----------------
Your rant was nothing but utterly childish. I will not tell you what to believe. But given what you do believe, it is quite clear that you have never mastered the fundamentals of professional translation.
----------------
You can continue to embarrass yourself on this blog, but 1) what have I said that is childish (you were the one questioned my world view of humanity. What fundamentals am I missing? I gave a physics/economics degree, and I have been translating for a few years now with happy clients. Can you send me the missing piece?
-----------
Someone like you will be among the first to be replaced by MT (even if it's as crappy as Google translation, because even that little "toy" produces a better translation, apparently).
------------
more name calling. yare yare...
---------
You mentioned an article in the TJ. Well, perhaps, you should have read this one more closely:
http://accurapid.com/journal/29computers.htm
Here it says, for example:
"Given the complexity of the phenomena that underlie the work of a human translator, it would be absurd to claim that a machine could produce a target text of the same quality as that of a human being."
---------------------------------
you can also read this one
http://accurapid.com/journal/13mt.htm
to make yoursefl feel better. He is also delusional.
----------------------------------
ah, but what about...
http://accurapid.com/journal/15mt.htm
"Still, the output may not be good enough for public display, so the question turns into: will the future of human translation be... proofreading computer output?
The bad news is yes. It all boils down to how long it will be until computers produce decent translations. All they need is basically here."
Well, apparently you have been quite "absurd" in claiming that many of us will be out of business in 5 to 10 years.
-------------
You see, they cannot be sure of the end product and don't know whether the machine translation will have to be re-translated from scratch at the full rate. So, they will simply give it to a professional human translator (one with a degree and all other necessary credentials and experience) and have it done right the first time round.
---------------
They will clearly know based on experience how much the MT can do, and know where near "from scratch" You keep talking about current MT,a nd Im talking about 5-10 years down the road. . Your idea that you will get anywhere close to 75% for editing the same document is off the wall.
----------
they demand a native speaker of the target language who has a degree in translation, is accredited and has plenty of experience. And these clients will NOT switch to MT. In fact, these companies are only too happy to pay a premium for quality HUMAN translation. They are not cutting corners now by assigning translations to wannabes and cheapies, so why should they change their behaviour later on?
-------------------
Because changing their behavior later on will save them enormous amounts of money. You still think in the present and will only discuss current MT. You laugh at the "toy" that is Babelfish, but that would have blown people away when you started out. If that is the toy in 2000, what do you think professionals will have in 2010? In the J/E field I have never heard of a clietn who demands a separate "translation degree." You get that knowledge from experience.
yamishogun
You started this thread by saying "Machine Translation does not work -- and never will." ne dit jamais jamais....
You added on your last post:
------------------
The authors of the article clearly explain that even after 2007 (or whatever date you'd like to place on it) no MT can ever do without the input of a human translator (if for no other reason than the fact that the "knowledge of socio-cultural aspects, that is, of the customs and conventions of the source and target cultures" cannot be translated into code and fed into the computer).
-----------------
You need to read up on how statistical MT works and why it has enoromous potential. The code doesnt attempt to input "customs and conventions" yet merely examines what hundreds of expert translators have _already_ done. This statistical learning isnt the same as how a human learns, although there are similarities. In this sense, your dream of translators finally entering the MT creation process is being realized in spades. It is only after many documents are translated that statistical translation works so effectively and with amazing speed and high accuracy. Not perfect, and there will be editors. Translators are vital for MT to become truly advanced, as it will in 5 -10 years time. 5 years may be early, but the translation industry will be spooked in 5 years with everything changed in another 5 years (give or take). Look at the Translation Journal over the past couple of years. They sure write a lot about MT if it is no threat.
You can whistle in the dark or start thinking of what you would like to do in a few years. Edit? maybe some novel translation? Seahorse farming?
A couple of points about the article I linked to earlier from The Transltion Jnl:
"A decent MT machine is just around the corner. The computer computes chess, while Kasparov plays chess. A computer will never understand, but it can translate, at least to some extent. And, since translation without understanding is meaningless, the future of the human translator is proof-sensing what a machine has pre-translated."
What this writer failes to mention , writing a couple of years ago, is the advance in statistical methodology for translation. This is an unfortunate omission. Furthermore he makes a common error that "translation without understanding is meaningless." We have already established that the toy known as Babelfish can translate fairly basic sentences _without understanding_
Multiply this power 1000 fold, which is happening with statistical techniques, and you have true MT. No toy.
The editing for most documents will be light and require little time.
yamishogun
You say those companies will change their behaviour in the future, because they will be able to save vast sums of money.
What do you base this on? Because I already mentioned that they can save tons of money now (by going to cheapies and wannabes), but they'd rather pay a premium for getting right.
So, if these companies (the second type I listed in my earlier post) don't do it now, why should they in the future? Your logic is totally warped!
Werner George Patels
-----------
What do you base this on? Because I already mentioned that they can save tons of money now (by going to cheapies and wannabes), but they'd rather pay a premium for getting right.
So, if these companies (the second type I listed in my earlier post) don't do it now, why should they in the future? Your logic is totally warped!
---------------
You insist on looking only 3 inches ahead for some reason. Firms can not normally save a ton of money because the "cheapies and wannabes," as you call them, are not consistently reliable. Yet a major patent firm in Tokyo just considered hiring Koreans with American editors to translate because the quality was decent and the cost low. That firm is sticking with mostly Japanese translating into English because the quality/cost ratio is more favorable. They also hire native English speakers but use Japanese in the "wrong direction." Why? Because it works for them.
As you must certainly know, the world is not cleaved nicely into "wannabes" and "professionals." There is a range of talent. Firms know this as well.
MT in 5-10 years will not be the same quality as the "cheapies," but better. It would take a "cheapie" a long time to do one patent. MT could do 10 pages a minute. Statistical MT should be able to instantly flag problem sentences as well. At this point, they hand it to someone with decent Japanese, not necesarily excellent Japanese. I know 5 Americans who work at a firm in Tokyo, and only two were good enough to quit the inhouse editing job and translate on their own.
There is going to be a major market squeeze when everyone becomes an editor or vies for the novel to tanslate. It might be different in the German world, but in Japan, very few novels are translated into English. The demand isnt there.
So where is my logic "whacked?" It only is by using your assumption that since a computer can't think, it can never turn out decent translations. OK, they will only be "transliterations" if you prefer. But a company would rather have those cheaply than pay thousands of dollars for a lengthy patent.
All of this is obvious.
Again, professional translators will not be out of work - especially those specializing in the more lucrative fields of law, business, finance, marketing and creative-type translations. Technical translations, as already discussed, are of lesser importance these days anyway, so anyone who does nothing but technical is already losing business as it is.
As the article in the TJ states, you are absolutely "absurd" in your beliefs. Dream on and have a nice life serving sushi pretty soon. You are the textbook case of a scientist living in an ivory tower and engaging in activities that you should really stay far away from ....
Werner George Patels
Again, professional translators will not be out of work - especially those specializing in the more lucrative fields of law, business, finance, marketing and creative-type translations. Technical translations, as already discussed, are of lesser importance these days anyway, so anyone who does nothing but technical is already losing business as it is.
----------------------
Not in the Japan field. Most I know translatign J/E do nothing but technical and make over $100,000/yr. Maybe that will change, but not yet.
-----------------------
As the article in the TJ states, you are absolutely "absurd" in your beliefs.
-----------------------
The article was written by three people who have done only language study their entire lives. Since they said it is "absurd" I guess it must be absurd. But note how they, like you, automatically say "human quality." I simply say much closer to human quality than today, and it will have a huge impact on the translation field. Another author in TJ writes:
"so the question turns into: will the future of human translation be... proofreading computer output? he bad news is yes. It all boils down to how long it will be until computers produce decent translations. All they need is basically here."
Is he an ivory tower, incompetent translator for stating this?
----------------------
Dream on and have a nice life serving sushi pretty soon. You are the textbook case of a scientist living in an ivory tower and engaging in activities that you should really stay far away from
-----------------------
Again, your ad homenum attacks on a successful translator do nothing to bolster your argument. I am not sure why you do not have the maturity to discuss this without the personal attacks, but it seems to be the case.
My, you are arrogant beyond belief: "ONLY language study"!
So, getting a degree in languages or translation is inferior to your degree in physics? With this statement of yours you have disqualified yourself once and for all. In fact, the people that are most qualified to comment on the success or failure of MT are practising and trained translators, linguists and other language experts - you, however, are none of the above.
Werner George Patels, M.A., C.Tran., C.Conf.Int.
My, you are arrogant beyond belief: "ONLY language study"!
-----------------
Once again, you cant post without name calling. Is there a spcial reason for that? Don't you respect TJ to keep it above the mud? Go look at your posts, constantlty belittling a translator who has satisfied clients.
-----------------
So, getting a degree in languages or translation is inferior to your degree in physics? With this statement of yours you have disqualified yourself once and for all.
-----------------
How about at least having the courtesy to not lie about what I write. It makes you look like a fool. I said nothing about any degree as superior to another. But people who only study languages often have insufficient technical training to understand what is on the horizon with respect to MT. Did you understand the recent statistical techniques? Not judging from your posts.
----------------------
In fact, the people that are most qualified to comment on the success or failure of MT are practising and trained translators, linguists and other language experts - you, however, are none of the above.
-----------------------
Im a translator, and I can tell quality as well as you in my language pair. How can you make such arrogant statements and not be embarrassed? The people most qualified to comment on MT are those developing it. I can guarantee that translators will continue to say "this is horrible" since they have every incentive to do so, but at some point in teh near future it will be obvious how advanced MT will be. The John Hopkins MT team leader thinks he can crack 100 language in 5 years. He seems like an expert to me. Maybe he is optimistic, but do you know more about stastical MT than this guy? And your response is to quote translators who simply use the strawman argument "it is absurd to think MT can translate as well as a human."
I'm sure my grandmother (a liguist) thinks the same thing.
You say you studied economics, business, etc at Translation U. Why then, do you not understand that this is not "all or nothing." Every post you put up is an extreme or absolute: "you know nothing about translation." "Only the cheapies will be affected" "MT is impossible because a computer can't understand."
Aren't you old enough to realize we live in a much more complex world than you make it out to be? You can laugh as hard as you want at Bablelfish, a free program that is 5 years old. You won't be laughing at what is coming along in 5 years.
I am not lying about what you write: you wrote that the authors of the article "only" studied languages. Well, you only studied science, but not languages - and certainly not professional translation (that is crystal-clear from YOUR writing).
Why do you have to belittle these qualified people? I suppose that's envy talking, because you probably got kicked out of language school.
Werner George Patels
I make spelling errors when typing fast. That is true. Again, all the name calling doesnt change the fact that extremely powerful MT is very close. I actually did attend a language school for a full year and did well.
My Japanese was ranked in the upper 10% of those who attended many years ago. Thanks for asking. Im sure your 2 years of Japanese is way beyond my 10 as you state. Good with in your new line of work in a decade. I doubt it will be in public relations. I hear there may be some editing needed. I will probably take a pass on that myself.
I'll illustrate your attitude and beliefs by giving you an example:
Some engineers want to create a brand new surgical equipment or procedure - but they reject all professional input from surgeons (the ones that would use this equipment or procedure on a daily basis). In fact, these engineers keep belittling surgeons and ridiculing their training and expertise. And when surgeons say that they don't like the equipment or procedure, the engineers say, "They only studied medicine. Their opinions don't count."
Translators (those with training) are the experts in translation. If an MT program is to be created, such project ought to be led by professional translators - with scientists, engineers and programmers serving as assistants only, sharing their expertise.
It makes one wonder why you, as a scientist-cum-economist, have to dabble in a field you have not been trained for. Couldn't cut it in your own discipline? Sorry, but that's the logical conclusion. I got my degree in translation, and that's the field I work in - I don't have to go and dabble like a fool in somebody else's area of expertise.
That you are not a professional translator is sufficiently evidenced by your beliefs about my profession - and your constant belittling of the work we (not you) do. To think that a complex - and human - process can be reduced to 0s and 1s is not only hubristic, but outright delusional.
Science always tries to imitate life, but it has never worked and never will. Even the best photocopy in the world is not as good as the original. Even digital and HD television still does not look "real" (and never will).
Language is something beautiful - not a code that you attempt to "crack" (as you keep saying). Language is so unpredictable (because every single person has his/her own "reality"). Translation requires understanding - without understanding there is no translation. Being able to translate the occasional simple sentence word for word and still have it come out right in the target language is only more proof of the above statement (the "infamous" exception to the rule). Even a cheapie/wannabe gets it right once in a while, and so will a machine, but that is no reason to put them "in charge" of the translation industry. In fact, it only reinforces the need for even more professional translators.
And if MT really comes, to the point where we're reduced to editing and re-translating, it will be even more pressing to have real professionals to do the job. As the authors of the TJ article point out, there is always that heightened risk of overlooking mistakes when editing somebody else's work. In other words, a good editor will be worth his weight in gold ...
Werner George Patels
------------
Translators (those with training) are the experts in translation. If an MT program is to be created, such project ought to be led by professional translators - with scientists, engineers and programmers serving as assistants only, sharing their expertise
-------------
Nobody is ridiculing translators nor would engineers ridicule doctors. But an engineer would rightly wonder why the doctor would know much about the mettalurgy
used. If the goal is to create a computer program, why in the world would the programmers be "assistants only?"
Current advances in MT rely on linguists but the "rule-based" platforms stalled in the 1990s. Statistical and example approaches are responsible for new breakthroughs.
---------------------
It makes one wonder why you, as a scientist-cum-economist, have to dabble in a field you have not been trained for. Couldn't cut it in your own discipline? Sorry, but that's the logical conclusion. I got my degree in translation, and that's the field I work in - I don't have to go and dabble like a fool in somebody else's area of expertise.
----------------------
You can't go entire post without insults. Is there a reason you can't stick with the argument? I suppose my grandmother wears army boots as well... Sad. Just because I have degrees in physics and economics does not imply "dabbling" as you call it. I would say I have been extremely fortunate with respect to my career. What can I say except that it has been rewarding, including translating. Thanks again for your concern.
--------------------
That you are not a professional translator is sufficiently evidenced by your beliefs about my profession - and your constant belittling of the work we (not you) do. To think that a complex - and human - process can be reduced to 0s and 1s is not only hubristic, but outright delusional.
----------------------
I translate for 16 cents/word, and I translate well. It is most definately work _I_ do. Why not email Yves Champollion at yves@champollion.net and call him "outright delusional" since he wrote in TJ what I have been arguing here. I have not once belittled any aspect of translation. (Unlike you, with your belittling "cheapies" and "wannabes" labels) I don't even belittle tranlators who do not understand how current MT limits do not imply future MT limits. But all translators need to know what developmentents are in progress, and many simply refuse to believe advances are coming. They say "a computer is dumb" and that makes them feel good inside. I'd like to be wrong about all of this, but my desire doesn't affect what is being developed around the world.
---------------------------
Science always tries to imitate life, but it has never worked and never will. Even the best photocopy in the world is not as good as the original.
-------------------------------
Right... the airplane was a complete flop. What were
the Wright Brothers thinking?? Lots of people laughed at them, too. Then more said, "Well, that contraption only went 12 feet! What a joke." That "joke" can now travel at Mach-5 and carry 500 passengers. But look at yuor point about the photo copier. Even if it doesn't reproduce "as good as the original" it serves a useful purpose to _every_ business in operation.
-------------
Being able to translate the occasional simple sentence word for word and still have it come out right in the target language is only more proof of the above statement (the "infamous" exception to the rule).
-------------
"The occasional simple sentence..." Babelfish is much better than this, and again is a FREE, 5 -year old program.
-----------------
And if MT really comes, to the point where we're reduced to editing and re-translating, it will be even more pressing to have real professionals to do the job. In other words, a good editor will be worth his weight in gold ...
-----------------
See, with fewer personal insults, you have more time to develop an argument. This last part gets at a couple important issues. It is almost certain a wide range of editors will be used to polish MT in the future. A firm in Tokyo employs several Westerners to edit Japanese who translate into English (wrong direction, but the clients are satisfied). Only one American is good enough to translate at a high level. The rest vary from poor to decent. Yet even the worst editor there is doing tons of work, and he has no translation degree or any degree that relates to the material. You can say that this is wrong, BUT IT SEEMS TO WORK FOR THE COMPANY.
So in the future, who will the editors be?
1) The "cheapies" as you call them, because many companies will be satisfied with the output.
2) translators with a legal, science, business degree,etc. Why? Future MT output will be very quick to edit, thereby allowing many translations to be edited in the time it used to take to fully translate one from scratch. Companies who need very high accuracy will employ those who also know the law/technology as well as the language. These people are not so rare.
3) general translators who do not mind becoming an editor and can compete with both the "cheapies" and the experts who hold degrees in the field they translate in.
-yamishogun
Cheapies will never be hired as editors. That would defeat the whole purpose. Now they are hired for "stage 1 rough drafts", which are then edited by real professionals. Why would these companies use MT and then have it "edited" by people that can't even tie their shoelaces?
Looking at today's limitations and drawing conclusions from them about future limitations: well, the thing is that there is no conclusion to be drawn. Just as the sun rises in the east and just as the sky is blue, there is nothing, absolutely nothing, that programmers, scientists or programmers can do to change the fact that language (like all other human emotions and thought processes) is "intangible" - a state that is unchangeable (like the blue sky or the sun rising in the east).
As a scientist, you should understand that there are certain principles and fundamentals in nature that cannot be changed. There is more to humans (and their languages) than science can ever hope to grasp. If it all were so simple, we would have created living, walking, moving "Frankenstein" monsters a long time ago (please don't tell me now that this is what geneticists do - assembling a being from body parts and bringing the whole "creation" back to life is a fairy tale, just like (functional) MT).
We cannot bring back the dead. We cannot create life from nothing. And language is part of life, and it's also very much alive. Nothing that has to do with life itself can be reproduced in code.
To you, a scientist, a person may be nothing but cells and that's it. Sorry to disappoint you - there is a WHOLE LOT MORE. Humans cannot be "cracked" and neither can their languages.
Google is as good as it gets.
Werner George Patels
I guess we can let any translators reading this decide which quote is more likely to be correct:
"Universal translation is one of 10 emerging technologies that will affect our lives and work 'in revolutionary ways' within a decade"
(Technology Review)
"Google is as good as it gets."
-Werner George Patels
For once, I can agree with you :-)
Werner George Patels
For further reading:
"Full Machine Translation (MT) only works well with simplified source texts, or to produce a "gist" version, for information only, of documents that, for economic reasons, would otherwise never be translated at all. There is certainly no prospect in the foreseeable future of human translators being made redundant by such systems! However, the core of our module is not MT but MAT: Machine-Assisted Translation. In this, the computer provides sophisticated aids to enhance the efficiency and consistency of the translation, but the human remains in charge and makes all the decisions."
"Although Machine Translation may suffice for rendering of technical or trade documents in native languages, many people choose not to read translated works of literature, because they believe that the act of translation compromises their integrity too much. In fact, not only Art and Culture may be beyond the reach of the Language Machine: Quality is also a watchword in business, and there will continue to be some business users of language who equate customer quality care with a human touch."
"I cannot believe that a Language Machine's overall command of English will match (let alone exceed) that of an advanced learner of English, even in 20 years' time. Machine Translation is already used where a perfect translation is not essential; for example, in scanning and searching multilingual World Wide Web documents to track down information [Pringle 1998]. Such use of the Language Machine is bound to grow; but this is an addition to, not a replacement for, current international flow of information. As the Language Machine improves in accuracy and acceptability, more users will flock to it; but most of these will be new users of multilingual resources, who might otherwise have been put off by the effort of learning a new language."
(The Language Machine, by Eric Atwell, Univ. of Leeds, School of Computing)
Werner George Patels
You continue to refer to either Bablefish (2000) or quote others who discuss the current state of MT (2004).
You can find many people including yourself who say 1. Translation requires understanding 2. "computers can't understand." Ergo, 3. "computers will never be able to translate." The problem is that Babelfish, a free program that is essentially mid 1990s technology already _does_ do some translation fine, and it doesn't understand. And it is more than the "occasional simple sentence." Again, the newer techniques (post Babelfish) are being developed and will create far more accurate translations at faster speeds. This is going to shake the industry much sooner than 20 or 100 years most guess. 5 is the tremor, 10 is the quake. (The birds are already acting funny...)
A computer does not think when playing, and chess is one of mankind's greatest and most complex games, but in 1997, Deep Blue beat the greatest human player in the world. It did not think, but it won. A computer won't think , but it will translate.
"Pong is as good as it gets"
-Werner George Patels, 1979
"640k should be enough for anybody."
-Bill Gates
For sure, internet can help translators do their job very much better.
As for me myself, I sometimes communicate with my friend overseas and get help finding the most appropriate meaning.
------------------------------------------
Machine Translation:
The point is that big companies use their
own simplified and standardized syntax and
grammar as well as defined vocabularies.
They have checker programs for texts.
Then MT is not so impossible any more.
At the TTE2004 conference in Savonlinna we
were given an excellent demonstration of
such a concept at work in a big company.
- Seppo Hovila
------------------------------------------
Whether blessing or a curse, Internet is unavoidable. We oldies remember life before Internet. I have posted a topic on this in my blog, see http://raghtransint.blogspot.com/2005/09/life-before-internet.html
Regards,
Dondu
(N.Raghavan)
Post a Comment
<< Home